STP 1543-2015

Comstock | Barbéris
Zirconium in the Nuclear Industry 17th International Symposium STP1543

ASTM INTERNATIONAL
Selected Technical Papers
Zirconium in
the Nuclear
Industry
17th International Symposium
ASTM INTERNATIONAL STP 1543
Helping our world work better Editors
Robert J. Comstock
Pierre Barbéris
ISBN: 978-0-8031-7529-7
Stock #: STP1543
www.astm.org
SELECTED TECHNICAL PAPERS
STP1543
Editors: Robert Comstock, Pierre Barbéris
Zirconium in the Nuclear

Industry: 17th International
Symposium
ASTM Stock #STP1543
ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19438-2959.
Printed in the U.S.A.
Library of Congress Cataloging-in-Publication Data
ISBN: 978-0-8031-7529-7
ISSN: 1050-7558
Copyright © 2015 ASTM INTERNATIONAL, West Conshohocken, PA. All rights reserved. This material
may not be reproduced or copied, in whole or in part, in any printed, mechanical, electronic, film, or
other distribution and storage media, without the written consent of the publisher.
Photocopy Rights
Authorization to photocopy items for internal, personal, or educational classroom use, or the internal,
personal, or educational classroom use of specific clients, is granted by ASTM International provided
that the appropriate fee is paid to the Copyright Clearance Center, 222 Rosewood Drive, Danvers, MA
01923, Tel: (978) 646-2600; http://www.copyright.com/
The Society is not responsible, as a body, for the statements and opinions expressed in this publication.
ASTM International does not endorse any products represented in this publication.
Peer Review Policy

Each paper published in this volume was evaluated by two peer reviewers and at least one editor. The
authors addressed all of the reviewers’ comments to the satisfaction of both the technical editor(s) and
the ASTM International Committee on Publications.
The quality of the papers in this publication reflects not only the obvious efforts of the authors and the
technical editor(s), but also the work of the peer reviewers. In keeping with long-standing publication
practices, ASTM International maintains the anonymity of the peer reviewers. The ASTM International
Committee on Publications acknowledges with appreciation their dedication and contribution of time
and effort on behalf of ASTM International.
Citation of Papers
When citing papers from this publication, the appropriate citation includes the paper authors, “paper title”,
STP title, STP number, book editor(s), page range, Paper doi, ASTM International, West Conshohocken, PA,
year listed in the footnote of the paper. A citation is provided on page one of each paper.
Printed in Bay Shore, NY

January, 2015
Foreword
This Compilation of Selected Technical Papers, STP1543, Zirconium in the Nuclear
Industry: 17th International Symposium, contains peer-reviewed papers that were
presented at a symposium held February 3–7, 2013 in Hyderabad, India. The sym-
posium was sponsored by ASTM International Committee B10 on Reactive and Re-
fractory Metals and Alloys and Subcommittee B10.02 on Zirconium and Hafnium.
The Symposium Co-Chairmen were Pierre Barbéris, Areva/Cezus Research
Centre, Ugine, France and Srikumar Banerjee, Atomic Energy Commission, Anush-
akti Bhavan, Mumbai, India.
The STP Editors are Robert J. Comstock, Westinghouse Electric Company,
Pittsburgh, PA, USA and Pierre Barbéris.
Contents
Overview xi
Kroll Papers
Reflections on the Development of the “f ” Texture Factors for Zirconium

Components and the Establishment of Properties of the Zirconium–Hydrogen
System 3
J. J. Kearns
Displacive and Diffusional Transformations of the Beta Phase in Zirconium Alloys 23

S. Banerjee
Schemel Award Paper
Effect of Hydrogen on Dimensional Changes of Zirconium and the Influence of

Alloying Elements: First-Principles and Classical Simulations of Point Defects,
Dislocation Loops, and Hydrides 55
M. Christensen, W. Wolf, C. Freeman, E. Wimmer, R. B. Adamson,
L. Hallstadius, P. Cantonwine, and E. V. Mader
Basic Metallurgy and Alloying Effects
Phase Field Modeling of Microstructure Evolution in Zirconium Base Alloys 95

G. Choudhuri, S. Chakraborty, B. K. Shah, D. Srivastava, and G. K. Dey
Thermodynamics of Zr Alloys: Application to Heterogeneous Materials 118

P. Barberis, C. Vauglin, P. Fremiot, and P. Guerin
Influence of Sn on Deformation Mechanisms During Room Temperature

Compression of Binary Zr–Sn Alloys 138
K. V. Mani Krishna, D. G. Leo Prakash, D. Srivastava, N. Saibaba,
J. Quinta da Fonseca, G. K. Dey, and M. Preuss
Impact of Iron in M5TM 159

D. Kaczorowski, J. P. Mardon, P. Barberis, P. B. Hoffmann, and J. Stevens
Microstructure and Properties of a Three-Layer Nuclear Fuel Cladding Prototype
Containing Erbium as a Neutronic Burnable Poison 184
J. C. Brachet, P. Olier, V. Vandenberghe, S. Doriot, S. Urvoy, D. Hamon,
T. Guilbert, A. Mascaro, J. Jourdan, C. Toffolon-Masclet, M. Tupin, B. Bourdiliau,
C. Raepsaet, J. M. Joubert, and J. L. Aubin
Characterizing Quenched Microstructures in Relation to Processing 225

P. Barberis, M. T. Tran, F. Montheillet, D. Piot, and A. Gaillac
Fabrication
Identification of Safe Hot-Working Conditions in Cast Zr–2.5Nb 259

J. K. Chakravartty, R. Kapoor, A. Sarkar, V. Kumar, S. K. Jha, N. Saibaba,
and S. Banerjee
A Numerical Study of the Effect of Extrusion Parameters on the Temperature

Distribution in Zr–2.5Nb 282
N. Saibaba, N. Keskar, K. V. Mani Krishna, V. Raizada, K. Vaibhaw, S. K. Jha,
D. Srivastava, and G. K. Dey
Study on Effect of Processing on Texture Development in Zirconium-2.5 %

Niobium Alloy Tubes 302
N. Saibaba, K. Kapoor, S. V. Ramana Rao, K. Itisri, S. K. Jha, C. Phani Babu,
G. N. Ganesha, B. Prahlad, and R. K. Mistry
Numerical Modeling of Fuel Rod Resistance Butt Welding 331

A. Gaillac, D. Duthoo, C. Vauglin, D. Carcey-Collet, F. Bay, and K. Mocellin
Application of Coating Technology on Zirconium-Based Alloy to Decrease

High-Temperature Oxidation 346
H.-G. Kim, I.-H. Kim, J.-Y. Park, and Y.-H. Koo
Corrosion and Hydrogen Pickup
Oxidation Mechanism in Zircaloy-2—The Effect of SPP Size Distribution 373

P. Tejland, H.-O. Andrén, G. Sundell, M. Thuvander, B. Josefsson, L. Hallstadius,
M. Ivermark, and M. Dahlbäck
Effect of Sn on Corrosion Mechanisms in Advanced Zr-Cladding for Pressurised

Water Reactors 404
P. G. Frankel, J. Wei, E. M. Francis, A. Forsey, N. Ni, S. Lozano-Perez,
A. Ambard, M. Blat-Yrieix, R. J. Comstock, L. Hallstadius, R. Moat,
C. R. M. Grovenor, S. Lyon, R. A. Cottis, and M. Preuss
Understanding of Corrosion Mechanisms of Zirconium Alloys after Irradiation:

Effect of Ion Irradiation of the Oxide Layers on the Corrosion Rate 438
M. Tupin, J. Hamann, D. Cuisinier, P. Bossis, M. Blat, A. Ambard, A. Miquet,
D. Kaczorowski, and F. Jomard
Effect of Alloying Elements on Hydrogen Pickup in Zirconium Alloys 479

A. Couet, A. T. Motta, and R. J. Comstock
Toward a Comprehensive Mechanistic Understanding of Hydrogen Uptake
in Zirconium Alloys by Combining Atom Probe Analysis With Electronic Structure
Calculations 515
M. Lindgren, G. Sundell, I. Panas, L. Hallstadius, M. Thuvander,
and H.-O. Andrén
Corrosion and Hydrogen Uptake in Zirconium Claddings Irradiated in Light

Water Reactors 540
S. Abolhassani, G. Bart, J. Bertsch, M. Grosse, L. Hallstadius, A. Hermann,
G. Kuri, G. Ledergerber, C. Lemaignan, M. Martin, S. Portier, C. Proff, R. Restani,
S. Valance, S. Valizadeh, and H. Wiese
In Reactor Performance
Oxidation and Hydrogen Uptake of ZIRLO Structural Components Irradiated to

High Burn-Up 577
J. M. García-Infanta, M. Aulló, D. Schrire, F. Culebras, and A. M. Garde
Performance and Property Evaluation of High-Burnup Optimized ZIRLOTM

Cladding 607
G. Pan, A. M. Garde, and A. R. Atwood
Corrosion, Dimensional Stability and Microstructure of VVER-1000 E635 Alloy

FA Components at Burnups up to 72 MWday/kgU 628
V. N. Shishov, V. A. Markelov, A. V. Nikulina, V. V. Novikov, M. M. Peregud,
A. Y. Shevyakov, I. N. Volkova, G. P. Kobylyansky, A. E. Novoselov,
and A. V. Obukhov
Corrosion and Hydriding Model for Zircaloy-2 Pressure Tubes of Indian Pressurised
Heavy Water Reactors 651
S. K. Sinha and R. K. Sinha
Oxide Surface Peeling of Advanced Zirconium Alloy Cladding after High Burnup
Irradiation in Pressurized Water Reactors 673
A. M. Garde, G. Pan, A. J. Mueller, and L. Hallstadius
The Effects of Microstructure and Operating Conditions on Irradiation Creep of

Zr-2.5Nb Pressure Tubing 693
L. Walters, G. A. Bickel, and M. Griffiths
Irradiation and Hydrogen Effects
Breakthrough in Understanding Radiation Growth of Zirconium 729

S. I. Golubov, A. V. Barashev, R. E. Stoller, and B. N. Singh
Microstructural Evolution of M5TM7 Alloy Irradiated in PWRs up to High

Fluences—Comparison With Other Zr-Based Alloys 759
S. Doriot, B. Verhaeghe, J.-L. Béchade, D. Menut, D. Gilbon,
J.-P. Mardon, J.-M. Cloué, A. Miquet, and L. Legras
Modeling Irradiation Damage in Zr-2.5Nb and Its Effects on Delayed Hydride
Cracking Growth Rate 800
G. A. Bickel, M. Griffiths, H. Chaput, A. Buyers, and C. E. Coleman
Understanding the Drivers of In-Reactor Growth of β -Quenched Zircaloy-2

BWR Channels 830
J. Romero, M. Dahlbäck, L. Hallstadius, M. Ivermark, and G. Ledergerber
Impact of Hydrogen Pick-Up and Applied Stress on C-Component Loops: Toward

a Better Understanding of the Radiation Induced Growth of Recrystallized
Zirconium Alloys 853
L. Tournadre, F. Onimus, J.-L. Béchade, D. Gilbon, J.-M. Cloué, J.-P. Mardon,
and X. Feaugas
High Temperature Transient Behavior
Contribution to the Study of the Pseudobinary Zr1Nb–O Phase Diagram

and Its Application to Numerical Modeling of the High-Temperature Steam
Oxidation of Zr1Nb Fuel Cladding 897
M. Négyesi, J. Krejčí, S. Linhart, L. Novotný, A. Přibyl, J. Burda, V. Klouček,
J. Lorinčík, J. Sopoušek, J. Adámek, J. Siegl, and V. Vrtílková
Experimental Comparison of the Behavior of E110 and E110G Claddings at High

Temperature 932
Z. Hózer, E. Perez-Feró, T. Novotny, I. Nagy, M. Horváth, A. Pintér-Csordás,
A. Vimi, M. Kunstár, and T. Kemény
Effect of Pre-Oxide on Zircaloy-4 High-Temperature Steam Oxidation

and Post- Quench Mechanical Properties 952
S. Guilbert, P. Lacote, G. Montigny, C. Duriez, J. Desquines, and C. Grandjean
Deviations From Parabolic Kinetics During Oxidation of Zirconium Alloys 979

M. Steinbrück and M. Grosse
Influence of Steam Pressure on the High Temperature Oxidation and Post-Cooling

Mechanical Properties of Zircaloy-4 and M5 Cladding (LOCA Conditions) 1002
M. Le Saux, V. Vandenberghe, P. Crébier, J. C. Brachet, D. Gilbon,
J. P. Mardon, P. Jacques, and A. Cabrera
Analysis of the Secondary Cladding Hydrogenation During the Quench–LOCA

Bundle Tests With Zircaloy-4 Claddings and its Influence on the Cladding
Embrittlement 1054
M. Grosse, J. Stuckert, C. Roessger, M. Steinbrueck, M. Walter, and A. Kaestner
Degradation and Failure Mechanisms
Effect of Hydride Distribution on the Mechanical Properties of Zirconium-Alloy

Fuel Cladding and Guide Tubes 1077
S. K. Yagnik, J.-H. Chen, and R.-C. Kuo
Mechanisms of Hydride Reorientation in Zircaloy-4 Studied in Situ 1107
K. Colas, A. Motta, M. R. Daymond, and J. Almer
Hydriding Induced Corrosion Failures in BWR Fuel 1138

D. Lutz, Y.-P. Lin, R. Dunavant, R. Schneider, H. Yeager, A. Kucuk, B. Cheng,
and J. Lemons
Author Index 1173
Subject Index 1177

Overview
This STP contains the papers presented at the 17th International Symposium on Zirco-
nium in the Nuclear Industry held in Hyderabad, Andhra Pradesh, India from Febru-
ary 3–7, 2013. The first symposium was held in Philadelphia in 1968 with subsequent
symposia held every two to three years. The proceedings of each symposium in the
series have been documented with an STP.
During this symposium, the William J. Kroll Zirconium Medal was presented to
John Kearns (2011 winner) and Srikumar Banerjee (2012 winner) for their unique
and lasting contributions to the technology of zirconium alloys. Both provided his-
torical perspectives of their research during the symposium and contributed papers
that are included in the STP.
The symposium was truly international; with approximately 130 participants
from 17 countries attending and representation from North and South America, Eu-
rope, and Asia. The 17th Symposium included 42 platform presentations along with
a session with 31 posters. This STP contains 38 peer reviewed papers along with the
papers from the two Kroll winners. In addition, the discussion of each platform pres-
entation provided an opportunity for further insight and understanding of the paper.
As in past symposia, the questions along with written responses by the authors are
included at the end of each paper.
The symposium is an opportunity to capture a snapshot of the current research
areas that are relevant to the nuclear industry. This symposium was no exception.
While the papers are grouped into seven categories (Basic Metallurgy and Alloying
Effects, Fabrication, Corrosion and Hydrogen Pickup, In Reactor Performance, Irra-
diation and Hydrogen Effects, High Temperature Transient Behavior, and Degradation
and Failure Mechanisms), there is often overlap between the topics as you will see
when you browse through the STP or delve into papers more deeply.
An important component of the symposium is the in-reactor performance of zir-
conium alloys with several papers presenting recent results from materials irradiated
to high burnups and fluence.
• Alloy E635 was irradiated to 72 MWd/kgU in VVER-1000 reactors with per-
formance data presented from both fuel cladding and structural components.
The behavior of the material was correlated to both temperature and neutron
fluence.
• The evolution of the microstructure of M5™ fuel rod cladding irradiated in
a PWR up to 7 cycles with fast fluence up to 17.1 × 1025 n/m2 was described.
In a separate paper, results from M5™ with 1000 ppm Fe (designated M5-Fe)
xi
included both oxide thickness measurements at burnups of about 65 MWd/tU
and free growth at fluences to 20 × 1025 n/m2 (E > 1Mev).
®
• ZIRLO structural components were characterized for both corrosion and di-
mensional changes following irradiation in PWRs with a maximum fluence of
13.6 × 1025 n/m2 (E > 1Mev).
• Optimized ZIRLO™ fuel cladding was irradiated beyond the license limit of
62 GWd/MTU to 70 GWd/MTU. Characterization of the cladding included
corrosion, dimensional changes, and mechanical properties.
• A detailed study was presented on the influence of temperature and micro-
structure on the irradiation creep behavior of Zr-2.5Nb pressure tubes.
A dominant theme in this STP is the role of hydrogen on the performance of zir-
conium alloy components. Issues discussed in this volume where performance was
dominated by hydrogen included the following:
• Failure of BWR fuel rods was attributed to the localization of hydrides fol-
lowing accelerated corrosion and subsequent cracking of the hydride lenses.
Despite an extensive investigation, the cause of the accelerated corrosion was
not definitively identified.
• The growth of beta-quenched Zircaloy-2 BWR channels was driven late in life
by accelerated hydrogen pickup that coincided with the dissolution of second
phase particles.
• As reorientation of hydrides plays an important role during dry storage, in-situ
measurements were performed to gain new insights into the reorientation of
hydrides in Zircaloy-4.
• Delayed hydride cracking (DHC) growth rate of in-service Zr-2.5Nb CANDU
pressure tubes was controlled by thermal and irradiation effects on the micro-
structure (e.g, decomposition and reconstitution of the beta phase controlling
hydrogen diffusion to the crack tip).
In addition to papers that highlight the impact of hydrogen, several papers fo-
cused on understanding the mechanisms of hydrogen ingress into the metal or
understanding the interaction of hydrogen with point defects and dislocation loops
in the matrix. The latter has potential implications related to hydrogen assisted ir-
radiation growth.
The US Nuclear Regulatory Commission proposal of a rule to amend the current
requirements governing emergency core cooling system for light nuclear power reac-
tors has prompted renewed activity in cladding ductility following a high tempera-
ture transient. Papers addressed different aspects of the high temperature oxidation
behavior of cladding, including secondary hydriding following rod burst, the detri-
mental role of nitrogen on oxidation kinetics, and the impact of a pre-oxide or steam
pressure on oxidation and subsequent cladding ductility. One paper demonstrated
xii
the improved oxidation performance of E110G relative to E110. Unfortunately, the
reason for the improvement remains an area for continued research.
In response to the Fukushima Daiichi nuclear accident in March, 2011, significant
attention has been given to improving the accident tolerance of zirconium alloy fuel
cladding. Researchers presented one approach through the application of coating
technology to improve the high temperature oxidation resistance of the cladding.
While significant development work remains, this is an area that will likely receive
continued attention to identify viable options.
The zirconium community continues to push the limits of state-of-the-art ana-
lytical techniques to characterize the microstructure of both non-irradiated and ir-
radiated zirconium alloys. Techniques such as synchrotron radiation (e.g., μ x-ray
absorption near edge spectroscopy, diffraction, and stress measurement), electron
back-scattered diffraction, atom probe tomography, secondary ion mass spectros-
copy, and electron energy loss spectroscopy have become routine analytical tools
in multiple laboratories around the world. Cold neutron prompt gamma activation
analysis was successfully used in one study to non-destructively measure hydrogen
content in corrosion coupons. Complementing the analytical characterization tech-
niques are modelling efforts designed to facilitate mechanistic understanding of per-
formance phenomena.
Following the symposium, a committee of technical experts covering a breadth
of experience in the zirconium nuclear industry selected the best paper based upon
technical excellence, relevance to the nuclear industry, and ‘groundbreaking’ research.
The winner of the John H. Schemel Best Paper Award was the paper entitled “Effect of
Hydrogen on Dimensional Changes of Zirconium and the Influence of Alloying Ele-
ments: First-Principles and Classical Simulations of Point Defects, Dislocation Loops,
and Hydrides” by M. Christensen, W. Wolf, C. Freeman, E. Wimmer, R. B. Adamson,
L. Hallstadius, P. Cantonwine, and E. V. Mader. Congratulations to the winners.
Robert J. Comstock
Pierre Barbéris
xiii
KROLL PAPERS
ZIRCONIUM IN THE NUCLEAR INDUSTRY: 17TH INTERNATIONAL SYMPOSIUM 3
STP 1543, 2015 / available online at www.astm.org / doi: 10.1520/STP154320120216
J. J. Kearns1
Reflections on the Development

of the “f ” Texture Factors for
Zirconium Components and the
Establishment of Properties of
the Zirconium–Hydrogen System
Reference
Kearns, J. J., “Reflections on the Development of the “f ” Texture Factors for Zirconium
Components and the Establishment of Properties of the Zirconium–Hydrogen System,”
Zirconium in the Nuclear Industry: 17th International Symposium, STP 1543, Robert Comstock
and Pierre Barberis, Eds., pp. 3–22, doi:10.1520/STP154320120216, ASTM International, West
Conshohocken, PA 2015.2
ABSTRACT
Results of work performed mostly by this author at the Bettis Atomic Power
Laboratory in the period from the 1950s through 1976 are combined with the
notable work of many others in the field to illustrate some of the metallurgical
properties important to the safety and reliability of reactor core components of
zirconium and its alloys. As implied by the title, this paper is not a review article
but a personal recollection of why the work in published papers by this author
was started and includes the challenges and some early problems and fortunate
decisions to provide historical context to this document.
Keywords
zirconium, texture, anisotropy, hydrogen, embrittlement, solubility, diffusion
coefficients, thermal-diffusion
Manuscript received December 26, 2012; accepted for publication May 30, 2013; published online September
26, 2014.
1
Retired from Bettis Atomic Power Laboratory, West Mifflin, PA 15122, United States of America.
2
ASTM 17th International Symposium on Zirconium in the Nuclear Industry on February 3–7, 2013 in
Hyderabad, India.
Copyright V
C 2014 by ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959.
4 STP 1543 On Zirconium in the Nuclear Industry
Introduction
This paper provides, primarily, a summary of work by this writer on the physical
metallurgy of zirconium alloys that was conducted at the Bettis Atomic Power Lab-
oratory (henceforth, Bettis) in the period from the late 1950s to 1976. Also included
are the highlights of a paper published in 2001 [1], which provided an extension of
the “f” texture concepts introduced in 1965 [2].
There were many worthy publications by others in the zirconium field, some of
which are cited here for their significant impact. The comprehensive coverage of
other work is provided in my original publications. Any oversights are regretted.
With the exception noted above and a few Bettis papers, what is not covered
are publications by others after 1976, when I was transferred from reactor core
materials and assigned work on plant materials.
Development of the “f” Texture Factor

The following mathematical development converts the distribution of the poles of a
given crystallographic plane displayed in stereographic projections of direct or
inverse pole figures into the effective fraction in each of the three principal direc-
tions of fabricated material. In zirconium and other important metals with the hex-
agonal crystal structure, such as hafnium and titanium, the most useful reference
pole is the basal pole, [0001], because it is most easily related to important proper-
ties. Because the document [2] on the original work is not easily obtained, the prin-
cipal portions of that report are provided herein.
The quantitative relationship that can be demonstrated between “f” factors and
properties refers to those in a class including thermal expansion and neutron dam-
age induced irradiation growth. However, the “f” factor and other texture indices
have found widespread use for correlations and material characterizations in work
on other texture-dependent properties of hexagonal materials. They are cited in Ref
1. Another example of “f” factor use is the work of Cook et al. [3] who used “f” to
monitor effects of variations in fabrication processes.
For calculations based on x-ray data, it is necessary to assume that the gross or
bulk property in a reference direction of a polycrystalline sample is the weighted
summation of this property in its individual crystals. For single hexagonal crystals,
the contribution to the bulk property depends on the angle between the reference
direction and the [0001] crystal direction, according to the following equation [4]:
(1) Pref ¼ P½0001 cos2 / þ P?½0001 ð1 cos2 /Þ
where Pref is the property in the reference direction, P[0001] and P\[0001] are the
single-crystal values parallel and perpendicular to [0001], and / is the angle
between the reference direction and [0001]. Thus, if the volume fraction, Vi, of crys-
tals oriented with their [0001] at a tilt angle, /i, to the reference direction is known
from x-ray data, Eq 1 gives their contribution to the property of the total sample.
KEARNS, DOI 10.1520/STP154320120216 5
Summation from /i ¼ 0 to /i ¼ p/2 gives the bulk property in the reference direc-
tion according to Eq 2
(2) Pref ¼ P½0001 RVi cos2 /i þ P?½0001 RVi P?½0001 RVi cos2 /i
Because RVi is unity,

(3) Pref ¼ P½0001 RVi cos2 /i þ P?½0001 1 RVi cos2 /i
An important step and fortunate choice in the original analysis was to define the
summation as an orientation parameter, f, as follows:
(4) f ¼ RVi cos2 /i
Insertion of this definition into Eq 3 produces the following equation that embodies
both the simplicity and exceptional utility of the f parameter:
(5) Pref ¼ fP½0001 þ ð1 f ÞP?½0001
Thus, “f” is a single number from which properties may be calculated from
single-crystal values. It is also useful as a quantitative index of the effective tex-
ture. As already stated, calculations of “f” will resolve the ordinary scatter of
[0001] orientations into an effective fraction aligned in each of the three princi-
pal directions of fabricated material. Values of “f” of zero and 1.0 indicate perfect
alignment of the [0001] crystal directions perpendicular or parallel, respectively,
to the direction of interest. It can be shown that the sum of “f” in the three
principal directions must be unity and that a value of 1/3 in each direction
defines the isotropic case.
A plot of Eq 5 is shown in Fig. 1 for four rolled plates of Zircaloy. The symbols
in the low, intermediate, and high “f” regions represent the longitudinal, transverse,
and normal directions, respectively. This figure represents the results of analytical
and experimental work that started in the late 1950s and culminated in the 1965
report in Ref 2.
Initial results were not so impressive. In fact, the first “f” values calculated with
Eq 4 were not sensible. There was nothing wrong with Eq 4. The failure was traced
to the omission of a geometric correction, the sin/ correction described by Cullity
and Freda [5].
As illustrated in Fig. 2, the volume fraction of poles in a D/ band depends on
the average density over 360 of the angle a and the size of the band, which is pro-
portional to sin/. In our initial calculation only the average pole density was used.
When the size of the band was added, everything fell into place. The volume frac-
tion of metal in a D/ band Vi in Eq 4 is then given by:
FIG. 1 Correlation of expansion and x-ray preferred orientation measurements.
ð /2
Ið/Þ sin / d/
/
(6) Vi or VDi / ¼ ð p=2
1
Ið/Þ sin / d/
0
If the intensity, I/, is expressed in times random units, the denominator of Eq 6 is

unity and Eq 4 can be expressed as
ð p=2
(7) f ¼ I/ sin / cos2 / d/
0
X-RAY METHOD FOR CALCULATION OF “F” FACTORS

All of the “f” calculations in Ref 2 were based on the intensities of the diffraction
reflecting planes of zirconium taken from simple 2h diffractometer traces. These
planes are identified in the standard projection of Fig. 3(b). The intensities were nor-
malized to multiples of random units by comparison with data from a pressed pow-
der random sample.
KEARNS, DOI 10.1520/STP154320120216 7
FIG. 2 (0001) Poles of random sample intersecting reference sphere.
FIG. 3 (a) Tilt between {0001} and {hki‘}, typified by {101‘}, and (b) standard projection
of diffracting planes of alpha zirconium, c/a ¼ 1.59.
FIG. 4 Intensity data from swaged rod as a function of tilt angle, /.
Because the reflecting planes are available at discrete points only along three
positions of the 30 a rotation, development of inverse pole figures required an esti-
mation procedure covering the entire space of the pole figure. This was accom-
plished by first plotting the intensities versus the tilt angle / from zero to 90 along
the three a lines as shown for example in the upper panels of Fig. 4. If it is assumed
that the intensity varies linearly with a in the 19.1 interval between 101‘ and 213‘
and the 10.9 between 213‘ and 112‘, then the average intensity at / angles above
58.3 is given by:
KEARNS, DOI 10.1520/STP154320120216 9

1 I1 þ I2 I2 þ I3
(8) I/ ¼ 19:1 þ 10:9
30 2 2
where I1, I2, and I3 refer to the intensities of 101‘, 213‘, and 112‘ curves, respec-
tively, at a fixed / angle. At / angles between 38.5 and 50.6 , a simple average of
101‘, and 112‘, curves is used. Below 38.5 , only the 101‘ curve is defined. There-
fore, it is necessary to assume that it represents the average value in this region. The
effect of the sparsity of intensity data in the upper region of the pole figure is greatly
mitigated by the sin / correction, which is shown clearly in the lower left panel of
Fig. 4. The reliability of the x-ray data processing described here is confirmed by
several measures, including the tight correlation with thermal expansion shown in
Fig. 1 and the required summation to unity of the “f” values in the three principal
directions of the fabricated material in this study. This measure varied from 0.961
to 1.03, with an average of 0.989.
The effectiveness of the inverse pole figure method that this study demon-
strated is a tribute to the researchers who contributed to its development. Six cita-
tions are provided in Ref 2.
The steps in the calculation of “f” factors are shown for example in Table 1.
TABLE 1 Example of “f” factor calculations for a swaged rod.
a
D/ I/ (avg.) I/ sin / VD/ VD/ cos2 /
Longitudinal section
0–10 3.27 0.285 0.048 0.0478
10–20 2.71 0.702 0.119 0.11
20–30 1.69 0.715 0.121 0.0993
30–40 1.35 0.775 0.131 0.0879
40–50 1.17 0.827 0.140 0.0700
50–60 0.97 0.794 0.134 0.0441
60–70 0.73 0.661 0.117 0.0209
70–80 0.62 0.600 0.101 0.0061
80–90 0.55 0.548 0.093 0.0007
5.91 1.00 0.488 ¼ f
Transverse section
0–50 0 0 0 0
50–60 0.11 0.090 0.014 0.0046
60–70 0.85 0.770 0.116 0.0208
70–80 2.43 2.35 0.353 0.0214
80–90 3.48 3.47 0.521 0.0040
6.68 1.00 0.0508 ¼ f
a
Defined by Eq 6.
LATER EXTENSIONS OF “f” FACTOR CONCEPTS

The work of Ref 1 examined published relationships among “f” factors for the basal
plane and the prism planes in the hexagonal crystal structure and introduced an
expression between f0002 and “f” for any plane in the system. This work also pro-
vided an inverse pole figure method for determining “f” factors for the prism
planes.
Kelly and Watson [6] observed that
(9) f0002 þ 2f1010 ¼ 1
and Bowen [7] found that
(10) f0002 þ f1010 ¼ f1120 ¼ 1
Through a single-cell model described in [1], f1010 and f1120 were shown to be equal
and, therefore, Eqs 9 and 10 are equivalent, at least analytically.
The expression that was formulated in [1] to determine “f” for any plane in the
system is the following:
1
(11) fhkil ¼ f0002 ð3 cos2 / 1Þ þ sin2 /
2
where f0002 is the value in the reference direction and / is the angle between the
{hkil} pole and the basal pole. Further analysis in Ref 1 shows that computation of
“f” for any one of the three principal planes is sufficient for evaluation of the other
two, or for any plane through the use of Eq 11.
An inverse pole figure method for the prism planes was developed in [1]
because in a round robin study on the consistency of texture measurements
reported by Lewis et al. [8], the only “f” values for the prism planes in the study
were obtained from direct pole figure measurements.
The reader is referred to Ref 1 for details of this method along with evaluation
of its reliability.
Establishment of Properties of the

Zirconium–Hydrogen System
HYDROGEN EMBRITTLEMENT
One of the ultimate goals of the voluminous work performed by the zirconium
community on the physical and mechanical properties of zirconium and its alloys
was the understanding and mitigation of hydrogen embrittlement in reactor core
components. Much of this work was spawned by two remarkable publications in
1963 by Marshall and Louthan [9,10]. That hydrogen lowers the ductility of zirco-
nium and its alloys has been known since as early as 1952. But how drastic this
effect could be at low hydrogen levels was not appreciated until Marshall and Lou-
than found that as little as 40 ppm could produce completely brittle behavior if the
KEARNS, DOI 10.1520/STP154320120216 11
hydrides were precipitated perpendicular to the axis of tensile specimens. This dis-
covery was doubly alarming because these workers found further that hydrides
tended to assume this orientation if the specimens were stressed during the process
of precipitation. In other words, the hydride platelets could be “stress-oriented.”
Office mate and friend Bill Babyak and I became quite aware of this work when
our Reactor Metallurgy manager, Dr. Ben Lustman, rushed into our office with the
first of the two papers in hand and sensed that the second was coming.
Ben assigned Bill the job of determining the habit plane of the hydride phase
and me other characteristics of potential importance.
Dr. Lustman not only assigned the work, he later contributed a theory on grain
size effects. He was that kind of manager.
Babyak [11] found a limited region for the habit plane, 5 to 25 from the basal
plane. Westlake in 1968 [12] found a (1017) habit plane, consistent with Babyak’s
results.
Habit planes close to the basal plane fit well with the effect of texture that was
found in the extensive study by Kearns and Woods [13] as indicated in the follow-
ing list of conclusions taken from that publication:
1. In annealed, unstressed samples, the orientation distribution of hydride plate-
lets is related generally to the crystallographic texture. The fraction of hydrides
with a given orientation in a region increases with the alignment of basal
planes (0001) in that region. This relationship is consistent with the hydride
habit planes in Zircaloy-4.
2. In samples stressed during precipitation, the orientation of the hydrides
depends on grain size and residual cold work, as well as texture. The stress
effect, that is, the change in orientation caused by stress, increases with
increasing cold work and decreasing grain diameter. The final “stressed” orien-
tation, therefore, depends on all three factors.
3. The basal texture-hydride orientation relationship and the general characteris-
tic in Zircaloy of basal planes being aligned parallel to positive strains and per-
pendicular to negative strains suggest that radially oriented hydrides in tubing
can be minimized by fabrication processes that produce relatively high values
of wall reduction.
4. Stress orientation is associated only with precipitation. Reorientation by disso-
lution of existing hydride and re-precipitation under isothermal conditions is
not to be anticipated.
The first and second conclusions are demonstrated in Figs. 5 and 6.
The first of these conclusions was further substantiated by later Bettis work
[14] in which micro-sphere indentations were made in individual grains of the pol-
ished transverse cross section of Zircaloy tubing that had been quenched to pro-
mote intragranular hydride precipitation. When the basal plane is parallel to the
polished surface, a spherical indenter produces a circular impression; but as the ba-
sal plane tilts out of the surface, the impression takes on an oval shape with the
long axis coinciding with the trace of the basal plane in the surface. As is clear in
Fig. 7, the intragranular hydride traces are near the basal plane traces because in
FIG. 5 Correlation of hydride orientations with basal pole textures in unstressed

samples.
typically fabricated Zircaloy tubing greater than 90% of the basal plane poles are
within 30 of the transverse surface, and the majority of these are within 10 [13].
Whereas the discussion here is centered on where the hydride habit
plane resides, the reader is referred to the original paper [14] for implications on
the process of stress-oriented hydride precipitation. The position put forth in Ref
14 is that the hydrides will preferentially precipitate in grains where the habit plane
is closest to being perpendicular to the direction of the applied stress.
HYDROGEN SOLUBILITY
Early published studies on this subject revealed considerable variability in the ter-
minal solid solubility (TSS). This appeared to be caused largely by the high capacity
of zirconium to supersaturate as observed by Erickson and Hardie [15]. Because of
KEARNS, DOI 10.1520/STP154320120216 13
FIG. 6 Effect of grain size on susceptibility to stress-oriented hydride precipitation.
this characteristic, the Bettis tests on TSS [16] were conducted by a method that
was simple, accurate, and, most importantly, immune to supersaturation.
The method consisted of determining the hydrogen concentration in the low
hydrogen side of diffusion couples made by resistance-welding hydride-bearing
samples to opposite ends of hydrogen-free samples and annealing to equilibrium in
the temperature range 260 C to 525 C. The hydrogen content of the high hydro-
gen part of the couples was sufficient (500 to 2000 ppm) to maintain a two-phase
mixture of hydride and saturated alpha solid solution during the diffusion anneal.
In this method, supersaturation of the low hydrogen part of the couple is avoided
because diffusion raises the level only to that of the alpha phase in the two-phase
mixture. To assure that the alpha phase in the mixture was not supersaturated, the
hydride samples were slowly cooled to room temperature prior to welding. Equilib-
rium was, therefore, approached on heating during the subsequent diffusion anneal.
These anneals, which were run in air, ranged from 10 days at 260 C to 2 days at
525 C. The diffusion parameter, Dt/‘2 , was always equal to or greater than 2, which
FIG. 7 Illustration of oval impressions made by a micro-sphere indenter on the

transverse section of Zircaloy-4 tubing (at 150 magnification).
assured that the hydrogen concentration was within 1% of its ultimate value. In this
parameter, D is the diffusion coefficient, t is the time, and ‘ is the specimen length.
This procedure assured that the hydrogen concentration in the low hydrogen
side of the couple represented the equilibrium TSS, free of any supersaturation.
Results of the Bettis TSS tests are compared with other published data in Figs. 8 and
9. Whereas the solubility originally derived by Sawatzky’s elegant thermal diffusion
analysis did not agree well with the Bettis diffusion couple results, later work by
Sawatzky and Wilkins with the thermal diffusion method [17] did agree. This later
work by Sawatzky also revealed little variation in the TSS among pure zirconium,
Zircaloy-2, and Zr-2.5 wt. % Nb. The Bettis work in Ref 16 also examined the likeli-
hood of partitioning of hydrogen in composites of zirconium and its alloys. This ex-
amination concluded that partitioning was unlikely to exceed 10%.
SUPERSATURATION AND HYDROGEN REDISTRIBUTION

Testing conducted at Bettis in the early 1970s on specimens precharged to obtain
uniformly distributed hydride phase, and then subjected to a thermal gradient,
KEARNS, DOI 10.1520/STP154320120216 15
FIG. 8 Compilation of terminal solubility data below 550 C for unalloyed zirconium
taken from Ref 16. Solid line is the best-fit linear curve.
failed to reveal evidence of redistribution, whereas hydrogen movement and heavy

hydride precipitation was expected at the cooled surfaces.
In the same period, a thermal gradient test by Maki [18] revealed only par-
tial redistribution. In this test, the temperature drop, DT, was 41 F, 586 F to
545 F. In the Bettis testing, it was 29 F, 593 F to 564 F. The conclusion drawn
from these tests is that significant redistribution, that is, precipitation of hydride
FIG. 9 Compilation of terminal solubility data below 550 C for Zircaloy-2 and Zircaloy-
4 taken from Ref 16. Solid line is best-fit linear curve through the Bettis data.
phase in the lower temperature region, requires the hydrogen activity in the
alpha phase at the higher temperature side to cause the limit of supersaturation
in the lower temperature region to be exceeded. The results in both tests are con-
sistent with the separation of the equilibrium and metastable solubility curves in
Figs. 8 and 9.
KEARNS, DOI 10.1520/STP154320120216 17
Whereas some hydrogen movement will occur across any thermal gradient
consistent with the heat of transport, significant build-up by hydride precipitation
will commence only when a critical DT is exceeded.
A notable aspect of these tests is that the metastable supersaturation exists even
in the presence of existing hydride phase. As a result of this early work, the hydro-
gen distribution expectations were modified to account for the dual solubility char-
acteristic of zirconium alloys.
The Bettis testing was later extended and reported first by Kammenzind and
Franklin in Ref 19. Results of related studies of hydrogen migration were reported
by Kammenzind [20] and Johnson [21].
DIFFUSION COEFFICIENT IN ALPHA ZIRCONIUM, ZIRCALOY-2, AND

ZIRCALOY-4
Prior to the Bettis study [22], the diffusion coefficient was determined by Sawatzky
in and Zircaloy-2 in unalloyed zirconium. Though the variability among these
determinations was not exceptionally large, it was large enough to raise concern
that material composition and perhaps metallurgical structure were affecting the
rate of hydrogen diffusion in these materials. Therefore, experiments were
conducted to examine the influence of these factors. Because there was some
evidence of anisotropy of diffusion, the effect of texture was also evaluated in the
Bettis work by utilizing specimens taken from the three principal directions in
rolled plate.
Diffusion coefficients were obtained by a conventional diffusion couple method
that consisted of measuring the hydrogen concentration along the length of diffu-
sion samples made by resistance-welding hydrogen-bearing studs to hydrogen-free
studs. After welding, the specimens were machined to smooth cylinders and
diffusion-annealed in air. The anneals were conducted in air purposely so that an
oxide film would form and act as a barrier to hydrogen loss from the surface. The
diffusion periods, which ranged from 90 min at 700 C to 5 days at 275 C, were
selected to produce an essentially constant diffusion zone of 3 cm.
An example of the well-behaved concentration versus distance in diffusion
zones is shown in Fig. 10 for a Zircaloy-2 specimen consisting of a cold worked ma-
terial on the low hydrogen side and annealed material on the high hydrogen side.
No effect of cold work is evident.
A comparison of Bettis results with earlier work is illustrated in the diffusion
coefficient, D, versus reciprocal temperature Arrhenius plot in Fig. 11. Because a tex-
ture effect was found, best-fit curves were drawn through the Bettis data represent-
ing the principal directions in fabricated plate.
The corresponding Arrhenius equations for, respectively, the longitudinal,
transverse, and through-thickness D values are, in cm2/s,

10; 830
(12) D ¼ 7:73 103 exp
RT
FIG. 10 Probability plot of concentration–penetration curve in a diffusion couple

composed of annealed and cold worked Zircaloy-2. Constant slope signifies
equal D values for each side.

3 10; 290
(13) D ¼ 5:84 10 exp
RT

10; 730
(14) D ¼ 7:90 103 exp
RT
The curves in Fig. 11 show that Someno’s results are closest to those of Bettis, likely
because the method was the same in the two studies.
The Bettis data supported the following conclusions [22]:
1. Within the limits of experimental error, the diffusion coefficient of hydrogen
in alpha zirconium is independent of grain size, cold work, and the composi-
tional variations encompassing Zircaloy-2 and Zircaloy-4.
2. The effects of the variations in crystallographic texture common in rolled sheet
are also small. Diffusivity of hydrogen is greater parallel to the c axis than per-
pendicular, but by a factor probably not >2.
The last statement is the result of an analysis akin to that relating texture and
thermal expansion in the first section of this document.
Because hydrogen diffusion is so rapid, the use of a diffusion coefficient de-
pendent only on temperature would likely be sufficiently accurate for most engi-
neering analyses. For such cases, the following relationship is provided, as in Ref 22.
KEARNS, DOI 10.1520/STP154320120216 19
FIG. 11 Comparison of Bettis data with earlier work on the diffusion of hydrogen in
alpha zirconium.

10; 620
(15) D ¼ 7:16 103 exp cm2 =s
RT
The exponential and pre-exponential constants are averages of those in Eqs 12

through 14.
ACKNOWLEDGMENTS
The writer was surrounded at Bettis by many people who were at the top both profes-
sionally and socially and contributed much to my work, which the writer considered
to be joyful. In the period covered by this paper, 1950s to 1970s, the people in man-
agement to be credited, in particular, are Kroll Medal recipient Ben Lustman who
interviewed (grilled) me, hired me, and kept me in his reactor metallurgy group for
many years, even after a slow start. In those early years, the writer was privileged to be
supervised by Don Thomas, a gentleman who looked after me and taught me (not so
gently) how to improve my technical writing, such as not “fast, or slow rate,” but
“high and low rate,” eliminating the redundancies many of us often use. Don and fel-
low manager Ken Goldman hold the Zircaloy patents. Another early supervisor was
George Salvaggio whose welcome style was to give an assignment then disappear,
except to drop in on Monday mornings to talk about the Sunday Steeler’s game. The
excellent management string continued through the 1970s with supervisor Jim
McCauley, Subdivision Manager and Kroll medal recipient, Fred Nichols, and Project
Manager, Ralph Frederickson. The engineering colleagues and friends in this period
were my first office mate Will Mudge who had lost a leg in an auto accident near Bet-
tis and later his life in another after leaving Bettis. Of special note was Bill Babyak
who was a technician when the writer started in 1952, but was an office mate later af-
ter receiving his degree at night, and then rose to positions in upper management
before retiring. As an office mate, we had many discussions trying to figure out the
puzzles of texture analysis. He unselfishly declined to be co-author of the 1965 report
on the “f” factor texture analysis. To have people around with the analytical talent of
Pete Kreyns and George Marino was very helpful to all of the engineering groups at
Bettis. Others of note were close friends and associates willing to offer their perspec-
tive on technical and other issues. These included Ted Chleboski, Ed Hillner, and
Owen Katz. Ed and Owen helped keep us healthy by walking the Bettis grounds every
day at lunch time. Also to be thanked are the dedicated technicians, who do not get
the recognition that those authoring reports do, but are so important to the experi-
mentalist. Those helping me were, in chronological order, Kenny Marsh, Hal Garvis,
Bob Rubino, and Norm Ferri. Helping everyone was Jim Korinko, who rose from
technician to engineer. If you had a problem with anything, particularly mechanical
or electronic, Jim is the guy you went to see. Finally, the help of Bruce Kammenzind
and Wendy Steffan of Bettis, and my son, James Kearns of York College of Pennsylva-
nia, was essential in the preparation of this document in a form suitable for
publication.
References
[1] Kearns, J. J., “On the Relationship Among “f” Texture Factors for the Principal Planes
of Zirconium, Hafnium, and Titanium Alloys,” J. Nucl. Mater., Vol. 299, 2001, pp.
171–174.
KEARNS, DOI 10.1520/STP154320120216 21
[2] Kearns, J. J., “Thermal Expansion and Preferred Orientation in Zircaloy,” Report WAPD-
TM-472, Bettis Atomic Power Laboratory, West Mifflin, PA, 1965, pp. 1–35.
[3] Cook, C. S., Sabol, G. P., Sekera, K. R., and Randall, S. N., “Texture Control in Zircaloy Tub-
ing Through Processing,” Zirconium in the Nuclear Industry: Ninth International Sympo-
sium, ASTM STP 1132, C. M. Eucken and A. M. Garde, Eds., ASTM International, West
Conshohocken, PA, 1991, pp. 80–95.
[4] Boas, W. and Mackenzie, J. K., “Anisotropy in Metals,” Progress in Metal Physics, Vol. 2,
B. Chalmers, Ed., Interscience, New York, 1950, pp. 90–120.
[5] Cullity, B. D. and Freda, A., “Quantitative Method for the Determination of Fiber Texture,”
J. Appl. Phys., Vol. 29, 1958, pp. 25–30.
[6] Kelly, P. M. and Watson, K. G., “A Simple Method for Determining Pole Figures of Zirco-
nium Alloy Tubing,” J. Nucl. Mater., Vol. 44, 1972, pp. 71–78.
[7] Bowen, A. W., “Quantitative Texture Analysis of Alpha Based Titanium and Titanium
Alloys—A Two Stage Process,” Titanium ’92, Science and Technology, F. H. Froes and I.
Caplan, Eds., Minerals, Metals and Materials Society, Warrendale, PA, 1993, pp. 271–276.
[8] Lewis, J. E., Schoenberger, G., and Adamson, R. B., “Texture Measurement Techniques
for Zircaloy Cladding: A Round-Robin Study,” Zirconium in the Nuclear Industry: Fifth
Conference, ASTM STP 754, D. G. Franklin, Ed., ASTM International, West Conshohocken,
PA, 1982, pp. 39–62.
[9] Marshall, R. P. and Louthan, M. R., Jr., “Tensile Properties of Zircaloy With Oriented
Hydrides,” ASM Trans. Quart., Vol. 56, 1963, p. 693.
[10] Louthan, M. R., Jr. and Marshall, R. P., “Control of Hydride Orientation in Zircaloy,” J.
Nucl. Mater., Vol. 9, 1963, p. 170.
[11] Babyak, W. J., “Hydride Habit Plane in Zirconium and in Unstressed and Stressed
Zircaloy,” Trans. TMS-AIME, Vol. 239, 1967, p. 252.
[12] Westlake, D. G., “The Habit Planes of Zirconium Hydride in Zirconium and Zircaloy,” J.
Nucl. Mater., Vol. 26, 1968, p. 208.
[13] Kearns, J. J. and Woods, C. R., “Effect of Texture, Grain Size, and Cold Work on the Pre-
cipitation of Oriented Hydrides in Zircaloy Tubing and Plate,” J. Nucl. Mater., Vol. 20,
1966, pp. 241–261.
[14] Kearns, J. J., “Micro-Sphere Probe of the Crystallographic Orientation of Hydrides in

Fine Grained Zircaloy Tubing,” J. Nucl. Mater., Vol. 33, 1969, pp. 114–118.
[15] Erickson, W. H. and Hardie, D., “Influence of Alloying Elements on the Terminal Solubility
of Hydrogen in Alpha Zirconium,” J. Nucl. Mater., Vol. 13, 1964, pp. 254–264.
[16] Kearns, J. J., “Terminal Solubility and Partitioning of Hydrogen in the Alpha Phase of Zir-
conium, Zircaloy-2, and Zircaloy-4,” J. Nucl. Mater., Vol. 22, 1967, pp. 292–303.
[17] Sawatzky, A. and Wilkins, B. J. S., “Hydrogen Solubility in Zirconium Alloys Determined
by Thermal Diffusion,” J. Nucl. Mater., Vol. 22, 1967, pp. 304–310.
[18] Maki, H., “Effect of Stress on Hydride Precipitation Behavior in Zircaloy-2,” J. Nucl. Sci.
Technol., Vol. 10(8), 1973, p. 470.
[19] Kammenzind, B. F. and Franklin, D. G., “Hydrogen Pickup and Redistribution in Alpha
Annealed Zircaloy-4,” 11th International Symposium on Zirconium in the Nuclear Indus-
try, ASTM STP 1295, ASTM International, West Conshohocken, PA, 1996, pp. 338–370.
[20] Kammenzind, B. F., “The Long Range Migration of Hydrogen Through Zircaloy-4 in
Response to Tensile and Compressive Stress Gradients,” 12th International Symposium
on Zirconium in the Nuclear Industry, ASTM STP 1354, ASTM International, West Consho-
hocken, PA, 2000, pp. 196–233.
[21] Johnson, B. K., Hydrogen Redistribution Promoted by Thermal Cycling, Experiment and
Modeling, B-T-3688,” Presented at Workshop on Hydride Re-Orientation in Zirconium
Alloys, Costa Mesa, CA, Jan 2007.
[22] Kearns, J. J., “Diffusion Coefficient of Hydrogen in Alpha Zirconium, Zircaloy-2, and Zir-
caloy-4,” J. Nucl. Mater., Vol. 43, 1972, pp. 330–338.
S. Banerjee1
Displacive and Diffusional

Transformations of the Beta
Phase in Zirconium Alloys
Reference
Banerjee, S., “Displacive and Diffusional Transformations of the Beta Phase in Zirconium
Alloys,” Zirconium in the Nuclear Industry: 17th International Symposium, STP 1543, Robert
Comstock and Pierre Barberis, Eds., pp. 23–51, doi:10.1520/STP154320130039, ASTM
International, West Conshohocken, PA 2015.2
ABSTRACT
Phase transformations of zirconium alloys in the solid state are essentially
governed by a few general tendencies. The high temperature b(bcc) phase has
the tendency to transform into the orthohexagonal structure, the orthorhombic
distortion from the hexagonal symmetry depending on the extent of super-
saturation with respect to b-stabilizing alloying elements. The b-phase also
exhibits the tendency to transform into the x-phase. The structures of the a, the
b, and the x phases are related through unique lattice correspondences, the
Burgers relation for b/a and that involving the collapse of a set of adjacent {222}
planes for b/x. In fact, the structural relationships between these phases are so
strong that the same relationships remain valid for both displacive and
diffusional transformations. Another strong tendency which is responsible for
several phase reactions in Zr-alloy systems is the clustering tendency in the b-
phase. Spinodal decomposition, phase separation, monotectoid decomposition,
metastable phase reactions during tempering of some martensites are all
consequences of this clustering tendency. Depending on the nature of alloying
additions, chemical ordering in the a, the b, and the x phase can be introduced.
The present paper illustrates how the interplay of these general tendencies can
generate the wide variety of phase transformations in zirconium alloys and how
some fundamental issues connected with phase transformation research are
elucidated by studying them.
Manuscript received March 7, 2013; accepted for publication May 21, 2014; published online October 14, 2014.
1
Bhabha Atomic Research Centre, Trombay, Mumbai - 400085, Maharashtra, India.
2
Hyderabad, India.
Copyright V
Keywords
Zirconium, martensite, phase transformation
Introduction
Pressure–temperature phase diagrams of single component systems comprising Group
IVA metals, Ti, Zr, and Hf show the three phases, a (hcp), b(bcc), and x (hexagonal),
in the solid state [1]. These phases are structurally related to each other by unique lat-
tice correspondences and their stability regimes are defined by their electronic density
of state. Introduction of alloying elements and/or change in external variables such as
pressure and temperature can induce a change in the electronic structure which in
turn is responsible for bringing about structural phase transitions [1–3].
Out of the three allotropes, the b phase with the bcc structure is associated with
the highest symmetry. This is also the first phase to appear from the liquid phase
on cooling. The wide variety of phase transformations encountered in zirconium
based alloys can be shown to result from some general tendencies as listed below:
1. The high temperature b (bcc) structure has a tendency to transform into the
orthohexagonal structure at lower temperatures. While in pure zirconium and
its dilute alloys, the low temperature phase is hcp, orthorhombic distortions
appear in alloys richer in b-stabilizing alloying elements which introduce a lat-
tice distortion leading to the formation of orthorhombic martensite [2,4–7].
2. There is a tendency for the b phase to undergo a transition into the hexagonal
x-phase in several alloys particularly when the b-stabilizing alloying elements
are present within a certain composition range. The b- and the x-structures
are so related that the collapse of a set of {111}b planes can generate the x-
structure [1–3,8–10].
3. The b-phase exhibits a strong phase separation tendency in several alloy sys-
tems containing b-stabilizing elements. The tendency manifests itself in spino-
dal decomposition, monotectoid reaction, formation of metastable b phase
during thermomechanical processing, tempering of martensites, and even
creep deformation [11–17].
4. Chemical ordering tendencies of the a-, b-, and x-phases resulting in the for-
mation of their corresponding superlattice structures [2].
The composition regime in which the aforementioned tendencies dominate are
illustrated in phase diagram (Fig. 1(a) with metastable a–x–b phase boundaries and
start temperatures for the martensitic, Ms(a0 ) and the x, Ms(x)- transformations.
The regime in which the b phase undergoes a phase separation process and in
which a spinodal decomposition is possible from the chemical free energy consider-
ation are indicated in Fig. 1(b)
Phase transformations in alloys are broadly classified on the basis of the trans-
formation mechanism into the displacive and the diffusional types. Transforma-
tions in titanium and zirconium are especially suitable for drawing comparisons
between the distinctive features of these modes of phase transformation. The fact
that both the b !a0 martensitic transformation and the diffusional process of the a
precipitation from the b matrix follow the same lattice correspondence provides
BANERJEE, DOI 10.1520/STP154320130039 25
FIG. 1 (a): Zr rich side of the Zr-Nb phase diagram Metastable b þ x phase field and
Ms(a0 ) and Ms (x) are also indicated by dashed lines. On quenching from the b
phase field, alloys with Nb content of up to about 7 % produce martensite a0 and
alloys with Nb content between 7 to 18 % transform into the bþ quenched-in x
structure. On ageing at temperatures below 773 k, the metastable b þ aged x
structure is seen in alloys with Nb content between 7 % and 32 %. (b) The Zr-Nb
phase diagram showing chemical and coherent spinodal lines. 1 and 2 are
computed from G-X plots [12] and derived from critical temperature and critical
concentration of the miscibility gap [2], respectively.
the opportunity for making a direct comparison between the martensitic and
diffusional modes. Similarly, the roles of diffusional and displacive atom move-
ments during the b!x transformation and during formation of ordered interme-
tallic x-related structures, as revealed by high resolution microscopy, provide an
insight into the mechanisms of these transformations.
The phase separation tendency exhibited by the b-phase gets manifested in sev-
eral ways. In many alloy systems, such as, Zr–Nb and Zr–Ta, the b phase decom-
poses into a two phase b1 þ b2 mixture in a certain temperature–composition
domain and a monotectoid reaction, b1 ! a þ b2 appears in the phase diagram
[12–15]. As a consequence, spinodal decomposition occurs in the b phase and the
metastable b phase (with nonequilibrium composition) forms or is retained under
situations such as tempering of martensite, creep deformation, in-reactor service
and cooling subsequent to hot working.
The present paper will bring out characteristic features of different types of phase
transformations encountered in zirconium based alloys and will demonstrate that the
variety of transformation related phenomena observed in these systems can be ration-
alized on the basis of the few general tendencies mentioned earlier. Finally, through
this paper, the author will attempt to convey to the readers how enriching the experi-
ence has been to study phase transformations in zirconium alloys. These studies have
not only revealed that nearly the entire spectrum of solid-solid phase transformations
can be studied in these systems but also helped in understanding the methods of tai-
loring microstructures to achieve the desired properties to meet the user’s demands.
As the author of the Kroll Award (2012) paper, I would like to mention that a
research group in the Bhabha Atomic Research Centre took up the work on various
aspects of phase transformation research with a focus on zirconium alloy systems
primarily to gain an understanding of structures, properties, and in-reactor behav-
ior of zirconium alloys in terms of physical metallurgy principles. Such a compre-
hensive approach has been very rewarding as many of the important issues related
to solid–solid phase transformations, in general, could be addressed and a better
understanding of several related phenomena gained. The credit, therefore, goes to
the members of the entire research group for their unstinted support for the realiza-
tion of this specific objective.
Martensitic Transformation in Zirconium Alloys

The martensitic transformation from the b phase to the low temperature orthohex-
agonal a0 or a00 phase can be viewed as a homogeneous distortion of the bcc lattice
into the orthohexagonal lattice as shown in Fig. 2(a). The lattice distortion B associ-
ated with the transformation is given by
2 3
g1 0 0
6 7
B¼4 0 g2 05
0 0 g3
BANERJEE, DOI 10.1520/STP154320130039 27
FIG. 2 (a) Lattice correspondence between the b and the a phase. Filled circles indicate
lattice points corresponding to (110)b plane, which is parallel to the basal (0001)a
plane, lattice points on which are indicated by open circles. (b) Strain ellipsoid
showing the lattice distortions.
rffiffiffi
3 aa 1
g1 ¼ ; g2 ¼ aa =ab and g3 ¼ ðca =ab Þ
2 ab 2
where g1 ; g2 , and g3 are distortions along the principal strain axes marked in [110]b
projection (Fig. 2(a)) and aa, ca, and ab are lattice parameters of the a and the b
phases.
Substituting the lattice parameter values for pure Zr, the lattice strains are
obtained to be approximately 10 % tensile, 10 % compressive, and 2 % tensile,
respectively, along g1 ; g2 , and g3 directions [2,4,5,18–20]. These lattice strains

nearly satisfy the invariant plane strain (IPS) condition, the deviation arising only
from the 2 % tensile strain in the g3 direction perpendicular to the basal plane of a.
The approximate crystallographic analysis (neglecting 2 % tensile strain) demon-
strates that the lattice distortion along with only a rigid body rotation that can yield
an IPS habit plane solution. As shown in the strain ellipsoid construction in
(Fig. 2(b)), there are two distinct possibilities of the rigid body rotation, anticlock-
wise around the [110]b vector to bring the undistorted vector OA0 to coincide with
its pretransformation position OA (or clockwise to bring OX0 to OX), which results
in yielding two solutions for crystallagrahically equivalent orientation relations:
ð110Þb ==ð0001Þa0 ; ½111b ==½1120a0 and
ð110Þb ==ð0001Þa0 ; ½111b ==½1210a0
To account for the deviation from IPS condition arising from 2 % tensile strain
along [110]b//[0001]a0 , the lattice invariant shear of a relatively small magnitude
needs to be introduced. This results in the formation of either internally twinned
plate or dislocated lath martensites. Microstructural studies have indeed demon-
strated the formation of the following 5 types (Fig. 3) which have been described in
detail elsewhere [6,7,18–22].
(i) Martensite laths belonging to the same crystallographic variant stacked in a
parallel array within a packet—a morphology usually observed in dilute
alloys with Ms temperatures higher than 923 K (Fig. 3(a)(i)).
(ii) Martensite laths of two twin related variants stacked alternately within a
packet-encountered in alloys with Ms temperature of about 873 K
(Fig. 3(a)(ii)).
(iii) Internally twinned martensite plates with thick twin segments and a corre-
sponding zig-zag habit. The ratio of thickness of the twin segments for this
type of morphology is 1:4, a value which is predicted from the
consideration of average habit plane satisfying the IPS condition. This
morphology is exhibited predominantly in primary martensite plates
(Fig. 3(b)(i)).
(iv) Internally twinned martensite plates with a stack of very thin twins. Such
plates are often seen as secondary plates in conjunction with primary plates
with thick twins (Fig. 3(b)(ii)).
(v) Group of three martensite variants mutually twin related which gives an
appearance of an indentation mark. The self-accommodation of strain
achieved in such a group is very high, degree of self-accommodation as
defined in Ref. [22]. being about 91 % (Fig. 3(c) and 3(d)).
The thick and thin twin morphologies have been found to be related to the
non-equivalent habit plane solutions. While the specific twin plane on which the
lattice invariant shear is operating remains nearly parallel to a {110}b type mirror
plane in case of thick twins, this condition is not fulfilled in martensite plates with
thin twins. In the former case, both the twin segments nearly satisfy the IPS
BANERJEE, DOI 10.1520/STP154320130039 29
FIG. 3 Different types of martensite morphologies and substructures. (a) Dislocated

lath martensites—(i) with small angle boundaries between adjacent laths and (ii)
alternate laths twin related. (b) Internally twinned plate martensites with (i) thick
twins and serrated habit plane and (ii) thin twins. (c) Partitioning of the parent
b-grain by primary maternsite plates of three variants. Maternsite plates of
continuously decreasing size forms in successive generations to achieve near
complete b to a0 transformation. (d) Indentation morphology of a group of three
variants.
condition and can therefore grow substantially, while such a growth is not energeti-
cally favourable for the thin twins in the latter.
Precipitation of Widmanstatten Alpha

in The Beta Matrix—A Case of Diffusional
Transformation
The b ! a transformation can also be induced in Zr alloys by bringing suitable
alloy samples from the high temperature b phase field to the (a þ b) phase field and
allow the transformation to occur in a thermally activated diffusional mode. The
crystallography of such a transformation has been studied by Perovic and Weath-

erly [23], Zhang and Weatherly [24] and Banerjee et al. [21]. The observed orienta-
tion relation, being close to the Burgers relation, suggests that the correspondence
between the b and a lattices remains to be the same as what is observed in the case
of martensitic transformation. This gives a unique opportunity to compare and
contrast the products of two modes of phase transformation in the same alloy sys-
tem maintaining essentially the same lattice correspondence. The crystallographic
features of a-laths in the b-matrix can be summarized as follows:
(i) The b/a orientation relation is very close to Burgers orientation relation
(ii) The habit plane pole (broad face containing the length and breadth direc-
tions) lies between {103}b and {131}b poles
(iii) The habit plane is decorated with periodic interfacial dislocations with a
line vector along the precipitate length direction. The average spacing
between the dislocations is about 8 to 10 nm and these dislocations are
associated with hc þ ai Burger Vector (Fig. 4).
(iv) The length direction closely matches with the invariant line strain (ILS)
direction h433ib
FIG. 4 (a) and (b) Widmanstatten a laths in the b matrix with interfacial dislocations
aligned along the long direction of lath.hc þ ai dislocations aligned along the
max growth direction surrounding the lath. (c) The structural model of the a/b
interface in a Ti–Cr alloy.
BANERJEE, DOI 10.1520/STP154320130039 31
These crystallographic observations bring out the essential differences between

the crystallography of precipitation from the martensite crystallography. Though
the lattice correspondence is similar in both cases, the habit plane in the diffusional
b ! a transformation does not satisfy the IPS condition. The criteria of selection of
the habit plane in diffusional transformations have been studied in several systems
and there is a general agreement that one of the vectors that define the habit plane
is the ILS direction [21,23,24]. The second vector fulfills one of the following
conditions:
(a) a vector that remains un-rotated
(b) a direction of low strain
(c) a direction of easy slip in the parent or the product phase
(d) a direction followed due to elastic anisotropy of the precipitate matrix
interface
(e) a direction along which structural ledges are aligned
Zhang and Purdy [25] using an extension of the O-lattice concept to a system
containing invariant lines have shown that a complete Burgers vector balance can
be accomplished at the a/b interface in Zr–2.5 Nb by a single set of dislocations
with the [010]b Burgers Vector and a dislocation spacing of about 10 nm. The inter-
facial structure shown in the present paper is consistent with the predictions of
Zhang and Purdy [25]. The observed interfacial structure in Zr–20 % Nb, however,
differs from that predicted from the model for atomistic matching at the a/b inter-
face in a similar system of Ti–Cr alloys, although the indices of macroscopic habit
plane and the orientation relation are quite similar. Matching of the atomic planes
across the b/a boundary in Ti–Cr alloy has been achieved by introduction of bi-
atomic structural ledges with Burgers Vector (1/6 hai and misfit compensating hci
type ledges [26]). In case of Zr–Nb alloys, the misfit along the hai and hci compo-
nents appears to have been compensated by a single set of hc þ ai dislocations lying
along the ILS direction [21].
b ! x Transformation
The x-phase, which is an equilibrium phase in Group IV A metals at high pres-
sures, also forms at the ambient pressure in several alloys based on Ti, Zr, and Hf,
and also in many other bcc alloys under metastable conditions. The unique mecha-
nism of the b ! x transformation has drawn attention of many research groups
for studying this transformation which can be induced by thermal and mechanical
treatments [1,2,27]:
(i) b-quenching of alloys in the composition range where the Ms(xÞ tempera-
ture is higher than the Ms(a0 ) temperature
(ii) Isothermal ageing at temperatures below about 773 K in a fairly wide com-
position range (electron irradiation is shown to assist the process)
(iii) Shock deformation of b-alloys [28,29]
Morphologies of the resulting x-phase and the mechanisms of their formation
under these treatments are different (Fig. 5). However, the lattice correspondence
FIG. 5 Different Morphologies of x phase in b-matrix. (a) Fine ellipsoidal particles of x

formed on quenching from the b phase field. (b) x – particles formed in the b –
matrix due to isothermal ageing. x - precipitates are aligned along h100ib. (c)
Plate like morphology of the x phase produced by application of shock pressure
(8 GPa) on the b Zr–20 % Nb alloy. (d) Bimodal size distribution of x particles in b
quenched Zr–17.5 % Ni alloy.
between the b and the x phase remains the same. The x-structure can be generated
from the parent b by collapsing of pairs of (222)b planes in a sequence as shown
in Fig. 6. The ABCABC stacking of the (222)b planes changes to the AB0 AB0 AB0
stacking of the basal planes of x phase as the B and C planes are collapsed on to the
middle position as designated as the BI plane, keeping the A plane unaltered. The
FIG. 6 The b–x lattice correspondence and schematic illustration of the lattice collapse
mechanism.
BANERJEE, DOI 10.1520/STP154320130039 33
periodic displacement of the (222)b planes as a longitudinal displacement wave,

where the displacement, up, of the pth plane is expressed as

up ¼ Ad sin Kx xp
where:
Ad ¼ the amplitude,
kx ¼ the wave vector (¼ 2p / 3d222) of the displacement wave, and
xp (¼ pd222) ¼ the distance of the pth (222) plane from the origin (Fig. 6).
The wave length, k of the longitudinal displacement wave is 3d222 and the
corresponding wave vector, kx, is equal to 2/3h111i*, where h111i* is a vector in
the reciprocal space. The merit of this description lies in the fact that it can as also
be applied to structures involving partial collapse of the (222)b planes by varying
the amplitude of the displacement wave.
The athermal b ! x transition is associated with an appearance of extensive
diffuse intensity distribution with the maximum intensity located close to the ideal
x reflections. The real time evolution of x reflections from the diffuse intensity dis-
tribution is shown in Fig. 7 in which the progressive change is recorded at 425 K
under the condition of radiation enhanced kinetics [30]. The pre-transition effects
such as diffused intensity distribution in diffraction patterns and a pronounced dip
in the phonon dispersion relations at wave vectors close to 2/3 h111i* have made
the b ! x transition a unique example of a first order transition with a strong
premonition.
When the b ! x transformation is introduced by a rapid quenching without
providing adequate thermal activation for atomic diffusion, the transformation is
accomplished only by the lattice collapse mechanism which involves only shuffle
of atoms without requiring a macroscopic homogeneous strain of the lattice. The
volume change due to the phase transformation is negligible and the x phase
remains fully coherent with the b matrix (Fig. 5(a)). No doubt such a transformation
is displacive in nature. The process is distinct from martensitic transformation
which requires a dominant lattice shear and corresponding macroscopic shape
strain.
The precipitation of the x-phase in the b-matrix induced by thermal ageing
involves atomic diffusion. A partitioning of alloying elements (depletion of
b-stabilizing elements), which is thermally activated, occurs as a precursor to the
lattice collapse for the b ! x transformation. Such concepts have earlier been intro-
duced [12] from thermodynamic arguments and have been established in Ti alloy
by high resolution electron microscopy and atom probe experiments [31].
Production of excess point defects under electron irradiation assists the atomic
diffusion and thereby induces the b ! x transition at temperatures as low as 300 K.
As the process occurs in a gradual manner, the stages of partial collapse of lattice
planes characterized by diffuse intensity distribution in diffraction patterns and the
progressive stages of transformation could be captured in real time.
FIG. 7 Electron irradiation induced b ! x transformation (a)–(e). Selected area

diffraction patterns ([110] zone axis) showing the real time progression of the
b ! x transformation on irradiating a sample of Zr–20 % Nb with 1 MeV electron
at 425 K at a dose rate of 10–3 dpa/s. (a) Before irradiation, (b) 15, (c) 60 s, (d)
300 s, (e) 450 s of irradiation, (f) dark field g ¼ (10–10)x image of fine x particles
formed after 450 s exposure.
In case of shock loading, a macroscopic shape strain is superimposed on the

b ! x transformation, which is induced by a thermodynamic driving force arising
out of PDV, where P is the applied pressure and DV is the change in specific
volume due to transformation. The transformation therefore occurs not by pure
shuffle. A macroscopic lattice strain is introduced and the observed crystallography
of {112}b habit plane is consistent with the IPS prediction. Figure 8 shows that a
macroscopic shear on the (112)b plane along a h111i direction, superimposed with
atomic shuffles, and can produce the x-structure [28,29].
BANERJEE, DOI 10.1520/STP154320130039 35
FIG. 8 Schematic showing the combination of macroscopic shear and atomic shuffle
resulting in the formation of x-plate in the shock pressure induced b to x
transformation.
Phase Separation in the b-Phase

Binary phase diagrams of Zr alloys (with b-stabilizing elements such as Nb, Ta)
exhibit a miscibility gap in the b-phase. The analysis of the Zr–Nb phase diagram
[12] has established the free energy versus composition plots for the b-phase. The
characteristic doubly inflected shape of these plots at temperatures below about
1200 K is consistent with the experimentally determined critical solution tempera-
ture. The temperature-composition regime in which spinodal decomposition of the
b-phase can occur has also been identified (Fig. 1(b)) from these plots. This result
also compares well with the experimental observations on spinodal decomposition
in b–Zr–Nb alloy by Flewitt [11]. Free energy–composition plots for the a-phase
have also been obtained from analysis of phase diagrams using the equilibrium crite-
ria between the a and the b phases. From a critical assessment of experimental phase
diagram data and results of thermodynamic modeling with regard to the Zr–Nb sys-
tem Abriata and Bolcich have shown a reasonable agreement between them [15].
The occurrence of monotectoid reactions in several Ti and Zr alloy systems is a
consequence of the presence of a high temperature continuous solid solution of the
b phase from pure Zr/Ti on one side to the b-stabilizing elements (Nb,Ta) on the
other, the allotropic transformation of the Zr/Ti rich b-phase to the hcp a phase
and the strong tendency of phase separation in the b phase at lower temperatures.
The free energy composition plots at the monotectoid temperature 883 K shows
that the a phase can set up equilibrium with the b phase of two widely varying com-
positions: bI(20 % Nb) and bII (85 %Nb) [12].
FIG. 9 Free energy – composition plots for the a- and the b-phase in the Zr–Nb
system at 850 K showing that a metastable/stable equilibrium can be set
up between the a phase on one side and the bI or the bII phase on the
other. (1) a0 ! a þ bI ! a þ bII , (2) a0 ! bI ! a þ bI ! a þ bII ,
(3) b ! spinodal ! bI þ bII ! bþx=bþa.
Metastable Equilibrium Between a and bI

The possibility of setting up metastable equilibrium between the a and the bI phase
even at temperatures lower than the monotectoid temperature (Fig. 9) leads to the
following metastable phase reactions during tempering of supersaturated a0 mar-
tensite [13]:
(i) Tempering of a0 in the Zr–2.5 alloy in the temperature interval of 773–883
results in precipitation of primarily bI phase (Zr–20 % Nb) and only a lim-
ited amount of the equilibrium bII phase (Zr–85 % Nb) (Fig. 10).
(ii) At temperatures below 773 K, the precipitates are of the equilibrium bII
type
(iii) On tempering, the Zr–5.5 % Nb martensite initially reverts back to the par-
ent b phase by a composition invariant process; subsequently, the reverted
b phase transforms into a structure consisting of Widmanstatten a-plates
in the bI matrix (Fig. 11).
These observations are rationalized in terms of calculated free
energy–composition plots for the a and the b phase at a temperature slightly below
the monotectoid temperature (Figs. 9 and 12). At such temperatures, two common
tangents can be drawn: one (line A) of which touches the a- and the bI curves, while
the other (line B) touches the free energy curves corresponding to the a- and the bII
phases. This implies that a metastable equilibrium between a- and bI is feasible at
BANERJEE, DOI 10.1520/STP154320130039 37
FIG. 10 Predominantly bI precipitation during tempering of a0 martensite at 773 K, while

bII precipitation is at 823 K.
such temperatures. It has also been argued that from kinetic considerations, the
precipitation of the bI phase is favoured at such temperatures [13]. The situation,
however, changes as the tempering temperatures is lowered below 773 K where
metastable equilibrium between a- and bI cannot be established.
The sequence of phase transformations can be correctly predicted using the
free energy–composition plots generated at different temperatures [12]. Figure 9
shows how the sequence changes at a given temperature for different composition.
While for a composition CI, (CI < C2, where C2 is the composition at which free
energy curves for a and b intersect) bI phase is precipitated during tempering, for a
composition C5 (C5 > C2) the sequence of transformation is a0 ! bI ðinheriting a0
compositionÞ ! a þ bI ! a þ bII . Compositions CI and C5 correspond to
Zr–2.5 % Nb and Zr–5.5 % Nb, experimental observations on which are summar-
ized in the preceding paragraph.
The metastable equilibrium between the a phase and the bI phase plays a major
role in dictating the microstructure of Zr–2.5 % Nb pressure tubes. The fabrication
FIG. 11 Reversal of martensite, a0 ! b, followed by Widmanstatten a plate precipitation

in the b matrix during tempering of Zr–5.5 % Nb a0 martensite at 823 K.
FIG. 12 Computed free energy versus Nb concentration plot for the Zr-rich side of the
Zr–Nb system.
schedule of Zr–2.5 % Nb pressure tubes involve hot extrusion at about 800 C (high
in the a þ b phase field) where the a phase remains in equilibrium with the b phase
with a composition of about 8 %–12 % Nb [16]. The resulting microstructure is the
two phase (a þ bI) elongated a-grains with b-stringers at a-boundaries (Fig. 13).
Depending on the cooling rate from the hot extrusion temperature, the volume
fraction of the b stringers is reduced with accompanying enrichment of Nb. The
final hot extruded material, therefore, tends to achieve the metastable a þ bI struc-
ture. During the subsequent forming operations, sufficient care must be taken not
to upset this metastable equilibrium. In case a single cold working step is intro-
duced in order to achieve the required dimensions of the pressure tube and also the
optimized cold work (necessary for controlling the in-reactor creep), the a þ bI
phase distribution remains unaltered. In contrast, if a two-step cold working pro-
cess is adopted for achieving a better control over the process and dimensions of
the product, the intermediate annealing condition should be so selected that the
elongated a þ bI morphology with the metastable bI phase is retained and at the
same time the extent of cold work is maintained at the predetermined level of about
25 %. Determination of the conditions at which the a/bI metastable equilibrium can
be maintained is, therefore, very important.
BANERJEE, DOI 10.1520/STP154320130039 39
FIG. 13 Evolution of microstructure during fabrication of Zr–2.5Nb pressure tubes.

Transmission electron micrographs of Zr–2.5 Nb alloy. (a) As extruded
microstructure showing elongated morphology of a-phase separated by
b-phase stringers. (b) First pilgered microstructure illustrating a very high
dislocation density. The dislocations are concentrated primarily at the a/b
interface. (c) Incomplete recrystallization of the a-stringers as evidenced by the
presence of a high density of dislocations after annealing at 773 K for 6 h. (d)
Coarsening of the a-grains caused by redistribution and agglomeration of
b-phase. The b-phase is located at the triple junctions of a-grains after annealing
at 873 K for 1 h. (e) Completely recrystallized a-lamellae obtained after
annealing at 823 K for 3 h. The lamellar morphology of the two phases is not
altered by this annealing treatment.
A recent study [17] pointed out that the competition between two equilibrium
conditions, one between a and bI and the other between a and bII, also influences
the creep behavior of Zr–2.5 Nb at temperatures slightly below the monotectoid
temperature. In contrast to the expectation, the equilibrium a þ bII structure exhib-
ited a creep rate considerably higher, over an order of magnitude, compared to that
for the metastable (a þ bI) structure. Detailed microstructural investigations have
revealed that during the creep deformation process the b phase gets redistributed to
a significant extent (Fig. 14). During the dissolution and re-precipitation of the b
phase, they undergo changes in chemical composition and b phase precipitates are
relocated at grain boundaries which are predominantly perpendicular to the tensile
FIG. 14 Redistribution of bI and bII precipitates after creep deformation at 800 K. Initial
microstructures of a þ bI: before creep (a) and after creep (b). Reduction in
volume fraction of b precipitates with a corresponding Nb enrichment of b
precipitates from 21 to 35 %. Initial microstructures of a þ bII: before creep
(c) and after creep (d). Increase in volume fraction of b precipitates with a
corresponding Nb depletion of b precipitates from 80 to 30 %. P indicates
compression facets of grains, from where b dissolved and re-precipitated at the
tensile (marked T) facets of a grains. The star sign marks a b phase trail
indicating the path of atom transfer during the creep-cum-re-precipitation
process. The tensile direction is shown with an arrow.
BANERJEE, DOI 10.1520/STP154320130039 41
stress axis. Re-precipitated b phase has been found to have a composition of around
35 %Nb.
This observation is analogous to that reported in Zr–2.5 Nb pressure tubes irra-
diated in a nuclear reactor. Griffith et al. [33] reported that the Nb content in the b
phase varied considerably (from 37 % to 75 %) in pressure tubes after 2–14 years of
service.
Given that the initial structure consists of equiaxed a grains with grain
boundaries and tri-junctions decorated by bI or bII precipitates, depending on the
equilibration temperature chosen, the creep deformation at around 800 K is
expected to induce atomic movements along the grain boundaries and sub bounda-
ries. Movement of both solute Nb and solvent Zr atoms is driven by the chemical
potential gradient setup between the facets of grain boundaries experiencing com-
pressive and tensile stresses. The difference in the creep rates of the two structures,
a þ bI and a þ bII stems from the fact that the volume per atom corresponding to
the bI and the bII phases are different—the former being larger than the latter. The
presence of a tensile stress, therefore, favours the formation of the bI in preference
to the bII phase. During creep deformation at temperatures close to 800 K, the dis-
solution of bII precipitates from the grain boundary regions experiencing compres-
sive stress and re-precipitation of bI (Nb content 35 %) at grain boundary facets
associated with tensile stress is expected to contribute an additional strain due to
increase in volume fraction of the precipitate phase. Unusually high creep rate
observed for the equilibrium a þ bII microstructure has thus been explained in
terms of this transformation from the equilibrium to the metastable structure.
Ordered Omega Formation—An Example of

Coupled Displacive and Replacive Ordering
A new class of transformation involving a combination of a displacive and a repla-
cive ordering has been encountered in binary Zr–Al and ternary Zr–Nb–Al alloys.
The equilibrium Zr2Al (B82) and Zr5Al3 (D88) structures can be viewed as chemi-
cally ordered derivatives of the x structure, the lattice correspondences between
them being illustrated in Fig. 15. The b ! B82 and the b ! D88 transition can be
accomplished by decorating the b (bcc) lattice with ordered arrangements of differ-
ent atomic species and vacancies and introducing a lattice collapse akin to the
b ! x transition [34–36]. The diffuse intensity distribution characteristic of the
b ! x transition is also seen in zirconium alloys in which B82 and D88 phases
form.
The phase evolution process during rapid solidification of an alloy close to
Zr3Al composition has been followed [34]. The quenched-in structure shows a
supersaturated b phase with composition modulation resulting from spinodal
decomposition (Fig. 16). On subsequent ageing, the amplitude of composition mod-
ulation along the h100ib directions increases and finally the aluminium enriched
regions of cuboidal shape undergo the b ! B82 transition. Each single cuboidal
FIG. 15 B82- and D88-ordered x-phases lattice correspondences. (a) and (b) The lattice
correspondence between ordered derivatives of bcc (B2) and the ordered
omega B82 structures. Different atomic sites and arrangements of the (222)bcc
planes is shown. Complete collapse of the two pairs of the (222) planes and
chemical ordering leading to changes in structural symmetry are shown. (c) The
atomic arrangement in the B2 unit cell. (d) The atomic arrangement in the D88
unit cell.
block is seen to transform into a single B82 particle. The absence of multiple
variants of B82 crystals within a cuboidal block, unlike that observed in the b ! x
transformation on quenching, suggests that the transformation is driven by an
instability towards the development of concentration and displacement waves.
The b ! D88 transformation has been encountered in Zr3Al–3 % Nb and in
Zr3Al–10 % Nb alloys. The observed transformation sequence is quite similar to the
case of the binary Zr3Al alloy as the tendency for phase separation in the b phase
and for the chemical and displacement ordering is present in these alloys.
Formation of c-Hydride—An Example

of a Bainitic Transformation
Precipitates of c-hydride appear as plate shaped products in both the a and b
matrix. The transformation is too rapid to be suppressed under water quenching
(104 K/s at a transformation temperature lower than 550 K). Self-diffusion of Zr
atoms at such temperatures being insignificant, the transformation mechanism nec-
essarily involves a shear of the a lattice accompanied by partitioning of hydrogen
atoms from a to c through an interstitial diffusion mechanism. The lattice corre-
spondence between the a, the b, and the c-hydride structures, as derived from the
observed orientation relation is described in Fig. 17 in which the (0001)a, the (110)b,
and the (111)c planes are superimposed. The crystallography of the transformation
BANERJEE, DOI 10.1520/STP154320130039 43
FIG. 16 Rapidly solidified Zr3Al-Nb alloy. (a) Dark field image shows cuboidal in bright
contrast with g ¼ (1100)x. (b) Dark field micrograph (g ¼ 1100) shows the second
phase in bright contrast. A tendency of h100ib alignment of the cuboidal phase
is noticed. (c) Bright field image showing modulations along h100ib directions.
(d) Matrix b phase with a distribution of fine x particles are in the bright
contrast with g ¼ (031)b. SAD pattern in the inset shows the diffuse intensity
distribution corresponding to the partially collapsed x structures in the b
matrix.
[37–40] has established that the IPS criteria correctly predict the habit plane. The
overall transformation, which qualifies to be a typical Bainitic transformation, is
characterized by a lattice shear of the a or the b lattice and accompanying hydrogen
partitioning by interstitial diffusion.
In view of the close lattice correspondence between the a, the b, and the c-
hydride structures and the fact that hydrogen is a strong b stabilizer, it is attractive
to envisage that the c-hydride formation in the a phase occur through an interme-
diate embryonic b structure. The overall c- precipitation in the a phase can also be
viewed as driven by a tendency of the a phase to transform into b due to localized
hydrogen enrichment and concomitant shearing of the b lattice into c-hydride due
to interstitial ordering.
FIG. 17 c-hydride precipitates in the a matrix. (a) Lattice correspondence between the
a0 , the b, and the c-hydride structures (b),(c), internally twinned c-hydride plate
in the a matrix.
Active Eutectoid Decomposition in Zirconium

Based Alloys
Active eutectoid systems [41] are those in which the b-phase decomposes into the
product phases so fast that it cannot be suppressed even by a rapid quenching. This
decomposition is observed in alloys of near-eutectoid compositions. The state of
aggregation of the decomposition products in such alloys is lamellar, so that a fine,
pearlite-like microstructure is obtained on rapid cooling. Systematic studies on alloy
systems exhibiting active eutectoid decompositions have revealed [42,43] that active
eutectoid decomposition occurs in those alloy systems where (i) the eutectoid
temperature is high, (ii) the eutectoid composition is solute lean, and (iii) the
intermetallic phase resulting from the reaction is rich in the base metal. The micro-
structural observations made on a b-quenched, near-eutectoid (Zr–1.6 wt. % Cu;
b ! a þ Zr2Cu) alloy, which represent a typical case of active eutectoid, could be
summarized as follows:
(i) The quenched material consists of several colonies with lamellar
structure, consecutive lamellae of the a-phase being separated by ribbon
BANERJEE, DOI 10.1520/STP154320130039 45
like features. In a given cell, the a-lamellae have a single crystallographic

orientation and so have the ribbons.
(ii) There are broadly two types of morphologies observed; one with straight
lamellae and other with wavy and broken lamellae. Figure 18(a)–18(c). In the
transition region, a one to one correspondence between the straight and
the wavy segments exists.
(iii) The average colony size and the average inter-lamellar spacing are about
2.0 and 0.1 lm, respectively.
Based on evidences provided by selected area diffraction and also by elemental
analysis of the phases, it was inferred that the ribbons comprised phase similar but
not identical to Zr2Cu (Fig. 18(d)–18(e)). The matching was better in regions with
wavy ribbons than in regions with straight ribbons. Assuming the phase constitut-
ing the ribbons to be the Zr2Cu phase, the following orientation relationship
between the two phases could be expressed as:
ð0001Þa ==ð013ÞZr2 Cu; ½1100a ==½031Zr2 Cu
FIG. 18 Microstructure of active eutectoid product in Zr–1.6 wt. % Cu (a) period and
straight lamellae of a þ b0 structure resembling internally twinned martensite
structure. (b) Straight lamellae degenerating into wavy lamellae. (c) Dark (b0
reflection) field image of a þ b0 structure. (d) Composite SAD pattern showing
orientation relationship between a and b0 phase and (e) key to the pattern.
This orientation relationship (Fig. 18(d)) is consistent with that observed

between the a–Ti and the Ti2Cu phases in the Ti–Cu system, which also exhibit
active eutectoid decomposition.
In the case of Zr–Cu alloys, it can be seen that the Zr2Cu phase can be
derived by chemical ordering of the b phase leading to the formation of a super-
lattice cell, which is made up of a stack of three bcc unit cells one over the other.
Using such relationships, and the observed orientation relationship between the
a and the Zr2Cu phases, the orientation relation between the parent b-and the
a-phases has been found to match with the Burgers relation. Using SAD pattern
analysis, it is also possible to show that the metastable phase constituting the
ribbons has attained a certain extent of long range order and this has resulted in
an approximately three fold increase in the dimension of the bcc unit cell in the
[001] direction. Based on these observations it has been suggested that, during
b-quenching, the b-phase decomposes into a mixture of the a-phase and a par-
tially ordered b0 phase. Regions containing straight and wavy lamellae in the
same colony possibly correspond to different levels of Cu enrichment, and there-
fore, to different degrees of long range order in the ordered b0 phase. The crys-
tallographic description of the lamellar product structure with respect to the
parent b structure is illustrated schematically in Fig. 19. As suggested in this
figure, presence of a lattice correspondence between the two product phases
decomposing from the b-phase essentially requires cooperative atom movements
and presence of partial order within the second phase, necessitating diffusion of
copper atoms.
As lattice site correspondence exists between the parent b phase on one side
and the two product phases, a and partially ordered Zr2Cu on the other, the
advancing transformation front is expected to maintain coherency at least partially.
However, redistribution of a copper atom ahead of the advancing front is a prereq-
uisite step which would cause 15 % contraction in the bcc unit cell and also intro-
duce long range ordering. In view of the fact that solubility of Cu in a–Zr is much
smaller than that in b–Zr and the anomalous diffusion possible in b–Zr, continuous
flow of Cu atoms ahead of the advancing transformation front is possible. A simpli-
fied analysis has been carried out with a view to comparing the observed reaction
rates with those estimated. It has been found that the observed and the estimated
growth rates are more or less of the same order at temperatures not far below the
eutectoid temperature. This analysis emphasizes the fact that solute partitioning
between the two product phases would be quite significant in spite of the extremely
fast reaction rates. This is possible due to the rapid anomalous diffusion in the b-
phase and the exceedingly small interlamellar spacing in the decomposition
product.
In the case of Zr3Fe, however, such a simple structural relationship is not appa-
rent. Nevertheless, from considerations of reaction kinetics and post-reaction
microstructure, the decomposition process does appear to be very similar in this
case too.
BANERJEE, DOI 10.1520/STP154320130039 47
FIG. 19 Lattice correspondence between the parent b and the product a and b0
structures across the transformation front. Microstructures of active eutectoid
product in Zr-3 wt. % Fe.
In short, in an active eutectoid decomposition underlying lattice corre-

spondence plays an important role. This helps in the formation of the glissile
interfaces and also in maintaining lattice registry during the phase
transformation.
Closing Remarks
Pursuing phase transformation research in zirconium alloys has been a rewarding
experience for our research group. When we started our work we had no idea of the
diversity of phase transformations in these alloys. As we probed into
transformations of different kinds in zirconium alloys, we met with many pleasant

surprises and gradually the whole field of phase transformation research was
unfolded before us. It is not only the variety which made these studies so attractive,
but also the amenability of these systems for providing answers to some fundamen-
tal queries. Usually, text books in physical metallurgy explain various aspects of
phase transformation citing examples from iron based alloys. We have shown in a
recently published monograph entitled Phase Transformations: Examples from Tita-
nium and Zirconium Based Alloys [2] and in a review paper [44] that the subject of
phase transformations can be treated taking examples only from titanium and zir-
conium alloys. The present paper is something like a preview of the extensive treat-
ment given in the monograph.
In retrospect, we realize the importance of these fundamental studies in under-
standing the process and mechanism of the microstructure development during
thermal and mechanical treatments. Such knowledge has been of great value in
devising fabrication schedules and in predicting in-service microstructural evolu-
tion of zirconium alloy components used in nuclear reactors.
ACKNOWLEDGMENTS
In this brief account of phase transformations in zirconium alloys, I have made an
attempt to summarize the work of our research group of Physical Metallurgy in
Bhabha Atomic Research Centre. I also take this opportunity to place on record the
close interactions we had with process metallurgists and reactor engineers particularly
with our colleagues in Nuclear Fuel Complex, Hyderabad. It is through the constant
exchange with them the concepts and ideas generated from research could be
deployed in engineering practice.
For introducing me to the subject of phase transformation in zirconium alloys
and for guiding me in the initial stages of my researches, I remain ever grateful to
Dr. R. Krishnan and Late Prof. P. R. Dhar. For encouraging me to pursue these
researches, I am gratefully indebted to Late Prof. H. I. Aaronson, Late Dr. M. K.
Asundi, Mr. K. Balaramamurthy, Late Prof. R. W. Cahn, Dr. C.K. Gupta, Prof. S.
Ranganathan, Prof. P. Rama Rao, Late Mr. P. R. Roy, and Late Prof. C. V. Sun-
daram. I had the opportunity of continuously interacting with a number of bright
colleagues over the years. This association and the innumerable discussions engen-
dered by it have benefited me immensely in terms of knowledge, perception, and
understanding. I am thankful to all of them. In the context of the Kroll lecture paper
and the ideas and results reported herein, it gives me great pleasure to put on record
my sincere appreciation of the contributions made by Late Dr. P. Mukhopadhyay,
Late Dr. S. J. Vijayakar, Dr. J. K. Chakravartty, Dr. G. K. Dey, Dr. R. Kishore, Dr. K.
Madangopal, Dr. E. S. K. Menon, Dr. D. Srivastava, and Dr. R. Tewari. Last but not
least, I express heartfelt gratitude to Bhabha Atomic Research Centre, the institution
which has nurtured and shaped me. I am really proud to belong to this great
institution.
BANERJEE, DOI 10.1520/STP154320130039 49
References
[1] Sikka, S. K., Vohra, Y. K. and Chidambaram, R., “Omega Phase in Materials,” Prog. Mater.
Sci., Vol. 27, 1982, pp. 245–310.
[2] Banerjee, S. and Mukhopadhyay, P., Phase Transformations in Alloys: Examples from Tita-
nium and Zirconium Alloys, Pergamon Press, London, UK, 2007.
[3] Banerjee, S., Tewari, R., and Mukhopadhyay, P., “Coupling of Displacive and Replacive
Ordering,” Prog. Mater. Sci., Vol. 42, Nos. 1–4, 1997, pp. 109–123.
[4] Gaunt, P. and Christian, J. W., “The Crystallography of the b-a Transformation in
Zirconium and in Two Titanium-Molybdenum Alloys,” Acta Metall., Vol. 7, 1959, pp.
534–543.
[5] Mackenzie, J. K. and Bowles, J. S., “The Crystallography of Martensite Transformations-

IV Body-Centred Cubic to Orthorhombic Transformations,” Acta Metall., Vol. 5, No. 3,
1957, pp. 137–149.
[6] Banerjee, S. and Krishnan, R., “Martensitic Transformation in Zirconium-Niobium Alloys,”

Acta Metall., Vol. 19, 1971, pp. 1317–1326.
[7] Banerjee, S. and Krishnan, R., “Martensitic Transformation in Zr–Ti Alloys,” Metall. Trans.,
Vol. 4, 1973, pp. 1811–1819.
[8] Sass, S. L., “The Structure and Decomposition of Zr and Ti b.c.c. Solid Solutions,” J. Less
Common Metals, Vol. 28, 1972, pp. 157–173.
[9] Silcock, J. M., Davies, M. H., and Hardy, H. K., “Structure of the x-Precipitate in Titanium-
16 Per Cent Vanadium Alloy,” Symposium on the Mechanism of Phase Transformations in
Metals, Institute of Metals, London, 1955.
[10] Moffat, D. L. and Larbalesiter, D. C., “The Competition Between Martensite and Omega in
Quenched Ti–Nb Alloys,” Metall. Trans., Vol. 19A, 1988, pp. 1677–1686.
[11] Flewitt, P. E. J., “Phase Transformations in Niobium 16 to 40 % Zirconium Alloys Above

the Monotectoid Temperature—I&II,” Acta Metall., Vol. 22, 1974, pp. 47–63.
[12] Menon, E. S. K., Banerjee, S., and Krishnan, R., “Application of Free Energy-Composition
Diagrams in Predicting the Sequences of Phase Transformations in Zr–Nb Alloys,” Metall.
Trans., Vol. 9A, 1978, pp. 1213–1220.
[13] Banerjee, S., Vijayakar, S. J., and Krishnan, R., “Precipitation in Zirconium-Niobium
Martensites,” J. Nucl. Mater., Vol. 62, 1976, pp. 229–239.
[14] Mukhopadhyay, P., Menon, S. K., Banerjee, S., and Krishnan, R., “Beta Precipitation in Zr-
5 Ta Martensite–Microstructure,” Z. Metallkunde, Vol. 69, 1978, pp. 725–731.
[15] Abriata, J. P. and Bolcich, J. C., “The Nb–Zr (Niobium–Zirconium) System,” Bull. Alloy
Phase Diag., Vol. 3, 1982, pp. 34–44.
[16] Srivastava, D., Dey, G. K., and Banerjee, S., “Evolution of Microstructure During Fabrica-
tion of Zr–2.5 Wt pct Nb Alloy Pressure Tubes,” Metall. Mater. Trans. A, Vol. 26A, 1995,
pp. 2707–2734.
[17] Kishore, R., Banerjee, S., and Rama Rao, P., “First Report on Observation of Abnormal
Creep in a Zr–2.5wt.%Nb Alloy at Low Stresses,” J. Mater. Sci., Vol. 44, No. 9, 2009, pp.
2247–2256.
[18] Srivastava, D., Mukhopadhyay, P., Banerjee, S., and Ranganathan, S., “Morphology and
Substructure of Lath Martensites in Dilute Zr–Nb Alloys,” Mater. Sci. Eng. A, Vol. 288,
2000, pp. 101–110.
[19] Srivastava, D., Banerjee, S., and Ranganathan, S., “The Crystallography of BCC to HCP
Martensitic Transformation in Dilute Zr-Nb Alloys: Part I: Application of Phenomenologi-
cal Theory of Martensitic Transformation,” Trans. Indian Inst. Metals, Vol. 57, No. 3, 2004,
pp. 205–223.
[20] Srivastava, D., Banerjee, S., and Ranganathan, S., “The Crystallography of BCC to HCP
Martensitic Transformation in Dilute Zr-Nb Alloys: Part II: Application of Phenomenologi-
cal Theory of Martensitic Transformation,” Trans. Indian Inst. Metals, Vol. 57, No. 5, 2004,
pp. 427–449.
[21] Banerjee, S., Dey, G. K., Srivastava, D., and Ranganathan, S., “Plate Shaped Transformation
Products In Zirconium Based Alloys,” Metall. Mater. Trans., Vol. 28, 1997, pp. 2201–2216.
[22] Srivastava, D., Madangopal, K., Banerjee, S., and Ranganathan, S., “Self Accommodation
Morphology of Martensite Variants in Zr–2.5wt%Nb Alloy,” Acta Metall. Mater., Vol. 41,
1993, pp. 3445–3454.
[23] Perovic, V. and Weatherly, G. C., “The Beta To Alpha Transformation In A Zr–2.5 wt-
Percent Nb Alloy,” Acta Metall., Vol. 37, 1989, pp. 813–821.
[24] Zhang, W.-Z. and Weatherly, G. C., “On the Crystallography of Precipitation,” Prog.
Mater. Sci., Vol. 50, 2005, pp. 181–292.
[25] Zhang, W.-Z. and Purdy, G. R., “A TEM Study of the Crystallography and Interphase
Boundary Structure of a Precipitates in a Zr–2.5 wt. % Nb Alloy,” Acta Metall. Mater., Vol.
41, 1993, pp. 543–551.
[26] Furuhara, T., Ogawa, T., and Maki, T., “Atomic-Structure of Interphase Boundary of an
Alpha-Precipitate Plate in a Beta–Ti–Cr Alloy,” Philos. Mag. Lett., Vol. 72, 1995, pp.
175–183.
[27] Banerjee, S., Tewari, R., and Dey, G. K., “Omega Phase Transformation—Morphologies
and Mechanisms,” Int. J. Mater. Res., Vol. 97, 2006, pp. 963–977.
[28] Dey, G. K., Tewari, R., Banerjee, S., Jyoti, G., Gupta, S. C., Joshi, K. D., and Sikka, S. K.,
“Formation of a Shock Deformation Induced x Phase in Zr–20Nb Alloy,” Acta Mater.,
Vol. 52, 2004, pp. 5243–5254.
[29] Dey, G. K., Tewari, R., Banerjee, S., Jyoti, G., Gupta, S. C., Joshi, K. D., and Sikka, S. K., “A
New High-Pressure Phase Transition in a Zirconium-Niobium Alloy,” Philos. Mag. Lett.,
Vol. 82, 2002, pp. 333–340.
[30] Dey, G. K., Singh, R. N., Tewari, R., Srivastava, D., and Banerjee, S., “Metastability of the
b-Phase in Zr-rich Zr-Nb Alloys,” J. Nucl. Mater., Vol. 224, 1995, pp. 146–157.
[31] Nag, S., Devaraj, A., Srinivasan, R., Williams, R. E. A., Gupta, N., Viswanathan, G. B., Tiley,
J. S., Banerjee, S., Srinivasan, S. G., Fraser, H. L., and Banerjee, R., “Novel Mixed-Mode
BANERJEE, DOI 10.1520/STP154320130039 51
Phase Transition Involving a Composition-Dependent Displacive Component,” Phys.

Rev. Lett., Vol. 106, 2011, 245701.
[32] Banerjee, S., Vijaykar, S. J., and Krishnan, R., “Strength of Zirconium–Titanium Marten-
sites and Deformation Behaviours,” Acta Metall., Vol. 26, 1978, pp. 1815–1831.
[33] Griffiths, M., Mecke, J. F., and Winegar, J. E., “Evolution of Microstructure in Zirconium
Alloys During Irradiation,” Zirconium in the Nuclear Industry: Eleventh International Sym-
posium, ASTM STP 1295, E. R. Bradly and G. P. Sabol, Eds., Sept 11–14, 1995, ASTM Inter-
national, West Conshohocken, PA, 1996.
[34] Banerjee, S. and Cahn R. W., “Formation of the Zr2Al Phase in b- Zr-Al Alloy: a Hybrid
Displacive and Replacive Ordering,” In Proceedings of International Conference on
Solid ! Solid Phase Transformations, Carnegie-Mellon, University, H. I. Aaronson,
D. E. Laughlin, R. F. Sekerka and C. M. Wayman, Eds., Pittsburgh, PA, USA, 1982,
pp. 1005–1009.
[35] Banerjee, S. and Cahn, R. W., “An Ordered x-Phase in the Rapidly Solidified Zr–27
at.%Al Alloy,” Acta Metall., Vol. 31, 1983, pp. 1721–1735.
[36] Tewari, R., Mukhopadhyay, P., Banerjee, S., and Bendersky, L. A., “Evolution of Ordered x
Phases in (Zr3Al)-Nb Alloys,” Acta Mater., Vol. 24, 1999, pp. 1307–1323.
[37] Perovic, V., Weatherly, G. C., and Simpson, C. J., “The Role of Elastic Strains in the For-
mation of Stacks of Hydride Precipitates in Zirconium Alloys,” Scr. Metall., Vol. 16, No. 4,
1982, pp. 409–412.
[38] Dey, G. K., Banerjee, S., and Mukhopadhyay, P., “Formation of Gamma Hydride in Alpha
and Beta Zirconium Alloys,” J. Phys., Vol. 43, No. C4, 1982, pp. C4-327–C4-332.
[39] Dey, G. K. and Banerjee, S., “Decomposition of the b-Phase in Zr-20 %Nb,” J. Nucl.
Mater., Vol. 124, 1984, pp. 219–227.
[40] Srivastava, D., Neogy, S., Dey, G. K., Banerjee, S., and Ranganathan, S., “Application of
Invariant Plane Strain (IPS) Theory to c Hydride Formation in Dilute Zr-Nb Alloys,” Mater.
Sci. Eng. A, Vol. 397, Nos. 1–2, 2005, pp. 138–144.
[41] Jaffee, R. I., “The Physical Metallurgy of Titanium Alloys,” Prog. Metal Phys., Vol. 7, 1958,
pp. 65–106.
[42] Mukhopadhyay, P., Menon, E. S. K., Banerjee, S. and Krishnan, R., “Active Eutectoid
Decomposition in a Near-Eutectoid Zirconium-Copper Alloy,” Metall. Trans., Vol. 10A,
1979, pp. 1071–1084.
[43] Kumar, L., Ramanujan, R. V., Tewari, R., Mukhopadhyay, P., and Banerjee, S., “Active
Eutectoid Decomposition in Zr-3 wt.% Fe,” Scr. Mater., Vol. 40, 1999, pp. 723–728.
[44] Tewari, R., Srivastava, D., Dey, G. K., Chakravarty, J. K., and Banerjee, S., “Microstructural
Evolution in Zirconium Based Alloys,” J. Nucl. Mater., Vol. 78, 2008, pp. 153–171.
SCHEMEL AWARD PAPER
M. Christensen,1 W. Wolf,1 C. Freeman,1 E. Wimmer,1

R. B. Adamson,2 L. Hallstadius,3 P. Cantonwine,4
and E. V. Mader5
Effect of Hydrogen on
Dimensional Changes of
Zirconium and the Influence
of Alloying Elements: First-
Principles and Classical
Simulations of Point Defects,
Dislocation Loops, and Hydrides
Reference
Christensen, M., Wolf, W., Freeman, C., Wimmer, E., Adamson, R. B., Hallstadius, L.,
Cantonwine, P., and Mader, E. V., “Effect of Hydrogen on Dimensional Changes of Zirconium
and the Influence of Alloying Elements: First-Principles and Classical Simulations of Point
Defects, Dislocation Loops, and Hydrides,” Zirconium in the Nuclear Industry: 17th
International Symposium, STP 1543, Robert Comstock and Pierre Barberis, Eds., pp. 55–92,
doi:10.1520/STP154320120170, ASTM International, West Conshohocken, PA 2015.6
ABSTRACT
Hydrogen-assisted irradiation growth may result in significant channel bow in
addition to regular fluence gradient-induced bow in boiling-water reactor (BWR)
fuel channels, especially at high exposures through “shadow corrosion,” if
hydrogen is picked up early in channel sides facing a control rod. This
phenomenon may be responsible for recent high channel bow observations. To
Manuscript received November 19, 2012; accepted for publication August 13, 2013; published online June 17,
2014.
1
Materials Design, Inc., Santa Fe, NM 87501, United States of America.
2
Zircology Plus, Freemont, CA 94538, United States of America.
3
Westinghouse Electric Sweden, Västerås, Sweden.
4
Global Nuclear Fuels, Wilmington, NC 28402, United States of America.
5
Electric Power Research Institute (EPRI), Palo Alto, CA 94304, United States of America.
6
Hyderabad, India.
Copyright V
develop a better understanding of the effect of hydrogen on dimensional changes

of channel materials, first-principles calculations combined with embedded-atom
molecular dynamics simulations have been performed under EPRI’s BWR channel
distortion program. The simulations reveal that: (1) H dissolved in zirconium
expands the lattice; (2) the volume effect of H in solution and as hydride is similar;
(3) regions under tensile strain attract hydrogen; (4) near Ni atoms the binding of
H is increased, and reduced near Sn and Nb; (5) 1 % Zr vacancies decrease the
volume by 0.44 % and 1 % Zr self-interstitial atoms (SIAs) expand the volume by
up to 1.2 %; (6) the bulk modulus of hydrides rises with increasing H concentration,
the shear modulus of hydrides is similar to that of pure Zr, while Young’s modulus
decreases; (7) coalescence of isolated vacancies into dislocation loops releases up
to 80 kJ/mol; (8) vacancy dislocation loops larger than 10–15 Å in diameter tend to
collapse thereby shrinking the lattice; (9) interstitial hydrogen is attracted to
isolated vacancies and vacancy loops and can retard or prevent their collapse; (10)
the diffusivity of interstitial Zr is higher than that of interstitial H atoms, diffusion
of vacancies is slower; (11) substitutional Fe and Cr atoms spontaneously swap
with interstitial Zr atoms and diffuse rapidly in the c direction; and (12) Nb
impedes the diffusion of Zr self interstitials, thus reducing the buildup of a-loops.
These simulations confirm some trends observed in material test reactors followed by
advanced transmission electron microscopy (TEM). Simulations can be used to help
optimize the materials properties for development of future channel alloys to
minimize their in-service distortion up to very high fluence.
Keywords
Zircaloy, computer modeling, zirconium, hydrogen effects, zirconium hydrides,
solubility, defects, diffusion, dislocation loops, atomistic simulation
Introduction
Zirconium alloys are the primary structural materials in the core of nuclear power
reactors. The combination of low neutron absorption cross section with good me-
chanical properties and corrosion resistance at 250 C–350 C (523–623 K) in water
allows zirconium alloyed with various combinations of Sn, Fe, Cr, Ni, Nb, and O to
be used as fuel-rod cladding and structural components in commercial reactors
throughout the world.
The corrosion reaction between water and Zr produces hydrogen, a portion of
which is absorbed into the metal. Hydrogen solubility in Zr is negligible at room
temperature and is in the range 30–130 ppm for typical reactor component temper-
atures. Above the solubility limit, zirconium hydride is formed having different
crystal structures depending on the ratio of hydrogen to zirconium. Typically, but
not exclusively, post-irradiation examinations reveal FCC delta hydrides with H/Zr
ratios of about 1.6. At any given temperature, a hydride will dissolve at the solubil-
ity value, but hydrogen in solution will precipitate only at an increased supersatu-
rated value [1]. The volume of delta hydride is about 16 % higher than the
zirconium matrix; therefore, extra energy is required to fit the hydride into the
CHRISTENSEN ET AL., DOI 10.1520/STP154320120170 57
zirconium lattice. As a result, hydrides tend to precipitate on grain boundaries and

on energy-favorable planes, which for the Zircaloys are f1017g, about 15 from the
basal plane [2].
Hydrides clearly have influence on mechanical properties, dramatically lower-
ing ductility of irradiated or un-irradiated zirconium alloys at temperatures below
about 200 C (473 K).
Delta hydride is expected to expand the lattice such that 1000 ppm H results in
a volume increase of about 1 %. Assuming isotropy following King et al. [3] (which
Carpenter [4] found not to be the case), a length change of 0.33 % would occur. Ex-
perimental data on non-irradiated material at room temperature indicate the
observed length change in the basal plane is about 0.2 % [3,5,6], which approxi-
mately accounts for the known anisotropy.
The fate of H in the Zr lattice is under increasing scrutiny. Attraction of H to
dislocations has been demonstrated by [7]. Increased mobility of dislocations in a
hydrogen environment has been shown [8]. Trapping of H by irradiation-produced
defects was indicated [9]. It has been reported that H is absorbed by second phase
precipitates (SPPs), such as the Laves-phase Zr(Fe,Cr)2 SPPs that occur in Zircaloy
[10].
More recent work indicates an attraction of H to irradiation-produced disloca-
tion loops. Detailed solubility measurements [11,12] indicate that the apparent sol-
ubility of H is increased by irradiation. Importantly, post-irradiation annealing in
the same temperature ranges that eliminate <a>-and/or <c>-component disloca-
tion loops tends to eliminate the observed increase in the apparent solubility of H.
A summary of the early results is available [13].
It is thought that hydrogen or hydrides can increase the irradiation growth rate
or decrease the threshold fluence for breakaway growth, but the mechanism has not
yet been defined. Russian scientists report that pre-hydriding alloy E110 with 200
or 500 ppm H increased the growth rate by a factor of 6 compared to non-hydrided
material [14]. Also, recent data from EPRI’s nuclear fuel industry research (NFIR)
dimensional stability project suggest that hydrogen increased irradiation growth
and that the effect was alloy-sensitive [15]. These results and others [16,17] lead to
hypotheses that hydrogen or hydrides may be interacting with <c>-component
dislocation loops,which are known to be correlated to enhanced growth in a variety
of zirconium alloys [18].
Hydrogen-enhanced irradiation growth may be a contributor to the occurrence
of significant channel bow observed in nearly half of operating U.S. BWRs [15].
This is particularly important at high exposures caused by enhanced hydrogen
pickup sometimes associated with shadow corrosion [15,19] and with dissolution of
Ni-bearing precipitates in Zircaloy-2 [20]. However, others have concluded that the
role of hydrogen in channel bow is driven mainly by the volume change associated
with hydrogen absorption [19]. To gain detailed insight into the mechanisms
underlying radiation-induced BWR channel distortion, first principles combined
with embedded molecular dynamics simulations were performed for zirconium in
the presence of lattice defects, alloying elements and hydrogen. Results are pre-
sented in this paper.
Computational Approach
The present project employs two different modeling approaches. The first uses ab
initio (or first principles) density functional theory (DFT) [21,22]. The high compu-
tational accuracy of the first-principles approach is needed to obtain reliable struc-
ture and energy data, including diffusion barriers and thermodynamic functions.
The other modeling approach employs a quasi-classical method to parameterize
interactions, specifically the embedded atom method (EAM) [23,24], and the
LAMMPS program [25] to run simulations. The EAM is used in the study of dislo-
cation loops and to investigate the more complex diffusion pathways for interstitial
diffusion.
The Vienna ab initio simulation package (VASP) version 5.2 [26–29] within
R
the MedeAV computational environment [30] has been used to compute the struc-
tural and thermodynamic properties of H in pure and alloyed zirconium. The com-
plex interactions between the electrons (exchange and correlation effects) in the
field of the positive nuclei are described by the generalized gradient approximation
(GGA) with the functional form proposed by Perdew-Burke-Ernzerhof (PBE) [31].
The density-functional or Kohn-Sham equations are solved with the projector-aug-
mented-wave (PAW) potentials [32], including for Zr the s and p semicore states,
for Cr and Nb the p semicore states, and for Sn the d semicore states. The plane
wave cutoff is 400 eV. Geometry optimizations and total energies are converged to a
maximum force of at most 0.02 eV/Å and to energy variations of at most
0.00001 eV, respectively, and the Brillouin zone is sampled with a regular C-
centered mesh of k-points with a density of about 0.15 Å1. For phonon calcula-
tions, transition state search, and configurational sampling of hydrides, the k spac-
ing is increased to about 0.3 Å1.
In this simulation, a solid material is modeled by a three-dimensional periodic
array of cells, which contain between about 40 and 50 up to about 180 atoms—such
as a Zr lattice with an interstitial H atom, a defect or an alloying element—for ab
initio calculations and up to about 10000 atoms and millions of configurations for
R
embedded atom molecular dynamics simulations. The MedeAV-VASP and
LAMMPS software system allows the construction of structural models, the full
relaxation of all atoms to their equilibrium position, the calculation of stress ten-
sors, the computation of equilibrium lattice parameters, molecular dynamics simu-
lations, and the analysis of the results.
The minimum energy paths for diffusion processes are mapped out by means
of the nudged elastic band approach (NEB) [33] making use of ab initio VASP cal-
culations. The transition states are identified by the climbing image technique
within NEB [30,34]. The thermodynamic partition functions and temperature-
dependent thermodynamic functions (internal energy, vibrational entropy, and free
energy) needed for solubility and diffusivity are computed from ab initio phonon
calculations [30,35]. Kinetic Monte Carlo simulations are used to explore diffusion
processes in complex energy hypersurfaces. For solubility calculations and deriving
solubility limits, the Gibbs free energy and chemical potential for HCP Zr and the
hydride phases are computed for a set of discrete concentrations from ab initio total
energy calculations. The temperature-dependent vibrational terms are obtained
from quasi-harmonic phonon calculations. In the case of vibrations of H atoms in
tetrahedral interstitial sites in Zr, the quasi-harmonic approach is augmented by a
numerical integration of the one-dimensional Schrödinger’s equation for the anhar-
monic potential energy hypersurface between two adjacent tetrahedral sites. An in-
dependent two-site model is applied to account for configurational entropy, which
assumes that for a given phase of the solid (i.e., HCP Zr or hydrides) and a given
temperature, the chemical potential is dependent on the concentration of the solute
(i.e., hydrogen) and on the site (i.e., tetrahedral or octahedral), but not on the spe-
cific configuration within the available sites (see Ref 36 for further details). The
temperature-dependent terms of the enthalpy and the entropy of hydrogen mole-
cules in the gas phase are taken from experimental data [37].
DFT on the GGA level typically reproduces experimental lattice constants
within a few %. For the computed structures of the dominant hydrides (e-, d-, c-
hydride), the maximum deviation from experimental lattice constants is 1.5 %. The
accuracy of computed energy differences is typically in the range 5–10 kJ/mol. Elas-
tic constants are computed ab initio from the stress tensors as described in Ref 38
R
and implemented in MedeAV [30], and are generally within 10–20 % from experi-
mental values, although quite often the variation between different measurements
of elastic coefficients has a similar spread. The polycrystalline elastic shear and
Young’s moduli are derived from the computed single crystal elastic constants by
means of the averaging scheme of Hill. Thermal expansion is not included in the
computations except where explicitly stated. For those cases, thermal expansion is
computed by means of the quasi-harmonic approach based on fitting to ab initio
total energy and phonon calculations for a set of expanded lattices.
Hydrides with hydrogen content between H/Zr ¼ 0.25 to 2.0 have been investi-
gated in the present study. Based on formation energy, symmetry, and phase char-
acteristics for hydrogen concentrations of H/Zr ¼ 0.25, 0.5, 0.75, 1.0, 1.25, 1.5, 1.75,
and 2.0, one hydride phase is chosen to represent each stoichiometry in the investi-
gation of hydride properties. It is satisfying that the experimentally observed
hydride phases are found to be the most stable for their respective stoichiometries.
The c-ZrH hydride with symmetry P42/mmc has the lowest electronic energy of
formation in the interval 0.75 1.25. f-hydride is the most stable Zr2H hydride
with ABA stacking of the Zr sublattice. However, the electronic energy is lower for
Zr2H hydride with ABC stacking of the Zr sublattice, indicating a meta-stable na-
ture of f-hydride. In this study, hydrides with HCP stacking of the host lattice (f-
hydride) is used up to H/Zr ¼ 0.5. The structures of c-, d-, and e-hydrides are
shown in Fig. 1.
FIG. 1 Atomic structures of (a) c-hydride (I-4m2), (b) c-hydride (P42/mmc), (c) d-
hydride (Fm-3 m), and (d) e-hydride (I4/mmm).
The effect by alloying elements is studied by separately adding the elements to

the Zr-H system. The present investigation includes the alloying/impurity elements
Sn, Nb, Cr, Fe, Ni, and O. Starting from zirconium and pure hydrides, the alloying
elements are added in concentrations of about 2.7–4.0 wt. % for Sn and Nb,
1.6–2.0 wt. % for Cr, Ni, and Fe, and 0.5 wt. % for O, depending on the element and
stoichiometry of the hydride. All elements except oxygen are inserted as substitu-
tional elements, replacing a Zr atom in the lattice. Oxygen is treated as an intersti-
tial element. The concentrations in the models are somewhat higher than in actual
Zircaloy for two reasons, namely computational efficiency and a clearer effect com-
pared with pure Zr. With supercell sizes between about 60 and 180 atoms, contain-
ing at least 36 metal atoms and for hydrides corresponding to the above
compositions, the interaction of impurity atoms between adjacent supercells is suf-
ficiently small so that the results can be extrapolated to the dilute cases.
The Zr EAM potential version V#3 optimized by Mendelev and Ackland [39]
for HCP Zr served as a starting point to develop EAM potentials for the Zr-H sys-
tem by including Zr-H and H-H potential terms. For the hydrogen embedding
energy function F, the density function qi for H atoms, and the pairwise interaction
term Vij as a function of the distance rij between two atoms, Zr-H and H-H, enter
the EAM energy expression.
X X
(1) E¼ F ðqi Þ þ V rij
i j>i
the following analytical forms are used:
(2a) F ðqÞ ¼ b1 q þ b2 q2 þ b3 q3 q 300

h 0 i1

F ðqÞ ¼ b1 q þ b2 q2 þ b3 q3 1 ec ðqc qÞ þ 1
0
h 0 i1
(2b) þ gea q ec ðqc qÞ þ 1 q > 300
(3) qðr Þ ¼ aH eaH r

h 00 i1
(4) VZrH ¼ D½1 þ bðr r0 Þebðrr0 Þ ec ðrrc Þ þ 1
TABLE 1 Optimized parameters for the Zr-H and H-H EAM potential including H embedding and
density functions as obtained from a fit to ab initio data.
Interaction D (eV) b (Å1) r0 (Å)
Zr-H 0.26360607905 4.1830297876 2.0458175862

H-H 0.25076933842 4.1394808138 2.6750531158
b1 [eV(arb. unit)1] b2 [eV/(arb. unit)2] b3 [eV/(arb. unit)3] aH (arb. units) aH (Å1)
8.9072864260e-03 2.1419447889e-04 7.5863505373e-07 9.5494498803eþ01 1.3004656587eþ00
The parameters a0 ¼ 0.01, g ¼ 0.3, c0 ¼ 0.04, qc ¼ 380, and c00 ¼ 5.0 and rc ¼ 6.5 for
both, Zr-H and H-H interactions, ensure proper asymptotic behavior and are kept
at fixed values. The other 11 parameters b1, b2, b3, DZr-H, bZr-H, r0Zr-H, DH-H, bH-H,
r0H-H, aH, and aH are determined in an iterative non-linear least-square procedure
such that the EAM results reproduced as close as possible a training set obtained
from quantum mechanical ab initio calculations. The training set includes surface
segregation energies, the trapping energy of H atoms clustering in a Zr vacancy as a
function of an increasing number of H atoms, and the energy difference between
octahedral and tetrahedral sites in the bulk and at surfaces. The resulting parame-
ters for the newly created Zr-H and H-H potential and the H embedding and den-
sity functions are summarized in Table 1. This parameterization reproduces the ab
initio data used for the fitting within a few kJ/mol and is able to reproduce other
properties such as the swelling of Zr as a function of H contents. As a limitation,
the potential is not well suited to reproduce the behavior of other Zr phases or
hydrogen-rich hydride phases.
EAM potentials for Zr including alloying elements are similarly developed by
fitting to ab initio molecular dynamics trajectories. The detailed functional form
and resulting parameters can be found in Ref 40.
Results and Analysis

The main objective of the present study is a clearer understanding of hydrogen-
enhanced irradiation growth in the Zr-H system. To this end, we first compute as
baseline the volume increase of Zr as a function of H concentration for defect-free
a-Zr and in hydrides. Subsequently, we investigate the impact of defects such as
vacancies, self-interstitials, and dislocations as well as the role of alloying elements
and O impurities.
VOLUMETRIC EFFECTS IN THE ZR-H SYSTEM WITHOUT DEFECTS

Hydrogen in Solution
At temperatures up to 1134 K and ambient pressure, Zr crystallizes in a hexagonal
close-packed (HCP) structure known as a-Zr. This structure provides one octahe-
dral and two tetrahedral interstitial sites per Zr atom. Experimental evidence
[41,42] indicates a site preference of H atoms for interstitial tetrahedral sites. This is
intuitively plausible because the end point of hydration of a-Zr is the fully stoichio-
metric hydride e-ZrH2, where all tetrahedral sites are occupied. Recent ab initio cal-
culations by Domain et al. [43] resulted in an energy preference of 5.5 kJ/mol for H
in tetrahedral sites based on total electronic energies. The present work reveals that
the picture is probably more subtle.
Thermodynamically, the equilibrium between different configurations is deter-
mined not only by the total electronic energy, but by the free energy, which depends
on the temperature. This has been computed in the present work including
temperature-dependent vibrational effects. The results show that the tetrahedral site
is thermodynamically favored by only 0.5 kJ/mol at T ¼ 0 K increasing to 8.6 kJ/mol
at T ¼ 600 K. As a consequence, a statistically significant fraction of interstitial H
atoms dissolved in a-Zr occupy octahedral sites. For example, at 600 K, about 6 % of
all dissolved H atoms are in octahedral sites. Nevertheless, the majority of H is in tet-
rahedral sites. The computational details of this analysis are described in Ref 36.
Up to the solubility limit, which is computed to be 0.006 wt.-ppm hydrogen at
25 C increasing to 41–60 wt.-ppm hydrogen at 300 C [11,36], hydrogen atoms
occupy interstitial sites in the Zr lattice. Beyond the solubility limit, additional
hydrogen leads to the precipitation of zirconium hydrides. Hydrogen both in solu-
tion and as hydride expands the Zr lattice. The overall behavior of the volume as a
function of H concentration from the dilute limit up to the fully hydrided system e-
ZrH2 is illustrated in Fig. 2.
FIG. 2 Volume change as a function of hydrogen concentration in wt. ppm and H/Zr
ratio (upper scale). The solid black curve is an analytical fit to the computed
values. The black squares are experimental data [4,44].
As can be seen in Fig. 2, the present ab initio calculations are in very good
agreement with the experimental results discussed below [4,44], thus validating the
approach. H has a larger effect on the volume expansion at low concentrations than
at higher concentrations. This is related to the fact that the lattice of a-Zr is some-
what too tight for the optimal Zr-H distance, whereas the expanded Zr-Zr distance
in hydrides makes it easier to insert additional H atoms without straining the lattice
too much.
At first it is perhaps surprising that there is no discontinuity between H dis-
solved in a-Zr and the hydrides. This can be rationalized by the fact that the local
environment of H atoms in a-Zr and in the hydride phase is quite similar. For
example, in c-ZrH, d-ZrH2-x, and e-ZrH2, the Zr atoms form (slightly distorted)
cubic close-packed structures with the tetrahedral interstitial sites being occupied
by H atoms. In a hexagonal lattice, adjacent tetrahedral sites share a common face,
which would result in an energetically unfavorable short H-H distance if two neigh-
boring sites were occupied. The system avoids this situation by shifting to a cubic
close-packed structure, where tetrahedral interstitial sites are separated by empty
octahedral sites. In other words, the solution of H in a-Zr and the formation of
hydrides can be seen as the gradual filling of all interstitial tetrahedral sites, the
ordering of the H atoms, and a transformation of the Zr host lattice from an HCP
to a (distorted) FCC lattice to avoid close contact of H atoms.
The computed volume change (%) caused by a concentration CH,sol (in wt.-
ppm) of H in solution in a-Zr can be conveniently expressed as
(5a) DV=V0 100 ¼ 1:7 103 CH;sol
and assuming isotropic behavior within the solubility limit (see Ref 36 for the com-
puted anisotropy data), the linear expansion (%) results in the expression
(5b) Dl=l0 100 ¼ 5:6 104 CH;sol
Of course, the concentrations and thus dimensional changes are bound by the solu-
bility limit of H in a-Zr, which increases with increasing temperatures [36]. It is
noted, therefore, that at operating temperatures dimensional changes can be larger
than measured at room temperature.
Volumetric Effect of Hydrides

The equilibrium volume of each hydride phase has been computed in the whole
stoichiometry range from f-hydride (Zr4H) to e-hydride (ZrH2). Precipitation of e-
ZrH2 (full occupancy of tetrahedral sites by H) results in a computed volume
expansion of 17.9 % compared with pure a-Zr. Carpenter reports a relative volume
difference of 17.2 % between a-Zr and d-hydride (ZrH1.66, with a ¼ 4.778 Å) [4].
The computations give a volume expansion of 16.6 % for d-ZrH1.5.
Numakura et al. give a relative volume difference of 13 % between c-hydride
and the Zr metal matrix [44]. This can be compared to the computed volume
expansion of the P42/mmc c-hydride, which was found to be the most stable
hydride in the composition range ZrH0.75 to ZrH1.25. For c-ZrH0.75, the relative vol-
ume difference is 10.1 % and for c-ZrH1.25, it is 13.6 %. Carpenter gives a relative
volume difference of 12.3 % for the c-hydride [4]. This volume expansion corre-
sponds to that of c-ZrH in the present study.
As shown in Fig. 2, the volume increase is non-linear and is tending to plateau
at ZrH2. The volume expansion is expressed as the change in volume as a function
of the hydrogen content in wt.-ppm. The solid curve is an analytic fit to the com-
puted values. The volume change (%) caused by a concentration CH,hydride of H in
hydride in wt.-ppm is

DV=V0 100 ¼ MZr CH;hydride = X MH 106 CH;hydride

23:15 1 e0:7452X ¼ 2095:2 1 e0:7452X

(6) CH;hydride = X 106 CH;hydride
where
MZr and MH ¼ the molar masses of Zr and H, and
X ¼ the H/Zr ratio of the hydride.
A closer analysis of the strain versus H concentration at small H concentrations
is shown in Fig. 3. Up to a concentration of a few hundred wt.-ppm H, there is no
significant difference between H being dissolved or precipitated as a hydride. On
the other hand, if d-hydride is formed, then the swelling effect will be smaller than
calculated for the metal. As an example, if 250 ppm of hydrogen is absorbed, of
which 50 ppm is in solution and 200 ppm is in hydrides (e.g., the d-phase), 200 ppm
H forming ZrH1.6 d-hydride gives a volume change of 0.18 %:

(7) DV=V0 100 ¼ 2095:2 ð1 e0:74521:6 Þ 200= 1:6 106 200 ¼ 0:18
whereas 50 ppm H in solution in a-Zr gives a volume change of 0.09 %:
(8) DV=V0 100 ¼ 1:7 103 50 ¼ 0:085
Hence, in this case, the volume expansion caused by hydrogen in solution is almost
twice the expansion as a result of hydrogen in the hydride (the amount of hydrogen
in hydrides is four times the amount in solution). The total volume expansion is
0.18 þ 0.09 ¼ 0.27 %. In general, if hydrogen is present partly in solution and partly
in hydrides, the combined effect on the volume expansion is
DV=V0 100 ¼ 2095:2 ð1 e0:7452 XÞ

(9) CH;hydride = X 106 CH;hydride þ 1:7 103 CH;sol
Measured data of the linear dimensional change versus hydrogen content have been
reported by King et al. [3]. The data, as reproduced in Ref 45 are shown in Fig. 3.
Assuming isotropic behavior (to compare with the data by King et al.), 1.7 % vol-
ume increase in pure Zr corresponds to 0.56 % linear increase and 1 % volume
FIG. 3 Computed linear expansion of a-Zr (solid line) and hydride (dashed line) versus
hydrogen content. Experimental points for polycrystalline samples of ZIRLO and
Zircaloy-4 are redrawn from Fig. 7-2 in Ref 45.
increase, which is the expected effect because of d-hydride to 0.33 % linear increase.
Whereas it is meaningful to describe anisotropy of the volume expansion as a func-
tion of hydrogen concentration in pure zirconium (lattice parameter c increases
more than lattice parameter a when hydrogen is inserted [36]), the precipitation of
hydride phases blur the anisotropy. For example, cubic zirconium hydride is iso-
tropic. If all of the hydrogen forms e-ZrH2, the calculated swelling would be 0.8 %
for 1000 ppm H, corresponding to 0.26 % linear increase. This actually coincides
with the largest experimental value of King et al. for this concentration.
Hence, depending on the stoichiometry of the hydride or the distribution of
hydrogen between the hydride and metal phase, volume increases between 0.8 %
and 1.7 % are calculated for 1000 ppm H, the former coinciding with the upper
bound of experimental data of King et al. for this concentration. The model calcula-
tions do not take into account defects, such as vacancies, dislocations, and grain
boundaries, which can trap hydrogen without noticeable volume effect. Such con-
siderations might explain why King et al. measured a linear increase of 0.23 % per
1000 ppm H, somewhat lower than the theoretical value of 0.33 % [46], and the
value of 0.56 % computed for a defect-free system. The fact that calculations pro-
vide an upper bound to measured volume increases is, therefore, rather reassuring
and permits an estimation of the fraction of H located in defects.
Elastic Properties
The non-linear dependence of the volume expansion as a function of H concentra-
tion shown in Fig. 2 indicates that hydrides become less compressible with
increasing hydrogen concentration. This is quantified by computing the elastic

moduli of a-Zr and zirconium hydrides. The result is shown in Fig. 4.
The computed bulk modulus increases almost linearly with the hydrogen con-
centration from 96 GPa for pure Zr to 132 GPa for ZrH2. Shear moduli of zirco-
nium hydrides are similar to the shear modulus of a-Zr, with about
30 GPa 6 10 GPa depending on the hydride stoichiometry. Young’s moduli of the
hydrides are typically somewhat lower than for a-Zr. The clear exception is ZrH1.25,
which is a c-hydride (P42/mmc) structure with high hydrogen content. The Young’s
modulus is 85 GPa 6 30 GPa depending on the stoichiometry.
Heat of Solution as Function of Strain

In a system containing H and Zr with three phases, namely a-Zr with dissolved H,
zirconium hydride, and a gas phase, one needs to consider the equilibrium between
the gas phase and H dissolved in a-Zr and the equilibrium between dissolved H and
the hydrides. Let us focus here on the equilibrium between the gas phase and dis-
solved H, which is determined by the Gibbs free energy of solution.
The computed Gibbs free energy of solution, i.e., the difference in Gibbs free
energy between hydrogen dissolved in Zr and as H2 molecules in the gas phase, as a
function of temperature, is shown in Fig. 5. As expected, the free energy of solution
becomes more positive with increasing temperature. This reflects the well-known
fact that heating a solid containing a dissolved gas causes outgassing.
In Fig. 5, the energy of solution is displayed for three different strain states of
the metal. The solubility decreases when the system is under compressive strain and
FIG. 4 Elastic moduli of a-Zr and the most stable Zr hydride structures for each
stoichiometry.
FIG. 5 Free energy of solution of interstitial H in Zr. The energies include corrections for
the thermal lattice expansion and non-harmonic effects in the tetrahedral site. A
volume change of 2 % corresponds to a linear strain of 0.66 % in each direction.
increases for tensile strain. In other words, regions under tensile strain attract
hydrogen. This is intuitively understandable, because H atoms bind preferentially
in tetrahedral interstitial sites, but the optimal H-Zr bond distance is slightly larger
than that available in pure Zr. Hence, an expansion of the lattice (positive strain)
creates the extra space for optimal H-Zr bonds, whereas compression expels H.
Earlier it was stated that the solubility limit of H in a-Zr increases with temper-
ature, which seems to be in contradiction with the outgassing mentioned above.
The explanation comes from the fact that the solubility limit is determined by the
equilibrium between the hydride phases and H dissolved in a-Zr. As explained in
detail in Ref 36, relative to the hydride phase the dissolved H becomes more stable
with increasing temperature.
Effect of Alloying Elements

The effect of alloying elements on the volumetric properties of both a-Zr and
hydrides has been studied here for Ni, Cr, Fe, Nb, and Sn, as well as for O
impurities. In systems where Zr hydrides coexist with pure Zr, it is of relevance to

investigate in which of the phases the alloying elements thermodynamically are
most stable. This is accomplished by computing the energy of inserting the ele-
ments into a-Zr and hydrides from their respective bulk phases.
The insertion energy of oxygen is strongly negative, which means that it is ener-
getically favorable for oxygen to move from the gas phase to interstitial sites in zir-
conium and the hydrides. Oxidation of zirconium hydrides is strongly exothermic.
Most alloying elements are more stable in metallic Zr than in hydrides, i.e.,
they inhibit hydride formation. Figure 6 shows the insertion energy of Ni, Cr, Fe,
Nb, and Sn in a-Zr and hydrides. The insertion energy, and thereby the destabiliza-
tion of the compound by alloying elements, increases with hydrogen concentration
being the largest for e-ZrH2. Metal elements that are more stable in hydride than in
pure Zr are Ni in Zr4H, Fe in Zr4H and Zr2H, Cr in Zr4H, Zr2H and Zr4H3, and Nb
in Zr4H. The element for which it is most energetically unfavorable to be in
hydrides is Sn. This is related to the strong repulsion of H by Sn in hydrides. For
oxygen, insertion in octahedral sites becomes successively harder when the hydro-
gen content increases. Insertion of oxygen in tetrahedral sites is less dependent on
the hydrogen content.
In the study of the effect of the alloying elements in a-Zr, the elements are dis-
solved in the Zr matrix. It should be kept in mind that the elements are likely to
have precipitated into intermetallic phases. However, it is known that Fe is released
very early from the secondary phase particles (SPPs) by irradiation. Ni and Cr are
FIG. 6 Insertion energies of metals into Zr hydrides from their respective bulk phases
versus hydride stoichiometry.
not strongly released until high fluence is reached. At high fluence, Ni is

dispersed in the matrix, whereas Cr remains in clusters [47]. The volume change
caused by alloying elements in the preferred a-Zr phase is approximately linear
with alloying element concentration. Analogous to the expansion for hydrogen
given by Eq 5, the volume change (%) of a-Zr caused by element Me, DVMe,a-Zr, is
given by
(10) DVMe;aZr =V0 100 ¼ KMe;aZr CMe;aZr
where
CMe,a-Zr ¼ the concentration of element Me in a-Zr (wt.-ppm).
Values of KMe,a-Zr are given in Table 2. Nickel, chromium, and iron cause a con-
traction whereas oxygen causes an expansion of the lattice. Nb and Sn have a less
pronounced effect and these elements are not expected to increase the hydrogen
solubility because of local stress. The expansion of the lattice caused by oxygen
could account for the fact that McMinn et al. found an increase in hydrogen solu-
bility in Zircaloy-4 with increased oxygen [11].
Adding alloying elements to Zr could also lead to a different effect by hydrogen
on the Zr lattice because of chemical interactions with H atoms. Therefore, the
interaction between alloying elements and hydrogen is assessed by computing the
interaction energy. The interaction energy is the change in energy when atoms of
an alloying element and hydrogen are brought into nearest neighbor contact in the
alloy, probing octahedral and tetrahedral sites. A negative value indicates that the
alloying element attracts hydrogen, which probably would lead to increased hydro-
gen solubility. The result is given in Table 2. It is found that Fe, Cr, and Ni increase
TABLE 2 Effect of alloying elements Me.
Cr Fe Ni Nb Sn O
Volume expansion 9.23 9.73 8.64 1.66 1.22 þ9.03

KMe,a-Zr (105)
Surface segregationa 41 32 64 5 82 36
(kJ/mol)
Attracts Hb (kJ/mol) 26 56 18 þ2 þ3 þ1
Attracts vacancyb 12 to þ42 74 to þ11 18 5 6 þ1
(kJ/mol)
Attracts interstitial Exch. Exch. Exch. 61 þ11 5
Zrb (kJ/mol)
a
A positive sign of the segregation energy indicates preference for the surface.
b
A positive sign means repulsion.
the binding of H, whereas Sn and Nb reduce it. There is no difference in binding

energy with oxygen.
For the elements that attract hydrogen, i.e., Fe, Cr, and Ni, the hydrogen atoms
closest to the alloying elements will not give the same contribution to the volume
expansion of the Zr lattice as the other hydrogen atoms in the lattice. In general,
the lattice around the alloying elements will be in a stretched state in which hydro-
gen atoms cannot exert as much pressure. Hence, hydrogen atoms closest to the
alloying atoms will give a smaller contribution to the volume expansion. The excep-
tion is hydrogen close to Nb, which gives a somewhat larger volume expansion (in
octahedral sites). However, this effect scales linearly with the concentration of the
alloying elements and is small for small concentrations.
The element with the largest interaction energies with hydrogen are Fe and Cr.
The electronic binding energy of hydrogen close to Fe is about 56 kJ/mol lower
(more stable) than in pure Zr. The binding of hydrogen close to Cr is about 26 kJ/
mol lower than in pure Zr. A second hydrogen atom added close to Cr also binds
stronger than in pure Zr, again by 25 kJ/mol. However, a third hydrogen atom will
bind weaker than in pure Zr by about 5 kJ/mol. Hence, a Cr atom can trap two
hydrogen atoms.
In hydrides, one element has a very pronounced behavior, namely Sn. Sn is an
element that strongly repels hydrogen in the hydrides. Hydrogen atoms close to an
Sn atom are displaced into neighboring sites. Large movements of hydrogen can
occur in all hydrides; however, the energetic cost becomes higher with increasing
atomic density. Sn is the element with the highest energetic cost for insertion into
ZrH2. On the other hand, Sn-containing phases have been observed to precipitate
in Zr alloys. In Zr-Nb-Sn-Fe, the precipitating phases are Zr4Sn and Zr5Sn3 (cited
in Ref 48). Pearson’s structural database [49] report the existence of Zr-Sn-Ni, Zr-
Sn-Fe, and Zr-Sn-Co hydrides.
The volume change of hydrides induced by the studied alloying elements in the
concentrations given above is shown in Fig. 7. Each bar corresponds to a specific H/
Zr ratio of the hydride from 0.25 to the left to 2.0 to the right for each alloying ele-
ment. The alloying elements Ni, Fe, Cr, and Nb always decrease the volume, and O
always increases volume. Increasing H content tends to increase the positive volume
effect. Lattice contraction is large for low H content and expansion is large for high
H content. Typically, the volume change is less pronounced in the hydrides com-
pared to pure Zr.
ZR-H WITH DEFECTS

Point Defects
In reality, a-Zr and Zr hydrides contain a certain level of defects, the concentration
of which depends on the history of the material, temperature, and level of irradia-
tion. The creation of Frenkel pairs during irradiation and their dynamics, formation
of loops, and interaction with hydrogen and other elements are of particular interest
FIG. 7 Effect of alloying elements Me on volume of hydrides. Each bar corresponds to a

specific H/Zr ratio of the hydride from 0.25 to the left to 2.0 to the right for each
alloying element.
for the induced growth behavior. Simulations addressing these questions have
therefore been performed and will be discussed next.
Calculations of formation energies of point defects show that it becomes
increasingly harder to form Zr interstitials when the hydrogen content increases.
Interstitials in hydrides typically prefer tetrahedral sites rather than octahedral sites.
The interstitial formation energy in e-hydride is very high (621 kJ/mol) where only
octahedral sites are available. That is a factor of 2 higher than in defect-free a-Zr
(315 kJ/mol). For c-hydride, the formation energy is between 280 kJ/mol and
470 kJ/mol depending on the hydrogen content. There is less variation in the for-
mation energies of Zr vacancies with hydrogen content. The formation energy is
193 kJ/mol in a-Zr and somewhat higher in most hydrides (248 kJ/mol to 258 kJ/
mol for c-hydride and 248 kJ/mol for e-hydride).
The induced volume change of zirconium hydrides for a particular defect
(vacancies or interstitials) is nearly independent of the hydrogen concentration in
the hydride. A Zr vacancy concentration of 1 at. % causes a lattice contraction with
a volume change of between 0.1 and 0.5 %, depending on the hydride. For pure
a-Zr, the volume change is 0.44 %. 1 at. % Zr interstitials in octahedral sites give a
lattice expansion with a volume change of between þ1.0 % and þ1.1 %. For tetra-
hedral sites, the corresponding volume expansion is between þ0.6 % and þ1.0 %.
The volume change in pure a-Zr is þ1.2 %.
Interaction between Point Defects and Alloying Elements

Computed interaction energies between point defects and alloying elements (see Ta-
ble 2) show that all elements except O attract defects. In the case of Fe and Cr, the
calculations reveal a delicate balance between the attraction of vacancies and the
magnetic state of the alloying element. In some configurations, there appears to be

a strong attraction of vacancies, especially in the case of Fe. For Ni, Nb, Sn, and O,
the interaction energy is small. Sn is experimentally reported to trap vacancies, but
the effect has been attributed to clusters of Sn, and it has been judged unlikely that
the effect could be ascribed to interactions of vacancies with isolated Sn atoms [49].
Hence, the present computed results are consistent with experiment.
Sn slightly repels Zr self-interstitials, whereas oxygen slightly attracts this
defect. However, the energies are small, so oxygen is not expected to trap self-
interstitials in zirconium. This is in agreement with experiments [50]. The interac-
tion of Ni, Cr, and Fe with Zr interstitials is particularly interesting. An analysis of
the atomic movements reveals that Zr does not remain in its interstitial site close to
these alloying elements. Instead, the interstitial Zr atom pushes away the alloying
element and takes its place in the Zr lattice. As a result of this swap, the alloying ele-
ment ends up as an interstitial. This is indicated with “Exch.” (exchange) in Table 2.
The Zr interstitial and the alloying element do not need to be nearest neighbors for
such replacements to occur. The interstitial can push a chain of intermediate Zr
atoms in a concerted motion up to the nearest neighboring Zr, which takes the
place of the alloying element. Hence, Fe, Cr, and Ni have an ability to heal Zr inter-
stitials in the lattice and thereby contribute to the stability of the Zr lattice.
The present results show that attraction between solutes and point defects can-
not be explained as a pure size effect of the solute, but also electronic interactions
play a role. This is the case not only for the attraction of point defects by solutes,
but also how vacancy formation close to the elements affects the Zr lattice. For
example, vacancy formation close to Sn leads to a lattice relaxation around the
defect that is different for Nb, although both elements are comparable in size to Zr.
In this respect, Sn behaves as an oversized atom, which is consistent with experi-
mental data for Sn in solid solution in a-Zr [51].
Trapping of Hydrogen in Point Defects

Zr vacancies may act as sinks for hydrogen atoms. Potential consequences for
hydrogen trapping include increased solubility and reduced dimensional changes
from hydrogen loading. It is known that there is increased apparent hydrogen solu-
bility in irradiated alloys [11]. In fact, the simulations reveal that a single vacancy
can accommodate up to nine hydrogen atoms. The binding energy in the vacancy is
lower (more stable) than in defect-free Zr for up to six hydrogen atoms. Thus, there
is a driving force for hydrogen to migrate to a vacancy with less than six hydrogen
atoms. In total, there are nine sites for hydrogen in the vacancy. When all nine sites
are occupied, hydrogen is repelled from the vacancy. The first-principles calcula-
tions accurately describe the binding energy of hydrogen in a Zr vacancy. The ex-
perimental value for the binding energy in a Zr vacancy is reported to be about
30 kJ/mol by [23], which can be compared to the computed value of 27 kJ/mol
(22 kJ/mol) using H in octahedral (tetrahedral) sites as reference. At elevated tem-
peratures, hydrogen atoms will tend to be increasingly distributed on interstitial
sites rather than trapped in local defects because of the configurational entropy.
There are many more interstitial sites for hydrogen than sites close to a defect.
However, the binding energy of 30 kJ/mol is quite large, so trapping is likely to
occur at elevated temperatures. The binding energy of H in a vacancy is not affected
to a large extent by the presence of solute elements close to the vacancy. The energy
difference with and without alloying elements is only a few kJ/mol.
The effect is then a smaller volume expansion than predicted for a given hydro-
gen concentration, but the possibility for hydrogen to cluster at defects leads to an
increased apparent solubility. The possible trapping of multiple hydrogen in vacan-
cies has also been demonstrated in the case of Pd, which can trap up to six hydro-
gen atoms [52].
Also, self-interstitial atoms (SIAs) have the potential to trap hydrogen, even
though hydrogen is less strongly bonded than in a vacancy. SIAs in tetrahedral sites
bind hydrogen by an energy between 13 and 18 kJ/mol. An SIA in an octahedral
site repel small amounts of hydrogen. A hydrogen atom near an octahedral Zr in-
terstitial is repelled by 14 kJ/mol. A second hydrogen atom added on the opposite
side of the interstitial is repelled by 9 kJ/mol. A third hydrogen atom added close to
the interstitial is attracted by 12 kJ/mol. Whereas one and two added hydrogen
atoms do not affect the lattice, adding the third hydrogen atom causes large atomic
rearrangements at the interstitial site where the interstitial is repelled by the hydro-
gen. The resulting atomic structure resembles the structure of ZrH2 (body-centered
tetragonal Zr lattice with tetrahedral sites filled with hydrogen). The large atomic
movements combined with the favorable binding energy of hydrogen when enough
hydrogen has accumulated at the interstitial could indicate the beginning of a phase
transformation where a hydride is formed. It is tempting to interpret the repulsion
of the first two hydrogen atoms as a barrier to the formation of a nucleation site for
zirconium hydride.
Dislocation Loops
The simulations based on the EAM potentials characterized previously show that it
is energetically favorable for isolated vacancies to coalesce and form dislocation
loops. The gain in energy per vacancy can reach 80 kJ/mol. It is favorable to form
both <a>- and <c>-type dislocation loops. This is the case also for interstitial
loops. The driving force to form <c> interstitial loops is about the same as of <a>
loops. The gain in energy per interstitial can reach 150 kJ/mol. However, forming
[11–20] <a> loops is only energetically favorable for interstitials, not for vacancies.
In experiments, preferences for specific orientations of the loops have been
observed. Adamson gives an account of loops in Zircaloy [53]. In Zircaloy, both va-
cancy and interstitial loops exist, but more than half have vacancy character. <a>
loops form early in the irradiation and the size of the loops increases with irradia-
tion temperature. The loops become unstable (start to disappear) at about 673 K
(400 C). The <c> loop is strictly a vacancy-type loop. It is relatively large
(>100 nm) and does not form until considerable irradiation effects have
accumulated. They are thermally stable to high temperature >560 C (>833 K). Dif-
ferences in preferred orientation between vacancy loops and interstitial loops could
be because of diffusional anisotropy. These experimental results are consistent with
the present simulations. Large dislocation loops are energetically more stable than
small loops. At elevated temperatures, the configurational entropy tends to favor
randomly distributed vacancies or interstitials over dislocation loops. Hence, as a
function of temperature there will be a certain fraction of vacancies or interstitials
in equilibrium with the dislocation loops. The entropic contribution to the Gibbs
free energy, which governs the equilibrium between condensed and dispersed
vacancies, diminishes at lower temperatures. In other words, dislocation loops are
thermodynamically favored at lower temperatures and can dissolve at elevated tem-
peratures; small loops disappear first whereas the larger and more stable loops per-
sist up to higher temperatures.
Vacancies accumulating in a plane typically only persist in an unrelaxed state
in small clusters. At a critical radius, the growing cluster can lower its energy by
rearranging the atoms and thereby reducing the volume, i.e., collapsing. The simu-
lations show that vacancy dislocation loops should collapse when they have grown
to a critical radius of 10 to 15 Å (in the absence of hydrogen). The collapse releases
energy by the removal of the internal surfaces in the void platelet by about 50 kJ/
mol per vacancy. Figure 8 shows the atomic structure of a vacancy cluster platelet
on the basal plane. The radius of the platelet is 15 Å, which corresponds to 85
vacancies. The initial structure is an unrelaxed structure in which the loop has not
collapsed and there is no strain in the system. After relaxation (molecular dynamics
at 600 K), the loop collapses into a fault loop structure (right panel in Fig. 8) with
significant strain. The collapsed loop is not symmetric, but has a directional “cone”
shape pointing in the [0001] direction. The collapse of the loop leads to a change of
the stacking sequence inside the loop structure from the hexagonal ABABA

FIG. 8 Atomic structure of (0001) vacancy dislocation loop with radius 15 A before and
after collapse.
stacking to AABAB stacking. The cone structure is limited by (20,21) atomic planes.
The elliptically shaped dislocation loops as observed in transmission electron mi-
croscopy (TEM) images may originate from planar cuts through the three-
dimensional cone-like structures.
The collapse is associated with a large volume decrease. Isolated vacancies give
volume shrinkage of 9 Å3 per vacancy. At loop sizes where the loops collapse, there
is an additional shrinkage of 14–16 Å3 per vacancy. This gives a total shrinkage of
23–25 Å3, which is very close to the volume of a Zr atom in bulk Zr. In other words,
after the collapse, the density of the system is very close to that of defect-free bulk
Zr. Also, interstitial loops affect the volume and there is a large driving force for
interstitials to form loops. Interstitial dislocation loops do not collapse at a critical
size in contrast to the vacancy dislocation loops. Rather, the growth of the disloca-
tion loop causes a monotonous increase of the volume with a volume change
roughly independent of loop size, about 22 Å3/interstitial. Hence, the volume
expansion caused by interstitial loops is of comparable size to the volume contrac-
tion caused by collapsed vacancy loops. Thus, if pairs of vacancies and interstitials
are created and both types of defects form loops, the respective volume changes
tend to be much less pronounced as compared to those of the individual defects.
One difference between the two types of loops is, however, that hydrogen may accu-
mulate inside vacancy loops. This can shift the balance and cause lattice distortion
and result in H-enhanced irradiation growth.
It is found that the presence of hydrogen in the loops can diminish the volume
decrease caused by the formation of loops, or even prevent the loops from collaps-
ing. The simulations show that for loops with the critical radius, less than 1 at. % of
hydrogen is needed to prevent the collapse. Hence, if hydrogen is present in the
loops or if hydrogen is elsewhere in the system, it could shift the volume balance
and be a mechanism for distortion and growth.
It is therefore of value to determine where the hydrogen is actually located in
the system. This question is first analyzed in terms of the energetics. There may be
many different types of defects in the material and hydrogen will have a different
propensity to be in each of these. For extended defects like dislocation loops, hydro-
gen may also have different stabilities in different regions of the defect. For the va-
cancy platelets, hydrogen prefers to decorate the internal surfaces. It is found that
hydrogen is more stable inside the (0001) vacancy loop than in random interstitial
sites by approximately 40 kJ/mol per hydrogen atom (for 10 hydrogen atoms inside
the loop). For the interstitial dislocation loops, hydrogen prefers to be at the rim of
the loop. Hydrogen is more stable at the rim of an interstitial dislocation loop than
in random interstitial sites by approximately 30 kJ/mol.
Thus, the simulations indicate that the most effective traps of hydrogen are va-
cancy loops rather than isolated vacancies. The vacancy loops can accommodate a
large amount of hydrogen. For the (0001) loop with radius 15 Å, the interior region
of the loop is energetically more favorable than outside the loop also for 25 hydro-
gen atoms. For the [1010] loop with radius 10 Å, the interior of the loop is
preferred for all studied hydrogen concentrations (up to 75 hydrogen atoms). How-
ever, isolated vacancies and interstitial loops could also trap hydrogen. If the mate-
rial contains alloying elements, they could also trap hydrogen. In particular, Cr for
which the trapping energy is similar to that of isolated vacancies, 25 kJ/mol is
computed by first-principles calculations. Again, one needs to keep in mind that at
elevated temperatures, entropy will tend to disperse trapped H atoms throughout
the lattice. At around 500 K (227 C), the configurational entropy of H dissolved in
the zirconium matrix can reach the same magnitude as the enthalpic preference for
H to be trapped in vacancies and vacancy dislocations. However, with increasing
temperature also, the entropy of H in vacancy loops is enhanced because of the
available space. Hence, it is plausible to assume that H atoms remain in vacancy
loops at temperatures above 500 K thus hindering the collapse of vacancy loops.
DIFFUSION
The results are concluded with a discussion of diffusion in the Zr-H system. The
ability of vacancy and interstitial point defects to aggregate into dislocation loops is
in part determined by their diffusivity in the zirconium lattice. The present simula-
tions have demonstrated that vacancy loops tend to collapse once a certain size is
reached and that this collapse can be prevented by hydrogen uptake in the loop.
Dislocation loop formation and collapse have rather drastic effects on the volumet-
ric properties of the zirconium metal, which may in addition be quite anisotropic.
Therefore, the relative diffusivities of vacancies, interstitials and hydrogen in the Zr
lattice are critical parameters, which in part determine whether hydrogen uptake of
vacancy loops is fast enough to prevent growing vacancy loops from collapsing. To
address this important question, ab initio and quasi-classical EAM simulations of
diffusion of vacancies, interstitials and hydrogen in Zr, and the influence by alloying
elements, have been performed.
Diffusion of Hydrogen
Dissolved hydrogen atoms in zirconium are located in interstitial sites. As already
mentioned in the section about hydrogen in solution above, the binding energy of
hydrogen to the zirconium lattice is nearly equal for octahedral and tetrahedral
sites. There are a variety of different diffusion paths available for H atoms to
migrate through the Zr lattice, which are summarized in Fig. 9. Basal plane diffusion
involves both tetrahedral and octahedral sites and has to overcome barriers associ-
ated with BTO transition states (TS) between these stable and metastable sites. Pris-
matic plane diffusion may involve octahedral sites only with basal octahedral
transition states, BO. An alternative path involves octahedral and tetrahedral sites
and energy barriers associated with BTO and basal tetrahedral BT transition states
(TS), which need to be overcome. The basal tetrahedral and basal octahedral transi-
tion states BT and BO were directly obtained by geometry optimization, for identifi-
cation of the transition state between tetrahedral and octahedral sites, BTO, a
nudged elastic band simulation was required. Diffusion along the c axis will be
FIG. 9 Diffusion paths of interstitial hydrogen and oxygen in zirconium.
referred to as axial diffusion and basal plane diffusion (along the a axis) will be
referred to as basal diffusion.
The lowest energy barrier (12.4 kJ/mol) is found between the tetrahedral sites,
i.e., the basal tetrahedral site BT. The transition state for migration of H from the
tetrahedral to the octahedral site, BTO, is much higher in energy, i.e., 39.8 kJ/mol
above the energy of the tetrahedral site and 34.7 kJ/mol above the octahedral site.
The direct transfer of H atoms from one octahedral site to the adjacent one exhibits
a transition state only slightly higher in energy, 41.2 kJ/mol. The effect of possible lat-
tice strain on the diffusion barriers is weak. The diffusivity decreases somewhat
with tensile stress. A change from a 1 % compression to a 1 % expansion of the Zr
lattice reduces the diffusion coefficient by a factor between two and three.
The temperature-dependent diffusion coefficient for hydrogen diffusion is
obtained by kinetic Monte Carlo simulations. For each of the interstitial and transi-
tion states, the phonon spectrum was calculated providing the temperature-
dependent thermodynamic functions, i.e., the partition functions required in transi-
tion state theory. The temperature-dependent free energy barriers obtained from
the phonon calculations are used to derive the jump rates on a grid of temperatures
for the kinetic Monte Carlo simulations. A linear least-squares fit to the data allows
evaluating an effective diffusion barrier summarizing the net effect of the atomistic
elementary jump processes. The hydrogen diffusion in Zr is found to be isotropic
within the numerical precision and is thereby described by

(11) DH ¼ 1:13 107 e41:9=ðRTÞ m2 =s
with the barrier given in kJ/mol. In analyzing the relative diffusivities of hydrogen,
vacancy defects and interstitial defects, these net diffusion coefficients are used. The
computed temperature-dependent diffusion coefficient for H diffusion is compared
to experimental results in Fig. 10 [54–57]. It is noted that the effect of thermal
expansion is not included in the current simulations. Thermal lattice expansion can
lead to reduced diffusion barriers and thereby faster diffusion bringing computed
values closer to experiments. This effect has been demonstrated in the case of
hydrogen diffusion in Ni [58]. In Ni, the thermal expansion from T ¼ 0 K to room
temperature is 0.3 % (linear), which reduces the diffusion barriers by 10 kJ/mol
thereby making the diffusion almost three times faster.
The effect of alloying elements on hydrogen diffusion has also been assessed.
Elements like Fe are known to have a low solubility in Zr [59] and alloying elements
are typically embedded in precipitated phases in Zircaloy. However, one of the
effects of irradiation on Zr is to disperse impurities such as Fe [60]. The discussion
of the influence by alloying elements is restricted in this study to elements in solu-
tion. In other words, the effect by single atoms of the elements, rather than clusters
(precipitates) is investigated. This is done introducing an alloying atom in close
proximity to diffusion pathways (by substituting a Zr atom) and computing the
resulting diffusion barriers. The diffusion pathways are the same (except for Sn) as
in pure Zr, however with possible local displacements of the atomic positions. For
example, iron atoms are highly mobile and are pushed aside by the diffusing H. The
barriers for H diffusion are significantly decreased close to Fe. For Ni and Cr, the
FIG. 10 Experimental and computed H diffusion coefficient. The experimental results

are taken from (a) Ref 54, (b) Ref 55, (c) Ref 1, (d) Ref 56, and (e) Ref 57, and
are redrawn from Ref 1 as solid lines. The dashed line shows the computational
result without thermal expansion.
effects are similar to those of Fe, but smaller. Also, Nb lowers hydrogen diffusion
barriers, but the Nb atoms themselves are not moved by nearby diffusing H atoms.
As noted previously, H is strongly repelled by Sn and diffusing hydrogen atoms are
pushed into octahedral sites close to Sn. The jump rate of H atoms from octahedral
to adjacent octahedral sites close to Sn is dramatically decreased. For other diffusion
pathways, e.g., in the basal plane, the proximity of Sn reduces the diffusion barriers.
For oxygen, the diffusion of H atoms is only slightly influenced by the proximity of
O atoms.
Hence, most diffusion barriers are much lower close to the alloying elements
than in pure Zr. Exceptions are diffusion barriers close to Sn and O, which are
larger than in pure Zr. From the temperature-dependent thermodynamic functions
obtained by computing phonon spectra, it is found that the temperature depend-
ence of the barriers is weak and can almost be neglected in the relevant temperature
interval. The temperature-dependent barriers were consequently used as input for
kinetic Monte Carlo simulations. It is found that hydrogen is faster in the vicinity
of Fe, Cr, Ni, Nb, and O. The effect is largest for the metals and not so pronounced
for oxygen. In contrast, hydrogen diffusion is slowed down close to Sn. Anisotropy
is introduced in that axial hydrogen diffusion is slowed down considerably more
than basal diffusion. The anisotropy increases with increasing temperature. It is im-
portant to note that the simulations only describe the diffusion of hydrogen in close
proximity to the alloying elements. Far from the element, the diffusion will be the
same as in pure Zr. The effect is thus depending on the concentration of alloying
elements. Given that in Zircaloy the concentration of alloying elements is very low,
one can expect that the direct effect of alloying elements on H diffusion is relatively
small.
Diffusion of Oxygen
Oxygen, like hydrogen, is an interstitial element in a-Zr and is the most important
impurity interstitially dissolved in zirconium [61]. Oxygen diffusion in pure a-Zr
has been studied using the same methodology as for hydrogen diffusion. The diffu-
sion pathways in Fig. 9 apply also for oxygen diffusion. Oxygen highly favors octa-
hedral interstitial sites to tetrahedral sites. The octahedral site is more stable than
the tetrahedral site by 85 kJ/mol. It is also evident from experiments of the
zirconium-oxygen binary solution that octahedral interstitial sites in hexagonal lat-
tice of zirconium are highly favored to receive oxygen atoms [61].
The diffusion coefficients for basal and axial oxygen diffusion are obtained as
(12a) DO;basal ¼ 6:13 105 e175:7=ðRTÞ m2 =s
(12b) DO;axial ¼ 4:64 105 e173:7=ðRTÞ m2 =s
The result agrees well with experimental measurements. Flinn et al. [62] used Auger
electron spectroscopy to measure oxygen diffusion away from the surface of single-
crystal zirconium. The resulting Arrhenius expression for diffusion has
D0 ¼ (4.14 6 1.92) 104 m2/s and Ea ¼ 199.1 6 2.6 kJ/mol and is a direct measure
of diffusion along the c axis. Oxygen diffusion in Zr single crystals has also been
measured by Hood et al. [63]. The measurements were done in the temperature inter-
val 610–870 K in directions both parallel and perpendicular to the c axis. It was found
that diffusion anisotropy is weak and that D is little affected by specimen impurity
content. The Arrhenius expression is D ¼ 4.92 105 exp(2.12 6 0.05 eV/kT) m2/s.
Figure 11 shows experimental data for oxygen diffusion together with the com-
puted results from this study [64–68]. The red and blue curves are the computed
results for axial and basal diffusion, respectively. The type of measurements are: (a)
internal friction and strain aging measurements, (b) AES, (c) XPS, (d) XPS, (e)
AES, and (f) AES.
Diffusion of Vacancies
The temperature-dependent diffusion coefficient for vacancy diffusion in pure a-Zr
was derived in a manner similar to that discussed above for hydrogen. For basal
and axial diffusion, the diffusion coefficient is:
(13a) Dvac;basal ¼ 8:62 106 e69=ðRTÞ m2 =s
(13b) Dvac;basal ¼ 9:87 106 e73=ðRTÞ m2 =s
FIG. 11 Experimental results (solid lines) for oxygen diffusion in Zr together with the
computed results (dashed lines) in basal (online: blue) and axial (online: red)
directions as labeled. The experimental results are taken from (a) Ref 64, (b) Ref
62, (c) Ref 65, (d) Ref 66, (e) Ref 67, and (f) Ref 68 and are redrawn from Ref
62. (c) to (f) and one plot (a) with smallest slope are grain-boundary diffusion
measurements; the lower envelope (a) and (b) plots are bulk diffusion.
The diffusivity of vacancies is almost isotropic and considerably slower than for
hydrogen. The vacancy migration barriers are about 50 % higher than for hydrogen
diffusion. The isotropic diffusion is in agreement with other computer simulation
studies [69]. Low temperature irradiation-damage annealing studies seem to indi-
cate diffusion barriers high enough for vacancies in Zircaloy to be immobile at tem-
peratures below 127 C (400 K) [70,71]. For a temperature of 1367 K, a simulation
was performed for vacancy diffusion in a-Zr with an H-concentration of 1.2 at. %
to check for a possible influence by hydrogen on the vacancy diffusion. H atoms
have a tendency to hinder the diffusion of vacancies, but the effect is rather small
and close to the statistical noise of the present calculations.
Next, the effect of alloying elements on the vacancy mobility is investigated. In
the simulations of vacancy diffusion in the presence of the alloying elements Fe, Cr,
Ni, Nb, and Sn, one Zr atom is substituted by the alloying element and the vacancy
is in a neighboring lattice site. It is found that the alloying elements reduce the dif-
fusion barriers and lead to faster vacancy diffusion. The effect is largest for Cr, fol-
lowed by Ni and Nb. Sn introduces anisotropy in the vacancy diffusion. The
diffusion is substantially increased within the basal plane close to Sn, but not axial
diffusion. Oxygen slows down vacancy diffusion. Once again, it is important to
remember that the discussed effect is local and only affects the close proximity to
the alloying elements.
Diffusion of Interstitials
The diffusion of Zr self-interstitials is computed directly from molecular dynamics
simulations using a quasi-classical embedded-atom method (EAM) rather than
using the ab initio approach. This choice is dictated by the fact that interstitial dif-
fusion of Zr involves the displacement of a cluster of Zr atoms (crowdions) in a
rather complex migration pattern, which are not suited for treatment by transition
state theory.
From molecular dynamics simulations, an average diffusion coefficient is
obtained with the values
(14) Dint ¼ 5:5 108 e9:4=ðRTÞ m2 =s
The diffusivity in the basal plane is approximately twice as high as in a direction

parallel to the c axis. At typical reactor temperatures, H diffuses about twice as fast
as vacancies and interstitials move 5–10 times faster than vacancies.
For two elements, namely Nb and Sn, the diffusion mechanism is similar to
that in pure Zr. The diffusion occurs by a collective motion of neighboring atoms
(crowdion). The present results indicate that the presence of substitution Nb
attracts and pins Zr interstitials and thereby slows down interstitial diffusion in Zr,
both basal and axial diffusion. Substitutional Sn seems to have a very small effect
on the diffusion of interstitial Zr at high temperatures, but a retarding effect at low
temperatures by repelling Zr interstitials. Interstitial Nb and Sn have a similar diffu-

sion behavior to interstitial Zr.
The 3 d transition metals Cr, Fe, and Ni behave very differently from Nb and Sn.
Cr, Fe, and Ni are interstitial diffusers that do not affect the motion of surrounding
Zr atoms much compared to Nb and Sn. These elements are smaller than Zr and
have a larger propensity to be in the interstitial sites than Zr. As mentioned before, a
Zr interstitial coming close to Cr, Fe, or Ni on substitutional sites will lead to an
exchange of positions where the Zr atom takes the position in the lattice and the
alloying element becomes an interstitial. The alloying element then becomes a fast in-
terstitial diffuser. Interstitial Cr, Fe, and Ni diffuse anisotropically with fast motion
along the c axis. The diffusion occurs by jumps between neighboring octahedral sites.
For Ni, the anisotropy is less pronounced. The fast axial diffusion is confirmed by
DFT computations of the diffusion barriers. The octahedral–octahedral barrier is
only 4 kJ/mol for Fe and Cr. The barrier is somewhat larger for Ni, 25 kJ/mol. The
alloying elements tend to cluster, which stops the fast diffusion. As clustering requires
diffusion both in the c direction and perpendicular to the c axis, Ni will cluster faster
than Cr and Fe. A more detailed analysis of the effect of alloying elements on self-
interstitial diffusion in Zr and the applied EAM potentials are provided by Ref 40.
Summary and Conclusions

In summary, the present atomistic calculations provide valuable insight into irradi-
ation growth of Zircaloy. Starting from the idealized case of pure a-Zr and hydrides,
the hydrogen-induced dimensional changes DVH,sol/V0 100 and DVH,hyd/
V0 100 (in %) from H in solution and in hydrides, respectively, are found to be
(15) DVH;sol =V0 100 ¼ 1:7 103 CH;sol

2095:2 ð1 e0:7452X ÞCH;hyd
(16) DVH;hyd =V0 100 ¼
X 106 CH;hyd
The term X is the atom ratio between H and Zr in the hydrides, i.e., X ¼ 1 for ZrH
and X ¼ 2 for ZrH2. CH,sol and CH,hyd are the concentrations of H in wt.-ppm in the
form of dissolved hydrogen and as hydride. The relative growth per H atom is
more pronounced at small concentrations and it tapers off toward the dihydrides.
The two contributions to the dimensional change are comparable in size, but are
somewhat larger than given by experimental data in Zircaloy. This is attributed to
trapping of hydrogen in defects such as vacancies, dislocations, and grain bounda-
ries in actual samples. For similar reasons, the solubility in a perfect a-Zr crystal is
smaller than observed values.
It is reasonable to assume that irradiation damage of Zircaloy leads to the gen-
eration of vacancies and interstitials. The calculations show that metallic zirconium
with Frenkel pairs has a larger volume than a perfect crystal. Hence, the initial effect of
irradiation is an expansion. The present simulations show that interstitial Zr atoms are
diffusing faster than interstitial H atoms whereas vacancies diffuse slower. At a temper-
ature of 300 C, vacancies diffuse about 0.1 mm per hour, hydrogen atoms twice as far,
and Zr interstitials about 5–10 times. There is evidence from the simulations that inter-
stitial Zr atoms diffuse faster in the a direction than in the c direction of the crystalline
grain. Furthermore, a diffusing particle has a probability of 6 out of 8 to hit a prismatic
plane at a given distance from a starting point, but only 2 of 8 to encounter a basal
plane. This implies that more interstitial a loops would form than c loops. As a conse-
quence, each crystallite would expand in the a direction and would shrink along the c
axis. The overall volume change is small. For textured materials with a preferred orien-
tation of the c axis of the crystallites, one could expect some macroscopic deformations
of the material. This would be a scenario without hydrogen, considering only the
migration and loop formation of vacancies and self-interstitials.
If we now assume that hydrogen is present in the Zr samples prior to irradia-
tion, then the scenario would be somewhat different. Provided a neutron flux pro-
duces a certain concentration of vacancies and self-interstitials, then we know from
the present simulations that self-interstitials diffuse faster than interstitial hydrogen
atoms, leading to the formation of interstitial a loops. On the other hand, if <c>-
loops are formed by the slowly and isotropically diffusing vacancies, then hydrogen
can reach these vacancy loops faster. The calculations reveal that such vacancy
loops can trap a fairly large number of H atoms. It is thus likely that vacancies dif-
fuse with trapped H atoms. By the time such a vacancy reaches a growing vacancy
dislocation loop, the H atoms are infused into the vacancy loop. The present simu-
lations show that in some cases hydrogen atoms prevent or retard the collapse of
vacancy loops. If these vacancy loops do not collapse because of accumulated
hydrogen, then the net result would be enhanced growth of the material. In other
words, samples pre-loaded with H or H formed by corrosion during irradiation
would show a larger irradiation growth than H-free samples. The collapse of H-
decorated vacancy loops would lead to regions that are supersaturated with hydro-
gen. This could nucleate the formation of hydrides, which would align parallel to
the vacancy <c>-loops. In fact, this is consistent with TEM images showing
hydrides in conjunction with <c>-loops. If the H-concentration in pre-loaded Zr
samples exceeds the solubility limit, then the samples prior to irradiation would al-
ready contain hydride precipitates. Hydrogen atoms bound in hydrides are rela-
tively stable and would not be directly available to contribute to the above
mechanism involving vacancy loops. However, binding energies of hydrogen atoms
in vacancies and vacancy loops are of the same magnitude than binding energies in
the hydrides. Consequently, once the zirconium matrix becomes hydrogen depleted
because of the above mechanism the hydrides may decompose, dissolving hydrogen
into the zirconium matrix to satisfy equilibrium conditions. In this sense, hydrides
may provide a reservoir of hydrogen atoms being pumped into vacancy <c>-loops,
and upon retardation of their collapse may contribute to irradiation growth.
In a zirconium alloy being irradiated in a light-water reactor, the effect of H
would obviously be more complicated because the concentration of hydrogen
will be a function of the corrosion rate, hydrogen pickup fraction (both of which
are dependent on the material and operating conditions), and the thickness of
the component. That is, the hydrogen concentration will be a function of operat-
ing time. Normally, in a Zircaloy-2 channel, the hydrogen concentration is
between 10–15 ppm after operating 2 years, 50–100 ppm after 4 years, and
>100 ppm after 6 years [19]. Thus, relative to the above scenario, an important
factor to consider is the relationship between the rate of H pickup and the rate
H is trapped by in vacancies or vacancy loops. In addition, it is possible there
could be a concentration threshold below which the trapped H may result in a
region of supersaturated H around a vacancy loop but not retard its collapse—
and thus not enhance growth.
Another important part of the present work is devoted to the role of the alloy-
ing/impurity elements Sn, Fe, Cr, Ni, Nb, and O. The simulations reveal that each
of these alloying elements and impurities has unique properties as can been seen
from the compilation in Table 2. With the exception of oxygen, all of the five alloy-
ing elements considered here are thermodynamically more stable on substitutional
sites of a Zr lattice than as interstitial atoms. Substituting a Zr atom by any of the
five alloying elements leads to a local contraction of the lattice. For the relatively
small Cr, Fe, and Ni atoms, this effect is most pronounced. Interestingly, also Nb
and Sn have a small tendency to contract the lattice. In contrast, interstitial oxygen
exerts a positive stress, i.e., it tends to expand the lattice. Except Nb, all elements
shown in Table 2 tend to segregate from the bulk to surfaces. Most notably this is
the case for Sn and Ni. This thermodynamic driving force also applies to “internal”
surfaces such as grain boundaries.
Cr, Fe, and Ni attract H atoms, whereas the other elements are indifferent. The
interaction of vacancies with substitutional Fe and Cr atoms is found to be closely
related and sensitive to the magnetic state of the transition metal atoms. Ni, Sn, Nb,
and O atoms do not show pronounced interactions with vacancies. If an interstitial
Zr atom gets close to a substitutional Fe, Cr, or Ni atom, then the alloying elements
swap their positions with the Zr self-interstitial. The Zr fills the lattice site and Fe,
Cr, and Ni become interstitial atoms. In contrast, Nb binds Zr self-interstitials quite
strongly, whereas Sn somewhat repels Zr self-interstitials. The alloying elements
locally change the mobility of diffusing vacancies and self-interstitials. Most nota-
bly, Nb retards the diffusion of Zr self-interstitials.
From these computed results, we can derive the following impact on loop for-
mation and are thus able to draw conclusions concerning irradiation growth. Fe,
Cr, and Ni eliminate rapidly diffusing Zr self-interstitial by swapping positions.
Once Fe, Cr, and Ni are interstitial atoms, they diffuse very rapidly in the c direc-
tion until they cluster, fill vacancies, and probably also accumulate in vacancy
loops. The solubility of Fe, Ni, and Cr in Zr is low, and these elements tend to seg-
regate out of bulk regions or form intermetallics. As a consequence, the build-up
of dislocation loops and associated dimensional changes of the material are
reduced by the presence of these alloying elements.
Niobium captures diffusing Zr-self interstitials, thus preventing them from

their rapid diffusion to form <a>-loops. Once a Zr-self-interstitial is held back by
an Nb atom, the probability for a slower diffusing vacancy to reach and neutralize
such a Zr self-interstitial is enhanced. In other words, Nb increases the recombina-
tion rate of Frenkel pairs decreasing the vacancy concentration, which is the source
for the vacancy <c>-loops that cause breakaway growth in Zircaloy. This is poten-
tially an explanation of the resistance to breakaway growth in various Zr-Nb alloys.
The interaction of Sn with Zr defects and diffusing H atoms is smaller, although
there is a tendency to slow down axial hydrogen diffusion close to Sn atoms.
Whereas at high temperatures there is little effect of Sn on diffusion of Zr self-
interstitials, at low temperatures a similar retarding effect on self-diffusion is seen
than for Nb, which is however a result of Sn atoms repelling Zr interstitials. The
action of Sn seems to be different. Possibly, the very strong tendency to segregate
from the bulk to surfaces or surface-like regions provides a clue. It is thermody-
namically favorable for Sn to segregate and possibly form clusters on surfaces, grain
boundaries, dislocations, or within vacancy loops.
The tendency of O atoms to segregate to surfaces is similar to that of Fe, Cr,
and Ni. Given the high mobility of interstitial Fe, Cr, and Ni atoms together with
the fact that O also diffuses via an interstitial pathway, one could speculate that
these features would lead to the formation of oxide layers similar to those found in
stainless steel, possibly at the boundary between metallic zirconium and the outer
corrosion film. In fact, this might also be the case for Sn, which furthermore could
influence the electrical conductivity of the oxide film and thus have electrochemical
consequences on the corrosion process. This, however, are rather speculative con-
clusions, which will require deeper analysis.
In summary, the present atomistic simulations provided a wealth of data ranging
from new materials property data such as H-induced volumetric changes, the chemi-
cal potential of H atoms trapped in vacancies, elastic coefficients of a range of zirco-
nium hydrides, diffusion coefficients of vacancies, and interstitial H and O. The
simulations provide new insight into the formation of dislocation loops and the
impact of alloying elements on their growth. Together, this body of systematic mate-
rials property data and atomic-level understanding contributes to a better and clearer
picture of the mechanisms involved in fuel channel distortion. As in any computer
simulation, the predictions are limited by the inherent simplification of the models as
well as by the limitations of the computational methods. Given the difficulties in col-
lecting and comparing existing field data and the substantial effort in time and
resources for carrying out systematic experimental programs on irradiation growth,
the present atomistic simulation work demonstrates that computational technology
has matured to a point of making valuable contributions of engineering value. With
the relentless progress in compute power and sophistication of software, it is clear
that atomistic computations based on the solid foundations of quantum mechanics
and statistical mechanics will play an increasingly important role in the analysis of
materials behavior and the design of new materials.
ACKNOWLEDGMENTS
This work is supported by the EPRI BWR Channel Distortion Program chaired by
Michael Reitmeyer and funded by Exelon, Entergy, TVA, PP&L Susquehanna, South-
ern Nuclear, Detroit Edison, Energy Northwest, FENOC, Iberdrola, PSEG, and Xcel
Energy. The writers express their thanks to the participants of the EPRI Channel Dis-
tortion Science Subcommittee for many fruitful discussions and continued support of
the effort.
References
[1] Kammenzind, B. F., Franklin, D. G., Peters, H. R., and Duffin, W. J., “Hydrogen Pickup and
Redistribution in a-Annealed Zr-4,” Zirconium in the Nuclear Industry: Eleventh Interna-
tional Symposium, ASTM STP 1295, E. R. Bradley and G. P. Sabol, Eds., ASTM Interna-
tional, West Conshohocken, PA, 1996, pp. 338–370.
[2] Chung, H., Daum, R., Miller, J., and Billone, M., “Characteristics of Hydride Precipitation
and Reorientation in Spent-Fuel Cladding,” Zirconium in the Nuclear Industry: Thirteenth
International Symposium, ASTM STP 1423, G. D. Moan and P. Rudling, Eds., ASTM Inter-
national, West Conshohoken, PA, 2002, pp. 561–582.
[3] King, S. J., Kesterson, R. L., Yueh, K. H., Comstock, R. J., Herwig, W. M., and Ferguson, S.
D., “Impact of Hydrogen on Dimensional Stability of ZIRLO Fuel Assemblies,” Zirconium
in the Nuclear Industry: Thirteenth International Symposium, ASTM STP 1423, G. Moan
and P. Rudling, Eds., ASTM International, West Conshohocken, PA, 2001, pp. 471–489.
[4] Carpenter, G., “The Dilatational Misfit of Zirconium Hydrides Precipitated in Zirconium,”
J. Nucl. Mater., Vol. 48, 1973, pp. 264–266.
[5] Kesterson, R. L., King, S. J., and Comstock, R. J., “Impact of Hydrogen in Dimensional
Stability of Fuel Assemblies,” Proceedings: International Topical Meeting on Light Water
Reactor Fuel Performance, American Nuclear Society, Park City, UT, April 10–13, 2000.
[6] Seibold, A., Garzarolli, F., and Manzel, R., “Material Development for Siemens Fuel Ele-
ments,” Proceedings: International Topical Meeting on Light Water Reactor Fuel Per-
formance, American Nuclear Society, Park City, UT, April 10–13, 2000.
[7] Roy, C., “Nucleation and Growth of Hydrides in Zirconium Alloys,” Report No. AECL-
2297, Atomic Energy of Canada, Chalk River, Ontario, 1965.
[8] Rowe, R., “Strain Rate Effects in Hydrogen Gas Cracking (HGC) of Zircaloy-2,” Scripta
Mater., Vol. 40, 1998, pp. 249–256.
[9] Shaltiel, D., Jacob, I., and Davidov, D., “Hydrogen Absorption and Desorption Properties
of AB2 Laves-Phase Pseudobinary Compounds,” J. Less-Common Metals, Vol. 53, 1977,
pp. 117–131.
[10] Lewis, M., “Deuterium-Defect Trapping in Ion-Irradiated Zirconium,” J. Nucl. Mater., Vol.
125, 1984, pp. 152–159.
[11] McMinn, A., Darby, E. C., and Schofield, J. S., “The Terminal Solid Solubility of Hydrogen
in Zirconium Alloys,” Zirconium in the Nuclear Industry: Twelfth International
Symposium, ASTM STP 1354, G. P. Sabol and G. D. Moan, Eds., ASTM International, West
Conshohocken, PA., 2000, pp. 173–195.
[12] Vizcaino, P., Banchik, A. D., and Abriata, J. P., “Solubility of Hydrogen in Zircaloy-4: Irradi-
ation Induced Increase and Thermal Recovery,” J. Nucl. Mater., Vol. 304, 2002, pp.
96–106.
[13] Adamson, R. B., “Recovery of Irradiation Damage by Post-Irradiation Thermal Anneal-

ing—Relevance to Hydrogen Solubility and Dry Storage Issues,” Technical Report
1013446, EPRI, Palo Alto, CA, 2006.
[14] Shishov, V. N., “The Evolution of Microstructure and Deformation Stability in Zr-Nb-(Sn,
Fe) Alloys Under Neutron Irradiation,” Zirconium in the Nuclear Industry: Sixteenth Inter-
national Symposium, ASTM STP 1529, M. Limback and P. Barberis, Eds., ASTM Interna-
tional, Chengdu, China, May 9–13, 2010.
[15] Mader, E. V., Reitmeyer, M., Garcia Sedano, P., Morris, J., Cantonwine, P., Hallstadius, L.,
Potts, G., Peters, H. R., and Li, W., “EPRI BWR Channel Distortion Program,” Water Reac-
tor Fuel Performance Meeting, Paper #2-001, Chengdu, China, Sept 11–14, 2011.
[16] McGrath, M. A. and Yagnik, S. Y., “Experimental Investigation of Irradiation Creep and
Growth of Recrystallized Zircaloy-4 Guide Tubes Pre-Irradiated in PWR,” ASTM Zr Sym-
posium, Chengdu, China, J. ASTM Int., Vol. 8, No. 3, 2010, Paper ID JAI103770.
[17] Blavius, D., Münch, C., and Garner, N., Dimensional Behaviour of Fuel Channels—Update
on the Operational Experiences and Evaluation Results, KTG Jahrestagung Kerntechnik,
Hamburg, Germany, 2008.
[18] Griffiths, M., “A Review of Microstructure Evolution in Zirconium Alloys during Irradi-
ation,” J. Nucl. Mater., Vol. 159, 1988, pp. 190–218.
[19] Mahmood, S. T., Cantonwine, P. E., Lin, Y., Crawford, D. C., Mader, E. V., and Edsinger, K.,
“Shadow Corrosion-Induced Bow of Zircaloy-2 Channels,” J. ASTM Int., Vol. 7, 2010, p. 954.
[20] Garzarolli, F., Adamson, R., and Rudling, P., “Optimization of BWR Fuel Rod Cladding Condi-
tion for High Burnups,” ANS LWR Fuel Performance Meeting, Paper 069, Orlando, FL, Sept
26–29, 2010.
[21] Hohenberg, P. and Kohn, W., “Inhomogeneous Electron Gas,” Phys. Rev., Vol. 136, 1964,
p. B864.
[22] Kohn, W. and Sham, L. J., “Self-Consistent Equations Including Exchange and Correlation
Effects,” Phys. Rev., Vol. 140, 1965, p. A1133.
[23] Daw, M. S. and Baskes, M. I., “Embedded-Atom Method: Derivation and Application to
Impurities, Surfaces, and Other Defects in Metals,” Phys. Rev. B, Vol. 29, 1984, pp.
6443–6453.
[24] Finnis, M. W. and Sinclair, J. E., “A Simple Empirical N-Body Potential for Transition Met-
als,” Phil. Mag. A, Vol. 50, 1984, pp. 45–55.
[25] Plimpton, S., “Fast Parallel Algorithms for Short-Range Molecular Dynamics,” J. Comp.
Phys., Vol. 117, 1995, pp. 1–19.
[26] Kresse, G. and Furthmüller, J., “Efficient Iterative Schemes for Ab Initio Total-Energy Cal-
culations Using a Plane-Wave Basis Set,” Phys. Rev. B, Vol. 54, 1996, pp. 11169–11186.
[27] Kresse, G. and Hafner, J., “Ab Initio Molecular Dynamics for Liquid Metals,” Phys. Rev. B,
Vol. 47, 1993, pp. 558–561.
[28] Kresse, G. and Joubert, D., “From Ultrasoft Pseudopotentials to the Projector Augmented-
Wave Method,” Phys. Rev. B, Vol. 59, 1999, pp. 1758–1775.
[29] Kresse, G. and Furthmüller, J., “Efficiency of Ab Initio Total Energy Calculations for Met-
als and Semiconductors Using a Plane-Wave Basis Set,” Comp. Mater. Sci., Vol. 6, 1996,
pp. 15–50.
R
[30] MedeAV; 2.10. (2011). Materials Design, Angel Fire, NM.
[31] Perdew, J. P., Burke, K., and Ernzerhof, M., “Generalized Gradient Approximation Made
Simple,” Phys. Rev. Lett., Vol. 77, 1996, p. 3865.
[32] Blöchl, P. E., “Projector Augmented-Wave Method,” Phys. Rev. B, Vol. 50, 1994, p.
17953.
[33] Henkelman, G., Johannesson, G., and Jonsson, H., “Methods for Finding Saddle Points
and Minimum Energy Paths,” Progress on Theoretical Chemistry and Physics, S. D.
Schwartz, Ed., Kluwer Academic, Boston, 2000, pp. 269–300.
[34] Henkelman, G., Uberuaga, B. P., and Jonsson, H., “A Climbing Image Nudged Elastic
Band Method for Finding Saddle Points and Minimum Energy Paths,” J. Chem. Phys.,
Vol. 113, 2000, pp. 9901–9904.
[35] Parlinski, K., Li, Z. Q., and Kawazoe, Y., “First-Principles Determination of the Soft Mode
in Cubic ZrO2,” Phys. Rev. Lett., Vol. 78, 1997, pp. 4063–4066.
[36] Christensen, M., Wolf, W., Freeman, C., Wimmer, E., Adamson, R. B., Hallstadius, L.,
Cantonwine, P. E., and Mader, E. V., “H in a-Zr and in Zirconium Hydrides: Solubility,
Effect on Dimensional Changes, and the Role of Defects” (to be submitted to Physi-
cal Review B).
[37] National Institute of Standards and Technology, 2006, http://webbook.nist.gov (Last

accessed 2 Sept 2013).
[38] LePage, Y. and Saxe, P., “Symmetry-General Least-Squares Extraction of Elastic Data for
Strained Materials from Ab Initio Calculations of Stress,” Phys. Rev. B, Vol. 65, 2002, p.
104104.
[39] Mendelev, M. I. and Ackland, G. J., “Development of an Interatomic Potential for the Sim-
ulation of Phase Transformations in Zirconium,” Phil. Mag. Lett., Vol. 87, 2007, pp.
349–359.
[40] Christensen, M., Wolf, W., Freeman, C., Wimmer, E., Adamson, R. B., Hallstadius, L., Can-
tonwine, P. E., and Mader, E. V., “Effect of Alloying Elements on the Zirconium-Hydrogen
System” (submitted to Journal of Nuclear Materials).
[41] Narang, P., Paul, G., and Taylor, K., “Location of Hydrogen in a-Zirconium,” J. Less-
[42] Khoda-Bakhsh, R. and Ross, D. K., “Determination of the Hydrogen Site Occupation in
the a Phase of Zirconium Hydride and in the a and b Phases of Titanium Hydride by
Inelastic Neutron Scattering,” J. Phys. F: Met. Phys., Vol. 12, 1982, pp. 15–24.
[43] Domain, C., Besson, R., and Legris, A., “Atomic-Scale Ab Initio Study of the Zr-H System.
I: Bulk Properties,” Acta Mater., Vol. 50, 2002, pp. 3513–3526.
[44] Numakura, H., Ito, T., and Koiwa, M., “Low-Frequency Internal Friction Study of Zr-H and
Zr-D Alloys,” J. Less-Common Metals, Vol. 141, 1988, pp. 285–294.
[45] Strasser, A., Adamson, R. B., and Garzarolli, F., “The Effect of Hydrogen on Zirconium
Alloy Properties,” Special Topical Report ZIRAT13/IZNA8, Vol. I, ANT International, Möln-
lycke, Sweden, 2008.
[46] Adamson, R., “The Zirconium-Hydrogen System,” Special Topical Report ZIRAT13, Vol. I,
ANT International, Mölnlycke, Sweden, 2008, pp. 2-1–2-31
[47] Valizadeh, S., Ledergerber, G., Abolhassan, S., Jädernäs, D., Dahlbäck, M., Mader, E. V.,
Zhou, G., Wright, J., and Hallstadius, L., “Effects of Secondary Phase Particle Dissolution
on the in-Reactor Performance of BWR Cladding,” Zirconium in the Nuclear Industry:
Sixteenth International Symposium, ASTM STP 1529, M. Limback and P. Barberis, Eds.,
ASTM International, Chengdu, China, May 9–13, 2010, pp. 729–753.
[48] Neogy, S., Srivastava, D., Dey, G. K., Chakravartty, J. K., and Baerjee, S., “Annealing Stud-
ies on Zr-1Nb and Zr-1Nb-1Sn-0.1Fe Alloys,” Trans. Indian Inst. Met., Vol. 57, 2004, pp.
509–519.
[49] Villars, P. and Cenzual, K., Pearson’s Crystal Data: Crystal Structure Database for Inor-
ganic Compounds (on CD-ROM); ASM International, Materials Park, OH, 2012.
[50] Dworschak, F., Dimitrov, C., and Dimitrov, O., “Interaction of Self-Interstitials With Oxy-
gen Atoms in Electron-Irradiated Zirconium,” J. Nucl. Mater., Vol. 82, 1979, pp. 148–154.
[51] Hood, G., “Point Defect Diffusion in a-Zr,” J. Nucl. Mater., Vol. 159, 1988, pp. 149–175.
[52] Vekilova, O. Y., Bazhanov, D. I., Simak, S. I., and Abrikosov, I. A., “First-Principles Study of
Vacancy-Hydrogen Interaction in Pd,” Phys. Rev. B, Vol. 80, 2009, p. 024101.
[53] Adamson, R. B., Garzarolli, F., and Patterson, C., “In-Reactor Creep of Zirconium Alloys,”
Special Topical Report ZIRAT14/IZNA9, ANT International, Skultuna, Sweden, 2009.
[54] Kearns, J. J., “Diffusion Coefficient of Hydrogen in a Zirconium, Zircaloy-2, and Zircaloy-
4,” J. Nucl. Mater., Vol. 43, 1972, pp. 330–338.
[55] Someno, M., “Determination of the Solubility and Diffusion Coefficient of Hydrogen in
Zirconium,” Nihon Kinzoku Gakkaishi, Vol. 24, 1960, pp. 249–253.
[56] Sawatzky, A., “The Diffusion and Solubility of Hydrogen in the a Phase of Zircaloy-2,” J.
Nucl. Mater., Vol. 2, 1960, pp. 62–68.
[57] Mallett, M. W. and Albrecht, W. M., “Low-Pressure Solubility and Diffusion of Hydrogen
in Zirconium,” J. Electrochem. Soc., Vol. 104, 1957, pp. 142–146.
[58] Wimmer, E., Wolf, W., Sticht, J., Saxe, P., Geller, C. B., Najafabadi, R., and Young, G. A.,
“Temperature-Dependent Diffusion Coefficients from Ab Initio Computations: Hydrogen,
Deuterium, and Tritium in Nickel,” Phys. Rev. B, Vol. 77, 2008, p. 134305.
[59] Perez, R. A. and Weissmann, M., “Ab Initio Approach to the Effect of Fe on the Diffusion
in HCP Zr,” J. Nucl. Mater., Vol. 374, 2008, pp. 95–100.
[60] Griffiths, M., Styles, R., Woo, C., Phillipp, F., and Frank, W., “Study of Point Defect Mobili-
ties in Zirconium During Electron Irradiation in a High-Voltage Electron Microscope,” J.
Nucl. Mater., Vol. 208, 1994, pp. 324–334.
[61] Yamanaka, S., Tanaka, T., and Miyake, M., “Effect of Oxygen on Hydrogen Solubility in
Zirconium,” J. Nucl. Mater., Vol. 167, 1989, pp. 231–237.
[62] Flinn, B. J., Zhang, C.-S., and Norton, P. R., “Oxygen Diffusion along the [0001] Axis in
Zr(0001),” Phys. Rev. B, Vol. 47, 1993, pp. 16499–16505.
[63] Hood, G., Zou, H., Herbert, S., Schultz, R., Nakajima, H., and Jackman, J., “Oxygen Diffu-
sion in a-Zr Single Crystals,” J. Nucl. Mater., Vol. 210, 1994, pp. 1–5.
[64] Ritchie, I. G. and Atrens, A., “The Diffusion of Oxygen in a-Zirconium,” J. Nucl. Mater.,
Vol. 67, 1977, pp. 254–264.
[65] Prieto, P., Galán, L., Sanz, J. M., and Reuda, F., “An XPS Study of the Kinetics of Dissolu-
tion of ZrO2 in a-Zr,” Surf. Interf. Anal., Vol. 16, 1990, pp. 535–539.
[66] De González, C. O. and Garcı́a, E. A., “Determination of the Diffusion Coefficients of Oxy-
gen in Zirconium by Means of XPS,” Appl. Surf. Sci., Vol. 44, 1990, pp. 211–219.
[67] Tanabe, T. and Tomita, M., “Surface Oxidation of Zirconium above Room Temperature,”
Surf. Sci., Vol. 222, 1989, pp. 84–94.
[68] Foord, J. S., Goddard, P. J., and Lambert, R. M., “Adsorption and Absorption of Diatomic
Gases by Zirconium: Studies of the Dissociation and Diffusion of CO, NO, N2, O2, and
D2,” Surf. Sci., Vol. 94, 1980, pp. 339–354.
[69] Woo, C. H., “Defect Accumulation Behaviour in HCP Metals and Alloys,” J. Nucl. Mater.,
Vol. 276, 2000, pp. 90–103.
[70] Hood, G. M. and Shultz, R. J., “The Influence of Sn Additions on the Recovery of Electron
Irradiated a-Zr: A Positron Annihilation Study,” Mater. Sci. Forum, Vols. 15–18, 1987, pp.
745–750.
[71] Eldrup, M., Hood, G. M., Pedersen, N. J., and Shultz, R. J., “A Positron Lifetime Study of
Defect Recovery in Electron Irradiated Zircaloy-2 and Zr-2.5Nb,” Mater. Sci. Forum, Vols.
105–110, 1992, pp. 997–1000.
DISCUSSION
Questions from Gargi Choudhuri, BARC Mumbai:—Can you comment on oxy-
gen diffusivity in Zr in different crystallographic directions?
Authors’ Response:—The oxygen diffusivity in Zr was investigated by ab-initio

simulations of the ground state, the metastable states and all involved transition
states along the interstitial migration paths, taking into account electronic energies
as well as vibrational contributions. These resulting T-dependent barrier heights
and jump rates are used in kinetic Monte Carlo simulations to obtain temperature
dependent diffusivities.
The anisotropy of the oxygen diffusivity is very small, such that one could
speak of an almost isotropic behavior. The expressions for the temperature depend-
ent diffusion coefficients in basal and axial direction are given in Eq. 12 in the paper
and shown graphically in Figure 11. Note: Eq 12 and Figure 11 need to be con-
firmed as correct in final version of paper.
Questions from S. Anantharaman, BARC Mumbai Q1:—What could be the

effect of trapping of hydrogen at point/planar defects on the solubility of hydrogen
in Zr? Will this hydrogen lead to the formation of hydrides at a later date?
Authors’ Response:—The solubility of hydrogen in bulk zirconium is an ideal

physical property and does not account for defects. We have calculated this prop-
erty purely based on ab-initio techniques and found good agreement with available
experimental data. Hydrogen is attracted by vacancy defects, vacancy loops and
grain boundaries. The presence of point and planar defects therefore increases the
apparent solubility of hydrogen in the material according to the defect’s capacity
for hydrogen and the defect concentration. Thus, defects provide a location of
enhanced hydrogen contents, and in particular healing the defect may result in a
supersaturation of hydrogen in Zr leading to nucleation of hydride phases. For
instance, the collapse of H-rich c-loops can lead to supersaturated Zr and nuclea-
tion of hydrides.
Q2:—Will deuterium make a difference?
Authors’ Response:—The isotope effect comes in via the vibrational contribu-

tion to the thermodynamics only, and thus is expected to be quite small. By far the
largest contribution is the chemical effect, which is not affected. In the simulation it
is quite straightforward to take into account the atomic mass for the phonon spec-
tra. To get an idea about the effect we refer to our previous work on ab-initio simu-
lation of the hydrogen diffusivity in nickel, where isotope effects were discussed in
detail (see E. Wimmer, W. Wolf, J. Sticht, P. Saxe, C. B. Geller, R. Najafabadi, and
G. A. Young, “Temperature-Dependent Diffusion Coefficients from ab initio Compu-

tations: Hydrogen, Deuterium, and Tritium in Nickel” Physical Review B 77, 134305
(2008)).
Q3:—<c> loops being similar to (1017), the habit plane of hydride may be the
reason for their preference to form on <c> loops <c>. Please comment.
Authors’ Response:—The preferred habit plane of hydrides is near (1017), which

is about 17 degrees off from the basal (0001) plane on which <c> loops form.
When <c> loops and hydrides are simultaneously imaged in TEM, they look to be
parallel to one another and sometimes appear to overlap. From our simulations it is
evident that hydrogen is attracted by loops and will accumulate at these defects. In
particular during loop collapse processes this may lead to supersaturation and
nucleation of hydrides. Therefore, the close proximity and overlap of hydrides and
<c> loops and the close orientational relationship between habit planes and loop
planes may be understood in terms of the computed attractive interaction between
hydrogen and <c> loops. A detailed understanding why exactly (1017) is the pre-
ferred habit plane would require direct simulation of the interface properties,
though.
Q4:—Can the trapped hydrogen be released by annealing at a higher

temperature?
Authors’ Response:—We haven’t conducted explicit simulations for hydrogen

trapped in defects at elevated temperatures, but certainly we expect that hydrogen
can be released by an annealing process.
BASIC METALLURGY
AND ALLOYING EFFECTS
Gargi Choudhuri,1 Sibasis Chakraborty,2 B. K. Shah,3

D. Srivastava,4 and G. K. Dey4
Phase Field Modeling

of Microstructure Evolution
in Zirconium Base Alloys
Reference
Choudhuri, Gargi, Chakraborty, Sibasis, Shah, B. K., Srivastava, D., and Dey, G. K., “Phase Field
Modeling of Microstructure Evolution in Zirconium Base Alloys,” Zirconium in the Nuclear
Industry: 17th International Symposium, STP 1543, Robert Comstock and Pierre Barberis, Eds.,
pp. 95–117, doi:10.1520/STP154320130045, ASTM International, West Conshohocken, PA
2015.5
ABSTRACT
Zirconium base alloys are used as critical components of water cooled
nuclear power reactors. The phase transformation and microstructural
evolution in these alloys are complex. Depending on soaking temperature
and cooling rate, the b phase of these alloys can transform into a variety of
microstructures, viz., allotriomorph alpha, Widmanstatten alpha with parallel-
plates or basket-weave morphology, martensitic microstructure, and omega
phase. An advanced numerical modeling technique, the “Phase field method”
(PFM) has been used to study morphological evolution of b–Zr phase in
dilute Zr–Nb alloys during b Zr ðBCC Þ ! a ZrðHCPÞ transformation in
mesoscopic scale. The nucleation events have been incorporated in the
model both explicitly and implicitly. The growth rate and morphology
selection have been investigated by varying the supersaturation, i.e.,
temperature and Nb content in Zr–Nb alloys. For Zr–2.5 %Nb with low
undercooling, the driving force for plate growth decreases and at 1054 K it
Manuscript received March 15, 2013; accepted for publication April 11, 2014; published online October 4, 2014.
1
Quality Assurance Division, Bhabha Atomic Research Centre, Mumbai 400085, India (Corresponding
author), e-mail: gargi@barc.gov.in
2
Radiometallurgy Division, Bhabha Atomic Research Centre, Mumbai 400085, India.
3
Quality Assurance Division, Bhabha Atomic Research Centre, Mumbai 400085, India.
4
Material Science Division, Bhabha Atomic Research Centre, Mumbai 400085, India.
5
Hyderabad, India.
Copyright V
shows formation of allotriomorph a through classical diffusional

transformation. Higher undercooling favours Widmanstatten lath formation.
The growth velocity of the Widmanstatten plate is constant. The change in
alloy composition changes the effective supersaturation at the a=b interface
and leads to a different tip velocity. When Nb content reduces below
200 ppm, lath morphology does not develop and the interface remains
planar only.
Keywords
phase field method, lath, interface, supersaturation, allotriomorph, Zirconium
Introduction
Zirconium–Niobium (Zr–Nb) alloys exhibit a wide variety of phase transforma-
tions. Depending upon the diffusion length of atomic species, different types of dif-
fusional transformation are encountered in Zr based alloys which lead to
generation of wide range of morphological products [1–4]. In Zr–Nb alloys, Nb acts
as a b stabilizer. At higher temperatures, complete solubility exists between Zr and
Nb in the b phase. At 893 K (620 C) (Fig. 1), the b–Zr containing 18.5 wt. % Nb
(BCC) transforms into equilibrium products a–Zr (HCP) and b–Nb (BCC). This
transformation (bZr to a and bNb) is very sluggish at lower temperature and com-
plete transformation does not occur. The metastable b–Zr progressively decom-
poses to a–Zr phase and b phase enriched with Nb and ultimately forms b-Nb
(Zr–85 wt. %Nb). Meta-stable x-phase precipitates as an intermediate step during
decomposition depending on the composition and temperature of decomposition.
During fast cooling b–Zr undergoes martensitic transformation up to about 8.0 at.
% Nb [5–7]. During intermediate or slow cooling rate, mainly two distinct mor-
phologies of a–Zr phase are found. One is allotriomorph a (Fig. 2) formed during
slow cooling or at lower undercooling and another one is lath shaped product
(Figs. 3(a), 3(b), and 3(c)) called Widmanstatten plate [2]. This plate morphology is
formed when the dilute Zr–2.5 Nb alloy is step quenched from x-phase and heat
treated isothermally at 823 K or during intermediate cooling rate [6]. The alpha zirco-
nium phase nucleating in beta phase always follows the Burgers orientation relation-
ship ((0001) hcp|| (011) bcc, (1–100) hcp|| (2–1–1) bcc and [11–20] hcp || [11–1] bcc)
irrespective of whether the transformation is diffusional or martensitic [1,7]. This
orientation relationship is responsible for formation of very low energy a=b inter-
face. In Zr–2.5Nb alloys, a interfaces consist of closely spaced (a) type dislocations
[7]. The interface structure of c phase in the a matrix of Zr–2.5Nb alloy has been
studied by Zhang et al. [8]. The misfit at the habit plane is accommodated by a sin-
gle set of dislocations lying along the invariant line. At the other side, the facet
planes consist of two or three different sets of dislocations. These dislocations are
required to relieve the misfit strain.
Several researchers [11,12] have studied pattern formation to design micro-
structures during crystal growth in a dynamical system. Widmanstatten plate
CHOUDHURI ET AL., DOI 10.1520/STP154320130045 97
FIG. 1 Zr–Nb phase diagram. The points A, B, D, E, and F specify the operating points,
i.e., the alloys with corresponding simulating temperature for Widmanstatten
plates and C for planar growth, respectively.
growth is a type of pattern formation problem during solid-solid phase transforma-

tion in both interstitial and substitutional metallic alloys. The diffusion controlled
growth of unstable planar surface generates a pattern due to interplay among sur-
face tension, bulk volume free energy, and kinetic phenomena like diffusion. The
investigations about Widmanstatten plate formation will remain incomplete if the
nucleation events along with growth process are not considered. Although many
researchers find a close relationship with the Mullin–Sekerka (M–S) type instability
FIG. 2 Typical grain boundary a Allotrimorph observed surrounding the prior b grain
along with lath shaped product formed in Zr–Nb alloy during cooling from b
phase [9].
FIG. 3 Typical Widmanstatten lath found in Zr–2.5Nb alloy during gas quenching from
a þ b phase [10].
theory [13–15], the second school of thought is the strain induced sympathetic
nucleation theory [16,17]. Experimental techniques like Laser Confocal Microscopy
[18] had been used to corroborate sympathetic nucleation theory of plate growth.
While in sympathetic nucleation theory strain plays a major role, for phase trans-
formations which are purely diffusion controlled in nature, the M–S instability
theory can be used in the nucleation stage to perturb the phase boundary before
progressing into the growth stage ignoring the strain energy contribution.
Over the last decade, phase-field method (PFM) has emerged as a powerful
computational tool for modeling and predicting the meso-scale morphological and
microstructure evolution during solid–liquid and solid–solid phase transformations
[19]. In PFM, predicting microstructure evolution at real length and time scales
requires that the fundamental model inputs be linked to thermodynamic and mo-
bility databases. Extensive work has been done by Wang et al. [20] in developing
and integrating thermodynamic modeling, mobility data base, and phase field simu-
lation to predict microstructural evolution in Ti base alloys mainly in Ti-6Al-4 V
alloy. Loginova et al. [21] studied Widmanstatten ferrite plate formation in binary
Fe–C interstitial solid solution system using a highly anisotropic interfacial energy.
To circumvent the ill-posed phase field equation due to highly anisotropic nature of
interfacial energy, the regularization method was adopted in the non-differential
domain of interfacial energy.
Limited work [22,23] has been done in modeling of microstructural evolution
in Zr–Nb alloys during various thermo-mechanical processing. Phase field
approach has been formulated for predicting microstructural evolution during athe-
rmal x-phase transformation [24] in Zr–Nb the alloy system. In our earlier paper
[25], the phase field model had been formulated during b Zr ðBCCÞ
! a ZrðHCPÞ phase transformation. Morphological evolution had been studied
for Zr–2.5Nb pressure tube material using pre-nucleated protrusion. In the current
paper, the formation of plate morphology during b Zr ðBCC Þ ! a ZrðHCPÞ

phase transformation has been modeled using stochastic nucleation events and fol-
lowed by deterministic growth using highly anisotropic interfacial energy as
adopted by Loginova et al. [21]. The nucleation phenomenon has been incorporated
both implicitly and explicitly [26–33]. The time dependent evolution of the spatially
dependent field variables is studied by solving the Allen–Cahn equation [34] for
non-conserved order parameter and Cahn–Hilliard equation [35] for conserved
concentration parameter. Using Gibbs free energy functional and diffusional mobil-
ity, different morphology of a–Zr phase evolution has been simulated in different
Zr–Nb binary alloys. Based on this model, the growth rate and morphology selec-
tion were investigated by varying the supersaturation, i.e., temperature and Nb con-
tent in different Zr–Nb alloys.
Mathematical Formulation
The formation of a–Zr plate occurs by nucleation-growth mechanism [1] which is a
first order transformation. To study the microstructure evolution during a–Zr
phase formation, phase field method has been used. This method is based on the
concept that a microstructure with compositional and structural heterogeneities
can be described by a continuous function, u, which is set to 0 in hexagonal closed
packed (HCP) a phase and 1 in body centered cubic (BCC) b phase. The interface
region has continuous values in between 0 and 1 and is diffuse in nature. The ther-
modynamic basis of the model is depicted through the free energy functional [25]
ð !
g ðu; c; T Þ e2c 2 e2u 2
(1) G¼ þ jrcj þ jruj dv
v Vm 2 2
where g ðu; c; T Þ, is the homogeneous free energy density at a given temperature

(T), non-conservative phase field order parameter (u), and conservative phase field
parameter concentration (c) (i.e., mole fraction of Nb in the alloy). The molar volume
(Vm ) is assumed to be constant for both the phases and eu and ec are the gradient
energy coefficient for order parameter and concentration in-homogeneities, respec-
tively. Simulations have been carried out in isothermal condition as rapid heat con-
duction has been assumed. Temperature (T) is expressed in Kelvin scale (K).
The homogeneous free energy density g ðu; c; T Þ associated with the BCC to
HCP phase transformation in Zr–Nb system is approximated with the following
functions, fp ðc; u; T Þ and kðuÞ [36,37].
1. fp ðc; u; TÞ is an interpolation function which combines the Gibbs free energy
expressions of the coexisting phases
(2) fp ðc; u; T Þ ¼ ð1 pðuÞÞGa ðc; T Þ þ pðuÞGb ðc; T Þ
where:
Ga ðc; TÞand Gb ðc; T Þ ¼ Gibbs free energy functions for the composition and
temperature dependence of the a phase and the b phase, respectively, and
pðuÞ ¼ a weight function expressed as

(3) pðuÞ ¼ u3 10 15u þ 6u2
so that when u ¼ 1in b-phase, pðuÞ ¼ 1 and when u ¼ 0; in a-phase,

pðuÞ ¼ 0.
2. kðuÞ is a double-well function where

(4) kðuÞ ¼ w qðuÞ ¼ w u2 2u3 þ u4
Hence w is the kinetic barrier between the two minima in the double well free
energy density representation and can be adjusted to fit the desired interfacial
energy. The w has been expressed as ð3 r=dÞ. The choice of w parameter has been
described elsewhere [25]. From Eq 4, it is seen that qðuÞ ¼ u2 2u3 þ u4 , which
indicates that qðuÞ ¼ 0 at both the phases and has the highest value at u ¼ 0:5 and
p0 ðuÞ ¼ 30qðuÞ.
Finally, the Gibbs free energy per mole, g ðc; u; T Þ becomes
(5) g ðc; u; T Þ ¼ fp ðc; u; T Þ þ kðuÞ
As the interface is diffuse in nature in the PFM technique, proper incorporation

of interfacial energy is most important. The bZr ðBCC Þ ! aZrðHCPÞ phase trans-
formation in Zr base alloys follows the Burgers Orientation relationship, which
results in formation of coherent/semi coherent interface with very low interfacial
energy. The parallel broad faces in Widmanstatten plates are coherent in nature
with very low interfacial energy and for perfectly coherent interface thickness is
zero. The tip of the Widmanstatten plate is incoherent in nature with relatively
higher interfacial energy. This anisotropy in interfacial energy is introduced in the
simulations through an analytical function, gðhÞ, where

uy
h ¼ arctan
ux
and is nearly equal to the angle between interface normal and x axis. The same
analytical function [38] can also depict the orientation dependence diffuse interface
thickness.
r ¼ r0 gðhÞ; d ¼ d0 gðhÞ
where r0 and d0 are the maximum interfacial energy and maximum interface
width of incoherent interface which are the input parameters of the model.
gðhÞ is a cosine function of h, i.e., g ¼ ðð1 þ cj cosðh h0 ÞjÞ=ð1 þ cÞÞ and is
regularized to satisfy Gibbs Thomson criteria [21]. Where h0 is the angle between
the interfacial plane having good crystallographic fit with the x axis, c is the extent
of anisotropy. The maximum interfacial energy of incoherent interface is r0 at
h ¼ h0 and the lowest interfacial energy r0 =ð1 þ cÞ of coherent interface occurs
when h h0 ¼ p=2;. i.e. c þ 1 becomes the ratio of interfacial energy of incoherent

interface to the coherent interface. The very high value of c ensures very low inter-
facial energy of coherent interface which emphasizes perfect match in between a
and b phases. During growth, in our model we have considered to have a good crys-
tallographic fit along an interfacial plane having the angle h0 ¼ 0 with the x axis. If
a plate with coherent sides develops it would grow with that angle (0 Þ toward the x
axis.
The energy penalty due to presence of order parameter in-homogeneities is a
function of gradient energy coefficient of order parameter ðeu Þ which can be
expressed as [21]
e2u ¼ 6rd ¼ 6r0 d0 gðhÞ2
For Widmanstatten plate formation, gradient energy coefficient for concentra-

tion in-homogeneities (ec ¼ 0Þ is neglected as this term will not contribute signifi-
cantly in kinetics of plate formation (for first order phase transformation), but this
term plays an important role in case of Spinoidal decomposition where slight
change or fluctuation in composition lead to appreciable decrease in free energy.
From here onwards, eu is denoted as e.
The simplest form of kinetic equation is the stochastic Langevin equation based
on the phenomenological time dependent Ginzburg–Landau kinetic equation [37],
which assumes that the rate of evolution of a field is a linear function with respect
to the transformation driving forces. This formulation has been used for simulating
implicit nucleation events.
@uðr; t Þ dG
(6) ¼ Mu þ nu ðr; t Þ
@t duðr; t Þ
where:
ðdG=duÞ ¼ the thermodynamic driving force, and
Mu ¼ the kinetic coefficient related to the interface mobility and chosen to
ensure diffusion controlled process. r is the spatial coordinate vector.
For anisotropic interface [21,38] where e is orientation dependent,
!
dG @g ðu; c; T Þ 2 2 @ 0 @u @ 0 @u
¼ e r uþ eeh eeh
duðr; t Þ @u @x @y @y @x
0
where eh ¼ ð@e=@hÞ.
nu ðr; tÞ is the Langevin random noise term [39]. nu ðr; tÞ is taken to be Gaus-
sian distributed and its correlation properties should meet the requirement of the
fluctuation–dissipation theorem.
0 0 0
0

(7) < nu ðr; t Þ; nu r ; t >¼ 2kB TMu d r r d t t
where kB is the Boltzmann constant.

In our implicit nucleation and growth model, we incorporate the stochastic

noise in a simple form to stimulate fluctuations at the planar interface of a –Zr. It is
introduced by a function which incorporates fluctuations in the local driving forces
using order parameter.

0 Gb Ga 0 w
(8) nu ¼ arn 16kðuÞMu p ðuÞ þ q ðu Þ
Vm Vm
where rn is a random number between 0 and 1, and a in Eq 8 is the amplitude
of fluctuation.
The temporal evolution of the concentration field can be obtained by solving
the Cahn–Hilliard diffusion equation
@cðr; tÞ 00 dG
(9) ¼ L r2 þ nc ðr; tÞ
@t dcðr; tÞ
where:
00
L ¼ a phenomenological constant that relates the flux with the driving force
and depicts the kinetic coefficient characterizing diffusional mobility of Nb atom
00
(MNb) in disordered phase b -phase through L ¼ cð1 cÞMNb [40,41], and
nc ðr; tÞ ¼ the Langevin noise term and is assumed to be zero.
hcp
MNb can be represented by phase field parameter, MNb ¼ ðMNb Þ1pðuÞ MNb bcc pðuÞ
,
hcp hcp bcc bcc hcp bcc
where MNb ¼ ðDNb =RTÞ and MNb ¼ ðDNb =RTÞ. DNb and DNb are the diffusivity of
the Nb atom in a (HCP) and b (BCC) phase, respectively (Table 1).
In order to account for nucleation in a meta-stable system, it is necessary to
consider fluctuations in the local driving forces for the reaction to occur. However,
coarse graining in mesoscale average out Langevin noises fluctuations of finer scale.
As the process of nucleation happens in lower time and length scale, it is computa-
tionally difficult to simulate nucleation and growth process in a single simulation
and at the outset it can be said that in mesoscale one can only simulate nucleation
TABLE 1 Simulation parameters for PFM.
Parameters’ Name Symbol (unit) Expression/Value
3
Molar Volume Vm (m /mol) 1:4060 105
Distance between atoms b (Å) 3.23
Incoherent Interface Thickness d0 (nm) 5
Interfacial Energy of incoherent r0 (J/m2) 0.3
interface
Extent of anisotropy c 120
hcp
Nb diffusion coefficient in a-Zr DNb (m2/s) 6:6 1010 eð15851:4=TÞ [42]
2
Nb diffusion coefficient b-Zr Dbcc
Nb (m /s) 9 109 ðT=1136Þ18:1
[43]
eð25100þ35:5ðT1136ÞÞ=ð1:98TÞ
Note: References cited in the Table are [42] and [43].

events (i.e., only equlibriated nucleus can be generated) not the physical process of
nucleation. This is done by artificially allowing the noise terms (nu in Eq 8) to have a
large amplitude at the beginning of the simulations to form a suitable number of via-
ble nuclei and subsequently turning off this noise term (nu ) to observe the growth.
The explicit nucleation algorithm developed by Simmons et al. [44] is another
way to allow nucleation events to occur depending on a criterion of nucleation rate
derived from the classical theory. From the classical theory [45], the nucleation rate is
s
DG
(10) J ¼ ZNb exp exp
kT t
where:
Z ¼ the non-equilibrium factor due to Zeldovich,
N ¼ the number of atoms in each phase field cell,
b* ¼ the frequency factor,
DG ¼ the activation free energy necessary to form a stable nucleus,
k ¼ Boltzmann’s constant,
s ¼ an incubation time, and
t ¼ time.
For simplicity, the nucleation rate can be approximated using the classical
nucleation theory without considering incubation.
In a two-dimensional model, DG is determined from following relationship
[44] where c is the interfacial energy (assumed as independent of temperature) and
DGa is the free energy change per unit area due to transformation.
pc2
(11) DG ¼
DGa
The expectation value of the number of nuclei that form in Dt time period is

J Dt. The probability of a nucleation event occurring can be approximated by
(12) Pn ¼ 1 expðJ Dt Þ
where expðJ Dt Þ is the zero event probability of a Poisson distribution.

In our model, nucleation events are explicitly incorporated by generating a uni-
form random variable at the a=b interface and comparing its value with Pn . If it is
more than Pn ; nuclei formed; otherwise they did not.
Equations 6 and 9 are transformed into a dimensionless form through the
introduction of reference length (l) and diffusion time (sÞ.
a
t RT MNb
(13) l ¼ :9d0 and diffusion time s ¼ 2
l
In non-dimensional form, the two coupled PDEs are solved by the finite ele-
ment method using an adaptive unstructured grid which reduces the computational
load without affecting the accuracy. Zero flux Neumann boundary conditions were
applied. The material parameters used in the simulations are summarized in Table 1.
The thermodynamic and kinetic input expressions are listed in the Appendix. It is
worth mentioning here that in reality, width of incoherent phase boundaries are
less than 1.5 nm, the model parameter d0 should be in between 0.5–1.5 nm. The
coherent interface is much thinner than incoherent interface. For perfect coherent
interface, thickness is zero. The phase field equation shows very stiff gradient at
the interface region and with realistic interface thickness the tasks of solving it
becomes challenging. Due to the requirement of huge computational power, we
have taken incoherent interface thickness d0 as 5 nm and in non-dimensional space
the interface regions have been discretized using five grid points and the rest of the
region with large triangles optimizing the computational load and accuracy of the
solution.
Results and Discussion

The Zr rich portion of Zr–Nb equilibrium phase diagram is given in Fig. 1 [46]. The
simulations were carried out at different temperatures and for different Nb contents
in Zr–Nb alloys. These temperatures and respective Nb contents in Zr–Nb system
were shown as A, B, C, D, E, and F in Fig. 1. In all simulations a thin layer of a–Zr
with planar interface was put along with b–Zr (Fig. 4(a)). The initial composition of
a–Zr was taken from the equilibrium phase diagram.
IMPLICIT NUCLEATION AND SUBSEQUENT GROWTH

At 890 K and below, simulations with a stochastic source term shows that the pla-
nar interface in Zr–1Nb, Zr–2.5Nb, Zr–5Nb, and Zr–7Nb becomes unstable and
undulations generate in the a=b diffuse interface. These undulations having favor-
able length scale grow and lead to the formation of stable protrusions (Fig. 4(b)
shown for Zr–2.5Nb). The formation of the stable protrusion causes solute to be
rejected laterally and pile up of solute at the root of the protrusion occurs which
restricts the further movement of the adjacent planar interface and growth of Wid-
manstatten lath is favored. The tip velocity of the lath depends on how fast solute
atoms diffuse away from the interface. A smaller tip radius allows the excess solute
atoms to diffuse away from the tip more rapidly, but also increases surface to vol-
ume ratio which leads to energy penalty in creating the surface. At an optimum ra-
dius, tip will grow with a maximum growth velocity. The earliest model for
lengthening of a plate was given by Zener [47] and further modified by Hillert [48].
The essential feature of the model was the assumption of a constant tip radius
which will give maximum growth rate (v) for a needle crystal [49] where
dr D C0 Ca X
(14) v¼ ¼2 ¼ 2D
dt r Cb Ca r
where:
Ca and Cb ¼ the equilibrium composition at particular temperature,
FIG. 4 Development of Widmanstatten lath morphology from planar front of

allotriomorph a–Zr in Zr–2.5Nb alloy using implicit nucleation. (a) Is the initial
microstructure. The dark region in (a) indicates pre-nucleated allotriomorph
a–Zr (u ¼ 0Þ. u ¼ 1 in the region of b–Zr. (b), (c), and (d) are microstructure at
different non-dimensional time steps (.035,.0531,.0663, respectively). Planar
front becomes unstable at 890 K. The colour code depicts the mole fraction of
Nb in Zr–Nb binary alloy. The domain size is 1.57 lm 5.14 lm.
D ¼ the diffusivity of the solute in the alloy having composition C0 , and

X ¼ the dimensionless supersaturation, described by ðC0 Ca =Cb Ca Þ.
After including the surface free energy effect [49], the above equation becomes
dr D C0 Ca ðrÞ X
(15) v¼ ¼2 ¼ 2D
dt r Cb ðrÞ Ca ðrÞ r
For a plate-like crystal the equivalent expression given by Ivantsov [49] is:

(16) X ¼ pP:5 expðPÞerf P:5
where P is the Peclet Numer P ¼ r=2D.

Livingston and Cahn [50] observed that the tip radius was larger than that
expected to give the maximum growth rate. As a result the growth rate was, of
course, less than the expected maximum velocity.
In this model, there is no need to specify tip radius, or tip velocity. Depending
upon operating temperature, composition of the alloy and degree of anisotropy, the
tip of undulation attains a characteristic velocity. When the degree of anisotropy
ðcÞ reduces below 20, lath morphology does not develop at or below 890 K and a
planar front is retained. c þ 1, which is the ratio of interfacial energy of incoherent
interface to the coherent interface, plays a vital role in formation of Widmanstatten
lath morphology. A higher value of c favours the formation of lath morphology at
higher undercooling. With increasing c, the plate lengthening rate increases while
the widening rate reduces. Beyond c ¼ 120, no further change in widening rate is
observed for single lath growth. In our simulations, the c value has been fixed at
120. A detailed description has been given in our earlier paper [25].
The diffusion field surrounding the initial protrusion (Fig. 4(c)) triggers the forma-
tion of other neighbouring protrusions and subsequently the entire planar interface
becomes unstable and the simultaneous growth of protrusions develops into parallel
laths giving rise to side plate morphology (Fig. 4(d)). Each lath of a–Zr including side
and tip are surrounded by b phase only, having higher concentration of Nb. This Nb
concentration will depend on time and temperature of transformation. The root of the
protrusion is always having high Nb concentration as shown in Fig. 4(d) (mole fraction
of Nb 0.16–b–Zr) whereas the tip is surrounded by b–Zr with low Nb content.
Decreasing temperature simulations show the formation of a–Zr side plates
with finer and finer spacing. As temperature is lowered, the interaction of diffusion
field of neighbouring protrusions is less; this allows growth of a large number of
protrusions leading to a finer lath width. With increasing undercooling, the tip ve-
locity also increases (Fig. 5).
When simulations were carried out at a higher temperature, i.e., with lower
undercooling (1054 K), no instability of planar interface is formed during the whole
simulation process and the entire interface moves in classical diffusion controlled
way giving the formation of allotriomorph a (Figs. 6(a) and 6(b)). At higher temper-
atures due to higher diffusivity, solute atoms get distributed more or less uniformly
at the tip as well as at the root of the protrusion, giving a uniform velocity of the
entire interface while planar growth is favored. With time, build-up of Nb concen-
tration has been observed ahead of the entire planar interface.
EXPLICIT NUCLEATION EVENTS AND SUBSEQUENT GROWTH

In the case of an explicit nucleation event, different numbers of equilibrated nuclei
were generated (Fig. 7(a)) on the a=b interface by generating a uniform random
variable. During growth with stochastic source term n ¼ 0, these equilibrated nuclei
have been grown into side plate morphology (Fig. 7(b)). Development of this mor-
phology is due to the interplay between temperature, interfacial energy, and the
degree of anisotropy in interfacial energy.
Interface Velocity
The a=b interface position has been fixed on the iso-line of u ¼ 0:5. In case of
Zr–2.5Nb alloy at 890 K, the plot of tip position (Fig. 8) at different time shows a
FIG. 5 Tip velocity as a function of undercooling in terms of (1/T) multiplied by 1000.

Lower temperature, i.e., higher undercooling favoured Widmanstatten lath
morphology.
linear trend for Widmanstatten plate growth, i.e., a constant velocity profile and
corresponding tip velocity is 20 lm=s. A constant lengthening rate is also experi-
mentally found [51]. The width of the single lath grows slowly compared to tip and
its growth rate in the initial stage is around 1.5 lm=s for the same c. The reasons
FIG. 6 Stable planar front grows from pre-nucleated allotriomorph a–Zr of Zr–2.5Nb
alloy at 1054 K. No stable perturbation has been formed at this temperature. The
images ((a) and (b)) were taken at 0.0, 0.107 non-dimensional time steps. The
colour code depicts the mole fraction of Nb in Zr–Nb binary alloy. The black line in
(b) denotes initial position of the interface. The domain size is 1.57 lm 5.08 lm.
FIG. 7 Development of Widmanstatten lath morphology at 890 K from allotriomorph

a–Zr in Zr–2.5Nb alloy using explicit nucleation. (a) is the initial microstructure
and (b) is the microstructure at 0.0814 non-dimensional time step. The
grayscale colour code depicts the mole fraction of Nb in Zr–Nb binary alloy. The
domain size is 1.8 lm 4.26 lm.
FIG. 8 Position of interface at different instants (dimensionless) for Widmanstatten

plate growth at 890 K indicating linear lengthening rate.
behind the low growth rate in width direction are that the broad faces of each lath
have a very good atomic fit with parent lattice and due to difficulty in joining of
individual atoms on the habit plane. Thickening of the plates are restricted, giving a
very low values of mobility. These broad faces move forward by sideways motion of
ledges. It was indicated by Aaronson [49] that ledged interfaces would always show
interface-controlled or mixed controlled growth and they grow more slowly than
expected for diffusion control. In addition, its tips grow linearly maintaining a con-
stant radius of curvature.
However, for classical planar growth at 1054 K, the interface position shows
(Fig. 9) a parabolic profile with diffusion time (s) and the velocity of the interface is
essentially proportional to s1=2 .
Effect of Supersaturation on Growth Velocity

The supersaturation of the alloy is defined as X0 ¼ ðcb=a eq co =cb=a eq ca Þ, where
cb=aeq is the equilibrium concentration of solute at the b=a interface, ca is the solute
concentration of the growing a phase and co is the initial alloy composition. It also
may be noted that the effective supersaturation at the interface deviates from X0
depending upon solute concentration on the two sides of the interface which are
determined by local equilibrium at the interface. Figure 10 shows the dependence of
tip velocity with supersaturation ðX0 Þ for Zr–0.8Nb, Zr–1Nb, Zr–2.5Nb, Zr–3.5Nb,
Zr–5.5Nb, and Zr–7Nb alloys at 890 K. The tip velocity increases steadily with
increase in supersaturation. At high super-saturation, the growth velocity
FIG. 9 Position of interface at different instants (dimensionless) for diffusion controlled

planar growth at 1054 K.
FIG. 10 Tip velocity as a function of supersaturation X0 at 890 K. At very high

supersaturation steeper slope presents and transition was observed at X0 0.83.
approaches a linear dependence with X0 .The work of Hillert [52] discusses the
appropriateness of the solutions of Zener–Hillert, Liu–Ågren, and Trivedi at differ-
ent supersaturation. The Trivedi’s solution is only valid at very high supersaturation
( X0 ¼ 0.99) i.e., close to 1.The Zener–Hillert proposes a velocity profile which
follows two different slopes (one from X0 ¼ 0.6–0.88 and another from 0.88–1). In
our case, the velocity profile is close to what was found by Zener–Hillert. The veloc-
ity profile has a lower slope up to supersaturation 0.83 and a steeper slope is seen
thereafter. There is a transition in slope of growth velocity with supersaturation.
FIG. 11 Contour plot of u ¼ 0.5 showing interface position at different instant. For
Zirconium with less than 200 ppm Nb, lath morphology does not develop and
initial protrusions decay and a planar interface formation favoured with very
high value of c (120).
The increase in tip velocity with supersaturation signifies that solute can easily be
removed from the tip region at higher supersaturation to favour plate growth.
Simulations also show that in pure Zr containing 200 ppm Nb or less, lath mor-
phology does not develop at 890 K and the initial protrusions gradually decays and
the formation of a planar interface is favoured with very high c (Fig. 11). Due to very
low concentration of solute atoms in the alloy, solute pile up at the tip of the protru-
sions as well as at the root of the protrusions is more or less the same. The tip of
the protrusion could not attain higher relative velocity with respect to planar inter-
face of pre-nucleated a and the protruded region decays and the growth of entire
planar interface takes place. This also established the fact that the influence of the
concentration field ahead of a=b interface is responsible for the growth of a group
of a lath of identical orientation giving parallel plate morphology.
Conclusions
The growth of a -phase from equilibrated nucleus has been modeled using phase
field method for Zr–Nb series alloys with different Nb content. Using these simula-
tions, the microstructure evolution during b to a diffusional phase transformation
can be predicted for Zr–Nb substitutional alloy system. The effect of nucleation
events has been incorporated through the introduction of noise term both implicitly
and explicitly. The equilibrium and kinetic data for the model input has been taken
from the literature and simulations were carried out at different isothermal temper-
atures. The anisotropic surface tension has a pronounced effect on the morphology
development apart from the mobility parameters. For Zr–2.5 %Nb with low under-
cooling, the driving force for plate growth decreases and at 1054 K it shows forma-
tion of allotriomorph a through classical diffusional transformation. Higher
undercooling favours Widmanstatten lath formation. The growth velocity of the
Widmanstatten plate is constant. The change in alloy composition changes the
effective supersaturation at the u ¼ 0:5 interface and leads to a different tip veloc-
ity. When Nb content reduces below 200 ppm, lath morphology does not develop
and the interface remains planar only.
ACKNOWLEDGMENTS
The writers are grateful to Shri S. Anantharaman, Head Quality Assurance Division
and Sri P.P Nanekar, Head NDT Validation Centre, Quality Assurance Division for
support and fruitful discussions during the course of work.
Appendix
Thermodynamic description of the Zr–Nb system [46,53]

GbZr ¼ 526:9 þ þ124:9457 T 25:607406 T logðT Þ 0:00034008415 T 2

25233
9:72897347 0:1 108 T 3 0:761428942 1010 T 4 þ
T
(A1)
GbNb ¼ 8519:35 þ 142:048 T 26:4711 T logðT Þ 0:000203475 T 2

93398:8
(A2) 3:50119 0:0000001 T 3 þ
T
Interaction parameters in b -phase
(A3) Lb0 ¼ 15911 þ 3:35 T
(A4) Lb0i ¼ 3919 1:091 T
iGbm ¼ c GbNb þ ð1 cÞ GbZr þ R T c logðcÞ þ R T ð1 cÞ logð1 cÞ

þ c ð1 cÞ ðLb0 þ Lb0i ð2 c 1ÞÞ
(A5)

34971
GaZr ¼ 7829 þ 125:649 T 24:1618 T logðT Þ 0:00437791 T 2
T
(A6)

GaNb ¼ 1480:65 þ 144:448 T 26:4711 T logðT Þ þ 0:000203475 T 2

93398:8
(A7) 3:50119 0:0000001 T 3 þ
T
Interaction parameter in a -phase
(A8) La0 ¼ 24411

iGam ¼ c GaNb þ ð1 cÞ GaZr þ R T c logðcÞ þ R T ð1 cÞ logð1 cÞ
þ c ð1 cÞ La0
(A9)
References
[1] Banerjee, S. and Mukhopadhyay, P., Diffusional Transformation, Phase Transformations:

Examples from Titanium and Zirconium Alloys, Vol. 12, Pergamon Materials Series,
Oxford, UK, 2007.
[2] Holt, R. A.,“The Beta to Alpha Phase Transformation in Zircaloy-4,” J. Nucl. Mater., Vol.
35, 1970, pp. 322–334.
[3] Perovic, V., Perovic, A., Weatherly, G. C., Brown, L. M., Purdy, G. R., Fleck, R. G., and Holt
R. A.,“Microstructural and Micro-Chemical Studies of Zr–2.5Nb Pressure Tube Alloy,”
[4] IAEA, “Waterside Corrosion of Zirconium Alloys in Nuclear Power Plants,” IAEA TECH-
DOC-996, IAEA, Vienna, 1998, pp. 124–169.
[5] Dey, G. K. and Banerjee, S., “Decomposition of the b-phase in Zr–20 %Nb,” J. Nucl.
Mater., Vol. 125, 1984, pp. 219–227.
[6] Dey, G. K., Singh, R. N., Tewari, R., Srivastava, D., and Banerjee, S.,“Meta-Stability of the
b-Phase in Zr-Rich Zr–Nb Alloys,” J. Nucl. Mater., Vol. 224, 1995, pp. 146–157.
[7] Tewari, R., Srivastava, D., Dey, G. K., Chakravarty, J. K., and Banerjee, S.,“Microstructural
Evolution in Zirconium Based Alloys,” J. Nucl. Mater., Vol. 383, Nos. 1–2, 2008, pp. 153–171.
[8] Zhang, W. Z., Perovic, V., Perovic, A., Weatherly, G. C., and Purdy, G. R.,“The Structure of
H.C.P.- B.C.C. Interfaces in a Zr–Nb Alloy,” Acta Mater., Vol. 46, No. 10, 1998, pp.
3443–3453.
[9] Misra P., Jathar, P. P., Srivastava, D., and Dey, G. K., unpublished work.
[10] Saibaba, N., Jha, S. K., Tonpe, S., Vaibhaw, K., Deshmukh, V., Ramana Rao, S. V., Mani
Krishna, K. V., Neogy, S., Srivastava, D., Dey, G. K., Kulkarni, R. V., Rath, B. B., Ramadasan,
E., and Anantharaman, S. A., “Microstructural Studies of Heat Treated Zr-2.5Nb Alloy for
Pressure Tube,” Proceedings of 16th International Symposium on Zirconium in the Nu-
clear Industry, ASTM STP 1529 M. Limbäck and P. Barbéris, Eds., ASTM International,
West Conshohocken, PA, 2010, pp. 349–360.
[11] Langer, J. S., “Instabilities and Pattern Formation in Crystal Growth,” Rev. Mod. Phys.,
Vol. 52, 1980, pp. 1–28.
[12] Pelcé, P. and Libchaber, A., Dynamics of Curved Fronts, Academic Press, New York, 1988.
[13] Mullins, W. W. and Sekerka, R. F., “Stability of a Planar Interface During Solidification of a
Dilute Binary Alloy,” J. Appl. Phys., Vol. 35, 1964, pp. 444–451.
[14] Mullins, W. W. and Sekerka, R. F., “Morphological Stability of a Particle Growing by Diffu-
sion or Heat Flow,” J. Appl. Phys., Vol. 34, 1963, pp. 323–329.
[15] Saarloos Wim, V., “Three Basic Issues Concerning Interface Dynamics in Non-
Equilibrium Pattern Formation,” Phys. Rep., Vol. 301, 1998, pp. 9–43.
[16] Aaronson, H. I., Spanos, G., Masamura, R. A., Vardiman, R. G., Moon, D. W., Menon, E. S.
K., and Hall, M. G., “Sympathetic Nucleation: An Overview,” Mater. Sci. Eng. B, Vol. 32,
No. 3, 1995, pp. 107–123.
[17] Shewmon, P. G., “Interfacial Stability in Solid–Solid Transformations,” Trans. TMS-AIME,

Vol. 233, 1965, pp. 736–748.
[18] Phelan, D., Stanford, N., and Dippenaar, R., “In-Situ Observations of Widmanstätten Ferrite
Formation in a Low Carbon Steel,” Mater. Sci. Eng. A, Vol. 407, Nos. 1–2, 2005, pp. 127–134.
[19] Chen, L. Q., “Phase-Field Models for Microstructure Evolution,” Ann. Rev. Mater. Res.,
Vol. 32, 2002, pp. 113–140.
[20] Wang, Y., Ma, N., Chen, Q., Zhang, F., Chen, S. L., and Chang, Y. A., “Predicting Phase
Equilibrium, Phase Transformation and Microstructure Evolution in Titanium Alloys,”
JOM, Vol. 57, 2005, pp. 32–39.
[21] Loginova, I., Ågren, J., and Amberg, G., “On the Formation of Widmanstatten Ferrite in
Binary Fe–C–Phase Field Approach,” Acta Mater., Vol. 52, No. 13, 2004, pp. 4055–4063.
[22] Dunlop, J. W. C., Bréchet, Y. J. M., Legras, L., and Zurob, H. S., “Modeling Isothermal and
Non-Isothermal Recrystallization Kinetics: Application to Zircaloy 4,” J. Nucl. Mater., Vol.
366, 2007, pp. 178–186.
[23] Massih, A. R. and Jernkvist, L. O., “Nucleation and Growth of Second-Phase Precipitates
Under Quenching and Annealing,” Comput. Mater. Sci., Vol. 39, 2007, pp. 349–358.
[24] Bin, T., Cui, Y. W., Chang, H., Hongchao, H., Li, J., and Zhou, L., “A Phase-Field Approach
to Athermal b to x Transformation,” Comput. Mater. Sci., Vol. 53, 2012, pp. 187–193.
[25] Choudhuri, G., Chakraborty, S., Srivastava, D., and Dey, G. K., “Phase Field Modeling of Wid-
manstatten Plate Formation in Zr–2.5Nb Material,” Results in Phys., Vol. 3, 2013, pp. 7–13.
[26] Karma, A. and Rappel, W. J., “Phase-Field Model of Dendritic Side Branching With Ther-
mal Noise,” Phys. Rev. E, Vol. 60, 1999, pp. 3614–3625.
[27] Elder, K. R., Drolet, F., Kosterlitz, J. M., and Grant, M., “Stochastic Eutectic Growth,” Phys.
Rev. Lett., Vol. 72, 1994, pp. 677–680.
[28] Pavlik, S. G. and Sekerka, R. F., “Forces Due to Fluctuations in the Anisotropic Phase-
Field Model of Solidification,” Phys. A, Vol. 268, 1999, pp. 283–290.
[29] Pavlik, S. G. and Sekerka, R. F., “Fluctuations in the Phase Field Model of Solidification,”
Phys. A, Vol. 277, 2000, pp. 415–431.
[30] Gránásy, L., Börzsönyi, T., and Pusztai, T., “Crystal Nucleation and Growth in Binary
Phase-Field Theory,” J. Cryst. Growth, Vols. 237–239, 2002, pp. 1813–1817.
[31] Jou, H. J. and Lusk, M. T., “Comparison of Johnson-Mehl-Avrami-Kologoromov Kinetics

With a Phase-Field Model for Microstructural Evolution Driven by Substructure Energy,”
Phys Rev. B, Vol. 55, No. 13, 1997, pp. 8114–8121.
[32] Castro, M., “Phase-Field Approach to Heterogeneous Nucleation,” Phys. Rev. B, Vol. 67,
2003, 035412.
[33] Li, J., Wang, J., and Yang, G., “Phase-Field Simulation of Microstructure Development
Involving Nucleation and Crystallographic Orientations in Alloy Solidification,” J. Cryst.
Growth, Vol. 309, 2007, pp. 65–69.
[34] Allen, S. M. and Cahn, J. W., “A Microscopic Theory for Anti Phase Boundary Motion and its
Application to Anti Phase Domain Coarsening,” Acta Metall., Vol. 27, 1979, pp. 1085–1095.
[35] Cahn, J. W. and Hilliard, J. E., “Free Energy of a Non Uniform System. I Interfacial Free
Energy,” J. Chem. Phys., Vol. 28, 1958, pp. 258–267.
[36] Wang, S. L., Sekerka, R. F., Wheeler, A. A., Murray, B. T., Coriell, S. R., Braun, R. J., and
McFadden, G. B., “Thermodynamically-Consistent Phase-Field Models for Solidification,”
Phys. D., Vol. 69, 1993, pp. 189–200.
[37] Provatas, N. and Elder, K., Phase-Field Methods in Material Science and Engineering,
Wiley-VCH Verlag GmbH & Co KGaA, Singapore, 2010.
[38] McFadden, G. B., Wheeler, A. A., Braun, R. J., Coriell, S. R., and Sekerka, R. F., “Phase-
Field Models for Anisotropic Interfaces,” Phys. Rev. E, Vol. 48, No. 3, 1993, pp.
2016–2024.
[39] Gunton, J. D., Miguel, M. S., and Sahni, P. S., “The Dynamics of First Order Transitions,”
Phase Transitions and Critical Phenomena, Vol. 8, C. Domb and J. L. Lebowitz, Eds., Aca-
demic Press New York, 1983.
[40] Singer, H. M., Singer, I., and Jacot, A., “Phase-Field Simulations of a to c Precipitations
and Transition to Massive Transformation in the Ti–Al Alloy, Acta Mater., Vol. 57, 2009,
pp. 116–124.
[41] Ågren, J., “Diffusion in Phase With Several Components and Sublattices,” J. Phys. Chem.
Solids, Vol. 43, No. 5, 1982, pp. 421–430.
[42] Dyment, F. and Libanati, C. M., “Self-Diffusion of Ti, Zr, and Hf in their HCP Phases and
Diffusion of Nb95 in HCP Zr,” J. Mater. Sci., Vol. 3, 1968, pp. 349–359.
[43] Federer, J. I. and Lundy, T. S., “Diffusion of Zr 95 and Cb 95 in BCC Zirconium,” Trans. Met.
Soc, AIME, Vol. 227, 1963, pp. 592–597.
[44] Simmons, J. P., Wen, Y., Shen, C., and Wang, Y. Z., “Microstructural Development Involv-
ing Nucleation and Growth Phenomena Simulated With the Phase Field Method,” Mater.
Sci. Eng. A, Vol. 365, 2004, pp. 136–143.
[45] Aaronson, H. I. and Lee, J. K., Lectures on the Theory of Phase Transformations, H. I. Aar-
onson, Ed., TMS, New York, 1975.
[46] Guillermet, A. F., “Thermodynamic Analysis of the Stable Phases in the Zr–Nb System
and Calculation of the Phase Diagram,” Z. Metalkundu, Vol. 48, 1991, pp. 478–487.
[47] Zener, C., “Kinetics of the Decomposition of Austenite,” Trans. Am. Inst. Mining Metall.
Eng., Vol. 167, 1946, pp. 550–595.
[48] Hillert, M., “Diffusion and Interface Control of Reactions in Alloys,” Metall. Trans. A, Vol.
6, 1975, pp. 5–19.
[49] Doherty, R. D., “Diffusive Phase Transformations in Solid State,” Physical Metallurgy, Vol.
2, 4th Revised and Enhanced ed., R. W. Cahn and P. Haasen, Eds., Elsevier Science, Am-
sterdam, The Netherlands, 1996.
[50] Livingstone, J. D. and Cahn, J. W., “Discontinuous Coarsening Of Aligned Eutectoids,”

Acta Metall., Vol. 22, No. 4, 1974, pp. 495–503.
[51] Aaronson, H. I., Enomoto, M., and Lee, J. K., Mechanisms of Diffusional Phase Transfor-
mation in Metals and Alloys, CRC Press, Boca Raton, FL, 2010.
[52] Hillert, M., Höglund, L., and Ågren J., “Diffusion-Controlled Lengthening of Widmanstat-
ten Plates,” Acta Mater., Vol. 51, 2003, pp. 2089–2095.
[53] Dinsdale, A. T., “SGTE Data for Pure Elements,” CALPHAD, Vol. 15, No. 4, 1991, pp.
317–425.
DISCUSSION
Question from Antoine Ambard, EDF R&D:—It seems that your modeling pre-
dicts Nb concentration between the Widmanstätten laths. Do you observe such
over-concentrations experimentally by TEM or any other technique?
Authors’ Response:—In dilute Zr-Nb alloy, Nb concentration between the Wid-

manstätten laths formed during air cooling was experimentally measured by EDS
[1] and was around 20 wt% Nb (b-Zr enrich with Nb). The Nb concentration in the
a-Zr matrix was lowered to the equilibrium concentration due to the slow cooling
rate. The authors model predicts microstructure evolution during b to a phase
transformation during isothermal holding. Local equilibrium was assumed to be
maintained at the a/b interface. Depending on kinetic parameters (temperature,
time, diffusivity), Nb concentration will be determined between the alaths as well as
within the lath. The values predicted at different temperature is surely not over-esti-
mated but in line with the phase diagram and supported by literature.
Ref.1. Yong Hwan Jeong, Kyoung Ok Lee, Hyun Gil Kim, Journal of Nuclear
Materials 302 (2002) 9–19.
Question from Hans-Olof Andrén, Chalmers University of Tech:—Based on

TEM micrographs, the authors suggest that the b/a phase boundary is around 5 nm
wide. Does this mean that there is an amorphous layer between the phases? I sug-
gest that in reality phase boundaries are sharp and that what is observed by TEM is
a tilted boundary with overlapping a and b grains, as evidenced by the Moiré con-
trast. Each grain is either a or b, nothing in-between.
Authors’ Response:—The authors agree. In reality, the width of incoherent

phase boundaries is less than 1.5 nm. The model parameter d0, which is the maxi-
mum thickness of an incoherent boundary, should be between 0.5–1.5 nm. The
coherent interface is much thinner than the incoherent interface. For a perfect
coherent interface, the thickness is zero. In the model, anisotropy in the thickness
and interfacial energy is introduced by a cosine function of h and then the phase
field equations are solved by finite element method using an adaptive unstructured
grid. The phase field equation shows a very steep gradient at the interface region
and with a realistic interface thickness, the task of solving it becomes challenging.
Due to the requirement of huge computational power, we have taken the incoherent
interface thickness d0 as 5 nm. In non-dimensional space, the interface regions have
been discretized using five grid points and the rest of the region with large triangles
optimizing the computational load and accuracy of the solution.
Ref. 2. W. Z. Zhang, V. Perovic, A. Perovic, G. C. Weatherly and G. R. Purdy,

“The structure of h.c.p.- b.c.c. interfaces in a Zr-Nb alloy”, Acta mater.
46(10) (1998) 3443–3453.
Question from K. Kapoor, NFC:—The anisotropy in the interface energy is also

related to crystallographic planes. Are the resulting growth rates in Zr alloy phases
related to the crystallographic planes? Is it possible to model them?
Authors’ Response:—The anisotropy in the interfacial energy is related to the

crystallographic planes. Hence growth rate in Zr alloy phases should be influenced
depending on extent of texture (degree of anisotropy). To model this phenomenon,
anisotropy in interfacial energy should be expressed using the proper function. This
interfacial energy and its orientation dependence in a phase-field model are
expressed via gradient energy coefficient (e). Using this coefficient, a phase field
model needs to be developed which can predict the rate of crystal growth as a func-
tion of crystallographic planes. Ref. 3 described a three-dimensional phase-field
model simulation using the interfacial energy expression and found convincing hcp
crystal morphologies consistent with evolution towards an equilibrium shape calcu-
lated on the basis of interfacial energy minimization.
Ref. 3: R. S. Qin, H. K. D. H. Bhadeshia, “Phase-field model study of the crystal

morphological evolution of hcp metals”, Acta Materialia 57 (2009) pp.
3382–3390.
P. Barberis,1 C. Vauglin,2 P. Fremiot,1 and P. Guerin1
Thermodynamics of Zr Alloys:
Application to Heterogeneous
Materials
Reference
Barberis, P., Vauglin, C., Fremiot, P., and Guerin, P., “Thermodynamics of Zr Alloys: Application
to Heterogeneous Materials,” Zirconium in the Nuclear Industry: 17th International Symposium,
STP 1543, Robert Comstock and Pierre Barberis, Eds., pp. 118–137, doi:10.1520/
STP154320120208, ASTM International, West Conshohocken, PA 2015.3
ABSTRACT
Thermodynamic databases are available now for zirconium alloys, and can be
used to design new alloys, master microstructure, or understand phenomena
occurring during the processing or the in-pile life. In the nuclear industry,
several components are not constituted of one homogeneous material, but in
some cases of several alloys intimately bonded (liner and duplex cladding, and
mixed welds, for example), or a unique alloy but with a metallurgical state
gradient (homogeneous welds). Thermodynamics computations were
performed, and bring new insights to some observed features. A
heterogeneous distribution of hydrogen is observed after autoclave corrosion
tests or in-pile service life in duplex cladding for pressurized-water reactors
(PWRs) or liner cladding for boiling-water reactors (BWRs). A heterogeneous
distribution is also observed in mixed welds. Whereas the usual interpretation
invokes some residual stresses at the bounding area, thermodynamic
computations clearly show that the hydrogen distribution is related to a
difference in the hydrogen chemical potential from one alloy to another. In the
same way, uphill diffusion of iron can be observed locally at the interface
between Zircaloy-2 and the Zr liner. It can be explained by the iron chemical
potential difference between both alloys. Last, a miscibility gap of the
Manuscript received December 18, 2012; accepted for publication August 1, 2013; published online June 17,
2014.
1
AREVA/CEZUS Research Center, Ave. Paul Girod, 73403 Ugine Cedex, France.
2
AREVA/Fuel Design 10 rue Juliette Récamier, 69456 Lyon, France.
3
Hyderabad, India.
Copyright V
BARBERIS ET AL., DOI 10.1520/STP154320120208 119
Zr(Nb,Fe,Cr)2 Laves phase has been observed and computed for some
compositions in ZrNbSnFeCr alloys.
Keywords
thermodynamics, chemical potential, hydrogen, iron, weld, liner, Laves phase,
miscibility gap, up-hill diffusion
Introduction
Two keys for understanding microstructural features of materials (specifically here,
zirconium alloys) either during processing or during in-pile service are thermody-
namics and kinetics. Some features that remained unexplained are the hydrogen (or
hydride) distribution in heterogeneous materials like liner or duplex cladding after
in-pile service. Typically, in addition to a gradient through the cladding thickness
toward the external interface, observed also on massive cladding, and explained by
the propensity of hydrogen to diffuse toward the cold surfaces, liner cladding shows
an accumulation of hydrides at the interface, on the inner side [1]. It can be
observed after irradiation, but also after autoclave test. This accumulation at the
interface has been tentatively explained by internal stresses though no stress mea-
surement is available, or more convincingly by a lower solubility of H in Zr than in
Zy2 [1].
On some welds after corrosion in autoclave, an uneven distribution of hydrides
has been observed Fig. 1. It may be attributed to difference in hydrogen pickup.
However, welds can also be considered as heterogeneous materials, even if the alloy
composition does not change, but because the metallurgical state and thus the
repartition of addition elements between the solid solution and the second phase
particles (SPP) may vary. Some welds are also made between different alloys, for
example between Zircaloy-4 and M5.4
In these cases, the diffusion of hydrogen is not primarily controlled by its con-
centration gradient, but by its chemical potential gradient. The chemical potentials
can nowadays be easily computed thanks to thermodynamic databases and appro-
priate software, using CALPHAD methodology [2].
We will show in this paper that in both cases (liner or duplex materials and
welds), the distribution of hydrogen obeys to the hydrogen chemical potential vari-
ation in the various materials.
In the same way, in duplex or liner cladding, the distribution of iron and nickel
in the vicinity of the interface is also governed by the difference in the iron or nickel
chemical potential between the two alloys: after extrusion, some uphill diffusion of
iron has been observed, that is an enrichment of iron at the interface, the iron com-
ing from the liner with an iron concentration lower than the Zircaloy-2 matrix.
4
M5 is a trademark of AREVA NP and is registered in the United States and in other countries.
FIG. 1 Micrograph of irradiated mixed resistance weld [4]: Zy4 plug (on the left), ZrNb1
cladding (on the right).
Last, the thermodynamic database has been used to confirm some unexplained
observations made in ZrNbSnFeCr alloys. These alloys are now investigated by sev-
eral authors because of their good corrosion and mechanical properties out of pile.
It has been reported that in some samples, two Laves phases with the same crystal-
lography (hexagonal C14 phase) but differing compositions were observed [3]. We
extended this observation on several alloys, and confirmed by thermodynamic com-
putations that a miscibility gap exists in this phase.
Materials and Computations

MATERIALS
The studied materials are well known, and their chemical composition is given
below (Table 1). Impurities are not given because they are not relevant here, and can-
not, generally, be taken into account in thermodynamic computations.
REMINDER ABOUT THERMODYNAMICS

As a reminder, the chemical potential of a species i (here for example hydrogen) in
a phase is defined as the derivative of the free energy G:
TABLE 1 Nominal alloy compositions (in wt. %).
Nb Sn Fe Cr Ni O
Zircaloy-2 1.5 0.13

Zircaloy-4 1.3 0.2 0.1 0.13
Zr-2.5Nb 2.5 0.06
Liner 0.01–0.4 0.06
M5 1 0.035 0.14
@G
(1) l¼
@ni
where:
ni ¼ the fraction of species i in the considered phase.
For dilute solutions, it can be expressed as
@G
(2) l¼ ¼ l0 þ RT ln cci
@ni
where:
l0 ¼ the reference chemical potential,
c ¼ the activity coefficient, and
ci ¼ the concentration of element i.
In the case of ideal solution, the activity is equal to 1. This assumption is often
made, specifically because the actual activity coefficient—and its dependence on
temperature and composition—is unknown. However, thermodynamic databases
allow to compute directly the chemical potential (or the activity) at a given temper-
ature and composition.
The chemical potential is the driving force for all phenomena dealing with ther-
modynamic evolution toward equilibrium. For example, as will be explained below,
the diffusion of a species always follows the downward slope of the chemical poten-
tial gradient. The general diffusion equation (without however thermal or electrical
gradient)—known also as second Fick’s law—is the following:
dci
(3) þ divðLgradlÞ ¼ 0
dt
with L a coefficient linked to the diffusion coefficient.

If there are in addition a thermal gradient and a stress gradient, corresponding
terms must be added under the divergence operator of the previous equation.
Note that for two different phases—or alloys—containing the same concen-
tration of a species i, the chemical potential can differ because the l0 can be dif-
ferent. There can thus be some diffusion of the species to equilibrate the
chemical potential—it is the well known uphill diffusion. In other terms, the evo-
lution of a system through an interface will tend to equalize (and lower) the
chemical potential—thus making it continuous across the interface, whereas the
concentration will be discontinuous. It will be shown in below that
l(Fe)liner > l(Fe)Zy2, which is equivalent to c(Fe)liner < c(Fe)Zy2.5 From Darken’s
theory [5], it is inferred that iron will migrate from the high potential (liner) to
the low potential (Zircaloy-2).
Last, it should be reminded that when two phases containing the same species
coexist in a system (for example hydrogen in solid solution in alpha Zr and
5
In this case, Eq 2 cannot be used because iron is not in solid solution. But the principle is the same.
hydrides), if the phases are at equilibrium, then the chemical potential of the species
(here the hydrogen) is equal in both phases.
SOFTWARE AND DATABASE

The TCC-S version of the Thermo-Calc software has been used with a database
that was slightly modified from the ZIRCOBASE [6,7]. Modifications mainly dealt
with the so-called end-members some of which had only default values in the pub-
lished database, and have been attributed more realistic values. If it improves some-
what some computations, it does not change the conclusions here.
The thermodynamic database was confronted to experimental results (observed
phases and phase diagrams involving Zr-Nb-Sn-Fe-Ni-Cr) in several earlier papers
[7,8]. Its validity as regards the hydrogen has been checked by computing the solu-
bility limit of hydrogen in several alloys, namely Zircaloy-4, Zircaloy-2, Zr-2.5Nb
and pure zirconium, and comparing it with the experimental data available in the
literature. The comparison is presented in Fig. 2. The hydrogen solubility in pure
zirconium is in excellent agreement with the literature data (see Ref 6). The hydro-
gen solubility in the tested alloys is very close to the solubility in pure zirconium; it
is slightly lower in Zircaloy-4 and Zr-2.5Nb. M5 is very close to pure zirconium.
The comparison with literature [9,10] shows the same trends, though computed
solubility seems less dependent on the alloy composition than the experimental
results.
Results
The evolution of the hydrogen chemical potential with the hydrogen content in two
alloys has been computed and is presented in Fig. 3. For low hydrogen content, the
potential increases as the logarithm of the concentration, until the solubility is
reached (the alloy then contains hydrides), and the potential becomes constant.
This will be of interest in the following section.
FIG. 2 Comparison of computed (left) and literature [9,10] (right) solubility of

hydrogen for various alloys.
FIG. 3 Computed chemical potential of hydrogen at 300 C as a function of the

hydrogen content for two alloys.
HYDROGEN DISTRIBUTION IN DUPLEX AND LINER CLADDINGS

Alloys of the same type will be considered here for sake of simplicity: indeed, duplex
or liner claddings are often constituted of Zircaloy-type material, with some varia-
tions for the liner or for the outer material in tin or iron, promoting lower tough-
ness (liner) or higher corrosion resistance (duplex).
The hydrogen chemical potential as a function of tin content has been com-
puted at 300 C in ZrSn0.2Fe0.1Cr0.13O and in Zr1NbSn0.1Fe0.14O alloys, for two
hydrogen contents: 10 wt. ppm (below the solubility limit) and 200 ppm (above the
solubility limit). Results are plotted in Fig. 4.
It is seen that the hydrogen chemical potential increases with the tin content in
a given alloy, and is higher for Nb-containing alloys than for ZrSn type alloys (for a
given Sn content). In other terms, in case of a heterogeneous junction, the hydrogen
FIG. 4 Hydrogen chemical potential at 300 C as a function of tin content in

ZrSn0.2Fe0.1Cr0.13O and in Zr1NbSn0.1Fe0.14O alloys, for H ¼ 10 ppm (left) and
H ¼ 200 ppm (right).
FIG. 5 Hydrogen chemical potential at 300 C as a function of tin content in ZrNb0.1Fe,

for H ¼ 10 ppm.
will go toward the lower tin containing material, and toward the non-niobium con-
taining alloy. The increase of the hydrogen chemical potential with the niobium
content is also visible in Fig. 5.
For a liner material in the absence of temperature gradient, the hydrogen will
migrate from the Zircaloy-2 or -4 toward the Zr(Fe) material. Whereas the solubil-
ity limit is not reached in the liner, the difference in equilibrium concentration can
be computed from the chemical potential:
(4) lðZy4ÞlðZrÞ ¼ 94100 ð94400Þ ¼ 300 ¼ RTlnðcZr=cZy4Þ
Hence,
(5) cZr ¼ cZy4 1:06
The difference is negligible.

However, when the solubility limit is reached in the Zr liner, the chemical
potential cannot be increased anymore until there is some remaining aZr, or
whereas the Zr is not totally hydrided. Thus hydrogen will continue to migrate to-
ward the Zr liner. The hydrides will be positioned on the liner side.
This explains also the hydride distribution in some duplex cladding after auto-
clave testing (Fig. 6). This cladding is constituted of two alloys differing only by the
tin content: the inner material is standard Zircaloy-4, the outer material is Zircaloy-
4 with 0.5 %Sn (to get a better corrosion resistance). After a 400 C steam corrosion
test, the hydride distribution is heterogeneous, with the hydrides located preferen-
tially in the low tin material, near the interface with the inner Zircaloy-4. A
hydrogen-depleted zone is also visible in the Zircaloy-4 toward the interface.
Because of the difference in chemical potential, there should be only a slightly
FIG. 6 Hydrogen distribution in a Zircaloy-4/Zr0.5Sn0.2Fe0.1Cr0.13O duplex after an

autoclave corrosion test in steam (140 days, 400 C, 10.5 MPa).
higher content of hydrogen in the low tin material during the corrosion test (the
equilibrium can be assumed, considering the high diffusion rate of hydrogen).
Upon cooling, the hydrides start to precipitate earlier in the low tin material, and
the hydrogen migrates from the Zircaloy-4 to the low tin alloy.
In more details, the process is the following for liner cladding (the low tin liner
material being at the inner side of the cladding), and assuming that the corrosion
occurs only from the outer surface:
• At the beginning, the hydrogen is picked up, and goes in solid solution;
whereas there are no hydrides, the hydrogen concentration in the liner is only
slightly higher than in the Zircaloy-2.
• At some concentration, as the solubility is lower in the liner, hydrides start to
precipitate in the liner, whereas hydrogen is still in solid solution in the Zirca-
loy. As the hydrogen chemical potential, fixed by the hydrides in the liner, is
lower in the liner than in the Zircaloy, it cannot increase and all the hydrides
will form in the liner.
• If the solubility is not reached during the test, then upon cooling, the precipi-
tation will start in the low tin layer—because of the lower chemical potential,
and hydrogen will migrate toward this layer. Depending on the cooling rate,
more or less hydrogen will be able to migrate to the low tin layer, and there-
fore the actual distribution will be changed.
A modeling of this phenomenon, taking into account the difference in solubil-
ity has been derived by Takagi [11].
It is worth noting that in Zr-1Nb alloys at 300 C, the hydrogen goes in aZr and
forms hydrides; but the bNb phase remains free from hydrogen, though hydrogen
is known as a b stabilizer. Again, it comes from the higher chemical potential of
hydrogen in bNb compared to hydrides.
FIG. 7 Hydrogen chemical potential at 300 C as a function of tin content in

ZrSn0.2Fe0.1Cr0.13O for H ¼ 10 ppm in an hyper-quenched state (no Laves
phases are present) and at equilibrium.
HYDROGEN DISTRIBUTION IN WELDS

In a weld, because of the fast quench from high temperature (beta phase or liquid
phase), the material is not at equilibrium, which means that the chemical potential
is higher than at equilibrium. Thus, for a homogeneous weld, hydrogen will diffuse
from the weld to the base metal. This can be seen in Fig. 7, where are computed the
chemical potentials of hydrogen as a function of the tin content in two metallurgical
states: the equilibrium (as in Fig. 4), and in a hyper-quenched state where the iron
and chromium are totally in supersaturated solid solution (they do not form Laves
phases). This last metallurgical state simulates a very high cooling rate in a weld.
This difference in chemical potential between the two metallurgical states is inde-
pendent of the tin content.
FIG. 8 Hydride distribution after 200 days autoclave corrosion in 400 C steam in
Zircaloy-4/M5 heterogeneous resistance welds.
TABLE 2 Heterogeneous weld couples and observed hydride location along the interface.
Clad Tube End Plug Hydrides Nb (%) Sn (%) Hydrides Nb (%) Sn (%)
Location Not In
M5 Zy4 M5 1 0 Zy4 0 1.4

Q Zy4 Q 1 0.3 Zy4 0 1.4
Q M5 M5 1 0 Q 1 0.3
Zy4 M5 M5 1 0 Zy4 0 1.4
Note: Q represents ZrNbSnFe alloys.
For an heterogeneous weld, the difference in chemical composition must be

taken into account. For example, the hydrogen will diffuse from the Zircaloy-4 to a
lower tin-containing alloy. Fig. 8 shows heterogeneous resistance welds corroded in
400 C steam autoclave for 200 days. Along the weld line, hydrides are precipitated
on the M5 side. On the Zircaloy-4 side, a depleted zone in the vicinity of the weld
line is seen. Both features indicate that the hydrogen migrated from the Zircaloy-4
to the M5. In the M5 cladding (Fig. 8, left), another line of precipitated hydrides,
more or less parallel to the weld line is seen. It could be indicative of the limit of the
heat affected zone, where bZr phase or higher niobium in solid solution is present,
rising the hydrogen chemical potential.
This phenomenon has been observed on several other alloy couples, as reported
in Table 2. This table shows that hydrides are present where the Sn content is the
lowest, whatever the Nb content.
IRON AND NICKEL DISTRIBUTION IN LINER CLADDINGS

The liner cladding processing includes a coextrusion of liner tube inserted in Zy2
matrix tube. This operation is done at about 600 C. After extrusion, the material is
cold pilgered and annealed to promote recrystallization (RXA TREX).
Optical micrographs performed on these RXA TREX show at the interface
some large precipitates on the liner side, with a 10 -lm-thick zone depleted of SPPs
FIG. 9 Optical micrograph of the Zircaloy-2/liner interface on a TREX sample.

FIG. 10 FEG-SEM observation and chemical analysis of an SPP at the Zircaloy-2/liner

interface.
still on the liner side (Fig. 9). Numerous small precipitates are visible in the
Zircaloy-2, whereas the grains are larger and more equiaxed in the liner, owing to
the faster recrystallisation rate of low tin-containing materials. EDX analysis done
on a FEG-SEM on the interface big particles show they contain iron and nickel; in
addition to zirconium (Fig. 10). TEM work shows they have an orthorhombic crystal
structure, and that their composition is Zr3(Fe,Ni) with Ni/Fe ¼ 0.35–0.4. EDX
analysis of smaller SPP in the liner further from the interface shows that there is
still nickel in the SPPs, but with a Ni/Fe ratio decreasing to less than 0.1 at about
70 lm from the interface (Fig. 11).
Table 3 shows the results of thermodynamic computations. The phases in the
Zircaloy-2 matrix and in the liner material are as expected
aZr þ Zr(Fe,Cr)2 þ Zr2(Fe,Ni), and aZr þ Zr3Fe, respectively. The chemical
potential of the nickel (at 600 C) is, as expected, much lower in the liner than
in the Zircaloy-2. Nickel will thus migrate from the Zircaloy-2 toward the liner.
FIG. 11 Analysis of SPP as a function of the distance from the interface. Position of the
analyzed particles (left), Ni/Fe ratio (right).
The chemical potential of the iron is independent of the liner iron content
(because Fe-containing SPP are present), and is lower than in the Zircaloy-2,
though the iron content in Zircaloy-2 is higher. Hence, the iron will migrate
from the liner toward the Zircaloy-2, which is observed: it is the so-called
uphill diffusion.
ZR(NB, FE, CR)2 LAVES PHASE MISCIBILITY GAP

The thermodynamic database was also used to explain some observations reported
in the literature, or observed in the lab on some ZrNbFeCr alloys. ZrNbSnFeCr(Ni)
alloys were melted at a lab scale in a levitation cold furnace, and processed into 1.5-
mm sheets by hot and cold rolling, with intermediate annealing at low temperature
(<600 C). The final heat treatment lasted 1 h at higher temperature (see Table 4).
SPP composition was determined on extractive replicas, the composition given in
TABLE 3 Computed chemical potentials at 600 C of Ni and Fe in some alloys.
Ni Content Fe Content Ni Chemical Fe Chemical

Equilibrium in the Alloy in the Alloy Potential Potential
Alloy Phases (ppm) (ppm) (kJ/mol) (kJ/mol)
Zy2 aZr þ Zr(Cr,Fe)2 649 1750 158 87.7

þ Zr2(Fe,Ni)
Liner aZr þ Zr3Fe 50 500 312 78.9
(500 ppm)
Liner aZr þ Zr3Fe 50 1000 315 78.9
(1250 ppm)
Liner aZr þ Zr3Fe 50 4000 322 78.9
(4000 ppm)
Zr3Fe particles Zr3Fe 0 16 wt. % 367 79.9
TABLE 4 Composition of the C14 hexagonal Laves phases obtained experimentally or by

computation.
Annealing
Temperature Phases Zr Nb Fe Cr
Zr1Nb0.5 650 C Zr (Nb,Fe,Cr)2 33.7 27.2 23.1 16.0

Sn0.35Fe Zr (Cr,Fe)2 33.1 5.0 29.3 32.6
0.25Cr
0.07Ni
Zr0.7Nb0.3 675 C Zr (Nb,Fe,Cr)2 33.8 20.2 29.2 16.8
Sn0.35Fe Zr (Cr,Fe)2 33.0 4.6 29.5 32.9
0.25Cr 700 C Zr (Nb,Fe,Cr)2 33.3 16.2 30.8 19.7
Zr (Cr,Fe)2 32.3 4.9 27.8 35.1
Zr1Nb0.25 675 C Zr (Nb,Fe,Cr)2 37.4 24.7 33.7 4.2
Fe0.05Cr Zr (Cr,Fe)2 33.6 7.9 40.0 18.5
(computation)
FIG. 12 Computed phase diagram section at 675 C of the Zr1NbFeCr system. The
stable phases are aZr and bZr in addition to the phases labelled in the figure.
C14-1 represents the C14 phase with low Nb content, C14-2 with high Nb
content. C15 is the cubic Laves phase.
the table being the mean of 10 particles. Zr(Nb,Fe,Cr)2 C14 Laves phases were
observed with two different compositions: one rich in niobium, the other poor in
niobium (Table 4). Such results have been obtained for example in X5A [3].
The question that arose was: are they both equilibrium phases?
A thermodynamic computation was performed, and resulted in only one C14
Laves phase being stable. However, a miscibility gap is predicted by the computa-
tion for slightly different compositions. An isothermal section of the phase diagram
(at 675 C) of the pseudo-ternary Zr1NbFeCr system is presented in Fig. 12. The sta-
ble phases are aZr and bZr in addition to the phases labeled in the figure. C14-1
represents the C14 phase with low Nb content, C14-2 with high Nb content. C15 is
the cubic Laves phase. It shows a domain where two C14 Laves phases coexist. In
the miscibility gap, for a composition Zr1Nb0.25Fe0.05Cr, the composition of both
Laves phases is given in Table 4. They are not very far from the experimental com-
position, except for the Cr level.
Some slight evolution of the database will be needed to get closer values. It
should be reminded that the database was set mainly with binary and ternary sys-
tems, and that this is the first time that quaternary ZrNbFeCr systems are studied
and computed.
This miscibility gap can also be interpreted in terms of ordering of the elements
on the three sub-lattices of the C14 P63/mmc crystallographic lattice, but there was
no attempt to determine this ordering. It should be clarified in future work. The
formation of those two C14 phases may also be interpreted in terms of up-hill diffu-
sion of some elements, here mainly Cr and Nb (see for example Chapters 5 and 7 in
Ref 12).
Conclusion
From thermodynamic computations and in agreement with experimental observa-
tions, it can be concluded that:
• for a given alloy family, the hydrogen chemical potential increases with the tin
content, meaning that the hydrogen will migrate from the high toward the
low-tin-containing alloy,
• for the same reason, it will migrate in an homogeneous weld from the weld to
the base metal,
• uphill iron diffusion has been observed from liner to Zircaloy-2, and explained
by the higher chemical potential of iron in the liner, and
• two hexagonal C14 Laves phases with different composition but the same crys-
tallography can be observed as stable phase in ZrNbFeCr alloys, because of a
miscibility gap of this phase.
Thermodynamic computations are essential to understand the microstructural
features and master both the product and the process. Some slight and continuous
improvements of the database are underway. The next step is to couple these com-
putations to kinetic modeling, to predict for example the actual distribution of the
elements.
References
[1] Ogata, K., Baba, T., Kamimura, K., Matsunaga, J., Nakatsuka, M., Sakamoto, K., and
Sawada, A., “Effect of Increased Hydrogen Content on the Mechanical Performance of
Irradiated Cladding Tubes,” Top Fuel Conference, Manchester, U.K., Feb 9, 2012.
[2] Sundman, B., Jansson, B., and Andersson, J. O., “The Thermo-Calc Databank System,”
Calphad, Vol. 9, 1985, pp. 153–190.
[3] Garde, A., Comstock, R., Pan, G., Hallstadius, L., Cook, T., and Carrera, F., “Advanced Zir-
conium Alloy for PWR Application,” J. ASTM Int., Vol. 7, 2010, Paper ID JAI103030.
[4] Gaillac, A., Duthoo, D., Vauglin, C., Carcey-Collet, D., Bay, F., and Mocellin, K., “Numerical
Modeling of Fuel Rods Resistance Butt Welding,” International ASTM Symposium on Zir-
conium in the Nuclear Industry, Hyderabad, India, Feb 4–7, 2013, p. 17.
[5] Darken, L. S., “Diffusion of Carbon in Austenite with a Discontinuity in Composition,”

Trans. AIME, Vol. 180, 1949, pp. 430–438.
[6] Dupin, N., Ansara, I., Servant, C., Toffolon, C., Lemaignan, C., and Brachet, J. C., “A Ther-
modynamic Database for Zirconium Alloys,” J. Nucl. Mater., Vol. 275(3), 1999, pp.
287–295.
[7] Toffolon-Masclet, C., Brachet, J. C., Servant, C., Joubert, J. M., Barberis, P., Dupin, N., and
Zeller, P., “Contribution of Thermodynamic Calculations to Metallurgical Studies of Multi-
Component Zirconium Based Alloys,” ASTM STP 1505, ASTM International, West Con-
shohocken, PA, 2008.
[8] Barberis, P., Dupin, N., Lemaignan, C., Pasturel, A., and Grange, J., “Microstructure and
Phase Control in Zr-Fe-Cr-Ni Alloys: Thermodynamic and Kinetic Aspects,” J. ASTM Int.,
Vol. 2, 2005, Paper ID JAI12771.
[9] Setoyama, D., Matsunaga, J., Ito, M., Muta, H., Kurosaki, K., Uno, M., Yamanaka, S.,
Takeda, K., and Ishii, Y., “Influence of Additive Elements on the Terminal Solid Solubility
of Hydrogen for Zirconium Alloy,” J. Nucl. Mater., Vol. 344, 2005, pp. 291–294.
[10] Tulk, E., Kerr, M., and Daymond, M. R., “Study on the Effects of Matrix Yield Strength on
Hydride Phase Stability in Zircaloy-2 and Zr 2.5 wt. % Nb,” J. Nucl. Mater., Vol. 425, 2012,
pp. 93–104.
[11] Takagi, I., Shimada, S., Kawasaki, D., and Higashi, K., “A Simple Model for Hydrogen Re-
Distribution in Zirconium-Lined Fuel Claddings,” J. Nucl. Sci. Technol., Vol. 39(1), 2002,
pp. 71–75.
[12] Banerjee, S. and Mukhopadhyay, P., Phase Transformations—Examples from Titanium

and Zirconium Alloys, R. W. Cahn, Ed., Elsevier Science, New York, Amsterdam, 2007.
DISCUSSION
Question from Arthur Motta, Penn State University:—Could the observations of
hydrogen precipitation near welds be influenced by the Soret effect (hydrogen
migration due to temperature gradient)?
Authors’ Response:—In the micrographs shown, the welds were autoclaved, and
the cooling rate was rather low. The temperature gradient is thus low and the Soret
effect negligible. In pile, the welds are rather far from the fuel pellets and should not
be submitted to a significant thermal gradient.
In the more general case the driving force for diffusion is a function of the (gra-
dient of the) chemical potential but also a function of the temperature (Soret effect),
stresses, and other terms like electric potential or others.
Question from Srikumar Banerjee, BARC, Mumbai:—Can the miscibility gap in

C14 Laves phase [Zr-(Nb,Fe,Cr)2] be viewed as due to the general tendency of the
b phase and its ordered derivatives to exhibit a positive deviation from ideality?
Authors’ Response:—As you proposed in your book (Phase Transformations,

Volume 12: Examples from Titanium and Zirconium Alloys, Pergamon Materials
Series, 2007), it can be viewed as an ordering of the elements (here mainly Nb and
Cr) on the sublattices of the crystallographic structure of the C14 phase, which is
actually a deviation from ideality. However, we did not attempt to check that point,
and to determine the ordering on each structure. It should be very difficult to differ-
entiate Nb from Zr. However, it could be the subject of future work.
Question from Hans-Olof Andrén, Chalmers University of Tech:—It is very good

news that there now exists so good thermodynamic databases that realistic calculations
can be made, e.g., to understand the uphill diffusion between Zircaloy-2 and liner. My
question is: How far are we from being able to model diffusion? What is the situation
when it comes to mobility (multi diffusion) databases for zirconium systems?
Authors’ Response:—To my knowledge, for zirconium alloys, only diffusion coeffi-

cients for single elements in pure zirconium or a few alloys are available in the litera-
ture. They are enough to make the first diffusion computations, especially if the actual
driving force (not the concentration gradient but the chemical potential gradient) in
the Fick’s equation is taken into account. This has been done (e.g., C. Corvaĺn-Moya et
al., J. Nucl. Mat. 400 (2010) 196-204) for oxygen at high temperature.
Question from Javier Romero, Westinghouse Electric Co.:—Why do you think

there are hydride depleted regions in the duplex cladding? Can there be a Sn gradi-
ent, or is it purely migration of H to the Sn poor layer?
Authors’ Response:—As shown by the microprobe observations, the Sn gradi-

ent is very sharp at the liner / matrix interface (and it would be the same at the
duplex / matrix interface). It is due to the very low diffusion coefficient of tin in
the alpha zirconium at the processing temperatures. Thus the H distribution is
only the result of the hydrogen chemical potential gradient between the two
layers.
Questions from R. K. Chaube, NFC Hyderabad:—What is the initial value of H

present in both layers, i.e., Zircaloy-2 and liner zirconium?
Authors’ Response:—The initial hydrogen content in both materials is 10 wt

ppm.
What is the effect of residual stresses present in both layers for H pickup?
Authors’ Response:—The residual stresses are likely to act more on the hydride
morphology (orientation…) than on their distribution. However, it was not
investigated.
Question from K. Kapoor, NFC:—In case of the low Sn clad, preferential precip-
itation of hydrides at the boundary of the clad is observed. How is this explained
by the thermodynamic basis?
Authors’ Response:—As explained in the paper and in the answer to JC Brachet

(below) the precipitation occurs during the cooling. The precipitation occurs first in
the low tin material, and hydrogen from the high tin material diffuses toward the
low tin material, and precipitates at the interface. It does not diffuse further since
there is no driving force for it to diffuse within the low tin material.
Question from Dave Ludlow, AMEC:—With respect to H chemical potential

test on the duplex alloy, it was observed that there was no thermal gradient. Why?
If there was a thermal gradient, how would this affect the results?
Authors’ Response:—There was no thermal gradient because the sample was

tested in autoclave (without heat flux), and with slow cooling.
In case of thermal gradient, we must consider the total driving force, that is the
gradient of the chemical potential (which may depend also on the temperature),
and of the Soret effect (temperature gradient itself). For example, during cooling,
with a temperature gradient constant across the cladding, the global potential (or
driving force) would decrease from the inner surface to the outer surface, with a
discontinuity at the interface, and globally the hydrogen would migrate toward the
surface of the cladding.
Question from N. Ramasubramanian, ECCATEC Inc. Canada:—Regarding the

effect of tin increasing the chemical potential of hydrogen, is it an effect of tin on
the l0 in the equation for calculating chemical potential?
Authors’ Response:—Yes, tin affects the l0 (or l0 þ RTln c, depending on the

way you may write the dependence of the chemical potential as a function of the
hydrogen concentration).
Question from S. K. Jha, Nuclear Fuel Complex:—What was the grain size of
base and liner material? What will be the effect of grain size on chemical potential
of hydrogen?
Authors’ Response:—The grain size is about 5 lm in the matrix, and larger

(about 10 lm) in the liner, due to the lower tin content and possibly the lower SPP
number. There is no effect of grain size on the chemical potential.
Question from S. Anantharaman, BARC, Mumbai:—Did you consider the effect

of residual stress on the observed hydride distribution, especially at the clad-end
plug resistance weld interface? Don’t you think this could have affected the
observed distribution of hydrides in addition to the difference in the concentration
of tin across the interface?
Authors’ Response:—As in answer to R.K. Chaube above, it seems that the residual
stresses are likely to act more on the hydride morphology (orientation…) than on their
distribution. The driving force for hydrogen migration under stress is rather low.
Question from Jean-Christophe Brachet, CEA-Saclay Nuclear Materials Dept.,

France:—There is potential confusion for the first example shown: The wall clad
thickness redistribution of hydrogen in Zircaloy-2/Zr lined cladding is mainly due
to the cooling rate from the autoclave test temperature (415 C). In fact the chemical
potential has only an indirect effect; it changes the Terminal Solid Solubility (TSS)
temperature thus promoting occurrence of hydride precipitation earlier in the liner
and thus inducing a hydrogen diffusion gradient from outer Zircaloy-2 layer. See
for example the modeling of this phenomenon by Takagi et al., J. Nucl. Sci. and
Tech 39 71–75 (2002).
Authors’ Response:—I agree with this explanation, which is not incompatible

with what I am saying. As explained in the paper, if the hydrogen is totally in solid
solution during the test (i.e., if the concentration is low enough), the difference in H
chemical potential between the two layers is rather low, so that the hydrogen con-
centration is nearly constant in the cladding. Upon cooling, the precipitation will
start in the low tin layer–due to the lower chemical potential, and hydrogen will
migrate toward this layer. Depending on the cooling rate, more or less hydrogen
will be able to migrate to the low tin layer. But actually, the driving force, and even
the TSS, are only functions of the chemical potential.
Question from Suresh Yagnik, EPRI:—The hydrogen distribution you showed

for liner Zircaloy-2 clad (slide 3) and duplex clad (slide 7) is not typical of in-service
hydriding due to two-sided gaseous charging (Zircaloy-2 liner clad) and two-sided
corrosion (duplex clad) in your samples. In-pile hydriding is from the OD side
only. Similarly for the weld region, your conclusion that hydrogen moves from
weld to base metal may not be typical of in-service conditions because there may be
microstructural changes in weld region which may impact local hydrogen pick-up
plus the long range axial diffusion of hydrogen since the welds are in the colder
parts of the fuel rod cladding in a fuel rod. Thus, while thermodynamic considera-
tions (chemical potential, etc.) are important, the mechanistic aspects of hydrogen
pickup and diffusional kinetics are equally, if not more important.
Authors’ Response:—I do agree with your comment. The objective of this paper is
to assess some basic thermodynamic properties of hydrogen, namely for example the
dependence of the hydrogen chemical potential on the tin content in the alloy. This
provides some driving force for the evolution of a system. In addition to that, and still a
thermodynamic property, the dependence of the hydrogen chemical potential on the
thermal gradient and on the stresses, not addressed in this paper, is necessary when
such gradients or stresses exist. Last, the evolution of a system is not only driven by
thermodynamics, but involves kinetic aspects. As you state, for the in-pile behaviour,
all these aspects should be taken into account (and also the influence of irradiation
which could shift thermodynamic equilibria…). However, if the basic thermodynamic
(for example the influence of the tin content on the hydrogen chemical) is not accu-
rately described, then some phenomena may be misinterpreted and mispredicted.
Question from Dr. J. K. Chakravartty, BARC, Mumbai:—Once hydride is

formed, it generates stresses in the surrounding matrix. While estimating chemical
potential do you consider the effect of generated stresses on it?
Authors’ Response:—No. This must be done in a future work, since the stresses
are rather large. Surface energy must also be included. But these aspects involve ei-
ther kinetics or localization of the hydrides (FEM or other space-dependent techni-
ques), which were not the subject of this paper.
Questions from Antoine Ambard, EDF R&D:—Is there any difference in texture
between liner and Zircaloy-2? If yes, it would be possible to estimate an order of
magnitude of the thermal stresses generated during cooling.
Authors’ Response:—The difference of texture between liner and the Zircaloy-2

matrix is very small, amounting to the texture gradient through the thickness of a
standard cladding (no supplementary effect due to the difference in composition).
Thus, though I did not compute the residual stress, I guess it is very small. As the
stress gradient has a small effect on hydrogen diffusion, I do not expect any signifi-
cant precipitation due to this texture difference.
In duplex material, H seems to precipitate at the Zircaloy-2/liner interface.
Have you made any observations related to the possibility that the interface traps
hydrogen?
Authors’ Response:—The interface in duplex (or liner) cladding is not a metallurgi-

cal interface, in the sense that we can see a crystallographic continuity across the inter-
face (some grains are continuous from one side to the other). There is thus no reason
that hydrogen be trapped at the interface as it could be on some grain boundaries.
How does the lH evolve with temperature? Difference between various metal is
already small at high temperature.
Authors’ Response:—As a first approximation for dilute solutions, the evolution

@G
of lH with the temperature is given by l ¼ @n i
¼ l0 þ RT ln cci (see the paper for
further details).
According to the presentation from Brachet et al., we can see that cooling rate
is an important parameter in hydrogen distribution. Could you detail the exact tem-
perature history of your autoclave test?
Authors’ Response:—The cooling rate has effectively an important influence on

the actual hydride distribution (see the answer to JC Brachet’s question in this sec-
tion). The samples underwent several cycles in autoclave (heating, corroding at con-
stant temperature, cooling). The cooling rate is not available, but rather low (several
hours to cool from 400 or 415 C to room temperature).
K. V. Mani Krishna,1 D. G. Leo Prakash,2 D. Srivastava,1

N. Saibaba,3 J. Quinta da Fonseca,2 G. K. Dey,1
and M. Preuss2
Influence of Sn on Deformation
Mechanisms During Room
Temperature Compression
of Binary Zr–Sn Alloys
Reference
Mani Krishna, K. V., Leo Prakash, D. G., Srivastava, D., Saibaba, N., Quinta da Fonseca, J., Dey,
G. K., and Preuss, M., “Influence of Sn on Deformation Mechanisms During Room Temperature
Compression of Binary Zr–Sn Alloys,” Zirconium in the Nuclear Industry: 17th International
Symposium, STP 1543, Robert Comstock and Pierre Barberis, Eds., pp. 138–158, doi:10.1520/
ABSTRACT
Role of Sn on the deformation mechanisms of Zr was investigated undertaking in
situ compression loading experiments using neutron diffraction and
complementary post mortem electron microscopy. Fully recrystallised binary
Zr–Sn alloys, displaying a typical rolling þ recrystallisation texture, were studied
in situ to determine intergranular strain development, monitor peak intensity
changes indicative of twinning, and peak broadening indicative of increased
dislocation density. Significant 1012 1011 type tensile twinning activity was
observed for all samples during compression loading. The critical stress for the
twin nucleation and the extent of twinning were found to be strongly influenced
by the Sn content. Critical plastic strain for the nucleation of twining, however,
was observed to be weakly dependent on the Sn content. Results demonstrate
Manuscript received March 15, 2013; accepted for publication January 19, 2014; published online June 17, 2014.
1
Materials Science Division, Bhabha Atomic Research Centre, Trombay, Mumbai 400085, India.
2
Manchester Materials Science Centre, The Univ. of Manchester, Grosvenor St., Manchester M1 7HS, UK.
3
Nuclear Fuel Complex, Hyderabad 500062, India.
4
Hyderabad, India.
Copyright V
MANI KRISHNA ET AL., DOI 10.1520/STP154320130044 139
significant plastic slip activity to be a necessary condition for the onset of

twinning regardless of the Sn content.
Keywords
Zr–Sn alloys, deformation mechanisms, intergranular strains, twinning, in situ
compression, neutron diffraction
Introduction
The wide spread use of Zr based structural materials in thermal nuclear reactors
arises from their unique combination of resistance to environmental degradation
(in the form of corrosion and mechanical degradation) combined with their high
transparency to thermal neutrons [1]. One of the key alloying elements in many
commercial Zr-alloys, such as Zircaloy-2, Zircaloy-4, E635, and ZIRLOTM, is Tin
(Sn), which was originally added to mitigate issues related to nitrogen on the aque-
ous corrosion resistance and to provide sufficient mechanical strength through solid
solution strengthening [2,3]. However, efforts in further improving corrosion resist-
ance of Zr cladding material have resulted in alloys with significantly reduced Sn
levels. For instance, Zircaloy-4 with improved corrosion resistance was developed
by reducing the Sn content. A similar trend has been pursued for ZIRLOTM with
the development of Optimized-ZIRLOTM [4–10]. Although there exists much litera-
ture on the role of Sn on corrosion behaviour of Zr based alloys [1], very limited in-
formation, if any, is available in the open literature on the role of Sn on the
deformation mechanisms in such materials. In the light of aforementioned efforts
to improve corrosion resistance by the reduction of Sn in Zr alloys, understanding
the associated effects brought in by the changes in Sn content on the deformation
behaviour of the alloy becomes pertinent. Since Zr based components are fabricated
through a series of thermomechanical treatments, involving a considerable degree
of deformation, a thorough knowledge of the operative deformation modes and the
changes in them, if any, as a function of Sn content becomes imperative for the
effective optimization of the fabrication procedures. Such knowledge is crucial for
the prediction of the evolution of microstructure and texture, which in turn helps
developing components with superior in-service performance.
Tin, unlike a majority of other common alloying elements used in nuclear
grade Zirconium alloys (such as Fe, Cr, Nb, etc.) shows relatively high solid solubil-
ity and thus contributes to appreciable solid solution strengthening [11–13]. Hence,
Sn has the potential of having a direct effect on the numerous possible slip systems
in Zr. Deformation by slip in materials with an hcp crystal structure, such as Zr and
Ti, is complicated by having easy hai type slip modes and more difficult to activate
slip modes that includes a hci component, i.e., hc þ ai slip [2,14–18]. Furthermore,
due to the difficulty of activating hc þ ai slip, Zr has a significant tendency for twin-
ning, which plays a major role in evolution of crystallographic texture of the mate-
rial [14]. However, existing information on how this twinning tendency in Zr is
influenced by alloying elements in general and by Sn in particular is rather limited,

forming the motivation to the present study. A particular focus of the present work
is the way Sn might affect twin nucleation and overall twin activity during the early
stages of deformation. It is worth mentioning here that twin nucleation criteria in
materials with an hcp crystal structure are still largely unknown. While in some
cases a simple stress criterion is assumed [19,20], similar to a critical resolved shear
stress for slip, there is evidence that partial dislocations are an additional require-
ment for twin nucleation [21,22].
A majority of previous deformation studies relied on slip trace analysis of the
single crystals for the study of the deformation behaviour and identification of the
active slip modes [16–18]. However, the stress state of individual grains within a
polycrystalline aggregate (which is of more practical importance from the industrial
application point of view) is significantly more complex compared to the uniaxial
stress state when loading a single crystal. Transmission electron microscopy (TEM)
based analysis has also been used on deformed samples to identify dominant defor-
mation mechanisms by identifying dislocations structures and types [15,23]. How-
ever, this technique suffers from poor statistics, can only be used in crystals with
low dislocation densities and cannot be used to derive the Critical Resolved Shear
Stress (CRSS) of the different slip systems. An elegant alternative to overcome the
above mentioned limitations is to use in situ neutron diffraction during mechanical
loading experiments, which acts as an efficient tool for recording the development
of intragranular strains of various grain families [24–32]. Such intragranular strain
developments can be seen as the fingerprint of certain deformation modes. Because
of the highly penetrating nature of the neutrons, a large volume of the sample can
be probed, making the measurements statistically reliable. The present study uses in
situ neutron diffraction extensively along with complementary microstructural
characterization tools of Electron back scattered diffraction (EBSD) and TEM to
evaluate the role of Sn on the deformation modes of Zr.
Experimental
MATERIAL AND PROCESSING OF SAMPLES
In order to study the role of Sn on the deformation mechanisms in Zr, four model
Zr–Sn binary alloys were prepared with nominal compositions of 0.15, 0.23, 0.33,
and 1.20 % Sn (amounts are in weight percentage), balance being Zr. In order to
keep the effects of the other trace elements to bear minimum, alloys were prepared
from the same relatively pure nuclear grade Zr sponge. In the light of the expected
role of the interstitial elements on the twinning behaviour [33], total interstitial
content variation was kept to a narrow range in these four alloys (see Table 1 for the
actual composition). The ingots were prepared by quadruple electron beam melting.
The cast structure of the ingots was broken by hot extrusion at 800 C. Blocks of
35 mm by 22 mm by 15 mm were cut from the extruded material and b heat treated
at 1050 C for 20 min followed by water quenching. The purpose of the heat
TABLE 1 Summary of the role of Sn on the yield stress, critical stress, and strains for the onset of
twins.
Alloy Sn content (Wt. %) 0.15 0.23 0.33 1.2

Interestital content (O þ C þ N) (530 þ 80 þ 48) (880 þ 80 þ 45) (830 þ 30 þ 49) (930 þ 120 þ 34)
in ppm by weight ¼ 658 ¼ 1005 ¼ 909 ¼ 1084
Macroscopic yield point (MPa) 260 281 303 352
Critical lattice strain for twin 1050 1150 1350 1550
nucleation (10–6)
Critical plastic strain for twin 1.3 1.5 1.6 2.1
nucleation (%)
Critical applied stress for twin 271 300 356 391
nucleation (MPa)
treatment was to reduce any potential microsegregation of Sn and generate a similar

starting microstructure with no preferred crystallographic texture. After b quench-
ing, the blocks were stress relieved at 500 C for 24 h followed by 1 C/min cooling.
Prior to hot rolling, the blocks were homogenized at 550 C for 40 min
and subsequently rolled at 550 C to a reduction of 65 %, followed by annealing
treatment at 550 C for 24 h. The purpose of this thermomechanical process was to
generate a fully recrystallised microstructure and a typical crystallographic texture
with the majority of basal poles aligned along the normal direction (ND) with some
spread along TD (transverse direction). The purpose of creating this microstructure
was to enable compression tests with the majority of the hci axis either aligned per-
pendicular (compression along the radial direction (RD)) or nearly parallel (com-
pression along ND) to the loading direction during the in situ neutron diffraction
studies. To date, only data from loading along RD have been fully analysed and are
presented here. The samples for the in situ compression experiments were
machined to cylindrical shape with a diameter of 6 mm and a length of 10 mm
(along the rolling direction).
IN SITU LOADING AND NEUTRON DIFFRACTION EXPERIMENTS

The time of flight neutron diffraction beamline ENGIN-X, at ISIS, Rutherford
Appleton laboratory, UK, was used for the in situ compression loading and diffrac-
tion experiments of the present study [34–36]. The schematic of the experimental
set up (Fig. 1) shows that the location of the detectors allows the capturing of diffrac-
tion data along two principle directions of the sample, e.g., along the rolling direc-
tion (longitudinal detector) and normal direction (transverse detector). The
diffraction peaks carry information regarding the lattice strain evolutions, and
deformation mechanisms through the changes in their peak positions (for intergra-
nular elastic lattice strains), intensity (for twinning), and FWHM (full width at half
maximum). For instance, onset of twinning can be detected by the sudden change
in the lattice strain evolutions of the grain family undergoing twinning (as shall be
FIG. 1 Schematic illustrating (a) in situ neutron diffraction experimental set up. (b) and
(c) show the orientation of the crystallographic planes sensed by the two
neutron detectors that are mounted on either side of the incident beam.
Essentially, crystallographic planes of the sample, because of diffraction, act like
mirrors to the incident neutron beam and deflect the incident beam on to the
detectors. Depending on the location of the detectors, the response of either
plane can be captured in the longitudinal (b) and transverse (c) direction in
respect to the loading direction. RD and ND in (a) indicate the orientation of the
sample’s “Rolling” and “Normal” directions, as used in the present study.
demonstrated in the following section). Similarly, significant changes in the FWHM

of the diffraction spectra indicate high slip assisted plastic deformation activity
from the grain families corresponding to the respective diffraction peaks.
In situ compression loading experiments were carried out using a 100 kN Ins-
tron compression rig. The samples were compressed along the rolling direction and
the data points in the elastic regime (up to 50 MPa below the macroscopic yield
point) were captured in constant stress state. This is followed by a continuous strain
controlled loading up to 0.18 true strain. In order to make effective use of beam
time, the data points in the plastic regime were captured at two different strain
rates: a slower rate of 7 106/s until 0.025 strain followed by 2.8 105/s until
0.18 strain. The frequency of measurement points was increased around the yield
point by the deliberate selection of low strain rate, as the onset of plasticity is a key
area of interest in understanding deformation mechanisms. In order to obtain a rea-
sonably good signal to noise ratio of the diffraction peaks, acquisition time for the
diffraction spectra at each of the measurement points was kept to around 5 min.
Due to the very slow strain rate, the sample temperature was constant throughout
the loading experiment. Furthermore, the energy range of the neutron spectrum at
ISIS is rather low to cause any radiation damage to the sample during the in situ
experiments. Thus the effects seen in the present study do not have any contribu-
tion from the radiation damage related phenomena.
MICROSTRUCTURAL CHARACTERIZATION
Both, undeformed and deformed samples were subjected to a detailed microstruc-
tural characterization using EBSD and TEM. Sample preparation for EBSD con-
sisted of standard metallographic preparation followed by electropolishing at 15 V
and a temperature of 5 C in an electrolyte of 80 % Methanol þ 20 % Perchloric
acid. TEM samples were also prepared using same electrolyte in a twin jet
electrolytic polisher at 40 C. The EBSD analysis was carried out on the FEI Sirion
FEG–SEM equipped with an HKL system. For texture analysis low spatial resolu-
tion (step size of 50 lm) large area maps were recorded covering an area of
70 mm2. Microstructural orientation maps were recorded using a step size of
0.3 lm covering 500 lm by 200 lm. Selected samples, subjected to low deformation
levels (0.02–0.03 true strain), were examined in a JEOL 2000FX TEM for the char-
acterization of early stages of twin formation and dislocation structure.
Results
STARTING MICROSTRUCTURE AND TEXTURE
Figure 2 depicts the typical microstructure and texture of the recrystallised material
(here Zr–1.2 %Sn) prior to in situ loading and diffraction experiments. No signifi-
cant microstructural differences were observed in the four alloys considered for the
present work. As it can be observed from the grain orientation map (inverse pole
figure colour code), the starting microstructure is completely recrystallised with an
average grain size of 4.5 lm (based on EBSD measurements). Grain size analysis of
the other three alloys gave 5.5 lm for Zr–0.15 %Sn, 4.5 lm for Zr–0.23 %Sn and
5 lm for Zr–0.33 %Sn. TEM micrographs (Fig. 2(b)) also show grains with low
dislocation density and well defined high angle grain boundaries, typical of recrys-
tallised material. As illustrated in Fig. 2(c), the starting texture is predominantly
basal poles aligned towards ND of the sample, with a 6 30 spread along TD.
Furthermore, there is also a slight tendency of the h1120i poles being preferentially
aligned towards RD.
MACROSCOPIC STRESS STRAIN BEHAVIOUR

The macroscopic stress strain behaviour of the four Zr–Sn alloys recorded during
the in situ compression experiments along the rolling direction is presented in
Fig. 3. As expected, with increasing Sn content the stress–strain curves exhibited
higher flow stress values. The yield point data can also be found in Table 1. In
addition, closer examination of the flow curves reveals the following
observations:
FIG. 2 Typical starting microstructure, (a) EBSD and (b) TEM, and (c) texture of the
samples subjected to in situ loading in neutron diffraction studies. In the present
case the results were recorded on the Zr–1.2 % Sn alloy.
1. The curves exhibited a comparatively flat stress-strain response regime (i.e.,

region of low strain hardening) during the initial stage of plastic deformation,
highlighted by a circle in Fig. 3. Such flat response was more prominent with
increasing Sn content.
2. The strain hardening rate in the high plastic strain domain (highlighted by
rectangle), increases with Sn content.
Both observations suggest that Sn does have an effect on deformation mecha-
nisms in these alloys.
EVOLUTION OF INTERGRANULAR ELASTIC STRAINS AS A FUNCTION OF SN

For further insights into the active deformation modes and the effect of changing
Sn content, we resort to analysis of the intergranular strain evolutions determined
by neutron diffraction during in situ loading. It is to be noted that the intergranular
strains presented in this study do not represent the absolute elastic strain, but the
relative change in strains from the unloaded condition. As described earlier, during
the loading experiment the diffraction signal is recorded in the rolling (longitudinal
detector) and normal (transverse detector) direction. Figure 4 shows the evolution
of the compressive elastic strain measured in the rolling (longitudinal) direction as
a function of the applied stress for different grain families of the four alloys consid-
ered in the present study. In other words, we can see the elastic strain development
FIG. 3 Macroscopic true stress–true strain curves recorded during the in situ
compression/diffraction experiments. Circled region near the lower strain values
indicates region of relatively low strain hardening. Regions highlighted by
rectangular box exhibit increasing strain hardening rate with increasing Sn
content.
of the grains that contribute to 0002, 1010, and 1011 reflections along the loading
direction. During the initial stage of loading, i.e., prior to the onset of plasticity, all
three grain families display a reasonably linear elastic strain response. The data
become particularly interesting in the moment of plasticity and the following
important observations can be made.
1. The 1010 grain family shows the highest degree of unloading (i.e., reduction in
elastic strain despite a further increase of applied stress) around the onset of
plastic yield and twinning, which is a signature of extensive plasticity in this
grain family. It is also noticeable that the initial degree of unloading increases
with increasing Sn content.
2. The 1011 grain family does not display a sharp unload at the early stage of
plastic yield but with increasing level of plasticity, this grain family takes up
less and less additional elastic strain despite a further increase of applied stress.
With increasing Sn content, the 1011 and 1010 elastic strain responses become
less distinguishable.
3. The 0002 grain family exhibits significant build-up of the lattice strain before a
sudden unload followed by a continued increase in build-up of lattice strain.
As discussed in the next section, this unloading can be related to twinning
type characterized by the orientation relationship f1012gh1011i commonly
known as “tensile twinning.” This twinning mode rotates the hci axis by about
85 from a tensile (here transverse loading direction, tensile stress is the result
of Poisson effect) towards a compression direction (here loading direction).
Therefore, after the onset of twinning, the 0002 grain family detected along
FIG. 4 Evolution of elastic lattice strains, along the loading direction, for the three
families of orientations (0002, 1010, and 1011) during in situ compressive loading
and neutron diffraction experiment for all the four Zr–Sn alloys used in the
present study. The horizontal dotted lines in the plots represent the
macroscopic applied stress at which onset of twinning was observed. Circled
regions highlight the fact that with increasing Sn content, the differences
between the 1010 and 1011 responses (i.e., separation of the respective curves)
decreases. Increased extent of unloading of the 1010 family of grains with
increasing Sn content can also be observed (marked with the arrows).
RD is made up of both pre-existing 0002 grains and the reoriented 0002

twinned volumes. The observed unloading is a result of relatively lower elastic
strain of the twinned volumes (twinning is essentially a stress relaxation
mechanism).
4. The stress at which the onset of twinning was observed, now after referred to
as critical stress for twin nucleation, (indicated with a horizontal line in the
Fig. 4) was found to increase with increasing Sn content.
Figure 5 presents the evolution of the lattice strains recorded along the normal
direction for the 0002 grain family for the Zr–0.15 %Sn and the Zr–1.2 %Sn alloy.
This grain family is expected to display tensile twinning, moving some of the dif-
fracting volume from normal to the rolling direction. Therefore, monitoring the
0002 reflection along ND (transverse to the loading direction) enables one to deter-
mine the intergranular strain at which twin nucleation starts. Figure 5 demonstrates
FIG. 5 Evolution of the (0002) lattice strains perpendicular to the loading (normal)
direction during in situ compression loading and neutron diffraction experiment
for Zr–0.15Sn and Zr–1.2Sn. The behaviour of the other two alloys is intermediate
to these two extremes. The critical lattice strain for the two alloys has been
marked by respective arrows.
that the critical elastic strain at which twinning initiates (indicated by arrow in
Fig. 5) is higher in the alloy with a high Sn content compared to the alloy with a low
Sn content. The critical lattice strains for the onset of the twinning for all four sam-
ples are compiled in the Table 1, which emphasises the correlation between Sn con-
tent and critical lattice strain for twin nucleation.
EVOLUTION OF INTEGRATED INTENSITY OF THE 0002 REFLECTION DURING

LOADING
Since twinning is associated with a rotation of the crystal structure, it can result
in detectable peak intensity variations of the 0002 reflection during plastic defor-
mation. In the present case, tensile twinning results in rotation of the hci axis
from a tensile towards compression direction. Since there is only one compres-
sion direction during compression loading and the f1012gh1011i type tensile
twin results in a 85 rotation of the hci axis, it is relatively easy to detect this
twinning mode in the present case by monitoring the integrated intensity of the
0002 reflection with straining. Here an increase of 0002 integrated intensity
means that tensile twinning is active (Fig. 6). When the change of 0002 integrated
intensity is plotted against true strain, Fig. 6(a), it can be seen that twinning ini-
tiates between a strain of 1.3 102 and 2.1 102. The data suggest that a
slightly higher strain is required for the activation of twinning with higher Sn
content (see Table 1). This is consistent with the higher value of intergranular
FIG. 6 Change in the integrated intensity of the 0002 reflection recorded using
detector along loading direction as a function of (a) true macroscopic strain (b)
true macroscopic applied stress.
strains measured at twinning for the alloys with more Sn. The observed increase
in critical strain values for the onset of twining, when viewed in the context of
the expected error in the strain measurement (due to machine compliance and
diffraction data being collected in continuous strain mode), suggests that critical
strain changes only modestly with Sn content, unlike the critical applied stress
which was found to increase significantly with the Sn content (Fig. 6(b) and Table
1). In general, twinning activity started in all of the samples between a strain of
1 102 and 2 102. Since the actual intensity of the peak corresponds to the
volume fraction of the orientations favourable for diffraction, and the 0002 inten-
sity increase in present case is due to twinning activity, one can easily infer that
samples with higher Sn content exhibit higher twinning activity. When plotted
against applied stress, it becomes apparent that the critical stress required for the
onset of twinning is strongly dependent on the Sn content, with Fig. 6(b) and
Table 1 providing a more quantitative picture of these observations.
MICROSTRUCTURE OF DEFORMED SAMPLES

In order to confirm the nature of the twins formed, and estimate their volume
fraction, extensive EBSD and TEM was carried out on the deformed samples.
Samples deformed in situ on ENGIN-X were used for the purpose of EBSD char-
acterization. An orientation map of Zr–1.2 %Sn deformed to a strain of 0.18 is
presented in Fig. 7(a) (as in Fig. 2(a), the inverse pole figure colour code has been
applied here). It should be noted that this map only represents a small crop of
the area scanned by EBSD. Figure 7(b) highlights > 15 misorientation boundaries
in grey and all twin boundaries associated with f1012gh1011i tensile twinning in
black, which shows that most of the red regions in Fig. 7(a) can be associated
FIG. 7 Zr–1.2 %Sn deformed to 0.18 true strain (a) grain orientation map recorded by
EBSD (IPF colour code) and (b) grain boundary map with high angle grain

boundaries in grey and 1012 1011 twin boundaries in black.
with the common tensile twin type. Note the absence of red in the orientation
map before compression (Fig. 2(a)). Similar observations have been reported pre-
viously [31] for a similar starting texture and loading direction arrangement.
Since the initial texture of the sample is such that majority of the basal poles are
along ND and compressive loading was applied to act along RD, a majority of
the grains will have their hci axis under tension leading to the formation of ten-
sile twins. It is notable, however, that the twinned regions in many cases do not
display a typical twin morphology. Instead, it appears that after 0.18 strain, entire
grains have been reoriented by the tensile twinning mode. Similar observations
have been made previously in Ti alloys [37].
TEM examination was also carried out for the characterization of the twins, but
on samples deformed to a plastic strain of 2 %–3 %, with an objective of under-
standing the nucleation stage of twins. This strain, as can be seen in the Fig. 6(a)
and Table 1, is just sufficient to nucleate the twins in different alloys. As shown in
the Fig. 8(a), twins of approximately 200 nm width were formed at this level of
deformation. More interestingly, it was observed that, the grains in which twins
could be observed have shown comparatively high dislocation density, indicating
significant plasticity by slip.
Figure 9 shows the twinning volume fraction as a function of the Sn content for
the samples deformed to 0.18 compressive strain measured by EBSD. Despite the
scatter of the data related to the comparatively small volume studied by EBSD, it is
evident that an increase in Sn content results in an increase in twin volume fraction.
This observation is in excellent agreement with the neutron diffraction
observations.
FIG. 8 TEM analysis with (a) bright field micrograph showing the formation of a twin in
the early stage of deformation. The image is from a sample subjected to only
3 % compressive deformation, and (b) composite diffraction pattern from the

twin and matrix region recorded along 24
23 zone axis of the matrix grain.

Misorientation between the matrix and twin is 140o@ 24
23 , which is

symmetrically equivalent to 85o@ 2110 tensile twins.
Discussion
The key findings of the present study with respect to the role of the Sn on the defor-
mation behaviour can be summarized in the following way:
• An anomalous flat response in stress–strain behaviour was observed during
the initial part of plastic deformation, the extent of which increased with the
Sn content.
FIG. 9 EBSD derived twin volume fraction as a function of Sn content for the samples
compressed to a strain of 0.18.
• Initial unloading of the intergranular strains of the 1010 grain family increased
with rising Sn content.
• Critical macroscopic strain for the nucleation of twinning was observed to be
a weak function of alloy Sn.
• Critical intergranular elastic strain and macroscopic stress for the nucleation
of the twins increased significantly with increasing Sn content of the alloy.
• The extent of twinning activity increased with Sn content.
An anomalous flat response of the macroscopic stress-strain curve during the

initial part of the plastic deformation has been observed by previous researchers
and was attributed to thermal intergranular stresses [32,38,39], which are a result of
the anisotropic thermal expansion coefficient for hai and hci direction [40]. Hence,
at room temperature the hci axis is tensioned while the hai axis is compressed
[32,38,39,41]. The present work clearly shows that the degree of initial flat response
depends on the level of Sn in the alloy. Therefore it seems unlikely that thermal
intergranular stresses are indeed the cause. It is very notable that during the flat
response the intergranular strain increases very dramatically indicating very signifi-
cant plasticity activity. However, activation of f1012gh1011i tensile twinning only
occurs once the flat response is overcome. EBSD and TEM analysis did not indicate
any other twin activity and therefore one might argue that the flat strain response
must be related to a slip burst (though strangely not resulting in any strain harden-
ing). In fact, there is a decrease in the 1010 elastic strains, indicating strain soften-
ing. It appears that when the yield point of the material increases, due to solution
strengthening of Sn, the initial strain burst also becomes more pronounced. It is
likely therefore that this strain burst is related to the ability of Sn to initially pin dis-
locations. Like carbon in steels, this initial yielding seems to be discontinuous, as
new dislocations are unpinned and therefore can move at a stress lower than the
initial slip resistance.
The fact that increasing levels of critical plastic strain and increased critical
applied stress for nucleation of twins was observed for the alloys of increasing Sn
suggests that a certain amount of plastic deformation is necessary before twinning
occurs in these alloys. This is in contrast to conventional opinion that slip and twin-
ning are mutually exclusive. The present work clearly shows that slip is a prerequi-
site for the onset of twinning. This could be because either a minimum dislocation
density is required to nucleate twins or that there is a need for the development of
intergranular strains in the parent grains before twinning starts.
Several previous studies have indicated that availability of twinning dislocations
is necessary for twin nucleation [21,22]. Hence, the nucleation probability of twins
can be thought to be proportional to the density of twinning dislocations, which in
turn are proportional to the extent of the plastic activity. Present results suggest
that at a strain between 1 and 2 %, a sufficient density of dislocations is produced,
giving rise to observed twin nucleation. Furthermore, the increased extent of
unloading of the 1010 grain family (see Fig. 4) with higher Sn content indicates
increased plastic activity of the twinning grain family. Therefore one might assume
a rise in dislocation density with increasing Sn content at a given macroscopic plas-

tic strain. This should in principle result in more nucleation sites for twins. In order
to further support this assumption, peak broadening (related to dislocation density)
was investigated of the twin prone grain family. It is well known that the grain
families, which undergo considerable plastic activity, give rise to diffraction peak
broadening due to increased dislocation activity [42]. Thus if indeed some critical
plastic strain is a prerequisite for onset of twinning, a corresponding signal should
be seen in the peak profile of the twinning grains. In principle, grains that have the
normal of either their first order or second order prismatic plane aligned with the
loading directions are the grains, which will twin, since their hci axis is orientated
in the transverse orientation. Figure 10 plots the change of FWHM of the 1120 (i.e.,
second order prismatic) reflection measured in the loading direction for
Zr–0.5 %Sn and Zr–1.2 %Sn. It can be clearly seen that there is considerably more
peak broadening in the case of Zr–1.2 %Sn compared to Zr–0.15 %Sn suggesting
higher slip activity in case of the higher Sn containing alloy.
The other important aspect is that Sn is an effective solid solution strengthen-
ing element in Zr alloys [43,44] and therefore the stresses for the onset of plasticity
increases with rising Sn content. Hence, with increasing Sn content, plasticity, and
therefore twinning, starts to occur at higher stress levels. Consequently, the stress
level during twinning as a function of plastic strain, keeping in mind that twinning
only starts after some plasticity by slip, will always have to be higher when the
material has been solution strengthened. Hence, there is a greater energy to nucleate
and grow twins with increasing strength of the material regardless of the potential
effect of Sn on for instance stacking fault energy.
FIG. 10 Variation in FWHM of 1120 peak (recorded along the loading direction) as a
function of the strain for two of the alloys considered in the present study.
In any case, the results suggests that increasing Sn content makes twin nuclea-
tion happen at higher stresses, but that, once it starts, it occurs more rapidly.
Conclusions
The present study made extensive use of in situ neutron diffraction and comple-
mentary microstructural characterization techniques to bring out the role of Sn on
the deformation behaviour of Zr. The influence of Sn on the twinning, in particular,
has been quantified in terms of the critical stresses and strains required for the onset
of the twins in Zr–Sn binary system. The following conclusions can be drawn from
the present study:
• Sn content has been observed to have a significant effect on the activation and
extent of different deformation modes of Zr, as revealed by the intergranular
strain, FWHM and change in integrated intensity data of the various grain
families during the in situ compression experiment using neutron diffraction.
• By far the most dominant twinning mode during our compression experi-
ments was the f1012gh1011i tensile twinning mode.
• The critical macroscopic stress and intergranular elastic strain required for
twin nucleation were observed to increase with Sn content.
• Despite the requirement of higher stresses for twin nucleation with increasing
Sn content, twinning becomes more dominant with increasing Sn content.
• A critical plastic strain is needed for the onset of the twinning in all alloys.
This was found to be between 1 %–2 %.
• Peak width analysis indicative of dislocation density suggests that with increas-
ing Sn content, an early strain burst results in significant dislocation density in
the grains that are orientated for tensile twinning.
Therefore, it appears that Sn content affects twinning simply by making defor-
mation by slip more difficult. It makes twinning nucleation more difficult but
enhances the twinning rate once twinning does start. The two factors that seem to
drive twinning rates after the initial nucleation are stress and dislocation densities
in the twinning grains.
ACKNOWLEDGMENTS
The present work is funded jointly by EPSRC, UK (EP/I012346/1) and the Depart-
ment of Atomic Energy, India, under the Indo-UK collaborative programme on
“Peaceful uses of Nuclear Energy”.
References
[1] Cox, B., “Some Thoughts on the Mechanisms of In-Reactor Corrosion of Zirconium
Alloys,” J. Nucl. Mater., Vol. 336, 2005, pp. 331–368.
[2] Linga Murty, K. and Indrajit, C., “Texture Development and Anisotropic Deformation of
Zircaloys,” Prog. Nucl. Energy, Vol. 48, 2006, pp. 325–359.
[3] Krishnan, R. and Asundi, M. K., “Zirconium Alloys in Nuclear Technology,” Sadhana, Vol.
4, No. 1, 1981, pp. 41–56.
[4] IAEA, “Waterside Corrosion of Zirconium Alloys in Nuclear Power Plants,” IAEA-TEC-
DOC-996, IAEA, Vienna, Austria, 1998.
[5] Seibold, A. and Woods, K. N., “Advanced PWR Cladding,” Proceedings of the Interna-
tional Topical Meeting on Light Water Reactor Fuel Performance, American Nuclear Soci-
ety, La Grange Park, IL, 1994, pp. 633–642.
[6] Garzarolli, F., Broy, Y., and Busch, R. A., “Comparison of the Long Time Corrosion Behav-
iour of Certain Zirconium Alloys in PWR, BWR and Laboratory Tests,” Zirconium in the
Nuclear Industry: Eleventh International Symposium, ASTM STP1295, E. R. Bradley and
G. B. Sabol, Eds., ASTM International, West Conshohocken, PA, 1996, pp. 850–863.
[7] Fuchs, H. P., Garzarolli, F., Weidinger, H. G., Boomer, R. P., Meier, G., Besch, O.-A., and
Lisdat, R., “Cladding and Structural Material Development for the Advanced Siemens
PWR Fuel Performance,” Fuel for the 90s, Proceedings of the International Topical Meet-
ing on LWR Fuel Performance, American Nuclear Society/European Nuclear Society, La
Grange Park, IL, 1991, pp. 682–690.
[8] Takeshi, I. and Masuto, Y., “Development of Highly Corrosion Resistant Zirconium-Base
Alloys,” Zirconium in the Nuclear Industry: Ninth International Symposium, ASTM
STP1132, C. M. Eucken and A. M. Garde, Eds., ASTM International, West Conshohocken,
PA, 1991, pp. 346–367.
[9] Charquet, D., “Improvement of the Uniform Corrosion Resistance of Zircaloy-4 in the
Absence of Irradiation,” J. Nucl. Mater., Vol. 160, 1988, pp. 186–195.
[10] Yueh, H. K., Comstock, R. J., Shah, H. H., Colburn, D. J., Dahlback, M., and Hallstadius, L.,
“Improved ZIRLOTM Cladding Performance Through Chemistry and Process Mod-
ifications,” Zirconium in the Nuclear Industry: Fourteenth International Symposium,
ASTM STP1467, P. Rudling and B. Kammenzind, Eds., ASTM International, West
[11] Okamoto H., “Sn–Zr (Tin-Zirconium),” J. Phase Equilib. Diff., Vol. 31, 2010, pp. 411–412.
[12] Carpenter, G. J. C., Ibrahim, E. F., and Watters, J. F., “The Aging Response of Zirconium-
Tin Alloys,” J. Nucl. Mater., Vol. 102, 1981, pp. 280–291.
[13] Charquet, D., Hahn, R., Ortlieb, E., Gros, J.-P., and Wadier, J.-F., “Solubility Limits and
Formation of Intermetallic Precipitates in ZrSnFeCr Alloys,” Zirconium in the Nuclear
Industry: Eighth International Symposium, ASTM STP1023, C. M. Eucken and L. F. P. Van
Swam, Eds., ASTM International, West Conshohocken, PA, 1989.
[14] Tenckhoff, E., “Review of Deformation Mechanisms, Texture, and Mechanical

Anisotropy in Zirconium and Zirconium Base Alloys,” J. ASTM Int., Vol. 2, No. 4, 1998, pp.
199–224.
[15] McCabe, R. J., Cerreta, E. K., Misra, A., Kaschner, G. C., and Tomé, C. N., “Effects of Texture,
Temperature and Strain on the Deformation Modes of Zirconium,” Philos. Mag., Vol. 86,
2006, pp. 3595–3611.
[16] Akhtar, A., “Basal Slip in Zirconium,” Acta Metall., Vol. 21, 1973, pp. 1–11.
[17] Akhtar, A., “Compression of Zirconium Single Crystals Parallel to the C-Axis,” J. Nucl.
Mater., Vol. 47, 1973, pp. 79–86.
[18] Akhtar, A., “Prismatic Slip in Zirconium Single Crystals at Elevated Temperatures,” Met.
Trans. A, Vol. 6, 1975, pp. 1217–1222.
[19] Abdolvand, H. and Daymond, M. R., “Internal Strain and Texture Development During
Twinning: Comparing Neutron Diffraction Measurements With Crystal Plasticity Finite-
Element Approaches,” Acta Mater., Vol. 60, No. 5, 2012, pp. 2240–2248.
[20] Abdolvand, H., Daymond, M. R., and Mareau, C., “Incorporation of Twinning Into a Crystal
Plasticity Finite Element Model: Evolution of Lattice Strains and Texture in Zircaloy-2,”
Int. J. Plast., Vol. 27, No. 11, 2011, pp. 1721–1738.
[21] Wang, J., Beyerlein, I. J., Hirth, J. P., and Tome, C. N., “Twinning Dislocations on {1011} and
{1013} Planes in Hexagonal Close-Packed Crystals,” Acta Mater., Vol. 59, 2011, pp.
3990–4001.
[22] Christian, J. W. and Mahajan, S., “Deformation Twinning,” Prog. Mater. Sci., Vol. 39, 1995,
pp. 1–157.
[23] Jones, I. P. and Hutchinson, W. B. “Stress-State Dependence of Slip In Titanium-6Al-4V

and Other H.C.P. Metals,” Acta Metall., Vol. 29, No. 6, 1981, pp. 951–968.
[24] Cai, S., Daymond, M. R., and Holt, R. A., “Deformation of High b-Phase Fraction Zr–Nb
Alloys at Room Temperature,” Acta Mater., Vol. 60, No. 8, 2012, pp. 3355–3369.
[25] Garlea, E., Clausen, B., Kenik, E. A., Ciurchea, D., Vogel, S. C., Pang, J. W. L., and Choo, H.,
“Intergranular Strain Evolution in a Zircaloy-4 Alloy With Basketweave Morphology,”
Metall. Mater. Trans. A, Vol. 41, No. 5, 2010 pp. 1255–1260.
[26] Mosbrucker, P., Daymond, M. R., and Holt, R. A., “In Situ Studies of Variant Selection Dur-
ing the a-b-a Phase Transformation in Zr–2.5Nb,” J. ASTM Int., Vol. 8, No. 1, 2011, 103066.
[27] Muránsky, O., Daymond, M. R., Bhattacharyya, D., Zanellato, O., Vogel, S. C., and Edwards,
L., “Load Partitioning and Evidence of Deformation Twinning in Dual-Phase Fine-Grained
Zr–2.5 %Nb Alloy,” Mater. Sci. Eng. A, Vol. 564, 2013, pp. 548–558.
[28] Pang, J. W. L., Holden, T. M., Turner, P. A., and Mason, T. E., “Intergranular Stresses in
Zircaloy-2 with Rod Texture,” Acta Mater., Vol. 47, No. 2, 1999, pp. 373–383.
[29] Xu, F., Holt, R. A., Daymond, M. R., Rogge, R. B., and Oliver, E. C., “Development of Inter-
nal Strains in Textured Zircaloy-2 During Uni-Axial Deformation,” Mater. Sci. Eng. A, Vol.
488, 2008, pp. 172–185.
[30] Rangaswamy, P., Bourke, M. A. M., Brown, D. W., Kaschner, G. C., Rogge, R. B., Stout,
M. G., and Tomé, C. N., “A Study of Twinning in Zirconium Using Neutron Diffraction and
Polycrystalline Modeling,” Metall. Mater. Trans. A, Vol. 33, 2002, pp. 757–763.
[31] Allen, V. M., Da Fonseca, J. Q., Preuss, M., Robson, J. D., Daymond, M., and Comstock,
R. J., “Determination and Interpretation of Texture Evolution During Deformation of a Zir-
conium Alloy,” ASTM STP 1505, West Conshohocken, PA, 2009, pp. 550–563.
[32] Holt, R. A., Daymond, M. R., Xu, F., and Cai, S., “Intergranual and Interphase Constraints in
Zirconium Alloys,” J. ASTM Int., Vol. 5, No. 6, 2008, pp. 776–795.
[33] Garde, A. M., Aigeltinger, E., and Reed-Hill, R. E., “Relationship Between Deformation
Twinning and the Stress–Strain Behavior of Polycrystalline Titanium and Zirconium at 77
K,” Met. Trans., Vol. 4, 1973, pp. 2461–2468.
[34] Santisteban, J. R., Daymond, M. R., James, J. A., and Edwards, L., “ENGIN-X: A Third-
Generation Neutron Strain Scanner,” J. Appl. Crystall., Vol. 39, No. 6, 2006, pp. 812–825.
[35] Daymond, M. R. and Priesmeyer, H. G., “Elastoplastic Deformation of Ferritic Steel and
Cementite Studied by Neutron Diffraction and Self-Consistent Modelling,” Acta Mater.,
Vol. 50, No. 6, 2002, pp. 1613–1626.
[36] Grant, B. M. B., Francis, E. M., Quinta Da Fonseca, J., Daymond, M. R., and Preuss, M.,
“Deformation Behaviour of an Advanced Nickel-Based Superalloy Studied by Neutron
Diffraction and Electron Microscopy,” Acta Mater., Vol. 60, 2012, pp. 6829–6841.
[37] Leo Prakash, D. G., Ding, R., Moat, R. J., Jones, I., Withers, P. J., Quinta da Fonseca J., and
Preuss, M., “Deformation Twinning in Ti-6Al-4 V During Low Strain Rate Deformation to
Moderate Strains at Room Temperature,” Mater. Sci. Eng. A, Vol. 527, 2010, pp.
5734–5744.
[38] MacEwen, S. R., Tome, C., and Faber J., “Residual Stresses in Annealed Zircaloy,” Acta
Metall., Vol. 37, 1989, pp. 979–989.
[39] Mac Ewen, S. R. and Tome, C., “Residual Stresses in Textured Zirconium Alloys,” ASTM-
STP-939, West Conshohocken, PA, 1987, pp. 631–652.
[40] Douglas, D. L., “Metallurgy of Zirconium,” Atomic Energy Rev., Vol. 1, 1963, pp. 73–74.
[41] Kearns, J., “Thermal Expansion and Preferred Orientation in Zircaloy,” Report WAPD-TM-
472, Bettis Atomic Power Laboratory, West Mifflin, PA, 1965.
[42] Cullity, B. D., Elements of X-Ray Diffraction, Addison-Wesley, Reading, MA, 1956.
[43] Trojanov, Z., Lukfi, P., Kral, F., and Kral, R., “Discontinuous Low Temperature Deformation
of Zr–Sn Alloys,” Mater. Sci. Eng. A, Vol. 137, 1991, pp. 151–155.
[44] Dlouh, A., Trojanovfi, Z., and Lukfic, P., “Thermally (Non-) Activated Deformation of
Zr–Sn Polycrystals,” Czech. J. Phys. B, Vol. 38, 1998, pp. 482–484.
DISCUSSION
Questions from Arthur Motta, Penn State University:—You showed a TEM ob-
servation showing a twin in a sample where you had seen a twin signal in diffrac-
tion. Could you specify how many sample conditions (alloy content, deformation
stage, twin or not) did you examine and did they correlate with your conclusions?
Authors’ Response:—We have examined all four alloys (i.e., samples with differ-
ent Sn contents) by TEM. In order to facilitate TEM studies of nucleation stage of
twins, samples were deformed 1% plastic strain above the critical strain sufficient
to initiate twinning (based on neutron diffraction data), i.e., samples were deformed
to 2–3% of plastic strain. As a whole, few tens of twinned and non-twinned grains
have been examined by TEM for each alloy. In general, twinned grains consistently
exhibited relatively higher dislocation densities in comparison to their non-twin-
ning counterparts, corroborating the observations made in in-situ neutron diffrac-
tion and deformation results.
Questions from Mark Daymond, Queen’s University:—You discussed that

increase in Sn was associated with an increase in twin volume fraction. Can you
clarify if that is due to an increase in the number of twins or an increase in the size
of the twins?
Authors’ Response:—EBSD results suggest that higher twinning fractions

observed in higher Sn samples were essentially due to significantly larger number of
twins while the twin width in fact decreased with increasing Sn content.
Question from AnandGarde, Westinghouse Electric Co.:—Based on your work,

can you comment on the possible impact of tin on dislocation channeling, a defor-
mation mechanism important for deformation of irradiated zirconium alloys?
Authors’ Response:—Since irradiation was not considered in the present study,

it will not be possible to comment on the expected role of Sn on the dislocation
channeling using the present results.
Question from Dinesh Srivastava, BARC, Mumbai:—What is the critical strain

for different texture of material?
Authors’ Response:—We have studied the same material by subjecting it to uni-

axial compression along different directions (along rolling, transverse, and normal
directions of the sample). The results (which shall be presented in future publication)
did indicate towards a significant influence of deformation direction (in other words
texture) on the critical strain for twinning. It was observed that, for samples deformed
along ND direction, twinning is rather difficult even at large plastic strains.
Question from Javier Romero, Westinghouse Electric Co.:—How constant was

the oxygen content in the Zr-Sn alloys? Have you looked at the effect of oxygen?
Authors’ Response:—The four alloys considered for the present study were care-
fully chosen to have insignificant variation in oxygen content. Their oxygen content
ranges between 600-900 ppm. Hence the observed systematic variation in proper-
ties (stress strain response and twinning behavior) of the four alloys is a reflection
of variation in Sn content only and not the oxygen.
Questions from Srikumar Banerjee, BARC, Mumbai:—Critical resolved shear

stress (CRSS) for twinning is generally weakly dependent on solute content in solid
solution. The variation essentially comes from the dependence of solute content on
elastic modulus. I have not understood the logic for the dependence of solute con-
tent on CRSS of twinning when the critical strain for twin nucleation is the same
for all of the alloys. Please clarify.
Authors’ Response:—Since the Sn is a solid solution strengthening element, the

stress required to achieve the same critical plastic strain for twin nucleation
increases with increasing Sn content. Hence, the stress level for twin nucleation
increases with increasing Sn content. The work suggests that considering only a
CRSS criterion for twinning might not necessarily be sufficient.
Questions from Dr. J. K. Chakravartty, BARC:—You did mention about the for-
mation of partial dislocations during deformation of Sn containing Zr-alloys. Did
you see stacking faults in deformed microstructure of the Zr-Sn alloys that were
studied?
Authors’ Response:—It is very difficult to resolve partial dislocations and image

stacking faults in deformed alloys. However, several theoretical works indicate
towards the role of partial dislocations in the twinning behavior of the Zr based
alloys.
D. Kaczorowski,1 J. P. Mardon,2 P. Barberis,3

P. B. Hoffmann,4 and J. Stevens5
Impact of Iron in M5TM6

Reference
Kaczorowski, D., Mardon, J. P., Barberis, P., Hoffmann, P. B., and Stevens, J., “Impact of Iron in
M5TM,” Zirconium in the Nuclear Industry: 17th International Symposium, STP 1543, Robert
Comstock and Pierre Barberis, Eds., pp. 159–183, doi:10.1520/STP154320120195, ASTM
ABSTRACT
To assess the effect of iron (Fe) content on the properties of M5TM, several
industrial-sized ingots with Fe content up to 1000 ppm and some smaller ingots
with Fe up to 1300 ppm have been manufactured, tested, and investigated. The
evolution of the microstructure with iron content has been both experimentally
determined and theoretically assessed with thermodynamic computations,
showing that iron in the investigated range has only a slight impact on the
balance between the two classical second-phase precipitates observed in M5TM:
b-Nb and Laves phase. Moreover, it was found that the impact of iron on texture
and mechanical properties, including thermal creep, is null. Out-of-pile autoclave
corrosion tests show, on the one hand, that the iron content has no effect in
360 C primary water environment and little effect in 400 C–415 C steam. On the
Manuscript received December 5, 2012; accepted for publication November 3, 2013; published online June
17, 2014.
1
Research Engineer, AREVA NP, Fuel, 10 Rue Juliette Récamier 69456 Lyon Cedex, France (Corresponding
author), e-mail: damien.kaczorowski@areva.com
2
AREVA Fellow Expert, 10 Rue Juliette Récamier 69456 Lyon Cedex, France,
e-mail: jean-paul.mardon@areva.com
3
AREVA Fellow expert, AREVA/ CEZUS Research Center, Ave. Paul Girod, F-73403 Ugine Cedex, France, e-
mail: pierre.barberis@areva.com
4
Senior Expert, AREVA NP GmbH, Paul-Gossen-Str. 100, D-91052 Erlangen, Germany, e-mail: petra-
britt.hoffmann@areva.com
5
Research Engineer, AREVA Inc., 3315 Old Forest Rd., OF70 Lynchburg, VA 24501, United States of America,
e-mail: jacqueline.stevens@areva.com
6
M5TM is a trademark of AREVA NP registered in the United States and in other countries.
7
Hyderabad, India.
Copyright V
other hand, iron content in the range of 300 to 1000 ppm improves the
resistance to corrosion induced by galvanic phenomena. This phenomenon has
been reproduced out-of-pile in galvanic coupling tests in 360 C oxygenated
water. In-pile, the corrosion resistance is equivalent, or even improved, under
demanding conditions when the iron content is increased. There is no adverse
effect of iron content on free growth kinetics and creep behavior. Under loss of
cooling accident (LOCA) conditions, the increase in iron content has no
significant impact on creep–burst resistance and ductility of the cladding. The
high-temperature steam oxidation kinetics, quench behavior, and post-quench
behavior of fresh and pre-hydrided materials are not affected. Finally, because of
the low hydrogen pickup of M5TM in service, which is unchanged by increased
iron content, the high performance of M5TM under reactivity-initiated accident
(RIA) conditions is not impacted.
Keywords
alloy composition, corrosion, dimensional properties, second-phase precipi-
tates, iron, M5TM, creep, mechanical properties, cladding
Introduction
M5TM was developed by AREVA NP to meet burnup extension demands and the use
of high-duty fuel management schemes for lower operating costs, higher flexibility,
and stronger safety margins required by today’s nuclear market. AREVA has a huge
in-pile experience database for M5TM at elevated burnups [1,2]
AREVA NP has successfully integrated customers need into M5TM develop-
ment in terms of:
• high reliability within a wide range of operational and hypothetical accidental
conditions and, more specifically, with respect to fuel assembly design and
cladding material,
• dimensional stability,
• high-corrosion resistance and low-hydrogen pickup,
• improved PCI (pellet-cladding interaction) behavior, and
• compliance with reactivity-initiated accident (RIA) and loss of cooling acci-
dent (LOCA) criteria.
Currently, the iron content in M5TM is targeted at 350/400 ppm. The impact of
the iron content on the properties of M5TM has been investigated by an extensive
R&D program performed both out-of-pile and in-pile with the help of specially
made industrial and experimental ingots covering a range of iron concentrations
from 150 to 1300 ppm.
In this paper, we successively discuss the effect of iron content on:
• the microstructure of the alloy (with particular focus on second phase
precipitates),
• the out-of-pile behavior (mainly corrosion, thermal creep, and mechanical
properties),
KACZOROWSKI ET AL., DOI 10.1520/STP154320120195 161
• the in-pile behavior (corrosion, hydriding, creep, and free growth), and,
finally,
• the behavior under accident conditions (LOCA, RIA).
Note that in the present document, M5TM will be used for material meeting the
current iron specification. Experimental materials, those with iron outside the speci-
fication, will be defined by their iron concentration (in weight ppm or wt. %), or as
M5TM-Fe for the specific ingot with 1000 ppm iron.
Materials and Methods

MATERIALS
Industrial-sized ingots with iron (Fe) content ranging from 140 up to 1000 ppm, as
well as some special ingots with imposed axial gradients of Fe up to 1300 ppm, were
manufactured into test specimens following the standard M5TM industrial process:
ingot forging, quenching, extrusion, several cycles of cold rolling, and annealing at
low temperature (<600 C). The final annealing leads to recrystallization.
OUT-OF-PILE CHARACTERIZATION PROGRAMS

Standard out-of-pile studies were performed to assess the influence of iron content
on M5TM, including:
• microstructural characterization using transition electron microscopy (TEM);
• corrosion tests in 360 C/18.6 MPa water, including elevated lithium (70 ppm)
and galvanic coupling;
• corrosion tests in steam at 415 C/10.5 MPa;
• biaxial thermal creep tests at 400 C, with two levels of hoop stresses (130 and
160 MPa); and
• tensile tests at room temperature and elevated temperature.
IN-PILE AND HOT CELL CHARACTERIZATION PROGRAM

M5TM experience covers a wide range of operating conditions, including: all fuel as-
sembly arrays used in pressurized water reactors (PWRs) (i.e., 14 14 through
18 18), cladding with different wall thicknesses, UO2 as well as MOX pellets, dif-
ferent coolant chemistry regimes, different cycle lengths, and different fuel duties.
The data also covers a wide range of thermal-hydraulic conditions. Thus, the accu-
mulated experience is now large and provides a reference database to assess the
effect of iron.
An experimental alloy referred to as M5TM-Fe, with 1000 ppm Fe, was manu-
factured into cladding tubes and irradiated in standard French and high duty Ger-
man plants in lead test assemblies. The properties of the cladding (corrosion,
hydriding, diametral creep, and fuel rod growth) were monitored.
On-site oxide thicknesses were measured by eddy current. Rod lengths were
deduced from the distances between nozzles and the lower and upper gaps or were
obtained on extracted fuel rods by comparison against a standard. Fuel rod
diameters were measured using linear variable displacement transducers. Various

test rods have also been irradiated to study irradiation growth, creep, and corrosion.
Test rods made of empty tubes were used to study free growth, whereas sealed clad-
ding segments and specific corrosion specimens were inserted into guide tubes of
selected fuel assemblies. The evaluation of length and diameter changes using eddy
current and diameter gauge, respectively, were performed annually. Corrosion sam-
ples were analyzed in hot cells at end of life.
In addition, non-destructive and destructive characterization to measure length,
diameter, oxide thickness, and hydrogen concentration of some fuel rods were per-
formed in hot cells.
Results
MICROSTRUCTURE AND SECOND-PHASE PARTICLES
Manufacture of Alloy M5TM according to the low-temperature processing route (in-
termediate and final annealing at temperature <600 C) leads to a final product
with a fully recrystallized microstructure, characterized by the presence of an a-Zr
matrix with two types of uniformly distributed phases: body-centered cubic b-Nb
precipitates and hexagonal Laves phase (Zr(Nb,Fe)2) intermetallic precipitates. Fig-
ure 1 provides TEM observations for samples with three different levels of iron (330,
412, and 900 ppm). The size of second phase precipitates are unaffected by the iron
content, as is the a-Zr matrix grain size.
The b-Nb precipitates are about 50 nm in diameter, whereas the Zr(Nb,Fe)2
Laves phase precipitates can reach about 100 nm in diameter. As the latter are
much less numerous, the mean second phase particles (SPP) size does not evolve
significantly with the iron content.
CORROSION
Long-Term Tests at 360 C
Corrosion tests were performed in simulated primary coolant water at 360 C/
18.6 MPa with 650 ppm of boron and 1.5 ppm of lithium. Several interims of expo-
sure were performed leading to a total exposure of 890 days. The oxidation kinetics
for various iron contents from 273 up to 1330 ppm are given in Fig. 2. The corrosion
phenomenon follows classical pre- and post-transition kinetics with a transition
thickness near 4.5 lm independent of iron concentration. The time at transition,
however, appears to be somewhat dependent on iron content based on the shift in
the post-transition curves.
Long-Term Tests at 415 C

The long-term uniform corrosion results in steam after 850 days at 415 C are pro-
vided in Fig. 3. A positive effect of iron concentration on corrosion resistance is
observed starting at 300 ppm with no further benefit of higher iron concentrations
up to 1330 ppm.
FIG. 1 Microstructure versus iron content (a) Fe ¼ 330 ppm, (b) Fe ¼ 412 ppm, and (c)
Fe ¼ 900 ppm.
FIG. 2 Influence of iron concentration on long-term corrosion in 360 C water.
Corrosion in Lithium-Containing Water

Autoclave corrosion tests in 360 C water with 70 ppm Li were run on cladding
tubes with iron contents between 273 and 1330 ppm. This condition, although
deemed to be unrepresentative of reactor conditions, is commonly used in the study
of zirconium alloy corrosion performance [4]. Figure 4 shows that the addition of
iron (from 300 to 1300 ppm) sharply improves its corrosion resistance in a
lithium-bearing environment by delaying the kinetic acceleration.
Galvanic Coupling Behavior

The corrosion behavior of Zr alloys with 1 % Nb by weight is sensitive to the pres-
ence of dissolved oxygen in the water [4]. Some autoclave tests were performed at
FIG. 3 Influence of iron concentration on long-term corrosion in 415 C steam.

FIG. 4 Influence of iron concentration on corrosion in lithium-bearing water at 360 C

and 70 ppm of lithium.
360 C in the presence of dissolved oxygen at a level of 100 lg/l. The oxide thick-
nesses observed were 1.5 times higher compared to reference conditions without
dissolved oxygen (presented previously). The iron concentration was found to have
no influence (Fig. 5).
This sensitivity to the presence of oxygen is increased when a galvanic cell is
formed with more noble metals such as stainless steel or nickel-based alloys.
FIG. 5 Influence of iron concentration in 360 C autoclave test after 360 days on the
corrosion in presence of dissolved oxygen (100 lg/l) and galvanic coupling
compared to reference conditions without oxygen or in presence of oxygen and
no coupling.
In the same corrosion test with dissolved oxygen, some samples were mechani-
cally connected to nickel-based alloys to form a galvanic cell. The electrical contact
in this configuration is very good compared to more realistic conditions encoun-
tered in-pile, where a partially insulating oxide is still present at the point of contact
[5]. The oxide thickness was found to be up to 3 times higher than only because of
the presence of oxygen, which means that it can be up to 4.5 times higher than
under the reference condition presented previously. The iron concentration was
found to have a beneficial impact, with the oxide thickness ratio measured on these
coupled samples decreasing with increasing iron concentration in M5TM up to
450 ppm. A significant improvement in the behavior of the alloy was observed
when the iron content was increased from 150 ppm to 450 ppm followed by a
quasi-plateau up to 1000 ppm.
TEXTURE
The basal plane Kearns factors, measured at mid-thickness on cladding tubes (Table 1),
are not affected by the iron content between 240 and 970 ppm.
MECHANICAL PROPERTIES
The tensile properties at room temperature for as-fabricated cladding tubes show a
slight hardening trend with iron content as it increases from 140 and 1000 ppm,
with little effect on ductility. However, the effect of iron remains slight and most
data fall within the range of scatter of industrially processed M5TM at both ambient
and elevated temperature (400 C) (Figs. 6–8).
CREEP
Thermal creep tests lasting 240 h at 400 C were performed at two hoop stress levels
(130 and 160 MPa) on cladding tubes for which the iron content varied from 150
to 900 ppm. The results of these tests (Fig. 9) show that the diametral strain meas-
ured on cladding tubes increases slightly with the iron content between 150 and
900 ppm for the tests performed at the highest stress level of 160 MPa, nevertheless
remaining within the range of scatter for industrially processed M5TM.
TABLE 1 Basal plane Kearns factors for various iron concentrations in cladding tube lots.
Fe (ppm) fr fa ft
240 0.607 0.072 0.321

282 0.616 0.074 0.310
306 0.594 0.071 0.335
322 0.599 0.077 0.324
456 0.581 0.084 0.336
970 0.594 0.087 0.319
FIG. 6 Influence of iron concentration on room temperature tensile properties. YS, Yield
strength; UTS, ultimate tensile strength.
In-Pile Results
CORROSION IN PWR
M5TM is successfully used as cladding material in all types of PWR arrays under a
wide range of thermal-hydraulic conditions including very high duty plants [2,3].
The global PWR corrosion experience of M5TM fuel rods is given in Fig. 10, which
includes data for the corrosion of cladding tubes with an iron content of 1000 ppm
(M5TM-Fe). More than 17000 singular data points are presented in this figure. Fig-
TM
ure 10 includes fuel rods equipped with M5 -Fe claddings containing 1000 ppm
FIG. 7 Influence of iron concentration on elevated temperature tensile properties. YS,

Yield strength; UTS, ultimate tensile strength.
FIG. 8 Influence of iron concentration on room temperature and elevated temperature

tensile properties (total elongation).
iron, irradiated to elevated burn-ups in a 17 17 French reactor and in a high-duty

German 16 16 reactor, and shows that the uniform corrosion resistance is unaf-
fected by, or is even positively affected by such iron content.
Axial oxide profiles have been measured in pool by eddy current for:
• an M5TM rod with an iron concentration of 350 ppm, see Fig. 11, and
• an M5TM-Fe rod with an iron concentration of 1000 ppm, see Fig. 12.
FIG. 9 Thermal creep strain at 400 C, 130 and 160 MPa after 240 h of cladding tubes
versus iron content.
FIG. 10 Corrosion experience of M5TM and M5TM-Fe rods in commercial PWR.
Both rods were irradiated in the same plant during the same cycles.
The axial profiles show an overall increase in oxide thickness with elevation in
accordance with the rod temperature. The small variations or sharp peaks observed
on both profiles are because of the eddy current sensor sensitivity and artifacts.
FIG. 11 Oxide axial profiles measured in hot cell after 55.6 GWd/tU on an M5TM rod.
FIG. 12 Oxide axial profiles measured in hot cell after 51.8 GWd/tU on an M5TM-Fe rod.
The axial measurements confirm the result obtained by individual pool meas-
urements: the maximum oxide thickness of M5TM-Fe is lower and the axial oxide
thickness profile is flatter compared to M5TM.
HYDROGEN PICKUP
Figures 13 and 14 show that the in-service hydrogen uptake is not adversely affected
by iron content up to 1000 ppm. The hydrogen concentration measured on an
M5TM-Fe rod at the hottest span with the highest corrosion is in good agreement
FIG. 13 Hydrogen pick up experience for M5TM and M5TM-Fe in commercial PWR.
FIG. 14 Metallographic observation of hydrides on a M5TM-Fe rod at the hottest span

for three angular positions (fuel rod burnup of 53.9 GWd/tU).
with the global M5TM experience. Few hydrides are observed on this rod, and those
that were present were homogeneously distributed within the wall of the cladding
(Fig. 14).
For this M5TM-Fe rod, the oxide thickness was 10 lm and the calculated hydro-
gen pick up fraction is between 8 % and 9 %, which is consistent with the M5TM ex-
perience [6].
DIAMETRAL CREEP OF CREEP SAMPLES AND CLADDING TUBES IN PWR

Diametral creep tests were performed with standard M5TM and M5TM-Fe with
1000 ppm of iron, in a PWR. Short, sealed cladding tube segments with 1 bar inter-
nal pressure were connected to a material test rod and irradiated in guide tubes of
commercial fuel assemblies. A hoop stress of -113 MPa resulted from the system
pressure at temperatures from 306 C up to 339 C. The creep experiment exhibited
similar diametral creep behavior (Fig. 15) regardless of iron content.
This result is confirmed by measurements performed on actual fuel rods oper-
ated in French reactors and one German reactor for the same two materials as those
tested in the creep experiment. Similar diametral creep behavior of the standard
M5TM rods and 1000 ppm iron M5TM-Fe rods was observed at the same initial in-
ternal pressure of 16 MPa (Fig. 16) prior to contact between cladding and fuel.
ROD GROWTH IN PWR

As shown in Fig. 17, the axial growth behavior of the M5TM-Fe rods (with 1000 ppm
of Fe) in the French 17 17 reactor Q and in the German 16 16 reactor C are
located within the scatter of global growth experience of M5TM rods. Moreover, as
usual because of slightly different fuel rod design, the 16 16 data is located at the
bottom part of this M5TM experience. For both reactor C and reactor Q, the growth
FIG. 15 Diametral creep strain in PWR. Comparison of M5TM with M5TM-Fe 1000 ppm
of iron.
of the M5TM-Fe rods tends to fall toward the bottom of the individual reactor popu-
lations for standard M5TM.
FREE GROWTH IN PWR

As shown in Fig. 18, no significant impact of iron content up to 1000 ppm is
observed on the M5TM-Fe free growth steady state conditions compared to
FIG. 16 Diametral creep strain in PWR versus iron content—comparison of M5TM with
M5TM-Fe 1000 ppm of iron in French (F) and German (G) reactors.
FIG. 17 M5TM and M5TM-Fe rod growth experience in commercial reactors with a focus
on the impact of iron.
M5TM up to 12 1025 n/m2 (Fig. 18). Beyond this, up to 20 1025 n/m2, no signif-
icant effect of iron content is noticed in the accelerated growth regime either.
Accident Condition Behavior

BEHAVIOR UNDER LOSS OF COOLANT ACCIDENT (LOCA) CONDITIONS
Thermal-Mechanical Behavior
High-temperature LOCA mechanical tests (creep and ramp temperature) were run
between 650 C and 1050 C to cover the three phase domains: a, b, and a þ b with
particular emphasis at the top of the a, domain. Based on this work, it can be stated
FIG. 18 Free growth of M5TM and M5TM-Fe in commercial reactors.

FIG. 19 Total elongation versus temperature for two levels of iron.
that the current M5TM correlations cover the behavior of M5TM with higher iron
contents.
The total elongations measured during the creep tests are given in Fig. 19 for
each temperature. It is noted that the increase in iron content up to 975 ppm does
not impact the ductility or the time to rupture.
For the ramp temperature test, the results at a ramp rate of 25 K/s show that
the increase in iron content up to 945 ppm has no effect on the ductility throughout
the investigated temperature and stress range (Fig. 20).
High-Temperature and Breakaway Oxidation

During a postulated LOCA event, fuel cladding tubes may be exposed to steam at
high temperature until they are quenched by an emergency core cooling system.
FIG. 20 Ramp of temperature at 25 K/s—effect of iron on the stress versus temperature

at rupture.
Numerous studies have demonstrated that at high temperature, the oxidation

kinetics of zirconium alloys is mostly parabolic (or even cubic), and time-
temperature oxidation correlations have been previously derived by other investiga-
tors [7–9]. The oxide formed under such conditions is sub-stoichiometric, black,
and dense. However, in some cases, and especially around 1000 C, a sharp increase
of the oxidation rate can be observed after an incubation period. This increase in
the oxidation rate coincides with a significant hydrogen uptake. This hydrogen
uptake is often used as an early clue that the so-called “breakaway” phenomenon
has occurred.
The increase in the oxidation rate combined with the associated hydrogen
pickup can embrittle the cladding. Numerous factors are assumed to influence the
incubation time and the breakaway oxidation: the oxidation temperature is the
first-order parameter, but also concerned are the chemical composition of the alloy
and/or its surface preparation for example.
To assess the influence of Fe content on the breakaway oxidation phenomenon,
cladding tubes with Fe content of 350 ppm, 195 ppm, and 715 ppm were one-side
oxidized in the DEZIROX 1 CEA device at 1000 C [10–13] for three oxidation
times (3270 s, 5292 s, and 9500 s) and directly quenched in water at room tempera-
ture. Post-oxidation hydrogen content measurement was performed on the speci-
men with the highest weight gain. The results [11] show that:
• The iron content has no impact on the pre-breakaway oxidation kinetics and
the time of occurrence of the breakaway transition (Fig. 21). In the three cases,
the transition occurs at about 5000 s, corresponding to a weight gain of
about 10 mg/cm2/single-side oxidation.
• After the breakaway transition, weight gains and hydrogen uptake data are
scattered (Figs. 21–23).
FIG. 21 Weight gain as a function of oxidation time at 1000 C in steam, comparison

between three iron concentrations.
FIG. 22 Post-oxidation hydrogen content as a function of oxidation time.
In summary, as demonstrated in our test up to 1000 C, an increase in iron con-

tent up to 715 ppm will have no impact on the oxidation kinetics, under LOCA
conditions; therefore, the Baker-Just regulatory correlation can be conservatively
applied [10].
RIA BEHAVIOR
We have shown that the in-service corrosion behavior and hydriding resistance as
well as high temperature ramp ductility of M5TM were unchanged by an increase in
iron content up to 1000 ppm.. Given that these three properties are major factors in
FIG. 23 Post-oxidation hydrogen content as a function of oxidation weight gain.

the performance of an alloy under RIA conditions, an increase in iron up to

700 ppm is expected to have no impact on its performance under RIA conditions.
Discussion
A comprehensive study of the influence of iron on the properties of Zr1 % Nb alloys
has not previously been published in the open literature. In the present paper, we
demonstrate that most of the properties of M5TM are at a minimum independent of
the iron content or even improved by iron additions, such as for corrosion under
very demanding conditions. Particular emphasis was placed on potential micro-
structural modifications that could have their origin in the iron content. To better
understand TEM observations, thermodynamic calculations have been performed.
The pseudo-binary Zr1 % Nb-Fe diagram has been computed with THERMO-
CALC and a database slightly modified (see paper by Barberis et al. in the same
conference) from the ZIRCOBASE [14–17]. This is presented in Fig. 24. Regardless
of iron content, <600 C the stable phases are the hexagonal a-Zr matrix, the
cubic b-Nb, and the hexagonal C14 Laves phase Zr(Nb,Fe)2. There is a slight dis-
crepancy between the computed transus temperature (613 C) and the experimental
one (600 6 5 C).
Except for very low iron content (below about 200 ppm), above this transus
(613 C) temperature, the stable phases are the a-Zr matrix, the cubic b-Zr, and the
C14 Laves phase. It is interesting to note that the C14 Laves phase dissolves at a
temperature that increases with the iron content, reaching about 700 C at
1000 ppm compared to 640 C at 300 ppm. Above this temperature, the alloy con-
tains only the two zirconium phases a-Zr and b-Zr. Iron content higher than
0.25 % is needed for another intermetallic (the cubic (Zr,Nb)4Fe2) to become stable
at the expense of b-Nb.
FIG. 24 Computed pseudo-binary Zr1 % Nb-Fe phase diagram.

FIG. 25 Computed fraction of b-Nb and C14 phase at 580 C in the pseudo binary Zr1 %
Nb-Fe phase diagram.
The computed fraction of the b-Nb and C14 phases at 580 C are plotted in Fig.
25. The b-Nb remains the main precipitated phase up to an iron content of
700 ppm, above which the C14 phase becomes the main one (in terms of mole frac-
tion, though, it is still by far the less numerous precipitate because of its larger size).
Because a first approximation Laves phase density seems the only parameter
that is dependent of the iron level, we can postulate that this precipitate participates
in the better corrosion resistance of M5TM when iron concentration increases. This
point is not demonstrated at the time being and needs deeper investigation. Never-
theless, the positive impact of increasing iron content on the corrosion of M5TM
has already been described and has also been observed for other zirconium alloys
such as Zr-1 % Nb-Sn-Fe [18,19].
Conclusion
An extensive R&D program has been launched by AREVA to fully assess the poten-
tial impact of an iron increase in the M5TM alloy with the objective of defining a
new upper limit at the industrial scale.
Specific laboratory-scale and industrial-sized ingots were fabricated for this
program, with a maximum content studied of 1300 ppm Fe.
For a maximum iron content of 1000 ppm, we observed a slight change in
terms of the number of iron-containing Laves phase precipitates with an increase in
iron content, as expected.
It was also shown that the main out-of-pile and in-pile M5TM properties are
not impacted or are even improved by an increase in the iron content up to
1000 ppm:
• no impact on creep and mechanical properties,
• no impact on hydrogen uptake and growth in reactor,
• equivalent or even improved general corrosion in-pile,

• improvement in resistance to galvanic corrosion,
• improvement in corrosion resistance under very demanding conditions, and
• no expected impact on accident behavior (LOCA and RIA) and licensing.
As equivalent, or even improved, global in-reactor performance and behavior

under accident conditions and no significant change in the microstructure in M5TM
(see Discussion) were observed, an increase in the iron content for M5TM is under-
way without any anticipated impact on licensing.
References
[1] Mardon, J. P., Charquet, D., and Senevat J., “Influence of Composition and Fabrication
Process on Out-of-Pile and In-Pile Properties of M5TM Alloy,” Zirconium in the Nuclear
Industry: Twelfth International Symposium, ASTM STP 1354, G. P. Sabol and G. D. Moan,
Eds., ASTM International, West Conshohocken, PA, 2000, pp. 505–524.
[2] Thomazet, J., Dalmais, A., Bossis, P., Godlewski, J., Blat, M., and Miquet, A., “The Corro-
sion of the Alloy M5TM: An Overview,” IAEA Technical Committee Meeting on Behavior of
High Corrosion Resistance Zr-Based Alloys, International Atomic Energy Agency, Buenos
Aires, Argentina, Oct 24–28, 2005.
[3] Bossis, P., Pêcheur, D., Hanifi, K., Thomazet, J., and Blat, M., “Comparison of the High
Burn-Up Corrosion on M5TM and Low Tin Zircaloy-4,” Zirconium in the Nuclear Industry:
Fourteenth International Symposium, ASTM STP 1467, P. Rudling and B. Kammenzind,
[4] “Waterside Corrosion of Zirconium Alloys in Nuclear Power Plants,” IAEA TECDOC 996,
International Atomic Energy Agency, Vienna, Austria.
[5] Howlader, M. M. R., Kinoshita, C., Shiiyama, K., Kutsuwada, M., and Inagaki, M., “In Situ
Measurement of Electrical Conductivity of Zircaloy Oxides and Their Formation Mecha-
nism under Electron Irradiation,” J. Nucl. Mater., Vol. 265, 1999, pp. 100–107.
[6] Kaczorowski, D., Chabretou, V., Thomazet, J., Hoffmann, P. B., Sell, H. J., and Garner, G.,
“Corrosion Behaviour of M5TM: Experience Feedback and Understanding of Phenomeno-
logical Aspects,” Water Reactor Fuel Performance Meeting, Seoul Korea, Oct 19–23,
2008, Paper ID 8102, ANS.
[7] Bossis, P., Verhaege, B., Doriot, S., Gilbon, D., Chabretou, V., Dalmais, A., Mardon, J. P.,
Blat, M., and Miquet, A., “In PWR, Comprehensive Study of High Burn Up Corrosion and
Growth Behavior of M5TM and Recrystallised Low Tin Zircaloy-4,” J. ASTM Int., Vol. 6,
No. 2, 2009, Paper ID JAI101314.
[8] Baker, L. and Just, L. C., “Studies of Metal-Water Reactions at High Temperatures. III: Ex-
perimental and Theoretical Studies of the Zirconium-Water Reaction,” Report No. ANL
6548, Argonne National Laboratory, Lemont, IL, 1962.
[9] Pawel, R. E., Cathcart, J. V., and McKee, R. A., “The Kinetics of Oxidation of Zircaloy-4 in
Steam at High Temperatures,” Electrochem. Sci. Technol., Vol. 126, No. 7, 1979, pp.
1105–1111.
[10] Mardon, J. P. and Waeckel, N., “Behavior of M5TM Alloy under LOCA Conditions,” TopFuel
2003, Wurzburg, Germany, March 16–19, 2003.
[11] Vandenberghe, V., et al., “Sensitivity to Chemical Composition Variations and Heating/
Oxidation Mode of the Breakaway Oxidation in M5TM Cladding Steam Oxidized at
1000 C (LOCA Conditions),” TopFuel 2012, Manchester, U.K., Sept 2–6, 2012, ENS.
[12] Portier, L., Bredel, T., Brachet, J., Maillot, V., Mardon, J., and Lesbros, A., “Influence of
Long Service Exposures on the Thermal-Mechanical Behaviour of Zy-4 and M5TM Alloys
in LOCA Conditions,” 14th International Symposium on Zirconium in the Nuclear Industry,
ASTM STP 1467, Stockholm, Sweden, June 13–17, 2004.
[13] Le Saux, M., et al., “Influence of Pre-Transient Oxide on LOCA High Temperature Steam
Oxidation and Post-Quench Mechanical Properties of Zircaloy-4 and M5TM Cladding,”
Water Reactor Fuel Performance Meeting, Chengdu, China, Sept 11–14, 2011.
[14] Grandjean, C. and Hache, G., “A State-of-the-Art Review of Past Programmes Devoted
to Fuel Behaviour Under Loss-of-Coolant Conditions. Part 3: Cladding Oxidation. Resist-
ance to Quench and Post-Quench Loads,” Report No. SEMCA 2008-093, IRSN, Fonte-
nay-aux-Roses, France, 2008.
[15] Toffolon-Masclet, C., Barberis, P., Brachet, J., Mardon, J., and Legras, L., “Study of Nb and
Fe Precipitation in a-Phase Temperature Range (400 to 550 C) in Zr-Nb-(Fe-Sn)
Alloys,” ASTM STP 1467, J. ASTM Int., Vol. 2, 2005, pp. 81–101.
Component Zirconium Based Alloys,” ASTM STP 1505, ASTM International, West Con-
[17] Toffolon-Masclet, C., Brachet, J. C., Servant, C., Legras, L., Charquet, D., Barberis, P., and
Mardon, J. P., “Experimental Study and Preliminary Thermodynamic Calculations of the
Pseudo-Ternary Zr-Nb-Fe-(O) System,” Zirconium in the Nuclear Industry: Thirteenth
International Symposium, ASTM STP 1423, G. Moan, and P. Rudling, Eds., ASTM Interna-
[18] Chabretou, V., Hoffmann, P. B., Trapp-Pritsching, S., Garner, G., Barberis, P., and Rebeyr-
olle, V., “Ultra Low Tin Quaternary Alloys In-Pile Performance—Impact of Tin Content on
Corrosion Resistance, Irradiation Growth, and Mechanical Properties,” J. ASTM Int., Vol.
8, No. 5, 2011, Paper ID JAI103013.
[19] Hoffmann, P. B. and Schmidt, C., “Erfahrungen in KKP 2 mit M5TM-Hüllrohren [M5TM Rod
Experience in KKP-2],” Proceedings of the Annual Meeting on Nuclear Technology, Karls-
ruhe, Germany, May 22–24, 2007, p. 436.
DISCUSSION
Questions from Arthur Motta, Penn State University:—What was the lowest
level of Fe you used?
Authors’ Response:—In our experimental out of pile program, the lowest iron
level was 140 ppm.
Do you have any idea as to why the Fe containing M5TM showed a flatter oxide
thickness profile than normal M5TM? The ups and downs are caused by lower tem-
perature caused by better heat transfer at the spacer grids, which should also be
present in high Fe M5TM?
Authors’ Response:—Today we have no explanation concerning our observa-

tions on the measured axial oxide profiles between M5TM and M5-Fe.
Is the hydrogen pickup fraction of high Fe M5 similar to normal M5TM?
Authors’ Response:—The hydrogen pickup fractions for M5TM and M5-Fe are
very similar: 10% for M5 and 9% for M5-Fe.
Do you plan to add more Fe to M5TM?
Authors’ Response:—As equivalent, or even improved, global in-reactor per-

formance and behavior under accident conditions and no significant change in the
microstructure in M5TM were observed, an increase in the iron content for M5TM
could be possible without any anticipated impact on licensing.
Question from Anand Garde, Westinghouse Electric Co.:—Did you observe any
visual appearance of non-uniformity after irradiation of these alloys? Westinghouse
has observed such non-uniformities for high corrosion resistant alloys that is
termed Oxide Surface Peeling.
Authors’ Response:—There is no impact of the iron concentration on the visual

aspect, especially in term of oxide spalling as you mentioned for other alloys.
Questions from K. Kapoor, NFC:—The effect of Fe on corrosion and hydrogen

pick-up in M5TM has been presented up to 1300 ppm Fe and no effect of Fe on
these properties is observed. What is the effect of Fe beyond 1300 ppm?
Authors’ Response:—We have no results beyond 1300 ppm of iron.

What is the source of Fe in the two alloys?
Authors’ Response:—Iron comes from the industrial process and is intentionally

added to reach 1000 ppm for M5-Fe.
What is the effect of size and distribution of Laves phase (containing Fe) on the
corrosion and hydrogen uptake?
Authors’ Response:—The corrosion resistance and hydrogen uptake seem

unchanged or slightly improved.
Question from Martin Steinbrück, Karlsruhe Inst. of Technology:—Have you

seen breakaway oxidation during high-temperature steam oxidation test at
1000 C?
Authors’ Response:—Breakaway oxidation was observed at 1000 C for M5TM

and the transition occurs after about 5000-5500s at this temperature under flowing
steam atmosphere and atmospheric pressure [1].
[1]- -‘Sensitivity to chemical composition variations and heating/oxidation
mode of the breakaway oxidation in M5V cladding steam oxidized at 1000 C
R
(LOCA conditions)’ V.Van denBerghe et al, Top Fuel, 2–6 September 2012,
Manchester, UK
Questions from Ron Adamson, Zircology Plus:—Do you have any in-pile results
for compositions where the C14 precipitate is not present, say around 200 ppm Fe
or below?
Authors’ Response:—As mentioned in the presentation and in the article, the

level of iron for standard M5TM is around 400 ppm. We do not have results where
C14 is not present.
How was the in-pile creep measured?
Authors’ Response:—Fuel rod diameters are measured using linear variable dis-
placement transducers (LVDT)
Questions from N. Ramasubramanian, ECCATEC Inc. Canada:—Comment:

The state of iron in un-irradiated and irradiated alloy is different. Under irradiation
iron moves away from Laves phase and is dispersed in the alloy matrix so the
microchemical property with respect to iron in irradiated alloy is different from
that in un-irradiated alloy. Comparison of M5TM with M5 Fe for corrosion, hydro-
gen pick-up, LOCA, etc., needs to be kept in two separate compartments - un-irra-
diated out-reactor and irradiated in-reactor conditions. Mechanisms under the two
conditions influencing the performance could be different.
Authors’ Response:—We agree with this comment.

Questions from Mirco Grosse, Karlsruhe Inst. of Technology:—What are the
sizes of the two types of precipitates?
Authors’ Response:—Lave phases mean diameter is 100nm and beta-Nb mean

diameter is about 50 nm.
Do they change during irradiation?
Authors’ Response:—The native bNb particles remain fully crystalline even after
irradiation up to 19.5 1025 n/m2. After a dose of 11 1025 n/m2 these native par-
ticles reach the “equilibrium” composition of 55% Nb under irradiation conditions.
Very few Laves phases Zr(Fe,Nb)2 were detected by TEM microanalyses on
M5TM alloy before irradiation and none of them after irradiation , except only rare
micro-crystallized and highly faulted particles appearing after irradiation [1]
[1]- -‘Microstructural evolution of M5TM alloy irradiated in PWRs up to high flu-
ences - Comparison with other Zr base alloys’ S. Doriot et al. 17th International Sym-
posium on Zirconium in the Nuclear Industry, Hyderabad, India, February 3-7, 2013
Do the precipitates survive the oxidation experiments at high temperatures?
Authors’ Response:—We have no data on the evolution of precipitates after

high temperature oxidation.
Questions from R. K. Chaube, NFC Hyderabad:—What is the oxidation rate

change at the weld joint of M5-Fe cladding and end plug?
Authors’ Response:—As these parts of the cladding are out of the heat flux and
their temperature is low compared to the rest of the tube, the oxidation is very lim-
ited. There is no significant difference on the expected oxidation rate of M5 and
M5-Fe in these parts of the cladding.
Questions from Suresh Yagnik, EPRI:—Comment: You showed no impact of Fe

content on out of pile corrosion in 360 C Li/B water tests. By contrast, there is a clear
effect of increasing Fe on in-pile corrosion (the latter being at the lower end of entire
M5 corrosion data). Thus I believe that autoclave testing to replicate or simulate in-
pile corrosion has, once again, remained elusive. Question: What is the motivation
behind this study? Is there a plan to introduce M5-Fe cladding material in the future?
Authors’ Response:—The objective of this work was to fully assess the influence
of iron concentration on the M5TM properties. As you mentioned it is important to
analyses both in pile and out of pile. It is of course easier to start by out of pile tests.
J. C. Brachet,1 P. Olier,2 V. Vandenberghe,2 S. Doriot,2

S. Urvoy,2 D. Hamon,2 T. Guilbert,2 A. Mascaro,2
J. Jourdan,2 C. Toffolon-Masclet,2 M. Tupin,3
B. Bourdiliau,3 C. Raepsaet,4 J. M. Joubert,5
and J. L. Aubin6
Microstructure and Properties

of a Three-Layer Nuclear Fuel
Cladding Prototype Containing
Erbium as a Neutronic Burnable
Poison
Reference
Brachet, J. C., Olier, P., Vandenberghe, V., Doriot, S., Urvoy, S., Hamon, D., Guilbert, T., Mascaro,
A., Jourdan, J., Toffolon-Masclet, C., Tupin, M., Bourdiliau, B., Raepsaet, C., Joubert, J. M., and
Aubin, J. L., “Microstructure and Properties of a Three-Layer Nuclear Fuel Cladding Prototype
Containing Erbium as a Neutronic Burnable Poison,” Zirconium in the Nuclear Industry: 17th
ABSTRACT
To increase cycle length and/or fuel burnup, several theoretical and experimental
studies have been performed at CEA. Among them, prospective neutronic
Manuscript received March 28, 2013; accepted for publication January 19, 2014; published online June 17,
2014.
1
CEA-DEN, Dept. for Nuclear Materials, Section for Applied Metallurgy Research, CEA-Saclay, F-91191
Gif-sur-Yvette Cedex, France (Corresponding author), e-mail: jean-christophe.brachet@cea.fr
2
CEA-DEN, Dept. for Nuclear Materials, Section for Applied Metallurgy Research, CEA-Saclay, F-91191
Gif-sur-Yvette Cedex, France.
3
CEA-DEN, Dept. for Nuclear Materials, Section for Research on Irradiated Materials, CEA-Saclay, F-91191
Gif-sur-Yvette Cedex, France.
4
CEA-DSM, IRAMIS/SIS2M/LEEL, CEA-Saclay, F-91191 Gif-sur-Yvette Cedex, France.
5
Chimie Métallurgique des Terres Rares, ICMPE, CNRS, Université Paris-Est, 2-8 rue H. Dunant, F-94320 Thiais,
France.
6
AREVA-CEZUS, BP21 44560 Paimboeuf, France.
7
Hyderabad, India.
Copyright V
BRACHET ET AL., DOI 10.1520/STP154320130053 185
calculations have shown that the addition of a few weight percents of erbium into
the cladding materials could be a promising alternative to the introduction of the
neutronic poison directly into the nuclear fuel pellets. Thus, fabrication of
homogeneous Zr-Er alloys has been assessed, at least up to 10 wt. % of erbium and,
based on the as-received mechanical properties, an optimum erbium concentration
ranging from 3 to 6 wt. % has been derived. However, because of the high-oxygen
thermodynamic affinity of erbium, thermal treatments have to be controlled during
the fabrication route to limit Er2O3 precipitation and coarsening, which may have
detrimental effects on the ductility/toughness of Zr-Er alloys. In parallel, to get more
fundamental insights into the underlying phase diagrams, thermodynamic studies
have been devoted to experimental assessment and modeling of the Zr-Er-(H-O)
system. Because of the detrimental influence of erbium on the corrosion resistance,
a three-layer sandwich clad prototype has been developed using corrosion-
resistant inner/outer Zr-1Nb layers to protect the intermediate Zr-Er layer from
direct water exposure. Compared to a reference Zr-1Nb(O) alloy that has been
subjected to the same fabrication route, the three-layer clad prototype shows
limited decrease in ductility because of pre-hydriding or after high-temperature
steam oxidation e.g., in the case of a loss-of-coolant accident). Moreover, the
studies performed so far have shown a spectacular hydride trapping capacity of the
intermediate Zr-Er layer both for hydrogen coming from nominal outer corrosion or
because of massive secondary hydriding in case of the direct access of water to the
Zr-Er intermediate layer. Using l-ERDA (elastic recoil detection analysis)
measurements, detailed studies of the hydrogen spatial redistribution upon thermal
cycling has been done. A simple model has been successfully used to characterize
the cooling rate influence on the through-wall clad thickness partitioning of
hydrogen/hydrides between the three layers, after cooling from a temperature
corresponding to full dissolution of hydrides.
Keywords
erbium, neutronic burnable poison, zirconium-erbium-hydrogen-oxygen
phase diagrams, CALPHAD, Thermocalc, Zircobase, hydrogen trapping, three-
layer-clad, l-ERDA
Introduction
The increase of cycle length and fuel burnup are key issues for nuclear light water
reactors (LWRs) to decrease the cost of electricity and reduce the amount of natural
uranium resources used. Specific studies have been initiated to evaluate the fuel
behavior and core loading to achieve burnups higher than 70 GWd/t (average
burnup of the most irradiated assemblies). This strategy necessarily implies an
increase of the initial reactivity reserve. Thus, it implies an ability of the control sys-
tems to counterbalance this reserve at the beginning of the cycle. For that, one can
use burnable (solid) poisons, such as gadolinium or erbium oxides (Gd2O3, Er2O3),
which are usually introduced into the fuel pellets [1–4]. However, this last solution
may induce a degradation of the thermal conductivity of the fuel pellet. It also leads
to a more complex quaternary (U,Pu,Gd/Er)Ox system in case of application to
MOX fuel and, for a given fuel pellet size, decreases the overall uranium quantity
available for the fission reaction. Moreover, when added inside the fuel as oxides,
the burnable poison is generally introduced in a few rods per assembly, which may
induce a heterogeneous power distribution within the LWR core.
The neutron capture reaction caused by erbium is as follows:
167
(1) 68 Er þ10 n !168
68 Er
Equation (1) indicates that, at the beginning of the in-core period of the clad
under neutron flux, the consumable erbium isotope, 167Er transforms to erbium iso-
tope 168Er, which is stable under neutron flux. Thus, no new and potentially delete-
rious chemical element is produced in situ and the concentration of elemental
erbium is kept constant until the in-service end-of-life. It is also necessary to men-
tion that, because of the above reaction, there is no production of detrimental
chemical elements/isotopes regarding the back-end recycling issues.
Thus, it has been proposed to incorporate the solid burnable poison directly into
the zirconium-base cladding tubes [5–13]. Based on preliminary neutronic calcula-
tions [14], introducing a few percents of erbium into the zirconium cladding tubes
seems to be enough to obtain the required properties. Compared to gadolinium, the
consuming rate of erbium as a poison is reduced as burnup proceeds; this characteris-
tic should enable achievement of higher burnup in LWRs. Then, prospective studies
have been developed at CEA, first to address the fabrication issues, and second, to
determine the characterictics of the obtained erbium-containing cladding materials.
Zr-Er-(H,O) Phase Diagrams Experimental

Studies and Modeling
Because of the lack of data in the literature and the potential strong chemical inter-
actions between zirconium, erbium, oxygen, and hydrogen (these two last chemical
elements coming from oxidation and hydrogen uptake in regular or incidental/acci-
dental conditions), several studies have been devoted to assess experimentally and
to derive some modeling of the phase diagrams of the Zr-Er-O-H system [15–19].
For the experimental study of the binary phase diagrams, numerous alloys have
been synthesized as 1–2 g samples by arc melting. Pure de-hafnied zirconium metal
(Van Arkel process) and pure distilled erbium metal were used. Moreover, to study
the ternary Zr-Er-H system, binary samples have been hydrogenated with carefully
controlled hydrogen uptake using a Sieverts’ device to produce ternary Zr-Er-H
compounds. In general, the synthesis of samples containing hydrogen was per-
formed by solid gas reaction at 350 C. The different systems have been assessed
with the CALPHAD method using Thermocalc software [20]. This has led to a
refinement of the Zircobase thermodynamic database dedicated to multi-alloyed
zirconium systems [21]. Furthermore, thanks to ab initio calculations of the stable
and metastable Zr, Er, and Zr-Er hydrides, the ternary Zr-Er-H system could be
reproduced by thermodynamic calculations [22].
Zr-Er BINARY PHASE DIAGRAM

Figure 1 shows the new assessed experimental Zr-Er phase diagram in comparison
with the previously published one. Several significant differences can be observed. For
example, it appears that at low temperatures, the erbium solubility in the a-zirconium
solid solution is highly superior to the one that could be deduced from the previously
published Zr-Er phase diagram. It is obvious that such an evolution may have signifi-
cant consequences on the actual microstructure of zirconium-erbium dilute alloys.
Er-O-Zr TERNARY PHASE DIAGRAM

Few hundreds wt. ppm oxygen are systematically present in zirconium alloys. Fur-
thermore, additional oxygen diffusion into the bulk matrix may occur because of
high-temperature oxidation upon incidental/accidental thermal transients. Thus,
the Zr-Er-O ternary phase diagram has been studied experimentally and thermody-
namically. It is well known that rare-earth chemical elements (e.g., Y, Er) have a
strong thermodynamic affinity with oxygen and are able to form very stable oxides.
Thus, Fig. 2 represents the Er-O-Zr partial isothermal section at both 800 C and
1100 C; from this figure, one can observe a limited co-solubility domain of erbium
and oxygen into the a-zirconium solid solution. This implies that, even at high tem-
peratures, small amounts of erbium alloyed with zirconium should promote
erbium-rich oxide (Er2O3) precipitation, reducing drastically the oxygen solubility
in the a-zirconium matrix (in pure zirconium, the oxygen solubility limit is very
high, up to 28–29 at. %).
FIG. 1 New experimental Zr-Er phase diagram compared to the one previously
published [15,17].
FIG. 2 Er-O-Zr partial isothermal sections at 800 C (a), and 1100 C (b) from [18].
Er-H-Zr TERNARY PHASE DIAGRAM

Nominal in-service corrosion and/or incidental-accidental situations [secondary
hydriding, post-breakaway oxidation conditions in loss-of-coolant accident
(LOCA)] may cause hydrogen uptake. Similarly to oxygen, hydrogen is known to
have a strong thermodynamic affinity with rare earth chemical elements like erbium
or yttrium. For example, from [23], the heat (enthalpy) of formation of ErH2
hydride value is 112 kJ/mol, whereas in the same reference, the enthalpy of forma-
tion of ZrH2 hydride in an a-zirconium matrix ranges between 82 and 94 kJ/
mol. It is obvious from these enthalpy values that erbium-rich hydrides may have a
stronger thermodynamic stability than the zirconium ones, and, thus, because of
in-service or accidental hydrogen uptake, one may anticipate some strong chemical
interactions between erbium and hydrogen in Zr-Er alloys, which should induce
some specific hydride secondary phase precipitation (SPP) reactions. Because of
higher thermodynamic affinity to hydrogen of erbium, in the ternary Zr-Er-H sys-
tem, the existing temperature range of erbium-containing hydrides is expected to
increase as compared to the binary Zr-H system, as will be shown in the next sec-
tion. Thus, one may assume that introduction of erbium in an a-zirconium matrix
should increase the terminal solid solution (TSS) temperature of (erbium-contain-
ing) hydrides.
However, literature review shows that there is nearly no data on the Er-H-Zr
ternary system and it was then decided to develop an experimental and thermody-
namic study of this system at the CEA in collaboration with CNRS-ICMPE/Paris-
Est University [16].
Figure 3 shows both experimental and calculated Zr-Er-H isothermal sections at
350 C deduced from the work performed so far. Experimental and theoretical
results are described in detail in Refs 16 and 19. From this result, one can conclude
that, for dilute Zr-Er alloys and for hydrogen content typical of hydrided zirconium
base claddings (i.e., a few at. % of hydrogen), two distinct hydride phases may pre-
cipitate in equilibrium conditions as a function of the relative Zr/Er/H ratio, that is:
d ðZr; ErÞHð2xÞ ðfccÞ phase

b ðEr; ZrÞHð2þ=xÞ ðfccÞ phase
It should be mentioned that even if these two hydrides have similar crystallographic
structures and present extensive mutual solubilities, a miscibility gap exists between
them.
Study of “Homogeneous” Semi-Industrial

Zr-Er-(O,Nb) Alloys
FABRICATION ROUTES
As mentioned previously, calculations performed at CEA [14] have demonstrated
that with only a few weight percents of erbium in addition to a zirconium base ma-
trix, neutronic properties of the cladding can allow a significant increase of the dis-
charge burnup. Thus, different alloys have been produced as thin 1-mm-thick
sheets, containing erbium ranging from 1.5 up to 10 wt. % [1] and low-alloyed
zirconium “Zr(D)” of industrial purity (i.e., zirconium “grade D” from CEZUS). A
few alloys based on a Zr1Nb(O) matrix have been also produced to check the
FIG. 3 Experimental and calculated Er-H-Zr isothermal sections at 350 C, from [16,20].
potential influence of additional alloying elements on the final properties. Two

main fabrication processes have been used:
– Arc melting in copper-base crucibles (100–200 -g ingots), followed by hot and
final cold rolling performed at CEA/SRMA/LTMEX laboratory; an illustration
of the final sheet obtained for a Zr-10 wt. %Er is shown in Fig. 4(a); and
– Induction heating [“skulls” products, 1–2-kg ingots, see Fig. 4(b)], followed by
hot forging and final cold rolling performed at Areva/CEZUS-Ugine laboratory.
FIG. 4 Optical macrographs of homogeneous semi-industrial Zr-Er alloys fabricated.
In both cases, a final full-recrystallization thermal treatment has been applied

to the alloys at 580 C for 5 h.
AS-RECEIVED MICROSTRUCTURE
As-cast microstructure of dilute Zr-Er-(Nb,O) alloys show good homogeneity, but
after high-temperature homogenization treatment in the b-phase temperature
range (T > 1000 C), an heterogeneity of the erbium distribution appears on EPMA
profiles as shown on Fig. 5.
Indeed, coarse erbium oxides precipitated during high-temperature thermal
tratments. This precipitation can be attributed to the amount of oxygen in solid so-
lution (between 500 and 1500 wt. ppm) and to the high thermodynamic stabil-
ity of erbium oxides. The erbium-rich oxide precipitates have been extracted on
carbon extractive replica [24]. EDS microanalysis and electron diffraction by TEM
on these coarse precipitates have shown that these SPPs constitute Er2O3 oxides,
with negligible amounts of zirconium and/or niobium in substitution to erbium.
Discussion
These results agree well with the Er-O-Zr phase diagram. As shown in Fig. 2(b), at
1100 C, even reduced amounts of erbium and oxygen promote Er2O3 precipitation
in the b-zirconium matrix. These observations are also consistent with several pre-
vious studies performed on titanium or zirconium alloys containing yttrium or er-
bium [25–27]. For example, in zirconium alloys, Wilkins and Chater [25] studied
Y2O3 [2] precipitation kinetics in the 800 C–1600 C temperature range. They came
to the conclusion that the Y2O3 precipitates formed by internal oxidation of their
dilute Zr-Y alloys display typical sizes ranging from 0.2 up to 5 lm in diameter,
which are quite consistent with the typical sizes of Er2O3 particles observed here—
and thus, were too large to strengthen the alloy. In the present study, it has been
FIG. 5 Electron-probe microanalysis (EPMA) erbium profiles throughout sheet

thickness of as-cast (a), and b-treated (b) Zr-1Nb-1.5Er-(O) alloy.
sometimes observed that Er2O3 precipitation induces softening of the materials,

because of trapping of the oxygen atoms by the erbium-rich oxides from the oxygen
supersaturated as-cast solid solution; thus, decreasing the well-known oxygen solid-
solution hardening of the a-zirconium matrix. These last observations confirm that
coarse and interface incoherent Er2O3 precipitates have limited strengthening
effects on the materials.
Finally, one has to take into account the potential detrimental effects of the
coarser Er2O3 precipitates on the mechanical properties (ductility, toughness). Coars-
ening of erbium oxide can be avoided by reducing the time spent at high temperature
during the thermal treatments. Figure 6 shows the microstructure thus obtained on a
Zr-1Nb-1.5Er-(O) alloy. The Er2O3 precipitate size reduction is obvious.
AS-RECEIVED MECHANICAL (TENSILE) PROPERTIES

Uniaxial tensile tests were performed at both room temperature (RT) and at 325 C
(close to in-service temperature). A flat specimen of 2-mm width, 1-mm thickness,
and 8-mm gauge length has been used. The applied relative strain rate was 1.5 %
min1. The yield strength (YS), ultimate tensile strength (UTS), uniform elongation
(UE), and total elongation (TE) have been deduced from the engineering stress–-
strain curves and are given in Table 1.
The tensile characteristics of the different Zr(D)-Er alloys as a function of the
erbium content are then compared to those of the Zr1Nb(O) reference alloy that
has been subjected to the same fabrication route (Fig. 7 and Fig. 8). These results
show that the introduction of erbium in low-alloyed zirconium base materials (zirco-
nium “grade D”) induces a significant strengthening and a decrease in the ductility at
both RT and in-service temperature. When compared to the Zr1Nb(O) reference
alloy, it seems that there is an optimum of the strength/ductility ratio for an opti-
mized erbium content ranging from 3 wt. % up to 6 wt. %. For example, at 325 C,
both strength and ductility tensile characteristics of Zr(D)-3 %Er alloy are very close
to those of the Zr1Nb(O) reference alloy. As discussed previously and because of the
low content of oxygen in Zr(D) materials (500–700 wt. ppm), only a limited frac-
tion of the total erbium content (i.e., less than 15 % of the overall erbium content) is
precipitated as Er2O3 particles. Thus, the observed erbium effect on the mechanical
tensile properties can be reasonably attributed to zirconium-a solid solution strength-
ening by the dissolved erbium atoms. This solid-solution-strengthening mechanism
appears to be more effective at 325 C than at RT, and one may anticipate a significant
impact of erbium addition on zirconium base alloys’ creep properties.
PARTIAL CONCLUSIONS ON HOMOGENEOUS SEMI-INDUSTRIAL Zr-Er

ALLOYS
The work performed has shown that fabrication of homogeneous Zr-Er alloys by a
conventional process is possible, at least up to 10 wt. % erbium in addition to
low-alloyed Zr(D) materials. For the alloys for which appropriate intermediate ther-
mal treatments have been applied to avoid coarse Er2O3 precipitation, tensile tests
FIG. 6 TEM micrographs on thin foils of as-received Zr-1Nb-1.5Er-(O) microstructures

showing the Er2O3 precipitates refinement thanks to the application of
optimized thermal treatments (Th.T) upon the fabrication route (top ¼ non-
optimized Th.T and bottom ¼ optimized Th.T).
have highlighted that there was an optimum erbium content range from 3 to
6 wt. %. Nevertheless, preliminary autoclave corrosion tests in pressurized water
(PWR relevant chemistry) performed at 350 C have shown very bad corrosion
properties for zirconium-erbium alloys. These observations are consistent with
studies performed in the early 1960 s, which pointed out that rare-earth elements
such as erbium may have detrimental influences on the corrosion properties of zir-
conium alloys [28]. Thus, a three-layer sandwich clad design has been developed to
TABLE 1 Yield strength (YS), ultimate tensile strength (UTS), uniform elongation (UE), and total
elongation (TE) of different Zr(D)-Er alloys as a function of the erbium content for two
uniaxial tensile testing temperatures (de/dt ¼ 1.5 % min1).
wt. %Er YS (Mpa) UTS (Mpa) UE (%) TE (%)
Room temperature
0 50 251 23.0 52.3
0 100 288 30.0 52.0
0 201 291 17.8 55.5
0 196 287 20.1 39.4
0 180 292 21.8 46.7
0 192 321 16.3 38.0
0 191 307 16.6 38.3
2.5 195 366 25.2 42.2
2.5 187 375 26.8 39.6
3 195 366 15.5 42.2
3 187 375 16.5 39.6
3 219 381 15.8 41.6
3 235 380 16.7 39.6
3 234 383 17.0 39.0
4 179 375 15.8 37.2
4 174 374 15.5 36.2
4 228 395 17.0 39.1
4 227 396 17.2 37.4
5 137 340 18.8 41.5
5 158 361 16.0 35.0
6 245 401 16.7 35.2
6 241 395 17.8 35.4
10 209 374 12.7 21.7
10 221 373 11.1 21.3
protect the intermediate Zr-Er layer from direct exposure to the outer/inner oxidiz-
ing environment. In the present study, conventional corrosion-resistant Zr-1Nb(O)
alloys have been employed as inner and outer protective layers.
Three-Layer Zr1Nb/Zr4Er/Zr1Nb Cladding

A schematic view of the three-layer clad design is shown in Fig. 9.
FABRICATIONS ROUTES
Co-extrusion was the basic process used to fabricate the multi-layered cladding pro-
totypes.8 A Zr1Nb(O) ingot supplied by CEZUS and extruded at CEA was used to
8
Attempts to fabricate a Zr-18 %Er alloy were unsuccessful because of the brittleness of this particular alloy
upon hot rolling.
TABLE 1 (Continued)
wt. %Er YS (Mpa) UTS (Mpa) UE (%) TE (%)
325 C
0 60 114 30.3 68.5
0 105 121 22.2 62.5
0 94 113 23.3 66.6
0 63 123 36.0 65.7
0 55 114 36.0 78.0
2.5 82 161 18.9 48.2
2.5 83 165 17.8 45
3 82 161 32.0 48.2
3 83 164 26.0 45.0
3 117 189 21.5 45.8
3 170 187 19.5 46.6
4 92 202 23.0 33.5
4 96 208 22.5 36.5
4 134 226 17.6 40.3
5 92 202 15 33.5
5 96 208 15.1 36.5
5 74 197 13.2 33.5
5 78 194 12.4 35.5
6 129 231 17.7 31.0
6 132 235 18.1 32.2
10 109 187 10.5 21.9
10 112 212 7.0 17.2
get the inner and outer layers of the three-layer clad. Two fabrication routes have
been investigated so far to produce the Zr-Er layer.
Powder-Metallurgy (PM) Fabrication Route

The first fabrication route performed on fabrication facilities available at CEA was
based on powder-metallurgy (PM) fabrication mode for the Zr-Er alloy. It was pro-
duced from elemental Zr and Er powders by using cold and hot isostatic pressing
followed by machining to the shape of a hollow cylinder. This Zr-Er alloy was
inserted between two thick Zr1Nb(O) hollow cylinders, encapsulated in a copper
can, and then extruded to obtain the three-layer clad. This non-conventional pro-
cess induced a heterogeneous microstructure mainly because of the precipitation of
coarse Er2O3 inclusions.9 This precipitation happened during the PM step as was
already observed in yttrium-containing alloys fabricated by the trough sintering
process. Consequently, the as-received mechanical properties achieved were quite
limited [29–32].
9
Because of analogies in their respective thermodynamic properties, Y2O3 can be considered as a surrogate
for Er2O3 precipitation.
FIG. 7 Tensile characteristics at RT of different Zr(D)-Er alloys as a function of the

erbium content, compared to a Zr1Nb(O) reference alloy.
Semi-Industrial (SI) Fabrication Route

In a second step, it was decided to use a semi-industrial (SI) fabrication route in col-
laboration with AREVA/CEZUS. In this case, the Zr-Er alloy was produced by arc
melting followed by hot and cold rolling to the shape of 2.5-mm-thick sheets, which
were bent to obtain half hollow cylinders. Figure 10 shows the different pieces used
just before the co-extrusion process was performed at the CEA. After the
co-extrusion step, the final cladding geometry (PWR representative inner/outer
diameters and thickness) was achieved by cold-pilgering performed at the AREVA/
CEZUS/Paimboeuf plant. A mono-layer Zr1Nb(O) clad was also produced with
FIG. 8 Tensile characteristics at 325 C of different Zr(D)-Er alloys as a function of the

erbium content, compared to a Zr1Nb(O) reference alloy.
fabrication route similar to reference materials. Finally, a full recrystallization heat

treatment of several hours at 580 C was applied.
AS-RECEIVED MICROSTRUCTURE OF THE SI CLAD PROTOTYPE

Clad Metrology, Layers Thicknesses, and Chemical Homogeneity
The dimensional measurements performed at different axial/circumferential posi-
tions on the final product show a good similarity of the final diameter and thickness
values as shown in Table 2 (note the standard deviation values). The optimum range
FIG. 9 Three-layer clad design schematic view.
for the erbium content was shown in the previous section to be 3–6 wt. %. In the
following, we chose to work on a Zr(D)-4 %Er alloy as an intermediate layer. This
minimizes the mechanical properties differences between intermediate Zr-Er and
inner/outer Zr1Nb layers. This is consistent with the microhardness measurements
obtained on the three different layers, respectively, showing nearly the same average
hardness value (200 HV25 g). Thus, one can assume that slight circumferential
and/or axial variations of the respective layers thicknesses may not induce
FIG. 10 Macrographs of the different semi-products used before hot extrusion for the
“semi-industrial” (SI) fabrication route.
TABLE 2 Average values of the layer thicknesses measured at different axial/circumferential posi-
tions of as-received Zr1Nb/Zr4Er/Zr1Nb clad prototype (SI fabrication route).
Mean Thicknesses (þ Standard Deviation)
Outer Zr1Nb Layer Intermediate Zr-Er Layer Inner Zr1Nb layer (lm) Overall Clad Wall Thickness (lm)
350 (15) 110 (5) 110 (8) 570 (8)
significant local evolution of the clad mechanical properties, as long as there is no

significant circumferential/axial variation of the overall wall clad thickness.
Figure 11 illustrates the typical microstructure obtained. The EPMA x-ray maps
presented on this figure illustrates much higher chemical homogeneity than
observed on the PM clad prototype [29].
Er2O3 Precipitate Sizes Distribution

Thanks to the use of controlled intermediate thermal treatments during the fabrica-
tion route, the as-received Er2O3 particles mean diameter of the SI Zr1Nb/Zr4Er/
Zr1Nb clad prototype is much thinner (see Fig. 12) than the one obtained previously
on the PM clad prototype (see [29] or the top micrograph of Fig. 6 for comparison
purposes), with an average value close to 300 nm.
AS-RECEIVED MECHANICAL PROPERTIES OF PM AND SI CLAD PROTOTYPES

Internal pressure burst tests have been performed on different clad segments taken
from the two types of Zr1Nb/Zr4Er/Zr1Nb clad prototypes. The relative
FIG. 11 As-received microstructure of Zr1Nb/Zr4Er/Zr1Nb clad prototype (SI

fabrication route): Optical micrograph of the clad wall thickness (left), back-
scattered electron (BSE) micrograph (center), and EPMA x-ray qualitative maps
of erbium and niobium (right).
FIG. 12 As-received Er2O3 precipitates typical sizes of the Zr1Nb/Zr4Er/Zr1Nb clad

prototype (SI fabrication route).
deformation rate applied was 2.5 104 s1. Figures 13 and 14 show the yield strength
and total elongation values measured at both RT and 350 C. It is obvious
from these figures and the values reported in Table 3 that the SI clad three-layer pro-
totype displays superior mechanical characteristics than for PM and similar proper-
ties to the reference Zr1Nb(O) clad that has been subjected to the same fabrication
route.
HIGH-TEMPERATURE STEAM OXIDATION AND POST-QUENCHING

MECHANICAL BEHAVIOR
As discussed previously, incursion of Zr-Er alloys at high temperature within the
bZr phase temperature range may induce fast coarsening of the Er2O3 precipitates.
Then the behavior of Zr1Nb/Zr4Er/Zr1Nb clad prototype during hypothetical
LOCA transients at high temperatures is questionable. To address this issue, several
high-temperature (HT) one-side outer steam oxidation tests have been conducted
in the CEA “Dezirox” facility [33,34] at different oxidation temperatures ranging
from 800 C up to 1200 C. Following the HT oxidation tests, post-quenching (PQ)
mechanical clad behavior has been addressed by ring compression and impact test-
ing. First of all, it has to be mentioned that, as expected, there were no significant
differences in the overall oxidation kinetics (i.e., weight gain evolution as a function
of the oxidation time) of Zr1Nb/Zr4Er/Zr1Nb clad prototype in comparison with
the reference Zr1Nb(O) materials. This is not surprising because, as a result of iden-
tical fabrication routes and chemical compositions, there is no reason that the outer
Zr1Nb(O) layer of the three-layer clad prototype should behave differently when
compared to the (mono-layer) Zr1Nb(O) reference alloy.
Figure 15 shows the PQ impact properties evolution as a function of the meas-
ured weight gain after one-side steam oxidation in the 1000 C–1200 C temperature
range. On this figure, we have indicated the conventional LOCA limits for the duc-
tile-to-brittle clad failure mode transition corresponding to an ECR of 17 %, and,
for the particular impact sample geometry used here, an impact energy threshold of
FIG. 13 Yield strength measured from burst tests performed at both RT and 350 C on
the two types of Zr1Nb/Zr4Er/Zr1Nb clad prototypes.
0.05 J/mm2 [34]. It can be observed that the PQ residual toughness of the Zr1Nb/
Zr4Er/Zr1Nb clad prototype is slightly lower than the reference Zr1Nb(O) materi-
als. However, after a LOCA prototypical HT oxidation time of at least 200 s at
1200 C, the RT failure mode of the three-layer prototype clad displays some resid-
ual ductility as highlighted by the SEM fractographs of Fig. 15.
Discussion
The limited but significant reduction in the PQ mechanical properties observed on
the Zr1Nb/Zr4Er/Zr1Nb clad prototype could be partially a result of the Er2O3 pre-
cipitates coarsening as shown in Fig. 16. One can make the assumption that, espe-
cially after oxidation at 1200 C, the coarse Er2O3 inclusions could promote local
stress concentration and associated incipient cracking at the precipitate/matrix
interfaces. Such an explanation has been already proposed by Rezek and Childs to
explain the deterioration in mechanical properties that they have observed on sin-
tered zirconium-yttrium dispersions for a few volume % of yttria additions with
FIG. 14 Total elongation measured from burst tests performed at both RT and at 350 C
and typical macrographs of the ballooned area of the two types of three-layer
clad prototypes.
TABLE 3 Yield strength (YS) and total elongation (TE) deduced from burst tests performed at RT
and at 350 C on as-received materials (de/dt ¼2.5 104 s1).
Room temperature 350 C
YS (Mpa) TE (%) YS (Mpa) TE (%)
Reference Zr1Nb(O) (PM) 572 4.2 211 36.2

577 4.1 217 25.6
Zr1Nb(O)/Zr-4Er/Zr1Nb(O) (PM) 533 4.2 196 1.5
537 5 197 4.4
206 12.2
198 12.2
Reference Zr1Nb(O) (SI) 649 21.3 244 31.8
Zr1Nb(O)/Zr-4Er/Zr1Nb(O) (SI) 621 10.6 250 43.8
715 20.4
FIG. 15 PQ impact properties evolution as a function of the measured weight gain after
one-side steam oxidation at 1000 C or 1200 C, and typical SEM fractographs of
a Zr1Nb/Zr4Er/Zr1Nb clad prototype segment that has experienced HT steam
oxidation for 200 s at 1200 C.
typical particle sizes very close to the Er2O3 particles measured in this study
[30–32].
EFFECTS OF HYDROGEN UPTAKE

Introduction
As highlighted by the underlying zirconium-erbium-hydrogen phase diagram stud-
ies briefly presented and discussed above, the presence of rare earth chemical ele-
ments like erbium in dilute zirconium alloys may have strong chemical interactions
with hydrogen atoms released during the regular and/or incidental/accidental cor-
rosion. In fact, regardless of these basic thermodynamic considerations, there have
been several other attempts in the past to develop hydrogen getters for mitigation
of the hydrogen embrittlement in zirconium-alloy fuel cans [35]. It was attempted
to mitigate the delayed hydride cracking (DHC) phenomena in Zr-2.5Nb pressure
tubes of CANDU reactors [36–41]. But the studies performed so far deal with yt-
trium additions only. For these applications, the choice of yttrium resulted in its
potential high-hydrogen trapping capacity and acceptable macroscopic capture
FIG. 16 Mean Er2O3 diameter evolution after steam oxidation at 800 C, 1000 C, and
1200 C and corresponding typical SEM micrographs.
cross-section for thermal neutrons with respect to neutronic considerations. In

other words, for pressure tube applications, previous authors wanted to develop
hydrogen getter materials with minimum impact on the overall resultant neutronic
properties. Thus, they did not consider chemical rare-earth elements with high-
neutron capture cross-section, as done in the present work. To our knowledge, this
is the first time that addition of erbium to dilute zirconium alloys is studied both as
a neutron burnable solid poison and, as described in the following sections, as a
potential hydrogen getter.
Hydriding Effect for Typical “End-of-Life” Hydrogen Content

as a Result of In-Service Corrosion
Depending on the nuclear fuel cladding materials used and discharge burnup
achieved in a PWR or a BWR, the maximum hydrogen pickup ranges from 100 up
to 700 wt. ppm at the end-of-life of the cladding. The two hydrogen target content
selected are representative of end-of-life: 300–400 wt. ppm and 600–700 wt. ppm.
The SI Zr1Nb/Zr4Er/Zr1Nb clad prototype was hydrided by hydrogen gaseous
charging performed at EDF R&D and at Areva-CEZUS laboratories.
Figure 17 shows optical micrographs of wall clad thickness of as-received hydro-

gen pre-charged materials. It is obvious from this figure that the intermediate Zr-Er
layers of Zr1Nb/Zr4Er/Zr1Nb clad prototypes have trapped most of the available
hydrogen, confirming the high-hydrogen trapping capacity of dilute zirconium
alloys containing small amounts of rare-earth chemical elements such as erbium.
To get some complementary insights into the observed hydrogen/hydrides redis-
tribution within the three-layer clad prototypes, thermal cycling treatments have
been applied, which consist of:
(1) heating at 10 C/min up to 580 C to achieve full dissolution of hydrides; and
(2) cooling down to RT at different cooling rates, ranging from 0.5 C/min up to
100 C/s.
Figure 18 shows optical micrographs of wall clad thickness of 650 wt. ppm pre-
hydrided SI Zr1Nb/Zr4Er/Zr1Nb clad segments that have experienced different cool-
ing rates from 580 C. These micrographs show a strong cooling-rate effect on the
hydrogen/hydrides trapping efficiency of the intermediate Zr-Er layer. For the fastest
cooling rate applied (i.e., 100 C/s), hydrogen does not seem to have enough time to
diffuse from the outer/inner Zr1Nb layers to the intermediate Zr-Er layer. This
FIG. 17 Optical micrographs of wall clad thickness of as-received hydrogen pre-

charged materials: reference Zr1Nb(O) clads (left side), and Zr1Nb/Zr4Er/Zr1Nb
clad prototypes (right side).
FIG. 18 Optical micrographs of wall clad thickness of 650 wt. ppm pre-hydrided SI
Zr1Nb/Zr4Er/Zr1Nb clad segments that have experienced different cooling
rates from 580 C.
induces negligible hydrides precipitation partitioning between the three layers. For
cooling rates lower than 1 C/min, most of the available hydrogen precipitates in
the intermediate Zr-Er layer. This results in the appearance of inner/outer hydride-
depleted layers.
l-ERDA (Elastic Recoil Detection Analysis) Measurements

To quantify more accurately the hydrogen/hydrides partitioning between the three
layers of the Zr1Nb/Zr4Er/Zr1Nb clad prototype as a function of the cooling rate
from 580 C, l-ERDA experiments have been performed thanks to the use of the
nuclear microprobe of DSM/IRAMIS/SIS2M/LEEL, CEA, Saclay, France [42,43].

The nuclear microprobe is equipped with a single-stage 3.75 MV Van de Graff ac-
celerator, which provides a 1Hþ, 3Heþ, 4Heþ beam with an energy ranging from
500 keV to 3.75 MeV, or a 2Hþ beam of 500 keV–2.1 MeV. A selecting magnet
insures the redirection of a monoparticle, monoenergetic beam in one of the two
microfocusing lines. The first line, situated at 90 with regard to the accelerating
section, has been used here. This line is devoted to analysis of un-irradiated materi-
als. The second one, at 45 , situated in a controlled and shielded area, has the
unique capacity to analyze irradiated materials. For un-irradiated materials, the typ-
ical spatial resolution (beam size on the target) of l-ERDA measurements is close
to a surface of 15 lm2. In our study, a 3 MeV 4Heþ incident beam with an intensity
of 800 pA was impinging on the target to be measured, at a glancing angle of 15
with respect to the incident beam direction. By elastic scattering collision, the inci-
dent particles can eject from the target a part of the hydrogen atoms, which are col-
lected in a Si barrier detector, placed at 15 from the incident beam direction and
shielded by a 15 lm aluminum foil. The magnitude of the H signal gives informa-
tion on the H sample content and the energy is related to the depth of the scattering
reaction, therefore, to the depth of the H atom location. Then, by selecting an opti-
mized energy range, it is possible to isolate the “deep” (i.e., bulk) H contribution
and avoid any surface contamination artifact, which takes place frequently on the
surface of zirconium base samples because of in situ or ex situ oxides/hydroxides
nanometers thick formation during the sample pre-polishing/conditioning.
Figure 19 shows the l-ERDA hydrogen maps and the corresponding wall clad
thickness hydrogen profiles thus obtained. It has to be mentioned that, because of
strong local hydrogen variations, the plotted hydrogen content of the intermediate
Zr-Er layer is a rough estimation of the average value. However, these measure-
ments allow the quantifying of the Zr-Er layer hydrogen enrichment and the corre-
sponding hydrogen depletion of the inner and outer Zr1Nb layers. They confirm
the high-hydrogen trapping efficiency of the Zr-Er intermediate layer and its sensi-
tivity to the cooling rate from a temperature where all the initial precipitated
hydrides have been dissolved.
Discussion
The Zr-Er layer hydrogen trapping phenomena observed here is quite similar to the
behavior of a low-alloyed zirconium liner of pre-hydrided BWR type Zircaloy-2
(Zry2) claddings (see Refs 44–46, for example). However, a Zr-Er layer shows an
apparently much higher hydrogen trapping efficiency. Based on BWR-type Zry2/
zirconium-lined claddings deuterium charging and analysis experiments [44],
Takagi et al. [47] have derived a model for hydrogen redistribution in zirconium-
lined fuel claddings. Basically, the hydrogen enrichment of the internal low-alloyed
zirconium layer is assumed to be because of its slightly higher terminal solid solu-
bility (TSS) temperature of hydrogen compared to that of the outer Zry2 layer one.
In this particular case, a difference of 10 C or less is generally considered between
FIG. 19 l-ERDA measurements performed on 650 wt. ppm pre-hydrided SI Zr1Nb/

Zr4Er/Zr1Nb clad segments that have experienced different cooling rates from
580 C: l-ERDA hydrogen maps and associated optical micrographs (left side)
and corresponding wall clad thickness hydrogen profiles (right side); because
of strong local hydrogen variations, the plotted hydrogen content of the
intermediate Zr-Er layer is a rough estimation of the average value.
the TSS temperatures of the two respective materials [45–48]. Then, for a slow cool-
ing rate from a temperature where all the hydrides have been dissolved, precipita-
tion of hydrides occurs earlier in the internal zirconium liner thus inducing a
hydrogen gradient within the outer Zry2 layer. As long as the hydride-precipitation
temperature of the outer Zry2 is not achieved, the hydrogen concentration gradient
induced will produce a net hydrogen-diffusion flux from the outer bulk Zry2 layer
to the zirconium/Zry2 interface (the so-called hydrogen “trapping” or “pumping”
effect). Because this hydrogen trapping effect is a diffusion-controlled mechanism,
it will depend on the cooling rate and especially on the time spent between the two
respective TSS temperatures of both materials.
In our case, the TSS temperature of Zr-Er materials is not known. Then we per-
formed thermodynamic simulations using refined CEA Zircobase database by
introducing zirconium-erbium-hydrogen thermodynamic interaction parameters
(see above). The evolution of the calculated TSS temperature of hydrogen as a func-
tion of the hydrogen content of a Zr-3 %Er alloy is compared to pure zirconium in
Fig. 20. The thermodynamic simulation performed indicates an increase of nearly
100 C of the TSS temperature of hydrogen because of the addition of 3 % erbium
to pure zirconium, which is at least 10 times higher than the reported value for
Zry2/zirconium-lined clad materials! To assess the order of magnitude of this value,
FIG. 20 Evolution of the calculated TSS temperature of hydrogen as a function of the

hydrogen content of a Zr-3 %Er alloy compared to pure zirconium.
differential scanning calorimetry (DSC) experiments have been performed on two

pre-hydrided low-alloyed Zr(D) with and without 3 % of erbium in addition. From
the DSC curves, it was possible to derive the terminal solid solubility temperatures
of hydrogen for hydride dissolution (TSSD) upon heating and hydride precipitation
(TSSP) upon cooling. Details of the experimental DSC curves obtained and of the
TSS determination can be found in [16]. The typical TSS temperatures derived
from the DSC experiments performed at 10 C/min are shown in Table 4. For an
overall hydrogen content of 600 wt. ppm, the values here obtained show an
increase in the TSSP and TSSD temperatures of at least þ20 C and þ35 C, respec-
tively, because of the addition of 3 % of erbium. These experimental values are sig-
nificantly lower than the calculated ones (Fig. 20). However, the DSC results
confirm a potential much higher hydrogen trapping efficiency of the Zr-Er layer
compared to a low-alloyed zirconium liner/Zry2 system, which is illustrated by a
higher TSS temperatures increase in the presence of erbium. This means a higher
thermodynamic stability of erbium-containing d(Zr,Er)H2x hydrides, as shown in
[16], in dilute zirconium-a alloys than conventional nearly pure d(Zr)H2x hydrides.
TABLE 4 Terminal solid solubility temperatures of hydrogen derived from DSC experiments for
hydride dissolution (TSSD) upon heating and hydride precipitation (TSSP) upon cooling
(heating/cooling rate ¼ 10 C/min).
Alloy TSSD ( C) TSSP ( C)
Zr(D) þ 600 wt. ppm H 536 489

Zr(D)-3 %Er þ 600 wt. ppm H 571 507
Note: Details of the measurements can be found in Ref 16.

One can now roughly estimate the hydrogen-diffusion profiles produced in the
pre-hydrided (650 wt. ppm) Zr1Nb/Zr4Er/Zr1Nb clad prototype as a function of
the cooling rate by considering the simplified assumptions below:
– Upon cooling, the time spent between the two respective TSS temperatures for
hydrides precipitation of both materials is given by Eq (2)
(2) t ¼ fhTSSPðZrErÞ hTSSPðZr1NbÞ g=ðdh=dtÞ
where: hTSSP(Zr-Er) and hTSSP(Zr1Nb) ¼ the TSSP temperatures of the inter-

mediate Zr-Er and inner/outer Zr1Nb layers, respectively, and
dh/dt ¼ the cooling rate applied.
– By considering rough discretization of the anisothermal (cooling) temperature
history, the corresponding effective isothermal temperature (Teff) for hydrogen
diffusion is the average value between hTSSP(Zr-Er) and hTSSP(Zr1Nb) (Eq 3):
(3) Teff ¼ ðhTSSPðZrErÞ þ hTSSPðZr1NbÞ Þ=2
– The typical hydrogen-diffusion path (XH) within the inner/outer bulk Zr1Nb
layers from the Zr1Nb/Zr-Er interfaces is given by Eq (4):
(4) XH ¼ 2ðD tÞ1=2
where:
D ¼ the thermal-diffusion coefficient of hydrogen in a-zirconium (Eq 5):
(5) D ¼ 3:2 107 expð4579:77=Teff Þðm2 =sÞðfrom Ref 49Þ
Figure 21 shows the simplified calculations methodology of hydrogen-diffusion

profiles because of the Zr-Er layer trapping effect, for the different cooling rates
studied. To perform the calculations, hTSSP(Zr-Er) and hTSSP(Zr1Nb) temperatures have
been roughly adjusted to obtain a reasonable fit on the experimental hydrogen pro-
files; values of 550 C and 490 C have been considered, respectively, which, how-
ever, are not so far from the experimental ones measured on the Zr(D) base alloys
presented in Table 4. It has to be mentioned that the results are more sensitive to
the hTSSP(Zr-Er) value than to the hTSSP(Zr1Nb) one. Taking into account the strong
approximations used, Fig. 21 shows a fairly good agreement between calculated and
experimental hydrogen profiles. Nevertheless, improvements of these simplified cal-
culations could be done by considering the full Takagi et al. [47] modeling men-
tioned previously. Moreover, for further modeling improvement, Fick’s law for
diffusion can be brought into a more fundamental form by introducing the chemi-
cal potential instead of the actual hydrogen concentrations profiles.
IMPACT OF PRE-HYDRIDING ON MECHANICAL PROPERTIES OF SI

Zr1Nb/Zr4Er/Zr1Nb CLAD PROTOTYPE
The mechanical tests have been performed on as-received hydrogen pre-charged
materials. Taking into account that, after gaseous charging, the specimens
FIG. 21 Simplified calculations of hydrogen-diffusion profiles as a result of the Zr-Er

layer trapping effect—scheme of the calculated hydrogen profiles methodology
(top), and comparison between calculated and experimental hydrogen profiles
for different cooling rates from 580 C (bottom).
experienced slow cooling, most of the hydrogen precipitated in the zirconium er-
bium layer (Fig. 17).
Ring tensile tests have been performed in the circumferential direction of the SI
Zr1Nb/Zr4Er/Zr1Nb clad prototype for the as-received (pre-hydrided) conditions.
A relative straining rate of 3.104 s1 has been applied. Figure 22 shows the engi-
neering mechanical characteristics thus obtained at room temperature (RT) as a
function of the overall hydrogen content and are compared to reference Zr1Nb
claddings that have experienced the same fabrication route and pre-hydriding pro-
cess (see Table 5 and Table 6 for the tabulated values). It can be observed that an
increase of the hydrogen content induces a slight strengthening and a limited
decrease in ductility, except for the Zr1Nb/Zr-Er/Zr1Nb clad prototype that has
been hydrogen pre-charged up to 650 wt. ppm. In this particular case, a more sig-
nificant total elongation (TE) decrease is observed and several post-ring tensile
FIG. 22 Ring tensile mechanical properties at RT of Zr1Nb/Zr4Er/Zr1Nb clad prototype

as a function of the overall hydrogen content, as-received (pre-hydrided)
conditions.
testing fractograph examinations have shown early delamination located at the

Zr1Nb/Zr-Er interfaces, probably because of local hydrides accumulations, as high-
lighted by optical micrographs and corresponding l-ERDA hydrogen maps per-
formed on as-received pre-hydrided samples.
However, the mechanical tests performed here confirmed that fully recrystal-
lized materials are relatively insensitive to pre-hydriding, as far as the overall hydro-
gen content is kept below a certain threshold value. Thus, further work would be
necessary to check the behavior of such pre-hydrided materials after neutron irradi-
ation because of the additional strengthening and potential embrittlement induced
by the irradiation defects (dislocations loops).
TABLE 5 Ring tensile test properties measured at RT of reference Zr1Nb(O) cladding (de/dt
¼ 3.104 s1, SI fabrication route).
Hydrogen Content (wt. ppm) YS (Mpa) UTS (Mpa) UE (%) TE (%)
Reference Zr1Nb(O) <10 437 495 8.6 41.0

cladding <10 443 495 8.8 41.5
<10 446 495 7.8 42.7
<10 448 491 8.3 43.7
<10 445 491 9.2 44.5
<10 444 490 7.9 44.3
350 499 553 7.7 38.3
350 498 555 7.8 38.7
350 493 546 8.2 38.8
650 501 550 8.0 37.1
650 506 555 7.4 37.5
650 507 555 7.1 38.8
TABLE 6 Ring tensile test properties measured at RT of Zr1Nb/Zr4Er/Zr1Nb clad prototype (de/dt
¼ 3.104 s1, SI fabrication route).
Hydrogen Content (wt. ppm) YS (Mpa) UTS (Mpa) UE (%) TE (%)
SI Zr1Nb/Zr4Er/Zr1Nb <10 477 522 7.2 38.7

clad prototype <10 483 523 9.9 37.2
<10 473 522 8.9 35.5
350 443 504 8.0 35.7
350 457 515 8.6 35.2
350 471 522 7.4 33.1
650 475 542 6.4 24.0
650 470 543 9.4 26.0
650 486 541 8.7 24.4
FIG. 23 Scheme of Zr1Nb/Zr4Er/Zr1Nb clad segment ends with direct water access to
the Zr-Er intermediate layer inducing fast local oxidation and potential massive
secondary hydriding.
FIG. 24 PM Zr1Nb/Zr4Er/Zr1Nb clad segment aspects after 10 days autoclave testing at

360 C in pressurized water showing full Zr-Er layer corrosion over 1–2 cm clad
length.
POTENTIAL (INCIDENTAL) MASSIVE SECONDARY HYDRIDING EFFECTS

Autoclave tests have been performed on short clad segments in pressurized water
(180 bar, PWR relevant chemistry) at 360 C. The purpose of these tests was to sim-
ulate incidental through-wall-clad cracking occurrence with direct access of the
water to the intermediate Zr-Er layer, here by direct contact between the clad seg-
ment ends and the pressurized water environment. In such conditions, Zr-Er alloys
would experience very fast oxidation because of their poor corrosion resistance
mentioned previously. One may anticipate huge local hydrogen production and,
consequently, some potential local massive hydrogen uptake as illustrated on the
scheme of Fig. 23. Thus, the anticipated dramatic oxidation and associated
FIG. 25 PM Zr1Nb/Zr4Er/Zr1Nb clad segments aspects after ring compression tests

performed at RT.
secondary hydriding of the Zr-Er layer may impact the resultant mechanical prop-
erties with, potentially, drastic embrittlement of the cladding.
To check the post-autoclave-test mechanical behavior and especially the resid-
ual ductility, ring compression tests (RCT) have been performed on PM Zr1Nb/
Zr4Er/Zr1Nb clad segments. The sample aspects after ring compression tests per-
formed at RT are illustrated in Fig. 24. It can be observed that no through-wall clad
cracking has occurred, and, even after complete corrosion of the intermediate Zr-Er
layer, the inner/outer Zr1Nb layers show a high residual ductility. As the Zr-Er
layer oxidize heavily (Fig. 24), one could expect a very high hydrogen uptake in the
inner and outer Zr-Nb layers. It is thus quite surprising that the overall ductility
should remain so high. For a better understanding of these quite surprising results,
complementary examinations of the clad thickness spatial distribution of oxygen/
oxide and hydrogen/hydrides have been done on the autoclaved PM Zr1Nb/Zr4Er/
Zr1Nb clad segment using EPMA and l-ERDA quantitative mapping. The results
illustrated in Fig. 26 confirm that the Zr-Er layer has been totally corroded with
coarse inclusions of ZrO2 oxides. Moreover, l-ERDA measurements show that
FIG. 26 BSE imaging, EPMA, and l-ERDA mapping of oxygen, erbium, and hydrogen
through the wall clad thickness of autoclaved PM Zr1Nb/Zr4Er/Zr1Nb clad segment.
nearly all of the hydrogen produced by the corrosion reaction is concentrated in the
Zr-Er layer and form very coarse hydrides (RIM) between the coarse ZrO2 inclu-
sions, thus preserving the inner and outer Zr1Nb layers from significant hydrogen
ingress. In other words, for the corrosion conditions tested here, the Zr-Er layer
plays a remarkable “sacrificial” role, avoiding any significant secondary hydriding
embrittlement of the inner/outer Zr1Nb materials, thus preserving more than 80 %
of the clad thickness from significant hydrogen embrittlement.
Conclusions and Recommendations for Further

Work
Prospective studies have been performed at the CEA to evaluate the possibility of
introducing erbium within zirconium nuclear fuel claddings for neutronic purposes.
To get more fundamental insights into the underlying zirconium-erbium-hydro-
gen-oxygen thermodynamic interactions, experimental study and modeling (CAL-
PHAD) of Zr-Er-(O,H) phase diagrams was developed and has been used to refine
the CEA Zircobase thermodynamic database for zirconium alloys.
Fabrication of homogeneous Zr-Er alloys has been assessed at least up to 10 wt.
% of erbium and, based on the as-received mechanical properties, an optimum er-
bium concentration ranging from 3 up to 6 wt. % has been derived. However,
because of the high-oxygen thermodynamic affinity of erbium, thermal treatments
have to be controlled during the fabrication route to limit Er2O3 precipitation and
coarsening, which may have detrimental effect on the ductility/toughness of Zr-Er
alloys.
Because of the detrimental influence of erbium on the corrosion resistance, a
three-layer sandwich clad prototype has been developed with use of corrosion-
resistant inner/outer Zr-1Nb layers to protect the intermediate Zr-Er layer from
direct water access.
Compared to a reference Zr-1Nb(O) alloy that has been subjected to the same
fabrication route, Zr1Nb/Zr4Er/Zr1Nb clad prototype shows limited decrease in duc-
tility because of pre-hydriding or after high-temperature steam oxidation (LOCA).
The studies performed so far have shown a spectacular hydride-trapping
capacity of the intermediate Zr-Er layer both for hydrogen coming from nominal
outer corrosion or because of massive secondary hydriding in case of direct access
of water to the Zr-Er intermediate layer. Thanks to l-ERDA measurements,
detailed studies of the hydrogen spatial redistribution upon thermal cycling have
been done. A very simple modeling has been successfully applied to support the
cooling rate influence on the through-wall clad-thickness partitioning of hydrogen/
hydrides between the three layers, after cooling from a temperature corresponding
to full dissolution of hydrides.
Further works—The behavior under neutron irradiation of Zr-Er alloys and
three-layer Zr1Nb/Zr-Er/Zr1Nb clad prototypes is an open issue. The study of the
following points should be of particular interest:
– The evolution of Er2O3 precipitates: one may speculate some dissolution/re-

precipitation phenomena under irradiation of these uncommon SPPs, as
observed, for example, in oxide-dispersion-strengthened (ODS) steels with
(Y,Ti)2O3 oxide dispersion, or, obversely, accelerated irradiation-assisted
coarsening.
– The post-irradiation behavior of such cladding materials in dry storage or
transportation conditions, i.e., does the intermediate Zr-Er layer act as an effi-
cient hydrogen/hydrides getter; thus preventing the embrittlement as a result
of hydride reorientation or delayed hydride cracking (DHC) phenomena of the
inner/outer Zr1Nb layer? Is the thermodynamic stability of erbium-enriched
hydrides modified by irradiation?
ACKNOWLEDGMENTS
Numerous people took part in this work at the CEA. The writers thank J. L. Béchade,
D. Nunès, C. Bernard, Ph. Bossis, P. Billaud, X. Averty, Q. Auzoux, J. Pegaitaz, C.
Chabert, and M. Auclair from the CEA for their contributions to this work. The writ-
ers are also grateful to M. Blat from EDF R&D, and B. Guerin from AREVA/CEZUS
for performing hydrogen charging, and P. Barberis for zirconium-erbium “skulls”
fabrication.
References
[1] Kim, H. S., Joung, C. Y., Lee, B. H., Kim, S. H., and Sohn, D. S., “Characteristics of GdxMyOz
(M ¼ Ti, Zr, or Al) as a Burnable Absorber,” J. Nucl. Mater., Vol. 372, 2008, pp. 340–349.
[2] Corcoran, E. C., Lewis, B. J., Thompson, W. T., Hood, J., Akbari, F., He, Z., and Reid, P.,
“Computed Phase Equilibria for Burnable Neutron Absorbing Materials for Advanced
Pressurized Heavy Water Reactors,” J. Nucl. Mater., 2008, doi: 10.1016/
j.jnucmat.2008.12.029.
[3] Shimoura, A., Miwa, S., Ushio, T., Hijiya, M., Mori, M., Yamasaki, M., and Takabatake, H.,
“Neutronics Design of an Erbia Bearing Super High-Burnup Fuel Assembly for Pressur-
ized Water Reactors,” Proceedings of the 2005 Water Reactor Fuel Performance Meet-
ing, Track No. 2, Paper No. 1024, Oct. 2-6, 2005, Kyoto, Japan.
[4] Yamanaka, S., Kurosaki, K., Katayama, M., Uno, M., Yamasaki, M., and Kuroishi, T.,
“Development of Erbia-Bearing Super High Burnup Fuel,” Proceedings of the 2008
Water Reactor Fuel Performance Meeting, Seoul, Korea, Oct 19–23, Paper No. 8054,
2008.
[5] L. V. Corsetti and S. R. Pati, “Zirconium Alloy Absorber Layer,” U.S. Patent No. 5,267,290
(Nov 30, 1993).
[6] L. N. Grossman, “Zirconium Alloy Containing Isotopic Erbium,” U.S. Patent No. 5,267,284
(Nov 30, 1993).
[7] S. R. Pati and L. V. Corsetti, “Corrosion Resistant Zirconium Alloy Absorber Material,”
U.S. Patent No. 5,241,571 (Aug 31, 1993).
[8] M. Noe, P. Beslu, J.-C. Brachet, P. Parmentier, and J. Porta, “Alliage de Zr et de Nb com-
prenant de l’erbium comme poison neutronique consommable, son procédé de prépara-
tion et pièce comprenant ledit alliage”, French Patent No. 2,789,404 (May 2, 1999).
[9] D. J. Senor, Johnson, N., Reid, B. D., Love, E. F., Larson, S., and Prichard, A. W.,
“Laminated Rare Earth Structure and Method of Making,” U.S. Patent No. 6,426,476B1
(July 30, 2002).
[10] D. J. Senor, Johnson, N., Reid, B. D., Love, E. F., Larson, S., and Prichard, A. W.,
“Laminated Rare Earth Structure and Method of Making,” U.S. Patent No. 2003/
0,000,926A1 (Jan 2, 2003).
[11] J. C. Brachet, P. Olier, C. Chabert, and S. Urvoy, “Alliage de zirconium comprenant de

l’erbium, ses procédés de préparation et de mise en forme et une pièce de structure
comprenant ledit alliage”, French Patent No. BD 1725 (Oct 16, 2006).
[12] Sridharan, K., “Incorporating of Integral Fuel Burnable Absorbers Boron and Gadolinium
into Zirconium-Alloy Fuel Clad Material,” 2003 Annual Report, Nuclear Energy Research
Initiative (NERI), Washington, D.C., 2003.
[13] Kahambwe, C. K., McDaniels, R. L., Hindin, B., and Grossbeck, M. L., “Investigation of
Zircaloy-4 Doped with Metallic Burnable Poisons,” Trans. Am. Nucl. Soc., Vol. 92, 2005,
pp. 498–499.
[14] Chabert, C., Brachet, J. C., and Olier, P., “Neutronic Study for Introduction of Erbium as a
Burnable Poison into the Fuel Cladding Tube to Enable PWR Core Control,” Proceedings
of ICAPP’08, Anaheim, CA, June 8–12, Paper 8159, 2008.
[15] Jourdan, J., 2010, “CEA Report CEA-R-6238,” Ph.D. thesis, CEA, Gif-sur-Yvette Cedex,
France (in French).
[16] Mascaro, A., 2013, “CEA Report CEA-R-6331,” Ph.D. thesis, CEA, Gif-sur-Yvette Cedex,
France (in French).
[17] Jourdan, J., Toffolon-Masclet, C., and Joubert, J. M., “Experimental Re-Determination
and Thermodynamic Assessment of the Erbium–Zirconium System,” J. Nucl. Mater., Vol.
402, 2010, pp. 102–107.
[18] Mascaro, A., Jourdan, J., Toffolon-Masclet, C., and Joubert, J. M., “Experimental Study of
the Er–Zr–O Ternary System at 800 C and 1100 C,” J. Nucl. Mater., Vol. 427, 2012, pp.
393–395.
[19] Mascaro, A., Toffolon-Masclet, C., Raepsaet, C., and Joubert, J. M., “Experimental Study
and Thermodynamic Assessment of the Erbium–Hydrogen Binary System,” CALPHAD,
Vol. 41, 2013, pp. 50–59.
[20] Sundman, B., Jansson, B., and Andersson, J. O., “The Thermo-Calc Databank System,”
CALPHAD, Vol. 9, 1985, pp. 153–190.
[21] Dupin, N., Ansara, I., Servant, C., Toffolon, C., Lemaignan, C., and Brachet, J. C., “A
Thermodynamic Database for Zirconium Alloys,” J. Nucl. Mater., Vol. 275, 1999, pp.
287–295.
[22] Joubert, J. M. and Crivello, J. C., “Stability of erbium hydrides studied by DFT calcu-
lations”, Int. J. Hydrog. Energy, Vol. 37(5), 2012, pp. 4246–4253.
[23] Griessen, R. and Riesterer, T., “Heat of Formation Models” Hydrogen in Intermetallic
Compounds: I. Topics in Applied Physics, Vol. 63, L. Schlapbach, Ed., Springer,
New York.
[24] Toffolon, C., Brachet, J. C., and Jago, G., “Studies of Second Phase Particles in Different
Zirconium Alloys Using Extractive Carbon Replica and an Electrolytic Anodic Dissolution
Procedure,” J. Nucl. Mater., Vol. 305, 2002, pp. 224–231.
[25] Wilkins, B. J. S. and Chater, J., “The Microstructures and Kinetics of Internally Oxidized
Dilute Zirconium-Yttrium Alloys,” J. Nucl. Mater., Vol. 31, 1969, pp. 288–299.
[26] Kampe, S. L. and Koss, D. A., “Deformation of Rapidly Solidified Dispersion Strength-
ened Titanium Alloys,” Metall. Trans. A, Vol. 20A, 1989, pp. 875–887.
[27] Kral, M. V., Hofmeister, W. H., and Wittig, J. E., “Interphase Boundary Precipitation in a
Ti-1.7 at. %Er Alloy,” Metall. Mater. Trans. A, Vol. 28A, 1997, pp. 2485–2497.
[28] Klepfer, H. H., Douglass, D. L., and Armijo, J. S., “Specific Zirconium Alloy Design Pro-
gram,” First Quarterly Progress Report GEAP-3979, U.S. Atomic Energy Commission,
Washington, D.C., 1962.
[29] Brachet, J. C., Olier, P., Urvoy, S., Vandenberghe, V., and Chabert, C., “Addition of Erbium
as a Neutronic Burnable Poison in Zirconium Base Nuclear Fuel Cladding Tube,” Pro-
ceedings of TopFuel 2009, Paris, Sept 6–10, Paper 2189, 2009.
[30] Rezek, J. and Childs, B. G., “Fabrication and Mechanical Properties of Sintered
Zirconium-Yttria Dispersions,” Report No. AECL-2633, Atomic Energy of Canada Lim-
ited, Ontario, Canada, 1967.
[31] Rezek, J. and Childs, B. G., “Structure and Properties of Yttria-Zirconium Dispersions,” J.
Nucl. Mater., Vol. 26, 1968, pp. 285–299.
[32] Childs, B. G., “Microstructures and Dispersoid Morphology in Yttria-Zirconium Dis-

persions,” Report No. AECL-2798, Atomic Energy of Canada Limited, Ontario, Canada,
1967.
[33] Brachet, J.-C., Pelchat, J., Hamon, D., Maury, R., Jacques, P., and Mardon, J. P.,
“Mechanical Behavior at Room Temperature and Metallurgical Study of Low-Tin Zy-4
and M5 (Zr-NbO) Alloys after Oxidation at 1100 C and Quenching,” Proceedings of the
TCM on Fuel Behavior under Transient and LOCA Conditions, IAEA, Halden, Norway,
Sept 10–14, 2001.
[34] Brachet, J. C., Maillot, V., Portier, L., Gilbon, D., Lesbros, A., Waeckel, N., and Mardon, J.
P., “Hydrogen Content, Pre-Oxidation and Cooling Scenario Influences on Post-Quench
Mechanical Properties of Zy-4 and M5 Alloys in LOCA Conditions—Relationship with the
Post-Quench Microstructure,” J. ASTM Int., Vol. 5(4), 2008, Paper ID: JAI101116.
[35] Spalthoff, W. and Wilhelm, H., “The Use of Hydrogen Getters for Prevention of Hydrogen
Embrittlement in Zirconium-Alloy Fuel Cans,” ASTM STP 458, ASTM International, West
Conshocken, PA, 1969, pp. 338–344.
[36] Coleman, C. E., Sagat, S., and Amouzouvi, K. F., “Control of Microstructure to Increase
the Tolerance of Zirconium Alloys to Hydride Cracking,” Report No. AECL-9524, Pro-
ceedings of the 26th Annual Conference of Metallurgists, Canadian Institute of Mining
and Metallurgy, Winnipeg, Manitoba, Aug 23–26, 1987.
[37] Cann, C. D., Sexton, E. E., Bahumurz, A. A., White, A. J., Balness, H. R., and Ledoux, G. A.,
“Gettering of Hydrogen from Zr-2.5Nb Pressure Tubes,” Proceedings of the International
Symposium on Metal-Hydrogen Systems, Banff, Alberta, Canada, Sept 2–7, 1990.
[38] Amouzouvi, K. F., Cann, C. D., Clegg, L. J., Sexton, E. E., Styles, R. C., and Ellis, R. B., “The
Effect of the Addition of 1.7 at. % Yttrium on the Mechanical Properties of Zr-2.5Nb
Alloy,” Scripta Metall. Mater., Vol. 25, 1991, pp. 1155–1160.
[39] Coleman, C. E., Cheadle, B. A., Cann, C. D., and Theaker, J. R., “Development of Pressure
Tubes with Service Life Greater than 30 Years,” Zirconium in the Nuclear Industry: Elev-
enth International Symposium, ASTM STP 1295, E. R. Bradley and G. P. Sabol, Eds., ASTM
International, West Conshohocken, PA, 1996, pp. 884–898.
[40] Batra, I. S., Singh, R. N., Sengupta, P., Maji, B. C., Madangopal, K., Manikrishna, K. V.,
Tewari, R., and Dey, G. K., “Mitigation of Hydride Embrittlement of Zirconium by
Yttrium,” J. Nucl. Mater., Vol. 389, 2009, pp. 500–503.
[41] Batra, I. S., Singh, R. N., Khandelwal, H. K., Mukherjee, A., Krishnamurthy, N., Gargi, C.,
and Shah, B. K., “Hydride Embrittlement and Oxidation Resistance of Some Zr-Nb-Y
Alloys,” J. Nucl. Mater., Vol. 434(1–3), 2013, pp. 389–394.
[42] Raepsaet, C., Khodja, H., Bossis, Ph., Pipon, Y., and Roudil, D., “Ion Beam Analysis of Ra-
dioactive Samples,” Nucl. Instr. Methods Phys. Res. B, Vol. 267, 2009, pp. 2245–2249.
[43] Raepsaet, C., Bossis, Ph., Hamon, D., Bechade, J. L., and Brachet, J. C., “Quantification
and Local Distribution of Hydrogen within Zircaloy-4 PWR Nuclear Fuel Cladding Tubes
at the Nuclear Microprobe of the Pierre Süe Laboratory from l-ERDA,” Nucl. Instr. Meth-
ods Phys. Res. B, Vol. 266, 2008, pp. 2424–2428.
[44] Takagi, I., Hashizumi, M., Yamagami, A., Maehara, K., and Higashi, K., “Effects of the
Solid–Solid Interface on the Thermal Behavior of Deuterium in Zircaloy Cladding Tubes,”
[45] Une, K. and Ishimoto, S., “Terminal Solid Solubility of Hydrogen in Unalloyed Zirconium
by Differential Scanning Calorimetry,” J. Nucl. Sci. Technol., Vol. 41, No. 9, 2004, pp.
949–952.
[46] Higuchi, T., Sakamoto, K., Etoh, Y., and Nakatsuka, M., “Power Ramp Test Simulation for
Fuel Cladding Tubes by Using Internal Heating and Pressure,” Proceedings of the 2008
Water Reactor Fuel Performance Meeting, Seoul, Korea, Oct 19–22, 2008, Paper No.
8053.
[47] Takagi, I., Shimada, S., Kawasaki, D., and Higashi, K., “Simple Model for Hydrogen Redis-
tribution in Zirconium-Lined Fuel Claddings,” J. Nucl. Sci. Technol., Vol. 39(1), 2002, pp.
71–75.
[48] Kearns, J. J., “Terminal Solubility and Partitioning of Hydrogen in the Alpha Phase of Zir-
conium, Zircaloy-2 and Zircaloy-4,” J. Nucl. Mater., Vol. 22, 1967, pp. 292–303.
[49] Someno, M., Nippon Kinzoku Gakkai-shi, 24, 249 (1960), [in Japanese].
DISCUSSION
Questions from Arthur Motta, Penn State University:—From a neutronics point
of view, a consumable neutron poison needs to change its concentration with irra-
diation dose, which should be proportional to the dose. How do you deal with such
large Er concentration variations? What is the nuclear reaction undergone by Er
when exposed to neutron irradiation? What concentration of Er would be needed
from a neutronics point of view to achieve criticality at the beginning of the cycle?
Authors’ Response:—We give in the paper the typical nuclear reaction under-
gone by the erbium efficient isotope when exposed to neutron irradiation in LWRs
(eq. 1). In fact, as discussed in the paper, equation 1 indicates that the consumable
erbium isotope transforms by radiative capture into another erbium isotope; no
new and potentially deleterious chemical element is produced in-situ and the con-
centration of elemental erbium is kept constant until the in-service end-of-life. The
only modification is the isotopic composition of erbium. Concerning the question
of “criticality at the beginning of the cycle”, one has to refer to [C. Chabert, et al.,
“Neutronic study for introduction of Erbium as a burnable poison into the fuel clad-
ding tube to enable PWR core control”, Proceedings of ICAPP ’08, Paper 8159,
Anaheim, CA USA, (June 8–12, 2008)] to have more details on the neutronics
calculations performed so far which support the Zr-Er clad design studied here.
Question from Javier Romero, Westinghouse Electric Co.:—How does the Er

layer behave when welded to a conventional Zircaloy end plug?
Authors’ Response:—Welding tests have been performed by AREVA. The Er

doped (3-layers) cladding tube has been welded to a conventional Zircaloy end
plug. AREVA managed the welding process in a way to avoid any disturbance of
the erbium layer although guaranteeing a full penetration.
Questions from Ron Adamson, Zircology Plus:—Can you speculate on the per-
formance in case of having a PCI crack? Is there any effect on overall PCI
resistance?
Authors’ Response:—As suggested in Reference 11 [J.C. Brachet, et al., French

patent: BD 1725 - (October 16, 2006)], one may take benefit from the 3-layer clad
design studied here by using PCI resistant internal liner materials (as done in BWR
lined-clad for example) to get some margin on the overall PCI resistance. However,
in case of having a PCI crack, it is anticipated that the 3-layer clad design studied
here may behave more or less as shown in the paper with the intermediate Zr-Er
layer playing a “sacrificial” role, thus preserving the integrity of the overall clad wall
thickness.
Questions from K. Kapoor, NFC:—What is the morphology and orientation of

the hydrides in the erbium containing layer?
Authors’ Response:—We did not perform detailed (TEM) studies of the hydride
morphology in the Zr-Er intermediate layer of the (pre-hydrided) 3-layer clad pro-
totype. However, on the one hand, from the Zr-Er-H basic thermodynamic studies
discussed in the present paper, one may anticipate that, depending on the relative
Zr/Er/H ratio in the Zr-Er layer, two distinct types of hydrides can be formed:
(Zr,Er)H2-x or _-(Er,Zr)H26x phases. Both have a basic fcc structure. On the other
hand, by analogy with yttrium rich hydrides in a dilute zirconium-niobium matrix
[I.S. Batra, et al., “Hydride Embrittlement and Oxidation Resistance of Some Zr-Nb-
Y Alloys”, Journal of Nuclear Materials, Volume 434, Issues 1–3, (March 2013),
389–394], one may suspect some differences in their morphology, compared to
more conventional zirconium hydrides.
Questions from Mirco Grosse, Karlsruhe Inst. of Technology:—You have demon-

strated the impressive trapping effect for hydrogen. Does oxygen act in the same
manner?
Authors’ Response:—As derived from the basic thermodynamic studies of the

Zr-Er-O system and as observed in the as-received structure of erbium containing
alloys, erbium has a strong thermodynamic affinity with oxygen thus inducing
internal oxidation with Er2O3 precipitation, even for low oxygen contents. In other
words, erbium addition tends to reduce drastically the oxygen solubility in both
alpha and beta zirconium matrix.
What can you say about swelling of the Er containing layer due to hydrogen
pickup and/or oxidation?
Authors’ Response:—For the hydrogen effect, taking into account the volumet-
ric density of the hydrides and the actual hydrogen content in the Zr-Er layer
(derived from l-ERDA measurements for example), it is possible to estimate the
swelling due to hydride precipitation within the layer. However, the Zr-Er layer
tends to concentrate the hydrogen/hydrides thus inducing hydrogen depletion in
the inner/outer Zr-1Nb respective layers (>80% of the overall wall clad thickness).
Then, it is anticipated that, even if an important Zr-Er swelling occurs due to
hydrogen ingress (and potentially associated local strong oxidation), the integrity
of the wall thickness of the 3- layers clad is preserved, as shown in the paper in
case of full corrosion (oxidation+hydriding) of the Zr-Er layer. In other words, in
case of important Zr-Er hydrogen pick-up and swelling, the inner and outer
Zr1Nb layers keep some significant ductility/toughness, due to their low hydrogen
content, and thus are able to mechanically accommodate the internal Zr-Er layer
swelling.
Questions from Zoltán Hózer, MTA EOC, Hungary:—What kind of mechanical

testing did you apply for the ductile/brittle limit, and how did you determine this
threshold?
Authors’ Response:—At CEA, after High Temperature (HT) steam oxidation in

LOCA conditions, the Post-Quenching (PQ) clad mechanical properties are studied
by Ring Compression, 3-Point Bending, and Impact Tests, as described in several
previous CEA papers. From all of these PQ mechanical tests and from additional
fractography analysis, it was possible to define accurately the Ductile-to-Brittle
threshold as a function of the materials type, the Equivalent Cladding Reacted
(ECR), the hydrogen content, the cooling scenario, etc (see for example Reference
34: [J.-C. Brachet, et al., “Hydrogen Content, Pre Oxidation and Cooling Scenario
Influences on Post-Quench Mechanical Properties of Zy-4 and M5 Alloys in LOCA
Conditions - Relationship with the Post-Quench Microstructure”, Journal of ASTM
International, Vol. 5, No. 4, Paper ID JAI101116, (2008)])
P. Barberis,1 M. T. Tran,2 F. Montheillet,2 D. Piot,2

and A. Gaillac1
Characterizing Quenched
Microstructures in Relation
to Processing
Reference
Barberis, P., Tran, M. T., Montheillet, F., Piot, D., and Gaillac, A., “Characterizing Quenched
Microstructures in Relation to Processing,” Zirconium in the Nuclear Industry: 17th
ABSTRACT
Quenching from the b phase occurs at several steps during the processing of
zirconium alloy cladding or flat products, from the ingot melting to the final
quench on some channel material, along with quenching after log forging. The
quenched microstructure is constituted of needles or platelets arranged in parallel
plates or basket-weave microstructures, in Burgers relationships with the former
b phase. Thanks to polarized micrographs or orientation imaging microscopy, a
quantification of the platelet or basket-weave character is derived, based on the
boundary length, normalized by the mean needle size. The number and fraction
of variant orientations in a former b grain is also measured. A thermal model has
been derived to compute the quenching rate in Zircaloy, taking into account not
only the fact that there is an exothermic phase transformation, but also its
dependence on temperature versus the quenching rate itself.
Keywords
quenching, variant selection, image analysis, modeling, EBSD, cooling rate,
microstructure, platelets, basketweave
Manuscript received December 19, 2012; accepted for publication March 14, 2014; published online
September 19, 2014.
1
AREVA/CEZUS Research Center, Ave. Paul Girod, 73403 Ugine Cedex, France.
2
École Nationale Supérieure des Mines de Saint-Étienne (SMS), Laboratoire Claude Goux (CNRS UMR 5146)
158 cours Fauriel, 42023 St Etienne Cedex 2, France.
3
Hyderabad, India.
Copyright V
Nomenclature
a ¼ adjustable parameter for computing the kinetic scaling factor in the trans-
formation equation
aa ¼ lattice parameter in the hexagonal plane of the a phase
ab ¼ lattice parameter of the cubic b phase
Cp ¼ specific heat capacity for a constant pressure, incorporating the allotropic
transformation heat
0
Cp ¼ standard specific heat capacity for a constant pressure without
transformation
ca ¼ lattice parameter along the c-direction perpendicular to the hexagonal
plane of the a phase
E ¼ the square of the norm of the Bain distortion tensor
e ¼ lamella thickness
F ¼ transformation-gradient tensor
fk ¼ volume fraction of the kth variant
I ¼ unit tensor
Ki ¼ Kearns parameter in direction i
k ¼ thermal conductivity or index referring to the numbering of possible
variants
N ¼ total number of variants
RGB ¼ a color scheme defined by the set of its three components, red, green, and
blue
S ¼ surface of acquisition
T ¼ temperature
Teq ¼ equilibrium temperature for the considered b fraction
T0 ¼ adjustable temperature for computing the kinetic scaling factor in the
transformation equation
t ¼ time
U ¼ one of the scalar products of two Bain distortion tensors
u ¼E – U
V ¼ left Cauchy–Green tensor
V ¼ one of the scalar products of two Bain distortion tensors
W ¼ one of the scalar products of two Bain distortion tensors
w ¼E – W
w ¼ stored energy by volume unit
X ¼ position vector at the initial state (b phase)
x ¼ position vector at the final state (a phase)
a ¼ angle between the b cube axes and the hexagonal a axes
b ¼ mass fraction of b phase
DHab ¼ mass enthalpy (heat) of a ! b transformation
db ¼ allowed tolerance around the blue color component
dg ¼ allowed tolerance around the green color component
dr ¼ allowed tolerance around the red color component

e ¼ average strain tensor within an ex-b grain
e ¼ strictly positive eigenvalue of the Bain distortion tensor after the virtual-
variant assumption
eB ¼ Bain distortion tensor
eext ¼ external prescribed strain tensor
ek ¼ Bain distortion tensor of the kth variant (k from 1 to 12)
e1 ¼ eigenvalue of Bain distortion related to the eigenvector V1
e2 ¼ eigenvalue of Bain distortion related to the eigenvector V2
e3 ¼ eigenvalue of Bain distortion related to the eigenvector Z
e== ¼ strain component in the free-surface plane
e? ¼ strain component perpendicular to the free-surface plane
hmin ¼ minimal misorientation between two crystals considering crystal
symmetries
j ¼ effective bulk modulus
k ¼ total grain-boundary length in the image
l ¼ shear modulus
l ¼ effective shear modulus
v ¼E – V
n ¼ kinetic scaling factor in the transformation equation
q ¼ density
rth ¼ thermoelastic stress tensor
r== ¼ thermoelastic stress scalar
v ¼ platelet:basket weave ratio
Introduction
Quenching from the b phase occurs at several steps during the processing of
zirconium-alloy cladding or flat products, from the ingot melting to the final
quench on some channel material, in addition to quenching after log forging or
welding. The term “quenching” for zirconium alloys is in fact usually used for
heating from the room-temperature a phase to the high-temperature b phase and
cooling back to the room-temperature a phase, without necessarily involving a high
cooling rate. This thermal treatment allows the following:
• homogenization of the microstructure by dissolving the second-phase particles
(SPPs) and homogenizing the additional element in solid solution (tin, oxy-
gen), and
• modification of the texture, which becomes pseudo-isotropic after quenching
(that is, Kearns factors are close to 1/3, which ensures isotropy of properties
such as elasticity or irradiation-induced growth, while some strong preferred
orientations are present in the material and plasticity appears anisotropic [1]).
The quenched microstructure is constituted of needles or platelets arranged in

parallel-plate or basket-weave microstructures, in a Burgers relationship with the
former b phase. The cooling rate determines the needle width [2,3] and the SPP
size [4].
During the allotropic phase transformation, one b grain can give birth to 12
equivalent orientations of a grains, called variants, responsible for the texture. Some
authors have noted that the crystallographic texture is strengthened by successive
quenches [5]. Other authors brought to light the fact that with an applied hoop
stress during heating, the texture is weakened after b ! a transformation. In this
case there is no variant selection [6]. Daymond et al. have observed variant selection
during heating as well as during cooling [5]. Recent investigations affirmed that
with an applied tensile or compressive stress during cooling, the variant selection is
effective [7]. However, this did not happen during heating in the same conditions.
In addition to the crystallographic texture, the microstructure morphology
(platelet or basket weave) is important for further deformation and recrystallization.
This morphology has long been evidenced [8], and several parameters have been
shown to influence it: the higher the cooling rate or the amount of chlorine or insol-
uble impurities, the more prominent the basket-weave character [9]. However, the
lack of quantitative measurement impedes accurate studies and process mastery.
Mastering the properties of the final or intermediate products thus calls for an
accurate knowledge and understanding of the following:
• the quenching rate,
• the morphology of the quenched structure (platelet-to-basket-weave ratio),
and
• the crystallographic texture, linked to the variant selection.
Although cooling-rate measurement is often difficult with industrial products,

modeling is a valuable tool for computation, and is rather straightforward. In addi-
tion to the heat equation, the exothermic phase transformation has been taken into
account. The phase transformation depends itself on the cooling rate [10], leading
to strong coupling between the equations.
A quantitative method for determining the platelet-to-basket-weave ratio v has
been developed via image analysis and applied to samples with different quenching
rates.
Finally, we performed a local study to understand the variant selection due to
quenching. In this regard, we quantified all the a fractions inside each b grain
before and after quenching.
Material and Techniques

MATERIAL
Various Zircaloy-4 samples were used in this study with sizes up to several centi-
meters in the lowest dimension in order to allow the growth of b grains. They were
taken from a quenched billet. The experiments conducted to determine and model
TABLE 1 Chemical composition, mass %.
Alloy Sn Fe Cr O
Zircaloy-4 1.35 0.21 0.10 0.13
the quenching rates were performed on samples 40 mm by 40 mm by 90 mm in

size. The nominal chemical composition is given in Table 1.
In some samples, thermocouples were inserted at various depths to follow the
temperature evolution during quenching. Samples were heated into the b phase
under air at temperatures between 1000 C and 1100 C, held at that temperature for
30 s to 10 min, and cooled in air or quenched in water at different temperatures
(typically 10 C to 60 C). These samples were used to check the cooling-rate
modeling and to analyze the influence of the cooling rate on the morphology of a
lamellae. The quenching rates measured at 17 C/s to 100 C/s are directly linked to
the various water temperatures.
Other samples, and specifically those used to follow the variant selection, were
encapsulated in quartz tubes evacuated and sealed under reduced argon pressure,
brought into b phase (typically at 1050 C), and quenched in water after the tube
had been broken. These samples showed a shiny surface after the quench, which
means that the unavoidable oxygen surface contamination was very low and could
be neglected.
QUANTIFICATION OF THE BASKET-WEAVE AND PARALLEL-PLATE

MICROSTRUCTURES
A quantification procedure for the v ratio must be easy for industrial application.
Thus the analysis of optical micrographs in polarized light was chosen, even though
electron back-scattered diffraction (EBSD) could give more accurate information.
Samples were polished with paper grit 1200 and then with colloidal silica (OPS)
for observation in bright field. The mean needle width was measured via the inter-
cept method at a magnification of 1000. An anodic oxidation was then performed
for polarized-light observation. Magnifications between 25 and 200 were used
depending on the quenching rate (and thus on the typical length scale of the
microstructure). Up to 10 images at a given set of parameters (sample, magnifica-
tion, polarizer orientation) were recorded to get not only the average v, but also its
standard deviation (characterizing the microstructure heterogeneity).
The v ratio is defined by
ke
(1) v¼
S
where:
k ¼ total grain-boundary length in the image, lm or square root of pixel,
e ¼ lamella thickness, lm or square root of pixel, and
S ¼ surface of acquisition, coherent area unit (lm2 or pixel).
The v ratio is dimensionless, and its value ranges from 0 (parallel plates) to
1.25 (individual needles) as derived in the Appendix.
The method uses an in-house image-analysis routine to draw the grain bounda-
ries between differently colored a colonies. The image is a raster-graphics image,
also called a bitmap, which means that each image is coded with a rectangular grid
of pixels. Each pixel, located by its coordinates in two dimensions, is associated
with a color in RGB scale (red, green, and blue). The main idea is to check, for each
pixel, the number of the first eight neighbors that have the same RGB value as the
checked pixel plus or minus the allowed tolerance (dr, dg, and db). If there are at
least six neighbors that have the same RGB value, then the pixel is considered as a
pixel within an a colony. It translates into a white pixel in the new image. If not, the
pixel is taken as a grain boundary and becomes a black pixel in the new image. In
the end, a black and white image is obtained from the grain-reconstruction code,
with the grain boundaries in black. One of the main inconveniences of this method
is that sometimes, with a polarized image, there is slight hue variation within a sin-
gle a colony (see zone A in Fig. 1). This variation leads to the detection of a grain
boundary that in reality does not exist. In order to correct this, the tolerances dr, dg,
and db can be increased. However, if that is done, when two a colonies (see zone B
in Fig. 1) have similar RGB values, then the grain boundary between them will not
be detected. Therefore, a compromise must be made. In this image analysis, the
more grain boundaries, the more pronounced the basket-weave character.
The v characterization can also be obtained through EBSD analysis. This is
indeed more accurate because the orientation of each pixel is given. However, the
observed surface is too small to be statistically representative, and the acquisition
time is quite long (approximately 31 h for a 2- to 3-cm2 surface). Nevertheless,
EBSD analysis gives a reference to validate the image treatment and compare the v
values.
FIG. 1 Optical micrograph with polarized light of the quenched Zircaloy-4.

FIG. 2 Ex-b grain in a quenched structure: location of the region of interest.
METHOD TO QUANTIFY VARIANT FRACTIONS IN ONE EX-b GRAIN BEFORE

AND AFTER HEAT TREATMENT
EBSD analysis was used to quantify the fraction of each crystallographic variant
within an ex-b grain. At room temperature, Zircaloy-4 is entirely composed of a
phase. Orientation imaging was mapped automatically in a JEOL 6500 FEG
scanning electron microscope equipped with HKL EBSD software.4 The samples
were 1-cm3 cubes, and the size of the orientation maps was tailored to the ex-b
grain size. The Widmanstätten lamellae had a thickness of about 5 lm, so the
selected step-size for the EBSD acquisition was 3 lm.
The samples were taken from a quenched billet and mechanically polished with
paper grit 2 500. Then they were submitted to electrolytic polishing with a mixed-
acid solution (80 vol. % of acetic acid and 20 vol. % of perchloric acid) for 10 s at
25 V. In order to find the same ex-b grain before and after heat treatment, four
Vickers indentations were made, locating the region-of-interest corners (see Fig. 2).
The procedure consists of quantifying the amount of each variant within one
ex-b grain. The specimen was sealed in an evacuated quartz tube that was then
filled with a reduced argon pressure, after which the sample was heat treated at
1050 C for 10 min and water quenched. Electrolytic polishing was performed on a
section of the sample before EBSD mapping. The specimen was then sealed again in
an evacuated quartz tube filled with a reduced argon pressure, heat treated at
1050 C for 10 min, and finally water quenched. A second quantification of a
variants was done inside the same ex-b grain, which was then a surface grain (see
Fig. 3). This experience was repeated on several grains in order to obtain a fair
statistical study.
The f0001g pole figure (PF) enables the selection of twin variants coming from
the same ð110Þb plane as in Fig. 4(a). The f1120g or f1100g PFs are useful for
separating those twin variants, misorientated at 10.5 around the c-axis [Fig. 4(b)],
4
HKL CHANNEL5, Oxford Instruments plc, Tubney Woods, Abingdon, Oxon OX13 5QX, UK.
FIG. 3 EBSD maps before and after heat treatment. Kn is the Kearns factor in the
normal direction (perpendicular to the surface).
as confirmed by the peak at 10.5 on the misorientation-angle graph [Fig. 4(c)]. The
inverse pole figure (IPF) map is used to observe the lamella morphology and also
their location inside the ex-b grain [Fig. 4(d)].
Modeling
Two models are presented here; the first one computes the cooling rate at each
location in a sample, and the second one deals with variant selection.
COOLING-RATE MODELING
As said before, two equations were implemented in COMSOL Multiphysics.5
5
COMSOL, WTC, 5 Place Robert Schuman, 38000 Grenoble, France.

FIG. 4 (a) {0001} PF. (b) 1120 PF. (c) Misorientation between the two selected variants. (d) IPF map showing the morphology of the two selected
variants. Percentages give the area fraction of the corresponding variants: 39.84 % for the two variants with the selected c-axis, which
decompose in one variant with an area fraction of 32.08 % and in the other with 7.74 %.
BARBERIS ET AL., DOI 10.1520/STP154320120211

233
The first was the heat equation,
@T
(2) qCp þ divðk gradTÞ ¼ 0
@t
where:
q ¼ density, kg/m3,
Cp ¼ specific heat capacity for a constant pressure, J/(kg K),
T ¼ temperature, K,
t ¼ time, s, and
k ¼ thermal conductivity, W/(m K).
More precisely, the heat capacity incorporates the phase-transformation heat
by means of
@b
(3) Cp ¼ Cp0 DHab
@T
where:
Cp ¼ specific heat capacity accounting for the transformation heat, J/(kg K),
0
Cp ¼ standard heat capacity if no transformation occurs, J/(kg K),
DHab ¼ mass enthalpy (heat) of a ! b transformation, J/kg,
b ¼ mass fraction of b phase (dimensionless), and
T ¼ temperature, K.
The second is the phase transformation equation,
@b
(4) ¼ n bð1 bÞ
@t
where:
b ¼ mass fraction of b phase (dimensionless),
t ¼ time, s, and
n ¼ kinetic scaling factor (s–1) given by n ¼ aðT Teq Þ exp½ðTeq TÞ=T0 ;
in which a (s–1 K–1) and T0 (K) stand for two adjustable constants and Teq (K) is
the equilibrium temperature for the considered b fraction.
These equations were slightly modified from equations in Ref 10 in order to
improve the numerical convergence.
VARIANT-SELECTION MODELING
Basic Principle and Assumptions
Upon cooling, b ! a phase transformation takes place according to the well-
known Burgers orientation relationships ð110Þb ==ð0001Þa and ½111b ==½1120a , as
illustrated in Fig. 5 [11]. This transformation is associated with the Bain distortion
tensor eB ¼ V I [12], where I is the unit tensor and
FIG. 5 b-to-a crystallographic relationships; Z is normal to the figure plane.
2 pffiffiffi 3
3 aa
6 pffiffiffi a 0 0 7
6 2 b 7
6 aa 7
(5) V¼6
6 0 0 7 7
6 ab 7
4 1 ca 5
0 0 pffiffiffi
2 ab
where:
V ¼ left Cauchy–Green tensor, expressed in the coordinate system (V1, V2, Z)
as defined in Fig. 5 (dimensionless),
aa ¼ lattice parameter of the a phase in the hexagonal plane, nm,
ab ¼ lattice parameter of the b phase, nm, and
ca ¼ lattice parameter of the a phase perpendicular to the hexagonal plane, nm.
The above definition of eB holds if the elements of V are “not too far” from
unity. For zirconium at the allotropic transformation temperature (863 C),
aa ¼ 0.3248 nm, ca ¼ 0.5198 nm, and ab ¼ 0.3609 nm [12,13], which leads to
2 3 2 3
e1 0 0 0:1022 0 0
(6) eB ¼ 4 0 e2 0 5 4 0 0:1000 0 5
0 0 e3 0 0 0:0184
where:
eB ¼ Bain distortion tensor with coordinate system ðV1 ; V2 ; ZÞ as defined in
Fig. 5 (dimensionless),
e1 ¼ eigenvalue of Bain distortion related to the eigenvector V1 (dimensionless),
e2 ¼ eigenvalue of Bain distortion related to the eigenvector V2 (dimensionless),
and
e3 ¼ eigenvalue of Bain distortion related to the eigenvector Z (dimensionless).
Although their norm, ðe21 þ e22 þ e23 Þ1=2 0:1442, is the same, the Bain
distortion tensors ek of the 12 variants are different when expressed in a common
reference frame, such as the cube axes of the “parent” b grain.
An external uniform strain eext can be prescribed to the material, associated, for
example, with thermoelastic stresses occurring during cooling, such as in the sec-
tion “Variant Selection under Thermoelastic Strain.”
Consider now the set of N variants (N 12) generated from the transformation
of a unique b grain (“parent” grain). The strain of each variant is ek þ eext ; and the
average strain in the former b grain is therefore
X
N
(7) e¼ fk ek þ eext
k¼1
where:
e ¼ average strain tensor within a former b grain (dimensionless),
k ¼ index referring to the numbering of possible variants,
N ¼ total number of possible variants,
fk ¼ volume fraction of the kth variant (dimensionless),
ek ¼ Bain distortion tensor of the kth variant (dimensionless), and
eext ¼ external prescribed strain tensor.
Assuming the transformation takes place instantaneously within the b grain,
the arrangement of variants is likely to minimize the above average strain [14]. This
is tantamount to saying that the variant selection will be predicted by minimization
P
of kek2 with respect to the values of fk under the condition Nk¼1 fk ¼ 1; meaning
that the transformation of b ! a is complete. This assumption can also be inter-
preted in terms of stored energy, as the equation for stored elastic energy associated
with e can be written in the form

1 2l
(8) w ¼ j ðtr eÞ2 þ l kek2
2 3
where:
w ¼ stored energy per unit volume, J/m3 (Pa),
j ¼ effective bulk modulus accounting for viscoplastic relaxations occurring
during the phase transformation, Pa,
l ¼ effective shear modulus accounting for the same relaxations, Pa,
tr ¼ trace operator (sum of the diagonal components), and
e ¼ average strain tensor within a former b grain (dimensionless).
Because tr e ¼ tr ek is the same for any variant combination, minimizing kek2 is
equivalent to minimizing the energy w .
Variant Selection Under Thermoelastic Strain

Results of an investigation conducted in the absence of external strain were partially
published elsewhere [14]. The case of a material undergoing b ! a phase
transformation under thermoelastic strain is now addressed. Figure 5 shows that

two distinct variants, referred to as twin variants, can be associated with each f110g
plane of the b phase according to the sign of the angle a. It can be shown that
a ¼ ½cos1 ð1=3Þ p=3=2 5:264 . To simplify the analytical derivations, six vir-
tual variants are considered here instead of the 12 real variants. A virtual variant is
obtained from any real one by setting a ¼ 0 (Fig. 5). It therefore fulfills the first
Burgers relationship, but only approximately the second. Two twin variants are
obviously associated with the same virtual variant. Moreover, considering that the
components of the Bain tensor verify e3 e1 je2 j; it is now assumed that e3 ¼ 0
and e1 ¼ e2 ¼ e ¼ 0:1: The Bain distortion tensors of the six virtual variants can
then be written in the b-grain reference system,
2 3 2 3
1 0 0 1 0 0
6 7 6 7
e1 ¼ 4 0 1=2 1=2 5e e2 ¼ 4 0 1=2 1=2 5e
0 1=2 1=2 0 1=2 1=2
2 3 2 3
1=2 0 1=2 1=2 0 1=2
6 7 6 7
(9) e3 ¼ 4 0 1 0 5e e4 ¼ 4 0 1 0 5e
1=2 0 1=2 1=2 0 1=2
2 3 2 3
1=2 1=2 0 1=2 1=2 0
6 7 6 7
e5 ¼ 4 1=2 1=2 0 5e e6 ¼ 4 1=2 1=2 0 5e
0 0 1 0 0 1
where:
ek ¼ Bain distortion tensor of the kth virtual variant (k from 1 to 6) expressed
in the reference frame ðX; Y; ZÞ as defined in Fig. 5 (dimensionless), and
e ¼ strictly positive eigenvalue of the Bain distortion tensor after the virtual var-
iant assumption (dimensionless).
An external uniform thermoelastic strain eext developed under cooling is
assumed to superimpose the Bain tensor. More specifically, an isotropic (positive or
negative) plane-stress state occurring at the specimen’s free surface is considered in
the form
2 3
r== 0 0
(10) rth ¼ 4 0 r== 05
0 0 0
where:
rth ¼ thermoelastic stress tensor due to cooling expressed in the reference axes
(s, t, n) [see Fig. 6(a)], Pa, and
r== ¼ thermoelastic stress scalar, Pa.
The associated strain tensor is then derived, assuming a first approximation
that the material is linearly elastic and isotropic.
FIG. 6 Set of reference orthogonal axes (s, t, n) linked to the specimen (a) and
orientation of the unit normal n to the surface of the specimen specified by two
angles h and u with respect to the h100i axes of the parent b grain (b).
2 3
e== 0 0
(11) e ext
¼4 0 e== 0 5
0 0 e?
where:
eext ¼ external thermoelastic strain tensor expressed in the reference frame (s, t,
n) (dimensionless),
e== ¼ strain component in the free-surface plane (dimensionless), and
e? ¼ strain component perpendicular to the free-surface plane (dimensionless).
The orientation of the unit normal n to the specimen surface is specified by
two angles, h and u, with respect to the h100i axes of the parent b grain, as shown
in Fig. 6(b); note that s and t are any couple of perpendicular vectors in the speci-
men surface. The transformation matrix from the b to the specimen axes can then
be written in the form
2 3
sin u cos h cos u sin h cos u
(12) P ¼ 4 cos u cos h sin u sin h sin u 5
0 sin h cos h
where:
P ¼ transformation matrix from the b to the specimen axes (dimensionless),
and
h; u ¼ angles defining the normal to the free surface, measured in degrees or
radians.
The expression of the thermoelastic strain in the b axes is then given by
T
eext
b ¼ Peext
ðs;t;nÞ P .
According to Eq 7, the volume fractions of the N present virtual variants
2
(N 6) are now determined via minimization of kf1 e1 þ þ fN eN þ eext k with
respect to fk. In view of the experimental data (see the section “Results on the a
Variant Fraction Quantification”), the case of N ¼ 6 is investigated here. This leads
to a six-by-six system of linear equations that can be easily solved, taking into
P
account 6k¼1 fk ¼ 1, using the simple relationship e== e? ¼ r== =ð2lÞ.
1 r==
(13) 8k 2 f1 6g; fk ¼ þ Yk ðh; uÞ
6 72le
where Yk is a specific trigonometric function (dimensionless).

When r== vanishes, one gets the trivial and unique solution f1 ¼
¼ f6 ¼ 1=6. However, those solutions are valid only if they are positive. Prescribing
that all values of fk be positive for any h and u angles amounts to defining a
restricted range of variation for the thermoelastic stress r== . It was found that this
pffiffiffiffiffi
domain of validity is ð6eÞ=11 r== =l ð6eÞ=ð1 þ 3 10Þ, or, assuming e ¼ 0:1;
0:055 r== =l 0:057: Such an interval is likely to include the expected thermo-
elastic stresses.
Each variant configuration can be easily characterized by its Kearns factor
associated with the normal n, defined as follows [15,16]: Let wk ¼ ðck ; nÞ denote
the angle between the c-direction of variant k (unit vector ck) and the surface
P
normal n. The Kearns factor Kn is then defined as Kn ¼ 6k¼1 fk cos2 wk : Analyti-
cal expressions for cos2 wk are easily derived. Using the above values of fk then
leads to
1 r==
(14) Kn ¼ þ T ðh; uÞ
3 72le
where T is a dimensionless specific trigonometric function.

Two other Kearns factors, Ks and Kt , can be defined for directions s and t,
respectively. However, it is easy to show that Kn þ Ks þ Kt ¼ 1: Furthermore, in
the present case Ks ¼ Kt for symmetry considerations, which leads to
Ks ¼ Kt ¼ ð1 Kn Þ=2: The value of Kn is related to the texture of the specimen:
high Kn factors mean that the c-axes of the selected variant set are close to the
normal to the specimen surface, whereas low Kn values mean they are close to the
surface plane. Kn factors given by Eq 14 are displayed in the diagrams in Fig. 7 as
functions of the specimen orientation.
Results
COOLING-RATE MODELING
The temperature evolution was recorded at the center and 2 mm from the surface
of samples 40 mm by 40 mm by 90 mm in size quenched in water at either 10 C or
60 C. The experimental and simulated curves are given in Fig. 8. The agreement
is accurate at the center. Toward the sample surface, the curve given by the thermo-
couple is closer to the simulated one 4 mm from the surface, because of the
imperfect heat transfer between the sample and the thermocouple. The phase-
transformation heat is seen at mid-thickness in Fig. 8, where the curves show a
“bump” at about 800 C. The cooling slows down because of the phase transforma-
tion. The bump is also visible near the surface of the sample in Fig. 8(b), while the
cooling rate is moderate (water at 60 C).
240
STP 1543 On Zirconium in the Nuclear Industry
FIG. 7 Kn factor maps; panel (a) is associated with a compressive-thermoelastic stress r== =l ¼ 0:05: The maximum value Kn ¼ 0.499 is achieved for
u ¼ 45 and h ¼ 54.74 (i.e., n==h111iÞ; whereas the minimum value Kn ¼ 0.278 corresponds to h ¼ 0 (i.e., n==h100iÞ. Panel b illustrates in turn the Kn
variations for a tensile stress r== =l ¼ 0:05: In this case, n==h100i corresponds to the maximum value Kn ¼ 0.389, whereas the minimum value
Kn ¼ 0.168 is reached for n==h111i:
FIG. 8 Evolution of temperature during quenching in 10 C (a) and 60 C (b) water;
“mid-thickness” and “near surface” denote the thermocouple location. x denotes
the distance from the center for simulated curves.
The influence of the dynamic phase transformation can be seen in Fig. 9. In

Fig. 9(a), the evolution of the phase fraction as a function of the temperature at three
locations in the sample thickness is plotted. The phase transformation occurred at a
lower temperature near the surface because of the higher quenching rate (as
described in Eq 4) and was not completed at 800 C. The computed time evolution
of the cooling rate at three locations along the sample thickness is plotted in
Fig. 9(b) (instantaneous cooling rate between 1000 C and 800 C) and Fig. 9(c)
(instantaneous cooling rate between 99 % and 1 % of the b phase, which appears
more metallurgical). At mid-thickness, the average cooling rate was 26.5 K/s
between 1000 C and 800 C, and it was 30.6 K/s between 99 % and 1 % of the b
fraction. In this case, the difference is relatively small.
FIG. 9 (a) Calculated b-phase fraction at different locations along the sample thickness
as a function of temperature when quenching in 10 C water; calculated
instantaneous cooling rate as a function of time at different locations along the
sample thickness from 1000 C to 800 C (b) and from 99 % to 1 % b phase (c).
RESULTS OF THE v CHARACTERIZATION

Validation of the Image Analysis
EBSD-OIM maps were acquired on a quenched Zircaloy-4 sample. In the IPF map
[see Fig. 10(a)], each color represents a given orientation. This IPF map used a
surrogate of the polarized light micrograph; this is not exactly a one-to-one
comparison of the techniques, but it is useful for validating the image analysis.
Figure 10(b) represents a grain-boundary map obtained from the orientation
information by the HKL software: the misorientation angles between the a lamellae
corresponding to the Burgers relationships (11 , 60 –63 , 90 ) are colored in green,
and the rest of the grain boundaries (between ex-b grains) are colored in black.
Figure 10(b) presents the boundary image obtained with the in-house image
FIG. 10 (a) IPF–EBSD map. (b) Grain boundary–EBSD map. (c) New image after
treatment for colony-boundary detection.
treatment applied to the IPF map. The v ratios obtained from the HKL software
and by the in-house treatment were, respectively, 0.69 and 0.71. These two values
were close enough to validate the in-house image-treatment.
Polarization Effect
Whether a colonies can be distinguished depends on the polarization angle (see
Fig. 11). Therefore, it is important to determine the influence of the polarization
angle on v. The v values of 10 images taken, each time, with two different angles of
polarization are presented in Fig. 11. The polarization angle was shown to have little
effect on v.
Platelet and Basket-weave Quantification

Three samples of Zircaloy-4 quenched bars with visually different v ratios were
selected: samples A through C, from the more basket-weave to the more platelet.
They were quenched with exactly the same conditions (heating time and tempera-
tures in the b phase, water temperature, geometry, and thus cooling rates). Such
variations in the microstructure are known to arise from small differences in impur-
ities such as Cl, P, Si, and C [8], which were not quantified here. On each sample, 1
FIG. 11 (a), (b) Influence of the polarizer angle on polarized light micrographs; same
area with two different polarizations, with the arrows showing differences in
contrast. (c) The v ratios of 10 different images taken, each time, with two
polarizer angles; the 11th ones correspond to the mean value of the 10 v values
with a given polarization angle. The data points correspond to the mean value,
and the error bar indicates 1 standard deviation of the 10 measurements.
picture was taken in polarized light at 25 magnification and 12 pictures were
taken at 100 to roughly cover the same zone. They were analyzed to assess the v
values (see Fig. 12). These v values agreed with the visual aspect of the microstruc-
ture: the greater the v value, the more prominent the basket-weave character. The
magnification had no significant influence on v (the value at 25 stayed within
FIG. 12 Micrographs of quenched Zircaloy-4 with visually different platelet/basket-

weave characters and the corresponding values of v based on 12 pictures for
each case; the error bars correspond to 1 standard deviation.
only 1 standard deviation of the 12 values around the average at 100) as long
as the observed area was representative (not too great magnification, or enough
pictures) and the needle size was not too small relative to the image resolution
(magnification not too low).
Influence of the Cooling Rate

Six series of 10 optical images were taken at two different magnifications (50 and
200) from Zircaloy-4 samples quenched at three cooling rates, 17 C/s, 30.4 C/s,
and 103 C/s, as computed in the section “Cooling-rate Modeling.” The lamella
mean thickness values were, respectively, 2.5 lm, 1.8 lm, and 1.7 lm. The same
image treatment was done for each micrograph, and the mean v value was calcu-
lated from each of the six series (see Fig. 13).
The v parameters were more scattered at 200 magnification than at 50. This
could be due to the fact that at higher magnification, the field is too small to be
FIG. 13 Influence of the cooling rate on v; e is the lamella mean thickness at a given
cooling rate. The error bar corresponds to 1 standard deviation.
statistically representative. As a result, the v value varies significantly from one

image to another. In contrast, at 50 magnification there is more than one ex-b
grain, so it is more homogeneous. The standard deviation gives additional informa-
tion about the microstructure homogeneity. The larger the standard deviation, the
more heterogeneous the sample.
The effect of cooling rate on the a morphology is not clear: although it is
reported that a high cooling rate should promote the basket-weave character
(because there are more nucleation sites for the needles), this cannot be concluded
from the current measurements. Either the cooling-rate range was too small in rela-
tion to the tested material, or some other phenomenon was occurring. Additional
studies are in progress.
RESULTS ON THE a VARIANT FRACTION QUANTIFICATION

A dozen ex-b grains, in their initial state, were analyzed. A variant is considered as
dominant if its volume fraction is at least 30 %. These dominant variants appear
randomly distributed from the normal axis (see Fig. 14).
Eight grains were observed in the initial state, which was a quenched billet, and
after a second quenching. The first observation thus concerns a phase transforma-
tion in the bulk of the material, and the second a transformation at the surface.
Figure 15 shows the f0001g PF of the grain G1 at its initial state; ½3
21 is the normal
direction. The normal direction is given in the b coordinate system for
FIG. 14 c-axis fraction distribution in the initial state.
confrontation with the model. Within grain G1, the three main variants (76 %,
10 %, and 5 %) were the farthest from the normal. After the second quenching, the
three main variants were the closest to the normal [see Fig. 15(b)]. Also, the Kearns
factor (of the normal direction) increased from 0.367 to 0.716.
In grain G2, the normal direction was ½111. This time, the three main variants
were already closest to the normal axis (see Fig. 16). After the second quenching, the
fraction of those three variants was larger. The Kearns factor (of the normal direc-
tion) increased slightly from 0.558 to 0.656. The difference was not as large as with
G1, but the Kearns factor was still increasing.
Other grains, not shown here, presented the same behavior. The evolution of
the Kearns factor of eight ex-b grains during a second quenching is plotted in
Fig. 17. For seven of the eight grains, the Kearns factor increased during the second
quenching (when the grain was at the sample surface). To sum up, a second
quenching tended to align the c-axis in the normal direction.
Discussion
In each b grain, a minimum of six variants were observed representing 60 % to
70 % of the grain surface, whereas 12 variants were observed in half of the
FIG. 15 (a) f0001g PF of the grain G1 in the initial state. (b) f0001g PF of the same grain
G1 after the second quenching. (c) Histogram of the variant fraction before and
after quenching; the six c-axes are sorted by increasing angle to the normal
direction.
investigated grains. A second quenching had no impact on that. It thus seems

that the surface energy between variants (neglected in the previous modeling) was
negligible in comparison with the thermodynamic (phase transformation) and
mechanical energies.
FIG. 16 (a) f0001g PF of the grain G2 in the initial state. (b) f0001g PF of the same grain
G2 after quenching. (c) Histogram of the variant fraction before and after
quenching.
MICROSTRUCTURE MORPHOLOGY
The quenched-microstructure morphology, characterized by v, has been shown not
to depend on the cooling rate (between 17 C/s and 100 C/s). At higher rates
(achievable mainly through welding or very specific techniques), v could increase
because of the higher nucleation rate in the material bulk.
FIG. 17 Local Kearns-factor evolution during second quenching.
The cooling rate does, however, affect the needle width, as shown above. The
obtained needle widths are in agreement with previous studies (at least in the scat-
ter band) [2–4].
The main parameter influencing v could be chlorine, or insoluble, impurities,
as reported in Ref 9. These impurities—at the wt. ppm level, or even below—are dif-
ficult to measure accurately and depend on the raw materials (sponge, scraps recy-
cling, etc.) and the melting conditions. Further studies are needed to establish an
accurate correlation.
Morphologic and crystallographic orientation can be very different: in some
cases, two variants crystallographically misoriented from 10.5 appeared as needles
(or platelets) morphologically disoriented from almost 90 [see Fig. 4(c)].
INFLUENCE OF THE SURFACE DURING QUENCHING

The second quenching applied to the samples allowed us to characterize grains at
the sample surface during this second quench after a first quench in which they
were in the bulk.
Comparison Between Electron Back-scattered Diffraction and

Light Microscopy
In the section “Results on the a Variant Fraction Quantification,” it is shown that
quenching tends to increase the Kearns factor of the normal direction. In Fig. 18,
FIG. 18 Evolution of v during a second quenching. 1, obtained by the HKL software. 3,

obtained by the image treatment for colony-boundary detection.
two already mentioned methods (see “Validation of the Image Analysis”) are com-
pared to show the evolution of v during quenching on the same grains discussed in
“Results on the a Variant Fraction Quantification.” It appears that v does not
depend on the computing method and does not evolve during a second quenching:
the variant morphology is not affected by the vicinity of the surface.
Comparison Between Model and Experimental Results

From EBSD analysis, it is possible to know each ex-b-grain orientation. From this
orientation, the h and u angles can then be extracted, which is necessary for calcula-
tion of the Kearns factor with the analytical model mentioned in the section
“Variant Selection under Thermoelastic Strain.” The Kearns factor of the normal
direction of each grain, labeled from 1 to 8, is represented in Fig. 19. These factors
are calculated with the model, with r== ¼ 0:05l (corresponding to e== ¼ 0:013
and e? ¼ þ0:012Þ: The model should apply because the six virtual variants are
observed experimentally except for grain G2 (and are assumed to be present in the
model). They also are computed from the EBSD analysis. Both calculations are
issued from grains analyzed after quenching. The comparison between model and
EBSD results shows that the stress applied to the observed surface was more likely a
compressive stress. Moreover, one can conclude that the stress r== ¼ 0:05l
FIG. 19 Comparison of the Kearns factors obtained on each grain from the EBSD map
and from the model.
assumed in the model is too small, considering the discrepancy between experimen-
tal and modeled Kearns factors. Indeed, in six of the eight cases, the model predicts
a Kearns factor smaller than the experimental one.
Conclusion
The main results obtained in this study are the following:
• We achieved accurate modeling of the cooling rate, accounting for the phase-
transformation kinetics and allowing one to compute the cooling rate either as
defined industrially (between 1000 C and 800 C) or during the actual phase
transformation.
• We presented a methodology for quantifying the platelet/basket-weave charac-
ter v of a microstructure (v varies from 0 for parallel plates to theoretically
1.25 but practically 1 for a basket weave).
• Within the investigated alloy and quenching-rates, v does not evolve signifi-
cantly with the quenching rate.
• The vicinity of the surface during quenching does not modify the morphology,
characterized by v, of the quenched state, but it does tend to promote variant
selection leading to higher local Kearns factors of the normal direction to the
surface. This is consistent with free distortion in this direction.
Appendix
Assume (see Fig. 20) an image with square dimensions A by A and needles with mean
sizes a by n by a (a being the width, n by a the length, and n the aspect ratio). The
FIG. 20 Image (left) and needle (right) sizes.
sizes a and na have to be understood as the mean width and length of individual
needles.
The image area is A2, the needle area is na2, and the needle perimeter is
2að1 þ nÞ: The v ratio is defined as v ¼ ðboundarylength needlewidthÞ=ðtotalareaÞ.
For a basket-weave structure, the extreme case contains individual needles. Thus
the number of individual needles in the image is A2 =na2 : The total perimeter is then
ð1=2Þ½A2 =ðna2 Þ½2að1 þ nÞ ¼ ½A2 ð1 þ nÞ=ðnaÞ: This leads to v ¼ ð1 þ nÞ=n: Typi-
cally, n varies from 4 to 10, and v from 1.25 to 1.1.
For parallel plates, the extreme case contains only one variant in the image; thus
the area is equal to the image area, the boundary length is equal to 4A, and the v ratio
is v ¼ ð4AaÞ=A2 ¼ ð4aÞ=A 0.
References
[1] Chauvy, C., Barberis, P., and Montheillet, F., “Microstructure Transformation during Warm
Working of Beta-treated Lamellar Zircaloy-4 within the Upper Alpha-range,” Mater. Sci.
Eng. A Struct. Mater., Vol. 431, Nos. 1–2, 2006, pp. 59–67.
[2] Woo, O. T. and Tangri, K., “Transformation Characteristics of Rapidly Heated and
Quenched Zircaloy-4 Oxygen Alloys,” J. Nucl. Mater., Vol. 79, No. 1, 1979, pp. 82–94.
[3] Sawatzky, A., Ledoux, G. A., and Jones, S., “Oxidation of Zirconium during a High-
temperature Transient,” Zirconium in the Nuclear Industry, ASTM STP 633, A. L. Lowe
and G. W. Parry, Eds., ASTM International, West Conshohocken, PA, 1977, pp. 134–149.
[4] Massih, A. R., Andersson, T., Witt, P., Dahlback, M., and Limback, M., “Effect of Quenching
Rate on the Beta-to-alpha Phase Transformation Structure in Zirconium Alloy,” J. Nucl.
Mater., Vol. 322, Nos. 2–3, 2003, pp. 138–151.
[5] Daymond, M. R., Holt, R. A., Cai, S., Mosbrucker, P., and Vogel, S. C., “Texture Inheritance
and Variant Selection Through an hcp-bcc-hcp Phase Transformation,” Acta Mater.,
Vol. 58, No. 11, 2010, pp. 4053–4066.
[6] Gey, N., Gautier, E., Humbert, M., Cerqueira, A., Bechade, J. L., and Archambault, P.,
“Study of the Alpha/Beta Phase Transformation of Zy-4 in Presence of Applied Stresses
at Heating: Analysis of the Inherited Microstructures and Textures,” J. Nucl. Mater., Vol.
302, Nos. 2–3, 2002, pp. 175–184.
[7] Mosbrucker, P., Daymond, M. R., and Holt, R., “In Situ Studies of Variant Selection during
the a-b-a Phase Transformation in Zr-2.5Nb,” J. ASTM Int., Vol. 8, No. 1, 2012, JAI103066.
[8] Charquet, D. and Alhéritière, E., “Influence of Impurities and Temperature on the Micro-
structure of Zircaloy-2 and Zircaloy-4 after the Beta-Alpha Phase Transformation,” Zir-
conium in the Nuclear Industry, ASTM STP 939, R. Adamson and L. Van Swam, Eds.,
ASTM International, West Conshohocken, PA, 1987, pp. 281–284.
[9] Charquet, D., “Microstructure and Properties of Zirconium Alloys in the Absence of Irra-
diation,” Zirconium in the Nuclear Industry, ASTM STP 1354, A. M. Garde and E. R. Brad-
ley, Eds., ASTM International, West Conshohocken, PA, 2000, pp. 3–14.
[10] Brachet, J. C., Portier, L., Forgeron, T., Hivroz, J., Hamon, D., Guilbert, T., Bredel, T., Yvon,
P., Mardon, J. P., and Jacques, P., “Influence of Hydrogen Content on the a/b Phase
Transformation Temperatures and on the Thermal-Mechanical Behavior of Zy-4, M4
(ZrSnFeV) and M5 (ZrNbO) Alloys during the First Phase of LOCA Transient,” Zirconium
in the Nuclear Industry, ASTM STP 1423, G. Moan and P. Rudling, Eds., ASTM Interna-
[11] Burgers, W. G., “On the Process of the Transition of the Cubic-body-centered Modifica-
tion into the Hexagonal-close-packed Modification of the Zirconium,” Physica, Vol. 1,
Nos. 7–12, 1934, pp. 561–586.
[12] Humbert, M. and Gey, N., “Elasticity-based Model of the Variant Selection Observed in
the b to a Phase Transformation of a Zircaloy-4 Sample,” Acta Mater., Vol. 51, No. 16,
2003, pp. 4783–4790.
[13] Gaunt, P. and Christian, J. W., “The Crystallography of the b!a Transformation in Zirco-
nium and in Two Titanium-Molybdenum Alloys,” Acta Metall., Vol. 7, No. 8, 1959, pp.
534–543.
[14] Barberis, P., Montheillet, F., and Chauvy, C., “Variant Selection in Zr Alloys: How Many
Variants Generated from One Beta Grain?” Solid State Phenom., Vol. 105, 2005, pp.
133–138.
[15] Kearns, J. J., “Thermal Expansion and Preferred Orientation in Zircaloy,” Report No.
WAPD-TM-472.1965, Bettis Atomic Power Laboratory, Pittsburgh, PA, 1965.
[16] Kearns, J. J., “On the Relationship among ‘f’ Texture Factors for the Principal Planes of
Zirconium, Hafnium and Titanium Alloys,” J. Nucl. Mater., Vol. 299, No. 2, 2001, pp.
171–174.
DISCUSSION
Questions from Arthur Motta, Penn State University:—
The results of increasing Kearns’ factor with greater alignment with the surface
are all derived from measurements performed on the surface. Do you think that
your results only reflect the surface layer and would the results be different in the
bulk?
Authors’ Response:—Yes, the results reflect only the surface state. The free sur-
face modifies the a-lamella nucleation. It could be also responsible for keeping the
same ex-b grain orientation and this is beneficial for studying the a-variant selec-
tion. It would be different in the bulk.
Questions from Ron Adamson, Zircology Plus:—What is the C and Si concentra-

tion of your material, and if it were different, would it affect your results?
Authors’ Response:—The Si and C content are about 100 and 120 ppm respec-
tively. It is known that such elements, as well as other volatile impurities like chlo-
rine may affect the quench microstructure and specifically the basketweave
platelets ratio. On the other hand, the evolutions presented here are obtained on
the same materials so that the evolution is meaningful even if the absolute value
may be impacted by the Si or C contents.
Questions from K. Kapoor, NFC:—What is the role of prior b grain size on the
variant selection and morphology of the a phase?
Authors’ Response:—The b-grain size in the present study was quite large (from
0.2 mm2 to 4mm2). Generally, the 12 a-variants are present. In smaller b-grains,
however, less a-variants would be expected and therefore the variant selection
would be different.
Question from Jean-Christophe Brachet, CEA-Saclay Nuclear Materials Dept.

France:—What is your under-cooling phase transformation (b ! a) kinetics
model?
Authors’ Response:—It is exactly the one you described in your earlier publica-
tion (see reference [10] in the paper).
Question from Srikumar Banerjee, BARC Mumbai:—Two morphologies,

namely, basket weave and parallel laths, are perhaps produced due to the combina-
tion of the following effects:
Strain energy minimization
Sequence of transformation in which laths/plates forming in the first genera-

tion influences the morphology of the second by providing surfaces for
nucleation of new generation and by providing autocatalytic nucleation by
straining the parent lattice.
The model may include both effects.
Authors’ Response:—The present model clearly addresses only the first effect:
an instantaneous and overall (i.e., within the whole b grain) phase transformation
is considered, which precludes the above-mentioned sequential effects. This is
admittedly a limitation of the model.
Question from Dinesh Srivastava, BARC Mumbai:—The model is based on

the bulk whereas the measurement was performed only on the surface. That may
be the reason for the model not matching the experimentally measured value.
Bulk texture should have been measured as Kearns’ factor increases after
quenching, i.e., it becomes more textured after quenching.
Authors’ Response:—1) The model considers instantaneous and overall phase

transformation where surface effects do not influence nucleation. However, variant
selection is affected by the surface through the plane stress state. 2) Bulk textures
have been measured, but were not available at the time of the conference.
Questions from Javier Romero, Westinghouse Electric Co:—What environment

was used during the thermal treatment? Was there any effect of oxygen?
Authors’ Response:—The specimens are sealed in a vacuumed quartz-tube

before the thermal treatment, such that the reaction with the oxygen is limited.
Questions from Javier Romero, Westinghouse Electric Co:—How much of the shape
and original boundaries of the ex-b grains is maintained during heating? It is
expected that the initial b grains will change with the quenching. How do you keep
track of one grain, does it not change shape/grain boundary characteristics during
heating, given the high temperature and long holding time?
Authors’ Response:—The shape and original boundaries of the ex-b grains change
a little bit but do not move significantly. This can be due to a surface effect and also to
the fact that the ex-b grains are quite large. The heat treatment consists to introduce
the specimen into the furnace at 1050 C, to hold it inside for 10 min in order to ho-
mogenize the temperature. The soaking time is long enough to complete the transfor-
mation into b phase. Vickers indents are used to track the ex-b grains.
FABRICATION
J. K. Chakravartty,1 R. Kapoor,1 A. Sarkar,1 V. Kumar,2

S. K. Jha,3 N. Saibaba,3 and S. Banerjee4
Identification of Safe Hot-Working

Conditions in Cast Zr–2.5Nb
Reference
Chakravartty, J. K., Kapoor, R., Sarkar, A., Kumar, V., Jha, S. K., Saibaba, N., and Banerjee, S.,
“Identification of Safe Hot-Working Conditions in Cast Zr–2.5Nb,” Zirconium in the Nuclear
Industry: 17th International Symposium, STP 1543, Robert Comstock and Pierre Barberis, Eds., pp.
259–281, doi:10.1520/STP154320120197, ASTM International, West Conshohocken, PA 2015.5
ABSTRACT
Cast Zr–2.5Nb was hot-deformed in the temperature range of 700 C to 1100 C
and in the strain-rate range of 103 to 10 s1 using compression tests. The stress-
versus-strain data generated at different rates and temperatures were used to
compute the strain rate sensitivity and generate processing maps. The map
showed a single domain that spanned the temperature range from 800 C to
1100 C and strain rates less than 0.1 s1 with a peak strain rate sensitivity of
0.3 at 1000 C and 103 s1. Regimes of flow instability were identified using
different criteria based on work-hardening rate and strain rate sensitivity, as well
as on parameters based on thermodynamics and extremum principles. Optical
metallography of the deformed samples revealed that the domain of high strain
rate sensitivity showed equiaxed grains. This, along with the high strain rate
sensitivity domain, suggested the occurrence of dynamic recrystallization in this
regime. At low temperatures (700 C and 750 C) and high strain rates of 1 and
10 s1, some macroscopic flow localization was observed. This was predicted by
one thermodynamics-based instability criterion that also represents regions of
low strain rate sensitivity. On the basis of the processing map generated and
Manuscript received December 7, 2012; accepted for publication July 2, 2014; published online September
25, 2014.
1
Mechanical Metallurgy Division, Bhabha Atomic Research Centre, Mumbai 400085, India.
2
RDCIS, Steel Authority of India Limited, Ranchi, Jharkhand 834002, India.
3
Nuclear Fuel Complex, Hyderabad 500062, India.
4
Dept. of Atomic Energy, Mumbai 400085, India.
5
Hyderabad, India.
Copyright V
metallographic observations, it was found that the cast structure could be

safely broken in both the a þ b and b phases when deformed at strain rates less
than 0.1 s1.
Keywords
hot deformation, Zr alloys, ingot breakdown
Introduction
Zr–2.5Nb is used as a pressure tube material for structural components in nuclear
pressurized heavy water reactors. At present, Zr–2.5Nb pressure tubes for Indian
pressurized heavy water reactors are manufactured through a series of hot-working,
stress-relieving, and cold-working steps, the first extrusion of the cast structure
being in the a þ b field. Cast alloys are hot-worked primarily to break down the
cast microstructure, shape the ingot, chemically homogenize the product, and
obtain a desired microstructure and texture. To identify the optimum hot-working
conditions, one usually constructs processing maps that map out either the strain
rate sensitivity or its function in a strain rate–temperature frame. Chakravartty and
co-workers [1–10] have studied the hot deformation behavior of Zr and its various
alloys, optimized the hot deformation parameters in terms of strain rate and tem-
perature, and correlated these with the microstructure. However, most of these
studies were carried out on wrought alloys whose microstructures might have
already been broken by a primary hot-working step. In one such study [4], hot
deformation of wrought Zr–2.5Nb was carried out in the a þ b phase field for two
different microstructures, one equiaxed a þ b and the other b quenched; in both
the optimum hot-working domain was around 800 C and 103 s1. However,
alloys with cast microstructures may have a different response and a different set of
optimum hot-workability conditions.
Apart from the optimization of hot deformation parameters in terms of strain
rate sensitivity, the conditions for unstable flow are also important to delineate.
This information can then be used to avoid instability and to obtain a defect-free
microstructure during hot-working. Microstructural instability includes cracking
and flow localization of both microscopic and macro-shear bands. There are differ-
ent methods for identifying instability. One is through flow localization based on
the work-hardening rate and strain rate sensitivity parameters; this is an extension
of Hart’s criterion for tensile flow (see, e.g., Refs 11–13). The other is through
the application of thermodynamics and extremum principles in which strain rate
sensitivities and temperature sensitivities of the flow stress are used (see, e.g.,
Refs 14–17).
The aims of the present work were to study the microstructural changes during
hot deformation of cast Zr–2.5Nb and to identify the optimum deformation param-
eters and the regimes of stable flow. These results are compared with those for
wrought Zr–2.5Nb [4,6].
CHAKRAVARTTY ET AL., DOI 10.1520/STP154320120197 261
Experimental
Zr–2.5 wt. % Nb in the cast condition was obtained from Nuclear Fuel Complex
(Hyderabad, India). The composition was 2.5 wt. % Nb, 0.08 to 0.1 wt. % Fe,
1000 ppm O, and the balance Zr. Cast Zr–2.5Nb alloys showed a Widmannstätten
type of a þ b microstructure (Fig. 1). The thickness of the a platelets was around
3 to 5 lm, and they were arranged in packets 50 to 70 lm in size. The prior b grain
size is much larger and is in the range of 4 to 6 mm. It is usually the finer micro-
structural feature that controls the mechanical behavior, which in this case would
be the a platelets (or the colonies of a), rather than the prior b grain size. The
Zr–2.5Nb with about 1000 ppm oxygen had an a þ b to b transformation tempera-
ture of 910 C. Hot compression tests were carried out in a thermo-mechanical sim-
ulator on 10-mm-diameter and 15-mm-length cylindrical samples from 700 C to
1100 C and from a true strain rate of 103 to 10 s1 in an argon atmosphere. After
the tests, the samples were cooled to ambient temperature at a cooling rate of
about 150 C/s to 200 C/s. Compression tests were carried out to a true strain of
0.7, with some tests conducted to a true strain of about 1.6 to check the microstruc-
tural evolution at large strains. Graphite foils were used as lubricant for tempera-
tures below 900 C, and tantalum foils were used for the higher temperatures. The
load–displacement data obtained from compression tests were converted to obtain
curves of true stress (r) versus true plastic strain (e) using standard equations. The
temperature was measured to 61 C using a thermocouple spot-welded on the sam-
ple surface. For strain rates of 1 s1 or more, the tests were considered adiabatic,
and the temperature increase of the sample was estimated by converting the work
FIG. 1 Microstructure of cast Zr–2.5Nb.

done to heat generated. The true stress versus estimated temperature after tempera-
ture rise correction was plotted, and the stress was interpolated at the given test
temperature. After deformation, the samples were cut longitudinally, mounted, and
polished using standard techniques. The polished samples were etched with 10 %
HF þ 45 % HNO3 þ 45 % H2O for observation through an optical microscope.

DEFORMATION BEHAVIOR IN COMPRESSION AND PROCESSING MAP
The flow stress behavior of Zr–2.5Nb at high temperatures (Fig. 2) showed signifi-
cant softening for all the strain rates at the lower temperatures of 700 C and 800 C,
at which the alloy was in the a þ b range. Similar flow softening was reported by
Chakravartty et al. [4] for b-quenched wrought Zr–2.5Nb tested in the a þ b phase
field and by Chauvy et al. [18] for Zircaloy-4 tested at the lower strain rates at
650 C. In the present work, at higher temperatures of about 900 C, where the
major phase was b, there was practically no softening. The stress-versus-strain
curves exhibited typical steady-state behavior immediately following a brief
FIG. 2 True stress versus true plastic strain plotted on a log-linear scale for cast
Zr–2.5Nb at 700 C, 800 C, 900 C, and 1100 C and at different strain rates as
listed next to each curve. True stress is plotted on a log scale so that the strain
rate sensitivities and normalized work-hardening rates may be easily
appreciated as discussed in the text.
FIG. 3 True stress at e ¼ 0.5 for Zr–2.5Nb plotted against (a) strain rates at different
temperatures and (b) temperature at different strain rates. The temperatures are
written next to the curves in (a), and strain rates are written next to the curves in
(b).
work-hardening stage. The wrought alloys also exhibited similar steady-state flow
behavior [4]. At 1100 C and a lower strain rate of 103 s1, softening did appear
when seen on a normalized scale (dln r/de), but the extent of softening was small
when seen on a linear scale (dr/de).
The variation of r with e_ at constant temperature and the variation of r with T
at constant e_ at e ¼ 0.5 were extracted from Fig. 2 and plotted in Fig. 3. The strain
rate sensitivity m is defined as
@ ln r
(1) m¼
@ ln e_
For cast Zr–2.5Nb, r increases with e_ [Fig. 3(a)], and at no condition is the slope
(m) observed to be negative. It is seen from the slopes of the plots in Fig. 3(a) that m
was high at 1000 C and at strain rates between 103 and 102 s1. The strain rate
sensitivities between 1 and 10 s1 were low. As seen from Fig. 3(b), stress decreased
with temperature up to about 950 C, beyond which its variation with temperature
was low. This shows that the a þ b phase field in Zr–2.5Nb has a greater magnitude
of temperature sensitivity than the b phase field at all strain rates, as seen in
Fig. 3(b). However, the strain rate sensitivity of Zr–2.5Nb in the b phase was greater
[Fig. 3(a)].
The strain rate sensitivity was calculated by fitting the data for log r versus log e_
to a cubic polynomial and taking the derivative (Eq 1) (see, e.g., Ref 6). The m
values obtained for Zr–2.5Nb were plotted as an iso-strain rate sensitivity contour
map as shown in Fig. 4. The map exhibits a single domain that spans the tempera-
ture range from 800 C to 1100 C encompassing the a þ b phase field and the b
FIG. 4 Iso-strain rate sensitivity contour plot for cast Zr–2.5Nb. The numbers on each
contour line are the m values.
phase field. The strain rate range of this domain ranges from 0.1 s1 to lesser values.
The peak m is centered on 1000 C and 103 s1, which is in the b regime. However,
the a þ b phase field also has a strain rate sensitivity domain of m ¼ 0.2 in the range
of 750 C to 800 C and 102 to 101 s1. Chakravartty et al. [4], working on
wrought Zr–2.5Nb, showed a high m domain in the temperature range from 700 C
to 900 C and the strain rate range from 103 to less than 101 s1 with a peak
m ¼ 0.3. It is seen that relative to the wrought alloy, the cast Zr–2.5Nb showed a
high m domain at higher temperatures, although the strain rate range was similar.
In the wrought alloy, the high m domain in the a þ b phase was attributed to
dynamic recrystallization.
TENSILE DUCTILITY
Tensile tests were carried out from 800 C to 1100 C over the entire strain rate
range in which the compression tests were carried out. A typical stress–strain curve
is shown in Fig. 5(a). It is seen that there was negligible uniform elongation, and
most of the ductility was taken up by non-uniform elongation after the onset of
necking. The stress–strain plots for the other temperatures are similar in shape,
except that with increasing temperature the ultimate tensile strength (UTS)
decreased. The early onset of localization is due to the low work-hardening rate at
these temperatures, as seen from the compression data. The large non-uniform
FIG. 5 (a) Typical tensile engineering stress versus engineering strain of Zr–2.5Nb. (b)
Ductility versus strain rate for different temperatures as marked on the plot.
elongation is due to the high strain rate sensitivity. The ductility at different tem-
peratures and strain rates is plotted in Fig. 5(b). The ductility was obtained by meas-
uring the length of the sample before and after tensile tests, rather than from the
stress–strain curve. It is seen that the ductility was highest at 1000 C and 1100 C
and at the low strain rate of 103 s1. This matches the high strain rate sensitivity
domain centered on 1000 C and 103 s1 seen in Fig. 4. The ductility increased
marginally with increasing temperature. A typical photograph of samples after ten-
sile testing is shown in Fig. 6. It is seen that at 800 C (a þ b phase field) the sample
failed via typical cup and cone fracture, whereas at higher temperatures of 1000 C
the failure was more of a chisel type (i.e., necking took place in one dimension
only). In some samples, such as that tested at 1000 C and 103 s1, a double neck
was observed, suggesting that localization, if initiated, was resisted and did not
propagate to failure. In the sample tested at 1000 C and 103 s1, the failure was
seen to occur at a neighboring neck formation. A similar prolonged neck with a
double neck and chisel-like features was seen at 1000 C and 102 s1. These condi-
tions are in the b phase and in the regime of high m; thus it is expected that strain
localization would be resisted and a prolonged neck would be formed. It should be
noted that at 900 C, in some cases chisel-type necking was observed. However, the
knowledge of m alone might not be sufficient to identify regimes of unstable flow.
THERMAL ACTIVATION PARAMETERS

_ and T data in or near the high m domain were fit to the power-law rate
The r, e;
equation e_ ¼ Arn expðQ=RTÞ, where Q is the activation energy, n is the stress
exponent, A is a constant, and R is the universal gas constant. The Zener–Hollomon
parameter Z e_ expðQ=RTÞ plotted against r on a log-log scale with appropriate
values of Q and n resulted in a straight line as seen in Fig. 7. The activation energy
266
FIG. 6 Some representative photographs of tensile samples after failure. The left-hand panel is for 800 C, and the right-hand panel is for 1000 C. The
ductility is marked as a percentage. Arrows show the formation of multiple necks.
FIG. 7 Plot of Zener–Holloman parameter versus stress for data within domain (as
mentioned in text). The activation energy Q and stress exponent n are written
on the plot. Also compared are Z–r plots for wrought Zr–2.5Nb in the a þ b [4]
and b [6] phases.
Q ¼ 270 kJ/mol for the a þ b phase in the high m domain, and Q ¼ 125 kJ/mol for
the b phase. These values were similar for the cast and wrought conditions. The n
values were 4.84 and 4.7 for a þ b and b phases, respectively. The activation energy
for self-diffusion in a Zr is 190 kJ/mol, and in b Zr it is 110 kJ/mol, whereas that
of Nb in b Zr is 210 kJ/mol. Q in the a þ b phase is much higher than the
self-diffusion energies, and n 5 suggests a dislocation-based, non-diffusive rate-
controlling mechanism. In the b phase, Q matches that for self-diffusion, suggesting
a diffusion-based rate-controlling mechanism.
INSTABILITY CRITERIA
Several approaches are available for identification of the e_ and T combinations that
result in flow instabilities. These approaches are described briefly here, and an
assessment is made to find the most suitable approaches for the cast Zr–2.5Nb
alloy.
Work-hardening and Softening Behavior

Earlier studies on plastic instability in tension were carried out by Hart [11] and
Campbell [12] in which the strain rate sensitivity m and the normalized work-
hardening rate c,
1 @r @ ln r
(2) c¼ ¼
r @e @e
were used to predict the start of instability. c can be visualized from the slope of r
versus e plotted on a log-linear scale as described through Eq 2 and seen in Fig. 2.
The formulation of strain localization in compression was done by Jonas et al.
[13], who derived the condition for the onset of strain localization as
(3) c>1m
It should be noted here that in compression the sign convention is r; e; and e_ < 0
[13]. Thus, for a material that work-hardens in compression, @r=@e > 0, which
makes c < 0. At steady state, @r=@e ¼ 0, and hence c ¼ 0. During flow softening,
@r=@e < 0; and because r < 0, the implication is that c > 0. Thus c is initially neg-
ative and becomes positive during flow softening. However, Eq 3 is not a sufficient
condition for the onset of flow localization and does not imply that the material
will actually localize. Instability manifests depending on how rapidly the flow local-
ization occurs. This is quantified using a parameter
c1
(4) a0 ¼
m
Although the onset of strain localization may occur when a0 > 0 [i.e., c > 1 (given
m > 0)], rapid strain localization in compression is expected when a0 > 5 [13]. A
typical plot of the variation of a0 with compressive strain is shown in Fig. 8. Also
shown in the plot is the curve for r versus e (both of which are shown as negative
according to the sign convention). In this case the material first work-hardens to a
minimum r (maximum jrj) and then softens (i.e., r increases). Correspondingly, a0
starts at a negative value, increases to a peak, and then decreases. This is a typical
feature of all r–e curves that exhibit flow softening. In compression there is a strain
window within which the material could show instability. With increasing jej, the
parameter a0 decreases and the material flow is no longer unstable. Thus, unlike
tension, which leads the instability to propagate with increasing strain, in compres-
sion it is not clear whether instability would propagate in the regime where a0 < 5.
Despite this ambiguity, the instability criteria based on a0 has been used to charac-
terize unstable flow [19–21]. It is seen from Eq 4 that if m < 0, then because c is
highly negative from the beginning of deformation, the implication is that a0 > 0
and may be greater than 5, thereby fulfilling the condition of instability as defined
by Eq 4. Thus,
(5) m<0
is a certain condition of instability.

The variation of a0 with strain for Zr–2.5Nb is shown in Fig. 9. It is seen that
the trend for all the conditions is same as in Fig. 8: a0 increases from a large negative
FIG. 8 A typical variation of a0 (shown on left y axis and as derived from Eq 4). These
data are for cast Zr–2.5Nb deformed at 700 C and 103 s1. The corresponding
true stress is shown on the right-hand y-axis. The derivative dr/de was taken at
discrete data points as shown by the symbols.
value, reaches a peak, and then decreases. It is seen that for all conditions a0 < 5,
implying that according to the criteria of Jonas et al. [13], Zr–2.5Nb exhibits stable
flow at all tested strain rates and temperatures covered in this study. However, it
should be noted that at the lower strain rates a0 does become greater than 0 for a
small strain window. This is a condition for the onset of strain localization. With
increasing jej; a0 again decreases to negative values, ensuring stable flow. There is
no condition in which a0 > 0 is sustained.
Zeigler’s Criteria of Continuum Mechanics and Extremum Principles

Instability can also be established through continuum principles. One way is
through Zeigler’s analysis [14] based on the rate of dissipation of work D during
deformation. It was stated that deformation is unstable if
dD D
(6a) <
de_ e_
which implies
d ln D
(6b) <1
d ln e_
FIG. 9 a0 as a function of strain for Zr–2.5Nb at different temperatures as indicated on

the plot. The strain is plotted on a negative axis according to the sign convention
as discussed in the text.
During plastic deformation, if the rate of dissipation of work is taken as the power
input rate itself, then D ¼ re;
_ and from Eq 6b,
d ln re_ d ln r
(7) <1) <0)m<0
d ln e_ d ln e_
This expression for unstable flow is same as that expressed by Eq 5.

Some researchers have interpreted D as J, the co-content of the power dissipa-
tion,6 which is expressed as [23]
ð e_
(8a) J ¼ P G ¼ re_ rd e_
0
where:
P ¼ power represented by the first term, and
G ¼ integral on the right-hand side of Eq 8a.
The calculation of J requires the integration of r versus e_ from 0 to the required
e._ However, as the data start from some minimum strain rate e_min and not 0, the
integral in Eq 8a can be split in two parts for convenience as [24]
ð e_min ð e_
(8b) J ¼ re_ rd e_ rd e_
0 e_min
If it is assumed that for 0 < e_ < e_min the constitutive behavior is the power law
r ¼ ke_m , then J can be computed as [24]
ð e_
re_
(8c) J ¼ re_ rd e_
m þ 1e_min e_min
The efficiency of power dissipation g is defined as [23]
2J
(9a) g¼
re_
which, using Eq 8c, becomes

ð
Pmin 1 1 e_
(9b) g¼2 1 rd e_
P m þ 1 P e_min
where Pmin is the power P at e_ ¼ e_min . Zeigler’s criteria for instability if D ¼ J is
dJ J
(10) <
d e_ e_
6
The physical significance of J is questionable, as argued in Ref 22. However, the purpose here was to use
and compare the various instability criteria used in the literature; hence the use of J and g.
FIG. 10 Instability condition as represented by Eq 11 shown as a hatched portion plotted

over the iso-m contour map of Fig. 4
Using the relation dJ ¼ edr,

_ and using Eqs 1 and 9a, the instability condition
expressed by Eq 10 can be simplified to [21]
(11) 2m < g
The instability regime of Zr–2.5Nb according to Eq 11 is shown in Fig. 10. The

instability domain occurs at high strain rates of 1 s1 from 750 C to 950 C. At 1 s1
the criteria show instability at only 700 C. The instability analysis of the wrought
Zr–2.5Nb alloy was carried out using a variant of Eq 11 [4]. In that study the insta-
bility domain was identified in the temperature range of 650 C to 750 C and the
strain rate range from 1 to 100 s1. The instability domain for cast Zr–2.5Nb
appears to be in a wider temperature domain. It should be noted that in the present
study, as well as the one for wrought alloy, the instability predicted by Eq 11 (or its
variant used in Ref 4) always fell in the regime where m was low.
Lyapunov Function
Yet another approach for identifying instability involves the concept of Lyapunov
stability [15]. The second method of Lyapunov uses a function called the Lyapunov
function, V(x), such that VðxÞ 0 with equality if and only if x ¼ 0, and defines
the stability as
_ dVðxÞ
(12) VðxÞ ¼ 0 with equality if and only if x ¼ 0
dt
This means that there exists a function that decreases with time to reach the point
of equilibrium. In a physical sense this can be considered as an energy function,
with the energy of the system decreasing with time and not being restored, and
eventually reaching its final resting state.
Gegel et al. [25] used this concept not as a function of time but as a function of
strain rate. They used two Lyapunov functions, g and s, both of which are functions
of ln e._ s is defined as
1 @ ln r @ ln r
(13) s¼ ¼
T 1 @ ln T
@
T
The instability criteria were considered as
@g
(14) >0
@ ln e_
@s
(15) >0
@ ln e_
Alexander [17] followed the above approach but used m (instead of g) and s as the
two Lyapunov functions of ln e_ to determine the instability criteria as
@m
(16) >0
@ ln e_
@s
(17) >0
@ ln e_
It is seen that by putting Eq 13 in Eq 17 and using Eq 1, one can rewrite the insta-
bility condition described by Eq 17 as
@m @m @m
(18) <0)T <0) <0
@ ln T @T @T
Equation 18 says that the domain of m will always be bisected with respect to the T
axis; that is, instability occurs if m decreases with T. Similarly, Eq 16 says that the
domain of m will always be bisected with respect to the ln e_ axis; that is, instability
occurs if m increases with the log of the strain rate.
The instability conditions as defined by Eqs 16 and 17 were computed and plot-
ted over the iso-m contour plots shown in Fig. 11 as hatched areas. It is seen that the
condition described by Eq 17 bisects the high m sensitivity domain into a stable and
an unstable regime.
FIG. 11 Instability condition as represented by Eqs 16 and 17 shown as a hatched portion

plotted over the iso-m contour map of Fig. 4
MICROSTRUCTURE
Metallographic investigations were carried out to observe changes in the micro-
structure after different deformation conditions and to identify features of
instability (if any) and match them to the different criteria described above. If possi-
ble, the best criteria for the identification of a safe hot-working domain were
identified.
Figure 12 shows the macroscopic view of some deformed samples corresponding
to the iso-strain rate sensitivity map. At high strain rates of 10 s1 and low tempera-
tures of 700 C and 750 C, the samples showed non-uniform flow, and possibly a
manifestation of macroscopic flow localization. The samples showed flow localiza-
tion along the diagonals of the sample, with regions of the sample remaining unde-
formed (such as the top and bottom). A similar but less pronounced feature was
also observed at 700 C and 1 s1. At the lower strain rates of 102 s1 and a higher
temperature of 800 C, the sample showed relatively uniform deformation. No
localized band formation was observed. At temperatures of 900 C and above, the
macroscopic view only showed large prior b grain boundaries, which appeared to
have deformed uniformly. Microstructures at higher magnifications of some sam-
ples showing high m are described below. No microscopic flow localization could
be observed in samples that showed macroscopic flow localization; hence those
micrographs are not shown.
FIG. 12 Macrographs of Zr–2.5Nb samples after deformation at different conditions.

The conditions are indicated by the arrows from the strain rate–temperature
axis and show the corresponding strain rate sensitivity.
The microstructure of Zr–2.5Nb samples after deformation at 750 C and

10 s1 showed an unbroken a þ b lamellar structure [Fig. 13(a)], indicating that
3
the microstructure was unaffected by these deformation conditions. The lamellae

deformed without major microstructural modification. In the map (Fig. 4) this is a
regime of low strain rate sensitivity (0.15) and also falls in the instability domain as
described by Eq 16. This is also in the regime of the parameter 0 < a0 < 5, which
implies the possible onset of strain localization but not confirmed instability. How-
ever, as seen in Fig. 13(a), the optical micrographs of this sample showed no obvious
features of instability.
At 800 C and 101 s1 and a large strain of 1.6, the sample showed fragmenta-
tion of lamellae resembling subgrain-like formation; some fragmentations are
shown by arrows in Fig. 13(b). This deformation condition also lies in the instability
domain, as implied by Eq 16; however, it is not certain whether fragmentation of
the initial lamellae microstructure represents instability, recovery, or a recrystalliza-
tion process. In terms of m it represents a reasonably moderate value of about 0.2,
and in terms of the Jonas criteria a value of a0 ¼ 1, which brings it out of the insta-
bility domain as per Eq 4.
FIG. 13 Optical micrographs of Zr–2.5Nb after deformation at (a) 750 C, 0.001 s1, and
a strain of 0.8 and at (b) 800 C, 0.1 s1, and a strain of 1.6. Grain fragmentation in
(b) is observed in many areas and is shown by arrows.
For the sample deformed at 900 C and 102 s1 and a large strain of 1.6, equi-
axed grains (prior b grains at the temperature of deformation) could be observed,
as seen in Fig. 14. This occurrence of equiaxed grains in an originally lamellae
structure represents dynamic recrystallization (DRX). This condition of 900 C and
102 s1 lies within the high-m domain with m ¼ 0.23 (Fig. 4). This is out of any
instability domain as described by Eqs 16 and 17. Further, for this test condition,
a0 ¼ 2 (Fig. 9) and is well outside the instability condition as defined by Eq 4.
FIG. 14 Optical micrograph of Zr–2.5Nb after deformation at 900 C, 0.01 s1, and a
strain of 1.6. Pockets of dynamically recrystallized grains are seen; some of
these are marked by arrows.
FIG. 15 Optical micrograph of Zr–2.5Nb after deformation at 1000 C, 0.001 s1, and a
strain of 0.8.
For the sample deformed at 1000 C and 103 s1 and a strain of 0.8, small
equiaxed grains were seen (Fig. 15), suggesting the occurrence of DRX. At this con-
dition, m ¼ 0.3 and a0 0, and Eq 11 is not satisfied, suggesting that no instability
should occur. However, according to Eq 17 this testing condition lies in an instabil-
ity domain, as shown by the hatched area of Fig. 11. The microstructure did not
show any instability features.
From the microstructural investigation and through comparison with the iso-m
contour map and the various instability conditions, the following can be inferred.
1. The instability conditions represented by Eqs 16 and 17 do not match with
microstructural observations. Samples deformed at 1000 C and 103 s1
showed DRX and no instability, although Eq 17 predicted it. It is likely that
the physical basis for choosing the Lyapunov functions (m and s) and their in-
dependent variable (ln e)_ may be questionable.
2. The instability criteria as defined by a0 > 5 showed that all conditions of strain
rate and temperature studied were stable. However, at low temperatures and
high strain rates, macroscopic flow localization was observed, which was not
captured by the a0 > 5 criteria.
3. The instability criteria as defined by Eq 11 (i.e., 2m < g) and shown in Fig. 10
predicted instability at high strain rates and lower temperatures. Samples
deformed at these conditions also showed macroscopic flow localization. This
appears to be a reasonable criterion for predicting instability. However, it
should be pointed out that at conditions where 2m < g, it was also seen that
m was low (Fig. 4), implying that the sample had a propensity for flow localiza-
tion. Further, the questionable physical basis of deriving Eq 11 suggests that
the matching of the instability domain of 2m < g might be due to low m
rather than Eq 11.
Conclusion
Hot deformation of cast Zr–2.5Nb showed a single domain that occurred over the
temperature range of 800 C to 1100 C and strain rates of less than 0.1 s1 with a
peak strain rate sensitivity of 0.3 at 1000 C and 103 s1. A regime of low m was
observed at low temperatures and high strain rates. Regimes of flow instability were
identified by different criteria. The criteria of a0 > 5 was not satisfied at any deforma-
tion conditions, implying that the entire strain rate and temperature space tested
represents stable flow. The criteria 2m < g predicted unstable flow at low tempera-
tures and high strain rates (Fig. 10), and macroscopic flow localization was indeed
observed at these conditions. The instability criteria as expressed by the Lyapunov
function (Eqs 16 and 17) bisected the high-m domain. However, microstructural
investigations in the predicted regime did not reveal instability; for example, for the
sample deformed at 1000 C and 103 s1, small grains representing DRX were
observed rather than any instability feature. It is likely that the Lyapunov function or
its variable used here might not be the correct choice. From the various criteria,
2m < g does appear to identify flow instability. However, conditions where 2m < g
is satisfied are also regimes of low m, and thus the instability seen might be due to
low m rather than the criteria 2m < g, which has a weak physical basis. Thus it could
just as well be said that m itself is a marker for instability; low m regimes would have
a propensity for flow localization. On the basis of the processing map generated and
metallographic observations, it was found that the cast structure can be safely broken
in both the a þ b and b phases when deformed at strain rates of less than 0.1 s1.
Deformation at low temperatures and high strain rates needs to be avoided.
References
[1] Chakravartty, J. K., Prasad, Y. V. R. K., and Asundi, M. K., “Processing Map for Hot Work-
ing of Alpha-Zirconium,” Metall. Trans. A, Vol. 22, No. 4, 1991, pp. 829–836.
[2] Chakravartty, J. K., Prasad, Y. V. R. K., and Asundi, M. K., “Processing Map and Hot
Working Characteristics of Zircaloy-2,” STP 1132, ASTM International, West Consho-
hocken, PA, 1991, pp. 48–60.
[3] Chakravartty, J. K., Banerjee, S., Prasad, Y. V. R. K., and Asundi, M. K., “Hot-Working Char-
acteristics of Zircaloy-2 in the Temperature Range of 650-950 C,” J. Nucl. Mater.,
Vol. 187, No. 3, 1992, pp. 260–271.
[4] Chakravartty, J. K., Dey, G. K., Banerjee, S., and Prasad, Y. V. R. K., “Dynamic Recrystalli-
sation During Hot Working of Zr-2-5Nb: Characterisation Using Processing Maps,” Mater.
Sci. Technol., Vol. 12, No. 9, 1996, pp. 705–716.
[5] Chakravartty, J. K. and Gupta, C., “Hot Working of Zirconium Alloys: Some Recent Devel-
opments,” Min. Process. Extract. Metall. Rev., Vol. 22, No. 1, 2001, pp. 197–220.
[6] Kapoor, R. and Chakravartty, J. K., “Characterization of Hot Deformation Behaviour of

Zr-2.5Nb in Beta Phase,” J. Nucl. Mater., Vol. 306, Nos. 2–3, 2002, pp. 126–133.
[7] Chakravartty, J., Kapoor, R., and Banerjee, S., “Characterization of Hot-Deformation
Behaviour of Zircaloy-2: A Comparison Between Kinetic Analysis and Processing
Maps,” Z. Metallk., Vol. 96, No. 6, 2005, pp. 645–652.
[8] Kapoor, R., Chakravartty, J. K., Gupta, C. C., and Wadekar, S. L., “Characterization of
Superplastic Behaviour in the (a+b) Phase Field of Zr-2.5wt.%Nb Alloy,” Mater. Sci. Eng.
A, Vol. 392, Nos. 1–2, 2005, pp. 191–202.
[9] Chakravartty, J. K., Kapoor, R., Banerjee, S., and Prasad, Y. V. R. K., “Characterization of Hot
Deformation Behavior of Zr-1Nb-1Sn Alloy,” J. Nucl. Mater., Vol. 362, No. 1, 2007, pp. 75–86.
[10] Chakravartty, J. K., Kapoor, R., Sarkar, A., and Banerjee, S., “Dynamic Recrystallization in
Zirconium Alloys,” J. ASTM Int., Vol. 7, No. 8, 2010, pp. 121–147.
[11] Hart, E. W., “Theory of the Tensile Test,” Acta Metall., Vol. 15, No. 2, 1967, pp. 351–355.
[12] Campbell, J. D., “Plastic Instability in Rate-Dependent Materials,” J. Mech. Phys. Sol., Vol.
15, No. 6, 1967, pp. 359–370.
[13] Jonas, J. J., Holt, R. A., and Coleman, C. E., “Plastic Stability in Tension and
Compression,” Acta Metall., Vol. 24, No. 10, 1976, pp. 911–918.
[14] Zeigler, H., “Some Extremum Principles in Irreversible Thermodynamics with

Applications to Continuum Mechanics,” Progress in Solid Mechanics, Vol. IV, I. N. Sneddon
and R. Hill, Eds., North-Holland Publishing Company, Amsterdam, 1963, pp. 93–192.
[15] Schultz, D. G. and Melsa, J. L., State Functions and Linear Control Systems, McGraw-Hill,
New York, 1967.
[16] Prigogine, I., “Time, Structure, and Fluctuations,” Science, Vol. 201, No. 4358, 1978, pp.
777–785.
[17] Alexander, J. M., “Mapping Dynamic Material Behavior,” Modelling Hot Deformation of
Steels, J. G. Lenard, Ed., Springer-Verlag, Berlin, 1989, pp. 101–114.
[18] Chauvy, C., Barberis, P., and Montheillet, F., “Microstructure Transformation During
Warm Working of b-Treated Lamellar Zircaloy-4 Within the Upper a-Range,” Mater.
Sci. Eng. A, Vol. 431, Nos. 1–2, 2006, pp. 59–67.
[19] Semiatin, S. L. and Lahoti, G. D., “Deformation and Unstable Flow in Hot Forging of
Ti-6Al-2Sn-4Zr-2Mo-0.1Si,” Metall. Trans. A Phys. Metall. Mater. Sci., Vol. 12A, No. 10, 1981,
pp. 1705–1717.
[20] Semiatin, S. L. and Lahoti, G. D., “The Occurrence of Shear Bands in Isothermal, Hot For-
ging,” Metall. Trans. A, Vol. 13, No. 2, 1982, pp. 275–288.
[21] Narayana Murty, S. V. S., Nageswara Rao, B., and Kashyap, B. P., “Instability Criteria for
Hot Deformation of Materials,” Int. Mater. Rev., Vol. 45, No. 1, 2000, pp. 15–26.
[22] Montheillet, F., Jonas, J. J., and Neale, K. W., “Modeling of Dynamic Material Behavior: A
Critical Evaluation of the Dissipator Power Co-content Approach,” Metall. Mater. Trans. A
Phys. Metall. Mater. Sci., Vol. 27, No. 1, 1996, pp. 232–235.
[23] Prasad, Y. V. R. K., Gegel, H. L., Doraivelu, S. M., Malas, J. C., Morgan, J. T., Lark, K. A., and
Barker, D. R., “Modeling of Dynamic Material Behavior in Hot Deformation: Forging of
Ti-6242,” Metall. Trans. A, Vol. 15, No. 10, 1984, pp. 1883–1892.
[24] Narayana Murty, S. V. S., Sarma, M. S., and Rao, B. N., “On the Evaluation of
Efficiency Parameters in Processing Maps,” Metall. Mater. Trans. A, Vol. 28, No. 7, 1997,
pp. 1581–1582.
[25] Gegel, H. L., Malas, J. C., Doraivelu, S. M., and Shende, V. A., “Modelling Techniques Used
in Forging Process Design,” Metals Handbook, ASM International, Materials Park, OH,
1987, pp. 417–438.
DISCUSSION
Questions from Pierre Barberis, CEZUS Research Center
Q1:—Does the deformation map depend on the strain, specifically when there
is softening? Does it depend on the initial as-cast microstructure (basket weave ver-
sus parallel plate)?
Authors’ Response:—The contour map plots strain rate sensitivity as function of

strain rate and temperature. The strain rate sensitivity can be determined in two
ways, one with instantaneous changes in strain rate and the other by comparing
flow stress curves at different strain rates. In the latter case, it is desired that the
flow stress be taken at large strains, preferably at steady state, where the microstruc-
ture evolution is also in steady state. Within the steady state regime, it is obvious
that the map will be independent of strain. In regions where the flow stress varies
with strain, such as during softening, the map may depend on strain depending on
the nature of the curves, i.e., whether or not they are parallel to each other. In the
present study where the curves (plotted on log stress axis) are parallel to each other,
the strain rate sensitivity will be independent of strain.
The steady state flow stress does not depend on initial microstructure and is deter-
mined by the final steady state microstructure. The initial microstructure may only
decide the path it takes to reach the steady state value, which is the evolution of the
microstructure from initial to final state.
Q2:—These deformation maps are established for uniform strain/strain rate or

stress. Can they be translated directly for forging or extrusion processes, where
strain rate and stress vary within the sample?
Authors’ Response:—Hot deformation experiments are carried out at uniaxial

conditions and what is determined is in effect the constitutive behavior where stress
is related to strain, strain rate and temperature. (Assuming that steady state is
reached at small strains, we have a relation between stress, strain rate and tempera-
ture only). During forging or extrusion, the state of stress and strain has all six com-
ponents of the tensor. To use the uniaxial data for hot working, finite element
methods (FEM) would need to be used. The constitutive behavior of the material
needs to be input in an FEM program and the effective (Von Mises) stress, strain
and strain rate need to be determined for the forging and extrusion process. The
Von Mises strain rate would vary within the work piece, however, an average (or a
range) of this effective strain rate could be identified. This effective strain rate can
then be achieved by choosing the appropriate working parameters, i.e., ram speed,
such that the effective strain rate and temperature lie within the domain in the
map.
N. Saibaba,1 N. Keskar,2 K. V. Mani Krishna,2 V. Raizada,3

K. Vaibhaw,3 S. K. Jha,3 D. Srivastava,2 and G. K. Dey2
A Numerical Study of the Effect

of Extrusion Parameters
on the Temperature Distribution
in Zr–2.5Nb
Reference
Saibaba, N., Keskar, N., Mani Krishna, K. V., Raizada, V., Vaibhaw, K., Jha, S. K., Srivastava, D.,
and Dey, G. K., “A Numerical Study of the Effect of Extrusion Parameters on the Temperature
Distribution in Zr–2.5Nb,” Zirconium in the Nuclear Industry: 17th International Symposium,
STP 1543, Robert Comstock and Pierre Barberis, Eds., pp. 282–301, doi:10.1520/
ABSTRACT
A three-dimensional finite element method (FEM) analysis of the hot extrusion
process of Zr 2.5 %Nb was carried out. Simulations were carried out for bringing
out the role of extrusion temperature, extrusion ratio, ram velocity, and profile of
die on the extrusion process. The effect of these process parameters on the
overall temperature distribution and hence on the resulting microstructural and
textural homogeneity or lack of it is discussed. FEM analysis allowed for the
quantification of the extent of differences in temperature distribution between
leading and trailing end and also across a given cross section.
Keywords
pressure tube, hot extrusion, simulation, FEM, Zr-2.5, wt. %Nb
Manuscript received January 25, 2013; accepted for publication April 12, 2014; published online October 6,
2014.
1
Nuclear Fuel Complex, Hyderabad, 500062 India (Corresponding author), e-mail address: nsai@nfc.gov.in
2
Material Science Division, Bhabha Atomic Research Center, Trombay, Mumbai, 400085 India.
3
Nuclear Fuel Complex, Hyderabad, 500062 India.
4
Hyderabad, India.
Copyright V
SAIBABA ET AL., DOI 10.1520/STP154320130023 283
Introduction
Zirconium alloys are extensively used in various types of fission reactors such as light
and heavy water based reactors as the proven structural material on account of good
high temperature mechanical properties and low neutron absorption cross-section
[1–3]. In pressurized heavy-water reactors (PHWRs) of Canada Deuterium Uranium
(CANDU) type, the fuel assemblies are placed inside tubes of Zr–2.5Nb alloy which
form the primary pressure boundary. These pressure tubes experience a high degree of
mechanical stresses along with the irradiation induced damage from high neutron flux
for a prolonged period of time [4–8]. The life of a pressure tube in a reactor depends
on the total deformation of the tube due to irradiation creep and growth. In order to
obtain an average lifetime of 20–25 years with acceptable levels of dimensional changes,
the tubes need to have appropriate texture and microstructure to resist the harsh service
environment. Typically textural and microstructural control is achieved by appropriate
thermomechanical fabrication processing [9–11]. In case of Zr 2.5Nb pressure tubes
fabrications, one of the most important fabrication steps that has a high bearing on the
final properties of the tube is hot extrusion [12–15]. Previous studies [16] show that the
crystallographic texture produced during the hot extrusion stage largely remains unal-
tered even after further cold work.
In addition, morphology and distribution of the a and b phase also get greatly
influenced during this hot extrusion stage only. These microstructural and textural
features are influenced by the actual temperature and stress distribution during the
hot extrusion process. In addition, extruded product quality in terms of variation
from leading end to tailing end, surface cracks, ovality, etc., also depend on the
presence of steep temperature gradients during the hot extrusion. Hence a detailed
knowledge of the role of process parameters such as billet preheat temperature,
extrusion speed, extrusion ratio, etc., on the temperature and stress distribution
during the hot extrusion process can help tailor the process [17–22]. However, the
design and control of the experiments to acquire such information during hot
extrusion process is rather difficult [21,22]. An alternative approach could be appli-
cation of computational and/or numerical analysis of the extrusion process for the
process optimization. In the recent past, the three-dimensional finite element analysis
for hot extrusion was being pursued in case of aluminium based materials [23–27].
Such studies in case of the Zr based alloys seem to be rather limited in the open litera-
ture, forming the motivation for the present study. Some of the key parameters of the
extrusion process such as ram velocity, billet pre-heat temperature, fillet radius, and
reduction ratio, have been studied for their role in determining the overall tempera-
ture distribution during hot extrusion, requirement of ram force, etc.
Hot Extrusion Model

The three main types of FE methods that are utilized in extrusion simulation are
Lagrangian, Eulerian, and Arbitrary Lagrangian Eulerian (ALE). The ALE method
is essentially an arbitrary combination of the Lagrangian and Eulerian methods and

attempts to bring the advantages with both Lagrangian and Eulerian formulations
together. In an ALE formulation, the displacements of material and mesh are
decoupled and the mesh can move independently of the material. The ALE formu-
lation is ideally suited for modeling of fluid-structure interaction and motion of free
surfaces in fluid mechanics. Since the heavy plastic deformation involved in extru-
sion resembles fluid flow, the ALE method is well suited for extrusion simulations.
The present work uses HyperXtrude, a commercial simulation software module
dedicated to simulation of extrusion process using the ALE approach [28]. The
models were prepared using a structured mesh with appropriate biasing to maintain
accuracy and continuity across the faces of different elements. Figure 1 shows the
geometry of the model along with the various boundaries and the way interactions
with surrounding are modeled. As can be seen, the model is made up of different
zones representing the container portion, die cavity, die land, and exit region. The
interaction of the billet material with the surroundings is modeled through appro-
priate heat transfer and friction boundaries as show in the figure. Error in heat bal-
ance and mass balance at successive simulation time steps were kept below 5 % and
1 %, respectively, in order to achieve good convergence in a reasonable time frame.
MATHEMATICAL FORMULATION
The fundamental equations that govern flow and heat transfer of incompressible
viscous fluids are derived from principles of conservation of mass, momentum, and
energy. These equations are written in terms of primitive variables (velocity, pres-
sure, and temperature) with reference to an Eulerian frame, i.e., a space-fixed sys-
tem of coordinates through which the fluid flows [28].
(1a) rU¼0
FIG. 1 Schematic explaining the meshing scheme, various components of the model
geometry and various interactions of the model with the bounding surfaces.
(1b) qðU rÞU ¼ r ðrÞ

(1c) qCP U rT ¼ r q þ u
where:
U ¼ the velocity vector,
r ¼ the total stress tensor,
T ¼ the temperature,
q ¼ the mass density,
CP ¼ the specific heat of the fluid at constant pressure,
q ¼ the heat flux vector, and
u ¼ internal heat generation rate due to viscous dissipation.
CONSTITUTIVE MODEL
For viscous incompressible fluids, the components of total stress tensor (r), can be
represented as the sum of viscous stress tensor (the deviatoric part of the stress ten-
sor) (s), and a spherical hydrostatic pressure (P).
(2) r¼sP
The material behavior is specified by the constitutive relations for the viscous stress
tensor, (s) and the heat flux vector (q) by the following equations [28]:
(3a) s ¼ 2lc
(3b) c ¼ ½ðrUÞ þ ðrUÞT

(3c) q ¼ krT
In the above equations, l is the viscosity, and k is the isotropic thermal conduc-
tivity of the material. This viscosity is a function of strain rate and temperature of
the material and typically described using sine-hyperbolic type constitutive relation:
r
(4a) l ¼ pffiffiffiffiffi
3_
1
(4b) e ¼ 23 ½c : cð3Þ
In case the flow stress values cannot be fit into such constitutive equation, as is
the case for Zr–2.5Nb alloys (the material of present study), tabulated experimental
data for flow stresses at different temperatures, strains, and strain rates can be uti-
lized as explained in the section titled “Material Properties Used for Simulations.”
The following are the boundary conditions assigned to various regions of the
simulation model (see Fig. 1):
• Convective Heat Transfer boundary: all tool face boundaries are assigned con-
vective heat transfer boundary, i.e., heat removal from these faces is through
convection.
• Friction Boundary: when shear stress over contact surfaces exceeds critical
shear stress, material starts to flow.
• Inflow Boundary: a constant velocity and temperature boundary condition is

assigned at billet ram face.
• Free Surface Boundary: free surface boundary is assigned at profile surface.
This boundary is assigned an insulated boundary, i.e., q (heat flux) ¼ 0.
• Outflow Boundary: this boundary is assigned to profile face.
• Traction forces ¼ 0 (as the extrudate is not pulled out of the die),
displacement ¼ 0 (as extrudate is free to expand), heat flux ¼ 0.
• Symmetric Boundary: this is assigned to symmetric faces.
Mathematically, these are expressed as billet-ram interface:
(6) U ¼ URam ; T ¼ TRam
billet-container interface:
(7) tS ¼ mrN ; T ¼ TContainer
die face:
(8) sS ¼ CðU UTool Þ; T ¼ TContainer
bearing surface:
(9) sS ¼ CðU UTool Þ
free surface:
(10) rn¼0
Since the surfaces of tools interacting with the billet (container, die cavity, etc.) are
represented through appropriate boundary conditions, they need not be included in
the analysis exclusively and hence are not meshed.
MATERIAL PROPERTIES USED FOR SIMULATIONS

The accuracy of the results of numerical simulations is determined by the extent to
which the subject material and the environment are accurately represented in the
model. The main inputs to any deformation modeling code are material properties
and boundary conditions. As for the mechanical properties of the material, the flow
stress data as a function of strain, strain rate, and temperature have to be intro-
duced to the FEM software. Conventionally, constitutive equations such as power
law or sine-hyperbolic inverse law are utilized by fitting the experimental flow prop-
erty data to suitable equations [29]. In the present case of Zr–2.5Nb alloys, flow
stress values were experimentally determined for a range of strains, strain rates, and
temperatures. However, these flow stress values could not be fitted to any single
constitutive equation. This is on account of the fact that, within the range of tem-
peratures of interest, complications due to phase transformation, variation of rela-
tive amounts of phase with respect to temperature render fitting mechanical
properties to a single equation difficult. Hence, a Look-Up Table (LUT) was used
which had experimentally measured flow stress values for compression tests at a
number of temperatures, strains and strain rates (see Table 1). These experiments
were carried out on a Gleeble Thermo-mechanical Simulator. Uniaxial compression
tests on homogenized cylindrical samples were conducted over a range of tempera-
tures 650 C –950 C (in steps of 50 C) and strain rates 0.001–100 s1 using a com-
pression test rig. The compression rig consists of high thermal conductivity grips.
The sample was heated by direct resistance heating leading to uniform temperature
throughout the sample. A graphite lubricant was used to minimize barreling and
obtain a uniform distribution of strain. The flow stress values for any intermediate
values of temperature, strain, and strain rate were calculated by trilinear interpola-
tion scheme. The extrusion force required was plotted against the ram displace-
ment. Such load profiles from simulations were compared to those obtained from
actual experimental shop-floor data. The shop-floor ram-force data is recorded by
an electronic data acquisition system. The set of friction and heat transfer coeffi-
cient values which showed the best fit to the experimental ram force profile (see
Fig. 2) was chosen for further simulations. As can be seen, a friction coefficient of
0.1 and a heat transfer coefficient of 100 Wm2K1 have given the best match for
the force versus time plots from simulations and the experimental values. Thus
these values were used for all of the rest of the simulations presented in the present
paper in which parametric variation of other extrusion variable such as extrusion
ratio, billet preheat temperature etc are studied.
Simulations
The geometry of the extrusion die, punch, and billet used for simulations is shown
in Fig. 3. The dimensions of the billet were chosen to correspond to that of the one
used in the actual pressure tube fabrication process at NFC, Hyderabad. In order to
systematically bring out the role of the important process parameters (such as
extrusion temperature, extrusion ratio, ram velocity, etc.) on the overall
TABLE 1 Look-Up Table (LUT) showing different experimental flow stress values for different
temperatures, strains and strain rates. Flow stress values for only one strain value are
shown here.
Flow Stress (Mpa) at Temperature ( C)
Strain Rate
Strain (s1) 650 700 750 800 850 900 950
0.5 0.001 72.7 50.3 33.5 15.1 13.8 10.9 8.1

0.01 122.3 80.8 61.8 36.5 27.8 18.2 16
0.1 143 118.4 75 66.4 44.2 30.2 25.8
1 332.8 180.1 129.7 92.7 60.4 45.7 39.3
10 303.1 234.6 179.6 120.5 88.1 78 70.7
100 286.3 264.1 222.4 151.2 120 100.7 87.3
FIG. 2 Optimizing convective heat transfer and friction coefficients to conform to

experimental ram force curve represented by solid black line, “Shop Floor Data.”
temperature distribution and hence the expected variability introduced across dif-
ferent regions (e.g., leading end and tailing end), simulations were carried out by
varying these process parameters to cover values both above and below values
reported in literature [19,22,30]. Simulations were carried out considering a con-
stant billet container temperature of 150 C. Since only one out of all the chosen
process parameters can be varied to understand its influence, one has to select a set
of fixed values for remaining parameters. These fixed values were chosen based on
the general values reported in the literature. These values are ram velocity (30 mm/
s), billet preheat temperature of 815 C (two phase region), included die angle of
90 , reduction ratio of 14.44, and fillet radius of 10 mm. For every simulation run,
FIG. 3 Schematic geometry of extrusion setup showing billet, mandrel, die angle, fillet
radius, outer diameter (OD), inner diameter (ID), wall thickness (WT).
the other invariant independent parameters were kept at the actual shop-floor val-
ues as mentioned above.
1. Effect of ram velocity: OD 119 mm, WT 6 mm, billet preheat temperature
815 C, included die angle 90 , fillet radius of 10 mm, reduction ratio 14.44,
ram velocities: 20 mm/s, 47 mm/s, 75 mm/s, 125 mm/s.
2. Role of reduction ratio: OD 119 mm, ram velocity 30 mm/s, fillet radius of
10 mm, billet preheat temperature 815 C, included die angle 90 , reduction
ratios: 6, 8, 10, 12, 14.44, 20.
3. Effect of preheat temperature: OD 119 mm, WT 6 mm, ram velocity 30 mm/s,
included die angle 90 , fillet radius of 10 mm, reduction ratio 14.44, billet pre-
heat temperatures: 775 to 855 C at intervals of 10 C.
4. Effect of fillet radius: OD 119 mm, WT 6 mm, ram velocity 30 mm/s, included
die angle 90 , billet preheat temperature 815 C, reduction ratio 14.44, fillet
radii: 1, 5, 10, and 20 mm.
Regions along the length of the tube were classified as leading and tailing ends.
This was done based on the time based variation of temperature at the point of
highest temperature (Fig. 4). The fillet region corresponds to the highest variation
in temperature and subsequently the highest variation in microstructure, and
thereby properties. As such, the temperature profile (mandrel to die, radially) at the
fillet region is plotted at two time instants: once at the beginning of extrusion and
once just before the end of extrusion. These represent the temperatures experienced
by the leading end and the tailing end of the extrudate, respectively. These tempera-
ture profiles are shown in Figs. 6, 13, and 15 for variations in ram velocity, pre-heat
temperature, and fillet radius, respectively.
FIG. 4 The temperature contours in a typical simulation. The location of the region of
highest temperature is shown. The variation of this temperature with time is
used to judge the expected variation in the leading and tailing end of the
resulting extrudate.
Results and Discussions

EFFECT OF VARIATION IN RAM VELOCITY
Figure 5 shows the ram force variation with ram displacement for different values of
constant ram velocities (30, 47, 75, and 125 mm/s), other parameters remaining
constant at shop-floor values. For each ram velocity, the billet preheat temperature
was kept at 815 C and the fillet radius was kept 10 mm. The required ram force, as
expected, is seen to increase with the increase in the ram velocity (Fig. 5). It can be
observed from the figure that peak ram force varied from 1836 tons for 20 mm/s
ram velocity to 2337 tons for a ram velocity of 125 mm/s, an increase of 27 % in
peak ram force, signifying the importance of the ram velocity on the overall extru-
sion load requirements. Furthermore, the experimentally measured force is
observed to be reasonably closer to the simulated curve for a ram speed correspond-
ing to 47 mm/s, particularly in terms of the peak force and steady state force, with a
deviation less than 5 % and 4 %, respectively. The measured force versus displace-
ments curve does show some oscillations during transition from the peak force to
steady state force mainly due to the variations introduced in the ram velocity under
practical conditions of the extrusion which are rather difficult to incorporate in the
simulations. Nevertheless, reasonably close predictions of the simulations for the
two characteristic loads of the extrusion process (i.e., peak and steady state ram
force) enable us to apply the simulation results for the analysis of temperature pro-
files across billet during the extrusion process. The observed ram force versus veloc-
ity curves for this set of simulations, are the reflection of balance achieved between
the two competing factors governing the force requirement. One is requirements of
FIG. 5 Ram force variation for various ram velocities.

higher force arising out of application of higher strain rate deformation (for higher
ram velocities) and the other is decreased force requirement resulting from
increased temperature due to higher strain rate deformation corresponding to
higher ram velocities, and lower heat losses due to lower residence time. Since the
first factor dominates during the initial stage of the extrusion, one would expect
more change in ram force as a function of ram velocity. This is clearly brought by
the simulations which predict 27 % of change in the peak rams force, in comparison
to 17 % of change in the steady state force requirement for the same change in the
ram velocities. Apart from ram force variation, ram velocity is expected to influence
the overall temperature distribution significantly. More importantly, differences
between the temperature profiles across a given cross section can also be influenced
which are important from the point of view of structural homogeneity and residual
stresses. Figure 6 is a depiction of the temperature profile across fillet region, at dif-
ferent times corresponding to the leading end and trailing end of the extrusion. As
can be observed from the figure, as the ram velocity increases, the temperature pro-
files of leading end and tailing end tend to converge. Such a convergence implies
the presence of a more homogeneous structure along the length of the tube/extru-
date because temperature is the defining parameter for microstructure develop-
ment, with the stress state being nearly constant. However, at the same time, this
FIG. 6 The leading (LE) and tailing end (TE) temperature profiles for ram speeds of (a)
20 mm/s, (b) 47 mm/s, (c) 75 mm/s and (d) 125 mm/s.
convergence comes at the cost of increasing DT, i.e., the temperature difference,
across the cross section of the tube (mandrel to die). This may introduce heteroge-
neity across the thickness of the tube.
Figure 6 shows that higher ram velocities give rise to steeper temperature gra-
dients across the cross section of a tube and the range of temperature difference
between the front end and back end at any given cross section was found to be in
the range of 100 to 140 K for the ram velocities considered in the present work.
Conversely, simulations that predict lower ram velocities should result in lower
temperature and hence microstructural variations across any given cross section. In
fact, low or negligible temperature and microstructural variations across the tube
cross section have been reported in literature [30] at low ram velocities; this is in
direct agreement with our present simulation results. However, lower ram velocities
also result in a more pronounced chilling effect of ram and container coupled with
decreased adiabatic heating as a consequence of lower strain rate deformations. It
may be pointed out that previous works have reported a drop in the ram side tem-
perature of the workpiece to the extent of 300 K, at ram velocities of the order of
3 mm s1 using a work piece much shorter than what is being considered in the
present study. This data, when extrapolated to work a piece of the dimensions of
the present study, indicates a similar chilling effect of ram (see Fig. 7). Two main
points emerge out of the present discussion: (1) The temperature variation across
the length is much more than across the section at lower ram velocities coupled
with higher residence time for the work piece at any given temperature. (2) In con-
trast, at higher ram velocities, temperature variation across cross section dominates
that of the one across length, with much lower residence time of the work piece at
any given temperature.
These observations, when viewed in the context of the well known sluggishness
of a ! b transformation, point out that higher temperature variations across a
FIG. 7 The leading and trailing end cross sectional temperature profiles at low ram
velocities of (a) 1 mm s1 and (b) 3 mm s1. At low ram velocities, the chilling
effect of ram dominates the deformation heating, leading to larger temperature
gradients along the length as compared to those across the cross section.
given cross section resulting from higher ram velocities may not translate in wide
microstructural variation due to the small residence time available for microstruc-
tural changes. On the contrary, significant temperature variations across the length
of the workpiece at lower ram velocities may result in appreciable microstructural
and textural changes across the length of the extrudate. Experimental evidence
from previous works also support this premise.
In view of the detrimental effects of cross sectional heterogeneity on the per-
formance of pressure tubes, imposition of low ram velocities seems advisable. How-
ever, ram stroke also contributes to adiabatic heat generation, i.e., heat generation
due to plastic deformation. In addition, the ram acts as a heat sink. Hence, very low
ram velocities will contribute to chilling of the billet. This effect will be most domi-
nant in the trailing end of the tube on account of longer residence time in the con-
tainer. The microstructural effects of this leading end to trailing end temperature
variation have been demonstrated in another study [30]. The present simulations
show that hot extrusions carried out at higher ram velocities (20–125 mm/s) show
much smaller variations from leading end to trailing end (see Fig. 6). Although
higher ram velocities lead to larger cross section variation, this variation is miti-
gated by the small residence time in the dies (15 s).
EFFECT OF VARIATION IN REDUCTION RATIO

Reduction ratio effectively denotes the amount of deformation that the workpiece is
subjected to during extrusion. Since deformation work is the most energy intensive
part of hot extrusion, one would expect that reduction ratio should have significant
influence on the required extrusion force [13]. Figure 8 brings out this strong
dependence of ram force on the extrusion ratio. Although the application of a larger
reduction ratio can give a thinner extrudate section (see Fig. 9), large deformations
have certain undesirable effects, too. As shown in Fig. 10, there exists a temperature
gradient from the ID (mandrel) to OD (Die) cross section. In case of hot extrusion,
two competing phenomena of heat transfer are in play. Firstly, the dies and the
mandrel absorb heat from the billet/workpiece; secondly, the workpiece gets locally
heated due to deformation work done and frictional effects. Figure 10(a) shows that
there is a drop in temperature (from the preheat temperature of 815 C) near the
mandrel, whereas near the fillet region, there is a sharp rise. The flow velocity of the
workpiece near the mandrel remains axial throughout the length of the die cavity.
On the other hand, the velocity near the die surface experiences a sudden change
near the fillet region. Such sudden velocity change translates into high strain rate
and corresponding rise in temperature. Thus, an inherent heterogeneity exists in
the extrudate cross section. However, the undesirable effects of such a temperature
profile can be alleviated if the temperature gradients are smooth and not steep. As
shown in Fig. 10, larger reduction ratios generate very steep temperature gradients
across the cross section (ID to OD). Such steep gradients can be precursors to high
residual stresses, which may lead to surface cracks. In the domain of the present
FIG. 8 Ram force variation for different reduction ratios.
FIG. 9 Sequence of extrudate profiles for reduction ratios of (a) 6, (b) 8, (c) 10, (d) 12,
(e) 14.4 and (f) 20.
FIG. 10 The leading (LE) and tailing end (TE) temperature profiles for Reduction Ratios
of (a) 6 and (b) 20.
study, it is seen that lower reduction ratios are desirable on account of lower tem-
perature gradients (Fig. 11).
EFFECT OF PREHEAT TEMPERATURE

Figure 12 shows ram force variation with ram stroke or displacement for preheat
temperatures ranging from 775 to 855 C at intervals of 10 C. Increasing the pre-
heat or soaking temperature lead to a decrease in the required ram force as
expected. A closer observation of the temperature profiles across a given region,
shown in Fig. 13, reveal that a given increase in the preheat temperature will not
result in same amount of increase in the peak temperature attained in that region.
For example, a change of 60 C in preheat temperature (between case of 795 C to a
FIG. 11 The temperature gradient profiles for Reduction Ratios of (a) 6 and (b) 20.
FIG. 12 Ram force variation for different values of preheat temperatures.
case of 855 C) resulted in the peak temperature to increase by an amount of 40 C.

This translates in to lower temperature variations across a given cross section, a
beneficial effect in terms of achieving microstructural homogeneity. This benefit is
in combination with the advantage of reduced ram force requirement. Thus the
present analysis indicates that using higher pre-heat temperatures within the tem-
perature domain considered is beneficial for overall homogeneity across cross sec-
tion of the extrudate.
EFFECT OF VARIATION IN FILLET RADIUS

Fillets are introduced in the die to soften steep flow velocity gradients. That is, fillets
are expected to lower strain rates, which are generated due to sharp or sudden
changes in material flow direction in any setup. These ensure a streamlined flow of
material and prevent damage to dies / flow channels. In addition, these ensure
homogeneity in the finished product by virtue of uniform filling up of dies. Hence,
a systematic variation in fillet radii was simulated to capture the effect of changing
the fillet radii on the extrusion process. As fillet radius was increased, the ram force
required increased (see Fig. 14). This is seen as elevation in the ram force profile
curves for larger radius of fillet. Although increased ram force was observed for the
increase in the fillet radii, the changes in the temperature profiles with the changing
fillet radii are seen to be confined to the surface near to fillet. In other words,
increasing fillet radius led to a temperature rise in the region directly in contact
with the die (fillet) while the temperature near the mandrel remained virtually unaf-
fected (see Fig. 15). This significant increase in temperature near the fillet surface,
can be attributed to an increase in the surface area, when the fillet radius is
FIG. 13 The leading (LE) and tailing end (TE) temperature profiles for billet preheat
temperatures of (a) 785, (b) 805, (c) 825, and (d) 855 C.
FIG. 14 Ram force variation for different values of fillet radii.

increased. The increased surface area leads to increased frictional drag causing
localized temperature rise. Except for the rise in the temperature near the region in
contact with the fillet, the temperature profile across a given cross section in the
fillet region remained reasonably unaffected other for all of the fillet radii consid-
ered in the present work.
Summary
In the present work, a systematic parametric study of various extrusion parameters
has been conducted using a three-dimensional finite element analysis. The results
have brought out the role of various process parameters on the zone specific tem-
perature distributions, and overall extrusion force requirements. Analysis of tem-
perature gradient arising out of variation in various process parameters has cast
light on the possible extent of inhomogeneity introduced during the hot extrusion
process. The results of the simulation can be summarized as follows:
• Lower ram velocities were found to result in appreciable temperature loss at
the tailing end, thus giving rise to more leading end to tailing end variation
while higher ram velocities were found to increase the variation across the
thickness.
• Higher reduction ratios were found to increase the temperature gradients
across the thickness due to increased strain gradients at die profile side.
• Frictional heating at the die profile was found to be more substantial than the
temperature drop due to mandrel chilling effect.
• Simulations showed that peak force of extrusion is stronger function of the
ram velocity than the steady state force.
• Simulations indicate that using of higher preheat temperature will have benefi-
cial effect in terms of overall homogeneity.
References
[1] Murty, K. L. and Charit, I., “Texture Development and Anisotropic Deformation of
Zircaloys,” Prog. Nucl. Energy, Vol. 48, No. 4, 2006, pp. 325–359.
[2] Krishnan, R. and Asundi, M., “Zirconium Alloys in Nuclear Technology,” Proc. Indian
Acad. Sci. Sect. C: Eng. Sci., Vol. 4, No. 1, 1981, pp. 41–56.
[3] Dodd, C., Hobson, D., Thoms, K., and van der Kaa, T., “Effects of Temperature and Pres-
sure on the In-Reactor Creepdown of Zircaloy Fuel Cladding,” ASTM Special Technical
Publication 754, ASTM International, Philadelphia, PA, 1982, pp. 173–192.
[4] Cox, B., “Hydride Cracks as Initiators for Stress Corrosion Cracking of Zircaloys,” ASTM
Special Technical Publication No. 681, ASTM International, Philadelphia, PA, 1979, pp.
306–321.
[5] Holt, R. and Ibrahim, E., “Factors Affecting the Anisotropy of Irradiation Creep and
Growth of Zirconium Alloys,” Acta Metall., Vol. 27, No. 8, 1979, pp. 1319–1328.
[6] Mani Krishna, K., Sain, A., Samajdar, I., Dey, G., Srivastava, D., Neogy, S., Tewari, R., and
Banerjee, S., “Resistance to Hydride Formation in Zirconium: An Emerging Possibility,”
Acta Mater., Vol. 54, No. 18, 2006, pp. 4665–4675.
[7] Kearns, J. and Woods, C., “Effect of Texture, Grain Size, and Cold Work on the Precipitation
of Oriented Hydrides in Zircaloy Tubing and Plate,” J. Nucl. Mater., Vol. 20, No. 3, 1966, pp.
241–261.
[8] Kim, Y., Perlovich, Y., Isaenkova, M., Kim, S., and Cheong, Y., “Precipitation of
Reoriented Hydrides and Textural Change of a-Zirconium Grains During Delayed Hydride
Cracking of Zr–2.5 %Nb Pressure Tube,” J. Nucl. Mater., Vol. 297, No. 3, 2001, pp.
292–302.
[9] Tenckhoff, E. and Rittenhouse, P., “Annealing Textures in Zircaloy Tubing,” J. Nucl. Mater.,
Vol. 35, No. 1, 1970, pp. 14–23.
[10] Ballinger, R., Lucas, G., and Pelloux, R., “The Effect of Plastic Strain on the Evolution of
Crystallographic Texture in Zircaloy-2,” J. Nucl. Mater., Vol. 126, No. 1, 1984, pp. 53–69.
[11] Mahmood, S., 1989, “Anisotropic Plastic Deformation, Formability and Crystallographic
Texture of Zircaloy-4 Sheet,” Ph.D. thesis, North Carolina State University, Raleigh, NC.
[12] Fujita, K., Kakuma, T., and Nagai, N., “Texture Control of Zircaloy Tubing During Tube
Reduction.,” ASTM Special Technical Publication No. 754, ASTM International, Philadel-
phia, PA, 1982, pp. 26–32.
[13] Lapovok, R., Barnett, M., and Davies, C., “Construction of Extrusion Limit Diagram for
az31 Magnesium Alloy by FE Simulation,” J. Mater. Process. Technol., Vol. 146, No. 3,
2004, pp. 408–414.
[14] Li, L., Zhou, J., and Duszczyk, J., “Determination of a Constitutive Relationship for AZ31B
Magnesium Alloy and Validation Through Comparison Between Simulated and Real
Extrusion,” J. Mater. Process. Technol., Vol. 172, No. 3, 2006, pp. 372–380.
[15] Malpani, M. and Kumar, S., “A Feature Based Analysis of Tube Extrusion,” J. Mater.
Process. Technol., Vol. 190, Nos. 1–3, 2007, pp. 363–374.
[16] Krishna, K. M., Sahoo, S., Samajdar, I., Neogy, S., Tewari, R., Srivastava, D., Dey, G., Das, G.
H., Saibaba, N., and Banarjee, S., “Microstructural and Textural Developments
During Zircaloy-4 Fuel Tube Fabrication,” J. Nucl. Mater., Vol. 383, No. 12, 2008, pp. 78–85.
[17] Picklesimer, M., “Deformation, Creep, and Fracture in a-Zirconium Alloys,” Electrochem.
Technol., Vol. 4, Nos. 7–8, 1966, pp. 289–300.
[18] Ells, C. E. and Cheadle, B. A., “The Anisotropy of Fracture Ductility in Flat Tension-Test
Bars of a-Zirconium Alloy,” ASTM Special Technical Publication No. 458, ASTM Interna-
tional, Philadelphia, PA, 1969, pp. 68–91.
[19] Alexander, W. K., Filderis, V., and Holt, R. A., “Zircaloy-2 Pressure Tube Elongation at the
Hanford N-Reactor,” ASTM Special Technical Publication No. 633, ASTM International,
Philadelphia, PA, 1977, pp. 344–363.
[20] Coleman, C. E., “Effect of Texture on Hydride Reorientation and Delayed Hydrogen
Cracking in Cold-Worked Zr-2.5Nb,” ASTM Special Technical Publication No. 754, ASTM
International, Philadelphia, PA, 1982, pp. 393–411.
[21] Holt, R. A. and Zhao, P., “Micro-Texture of Extruded Zr-2.5Nb Tubes,” J. Nucl. Mater., Vol.
335, 2004, pp. 520–528.
[22] Holt, R. A. and Aldridge, J., “Effect of Extrusion Variables on Crystallographic Texture of
Zr-2.5 wt% Nb,” J. Nucl. Mater., Vol. 135, 1985, pp. 246–259.
[23] Yang, D. and Kim, K., “Design of Processes and Products Through Simulation of Three-
Dimensional Extrusion,” J. Mater. Process. Technol., Vol. 191, 2007, pp. 2–6.
[24] Duan, X., Velay, X., and Sheppard, T., “Application of Finite Element Method in the Hot
Extrusion of Aluminium Alloys,” Mater. Sci. Eng., Vol. A369, 2004, pp. 66–75.
[25] Bontcheva, N., Petzov, G., and Parashkevova, L., “Thermomechanical Modelling of Hot
Extrusion of Al-Alloys, Followed by Cooling on the Press,” Comput. Mater. Sci., Vol. 38,
2006, pp. 83–89.
[26] Yang, D., Lee, C., and Yoon, J., “Finite Element Analysis of Steady-State Three-
Dimensional Extrusion of Sections Through Curved Dies,” Int. J. Mech. Sci., Vol. 31,
1989, pp. 145–156.
[27] Lee, C., Yang, D. Y., and Kim, M., “Numerical Analysis of Three-Dimensional Extrusion of
Arbitrarily Shaped Sections by the Method of Weighted Residuals,” Int. J. Mech. Sci., Vol.
32, 1990, pp. 65–82.
[28] Mahender, P. R., Mayavaram, R., Durocher, D., Carlsson, H., and Bergquist, O., “Analysis
and Design Optimization of Aluminum Extrusion Dies,” Proceedings of the 8th Interna-
tional Extrusion Technology Seminar, Orlando, FL, May 18–21, 2004.
[29] Li, L., Rao, K., Lou, Y., and Peng, D., “Hot Deformation Characteristics of Ti-6Al-4V,” Int.
J. Mech. Sci., Vol. 44, 2002, pp. 2415–2425.
[30] Li, Y., Rogge, R., and Holt, R., “Development of Local Microstructure and Crystallographic
Texture in Extruded Zr-2.5Nb Tubes,” Mater. Sci. Eng. A, Vol. 437, No. 1, 2006, pp. 10–20.
DISCUSSION
Questions from Alexis Gaillac, CEZUS/AREVA
Q1:—Is the temperature of the tools (container and die) constant in your
model?
Authors’ Response:—Yes, the temperatures of the container and die were kept
constant at 150̄C. The temperature was selected based on shop floor practice.
Q2:—Did you evaluate the influence of tool temperature on maximum tube

temperature, and especially the difference from start to end of extrusion?
Authors’ Response:—Since in the present model, the temperature of the tool is

considered constant, its influence on maximum tube temperature has not been
investigated.
Questions from Pierre Barberis, CEZUS Research Center
Q1:—You mentioned briefly lubrication during extrusion. Which lubricant is

used (e.g., copper, glass)?
Authors’ Response:—The copper jacketing used on Zr-2.5Nb extrusion is for

protection against oxidation. For lubrication purposes, oil based graphite lubricant
is used.
Q2:—Does the friction coefficient remain constant with the temperature?
Authors’ Response:—In the present model, the friction coefficient was kept con-
stant with temperature.
Question from Malcolm Griffiths, AECL:—You mentioned all microstructure

parameters except dislocation density. Have you measured it and what values do
you get?
Authors’ Response:—No. Dislocation density has not been measured.

N. Saibaba,1 K. Kapoor,1 S. V. Ramana Rao,1 Kalpana Itisri,1

S. K. Jha,1 C. Phani Babu,1 G. N. Ganesha,1 B. Prahlad,1
and R. K. Mistry1
Study on Effect of Processing

on Texture Development in
Zirconium-2.5 % Niobium
Alloy Tubes
Reference
Saibaba, N., Kapoor, K., Rao, S. V. Ramana, Itisri, Kalpana, Jha, S. K., Babu, C. Phani, Ganesha,
G. N., Prahlad, B., and Mistry, R. K., “Study on Effect of Processing on Texture Development in
Zirconium-2.5 % Niobium Alloy Tubes,” Zirconium in the Nuclear Industry: 17th International
ABSTRACT
In the present study, the effect of several processing steps like ingot working, hot
extrusion, cold working, and thermo-mechanical treatment of Zr-2.5 wt. %Nb
tube material on the texture evolution has been studied in the light of the
microstructure developed during the different stages. The role of microstructure
during the deformation on the texture development has been investigated in
detail. By comparing the different processing methods, an understanding on the
role of major processing steps has been developed. An explanation of the
observed results of the texture and microstructure has been made by
comparison to existing models.
Keywords
Zr-2.5Nb, crystallographic texture, thermo-mechanical-treatment, pilgering
Manuscript received March 21, 2013; accepted for publication September 4, 2014; published online October
14, 2014.
1
Nuclear Fuel Complex, ECIL (PO), Hyderabad, 500062 India.
2
Hyderabad, India.
Copyright V
Introduction
The study of texture development in zirconium alloy structural components is
important due to its significant influence on a host of properties like mechanical,
corrosion, hydride orientation, diametric creep, stress free irradiation growth, ther-
mal expansion, etc. During processing of tubes, texture development is influenced
by variables like temperature of deformation, reduction ratio, nature of working
(extrusion or forging at elevated temperature and pilgering/drawing at ambient
temperature), ratio of wall thickness to o.d. reduction (Q-ratio), and heat-
treatment. The role of these variables on texture development in the processing of
Zr-2.5 wt. %Nb (Zr-2.5Nb) material is studied in detail. The most common method
of characterizing texture is by obtaining pole-figures and orientation distribution
function (ODF) by X-ray diffraction (XRD) or more recently by electron backscat-
ter diffraction (EBSD). The pole figure gives a qualitative description of the grain
orientations, while a more quantitative description of texture is possible through an
evaluation of ODF. For the texture measurement, XRD data were acquired for
(0 0 0 2), (1 1 2 0), (1 1 2 2), (1 1 2 4), and (1 0 1 3) planes using Shultz technique
to generate the ODF. In addition, EBSD data were acquired for characterization of
microtexture during the various stages of fabrication. In the present study, samples
were collected at various stages during fabrication of Zr-2.5Nb pressure tube
processed with different routes. Results of texture development during processing
of Zr-2.5Nb pressure tubes were studied by comparing the texture formation during
the major processing stages, i.e., ingot processing, hot extrusion, cold working
(comparison of 3 different routes), and heat-treatment.
It is observed that initial processing (extrusion/forging) of ingots does not influ-
ence texture at the final stage provided b-quenching is carried out before extrusion
to tube hollows. During hot extrusion to mother hollows, one dominant orientation
is observed which sharpens after cold deformation and weakens with annealing.
The presence of the secondary b-phase surrounding the primary a-phase plays a
crucial role in texture evolution. The texture evolution in the cold worked stress
relieved route is different from heat-treated route due to the effect of microstructure
and processing. The evolution of texture has been studied by applying existing
texture models which incorporate the microstructural effect on the texture
development.
Experimental
OPTICAL MICROSCOPY
The optical microscopy was carried out using Olympus GX-71 (Olympus Optical
Company Ltd, Tokyo, Japan) metallurgical microscope with digital camera attach-
ment Olympus DP-12 (Olympus Optical Company Ltd, Tokyo, Japan). The modes
of observations were bright field and polarized light with magnifications upto
1000X.
TRANSMISSION ELECTRON MICROSCOPY

Thin foils for transmission electron microscopy (TEM) were prepared by electro-
polishing using double-jet thinning in a solution of 6 % sulfuric acid in methanol
maintained at 223 K. Thin foils were examined in a PEM430T (FEI Electron Optics
Eindhoven, Netherlands) TEM operating at 300 kV and equipped with an EDAX
(EDAX Inc., Mahwah, NJ, USA) Energy Dispersive X-ray Analyser SiLi detector
(Be-window) capable of qualitative and quantitative compositional measurements
from Na onwards.
SCANNING ELECTRON MICROSCOPY

Scanning electron microscopy (SEM) was done using Philips XL-30 (FEI Electron
Optics Eindhoven, Netherlands) scanning electron microscope to examine second-
ary and backscatter electron images at 30 kV operating voltage. The microscope was
equipped with an Energy Dispersive X-ray Analyzer EDAX Genesis (EDAX Inc.,
Mahwah, NJ, USA) with an ultra-thin window capable of qualitative and quantita-
tive compositional measurements from B onwards and an EBSD attachment.
X-RAY DIFFRACTION
X-ray diffraction measurements were performed using a Rigaku D-max 2200
(Rigaku International Corporation, Tokyo, Japan) X-ray diffractometer fitted with a
vertical goniometer. Cu Ka (k ¼ 0.154 nm) radiation from an X-ray tube operating
at 30 kV and 30 mA was used as the source. A curved crystal monochromator was
used at the receiving end. For texture measurement, Rigaku D-max 2200 (Rigaku
International Corporation, Tokyo, Japan) X-ray diffractometer fitted with a hori-
zontal goniometer was used. Cu Ka (k ¼ 0.154 nm) radiation from an X-ray tube
operating at 30 kV, 30 mA with a Ni filter at the source was used. The equipment
was fully automated using a microprocessor-controlled goniometer along with the
driver and processing software to handle necessary data processing. Schultz reflec-
tion technique [1] was used to measure pole figures from five high intensity peaks
of the samples. The incomplete pole figure data were generated by using tilt steps of
10 (0–70 range) and rotation stepsof 5 (0–360 for each tilt).
Determination of Quantitative Texture Parameter (Kearns f) Using

Diffraction Data
This method uses the diffraction data from the three samples taken from the
tube in the radial (RD), transverse (TD), and axial directions (AD). Scan speed of
1 /min, scan step of 0.01 /step and Ka2 stripping intensity ratio of 0.5 were used.
The raw data were smoothed (five points) using the Savitzky Golay’s method [2]
and corrected for background using Sonneveld–Visser’s method [3]. The eighteen
unique poles accessible using Cu radiations were used to compute the Kearns f
parameter as per standard procedure [4,5]. The samples were prepared with cut
pieces stacked together. The specimens were mechanically polished and etched with
45 % H2O þ 45 % HNO3 þ 10 % HF solution. This parameter gives the effective

fraction of basal poles aligned along the designated sample direction [4,5].
Determination of Crystallographic Orientation Distribution Function

Crystallographic orientation distribution function (CODF) was generated from
pole figures using the Labotex software (LaboSoft s.c., Krakow, Poland). The
software uses arbitrary defined cell method for determination of CODF from
pole figures [6]. Defocusing correction for texture data have been performed
with the standard procedure during CODF generation using the Zr-powder
data. The corrected pole figures (stereographic projection) were obtained from
the incomplete experimental pole figures through CODF. CODF is the quantita-
tive representation of the differently oriented crystallites in polycrystalline mate-
rials in three-dimensional space, i.e., the description of the frequency of
occurrence of particular orientation in three-dimensional (Euler) orientation
space. An orientation is a point in the three-dimensional orientation space,
expressed by three parameters, e.g., the three Eulerian angles u1, /, and u2.
The Eulers angles are related to (h k i l) hu v t wi in the Bunge system [7]. Due
to sample symmetry, the individual orientations measured on rotation about a
specific sample direction occur with equal statistical frequency. Its knowledge is
very important for good texture analysis. In the present analysis, sample sym-
metry was determined by first choosing lowest symmetry (triclinic) and the
CODF was determined. A number of symmetric orientations were found within
the basic region; hence sample exhibits higher symmetry. With further iterative
process, it was concluded that the sample has orthorhombic symmetry (the
extent of dissymmetry was less than 5 ). For hexagonal crystal symmetry and
orthorombic sample symmetry, the dimension of the unit cube is defined as
{0 u1 90 , 0 / 90 , 0 u2 60 }. A texture component can then be
described by f(g) with {u1, /, u2} being the orientation. f(g) is normalized in
such a way that f(g)dg represents the volume fraction of crystallite having an
orientation in the range dg defined as {Du1 ¼ 10 , D/ ¼ 15 , Du2 ¼ 10 }. The
contour line representation of f(g) is by sections of the cube perpendicular to
u2 for the case of hexagonal close packed (HCP) structure.
Calculation of CODF from Pole Figure. The pole figure is an average of the orienta-
tion with rotation about a given axis. An orientation in three-dimensional space is
represented by Euler angles as u1hkl, /hkl, u2hkl. The pole density in a pole figure
may be expressed in terms of pole density function as,
ð
1 hkl
(1) Phkl ða; bÞ ¼ f ðuhkl hkl hkl
1 ; / ; u2 Þdu2
2p
where:
Phkl(a,b) ¼ the pole density function, and
f(u1hkl, /hkl, u2hkl) is the orientation distribution function.
When several pole figures (five in the present case) belonging to different lattice
planes (hkil) are considered, the Euler angles are transformed from specific orienta-
tion u1hkl, /hkl, u2hkl to general angles u1, /, u2. The calculation of the three
dimensional function f(g) from several two dimensional projections Phkl(a,b) is
called the pole figure inversion. The pole figure inversion can be done using any of
the mathematical methods [8]. In the present study, the discretization method of
CODF calculation has been used. In the discretization or Arbitrary Defined Cell
(ADC) method [6,9,10], both the pole figures and orientation space are discretized.
The pole figure density functions corresponding to discrete elements of the pole
figure are connected to projection tubes with cells, which are discrete elements of
the orientation space.
Calculation of Volume Fraction from CODF. The volume fraction of components g

were calculated by integrating over the appropriate sub-volume in the orientation
space [11]. For volume fraction calculation, the angular sector “dg” was chosen with
u1 ¼ 10 , / ¼ 15 , u2 ¼ 10 .
The texture characterization carried out using different techniques and the pa-
rameters used for quantification from these techniques are summarized in Fig. 1.
Results
In this study, three ingot processing methods were used to manufacture the tubes
(Table 1). The raw material for all three processing methods was from quadruple
vacuum arc melted ingots. In stage 1 of the processing, ingots were processed to
FIG. 1 Macro and micro texture measurement and related quantification, (CPF:
Calculated Pole Figure, IPF: Inverse Pole Figure).
TABLE 1 Four processing routes with different billet-process schedules for manufacture of
Zr-2.5Nb pressure tubes (ER: Extrusion ratio).
ROUTE 1 (a and b) ROUTE 2 ROUTE 3
Ingot Processing: Vacuum Arc melting (350 mm Vacuum Arc melting Vacuum Arc melting
Stage 1 diametermeter) (800 mm diameter) (650 mm diameter)
; ; ;
Hot Extrusion (800 C) to Billet 220 mm b Forge I to 350 mm Double Forged (b
diameter diameter forge followed by
; ; a þ b forge) to
b quenching (1000 C) b quenching 220 mm diameter
(1000 C)
; ; ;
a þ b Forge II to b quenching
220 mm diameter (1000 C)
Hot Processing: ; ;
Stage 2 Hot Extrusion (800 C) to Tube Hollow Hot Extrusion Hot Extrusion
(ER 1:8) (ER 1:8) (800 C) to Tube (800 C) to Tube
Hollow (ER 1:8) Hollow (ER 1:12)
; ; ;
Cold Processing Cold Pilger I (50 %–60 % CW) Cold Pilger I Cold Pilger I
and HT: Stage 3 (50–60 % CW) (20–25 %CW)
; ; ;
Annealing a þ b Quenching Annealing
(in a-phase) 883 C/0.5 h (in a-phase)
; ; ;
Cold Pilger II Cold Pilger I Cold Pilger II
(20 %–25 % CW) (20–25 % CW) (20 %–25 % CW)
; ; ;
Ageing
(540 C/24 h)
;
Autoclave Autoclave Autoclave Autoclave
(400 C/24 h) (290 C/120 h) (400 C/24 h) (400 C/24 h)
billets with three different methods. In route 1(a and b), the ingots were hot
extruded in the two-phase a þ b phase region to billet size and then b quenched at
1000 C. In route 2, the ingots were hot forged to an intermediate size and then b
quenched. These were finally reduced to billet size by a second hot forging. The first
forging was carried out in the single-phase b region, whereas the second forging
was done in the two-phase a þ b region. After the second forging the billets were
air-cooled and machined. Route 3 was identical to route 2, except that the b
quenching step was carried out after the second hot forging. All three routes have
been used elsewhere for the manufacture of the billets for Zr-2.5Nb pressure tubes
[12,13]. Billets from these three methods were hot extruded and processed by using
two different extrusion ratios (ER) in stage 2 as described in the flow sheet (Table 1).
In stage 3, the extruded tube hollows were processed to tubes using two-stage pil-
gering route with an intermediate anneal (process routes 1a and 2), single stage pil-
gering (route 3) and by following a thermo-mechanical processing route with a þ b
quenching and ageing (route 1b). Details are provided in Table 1.
Samples were collected at different stages of the processing for analysis. Optical
microscopy and X-ray diffraction were performed on the as-cast ingot. X-ray dif-
fraction, optical, and TEM were performed on the billet at the end of stage 1. X-ray
diffraction and microstructure evaluations were performed on extrusion samples
taken at the end of stage 2. Samples were analyzed for crystallographic texture and
by TEM at the end of stage 3.
EFFECT OF INGOT PROCESSING ON TEXTURE AND MICROSTRUCTURE

DEVELOPMENT
X-Ray Diffraction
Figure 2(a) shows the X-ray diffraction pattern of the cast ingot after vacuum arc
melting, i.e., stage 1. The sample was oscillated to 65 mm in order to cover a larger
area as the prior beta grains are coarse in the as-cast material. This pattern corre-
sponds to a-Zr (space group: P63/mmc). However, the observed peak intensities do
not match the JCPDS standard data file (card number: 5-665) for a Zr sample with
random texture (Fig. 2(b)), thus confirming presence of texture in the cast ingot.
XRD diffraction scans of the b-quenched billets from routes 1 and 3 revealed
relative peak intensities that were similar to those in Fig. 2(b) for a random Zr
sample. This was in contrast to the billet from route 2 where the final forging was
performed in the a þ b phase. Some preferred orientation was present as relative
peak intensities from the route 2 billet did not match those of the random sample
(Fig. 2(b)). To confirm this, the sample was carefully powdered to fine size by mill-
ing. The X-ray diffraction pattern for this powder showed that the relative inten-
sities of the various peaks were similar to the standard (un-textured) data (Fig. 2(b)),
with some peak broadening due to cold work. This confirmed that beta-quenching
randomizes the texture. Finally, XRD scans after extrusion to the tube hollows
(stage 2) showed the presence of both a-Zr and b-Zr for all three processing routes,
where the extrusion temperature of 800 C was in the a þ b phase range (XRD scans
not shown in figure).
Optical and TEM

Optical micrographs of the sample after vacuum arc melting (Fig. 3) show a-phase
with “parallel plate” morphology. The starting microstructure of this as-cast ingot
was similar in the three processing routes. Optical and electron micrographs of
samples taken at the billet stage (i.e., at the end of stage 1) from routes 1, 2, and 3
FIG. 2 XRD patterns for three processing routes (see Table 1) at their different stages:
(a) at starting ingot, stage 1 (b) the standard JCPDS file, its comparisons with (a)
shows no matching of relative intensities, due to texture.
are shown in Figs. 4, 5, and 6, respectively. Fine acicular martensitic structures with
a “basket weave” pattern are seen in the optical micrographs in Figs. 4(a) and 6(a)
corresponding to the billet stage of process routes 1 and 3, respectively. The corre-
sponding TEM structures show plates of martensite (Figs. 4(b) and 6(b)). A twinned
primary martensitic plate (marked as “A”) surrounded by secondary martensite
(marked as “B”) plate is seen in Fig. 4(b). This martensitic structure with primary
and secondary martensite plates was reported by Banerjee and Krishnan [14]. As
shown by the optical micrograph in Fig. 5(a), the sample from processing route 2,
taken at the end of stage 1, reveals a two-phase structure consisting of (light) a
phase, surrounded by (dark) b phase. In the corresponding TEM structure, clear a
and b phase separation (Fig. 5(b)) is observed. The b phase is present as a thin film
surrounding the a phase. A fine dislocation network inside a grains marked as “/”
is also observed. This is a typical slowly cooled structure.
FIG. 3 Optical microstructure of as-cast Zr-2.5Nb ingot, showing parallel plate

martensitic morphology.
EFFECT OF HOT WORKING OF BILLETS ON TEXTURE AND MICROSTRUCTURE

DEVELOPMENT IN TUBE HOLLOW
By comparing the three routes given in Table 1, the effect of hot extrusion of the
billets was studied in terms of (a) the effect of microstructure at the billet stage (i.e.,
before extrusion) and (b) effect of extrusion ratio on the resulting microstructure
and texture. The SEM backscatter electron micrographs after extrusion at stage 2
are shown in Fig. 7. The dark regions represent the a phase while thin, elongated,
light stringers represent the b phase.
FIG. 4 Billet microstructure for processing route 1 taken at the end of stage 1 (see
Table 1): (a) optical micrograph showing martensitic structure and (b) TEM
structure showing a twinned primary martensitic plate, A, surrounded by
secondary martensite, B.
Table 1): (a) optical micrograph showing air-cooled two-phase structure and (b)
TEM structure showing light a -zirconium grains, containing dislocation network,
/, surrounded by a film of dark b -zirconium phase.
The structure resulting from processing route 1 (Fig. 7(a)) exhibits evenly
spaced distinct b stringers while those from processing route 2 (Fig. 7(b)) are seen to
be broken with varying thickness and spacing. The b phase stringers for the sample
from processing route 3 (Fig. 7(c)) are similar to those observed in process route 1.
Table 2 gives the quantitative analysis of the b-phase from the three routes.
In order to study the effect of the extrusion ratio, EBSD of samples from routes
1 and 3 was carried out. The resulting textures determined by EBSD for the two
routes are given in Fig. 8. The Kearns f parameter for the two routes (Table 3) show
that with higher extrusion ratio, the basal poles become more intense in the trans-
verse direction. The results from EBSD and ODF substantiate this observation.
Table 1): (a) Optical micrograph and (b) TEM similar to Fig. 4(b).
FIG. 7 Microstructure of as extruded material at stage 2 (see Table 1). Backscatter SEM
image showing a two-phase structure containing thin, white b phase in matrix of
a phase for: (a) Processing route 1, (b) processing route 2 and (c) processing
route 3.
EFFECT OF COLD PROCESSING ON TEXTURE AND MICROSTRUCTURE

DEVELOPMENT
Figures 9–11 are TEM micrographs of the samples obtained from processing routes
1a, 2, and 3 at stage 3, after autoclaving. Figure 9 shows the presence of elongated a
grains oriented in the working direction; and the insert on left shows selected area
diffraction (SAD) pattern with [1 1 2 0] as zone axis (ZA) plane of a-Zr. During the
processing of this material, the intermediate anneal and final stress relieving treat-
ments result in the dissociation of b-Zr (that exists in the as-extruded tube hollow
stage) to stable b-Nb and a-Zr phases [12,13]. The right insert in Fig. 9 shows SAD
of a b-Nb precipitate with [1 1 1] as ZA. The number of precipitates in this case are
TABLE 2 Quantitative shape analysis of the b-phase in as-extruded stage (Fig. 7).
Process Average length of the b stringers (lm) Aspect ratio (length/diameter) of b
Route 1 2.2 10.1

Route 2 0.9 8.2
Route 3 1.7 9.0
FIG. 8 EBSD of as extruded material at stage 2 showing the acicular alpha phase
for: (a) Processing route1- longitudinal section (LS) (b) Processing
route 1 - transverse section (TS), (c) Processing route 3 - LS and (d) Processing
route 3 - TS. The EBSD computed pole figures and ODF u1 section at 60 for
samples after extrusion for (e) route 1 and (f) route 3.
few. In the case of the sample taken from processing route 2 at stage 3, the TEM
micrograph reveals the presence of a large number of globular precipitates at the
grain boundary of a-Zr (Fig. 10). SAD of a-Zr with [1 1 2 0] as ZA is shown in the
left insert in Fig. 10. Right insert in Fig. 10 shows SAD of b-Nb precipitate at
the grain boundary with [l 0 0] as ZA. Figure 11 is a TEM micrograph of the sample
taken at stage 3 from processing route 3. This reveals elongated a-Zr grains with
some b-Nb precipitates at the a grain boundaries. The number of precipitates is
similar to that observed for processing route 1a. SAD from a-Zr and b-Nb are
shown as left and right inserts, respectively, in Fig. 11.
TABLE 3 Kearns f parameter and volume fraction of strong orientations (from ODF) for the sam-
ples taken at stage 2 (after hot extrusion) with varying extrusion ratio (ER).
Route 1: Hot Extruded (ER 1:8) Route 3: Hot Extruded (ER 1:12)
Kearns f
fr ft fa fr ft fa
0.38 0.56 0.06 0.27 0.69 0.04
Volume Fraction of Orientations (From ODF)
(11–22) h1–100i (11–24)h1–100i (33–61) h1–100i (11–22) h1–100i (11–24) h1–100i (33–61) h1–100i
18.8 % 2.8 % 3.7 % 10.1 % 4.2 % 4.7 %
Crystallographic Texture
The crystallographic texture was determined at the final stage of the three process-
ing routes. The ODF for the sample taken at stage 3 shows a very strong orientation
fiber of the type (hkil)[h1 0 1 0i (u1 ¼ 0 , / ¼ 0–90 , u2 ¼ 0 ) in all three processing
routes. This orientation fiber is parallel to / direction. Figures 12(a), 12(b), and 12(c)
show the ODF section taken at u2 ¼ 0 for the three processing routes 1a, 2, and 3,
respectively. These figures shows the prominent orientation fiber of the type
(hkil)h1 0 1 0i, with maximum strength at / ¼ 90 . The (0 0 0 2) pole figure
obtained from the ODF shows that the basal poles are very strongly oriented in the
FIG. 9 TEM micrograph at stage 3 for processing route 1a (see Table 1), showing typical
bamboo-tree structure with small amount of b-Nb precipitated from b-Zr phase
at a-Zr phase boundary (see arrow). Left insert: SAD from a-Zr [1 1
2 0] as ZA,
right insert: SAD from b-Nb precipitate [1 1 1] as ZA.
FIG. 10 TEM micrograph at stage 3 for processing route 2 (see Table 1), showing larger
amount of b-Nb precipitated at a-Zr phase boundary (see arrow). Left insert:
SAD from a-Zr [1 1
2 0] as ZA, right insert: SAD from b-Nb [1 0 0] as ZA.
FIG. 11 TEM micrograph at stage 3 for processing at route 3 (see Table 1) showing some
b-Nb precipitate at a-Zr phase boundary (see arrow), Left insert: SAD from a-Zr
[1 1 2 0] as ZA and right insert SAD from b-Nb [1 1 1] as ZA.
FIG. 12 (a)–(c) ODF u2 ¼ 0 (u2 sections) for the material at the final stage of processing
according to the three processing routes 1a, 2 and 3, respectively (see Table 1).
(d), (e) and (f) (0002) basal pole figures were obtained from the ODF for the
three processing routes 1a, 2 and 3 respectively, showing that the final texture
developed is very much identical in the three routes.
transverse direction. The ODFs and basal pole figures from the three processing
routes are very much identical. The quantitative data are obtained by measuring the
Kearns f parameter and volume fraction of orientations (Table 4). Kearns f parame-
ter measured after extrusion (Table 3) and cold working after extrusion (Table 4) is
observed to be similar, i.e., there is no change in Kearns f after cold deformation for
both routes 1a and 3. The basal pole fraction in the transverse direction (ft) is higher
for the route with high extrusion ratio and the same is retained even after cold
working. The effect of two-pass pilgering with higher reduction or single-pass
pilgering does not seem to influence the texture formed after extrusion.
TABLE 4 Kearns f parameter and volume fraction of strong orientations for route 1a and 3. (ER:
extrusion ratio).
Route 1a: Hot Extruded (ER 1:8) þ 2-Pass Pilgered Route 3: Hot Extruded (ER 1:12) þ 1-Pass Pilgered
Kearns f
fr ft fa fr ft fa
0.41 0.57 0.02 0.29 0.68 0.03
Volume Fraction of Orientations (from ODF)

2 h1 1 0 0i
112 h1100i
1124 336 1 h1 1 0 0i 2 h1 1 0 0i
112 11
2 4 h1 1 0 0i 1 h1 1 0 0i
336
27.5 % 8.1 % 10.1 % 19.9 % 7.6 % 15.8 %
FIG. 13 (0002) pile figures and Orientation Distribution Function of Zircaloy-4 samples
in u2 sections: (a), (d) extruded, section u2 ¼ 30 , (b), (e) Cold pilger I and
annealed, section u2 ¼ 0 and (c), (f) Cold pilger II and autoclaved, section
u2 ¼ 0 .
To further study the effect of microstructure on the texture formation, a

Zircaloy-4 sample was processed with identical processing schedule as route 1a.
Basal (0 0 0 2) pole figures and ODFs of single-phase Zircaloy-4 tube material are
presented in Fig. 13 for three different conditions. The ODF of the single phase
Zircaloy-4 material, at the extruded stage, shows component (0 2 2 3)h2 1 1 0i at
{u1 ¼ 0 , / ¼ 50 , u2 ¼ 30 } (Fig. 13(d)). After cold pilger I and annealing, a new
texture component (1 2 1 6)h1 0 1 0i at {u1 ¼ 0 , / ¼ 28 , u2 ¼ 0 } is formed, which
reveals a sharp change in the orientation. With the cold work and annealing, the
basal poles move towards RD (/ changing from 50 to 28 ).
Moreover, a rotation of the crystal occurs by 30 about the ‘c’ axis, which is
known to be the result of annealing after the cold work (u2 changing from 30 to
0 ) (Fig. 13(e)). After the cold pilger II, (1 2 1 7)h1 0 1 0i texture at {u1 ¼ 0 ,
/ ¼ 25 , u2 ¼ 0 } is seen (Fig. 13(f)) with further movement of basal poles towards
RD (/ changing from 28 to 25 ). Thus, from basal pole figures (Fig. 13), in the
extruded sample, the basal pole appears at 50 tilt from RD, which reduces to 28
in cold pilger I and annealed and further to 25 after cold pilger II and stress relief.
TEXTURE EVOLUTION IN THERMO-MECHANICALLY HEAT-TREATED ROUTE

Texture evolution in case of thermo-mechanically heat-treated route 1b was stud-
ied. It was shown earlier [15] that the crystallographic texture of fabricated tube is a
strong function of the amount of primary a phase present after water quenching
(WQ) treatment, which in turn is dependent on the soaking temperature in the
a þ b phase field [15]. Detailed characterization of the microstructure at all the
stages of thermo-mechanically heat-treated (TMT) route was carried out to explain

the evolved texture.
Microstructure Development During the TMT Route

In the present study, the amount of primary or retained a phase was determined
through SEM study of the water quenched tube samples. The volume fraction of
the primary a was found to be in the range of 20 %–25 % (Fig. 14). It could be seen
that all prior b phase has transformed to the lath martensite which is surrounding
the retained a phase. This microstructure consisting of fine a0 martensite phase
along with 20 %–25 % primary a was used for further processing as described in
the fabrication flow chart (route 1b in Table 1). The quenched tubes were cold
pilgered with a reduction of 23 %. The purpose of this cold deformation process
was to achieve the final dimension of the tube within a narrow range. Additionally,
during this fabrication stage, a complex network of dislocations was introduced in
both primary a and martensitic phases. One of the consequences of the thermome-
chanical treatments (quenching followed by cold deformation) was the creation of a
large number of nucleation sites at the interfaces of the fine martensitic microstruc-
ture, thereby resulting in fine precipitation of stable b-Nb phase during ageing.
Cold deformation of martensitic microstructure resulted in considerable increase in
the dislocation density and dislocation substructure. These resulted in dislocation
enhanced diffusion, possibly by pipe diffusion mechanism; thus leading to precipi-
tation of b-Nb from the supersaturated martensitic phase during subsequent ageing
process. Therefore, an ageing treatment at 540 C/24 h was given, which is below
the monotectoid temperature (610 C), to obtain fine b precipitates having composi-
tion close to equilibrium along with retained primary a. The microstructures are
presented in Fig. 14(b), which shows fully a recovered structure with no
FIG. 14 Microstructures of the TMT route 1b (in Table 1) at different stages a) As

quenched (from a þ b phase) showing retained a grains (20-25 % in volume
fraction) embedded in matrix of martensitic a0 phase. (b) showing
microstructure after 23 % CW followed by aging.
TABLE 5 Kearns f parameters for the TMT processing route 1b (Table 1).
Condition fr ft fa
After second extrusion 0.38 0.56 0.06

As (a þ b) water quenched (883 C/0.5 h) 0.30 0.33 0.37
As aged 540 C/24 h, (after WQ) 0.36 0.50 0.14
recrystallization of the microstructure after ageing at 540 C, resulting in complete

tempering of martensite. Analysis of the b phase was carried out by energy disper-
sive spectroscopy (EDS) showed that the average composition in the aged
sample was 75–80 %Nb (estimation without considering matrix effect). Finally an
autoclaving treatment at 290 C for 120 h did not modify the microstructure to any
noticeable extent.
Texture Evolution During TMT Route

Table 5 lists the Kearns f parameter at various stages. The tubes were soaked at
883 C, and quenched using argon gas and water. Water quenching was performed
in horizontal as well as vertical type furnace. In the former case, sample transfer
time was longer and effective cooling rate was lower than the vertical furnace. For
the present tests, vertical quenching was used which gave fast quenching rates.
The main observations from Table 5 with respect to texture evolution during
quenching can be summarized as the resultant texture upon water quenching is
near random. In an earlier study [16], it was concluded that the volume fraction of
the retained alpha phase controls the texture subsequent to quenching. These
results highlight the importance of the volume fraction of primary alpha phase at
the quenching temperature on the quenching texture. There was a marginal modifi-
cation of texture during the cold work plus ageing treatment. The tubes aged at
540 C had the second phase b present only within the transformed a0 phase (see
Fig. 14(b)). This was unlike the other routes (1a, 2, and 3) where the second phase
was present at the a-phase boundary. There was a direct grain-to-grain contact at
the a grains (without any b-phase present at the grain boundary) in the TMT route.
This suggested that with this microstructure, the deformation carried out at room
temperature was able to modify texture significantly unlike routes 1a, 2, and 3. The
(0002) pole figure and ODF u2 ¼ 0 section for the WQ and aged material is shown
in Fig. 15.
Discussion
SIGNIFICANCE OF INGOT PROCESSING PARAMETERS IN THE
SELECTED ROUTES
The structure at ingot stage was “plate type” morphology (Fig. 3). As the ingot mate-
rial used for the three routes was melted and cast using similar melting parameters,
FIG. 15 Final Texture for the TMT processing route 1b (Table 1) showing a) the (0002)
pole figure having distribution of basal poles in TD-RD direction and b) ODF
u2 ¼ 0 section showing major texture component along the u direction and a
minor component at u ¼ 0 and u2 ¼ 60 leading to some orientations with
basal plane in axial direction.
the microstructures at the starting ingot stage for the three routes were similar.
During further processing of the ingots, forging or extrusion was used to break the
cast structure. The selection of deformation temperature in a þ b or b phase region
is based on the extent of deformation. The deformation behavior in the b phase
region [17] and in a þ b phase region [18] has been studied using the processing
maps. High reduction during hot working is preferred in the b phase region than
that in the a þ b region. The factors affecting hot working in the two-phase (a þ b)
or single phase (b) region was reviewed by Bournell and McQueen [19]. One factor
for selection of the first hot forging in the b phase region was to completely break
the cast structure by a large amount of deformation.
During hot working in the b phase region, dynamic recrystallization with large
flow softening with strain was found to occur and could facilitate large deformation.
In contrast, at lower temperatures in the a þ b phase field, workability was noted to
be relatively less. In the first step of process routes 2 and 3, a high reduction ratio of
6 and 8, respectively, was used requiring forging to be done in the b phase region.
In the second forging step in process routes 2 and 3, or first extrusion in the case of
process route 1, the reduction is low. Hence working at lower temperature in the
(a þ b) region was carried out. At the billet stage for process route 2, the micro-
structure is two-phase, which is different from that observed in process routes 1
and 3. Thus, it becomes necessary to analyze if the difference in the initial working
stages of the ingots and the subsequent heat treatment would affect the properties
of the tubes.
The evolution of texture in the ingot is attributed to solidification occurring
along the length of the ingot. One of the purposes of heat treatment (b quenching)
during the ingot processing is to remove this cast structure and obtain texture free
starting material in stage 2 of the processing. Starting with an ingot of heavily
textured structure, a randomly oriented structure was obtained in the b quenched

billets. On quenching, b transforms to a0 (martensitic HCP phase). This phase
transformation follows a Burgers relationship (0 0 0 1)a//{0 1 1}b; h1 1 2 0ia//
h1 1 1ib with twelve crystallographic variants. During this transformation, fine a0
acicular phase is nucleated at the prior b-phase boundary, with each plate having
equal probability of spontaneous variant selections among the twelve possibilities.
This leads to fine a0 acicular phase having random orientation within each prior b-
grain. This texture can be called as pseudo-isotropic rather than texture free. How-
ever, the process of variant selection also depends on the cooling rates. For very fast
cooling rates, the variant selection is rather limited. In the present case, due to large
mass of the billet, the quenching rate was not very high as compared to a small
sample. Hence, the cooling rates during quenching are optimum for spontaneous
variant selection in the present case.
MORPHOLOGICAL EVOLUTION OF b-PHASE DURING THERMO-MECHANICAL

PROCESSING
The microstructural evolution in the three routes at stage 2 and stage 3 is given
in Figs. 9–11 and discussed in this section. On soaking before extrusion, an addi-
tional amount of b phase is nucleated in the Zr-2.5Nb material and the structure
consists of equal volume mixtures of a and b phases. In the case of billets with
martensitic structure (process routes 1 and 3), this b phase nucleates randomly at
the martensitic plate boundaries. In the case of a þ b structure, the b phase grows
at the preexisting b-phase. After extrusion and cooling to room temperature, the
structure consists of b-Zr phase as thin stringers along the extrusion direction at
the grain boundary of elongated a-Zr grains. During subsequent processing, the
various thermo-mechanical treatments result in stable b-Nb from metastable b-
Zr. One of the major differences in the final microstructures of the tubes from
the three processing routes is the quantity and distribution of this b-Nb phase
(Figs. 9–11). As seen in the TEM micrographs, the quantity of this phase observed
in route 2 is much lower than that observed in routes 1a and 3 and has smaller
aspect ratio. The likely explanation for this is as follows. In route 2, the b-Zr
phase already exists in the pre-extrusion stage while in processes 1a and 3, b-Zr
precipitates during preheating before extrusion. The structure at the extrusion
temperature has a bearing on further morphology of b-phase. This effect could
be due to the difference in diffusivity of niobium in the two cases; the diffusivity
of niobium is at least five orders lower in a phase as compared to that in the a–b
interface [20,21]. Hence, the pre-existing a þ b structure at the start of stage 2 for
processing route 2 will enhance decomposition of b-Zr into stable b-Nb and a-Zr
phases. In the diffusion (Fick’s) equation, the diffusion coefficient (measured with
tracers) [20,21] must be multiplied by the gradient of the (local) chemical poten-
tial which is not likely to be the same in microstructures from route 1a and 3.
However, this contribution is much lower that the difference of diffusivity in the
two phases. The morphology and distribution of b-phase can be explained by

using a model given below.
Model for Morphological Evolution of b-Phase

The morphology of the b phase is observed to change from the extrusion to final
stage. This is due to unstable elongated grain shape and metastable nature of this
phase. In a study by Jovanovic et al. [22], they determined the effect of annealing
on the continuity of b phase. After 1 h at 400 C, the b phase was thin, relatively
long, but totally continuous. After 24 h, the b phase was somewhat fragmented and
composed of distinct particles. This breakup was complete when the sample was
annealed for 1000 h. Thus, this rod shaped microstructure is noted to be unstable
and changes as discrete particles as a result of the decrease in interfacial energy
through lower surface-to-volume ratio. As pointed out by Cline [23], such processes
would be activated by decomposition of the phase, which is similar to Oswald rip-
ening. In the theoretical treatment (Fig. 16) of Cline [23] and Ho and Weatherly
[24], a long fiber (length l) of radius d is eventually replaced by a string of spheres
(referred to as Rayleigh instability) where the sphere radius and the spacing, k,
depend on the active kinetic processes. In the case of infinite fibers, the drop
detachment at the end of the fibers was a faster process (Fig. 16(a)). For finite fibers
(with l/d > 7.2), the rod shape particle would breakup into several spherical
FIG. 16 Model by Cline [23], Ho and Weatherly [24] to explain the evolution of
cylindrical particles undergoing modification in shape as a function of time due
to Oswald ripening.
FIG. 17 Model proposed to explain the origin of the morphology of the b phase in the
final stage for the three processing routes adopted here (Table 1).
particles (Fig. 16(b)), while for l/d less than 7.2, the shape relaxation to a single
sphere is predicted (Fig. 16(c)). In the present case, the aspect ratio calculated for the
b phase in all the three cases (Table 2) is >7.2. With this model, each of the b-phase
stringer is likely to transform into a series of spherical particles, which is evident
from Fig. 11(a) (see marked arrow).
The microstructure is not at equilibrium and based on the microstructural
examinations at various stages for all three routes, a model is proposed for evolu-
tion of the morphology of the b-phase in Fig. 17. The starting microstructure for
stage 2 could be either martensitic (route 1a or 3) - case 1, or air-cooled (route
2)–case 2. The martensitic microstructure with Nb in supersaturated solid solution
is likely to have higher free energy than the route 2 microstructure. Thus, during
soaking before extrusion for the case of a martensitic structure, the nucleation of
the b phase is expected to be random at the martensitic plate boundaries. For
the case of the preexisting a þ b structure, the b phase would grow at the existing
a-phase boundary. On cooling after extrusion, part of the b phase would dissociate
and thus the b stringers could be somewhat broken. This is in contrast to case 1,
where the b phase will be continuous. More decomposition in both cases will occur
in the b phase during further thermo-mechanical processing. Thus, decomposition
of b phase to b-Nb leading to nearly full spheroidsation and fragmentation in case
2, and partly in case 1, are suggested to occur.
TEXTURE EVOLUTION IN SINGLE-PHASE AND TWO-PHASE

MATERIALS—JUSTIFICATION USING EXISTING TEXTURE MODELS
The above results show that deformation at ambient temperature produces a grad-
ual change in texture for single-phase Zircaloy-4 material. In contrast, there is no
change in texture following deformation at ambient temperature for the two-phase
Zr-2.5Nb material. The following two models have been applied to Zr, Ti, and other
HCP systems to predict the texture.
1. Taylor model [25] referred to as full constraint (FC) or relaxed constraint
(RC) [26–30] version.
2. Visco-plastic-self-consistent (vpsc) [31–35].
The Taylor model assumes that each grain deforms like a poly crystal, i.e., the
microscopic plastic deformation is equal to the macroscopic in each grain. Further-
more, the plastic strain is homogenous “within” the limits of a grain. This means that
the surrounding imposes full constraint to the individual grain, which deforms by
slip and/or twinning. This situation is true only in the case of a single-phase material.
Thus, the FC or the modified version RC, have given satisfactory results in case of
single-phase material [28,29]. Bechade et al. [30] have used FC and RC to predict the
texture in case of rolling process for Zircaloy-4 sheet. In case of the two-phase mate-
rial, the surrounding will have the influence on the constraint developed within the
grain. The VPSC model on the other hand gives great deal of importance to the grain
and surrounding interaction. In fact, it takes into account the extent of the plastic
strain accumulated in the grain and amount transferred to the surrounding. The rel-
ative amounts depend on the plastic compliance of the grain and the surrounding. In
the two-phase material, with the soft grain boundary phase, this model would be
more appropriate. A comparison of the application of both the models to single and
two-phase material has been already done by Lebensohn et al. [34]. They show the
results of the comparison of the two models for Zircaloy-2 and Zr-2.5Nb. The experi-
mental and predicted intensity of the basal and prism poles with respect to the tilt
angle after asymmetric tensile deformation are compared. In their work, they rightly
claim that the VPSC model exactly reproduces all features of the experimental curve
for both single-phase (Zircaloy-2) and two-phase (Zr-2.5Nb) materials. However,
they fail to explain why the FC model is not appropriate for two-phase material.
Our experimental observations and its justification for the evolution of the
texture in case of the single-phase and two-phase material fits correctly in the two
models. In case of the strong surrounding influence (two-phase material), the
Taylor model is not applicable while, the VPSC will still hold good. We believe that
this is because of the presence of bcc b phase at the grain boundary, which on
deformation can modify the intergranular constraints developed among the a
grains. This is exactly what is presented in the work by Lebensohn et al. [34]. For
the two models, they have predicted active deformation systems. In the case of
VPSC model, fewer deformation systems are required (only 3) and soft deformation
modes like prismatic slip are favored while for the FC model, more (five) deforma-
tion systems are required, which have with high critically resolved shear stress
(CRSS). In the case of two-phase material with soft phase at the grain-boundary,
where the VPSC model correctly fits, the relative amount of deformation at the
grain boundary phase is likely to be higher. This is due to activation of deformation
system in the grain boundary b-phase, which has lower CRSS than the a-phase
grain.
CORRELATION OF TEXTURE AND MICROSTRUCTURE DEVELOPED

DURING PROCESSING
In an earlier study carried out by Saibaba et al. [36], a comparison of process routes
was done for texture and microstructure. The present work provides an in depth
analysis for the new results detailed above.
Hot Working
During annealing and thermo-mechanical treatments, some changes in grain shape
and orientation in material take place, which result in the development of texture.
The Zr-2.5Nb material at the extrusion temperature (800 C) contains equal volume
fractions of a and b phases. The texture developed during hot deformation may be
a combination of the deformation mechanisms operating in the constituent phase
of this two-phase mixture. Holt and Aldridge [37] reported a detail analysis of
extrusion variables on texture formation. It was shown that the increases in the
extrusion ratio and extrusion temperature increase the proportion of basal poles in
the transverse direction. In case of two-phase alloys, the deformation at elevated
temperature can be explained by combination of the following mechanisms:
1. Deformation in a-Zr phase by prismatic slip h1 1 2 0i (1 0 1 0) and twinning
[38].
2. Reorientation of Widmanstatten a plates by phase boundary sliding to lie nor-
mal to major compressive strain [39].
3. Stress induced b!a phase transformation, where the orientation change
occurs along the direction of compressive strain [40]. This is not likely to con-
tribute in the present case.
The first mechanism results in basal pole orientation in radial direction, while
the second mechanism result in transverse orientation. The reason for this trans-
verse texture in Zr-2.5Nb material is as follows, formation of sub-grains with low
angle boundaries by dynamic recovery was observed in the extruded Zr-2.5Nb ma-
terial. This structure allows sliding at sub-grain and second phase boundaries,
which favors transverse texture. In case of the single phase Zircaloy-4 material,
dynamic recrystallization is more likely to occur during deformation [17]. In the
case of the extruded Zircaloy-4 material, fully recrystallised grains with traces of
twins were observed. This suggests that, in single phase alloy at elevated tempera-
ture deformation by glide and twinning (first mechanism) is predominant which
favors radial texture. The present observations about the texture of single-phase
Zircaloy-4 material are in line with that of Tenckhoff and Rittenhouse [41],
where it was observed that the basal plane normal after extrusion lie midway
between the circumferential and radial directions (which is similar to our observa-
tions in Fig. 13 (a)).
Cold Working
On cold working, the a grains in two-phase Zr-2.5Nb alloy become more elongated.
With intermediate annealing (550 C) and final stress relieving (400 C), b phase
partly dissociates into a-Zr and b phase rich in niobium. The deformation mecha-
nisms operating at room temperature in the primary a-Zr include prismatic slip
and {1 0 1 2}h1 0 1 1i twinning [42]. The texture developed during cold working is
strongly dependent on the concurrent microstructure during deformation. The tex-
ture development in Zr-2.5Nb two-phase material is quite different from that of
single-phase Zircaloy-4 material. In single-phase material, there occurs grain-to-
grain interaction and each grain undergoes changes in shape and rotation. Since all
grains are in contact with each other, they impose mutual constraint in their move-
ments due to deformation. Rotation and orientation of grains in order to accommo-
date the deformation may result in new components of texture instead of
intensifying the texture of the former grains as a result of constricted movement of
adjoining grains in single-phase material.
In the case of two-phase material, the grain of the primary phase (a) is mostly
in contact with the softer secondary phase (b). On deformation, the rotation and
change in shape of the grains of the harder phase occur with minimum obstruction
from the softer phase in its contact. Hence, no new texture component is evolved,
only the existing ones are intensified.
Conclusions
The present study of texture development in Zr-2.5Nb material processed from dif-
ferent routes led to a number of useful results as concluded below.
1. Heat treatment (b-quenching) of the billets of Zr-2.5Nb during the processing
of the ingots is a critical step. The stage at which it is carried out influences
the final microstructure and decomposition of b-phase. This heat-treatment
randomizes the texture with the result that the effect of prior processing
(extrusion or forging) is not significant.
2. The texture developed during hot extrusion remains stable during the subse-
quent cold working of the two-phase Zr-2.5Nb alloy whereas it changes con-
tinuously in case of single-phase Zircaloy-4. The difference in the texture
evolution of the two materials is attributed to the presence of the second phase
b on the grain boundaries in the former alloy. The softer b phase deforms and
reduces the constraints that otherwise would be faced by two adjoining grains
of harder a-phase.
3. In case of Zr-2.5Nb material, the effect of a higher extrusion ratio is to
increase the transverse basal pole texture which is retained after subsequent
cold working. A strong fiber texture parallel to the axial direction of the
type (h k i l)h1 1 0 0i is formed during extrusion which is retained during
further processing.
4. The texture development in the thermo-mechanically treated Zr-2.5Nb tubes
(process 1b) depends on the retained a-phase and transformed a þ b phase
present in the material. The second phase present in this case is within the a-
phase and thus does not contribute in the texture formation as there is a direct
grain to grain contact of the a-phase.
References
[1] Schulz, L. G., “Determination of Preferred Orientation in Flat Transmission

Samples Using a Geiger Counter X-Ray Spectrometer,” J. Appl. Phys., Vol. 20, 1949,
pp. 1033–1036.
[2] Savitzky, A. and Golay, M. J. E., “Smoothing and Differentiation of Data by Simplified
Least Squares Procedures,” Anal. Chem., Vol. 36, No. 8, 1964, pp. 1627–1639.
[3] Sonneveld, E. J. and Visser, J. W., “Automatic Collection of Powder Data from
Photographs,” J. Appl. Cryst., Vol. 8, 1975, pp. 1–7.
[4] Kearns, J. J., “Thermal Expansion and Preferred Orientation in Zircaloy,” Report
WAPD-TM- 472, Westinghouse Co., Pittsburgh, PA, 1965.
[5] Kearns, J. J., “Reflections on the Development of the “f” Texture Factors for Zirconium
Components and the Establishment of Properties of the Zirocnium–Hydrogen System
Including Solubility, Diffusivity and Embrittlement,” ASTM STP 1543, ASTM International,
West Conshohocken, PA, 2013.
[6] Pawlik, K., Pospiech, J. and Luke, K., “Determination of the Orientation Distribution
Function from Pole Figures in Arbitrarily Defined Cells,” Proceedings of the ICOTOM-8,
Santa Fe, NM, Sept.–25, 1987.
[7] Bunge, H. J., Texture Analysis in Materials Science, Butterworths Press, London, 1982.
[8] Bunge, H. J., “Preferred Orientation in Deformed Metals and Rocks: An Introduction to
Modern Texture Analysis,” Int. Mater. Rev., Vol. 32, 1987, pp. 265–291.
[9] Pawlik, K., Pospiech, J. and Luke, K., “The ODF Approximation from Pole Figures With
the Aid of the ADC-Method,” Proceedings of the ICOTOM-9, Avignon, France, Sept.
17–21, 1990.
[10] Wenk, H. R., Pawlik, K., Pospiech, J. and Kallend, J. S., “Deconvolution of Superposed
Pole Figures by Discrete ODF methods: Comparison of ADC and WIMV for Quartz and
Calcite with Trigonal Crystal and Triclinic Specimen Symmetry,” Text. Microstruct., Vol.
22, 1994, pp. 233–260.
[11] Flowers, J. W., “Volume Fractions of Texture Components of Cubic Materials,” Text.
Microstruct., Vol. 5, 1983, pp. 205–218.
[12] Fleck, R. G., Perovic, V. and Ho, E. T. C., “Development of Modified Pressures Tubes,” Ont.
Hydr. Res. Rev., Vol. 8, 1993.
[13] Causey, A. R., Cristodoulou, N., Davis, P. A., Griffiths, M., McDougall, G. M., Moan, G. D.,
Ploc, R. A. and Puls, M. P., “Deformation of Zr-2.5Nb Pressure Tubes,” Rare Metall. Mater.
Eng., Vol. 30, 2001.
[14] Banerjee, S. and Krishnan, R., “Martensitic Transformation in Zirconium-Niobium Alloys,”

[15] Coleman, C. E., Griffiths, M., Grigoriev, V., Ksieliov, V. and Rodchenkov, B., “Mechanical
Properties of Zr-2.5Nb Pressure Tubes Made from Electrolytic Powder,” Zirconium in Nu-
clear Industry, STP 1505, ASTM International, West Conshohocken, PA, 2009, pp. 699–722.
[16] Saibaba, N., Jha, S. K., Tonpe, S., Vaibhaw, K., Deshmukh, V., Ramana Rao, S. V., Mani
Krishna, K. V., Neogy, S., Srivastava, D., Dey, G. K., Kulkarni, R. V., Rath, B. B., Ramadasan,
E., Anantharaman, S. A., “Microstructural Studies of Heat Treated Zr-2.5Nb Alloy for
Pressure Tube Applications,” J. ASTM Int., Vol. 8, 2011, 103213.
[17] Chakravartty, J. K., Banerjee, S. and Prasad, Y. V. R. K., “Superplasticity in b Zirconium: A

Study Using Processing Map,” Scr. Metall. Mater., Vol. 26, 1992, pp. 75–.78
[18] Chakravartty, J. K., 1993, “Optimisation of Hot Workability and Control of

Microstructure in Zr and Zr-alloys Using Processing Maps,” Ph.D. thesis IISc,
Bangalore, India.
[19] Bournell, D. L. and McQueen, H. J., “Thermomechanical Processing of Titanium, Zirco-

nium, Magnesium, and Zinc in the hcp Structure,” J. Mater. Shaping Technol., Vol. 5,
1987, pp. 163–189.
[20] Horvath, J., Dyment, F. and Mehrer, H., “Anomalous Self-Diffusion in a Single Crystal of
a-Zirconium,” J. Nucl. Mater., Vol. 126, 1984, pp. 206–214.
[21] Piotrkowski, R., “Comments on a-b Phase Boundary Self-Diffusion in Zr-2.5 % Nb

[22] Jovanovic, M. T., Eadie, R. L., Ma, Y., Anderson, M., Sagat, S. and Perovic, V., “The Effect
of Annealing on Zr-2.5Nb,” Mater. Charact., Vol. 47, 2001, pp. 259–268.
[23] Cline, H., “Shape Instabilities of Eutectic Composites at Elevated Temperatures,” Acta
Metall., Vol. 19, 1971, pp. 481–490.
[24] Ho, F. and Weatherly, C. G., “Interface Diffusion in the Al-CuAl2 Eutectic,” Acta Metall.,
Vol. 23, 1975, pp. 1451–1460.
[25] Taylor, I. G., “Plastic Strain in Metals,” J. Inst. Metals, Vol. 62, 1938, pp. 307–325.
[26] Honnef, H. and Mecking, H., “General Problems in Reproducing the Orientation Distribu-
tion Function from Pole Figures,” Proceedings of the ITCOM 6, Orlando, FL, Sept. 7–11,
2003.
[27] Phillippe, M. J., “Texture Formation in Hexagonal Materials,” Mater. Sci. Forum, Vol.
157–162, 1994, pp. 1337–1350.
[28] Fundenberger, J. J., Phillippe, M. J., Wagner, F. and Esling, C., “Modelling and Prediction
of Mechanical Properties for Materials with Hexagonal Symmetry (Zinc, Titanium and
Zirconium Alloys),” Acta Mater., Vol. 45, 1997, pp. 4041–4055.
[29] Phillippe, M. J., Bouzy, E. and Fundenberger, J. J., “Textures and Anisotropy of Titanium
Alloys,” Mater. Sci. Forum, Vols. 273–275, 1998, pp. 511–522.
[30] Bechade, J. L., Bacroix, B. and Guillen, R., “The Effects of the Cold Rolling Process on the
Texture of Zircaloy-4 Sheets,” Mater. Sci. Forum, Vols. 157–162, 1994, pp. 617–626.
[31] Molinari, A., Canova, G. R. and Ahzi, S., “A Self Consistent Approach of the
Large Deformation Polycrystal Viscoplasticity,” Acta Metall., Vol. 35, 1987, pp.
2983–2994.
[32] Tomé, C. N. and Knocks, U. F., “The Yield Surface of h.c.p. Crystals,” Acta Metall., Vol. 33,
1985, pp. 603–621.
[33] Tomé, C. N., Lebensohn, R. A. and Knocks, U. F., “A Model for Texture Development
Dominated by Deformation Twinning: Application to Zirconium Alloys,” Acta Metall.
Mater., Vol. 39, 1991, pp. 2667–2680.
[34] Lebensohn, R.A. and Tomé, C. N., “A Self-Consistent Anisotropic Approach for the Simu-
lation of Plastic Deformation and Texture Development of Polycrystals: Application to
Zirconium Alloys,” Acta Metall. Mater., Vol. 41, 1993, pp. 2611–2624.
[35] Lebensohn, R. A. and Tomé, C. N., “A Study of the Stress State Associated With Twin
Nucleation and Propagation in Anisotropic Materials,” Philos. Mag., Vol. A67, 1993, pp.
187–206.
[36] Saibaba, N., Kumar Vaibhaw Neogy, S., Mani Krishna, K. V., Jha, S. K., Phani Babu, C.,
Ramana Rao, S. V., Srivastava, D. and Dey, G. K., “Study of Microstructure, Texture and
Mechanical Properties of Zr-2.5Nb Alloy Pressure Tubes Fabricated with Different Proc-
essing Routes,” J. Nucl. Mater., Vol. 440, 2013, pp. 319–331.
[37] Holt, R. A. and Aldridge, S. A., “Effect of Extrusion Variables on Crystallographic Texture
of Zr-2.5 wt. % Nb,” J. Nucl. Mater., Vol. 135, 1985, pp. 246–259.
[38] Cheadle, B. A. and Ells, C. E., “The Effect of Rolling Temperature on the Texture Devel-
oped in Rolled Zirconium Rich Alloys,” J. Nucl. Mater., Vol. 24, 1967, pp. 240–244.
[39] Bocek, N., Hofman, P. and Peterson, C., “Superplasticity of Zircaloy-4,” Zirconium in
Nuclear Industry: 5th International Symposium, ASTM STP 663, ASTM International,
[40] Sills, H. E. and Holt, R. A., “Predicting High-Temperature Transient Deformation from
Microstructural Models,” Zirconium in Nuclear Industry, 6th International Symposium,
ASTM STP 681, ASTM International, Philadelphia, PA, 1979, pp. 325–341.
[41] Tenckhoff, E. and Rittenhouse, P. L., “Texture Gradients in Thin Walled Zircaloy Tubing,
Applications Related Phenomena for Zirconium and Its Alloys,” Zirconium in Nuclear
Application, ASTM STP 458, ASTM International, Philadelphia, PA, 1969, pp. 50–67.
[42] Tenckhoff, E., “Operable Deformation Systems and Mechanical Behavior of Textured Zir-
caloy Tubing,” Zirconium in Nuclear Applications, ASTM STP 551, ASTM International,
DISCUSSION
Question from Malcolm Griffiths, AECL:—You mentioned getting a high trans-
verse texture with a higher Q-ratio. What temperature was the extrusion and what
do you estimate the relative volume fractions of a and b-phases?.
Authors’ Response:—The extrusion temperature was 800 C and at that temper-

ature, there is nearly equal volume fractions of the two phases.
A. Gaillac,1 D. Duthoo,2 C. Vauglin,3 D. Carcey-Collet,1

F. Bay,4 and K. Mocellin4
Numerical Modeling of Fuel Rod

Resistance Butt Welding
Reference
Gaillac, A., Duthoo, D., Vauglin, C., Carcey-Collet, D., Bay, F., and Mocellin, K., “Numerical
Modeling of Fuel Rod Resistance Butt Welding,” Zirconium in the Nuclear Industry: 17th
ABSTRACT
Resistance butt welding (RBW) is the main process used by AREVA to weld
end plugs to fuel rod cladding. The combination of current, electrical resistivity
of the plug–cladding contact, and applied pressure is set so as to weld the
parts without fusion of the materials. Because this is a solid state welding
process, one of its main advantages is that the corrosion behavior of the welds
is very good, because it is not affected by any potential contamination from the
welding atmosphere or other components. The RBW welds are geometrically
characterized by plastic deformation of the welded interface that depends on
process parameters such as applied current and pressure, welding time, and
initial geometry of the parts. Microstructural evolutions and extension of the
heat-affected zones are also consequences of these parameters. In order to
better understand the multiphysical phenomena involved during the very short
welding time of this process, numerical modeling of RBW was developed. The
model is also very useful for investigating new configurations and further
Manuscript received December 21, 2012; accepted for publication September 27, 2013; published online
September 26, 2014.
1
AREVA/CEZUS Research Center, Ave. Paul Girod, F-73403 Ugine Cedex, France.
2
AREVA/FM, FBFC BP 1114, 26104 Romans sur Isère, France.
3
AREVA/FD, 10 rue Juliette Récamier, 69456 Lyon, France.
4
MINES ParisTech, CEMEF, Center for Material Forming, CNRS UMR 7635, 1 Rue Claude Daunesse, BP 207, F-
06904 Sophia Antipolis Cedex, France.
5
Hyderabad, India.
Copyright V
optimizing weld quality with a reduced number of welding samples. The RBW
numerical model was implemented within the finite element code FORGE, and
electrical phenomena were coupled with thermal and mechanical ones.
Mechanical properties of the tube and the plug were described as a function of
strain, strain rate, and temperature. The temperature and pressure dependence
of the electrical contact resistance were also determined and taken into
account. Thanks to a good description of the thermal and mechanical history of
the weld, numerical results, such as weld geometry, plastic strain, and heat-
affected zones, were in good agreement with experimental ones. The model
was primarily used to understand the effects of process parameters on
geometry and temperature, but it was also used to optimize welding
conditions.
Keywords
zirconium alloys, fuel rods, welding, numerical modeling, finite element
Introduction
Resistance butt welding (RBW) is the main process used by AREVA to weld end
plugs to fuel rod cladding. The welding cycle includes three steps (Fig. 1):
• The end plug electrode applies a force on the plug, which comes in contact
with the tube. This force is maintained until the end of the welding.
• An alternating or dc current is injected by the cladding electrode, in contact
with the external surface of the cladding tube. At the tube–plug interface, the
contact resistance and this current act as a heat source. In combination with
the applied force, this heat source is responsible for the deformation and weld-
ing of the tube and plug.
• After the current is cut off, the force is maintained until the weld is cooled and
the deformation ceases.
This solid state welding process has two main advantages:
• The corrosion behavior of the welds is very good, because it is not affected by
any potential contamination from the welding atmosphere.
• Volumetric defects due to solidification shrinkage are avoided and x-ray
examinations are then unnecessary.
RBW welds are geometrically characterized by plastic deformation of the
welded interface that depends on process parameters such as applied current and
pressure, welding time, and initial geometry of the parts. Microstructural evolution
and extension of the heat-affected zones are also consequences of these
parameters.
In order to better understand the multiphysical phenomena involved during
the few milliseconds over which this process takes place, numerical modeling of
RBW has been used since 2007. This is also very useful in the study of new configu-
rations and for further optimization of the weld quality with a reduced number of
welding samples, thanks to a well-designed numerical test matrix.
GAILLAC ET AL., DOI 10.1520/STP154320120214 333
FIG. 1 The resistance butt welding process: (a) setup of the different parts and tools;
(b) the welding cycle.
Description of the Numerical Model

MECHANICAL, THERMAL, AND ELECTRICAL FORMULATIONS
The RBW numerical model was implemented within the finite element code
FORGE, and electrical phenomena were coupled with thermal and mechanical
ones.
The electric model is based on the Poisson equation, where req(T) is the
temperature-dependent electrical conductivity and V is the electric potential field.
divðrelec ðT ÞgradðV ÞÞ ¼ 0
The solid mechanics model is based on the classical virtual work principle,
which expresses the equilibrium of a workpiece undergoing thermomechanical
loads, as well as on a general constitutive law based on the material mechanical
properties. More details can be found in Ref 1.
Temperature evolution in the workpiece is governed by the classical heat trans-
fer equation, where q is the material density, C is the specific heat, k is the thermal
conductivity, Pelec is the heat source term representing the heat dissipated by elec-
tric current because of the volumic Joule effect, and Pmech is the heat source term
representing the heat dissipated by the plastic strain.
@T
qC ðT Þ divðkðT ÞgradðT ÞÞ ¼ Pmech þ Pelec
@t
Pmech ¼ r : e_
Pelec ¼ relec ðT ÞkgradðV Þk2
At the free surfaces, the model takes into account convection and radiation
heat exchanges. The convection exchange coefficient is represented by h, eT is the
emissivity, and r is the Stefan constant.
dT
k ðT Þ ¼ hðT Text Þ þ eT r T 4 Text
4
dn
At the interface of tube, plug, and electrodes, the thermal contact resistance
Rc_therm and the electrical contact resistance Rc_elec are taken into account, as well as
their variation with respect to contact pressure and temperature.
dT 1 1
kðT Þ ¼ ðT1 T2 Þ þ ðV1 V2 Þ2
dN Rc therm Rc elec
THERMO-ELECTRO-MECHANICAL COUPLING
A weak coupling procedure is used. At each time step, the electric potential is first
computed. The temperature is then computed using the source term coming from
the volume and surface Joule effects. Finally, the mechanical problem is solved. Ma-
terial and interface properties are then updated for the next time step. To avoid any
convergence problems of the model, the maximum time step was fixed at 0.01 ms.
FINITE ELEMENT MODEL

The parts are discretized by triangular finite elements. The thermal and electric
problems are solved with a P1 approximation of the temperature and the electric
potential. The mechanical problem is solved by a mixed velocity/pressure formula-
tion using a P1þ/P1 element. The nonlinearity arising from the rheological and
FIG. 2 Microstructure of an RBW weld (radial/longitudinal plane).
contact behavior is solved using a Newton–Raphson algorithm. All numerical

methods for the thermomechanical model are detailed in Ref 2.
The mesh is refined in the areas where the main strain will take place, that is, at
the plug–tube interface, mainly in the tube (Fig. 3). The mesh size is also reduced at the
interfaces with the electrodes in order to provide better resolution of the temperature
and electrical potential gradients in these regions. For this purpose, “mesh boxes” were
used to define the local mesh size. A test matrix was used to optimize the mesh size, in
order to reduce the calculation time without decreasing the quality of the results.
MATERIAL PROPERTIES
Elastic-plastic properties of the tube and the plug are described as a function of
strain, strain rate, and temperature, using compression, tensile, and torsion tests on
Zircaloy-4 specimens taken from recrystallized bars (Fig. 4). Rigid behavior is
assumed for the electrodes, which are not plastically deformed during welding.
Thermal properties (specific heat and thermal conductivity) are functions of
the temperature and take into account the phase changes for the tube and the plug
(Fig. 5). The specific heat was calculated using the TCC-S version of THERMO-
CALC, with a database that was slightly modified from the ZIRCOBASE [3,4]. For
the copper electrodes, thermal properties were taken from the literature.
FIG. 3 Finite element mesh of the parts.

FIG. 4 Equivalent stress as a function of equivalent plastic strain and temperature for
0.1 s1 and 10 s1 strain rates.
Electrical properties (electrical conductivity) of copper and zirconium (Fig. 6)

are also a function of the temperature and were taken from the literature.
BOUNDARY AND INTERFACE PROPERTIES

• Mechanical properties: At the interface between tube and plug, a sticking con-
tact is assumed, because of the fast welding of these two surfaces. At the inter-
faces with electrodes, a sliding contact is assumed, with a friction coefficient
taken from the FORGE database.
• Thermal properties: At the free surfaces, the convection coefficient and emis-
sivity were taken from the FORGE database. At the interface of tube, plug, and
electrodes, thermal resistance values were also taken from the FORGE
database.
• Electrical properties: The current is applied with a Neumann condition on the
cladding electrode. A Dirichlet condition is applied on the plug electrode for
the current exit. At the interface between plug and tube, the electrical resist-
ance is defined as a function of pressure and temperature. Values come from
hot compression tests of copper and Zircaloy-4 specimens and are between
1011 X m2 and 108 X m2. Details of these tests are presented in Ref 5.
FIG. 5 Thermal properties of the tube and plug as a function of temperature.

FIG. 6 Thermal conductivity of zirconium and copper as a function of temperature.
Validation of the Numerical Model

The model was first tested on a standard welding configuration in order to validate
it through comparison with experimental results. Two different welding currents
were used in order to test the model validity. Both were pseudo-continuous cur-
rents, and their respective effective values were 8 kA and 16 kA. These currents
were measured during the welding tests as a function of time and were introduced
as such into the numerical model (Fig. 7).
Figure 8 shows the numerical initial and final geometry of the parts with these
two different welding currents.
To validate the model, the weld length and weld angle change (with respect to
the initial plug angle) were compared to experimental measurements as well as the
displacement of the plug, which was measured for each welding (Fig. 9). The nu-
merical results were in very good agreement with experimental ones, without any
FIG. 7 Current applied for the two validation computations.

FIG. 8 Geometries of standard welding configuration: (a) initial geometry of the parts;
(b) final geometry of the weld for an 8-kA welding current; and (c) final
geometry of the weld for a 16-kA welding current.
FIG. 9 Validation of the weld dimensions and plug displacement by experimental

measurements for the two applied currents.
adjustment of the model parameters (neither boundary conditions nor material

parameters).
Application
UNDERSTANDING THE RESISTANCE BUTT WELDING PROCESS AND THE
INFLUENCE OF CURRENT
The numerical model, validated against experimental measurements, can thus be
used to better understand the RBW process.
The effect of current is clear: the higher the current, the greater the temperature
increase in the weld (Fig. 10). The consequence is a reduced strength of the material
and, with a constant load, greater strain in the weld and greater displacement of the
plug. The maximum temperature in the weld was reached after 3 ms for the 8 kA
current. This time corresponds to the maximum current value. For the 16 kA cur-
rent, the maximum temperature was reached after 6 ms. To understand this, we fur-
ther analyzed the location and the evolution of the temperature for the 16 kA case.
To locate the maximum temperature during the welding, three numerical sen-
sors were used. These sensors were placed at the outer side, at mid-thickness, and at
the inner side of the final weld geometry.
The temperatures recorded by these three sensors gave the following informa-
tion (Fig. 11):
• Up to 3 ms, the maximum temperature was located at the inner side of the
weld.
• Then the maximum temperature moved to the outer side of the weld.
• The maximum temperature was reached after 6 ms.
• Then, the temperature decreased while the current was still high.
• The temperature finally decreased faster when the current decreased from
18 ms to the end of welding.
FIG. 10 Maximum temperature during the welding.

FIG. 11 Temperature during the welding for the 16-kA current at the inner side of the
weld, at mid-thickness, and at the outer side of the weld.
These numerical temperature results are very difficult to validate with experi-
mental measurements, as the heated zone is very small and the welding time very
short. However, extreme welding conditions (very high current) experimentally and
numerically tested showed that when superficial fusion occurs, it is always located
at the outer side of the weld.
To explain this, the current density and temperature at 0.7 ms (at the beginning
of welding) and 6.2 ms (when the maximum temperature was reached at the outer
side of the weld) are shown in Figs. 12 and 13.
At the beginning of welding, the tube is in contact with the plug on the inner
side. The current coming from the outer surface of the tube is then heating the tube–-
plug interface, and the temperature increases rapidly. For this reason, at the begin-
ning of welding, the maximum temperature is located at the inner side of the weld.
FIG. 12 Current density (A/m2) at two welding times for the 16-kA current case.
FIG. 13 Temperature ( C) at two welding times for the 16-kA current case.
After 4 ms, the weld interface is partially formed. At this moment, the current
travels along the shorter path—that is, along the outer side of the welding interface,
because the plug electrode is on the outer surface of the plug.
At 6 ms, the plug comes in contact with the cladding electrode, forming a short
circuit. The current is then primarily flowing from the cladding electrode to the
plug, without passing through the weld interface, because the electric contact resist-
ance of the copper–zirconium interface is lower than that at the zirconium–-
zirconium interface. For this reason, even if the current is still high, the
temperature starts to decrease.
After 6 ms, the current keeps heating the plug–electrode interface and, to a
lesser extent, the plug–tube interface. But because the two contact surfaces are get-
ting larger and larger, the temperature slowly decreases.
After 18 ms, the current decreases and the weld cools down until the end of
welding.
OPTIMIZING THE WELD QUALITY

The two welds used to validate the model were also mechanically tested at room
temperature with a burst test (inner pressure applied in the tube). This test showed
good resistance of the 16-kA weld (rupture in the tube), whereas the 8-kA weld
exhibited a fracture on the welding surface for a lower burst pressure. The reasons
for the 8-kA weld’s lack of strength are its smaller size (shorter welded length than
the 16-kA weld) and its lower “welding quality” due to a lower welding temperature
and thus a lack of sufficient diffusion bonding at the plug–tube interface.
The finite element model of the RBW process, coupled with a burst test model,
is now used to simulate experimental good and bad welds in order to determine the
geometry and temperature criteria defining weld quality. Once this is achieved, the
model will be a very useful tool for optimizing weld quality and designing new
welding configurations.
Summary and Conclusion

The RBW numerical model was implemented within the finite element code FORGE,
and electrical phenomena were coupled with thermal and mechanical ones. Mechani-
cal properties of the tube and the plug were described as a function of strain, strain
rate, and temperature. The temperature and pressure dependence of the electrical
contact resistance were also taken into account. Thanks to a good description of the
thermal and mechanical history of the weld, numerical results, such as the weld ge-
ometry, were in good agreement with experimental ones. The model was first used to
understand the effect of process parameters on geometry and temperatures reached
during welding, analyzing the current density during the welding time.
The RBW model, coupled with a burst test model, is now used to select
parameters for new welding configurations in order to optimize the weld quality.
ACKNOWLEDGMENTS
The writers thank all the people at FBFC Romans who helped with the RBW experi-
ments presented in this paper and used to validate the model. The present work also
relied largely on the work performed by graduate students at Cemef, Larbi Arbaoui,
Aurélien Milhe, Christophe Pradille, and Mohamed Ali Mzoughi, who created and
improved the RBW finite element model within the FORGE software. Industrial appli-
cation of this model was also a success thanks to the work of other students at the
Cezus research center: Mounir Ba, Julien Armand, and Xavier Cerutti.
References
[1] Mocellin, K., Jarleton, L., Dahan, Y., and Bay, F., “A Numerical Model Coupling Electrical,
Thermal and Mechanical Effects for Analyzing Material Deformation,” 9th International
ESAFORM Conference on Material Forming, N. Juster and A. Rosochowski, Eds., Akapit,
pp. 527–531.
[2] Chenot, J.-L. and Bay, F., “Modeling of Metal Forming Processes and Multi-Physics
Coupling,” COMPLAS VIII International Conference on Computational Plasticity, Barce-
lona, Spain, Sept 2005.
[3] Dupin, N., Ansara, I., Servant, C., Toffolon, C., Lemaignan, C., and Brachet, J. C., “A Thermo-
dynamic Database for Zirconium Alloys,” J. Nucl. Mater., Vol. 275, No. 3, 1999, pp. 287–295.
Component Zirconium-Based Alloys,” ASTM STP 1505, ASTM International, West Con-
[5] Pradille, C., Mocellin, K., and Bay, F., “An Experimental Study to Determine Electrical
Contact Resistance,” 25th ICEC and 56th IEEE Holm Conference, Charleston, SC, Oct.
4–7, 2010, pp. 329–333.
DISCUSSION
Question from Jean-Christophe Brachet, CEA:—What is the typical heating rate
of the process? It could be interesting to look at the local microstructure because
on-heating, the a ! b phase transformation occurs under nonconventional, very
dynamic conditions (e.g., occurrence of martensitic or “massive” type a ! b phase
transformation, no micro chemical partitioning, . . .).
Authors’ Response:—In Figure 10, the maximum temperature of 1500 C is

reached after 6 ms, so a heating rate of 250000 C/s can be estimated from this
result. The a ! b phase transformation occurs under a very dynamic condition.
Simultaneously, very large deformations occur and then the b ! a phase transfor-
mation during cooling. As a consequence of this, microstructure of welds is very
complex and not only a consequence of the a ! b phase transformation.
Question from J. Romero, Westinghouse Electric Co.:—Is the holding force dur-
ing welding constant? What is the effect of changing the force on temperature, ma-
terial flow, etc.?
Authors’ Response:—The holding force during welding is relatively constant. A

higher applied force results in greater material flow and a larger the weld, whereas
the temperature remains constant.
Question from K. Kapoor, NFC, kapoork@nfc.gov.in:—During heating in RBW

process, the temperature rise is very rapid. The isothermal conditions used in the
thermal model are simplistic in nature. However, in the real situation, conditions
are more complex and to some extent adiabatic. How does the model handle these
conditions during welding?
Authors’ Response:—In the present model, the heating rate does not have an
effect on the thermal properties of the material (for example on the phase transfor-
mation temperature). However, the diffusion of the heat coming from the electrical
contact resistance and from the plastic strain is taken into account. In our case,
heating is fast and the diffusion of temperature in the parts (tube, plug and elec-
trode) is quite low during the 20 ms welding time.
Questions from J Jha, Nuclear Fuel Complex, Hyderabad, IN
Q1:—What will be the strain rate during butt welding simulation?
Authors’ Response:—Maximum strain rate during welding can be up to 1000s1.
Q2:—Can you relate welding parameters with strain rate?

Authors’ Response:—As the applied force is constant, strain rate will be mainly
a function of the heating rate. Maximum strain rates will be reached for the higher
currents, which produce very fast temperature increase.
Questions from R. K. Chaube, Nuclear Fuel Complex
Q1:—Is the model able to take care of geometrical changes with respect to plug
geometry?
Authors’ Response:—Yes, the model can be used to quantify the influence of the
geometry, such as the plug angle, the electrode shape, etc.
Q2:—What is the probability of getting leak defects from the weld?
Authors’ Response:—Our experience is that we never have seen any leaking

rods in core related to a defective weld with this process. As for any welding pro-
cess, the cleanliness of the components and the monitoring of the process parame-
ters are essential factors to get such performance.
Questions from GVS Hemantha Rao, Nuclear Fuel Complex,

hemanth@nfc.gov.in
Q1:—Does the model considered argon, helium gas and its effect on heat
transfer?
Authors’ Response:—A convection coefficient is used for the heat exchange with
inert gas during welding. Since the welding time is very short, this coefficient has a
low effect on the calculated temperature and geometry of the weld.
Q2:—The actual weld current is cyclic (preheat, weld, post heat). Does the
model account for this?
Authors’ Response:—Every kind of current can be used in the model, since the
current is described in a data file as a function of time. In our case, the applied cur-
rent is described in Figure 7.
Q3:—The electrodes are taken as copper. Actually the electrodes are RWMA
Class 12 or Class 2 where properties are very much different from copper.
Authors’ Response:—In our model, electrodes are taken as pure copper. Differ-
ent copper alloys have been experimentally tested without significant effect on
the weld characteristics. Material is chosen mostly based upon the electrode s
lifetime.
Questions from Vikrant Raizada, Nuclear Fuel Complex
Q1:—Is the study a 2D or 3D analysis? If it is a 3D analysis, then what type of

element is used?
Authors’ Response:—This is a 2D axisymetric model.
Q2:—Can the modeling be performed using any other general purpose software
such as ANSYS?
Authors’ Response:—To model the RBW process, it is necessary to couple the

mechanical, thermal and electrical formulations. I do not know if it is possible with
the ANSYS software.
Q3:—How are the displacement boundary conditions taken care of in the

model?
Authors’ Response:—The plug is free of axial displacement. It will be a function

of the applied force and the reached temperature. The tube is blocked by the tube
electrode, assuming a sticking contact between these two parts.
Hyun-Gil Kim,1 Il-Hyun Kim,2 Jeong-Yong Park,2

and Yang-Hyun Koo2
Application of Coating
Technology on Zirconium-Based
Alloy to Decrease
High-Temperature Oxidation
Reference
Kim, Hyun-Gil, Kim, Il-Hyun, Park, Jeong-Yong, and Koo, Yang-Hyun, “Application of Coating
Technology on Zirconium-Based Alloy to Decrease High-Temperature Oxidation,” Zirconium
in the Nuclear Industry: 17th International Symposium, STP 1543, Robert Comstock and Pierre
Barberis, Eds., pp. 346–369, doi:10.1520/STP154320120161, ASTM International, West
ABSTRACT
Since the Fukushima accident, it has been recognized that a hydrogen-related
explosion is one of the major concerns regarding reactor safety during the high-
temperature oxidation of zirconium alloys. To decrease the high-temperature
oxidation rate of zirconium-based alloy, a coating technology for the zirconium
alloy surface was considered. The selection of coating materials was based on the
neutron cross-section, thermal conductivity, thermal expansion, melting point,
phase transformation behavior, and high-temperature oxidation rate. After
consideration of these factors, silicon was selected as a coating material for the
first surface coating of zirconium-based alloy. A plasma spray and laser beam
scanning were selected for the coating method, as both can be applied to a long
tube shape without high-vacuum and high-temperature environments during the
coating process. After Si-coated samples on Zircaloy-4 sheet had been prepared
via plasma spray and combined plasma spray–laser beam scanning treatments,
Manuscript received November 19, 2012; accepted for publication August 31, 2013; published online
September 15, 2014.
1
LWR Fuel Technology Division, KAERI, 989-111 Daedeok-daero, Yuseong-gu, Daejeon, 305-353, Republic of
Korea (Corresponding author), e-mail: hgkim@kaeri.re.kr
2
LWR Fuel Technology Division, KAERI, 989-111 Daedeok-daero, Yuseong-gu, Daejeon, 305-353, Republic of
Korea.
3
Hyderabad, India.
Copyright V
KIM ET AL., DOI 10.1520/STP154320120161 347
the samples were tested in a high-temperature steam environment at 1200 C for

2000 s. The adhesion property of the Si-coated layer prepared via plasma spray
was insufficient for the high-temperature application, whereas the stability of the
Si-coated layer prepared with both plasma spray and laser beam scanning
treatments was maintained without peeling after a high-temperature oxidation
test. A very thin oxide layer, which was a few micrometers in thickness, was
formed on the Si-coated surface, whereas a ZrO2 layer 110 lm in thickness was
observed on the Zircaloy-4 surface. From this result, it was determined that the
Si-coated layer successfully acted as corrosion barrier layer to resist the high-
temperature oxidation of zirconium-based alloy.
Keywords
zirconium, high-temperature oxidation, coating
Introduction
The development of zirconium alloys has recently focused on decreasing the corro-
sion rate and hydrogen pickup under normal operating conditions in order to
increase the operation economy and safety margin [1–3]. After the Fukushima acci-
dent, it was recognized that hydrogen generation is one of the major concerns
regarding reactor safety, as serious reactor damage can be caused by a hydrogen
explosion. Hydrogen is generated by the corrosion reaction of zirconium alloys
found in reactor elements such as the fuel cladding, spacer grid, and channel box,
and the corrosion reaction can be considerably increased with an increase in the
environmental temperature [4,5]. Thus, a decrease of the high-temperature oxida-
tion rate of zirconium alloys was the key to decreasing the hydrogen generation
during the nuclear power plant accident.
The current method used to increase the corrosion resistance of zirconium alloy
for nuclear applications basically adjusts alloying elements such as Nb, Sn, Fe, and Cr
and their ratios. However, the oxidation rate of zirconium alloys at a high tempera-
ture of 1200 C is not considerably changed with the alloy composition [6,7]. Thus, it
is a problem that a decrease in the oxidation rate of zirconium-based alloys at high
temperature is difficult to achieve using commercial materials. To overcome the
acceleration of the high-temperature oxidation of zirconium alloys, research into new
materials and concepts has been suggested [8,9]. A SiC-based ceramic cladding is
being investigated to replace zirconium alloys, because SiC is durable up to 1500 C
without a loss of strength and shows very good oxidation resistance in high-
temperature environments [8]. A novel design concept of a Mo–Zr duplex and
Mo–Zr triplex metal composite cladding is proposed as the best candidate for further
development [9]. As a separate method, the coating of incorrodible materials on a
zirconium alloy surface can be considered. Coating technology is widely applied with
other industrial materials to reduce corrosion and wear damage, as resistance to cor-
rosion and wear can be easily conferred by a coating without a change in the base
material. Based on this, we considered a comparison between technical issues and the
FIG. 1 Consideration of the technical issues and time to development of the accident-
tolerant fuel cladding.
development time of the fuel cladding as shown in Fig. 1 to improve the accident tol-
erance of the fuel cladding. Thus, surface-modified zirconium alloy was selected for
this work after technical deliberation for a decrease in the high-temperature oxidation
rate, near-term application, easy fabrication, economic benefit, and easy verification,
although the high-temperature strength was reduced more than for other suggested
technologies involving hybrid and full ceramic materials.
Among the surface-coating methods, plasma spray-coating technologies, which
deposit melted material powders on the substrate using a heat source generated by
a plasma torch nozzle, have been used to increase the high-temperature oxidation
resistance of various materials [10–12]. Also, laser cladding techniques have been
applied to the surface cladding of ceramics and intermetallic compounds on the
metal base and ceramic-based components used to increase the resistance to corro-
sion and wear [13,14]. However, an optimized technology for the coating materials
and coating methods must be developed for nuclear applications, as coating tech-
nologies to increase the high-temperature oxidation resistance of zirconium-based
alloys have not been determined at the present time. Thus, this work studied coat-
ing techniques for both coating methods and coating materials to reduce the oxida-
tion rate of zirconium-based alloy in a high-temperature steam environment.
Experimental Procedures
COATING MATERIALS
A Zircaloy-4 (Zr-1.5Sn-0.2Fe-0.1Cr in wt. %) alloy sheet with dimensions of 95 mm
by 25 mm by 2.54 mm was used as a substrate, as this material has been used in fuel
KIM ET AL., DOI 10.1520/STP154320120161 349
cladding tubes, spacer grids, and guide thimble tubes in the fuel assemblies of light-
water reactors. The initial state of the Zircaloy-4 sheet was recrystallized with a
mean grain size of 7 lm. Before the plasma spray, the substrates were cleaned with
alcohol to remove any stains or contamination on the surface and then dried.
The selection of coating materials was based on the neutron cross-section, ther-
mal conductivity, thermal expansion, melting point, phase transformation behavior,
and high-temperature oxidation rate. The metal base materials Si and Cr were
selected for coating the surface of zirconium-based alloy. Si was the focus of the
current work. The mean size of the Si powders (3 N purity) used as a raw material
for coating was 90 lm. A Si wafer with dimensions of 10 mm by 10 mm by
0.625 mm was used to evaluate the high-temperature oxidation behavior, because
the high-temperature oxidation test was impossible using the Si powders. Also, Cr
and SiO2 with dimensions of 10 mm by 10 mm by 1 mm was used to evaluate the
high-temperature oxidation behavior.
COATING METHODS
Two applied coating techniques were used: for the first, pure Si powders were
attached to the Zircaloy-4 sheet surface via a plasma spray (PS) coating method,
and for the second, a Si-coated layer applied via PS was treated by laser beam scan-
ning (LBS) to increase the adhesion between the Zircaloy-4 matrix and the Si layer.
Figure 2 shows schematic drawings of the PS coating method and the LBS method
used to make a Si-coated layer on the Zircaloy-4 sheet surface and the appearance
of the Si-coated layers obtained via the two methods.
For PS coating, a thermal spray machine (LCP Rev. A model, Sulzer Metco
Co.) was used for the plasma coating, and Ar was used as the inactive gas. With the
use of a feeder, 90 -lm Si powders were propelled through an Ar inlet under a pres-
sure of 100 MPa. One, three, and six times the PS coating, each for 20 s, were
applied to control the thickness of the coating layer. The plasma gun and the sam-
ple were maintained at a 100-mm distance from each other.
The Si layer applied to a Zircaloy-4 sheet by applying the PS three times was
scanned using a continuous wave diode laser with a mean power of 300 W (PF-
1500 F model, HBL Co.). The LBS parameters applied to the Si-coated Zircaloy-4
alloy sheet were developed based on a preliminary study. Finally, the applied power
for the LBS treatment ranged from 80 to 150 W, and the scanning speed was
10 mm/s. A schematic drawing of the LBS treatment is shown in Fig. 2. To prevent
oxidation during LBS, an inert gas (Ar) was continuously bellowed into the melted
zone.
MICROSTRUCTURE CHARACTERIZATION
After the coating treatments of the samples—both PS coating and LBS treatment af-
ter PS coating (PS þ LBS)—the microstructure and composition of a Si-coated layer
in the cross-sectional direction was determined using scanning electron microscopy
(SEM) with energy dispersive spectrometry (EDS), and transmission electron
FIG. 2 Schematic drawings of the plasma spray (PS) coating method and laser beam
scanning (LBS) method used to make a Si coating layer on the Zircaloy-4 sheet
surface. the bottom panel shows the surface appearance of Si-coated layers
created using the two methods.
microscopy (TEM) analysis was also conducted. The samples for the TEM observa-
tion were prepared using focused ion beam equipment.
To investigate differences in the oxide properties between Zircaloy-4 and the
Si-coated layer, cross-sectional analyses using SEM and TEM were performed after
a high-temperature oxidation test. An evaluation of the oxidation thickness of the
tested samples was carried out using SEM images and an EDS profile of alloying
elements such as Si, Zr, and oxygen for the cross-sectional direction.
HIGH-TEMPERATURE OXIDATION TEST

The Zircaloy-4 and Si-coated samples prepared via PS and PS þ LBS as shown in
Fig. 2 were cut to dimensions of 10 mm by 10 mm, and the cut surface was ground
with silicon carbide (SiC) paper. The ground samples were washed via ultrasonic
cleaning in alcohol for 5 min and then dried. The dried samples were mounted in
the test equipment for high-temperature oxidation, and a mixture of steam and Ar
gas was then introduced at a flow rate of 10 ml/min. If the mixed gas flow was faster
than 10 ml/min, the in-suit measurement of weight gain data was not reasonable
because of the high fluctuation in weight change. The temperature of the samples
rose 50 C/min, and the temperature was maintained at 1200 C for 2000 s. During
the heating ramp, only Ar was continuously supplied in order to prevent oxidation,
and a mixture of steam and Ar gas was supplied at the target test temperature. The
KIM ET AL., DOI 10.1520/STP154320120161 351
high-temperature oxidation test was performed on two samples for each material,
and the in-suit weight gain data were acquired every 5 s during the test.
Results
MICROSTRUCTURE OF THE SI-COATED LAYER
Figure 3 shows a cross-sectional SEM image together with an EDS analysis of a Si-
coated layer created via PS on a Zircaloy-4 sheet. The thickness of the Si-coated
layer was determined from the SEM image and EDS profile. For this sample, the
thickness of the Si-coated layer was increased by means of repeated spraying (one
to six passes). After the first pass, the mean thickness of the Si-coated layer was
15 lm, and the variation of the thickness was 63 lm. Thus, a uniform layer thick-
ness cannot be obtained via a one-pass PS treatment with Si powder. The mean
thickness of the Si-coated layer after three and six passes was 70 and 130 lm,
respectively. Many irregularly shaped pores were observed in the Si-coated layer.
Although many aspects of the PS coating method, such as the distance between the
plasma nozzle and the specimen, the plasma power, and the Ar gas pressure, were
varied in an attempt to decrease the pore density in the Si-coated layer, the sound-
ness of the Si-coated layer could not be improved. Because many pores formed in
FIG. 3 Upper panels: cross-sectional SEM observation of Si-coated layers on Zircaloy-4

sheet created via one to six PS passes. Lower panels: EDS analysis results for the
Zr and Si element profiles.
FIG. 4 Cross-sectional SEM observation (upper panel) and EDS profile analysis for the
Zr and Si element profiles (lower panels) enabling comparison of the
microstructure characteristics obtained with PS coating and LBS treatment
after PS coating.
the PS Si-coated layer in this work, it is thought that a Si-coated layer applied via
PS cannot be directly applicable as a fuel cladding coating. From SEM EDS analysis
of the interface between the Si-coated layer and the Zircaloy-4 substrate, it can be
assumed that an alloy mixing zone between Zr and Si elements did not form at the
interface. Thus, it is known that the PS method created a pure Si layer with pores
that was attached to the Zircaloy-4 matrix without a diffusion zone between Si and
Zircaloy-4.
LBS treatment was applied after PS coating (PS þ LBS) on samples of a three-
pass Si-coated layer to remove pores and to improve adhesion to the Zircaloy-4
substrate, as shown in Fig. 2. The optimized laser power of 130 W was determined
after various changes in power for a three-pass Si-coated layer of 70 -lm thickness.
The Si-coated layer was not fully melted in the low laser power condition; therefore,
a diffusion zone between Si and Zr was not formed, and the pores observed in the
Si-coated layer created via PS were not completely removed. In the high laser power
condition, the PS Si-coated layer was broken down by the excessive mixing between
Si and Zr elements. Figure 4 shows SEM and EDS images of the boundary between
KIM ET AL., DOI 10.1520/STP154320120161 353
the PS and PS þ LBS layers, enabling comparison of the microstructure characteris-

tics. In contrast to the Si-coated layer created via PS, shown in the left-hand part of
the SEM image, after LBS treatment the pores were fully removed, and a diffusion
zone between Si and Zircaloy-4 was formed. It was determined that the average Si
content of the Si–Zr mixed area was about 70 at. %; thus the pure Si layer was
changed to a compositionally mixed layer of Si and Zr by the LBS treatment. Thus,
LBS after PS coating of a three-pass Si-coated layer of 70 -lm thickness allowed us
to obtain a compositionally mixed and poreless layer.
Figure 5 shows a cross-sectional TEM image of the interface between the
Zircaloy-4 substrate and the Si-coated layer prepared via PS þ LBS. At the interface
between Si and Zircaloy-4, defects such as cracks and pores were not observed after
the LBS treatment. From the Zircaloy-4 matrix having a 7 -lm grain size before the
coating, new grains ranging from 100 to 500 nm in size were formed by the
PS þ LBS treatments, as shown in upper part of Fig. 5, and the new grains had crys-
talline characteristics, as identified from a selected area diffraction image. In terms
of compositional characteristics, the new grains containing the Si element ranged
from 5 to 20 at. % at the observed area of the interface, and the Si content was
increased when the distance from the interface between Zircaloy-4 and Si was
increased. Thus, a sound interface without defects and with a compositional gradi-
ent can be obtained by using an additional LBS treatment on a pure PS Si-coated
layer.
FIG. 5 Cross-sectional TEM observation of the microstructure and composition of the

interface region between Zircaloy-4 substrate and the Si–Zr mixed layer
prepared via LBS treatment after PS coating.
HIGH-TEMPERATURE OXIDATION BEHAVIOR OF COATING MATERIALS

AND SI-COATED SAMPLES
To evaluate the oxidation behavior, we performed a high-temperature oxidation
test for the considered coating materials of an Si wafer and metal Cr, the reference
materials of a Zircaloy-4 sheet and an SiO2 block, and Si-coated Zircaloy-4 sheet
prepared via PS and PS þ LBS treatments. Figure 6 shows the high-temperature oxi-
dation behaviors of samples tested at 1200 C for 2000 s as a function of the expo-
sure time. Because of the large discrepancies in weight gain among the tested
samples shown in Fig. 6(a), a semi-log scale graph was used to evaluate the oxida-
tion behavior of the reference materials. The oxidation behaviors of the tested sam-
ples of Zircaloy-4, Si wafer, metal Cr, and SiO2 block followed the parabolic rate
law as shown in Fig. 6(a). It was observed that the weight gain of 2500 mg/dm2 of
the recrystallized Zircaloy-4 sheet in this work was about 15 % lower than that
observed by Baek and Jeong [7] under the same test conditions, as well as the
Cathcart–Pawel oxidation prediction for Zircaloy-4 after oxidation testing for
2000 s at 1200 C [5]. It was thought that such discrepancies in weight gain might
have been caused by some differences in the experimental procedure between this
work and other studies, such as a different specimen shape, a final annealing condi-
tion of Zircaloy-4, and a mixed gas condition including steam and Ar. However,
the comparison of weight gain data between Si-coated samples and uncoated refer-
ence samples is acceptable because the shape and oxidation conditions of all sam-
ples were the same in this study.
In the comparison of the weight gain of tested samples, SiO2, Si, and Cr showed
superior oxidation resistance relative to Zircaloy-4 under a high-temperature steam
environment. In particular, the SiO2 block was very stable under the test conditions,
and the weight gains of the Si wafer and metal Cr were shown to be 30 and 75 mg/
FIG. 6 High-temperature oxidation behaviors and surface appearances of various

materials: (a) coating and reference materials (Si wafer, metal Cr, SiO2 block,
Zircaloy-4); (b) Si-coated Zircaloy-4 sheets (PS and PS þ LBS methods) tested
in a steam environment at 1200 C for 2000 s.
KIM ET AL., DOI 10.1520/STP154320120161 355
dm2, respectively. Thus, it is known that the oxidation resistance of Si is about 80

times greater than that of Zircaloy-4, and the oxidation resistance of Si is about
twice that of Cr.
Figure 6(b) shows the oxidation behaviors of a Si-coated Zircaloy-4 sheet pre-
pared using PS and PS þ LBS methods. The oxidation behaviors of the Si-coated
samples prepared via both methods varied with the number of coating passes. The
Si-coated sample prepared via PS þ LBS showed superior oxidation resistance rela-
tive to Zircaloy-4 and PS Si-coated samples based on a comparison of the weight
gain of tested samples. The interesting point is that the weight gain of PS Si-coated
samples changed with the number of coating passes. At the initial stage of the oxi-
dation test, the weight gain of the PS Si-coated samples was less than that of the
Zircaloy-4 sample. However, the weight gain of PS Si-coated samples increased
when the number of coating passes decreased from six to one. In surface observa-
tions of the tested one-pass PS Si-coated sample, we noted that a white oxide had
formed on the edge of sample, like on the Zircaloy-4 sample. It is assumed that this
edge/side oxidation can occur because of the oxidation behavior of PS Si-coated
samples. In contrast, the white oxide at the edge of the sample was not observed in
the sample coated via PS þ LBS after the oxidation test. In addition, the weight gain
measuring method for the Si-coated samples cannot clearly represent the oxidation
behavior during the high-temperature oxidation test, because the Si-coated area is
only one surface of the hexahedron shape of the tested samples. Thus, a cross-
sectional observation was performed on the Si-coated samples to evaluate the high-
temperature oxidation behavior.
MICROSTRUCTURE OF THE SI-COATED SAMPLE AFTER HIGH-TEMPERATURE

OXIDATION TEST
Cross-sectional SEM observations of the surface region of oxidized PS Si-coated
Zircaloy-4 samples were obtained in order to evaluate the oxidation characteristics
(Fig. 7). The microstructural observations after oxidation testing were compared
with the cross-sectional observations of PS Si-coated layers before oxidation as
shown in Fig. 3. The SEM EDS profiles for Si, Zr, and oxygen corresponded to the
one-pass Si-coated sample. It was determined that the oxygen content peak of the
Si-coated layer was similar to the background of the SEM EDS oxygen profile, and
the oxide layer thickness on the Si-coated layer was less than a few micrometers;
thus the oxidation of the Si-coated layer did not progress much after tests at
1200 C for 2000 s like Zircaloy-4. However, an intermediate layer was clearly
observed between the Si-coated layer and the Zircaloy-4 substrate, in contrast to the
structure before the oxidation test, and this intermediate layer was identified as a
ZrO2 phase from the SEM EDS analysis. It is thought that the formation of the in-
termediate ZrO2 phase was caused by oxygen diffusion along the interface between
the Si-coated layer and the Zircaloy-4 substrate, as the formation of the intermedi-
ate ZrO2 phase started at the edge of sample, as shown in the image for the six-pass
PS condition in Fig. 7. During the oxidation test, spalling of the Si-coated layer was
FIG. 7 Cross-sectional SEM observation with the spraying times (upper panels) and
EDS analysis of Si, Zr, and oxygen for the one-pass PS Si-coated layer (lower
panels) after the high-temperature oxidation test at 1200 C for 2000 s in a
steam environment.
shown at the edge region in the PS coated samples, and this phenomenon was fre-
quently observed more on single-pass coated samples than on multi-pass coated
samples. The white oxide was observed at the edge of oxidation-tested samples as
shown in Fig. 6(b). In addition, it was observed that the intermediate ZrO2 phase
thickness was greater at the edge region than at the center region of the one-pass
Si-coated sample. On the other hand, the oxygen diffusion through the Si-coated
layer can affect the formation of the intermediate ZrO2 phase layer. Because many
pores of irregular shape were observed in the PS Si-coated layer, as shown in Fig. 3,
the oxygen diffusion could progress through the pore network in the Si-coated
layer. Thus, the intermediate ZrO2 phase layer can be grown easily in the case of a
thin layer thickness of the Si-coated layer. For this reason, the weight gain of PS Si-
coated samples increased with decreasing numbers of coating passes.
KIM ET AL., DOI 10.1520/STP154320120161 357
FIG. 8 Upper panels: cross-sectional SEM observation of the Zircaloy-4, PS Si-coated

layer, and Si–Zr mixed layer created via PS þ LBS. Lower panels: EDS analysis
profiles for each prepared specimen after the high-temperature oxidation test at
1200 C for 2000 s in a steam environment.
The oxidation reaction along the interface between the Si-coated layer and the
Zircaloy-4 and throughout the pore network of the PS Si-coated layer cannot be
prevented when the interface is exposed to a steam environment. However, it was
determined that the oxidation reaction of the Si-coated layer was much less than
that of the Zircaloy-4 matrix, as was also shown in the Si wafer oxidation test. Thus,
it is known that the higher oxidation rate of the Si-coated samples, shown in Fig.
6(b), was caused by the Zircaloy-4 oxidation of the non-coated area and the coated
interface region between the Si-coated layer and the Zircaloy-4 matrix.
Figure 8 shows the cross-sectional SEM analysis results for three different sur-
face conditions. The thickness of the observed layers was determined via a combi-
nation of SEM image and EDS profile. At the Zircaloy-4 surface, two layers in both
the oxide having a columnar shape and the a-Zr(O) having a large grain size were
observed, and the thickness of both layers was about 220 lm (the ZrO2 phase was
110 lm) at the center region of the oxidation-tested sample. These two layers were
generally formed in zirconium-based alloys during the high-temperature oxidation
test. However, a triangular oxide formed at the edges of samples and was identified
as a white oxide from the surface appearance after the oxidation test as shown in
Fig. 6.
At the PS Si-coated layer, the Si layer with pores was maintained without a
severe oxidation reaction, but an intermediate ZrO2 phase of 180 -lm thickness was
formed between the Si-coated layer and the Zircaloy-4 substrate, which was men-
tioned in relation to Fig. 7. At the LBS-treated Si-coated layer, a Si–Zr mixed layer
was shown without the formation of an intermediate ZrO2 phase layer between the
Si–Zr mixed layer and the Zircaloy-4 substrate, and a thin oxidation layer a few
micrometers in thickness was observed at the surface of the coated layer in the EDS
analysis. Based on this, the Zr–Si mixed layer prepared via PS þ LBS showed good
oxidation resistance at high temperature relative to the Zircaloy-4.
To confirm adhesion of the Si-coated layer at high temperature between PS and
PS þ LBS layers, a high-temperature oxidation test and microstructure observation
were performed on the Si-coated sample as shown in Fig. 4. Figure 9 shows a cross-
sectional SEM image of the boundary between the PS and PS þ LBS layers after the
high-temperature oxidation test at 1200 C for 2000 s. The Si–Zr mixed layer pre-
pared via PS þ LBS treatment was maintained without severe oxidation and dam-
age, whereas the Si-coated layer applied via PS was lifted up and spalled by the
formation of an intermediate ZrO2 layer. From this, it is known that the stability of
a Si–Zr mixed layer prepared via PS þ LBS treatments can be maintained without
peeling after a high-temperature oxidation test.
Figure 10 shows a cross-sectional TEM image of the Si–Zr mixed layer prepared
via PS þ LBS after the high-temperature oxidation test at 1200 C for 2000 s. This
observation was performed to evaluate the oxidation behavior of the Si–Zr mixed
layer coated on the Zircaloy-4 matrix, because the oxidation resistance of the Si–Zr
FIG. 9 Cross-sectional SEM observation to compare the microstructure characteristics

obtained with PS coating (PS) and LBS treatment after PS coating (PS þ LBS)
after the high-temperature oxidation test at 1200 C for 2000 s in a steam
environment.
KIM ET AL., DOI 10.1520/STP154320120161 359
FIG. 10 Cross-sectional TEM observation of the microstructure and composition of the

surface region of the Si–Zr mixed layer prepared via LBS treatment after PS
coating after the high-temperature oxidation test at 1200 C for 2000 s in a
steam environment.
mixed layer was considerably increased relative to that of the Zircaloy-4 alloy. In
terms of grain morphologies, two grain shapes were observed at the surface of the
Si–Zr mixed layer after the oxidation test. Large grain regions of 1 -lm thickness
were observed at the outer surface region, and the composition of this region was
30 to 40 at. % oxygen, 5 to 20 at. % Si, and 40 to 65 at. % Zr. Also, a wavy elongated
grain region a few micrometers thick was observed at the inner part of the large
grain region, and the compositional characteristics of this layer were changed by a
decrease of 10 at. % oxygen and an increase of 10 at. % Zr relative to the composi-
tion of the large grain region. As the oxygen content of the Si–Zr mixed layer
decreased with increasing depth from the surface, the oxidation behavior of the
Si–Zr mixed layer followed the diffusion process of oxygen. The crystalline charac-
teristics were maintained in these two regions. The observed grain shapes of the
Si–Zr mixed layer after oxidation were very different from the grain shape of the co-
lumnar ZrO2 grains formed on the zirconium-based alloys at 1200 C [6], and the
oxygen content of the Si–Zr mixed layer was lower than the oxygen content of
zirconium-based alloys (about 66 at. % oxygen) after oxidation. From the TEM ob-
servation of the oxidized area of the Si–Zr mixed layer prepared via PS þ LBS, it
was recognized that the oxidation process of the Si–Zr mixed layer was very differ-
ent from that of Zircaloy-4 in terms of oxygen content and grain morphologies.
Based on these characteristics, the Si–Zr mixed layer showed better oxidation resist-
ance than the Zircaloy-4 substrate in the high-temperature steam environment.
Discussion
CONSIDERATION OF COATING METHODS AND MATERIALS
Many coating methods such as a PS, chemical/physical vapor deposition, and LBS
have been developed and applied in the commercial industry field. From among
these, we considered direct application methods for fuel assembly components hav-
ing a complex shape (e.g., cladding tube, guide tube, and spacer grid). Because the
length of the cladding and guide tubes is 4 m, a coating method having high vac-
uum control is not acceptable from an economical point of view. Also, it had to be
possible to apply the coating to an irregular surface shape, because the area to be
coated was not flat (tube and irregularly formed grid). For these reasons, the PS and
LBS methods were considered in this study. Because pores can form in PS-applied
coated layers owing to an inherent deficiency in the spray technology [15], laser
surface treatment of the PS-coated layer was applied to remove defects such as
inclusions and pores [16]. Thus, two types of coating methods, PS and PS þ LBS,
were selected in this work using the Si element. In this study, the PS coating tech-
nique was associated with problems such as the formation of pores in the Si-coated
layer and severe interface oxidation during the high-temperature oxidation test at
1200 C. The LBS treatment was applied to the PS Si-coated surface to remove the
pores and to increase the adhesion; as a result, the pores formed in the Si-coated
layer were clearly removed, and the interface oxidation was suppressed by the diffu-
sion bonding between the Zircaloy-4 substrate and the Si-coated layer, as shown in
Fig. 4. However, the pure Si was changed to a Si–Zr mixed phase by the LBS.
In order to apply the material to the nuclear field, it is necessary to check the
neutron cross-section and other physical properties (e.g., phase transformation
temperature, melting point, thermal expansion coefficient, and thermal conductiv-
ity). Thus, we checked out the coating material properties shown in Table 1. Among
the materials considered, the ceramics (Y2O3, SiO2, ZrO2, TiO2, Cr2O3,), carbides
(Cr3C2, SiC, ZrC), and nitrides (ZrN, TiN) are generally used as coating materials
for high-temperature application components because they have a high melting
point and high corrosion and wear resistance. However, it proved difficult to con-
trol the stoichiometry of the ceramics, carbides, and nitrides during the general
coating process, and another problem was weak adhesion to zirconium-based metal
caused by the technical difficulty of making a diffusion zone. Although the ceramics
showed good corrosion resistance, their thermal conductivity was considerably
decreased relative to that of the other metals, and the ceramics generally had low
impact strength. SiC was recently considered for application as a fuel cladding in a
light water reactor because of its strength and very good oxidation resistance in
very high temperatures [8]. However, the manufacturing technology needed in
order to make SiC cladding had not been developed [9], and the dissolution
KIM ET AL., DOI 10.1520/STP154320120161 361
TABLE 1 Summary of the basic properties of coating materials.
Thermal Thermal
Phase Expansion Thermal Neutron
Transformation Melting Coefficient Conductivity, Absorption,
Materials Temperature, C Point, C (106 K) W/mK Barns
Y2O3 None 2690 8.1 1.0 1.28(Y)

0.0002(O)
SiO2 Depends on 1600 12.3 1.3 0.177(Si)
pressure
0.0002(O)
ZrO2 M(970)/T(1205)/ 2130 10.1 1.8–3.0 0.182(Zr)
cubic
0.0002(O)
Cr2O3 None 2400 90 - 3.05(Cr)
0.0002(O)
Cr3C2 None 1895 10.3 13 3.05(Cr)
0.0035(C)
SiC (CVD) None 2545 <5 330 0.177(Si)
0.0035(C)
ZrN None 1960 7.24 10 0.185(Zr)
1.9(N)
ZrC None 3540 7.01 12 0.185(Zr)
0.0035(Cr)
Cr None 1907 4.9 93.9 3.05(Cr)
Si None 1414 2.6 149 0.177(Si)
Zr HCP(863)/BCC 1850 7.2 22 0.185(Zr)
behavior of SiC has been shown in a water environment at 360 C [17]. Because it
has been reported that several pores are formed through the formation of CO2 gas
during the oxidation of the ZrC phase at 600 C [18], the stability of the ZrC phase
cannot be guaranteed when it is applied as a coating material for zirconium alloy.
Thus, pure metal Si was selected as a coating material in our initial work. It can be
anticipated that Si will change to the SiO2 phase (quartz) during oxidation; there-
fore, a SiO2 layer can be made using pure Si coating on a zirconium matrix. How-
ever, because the thermal expansion coefficients of Si (2.6) and Zr (7.2) are very
different, as shown in Table 1, the increased adhesion between the two materials is a
key point in the coating technology. This technical problem can be solved by the
formation of a diffusion zone between the substrate and the coating material using
LBS.
HIGH-TEMPERATURE OXIDATION BEHAVIOR OF COATING MATERIALS

AND SI-COATED SAMPLES
From the coating material test in a high-temperature steam environment at
1200 C, as shown in Fig. 6(a), it was revealed that SiO2 showed the greatest
oxidation resistance among the tested materials, and that Si was more effective than
Cr in terms of oxidation resistance. Regarding the potential for oxidation resistance
under a high-temperature steam condition, three of the tested materials can be con-
sidered as coating materials to prevent the severe oxidation of zirconium-based
alloys, as these three materials showed greater oxidation resistance than the
Zircaloy-4 sheet.
After Si coating using the PS and PS þ LBS methods, the oxidation behaviors
were evaluated based on cross-sectional microstructure observations of the Si-
coated layer, as shown in Figs. 7 through 9. It is known that the oxidation resist-
ance of the Si-coated layer is somewhat improved relative to that of the Zircaloy-
4 sheet because the severe oxidation reaction was suppressed in the layer coated
via multiple PS applications. However, the adhesion property of the PS Si-coated
layer was considerably decreased during the high-temperature oxidation test at
1200 C, and severe oxidation progressed at the interface between the Si-coated
layer and the Zircaloy-4 substrate at the side of the tested samples. In addition,
the formation of a poreless layer is an important factor in Si coating because the
intermediate ZrO2 phase can grow along the pore network formed in a PS Si-
coated layer. If a poreless Si coating covers all surfaces of a sample, the formation
of the intermediate ZrO2 phase should be suppressed. However, the adhesion
property of the PS method cannot be guaranteed because of the large difference
between the thermal expansion coefficients of Si and Zr. In order to improve the
bonding strength and remove the pores formed in the PS Si-coated layer, an
alloying bond layer between the Si and the Zr can be formed and the pores can
be fully removed by applying LBS treatment to the Si-coated Zircaloy-4 samples.
However, it was observed that the pure Si layer was changed to an Si–Zr mixed
layer by LBS, contrary to our intention. From cross-sectional observations of the
Si–Zr mixed layer after the oxidation test, it is known that the oxidation resist-
ance of the Si–Zr mixed layer was superior to that of Zircaloy-4, and the adhesion
property of the PS Si-coated layer was considerably increased by the additional
LBS treatment.
FUTURE PLAN
As a result of this study, the hydrogen generation of zirconium alloy caused by an
excess oxidation reaction in a high-temperature steam environment can be consid-
erably reduced by the application of Si coating via PS þ LBS treatments. However,
the coating thickness for the fuel cladding will have to be determined after various
additional tests. In addition, the possibility of other coating materials such as Cr
will be studied, because weight loss of Si under normal reactor conditions would be
anticipated based on the results of the SiC dissolution phenomenon in water for a
360 C test [17].
If this technology for coated cladding shows promise for nuclear fuel cladding,
irradiation effects such as the neutron cross-section across the whole spectrum of
energy and the variation in material properties induced by irradiation will be
KIM ET AL., DOI 10.1520/STP154320120161 363
considered for the candidate coating material and coating methods. In order to
apply the accident tolerant fuel, the coated cladding will be tested in the in-pile con-
dition, as well as in a severe oxidation test such as a steam condition and in mixed
conditions between steam and air at temperatures higher than 1200 C. In addition,
the surface-modified cladding will be evaluated for various items such as bursting
and ballooning and oxidation of the inner surface of the cladding to simulate the
integrated physical properties during loss-of-coolant accidents.
Summary
Coating methods and coating materials that could be used to reduce the oxidation
rate of zirconium-based alloy in a high-temperature steam environment were stud-
ied. Two technologies, plasma spray (PS) and laser beam scanning (LBS) after a PS,
were selected as the coating methods, and Si was selected as a coating layer for the
surface coating material on zirconium-based alloy.
Regarding the coating materials tested at a 1200 C in steam environment, the
SiO2 showed the greatest oxidation resistance among the tested materials, and Si
was more effective than Cr in terms of oxidation resistance.
The PS coating technique has problems, such as the formation of pores in the
Si-coated layer and interface oxidation caused by low adhesion during the high-
temperature oxidation test at 1200 C. To remove the problems associated with the
PS coating method, the LBS treatment was applied to the PS-coated surface.
Although the pure Si-coated layer was changed to a Si–Zr mixed layer, the pores in
the Si-coated layer were successfully removed, and the interface oxidation was sup-
pressed by the formation of diffusion bonding between the Zircaloy-4 substrate and
the Si–Zr mixed layer.
Based on the cross-sectional observation of Si coating applied via PS þ LBS
treatment after the oxidation test, the oxidation resistance of the Si–Zr mixed layer
was superior to that of Zircaloy-4, and good adhesion was obtained. Thus, the
hydrogen generation of zirconium alloy undergoing excess oxidation in a high-
temperature steam environment can be considerably reduced by the application of
Si coating via PS þ LBS technology.
ACKNOWLEDGMENTS
This work was supported by a National Research Foundation of Korea (NRF) grant
funded by the Korean government (MSIP) (Grant No. 2013M2A8A5000702).
References
[1] Garde, A. M., Comstok, R. J., Pan, G., Baranwal, R., Hallstadius, L., Cook, T., and Carrera,
F., “Advanced Zirconium Alloy for PWR Application,” J. ASTM Int., Vol. 7, No. 9, 2011, pp.
784–799.
[2] Chabretou, V., Hoffmann, P. B., Trapp-Pritsching, S., Garner, G., Barberis, P., Rebeyrolle,
V., and Vermoyal, J. J., “Ultra Low Tin Quaternary Alloys PWR Performance—Impact of
Tin Content on Corrosion Resistance, Irradiation Growth, and Mechanical Properties,” J.
ASTM Int., Vol. 8, No. 5, 2011, pp. 801–826.
[3] Kim, H. G., Choi, B. K., Park, S. Y., Jung, Y. I., Park, D. J., and Park, J. Y., “Post-irradiation
Examination of HANA Claddings after Research Reactor Test up to 34 GWD/MTU,” J.
Nucl. Mater., Vol. 426, Nos. 1–3, 2012, pp. 173–181.
[4] Baker, L. and Just, L. C., “Studies of Metal-Water Reactions at High Temperatures. III. Ex-
perimental Studies of the Zirconium-Water Reaction,” Report No. ANL-6548, Argonne
National Laboratory, Argonne, IL, 1962.
[5] Cathcart, J. V., Pawel, R. E., McKee, R. A., Druschel, R. E., Yurek, G. J., Campbell, J. J., and
Jury, S. H., “Zirconium Metal-Water Oxidation Kinetics. IV. Reaction Rate Studies,”
Report No. ORNL/NUREG-17, Oak Ridge National Laboratory, Oak Ridge, TN, 1977.
[6] Kim, H. G., Kim, I. H., Jung, Y. I., Park, J. Y., and Jeong, Y. H., “Properties of Zr Alloy Clad-
ding after Simulated LOCA Oxidation and Water Quenching,” Nuclear Engineering and
Technology, Vol. 42, No. 2, 2010, pp. 193–202.
[7] Baek, J. H. and Jeong, Y. H., “Breakaway Phenomenon of Zr-based Alloys during a High-
temperature Oxidation,” J. Nucl. Mater., Vol. 372, Nos. 2–3, 2008, pp. 152–159.
[8] Hallstadious, L., Johnson, S., and Lahoda, E., “Cladding for High Performance Fuel,”
Prog. Nucl. Energy, Vol. 57, 2012, pp. 71–76.
[9] Cheng, B., “Fuel Behavior in Service Accidents and Mo-alloy-based Cladding Designs to
Improve Accident Tolerance,” TopFuel 2012, Manchester, UK, Sept 2–6, European Nuclear
Society, 2012.
[10] Sidhu, B. S. and Prakash, S., “Evaluation of the Behavior of Plasma-sprayed Ni3Al Coat-
ings on Steel in Oxidation and Molten Salt Environments at 900 C,” Surf. Coat. Technol.,
Vol. 116, No. 1, 2003, pp. 89–100.
[11] Kamal, S., Jayaganthan, R., and Parkash, S., “High Temperature Oxidation Studies of Det-
onation-gun-sprayed Cr3C2–NiCr Coating on Fe- and Ni-based Superalloys in Air under
Cyclic Condition at 900 C,” J. Alloys Compd., Vol. 472, 2009, pp. 378–398.
[12] Matthews, S., James, B., and Hyland, M., “The Role of Microstructure in the High Temper-
ature Oxidation Mechanism of Cr3C2–NiCr Composite Coatings,” Corros. Sci., Vol. 51, No.
5, 2009, pp. 1172–1180.
[13] Lusquiños, F., Pou, J., Quintero, F., and Pérez-Amor, M., “Laser Cladding of SiC/Si Com-
posite Coating on Si-SiC Ceramic Substrates,” Surf. Coat. Technol., Vol. 202, 2008, pp.
1588–1593.
[14] Anandkumar, R., Almeida, A., Colaço, R., Vilar, R., Ocelik, V., and De Hosson, J. Th. M.,
“Microstructure and Wear Studies of Laser Clad Al-Si/SiC(p) Composite Coatings,” Surf.
Coat. Technol., Vol. 201, No. 24, 2007, pp. 9497–9505.
[15] Venkataraman, R., Das, G., Singh, S. R., Pathak, L. C., Ghosh, R. N., Venkataraman, B., and
Krishnamurthy, R., “Study on Influence of Porosity, Pore Size, Spatial and Topological
Distribution of Pores on Microhardness of as Plasma Sprayed Ceramic Coatings,” Mater.
Sci. Eng., A, Vols. 445–446, 2007, pp. 269–274.
KIM ET AL., DOI 10.1520/STP154320120161 365
[16] Ibrahim, A., Salem, H., and Sedky, S., “Excimer Laser Surface Treatment of Plasma
Sprayed Alumina-13 % Titania Coatings,” Surf. Coat. Technol., Vol. 203, No. 23, 2009, pp.
3579–3589.
[17] Kim, W. J., Hwang, H. S., Park, J. Y., and Ryu, W. S., “Corrosion Behaviors of Sintered and
Chemically Vapor Deposited Silicon Carbide Ceramics in Water at 360 C,” J. Mater. Sci.
Lett., Vol. 22, No. 8, 2003, pp. 581–584.
[18] Shimada, S., “Interfacial Reaction in Oxidation of Carbides with Formation of Carbon,”
Solid State Ionics, Vols. 141–142, 2001, pp. 99–104.
DISCUSSION
Questions from K. Chaube, N.F.C. Hyderabad
Q1:—How does coating of Zr-Si affect the weld qualities during the making of
the fuel bundle?
Authors’ Response:—The coating layer of the Zr-Si phase might be affected by

the welding process for making the fuel rods. To decrease this problem, we feel that
the coating of the Si element for the welded area can be re-tried at the tube end cap
welded region. We already started the improved coating method, which supplies Si
powder during LBS, for direct application to the welded region without a PS
process.
Q2:—What is the stability of this layer during in-pile operations?
Authors’ Response:—The stability of this layer in the in-pile is unknown,

although the selection of the coating material was based on the neutron cross sec-
tion. Thus, we have a plan for an irradiation test of coated samples in the research
reactor.
Questions from Suresh Yagnik, EPRI
Q1:—What is the reason behind differences in oxidation behaviors of PS and

PS+LBS coatings? Does it have to do with changes in permeability of the coating
due to melting of the PS’ed Si?
Authors’ Response:—The major difference in the oxidation behaviors between

the PS and PS+LBS is the formation of an intermediate oxide layer between the
coated layer and the matrix. In the case of the PS coating of Si, an intermediate oxide
layer was formed during the oxidation test, whereas, such an oxide layer was sup-
pressed when the coated layer was prepared by PS+LBS treatments. The changes in
the coated layer are the removal of pores in the coated layer and the formation of a
Zr-Si mixed layer in the interface by additional LBS treatment after the PS coating.
Q2:—The weight gain plot shows very little difference in the corrosion behavior
of Zircaloy-4 with no coating, with PS, and with PS+LBS. Which coating is being
pursued for further work and why?
Authors’ Response:—The slight difference in the corrosion behavior of the three

kinds of samples is related to the test sample preparation because the coated layer
was only on one side for the sheet samples. Thus, we showed the cross-sectional ob-
servation for the oxide layer after the oxidation test. We identified that the oxida-
tion behavior was improved by the PS+LBS treatment of the Si element. Thus,
KIM ET AL., DOI 10.1520/STP154320120161 367
further work is initially focused on the PS+LBS treatment, and an improved coating
method, which is the LBS supplied with Si powder, will be developed.
Question from Jean-Christophe Brachet, CEA:—Do you have any idea regarding
the behavior of Zr-Si intermetallic compounds under neutron irradiation?
Authors’ Response:—We have no idea what to expect of the irradiation effect of

Zr-Si intermetallic compounds. We have a plan to conduct an irradiation test of
our coated samples in a research reactor.
Question from Mirco Grosse, Karlsruhe Inst. Of Technology:—If the severe acci-
dent starts with LOCA or cracks through the cladding tube wall occurs, steam pene-
trates into the cladding tube and inner oxidation occurs. Is it possible to produce a
coating on the inner surface?
Authors’ Response:—We recognized this point because the inner-side oxidation

of the cladding can be expected after the cladding balloons and ruptures. However,
the inner-side coating method of the full scale cladding has not been prepared at
the present time.
Questions from Arthur Motta, Penn State University, atm2@psu.edu
Q1:—How uniform and free of defects is your deposited coating layer?
Authors’ Response:—From the observation and analysis of the coated layer using
SEM-EDS, the coated layer entirely showed a uniform composition at the outer de-
posited part, and was defect free after the LBS treatment for the Si-coated layer by PS.
Of course, the composition of the Zr-Si mixed layer at the interface between the
coa:—ted layer and matrix is different when compared to the outer deposited part.
Q2:—Have you subjected your coated samples to mechanical stability tests,

such as bending or scratching?
Authors’ Response:—Mechanical stability tests such as bending or scratching

have not been performed at the present time; however, we have a plan to do such
tests. We confirmed the adhesion stability of the coated layer using a thermal shock
test. The heating and cooling rates were 50oC/s to 1000oC/s, and the tests were
repeated 10 times during the thermal shock test. From this, it was confirmed that
the coated layer showed a good adhesion property by the PS+LBS treatment,
whereas a peeling off phenomenon was observed when the coated layer was depos-
ited by PS.
Question from Pierre Barberis, CEZUS Research Center:—For choosing Cr and

Si as metallic coating, you considered the high melting points of the Zr-Cr or Zr-Si
intermetallic compounds. However, there are also deep eutectics which would lead
to liquid phases in the case of LOCA. Could you comment?
Authors’ Response:—This is a good question. Our target of a coated material phase

is Zr-Cr or Zr-Si intermetallic compounds which have a high melting temperature.
However, the composition of deep eutectics can be formed especially at the interface
region between the coated layer and the matrix. We considered two points regarding
the high-temperature property of the cladding. The first is that the maximum cladding
temperature is 1200 C in the LOCA criteria, whereas, the temperature of the deep
eutectics of Zr-Cr or Zr-Si intermetallic compounds is higher than 1200 C. The second
is that the cladding temperature can be increased by the severe oxidation at the outer
surface, which can be mentioned as an exothermic reaction during the zirconium alloy
oxidation, as well as a decrease of the thermal conductivity from the formation of the
thick oxide by a high oxidation rate of the zirconium base alloy. Thus, we thought that
high-temperature stability of the Zr-Cr or Zr-Si intermetallic phase coated on the zirco-
nium alloy cladding is more improved than for a commercial zirconium alloy cladding.
Question from R. N. Jayaraj, Dept. of Atomic Energy, India:—What is the effect

of coatings on welding of end-caps of fuel rod?
Authors’ Response:—We did not test the welding point of the cladding. How-
ever, we considered re-trying the coating of the Si element for the welded area at
the tube end cap welded region.
Questions from Damien Kaczorowski, AREVA
Q1:—Have you tested (e.g., corrosion and creep) coated rods under normal
conditions?
Authors’ Response:—No, we did not test under normal conditions for the corro-
sion and creep. We have a plan for a normal operation condition after the optimiza-
tion of the coating materials and methods.
Q2:—What is the thickness of the coating?
Authors’ Response:—At the present time, the coated layer thickness of the tested
samples ranges from 15 to 800 lm.
Q3:—Are the residual stresses in the coatings tensile or compression?
Authors’ Response:—We did not check the residual stress in the coated layer.
Question from Antoine Ambard, EDF:—Basically the coated alloy still oxidizes.
In that case, the relative value could be the additional time gained before the
KIM ET AL., DOI 10.1520/STP154320120161 369
specimen is totally oxidized compared to the non-coated specimen. Your tests are
of short duration (2000 seconds). Do you have an idea of this potential gain in
time? Is it hours, days or weeks? Do you plan to perform such tests?
Authors’ Response:—We recognized that the test time is very short. Thus, our
test duration will be extended up to 7000 s at 1200 C using the TGA equipment.
However, further extension is limited because of the TGA equipment specifications.
Question from N. Ramasubramanian, ECCATEC Inc. Canada:—During laser

beam scanning, does the plasma-sprayed silicon/chromium melt and become ho-
mogenized? Does this change the Zircaloy microstructure to a depth at the coat-
ing-substrate interface?
Authors’ Response:—We repeated the sample test to obtain the homogenized

coated layer of Si and Cr using the LBS treatment. Thus, both the composition and
microstructure of the homogenized layer can be obtained by the control of the laser
power, particle size, scan speed, and inert gas flow.
The Zircaloy microstructure at the coated-substrate interface was changed

from recrystallized grains into a Widmanstätten structure by a laser heat source.
The changed phase thickness can be reduced by the control of both the laser beam
scanning conditions and the cooling method of the substrate.
CORROSION
AND HYDROGEN PICKUP
Pia Tejland,1 Hans-Olof Andrén,2 Gustav Sundell,2

Mattias Thuvander,2 Bertil Josefsson,3 Lars Hallstadius,4
Maria Ivermark,4 and Mats Dahlbäck4
Oxidation Mechanism in
Zircaloy-2—The Effect of
SPP Size Distribution
Reference
Tejland, Pia, Andrén, Hans-Olof, Sundell, Gustav, Thuvander, Mattias, Josefsson, Bertil,
Hallstadius, Lars, Ivermark, Maria, and Dahlbäck, Mats, “Oxidation Mechanism in Zircaloy-2—
The Effect of SPP Size Distribution,” Zirconium in the Nuclear Industry: 17th International
ABSTRACT
The metal/oxide interface region in Zircaloy-2 oxidized in autoclave was studied
with transmission electron microscopy (TEM) and atom probe tomography. In
addition to waviness on the micrometer scale the metal/oxide interface was
found to have irregularities on a finer scale, and metal islands were found
especially at metal hills (delayed parts of the oxidation front). The thickness of
the sub-oxide layer varies considerably along the interface in the same sample,
from 100 to virtually 0 nm. The sub-oxide composition may vary on a very fine
scale (down to 5 nm), and it can sometimes be a mixture of sub-oxides with
different oxygen content. The metal matrix in contact with the sub-oxide is
saturated with up to 32 at. % oxygen, and the oxygen diffusion profile in the
metal is in approximate agreement with literature data for pure Zr. However, the
diffusion length appears to be somewhat larger at interface metal hills than
under valleys, probably for both geometrical and stress state reasons. Hydride
Manuscript received March 27, 2013; accepted for publication October 1, 2013; published online September
19, 2014.
1
Studsvik Nuclear AB, SE-611 82 Nyköping, Sweden (Corresponding author), e-mail: pia.tejland@studsvik.se
2
Chalmers Univ. of Technology, Dept. of Applied Physics, SE-412 96 Göteborg, Sweden.
3
Vattenfall Nuclear Fuel AB, SE-169 92 Stockholm, Sweden.
4
Westinghouse Electric Sweden, SE-721 63 Västerås, Sweden.
5
Hyderabad, India.
Copyright V
precipitates, hardly visible in conventional TEM, give a good image contrast

when employing high angle annular dark field imaging. A model for the
oxidation process is presented, where the creep deformation of the metal close
to the interface and the formation of lateral cracks in the oxide are of highest
importance. The effect of second phase particle (SPP) size is suggested to be
twofold: Small and numerous SPPs give a stronger metal and therefore higher
stress in the oxide. Small SPPs also nucleate many more lateral cracks in the
oxide, which gives a weaker oxide. Together this leads to formation of large
cracks associated with transition in the oxidation rate at an earlier time than for a
material with larger and fewer SPPs, and thereby a higher oxidation rate.
Keywords
Zircaloy-2, oxidation, SPP, TEM, APT
Introduction
Extensive research has been carried out over the years concerning the oxidation of
zirconium alloys used in nuclear reactors. However, although many pieces of the
puzzle have been found, no complete mechanism has yet been presented explaining
the entire oxidation process. One parameter that early on was found to be impor-
tant for the oxidation process was the size distribution of second phase particles
(SPPs) [1–4]. However, although the optimal size distribution for a good corrosion
resistance has been found, the mechanisms behind the SPP influence have not yet
been determined.
In an attempt to understand and explain these mechanisms, microstructural
investigations have previously been performed on two Zircaloy-2 materials, with
the main difference being their SPP size distribution. The mean SPP diameter of
the two Zircaloy-2 materials investigated in this study, material A and B, was 22 nm
and 84 nm, respectively, and the oxidation behavior of the two materials in both
steam autoclave and boiling water reactor (BWR) conditions are shown in Fig. 1. It
is apparent that material B, containing larger but fewer SPPs, shows the better uni-
form corrosion resistance. The results from the previous microstructural investiga-
tions are presented in this section.
It is a well-known fact that the metal/oxide interface in zirconium alloys is
undulating. These undulations were further investigated, and the main results were:
1. The interface is undulating on a micrometer scale. There is no difference in
the undulations of the interface of the metal and a 1 -lm-thick oxide scale in
materials A and B; they both show a periodicity of around 1 lm and an ampli-
tude of around 100 nm, see Fig. 2(a) [5–7].
2. There is a difference in amplitude of the metal/oxide interface undulations in
materials with different oxide thicknesses. It seems from Fig. 2 that the ampli-
tude of the interface undulations is larger in the sample with 2 lm oxide than
in the 1 -lm-oxide sample. However, the sample with a 9 -lm-oxide scale does
not show significantly larger amplitude than the two samples with thinner
TEJLAND ET AL., DOI 10.1520/STP154320130052 375
FIG. 1 Oxidation behavior of material A, material B, and ZIRLO during static steam
autoclave corrosion testing (top) [5] and of material A and material B in BWR
(bottom) [6].
oxide scales. This suggests that the undulation amplitude does not continue to
increase monotonically during oxidation, but rather increases and decreases in
a cyclic manner [5]. It has been reported in the literature that the amplitude
increases up until transition, where it decreases drastically and the process
starts over again until the next transition [8,9].
3. The combination of an undulating interface and the volume expansion con-
nected to the oxidation (Pilling-Bedworth ratio is 1.55 [10], with 54 % expan-
sion in the radial direction of the tube and 0.5 % each in the axial and
tangential directions [11]) gives rise to large stresses in the very vicinity of the
FIG. 2 Metal/oxide interface undulations in material B under (a) 1 lm, (b) 2 lm, and (c)
9 lm oxide thickness. The inversed contrast in the images comes from using
SEM with an annular dark field (ADF) detector in (a), and TEM with a high angle
annular dark field (HAADF) detector in (b) and (c), both giving atomic number
contrast but using different energies [5]. The acicular-shaped precipitates seen
in the metal in (a) are the hydrides treated in the text.
metal/oxide interface. A simple diagram showing the expansion of the volume

of metal above a metal valley is shown in Fig. 3. The volume of metal between
A and B is oxidized and has therefore expanded 54 % until C. At the metal hill
(D), no oxidation has yet occurred. If the metal could deform under a negligi-
ble stress (like a liquid), no stress perpendicular to the interface would be cre-
ated. However, the metal requires a high stress to deform, and large
compressive stresses perpendicular to the interface will be created at the val-
leys both in oxide and metal. At the hills large tensile stresses will be created
perpendicular to the interface, locally above 800 MPa in the oxide as calculated
by Parise et al. [11]. Parallel to the interface, compressive stresses exist in the
oxide and tensile stresses in the metal because of the much smaller expansion
(0.5 %) along the interface.
Because the most prominent difference between materials A and B is the num-
ber of lateral cracks in the oxide, see Fig. 4, these cracks have been thoroughly stud-
ied. In Ref 7, a mechanism for how SPPs act as nucleation sites for cracks is
presented, based on the voids being created above each SPP because of the absence
of volume increase caused by the delayed oxidation of SPPs compared to the metal
matrix, see Fig. 5. In Ref 5, more investigations were performed concerning the lat-
eral cracks, and a mechanism for crack formation was presented as well as a corre-
lation between SPP size distribution and crack formation.
FIG. 3 Diagrams showing (a) the expansion of the volume of metal above a metal
valley, and (b) the stresses created as a result of the volume change [5]. Fig. 3(b)
redrawn after [11].
The oxygen ingress into the metal has been investigated, and results from both
transmission electron microscopy (TEM) and atom probe tomography (APT) show
the presence of a sub-oxide with an oxygen content of 45–55 at. %, see Figs. 6 and 7.
In some specimens also a step to an oxygen level of 30 at. % was found, suggesting
that there is a region present with a composition close to the oxygen solubility level
in zirconium [7].
A plastically deformed layer in the metal underneath the oxide scale was found,
manifested by areas of dislocation tangles and patches, sub-grain formation, and
sometimes twinning, see Fig. 8. It has been shown that this deformation is an
effect of the oxidation process, during which large stresses are introduced in the
material [12].
In this paper, previously reported work will be combined with new results to
form a more cohesive view of the oxidation process of Zircaloy-2 and the influence
FIG. 4 Bright field TEM images of (a) material A, and (b) material B, both with 1 lm
oxide, showing the difference in crack density [5].
FIG. 5 Schematic diagram of void and crack formation at SPPs. When the metal layer
between the full and dashed line in (a) oxidizes, it expands perpendicular to the
interface and forms the layer between the dashed and full line in (b). Because
the SPPs are not oxidized, a void (left) is formed or, if the oxide first adheres to
the SPP, a crack in the oxide and a smaller void form (right) [7].
FIG. 6 Zr and O TEM EDX line scan showing an indication of a 100-nm-thick layer of
sub-oxide at the metal/oxide interface of material B with an oxide thickness of
1 lm [7].
FIG. 7 APT concentration profile through the metal/oxide interface in material B with
an oxide thickness of 1 lm. Note the presence of a sub-oxide and the sharp step
in oxygen (lower curve) content to the max solubility in the matrix, around 30 at.
%. Proximity histogram with origin at the interface (iso-concentration surface
with 40 at. % non-molecular Zr ions) [7].
FIG. 8 Material B with 1 -lm-thick oxide. Some of the areas with dislocation patches are
encircled by white lines [12].
of the SPPs on the same. A suggestion for a transition mechanism will be presented,
as well as a model for the whole oxidation process, including a connection between
SPP size distribution and oxidation behavior.
Experimental Details
MATERIALS
Zircaloy-2 is a material that is and has been extensively used in boiling water reac-
tors (BWRs). In this study, two different commercial Zircaloy-2 alloys were chosen,
for which there exists long-term in-pile data. The two materials have approximately
the same chemical composition, as can be seen in Table 1. The somewhat higher iron
TABLE 1 Chemical composition (provided by the manufacturer, Sandvik AB) and average SPP size
(determined in [5]) of the two studied Zircaloy-2 materials.
Material A Material B
Chemical composition
Sn (%) 1.46 1.32
Fe (%) 0.12 0.17
Cr (%) 0.10 0.10
Ni (%) 0.05 0.05
Si (ppm) 80 70
O (ppm) 1240 1300
N (ppm) 40 50
SPP size
Average SPP diameter (nm) 22 84
TABLE 2 Mechanical properties from tensile testing at 300 C of material BA and material B, where
material BA has the chemical composition of material B but the heat treatment (and
thereby SPP size distribution) of material A.
Mechanical Properties Material BA Material B
Yield stress (MPa) 215 180

Fracture stress (MPa) 351 320
Note: Data provided by the manufacturer, Sandvik AB.
content in material B gives only a slightly higher volume fraction of SPPs, 1.6 vol. %
against 1.3 vol. % in material A [5]. They do, however, exhibit substantially different
oxidation behavior, both in-pile and in-steam autoclave, see Fig. 1. This difference is
mainly associated with their SPP size distribution. Because of different final heat
treatments, material A contains smaller but more numerous SPPs, while material B
contains larger but fewer SPPs. The average SPP diameters can be found in Table 1.
The large difference in SPP size gives a huge difference in SPP number density,
560 lm3 in material A and 39 lm3 in material B [5]. Both materials were in the
form of recrystallized cladding tubes. For more detailed information about the stud-
ied materials, see Refs 5 and 7. The materials were thus chosen to isolate the influ-
ence of a difference in SPP size distribution on the oxidation process of Zircaloy-2.
As discussed in Ref 12, the difference in SPP size distribution also affects the
mechanical properties of the materials, through particle hardening. Displayed in
Table 2 are results from tensile testing performed at 300 C of material B and mate-
rial BA, where the latter is a material with the same chemical composition as mate-
rial B, but with the final heat treatment (and thereby SPP size distribution) of
material A. Both materials were fully recrystallized. Because the only difference
between material BA and material B is the heat treatment, the results clearly show
the difference in mechanical properties as a function of SPP size distribution, with
30 MPa higher values of yield and fracture stress for material BA.
Another contribution to the hardening of the materials comes from solid solu-
tion hardening by tin. The tin content of material A is 0.14 % higher than that of
material B, but as was concluded in Ref 12, this difference only gives a very small
additional contribution to the hardening. The main reason for the higher strength
of material A thus comes from having more SPPs.
As described below, during oxidation of zirconium alloys, large local stresses
will develop in the metal that will deform by creep. This means that the creep
strength of the materials is of importance. Because temperatures and stresses are
such that dislocation creep occurs, the creep strength is given mainly by obstacles
to (thermally activated) dislocation glide and climb. The yield stress at high temper-
ature is a measure of the resistance to glide only, but the resistance to climb usually
scales well with the yield stress because both are dominated by obstacles to
TABLE 3 Autoclave data and the resulting weight gains and corresponding oxide thicknesses (cal-
culated assuming that 15 mg/dm2 corresponds to 1 lm).
Alloy Temp ( C) Time (Days) Weight Gain Oxide Thickness

(mg/dm2) (lm)
Material A 400 3 18.9 1.3

400 6 33.9 2.3
415 30 132.5 8.8
Material B 400 3 15.0 1.0
400 15 27.5 1.8
415 150 135 9.0
dislocation movement, i.e., SPPs and grain boundaries. Thus, even if creep strength
is lower than high-temperature yield stress, the relative creep strength of a fully
recrystallized material can be judged from its high-temperature yield stress.
The materials were oxidized in static steam autoclaves at temperatures of
400 C and 415 C (where the higher temperature was used for obtaining thick
oxides within a reasonable time frame), a pressure of 10.3 MPa and exposure
times ranging from 3 to 150 days to obtain different oxide thicknesses, see
Table 3. From this table, it is apparent that material B has the better oxidation
behavior. This could in principle be because of the slightly lower tin content
of material B, because it is well known that a lower tin content improves
corrosion resistance. However, because this improvement is not seen for thin,
pre-transition oxides [13] and we see a large difference in corrosion resistance al-
ready for 1 - and 2 -lm-thick oxides, it is reasonable to ascribe the difference in
corrosion resistance to the very large difference in SPP size between materials A
and B. Besides the autoclaved materials, with oxide thicknesses of roughly 1, 2,
and 9 lm, also un-oxidized material has been available for observation. In the ta-
ble, all available materials are presented, with the materials appearing in figures
in this work highlighted in bold.
TEM
TEM specimens of the metal/oxide interface were prepared with the focused ion
beam (FIB) in situ lift-out technique using an FEI Strata 235 Dual Beam system; for
details, see Ref 7.
The TEM work was carried out on three different instruments: a Philips CM
200 field emission gun (FEG) TEM, an FEI Tecnai T20 with a LaB6 crystal electron
source and an EDAX energy-dispersive X-ray spectroscopy (EDX) system, a high-
angle annular dark field (HAADF) detector, and an FEI Titan 80-300 FEG TEM
equipped with a monochromator and a Gatan image filter (GIF) for electron
energy-loss spectroscopy (EELS) analysis.
APT
APT specimens were prepared using the FIB-SEM; for details, see Ref 7. The APT
analyses were performed with an Imago LEAP 3000X HR instrument in laser
mode, using a specimen set temperature of 70–80 K, laser pulse frequency of 200
kHz, and laser pulse energy of 0.5 nJ. The APT data were evaluated with the Imago
IVAS 3.4.3 software.
For presentation of the data, so-called heat maps were used. These were con-
structed as follows: The dataset was divided into cubes 1 nm3 in size, and the oxy-
gen content in each cube was calculated. A slice, a few nm in width, was then made
through the dataset and the average concentration in a column of cubes perpendic-
ular to the slice was projected onto one surface of the slice. The resulting 1 nm2
pixel pattern was then filtered to get a smooth concentration variation, and finally
the results were color coded with red corresponding to the highest and blue to the
lowest oxygen concentration, according to the color scales inserted in the figures.
Results
INTERFACE MORPHOLOGY
The waviness of the metal/oxide interface has been reported elsewhere [5,7,14–17].
However, TEM images using HAADF contrast reveal that the interface region is
not only wavy on a micrometer scale, but also has an even more complex morphol-
ogy. In Fig. 9, the metal/oxide interface underneath a 1 -lm-thick oxide scale in ma-
terial B is shown. It is apparent that besides the undulations, there is also a mixture
of oxidized and un-oxidized parts, especially above the metal hills. This appearance
is even more pronounced underneath thicker oxide scales, as can be seen in Fig. 10,
where the metal/oxide interface underneath a 9 -lm-thick oxide scale in material B
is shown. The same complex interface structure can be seen in the APT heat maps.
One example is shown in Fig. 11, where the interface region of material B with a
1-lm-oxide scale is displayed.
FIG. 9 HAADF TEM image showing the complex morphology of the metal/oxide
interface of material B with a 1 -lm-thick oxide scale.
FIG. 10 HAADF TEM image showing the complex morphology of the metal/oxide
interface of material B with a 9 -lm-thick oxide scale. This figure shows part of
Fig. 2(c) in higher magnification and therefore better resolution.
SUB-OXIDE LAYER
APT has been used in this study for analyzing the sub-oxide to gain a better quanti-
fication of the oxygen levels than can be obtained from TEM studies. However,
because APT is a very local technique and the metal/oxide interface has a very com-
plex morphology, information gained from only one or a few specimens is not
enough to completely determine the nature of the sub-oxide layer. For this reason,
15 APT specimens from the same autoclave sample (material B with 1 -lm-thick
oxide) were studied, although only a few representative images are shown here.
In Fig. 12, an oxygen heat map of the interface region in one of the specimens is
shown, demonstrating that no sub-oxide is present, but instead there is a sharp step
from the oxygen level in the main oxide of >60 at. % down to below 32 at. %. In
FIG. 11 APT oxygen heat map showing the complex morphology of the metal/oxide
interface of material B with a 1 -lm-thick oxide scale.
FIG. 12 APT oxygen heat map of the metal/oxide interface region of material B with a 1 -
lm-thick oxide scale. No sub-oxide is present in this specimen.
Fig. 13, two 2D slices of another specimen are shown, demonstrating variations in
sub-oxide layer thickness even within the same specimen, as the layer thickness
varies between 20 and 100 nm before there is again a sharp step down to below 32
at. % oxygen. In general, the results show that there are large variations in the thick-
ness and morphology of the sub-oxide layer also within the same material with the
same oxide thickness. It is, however, not possible to relate the variations of sub-
FIG. 13 APT oxygen heat maps of two orthogonal 2D slices of one APT specimen of
material B with a 1 -lm-thick oxide scale, showing that within the same
specimen the thickness of the sub-oxide layer varies between 20 and 100 nm.
FIG. 14 APT 3D iso-concentration surface representation of the sub-oxide region in

material B with a 1 -lm-thick oxide scale. The iso-concentration surface
separates volumes with a higher oxygen concentration than 57 at. % (right)
from volumes with a lower oxygen concentration (left).
oxide layer thickness to a specific feature in the metal/oxide interface, such as a

metal hill or valley.
Not only does the interface region consist of mixtures of oxidized and un-
oxidized parts as reported in previously, but also within the sub-oxide there is
sometimes a mixture of sub-oxide volumes with different oxygen content. This
kind of mixed region is displayed in Figs. 14 and 15. In Fig. 14, an oxygen iso-
concentration surface of 57 at. % has been applied, enclosing all regions with a
higher oxygen level than this. In Fig. 15, a 2D slice of the specimen has been cut out
to show variations in oxygen content in a heat map. Both of these images (Figs. 14
and 15) show “fingers” of higher oxygen concentration (>57 at. %) penetrating into
a volume of lower oxygen concentration (<57 at. %). The fingers are around 5 nm
in diameter and perhaps 50 nm in length. In other words, there is not a clear inter-
face between the stoichiometric ZrO2 and the ZrO sub-oxide, but rather there is a
mixture of sub-oxide with 60 at. % and sub-oxide with 50 at. % close to the ox-
ide/sub-oxide interface. The average value in this region is approximately 55 at. %.
FIG. 15 APT oxygen heat map of material B with a 1 -lm-thick oxide scale showing
“fingers” of higher oxygen content penetrating into a volume of lower oxygen
content.
FIG. 16 TEM EDX line scans of specimen 1 from material B with a 1 -lm-thick oxide scale.
OXYGEN DIFFUSION INTO THE METAL

In an attempt to link the lm-scale waviness of the interface to oxygen diffusion, the
oxygen ingress into the metal in two different TEM specimens taken from the same
autoclave sample (material B with a 1 -lm-oxide scale) was investigated using EDX.
In both specimens (specimen 1 and specimen 2), two line profiles were made, one
at a metal hill and one at a metal valley.
The oxygen concentration profiles were evaluated assuming that they follow
the conventional error function solution to Fick’s second law of diffusion. Accord-
ing to this solution, one diffusion length (2(Dt)[1/2]) is equal to the distance from an
interface to a point in the profile where the oxygen content has decreased to 16 %
of the value at the interface. When evaluating the profiles it was assumed that the
maximum oxygen content in the metal in equilibrium with the oxide was 28.6 at. %
[18]. The oxygen content in the metal matrix is 0.7 at. % (1300 wt. ppm). Therefore,
the full diffusion profile ends not at zero but at 0.7 at. %, and the diffusion lengths
were evaluated between the locations along the profiles where the oxygen content
decreases from 28.6 to 5.2 at. % (0.16 (28.6 – 0.7) þ 0.7 ¼ 5.2).
The oxygen line profiles from specimen 1 (subscript H denotes metal hill and
V denotes metal valley) are displayed in Fig. 16, both showing an oxygen concentra-
tion gradient from an approximate value of 50 at. %. However, the oxygen levels in
the metal matrix are somewhat higher than expected, probably an artifact caused by
surface oxide on the thin foil specimens. Before evaluating the diffusion lengths, a
background oxygen signal equal to the approximate mean oxygen concentration in
the metal, 4 at. %, was subtracted. The measured diffusion length in profile 1 H was
130 nm and in 1 V 100 nm. Also, in profile 1 H, there is a hint of a step at an ox-
ygen level of 30 at. %.
FIG. 17 TEM EDX line scans of specimen 2 from material B with a 1 -lm-thick oxide
scale.
The oxygen line profiles from specimen 2 are displayed in Fig. 17. In these pro-
files, there is some problem with the EDX quantification, manifested by lower maxi-
mum oxygen levels than expected. However, it is still possible to study the shape
and lengths of the gradients. The diffusion length at the metal hill, 2 H, was meas-
ured to 220 nm in comparison to the much shorter diffusion length under the metal
valley, 2 V, which was measured to 100 nm. Both profiles show an unexpected
shape; the oxygen content decreases stepwise rather than resembling a classic error
function type diffusion profile.
From the EDX profiles, the conclusion can be drawn that there appears to be a
difference in oxygen profile depending on the location along the metal/oxide inter-
face, and that there might exist some steps in the oxygen profiles. However, it can
also be concluded that the EDX profiles are not satisfactory to confirm the presence
of a step in the profiles because of quantification difficulties, a too large step size
(25 nm), and limited spatial resolution (around 30 nm because of beam broadening
in the thin foil specimen). To gain more detailed information about the stepwise
diffusion profiles, APT was used.
The APT investigation shows that although the existence and thickness of the
sub-oxide layer varies substantially, the aforementioned step to <32 at. % oxygen
was present in all of the APT specimens, indicating that such a region is present
everywhere along the interface. Following this region is an oxygen gradient of vary-
ing width into the base metal.
HYDRIDE PRECIPITATION
The localized creep deformation of the metal underneath oxide scales has been
reported and explained elsewhere [12]. However, besides the features showing
FIG. 18 STEM images of small hydrides (needle shaped features) in the metal
underneath the oxide, in material B with a 1 -lm-thick oxide scale.
evidence of creep deformation (mainly areas with a high dislocation density), one
other type of feature is apparent in the microstructure of the metal. These features
are small needle-shaped structures, which are most clearly seen using HAADF con-
trast in scanning TEM (STEM) mode (see Fig. 2(a)), but, as can be seen in Fig. 18,
can also be imaged using bright field (BF) STEM. They are, however, more difficult
to distinguish using regular BF TEM. These features were investigated using EELS
and it was found that between the matrix and the needle shaped structures, there is
a shift in the plasmon peak in the EELS spectrum (see Fig. 19) indicating that the
needles contain hydrogen. The peak shift is around 2.1 eV, which indicates that
they are d-type hydrides [19].
Discussion
INTERFACE MORPHOLOGY
Many previous TEM studies of oxidized zirconium alloys have shown an undulat-
ing metal/oxide interface on a micrometer scale [5,7,14–17]. Parise et al. modeled
oxygen diffusion in monoclinic zirconia with 25-nm-wide columnar oxide grains,
assuming that diffusion was restricted to oxide grain boundaries and that the diffu-
sion coefficient varied randomly with a factor of up to 100 between individual grain
boundaries. They were able to explain the micrometer-sized waviness of the inter-
face as being caused by these large differences in oxygen grain boundary diffusion
rate [20].
In this study, an additional observation concerning the morphology of the
interface is added, as a lobed or mixed structure, especially frequent above the metal
FIG. 19 EELS spectrum from matrix and hydride, showing a shift in the plasmon peak of
2.1 eV.
hills is presented. This structure has been observed but not explained in previous
studies [21,22]. The recent observation by Sundell et al. of iron and nickel decorated
sub-grain boundaries in the matrix, extending into the oxide [23], suggests an addi-
tional route for oxygen diffusion in the oxide [24], adding to the uneven oxygen dif-
fusion through the oxide layer. The metal hills thus represent areas where most
oxide grain boundaries have a low diffusion coefficient for oxygen, leading to slower
oxygen ingress at the interface and slower oxidation of the metal. Islands or lobes of
metal embedded in the oxide can therefore exist for some time, before eventually
being oxidized. It should be noted that, in a separate paper, we have demonstrated
that lateral cracks form in the oxide close to metal hills. Because surface diffusion is
faster than grain boundary diffusion, we suggested that these (non-connected)
cracks act as short circuits for oxygen diffusion, enabling oxygen to move to grain
boundaries with a higher diffusion coefficient [5]. This mechanism should increase
the total flux of oxygen through the oxide and thus give a higher oxidation rate. It
should also decrease the areas of very slow oxygen ingress and therefore decrease
the size of metal islands, which otherwise would have been even larger and more
frequent above metal hills.
SUB-OXIDE LAYER
Below the bulk oxide of monoclinic zirconia (ZrO2) a sub-stoichiometric oxide has
been found close to the metal/oxide interface [21,25–33]. This sub-oxide reportedly
has an oxygen content of 45–55 at. %, i.e., ZrO. However, although various
researchers agree that this sub-oxide exists, there is no agreement on its crystal
structure, and it is not always clear how thick the layer is.
The sub-oxide has been reported to show a thickness variation connected to

the transition in corrosion rate [22,33]. However, because of the complex morphol-
ogy of the metal/oxide interface, before comparing different materials and different
oxide thicknesses, it is important to fully understand the sub-oxide nature of one
material and one oxide thickness. In this study, 15 APT specimens prepared from
the same autoclave sample were studied, and indeed the investigation shows large
variations both in the sub-oxide layer thickness (ranging from 100 to virtually
0 nm) and in the morphology of the interface region.
The structure of the 5-nm-thick “fingers” of sub-oxide with higher oxygen level
penetrating into volumes of sub-oxide with lower oxygen content found in this
study shows that the sub-oxide may contain a mixture of sub-oxides with different
oxygen levels. Further, it also suggests that there could be a mixture of different
crystal structures/phases, a finding that might explain why it has proven so difficult
to determine the crystal structure of the sub-oxide layer. For example, Ni et al. stud-
ied the sub-oxide layer using electron diffraction, and found several phases: a-Zr,
x-Zr, tetragonal and monoclinic ZrO2, and also unidentified phases [22].
TRANSITION MECHANISM
It is a well-known fact that there is a transition in the oxidation kinetics occurring
at an oxide thickness of approximately 2–3 lm, depending on alloy type. This tran-
sition is associated with the creation of large interconnected lateral cracks that
somehow allow water to penetrate deeper into the oxide, thus increasing the oxida-
tion rate. Because radial cracks connecting the lateral cracks to the surrounding
water are rarely seen, it has been suggested that interconnected pores in oxide grain
boundaries, observed at some distance from the metal/oxide interface, might be the
pathway for water ingress [8]. These large interconnected lateral cracks are most
likely created in two steps, where the first is the formation of small lateral cracks
and the second is when at a certain point, these small cracks propagate and com-
bine to form the large cracks responsible for the onset of transition.
The creation of the small lateral cracks has already been mentioned in previous
sections in this paper, and is explained in detail in Ref 5, according to the following
scenario:
— The undulations of the metal/oxide interface give rise to large stresses in the
direction of the oxide growth, i.e., perpendicular to the interface, with tensile
stresses above metal hills and compressive stresses in between.
— The tensile stresses above metal hills cause lateral cracks to form preferentially
in these locations.
— SPPs act as nucleation sites for cracks, because voids are created above each
SPP as a result of the absence of volume increase caused by the delayed oxida-
tion of SPPs compared to the metal matrix.
The driving force for the growth of the cracks is not established. However, the
following scenario seems likely: Due to the high stresses and small grain size in the
oxide close to the metal/oxide interface, tetragonal oxide is stabilized to some extent
in this area [33,34]. However, the stresses in the oxide decrease with increasing dis-
tance from the metal/oxide interface [11,20,33], causing the tetragonal oxide grains
to transform into monoclinic grains. Because monoclinic grains are larger than tet-
ragonal ones, there will be a substantial increase in the stress levels as this
transformation-induced volume increase occurs, causing the existing cracks to
propagate and form an interconnected band of large cracks. At this point, the oxide
is no longer protective and the corrosion kinetics start over again.
The APT investigation shows that there are very large thickness variations in the
sub-oxide layer along the interface, ranging from 100 to virtually 0 nm. Assuming that
the existence and thickness of the sub-oxide layer is connected to the transition mecha-
nism (the oxidation rate increases drastically at transition, affecting the formation of sub-
oxide), this finding suggests that the transition in Zircaloy-2 is a localized phenomenon.
The idea of a localized transition mechanism in Zircaloy-2 agrees well with
findings in the literature, because the cyclic behavior of the oxidation process mani-
fested by regular intervals of large lateral cracks that exist for many zirconium alloys
during autoclave experiments [31,34–37] has not been reported to the same extent
for Zircaloy-2. Only one or, at the most, two cycles have been reported, followed by
a more linear oxidation rate [38], which would support the idea of a localized tran-
sition in this alloy type.
OXYGEN SATURATION
In all APT specimens, a region containing 30–32 at. % oxygen was found. A simi-
lar composition was found by Ni in ZIRLO after autoclave testing to around 2 and
3 lm by TEM EELS, and interpreted as “oxygen-saturated Zr metal” [39]. As can be
seen in Fig. 1, ZIRLO has a similar weight gain curve as material B during static
steam autoclave oxidation. Using electron diffraction, Ni finds Zr3O at some dis-
tance from the interface (up to 400 nm) and interprets this as the crystal structure
of the oxygen saturated metal [22,39]. In the literature, several zirconium-rich Zr-O
phases are described, mainly hexagonal phases with different oxygen ordering [18].
The maximum oxygen content in these phases is 28.6 at. % [18]. Our APT data on
the oxygen content are slightly higher than this, but it is still possible that the phase
we observe is oxygen-saturated a-Zr or “Zr3O1þx.” Another phase that is not usu-
ally included in the phase diagram is the hexagonal Zr2O1x phase [40], containing
32.4 at. % oxygen. The distance between zirconium atoms is slightly lower in Zr2O
than in hexagonal zirconium (a-Zr) [40], leading to a volume contraction of 7 %
upon oxygen absorption. Because of the compressive stresses underneath metal val-
leys, it is possible that this Zr2O phase is preferentially stabilized in these regions.
The high oxygen level might also be assisted by two other factors:
1. The high dislocation density underneath the oxide scale offers more interstitial
positions for oxygen.
2. All of the phases mentioned above are based on the system of pure zirconium
and oxygen, and not the actual chemistry of Zircaloy-2 and oxygen. Therefore,
it is possible that this system might be able to hold more oxygen.
DIFFUSION PROFILES
It is interesting to compare the measured oxygen diffusion lengths into the metal
with data from the literature. Béranger and Lacombe studied the oxidation of
pure zirconium in oxygen at atmospheric pressure in the temperature range
550 C–850 C and evaluated the resulting oxygen profiles by microhardness
measurements [41]. Using the fact that the diffusion profile into the metal
depends only on the diffusion coefficient for oxygen in the metal, the time for
diffusion, and the oxygen concentration in the metal at the metal/oxide interface
in equilibrium with the oxide—but not on the speed on the advancing metal/ox-
ide interface—they could determine the diffusion coefficient in the temperature
interval 500 C to 850 C. Extrapolating their diffusion data to 400 C we find that
one diffusion length (2(Dt)[1/2]) for a diffusion time t of 3 days (259 200 s) is pre-
dicted to be 205 nm. This is only somewhat larger than the average of the meas-
ured diffusion lengths and suggests that the diffusion coefficients for oxygen in
pure Zr and Zircaloy-2 are similar. The main difference between the matrix of
Zircaloy-2 and pure zirconium is the presence of approximately 1.0 or 1.1 at. %
tin and 0.7 at. % oxygen; the concentration of all other atomic species are very
low. However, Béranger and Lacombe [41] assumed that the oxygen diffusion
coefficient did not depend on the oxygen concentration, so the only difference
that could affect diffusion is the tin content, and thus it seems that the effect of
1 at. % Sn is small.
The possible difference in diffusion profiles between metal hills and valleys, as
observed using TEM EDX, can be explained both in terms of geometric effects
(lower oxygen supply under a convex (metal valley) than a concave (metal hill)
interface because of divergent and convergent oxygen flux, respectively) and in
terms of stresses in the material perpendicular to the metal/oxide interface. It has
previously been shown that both the oxide and the metal are in tension at metal
hills and in compression at metal valleys. The tensile stresses at the hills will cause a
somewhat stretched lattice and therefore facilitate diffusion [20]. A higher diffusion
rate would result in a wider diffusion profile, which is also what was suggested in
the measured EDX line scans, as the diffusion profiles were somewhat longer at
metal hills than at valleys, 130 versus 100 nm in specimen 1 and 220 versus 100 nm
in specimen 2. This higher diffusion rate underneath metal hills might even out the
waviness somewhat, or at least prevent large parts of metal from remaining un-
oxidized inside the oxide scale.
Ni has studied the occurrence of sub-oxide in autoclave-tested samples of
ZIRLO using TEM EELS for oxygen analysis [39]. In one specimen close to transi-
tion (oxide thickness around 3 lm) Ni finds a sub-oxide that is 150 nm wide, fol-
lowed by a diffusion profile into the metal. From the diagram in Fig. 6.9a in Ref 39,
the diffusion length can be estimated to approximately 300 nm. This oxygen profile
was recorded under a metal hill of the undulating interface. In the same specimen,
underneath a metal valley, no sub-oxide was found, only a diffusion profile under
the zirconia layer, with a diffusion length that can be estimated to around 150 nm
(Fig. 6.9b in Ref 39). Apparently, also in this Zr-Sn-Nb alloy, the oxygen diffusion
length is larger under metal hills. Similar results for ZIRLO can be seen (but are not
commented on) in Fig. 9 in Ref 22.
HYDRIDE PRECIPITATION
During autoclave corrosion, hydrogen is generated and some of it is absorbed by
the metal. The small hydrides precipitated in the metal underneath the oxide scale
do not form during oxidation, but are a result of cooling from autoclave tempera-
tures to room temperature. However, it is still important to determine what the fea-
tures are to fully understand the microstructure of the metal underneath the oxide
scale. HAADF contrast in the TEM provides a convenient method for the visualiza-
tion of these small hydride precipitates.
INFLUENCE OF SPPS ON THE OXIDATION PROCESS

The average oxidation rate depends both on the actual oxygen transport rate
through the oxide and on the time to transition in oxidation rate. The oxygen trans-
port rate is believed to be higher in the material with many and small SPPs, because
this material (material A) exhibits an oxide with many more isolated lateral cracks
that act as short-cuts for oxygen diffusion, as described previously. The transition
mechanism has been thoroughly treated previously and is believed to be associated
with cracking of the oxide scale.
The initiation sites for the smaller lateral cracks in the oxide are un-oxidized
SPPs in the oxide, meaning that a material with more SPPs will have more such
cracks. This link is further explained and discussed in Ref 5.
Because of particle hardening, a material with more SPPs also has higher yield
strength. Therefore, less stress relaxation by plastic deformation of the metal will
take place [12] and more stresses will be retained in the material. The high stress
levels will cause the oxide to crack more easily.
In conclusion, the onset of transition will be assisted by a weaker oxide scale
(with more lateral cracks) and a stronger metal. This means that material A will
reach transition earlier than material B, and thus exhibit a poorer corrosion re-
sistance. This is consistent with corrosion data from autoclave and reactor, see
Fig. 1.
The influence of SPP size distribution on sub-oxide formation and oxygen
ingress has not been investigated in this work.
A Model of the Oxidation Process

Summarizing new and previous results on the oxidation behavior of Zircaloy-2 in
steam autoclave, the following picture emerges:
1. Laterally uneven oxygen transport through the oxide layer leads to an undu-
lated metal/oxide interface on a micrometer scale. At a still finer length scale a
further unevenness exists, leading to islands of metal in the oxide.
2. An expansion upon oxidation of 0.5 % parallel to the metal/oxide interface

induces compressive stresses in the oxide and tensile stresses in the metal par-
allel to the interface. In addition, the waviness of the interface leads to very
large local stresses in the oxide and in the metal perpendicular to the interface,
as a consequence of the large expansion upon oxidation, 54 %.
3. SPPs oxidize more slowly than the surrounding matrix, leading to voids in the
oxide at SPPs. These act as nucleation sites for isolated lm-sized lateral cracks
in the oxide, formed by the action of the large tensile stresses in the oxide per-
pendicular to the interface (mainly over metal hills in the metal/oxide inter-
face). A material with a high number density of SPPs (i.e., small SPPs) has an
oxide with many more lateral cracks. It is suggested [5] that these cracks act as
short-circuits for oxygen diffusion paths (oxide grain boundaries) and there-
fore increase the oxidation rate.
4. A sub-oxide layer with around 45–55 at. % oxygen forms at the metal/oxide
interface. Its composition can vary somewhat (50–60 at. %) on a fine (5 nm)
scale, and its thickness varies in the same autoclave tested material from 100
to virtually 0 nm.
5. The sub-oxide (or, if absent, the oxide) is in equilibrium with oxygen saturated
metal matrix. Often a somewhat higher oxygen level (30–32 at. %) is measured
than the literature value for pure Zr, 28.6 at. %. From this level there is an oxy-
gen diffusion profile into the metal, and the diffusion length is in fair agree-
ment with literature data for pure Zr. However, the diffusion length at metal
hills on the undulating metal/oxide interface is considerably larger than that at
metal valleys. This is probably because of both geometrical effects and the dif-
ferent stress situation in the metal perpendicular to the interface (tensile at
hills and compressive at valleys).
6. The locally high tensile stresses in the metal close to the metal/oxide interface
give rise to extensive creep deformation during the oxidation process, evi-
denced by dislocation tangles and patches, sub-grain formation, and some-
times twinning.
7. As the oxide layer grows, stress build-up leads to joining of lateral cracks and
eventually to the formation of large cracks connected to the surrounding water.
This is the cause of the transition in oxidation rate. As suggested in the literature,
this process may be connected to the transformation of stress-stabilized tetragonal
zirconia grains to monoclinic structure, which is associated with a volume
increase. It appears that transition is a more local phenomenon in Zircaloy-2 than
in most other zirconium alloys, because weight gain curves for Zircaloy-2 usually
show a more linear than cyclic oxidation behavior.
8. The effect of SPP size on corrosion rate is twofold: An alloy with many small
SPPs has higher yield strength and will creep deform less than an alloy with
few large SPPs. This leads to a higher stress in the oxide, which has many
more isolated lateral cracks and therefore undergoes major cracking (transi-
tion) at an earlier time.
9. In short, small SPPs give a stronger metal and a weaker oxide, both of which
give shorter time to transition and a higher oxidation rate. In addition, oxygen
diffusion in the weaker oxide is faster.
Conclusions
1. The metal/oxide interface is wavy on a micrometer scale, and has also a finer
sub-structure with metal islands forming especially at wave crests (metal hills).
2. The sub-oxide layer (with an oxygen content of 45–55 at. %) at the metal/ox-
ide interface varies widely in thickness and morphology even within the same
material with the same oxide thickness, from 100 to virtually 0 nm.
3. The sub-oxide layer is sometimes nanostructured and appears to be a mixture
of small (down to 5 nm) volumes with different oxygen content (e.g., 50 and
60 at. %).
4. The metal matrix adjacent to the sub-oxide is oxygen saturated with up to 32
at. % oxygen.
5. The oxygen concentration profile in the metal is in fair agreement with diffu-
sion data for pure zirconium.
6. The oxygen diffusion profile appears to be longer at metal hills than at valleys,
and it is suggested that this is caused by tensile stresses in the metal under hills
and compressive stresses in the metal under valleys, and also by geometrical
effects.
7. Hydride precipitates can be imaged with high contrast in the TEM by using
HAADF imaging.
8. A model for the oxidation is suggested involving creep deformation of the
metal close to the interface and formation of lateral cracks in the oxide. Both
phenomena are caused by the high stresses induced by the metal expansion
upon oxidation and the undulating interface.
9. Observations suggest that the mean SPP size affects the oxidation process in
two ways: A high number density of small SPPs gives a more creep resistant
metal, inducing a higher stress level in the oxide. Many small SPPs also nucle-
ate a higher density of lateral cracks, making the oxide weaker. Both these facts
lead to earlier formation of large cracks in contact with water, and thus an ear-
lier transition in oxidation rate.
ACKNOWLEDGMENTS
The writers thank Stefan Gustafsson, Chalmers University of Technology, for helping
with the EELS investigation of the hydride precipitates, and Sandvik Materials Tech-
nology for supplying the materials and performing the autoclave testing. This work
was carried out with the support of Westinghouse Electric Sweden AB, Sandvik AB,
Vattenfall Nuclear Fuel AB, OKG AB, and the Swedish Radiation Safety Authority.
References
[1] Rudling, P. and Wikmark, G., “A Unified Model of Zircaloy BWR Corrosion and Hydriding
Mechanisms,” J. Nucl. Mater., Vol. 265, 1999, pp. 44–59.
[2] Garzarolli, F., Steinberg, E., and Weidinger, H. G., “Microstructure and Corrosion Studies
for Optimized PWR and BWR Zircaloy Cladding,” Zirconium in the Nuclear Industry,
Eighth International Symposium, ASTM STP 1023, L. F. P. Van Swam and C. M. Eucken,
[3] Eucken, C. M., Finden, P. T., Trapp-Pritsching, S., and Weidinger, H. G., “Influence of
Chemical Composition on Uniform Corrosion of Zirconium-Base Alloys in Autoclave
Tests,” Zirconium in the Nuclear Industry, Eighth International Symposium, ASTM STP
1023, L. F. P. Van Swam and C. M. Eucken, Eds., ASTM International, West Conshohocken,
PA, 1989, pp. 113–127.
[4] Garzarolli, F., Stehle, H., and Steinberg, E., “Behavior and Properties of Zircaloys in Power
Reactors: A Short Review of Pertinent Aspects in LWR Fuel,” Zirconium in the Nuclear
Industry: Eleventh International Symposium, ASTM STP 1295, E. R. Bradley and G. P.
Sabol, Eds., ASTM International, West Conshohocken, PA, 1996, pp. 12–32.
[5] Tejland, P. and Andrén, H.-O., “Origin and Effect of Lateral Cracks in Oxide Scales
Formed on Zirconium Alloys,” J. Nucl. Mater., Vol. 430, 2012, pp. 64–71.
[6] Tägtström, P., Limbäck, M., Dahlbäck, M., Andersson, T., and Pettersson, H., “Effects of
Hydrogen Pickup and Second Phase Particle Dissolution on the In-Reactor Corrosion
Performance of BWR Claddings,” Zirconium in the Nuclear Industry, Thirteenth Interna-
tional Symposium, ASTM STP 1423, G. D. Moan and P. Rudling, Eds., ASTM International,
[7] Tejland, P., Thuvander, M., Andrén, H.-O., Ciurea, S., Andersson, T., Dahlbäck, M., and Hall-
stadius, L., “Detailed Analysis of the Microstructure of the Metal/Oxide Interface Region
in Zircaloy-2 after Autoclave Corrosion Testing,” J. ASTM Int., Vol. 8, No. 6, 2011, Paper ID
JAI102956.
[8] Ni, N., Lozano-Perez, S., Sykes, J. M., Smith, G. D. W., and Grovenor, C. R. M., “Focussed
Ion Beam Sectioning for the 3D Characterization of Cracking in Oxide Scales Formed on
Commercial ZIRLO Alloys during Corrosion in High Temperature Pressurized Water,”
Corr. Sci., Vol. 53, 2011, pp. 4073–7083.
[9] Harada, M. and Wakamatsu, R., “The Effect of Hydrogen on the Transition Behavior of
the Corrosion Rate of Zirconium Alloys,” J. ASTM Int., Vol. 5, No. 3, 2008, Paper ID
JAI101117.
[10] Pilling, N. B. and Bedworth, R. E., J. Inst. Met., Vol. 29, 1923, pp. 529–591.
[11] Parise, M., Sicardy, O., and Cailletaud, G., “Modelling of the Mechanical Behavior of the
Metal-Oxide System During Zr Alloy Oxidation,” J. Nucl. Mater., Vol. 256, 1998, pp. 35–46.
[12] Tejland, P. and Andrén, H.-O., “Oxidation Induced Localized Creep Deformation in Zirca-
loy-2,” J. Nucl. Mater., Vol. 444, 2014, pp. 30–34.
[13] Wei, J., Frankel, P., Polatidis, E., Blat, M., Ambard, A., Comstock, R. J., Hallstadius, L., Hud-
son, D., Smith, G. D. W., Grovenor, C. R. M., Klaus, M., Cottis, R. A., Lyon, S., and Preuss,
M., “The Effect of Sn on Autoclave Corrosion Performance and Corrosion Mechanisms in
Zr-Sn-Nb Alloys,” Acta Mater., Vol. 61, 2013, pp. 4200–4214.
[14] Blank, H., Bart, G., and Thiele, H., “Structural Analysis of Oxide Scales Grown on Zirco-
nium Alloys in Autoclaves and in a PWR,” J. Nucl. Mater., Vol. 188, 1992, pp. 273–279.
[15] Hutchinson, B. and Lehtinen, B., “A Theory of the Resistance of Zircaloy to Uniform
Corrosion,” J. Nucl. Mater., Vol. 217, 1994, pp. 243–249.
[16] Bossis, P., Thomazet, J., and Lefebvre, F., “Study of the Mechanisms Controlling the Ox-
ide Growth under Irradiation: Characterization of Irradiated Zircaloy-4 and Zr-1 Nb-O
Oxide Scales,” Zirconium in the Nuclear Industry: Thirteenth International Symposium,
ASTM STP 1423, G. D. Moan and P. Rudling, Eds., ASTM International, West Consho-
hocken, PA, 2002, pp. 190–221.
[17] Motta, A. T., Gomes da Silva, M. J., Yilmazbayhan, A., Comstock, R. J., Cai, Z., and Lai, B.,
“Microstructural Characterization of Oxides Formed on Model Zr Alloys Using Synchro-
tron Radiation,” J. ASTM Int., Vol. 5, No. 3, 2008, Paper ID JAI101257.
[18] Predel, B., “O-Zr (Oxygen-Zirconium),” O. Madelung, Ed., SpringerMaterials, The Landolt-
Börnstein Database, http://www.springermaterials.com (Last accessed 28 Oct 2013).
[19] Zhao, Z., Blat-Yrieix, M., Morniroli, J.-P., Legris, A., Thuinet, L., Kihn, Y., Ambard, A., and
Legras, L., “Characterization of Zirconium Hydrides and Phase Field Approach to a
Mesoscopic-Scale Modeling of Their Precipitation,” J. ASTM Int., Vol. 5, No. 3, 2008, Pa-
per ID JAI101161.
[20] Parise, M., Foerch, R., and Cailletaud, G., “Coupling Between Diffusion and Mechanics
During the Oxidation of Zircaloy,” J. Physique IV, Vol. 9, No. 9, 1999, pp. 311–320.
[21] Hudson, D., Ni, N., Lozano-Perez, S., Saxey, D., English, C., Smith, G. D. W., Sykes, J., and
Grovenor, C., “The Atomic Scale Structure and Chemistry of the Zircaloy-4 Metal-Oxide
Interface,” 14th International Conference on Environmental Degradation of Materials in
Nuclear Power Systems, Virginia Beach, VA, Aug 23–27, 2009, American Nuclear Society,
La Grange Park, IL.
[22] Ni, N., Hudson, D., Wei, J., Lozano-Perez, S., Smith, G. D. W., Sykes, J. M., Yardley, S. S.,
Moore, K. L., Lyon, S., Cottis, R., Preuss, M., and Grovenor, C. R. M., “How the Crystallog-
raphy and Nanoscale Chemistry of the Metal/Oxide Interface Develops During the Aque-
ous Oxidation of Zirconium Cladding Alloys,” Acta Mater., Vol. 60, 2012, pp. 7132–7149.
[23] Sundell, G., Thuvander, M., and Andrén, H.-O., “Enrichment of Fe and Ni at Metal and Ox-
ide Grain Boundaries in Corroded Zircaloy-2,” Corr. Sci., Vol. 65, 2012, pp. 10–12.
[24] Lindgren, M., Sundell, G., Panas, I., Hallstadius, L., Thuvander, M., and Andrén, H.-O.,
“Towards a Comprehensive Mechanistic Understanding of Hydrogen Uptake in Zirco-
nium Alloys by Combining Atom Probe Analysis With Electronic Structure Calculations,”
J. ASTM Int. (to be published).
[25] Moseley, P. T. and Hudson, B., “Phases Involved in the Corrosion of Zircaloy by Hot Water
350 C,” J. Nucl. Mater., Vol. 99, 1981, pp. 340–344.
[26] Iltis, X. and Michel, H., “Transmission Electron Microscopy Study of a Locally Ordered Zr-
O Solid-Solution Obtained by an Oxidation Treatment of a Zircaloy-4 Alloy,” J. Alloys
Comp., Vol. 177, 1991, pp. 71–82.
[27] Wadman, B., Lai, Z., Andrén, H.-O., Nyström, A.-L., Rudling, P., and Pettersson, H.,
“Microstructure of Oxide Layers Formed During Autoclave Testing of Zirconium Alloys,” Zir-
conium in the Nuclear Industry: Tenth International Symposium, ASTM STP 1245, A. M. Garde
and E. R. Bradley, Eds., ASTM International, West Conshohocken, PA, 1994, pp. 579–596.
[28] Nishino, Y., Krauss, A. R., Lin, Y. P., and Gruen, D. M., “Initial Oxidation of Zirconium and
Zircaloy-2 With Oxygen and Water Vapor at Room Temperature,” J. Nucl. Mater., Vol.
228, 1996, pp. 346–353.
[29] Anada, H. and Takeda, K., “Microstructure of Oxides on Zircaloy-4, 1.0Nb Zircaloy-4, and
Zircaloy-2 Formed in 10.3-MPa Steam at 673 K,” Zirconium in the Nuclear Industry: Elev-
[30] Bossis, P., Lelièvre, G., Barberis, P., Iltis, X., and Lefebvre, F., “Multi- Scale Characteriza-
tion of the Metal-Oxide Interface of Zirconium Alloys,” Zirconium in the Nuclear Industry:
Twelfth International Symposium, ASTM STP 1354, G. P. Sabol and G. D. Moan, Eds.,
[31] Yilmazbayhan, A., Breval, E., Motta, A. T., and Comstock, R. J., “Transmission Electron Mi-
croscopy Examination of Oxide Layers Formed on Zr Alloys,” J. Nucl. Mater., Vol. 349,
2006, pp. 265–281.
[32] Hutchinson, B., Lehtinen, B., Limbäck, M., and Dahlbäck, M., “A Study of the Structure
and Chemistry in Zircaloy-2 and the Resulting Oxide after High Temperature Corrosion,”
J. ASTM Int., Vol. 4, No. 10, 2007, Paper ID JAI101106.
[33] Preuss, M., Frankel, P., Lozano-Perez, S., Hudson, D., Polatidis, E., Ni, N., Wei, J., English,
C., Storer, S., Chong, K. B., Fitzpatrick, M., Wang, P., Smith, J., Grovenor, C., Smith, G.,
Sykes, J., Cottis, B., Lyon, S., Hallstadius, L., Comstock, B., Ambard, A., and Blat-Yrieix,
M., “Studies Regarding Corrosion Mechanisms in Zirconium Alloys,” J. ASTM Int., Vol. 8,
No. 9, 2011, Paper ID JAI103246.
[34] Yilmazbayhan, A., Motta, A. T., Comstock, R. J., Sabol, G. P., Lai, B., and Cai, Z., “Structure
of Zirconium Alloy Oxides Formed in Pure Water Studied With Synchrotron Radiation
and Optical Microscopy: Relation to Corrosion Rate,” J. Nucl. Mater., Vol. 324, 2004, pp.
6–22.
[35] Park, J.-Y., Yoo, S., Choi, B.-K., and Jeong, Y., “Oxide Microstructure of Advanced Zr
Alloys Corroded in 360 C Water Loop,” J. Alloys Comp., Vol. 437, 2007, pp. 274–279.
[36] Qin, W., Nam, C., Li, H. L., Szpunar, J. A., “Tetragonal Phase Stability in ZrO2 Film Formed
on Zirconium Alloys and Its Effects on Corrosion Resistance,” Acta Mater., Vol. 55, 2007,
pp. 1695–1701.
[37] Bouineau, V., Ambard, A., Bénier, G., Pêcheur, D., Godlewski, J., Fayette, L., and Duver-
neix, T., “A New Model to Predict the Oxidation Kinetics of Zirconium Alloys in a Pressur-
ized Water Reactor,” J. ASTM Int., Vol. 5, No. 5, 2008, Paper ID JAI101312.
[38] Griggs, B., Maffei, H. P., and Shannon, D. W., “Multiple Rate Transitions in the Aqueous
Corrosion of Zircaloy,” J. Electrochem. Soc., Vol. 109, No. 8, 1962, pp. 665–668.
[39] Ni, N., “Study of Oxidation Mechanisms of Zirconium Alloys by Electron Microscopy,”
Ph.D. thesis, University of Oxford, Oxford, U.K., 2011.
[40] Steeb, S. and Riekert, A., “Ermittlung der Struktur des Zirkoniumsuboxides Zr2O mittels
Elektronenbeugung [Determination of the structure of the zirconium suboxide Zr2O by
electron diffraction],” J. Less-Common Metals, Vol. 17, 1969, pp. 429–436 (in German).
[41] Béranger, G. and Lacombe, P., “Contribution a l’étude de la cinétique de l’oxydation du

Zirconium a et de la Diffusion de l’oxygène dans le Métal sous-jacent a l’oxyde [Contri-
bution to the study of the kinetics of the oxidation of a-zirconium and diffusion of oxy-
gen into the underlying metal oxide],” J. Nucl. Mater., Vol. 16, 1965, pp. 190–207
(in French).
DISCUSSION
Questions from Arthur Motta, Penn State University
Q1:—Do you think that the cracks you observe near the precipitates are formed
at temperature or upon cooling? What is the evidence either way?
Authors’ Response:—At temperature, since we believe that they are caused by

the expansion of the material perpendicular to the metal/oxide interface upon oxi-
dation. The wavy interface gives rise to large tensile stresses over metal hills and
large compressive stresses over metal valleys. These tensile stresses perpendicular to
the interface give rise to cracks over the hills, as demonstrated by N. Ni [ref. 8 in
the paper]. Compared to these stresses, the difference in thermal expansion between
oxide and metal is small.
Q2:—You mention that a harder metal matrix (caused by precipitates) will

cause the transition to happen earlier. The mechanical properties of Zircaloy tend
to be controlled by cold work and the oxygen dissolved in the oxide, since the pre-
cipitate spacing is too wide to cause hardening. Any comment on the relative role
of these effects on mechanical properties?
Authors’ Response:—There is very little difference in oxygen content (60 ppm)

and no difference in amount of cold work in the two studied materials. As is dis-
cussed in Section 2.1, and seen in Table 2, the SPPs do give a large contribution to
the hardening of the material.
Q3:—Do you also see segregation of Ni and Cr, in addition to Fe, to grain
boundaries?
Authors’ Response:—We see segregation of Fe and Ni, not Cr.
Questions from N. Ramasubramanian, ECCATEC Inc. Canada
Q1:—Oxidation of SPPs is reported to initiate the transformation of tetragonal

to monoclinic (in the barrier layer) and initial cracks. You are proposing a new
mechanism of cavity formation around SPPs in the oxide. Do you imply a solid
state oxidation of SPP that results in cavity formation?
Authors’ Response:—No, SPPs are nobler than the matrix and therefore oxidize
slower. The cavities are formed because the surrounding matrix expands 55% upon
oxidation, whilst SPPs do not oxidize and therefore do not expand.
Q2:—Have you observed unoxidized SPPs embedded in the oxide in your TEM
examination of thin foils?
Authors’ Response:—Yes, we have. This was presented at the last ASTM Zir-
conium in Nuclear Industry Syposium, see Ref. [7].
Q3:—Will your mechanism be applicable to corrosion in-reactor where SPPs

undergo amorphization and dissolution?
Authors’ Response:—Yes, at least up until the point where the SPPs in the ma-
trix dissolve before being oxidized.
Questions from Ron Adamson, Zircology Plus, USA
Q1:—Normally we have not seen a significant influence of SPP size/density on

strength of unirradiated Zircaloy-2. We have seen an effect of irradiation - higher
strength for small SPP. Are you sure the strength differential you report is due to
SPPs, or might there be another factor, like Sn?
Authors’ Response:—We are sure that the effect comes mainly from the SPPs, as
is discussed in Section 2.1, although the slightly higher Sn content in Material A
also gives a small additional contribution.
Q2:—You observed Fe in the oxide grain boundaries (by atom probe). Did you
see evidence of Ni, and can you say anything about the oxidation state?
Authors’ Response:—We do also see Ni, but cannot determine the oxidation state.
Question from Yang-Pi Lin, Global Nuclear Fuel:—You show Fe segregation to

grain boundaries. (This is very good result). There is also a significant difference in
Fe content between A and B. Can you elaborate on the role of Fe content with
respect to corrosion behavior?
Authors’ Response:—Since the solubility of Fe in the matrix is very small, the main
difference between materials of different Fe content is the volume fraction of SPPs. We
believe that the mean SPP size is more important than the SPP volume fraction.
Question from Johannes Bertsch, Paul Scherer Institute:—Comment: Crack for-

mation at SPP can also be found in irradiated cladding as long as the SPP still exists
in the oxide layer.
Q1:—Can small cracks in the oxide help to reduce the overall stresses in the ox-
ide layer?
Authors’ Response:—Yes they can. This is also what we say in the paper, that
the small lateral cracks help release the tensile stresses created in the oxide above
metal hills.
Question from Jean-Christophe Brachet, CEA-Sailey Nuclear Materials Dept.

France:—How many samples did you prepare before obtaining a sample located at
the ZrO2/metal interface for the APT?
Authors’ Response:—Using the FIB system it is possible to locate the interface

rather precisely so we had no problems to locate the specimen at the desired posi-
tion. So far we have prepared and analyzed some 50 APT specimens containing the
metal/oxide interface region.
Question from Igor Evdokimov, SRC RF Triniti, Russia:—You have reported

redistribution of Fe near the oxide/metal interface at a small scale - it leads to segre-
gation of iron at boundaries of ZrO2 sub grains. It is hypothesized that due to stress
gradient, Fe migration near the interface can take place at a larger scale, comparable
to the undulation period. Did you try to check if there were noticeable variations in
Fe content in the metal along the undulated metal/oxide interface?
Authors’ Response:—No, not at such a large scale.
Question from K. Kapoor, NFC:—The compressive stress in oxide is likely to

nucleate tetragonal phase in oxide. Please comment on the effect of this tetragonal
phase on crack nucleation in the oxide.
Authors’ Response:—In the paper we discuss that the transformation from tet-
ragonal to monoclinic oxide affects crack propagation rather than initiation. How-
ever, it is possible that this transformation may also be a contributing factor to the
initiation, as suggested by other authors.
Question from Dave Ludlow, AMEC:—Have you observed a variation in crack

morphologies in the oxide (e.g., crack formation at SPP or undulation peak)? What
are your thoughts on the mechanisms that may cause the variations in crack
morphologies?
Authors’ Response:—As explained above, the wavy interface gives rise to a large
tensile stress perpendicular to the interface above metal hills and a large compres-
sive stress above metal valleys. In Ref. [5] we suggest that both a tensile stress and a
large SPP is required to initiate cracks, which explains that cracks are most often
observed over metal hills.
Question from S. Anantharaman, BARC, Mumbai, India:—This question is

regarding the hydride observed at the metal-oxide interface. How was it ensured
that the observed feature was indeed a hydride? What is the effect of such hydrides
on the corrosion reaction?
Authors’ Response:—It was ensured using EELS, where a shift in the plasmon
peak of 2.1 eV was found between the metal and the hydride. This shift indicates
the presence of hydrogen and further that the hydride is of d type. These hydrides
should have no effect on the corrosion rate, as they are most likely created during
cooling from autoclave temperature.
P. G. Frankel,1 J. Wei,1 E. M. Francis,1 A. Forsey,1 N. Ni,2

S. Lozano-Perez,2 A. Ambard,3 M. Blat-Yrieix,3
R. J. Comstock,4 L. Hallstadius,5 R. Moat,6
C. R. M. Grovenor,2 S. Lyon,1 R. A. Cottis,1 and M. Preuss1
Effect of Sn on Corrosion
Mechanisms in Advanced
Zr-Cladding for Pressurised
Water Reactors
Reference
Frankel, P. G., Wei, J., Francis, E. M., Forsey, A., Ni, N., Lozano-Perez, S., Ambard, A., Blat-Yrieix,
M., Comstock, R. J., Hallstadius, L., Moat, R., Grovenor, C. R. M., Lyon, S., Cottis, R. A., and
Preuss, M., “Effect of Sn on Corrosion Mechanisms in Advanced Zr-Cladding for Pressurised
Water Reactors,” Zirconium in the Nuclear Industry: 17th International Symposium, STP 1543,
Robert Comstock and Pierre Barberis, Eds., pp. 404–437, doi:10.1520/STP154320130006,
ASTM International, West Conshohocken, PA 2015.7
ABSTRACT
The desire to improve the corrosion resistance of Zr cladding material to allow
high burnup has resulted in a general trend among fuel manufacturers to
develop alloys with reduced levels of Sn. Whereas the detrimental effect of Sn on
high-temperature aqueous corrosion performance is widely accepted, the reason
for it remains unclear. High-energy synchrotron x-ray diffraction was used to
characterise the oxides formed by autoclave exposure on Zr-Sn-Nb alloys with
tin concentrations ranging from 0.01 to 0.92 wt. %. The alloys studied included
the commercial alloy ZIRLO and two variants of ZIRLO with significantly lower tin
Manuscript received January 7, 2013; accepted for publication May 23, 2013; published online June 19, 2014.
1
Materials Performance Centre, School of Materials, Univ. of Manchester, M13 9PL, United Kingdom.
2
Dept. of Materials, Univ. of Oxford, Parks Rd., Oxford OX1 3PH, United Kingdom.
3
EDF R&D, MMC Dept., Les Renardières, Route de Sens, Ecuelles, 77818 Moret sur Loing Cedex, France.
4
Science and Technology Dept., Westinghouse Electric Co., 1340 Beulah Rd., Pittsburgh, PA, United States of
America.
5
Westinghouse Electric Sweden AB, S-721 63 Västerås, Sweden.
6
The Open Univ., Walton Hall, Milton Keynes, MK7 6AA, United Kingdom.
7
ASTM 17th International Symposium on Zirconium in the Nuclear Industry on February 3-7, 2013 in
Hyderabad, India.
Copyright V
FRANKEL ET AL., DOI 10.1520/STP154320130006 405
levels, referred to here as A-0.6Sn and A-0.0Sn. The nature of the oxide grown
on tube samples from each alloy during autoclave testing at 360 C was
investigated by cross-sectional scanning and transmission electron microscopy
(SEM and TEM). Non-destructive synchrotron x-ray diffraction analysis on the
oxides revealed that the monoclinic and tetragonal oxide phases display highly
compressive in-plane residual stresses with the magnitudes dependent on both
phase and alloy. Additional in situ synchrotron x-ray diffraction experiments
during oxidation at 550 C provided further confirmation of the trends seen for
autoclave-tested samples and demonstrated the presence of elevated levels of
tetragonal phase in the initial stages of oxidation. In situ and ex situ
measurements demonstrate unambiguously that the amount of tetragonal phase
present and, more importantly, the degree of transformation from tetragonal to
monoclinic oxide both decrease with decreasing tin levels, suggesting that tin
stabilises the tetragonal phase. It is proposed that in Zr-Nb-Sn alloys with low
Sn, the tetragonal phase is mainly stabilised by very small grain size and,
therefore, remains stable throughout the corrosion process. By contrast, in alloys
with higher tin levels, larger, stress stabilised, tetragonal grains can form initially,
but then transform as the corrosion front progresses inward and stresses in the
existing oxide relax.
Keywords
Zirconium alloys, ZrO2, corrosion mechanisms, synchrotron X-ray diffraction, In-
situ oxidation, phase transformation
Introduction
The importance of zirconium alloys in pressurised water reactors (PWRs) is a result
of their low neutron cross section, high corrosion resistance, and sufficient mechani-
cal performance during reactor service. During the initial stage of corrosion in high-
temperature pressurised water, zirconium alloys deviate from the expected parabolic
rate law with a tendency towards cubic growth kinetics [1–5]. In general, once the ox-
ide has grown to a few lm (pre-transition), a brief acceleration of corrosion kinetics
is observed (post-transition), followed by a second cycle of parabolic/cubic growth
rates. These cycles are repeated until accelerated linear corrosion occurs (breakaway)
[3,6–16]. Knowledge of these corrosion kinetics, and especially a more mechanistic
understanding of transition and breakaway, is of great importance for undertaking
accurate lifetime predictions for cladding tubes. During the gradual oxidation, a large
compressive stress is expected to build up inside the oxide because of a Pilling-
Bedworth ratio of 1.56 for the formation of zirconia from zirconium [1,17]. This
stress is believed to help stabilise the metastable tetragonal ZrO2 [7,18–20], and has
been reported to form and accumulate near the oxide/metal interface [16,19,21,22].
Compressive stress in the oxide must be balanced by tensile stress in the metal sub-
strate and it has been argued that this will result in creep of the metal substrate with
prolonged corrosion [19,23], relaxing the compressive stresses in the oxide and gen-
erating a stress gradient with the highest stresses near the oxide/metal interface
[19,24]. As a result of this stress relaxation over time, the tetragonal ZrO2 grains can
become destabilised, resulting in a martensitic phase transformation to monoclinic
ZrO2 associated with a 5 % volume expansion [25]. It has been suggested that this
phase transformation results in the formation of cracks [26], which will eventually
lead to breakdown of the barrier properties of the inner oxide. However, experimental
evidence for a direct link between the tetragonal to monoclinic phase transformation
and crack formation remains inconclusive. Other proposed factors that may contrib-
ute to stabilisation of the tetragonal phase include the effect of small oxide grain size
(smaller than 25–30 nm) [27–30] and defects induced by doping, but the concentra-
tion of dopants in the oxide is generally considered to be too low to play any signifi-
cant role.
Tin is a common addition in the zirconium alloys widely used as fuel cladding
material in light water reactors, and was originally introduced as an alloying element
countering the poor corrosion resistance caused by nitrogen [31]. In PWR reactors, it
is well known to have both positive and negative impacts on the in-reactor corrosion
rate and growth acceleration at high burnups. A reduction of the tin content generally
improves the corrosion resistance, e.g., for Zircaloy-4 [31] and advanced alloys like
Westinghouse ZIRLO8 [32,33]. This insight has been utilized over the years to improve
cladding alloys by reducing tin content. However, tin is also known to provide a
robustness and predictability of the corrosion behaviour, e.g., it protects against
enhanced corrosion in high lithium and abnormal chemistry conditions [34]. Tin
additions also reduce the corrosion enhancement around welds, and reduce the effect
of irradiation on accelerating oxidation of zirconium alloys at high burnups believed
to be caused by hydrogen [35]. These contradictory effects of tin have been balanced
in the latest generation of advanced alloys, e.g., Westinghouse Optimized ZIRLO.9
The relationship between tin content and the volume fraction of tetragonal phase
in the oxide is considered to be of great importance in terms of understanding its
effect on corrosion [27,36,37]. It has been reported that decreasing tin levels lead to
the formation of a thicker tetragonal barrier layer at the oxide/metal interface
improving corrosion performance [37]. However, other researchers have found that
the decrease of tin actually reduces the tetragonal phase fraction in the oxide grown
on Zr alloys [27,36]. The latter findings are further supported by studies on the stabi-
lisation of tetragonal phase in structural tetragonal zirconia by tin, although in this
case the required tin levels are comparatively high because of the larger oxide grain
size [38,39]. In general, the observations reported in these studies were all based on a
very limited number of samples and used laboratory x-ray diffraction rather than
high flux synchrotron x-ray diffraction. Considering the extremely low intensity of
8
ZIRLO is a registered trademark of Westinghouse Electric Company LLC in the United States and may be
registered in other countries throughout the world. All rights reserved. Unauthorized use is strictly
prohibited.
9
Optimized ZIRLO is a trademark of Westinghouse Electric Company LLC in the United States and may be
registered in other countries throughout the world. All rights reserved. Unauthorized use is strictly
prohibited.
the tetragonal reflections used in such studies, the uncertainty from those findings
must have been very high. To date, mechanistic studies on the effect of tin on corro-
sion performance have only been carried out in more traditional Zr alloys that do
not contain any Nb. However, Nb containing alloys such as ZIRLO and M510 have
now widely replaced traditional Zircaloys in fuel assemblies for PWRs.
In the present study, three Zr-Nb-Sn alloys with different tin contents (ZIRLO,
A-0.6Sn, and A-0.0Sn) were corrosion tested in an autoclave operating in simulated
primary water conditions [40]. Autoclave tested materials were examined after dif-
ferent exposure times by high-energy synchrotron x-ray diffraction to study the
stress state and the tetragonal/monoclinic ZrO2 ratio in the oxide. (Preliminary
results from these measurements were presented in [41].) The oxide microstruc-
tures were characterised by scanning electron microscopy (SEM) and transmission
electron microscopy (TEM).
Additionally, in situ synchrotron x-ray diffraction was carried out during oxida-
tion of ZIRLO and A-0.0Sn samples in air at 550 C to provide information about the
state of the oxide during the early stages of formation. This investigation also elimi-
nated much of the uncertainty associated with the limited number of measurements
at different oxidation times possible for autoclave tested samples, as well as the sam-
ple to sample variation arising from using different autoclave tested samples to repre-
sent different exposure times. In situ testing allowed a single position on a single
sample to be tracked while continuously collecting diffraction data.
Experimental
MATERIALS PREPARATION
The alloys investigated in this study were recrystallized ZIRLO and two low tin ver-
sions (A-0.6Sn and A-0.0Sn). Corrosion samples of 30 mm in length (9 mm outer
diameter, 0.8 mm thickness) were cut from tubes provided by Westinghouse. For in
situ oxidation, 2 mm strips were prepared from the alloys with highest and lowest
tin content. Chemical compositions of the alloys were determined at the EDF
Research and Development facility in Moret-Sur-Loing, France and can be found in
Table 1. Prior to corrosion testing, all samples were pickled at 60 C in a solution of
10 vol. % hydrofluoric acid, 45 vol. % nitric acid, and 45 vol. % distilled water for
3 min to remove inorganic contamination from the fabrication process.
AUTOCLAVE CORROSION
To record oxidation weight gain as a function of autoclave exposure, as well as gen-
erating samples that have experienced different duration of corrosion, a set of four
samples from each alloy were introduced into the autoclave every 20 days, provid-
ing a total of 60 samples, subsequent 100-day exposures then provided samples
over a range of times. Corrosion experiments were conducted in static isothermal
autoclaves operating at 360 C in simulated primary water chemistry (Li and B were
10
M5 is a registered trademark of Areva NP.
TABLE 1 Measured chemical composition of the materials.
Alloying Elements
Alloy Zr Fe (wt. %) Sn (wt. %) Nb (wt. %) Hydrogen

(wt. ppm)
Recrystallised Balance 0.09 0.92 0.91 10

ZIRLO
A-0.6Sn
A-0.0Sn
added in the form of lithium hydroxide and boric acid, Li ¼ 2 wt. ppm, B ¼ 1000 wt.
ppm), at a saturation pressure of around 18 MPa (360 C). Careful weighing of all
samples at regular intervals allowed the pre-transition corrosion kinetics of each
alloy to be determined, with approximately 20-day increments. The results of the
autoclave tests are summarised in Table 2. The tests were performed consistent with
the ASTM G2 standard and are discussed in detail elsewhere [40]. The corrosion
experiments aimed to provide samples and weight gain data from the early stage of
corrosion until after the first transition.
OXIDE MICROSTRUCTURE CHARACTERISATION

For imaging of the oxide after corrosion testing, cross-sectional samples were pre-
pared for SEM. To protect the oxide during sample preparation, a Ni coating was
applied to each sample before cutting. Cross-sectional specimens were ground and
polished carefully following standard metallographic procedures and etched in a so-
lution of 5 vol. % hydrofluoric acid, 45 vol. % nitric acid, and 50 vol. % distilled
water at room temperature. A carbon layer (5 nm thick) was deposited on freshly
etched sample surfaces using a Gatan Model 682 PECS precision etching and coat-
ing system. The carbon layer improved the surface conductivity, preventing the
accumulation of electrostatic charge, thereby minimising image artefacts. Imaging
of the cross-sectional samples was performed using a Philips XL30 FEG-SEM.
Further characterisation was carried out on ZIRLO samples by transmission
electron microscopy (TEM). Thin specimens were extracted from samples with a
TABLE 2 Summary of autoclave corrosion results.
Alloy Time to Transition (days) Max. t (days) Max. DW (mg/dm2)
A-0.0Sn > 360 540 78.7 6 0.8

A-0.6Sn 140–240 260 56.2 6 3.7
ZIRLO 140 160 43.2 6 2.9
range of exposure times using an FEI FIB200 instrument (5–30 kV gallium incident
beam with beam currents of 50– 5000 pA) to give electron-transparent areas approx-
imately 10 5 lm containing the whole oxide scale and some metal substrate. Sam-
ples were carefully thinned to provide a uniform thickness of < 100 nm, to avoid the
complex images that result with specimens much thicker than the oxide grain size.
TEM characterization was carried out using a Philips CM20 microscope operated at
200 kV. Several TEM specimens were investigated from each sample to ensure the
reproducibility of observations.
SYNCHROTRON X-RAY DIFFRACTION

The advantage of using synchrotron x rays over laboratory x rays is the high flux
and high peak position accuracy in conjunction with a low background, which ena-
bles the characterisation of minor phases such as the tetragonal phase in the present
case. In this study, two types of diffraction experiments are discussed, in each case
the penetration depth of the x ray was such that measurements represented an aver-
age over the entire oxide thickness. This was also demonstrated by the presence of
FIG. 1 Schematic layout of (a) the main beamline components at EDDI, BESSY II [42] and
(b) the in situ oxidation setup used at I11, Diamond Light Source.
FIG. 2 Diffraction results, showing (a) a 2D plot intensity variation with tilt w angle,
captured at BESSY II EDDI beamline, on the oxide of a ZIRLO sample at
diffraction angle 2h ¼ 6 and (b) diffraction rings recorded for a ZIRLO sample
during in situ oxidation at I12, Diamond Light Source.
strong diffraction signal from the Zr substrate for all measurements, as shown in
the typical diffraction patterns in Fig. 2.
AUTOCLAVE SAMPLES
Experiments to characterise the residual stresses in the oxide layer (monoclinic and
tetragonal ZrO2), the oxide layer, and the metal substrate, together with the tetrago-
nal/monoclinic phase fraction, were carried out on the EDDI beam line at BESSY,
Berlin, an energy dispersive instrument dedicated to material science. A detailed
description of the EDDI beam line has been given by Genzel et al. [42], and the gen-
eral layout is shown in Fig. 1(a). Where the angle / determined the component of in-
plane stress measured, and w ¼ h to ensure that when u ¼ 0, the crystal planes paral-
lel to the specimen surface were sampled. The residual stress measurements were
undertaken using the classical sin2 w approach [43] assuming a biaxial stress state in
a Cartesian coordinate system. The elastic properties used for each phase are shown
in Table 3 [24,44]. Based on single crystal values of the monoclinic zirconia phase,
Polatidis et al. [24] calculated polycrystalline diffraction elastic constants using an
TABLE 3 Elastic constants used for stress calculation.
Phase Poisson’s Ratio Young’s Modulus (GPa)
ZrO2 [39, 41] 0.282 253

a-Zr [40] 0.34 96
elasto-plastic self-consistent (EPSC) model [45]. EPSC modelling simulates deforma-

tion of polycrystalline aggregates by predicating the response of each grain within an
infinite medium to predict the overall response of a given diffraction plane within
the polycrystalline material. Because of the lack of reliable elastic properties for the
tetragonal phase at room temperature (yttrium-stabilised zirconia data was consid-
ered inappropriate), the current study applied the same monoclinic elastic constants
to calculations regarding the tetragonal phase.
Because all measured samples had layers of oxide that were only a few microns
thick, it is reasonable to assume that the out of plane stress in the oxide is zero. This
assumption, in combination with the sin2 w technique, eliminates the need for a
stress-free lattice parameter. Measurements were carried out with an incident beam
and the EDX detector placed at 3 (i.e., 2h ¼ 6 ) to the surface when the sample was
at w ¼ 0 . Twenty-one individual w angles ranging from 22 to 66 were selected
based on the intensity of the spectrum at each angle, Fig. 2(a). This range corre-
sponds to a sin2 w range of 0.14 to 0.84. All stress data presented here are represen-
tative of the hoop direction of the cladding tube. Initial measurements carried out in
the both axial and hoop directions showed no significant differences between the
stresses in the two directions. A BESSY II in-house analysis program written in
MATHEMATICA was used for processing the output files. It applies pseudo-Voigt
peak fitting for a selected energy range and produces plots of sin2 w, peak intensity,
peak position, and widths at half maximum of each peak (FWHM). In general, accu-
rate peak analysis of the diffraction signal from a thin monoclinic/tetragonal oxide
layer is inherently difficult because of many overlapping peaks and the nano-sized
grain structure resulting in peaks that are both low in intensity and broad. In the
present case, the reflections from the oxide that could be fitted with the highest cer-
tainty were the (111) reflection of the monoclinic phase and the (101) of the tetrago-
nal phase. In addition, the (1011) reflection of the a-zirconium phase and (111)
reflection of the monoclinic phase were also analysed. A typical diffraction spectrum
of autoclave tested material can be seen in Fig. 2, highlighting the limited quality of
the tetragonal ZrO2 (101) reflection. Note that hydrogen generated during prolonged
autoclave testing led to the formation of hydrides when the material was cooled to
room temperature. This resulted in the observation of a weak (111) reflection of the
d Zr-hydride phase. To minimise texture effects in the oxide when determining the
integrated intensity (I) of each phase, the data were averaged along the range of w
angles. A more detailed description of the data analysis carried out during such stud-
ies can be found in [24].
IN SITU OXIDATION
To provide a higher density of measurements as a function of oxide growth, and
thereby increased confidence in the trends observed in autoclave tested samples, syn-
chrotron XRD measurements were performed while samples were exposed to air at
550 C. This temperature was chosen to maximise oxide growth in the beamtime
available. To achieve in situ oxidation, an electro-thermal mechanical tester (ETMT)
was mounted onto a large sample stage at the I12 beamline, Diamond Light Source.
Resistive heating in the ETMT ensured excellent temperature control of samples
while still allowing easy access for the x rays onto the oxides as they formed. A sche-
matic of the setup is shown in Fig. 1. Diffraction measurements were carried out
directly opposite to the position where thermocouples were spot welded to the sam-
ples, ensuring careful control of the oxidation temperature in the region of interest.
Diffraction patterns produced by the sample from a monochromatic incident beam
(79 keV), were recorded every 5 s using a high-resolution 2D detector (Thales Pixium
RF4343) as the sample was heated in air. Figure 2 shows a typical diffraction pattern
obtained after only 4 s exposure time, when the oxide peaks became visible. To
improve the peak intensities for fitting, images were summed together in sets of 20,
resulting in a final time resolution of 100 s for the data presented here. Once summed,
images were integrated circumferentially to provide diffraction profiles at every five
degrees relative to the sample surface. From the resulting profiles, pseudo-Voigt peak
fitting was applied to the monoclinic (111)/(111) and tetragonal (101) peaks
using MatlabTM scripts. This procedure provided time-resolved peak intensity and
position for each peak as a function of angle to the sample surface. By applying the
sin2 w technique, as discussed earlier, it was possible to evaluate the average stress in
the two oxide phases as a function of time. The relative intensities of the tetragonal
and monoclinic peaks were used to calculate the evolution of tetragonal phase
fraction.
It should be noted that it was not possible to measure the weight gain of the in
situ oxidised samples, so only the oxidation time can be plotted for these experi-
ments. Further controlled exposures are planned in the future to correlate oxidation
time to the thickness of oxide produced, thereby allowing the evolution of stress
and tetragonal phase to be related directly to oxide thickness.
Results
AUTOCLAVE CORROSION
The corrosion kinetics of autoclave tested ZIRLO, A-0.6Sn, and A-0.0Sn samples
are presented in Fig. 3 as corrosion weight gain against autoclave exposure time.
Note that instead of presenting averaged values for each group of notionally identi-
cal samples, data for each individual sample are plotted in Fig. 3, illustrating the
level of scatter observed during the experiment.
The tested materials exhibited typical corrosion performance with an indication
of a change from pre- to post-transition kinetics at around 140 days in the case of
ZIRLO and 140–240 days in the case of A-0.6Sn. Before transition, the oxide
appeared black in all samples. During transition, specimens exhibited patches where
the oxide had become grey in colour. Note that in the case of A-0.6Sn, a very long
transition phase of about 100 days was detected. Interestingly, corrosion tests carried
out on material with identical chemistry, but in a partially recrystallized condition,
did not show such a long transition phase [40].
FIG. 3 Autoclave corrosion weight gain profile of recrystallized ZIRLO up to 180 days of
autoclave exposure, recrystallized A-0.6Sn, and recrystallized A-0.0Sn up to
260 days. Note that measurements above 300 days were taken from a separate,
long-term autoclave test running under similar conditions, but with fewer
interruptions.
With up to 260 days of autoclave exposure, A-0.0Sn showed no sign of a transi-

tion in oxidation kinetics. A universally black oxide was observed on pre-transition
samples as early as after 20 days and as late as after 260 days of corrosion testing. A
second series of autoclave exposures aimed at recording long-term corrosion
kinetics also confirmed that A-0.0Sn showed no sign of transition up to 360 days of
exposure (i.e., 45 mg/dm2 weight gain), However, after 540 days of exposure, the
oxides appeared to be post-transition in nature because of the presence of clearly
delineated dark and light patches on their surfaces and 80 mg/dm2 weight gain.
OXIDE MICROSTRUCTURE
SEM analysis showed that in the early stages of corrosion (10 days of exposure), a
continuous oxide layer (1 lm) with almost no visible cracks and a smooth oxide/
metal interface was present for all three alloys, Figs. 4(a), 4(c), and 4(e). The oxide
thickness, measured by using the FEG-SEM images, was consistent with that calcu-
lated from oxidation weight gain (15 mg/dm2 ¼ 1 lm) [46,47].
Figure 4(b), 4(d), and 4(f) show cross-sectional FEG-SEM images taken when each
alloy had gained about 30 mg/dm3 weight. This corresponded to an exposure time in
the autoclave exceeding 120 days in the case of ZIRLO samples, 160 days in the case of
A-0.6Sn, and 180 days in the case of A-0.0Sn. In the case of ZIRLO and A-0.6Sn, the
examined samples were near to the exposure time where transition occurred. It can be
seen that at this stage the oxide had developed some short and medium-sized cracks.
In A-0.0Sn alloy, where the examined sample was still more than 180 days prior to the
first transition, the oxide displayed only very few small cracks.
FIG. 4 Backscattered electron mode SEM images representative of the oxide (cross-
section) (a) ZIRLO after 10 days of autoclave exposure, (b) ZIRLO (120 days
exposure), (c) A-0.6Sn (10 days exposure), (d) A-0.6Sn (160 days exposure), (e)
A-0.0Sn (10 days exposure), and (f) A-0.0Sn (180 days exposure). Note that an
Au layer was deposited on top of the oxide prior to Ni-coating to improve the
electrical conductivity of the surface for subsequent electroplating.
A more detailed illustration of the pre-transition oxide microstructure found in

the ZIRLO samples is provided by the TEM bright field images in Figs. 5 and 6. The
oxide thicknesses observed by TEM show good agreement with those from SEM
and weight-gain measurements. The images display the typical columnar oxide
grain structure often reported for zirconium alloys over most of the oxide. The co-
lumnar grains were always observed to grow roughly perpendicular to the metal/ox-
ide interface, with an average grain size of about 14–40 nm 135–230 nm,
occasionally interrupted by smaller equiaxed grains 10–30 nm in diameter (Fig.
6(b)). Additionally, a region approximately 200 nm thick was observed at the outer-
most part of all the samples studied (i.e., the first oxide to form) consisting almost
entirely of equiaxed grains and usually associated with extensive porosity (Fig. 6(a)).
EX-SITU RESIDUAL STRESS AND PHASE FRACTION ANALYSIS

A number of autoclave tested specimens were investigated using synchrotron x-ray
diffraction as described above and in [24,48]. Specimens were selected from rela-
tively early stage of corrosion, i.e., before and shortly after the first transition.
Because samples at transition display a patchy surface of grey (post-transition) and
black (pre-transition) oxides, measurements were carried out in both areas. Meas-
urements of the residual stresses in the metal substrate consistently showed tensile
stresses of up to þ150 MPa. In Fig. 7, the measured oxide stresses (monoclinic and
tetragonal phase) are plotted as a function of corrosion weight gain for ZIRLO, A-
0.6Sn, and A-0.0Sn. In both ZIRLO and A-0.6Sn, the oxide stresses in the
FIG. 5 TEM bright field images showing morphology of oxides formed on ZIRLO after
(a) 34 days, (b) 60 days, and (c) 100 days of autoclave exposure.
monoclinic phase were about 700 MPa to 1000 MPa regardless of the corrosion
weight gain, whereas stresses in A-0.0Sn samples were slightly lower in magnitude
(500 to 900 MPa) and accompanied by greater scatter. It should be noted that
whereas detailed comparisons of the oxide texture are not given here, the diffrac-
tion patterns from different oxides showed no significant differences in terms of
their crystallographic texture.
FIG. 6 TEM images of the oxide-formed ZIRLO after 100 days of autoclave exposure,
showing (a) the high density of small equiaxed grains present at the outermost
part of the oxide and (b) the less commonly seen equiaxed grains occasionally
interrupting the columnar grains further away from the outer oxide surface.
FIG. 7 The evolution of hoop stresses in (a) the monoclinic, and (b) tetragonal oxide
phases formed on ZIRLO, A-0.6Sn, and A-0.0Sn during autoclave exposure at
360 C water.
The stresses measured in the tetragonal oxide phase had considerably higher
uncertainty than the stresses measured in the monoclinic phase because of a very
weak tetragonal (101) reflection, see Fig. 2. The stresses determined in the tetragonal
phase were generally between 1300 to 2100 MPa for ZIRLO and A-0.6Sn sam-
ples. It can be seen from Fig. 7(f) that A-0.0Sn displayed very large stress variations
in the tetragonal phase, with compressive stresses as low as 500 MPa and as high
as 2400 MPa.
Figure 8 shows the calculated tetragonal phase fraction and the equivalent layer
thickness of the tetragonal phase as a function of weight gain. The calculation is
based on the Garvie-Nicholson equation [49]:

Tetragonal Fraction ¼ ltetð101Þ =lmonoð111Þ þ lmonoð111Þ
However, in this case, the Garvie-Nicholson equation was modified to include the
integrated intensity of I averaged over all the w angles, thereby reducing the
FIG. 8 Calculated (a) phase fraction and (b) equivalent layer thickness of tetragonal
phase plotted as a function of weight gain measured on ZIRLO, A-0.6Sn, and A-
0.0Sn.
influence of any crystallographic texture present in the tetragonal phase. An equiva-

lent layer thickness of the tetragonal phase was calculated by normalising the calcu-
lated tetragonal phase fraction with the measured corrosion weight gain. In contrast
to the calculated tetragonal fraction, the theoretical equivalent tetragonal phase
thickness aims to take the oxide thickness into consideration. Figure 9 shows sche-
matically, in a simplified manner, the expected distribution of tetragonal and
monoclinic ZrO2 based on observations such as [16,19]. It highlights that with con-
tinuous growth of the oxide, the ratio of tetragonal/monoclinic phase is expected to
decrease, although the absolute volume of tetragonal oxide might remain
unchanged. Such a trend is indeed observed in our new data, particularly for ZIRLO
and to a slightly lesser degree for A-0.6Sn, Figs. 8(a) and 8(b), although during the
early stages of corrosion the tetragonal phase fraction is significantly lower in A-
0.6Sn compared to ZIRLO. In contrast, A-0.0Sn shows hardly any change of the tet-
ragonal/monoclinic ratio with increasing weight gain and a very low tetragonal
phase fraction from the first measurement point, i.e., after only 10 days of corro-
sion. The present results, based on a large number of samples, clearly show that tin
increases the tetragonal phase fraction in agreement with observations by [27,36],
but contradicts the observations of [37]. Finally, a decrease of the tetragonal phase
fraction from pre- to post-transition conditions was observed in A-0.6Sn and A-
0.0Sn.
FIG. 9 Schematic illustrating three possible distributions of tetragonal phase in the

oxide during the inward corrosion of Zr, where the tetragonal phase is (a)
uniformly distributed, (b) concentrated only near the oxide/metal interface, and
(c) concentrated at the interface but also remains present further away (note
that the shaded region represents the presence of tetragonal phase). The
evolution of the average tetragonal phase fraction that would be observed for
each is shown, as well as the volume of tetragonal phase this represents.
IN SITU RESIDUAL STRESS AND PHASE FRACTION ANALYSIS

The evolution of tetragonal phase fraction for ZIRLO and A-0.0Sn during oxidation
in air at 550 C is shown in Fig. 10. These results confirm the observations made for
ex situ measurements, where a reduced fraction of the tetragonal phase was seen for
the lower tin alloys. As with the measurements on autoclave samples, the tetragonal
phase fraction for the oxides grown on ZIRLO is about twice that present in the low
tin alloy (A-0.0Sn). However, for oxidation at 550 C in air, the tetragonal phase
fractions were significantly higher than found in the autoclave samples, with aver-
age fractions of around 30 % and 15 % for ZIRLO and A-0.0Sn respectively after
14 h exposure. In addition to the data after prolonged oxidation, the in situ experi-
ment was also able to track the evolution of the tetragonal phase in the very early
stages of oxidation, providing information not available from previous ex situ meas-
urements. This early corrosion data shows that initially the oxides formed on both
alloys contain very high fractions of the tetragonal phase. For ZIRLO, the tetragonal
phase makes up around 65 % of the initial oxide formed, whereas for A-0.0Sn the
initial tetragonal fraction is as high as 40 %. This high initial tetragonal content
appears to correlate with TEM images from the upper (first formed) oxide, which
show a high concentration of equiaxed grains. In both cases, the tetragonal fraction
drops rapidly in the first 2 h of oxidation, and for A-0.0Sn the tetragonal fraction
appears to plateau after around 6 h. The tetragonal fraction for ZIRLO, however,
seems to continue to gradually fall for the duration of the experiment, resulting in a
difference of only about 5 % between the two alloys after 14 h of oxidation.
Figure 11 illustrates the evolution of the in-plane residual stresses present in the
monoclinic oxide for the two alloys. In contrast to the results shown in Fig. 7(a) for
FIG. 10 The evolution of tetragonal phase fraction during the oxidation of ZIRLO and A-
0.0%Sn at 550 C for 14 h.
FIG. 11 The evolution of residual stresses in the monoclinic phase of the oxides formed
during the oxidation of ZIRLO and A-0.0 %Sn at 550 C in air for 14 h.
the autoclave tested samples, a clear difference is seen between the two alloys when
oxidised at 550 C in air. For ZIRLO, these compressive stresses are fairly constant
at 1300–1400 MPa over the entire 14-h exposure time. However, for the A-0.0Sn
alloy, the monoclinic stresses are significantly lower and relax significantly as oxida-
tion proceeds, from around 900 MPa initially to about 600 MPa after 14 h exposure.
For the stresses measured in the tetragonal phase for the two alloys, Fig. 12, the
trend is more similar to that seen for autoclave tested samples, with significantly
FIG. 12 The evolution of residual stresses in the tetragonal phase of the oxides formed
during the oxidation of ZIRLO and A-0.0 %Sn at 550 C in air for 14 h.
lower compressive stresses for A-0.0Sn (0–1500 MPa) than for ZIRLO
(1500–3500 MPa). As with the ex situ measurements, the relatively low intensity of
the tetragonal peak results in a high degree of scatter. However, the large number of
data points recorded demonstrates that there is a clear difference between the stress
state of the tetragonal phase formed on these two alloys. Both alloys appear to show
a gradual relaxation of the compressive stresses in the tetragonal phase over the
course of the experiment.
Discussion
It should be noted that varying the tin content of an alloy does not necessarily result
in a change in the tin level of the resulting oxide. However, recent 3-D atom probe
analysis has confirmed that during corrosion of Zr alloys, the tin/zirconium ratio is
similar in the metal and oxide (i.e., tin present in the metal passes into the oxide
without partitioning as corrosion progresses) [48,50]. This is a very important ob-
servation as it shows that differences in tin levels between alloys will be carried over
into the oxide, where they can have an influence on the relative stabilities of the ox-
ide phases.
TIN CONTENT AND CORROSION

The improved resistance to aqueous corrosion achieved in Zr alloys by reducing the
tin content has been previously demonstrated for Zircaloy-4 [31] and advanced
alloys like Westinghouse ZIRLO [32,33]. The corrosion kinetics recorded for the
three Zr-Nb-Sn alloys discussed here also confirm this trend [40]. Despite very sim-
ilar pre-transition corrosion kinetics, a significant delay in the onset of transition
was observed with reduced tin content. The ZIRLO showing signs of transition
(defined by the acceleration of corrosion kinetics and the appearance of grey
patches on the oxide surface) after only 140 days, whereas for the lowest tin alloy
(A-0.0Sn) transition was only reached after more than 360 days.
It should be noted that despite these apparent benefits of reducing tin levels, it
is also well known that the addition of tin is advantageous for in-reactor perform-
ance, providing robustness and predictability of the corrosion behaviour in high
lithium and abnormal chemistry conditions [34]. A better understanding of the
mechanism by which tin affects corrosion would allow these competing factors to
be more effectively balanced.
THE IMPORTANCE OF TETRAGONAL PHASE FRACTION

Understanding the structure of the oxide that forms on Zr alloys is vital to deter-
mining the processes that control the inward diffusion of oxygen ions across the
oxide film and thereby define the pre-transition corrosion kinetics [12]. It remains
unclear whether diffusion of oxygen along crystallite boundaries or bulk diffusion
through a perfect oxide are the dominant factors [51]. However, it is generally
accepted that the relationship between the two oxide phases may play an
important role. The tetragonal phase has often been reported to accumulate near
the oxidation progression front, the oxide/metal interface. This observation has
often led to suggestions that the tetragonal phase acts as a barrier layer for oxygen
diffusion [10,16,19,21,22,52]. Consequently, the improved corrosion resistance
observed with reduced tin content for a range of Zircaloy-4 type alloys was attrib-
uted by Takeda and Anada [37] to a degradation of the tetragonal phase. Their
laboratory-based x-ray measurements, however, suggested that the tetragonal
phase fraction increases as the tin content is reduced. According to the degrada-
tion model put forward, the tetragonal to monoclinic phase transformation is
accelerated by oxidation of the tin segregated within the oxide, resulting in
increased porosity. However, combining the facts that (i) the tetragonal phase
fraction was only determined for a single sample in each alloy (and considering
the experimental uncertainty when measuring the volume fraction of this phase
by laboratory x-ray diffraction), (ii) that in the very large number of published
TEM studies of oxides on Sn-containing zirconium alloys no unambiguous identi-
fication of tin oxide phases has been reported, and (iii) the general observation
that tin does in fact stabilise the tetragonal phase, it seems unlikely that this model
is correct. One of the key findings of the current work has been that the tetrago-
nal/monoclinic ratio is strongly depending on the thickness of the oxide, which
demonstrates that comparisons between the tetragonal phase fractions of different
alloys are only meaningful when multiple measurements are conducted as a func-
tion of corrosion weight gain.
EVALUATING TETRAGONAL PHASE FRACTION

Evaluating differences in the evolution of tetragonal phase fraction between differ-
ent alloys can be challenging even when comparing trends for multiple measure-
ments over a range of corrosion weight gains. For the measurement technique used
here, the tetragonal fraction obtained represents an average over the entire oxide, so
from a single measurement it is impossible to distinguish between tetragonal phase
that is evenly distributed over the entire oxide and that which may be concentrated
near the metal/oxide interface. The schematic shown in Fig. 9 illustrates three possi-
ble scenarios for the distribution of tetragonal phase as the oxide layer grows, along
with the evolution of tetragonal phase fraction that might be observed for each. The
schematic demonstrates that when the tetragonal phase is evenly distributed
throughout the oxide, a constant fraction would be measured regardless of oxide
thickness. Conversely, if most of the tetragonal phase was concentrated near the ox-
ide/metal interface, as has been often suggested, the measured tetragonal fraction
would fall with increasing oxide thickness, even though the total volume of tetrago-
nal phase present may remain constant. It is important to consider these effects
when evaluating measurements of tetragonal phase fraction.
Measurements on the autoclave tested samples show that the oxide formed on
ZIRLO contains in the early stages of oxidation the highest tetragonal phase
fraction. This fraction falls significantly as corrosion proceeds (Fig. 8(a)), suggesting
that the tetragonal phase is not uniformly distributed over the oxide. This is further
confirmed by the near horizontal trend calculated for equivalent tetragonal layer
thickness shown in Fig. 8(b). Comparing the evolution of tetragonal phase fraction
for the three alloys in Fig. 8, it is evident that as the tin content is reduced the initial
tetragonal/monoclinic ratio also decreases. Furthermore, the drop in tetragonal
phase fraction with overall weight gain (i.e., the gradient in Fig. 8) also reduces with
decreasing tin. For the lowest tin alloy, A-0.0Sn, the tetragonal phase fraction
appears to be nearly constant with weight gain, suggesting that for this alloy the tet-
ragonal phase approaches a uniform distribution throughout the oxide. The data
from in situ oxidation experiments (Fig. 10) confirms the trends observed for
autoclave-tested samples, with the oxide formed on ZIRLO showing a much higher
tetragonal phase fraction than that on A-0.0Sn. The latter stages of the experiment,
i.e., after 6 h oxidation, also suggest a continued gradual decrease in the tetragonal
phase fraction for ZIRLO, whereas for A-0.0Sn a plateau seems to have been
reached. However, it is in the initial stage of oxidation where the in situ experiment
may be able to provide additional insight into the corrosion mechanisms. Very high
initial tetragonal phase fractions were recorded, even for the low tin alloy, which
reduced rapidly over the first 2 h of oxidation. This observation would suggest high
levels of phase transformation may be taking place in the early stages of oxidation.
Importantly, even at this early stage of corrosion, ZIRLO still showed significantly
higher tetragonal phase fraction.
To understand this apparent variation in tetragonal phase distribution, it is im-
portant to consider the mechanisms by which the tetragonal phase can be stabilised.
The mechanism most commonly used to explain the stabilisation of tetragonal
grains near the oxide/metal interface is that of high compressive stress
[16,24,53–58]. This compressive stress is related to the volume increase when Zr
transforms to ZrO2 (Pilling-Bedworth ratio is 1.56) [57,59]. However, this stress
has been shown to relax in the oxide away from the oxide/metal interface as corro-
sion progresses [24], therefore its stabilising effect is limited to the oxide directly ad-
jacent to the interface resulting in higher levels of tetragonal phase in this region.
If stress was the only factor affecting the stability of the tetragonal phase, it
would be expected that as corrosion progresses, and the stresses in the existing ox-
ide relax, all the tetragonal phase present would eventually transform to monoclinic
oxide. However, there is sufficient experimental evidence from measurements on
thick oxides [19], that even in regions with low stresses some tetragonal phase
remains present. For these regions it is believed that the grains of the tetragonal
phase are small enough to result in stabilisation without the need for an external
compressive stress. In the case of an Nb-free Zr alloy, it has been suggested that this
critical grain size is about 30 nm [27], as can readily be observed in TEM images
like that shown in Fig. 6(a).
In the case of ZIRLO and A-0.6Sn, stresses measured in the tetragonal phase
were generally between 1500 MPa and 2000 MPa, for both autoclave samples
and in situ oxidation (Fig. 7 and Fig. 11). Stress measurements of the tetragonal phase
in oxides on A-0.0Sn showed significantly more scatter, but generally the observed
stresses were similar to those for the monoclinic phase, i.e., less compressive than
the tetragonal stress values for the other alloys. This difference suggests that the ma-
jority of tetragonal phase in A-0.0Sn is not stabilised by stress but by grain size. It
should be noted that the high volume of data available from the in situ experiment
is important as it provides additional confidence that this observation is valid de-
spite the large scatter in the stress data for the tetragonal phase on A-0.0Sn.
The results suggest that in the presence of tin, tetragonal grains can be readily
stress-stabilised, but that when tin is below a critical level the stresses alone are not
sufficient to stabilise grains that exceed a certain critical grain size. The high initial
tetragonal phase fraction observed during in situ oxidation may be related to the
combination of a highly stressed oxide with the high proportion of small equiaxed
oxide grains present at this early stage of oxidation. The presence of a high propor-
tion of small equiaxed grains early on in oxidation has been confirmed for the auto-
clave tested ZIRLO samples by the TEM images in Figs. 5 and 6.
While the higher temperature air oxidation used for the in situ experiment
may not provide the ideal comparison to the autoclave exposures, it is encour-
aging that despite some differences in absolute tetragonal fraction and stress
levels, very similar trends are observed between different alloys for in situ and
ex situ oxidation. Further experiments are planned with oxidation to be car-
ried out in situ at temperatures more similar to those used for autoclave test-
ing, and the possibility of in situ autoclave exposures is currently being
investigated.
It should be noted that some differences are observed between the two experi-
ments in terms of the monoclinic residual stresses measured for each alloy. For
the lower temperature autoclave exposures, the monoclinic stresses are very simi-
lar for all three alloys. However, when oxidation was performed at 550 C in air,
the stresses measured in the monoclinic phase the A-0.0 %Sn oxide was signifi-
cantly lower. This reduced stress may be attributed to the significant difference in
creep properties of the two alloys at this temperature, because of the reduction in
the solid solution strengthening provided by tin. As the creep strength of the alloy
decreases, the level of constraint is reduced and less compressive stress can be
maintained in the oxide. However, the difference in residual stresses for the tetrag-
onal phase between the two alloys appears to be independent of oxidation temper-
ature, and is therefore not expected to be related to the mechanical properties of
the alloys.
As discussed earlier, accurate elastic properties are not available for tetragonal
zirconia at the temperatures studied here. For this reason, monoclinic elastic prop-
erties have been applied to both phases. This may introduce an added uncertainty
with regard to the absolute magnitude of the stresses present in the tetragonal
phase. However, because the current work is concerned with how the stress state
changes over time for the oxide phases of each alloy and how this relates to the
FIG. 13 Schematic description of corrosion mechanisms in material with higher tin

content and material with lower tin content. The proposed extreme cases are
presented. The amount of tetragonal and monoclinic grains is only a schematic,
not a quantitative representation.
tetragonal phase fraction, the lack of precise elastic properties should not impact
greatly on the findings presented.
PROPOSED MODEL FOR TRANSITION

Based on the current observations and discussions, we propose an updated model
to describe the oxidation of Zr-Nb-Sn alloys, illustrating one factor that contributes
to the transition from slow to accelerated growth kinetics, Fig. 13.
When ZrO2 first nucleates at the metal surface, a large proportion of new oxide
grains may be tetragonal, stabilised by a combination of their small grain size, the
large compressive stress present, and by any tetragonal stabilising elements present
in the alloy. Those grains that are suitably oriented will begin to grow epitaxially as
corrosion proceeds. If these grains are tetragonal, and a tetragonal stabiliser such as
tin is present, they may continue to grow while under the influence of a compres-
sive stress. These grains will, however, transform, to monoclinic oxide if the com-
pressive stress is relaxed over time. As a result, stress-stabilised tetragonal grains
will only maintain their crystal structure near the interface, resulting in a layer with
high tetragonal/monoclinic fraction as shown in Figs. 9(b)–9(c). Tetragonal grains
that do not have the correct orientation will not grow, and may stay below a critical
grain size that enables them to remain tetragonal in the absence of a compressive
stress. These size-stabilised tetragonal grains may therefore be uniformly distributed
throughout an oxide. Oxides with higher levels of phase transformation may be
subject to more extensive crack formation according to the shear model previously
described by a number of researchers [3,26,60].
During the pre-transition phase of corrosion, the rate of oxide growth slows
down over time, so the rate at which new, highly-stressed oxide forms also
decreases. However, relaxation mechanisms may still be acting on the oxide. The
combination of these factors would result in a gradual reduction in the thickness of
the region where the oxide is highly stressed and where the tetragonal phase will
remains stable. Eventually, the damage in the oxide caused by phase transformation
will occur close enough to the interface to produce a breakdown in the integrity of
the oxide, initiating the observed transition in oxidation kinetics.
A critical aspect of this model is the distinction between grain size and stress-
stabilised tetragonal grains, and that the Zr alloy chemistry affects the volume frac-
tion of tetragonal grains that can be stress stabilised and can transform once the
stress has decreased. The epitaxial growth of tetragonal grains is rare in material
with lower tin content, as transformation would occur as soon as the grain becomes
enlarged. When the interface moves away from the previously formed oxide, stress-
stabilised tetragonal grains may still transform into monoclinic oxide, but the total
volume of material transforming would be much smaller than in material with
higher tin content. The reduced amount of transformation brings about slower for-
mation of a smaller volume of cracks (Fig. 4(f), and subsequently a longer pre-
transition period (Fig. 3(c)). For the three alloys investigated in the present work,
i.e., Nb-containing Zr alloys, the proposed model seems to be responsible for the
different time it takes for transition to occur. However, it remains to be confirmed
that this model is also valid for other alloy systems.
It has not yet been possible to directly relate the level of tetragonal/monoclinic
phase transformation to observations of oxide breakdown in the different alloys.
Transmission electron microscopy studies (TEM) may be a possible way to attempt
this, and observations on ZIRLO have already suggested the presence of nano-
porosity at oxide grain boundaries, which appears to become more linked up nearer
to the interface as transition is approached [61,62]. However, such investigations
will only be able to evaluate the differences in how damage (i.e., porosity) evolves
over time for different tin levels. It is not possible to accurately determine the pres-
ence of stress-stabilised tetragonal grains by TEM because the preparation of the
thin foil samples needed for TEM analysis would necessarily result in the relaxation
of stresses in the oxide and thereby transformation of any such tetragonal grains.
Other diffraction-based studies have been able to successfully determine the distri-
bution of the tetragonal phase within the oxide by studying cross-sectional samples
[16,63]. This technique only interrogates material near to the surface of cross-
sectional specimens or through very thin cross-sections, and therefore may not
always represent the stress state and phase distribution present in the bulk oxide,
because of the effect of sample preparation on the constraint present. However, the
results from these studies are in good agreement with the model proposed here, as
they suggest that oxides with higher levels of tetragonal phase near to the oxide/
metal interface are likely to reach transition earlier. The current work reinforces
these findings by comparing the evolution of tetragonal phase for each alloy over a
range of exposure times.
It should also be noted that tin may also affect the transport of hydrogen in zir-
conium alloys during oxidation, thereby impacting the corrosion rate after extended
exposure. Initial observations for samples cathodically charged to induce a hydride
rim before corrosion testing, suggest that alloys with low tin levels may experience
some outward depletion of their hydride rim through the oxide layer during corro-
sion [40]. Further tests to investigate this phenomenon are underway, tracking the
movement of hydrogen during corrosion by performing NanoSIMS on selected
specimen partially corroded in isotopically spiked water.
Summary
The mechanisms of oxide formation on three recrystallized Zr-Sn-Nb alloys with
different tin contents (ZIRLO, A-0.6Sn, and A-0.0Sn) have been investigated by
performing aqueous corrosion at 360 C and 18 bar in simulated primary water and
by in situ oxidation in air at 550 C. The results shed new light on the oxidation
mechanism that leads to transition of the corrosion rate. The following observations
have been made:
1. Corrosion resistance during autoclave exposure improves significantly with
decreasing tin content.
2. Compressive stresses in the monoclinic oxide phase were within the range of
700 to 1000 MPa in the case of ZIRLO and A-0.6Sn, whereas for A-0.0Sn
lower stresses of 500 to 900 MPa were measured. Compressive stresses in
tetragonal phase were 1500 to 2000 MPa in both ZIRLO and A-0.6Sn. A-
0.0Sn displayed stresses in the tetragonal phase similar to the stresses in the
monoclinic phase.
3. Oxidation in air at 550 C produced slightly different stresses, but the trends
between alloys remained the same. Monoclinic stresses were within the range
of 1200 to 1500 MPa for ZIRLO and 600 to 1000 MPa for A-0.0Sn,
whereas the tetragonal stresses were 1500 to 3000 MPa for ZIRLO and
1500 to 0 for A-0.0Sn.
4. Tetragonal phase fractions and equivalent layer thickness of the tetragonal
phase were calculated for autoclave-tested samples. It was possible to demon-
strate that tin contributes to the stabilisation of the tetragonal phase. It was
also inferred that the rate of tetragonal to monoclinic phase transformation
reduces with decreasing tin content.
5. In situ synchrotron x-ray diffraction measurements during oxidation pro-
vided, for the first time, information about mechanisms at work during early
oxidation. These measurements demonstrated that, regardless of tin content,
elevated levels of tetragonal phase are present when oxidation first initiates,
but that these levels are significantly higher for ZIRLO than for A-0.0Sn.
6. A model explaining the corrosion mechanisms leading to the rate-transition
phenomenon has been proposed highlighting the distinction between grain
size-stabilised and stress-stabilised tetragonal grains. It assumes that stress sta-
bilisation of tetragonal oxide phase is only possible in combination with some
chemical stabilisation, which in the present case is by tin.
7. The model identifies the stress-stabilised tetragonal phase as detrimental to
the corrosion performance because of stress relaxation in the oxide-enabling
tetragonal to monoclinic phase transformation with an associated shear strain
and volume expansion.
8. The observation of stress-stabilised tetragonal phase is only possible via a bulk
technique, where the material constraint is maintained. Therefore, the pres-
ence of such grains cannot be confirmed by TEM analysis.
ACKNOWLEDGMENTS
The writers thank the Engineering and Physical Science Research Council (EPSRC)
and MoD in the U.K. for funding the research as part of the Materials for Energy call
(research code: EP/E036171/1). The project would not have been possible without the
strong support of our industrial project partners (in alphabetical order) EDF, National
Nuclear Laboratory, Rolls-Royce, Serco, and Westinghouse, who have provided sub-
stantial in-kind and financial support, and our academic project partners from the
Open University. The provision of synchrotron beam-time at the Bessy II facility (EU
proposal 2010_1_90919) and Diamond Light Source (EE7900) is gratefully acknowl-
edged, as is the help provided by the beamline scientists at the EDDI instrument and
I11, in particular that of Manuela Klaus, Christina Reinhard, and Richard Atwood.
References
[1] Arima, T., Miyata, K., Inagaki, Y., and Idemitsu, K., “Oxidation Properties of Zr-Nb Alloys
at 500-600 C under Low Oxygen Potentials,” Corros. Sci., Vol. 47(2), 2005, pp.
435–446.
[2] Cox, B., “Comments on ‘Aqueous Corrosion of the Zircaloys at Low Temperatures.’” J.
Nucl. Mater., Vol. 30(3), 1969, pp. 351–352.
[3] Cox, B., “Some Thoughts on the Mechanisms of in-Reactor Corrosion of Zirconium
Alloys,” J. Nucl. Mater., Vol. 336(2–3), 2005, pp. 331–368.
[4] Motta, A. T., Gomes da Silva, M. J., Yilmazbayhan, A., Comstock, R. J., Cai, Z., and Lai, B.,
“Micostructural Characterisation of Oxides Formed on Model Zr Alloys Using Synchro-
tron Radiation,” 15th International Symposium on Zirconium in the Nuclear Industry, B.
Kammenzind and M. Limback, Eds., ASTM International Sunriver Resort, Oregon, June
24–28, 2007, pp. 486–506.
[5] Sabol, G. P. and Dalgaard, S. B., “The Origin of the Cubic Rate Law in Zirconium Alloy
Oxidation,” J. Electrochem. Soc., Vol. 122(2), 1975, pp. 316–317.
[6] Bossis, P., Pecheur, D., Hanifi, L., Thomazet, J., and Blat, M., “Comparison of the High
Burn-up Corrosion on M5 and Low Tin Zircaloy-4,” 14th International Symposium on Zir-
conium in the Nuclear Industry, P. Rudling and B. Kammenzind, Eds., ASTM International,
Stockholm, Sweden, 2005, pp. 494–525.
[7] Bouvier, P., Godlewski, J., and Lucazeau, G., “A Raman Study of the Nanocrystallite Size
Effect on the Pressure-Temperature Phase Diagram of Zirconia Grown by Zirconium-
Based Alloys Oxidation,” J. Nucl. Mater., Vol. 300(2–3), 2002, pp. 118–126.
[8] Bryner, J. S., “The Cyclic Nature of Corrosion of Zircaloy-4 in 633 K Water,” J. Nucl.
Mater., Vol. 82(1), 1979, pp. 84–101.
[9] Garde, A. M., “Enhancement of Aqueous Corrosion of Zircaloy-4 due to Hydride Precipi-
tation at the Metal–Oxide Interface,” 9th International Symposium on Zirconium in the
Nuclear Industry, C. M. Eucken and A. M. Garde, Eds., ASTM International, Kobe, Japan,
1991, pp. 566–592.
[10] Godlewski, J., “How the Tetragonal Zirconia Is Stabilized in the Oxide Scale that Is
Formed on a Zirconium Alloy Corroded at 400 C in Steam,” 10th International Sympo-
sium on Zirconium in the Nuclear Industry, A. M. Garde and E. R. Bradley, Eds., ASTM
International, Baltimore, MD, June 21–24, 1993, pp. 663–683.
[11] Griggs, B., Maffei, H. P., and Shannon, D. W., “Multiple Rate Transitions in the Aqueous
Corrosion of Zircaloy,” J. Electrochem. Soc., Vol. 109(8), 1962, pp. 665–668.
[12] Hauffe, K., Oxidation of Metals, Plenum, New York, 1965.
[13] Kass, S., “The Development of the Zircaloys, Corrosion of Zirconium Alloys,” ANS Winter
Meeting, ASTM International, West Conshohocken, PA, 1964.
[14] Lustman, B. and Kerze, F., The Metallurgy of Zirconium, 1st ed., National Nuclear
Energy Series, Vol. 4, McGraw-Hill, New York, 1955, p. 776.
[15] Peters, H. R., “Improved Characterization of Aqueous Corrosion Kinetics of Zircaloy-4,”

6th International Symposium on Zirconium in the Nuclear Industry, D. G. Franklin and R. B.
Adamson, Eds., ASTM International, Vancouver, British Columbia, Canada, 1984, pp.
507–519.
of Zirconium Alloy Oxides Formed in Pure Water Studied with Synchrotron Radiation
and Optical Microscopy: Relation to Corrosion Rate,” J. Nucl. Mater., Vol. 324(1), 2004,
pp. 6–22.
[17] Roy, C. and Burgess, B., “A Study of the Stresses Generated in Zirconia Films During the
Oxidation of Zirconium Alloys,” Oxid. Met., Vol. 2(3), 1970, pp. 235–261.
[18] Block, S., Da Jornada, J. A. H., and Piermarini, G. J., “Pressure-Temperature Phase Dia-
gram of Zirconia,” J. Am. Ceram. Soc., Vol. 68(9), 1985, pp. 497–499.
[19] Preuss, M., Frankel, P., Lozano-Perez, S., Hudson, D., Ni, N., Wei, J., English, C., Storer, S.,
Chong, K. B., Fitzpatrick, M., Wang, F., Smith, J., Grosvenor, C., Smith, G. D. W., Sykes, J.,
Cerezo, A., Cottis, B., Lyon, S., Hallstadius, L., Comstock, B., Ambard, A., and Blat-Yrieix,
M., “Towards a Mechanistic Understanding of Corrosion Mechanisms in Zirconium
Alloys,” 16th International Symposium on Zirconium in the Nuclear Industry, ASTM Inter-
national, ChengDu, People’s Republic of China, 2010.
[20] Zhilyaev, A. P. and Szpunar, J. A., “Influence of Stress Developed due to Oxide Layer For-
mation on the Oxidation Kinetics of Zr-2.5 %Nb Alloy,” J. Nucl. Mater., Vol. 264(3), 1999,
pp. 327–332.
[21] Cox, B., Are Zirconium Corrosion Films a Form of Partially Stablized Zirconia? Atomic
Energy of Canada, Reactor Materials Division, Chalk River Nuclear Laboratories, Chalk
River, Ontario 1987.
[22] Godlewski, J., Bouvier, P., Lucazeau, G., and Fayette, L., “Stress Distribution Measured
by Raman Spectroscopy in Zirconia Films Formed by Oxidation of Zr-Based Alloys,”
12th International Symposium on Zirconium in the Nuclear Industry, G. P. Sabol and G.
D. Moan, Eds., ASTM International, Toronto, Canada, 2000, pp. 877–899.
[23] Blat-Yrieix, M., Ambard, A., Foct, F., Miquet, A., Beguin, S., and Cayet, N., “Toward a Bet-
ter Understanding of Dimensional Changes in Zircaloy-4: What Is the Impact Induced by
Hydrides and Oxide Layer?” J. ASTM Int., Vol. 5(9), 2008, p. 16.
[24] Polatidis, E., Frankel, P., Wei, J., Klaus, M., Comstock, R. J., Ambard, A., Lyon, S., Cottis,
R. A., and Preuss, M., “Residual Stresses and Tetragonal Phase Fraction Characterisation
of Corrosion Tested Zircaloy-4 Using Energy Dispersive Synchrotron X-Ray Diffraction,”
J. Nucl. Mater., Vol. 432(1–3), 2013, pp. 102–112.
[25] Heuer, A. H. and Rühle, M., “Overview No. 45: On the Nucleation of the Martensitic Trans-
formation in Zirconia (ZrO2),” Acta Metall., Vol. 33(12), 1985, pp. 2101–2112.
[26] Cox, B., Kritsky, V. G., Lemaignan, C., Polley, V., and Ritchie, I. G., Waterside Corrosion of
Zirconium Alloys in Nuclear Power Plants, IAEA, Vienna, 1998.
[27] Barberis, P., “Zirconia Powders and Zircaloy Oxide Films: Tetragonal Phase Evolution
during 400 C Autoclave Tests,” J. Nucl. Mater., Vol. 226(1–2), 1995, pp. 34–43.
[28] Bouvier, P. and Lucazeau, G., “Raman Spectra and Vibrational Analysis of Nanometric
Tetragonal Zirconia Under High Pressure,” J. Phys. Chem. Solids, Vol. 61(4), 2000, pp.
569–578.
[29] Djurado, E., Bouvier, P., and Lucazeau, G., “Crystallite Size Effect on the Tetragonal-
Monoclinic Transition of Undoped Nanocrystalline Zirconia Studied by XRD and Raman
Spectrometry,” J. Solid State Chem., Vol. 149(2), 2000, pp. 399–407.
[30] Garvie, R. C., “The Occurrence of Metastable Tetragonal Zirconia as a Crystallite Size
Effect,” J. Phys. Chem., Vol. 69(4), 1965, pp. 1238–1243.
[31] Garde, A. M., Pat, S. R., Krammen, M. A., Smith, G. P., and Endter, R. K., “Corrosion Behavior
of Zircaloy-4 Cladding with Varying Tin Content in High-Temperature Pressurized Water
Reactors,” 10th International Symposium on Zirconium in the Nuclear Industry, A. M. Garde
and E. R. Bradley, Eds., ASTM International, Baltimore, MD, 1994, pp. 760–778.
[32] Yueh, H. K., Kesterson, R. L., Comstock, R. J., Shah, H. H., Colburn, D. J., Dahlback, M., and
Hallstadius, L., “Improved ZIRLO (TM) Cladding Performance Through Chemistry and Pro-
cess Modifications,” Zirconium in the Nuclear Industry: 14th International Symposium, P. K.
B. Rudling, Ed., ASTM International, Stockholm, Sweden, June 13–17, 2004, pp. 330–346.
[33] Yueh, H. K., Kesterson, R. L., Comstock, R. J., Shah, H. H., Colburn, D. J., Dahlback, M.,
and Hallstadius, L., “Cladding Optimization for Enhanced Performance Margins,” TOP-
FUEL 2006, Salamanca, Spain, 2006, p. 67.
[34] Sabol, G. P., Kilp, G. R., Balfour, M. G., and Roberts, E., Development of a Cladding Alloy
for High Burnup, ASTM STP 1023, 1989, pp. 227–244.
[35] Chabretou, V., Hoffmann, P. B., Trapp-Pritsching, S., Garner, G., Barberis, P., Rebeyrolle, V.,
and Vermoyal J. J., “Ultra Low Tin Quaternary Alloys PWR Performance—Impact of Tin
Content on Corrosion Resistance, Irradiation Growth, and Mechanical Properties,” J. ASTM
Int., Vol. 8(5), 2011.
[36] Beie, H.-J., Garzarolli, F., Mitwalsky, A., Ruhmann, H., and Sell, H., “Examinations of the
Corrosion Mechanism of Zirconium Alloys,” 10th International Symposium on Zirconium
in the Nuclear Industry, A. M. Garde and E. R. Bradley, Eds., ASTM International, Balti-
more, MD, June 21–24, 1993, pp. 615–643.
[37] Takeda, K. and Anada, H., “Mechanism of Corrosion Rate Degradation due to
Tin,” 12th International Symposium on Zirconium in the Nuclear Industry, G. P.
Sabol and G. D. Moan, Eds., ASTM International, Toronto, Canada, 2000, pp.
592–608.
[38] Kim, D.-J., Jang, J.-W., and Lee, H.-L., “Effect of Tetravalent Dopants on Raman Spectra
of Tetragonal Zirconia,” J. Am. Ceram. Soc., Vol. 80(6), 1997, pp. 1453–1461.
[39] Li, P., Chen, I. W., and Penner-Hahn, J. E., “Effect of Dopants on Zirconia Stabilization—
An X-Ray Absorption Study. II: Tetravalent Dopants,” J. Am. Ceram. Soc., Vol. 77(5),
1994, pp. 1281–1288.
[40] Wei, J., Frankel, P., Blat, M., Ambard, A., Comstock, R. J., Hallstadius, L., Lyon, S., Cottis, R.
A., and Preuss, M., “Autoclave Study of Zirconium Alloys With and Without Hydride Rim,”
Corros. Eng., Sci. Technol., Vol. 47(7), 2012, pp. 516–528.
[41] Frankel, P., Wei, J., Lyon, S., Cottis, R. A., Preuss, M., Ambard, A., Comstock, R. J., and
Hallstadius, L., “Understanding the Importance of Sn Content for the Corrosion Behav-
iour of Zr-Nb-Sn Alloys for Cladding Material,” Top Fuel 2012, European Nuclear Society,
Manchester, 2012.
[42] Genzel, Ch., Denks, I. A., Gibmeier, J., Klaus, M., and Wagener, G., “The Materials Science
Synchrotron Beamline EDDI for Energy-Dispersive Diffraction Analysis,” Nucl. Instrum.
Methods Phys. Res. Sect. A: Accelerators, Spectrometers, Detectors Associated Equip-
ment, Vol. 578(1), 2007, pp. 23–33.
[43] Withers, P. J., Preuss, M., Steuwer, A., and Pang, J. W. L., “Methods for Obtaining the
Strain-Free Lattice Parameter When Using Diffraction to Determine Residual Stress,” J.
Appl. Crystallogr., Vol. 40(5), 2007, pp. 891–904.
[44] Northwood, D. O., London, I. M., and Bähen, L. E., “Elastic Constants of Zirconium
Alloys,” J. Nucl. Mater., Vol. 55(3), 1975, pp. 299–310.
[45] Tomé, C. N. and Oliver, E. C., Computer Program: Elasto-Plastic Self-Consistent (EPSC)
Model, Los Alamos National Laboratory, Los Alamos, NM, 2010.
[46] Hillner, E., Franklin, D. G., and Smee, J. D., “Long-Term Corrosion of Zircaloy Before and
After Irradiation,” J. Nucl. Mater., Vol. 278(2–3), 2000, pp. 334–345.
croscopy Examination of Oxide Layers Formed on Zr Alloys,” J. Nucl. Mater., Vol. 349(3),
2006, pp. 265–281.
[48] Wei, J., Frankel, P., Polatidis, E., Blat, M., Ambard, A., Comstock, R. J., Hallstadius, L., Hud-
son, D., Smith, G. D. W., Grosvenor, C., Klaus, M., Lyon, S., Cottis, R. A., and Preuss, M.,
“The Effect of Sn on Autoclave Corrosion Performance and Corrosion Mechanisms in Zr-
Sn-Nb Alloys,” Acta Mater., Vol. 61, 2013, pp. 4200–4214.
[49] Garvie, R. C. and Nicholson, P. S., “Phase Analysis in Zirconia Systems,” J. Am. Ceram.
Soc., Vol. 55(6), 1972, pp. 303–305.
[50] Hudson, D., Cerezo, A., and Smith, G. D. W., “Zirconium Oxidation on the Atomic Scale,”
Ultramicroscopy, Vol. 109(5), 2009, pp. 667–671.
[51] Cox, B. and Pemsler, J. P., “Diffusion of Oxygen in Growing Zirconia Films,” J. Nucl.
Mater., Vol. 28(1), 1968, pp. 73–78.
[52] Schwartz, C. M., Vaughan, D. A., and Cocks, G. G., “Report BMI-793 Dec 17, 1952,” The
Metallurgy of Zirconium, B. Lustman and F. Kerze, Eds., McGraw-Hill, New York, 1955, p.
562.
during 400 C Autoclave Tests,” J. Nucl. Mater., Vol. 226, 1995, p. 34.
[54] Lin, J., Li, H., Nam, C., and Szpunar, J. A., “Analysis on Volume Fraction and Crystal Ori-
entation Relationship of Monoclinic and Tetragonal Oxide Grown on Zr-2.5Nb Alloy,” J.
Nucl. Mater., Vol. 334, 2004, p. 200.
[55] Lin, J., Li, H., Szpunar, J. A., Bordoni, R., Olmedo, A. M., Villegas, M., and Maroto, A. J. G.,
“Analysis of Zirconium Oxide Formed during Oxidation at 623 K on Zr–2.5Nb and Zirca-
loy-4,” Mater. Sci. Eng., Vol. 381, 2004, p. 104.
[56] Roy, C. and David, G., “X-Ray Diffraction Analyses of Zirconia Films on Zirconium and
Zircaloy-2,” J. Nucl. Mater., Vol. 37, 1970, p. 71.
[57] Gosmain, L., Valot, C., Ciosmak, D., and Sicardy, O., “Study of Stress Effects in the Oxida-
tion of Zircaloy-4,” Solid State Ionics, Vols. 141–142, 2001, pp. 633–640.
[58] Zhang, H. X., Fruchart, D., Hlil, E. K., Ortega, L., Li, Z. K., Zhang, J. J., Sun, J., and Zhou, L.,
“Crystal Structure, Corrosion Kinetics of New Zirconium Alloys and Residual Tress Analy-
sis of Oxide Films,” J. Nucl. Mater., Vol. 396, 2010, p. 65.
[59] Qin, W., Nam, C., Li, H. L., and Szpunar, J. A., “Tetragonal Phase Stability in ZrO2 Film
Formed on Zirconium Alloys and Its Effects on Corrosion Resistance,” Acta Mater., Vol.
55, 2007, p. 1695.
[60] Ploc, R. A. and Newcomb, S. B., “Microscopy of Oxidation 3,” Third International Confer-
ence on the Microscopy of Oxidation, Institute of Materials, Minerals and Mining, Trinity
Hall, Cambridge, 1996.
[61] Ni, N., Hudson, D., Wei, J., Wang, P., Lozano-Perez, S., Smith, G. D. W., Sykes, J. M., Yardley,
S. S., Moore, K. L., Lyon, S., Cottis, R., Preuss, M., and Grovenor, C. R. M., “How the Crystallog-
raphy and Nanoscale Chemistry of the Metal/Oxide Interface Develops during the Aqueous
Oxidation of Zirconium Cladding Alloys,” Acta Mater., Vol. 60(20), 2012, pp. 7132–7149.
[62] Ni, N., Lozano-Perez, S., Jenkins, M. L., English, C., Smith, G. D. W., Sykes, J. M., and
Grovenor, C. R. M., “Porosity in Oxides on Zirconium Fuel Cladding Alloys, and Its
Importance in Controlling Oxidation Rates,” Scripta Mater., Vol. 62(8), 2010, pp. 564–567.
[63] Yilmazbayhan, A., Motta, A. T., Comstock, R. J., Sabol, G. P., Lai, B., and Cai, Z.,
“Characterization of Oxides Formed on Model Zirconium Alloys in 360 C Water Using
Micro-Beam Synchrotron Radiation,” 12th International Conference on Environmental
Degradation of Materials in Nuclear Power System, TMS (The Minerals, Metals and Mate-
rials Society), Snowbird, UT, 2005.
DISCUSSION
Question from Philippe Bossis, CEA, France:—Was a difference in size of oxide
grains observed between the two alloys?
Authors’ Response:—At the time of writing, TEM analysis has only been com-
pleted for the ZIRLO material. Further analysis is planned for other oxides. In terms
of the diffraction data, it is possible to make inferences about the grain size of the
different oxides, as the width of diffraction peaks are related to the grains size.
However, for quantitative grain size analysis from diffraction data, careful calibra-
tion measurements would be needed to separate the grain size broadening from
other contributing factors. Furthermore, due to the geometry in which the measure-
ments were carried out, only the grain size parallel to the direction of the X-ray
beam would contribute to this broadening. This means that if a columnar grain
morphology exists in the oxides, as is usually suggested, only the widths of the col-
umns can be estimated by this technique. Despite these uncertainties, the data col-
lected suggests that the “diameter” of the monoclinic oxide columns is about 30%
greater for the A0.0% Sn alloy than for ZIRLO.
Question from N. Ramasubramanian, ECCATEC Inc. Canada:—Are the stress

values calculated ones from the synchrotron x-ray data?
Authors’ Response:—Yes, all of the stress data presented have been calculated
from the observed synchrotron x-ray peak shifts, and represent the stresses aver-
aged over the entire oxide thickness.
Questions from Arthur Motta Penn State University
Q1:—Although you measure a higher tetragonal fraction in Sn containing

ZrO2, the overall fraction remains less than 10%. Do you not think that
you would require a higher tetragonal fraction in order to affect the over-
all transition process over the whole protective layer?
Authors’ Response:—It is difficult to know what the critical phase fraction

would be to affect transition. It is important to remember that the phase fraction
shown here is averaged over the entire oxide thickness, there may be localized
regions where the tetragonal fraction is significantly higher. The TEM studies car-
ried out for ZIRLO, suggest that as transition is approached intergranular porosity
is observed closer to the interface. Furthermore it is proposed that one factor driv-
ing the interlinking of porosity is tetragonal to monoclinic phase transformation
and that once the interlinked porosity reaches the interface a pathway is provided
for faster corrosion and transition is initiated. This is a gradual process, with phase
transformation occurring continuously; therefore even if the fraction of tetragonal
phase measured at any time is not very high the level of transformation may still be
significant.
Q2:—What is the reason for the higher tetragonal fraction in the initial oxide in
the Sn containing alloy?
Authors’ Response:—The proposed model suggests that the presence of Sn

allows larger grains to remain stabilized as tetragonal phase. In the absence of Sn,
only tetragonal grains below a critical size can remain stable. In the early stages of
corrosion a high proportion of small grains is expected to be present, allowing a
higher tetragonal fraction. Whilst the presence of an initial smaller grain morphol-
ogy has been commonly observed in the past, it is true that this needs to be proven
by further TEM analysis of the current samples.
Q3:—What is the orientation of the stress you determine? If it is in-plane

stress, how would it lead to linking to porosity? Also, similar in-situ
observations of early formation of tetragonal phase in Zircaloy-4 and M5
had been made in the CEA, which agree with your results (J. L. Bechade).
Authors’ Response:—The measured stresses represent the in-plane stress aver-

aged over each of the phases. It is true that it might seem strange to have porosity
building up under a compressive stress field. However, the average stresses meas-
ured cannot describe the localized stress state that might be produced at the boun-
daries of a grain that has undergone the transformation from tetragonal to
monoclinic phase. This localized stress field is dependent on the misfit produced
during the transformation. Finite element modeling is ongoing to try and simulate
this complex effect and determine whether a scenario is possible where such a
transformation could lead to redistribution of existing grain boundary porosity in
such a way as to produce the interlinked porosity observed by TEM.
Questions from P. Barberis, CEZUS Research Center
Q1:—XRD allows one to measure strains. To convert them to stress, we need

to use elastic constants, which can be very different from monoclinic to
tetragonal phases, and also depend a lot on the crystallographic direction
(texture effect). How did you proceed?
Authors’ Response:—It is correct that the elastic constants chosen in order to

calculate stress have an important effect. For the monoclinic phase plane specific
elastic constants were calculated from single crystal values using Elasto-platic Self
Consistent (EPSC) modeling to give elastic constants representative of a polycrys-
talline material. Using this model, the effect of texture was also investigated. It was
found that the (-111) diffraction planes were among those least prone to
uncertainties due to the typical oxide texture (10%). During stress measurements
using the sin2? technique, it is also possible to see the oscillation in the measured d-
spacing as a function of tilt angle, and thereby estimate the effect of texture. The os-
cillation was similar for all samples and did not contribute significantly to the
results. Furthermore, it is important to bear in mind that the findings presented
here, as well as the proposed model proposed, do not rely on the absolute value of
residual stresses present in the oxides, but on the changes in these stresses relative
to the phase fractions measured. Therefore, factors such as uncertainties in the elas-
tic properties would not affect the findings. Similarly, the use of monoclinic elastic
properties for both phases, may well introduce an added uncertainty for the abso-
lute stresses measured for the tetragonal phase. However, this does not affect the
clear trends observed. Furthermore, while it may be possible to obtain elastic con-
stants for tetragonal ZrO2 at the temperatures of interest, such data would have
similar levels of uncertainty associated with them, since they would be obtained by
extrapolation from high temperature values, by using chemically stabilized tetrago-
nal or by theoretical calculation due to the instability of the phase under normal
conditions.
Q2:—You mentioned metal creep. Did you try to measure the overall sample
deformation during or after the oxidation? (see for example P. Barberi’s
et al., ASTM STP Sun River 2007, “CASTADIVA”) NEED PROPER
REFERENCE
Authors’ Response:—Overall sample deformation has not been measured in this

study. While this would be of interest, it was believed that the uncertainty associ-
ated with such measurements on the relatively small samples used would be too
large for useful conclusions.
Questions from Mirco Grosse, Karlsruhe Institute of Technology
Q1:—How do you determine the stress free lattice parameters? It strongly

depends on the stoichiometric ratio? This ratio changes through the oxide
thickness. In the sin2 method, the weight of the contribution of different
through thickness positions in the oxide to the measured XRD pattern
changes. It means that each is related to a different stress free lattice pa-
rameter. Please comment.
Authors’ Response:—For the thin oxides investigated here, a plane stress condi-
tion is assumed. In such a situation the sin2 method does not require a stress free
lattice parameter. Also, the measurements presented here only attempt to show the
average stresses over the entire oxide. Further work is planned to investigate the na-
ture of any stress gradients in the oxides. From our measurements, we have not
seen significant variations in stoichiometry for these oxides.
Q2:—At high temperatures (1000 C - 1400 C), we found a redistribution of

Sn. Do you observe a redistribution at 550 C?
Authors’ Response:—While we have not looked at the Sn distribution for in situ

experiments carried out at 550 C, atom probe measurements for oxides grown at
360 C have not shown any Sn redistribution.
Questions from K. Kapoor, NFC
Q1:—Did you observe any difference in the formation of cracks in the two
alloys?
Authors’ Response:—SEM images of each of the oxides are shown in Figure 4 of

the manuscript. Significantly less lateral cracks were observed for the oxide formed
on A-0.0%Sn, with the other alloys showing more cracking as the oxide thickened.
It should be noted that most of the larger cracks were not adjacent to the metal/ox-
ide interface, and would not be expected to provide a pathway for oxidizing species.
The fine scale porosity believed to be important for the proposed mechanism would
not be visible via SEM analysis. Further TEM studies are planned in order to make
comparisons of the morphology of nano-porosity for each alloy.
Q2:—How does the presence (nucleation and formation) of the cracks in the
oxide affect the stress present in the oxide?
Authors’ Response:—For the relatively thin oxides studied, it is not believed that
the lateral cracks have a significant effect on the average stress state in the oxides.
This is demonstrated by the relatively constant level of monoclinic stress observed
for all the oxides. On a finer scale, it is believed that larger cracks will potentially
lead to localized stress variations which can influence the stress state of adjacent
grains and also impact on the stability of tetragonal grains adjacent to such cracks.
Such localized variations would be very difficult to measure experimentally, but the
development of models to predict these stress fields may shed further light on their
influence.
Marc Tupin,1 Joel Hamann,2 Damien Cuisinier,2 Philippe

Bossis,2 Martine Blat,3 Antoine Ambard,3 Alain Miquet,4
Damien Kaczorowski,5 and François Jomard6
Understanding of Corrosion
Mechanisms of Zirconium Alloys
after Irradiation: Effect of Ion
Irradiation of the Oxide
Layers on the Corrosion Rate
Reference
Tupin, Marc, Hamann, Joel, Cuisinier, Damien, Bossis, Philippe, Blat, Martine, Ambard, Antoine,
Miquet, Alain, Kaczorowski, Damien, and Jomard, François, “Understanding of Corrosion
Mechanisms of Zirconium Alloys after Irradiation: Effect of Ion Irradiation of the Oxide
Layers on the Corrosion Rate,” Zirconium in the Nuclear Industry: 17th International
ABSTRACT
The irradiation damage in the fuel cladding material is mainly caused by the
neutron flux resulting from the fission reactions occurring in the fuel. From an
experimental point of view, the neutrons have the disadvantage to activate
materials by neutron capture rendering them difficult to handle. To avoid these
constraints inherent in the handling of radioactive material, the radiation effects
on the corrosion resistance of zirconium alloys can be studied by irradiating the
Manuscript received December 13, 2012; accepted for publication September 15, 2013; published online
September 19, 2014.
1
DEN, Section for Research on Irradiated Material, CEA/Saclay, 91191 Gif-sur-Yvette Cedex, France,
e-mail: marc.tupin@cea.fr
2
DEN, Section for Research on Irradiated Material, CEA/Saclay, 91191 Gif-sur-Yvette Cedex, France.
3
EDF, EDF R&D, Centre des Renardières, Ecuelles, 77818 Moret-sur-Loing Cedex, France.
4
EDF, EDF/SEPTEN, 69628 Villeurbanne Cedex, France.
5
AREVA, AREVA NP, SAS, Fuel Business Unit, 10 rue Juliette Récamier, 69456 Lyon Cedex 06, France.
6
CNRS UMR 8635, Groupe d’Etude de la Matière Condensée, 92195 Meudon Cedex, France.
7
Hyderabad, India.
Copyright V
TUPIN ET AL., DOI 10.1520/STP154320120199 439
materials with ions. A new experimental approach using ion irradiation was
performed in the Microscopy and Irradiation Damage Studies Laboratory of the
CEA in Saclay, with the aim to study more specifically the influence of the
irradiation damages in the oxide on the corrosion rate of the zirconium alloys.
This study was, moreover, focused on a particular distribution of defects in the
oxide layer, basically, localised close to the metal/oxide interface. From the
results of the irradiation of the metal/oxide interface, it was clearly shown that,
whatever the incident ion, the irradiation of the internal interface results in a
significant increase of the oxygen diffusion flux ratios between the most
irradiated Zircaloy-4 and the unirradiated one, whereas that of the oxide formed
on M5TM induces a big decrease of the oxygen diffusion flux in the film. These
effects are less marked with helium ions compared to protons (M5TM is a
trademark of AREVA NP registered in the United States and in other countries).
Finally, the oxide irradiation impact on the oxygen diffusion through the layer
could explain the corrosion acceleration factor observed on Zy4 during the first
cycles of irradiation, but cannot alone explain observed corrosion accelerations
under high burn-up conditions. The discussion on the oxide irradiation effects
puts forward the probable role of the residual charge left by ion implantation.
Keywords
irradiation, corrosion, diffusion
Introduction
EDF, AREVA, and CEA are continuously improving their knowledge of the behav-
iour of zirconium alloys in primary water under irradiation. In this framework,
the following work is focused on the effect of irradiation damage on corrosion.
This damage in the fuel cladding material is mainly because of the neutron flux
resulting from the fission reactions occurring in the fuel. From an experimental
point of view, the neutrons have the disadvantage to activate materials by neutron
capture rendering them difficult to handle. Work on these radioactive materials is
extremely expensive and very constraining owing to the difficulty in terms of se-
curity and safety to handle them. To avoid these constraints inherent in the radio-
active material handling, the radiation effects on the corrosion resistance of
zirconium alloys can alternatively be studied by irradiating the samples with ions.
For sufficiently weak energies of incident ions, the materials are not activated
under these operating conditions.
Very few academic studies of the irradiation impact on corrosion have been
carried out up to now. That is the reason why a research program on these effects
was performed at the CEA in Saclay with the aim to study more specifically the
influence of irradiation damage in the oxide on the corrosion rate of the zirco-
nium alloys. We were also interested in a particular distribution of defects in the
layer localised near the metal/oxide interface. After a short presentation of the
industrial problems, this paper is divided into four parts: the state of the art, the
experimental approach, the results, and the discussion.
In-Pile Corrosion Issues of Zirconium Alloys

Investigated in This Work and General
Experimental Method Chosen
In the core of the nuclear power plant, the fuel rods are exposed to high neutron,
electronic, and photonic fluxes, inducing consequent irradiation damage in the
cladding material.
In Refs 1 and 2, it has been shown that the neutron flux increases the in-reactor
corrosion kinetics of Zircaloy-4 in primary water by a factor of 2. This acceleration
might result from the irradiation damage generated into the oxide. That hypothesis
will be explored in the following work. Should it be validated, it will support the
modelling used in fuel performance codes.
From a corrosion resistance point of view, there exists, as illustrated in Fig. 1,
marked differences in corrosion behaviour between the Zircaloy-4 (Zy4) and M5TM
alloys in a PWR environment. In particular, beyond 35 GWd/t, the corrosion rate
of Zy4 increases drastically in a quasi-exponential way, whereas M5TM maintains a
nearly constant rate of oxide growth [3].
Numerous hypotheses are proposed in the literature to explain the acceleration,
for instance, the hydride precipitation in the matrix [1] below the metal/oxide inter-
face. It is possible that this acceleration, often designated as “high burn-up acceler-
ation,” could also be partially caused by the accumulation of the irradiation damage
in the cladding, both metal and/or oxide.
FIG. 1 Comparison of the corrosion kinetics of Zircaloy-4 and M5TM in pile [2].
This impact and, more particularly, this accelerating effect of irradiation on the
oxidation kinetics of Zircaloy-4 beyond a given burn-up were confirmed by com-
paring the kinetic curves obtained on claddings in core (orange curve) and in corro-
sion loop (red curve) at the same wall temperature (Fig. 2) [4]. The neutron
irradiation affects simultaneously the matrix, the oxide, and the corrosion medium
because of the water radiolysis. It is thus difficult, starting from the previous curves,
to identify the major cause of this acceleration. This work is specifically focused on
irradiation’s effect on zirconia.
The zirconia, being an n-type semiconductor with anion vacancies, grows inwards.
New cells of zirconia are thus formed at the metal/oxide interface. During the pre-
transition stage, oxidation of Zy4 can be possibly divided into five steps: adsorption (1)
and reduction of water (2) on the surface desorption of hydrogen to the external inter-
face (3), oxygen diffusion through a vacancy mechanism (4), and oxidation of zirco-
nium at the metal/oxide interface (5). Considering the Wagner’s model [5] that
assumes local electroneutrality and a very low diffusion coefficient of vacancies in com-
parison to that of the electrons, the oxidation rate can be written as follows (Eq 1):
(1) dX=dtaJvaDv ðDCv =XÞ ¼ Dv ðCvint Cvext Þ=X DðCvint =XÞ
where:
Jv ¼ the diffusion flux of the vacancies,
Dv ¼ the diffusion coefficient of the vacancies,
FIG. 2 Oxide thickness versus oxidation time, deduced from tests carried out at the
CEA for Zircaloy-4 oxidized under various conditions of corrosion: in autoclave
(without heat gradient, with two different temperatures); in-loop (with electric
heating in the cladding to simulate the heat flow); in-pile (with heat flow and
irradiation) [3].
Cv-int and Cv-ext ¼ the oxygen vacancy concentrations at the metal/oxide interface
and at the surface, respectively (this last term can be neglected because of the
extremely high value of the equilibrium constant of oxidation) [6],
X is the oxide thickness, and
a ¼ “proportional to.”
The corrosion rate is thus proportional to the vacancy concentration at the in-
ternal interface. Given Eq 1, the experimental approach was to increase the gradient
of oxygen vacancy concentration by irradiating the sample to create anionic defects
at the metal/oxide interface.
Before turning to the description of the experimental devices, a short review of
bibliographic results gained on irradiation effects is presented.
Irradiation Effect on Zirconia and Kinetic Impact

on the Corrosion Rate
GENERALITIES
Zirconia is an extremely stable oxide owing to the very high free enthalpy
of reaction in absolute value over a broad temperature range. At ambient
temperature, the stable allotropic phase is monoclinic. Beyond 1400 K, a martensitic
type transformation occurs from the monoclinic into the tetragonal form [7].
Last, the cubic structure of the fluorite type is stable at very high temperature
(>2500 K). The density of these phases increases with the symmetry of the crystal
lattice. Typically, the increase in volume during the transformation of tetragonal
into monoclinic zirconia is about 4 %. In addition, the differences in cell parame-
ters between these three crystallographic structures being relatively low, a very
weak displacement allows passage from one structure to another. Finally, there
are three dominating factors of stabilization of the cubic or tetragonal phase at
ambient temperature: the reduction of the grain size [8], the compressive stresses
[9,10], and the doping by trivalent chemical elements (Y3þ) [11] or divalent
(Ca2þ), i.e., a valence lower than that of zirconium and also a size higher than
that of this cation.
According to certain authors [12], the “thermodynamic” stability of tetragonal or
cubic zirconia doped by lower valence elements would result in an increase of oxygen
vacancy concentrations imposed by the electroneutrality in the matter.
FORMATION ENERGY OF IRRADIATION DEFECTS AND DISPLACEMENT

ENERGIES
The formation energy of vacancies decreases gradually with the vacancy ionization.
The formation energy of anti-Frenkel pairs decreases indeed from 7.3 to
5.4 eV when the two electrons trapped on the vacancy are delocalized on the inter-
stitial oxygen. Moreover, that varies slightly with the allotropic variety of zirconia
[12,13].
No direct measurement of these displacement energies has been performed in

the past. Values deduced from simulations to describe the evolution of the defect
production of the type Fþ (see the next paragraph), which corresponds to an ion-
ized vacancy, are, however, available. Costantini and Beuneu [14] found that these
energies for the anions and cations are, respectively, about 120 and 80 eV.
However, the values chosen during this study are those adopted for monoclinic
zirconia polycrystals by Simeone et al. [15]. All TRIM (transport of ion in matter)
simulations presented in this paper were so performed by taking displacement ener-
gies of 20 eV for Zr and 60 eV for O.
TYPES OF IRRADIATION DEFECTS

Only a few experimental techniques are capable of characterizing the defects pro-
duced by the irradiation at an atomic scale. Among them, electron paramagnetic
resonance (EPR) allows the identification of stable paramagnetic centers at ambient
temperatures [14,16]. This technique is generally used on model materials like
monocrystals of cubic zirconia stabilized with yttrium (9.5 %).
These studies highlight three types of paramagnetic centers: the Fþ centers
related to the production of ionized oxygen vacancies, the T centers corresponding
to defects of the Zr3þ type of trigonal geometry in terms of coordinance with the
oxygen anions close to this center, and the electron holes. The first type, Fþ centers,
would preferentially be produced by collision cascades, and their concentration (in
log–log scale) would grow almost linearly with the irradiation damage contrary to
the T centers, which are potentially generated more probably by the electronic exci-
tations [14,16].
IRRADIATION DEFECT STABILITY WITH TEMPERATURE

According to the work of Costantini and Beuneu [17], the Fþ centers corresponding
to ionized vacancies transform beyond 550 K. The T centers would only be restored
beyond 800 K. These fundamental studies are very useful to identify the dominating
defects produced by the irradiation and to know the evolution of their concentra-
tion with fluence, their stability as a function of temperature, and in particular
under conditions representative of PWR operation.
ELECTRONIC IRRADIATION EFFECT ON THE RESIDUAL CHARGE IN ZIRCONIA

In addition to the defects produced by the irradiation previously mentioned, the
irradiation will also induce a modification of the charge distribution in zirconia.
Thomé et al. [18] have measured the ratio between the current of secondary elec-
trons emitted by a monocrystal of stabilized zirconia, previously exposed or not to
1 MeV electron irradiation and the incident current of the electron beam using an
instrumented electron microscope. This study shows in particular that the irradia-
tion of zirconia by electrons of 1 MeV results in the accumulation of positive charge
on the surface generating an electric field directed towards the outside, which
reduces the flux of secondary electrons. Consequently, this ratio, whatever the
electron incident beam energy, is lower on irradiated material compared to the val-
ues obtained on an unirradiated one. The authors quantified the fields induced by
this accumulation of charge coming from the electron irradiation. At the surface,
this field directed towards the outside is about 103 V/m, whereas it reaches values of
about 0.5–1.106 V/m when the space charge zone extends only approximately 1 lm
under the surface. Unlike the internal field that disappeared above 600 K, the sur-
face field is still present until 800 K as the T centers. The presence of these field
induced by electron irradiation is likely to not only modify the rates of the interface
reactions but also the flux of oxygen diffusion in the layer.
ELECTRONIC INTERACTION EFFECT ON THE DAMAGE PRODUCTION

Irradiation with ions produces defects by electronic excitations that significantly
affect the surface of zirconia, unlike damages caused by collision cascades [19,20].
As illustrated in Fig. 3, this impact of ions results in the formation of protrusion on
the surface of about a few tens of nanometers (similar to volcanic eruptions), which
changes significantly the reactive area [20].
Second, the trace left inside the zirconia by high-energy incident ions can be
observed by TEM in high-resolution mode. These traces evidenced in Fig. 4 reveal a
local degradation of the crystalline lattice structure similar to an amorphisation, and
their diameter is strongly dependent on the electronic energy loss in the medium,
and thus on the interaction between the incident ion and matter [20]. And independ-
ently of the amorphisable character of the material, these traces would only be
observed for values of electronic stopping power (Se) higher than 10 keV/nm.
FIG. 3 TEM micrography of YSZ sample after irradiation with Pb ion at 940 MeV up to a
fluence of 5.1011 cm2 [20].
FIG. 4 Atomic force microscopy micrography of the surface of a zirconia sample

stabilized by yttrium after irradiation with Pb of 940 Mev up to a fluence of
1010 cm2 [20].
Finally, the damage at low fluences of irradiation grows linearly with the
received dose according to the results obtained by Rutherford backscattering
and channeling spectroscopy (RBS/C) and at high fluence tends to a satu-
rated damage level depending on the type of ion. The monoclinic zirconia
under irradiation can also be transformed to the tetragonal form.
ALLOTROPIC TRANSFORMATION UNDER IRRADIATION

Despite the fact that the stable phase at 573 K under atmospheric pressure is the
monoclinic form, the oxide formed on zirconium alloys is made up of a mixture of
monoclinic and tetragonal zirconia (between 5 % to 20 % of tetragonal zirconia
depending of the type of alloy) [21–23], because of the compressive stresses, the
small grain size and/or the doping elements present in the oxide, as already men-
tioned. In core, Gibert et al. [24] showed that the proportion of the tetragonal phase
was relatively homogeneous in the layer and more than 28 %, whereas, without irra-
diation, the tetragonal phase is essentially concentrated close to the metal/oxide
interface [25,26]. This microstructural evolution was reproduced by ionic irradia-
tion. Indeed, according to the works of Simeone et al. [15], the ion irradiation indu-
ces an allotropic transformation of the monoclinic phase into the tetragonal one,
beyond a certain dose. Figure 5 illustrates this process of phase transition, which
starts from around 0.5 dpa depending on the ion used and is complete around 12
dpa. This transformation observed after irradiation would be produced by the for-
mation of oxygen vacancies, according to Simeone et al. [7], which would generate
a sufficiently high local stress field to initiate this transformation.
In addition, considering that the fraction of tetragonal zirconia varies linearly
with the concentration of anion defects produced by irradiation, and taking into
account the fact that the defects of the type Fþ are eliminated beyond approxi-
mately 600 K, these authors, using a relatively simple kinetic model with only one
stage, manage to describe satisfactorily the kinetics of reverse transformation
FIG. 5 Normalized volume fraction of the tetragonal phase versus dpa—dark square for
Bi of 800 keV—cross for Xe of 400 keV [15].
occurring during isochronous annealings (30 min) at various temperatures as

shown in Fig. 6. However, according to the XRD (X-ray diffraction) patterns
obtained by these authors, it seems that a residual disorder remains up to 800 K as
for the residual charge mentioned in a previous paragraph. Last, the displacement
of the anions to interstitial positions is accompanied by the formation of electric
dipoles likely to modify the field locally and to change the energy of migration of
the anions. Concerning the extended defects, neither dislocation loop nor a cluster
of defects are observed on oxides with submicronic grain sizes. This behaviour is
similar to that of the zirconia films formed on a zirconium alloy under normal
operating conditions. This is because of the fact that the grain boundaries acting as
sinks, preventing the formation of broad clusters of defects.
Finally, the monoclinic zirconia under irradiation to even a relatively high dam-
age level does not tend to amorphise, as many materials do. Rather it appears to
adopt a more symmetrical structure, which is surprising knowing that the irradiation
produces some disorder at a nanometric scale because of the atomic displacement.
OXIDE IRRADIATION EFFECT ON THE DIFFUSION AND THE CORROSION RATE

As stated in the introduction, the corrosion rate of Zircaloy-4 under neutron irradi-
ation is twice that in an autoclave at similar temperature [1,2]. Bérerd et al. [27]
irradiated the Zircaloy-4 alloy in situ by a beam of 129Xe19þ (50 MeV and 2.61010
Xe.cm2.s1) during oxidation at 480 C and 280 C under a low partial pressure of
oxygen. In these conditions, the oxidation kinetics under irradiation are very close,
independent of the temperature, and linear as a function of time. Therefore, the
temperature does not influence the corrosion rate under irradiation by ions. The
FIG. 6 Normalized volume fraction of the tetragonal phase versus temperature for
isochronous thermal treatment (30 mn)—continuous line came from the model
used by the authors [15].
apparent diffusion coefficient of oxygen under irradiation was estimated at 3.1013

cm2/s at 480 C, approximately 10 times greater than that evaluated without irradia-
tion (about 3.1014) [28]; the difference was clearly higher at 280 C, typically 4
orders of magnitude between unirradiated material and that exposed to radiation.
In addition, these irradiated oxides contain very high proportions of tetragonal
phase, approximately 50 % at 480 C and about 100 % at 280 C, which is consistent
with the results obtained by Simeone et al. [15] at high fluence. It will thus be im-
portant to check thereafter if the irradiated oxide evolved or not after irradiation
from a microstructural point of view.
In conclusion, one should retain the following observations and established
phenomena as likely to influence the process of oxygen diffusion:
• Two main types of defects are produced during irradiation, the Fþ and T cen-
ters, annealed, respectively, beyond 600 K and 800 K.
• Electron irradiation induces the persistence of a residual field after interrup-
tion of the flux, which could modify the kinetic constants of the interface reac-
tions and diffusion fluxes of the ions in the oxide. Unlike the internal field that
disappeared behind 600 K, the surface field is still present up to 800 K as the T
centers.
• In the field of the electronic energy loss, the trace left by the particles results
initially in an atomic disorder localised around the trace and protrusions at
the surface.
• Irradiation of monoclinical zirconia by ions induces, beyond 0.5 dpa, a phase
transition to the tetragonal form related to the production of anion vacancies.
• The simultaneous irradiation of oxide and metal during the corrosion process
increases drastically the flux of oxygen diffusion through the layers.
• Irradiation fluxes also have an effect on the corrosion rate of zirconium alloys.
The following section describes the techniques used during this study and
the approach developed to obtain information from the experimental results.
Experimental Approach and Technical Tools

EXPERIMENTAL APPROACH
The approach adopted during this study is divided into six stages:
1. Preparation and oxidation of the samples: the samples are mirror polished
plates of recrystallized Zircaloy-4 (1.46 wt. %Sn-0.22 wt. %Fe-0.11 wt. %Cr-
0.12 wt. %O) and M5TM (1.03 wt. %Nb-0.035 wt. %Fe-0.005 wt. %Cr-0.13 wt.
%O) completed with OPS. These samples are corroded in static autoclaves at
633 K and 18.7 MPa in light primary water (H2O, 2 ppm Li, 1000 ppm B). The
goal of this pre-oxidation stage is to obtain approximately 1.5 lm of oxide on
the surface of the metallic materials.
2. Simulation of irradiation’s effect with the TRIM code. Before irradiation, the
simulation code TRIM is used to determine, first the type of ions, and, second
the incident energy of these ions to obtain the desired distribution of defects.
3. Ion irradiation of the samples. The samples are then irradiated with two types
of light ions, Hþ and Heþ, at various energies. The parameters of irradiation
and the distribution of the produced defects will be specified thereafter for
each irradiation.
4. Characterizations after irradiation: impedance spectroscopy and SIMS. To evalu-
ate the irradiation effect on the oxide properties, the samples have been character-
ized after irradiation by electrochemical impedance spectroscopy (EIS) to estimate
the variations of the electrical conductivity in the oxide layer and by secondary
ions mass spectrometry (SIMS) to determine the local implantation of the ions
and compare it to the distribution given by the TRIM code.
5. Isotopic exchange: re-oxidation in a medium H218O/D216O. The irradiated sam-
ples were corroded again at 360 C and under 18.7 MPa in a solution with lith-
ium (2 ppm) and boron (1000 ppm) containing 20 % of H218O and 80 % of
D2O in volume fraction. The time exposures of the isotopic exchange are rela-
tively short and do not exceed 24 h to limit the defect recombination. It is worth
noting that the heating time is around 45 min to reach 360 C. The cooling time
is of course much longer. After 1 h of cooling, the temperature is lower than
300 C and after 3 h, the temperature is below 200 C. Under these low tempera-
ture conditions, the oxygen diffusion rate is very slow and we can consider that
the diffusing species no longer move. Consequently, the oxygen penetration
depth through the oxide during the cooling time can be considered as negligible
compared to that during the exposure time in isothermal conditions.
6. SIMS analyses. The penetration profiles of oxygen 18 and deuterium in the
oxides are finally obtained by SIMS analyses (secondary ion mass spectrome-
try) to get the diffusion fluxes of these elements through the layers.
This general approach with several steps was followed whatever the desired dis-
tribution of defects or ion used for irradiation.
TECHNICAL TOOLS
Irradiation Facilities
The JANNUS (Joint Accelerators for Nanosciences and NUclear Simulation) plat-
form for multi-irradiation consists of two experimental facilities.
One is located at Orsay in the Center of Nuclear Spectrometry and Mass Spec-
trometry (CSNSM) and is attached to CNRS and IN2P3 of the Université Paris Sud.
These facilities are composed of a tandem accelerator of 2 MV (ARAMIS), another
one of 190 kV (IRMA) and a transmission electron microscope with a 200 kV accel-
eration voltage. Only the first one was used for this study.
The other facility is installed at the CEA-Saclay and is managed by the Service
for Research on Physical Metallurgy. Three electrostatic accelerators of ions can be
used simultaneously: a Pelletron type machine of 3 MV named EPIMETHEE
equipped with a source of multi-charged ions (ECR with Electron Cyclotron Reso-
nance), a simple stage VAN DE GRAAFF of 2.5 MV (YVETTE) coming from the
National Institute from Nuclear Sciences and Techniques (INSTN), and a tandem
one, called JAPET, of 2 MV, which completes this device. It is consequently possible
with this installation to carry out irradiations simultaneously with three beams of
ions and to simulate, in this way, the conditions of irradiation close to those met in
the core of the nuclear power plant. The beams of these three accelerators converge
actually towards the same target in an experimental room under ultra-high
vacuum (pressure ranging between 107 and 109 mbar), which can be heated
up to 1000 C or cooled until approximately the temperature of liquid
nitrogen (196 C). Strong rates of damage—up to 100 dpa in 1 day—are likely to
be reached with heavy ions produced by the 3 MV linear accelerator of EPIME-
THEE. The high voltage of acceleration multiplied by the charge confers to the
ions such a kinetic energy that a penetration of few microns in the solids can be
reached.
Two of the available accelerators were used during this study, ARAMIS in
Orsay and YVETTE at JANNUS Saclay. Figure 7 shows the experimental setup
used in the ARAMIS facility. Two plates of 25 mm length and 10 mm width
(10 25 mm2) were set side by side and a small part of the samples was abraded
to get good thermal contact between them and the thermocouple to check the
temperature. The irradiation area is indicated in Fig. 7 and is limited to the dark
part of the samples, i.e., the area coated by the oxide layer.
The following paragraphs describe the techniques used to characterise the
materials after irradiation and/or oxidation, EIS and SIMS.
ELECTROCHEMICAL IMPEDANCE SPECTROSCOPY (EIS)

Electrochemical impedance spectroscopy is an experimental technique providing
information on the electrical properties of the oxides formed on the surface of the
FIG. 7 Experimental irradiation device used in the ARAMIS accelerator.
zirconium alloys (thickness, resistivity, etc.). This technique consists of imposing a

sinusoidal voltage at various frequencies (f) (of from 103 Hz to 105 Hz) between
two electrodes, and measuring the resulting sinusoidal current. By forming the ratio
between the variations of the imposed sinusoidal voltage and those of the current
obtained, the complex impedance characteristic of the studied system is obtained:
ZðwÞ ¼ DUðwÞ=DIðwÞ ¼ ReðZÞ þ j:ImðZÞ ¼ IZðwÞI exp ðjUÞ
where w is the pulsation (w ¼ 2pf), Re(Z) and Im(Z) are, respectively, the real and
imaginary part of the complex impedance, and U is the phase.
The experimental device is made up of three electrodes. The working electrode
is the mirror polished zirconium alloy plates pre-oxidized and irradiated or not.
The reference electrode is a saturated mercury sulphate electrode ((Hg/HgSO4/
K2SO4sat), E ¼ 0.658 V/NHE). The counter-electrode is made of Zy4 in the metal-
lic state whose geometry is an empty cylinder, made from TREX. We have experi-
mentally checked that the use of TREX Zy4 as counter electrode leads to similar
results as those obtained with a platinum electrode. The three electrodes are
immersed in a solution of KOH at 1.25 102 mol/l, H3BO3 at 1.8 101 mol/l, and
K2SO4 at 2.5 102 mol/l and the samples are placed in the center of that cylindrical
device. The impedance spectra are obtained with GAMRY software (version 3.2)
and the open circuit potential is measured before acquiring these spectra. The
potential oscillated 6 10 mV around the open circuit potential. The immersed sur-
face of the working electrode is around 150 mm2 (15 mm long and 10 mm wide).
SECONDARY ION MASS SPECTROMETRY

The secondary ion mass spectrometry (SIMS) is a technique for elemental analysis
that consists of sputtering the surface of the sample by an ion beam. Part of the
sputtered matter is ionized, and the secondary ions coming from the studied matter
are then accelerated towards a mass spectrometer where the relative elemental or
isotopic composition is measured as a function of the sputtering time of the sample.
SIMS is an extremely sensitive technique of analysis in particular to detect trace ele-
ments. To be rigorous, the quantification requires precautions of calibration by
using generally implanted samples. Within the framework of our study, the device
used is an ionic analyzer IMS 4 F CAMECA of the CNRS Bellevue in Meudon. It is
composed of a source of primary ions, an optical system, a lens for extraction of
secondary ions, and a mass spectrometer. The primary ion beam for sputtering can
consist of caesium (Csþ), oxygen (O2þ), argon (Arþ), or gallium (Gaþ) ions. The
chemical nature of the incident ions of the beam has a considerable influence on
the degree of ionization of the elements present in the sample. Typically, if the sec-
ondary ions of interest are more easily negatively charged, as is the case for deute-
rium and oxygen 18, the most efficient primary ion is caesium (Csþ), and thus it
was the one chosen to be used in our study.
Experimental Results
EFFECT OF PROTON IRRADIATION ON THE OXYGEN DIFFUSION RATE IN
OXIDE
Oxidation studies of the Zircaloy or M5TM alloy at high-pressure steam during the
pre-transition stage evidenced the existence of different rate-limiting steps accord-
ing to the nature of alloy [6,30,31]. As previously discussed, a comparison of the
corrosion behaviour of these two alloys in autoclave after irradiation of the metal/
oxide interface is likely to bring determining information on their oxidation mecha-
nisms during the pre-transition stage.
Preoxidation of the Samples

The studied samples are mirror-polished plates of Zircaloy-4 and M5TM. For each
alloy, four samples were oxidized in autoclave during 35 days for M5TM and 43
days for Zircaloy-4, which produced, respectively, 1.5 - to 1.6 -lm- and 1.3 - to 1.4 -
lm-thick corrosion films. The oxide thickness variations formed on these materials
as a function of time were deduced from the mass gain according to the following
classical conversion factor: 15 mg/dm2 1 lm. The oxidation kinetics of these two
materials are presented in Fig. 8.
As often observed on plates, the corrosion rate of M5TM is slightly higher than
that measured on Zy4 during pre-transition.
TRIM Calculations
As already mentioned, the TRIM code makes it possible to predict the distribution
of the damage but also the concentration of cation and anion vacancies produced
by irradiation in the matter. The objective is initially to observe the consequences of
the irradiation damage at the metal/oxide interface. TRIM simulations were carried
out to determine the type of ion and the energy of the ions needed to get a relatively
FIG. 8 Oxide thickness variations of the samples of Zircaloy-4 and M5TM versus
oxidation time.
sharp damage peak localised around this interface. These simulations were per-
formed considering a matrix of Zircaloy-4 with an oxide layer of 1.3 lm and one of
M5TM with a 1.6 lm oxide thickness according to the results of pre-oxidation (Fig.
8). Several incident ions were tested, in particular, 250 keV protons, helium ions of
650 keV, and neon ions of 2250 keV.
As shown in Fig. 9, the irradiation by protons with a 250 keV energy leads to a
vacancy distribution in agreement with the objectives. These conditions were conse-
quently retained for this study.
Irradiation Step
The oxidised samples were irradiated by H2þ ions of 500 keV, equivalent to protons
of 250 keV because the molecular ion is divided at the entry into the matter into
two protons of equivalent kinetic energies. The irradiations were carried out thanks
to the ARAMIS accelerator in VAN DE GRAAFF mode with a Penning type source
of ions and at relatively low temperature. A thermocouple installed in contact with
the samples subjected to the greatest fluence (and thus irradiated during the longest
time of about 6 h) indicated a maximum temperature of 80 C. Table 1 indicates the
fluence of irradiation undergone by the samples as well as the number of dpa and
the percentage of anion vacancies deduced from TRIM simulations by supposing a
linear growth of the production rate of defects as a function of the fluence.
It should be noticed that, first, the damage levels and the concentrations of
vacancies correspond to those obtained at the maximum of defect production (i.e.,
close to the metal/oxide interface) and, second, the damage level is much weaker
than the one beyond which Simeone [15] observed a transition phase of zirconia.
FIG. 9 Distributions of anion vacancies simulated by TRIM for an irradiation by protons

of 250 keV in the case of Zircaloy-4 (a), and M5TM (b).
Therefore, the present irradiation with protons should not induce this allotropic
transformation.
Characterization of Hydrogen Implantation

To validate the hydrogen implantation simulated with the TRIM code, the proton
distributions were analyzed by SIMS. In addition, the distribution of the vacancies
produced by irradiation is actually similar to the implantation of the protons
because the defects are essentially generated at the end of the trajectory of the pro-
tons. One can thus suppose that, if TRIM simulation describes satisfactorily SIMS
profiles of hydrogen, it will be the same for the distribution of the oxygen vacancies.
The hydrogen SIMS profiles were performed with a beam of Csþ primary ions.
As shown in Figs. 10 and 11, hydrogen is indeed localised close to the metal/oxide
interface, and the width of the peak is similar to that indicated by the TRIM
TABLE 1 Fluences of irradiation undergone by the materials, numbers of dpa and theoretical per-
centage of vacancy concentrations deduced from SRIM simulations for irradiations with
protons of 250 keV.
Nature Zy4 Zy4 Zy4 Zy4 M5TM M5TM M5TM M5TM
Fluence — 1015 1016 1017 — 1015 1016 1017

(Hþ/cm2)
dpa — 2.2 103 2.2 102 2.2 101 — 3.0 103 3.0 102 3.0 101
% Vo — 0.07 0.7 7 — 0.1 1.1 11
(TRIM)
FIG. 10 Comparison of SIMS profiles of hydrogen in the oxide formed on M5TM with
those obtained from the TRIM code (full dots: SIMS profiles, empty dots: TRIM
calculations).
calculations. Agreement between the experiment and simulation is however more

satisfactory for M5TM than for Zy4. This difference is probably due to the fact that
hydrogen is mainly in the oxide in the case of M5TM and more in the metal for Zy4.
As the SIMS abrasion rates and the hydrogen ionization rates are different in the
oxide and the matrix, these discrepancies may involve a widening of the hydrogen
peak and a ratio of the maximum magnitudes as a function of the fluence lower
than that given by the TRIM simulation in the case of Zy4.
FIG. 11 Comparison of SIMS profiles of hydrogen in the oxide formed on Zy4 with those
obtained with the TRIM code (full dots: SIMS profiles, empty dots: TRIM
calculations).
Study by Electrochemical Impedance Spectroscopy of the

Consequences of the Irradiation: Evolution of the Conductivity in the
Oxide
Electrochemical impedance measurements at the free corrosion potential were used
before and after irradiation of the samples to determine the evolution of the con-
ductivity in the oxide.
A representation commonly used for electrochemical impedance diagrams con-
sists in drawing the imaginary component of the complex impedance on the ordi-
nate versus the real part of the complex impedance on the abscissa (Nyquist
diagram). The qualitative comparison of these diagrams obtained before and after
irradiation and illustrated in Fig. 12 shows that the real portion of the complex im-
pedance measured on Zircaloy-4 increases with increasing received dose. This result
is quite surprising because the production of defects, in particular anion vacancies,
is expected to result in a significant increase of the oxide conductivity, which is not
the case as indicated by the values of the real impedances given at various frequen-
cies and fluences in Table 2.
These values seem to indicate that the real portion of the complex impedance
of the layer formed on Zy4 is apparently less the result of ionic transport than elec-
tronic conduction through the oxide film. Last, according to the preceding biblio-
graphic review, the irradiation results in a modification of the residual charge
distribution in the layer, which could modify the kinetics of charge transfer at the
interfaces and/or the electronic conduction through the layers.
The results obtained on M5TM are given in Fig. 13 and Table 3. By comparing
TM
Figs. 12 and 13, the resistivity of the oxide formed on the unirradiated M5
6
(7 10 ohm) at low frequency is definitely higher than that measured on the layer
FIG. 12 Comparison of the Nyquist diagrams of irradiated Zy4 samples acquired at the
free corrosion potential (black full square: unirradiated Zy4 (Zy4-NI), blue full
triangle: Zy4 irradiated up to 1015 Hþ/cm2, green full diamond: Zy4 irradiated up
to 1016 Hþ/cm2, red full circle: Zy4 irradiated up to 1017 Hþ/cm2).
TABLE 2 Values of the real impedances for various frequencies and irradiation fluences on unirradi-
ated and irradiated Zy4 samples.
Re(Z) Re(Z) Re(Z) Re(Z)

f (Hz) Zy4-NI Zy4-1015 H+/cm2 Zy4-1016 H+/cm2 Zy4-1017 H+/cm2
100 9.3 102 1.15 103 1.4 103 2.7 103

1 1.2 105 1.4 105 1.9 105 3.1 105
0.1 1.5 106 1.7 106 2.1 106 2.6 106
of unirradiated Zy4 (1.5 106 ohm). Unlike Zy4, the real electrical conductivity of
the layer formed on M5TM at low frequency (Fig. 13) increases after irradiation.
Moreover, as indicated in Table 3, the change in conductivity as a function of the
fluence are not monotonic; it goes indeed through a maximum at a fluence of
1016 Hþ/cm2. On the other hand, the high frequency real impedance increases after
irradiation as does Zy4.
The results at low frequencies on M5TM could be interpreted as an increase of
the contribution of ionic transport to the electrical conductivity because of the va-
cancy production during irradiation. The second possible hypothesis is a strong
contribution of the implanted protons to the conductivity of the oxide layer for this
material. Last, the disturbance induced by the irradiation of the electrical potential
distribution in the layer could, contrary to what is observed on the Zy4 alloy,
enhance electronic conduction or the charge transfer at the surface.
In conclusion, whereas the electrical conductivity of the oxides formed on Zy4
decreases as the irradiation fluence increases whatever the frequencies, the low
FIG. 13 Comparison of the Nyquist diagrams of irradiated M5TM samples acquired at

the free corrosion potential (black full square: unirradiated M5TM (M5TM-NI),
blue full triangle: M5TM irradiated up to 1015 Hþ/cm2, green full diamond: M5TM
irradiated up to 1016 Hþ/cm2, red full circle: M5TM irradiated up to 1017 Hþ/cm2).
TABLE 3 Values of the real impedances for various frequencies and irradiation fluences on unirradi-
ated and irradiated M5TM samples.
Re(Z) Re(Z) Re(Z) Re(Z)

f (Hz) M5-NI M5-1015 H+/cm2 M5-1016 H+/cm2 M5-1017 H+/cm2
100 2.2 103 2.7 103 3.8 103 3.5 103

1 6.7 105 7.3 105 7.4 105 6.5 105
0.1 6.3 106 5.0 106 3.2 106 4.5 106
frequency impedance measured on M5TM tends to decrease significantly with

irradiation.
STUDY OF THE OXYGEN DIFFUSION BY ISOTOPIC EXCHANGE

Reminder on the Principle of Isotopic Marker
The techniques of isotopic exchange has largely been used during these last decades
for the study of the oxidation mechanisms. The principle is based on the alternated
oxidation of a sample: the first oxidation stage is carried out under 16O2, C16O2, or
H216O environment and the second one is performed in 18O2, C18O2, or H218O
media. After the formation of an oxide under light water (H216O), the 18O isotope
will occupy, during the stage of exchange whole or in part, the sites of 16O. The pro-
file of the oxygen 18 distribution obtained by SIMS gives therefore information
about the diffusion mechanisms of oxygen for long times of exchange [32] and
allows one to estimate under certain conditions the apparent diffusion coefficient of
oxygen in the oxides for short times of exchange.
Reoxidation under D2O/H218O

The six irradiated samples and the two reference materials were oxidized for 24 h at
360 C and 18.6 MPa in a mixture of D2O/H218O (80/20 %) with 1000 ppm of bo-
ron and 1 ppm of lithium. The advantage in using a D2O/H218O mixture is to get
simultaneously the 18O and D diffusion profiles by SIMS and to know at the same
time the irradiation impact on the hydrogen amount absorbed in the oxide layer.
SIMS Profiles after Isotopic Exchange

To separate the signals from the ions of equivalent mass such as 18O and 16OD by
mass spectrometry, the SIMS profiles have been acquired in high resolution mass
mode.
For short times of isotopic exchange, the diffusion depths are related to the
apparent oxygen diffusion coefficients through the oxide layer in accordance with
Fick’s second law. Only short times of exchange, typically 24 h, were performed
during this study to limit the time of recombination/annihilation of the defects dur-
ing annealing in the autoclave. Moreover, the 18O SIMS profiles were treated to
know the distribution of this element that we should get if the isotopic exchange
had been carried out in a water solution enriched with 100 % of oxygen 18. The so-
lution is made up of 20 % H218O and 80 % D216O, and the intensities of the profiles
in 18O and 16O presented thereafter were reconstituted in the following way [33]:
It ð18 OÞ ¼ 5 ðIb ð18 OÞ Inat ð18 OÞÞ

It ð16 OÞ ¼ Ib ð16 OÞ 4 ðIb ð18 OÞ Inat ð18 OÞÞ
where It is the intensity of the curves reconstituted by considering an isotopic

exchange with water enriched with 100 % of oxygen 18, Ib is the intensity of the
rough curves, and Inat is the intensity related to the natural percentage of oxygen
18.
Last, to compare all of the obtained curves, normalization of the rough inten-
sities to a value of 10 106 counts for oxygen 16 was performed. Consequently, an
18
O intensity of 106 cps/s corresponds to a fraction of oxygen 18 in the layer of
10 % by supposing obviously that the ionization rate of this isotope is equivalent to
that of oxygen 16.
First of all, as shown in Fig. 14, the integrated amount of 18O in the oxide layer
is definitely higher in the most irradiated sample of Zy4 than that of the unirradi-
ated sample. Moreover, the penetration depth of 18O and its concentration on the
surface are clearly increased after irradiation to the highest fluence in comparison
to non-irradiated Zy4. Unlike for Zircaloy-4, the quantity of absorbed 18O, the
18
FIG. 14 O diffusion profiles obtained by SIMS normalized to 16O counts for Zircaloy-4
samples irradiated at different fluences by protons (black full square:
unirradiated Zy4 (Zy4-NI), blue full triangle: Zy4 irradiated up to 1015 Hþ/cm2,
green full diamond: Zy4 irradiated up to 1016 Hþ/cm2, red full circle: Zy4
irradiated up to 1017 Hþ/cm2).
18
FIG. 15 O diffusion profiles normalized to 16O counts, obtained by SIMS for M5TM
samples irradiated at different fluences by protons (black full square:
unirradiated M5TM (M5TM-NI), blue full triangle: M5TM irradiated up to 1015 Hþ/
cm2, green full diamond: M5TM irradiated up to 1016 Hþ/cm2, red full circle: M5TM
irradiated up to 1017 Hþ/cm2).
diffusion depth, and the surface concentration of 18O decreases as the dose received
by the M5TM samples increases (Fig. 15).
Calculations of the apparent diffusion coefficients, the 18O surface concentration
ratios, and estimates of the oxygen diffusion flux ratios between the irradiated and
non-irradiated conditions are indicated for Zy4 and M5TM in Tables 4 and 5, respec-
tively. Two methods (explained briefly in the footnotes) were used to estimate the ratios
of the diffusion fluxes.8,9 The quantitative exploitation of these profiles with both meth-
ods, and the assumptions of the models used are explained in the Appendix.
According to Table 4 for Zircaloy, the apparent diffusion coefficient of oxygen
in the oxide irradiated up to a fluence of 1017 Hþ/cm2 is approximately twice that
observed in the non-irradiated reference condition. In addition, the 18O surface
concentration for the most irradiated Zy4 is around 40 % higher compared to the
unirradiated material. The flux ratio is between a factor of 2 and 3 higher for the
irradiated condition depending on the analysis method followed.
8
Method 1: Diffusion flux ratios are estimated by taking the ratios of products of the surface concentration
and the 18O apparent diffusion coefficient of irradiated material at a given fluence compared to that of the
sample not irradiated.
9
Method 2: Oxygen diffusion flux ratios correspond to the ratios of the total oxygen quantities integrated in
the oxide obtained by integration of the 18O SIMS profiles, after normalization from 16O counts given by
SIMS profiles.
TABLE 4 Apparent diffusion coefficients of oxygen 18 in the oxide formed on Zircaloy-4 irradiated
at different fluences, 18O surface concentration ratios, and oxygen diffusion flux ratios cal-
culated from two different methods.
Nature Zy4 Zy4 Zy4 Zy4
Fluence (H+/cm2) — 1015 1016 1017

DO (cm2/s) 1.1 1015 8.5 1016 1.3 1015 2.2 1015
CO18-e/CO18-e (NI) 1 1.05 1.05 1.4
J/JNI (method 1) 1.0 0.9 1.4 3.1
J/JNI (method 2) 1.0 0.8 1 1.9
As a conclusion, the irradiation with the protons beyond a certain dose results
in a marked increase of the rate of oxygen diffusion through the oxide formed on
Zircaloy-4 by increasing not only the 18O surface concentration as expected but
also the apparent diffusion coefficient.
As indicated in Table 5 for M5TM, the apparent diffusion coefficients of oxygen and
the surface concentrations of 18O decrease as the irradiation dose received by the sam-
ple increases, unlike Zy4. Quantitatively, for the most irradiated material, both the sur-
face concentration and the diffusion coefficient are typically reduced by a factor of two
for the M5TM material, which results in a reduction by a factor ranging from 3 to 4 in
terms of oxygen diffusion fluxes through the oxide. The irradiation of the oxide formed
on M5TM with the protons results in a very marked decrease in the diffusion rates of
oxygen because of the apparent reduction in both the diffusion coefficient and the 18O
surface concentration. These results put forward once again the wide differences in
terms of corrosion behaviour between theses two alloys.
Study of the Deuterium Diffusion

As illustrated by the deuterium SIMS profiles shown on Fig. 16, no significant trend
concerning an irradiation impact on either the diffusion rates or amount of absorbed
TABLE 5 Apparent diffusion coefficients of oxygen 18 in the oxide formed on M5TM irradiated at dif-
ferent fluences, 18O surface concentration ratios, oxygen diffusion flux ratios calculated
from two different methods, according to the fluence of the protons.
Reference M5TM M5TM M5TM M5TM
Fluence (H+/cm2) — 1015 1016 1017

15 15 15
DO (cm /s)2
1.7 10 1.2 10 1.0 10 9 1016
CO18-e/CO18-e (NI) 1 1 0.8 0.5
J/JNI (method 1) 1.0 0.8 0.5 0.25
J/JNI (method 2) 1.0 0.9 0.6 0.3
FIG. 16 SIMS profiles of deuterium in the oxide formed on Zy4 irradiated or not (black
full square: unirradiated Zy4 (Zy4-NI), blue full triangle: Zy4 irradiated up to
1015 Hþ/cm2, green full diamond: Zy4 irradiated up to 1016 Hþ/cm2, red full circle:
Zy4 irradiated up to 1017 Hþ/cm2).
deuterium in the oxide formed on Zircaloy-4 was observed. On the other hand, an irra-
diation effect on the deuterium diffusion appears clearly on the sample of M5TM irradi-
ated at a fluence of 1017 ions/cm2 (Fig. 17). Indeed, according to Fig. 17, whereas the
slightly irradiated sample with 1015 ions/cm2 presents a penetration profile of deute-
rium similar to that obtained on the reference material, deuterium in the oxide formed
on the most irradiated M5TM alloy is more concentrated at the external interface and
the penetration depth is lower. One may speculate that the high quantity of implanted
protons close to the metal/oxide interface could reduce the deuterium diffusion by lim-
iting the hydrogen concentration gradient in the oxide.
For the Zy4 alloy, these observations show that hydrogen diffusion is not
coupled to the oxygen flux in the oxide layer.
The protons were selected because of the smoothness of their peak of defect
production close to the metal/oxide interface. However, it has been proven that a
strong density of hydrides under this interface could significantly increase the oxi-
dation rate of Zy4 [34,35]. Therefore, to confirm the irradiation impact on the in-
ternal interface, the previous experimental approach was reproduced by
implanting, this time, helium ions at the metal/oxide interface. This is the subject of
the following section.
OXIDE IRRADIATION EFFECT BY HELIUM IONS ON THE OXYGEN DIFFUSION

RATE
Preoxidation of the Samples
The operating conditions are the same as previously identified (static autoclaves at
633 K and 18.7 MPa) in light primary water (H2O, 2 ppm Li, 1000 ppm B). The
FIG. 17 SIMS profiles of deuterium in the oxide formed on M5TM irradiated or not (black
full square: not irradiated M5TM (M5TM-NI), blue full triangle: M5TM irradiated up
to 1015 Hþ/cm2, red full circle: M5TM irradiated up to 1017 Hþ/cm2).
kinetics of pre-oxidation of the mirror polished samples of Zy4 and M5TM are pre-
sented in Fig. 18.
The corrosion rates of M5TM is again higher than that measured on Zy4 during
the pre-transition stage as in Fig. 8. The average thicknesses of the oxide layers
FIG. 18 Oxide thickness variations of the samples of Zircaloy-4 and M5TM as a function
of the oxidation time.
FIG. 19 Distributions of anion vacancies simulated by TRIM for an irradiation by helium

ions of 700 keV in the case of Zircaloy-4 (a), and M5TM (b).
obtained on Zy4 and M5TM are, respectively, of 1.4 and 1.5 lm. The corrosion rate
of M5TM is quite higher than that obtained previously during the campaign of pro-
ton irradiations. It is likely that the surface state of the samples before oxidation
was not exactly the same for the two campaigns in spite of our wish to perform the
most reproducible polishing.
TRIM Calculations
TRIM calculations were carried out for matrixes of Zircaloy-4 with a layer of 1.4 lm of
oxide and of M5TM with a thickness of 1.5 lm zirconia to have an implantation peak of
helium and thus of defect production close to the metal/oxide interface. As illustrated
by the TRIM simulations reported in Fig. 19, performed with an incidence angle of 15
(because of the configuration of the irradiation setup of the JANNUS platform at the
CEA of Saclay), helium ions with an energy of 700 keV provide the desired defect dis-
tribution. These irradiations were carried out thanks to the YVETTE accelerator.
Irradiation Step
The objective was to perform irradiation campaigns similar to the previous one in
terms of irradiation damage level. So three fluences, 1014, 1015, and 1016 Heþ/cm2,
were selected to obtain these comparable damage levels, the highest fluence produc-
ing approximately 0.3 dpa near the metal/oxide interface as previously. These fluen-
ces are thus 10 times lower than those reached with protons. Table 6 summarizes
the irradiation fluences undergone by the samples as well as the dpa and the theo-
retical percentage of anion vacancies deduced from TRIM simulations.
As helium atoms are not detectable by SIMS in negative secondary ions, it was
not possible to determine their distribution in the materials to check the coherence
between SIMS profile and helium implantation given by the simulation.
TABLE 6 Fluences of irradiation undergone by the materials, dpa, and percentage of theoretical va-
cancy concentrations deduced from TRIM simulations for irradiations with the helium of
700 keV.
Nature Zy4 Zy4 Zy4 Zy4 M5TM M5TM M5TM M5TM
Fluence — 1014 1015 1016 — 1014 1015 1016

(Heþ/cm2)
Dpa — 2.4 103 2.4 102 2.4 101 — 3.0 103 3.0 102 3.0 101
(TRIM)
% Vo — 0.07 0.7 7 — 0.1 1.3 13
(TRIM)
Oxygen 18 SIMS Profiles after Isotopic Exchange

The six irradiated samples and the two non-irradiated reference samples were oxi-
dized for 24 h at 360 C and 18.6 MPa in a mixture of D2O/H218O (80/20 % in vol-
ume fraction).
As observed in Fig. 20, the amount of 18O integrated through the oxide is clearly
higher on the most irradiated Zy4 sample compared to that of the non-irradiated
sample, as obtained after proton irradiation. Moreover, the penetration depth and
the 18O concentration on the surface of the material irradiated to the highest
FIG. 20 Normalized 18O SIMS profiles for Zircaloy-4 samples irradiated at different
fluences by helium ions (black full square: unirradiated Zy4 (Zy4-NI), blue full
triangle: Zy4 irradiated up to 1014 Heþ/cm2, green full diamond: Zy4 irradiated up
to 1015 Heþ/cm2, red full circle: Zy4 irradiated up to 1016 Heþ/cm2).
FIG. 21 Normalized 18O SIMS profiles for M5TM samples irradiated at different fluences by
helium ions (black full square: unirradiated M5TM (M5TM-NI), blue full triangle: M5TM
irradiated up to 1014 Hþ/cm2, green full diamond: M5TM irradiated up to 1015 Hþ/
cm2, red full circle: M5TM irradiated up to 1016 Hþ/cm2).
fluence is clearly increased compared to those of referenced non-irradiated Zy4.

The effect of irradiation by helium ions is however less marked than the irradiation
impact of protons.
As previously obtained after hydrogen irradiation, and illustrated in Fig. 21, the
quantity of 18O in the oxide, the diffusion depth, and the 18O concentration at the
surface of the most highly irradiated M5TM sample decreases significantly after he-
lium irradiation compared to those evaluated on the unirradiated material.
Calculations of the apparent diffusion coefficients, the ratios of the 18O surface
concentrations compared to that of the non-irradiated reference material, and the
TABLE 7 Apparent diffusion coefficients of oxygen 18 in the oxide formed on Zy4 irradiated at dif-
ferent fluences, 18O surface concentration ratios, and oxygen diffusion flux ratios calcu-
lated from two different methods as a function of the fluence of helium ions.
Nature Zy4 Zy4 Zy4 Zy4
+ 2 14 15
Fluence (He /cm ) — 10 10 1016
15 15 15
DO (cm /s)2
1.3 10 1.5 10 1.5 10 1.8 1015
CO18-e/CO18-e (NI) 1 1 1.1 1.3
J/JNI (method 1) 1 1.2 1.3 1.8
J/JNI (method 2) 1 1.1 1.2 1.6
TABLE 8 Apparent diffusion coefficients of 18 oxygen in the oxide formed on M5TM irradiated at
different fluences, 18O surface concentration ratios, oxygen diffusion flux ratios calculated
from two different methods, according to the fluence of helium ions.
Nature M5TM M5TM M5TM M5TM
Fluence (He+/cm2) — 1014 1015 1016

DO (cm2/s) 1.6 1015 1.7 1015 1.3 1015 0.7 1015
CO18-e/CO18-e (NI) 1 1 1 0.9
J/JNI (method 1) 1 1 0.8 0.4
J/JNI (method 2) 1 1 0.9 0.5
ratios of the oxygen diffusion fluxes are presented in Tables 7 and 8, respectively, for
Zy4 and M5TM.
According to Table 7, the increase of the apparent oxygen diffusion coefficient is
observed beyond 0.1 dpa after helium irradiation of Zy4. Second, the 18O surface
concentration is about 30 % higher than that observed in the unirradiated sample,
and is similar to that obtained by proton irradiation. The ratio of the fluxes is in
this case between 1.5 and 2 depending on the method of evaluation.
In conclusion, the irradiations with helium beyond a certain dose confirms the
accelerating role of the irradiation defects on the oxygen diffusion rate through the
oxide formed on Zircaloy-4. However, the effect is smaller than with the protons.
For M5TM, the irradiation with helium ions leads to similar results to those
obtained with protons but with a lower magnitude in terms of reduction of the
apparent diffusion coefficients and 18O surface concentration. Quantitatively, as
indicated in Table 7, the oxygen diffusion flux through the most irradiated layer is
divided by a factor of 2 to 3 compared to that of the unirradiated material.
Finally, these results are very similar to those obtained by proton irradiation.
Concerning hydrogen diffusion, results not presented in this paper for the sake
of length have shown that He ion irradiation leads to similar conclusions as for pro-
ton irradiation.
Discussion
Despite the extent of the results, this experimental study alone does not make it
possible to provide a model describing the corrosion behaviour of zirconium alloys
under irradiation. Although we understand well the nuclear collisions through the
kinematics of billiard balls, the inelastic collision process linked to the electronic
stopping power is more difficult to understand in a straightforward manner. Thus,
the results in this study are not easy to interpret in terms of the differing impact of
irradiation on the diffusion flux through oxide films formed on Zircaloy and M5TM.
However, within the framework of the understanding of the results obtained and
their interpretation, this discussion will be focused on analyzing the probability of
occurrence of certain phenomena and determining which are potentially brought

into play.
At the beginning, knowing that the nuclear collisions create anti-Frenkel pairs

like (VO -O00i ), our objective was initially to produce by irradiation vacancies at
the metal/oxide interface of the oxide layer formed on Zy4 to increase the vacancy
concentration gradient in the film. This interpretation supposed implicitly that
the rate-limiting step of the corrosion process is oxygen diffusion and that it
occurs via a vacancy mechanism. But, the Fþ centers corresponding to ionized ox-
ygen vacancies are quickly restored beyond 550 K, according to Costantini and
Beuneu [15], which means that this type of defect produced by irradiation could
not be responsible for the increase of the oxygen diffusion flux at the temperature
of the corrosion test.
Second, the inelastic collisions may theoretically produce atom displacements if
the electric field induced by the ion trace is sufficiently intense to impose a local deg-
radation of the atomic order. Apparently, according to the bibliographic data, this
type of disorder would be observable only for strong electronic decelerations
(>10 keV/nm) [20] and would be accompanied by a local amorphisation of the crys-
tal. Though the electronic stopping powers investigated during this study are gener-
ally lower than 1 keV/nm, it is an a priori possibility that the fine trace left by the
incident ion constitutes a preferential way of diffusion. Considering that the oxygen
transport occurs preferentially along grain boundaries as usually suggested in these
conditions for Zy4, this trace could actually play a role of a short-circuit of diffusion
similar to the grain boundaries, but probably with a migration energy quite different
because of the local state of charge induced by the way of the particle. In this scenario,
the traces generated by ion irradiation in the oxide could increase the diffusion flux
of oxygen by growing the density of the short-circuit of diffusion.
Third, it is extremely difficult, as already suggested, to grasp the consequences
of irradiation in terms of electrical potential changes in the layer. It is actually
impossible to know without dedicated experimental study as that performed by
Thomé and co-workers [18] if there remains a residual charge after irradiation and
how the field induced by the new distribution of electrical potential modifies the
reaction equilibrium and the diffusion process. But we can try to establish a link
between the temperature stability of irradiation defect and the electrical field in the
oxide. Indeed, whereas no Fþ center is observed at 600 K, there remains, according
to certain authors [7], a residual disorder until 800 K hypothetically related to the
dipoles produced by the anti-Frenkel pairs. Others showed that the T centers, trigo-
nal cationic sites, Zr3þ, related to three oxygens, are restored only beyond 650 K
[17] (>633 K). And Thomé showed also that the surface electrical field of irradiated
stabilized zirconia is only annealed beyond 800 K. The charge distribution induced
by irradiation seems thus to be relatively persistent at high temperature and could
be potentially linked to the presence of the T centers, i.e., trigonal cationic sites in
the oxide. However, the electrical field induced by ion implantation can modify in
one sense or in another the kinetic constants and the diffusion fluxes taking into
account the Wagner’s theory. We can speculate that the discrepancies of the varia-
tion of the electrochemical impedance, because of the irradiation and probably to
the residual charge left by the incident ions in the oxide, between Zy4 and M5TM,
could be at the origin of the difference of the oxygen diffusion flux evolution.
Finally, a question of interest from an industrial point of view is to determine
whether or not the ion irradiation damages in the oxide are likely to reproduce the
neutron effect on the corrosion rate. It has been shown in references [1,2] that the
acceleration factor on Zy4 between PWR conditions and autoclave is around 2
before the phase III acceleration. During the works presented here, the oxygen dif-
fusion flux ratios between the most irradiated samples and the reference ones are
comprised between 1.5 and 3 depending on the considered ion. This value is there-
fore quite consistent with that obtained on claddings burned up to three cycles in
PWRs. Ion irradiation of the oxide seems thus to reproduce quite well the observed
effect of neutron damage on corrosion kinetics for up to three cycles of exposure.
During the high burn-up acceleration stage, the corrosion rate increase can
reach acceleration factors greater than 10, which is clearly not the same order of
magnitude as those observed in this study. It could be because of the discrepancy of
irradiation damage level between this work and the cladding state at relatively high
burn-up (above four cycles, which correspond approximately to 5 dpa in the outer
part of the oxide) but also to the change of matrix composition and more specifi-
cally to the hydride precipitation under the metal/oxide interface.
In conclusion, oxide irradiation, whichever particle has an accelerating effect
on the oxygen diffusion flux through the film formed on Zy4, could be fully respon-
sible for the observed effect of neutron damage on corrosion kinetics for up to three
cycles of exposure, but only partially responsible for the phase III acceleration.
With regard to M5TM, we can speculate that it’s good “resistance” to irradiation
from a corrosion kinetics point of view could prevent, delay, or inhibit the accelera-
tion stage at high burn-up.
Conclusion
During this research program, a new technical approach was carried out to get in-
formation about the irradiation effects on the corrosion rate. This study was specifi-
cally focused on a particular distribution of defects in the oxide layer, typically,
localised close to the metal/oxide interface.
From the results of the irradiation of the metal/oxide interface on Zy4, we
obtain the following conclusions:
• The irradiation with the protons of the internal interface is not accompanied
by an increase of the electrical conductivity in the layer but, on the contrary,
by an enhancement of its resistivity.
• Whatever the incident ion, the irradiation of the internal interface of the oxide
formed on Zy4 results in an increase of oxygen absorbed by the layer beyond a
threshold of damage ranging between 0.01 and 0.1 dpa;
• The 18O surface concentration increases by about 30 % to 40 % for a damage

level of approximately 0.2–0.3 dpa close to the metal/oxide interface, inde-
pendent of the incident particle.
• The penetration depth and thus the apparent coefficient of diffusion of oxygen
are appreciably increased after irradiation with the protons, this effect being
definitely less marked with helium ions.
• Quantitatively, the ratios of average diffusion fluxes of oxygen between the
most irradiated materials and the non-irradiated reference sample are around
2 for proton irradiation and approximately 1.5 for helium ion irradiation.
Concerning the results on M5TM, we obtain the following conclusions:
• The irradiation with protons of the internal interface modifies the electric con-
ductivity of the layer which varies in a non monotonic way with the frequency
and the fluence of irradiation.
• Whatever the incident ion, the irradiation of the internal interface of the oxide
formed on M5TM results in a marked reduction in the amount 18O integrated
through the oxide layer; in the 18O surface concentration, the apparent diffu-
sion coefficient of oxygen decreases significantly at high fluence. The reduction
of these parameters is less marked with helium ion irradiation.
• Quantitatively, the diffusion fluxes of oxygen through the oxide film are
respectively reduced after irradiation at the highest fluences by a factor,
between 3 and 4 for proton irradiations, and, between 2 and 2.5 for helium ion
irradiations.
• For deuterium, a significant increase of the surface concentration is associated
with a penetration depth reduction in the layer after irradiation of the internal
interface.
Finally, the oxide irradiation impact on the oxygen diffusion can probably
explain the acceleration factor observed on Zy4 during the first cycles in pile and
has been discussed in putting forward the probable role of the residual charge left
by ion implantation.
ACKNOWLEDGMENTS
The writers would like to warmly thank all the people who have contributed to these
works and in particular the staffs of ARAMIS and YVETTE accelerators of, respec-
tively, JANNUS-Orsay and JANNUS Saclay platforms.
Appendix: Description of the Calculations

Performed with Method 1 and 2
METHOD 1
Quantitatively, to analyze the results, the 18O surface concentration, cs is considered as
constant during the isotopic exchange:
t > 0; x ¼ 0; c ¼ cs
FIG. 22 Kaleidagraph fitting compared to the experimental points from the 18O SIMS
profile obtained on the M5TM sample irradiated up to 1017 Hþ/cm2.
18
And the initial concentration in oxide is that of the O natural concentration, c0
(0.2 %):
t ¼ 0; x > 0; c ¼ c0
where cs is the 18O surface concentration, c0, the 18O natural concentration, t the (re-)
oxidation time, and x, the depth of the oxide layer.
Taking into account the following boundary conditions, the solution of the sec-
ond law of Fick is given by the following expression:

@c @2c x
¼ Da 2 ) c c0 ¼ ðcs c0 Þ 1 Erf pffiffiffiffiffiffiffi
@t @x 2 Da t
where Da is the apparent diffusion coefficient of 18O, c0, the natural concentration of
18
O, and cs, the 18O surface concentration.
Using the Kaleidagraph software, the experimental dots can be fitted with the pre-
vious law to obtain the value of the apparent diffusion coefficient and the 18O surface
concentration. As shown on Fig. 22, the points and the curve correspond, respectively,
with the experimental dots coming from the 18O SIMS profile and the fitting obtained
with Kaleidagraph with m2, the apparent diffusion coefficient, and m1, the 18O sur-
face concentration.
Finally, from these fitting parameters, it is then possible to evaluate the ratio of
oxygen diffusion fluxes between irradiated material and the reference one by consider-
ing that this flux verifies the expression coming from the Wagner’s theory:
Dc J1017 ðDa cs Þ1017

JaDa ) ¼
X JNI ðDa cs ÞNI
where J1017 is the diffusion flux of 18O after irradiation until the fluence of 1017 ions/cm2,
JNI , the 18O diffusion flux of the unirradiated material, and X, the oxide thickness.
METHOD 2
In this case, the global amount of 18 oxygen absorbed in the oxide layer is estimated
in integrating the 18O signal of the SIMS profile as a function of the oxide depth. By
taking the ratios of these integrated signals, the ratio of the average fluxes during the
isotopic exchange can be evaluated by this way:
ðx
I18O dx a hJ18O i Dtexchange
ð0x
I18O dx
1017 hJ18O i1017
ð0x ¼
hJ18O iNI
I18O NI
dx
0
Where I18 O 1017 and hJ18 O i1017 are, respectively, the 18O signal deduced from the SIMS pro-
file and the average diffusion flux of 18O after irradiation until the fluence of 1017 ions/
cm2 and I18 O NI and hJ18 O iNI are, respectively, the 18O signal deduced from the SIMS
profile and the average diffusion flux of 18O of the reference material and X is the oxide
thickness.
References
[1] Bouineau, V., Ambard, A., Bénier, G., Pêcheur, D., Godlewski, J., Fayette, L., and Duver-
neix, T., “A New Model to Predict the Oxidation Kinetics of Zirconium Alloys in a Pressur-
ized Water Reactor,” J. ASTM Int., Vol. 5, No. 5, 2008, JAI101312.
[2] Bouineau, V., Bénier, G., Pêcheur, D., Thomazet, J., Ambard, A., and Blat, M., Nucl. Tech-
nol., “Analysis of the Waterside Corrosion Kinetics of Zircaloy-4 Fuel Cladding in French
PWRs,” Vol. 170, No. 3, 2010, p. 444.
[3] Garner, G. L., Hilton, B. A., and Mader, E., “Performance of Alloy M5 Cladding and
Structure,” Proceedings of the LWR Fuel Performance Meeting/Top Fuel, San Francisco,
CA, Sept 30–Oct 3, 2007.
[4] Gilbon, D., “Les matériaux de gaines et d’assemblage,” Les Combustibles Nucléaires,
Monographie de la DEN, Editions du Moniteur, Paris, 2008, p. 38.
[5] Wagner, C., “Contributions to the Theory of the Tarnishing Process,” Z. Phys. Chem., Vol.
B21, 1933, pp. 25–41.
[6] Tupin, M., Pijolat, M., Valdivieso, F., Soustelle, M., Frichet, A., and Barberis, P., “Differences
in Reactivity of Oxide Growth During the Oxidation of Zircaloy-4 in Water Vapour
Before and After the Kinetic Transition,” J. Nucl. Mater., Vol. 317, Nos. 2–3, 2003, pp.
130–144.
[7] Simeone, D., Baldinozzi, G., Gosset, D., Le Caër, S., and Mazerolles, L., “Impact of
Radiation Defects on the Structural Stability of Pure Zirconia,” Phys. Rev. B, Vol. 70, 2004, p.
134116.
[8] Méthivier, A., “Etude Experimentale et Theorique de l’evolution Texturale et Structura-

lede Poudres de Zircone Pures et Dopees,” Ph.D. thesis, Institut National Polytechnique
de Grenoble, Grenoble, France, 1992.
[9] Abriata, J. P., Garcés, J., and Versaci, R., “The O-Zr (Oxygen-Zirconium) System,” Bull.
Alloy Phase Diagr., Vol. 7, No. 2, 1986, p. 116.
[10] Godlewski, J., Bouvier, P., Lucazeau, G., and Fayette, L., “Stress Distribution Measured by
Raman Spectroscopy in Zirconia Films Formed by Oxidation of Zr-Based Alloys,” 12th
International Symposium of Zirconium in the Nuclear Industry, ASTM STP 1354, ASTM
[11] Kountouros, P. and Petzow, G., Advances in Ceramics 3, Science and Technology of Zirco-
nia, A. H. Heuer and L. W. Hobbs, Eds., American Ceramic Society, Columbus, OH, 1981,
pp. 202–213.
[12] French, R. H., Glass, S. J., and Ohuchi, F. S., “Experimental and Theoretical Determina-
tion of the Electronic Structure and Optical Properties of Three Phases of ZrO2,” Phys.
Rev. B, Vol. 49, No. 8, 1994, pp. 5133–5142.
[13] Jomard, G., Petit, T., and Pasturel, A., “First-Principles Calculations to Describe Zirconia
Pseudopolymorphs,” J. Phys. Rev. B, Vol. 59, No. 6, 1999, pp. 4044–4052.
[14] Costantini, J. M. and Beuneu, F., “Threshold Displacement Energy in Yttria-Stabilised

Zirconia,” Phys. Stat. Solidi C, Vol. 4, No. 3, 2007, pp. 1258–1263.
[15] Simeone, D., Gosset, D., Bechade, J. L., and Chevarier, A., “Analysis of the Monoclinic-
Tetragonal Phase Transition of Zirconia Under Irradiation,” J. Nucl. Mater., Vol. 300,
2002, pp. 27–38.
[16] Costantini, J. M. and Beuneu, F., “Point Defect Induced in Yttria-stabilised Zirconia by
Electron and Swift Heavy ion Irradiations,” J. Phys.: Conden. Matter, Vol. 23, 2011, pp.
115902–115911.
[17] Costantini, J. M. and Beuneu, F., “Thermal Recovery of Colour Centres Induced in Cubic
Yttria-stabilised Zirconia by Charged Particle Irradiations,” J. Phys.: Condens. Matter,
Vol. 18, 2006, pp. 3671–3682.
[18] Thomé, T., Braga, D. G., Blaise, G., Cousty, J., Pham van, L., and Costantini, J. M.,
“Charging Kinetics in Virgin and 1 MeV-electron Irradiated Yttria-stabilised Zirconia in
the 300-1000K Range,” Mater. Sci. Eng. B, Vol. 130, 2006, pp. 177–183.
[19] Moll, S., Thomé, L., Garrido, F., Vincent, L., Sattonnay, G., Costantini, J. M., Jagielski, J.,
Benayoub, A., and Behar, M., “Radiation Effects in Yttria-stabilized zirconia: Comparison
Between Nuclear and Electronic Processes,” Nucl. Instrum. Methods Phys. Res. B, Vol.
266, 2008, pp. 3048–3051.
[20] Moll, S., Thomé, L., Garrido, F., Vincent, L., Sattonnay, G., Costantini, J. M., Jagielski,
J., Benayoub, A., and Behar, M., “Damage Induced by Electronic Excitation in Ion-
Irradiated Yttria-stabilised Zirconia,” J. Appl. Phys., Vol. 105, 2009, pp.
023512–023524.
[21] Petigny-Putigny, N., “Comparaison de l’oxydation de deux Alliages de Zirconium par Dif-
fraction des Rayons X in-situ et ex-situ: Texture, Phase, Contrainte,” Ph.D. thesis, Univer-
sité de Dijon, Dijon, France, 1998.
[22] Roy, C. and David, G., “X-Ray Diffraction Analyses of Zirconia on Zirconium and Zircaloy-
2,” J. Nucl. Mater., Vol. 37, 1970, pp. 71–81.
[23] Parise, M., 1996, “Mecanisme de Corrosion des Alliages de Zirconium : Etude des Cine-
tiques Initiales d’oxydation et du Comportement Mecanique du Systû´me metal-oxyde,”
Ph.D. Thesis, Ecole des Mines de Paris, Paris, France.
[24] Gibert, F., Couvreur, C., Damien, D., Gautier-Soyer, M., Thromat, N., Guittet, M.J., Serruys,
Y., Bouffard, S., and Elkaim, E., “Study of Irradiation Effects on the Crystallographic Na-
ture of Zirconia,” http://www.iaea.org/inis/collection/NCLCollectionStore/_Public/
30/060/30060381.pdf (Last accessed 30 April 2012).
[25] Godlewski, J., Gros, J. P., Lambertin, M., Wadier, J. F., and Weidinger, H., “Raman
Spectroscopy Study of the Tetragonal-to-Monoclinic Transition in Zirconium
Oxide Scales and Determination of Overall Oxygen Diffusion by Nuclear Micro-
analysis of 18O, Zirconium in the Nuclear Industry,” Ninth International Sympo-
sium, ASTM STP 1132, C. M. Eucken and A. M. Garde, Eds., Philadelphia, 1991, pp.
416–436.
[26] Cox, B., “Some Thoughts on the Mechanisms of in-reactor Corrosion of Zirconium
[27] Bérerd, N., Chevarier, A., Moncoffre, N., Jaffrézic, H., Balanzat, E., and Catalette, H.,
“Zirconium Oxidation Under High-Energy Heavy-ion Irradiation,” J. Appl. Phys., Vol. 97,
2005, p. 083528.
[28] Bérerd, N., “Effets d’irradiation sur l’oxydation du Zirconium et la Diffusion de l’uranium
dans la Zircone,” Ph.D. thesis, Université de Lyon, Lyon, France, 2003.
[29] Cox, B. and Pemsler, J. P., “Diffusion of Oxygen in Growing Zirconia Films,”J. Nucl. Mater.,
Vol. 28, 1968, pp. 73–78.
[30] Tupin, M., Pijolat, M., Valdivieso, F., and Soustelle, M., “Oxidation Kinetics of ZrNbO in
Steam : Differences Between the Pre- and Post-Transition Stage,” J. Nucl. Mater., Vol.
342, 2005, pp. 108–118.
[31] Dali, Y., “Etude Expérimentale de l’oxydation des Alliages de Zirconium à Haute Pression
de Vapeur d’eau et Modélisation des Mécanismes,” Ph.D. thesis, Ecole des Mines de
Saint Etienne, St. Etienne, France, 2008.
[32] Basu, S. N. and Halloran, J. W., “Tracer Isotope Distribution in Growing Oxide Scales,”
Oxidation of Metals, Vol. 27, No. 3/4, 1987, pp. 143–155.
[33] Dali, Y., Tupin, M., Bossis, P., Pijolat, M., Wouters, Y., and Jomard, F., “Corrosion Under
High Pressure Steam of Pure Zirconium and Zircaloy-4 Followed by in-situ
Thermogravimetry,” J. Nucl. Mater., Vol. 426, 2012, pp. 148–159.
[34] Blat, M. and Noel, D., “Detrimental Role of Hydrogen on the Corrosion Rate of Zirconium
Alloys,” 11th International Symposium of Zirconium in the Nuclear Industry, ASTM STP
1295, E. R. Gradley and G. P. Sabol, Eds., West Conshohocken, PA, 1996, pp. 319–337.
[35] Bisor, C., “Compréhension des Mécanismes de Prise d’hydrogéne des Alliages de Zr en
Situation de Corrosion – Impact des Hydrures sur la Cinû
`tique de Corrosion,” Ph.D. the-
sis, Université d’Evry, Evry, France, 2010.
DISCUSSION
Question from Mark Daymond, Queen’s University:—In your figures, both pre-
dicted and experimental location of ions/damage is in different locations relative to
the interface, and of different shapes, for Zircaloy-4 and M5. Can you explain and
comment on potential impact on your results?
Authors’ Response:—Undoubtedly, the proton implantation of M5TM samples is

not exactly located right at the metal/oxide interface. However the production of
irradiation defects just before the internal interface will increase the vacancy con-
centration gradient in the outer part of the oxide as for Zircaloy-4 and should
increase the oxygen diffusion rate in the oxide layer. This was not observed for M5.
What makes us confident in these results is that the trend concerning the decrease
of the oxygen diffusion is also noticed with helium ions. In this case the peak of
irradiation defect is well located at the metal/oxide interface as indicated in the pa-
per. Thus based on this comparison, we concluded that proton implantation in the
oxide near the internal interface on M5 is representative of the irradiation effect
produced right at the metal/oxide interface.
Question from Philipp Frankel, University of Manchester:—The peak of defects

introduced by irradiation is different relative to the metal/oxide interface. Given
that, the oxide closest to the interface is often considered the most important in
terms of protectiveness, will the fact that the peak does not reach the interface for
M5 impact the results, i.e., is the damage of less consequence when it is mostly
away from the interface?
Authors’ Response:—This question is similar as the previous one. The results

obtained with helium ions compared to protons show that the oxygen diffusion flux
evolution due to irradiation is not changed by the small difference of irradiation
defect peak location. Thus the fact that the proton peak does not reach the interface
for M5 does not impact the trend of our results.
Question from Ron Adamson, Zircology Plus, USA
Q1:—Do you think your main results could be affected by another effect? Dur-
ing irradiation, the conductivity of the oxide could significantly increase; therefore
electron transport may be rate limiting rather than oxygen diffusion.
Authors’ Response:—Due to the fact that the oxides that were studied were rela-
tively thick, more than 1 lm, the vacancy flux during the corrosion process out-
core is proportional to the electron flux according to the Wagner theory. It is
impossible to know which diffusion process is the rate limiting step. We just know
that they tend to be equal.
Moreover the conductivity of the oxide formed on Zircaloy-4 decreases signifi-

cantly after irradiation while the oxygen diffusion flux is increased. So in our case,
the increase of the oxygen diffusion flux is not linked to an increase in conductivity.
Moreover, it seems that the conductivity is preferentially due to electronic transport
than ionic transport.
However, as mentioned in the paper, the residual charge potentially left in the
oxide during ion irradiation could significantly modify both electron and oxygen
diffusion fluxes.
Q2:—How might your post-irradiation results be different than during irradia-

tion results?
Authors’ Response:—The recombination rate of irradiation defects beyond a

certain time is equal to the production rate of these defects under irradiation while
it is not the case in our experimental approach by steps. Consequently the concen-
tration of irradiation defects is probably not the same. The results may quantita-
tively be different but the trend should not change.
Moreover, the irradiation defect distribution is uniform in core while defect

production is located at the metal/oxide interface in this study. This difference does
not seem to play a major role since we have recently obtained similar results after
generating uniform concentration of defects across the oxide.
Question from Shiori Ishino, (Retired) Univ. of Tokyo:—
You discuss the effect of irradiation only by considering vacancies. However,

interstitials can contribute to diffusion. This should also be considered. Please
comment.
Authors’ Response:—That is right that the reasoning is based on the vacancy

production by irradiation without taking into account the simultaneous creation of
interstitials. During the post-irradiation isotopic exchange experiment, two ways of
defect annealing are possible: recombination or annealing on the sinks. If the
recombination rate was very high, we should have no effect of irradiation damage
on the corrosion rate which is not observed. Moreover, if both defects were
annealed very quickly on the sinks, no effect of irradiation damage on the corrosion
rate would have been observed. Consequently, the annealing rate and the migration
rate of these defects have to be different between interstitials and vacancies. We
suppose implicitly that interstitials due to their large size diffuse very slowly and do
not participate in the migration process of oxygen through the oxide, which is con-
trolled by vacancies.
Questions from Javier Romero, Westinghouse Electric Co.:—Do you see differen-
ces in intermetallic size/density between non-irradiated and irradiated material?
How much evolution of intermetallics will affect impedance and oxygen diffusion,
in order to explain the reverse result in M5?
Authors’ Response:—This aspect has not yet been investigated. It will be

however difficult to relate SPPs distribution to impedance and/or oxygen diffu-
sion. We are working on characterization by TEM of the samples to make pro-
gress in the understanding of the irradiation impact on the nano- and
microstructure of the oxide.
Question from B. K. Shah, BARC Mumbai:—Is there any short-term accelerated

autoclave corrosion test which can be used for ranking of Zr alloys for their in-reac-
tor corrosion behavior?
Authors’ Response:—In the case of Zircaloy-4, the experimental approach used

during this work seems relevant to qualitatively study the in-core corrosion behav-
iour, especially at low irradiation damage (less than 30 GWd/tU). The accelerating
factor between in-core and autoclave test is around 2 for Zircaloy-4. This value is
relatively close to the results we obtained here with two light ions on Zircaloy-4.
Our results in terms of kinetic effect are finally quite consistent with the neutron
irradiation impact on the oxidation rate for low burn-up.
Regarding M5TM, the accelerating factor between in-core and autoclave test has
been estimated at around 2 by V. Bouineau (ref. [2] in the paper). In this case, our
results are not able to explain the in-pile accelerating factor observed on this mate-
rial. This could be due to a factor (e.g., radiolysis, thermo hydraulic conditions)
other than irradiation. Moreover, a study on Zr-2.5%Nb showed a decrease of the
in-pile corrosion rate compared to non irradiated material1. The authors attributed
that kinetic effect to the matrix evolution while our study shows that the oxygen dif-
fusion flux reduction is due to oxide irradiation.
Question from N. Ramasubramanian, ECCATEC Inc. Canada
Q1:—Following ion implantation in the pre-transition oxide, have you per-

formed short-term corrosion (a few hours) in 18O/2D (H218O/D216O) enriched
aqueous environment and checked the SIMS profiles for oxygen?
Authors’ Response:—The time of isotopic exchange has been fixed to 24 hours

to get a sufficient penetration depth of oxygen. Below this time, the penetration
depth would have been too low and we could not separate the diffusion profiles
between irradiated and unirradiated material.
Q2:—Would 18O profile be only surface deep? Does the 18O profile change to a
diffusion type profile with longer term oxidation?
Authors’ Response:—It is quite difficult to anticipate the evolution of the SIMS

profiles obtained for longer terms of isotopic exchange. It would be nevertheless
interesting to study this evolution but the irradiation effect should reduce with time
due to the recombination of the irradiation defects with exposure time. That is the
reason why we had first chosen a short time of isotopic exchange.
1. V. F. Urbanic, J. E. Lesurf and A. B. Johnson, “Effect of Ageing and Irradia-

tion on the Corrosion of Zr-2.5 Wt% Nb,” CORROSION-NACE, vol. 31, n 1
(1975)15
Adrien Couet,1 Arthur T. Motta,2 and Robert J. Comstock3
Effect of Alloying Elements on

Hydrogen Pickup in Zirconium
Alloys
Reference
Couet, Adrien, Motta, Arthur T., and Comstock, Robert J., “Effect of Alloying Elements on
Hydrogen Pickup in Zirconium Alloys,” Zirconium in the Nuclear Industry: 17th International
ABSTRACT
Although the optimization of zirconium-based alloys has led to significant
improvements in hydrogen pickup and corrosion resistance, the mechanisms by
which such alloy improvements occur are still not well understood. In an effort to
understand such mechanisms, we conducted a systematic study of the alloy
effect on hydrogen pickup, using advanced characterization techniques to
rationalize precise measurements of hydrogen pickup. The hydrogen pickup
fraction was accurately measured for a specially designed set of commercial and
model alloys to investigate the effects of alloying elements, microstructure, and
corrosion kinetics on hydrogen uptake. Two different techniques for measuring
hydrogen concentrations were used: a destructive technique, vacuum hot
extraction, and a non-destructive one, cold neutron prompt gamma activation
analysis. The results indicate that hydrogen pickup varies not only from alloy to
alloy, but also during the corrosion process for a given alloy. These variations
result from the process of charge balance during the corrosion reaction, such
that the pickup of hydrogen decreases when the rate of electron transport or
Manuscript received December 25, 2012; accepted for publication June 26, 2013; published online June 17,
2014.
1
Dept. of Mechanical and Nuclear Engineering, Penn State Univ., University Park, PA 16802, United States of
America (Corresponding author), e-mail: adrien.couet@gmail.com
2
America.
3
Westinghouse Electric Company LLC, Pittsburgh, PA 15235, United States of America.
4
Hyderabad, India.
Copyright V
ox
oxide electronic conductivity (re ) through the protective oxide increases.
According to this hypothesis, alloying elements (either in solid solution or in
ox
precipitates) would affect the hydrogen pickup fraction by modifying re . Because
the mechanism whereby these alloying elements are incorporated into the oxide
layer is critical to changing electron conductivity, the evolution of the oxidation state
of two common alloying elements, Fe and Nb, when incorporated into the growing
oxide layers of two commercial zirconium alloys (Zircaloy-4 and ZIRLO) and model
alloys (Zr-0.4Fe-0.2Cr and Zr-2.5Nb) was investigated using x-ray absorption near-
edge spectroscopy with microbeam synchrotron radiation on cross-sectional oxide
samples. The results show that the oxidation of both Fe and Nb is delayed in the
oxide layer relative to that of Zr, and that this oxidation delay is related to the
variations of the instantaneous hydrogen pickup fraction with exposure time.
Keywords
hydrogen pick-up, zirconium alloys, CNPGAA, XANES, alloying elements, elec-
tronic conductivity
Introduction
As fuel burnup and reactor residence times increase, the uniform corrosion of zir-
conium alloy nuclear fuel cladding (and associated hydrogen pickup) can become a
limiting factor for the use of high-burnup fuel rods in existing and advanced light water
reactors [1,2]. Several factors can control the uniform corrosion of zirconium alloys [3].
Although alloy optimization of zirconium-based alloys used for nuclear fuel cladding
has been a key to increasing corrosion resistance and reducing hydrogen pickup, a
complete understanding of the role of alloying elements in the corrosion and hydrogen
pickup mechanisms is still lacking.
Because very small alloying element differences cause significant differences in
corrosion and hydrogen pickup, it is of interest to examine the effect of alloying ele-
ments on the corrosion and hydrogen pickup mechanisms of zirconium alloys,
which can, in turn, yield significant insights potentially leading to validation of a
general hypothesis on the underlying mechanisms.
The overall zirconium corrosion reaction is written as
(1) Zr þ 2H2 O ! ZrO2 þ 2H2
Some of the hydrogen generated by the corrosion reaction diffuses through the ox-
ide and enters the metal, where it eventually can precipitate as hydrides, causing
cladding embrittlement [1]. For comparison between alloys, the total hydrogen
pickup fraction fH is defined as the ratio of the hydrogen absorbed from the begin-
ning of the corrosion test, Dt0 Habsorbed , to the total amount of hydrogen that has
been generated by the corrosion reaction, Dt0 Hgenerated .
Dt0 Habsorbed
(2) fH ¼
Dt0 Hgenerated
COUET ET AL., DOI 10.1520/STP154320120215 481
The instantaneous hydrogen pickup fraction fHi is defined as the ratio of the hydro-
gen absorbed from a time t to a time t þ Dt to the total amount of hydrogen that
has been generated by the corrosion reaction during the same period.
DtþDt Habsorbed
(3) fHi ¼ t
DttþDt Hgenerated
Few measurements of instantaneous hydrogen pickup fractions have been

reported in the literature, as they require precise hydrogen measurements at suc-
cessive small exposure time intervals. The determination of fH (and, to a lesser
extent, fHi ) on various zirconium alloys has been the subject of extensive
research, but the mechanisms of hydrogen pickup and, especially, the influence
of the alloy composition and microstructure on fH are still not well understood
[4–7]. Studies have shown that fH depends strongly on alloy composition [4,6],
alloy microstructure [8,9], and corrosion conditions [10]. It has been shown in
the protective regime that pure Zr picks up about 20 % to 30 % of the hydrogen
generated during the corrosion reaction, more than commercial alloys [11]. In
general, alloying elements decrease hydrogen pickup, with the exceptions of Ni,
which increases fH, and Sn, which is reported to have no effect [11,12]. Previous
measurements of instantaneous hydrogen pickup have shown that fHi may vary
at different stages of oxide film growth [13], but no clear understanding yet
exists. Some of the literature is reviewed in the following sections.
Corrosion and Hydrogen Pickup Mechanisms

RATE-LIMITING STEP IN UNIFORM ZIRCONIUM ALLOY CORROSION
The zirconium oxide formed on Zr alloys exhibits a protective character, such that
after the formation of the oxide there is no direct contact between the metal and
the water, and the corrosion reaction cannot happen directly, so the oxidizing spe-
cies have to travel through the oxide layer. In order for zirconium oxidation to
occur, either the cations or the anions have to be transported through the oxide
layer. The corrosion of zirconium and its alloys in high-temperature environments
occurs via oxygen anion migration through the corrosion film, with the formation
of new oxide occurring at the metal–oxide interface [14,15]. The sub-
stoichiometric gradient of the zirconium oxide would be the primary driving force
for the oxygen anion diffusion, although this value has not been precisely
determined.
The oxidation process can be divided into several steps, as presented in Fig. 1.
First, oxygen in the water molecule dissociates and is adsorbed onto the oxide layer
surface.
dissociation
(4) 2H2 O ! 4Hþ þ 2O2
adsorbed
FIG. 1 Schematic of the oxidation process of zirconium alloys. The chemical reactions
are numbered, and the fluxes of charged species through the oxide are
represented by arrows.
absorption
(5) O2
adsorbed þ VO
2
€ ! Oabsorbed
Because of the defect concentration gradient, the oxygen anions diffuse either through
the bulk of the oxide or along the oxide grain boundaries. When the oxygen anion
reaches the oxide–metal interface, it reacts with Zr cations to form new oxide.
oxidation
(6) Zr ! Zr4þ þ 4e
oxide
formation
(7) Zr4þ þ 2O2
absorbed ! ZrO2
The formation of this new oxide releases electrons, which then migrate through the
oxide to reduce the hydrogen ions at the cathodic site.
reduction
(8) 4Hþ þ 4e ! 2H2
It is well known that the corrosion rate of zirconium alloys decreases as the thick-
ness of the oxide layer increases [16]. Because of this, it is considered that either ox-
ygen anion diffusion or electron diffusion is the rate-limiting step. This assumption
leads to the parabolic scaling law for the oxidation kinetics: d ¼ Ktn, with n ¼ 0.5
[17]. However, it has been observed that the oxidation of zirconium alloys is fre-
quently sub-parabolic [18–21].
In the absence of externally applied potentials on the specimen, the net current
through the oxide is zero, which means that the negative and positive oxidation
currents must be equal and opposite. In the case of a deviation from this balance, a
potential gradient develops across the oxide thickness that equalizes these currents.
If the transport of electrons through the oxide is the rate-limiting step, a positive
electric gradient through the oxide layer [the potential is negative at the metal–-
~
oxide interface relative to the oxide–water interface, ðr:VÞ > 0] appears and indu-
ces a negative electric field across the oxide layer. This field slows down the
diffusion of the positively charged oxygen vacancies and increases the diffusion of
electrons toward the oxide–water interface through the Coulomb force ~ F
[~ ~
F ¼ eZs ðr:VÞ; where Ze ¼ 1 and ZVO€ ¼ 2]. Hence this negative electric field
will tend to decrease the corrosion rate. Thus, a negative electric field tends to
increase the transport of electrons and decrease the transport of oxygen vacancies.
By these effects, the magnitude of the negative electric field is in turn reduced, so
that the variations effected by the electric field are in a direction to decrease the rate
of change in the field. Therefore, a stable situation in which the currents produce
no further changes in the field is reached, and the net transport of charges at any
location in the oxide is zero. Of course, the opposite is observed if the transport of
oxygen vacancies is the rate-limiting step (and thus the electric field in the oxide is
positive).
Many studies have observed negative potential at the metal–oxide interface rel-
ative to the oxide–water interface (the electric field in the oxide then being negative)
[3,22–27]. From measurement of that potential as a function of exposure time, it
also appears that electronic and ionic resistivities become more balanced as the ox-
ide thickens, but the electric field normally remains negative. This indicates that the
transport of electrons is normally slower than the transport of oxygen vacancies,
making it the rate-limiting step in most circumstances. Although the mechanism of
electron transport is still unclear [24,25,28,29], it seems reasonable to consider that
both bulk and localized conduction can contribute to the overall rate [29–32]. The
preceding arguments indicate that oxide electronic conductivity is a key parameter
in controlling zirconium alloy oxidation [33,34].
Clearly, if electron transport controls the oxidation kinetics, the precipitates
embedded into the growing oxide and the extrinsic compensating defects due to
aliovalent cations in solid solution could have a significant effect on the corrosion
kinetics by affecting the oxide electronic conductivity [35–37].
HYDROGEN PICKUP IN ZIRCONIUM ALLOYS

Hydrogen pickup in zirconium alloys has been the subject of many studies in the
past 50 years, and several hydrogen pickup mechanisms have been proposed.
It has been proposed that the diffusion of hydrogen through the oxide layer
would occur via solid-state diffusion through oxygen vacancies [10,38,39], with pos-
sible localized diffusion at the second phase precipitates [9]. However, the hydrogen
pickup fraction would be not a function of the corrosion kinetics, but rather the
result of two competing processes: hydrogen discharge (hydrogen evolution) and
the metal hydrogen uptake [5,40], with the alloying elements embedded in the
oxide layer affecting both of these processes.
It is well accepted in the literature that Nb has a beneficial effect on hydrogen
pickup, significantly decreasing the hydrogen pickup fraction [41–43]. This can be
explained by the doping effect of the donor Nb5þ at the interface with the aqueous
medium, which increases the electron concentration at the oxide–water interface,
thereby promoting the hydrogen discharge process [26,44]. Electrochemical meas-
urements have also shown that oxide electronic conductivity is increased by Nb
additions, promoting hydrogen evolution at the oxide–water interface and reducing
hydrogen uptake [34]. However, these effects cannot readily explain the observed
variations of fHi as a function of exposure time [13]. The effect of the addition of
transition metals such as Ni, Fe, and Cr on hydrogen pickup has also been studied,
and it has been shown that for a given oxide thickness, Zircaloy-2 absorbs more
hydrogen than Zircaloy-4. Also, the conclusion has been drawn that Ni has a detri-
mental effect on hydrogen pickup [4,8,45]. The effect of second-phase precipitates
on hydrogen pickup is, however, still unclear [7], with different results depending
on the precipitate size and distribution [9,33].
OBJECTIVE OF THE STUDY

The foregoing illustrates the need to understand the effects of alloying elements on
hydrogen pickup so as to design better alloys. In order to contribute to this under-
standing, we used a non-destructive technique called cold neutron prompt gamma
activation analysis (CNPGAA) to quantitatively assess hydrogen concentrations in zir-
conium alloys. Using this nondestructive technique coupled with a conventional vac-
uum hot extraction (VHE) technique, we systematically measured the hydrogen
pickup fraction at different exposure times for a set of zirconium alloys with specific
chemistries and microstructures. The evolution of the chemical states of two alloying
elements, Fe and Nb, when incorporated into the oxide layers formed during autoclave
testing of various zirconium alloys was examined using micro x-ray absorption near-
edge spectroscopy (lXANES).
ALLOYS
The alloys used in this study are listed in Table 1 and included production materials
Zircaloy-4 and ZIRLO and model alloys Zr-2.5Nb, pure sponge Zr, and Zr-0.4Fe-
0.2Cr. All alloys were in the recrystallized state except for Zircaloy-4 tube, which
was in a cold work stress relieved state. The starting hydrogen concentration in the
alloys prior to autoclave testing was 10 to 15 wt. ppm.
The zirconium alloy samples were in the form of corrosion sheet coupons (25
mm by 20 mm by 0.8 mm) or tubing as shown in Table 1. The alloys were processed
following a procedure described in Ref 46. Two of the alloys were processed at
580 C, to minimize grain growth for pure zirconium (sponge) and to maintain
small precipitates in the Zr-0.4Fe-0.2Cr (L) alloy. A high process temperature of
TABLE 1 Description of the chemical composition and geometry of the commercial and model
alloys used in this study. (L) and (H) denote small and large precipitate size, respectively
(see text) [21].
Alloy System Alloy Composition, wt. % Sample Geometry
Model alloy Pure Zr Zr sponge Sheet

Zr–Fe–Cr Zr-0.4Fe-0.2Cr (L) Sheet
Zr-0.4Fe-0.2Cr (H)
Zr–Nb Zr-2.5Nb Tube
Commercial alloy ZIRLO Zr-1.0Nb-1.0Sn-0.1Fe Sheet and tube

Zircaloy-4 Zr-1.45Sn-0.2Fe-0.1Cr Sheet and tube
720 C was used to grow large precipitates in the Zr-Fe-Cr (H) alloy. The average
precipitate diameter determined via synchrotron diffraction in Zr-0.4Fe-0.2Cr (L)
and Zr-0.4Fe-0.2Cr (H) is equal to 40 nm and 110 nm, respectively [47].
CORROSION TEST
Corrosion tests were performed at Westinghouse Electric Co. autoclave facilities
in Churchill, PA. The samples were corroded in 360 C pure water in saturated
pressure conditions at 18.7 MPa (2708.6 psi) according to ASTM G2 [48]. After
every exposure cycle, the autoclave was opened and the water was refreshed in
order to maintain a low hydrogen concentration in the water. In addition, all
samples were weighed at the end of each cycle to measure weight gain as a func-
tion of exposure time. Sister samples from each alloy were periodically archived
for subsequent destructive measurement of hydrogen content. Some selected
samples were used for the nondestructive measurement of hydrogen content via
CNPGAA. After hydrogen content measurements, these samples were returned
to the autoclave for additional exposure, followed by further CNPGAA measure-
ments of hydrogen concentrations, so that the evolution of the hydrogen pickup
fraction was characterized on single samples.
HYDROGEN MEASUREMENTS
The hydrogen pickup fraction was determined at different exposure times using a
combination of two different techniques, VHE and CNPGAA, described in the fol-
lowing paragraphs.
VHE was performed by LUVAK, Inc. (Boylston, MA). An NRC Model 917 ap-
paratus [49] was used for VHE. The technique is described in detail elsewhere [50].
Because the technique is destructive, it is necessary to perform several measure-
ments on different samples (sister samples), which is intrinsically a cause of mea-
surement dispersion. Also, the experimental error is not well characterized, and the
average experimental error is approximately 65 %, as confirmed by different stud-
ies [13,51]. The dimensions of the analyzed samples were 8 mm by 8 mm for sheet
coupons and a 3-mm-long ring in the case of tubes. In most cases, several samples
from the same coupons were analyzed to improve the accuracy of the results.
The procedure for CNPGAA measurements in zirconium alloys is discussed in
detail elsewhere [50]. CNPGAA on ZIRLO and Zircaloy-4 sheet samples was per-
formed at the National Institute of Standards and Technology (NIST) (Gaithersburg,
MD) in one of the cold neutron beam lines. The background noise at the hydrogen
gamma ray energy at this beamline is extremely low, so concentrations as low as 5
wt. ppm of hydrogen in zirconium alloys are detectable.
HYDROGEN PICKUP FRACTION

The hydrogen pickup fraction of a sample was calculated from the measurement of
its weight gain and its hydrogen content. The calculation assumed no loss of oxide
during corrosion. This assumption was verified by the good correspondence between
weight gain and oxide layer thickness measured using scanning electron microscopy
(SEM) on polished cross-sections of the corrosion sample. In addition, the weight
gain was assumed to be due only to oxygen and to not take into account the hydro-
gen uptake. This is a good assumption for low hydrogen pickup fractions, but one
has to keep in mind that the hydrogen uptake resulting from a theoretical hydrogen
pickup fraction of 100 % could account for 11.1 % of the total weight gain. Because
the autoclave is opened at least every 30 days, no significant build-up of hydrogen
gas is observed (during the early stages of corrosion, in which corrosion rates are
higher and thus hydrogen gas releases are more significant, the autoclave was opened
more often).
Following these assumptions, fH is equal to [46]
106 ðmts CHt mis CHi Þ

(9) fH ¼
ðmt mis Þ
2 s MH
MO
where:
mts ¼ mass of the sample at the time of the measurement,
mis ¼ mass of the bare sample,
MO ¼ atomic mass of the oxygen atom,
MH ¼ atomic mass of the hydrogen atom,
CHi ¼ initial concentration of hydrogen in the sample, wt. ppm, and
CHt ¼ concentration of hydrogen at the time of measurement, wt. ppm.
The instantaneous pickup fraction fHi is given by
106 ðmtþDt CHtþDt mts CHt Þ

(10) fHi ¼ s
tþDt
ðm mts Þ
2 s MH
MO
The hydrogen pickup fraction errors originating from weight gain and hydrogen
concentration measurements have been evaluated using an error propagation
formula.
X-RAY ABSORPTION NEAR-EDGE SPECTROSCOPY EXPERIMENT

Micro-XANES experiments were performed at the 2ID-D beamline of the
Advanced Photon Source (APS) at Argonne National Laboratory [52]. Figure 2
shows a schematic of the scattering and data-acquisition geometry for the XANES
experiments: a focused synchrotron microbeam (0.2 lm in diameter) is placed in
frontal incidence on the region of interest of the cross-sectional oxide specimen.
The XANES signals are recorded in fluorescence. The XANES scan energy step is
0.5 eV, and the chosen energy windows are 100 eV (from 7.09 keV to 7.19 keV) for
Fe (the Fe K-edge is located at 7.12 keV) and 140 eV (from 18.94 keV to 19.08 keV)
for Nb (the Nb K-edge is located at 18.98 keV). All measurements were performed
at room temperature.
The experimental procedure consisted of first recording a fluorescence scan of
the oxide layer Zr L-edge (in case of Fe) or Zr K-edge (in case of Nb) to determine
the positions of the oxide–metal and oxide–water interfaces. XANES scans were
performed at various distances from the metal–oxide interface, both in the oxide
and in the metal. In order to rule out specimen drift during the experiment, after
each XANES scan an extra fluorescence scan was obtained to confirm the oxide
position.
A list of the corrosion samples that were archived and used in the XANES
experiments with the exact exposure time and oxide thickness (determined from
weight gains) is shown in Table 2.
In order to determine the chemical state of Fe and Nb in the zirconium oxide
layer, Fe K-edge and Nb K-edge XANES spectra of reference standards were meas-
ured. The processing of XANES spectra was performed using Athena software (ver-
sion 0.8.061, Ifeffit 1.2.11c) [53].
FIG. 2 Schematic drawing showing the geometry of data acquisition at the

synchrotron beamline.
TABLE 2 List of the samples used for the XANES experiments. Their exposure time (in days) and
their oxide thickness (in micrometers, derived from the weight gain) are indicated.
Exposure Oxide
Time, days Thickness, lm
Fe Zircaloy-4 Before first 45 1.7

transition 60 1.9
Between first and 90 2.8
second transitions 105 3.4
120 3.6
135 3.9
150 4.1
165 4.2
Between second and 225 5.9
third transitions 255 6.4
Zr-0.4Fe-0.2Cr(H) Pre-breakaway 173 2.4
463 3.2
R
V
ZIRLO Before first 30 1.7
transition 60 2.3
90 2.7
R
V
Nb ZIRLO Before first 30 1.7
transition 60 2.3
90 2.7
Between first and 240 5.5
second transitions
Zr-2.5Nb Before first 60 1.7
transition 120 2.7
150 3.2
The different Fe and Nb standards used for the XANES experiments are
detailed at length elsewhere [52], and the list is presented in Table 3.
The oxide samples were in the form of small transverse cross-sections (the
transverse direction being normal to the cross-section) of corroded sheet coupons
prepared according to a procedure detailed elsewhere [52].
A schematic drawing of the final sample configuration is shown in Fig. 3. The
sample thickness was approximately 200 lm. The incident beam being parallel to
the transverse direction, the measured XANES signal as a function of distance from
the oxide–metal interface includes contributions from the alloying element (Fe or
Nb) both in precipitates and in solid solution, as the x-ray attenuation length is
approximately 12.5 lm at 7.12 keV and 28 lm at 18.98 keV [54], which results in a
sampled volume that is tens of micrometers deep.
For Zircaloy-4, assuming that the ratios of Fe and Cr in the precipitates and in the
alloy are the same (which has been found to be true for Feðwt: %Þ=Crðwt: %Þ < 4
TABLE 3 List of the different standards used in the fitting process for XANES spectra [52].
Alloy Metal Standard [52] Oxide Standard
Fe Zircaloy-4 Fe bcc Fe2O3 powder

Fe in Zircaloy-4 metal Fe3O4 powder
Zr-0.4Fe-0.2Cr Fe in Zr(Fe,Cr)2 SPP FeO powder
Fe in pure Zr
ZIRLO Fe bcc Fe2O3 powder
Fe in ZIRLO metal Fe3O4 powder
Fe in pure Zr FeO powder
Nb ZIRLO Nb powder NbO powder
Nb in ZIRLO metal NbO2 powder
Nb2O5 powder
Zr-2.5Nb Nb powder NbO powder
Nb in Zr-2.5Nb metal NbO2 powder
Nb2O5 powder
[55]) and the concentration of Fe in solid solution ranges between 50 and 200 wt. ppm
[56], the volume fraction of precipitates is approximately 0.5 %, and the total amount
of Fe in second-phase precipitates is approximately 10 to 40 times the amount of Fe in
solid solution. Although the fluorescence signal is not directly proportional to the con-
centration, we can conclude that the contribution of alloying elements in solid solution
FIG. 3 Schematic drawing showing the geometry of the samples used for XANES in a
transverse cross-section top view.
in the XANES signal coming out from the bulk samples cannot be neglected. The prob-
ability that the x-ray beam will hit a single precipitate as a function of oxide depth in
our samples as calculated in Ref 52 is equal to 100 % for samples as thick as 200 lm.
Thus the recorded XANES spectra are a convolution of signals from Fe in both solid
solution and precipitates, so that the deconvolution of Fe signals from solid solution
and precipitates during the processing of XANES spectra is impossible.
Results
CORROSION TEST
The corrosion weight gains for the different alloys as a function of exposure time
are plotted in Fig. 4. The plotted weight gains represent an average of the whole set
of sister samples. The standard deviations of weight gain measurements among sis-
ter samples of a given alloy at a given exposure time are less than 0.5 mg/dm2. The
samples indicated by arrows have been archived and used for XANES experiments.
The commercial alloys show the well-defined transition-type corrosion behavior
that has been observed previously in Zr alloys [20]. The oxide thicknesses at the
first transition are 2.1 lm for Zircaloy-4, 2.9 lm for ZIRLO, and 3.5 lm for Zr-2.5Nb
(1 lm ¼ 14.77 mg/dm2).
The Zr–Fe–Cr model alloys did not show a transition up to a corrosion time of
463 days. The corrosion data of pure Zr are not shown after 14 days because the
alloy underwent a sudden breakaway and loss of protectiveness, as confirmed by
SEM characterizations of the oxide layers, which also showed lateral and longitudi-
nal cracks and significant preferential oxide growth in the zirconium metal (in the
form of dendrites).
TOTAL HYDROGEN PICKUP FRACTION

The hydrogen contents (mg/dm2) of zirconium alloys as a function of weight gain
(mg/dm2) are plotted in Figs. 5 and 6. The dashed lines represent the hydrogen con-
tent as a function of weight gain for constant fH of 10 %, 20 %, and 30 %. The
hydrogen contents measured via CNPGAA are marked by a star on the expanded
view of Fig. 5. At a given weight gain wide variations are seen in fH between alloys.
Additionally for a given alloy, fH significantly changes (increases) with oxide thick-
ness for a given alloy.
Figure 5 shows that the total hydrogen pickup fraction of pure Zr before break-
away is equal to 18 %, whereas that for Zircaloy-4 in the pre-transition regime is
lower than 10 %, suggesting that the presence of precipitates reduces hydrogen
pickup, as Sn is known to have almost no effect on hydrogen pickup [4,6]. This is
confirmed by the pickup fraction of approximately 10 % measured for the model
alloy ZrFeCr for the same oxide thickness (see Fig. 6). The fH of ZIRLO is consis-
tently lower than that of Zircaloy-4, whereas the fH of Zr-2.5Nb at a given oxide
thickness is the lowest of the samples measured, which suggests that the addition of
Nb decreases the hydrogen pickup fraction.
FIG. 4 Weight gain as a function of exposure time for the following alloys: (a) ZIRLO
sheet; (b) Zircaloy-4 sheet; (c) Zr-2.5Nb; (d) pure sponge zirconium; (e) Zr-
0.4Fe-0.2Cr (L); and (f) Zr-0.4Fe-0.2Cr (H). The arrows indicate the samples
that were archived and studied using XANES.
It is apparent from Figs. 5 and 6 that for all the alloys studied, the overall fH var-
ied significantly with oxide thickness. One point to note is that the fH was higher
for the weight gain acquired between the first and second transitions than before
the first transition, and still higher between the second and third transitions. This is
in spite of considerable evidence that the corrosion kinetics are repetitive and the
oxide layer reforms itself in each transition. This suggests that the presence of the
FIG. 5 Hydrogen content as a function of weight gain (and oxide thickness) for the
alloys studied. The corrosion data are available in Fig. 4. The dashed lines
correspond to constant total hydrogen pickup fractions of 10 %, 20 %, 30 %, and
40 %. An expanded view of early exposure time is also displayed. Hydrogen
contents of samples marked by a star have been measured via CNPGAA.
porous oxide formed during previous transitions might act to increase hydrogen
pickup although this oxide is permeable to water.
At the end of the corrosion test, after 375 days of corrosion, fH was equal to
25 % for Zircaloy-4 and 19 % for ZIRLO.
The value of fH increased significantly from the pre-transition period to the first
transition period (the transitions are marked by a sudden increase in weight gain,
as seen in Fig. 4; the first transition period is defined as the period between the first
and second transitions) and, to a lesser extent, from the first transition period to
the second transition period. It also appears that fH varied within the transition
periods, which would suggest that there are significant variations of fHi even though
the corrosion kinetics changes only smoothly (see Fig. 4). This particular point is
discussed in more detail in the next section.
The value of fH also depends on the alloy microstructure. The hydrogen content
as a function of weight gain is plotted for the two Zr–Fe–Cr model alloys in Fig. 6.
Before reaching a thickness of approximately 3 lm, both alloys have similar fH val-
ues (between 10 % and 15 %). After the oxide thickness reached 3 lm, we observed
an increase in fH, whereas the corrosion kinetics remained unchanged (see Fig. 4).
FIG. 6 Hydrogen content as a function of weight gain (and oxide thickness) for the
alloys Zr-0.4Fe-0.2Cr (L) and Zr-0.4Fe-0.2Cr (H). The corrosion data are
available in Fig. 4. The dashed lines correspond to constant total hydrogen
pickup fractions of 10 %, 20 %, and 30 %. The hydrogen content of the sample
marked by a star has been measured via CNPGAA.
Even though both model alloys show that fH increased once the oxide was approxi-
mately 3 lm thick, the alloy with bigger precipitates was less sensitive to that
increase. This suggests that for a given volume fraction and corrosion rate, alloys
with bigger Zr(Fe,Cr)2 precipitates tend to pick up less hydrogen than alloys with
smaller precipitates.
As discussed in the preceding section, the zirconium alloys’ oxidation kinetics
follow a power law of the form Ktn. We determined the exponent n by fitting the
weight gain curves by a power law. The R2 of the power law fitting was kept above
0.999 to ensure a good fit. The total hydrogen pickup fraction at the last fitted point
(approximately 50 days of exposure time) is plotted in Fig. 7 as a function of the
exponent n. An inverse relationship between the corrosion kinetics and fH is
observed: the lower the value of n, the greater the value of fH. Relative to pure Zr,
the addition of Zr(Fe,Cr)2 precipitates tends to increase the kinetics and lower the
fH. Nb-containing alloys have the fastest corrosion kinetics but the lowest hydrogen
pickup. Even though they are not discussed in this paper, we also added Zr-0.5Cu
and Zr-2.5Nb-0.5Cu alloys to this plot. It appears that relative to pure Zr, the addi-
tion of Cu slows down the kinetics but increases the hydrogen pickup fraction.
FIG. 7 Total hydrogen pickup fraction at the last fitted point (generally 50 days of
exposure) as a function of the exponent n from the power law fit of the weight
gain wg ¼ ktn for various zirconium alloys.
INSTANTANEOUS HYDROGEN PICKUP FRACTION

We calculated the instantaneous hydrogen pickup fraction of commercial alloys in
order to better characterize the evolution of hydrogen pickup as a function of corro-
sion kinetics. The results are presented in Fig. 8 for the ZIRLO sheet. All results
came from VHE measurements on sister samples. The weight gains of the samples
that were analyzed for hydrogen are also indicated in the figures. Two plots of fHi
are superimposed in the figures because half of the sister samples in the autoclave
had 15 more days of corrosion than the other half. Thus, with the autoclave being
opened every 30 days, half of the samples had a corrosion time of t and the other
half had a corrosion time of t þ 15 days. Two fHi curves are plotted, one for each set
of sister samples. The time Dt between two measurements of fHi was generally equal
to 30 days.
The fHi error bars are rather large because of the small Dt and the errors from
the VHE measurements, which led to more uncertainties, especially when the cor-
rosion rate was low. However, a general trend is observed in these plots. Similar
trends have been observed for Zircaloy-4, Zr-2.5Nb-0.5Cu, and Zr-2.5Nb alloys
[46]. It is clear that fHi varies as a function of exposure time. At first, when the cor-
rosion rate is high, fHi is low (around 5 % in the pre-transition regime). As the expo-
sure time increases, the corrosion rate slows down and fHi increases significantly.
The instantaneous hydrogen pickup fraction keeps on increasing until the sample
reaches transition. At transition, the corrosion rate increases again, and fHi drops.
The process then repeats itself in the second transition regime, and so on, following
the periodicity of corrosion kinetics. The instantaneous hydrogen pickup fraction
evolution has been confirmed by following the pickup fraction of a given sample
using CNPGAA to measure its hydrogen content at regular time intervals [46], so
FIG. 8 Instantaneous hydrogen pickup fraction and weight gain as a function of

exposure time (in days) for ZIRLO sheet alloy.
that the dispersion of results due to sister sample variations and spot-to-spot varia-
tions are minimized.
After the first transition, fHi decreases to approximately 10 % to 15 %, and not
to the 5 % observed at the beginning of corrosion. Similarly, fHi at the second transi-
tion reaches higher values than at the first transition. These increases in fHi from
one transition period to another lead to the observed increases in fH in Fig. 5.
FE AND NB OXIDATION STATES

To quantify the Fe and Nb chemical states as measured by lXANES, the spectra
were fit using the linear combination fitting algorithm included in the Athena soft-
ware [53] (in this case, 14 spectra were used, including Fe2O3, Fe3O4, bcc-Fe, FeO,
Fe in Zr, etc. [see Table 3]). Through the fitting, we quantified the fraction of alloy-
ing element that had been oxidized in the oxide. In all alloys, the fractions of oxi-
dized Fe and Nb were close to zero at the metal– oxide interface and gradually
increased in the protective oxide (the oxide formed since the previous transition),
finally reaching 100 % some distance from the oxide–metal interface, indicating
that all of the Fe and Nb was oxidized in the outer part of the oxide. The oxidation
of Fe and Nb (both in precipitate and in solid solution) is delayed relative to the
oxidation of Zr, so that a significant fraction of Fe and Nb remains metallic in the

oxide layer, as already observed [57,58].
To compare the evolution of alloying element oxidation states between samples
of different oxide thicknesses (i.e., at different exposure times), we chose the param-
eter dmet, defined as the length of the oxide layer in which the fraction of metallic
Fe or Nb is above 50 %. Thus dmet is an indication of the evolution of the oxygen
potential in the oxide layer. We also define dp as the protective oxide thickness
(defined as the oxide formed since the previous transition). The ratio dmet/dp is an
indicator of the fraction of the oxide in which alloying elements are mainly unoxi-
dized. Values of dmet for Fe in Zircaloy-4 and ZIRLO and for Nb in ZIRLO and Zr-
2.5Nb as a function of dp are shown in Fig. 9. The transition thickness is represented
by the shaded area. Because there are some variations in transition thickness from
one transition period to another, the transition thickness is represented by a shaded
area for ZIRLO and Zircaloy-4 in Fig. 9, and not by a straight line. Also, because the
FIG. 9 dmet (in microns) as a function of the protective oxide thickness dp (in microns)
for (a) Fe in Zircaloy-4 sheet, (b) Fe and Nb in ZIRLO tube, (c) Nb in Zr-2.5Nb,
and (d) Fe in Zr-0.4Fe-0.2Cr (H). The shaded area shows the transition
thickness for the different alloys.
oxide thickness increases rapidly right after transition, it is difficult to archive a

sample with small dp; hence the lack of data at small values of dp.
Generally, at first, dmet increases up to a threshold value of approximately 0.8
lm for Zircaloy-4, 1 lm for ZIRLO, and 1.2 lm for Zr-2.5Nb. The threshold value
for the model alloy Zr-0.4Fe-0.2Cr (H) is at least 1.25 lm. Once dmet reaches its
threshold value, it remains constant up to the time when the oxide reaches its tran-
sition thickness. At the transition, it is believed that dmet drops, to zero. Interest-
ingly, between the points at dp equal to 2.4 lm and dp equal to 3.2 lm for Zr-0.4Fe-
0.2Cr (H), we observed a decrease in dmet suggesting that although the corrosion
kinetics do not change, a loss of protectiveness might be imminent.
As expected, dmet does not depend on the transition period (see the results for
Zircaloy-4 and ZIRLO), as it is directly dependent on the oxygen potential in the
oxide layer and it is known that the corrosion of Zircaloy-4 is periodic [20]. Thus
the evolution of the oxidation of alloying elements is periodic from one transition
period to another.
Discussion
The results above indicate that hydrogen pickup results from the need to balance
ox
charge such that when oxide electronic conductivity (re ) decreases, a driving force
exists for hydrogen ingress.
RELATIONSHIP BETWEEN OXIDATION KINETICS AND HYDROGEN PICKUP

FRACTION
As discussed previously, different alloys have different oxidation kinetics, with Nb
alloys being closer to parabolic, whereas pure Zr and Fe/Cr-based alloys are sub-
cubic. Similar variations in the oxidation kinetics of zirconium binary alloys have
already been noticed [19]. If we assume that the electron transport is rate limiting,
the hydrogen pickup mechanism will be linked to the corrosion kinetics through
the electron flux: the higher the oxide electronic conductivity, the higher the elec-
tron flux (and thus the higher the corrosion rate) and the lower the hydrogen
pickup fraction. This is also in accordance with results reported on binary Zr–Fe
alloys [59] and on Zr–Sn–Fe–Cr alloys [60].
Striking and consistent variations in fHi were observed in this work, with fHi
increasing consistently before transition, dropping at transition, and increasing in
the following transition period, in agreement with previous results [13]. One of the
possible hypothesis for the variations of fHi could be that the hydrogen pickup frac-
tion is only a phenomenological characteristic and that the rate of hydrogen picked
up by the metal is actually constant, so that fH appears higher at lower corrosion
rates. Using CNPGAA on given samples to measure the quantity of hydrogen
picked up by the metal as a function of exposure time, it is possible to show that the
actual quantity of hydrogen picked up varies as a function of exposure time, which
negates this hypothesis [46]. The present results show that the hydrogen pickup
fraction is linked to the corrosion kinetics, although it does not follow this kinetics.
A likely explanation is that initially the corrosion rate is high, because electron
transport is easy. When the evolution of alloying elements in the oxide layers for
example by oxidation of precipitates causes the electronic conductivity to decrease,
the corrosion rate decreases, and concomitantly, a driving force is established that
enhances hydrogen ingress. Thus alloying elements are a key parameter of the
hydrogen pickup mechanism, in agreement with the fact that fH changes from alloy
to alloy. Their effect on hydrogen pickup is discussed in the next sections in terms
of the effect of Nb and the effect of precipitates.
Effect of Nb Addition on fH
Nb additions decrease the hydrogen pickup fraction. Kiselev et al. studied the
effect of Nb additions on fH in binary Zr–Nb alloys [41] and concluded that as
the Nb concentration increases, the corrosion rate and therefore the amount of
hydrogen produced increase, but the hydrogen pickup fraction decreases, espe-
cially in the solid solution range (up to 0.5 wt. % [61]). In our study, we also
observed that Zr-2.5Nb alloy showed both the lowest fHt throughout the experi-
ment and the highest corrosion rate. The presence of Nb5þ in the oxide is indi-
cated by the fact that the Nb XANES spectra showed absorption edges above the
Nb4þ edge, although it is possible that a contribution of lower valences also exists
[58,62]. The presence of the Nb2O5 phase in Zr-2.5Nb oxides has also been con-
firmed by photoelectrical analysis of passive zirconium niobium oxide layers [63].
Oxidized Nb atoms dissolved in the ZrO2 solid solution would dope the oxide
layer and act mostly as donors. If we do not consider the aggregation of alloying
elements and the formation of complex defects, the compensating defect of the
Nb positive charge will be either zirconium vacancies or electrons. Given that the
conduction band of ZrO2 is formed of zirconium 3d empty states and the zirco-
nium vacancy is highly positively charged, it is believed that electrons are the pre-
ferred compensating defects.
According to this picture the oxide electronic conductivity would increase as a
result of the increase in the free electron concentration. This result is also con-
firmed by electrochemical measurements on zirconium alloys, with Zr-2.5Nb
showing by far a lower electronic resistance than other Zr alloys [33]. As a result
it is believed that an increase in oxide electronic conductivity would reduce the
hydrogen pickup fraction. As it is believed that electron transport is the rate-
limiting step in zirconium alloy oxidation (see the section “Rate-limiting Step in
Uniform Zirconium Alloy Corrosion”), the increase in oxide electronic conductiv-
ity could also explain the faster kinetics observed in Zr–Nb alloys, which is closer
to parabolic than cubic. This hypothesis could also explain why the size and vol-
ume fraction of Nb precipitates do not play a significant role in determining the
hydrogen pickup fraction and corrosion resistance of Zr–Nb alloys, as the domi-
nant effect would be a low electronic resistivity due to the donor effect of Nb in
solid solution.
Effect of Precipitates on fH
The presence of Zr(Fe,Cr)2 precipitates appears to reduce the hydrogen pickup frac-
tion relative to pure Zr. In agreement with previous observations, the lXANES
results show that precipitates remain metallic when embedded in the growing zirco-
nium oxide layer up to a certain distance from the metal–oxide interface. Metallic
precipitates would likely enhance the electronic conductivity of the oxide layer,
which would in turn reduce the hydrogen pickup relative to pure Zr. It is believed
that a material with a homogeneous distribution of fine precipitates has a higher ox-
ide electronic conductivity than pure Zr [34,37].
For a given volume fraction and corrosion rate, the fH of the alloy with larger
Zr(Fe,Cr)2 precipitates is lower than the fH of the alloy with smaller Zr(Fe,Cr)2 pre-
cipitates. Electrochemical measurements have shown that for a given volume frac-
tion, Zircaloy-4, with bigger Zr(Fe,Cr)2 precipitates, would have a greater oxide
electronic conductivity than alloys with smaller precipitates [33], resulting in a
lower fH. Metallic precipitates can act as local electric shortcuts favoring electronic
conduction, thus promoting hydrogen evolution at the oxide–water interface and
resulting in a low hydrogen pickup fraction. However, lXANES results have shown
that precipitates do not remain metallic and oxidize after a certain distance from
the oxide– metal interface, so that they could not act as local electric shortcuts
throughout the oxide layer.
Precipitate oxidation and its effect on hydrogen pickup fraction are the subject
of the next sections.
Oxidation Model of Precipitates

The lXANES measurements in Zircaloy-4, ZIRLO, Zr-2.5Nb, and Zr-0.4Fe-0.2Cr
alloys reported in this work [64,65] show that the oxidation of Fe and Nb in precip-
itates is delayed relative to the oxidation of Zr. This is in accordance with previous
transmission electron microscopy observations [65] and electrochemical measure-
ments [66]. Zirconium is preferentially oxidized and alloying elements in precipi-
tates are protected up to a certain distance from the oxide–metal interface where
the oxygen potential is high enough to oxidize Fe or Nb in precipitates. As reported
in the literature, first Cr is oxidized and metallic Fe segregates to form bcc Fe [52].
In general, once the oxidation potential of alloying elements in precipitates is
reached, they start oxidizing.
The oxidation model based on the lXANES results proposed in Ref 52 is
recalled here. Right after the oxide transition, the oxide layer is fully protective and
a continuous oxygen potential gradient across the oxide layer is established. The
boundary conditions at the interfaces fix the oxygen potentials at these locations.
As dp increases, the oxygen potential gradient decreases, so that dmet also increases,
but the ratio dmet/dp remains constant [see Fig. 10(a)]. As dp continues to increase,
dmet reaches a threshold value and remains constant. Thus, the ratio dmet/dp
decreases and the oxygen potential in the outer part of the protective oxide
increases [see Fig. 10(b)]. The threshold value of dmet depends on the alloy and is
FIG. 10 Schematic evolution of the oxidation of precipitates in the zirconium oxide layer
as a function of the oxygen partial pressure across the oxide: (a) before dmet
reached its threshold value; (b) after dmet reached its threshold value.
smaller for Zircaloy-4 (800 nm) than for ZIRLO (1.0 lm), and the value for
ZIRLO is smaller than dmet in Zr-2.5Nb (1.2 lm) and Zr-0.4Fe-0.2Cr (H) alloy
(at least 1.25 lm). The oxygen boundary conditions are likely to be the same among
the different alloys, and the different oxidation potential of precipitates cannot
explain the observed variations in the dmet threshold value [67]. Thus, at a given ox-
ide thickness, the oxygen partial pressure in the Zr-2.5Nb oxide layer is lower than
the oxygen partial pressure in ZIRLO, which is in turn lower than the oxygen par-
tial pressure in Zircaloy-4 oxide layers. When the oxide transition occurs and the
oxide is no longer protective and is permeable to water, so that dmet drops to zero
(Fe and Nb are fully oxidized).
It is thought that the development of porosity in the protective oxide layers is
responsible for the fact that dmet reaches a threshold. The evolution of the micro-
porosity in zirconium oxide layers as a function of oxide thickness has been exten-
sively studied [26,68–72] and the reported values of an oxide thickness free of
interconnected pores (0.8 lm to 1.2 lm) are in agreement with our reported values
of dmet as a function of dp. Before transition, the formation of a connected network
of pores would increase the partial pressure of oxygen in the outer part of the pro-
tective oxide layer. Once the precipitates are embedded in the outer part of the ox-
ide layer where pores are interconnected, they will oxidize. dmet represents the
boundary between the inner part of the oxide, free of interconnected pores, and the
outer part in which pores are interconnected. It is of course tempting to relate dmet
to the concept of a barrier layer as defined by other authors [11], but it is necessary
to carefully define the barrier layer concept, as different techniques will yield differ-
ent values of dmet.
Yilmazbayhan et al. have shown that for the same alloys used in this study, the
higher the post-transition corrosion rate, the smaller the oxide thickness at transi-
tion [20]. The Zircaloy-4 transition being the earliest (and its corrosion rate the
highest) among the alloys studied, its level of porosity at a given oxide thickness
would be the highest among the considered alloys [71]. Different levels of porosity
among the alloys (highest for Zircaloy-4 and lowest for Zr-2.5Nb) would explain
the different threshold values of dmet (smallest for Zircaloy-4 and highest for
Zr-2.5Nb).
Effect of Precipitate Oxidation on Hydrogen Pickup Fraction

The ratio dmet/dp and the instantaneous hydrogen pickup fraction fHi are plotted in
Fig. 11 as a function of dp for all the alloys (fHi is plotted in a similar fashion as in
Fig. 8). The transition thickness for Zr-0.4Fe-0.2Cr (H) is unknown, and the third
transition of Zircaloy-4 is not precisely known. The total hydrogen pickup fraction
is plotted in the case of Zr-0.4Fe-0.2Cr (H). In Fig. 11(b), the pickup fraction at 0 lm
is the total hydrogen pickup fraction from t ¼ 0 days to the time at the first transi-
tion [Fig. 11(c) shows the same for the second transition].
It appears that variations in dmet/dp and fHi are connected. At first, the concen-
tration of metallic precipitates in the protective oxide layer, represented by the ratio
ox
dmet/dp, is constant, so that re is constant until dmet reaches its threshold value.
According to the hypothesis that hydrogen pickup results from the need to balance
charge, fHi would remain constant as observed in Figs. 11(b) and 11(d). As explained in
the preceding section, because of the development of microporosity in the outer
part of the protective oxide layer, the ratio dmet/dp decreases and the concentration
of metallic precipitates acting as local electric shortcuts in the oxide layer decreases.
ox
As a result, re would also decrease, and fHi would increase up to transition. This
effect is observed in all the plots in Fig. 11. Eventually, the same mechanism will take
place in the next transition period.
However, the observed increase in fH from one transition period to the next
cannot be fully explained by this mechanism, because the growth of the oxide layer
and the oxidation of alloying elements are periodic processes. The likely explana-
tion for this difference is that one of the boundary conditions is changing after tran-
sition. One possibility is that the presence of the outer porous oxide changes the
FIG. 11 Instantaneous hydrogen pickup fraction (at the bottom) and the ratio dmet/dp of
Fe and/or Nb (at the top) as a function of the protective oxide thickness dp (in
micrometers) for various alloys and transition regimes: (a) Zircaloy-4 pre-
transition (b) Zircaloy-4 in the transition period; (c) Zircaloy-4 before the
second transition period; (d) ZIRLO; (e) Zr-2.5Nb; and (f) Zr-0.4Fe-0.2Cr (H).
The shaded area shows the transition thickness for the different alloys
(unknown for Zr-0.4Fe-0.2Cr (H) and Zircaloy-4 at the third transition). In the
case of Zr-0.4Fe-0.2Cr (H), the total hydrogen pickup fraction is plotted instead
of the instantaneous hydrogen pickup fraction.
boundary condition at the outer limit of the protective oxide. Previous studies have
indicated that hydrogen overpressure increases hydrogen pickup in zirconium
alloys [10,73]. In the pre-transition regime, the cathodic site is directly in contact
with water, so that the hydrogen pressure at the cathodic site and that in the water
are equal. However, when a non-protective oxide layer is present on the top of the
growing protective oxide, the hydrogen evolved at the cathodic site has to diffuse
through these layers to finally reach the water. As a result, a hydrogen pressure gra-
dient will be established across the non-protective oxide layers. This will cause
hydrogen overpressure at the hydrogen evolution site to build up, leading to an
increase in proton concentration at this location. Higher concentrations of protons
at the interface could lead to an increase in hydrogen pickup fraction such as seen
in Fig. 5.
Conclusion
Detailed measurements were performed for hydrogen pickup and oxide growth as a
function of exposure time for a set of chosen zirconium alloys with specific chemis-
tries and microstructures using vacuum hot extraction (VHE) and cold neutron
prompt gamma activation analysis (CNPGAA). The variations of the oxidation
states of Fe and Nb as a function of oxide depth in these samples were investigated
by means of x-ray absorption near edge spectroscopy (XANES) using microbeam
synchrotron radiation.
1. The hydrogen pickup fraction is linked to the corrosion kinetics but does not
follow it exactly. Results are consistent with the hypothesis that the higher the
oxide electronic conductivity, the higher the corrosion rate and the lower the
pickup fraction. Hydrogen pickup during corrosion results from the need to
balance charge, causing hydrogen pickup to increase when the rate of electron
transport through the protective oxide decreases. According to this, oxide elec-
ox
tronic conductivity (re ) plays a key role in the hydrogen pickup mechanism,
ox
such that alloy oxides with higher re result in a lower hydrogen pickup frac-
tion, and vice versa.
2. Nb additions generally reduce the hydrogen pickup fraction. Following the
previously stated hypothesis, it is proposed that the donor effect of Nb in
ox
solid solution increases re , thereby reducing hydrogen pickup. The oxida-
tion of Fe and Nb when incorporated into the oxide is delayed relative to
zirconium oxidation. Thus metallic precipitates are embedded in the grow-
ing oxide up to a thickness of dmet. It is proposed that metallic precipitates
embedded in the protective oxide reduce hydrogen pickup by increasing
ox
re , possibly by acting as local electric shortcuts.
3. An oxidation model of precipitates has been developed: an inner layer
(thickness ¼ dmet) in which most of the alloying elements in precipitates are
still metallic develops as the oxide grows. At first dmet increases as the protec-
tive oxide layer thickens. After the oxide layer grows up to a threshold value of
around 1 lm to 1.5 lm, dmet reaches a constant value that lasts until the oxide
reaches transition, when dmet drops to zero. This threshold value is alloy
dependent and has been connected to the development of microporosity in

the oxide. The alloys scale as follows: the higher the corrosion rate, the higher
the porosity and the lower the dmet.
4. The evolution of dmet/dp is connected to the evolution of the instantaneous
hydrogen pickup fraction in all the studied alloys. The increase in instantane-
ous hydrogen pickup fraction as the corrosion rate slows down is rationalized
to the delayed oxidation of precipitates in the protective oxide layer.
ACKNOWLEDGMENTS
The writers thank R. Paul at NIST for CNPGAA measurements and Z. Cai for his
expert assistance in lXANES experiments at APS. This research was funded by EPRI
and Westinghouse Electric Co. LLC. Usage of the Advanced Photon Source was sup-
ported by the U.S. Department of Energy, Office of Basic Energy Sciences, under Con-
tract No. DE-AC02-06CH11357. The writers also thank the community of the
MUZIC-2 program for support and helpful discussions. The writers thank K. Saka-
moto from Nippon Nuclear Fuel Development and G. Kuri from Paul Scherrer Insti-
tute for helpful discussions. The writers thank Benoit de Gabory and Aditya
Shivprasad for their assistance in performing the XANES experiments.
References
[1] “Corrosion of Zirconium Alloys in Nuclear Power Plants,” IAEA-TECDOC-684, Interna-

tional Atomic Energy Agency, Vienna, 1993.
[2] “Waterside Corrosion of Zirconium Alloys in Nuclear Power Plants,” IAEA-TECDOC-996,

International Atomic Energy Agency, Vienna, 1998.
[3] Cox, B., “Some Thoughts on the Mechanisms of In-reactor Corrosion of Zirconium
[4] Kass, S., “Hydrogen Pickup in Various Zirconium Alloys during Corrosion Exposure in
High-Temperature Water and Steam,” J. Electrochem. Soc., Vol. 107, 1960, pp. 594–597.
[5] Klepfer, H. H., “Hydrogen Uptake of Zirconium Alloys during Water and Steam
Corrosion,” Corrosion, Vol. 19, 1963, p. 285.
[6] Berry, W. E., Vaughan, D. A., and White, E. L., “Hydrogen Pickup During Corrosion of Zir-
conium Alloys”, Corrosion, Vol. 17, 1961, p. 109.
[7] Cox, B., “A Mechanism for the Hydrogen Uptake Process in Zirconium Alloys,” J. Nucl.
Mater., Vol. 264, 1999, pp. 283–294.
[8] Kass, S. and Kirk, W. W., “Corrosion and Hydrogen Absorption Properties of Nickel-free
Zircaloy-2 and Zircaloy-4”, ASM Trans. Q., Vol. 55, 1962, pp. 77–100.
[9] Hatano, Y., Sugisaki, M., Kitano, K., and Hayashi, M., “Role of Intermetallic Precipitates in
Hydrogen Transport through Oxide Films on Zircaloy,” Zirconium in the Nuclear Industry:
12th International Symposium, ASTM STP 1354, ASTM International, Philadelphia, PA,
2000, pp. 901–917.
[10] Hillner, E., Hydrogen Absorption in Zircaloy During Aqueous Corrosion, Effect of Environ-
ment, WAPD-TM-411, AEC Research and Development, Pittsburgh, PA, 1964.
[11] Adamson, R., Garzarolli, F., Cox, B., Strasser, A., and Rudling, P., Corrosion Mechanisms in
Zirconium Alloys, A.N.T. International, Sweden, 2007.
[12] Kakiuchi, K., Itagaki, N., Furuya, T., Miyazaki, A., Ishii, Y., Suzuki, S., Terai, T., and Yama-
waki, M., “Effect of iron on hydrogen absorption properties of zirconium alloys”, J. Phys.
Chem. Solids, Vol. 66, 2005, pp. 308–311.
[13] Harada, M. and Wakamatsu, R., “The Effect of Hydrogen on the Transition Behavior of the
Corrosion Rate of Zirconium Alloys,” Zirconium in the Nuclear Industry: 15th International
Symposium, ASTM STP 1505, ASTM International, West Conshohocken, PA, 2008, p. 384.
[14] Grandjean, A. and Serruys, Y., “Metal and Oxygen Mobilities During Zircaloy-4 Oxidation
at High Temperature,” J. Nucl. Mater., Vol. 273, 1999, pp. 111–115.
[15] Cox, B. and Pemsler, J. P., “Diffusion of Oxygen in Growing Zirconia Films,” J. Nucl.
Mater., Vol. 28, 1968, pp. 73–78.
[16] Lemaignan, C. and Motta, A. T., Zirconium Alloys in Nuclear Applications, B. R. T. Frost,
Ed., Vol. 10B, Material Science and Technology Series, VCH, New-York, R. W. Cahn, P.
Haasen and E. J. Kramer, Eds., 1994, pp. 1–51.
[17] Hauffe, K., Oxidation of Metals, Plenum Press, New York, 1965.
[18] Sabol, G. P. and Dalgaard, S. B., “The Origin of the Cubic Rate Law in Zirconium Alloy
Oxidation,” J. Electrochem. Soc., Vol. 122, 1975, pp. 316–317.
[19] Porte, H. A., Schnizlein, J. G., Vogel, R. C., and Fischer, D. F., “Oxidation of Zirconium and
Zirconium Alloys”, J. Electrochem. Soc., Vol. 107, 1960, pp. 506–515.
of Zirconium Alloy Oxides Formed in Pure water Studied with Synchrotron Radiation
and Optical Microscopy: Relation to Corrosion Rate,” J. Nucl. Mater., Vol. 324, 2004, pp.
6–22.
[21] Motta, A. T., Gomes Da Silva, M. J., Yilmazbayhan, A., Comstock, R. J., Cai, Z., and Lai, B.,
“Microstructural Characterization of Oxides Formed on Model Zr Alloys Using Synchro-
tron Radiation,” Zirconium in the Nuclear Industry: 15th International Symposium, ASTM
STP 1505, ASTM International, West Conshohocken, PA, 2009, p. 486.
[22] Bradhurst, D. H., Draley, J. E., and Van Drunen, C. J., “An Electrochemical Model for the
Oxidation of Zirconium,” J. Electrochem. Soc., Vol. 112, 1965, pp. 1171–1177.
[23] Cox, B., “Rate Controlling Processes During the Pre-transition Oxidation of Zirconium
Alloys”, J. Nucl. Mater., Vol. 31, 1969, pp. 48–66.
[24] Cox, B., Rate Controlling Processes During the Oxidation of Zirconium Alloys, AECL-
2777, Atomic Energy of Canada Limited, Chalk River, ON, Canada, 1967.
[25] Frank, H., “Transport Properties of Zirconium Alloy Oxide Films”, J. Nucl. Mater., Vol.
306, 2002, pp. 85–98.
[26] Ramasubramanian, N., Billot, P., and Yagnik, S., “Hydrogen Evolution and Pickup During
the Corrosion of Zirconium Alloys: A Critical Evaluation of the Solid State and Porous
Oxide Electrochemistry,” Zirconium in the Nuclear Industry: 13th International Sympo-

sium, ASTM STP 1423, ASTM International, Philadelphia, PA, pp. 222–244, 2002.
[27] Beie, H.-J., Mitwalsky, A., Garzarolli, F., Ruhmann, H., and Sell, H. J., “Examinations of the
Corrosion Mechanism of Zirconium Alloys,” Zirconium in the Nuclear Industry: 10th Inter-
national Symposium, ASTM STP 1245, ASTM International, Philadelphia, PA, 1994, pp.
615–643.
[28] Shirvington, P. J., “Electron Conduction Through Oxide Films on Zircaloy,” J. Nucl. Mater.,
Vol. 37, 1970, pp. 177–202.
[29] Howlader, M. M. R., Shiiyama, K., Kinoshita, C., Kutsuwada, M., and Inagaki, M., “The Elec-
trical Conductivity of Zircaloy Oxide Films,” J. Nucl. Mater., Vol. 253, 1998, pp. 149–155.
[30] Hartman, T. E., Blair, J. C., and Bauer, R., “Electrical Conduction through SiO Films,” J.
Appl. Phys., Vol. 37, 1966, pp. 2468–2474.
[31] Simmons, J. G., “Poole-Frenkel Effect and Schottky Effect in Metal-Insulator-Metal Sys-
tems,” Phys. Rev., Vol. 155, 1967, pp. 657–660.
[32] Ramasubramanian, N., “Localised Electron Transport in Growing Zirconium Alloys,” J.

Nucl. Mater., Vol. 55, 1975, pp. 134–154.
[33] Baur, K., Garzarolli, F., Ruhmann, H., and Sell, H.-J., “Electrochemical Examinations in
350 C Water with Respect to the Mechanism of Corrosion-hydrogen Pickup,” Zirconium
in the Nuclear Industry: 12th International Symposium, ASTM STP 1354, ASTM Interna-
[34] Gohr, H., Schaller, J., Ruhmann, H., and Garzarolli, F., “Long-term In Situ Corrosion Inves-
tigation of Zr Alloys in Simulated PWR Environment by Electrochemical Measurements,”
Zirconium in the Nuclear Industry: 11th International Symposium, ASTM STP 1295, ASTM
[35] Harding, J. H., “The effect of alloying elements on Zircaloy corrosion”, J. Nucl. Mater.,
Vol. 202, 1993, pp. 216–221.
[36] Taylor, D. F., “An Oxide-semiconductance Model of Nodular Corrosion and its Applica-
tion to Zirconium Alloy Development”, J. Nucl. Mater., Vol. 184, 1991, pp. 65–77.
[37] Urquhart, A. W., Vermilyea, D. A., and Rocco, W. A., “Mechanism for the Effect of Heat-
Treatment on the Accelerated Corrosion of Zircaloy-4 in High Temperature, High Pres-
sure Steam,” J. Electrochem. Soc., Vol. 125, 1978, pp. 199–204.
[38] Smith, T., “Kinetics and Mechanism of Hydrogen Permeation of Oxide Films on
Zirconium,” J. Nucl. Mater., Vol. 18, 1966, pp. 323–336.
[39] Malki, B., Le Bacq, O., and Pasturel, A., “Ab initio Study of Hydrogen Related Defect in
ZrO2: Consequences on Dry and Aqueous Oxidation,” J. Nucl. Mater., Vol. 416, 2011, pp.
362–368.
[40] Draley, J. E. and Ruther, W. E., “Some Unusual Effects of Hydrogen in Corrosion Reac-
tions,” J. Electrochem. Soc., Vol. 104, 1957, pp. 329–333.
[41] Kiselev, A. A., Research on the Corrosion of Zirconium Alloys in Water and Steam at High
Temperature and Pressure, AECL-1724, Atomic Energy of Canada Limited, Chalk River,
Ontario, CA 1963.
[42] Choo, K.-N., Pyun, S.-I., and Kim, Y.-S., “Oxidation and Hydrogen Uptake of Zr Based Nb Alloys
at 400 C Under 10 MPa H20 Steam Atmosphere”, J. Nucl. Mater., Vol. 226, 1995, pp. 9–14.
[43] McIntyre, N. S., Davidson, R. D., Weisener, C. G., Good, G. M., Mount, G. R., Warr, B. D.,
and Elmoselhi, M., “Migration of hydrogen through thin films of ZrO2 on Zr-Nb alloy”
J. Vac. Sci. Technol. A, Vol. 9, 1991, pp. 1402–1405.
[44] Bossis, P., Pecheur, D., Hanifi, K., Thomazet, J., and Blat, M., “Comparison of the High
Burn-up Corrosion on M5 and Low Tin Zircaloy-4,” Zirconium in the Nuclear Industry:
14th International Symposium, ASTM STP 1467, ASTM International, West Conshohocken,
PA, 2005, pp. 494–524.
[45] Cox, B., “Hydrogen Absorption by Zircaloy-2 and Some Other Alloys during Corrosion in
Steam”, J. Electrochem. Soc., Vol. 109, 1962, pp. 6–12.
[46] Couet, A., Motta, A. T., and Comstock, R. J., “Hydrogen Pickup Measurements in Zirco-
nium Alloys: Relation to Corrosion Rate,” J. Nucl. Mater. (submitted).
[47] Gomes Da Silva, M. J., 2007, “Influence of Oxide Microstructure on Corrosion Behavior
of Zirconium-based Model Alloys,” Ph.D. thesis in Nuclear Engineering, Penn State Uni-
versity, University Park, PA.
[48] ASTM G2/G2M-06: Standard Test Method for Corrosion Testing of Products of Zirco-
nium, Hafnium, and Their Alloys in Water at 680 F or in Steam at 750 F, Annual Book of
ASTM Standards, ASTM International, West Conshohocken, PA, 2011.
[49] ASTM E146-83: Methods of Chemical Analysis of Zirconium and Zirconium Alloys (Sili-
con, Hydrogen, and Copper), Annual Book of ASTM Standards, ASTM International, West
Conshohocken, PA, 1989.
[50] Couet, A., Motta, A. T., Comstock, R. J., and Paul, R. L., “Cold Neutron Prompt Gamma
Activation Analysis, a Non-destructive Technique for Hydrogen Level Assessment in Zir-
conium Alloys,” J. Nucl. Mater., Vol. 425, 2012, pp. 211–217.
[51] Wiese, H., 1999, “Fractionated Determination of Hydrogen in Corroded Zirconium

Alloys,” Ph.D. thesis, Paul Scherrer Institute, Villigen.
[52] Couet, A., Motta, A. T., De Gabory, B., and Cai, Z., “X-ray Absorption Near-edge Spec-
troscopy Study of Fe and Nb Oxidation States Evolution in Zircaloy-4, ZIRLO, Zr-2.5Nb
and Zr-0.4Fe-0.2Cr Oxide Layers,” J. Nucl. Mater. (submitted).
[53] Ravel, B. and Newville, M., “Athena, Arthemis, Hephaestus: Data Analysis for X-Ray Absorp-
tion Spectroscopy Using IFEFFIT,” J. Synchrotron Radiat., Vol. 12, 2005, pp. 537–541.
[54] Henke, B. L., Gullikson, E. M., and Davis, J. C., At. Data Nucl. Data Tables, Vol. 54, 1993,
pp. 181–342.
[55] Charquet, D., Hahn, R., Ortlieb, E., Gros, J.-P., and Wadier, J.-F., “Solubility Limits and For-
mation of Intermetallic Precipitates in ZrSnFeCr Alloys,” Zirconium in the Nuclear Indus-
try: 8th Symposium, ASTM STP 1023, ASTM International, West Conshohocken, PA, 1988,
pp. 405–422.
[56] Yilmazbayhan, A., Delaire, O., Motta, A. T., Birtcher, R. C., Maser, J. M., and Lai, B.,
“Determination of the alloying content in the matrix of Zr alloys using synchrotron radia-
tion microprobe X-ray fluorescence”, J. Nucl. Mater., Vol. 321, 2003, pp. 221–232.
[57] Sakamoto, K., Une, K., and Aomi, M., “Chemical State of Alloying Elements in Oxide
Layer of Zr-based Alloys,” ANS LWR Fuel Performance/Top Fuel/WRFPM, Orlando, FL,
Sept 26–29, 2010.
[58] Sakamoto, K., Une, K., Aomi, M., and Hashizume, K., “Depth Profile of Chemical States of
Alloying Elements in Oxide Layer of Zr-based Alloys” Prog. Nucl. Energy, Vol. 57, 2012,
pp. 101–105.
[59] Murai, T., Isobe, K., Takizawa, Y., and Mae, Y., “Fundamental Study on the Corrosion
Mechanism of Zr-0.2Fe, Zr-0.2Cr and Zr-0.1Fe-0.2Cr Alloys,” Zirconium in the Nuclear
Industry: 12th International Symposium, ASTM STP 1354, ASTM International, West Con-
shohocken, PA, 2000, pp. 623–640.
[60] Broy, Y., Garzarolli, F., Seibold, A., and Van Swam, L. F., “Influence of Transition Elements
Fe, Cr, and V on Long-time Corrosion in PWRs,” Zirconium in the Nuclear Industry: 12th
International Symposium, ASTM STP 1354, ASTM International, West Conshohocken, PA,
2000, pp. 609–622.
[61] Abriata, J. P. and Bolcich, J. C., “The Zr-Nb (Zirconium-Niobium) System,” Bull. Alloy
Phase Diagrams, Vol. 3, 1982, pp. 1710–1712.
[62] Froideval, A., Abolhassani, S., Gavillet, D., Grolimund, D., Borca, C., Krbanjevic, J., and
Degueldre, C., “Microprobe Analysis of Neutron Irradiated and Autoclaved Zirconium Ni-
obium Claddings Using Synchrotron-based Hard X-ray Imaging and Imaging,” J. Nucl.
Mater., Vol. 385, 2009, pp. 346–350.
[63] Kim, B.-Y., Park, C.-J., and Kwon, H.-S., “Effect of Niobium on the Electronic Properties
of Passive Films on Zirconium Alloys,” J. Electroanal. Chem., Vol. 576, 2005, pp.
269–276.
[64] Pêcheur, D., “Oxidation of Beta-Nb and Zr(Fe,V)2 Precipitates in Oxide Films Formed on
Advanced Zr-based Alloys,” J. Nucl. Mater., Vol. 278, 2000, pp. 195–201.
[65] Pêcheur, D., Lefebvre, F., Motta, A. T., Lemaignan, C., and Wadier, J. F., “Precipitate Evolu-
tion in the Zircaloy-4 Oxide Layer,” J. Nucl. Mater., Vol. 189, 1992, pp. 318–332.
[66] Barberis, P., Ahlberg, E., Simic, N., Charquet, D., Dahlback, M., Limback, M., Tagstrom, P.,
Lemaignan, C., Wikmark, G., and Lehtinen, B., “Role of Second Phase Particles in Binary
Zirconium Alloys,” Zirconium in the Nuclear Industry: 13th International Symposium,
ASTM STP 1423, ASTM International, West Conshohocken, PA, 2001, pp. 33–58.
[67] Proff, C., Abolhassani, S., and Lemaignan, C., “Oxidation Behaviour of Zirconium Alloys
and Their Precipitates — A Mechanistic Study,” J. Nucl. Mater., Vol. 432, 2013, pp. 222–238.
[68] Bossis, P., Lelievre, G., Barberis, P., Iltis, X., and Lefebvre, F., “Multi-scale Characterization
of the Metal-Oxide Interface of Zirconium Alloys,” Zirconium in the Nuclear Industry:
PA, 2000, pp. 918–947.
[69] Cox, B., “Pore Structure in Oxide Films Irradiated and Unirradiated Zirconium Alloys,” J.
Nucl. Mater., Vol. 148, 1987, pp. 332–343.
[70] Ploc, R. A. and Newcomb, S. B., “Porosity in Zr-2.5Nb Corrosion Films,” Microscopy of
Oxidation 3: Proceedings of the Third International Conference on the Microscopy of Oxi-
dation, Cambridge, UK, Sept 16–18, 1996, pp. 475–487.
[71] Ni, N., Lozano-Perez, S., Jenkins, M. L., English, C., Smith, G. D. W., Sykes, J. M., and Gro-
venor, C. R. M., “Porosity in Oxides on Zirconium Fuel Cladding Alloys, and its Impor-
tance in Controlling Oxidation Rates”, Scr. Mater., Vol. 62, 2010, pp. 564–567.
[72] Une, K., Sakamoto, K., Aomi, M., Matsunaga, J., Etoh, Y., Takagi, I., Miyamura, S., Kobaya-
shi, T., and Ito, K., “Hydrogen Absorption Mechanism of Zirconium Alloys Based on Char-
acterization of Oxide Layer,” Zirconium in the Nuclear Industry: 16th International
Industry, ASTM STP 1529, ASTM International, West Conshohocken, PA, 2011, pp.
401–432.
[73] Cox, B., “Assessment of In-Reactor Corrosion Models and Data for Zircaloys in Water,”
2nd International Symposium on Environmental Degradation of Materials in Nuclear
Power Systems—Water Reactors, Monterey, CA, Sept 9–12, 1985, pp. 219–226.
DISCUSSION
Q1:—How does the differential hydrogen pickup fraction vary with weight
gain?
Authors’ Response:—The differential hydrogen pickup fraction is defined as the

instantaneous hydrogen pickup fraction fHi , which is equal to:
dHabsorbed
DtþDt Habsorbed dHabsorbed
fHi ¼ dHgenerated
dt
lim t
1
Ds!0 DtþDt Hgenerated dd
dt t
Variations of this derivative as function of exposure time can be observed in

Figure 8 on ZIRLO. On the same figure, weight gain is plotted so that we can have a
general idea of how fHi behaves as a function of weight gain wg . To have a better
description of this dependence, in the following figure we plotted the fit of fHi as
function of the fit of weight for ZIRLO (the fits details are provided in [1])
Instantaneous hydrogen pickup fraction as function of oxide thickness (or

weight gain) follows a general trend common to every alloy studied in this paper:
* At small protective oxide thickness (<40% of oxide thickness at transition),

fHi increases with oxide thickness.
* Between 40% and 70% of oxide thickness at transition, f i reaches a plateau.
H
* Above 70% of oxide thickness at transition, f i starts to steadily increase
H
again, reaching a peak and decrease right before oxide transition.
Q2:—Does hydrogen evolution, assisted by electron transport via segregated

iron, occur at the oxide coolant interface?
Authors’ Response:—It is believed that the hydrogen evolution reaction leading

to the formation of hydrogen gas occurs at the oxide/coolant interface. However,
some hydrogen from the corrosion reaction enters the oxide under the form of pro-
tons and recombine either in the oxide or at the metal/oxide interface and not at
the oxide/coolant interface. The location of these cathodic sites inside the oxide and
the fate of hydrogen evolved in the oxide are not known at the moment.
Q3:—Protectiveness and pickup fraction: the more protective the oxide, the
higher the pickup fraction? Does this mean protective for electron transport
(insulating)?
Authors’ Response:—Indeed, the conclusions of our study seem to show this

correlation: the more insulating the oxide is, the higher the hydrogen pickup frac-
tion since the transport of electrons is more difficult. The difficulty of electron
transport is transcribed as a higher electric potential across the oxide, believed to be
the driving force for protons to get into the oxide. By adding a donor such as Nb5+
and/or metallic second phase particles in the insulating oxide, the insulating oxide
could be more conductive for electrons, lowering the electrical potential across the
oxide, so that the driving force for hydrogen pickup would be reduced.
Question from B. K. Shah, BARC Mumbai
Q1:—Does Cold Neutron Prompt Gamma Activation Analysis for non-destruc-

tive measurement of hydrogen in Zr alloys measure hydrogen in the surface layer
or is it a bulk analysis? What is the uncertainty in hydrogen measurement by this
method?
Authors’ Response:—This method measures the hydrogen concentration in the

whole thickness of the sample as long as the sample is not too thick (not thicker
than 1 cm or so). Indeed if it is too thick, neutron beam energy would vary inside of
the sample due to neutron thermalization and thus the reaction rate between
hydrogen and neutrons would change as function of sample depth, leading to erro-
neous hydrogen concentrations. Correction factors need to be precisely determined
before proceeding on thick samples. The beam size is approximately 2 cm in diame-
ter. Thus an average bulk concentration of hydrogen is obtained in the region where
the beam covers the sample [2].
The uncertainty depends on many parameters (geometry at the beamline, sig-
nal to noise ratio and fitting procedure of the results) but was on the order of
5wt.ppm at the maximum in our study [2–4].
Q2:—What is the effect of Fe impurity in Zr-2.5Nb alloy on hydrogen pickup?
Authors’ Response:—See response to J.-C. Brachet in the following.
Question from S. K. Jha, Nuclear Fuel Complex, Hyderabad:—What is the effect

of Fe in Zr-2.5Nb as far as hydrogen pickup is concerned?
Authors’ Response:—See response to J.-C. Brachet in the following.
Question from J. -C. Brachet, CEA Saclay, Nuclear Materials Dept. France:—
The presentation from D. Kaczorowski showed no significant effects of Fe content
on hydrogen pickup of M5 from 200 wt.ppm up to 1000 wt.ppm, suggesting a
factor of approximately five between the ZrNbFe volumetric fractions. How do you
explain this?
Authors’ Response:—The conclusions of the study show that oxidations of Nb

and/or Fe in precipitates (either Zr(Fe,Cr)2, ZrNb or ZrNbFe) are delayed. Both ele-
ments are incorporated in a metallic state in the oxide layer (either in the form of
precipitates as shown in the present paper or in solid solution in the case of Fe [5])
and their delayed oxidations are correlated to the increase in instantaneous hydro-
gen pickup fraction (see Figure 11). The fHi evolution is similar for every alloy, thus,
not the type of precipitates but rather their delayed oxidation (observed on every
studied alloys) has an impact on the increase in instantaneous hydrogen pickup
fraction.
We did not perform XANES on Fe impurity in Zr-2.5Nb. However it is
believed that Fe as an impurity in Zr-2.5Nb has little effect on hydrogen pickup, its
effect being screened by the dopant effect of Nb saturated in solid solution and the
metallic ZrNb precipitates. If there is sufficient Nb in solid solution, the Fe content
as an impurity (several hundreds of wt.ppm) would not play a significant role in the
hydrogen pickup mechanism in autoclave conditions. This is why it is mainly in its
solid solution range that Nb concentration has an effect on electronic conductivity
[6] and hydrogen pickup [7]. D. Kaczorowski’s paper included in these proceedings
confirms that Fe concentration up to 1000 wt.ppm in M5 (Zr-1.0Nb) has no no-
ticeable effect on corrosion and hydrogen pickup. A thorough study of the elec-
tronic levels of metallic/oxidized isolated elements and clusters in the zirconium
oxide band gap would be necessary to assess their effect on electronic transport.
Question from S. Ortner, Nuclear National Laboratories, U.K.:—Please clarify

the range of time over which dmes is increasing so we can compare with the time
over which the instantaneous hydrogen pickup fraction fHi is shown to be approxi-
mately constant, increasing or decreasing.
Authors’ Response:—Since the range of time, dmes , and fHi are upon the alloy de-
pendent, let us consider the alloy for which we have the most experimental data:
Zircaloy-4.
Up to dp 1.3 lm (19 days of exposure time), dmes increases up to 0.8 lm
(see Figure 9.a). In that range of time, dmes increases linearly with dp meaning that
the fraction of metallic particles in the oxide layer is constant (ddmes p
is constant as
seen in Figure 11.e). During the same period of time fHi is also approximately con-
stant (see Figures 11).
From dp 1.3 lm to transition at fHi 2.2 lm (from 19 days to 85 days of expo-
sure time), dmes is constant (0.8 lm for Zircaloy-4) but dp keeps on increasing. Thus
the ratio ddmes
p
decreases meaning that the fraction of metallic particles in the oxide layer
is decreasing. At the same time, fHi increases sharply as observed in Figures 11.
Eventually, the alloy reaches transition and dmes drops to zero. Even though
this is not seen on Zircaloy-4 since the time range in which dmes drops to zero is
extremely short (the alloy undergoes a sudden transition, confirmed by the weight
gain data in Figure 4.b), the decrease in dmes before transition appears in the other
alloys in Figure 9. Eventually the process repeats itself in the next transition regime
because the oxidation kinetics and hydrogen pickup kinetics are periodic.
Question from M. Griffiths, AECL:—You showed results for the copper-con-

taining alloy with high pickup fraction and low corrosion kinetics compared with
all other Fe containing alloys. What can you say about any other differences in the
microstructure (precipitate distribution in particular) that could affect the
behavior?
Authors’ Response:—SEM and TEM microscopy on Zr-2.5Nb-0.5Cu tube alloy

have been performed. Synchrotron X-ray diffraction experiments have also been
performed on that alloy and b-Nb, C14 Zr(Nb,Fe)2 and Zr(Nb,Cu)2 Laves phase
and tetragonal Zr2Cu phases have been detected [8]. Using scanning TEM and
SEM, Cu precipitates were characterized. Their sizes vary between 0.2 to 1 lm.
They are elongated along the rolling direction (parallel to oxide/metal interface)
and homogeneously distributed on the cross-section plane. The precipitate diameter
in a Zr-0.5Cu binary alloy has been determined by synchrotron X-ray diffraction
experiments and is on the order of 200 nm.
[1] A. Couet, et al., “Hydrogen Pickup Measurements in Zirconium Alloys:

Relation to Oxidation Kinetics,” Journal of Nuclear Materials, accepted.
[2] A. Couet, et al., “Cold neutron prompt gamma activation analysis, a non-
destructive technique for hydrogen level assessment in zirconium alloys,”
Journal of Nuclear Materials, vol. 425, pp. 211–217, 2012.
[3] E. A. Mackey, et al., “Sources of uncertainties in prompt gamma activation

analysis,” Journal of Radioanalytical and Nuclear Chemistry, vol. 265, pp.
273–281, 2005.
[4] R. Paul, “Hydrogen Measurement by Prompt Gamma-ray Activation Anal-

ysis: A Review,” Analyst, vol. 122, pp. 35R–41R, 1997.
[5] A. Couet, et al., “Title,” unpublished.
[6] P. Chan-Jin, et al., “Effect of niobium on the electronic properties of passive

films on zirconium alloys,” Journal of Electroanalytical Chemistry, vol. 576,
pp. 269–76, 2005.
[7] A. A. Kiselev, “Research on the Corrosion of Zirconium Alloys in Water

and Steam at High Temperature and Pressure,” Atomic Energy of Canada
Limited1963.
[8] A. T. Motta, et al., “Microstructural Characterization of Oxides Formed on

Model Zr Alloys Using Synchrotron Radiation,” in Zirconium in the Nu-
clear Industry: 15th International Symposium, 2009, p. 486.
Mikaela Lindgren,1 Gustav Sundell,2 Itai Panas,1

Lars Hallstadius,3 Mattias Thuvander,2
and Hans-Olof Andrén2
Toward a Comprehensive
Mechanistic Understanding
of Hydrogen Uptake in Zirconium
Alloys by Combining Atom Probe
Analysis With Electronic
Structure Calculations
Reference
Lindgren, Mikaela, Sundell, Gustav, Panas, Itai, Hallstadius, Lars, Thuvander, Mattias, and
Andrén, Hans-Olof, “Toward a Comprehensive Mechanistic Understanding of Hydrogen
Uptake in Zirconium Alloys by Combining Atom Probe Analysis With Electronic Structure
Calculations,” Zirconium in the Nuclear Industry: 17th International Symposium, STP 1543,
Robert Comstock and Pierre Barberis, Eds., pp. 515–539, doi:10.1520/STP154320120164, ASTM
ABSTRACT
The ability of a zirconium alloy to resist corrosion relies on a compromise
between two opposing strategies. Minimizing the hydrogen pickup fraction
(HPUF) by invoking metallic electron conduction in the barrier oxide results in
rapid parabolic oxide growth. On the other hand, slow sub-parabolic barrier
oxide growth, as reflected in rate limiting electron transport, may result in a high
HPUF. The objective of the present study is to offer mechanistic insights as to
how low concentrations of different alloying elements become decisive for the
Manuscript received November 19, 2012; accepted for publication June 2, 2014; published online September
22, 2014.
1
Department of Chemical and Biological Engineering, Chalmers Univ. of Technology, 412 96 Gothenburg,
Sweden.
2
Department of Applied Physics, Chalmers Univ. of Technology, 412 96 Gothenburg, Sweden.
3
Westinghouse Electric Sweden AB, 721 63 Västerås, Sweden.
4
Hyderabad, India.
Copyright V
overall corrosion behavior. Combining atomistic microanalysis with first

principles modeling by means of density functional theory, the speciation and
redox properties of Fe and Ni towards hydrogen evolution are firstly explored.
Complementary atom probe microanalysis at the metal–oxide interface provides
evidence for Fe and Ni segregation to grain boundaries in Zircaloy-2 that
propagates into the ZrO2 scale. Descriptors for how alloying elements in ZrO2
control electron transport as well as catalytic electron-proton recombination in
grain boundaries to form H2 are determined by means of theory. The findings are
generalized by further atomistic modeling, and are thus put in the context of
early reports from autoclave experiments on HPUFs of zirconium with the
alloying elements Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, and Nb. A shunting mechanism
which combines inner and outer hydrogen evolution mechanisms is proposed.
Properties of the transient zirconium sub-oxide are discussed. A plausible
atomistic overall understanding emerges.
Keywords
corrosion, hydrogen pickup, density functional theory, atom probe tomography,
zirconium, alloys, suboxide, hydride, HPU, HPUF
Introduction
It was recognized very early that the nuclear fuel cladding should constitute a zirco-
nium alloy, achieving both low cross section for neutron scattering of Zr and
improved mechanical properties. The influence of the alloying elements on the
rate of oxidation in reactor water was addressed soon after, as was the impact of
alloying elements on the resulting hydrogen pickup (HPU). In particular, depend-
ences of the hydrogen pickup fraction (HPUF) on transition metal (TM) addition—
including the series of 3d elements from titanium to zinc—in binary Zr alloys were
determined, see, e.g., Refs. [1,2].
This alloying addition dependence of the hydrogen pickup fraction may poten-
tially be of crucial importance in future strategies to improve on cladding perform-
ance. Consequently, the present effort aims to contribute a detailed mechanistic
understanding of this effect, which would serve as a platform for further investiga-
tions. Rather than a final answer, we seek an internally consistent working hypothe-
sis. In doing so, we employ empirical observations complemented by corresponding
first principles calculations. Any valid such understanding must thus connect to at
least four outstanding observations. One is the Zr–O phase diagram [3]. Hence, for
the temperatures relevant to light water reactors, thermodynamically stable solid
solutions of oxygen in hexagonally close packed zirconium is possible up to ZrOy
y < 0:5, i.e., molar concentration XðOÞ < 33 at: % Yet, at different stages during
oxidation all transient oxygen concentrations 0 X 66% including super-
saturation are observed locally [4]. Second is the quasi-periodic nature of the mass
gain curve [4,5]. Indeed, the thermodynamic limit, eventually resulting in Zr oxida-
tion by water to form ZrO2 and H2, is ideally approached in periodically repeating
LINDGREN ET AL., DOI 10.1520/STP154320120164 517
cycles [4,5]. Distinctly different oxygen concentration profiles through the ZrO2/Zr
interface early in the period versus late have been reported, i.e., fairly sharp early
and more diffuse late in each period [4]. Third is recent findings concerning the
correlation between the rates of oxidation and hydrogen pickup [4,6]. The oxida-
tion of zirconium by water has hydrogen evolution (HE) as well as HPU as neces-
sary consequences. Interestingly, an anti-correlation between mass gain and
instantaneous HPUF emerges from Refs. [4,6], where in the case of Ref. [4], the
hydrogen concentration profile was also found to display a distinct hump in the
vicinity of the ZrO2/Zr interface. Early, this hump is situated at the interface, while
just prior to transition it moves from the effective interface into the Zr suboxide.
This relative displacement is apparently associated with a peak in the instantaneous
HPUF. Four is the sensitive dependence of HPUF on alloying elements in binary Zr
model alloys [1,2] which is shown below and in Ref. [7] to correlate to the ener-
getics for the hydrogen evolution reaction channel, which has hydride H– reacting
with proton Hþ to form H2 at hydroxylated ZrO2 grain boundaries in the vicinity
of the ZrO2/Zr interface.
This paper is outlined as follows: the second section describes the fundamental
experimental observations including the speciations of aliovalent ions. In addition, it
explains how the employed transition metals decorated hydroxylated ZrO2 grain
boundary is modeled. The experimental and computational details are also presented.
The third section provides the characteristics for hydrides in hydroxylated ZrO2 grain
boundaries associated with transition metal additions, i.e., electronic structures for
understanding and vibrational frequencies for identification. In addition, in the third
section, the computed addition dependent energetics for hydrogen evolution resulting
from hydride–proton recombination is presented, and correlated to corresponding
experimental addition dependent hydrogen pickup fractions. For completion, ener-
getics for Zr(Ox) suboxide formation from Zr and ZrO2 is touched upon in the fourth
section, where the impact on H dissolution in some suboxides is provided in conjunc-
tion with characteristic vibrational frequencies for identification. The fifth section
addresses the influence of Nb doping on the electronic conduction properties of the
oxide where Nb is taken to substitute for Zr in ZrO2. The sixth section seeks to illus-
trate the emerging internally consistent working hypothesis by connecting the
obtained results to observations that have been made over the years.
Density functional theory (DFT) is utilized as the central tool in this endeavor
providing “gedanken experiments” as well as illustrations of the emerging concep-
tual understanding, while relevance is ensured here by atom probe tomography
(APT) measurements.
Modeling Considerations
Detailed understanding of the atomistic origin of the HPUF [1,2] implies resorting
to electronic structure calculations. The complexity of the problem is reflected in
that the local electrochemistry of protons with redox active metal ions presenting
“impurity states” in the large band gap ZrO2 material is sought. The atomic origin
of the HPUF, and in particular the relevance of the local TM associated electronic
structure, was proposed in Ref. [2]. Connection to catalysts was also made in
Ref. [2], and it was pointed out that “… we must take the unalloyed zirconium as
the reference basis for evaluating the influence of other elements on the properties
of ZrO2.” Inspired partly by these speculations, in what follows, robust characteris-
tics for the addition dependent hydrogen pickup and HPUF are sought at the
atomic level (Fig. 1). Overall, we understand water to penetrate to the metal–oxide
interface by hydrolysis, resulting in hydroxylated grain boundaries (e.g., Refs. [8,9]).
This is schematically represented in Figs. 1(a)–1(d) and 1(g), where an oxy-bridge is
hydrolyzed (Figs. 1(a) and 1(b)) followed by atomic oxygen uptake in Zr leaving two
electrons behind which in turn accommodate a proton (Figs. 1(c), 1(d), and 1(g)). By
including the protons explicitly in the model, the crucial connection between the
Zr(s) oxidation and the complementary proton reduction to form H is thus articu-
lated. Beyond Figs. 1(d) and 1(g), the proton reduction pathway splits into (a) a
hydrogen evolution pathway (Figs. 1(e) and 1(f)) where proton recombines with
hydride to form H2 (Figs. 1(d) and 1(e)) followed to the regeneration of the original
grain boundary (Figs. 1(e) and 1(f)), and (b) the complementary hydrogen pickup
channel (Figs. 1(h) and 1(k)) where the pickup of the first hydrogen is followed by
the second proton transforming into a grain boundary hydride (Figs. 1(g) and 1(h));
finally this second hydrogen becomes incorporated in the Zr matrix allowing for
the regeneration of the original grain boundary oxide (Figs. 1(h) and 1(i)). Hence, a
route for hydrogen pickup is provided but in addition we allow for the possibility of
H in the grain boundaries to be intercepted by incoming Hþ ions thus forming H2
in the grain boundary.
It is repeatedly stressed that for every oxygen atom that reacts with Zr to form
ZrO2, either one molecule of H2 is produced according to
(1) Zr þ 2H2 O ! ZrO2 þ 2H2
or two hydrogen atoms are picked up in the cladding. In what follows, we show
how this reaction occurs by formation of intermediate TM–H–Zr hydride, which
subsequently recombines with a hydroxide proton to form H2. These H2 molecules
are understood to experience constrained diffusion in the grain boundary.
GRAIN BOUNDARY MODEL

The employed computational model of a hydroxylated ZrO2 grain boundary is
composed of a super-cell, which contains about 50 atoms. It can be understood as a
transient hydroxylated sub-nanometer sized interface between oxide grains or as a
hydroxylated grain boundary. It is constructed by inserting one unit cell of mono-
clinic ZrO(OH)2 (5.4 Å by 10 Å by 5.4 Å) in between two supercells of monoclinic
ZrO2 (5.4 Å by 10.8 Å by 5.4 Å), where the unit cell doubling is in the b direction.
The starting ZrO(OH)2 structure in turn was obtained by replacing Liþ with Hþ in
Li2ZrO3. Interestingly, an excellent proton conductor is obtained experimentally
FIG. 1 Schematic representation of overall scenario. Region between vertical lines is

boundary between ZrO2 grains. Region below horizontal line is Zr metal. ((a) and
(b)) hydrolysis of grain boundary; ((c), (d), and (g)) displacement of oxygen
atom into Zr; (d) and (g) formation of hydride; (d) and (e) proton transfer,
resulting in H þ Hþ recombination; (e) and (f) H2 evolution and recovery of the
original grain boundary, and (g) and (i) hydrogen pickup path. Note that (e) does
not exist as an intermediate but only as a transition state, as the reaction
continues immediately to form the product (f). The hydrogen conversion ratio of
paths (d)–(f) and (g)–(i) determines the hydrogen pickup fraction (HPUF). TM(II)
represents a transition metal addition in oxidation state þII, the impacts of which
determines said ratio.
[10] by performing ion exchange on Li2ZrO3, replacing Liþ by Hþ. The obtained
hydroxylated grain boundary model utilized in the present study is displayed in
þ þ
Fig. 2. Additional justification for treating H and Li as interchangeable ions in
reactor is provided, e.g., in Refs. [8,9].
SPECIATION OF ALIOVALENT IONS

This study concerns the electrochemistry of dispersed metal ions at ZrO2 grain
boundaries. It is conceivable that as long as electric contact with the Zr metal is
maintained, all transition metal adatoms considered in the present study would be
in oxidation state 0. The fact that Zr ions in the model barrier oxide a priori display
formal oxidation state þIV, i.e., as reflecting local ZrO2 or ZrO(OH)2 environ-
ments, strongly suggests that a sub-monolayer of guest atoms would also be ionic
and furthermore occupy interstitial sites in the hydroxylated grain boundary owing
FIG. 2 Illustration of the hydroxylated grain boundary model in the electronic structure
calculations. It is constructed by fusing monoclinic ZrO2 with H2ZrO3 (see text).
Red: oxygen; blue: zirconium; white: hydrogen atom; dashed lines indicate
hydrogen bonding.
to their lower oxidation numbers than Zr(IV). Inevitably, the ionic adatoms become
coordinated to oxygen ions or hydroxide ions at the hydroxylated interface, and
while distinguishing between substitutional and interstitial sites at the hydroxylated
grain boundary is meaningless; still, when preference of aliovalent ions for the
interface relative to the ZrO2 bulk is observed, it underlines their preference for
non-substitutional sites. In case of Ti, V, Cr, and Mn several different oxidation
states are possible. In case of Co, Fe, and Ni, mainly þII and þIII are possible
owing to the low crystal field. In case of Cu and Zn only oxidation state, þII is
taken to be relevant. In the calculations, the energetics for HE is systematically
evaluated for the TMs in oxidation states þII and þIII. The oxidation states were
controlled by the appropriate removal of hydrogen atoms in the grain boundary
[11]. Thus, neutral unit cells are considered exclusively. In addition, substitutional
Nb(IV) doped bulk ZrO2 was examined, support for which was provided by an
experimental observation, see below.
COMPUTATIONAL DETAILS
The route taken to gain a fundamental understanding of key features in the oxida-
tion of Zr by water utilizes solid state electronic structure calculations by means of
DFT. This formalism builds on the Hohenberg–Kohn theorems [12], which state
that (a) all the properties of a system in its ground state may be extracted from the
electron density including the energy of the system, and (b) the electron density
uniquely determines the external potential, here the positions of the nuclei in
the solid. We resort to the former in order to compute reaction energies, and to the
latter when we ask for structural information. We employ Kohn–Sham DFT [13].
It implies that the electron density is formed from the simplest possible first-
principles wave function ansatz, which assumes the coordinates of each electron to
be independent of any other electron’s coordinates. It becomes the task of the
so-called exchange-correlation potential to extract the detailed energy contributions
owing to the detailed electron correlation. In the present study one of the most
commonly used functionals is employed comprising the Perdue et al. [14] general-
ized gradient approximation PBE GGA, which improves on the local density
approximation LDA, as implemented in the DMOL3 engine in the Material Studios
program package [15]. A double-f numerical basis set with extra polarization func-
tion on each heavy atom and a p-function on each hydrogen atom was employed in
all calculations on the hydroxylated ZrO2 grain boundary. Systematic spin polarized
calculations were performed where a 4 by 4 by 1 k-point set for sampling the
Brillouin zone was compared to a 2 by 2 by 1 k-point set, and the latter found to
suffice. In all cases the spin states of reactant and product corresponding to the
lowest energies were determined. Sometimes the spin state of the product did not
correlate with the spin state of the reactant in its ground state. In such cases, cancel-
lation of errors cannot be assumed owing to the so-called self-interaction error. In
those cases excited states were considered where such cancellations would be
expected. If the corresponding excitation energies come out small in comparison to
the overall reaction energy then the resulting reaction energy is taken to be qualita-
tively robust [16]. In order to reduce the computational effort, inert electrons were
described effectively by means of the semi-core pseudopotentials. For the explora-
tory calculations on Zr(s), bulk ZrO2, the suboxides Zr(Ox), and impact on hydro-
gen dissolution the CASTEP program package within the Material Studios
framework [17] was utilized and the PBE GGA functional [14] was employed. Core
electrons were described by ultra-soft pseudo-potentials [18] in conjunction with
300 eV cut-off energy for the plane wave basis set employed to span the explicitly
treated valence electrons. The cut-off energy was chosen after a computational
cost-benefit analysis, and care was taken throughout to employ suitable k-point sets
for sampling the Brillouin zone.
EXPERIMENTAL CONSIDERATIONS
Experimental results from APT are presented here. These key observations form
the basis as well as render relevant the theoretical study in general and the used
model in particular. For experimental details, see the section below. Firstly, APT of
the metal–oxide interface region in Zircaloy-2 reveals that Fe and Ni segregate to
planar features in the metal, which continue uninterrupted into the oxide. The Fe
quantity in the oxide grain boundary corresponds to 0.30 close-packed fcc planes,
and Ni quantity corresponds to 0.15 close-packed fcc planes. It should be noted
that these numbers may be slight underestimations of the true grain boundary con-
tent due to overlaps in the mass spectrum. The segregation is depicted in Fig. 3(a).
FIG. 3 APT observation of segregation of Fe (purple) and Ni (green) in the metal and in
the oxide (blue) in corroded pre-transition Zircaloy-2. The sizes of the boxes are
80 by 80 by 300 nm3 in (a), and 80 by 80 by 30 nm3 slice in (b). (a) depicts the
network of sub-grain boundaries that are typically found in the metal close to
the metal–oxide interface. A branch of the sub-grain boundary is seen to
continue uninterrupted into the oxide (blue). (b) depicts a triple grain boundary
found in the oxide approximately 100 nm from the metal–oxide interface.
These planar features are interpreted as sub-grain boundaries caused by creep

deformation of the metal, and their chemistry is subsequently inherited by
oxide grain boundaries as the metal is consumed [19]. There is no indication of
redistribution of Fe and Ni as the oxidation proceeds, as shown in Fig. 3(b) where
the decorated grain boundary continues for over 100 nm into the barrier layer of
the oxide. It should also be noted that no segregation of Cr to grain boundaries is
observed. The facts that (a) Fe and Ni segregate to the same grain boundaries and
(b) the difference in HPUF displayed by the Fe and Ni binary alloys is maintained
down to very low concentrations [20] could be taken to suggest that this difference
is of local and possibly atomic origin. This possibility is validated in the third sec-
tion of this paper by means of atomistic calculations.
Distribution of Nb in the metal–oxide interface region of ZIRLO is presented
in Fig. 4. In contrast to Fe, Cr and Ni, Nb does have some solubility in the Zr and
ZrO2 matrices, and a concentration of approximately 0.15 at. % is measured in the
oxide in the vicinity of the interface. Also in contrast to Ni and Fe, no segregation
of Nb in grain boundaries is found. Substitutional Nb doping in ZrO2 is studied
below in order to contrast the grain boundary controlled HPUF-scenario with a
bulk electron conductivity channel for HE at the water/oxide interface. Yet it must
be emphasized that much of the Nb is found as large b–Nb secondary phase precip-
itates (SPPs) present in the material.
FIG. 4 APT analysis of the metal–oxide interface region of corroded pre-transition

ZIRLO. Shown are Nb distributions in the metal and in the oxide. These are
detected as Nb (red dots) and NbO2 (blue dots), respectively. The size of the
box is 100 by 100 by 450 nm3. In the figure, the oxide–metal interface is
highlighted by an iso-surface corresponding to 50 at. % oxygen.
In addition, the oxygen concentration profiles at the metal–oxide interface are

addressed. This region acts as anode providing electrons to the cathode process
where protons are reduced. Also, any HPU must occur by transport through this
interface. Stability of dissolved oxygen in Zr as a function of oxygen concentration
is computed below. Furthermore, for each oxygen concentration, the energetics for
hydrogen uptake are also provided.
Variable oxygen concentrations across the metal–oxide interface are observed
by APT. Typical concentration profiles obtained from analyses normal to the mac-
roscopic interface are presented in Fig. 5. Clearly, the oxygen ingress into the metal
may vary greatly depending on which region is sampled. The oxygen concentration
in the innermost part of the oxide is typically approximately 55–60 at. %, which is
slightly sub-stoichiometric as compared to the pure ZrO2 phase. In some cases,
oxygen levels rapidly drop down to a base level of a few %, whereas in other cases a
diffusion profile in excess of 500 nm is present. The investigated materials are all in
the macroscopic pre-transition domain of the corrosion process, in accordance
with the characteristic cyclic oxide growth rate of Zr alloys [4,5]. The large varia-
tions are in line with a diffuse pre-transition oxygen concentration profile reported
in Ref. [4].
EXPERIMENTAL DETAILS
APT is carried out using a local electrode atom probe of model Imago LEAP 3000X
HR. The instrument is equipped with a green laser (wavelength 532 nm), which
allows for laser pulsing at a frequency of 200 kHz. The pressure in the analysis
FIG. 5 Oxygen concentration profiles in corroded pre-transition Zircaloy-2. Profiles are

sampled normal to the macroscopic metal–oxide interface.
chamber is maintained at approximately 1.3 1011 Torr during the experiments.

The software IVAS 3.4.3 is used for evaluation of the experimental data.
The analyzed material was a commercial Zircaloy-2 containing 1.32 % Sn,
0.17 % Fe, 0.10 % Cr, and 0.05 % Ni (wt. %), and for the Nb-analysis in the previous
section, a stress relieved ZIRLO containing approximately 1 % Sn, 0.1 % Fe, and
1 % Nb is examined. The Zircaloy-2 material was oxidized in a static steam auto-
clave at 400 C and 10.3 MPa for three days, leading to an oxide thickness of
approximately 1 lm. The ZIRLO sample were subjected to aqueous corrosion in an
autoclave at 360 C and 18 MPa for 100 days, yielding an oxide thickness of approxi-
mately 1.2 lm. The materials are subjected to autoclave testing in steam, which has
led to the formation of an oxide of approximate thickness 1.2 lm on the surface.
Atom probe tips were prepared using a focused ion beam (FIB) workstation, which
is equipped with a scanning electron microscope (SEM). This technique allows for
site-specific sample preparation, where the apex of the atom probe needle can be
located at metal–oxide interface of the sample. A detailed description of this sample
preparation technique is presented elsewhere [21].
Influence of Alloying Elements

on the Hydrogen Pickup Fraction
The present section articulates the possible understanding that HE occurs by (i) the
interception of two electrons by a proton at a metal site thus forming a local
hydride, and (ii) a second proton recombining with the hydride to form H2 (see
Figs. 1(e) and 1(f)). Both protons originate from corresponding hydroxide ions. The
metal sites involve both Zr and various TMs present in a hydroxylated grain
FIG. 6 (a) Ni(II) in grain boundary, Ni-H-Zr hydride encircled, and (b) electron density
for analogous Fe–H–Zr hydride. Red: oxygen; light blue: zirconium; dark blue:
TM; white: hydrogen; green: electron density.
boundary (GB). The bonding in the TM associated hydrides can to some extent
be appreciated in Fig. 6. Thus, Fig. 6(a) displays the most stable structure of the
Ni(II)–H–Zr three centre hydride, while the electron density “halo” centered on H
in Fig. 6(b) emphasizes the hydride character. Here it is noted that the lower
oxidation states of the TMs translates into lower TM–O coordination numbers as
compared to Zr(IV), which renders substitutional dissolution in ZrO2 bulk unfavor-
able. Thus, the TMs accumulate in the hydroxylated grain boundaries where the
electroneutrality constraint is satisfied by the variable proton concentration. This is
discussed further in the next subsection, where the electronic characteristics of these
intermediate hydrides are addressed. In the hydride formation subsection, the com-
puted energetics for the hydride-proton recombination reactions in the presence of
TM ¼ Ti, V, Cr, Mn, Fe, Co, Ni, Cu, and Zn, is displayed. The energetics is decided
by the cost to form the hydride, paid by the reducing potential of Zr, and realized
in the release of H2 upon hydride–proton recombination [7]. Connection to the
experimental HPUF observations [1,2] is suggested.
TM ASSOCIATED HYDRIDE CHARACTERISTICS IN HYDROXYLATED

ZrO2 GRAIN BOUNDARY
Electronic structure characteristics of two hydrides in a generic ZrO2 grain bound-
ary (Fig. 6) are extracted from the density of states (DOS) as displayed in Fig. 7, one
corresponding to Fe(II) (Figs. 7(a) and 7(b)) and the second to Ni(II) (Figs. 7(c) and
7(d)) where the latter in particular is known to display a large hydrogen pickup frac-
tion. Thus, the total electronic DOS for each system is projected on the atomic
eigenfunctions. Atom resolved partial density of states is obtained by summing up
these contributions for each atom. It emerges that states residing on O(-II) domi-
nate the valence band (mainly of 2p character), while states predominantly residing
on Zr(IV) ions (mainly 4d states) dominate the conduction band.
In addition, we find that the 3d transition metals (e.g., Fe and Ni) offer states in
the valence band as well as at the Fermi energy EF, and in the Zr(IV) conduction
FIG. 7 (a) Top left: Atom resolved partial density of states (PDOS) for Fe–H–Zr hydride
decorated ZrO2 grain boundaries (see Fig. 6); (b) top right: atom resolved PDOS
focusing on the Fe–H–Zr hydride; (c) bottom left: atom resolved PDOS for
Ni–H–Zr hydride decorated ZrO2 grain boundaries (see Fig. 6); (d) bottom right:
atom resolved PDOS focusing on the Ni–H–Zr hydride. In all cases, the EF is
identified by the vertical dashed line.
band. In particular, both Fe(II) and Ni(II) introduce states in the band gap of ZrO2.
In the case of Fe(II), the high density of 3d states at EF reflects the high degree
of degeneracy among the filled t2g orbitals corresponding to the low spin 3d6
electronic structure. In contrast, in the case of Ni(II), a more complex electronic
structure is found, corresponding to six electrons in t2g and two electrons in eg*.
The resulting 3d8 electronic structure has two coupled unpaired electrons, display-
ing a triplet state. The larger 3d electron occupation number in case of Ni(II) as
compared to Fe(II), can be appreciated by their respective contributions to the
Zr(IV) conduction band. Thus, significantly less residual 3d partial density of states
in the conduction band is found for Ni(II) than for Fe(II). In the following subsec-
tion, the reactivities of the obtained hydrides towards hydride–proton recombina-
tion is found to correlate with the corresponding hydrogen pickup fractions. Thus,
besides electronic characterization, it becomes relevant to provide vibration spec-
troscopy fingerprints of each of the species—see Table 1 where the highest hydride
associated frequencies are listed. Repeatedly taking the computed hydrides to be
TABLE 1 The highest frequencies associated with the TM(II) hydrides for Ni and Fe for two stable
three-center configurations, type 1 (straight) and type 2 (bent). Experimental frequencies
are provided for comparison. The Ca(II) H–Zr utilizes Ca(II) as a dummy in order to provide
grain boundary Zr–H vibrational frequencies analogous to those for the TM(II) associated
hydrides.
Hydride in Grain Boundary Frequency (cm1) Experiment Frequency (cm1)
Ni–H–Zr (1) 1887 Mg2NiH4 [22] 1645

Ni–H–Zr (2) 1529
Fe–H–Zr (1) 1967 Mg2FeH6 [22] 1873
Fe–H–Zr (2) 2099
Ca(II) H–Zr 1481 ZrO2 on Zr–2.5Nb [8] 1400–1600
mainly local in characters (Fig. 6, again) allows for comparing with experimental
vibrational spectra for hydrides in related compounds [22], which are also provided
in Table 1.
That fact that typical frequencies for OH are found in the range
2500–3700 cm1, i.e., qualitatively different from 1400–2000 cm1 for the grain
boundary hydrides, will simplify the experimental analysis regarding hydrogen spe-
ciation. Indeed, said grain boundary hydrides may already have been observed by
Ramasubramanian et al. [8], who assigned two transient vibrational bands at
1400–1600 cm1 to carbonates. A possible alternative interpretation is that Liþ
uptake in the grain boundaries renders residual transient hydride ions stability by
blocking the hydride-proton recombination reaction. The disappearance of these
bands upon treatment with weak nitric acid is explained here by Liþ–Hþ ion
exchange followed by hydride proton recombination to form H2. The above charac-
teristics are discussed in more detail elsewhere [7].
HYDRIDE FORMATION AND H2 EVOLUTION BY HYDRIDE–PROTON

RECOMBINATION
The mechanism whereby hydrogen is incorporated in the zircaloy cladding is
still to a significant extent unknown. In order to address this enigma, a working
hypothesis is formulated (see Fig. 1). Internal consistency of the proposed under-
standing is demonstrated by means of DFT on model systems. Here, a route to HE
in a grain boundary of the ZrO2 “barrier oxide” is proposed, and its dependence on
alloying elements evaluated. Having proposed a transient route for proton ingress
by hydrolysis whereby H2O attack ZrIV–O–ZrIV bridges forming sub-nanometer
sized hydroxylated interfaces in the oxide to form ZrIV(OH–OH)ZrIV passages, the
fates of these protons are either to result in H2 evolution or HPU. It is assumed
here that hydride intermediates are formed irrespective of reactions, whether HE or
HPU, i.e., according to Eq 2

OH þ ZrðIVÞ O2 TMðXÞ þ 2e ! 2O2
(2)
þ ½ZrðIVÞ H TMðXÞþ X ¼ II; III
Consequently, the HE must result from H – Hþ recombination according to Eq 3

(3) ½ZrðIVÞ H TMðXÞþ þOH ! ZrðIVÞ O2 TMðXÞ þ H2
where the fate of H2 is to diffuse out of the oxide scale. Both TM(II) and TM(III)
were found to be relevant for HE [7]; however, while Eq 2 occurs spontaneously in
case of TM(II), the formation energies of the TM(III) hydride intermediates depend
sensitively on the choice of TM(III) as displayed in Fig. 8(b). It is suggested that the
efficiency of the HE reaction benefits from the TM(III) channel in that, when
allowed, it will be preferred as it can be sustained deeper into the oxide where the
FIG. 8 (a) Top left: the energetics for H2 formation from TM hydride for TM(II) (squares)
and TM(III) (dots), respectively. (b) Top right: the energetics for formation of
TM(III) hydrides, telling of the feasibility of this channel for H2 evolution.
Negative numbers are favorable, while positive numbers disfavor this channel.
(c) Bottom left: weighted average (dashed) between TM(II) (blue) and TM(III)
(red) channels at T ¼ 700 K. Horizontal dashed line corresponds to HE from
Zr(IV) hydride at GB with Ca(II) as spectator (see text). (d) Bottom right:
comparison of theoretical data at T ¼ 700 K and experimental data; o from Ref.
[1] and * from Ref. [2]. Lower horizontal dashed line is same as in (c), while the
upper horizontal dashed line is HPUF in pure ZrO2 from Ref. [2].
higher oxidation state is expected to be predominant. However, Fig. 8(b) implies

that the TM(III) route is open only for Ti, V, Cr, and possibly also for Mn, and
switched off for Fe, Co, and Ni. In addition, TM(III) is unrealistic for Cu and Zn.
This understanding is further realized by plotting the weighted average between the
DEfTMðIIÞg and DEfTMðIIIÞg in Fig. 8(a) according to
DEfTMg ¼ W DEfTMðIIIÞg þ ð1 W Þ DEfTMðIIÞg W ¼ eDDE=k1B T þ1
where DDE in W refers to the differences in the energetics upon H2 evolution for
the two oxidation states (see Fig. 8(b)). The temperature is taken to be 700 K.
The resulting DEfTMg is displayed in Fig. 8(c). An estimate of the hydride-
proton recombination energy in Zr “sponge” is arrived at by employing Ca(II) as
spectator TM(II) ion (dashed horizontal line at 0.4 eV in Fig. 8(c)). The choice of
Ca(II) as dummy is because it displays the appropriate oxidation number while not
providing chemical bonding to the hydride intermediate. Taking the exothermicity
for HE to represent a geometric tolerance for the HE channel it becomes a measure
of the energy/geometry window, which closes upon diminishing exothermicity for
HE. This interpretation is articulated in Fig. 8(d), where experimental [1,2] and
theoretical data (here and Ref. [7]) for the dependence of HPUF on choice of TM
in binary Zr alloys are superimposed. Striking consistency between experiment and
model emerges.
ZrO2 Dissolution in Zr(s) and Energetic

for H Incorporation
In the previous section, renewed support (by means of first principle calculations in
conjunction with a quite general mechanistic understanding) was provided to old
observations [1,2] that report sensitive dependence of HPUF on choice of alloying
transition metal. Necessary requirements for achieving coherence between experi-
ment and theory comprise availability of protons—in the form of hydroxide ions at
oxide grain boundaries—onto which the excess electrons resulting from the oxida-
tion of Zr by O2 may be deposited. In spite of the demonstrated sensitivity to TM,
it is noted that the centre of gravity of the TM effect is represented by pure Zr(IV),
see, e.g., Fig. 8(d). Given that ZrO2 is a large band gap insulator, this suggests that
the corresponding HE process occurs in the vicinity of the Zr/ZrO2 interface, which
is continuously converted into a Zr(Ox)/ZrO2 interface (Fig. 5). Consequently, the
purpose of the present section is to offer insight into relevant electronic and chemi-
cal properties of this interface, known to exhibit an oxygen inwards concentration
profile X(O), which decays from 66 to 0 at. % (see Fig. 5). In particular, the drives
for ZrO2 dissolution and simultaneous hydrogen absorption are addressed. Hence,
the energetics for Zr(Ox) suboxide formation by dissolution of ZrO2 in Zr(s) are
presented in Table 2 (see Ref. [23]). Negative numbers imply that the dissolution of
ZrO2 to form the corresponding suboxide concentration is exothermic. It becomes
TABLE 2 Energetics for Zr(Ox) formation from ZrO2 and Zr for different concentration of oxygen.
Suboxide Reaction DE (eV)
Zr8O 1 15 0.27
ZrO2 þ Zr8 ! Zr8 O
2 16
Zr8O2 1 7 0.25
ZrO2 þ Zr8 ! Zr4 O
2 16
Zr8O3 1 13 0.12
ZrO2 þ Zr8 ! Zr8 O
2 48 3
Zr8O4 1 3 0.02
ZrO2 þ Zr8 ! Zr2 O
2 16
Zr8O5 1 11 0.06
ZrO2 þ Zr8 ! Zr8 O
2 80 5
Zr8O6 1 5 0.13
ZrO2 þ Zr8 ! Zr43 O
2 48
Zr8O7 1 9 0.21
ZrO2 þ Zr8 ! Zr8 O
2 112 7
Zr8O8 1 1 0.19
ZrO2 þ Zr8 ! ZrO
2 16
interesting to monitor the evolution of the electronic density of states as the con-
centration of oxygen increases until the fully gapped insulator state of ZrO2 is
reached. This is done in Ref. [23] where additional detailed properties of the subox-
ide are included. Here it suffices to state that the band gap in ZrO2 develops by the
oxygen associated removal of states in the conduction band at the Fermi energy,
and that the reduced mass gain as function of time, owing to oxidation by water, is
what causes the gradual transformation of Zr(Ox) to ZrO2 at the interface for a
steady state concentration of oxygen close to 30 at. %, i.e. corresponding to the satu-
ration limit.
In Table 3, the hydrogenation of Zr(Ox) is addressed as well as that of ZrO2 for
Ref. [23]. Similar to the oxygen dissolution, the introduction of H in the suboxide is
associated with the titration of Zr metal states at the Fermi energy. The fact that the
TABLE 3 Hydride formation energies and resulting hydride associated vibrational frequencies. The
energetics employs H2 in grain boundary as reference.
Hydride Reaction DE (eV) Frequency (cm1)
Zr8H2 1 1.19 1235

Zr4 þ H2 ! HZr4
2
Zr8OH2 1 1 1 1.22 1395
Zr8 Oþ H2 ! H2 Zr8 O
2 2 2
Zr8O4H2 1 1.23 1329
Zr4 O2 þ H2 ! HZr4 O2
2
Zr8O8H2 1 1.16 1268
Zr4 O4 þ H2 ! HZr4 O4
2
Zr4O8H2 Zr4 O8 þH2 ! HZr4 O7 ðOHÞ 0.98 1281
energetics associated with the dissolution of H in Zr(Ox) is almost constant as long

as there are metallic Zr states at the Fermi level is a crucial finding. ZrO2 hydrogen-
ation by H2 is included in Table 3 for reference, as it produces both hydride and
hydroxide moieties where the vibrational frequency of the hydride is sought. As a
byproduct, this reaction serves as a consistency check for the dependence of the
hydrogen pickup fraction on transition metal addition presented in the Hydride
Formation subsection above. Thus, the negative of the obtained endothermicity
would reflect a recombination energy of 0.98 eV for hydride with proton, which in
turn would imply 0 % pickup fraction (see Fig. 8(d)). This has as consequence that
HE from hydrogenated ZrO2 bulk is effectively ruled out as a decisive reaction
channel for HPUF because it contradicts the 50 % HPUF reported for Zr from
experiment. Theoretical vibrational frequencies were calculated for hydrides in
Zr(s), in some representative Zr(Ox) suboxides as well as in ZrO2 bulk. The fact
that the vibrational spectra of H in Zr(Ox) deviate significantly from those at inter-
grain sites (see Tables 1 and 3) is promising in that IR/Raman spectroscopy may be
utilized to distinguish between these different species. The fact that the frequencies
of the hydrides are often well separated from those of the oxide as well as any
hydroxides in the grain boundaries suggests that IR/Raman spectroscopies could
indeed prove useful for tracking the passage of H into the metal.
Characteristics of Substitutional Nb
Doping in ZrO2
Mitigation of hydrogen pickup is often achieved by doping Zr with Nb [4]. Yet,
experiment reports little or no accumulation of Nb in grain boundaries of ZrO2 (see
Fig. 4). This supports the understanding that the impact of Nb doping on the HPUF
is different from those of the TMs addressed in the third section of the paper, and
this may be owing to Nb substituting for Zr in ZrO2. In the present section, the
resulting electronic structure owing to substitutional Nb doping in ZrO2 is
described (see Fig. 9), where the Nb doping concentration is 2.3 at. %; this is real-
istic. From the DOS alone, it appears as though the Nb doping indeed produces a
metallic conductor, i.e., displaying significant DOS at the Fermi energy EF. How-
ever, upon plotting the orbitals corresponding to the states at EF, it emerges that
these states are local in character and strongly dominated by the Nb(IV) ions (see
Figs. 9(c) and 9(d)). This understanding is supported by the PDOS for Nb and Zr in
Nb doped ZrO2 (see Fig. 9(b)), which clearly show that the states in the vicinity of
the EF are almost exclusively of Nb origin, and that there is an effective band gap of
1 eV between the local Nb(IV) states at EF, and the Zr(IV) conduction band.
The small residual Zr related DOS at EF is understood to result from local
spill-over to Zr nearest neighbors, some of which may originate from a flaw in
present-day density functional theory comprising the self-interaction error. Thus,
the Zr(IV) ions which are proximal to Nb(IV) become contaminated by the Nb
valence state. Significant reduction in Zr(IV) PDOS at EF is observed already for
FIG. 9 (a) Top left: atom resolved PDOS for Nb(IV) doped ZrO2. (b) Top right: atom
resolved PDOS for Nb and Zr in Nb(IV) doped ZrO2, where the PDOS(Zrtot) is
further subdivided into contributions from Zr1 nearest neighbors to Nb(IV) and
the remaining Zr(IV) ions, comprising Zr2. (c),(d) Bottom: two representative
electron states (green) at EF for Nb(IV) doped ZrO2. Clearly, nearly all weights
are located on the Nb sites.
next-nearest neighbors (see Fig. 9(b)). The Nb(IV) state sits at 1 eV below the con-
duction band (see Fig. 9(b)), which is consistent with the observation of Ramasubra-
manian et al. [24]. It can be understood as an impurity state that is tailored to add a
narrow band semiconductor property to an a priori large band gap insulator. Thus
Nb doping offers a significant reduction of the original 4–5 eV band gap in pure
ZrO2. The locality of the impurity states and possible spill-over implies that
electron conductivity in Nb doped ZrO2 would depend strongly on inter-Nb dis-
tance, i.e., the Nb concentration. This electron transport channel is understood to
work as a shunt, offering an electron transport channel through the oxide bulk to
the water–oxide interface, which becomes competitive at later stages in the Zr
oxidation cycles. Ideally, from here on the shunting current owing to Nb doping
would maintain a constant oxidation rate and a constant HPUF. Thus, the lower
bound to the oxidation rate owing to the Nb doping would become effective when
the instantaneous hydrogen pickup fraction peaks. In reality, this limiting behavior
is interrupted owing to the cracking of the barrier oxide as manifested by a disconti-
nuity in the mass gain curve. The initial acceleration in mass gain is subsequently
followed by the retardation phase until the barrier oxide again cracks. Below, this
effect is proposed to reflect a supersaturated Zr(Ox) repeatedly accessing the misci-
bility of the Zr–O phase diagram resulting in the cracking of the barrier oxide, pos-
sibly with hydrides rendering it brittle. The cracks are hydrated, and the oxidation
resumes anew by zirconium oxidation by water, employing O2 oxidation as anode,
and Hþ reduction as cathode.
Discussion and Concluding Remarks

The semi-quantitative agreement between experimental and theoretical HPUFs, as
inferred from the theoretical energetics for the HE reaction (Fig. 8(d)), in conjunc-
tion with the suboxide hydrogenation energies, which come out similar to that of
bulk zirconium (Table 3) offer a qualitative interpretation of experiment. This
interpretation implies an inner HE channel taking place at the interface between
the oxygen saturated Zr metal and the temporarily formed ZrO2 barrier oxide.
Moreover, an understanding of the HPUF in the binary alloys is arrived at reflecting
detailed quantitative TM control of the qualitative behavior for pure zirconia on zir-
conium. Taken literally, this implies that the HPUF reflects detailed electronic
properties of transition metal ions directly associated with the proton reduction,
i.e., attenuating the effective cathode process taking place in oxide grain boundaries
and/or possibly in transient sub-nanometes sized hydroxylated interfaces in the
barrier oxide.
The energetics for HE and HPU in Zr(Ox) comes out similarly (see Table 3 and
Fig. 8(d)). This implies that any competition or indeed cooperation between H and
O in forming HyZr(Ox) compounds by utilizing metal states at EF in Zr(Ox) cannot
be excluded. The latter would offer a candidate origin for apparent oxygen solubility
in Zr greater than the thermodynamic limit of the pure Zr–O system, rendering the
resulting supercritical material unstable. This possible understanding would render
speciation of different hydrides an interpretational tool, and as such, it invites com-
plementary experiment by means of vibrational spectroscopy. Having said this, it
may well be that the sought grain boundary hydrides have already been observed,
i.e., causing the bands observed by FTIR at 1400–1600 cm1 in Liþ decorated ZrO2
grain boundaries (see Ref. [8] and Table 1 again).
While the above lends possible crucial significance to TM associated properties
at inter-grain compartments in the vicinity of the metal–oxide interface, it was
argued that Nb doping has a complementary role in the design of HPUF mitigating
zirconium alloys. Thus the tentative understanding that emerges suggests the exis-
tence of two hydrogen evolution channels, one inner and one outer. Irrespective of
which, the two factors determining the pre-transition HPU comprise
i. The drive for oxidation
ii. The ability to dispose electrons resulting from the Zr oxidation process
Compare
(4) aZr þ O2 ! Zra ðOÞ þ 2e
In case of the inner hydrogen evolution channel (IHEC), the ability of hydroxide
and transition metal decorated grain boundaries to accommodate these residual
electrons is decisive, (see Fig. 1 and Eqs 2 and 3). However, while zirconium oxida-
tion by IHEC requires means to dispose said electrons in zirconia grain boundaries,
the way this is accomplished during an oxidation cycle is different at different
stages:
• Early during the oxidation cycle, protons intercept the excess electrons to
form hydrides, which subsequently undergo hydrogen evolution HE by
hydride–proton recombination as described in Eqs 1 and 2 and in Figs. 1(e)
and 1(f). At this stage, the oxygen incorporation is rapid, while HPUF may be
small.
• Later, the oxidation rate is slowing down to the extent that it becomes sub-
parabolic. This is partly owing to the depletion of protons in the grain bounda-
ries, and the growing electronic resistance in the fresh barrier oxide. Were the
electron conductivity not limiting, parabolic mass gains would always be
observed since in that case the oxygen ions diffusion be rate limiting. While
hydrides are still formed, as reflected in the oxide growth, the IHEC is closing
due to lack of protons. This renders the instantaneous HPUF large, see
Figs. 1(h)-1(i).
• The oxidation cycle is completed when the hydride buffer is saturated, and the
HyZr(Ox) has become supercritical with respect to O and H dissolution.
Would the electron conductivity in the grain boundary be improved suffi-
ciently, then additional electron buffering capacity would be achieved owing to
access to a larger reservoir of protons in the vicinity of the outer non-barrier oxide.
A limiting behavior would comprise an outer hydrogen evolution channel (OHEC),
complementary to IHEC. This channel offers HE at the interface between the bar-
rier oxide and the non-protecting oxide where we get
(5) 2e þ 2H2 O ! H2 þ 2OH
Indeed, such an overall understanding involving parallel hydrogen evolution

processes at the metal–oxide interface and at the interface between protective and
non-protective oxide was advocated in Ref. [24]. In order for this to happen, elec-
tron conductivity through the barrier oxide (e.g., as accomplished by substitu-
tional Nb doping in the ZrO2 barrier oxide), where the concentration of Nb is
taken to control the effective electron mobility, must be present. The resulting re-
sidual band gap difference between Nb(IV) and Zr(Ox) conduction bands ener-
gies of 1 eV is understood to translate into an activation energy for hopping
between Nb(IV) sites.
Besides mitigating HPU, any successful fuel cladding must maintain a low net
oxidation rate. Thus, any shunting between the inner and the outer hydrogen
evolution channels, IHEC and OHEC, respectively, must maintain low net electron
current as it is proportional to the oxidation rate. This corrosion resistance condi-
tion implies that while HPU may be minimized it cannot be completely avoided.
Early in each oxidation cycle, HPUF is low. Metal oxidation is maintained by

hydride formation and hydride–proton recombination. The ratio between initial
proton concentration and accumulating residual hydrides in the grain boundaries
is decided by the efficiency of the IHEC, the TM dependence of which is shown in
Fig. 8(d). Resulting depletion of protons, as well as increased electronic resistance,
translates into reduced oxidation rate. From this point, the electrons released in the
oxidation process (Eq 4) predominantly produce hydrides. This is when the instan-
taneous HPUF peaks [4,6]. At the stage when the IHEC becomes inefficient or the
hydride sink approaches saturation, the OHEC may offer a complementary route
for maintaining the oxidation process; thus delaying the transition. Tailored shunt-
ing would be accomplished by balancing the OHEC against IHEC in such a way
that the OHEC would decide the lower bound for the net pre-transition oxidation
rate. This lower bound would in turn be decided by the concentration of Nb dopant
in ZrO2. However, different Nb concentrations match different transition metal
additions. Repeatedly, this tailoring depends crucially on choice of TM additions.
This understanding may be illustrated in the following somewhat oversimplified
exercise:
1. Turn on the IHEC while keeping the OHEC closed. High Fe concentration
implies high initial oxidation rate due to the efficiency of the IHEC. Yet, in
spite of low HPUF, the net HPU may be large depending on the amount of
oxide formed.
2. Respond by reducing the Fe concentration implies slower oxidation rate owing
to the suppression of the IHEC. This method, though good, comes at the
expense of larger HPUF. This situation reflects more of the performance of the
native Zr, i.e., increased early accumulation of hydrides.
3. Keep the reduced Fe concentration, but turn on the OHEC, with Nb doping
for example. The original oxidation rate corresponding to the high Fe concen-
tration is recovered, however the HPU is much reduced.
4. Finally, add small amounts of Ni, by forming an Fe–Ni–Nb alloy. By this, a
handle on the rate of oxidation may be offered owing to the inability of nickel
to sustain IHEC. Thus, ideally, the low oxidation rate of Zircaloy-2 in conjunc-
tion with minimal HPU would be achieved—see the new developed corrosion
resistant alloy (NDA), in Ref. [4].
Finally, we shall briefly discuss the generic cyclic oxidation of Zr alloys. We
suggest that this spatio-temporal behavior is driven by the oxidation of Zr by
water. It occurs by periodic supersaturation with respect to the phase diagram of
Zr–O. We speculated that the sub-oxide becomes supercritical owing to the
HPU and that the resulting phase comprises HyZr(Ox). Continuous dissolution of
oxygen and hydrogen into the suboxide translates into barrier oxide growth
on-top of supercritical HyZr(Ox) sub-oxide. Discontinuous recovery of the sharp
metal–oxide interface [4] occurs when the barrier oxide cracks as a means to sus-
tain continued Zr oxidation. Further detailed fine tuning of the Zr alloy, besides
the shunting between IHEC and OHEC, would aim for control of these repeated
transitions.
In conclusion, a mechanistic framework from first principles has been articu-

lated aiming at providing new opportunities for the improvement of material per-
formance in nuclear fuel claddings. Future work will include:
1. Consolidation of the present tentative findings concerning the HPUF and the
Nb induced shunting between the IHEC and OHEC,
2. Development of a simulation tool which utilizes the above understanding to
describe the HPU and HPUF as function of mass gain rate.
A potentially important additional parameter would comprise control of the
robustness of the metal–oxide interface towards repeated cracking. Future research
into new alloys that delay said transition or, indeed, attempt to effectively “close”
the Zr–O miscibility gap, while maintaining low oxidation rates, are encouraged.
ACKNOWLEDGMENTS
The Swedish Research Council, Westinghouse Electric Sweden, Sandvik Materials
Technology, Vattenfall, and the EPRI are gratefully acknowledged for financial
support.
References
[1] Cox, B., “Oxidation of Zirconium and Its Alloys,” Advances in Corrosion Science and Technol-
ogy, Vol. 5, M. G. Fontana and R. W. Staehle, Eds., Plenum, New York, 1976, pp. 173–391.
[2] Parfenov, B. G., Gerasimov, V. V., and Venediktova, G. I., “Corrosion of Zirconium and
Zirconium Alloys (Korroziya Tsirkoniya i Ego Splavov),” Israel Program for Scientific
Translations, Jerusalem, 1969, pp. 118–120.
[3] Abriata, J. P., Garcés, J., and Versaci, R. “The O–Zr (Oxygen–Zirconium) System,” Bull.
Alloy Phase Diag., Vol. 7, No. 2, 1986, pp. 116–124.
[4] Harada, M. and Wakamatsu, R., “The Effect of Hydrogen on the Transition Behavior
of the Corrosion Rate of Zirconium Alloys,” Proceedings of the 15th International
Symposium, Zirconium in the Nuclear Industry, ASTM STP 1505, ASTM International,
[5] Hillner, E., “Corrosion of Zirconium-Base Alloys—An Overview,” Zirconium in the Nuclear
Industry, ASTP STP 633, A. L. Lowe, Jr, and G. W. Parry, Eds., ASTM International, Phila-
delphia, PA, 1977, pp. 211–235.
Activation Analysis, a Non-Destructive Technique for Hydrogen Level Assessment in
Zirconium Alloys,” J. Nucl. Mater., Vol. 425, 2012, pp. 211–217.
[7] Lindgren, M. and Panas, I., “Impacts of Additives on Hydride Formation and H2 Recombi-
nation During Zirconium Oxidation by Water,” RSC Adv, Vol. 3, No. 44, 2013, pp.
21613–21619.
[8] Ramasubramanian, N. and Balakrishnan, P. V., “Aqueous Chemistry of Lithium Hyrdoxide

and Boric Acid and Corrosion of Zircaloy-4 and Zr-2.5Nb Alloys,” Zirconium in the Nu-
clear Industry: Tenth International Symposium, ASTM STP 1245, ASTM International,
[9] Ramasubramanian, N., Perovic, V., and Leger, M., “Hydrogen Transport in the Oxide and
Hydrogen Pickup by the Metal During Out- and In-Reactor Corrosion of Zr-2.5Nb Pres-
sure Tube Material,” Zirconium in the Nuclear Industry: Twelfth International Symposium,
[10] Hodeau, J. L., Marezio, M., Santoro, A., and Roth, R. S., “Neutron Profile Refinement of the
Structures of Li2SnO3 and Li2ZrO3,” J. Solid State Chem., Vol. 45, No. 2, 1982, pp. 170–179.
[11] Busch, M., Ahlberg, E., and Panas, I., “Validation of Binuclear Descriptor for Mixed Transi-
tion Metal Oxide supported Electrocatalytic Water Oxidation,” Catal. Today, Vol. 202,
2013, pp. 114–119.
[12] Hohenberg, P. and Kohn, W., “Inhomogeneous Electron Gas,” Phys. Rev., Vol. 136, No. 3B,
1964, pp. B864–B871.
[13] Kohn, W. and Sham, L. J., “Self-Consistent Equations Including Exchange and Correlation
Effects,” Phys. Rev., Vol. 140, No. 4A, 1965, pp. A1133–A1138.
[14] Perdew, J. P., Burke, K., and Ernzerhof, M., “Generalized Gradient Approximation Made
Simple,” Phys. Rev. Lett., Vol. 77, 1996, pp. 3865–3868.
[15] Materials Studio 6.0. (2011), Accelrys Inc., San Diego, CA.
[16] Busch, M., Ahlberg, E., and Panas, I., “Electrocatalytic Oxygen Evolution From Water on
a Mn(III–V) Dimer Model Catalyst—A DFT Perspective,” Phys. Chem. Chem. Phys., Vol.
13, 2011, pp. 15069–15076.
[17] Clark, S. J., Segall, M. D., Pickard, C. J., Hasnip, P. J., Probert, M. J., Refson, K., and Payne, M. C.,
“First Principles Methods Using CASTEP,” Z. Kristall., Vol. 220, Nos. 5–6, 2005, pp. 567–570.
[18] Vanderbilt, D., “Soft Self-Consistent Pseudopotentials in a Generalized Eigenvalue

Formalism,” Phys. Rev. B, Vol. 41, 1990, pp. 7892–7895.
[19] Sundell, G., Thuvander, M., and Andrén, H.-O., “Enrichment of Fe and Ni at Metal and
Oxide Grain Boundaries in Corroded Zircaloy-2,” Corros. Sci., Vol. 65, 2012, pp. 10–12.
[20] Berry, E. B., “Hydrogen Pickup During Aqueous Corrosion of Zirconium Alloys,”
Corrosion, Vol. 17, 1961, pp. 109–117.
[21] Larson, D. J., Ford, D. T., Petford-Long, A. K., Liew, H., Blamire, M. G., Cerezo, A., and
Smith, G. D. W., “Field-Ion Specimen Preparation Using Focused Ion-Beam Milling,”
Ultramicroscopy, Vol. 79, 1999, pp. 287–293.
[22] Parker, S. F., Williams, K. P. J., Smith, T., Bortz, M., Bertheville, B., and Yvon, K.,
“Vibrational Spectroscopy of Tetrahedral Ternary Metal Hydrides: Mg2NiH4, Rb3ZnH5 and
Their Deuterides,” PCCP, Vol. 4, 2002, pp. 1732–1737.
[23] Lindgren, M. and Panas, I., “On the Fate of Hydrogen During Zirconium Oxidation by
Water: Effect of Oxygen Dissolution in a-Zr,” RSC Adv., Vol. 4, No. 22, 2014, pp.
11050–11058.
[24] Ramasubramanian, N., Billot, P., and Yagnik, S., “Hydrogen Evolution and Pickup During
the Corrosion of Zirconium Alloys: A Critical Evaluation of the Solid State and Porous
Oxide Chemistry,” Zirconium in the Nuclear Industry: Thirteenth International
Symposium, ASTM STP 1423, ASTM International, West Conshohocken, PA, 2002, pp.
222–244.
DISCUSSION
Question from Malcolm Griffiths, AECL:—You showed that Ni is unusual in its
hydrogen pickup relative to other elements. Were you also aware that pure Ni in a
nuclear reactor also generates hydrogen by the (n,p) reaction (50 appm in the first
year of reactor operation and accelerating to 500 appm at end-of-life)? Whereas
this is insignificant in a global sense, the local hydrogen generation at Ni-rich pre-
cipitates should be noted.
Authors’ Response:—No, I was not aware of this. Things just keep getting
“curiouser and curiouser”!
Question from Rishi Sharma, IIT Bombay, India:—Can we say that Nb reduces
the porosity of the Zr oxide layer?
Authors’ Response:—Yes, in two ways. Firstly, interception of protons with elec-

trons closer to the interface between barrier oxide and non-barrier oxide renders
the hydroxylation depth shallower. Secondly, avoiding the inner hydrogen evolu-
tion channel mitigates the kind of porosity that results from the H2 release from the
metal-oxide interface.
Question from Philippe Bossis, CEA:—From this approach, which is the best to
reduce hydrogen pick-up: big or small precipitates?
Authors’ Response:—Indeed, the more we can separate the effective anode pro-
cess at the metal-oxide interface from the effective cathodes where protons and
electrons recombine inside the barrier oxide, the less likely is the hydrogen pick-up
process. Metallic precipitates protruding into the oxide would do this job and so,
the bigger the better. However, the price for avoiding hydrogen pick-up in this way
is rapid oxidation followed by oxide cracking etc. and so there is no one simple
cure. Conceptually, we feel that we are getting there but before we dare to predict
effective performances of hypothetical alloys, we need to construct a simulation tool
to help us think.
Question from N. Ramasubramanian, ECCATEC Inc. Canada:—How would

the model explain the difference between un-irradiated and irradiated materials?
In the irradiated material, transition metal additions are dispersed in the alloy ma-
trix and so can easily be segregated to oxide grain boundaries but not so in un-irra-
diated material. However, both irradiated and un-irradiated materials show pick-
ups with wide spreads.
Authors’ Response:—We have put all our efforts in extracting an electro-cata-

lytic understanding of hydrogen pick-up. Formation of new oxide implies electron
release thus fixing the anode at the metal-oxide interface. Protons are delivered in
hydroxylated grain boundaries. Intercepting protons by electrons produces hydride
ions and subsequent hydride-proton recombination results in H2 formation. When
the latter step does not happen, we have hydrogen pick-up. Indeed, before we go
ahead and fine-tune a simulation tool for predicting alloy performances, we need to
learn to what extent experiments in the autoclave are valid for modeling what is
going on in the reactor. This is why we are currently starting up experiments on
irradiated samples.
Sousan Abolhassani,1 Gerhard Bart,2 Johannes Bertsch,2

Mirco Grosse,3 Lars Hallstadius,4 Armin Hermann,2
Goutam Kuri,2 Guido Ledergerber,5 Clément Lemaignan,6
Matthias Martin,7 Stéphane Portier,7 Christian Proff,2
Renato Restani,7 Stéphane Valance,2 Sima Valizadeh,4
and Holger Wiese7
Corrosion and Hydrogen Uptake

in Zirconium Claddings Irradiated
in Light Water Reactors
Reference
Abolhassani, Sousan, Bart, Gerhard, Bertsch, Johannes, Grosse, Mirco, Hallstadius, Lars,
Hermann, Armin, Kuri, Goutam, Ledergerber, Guido, Lemaignan, Clément, Martin, Matthias,
Portier, Stéphane, Proff, Christian, Restani, Renato, Valance, Stéphane, Valizadeh, Sima, and
Wiese, Holger, “Corrosion and Hydrogen Uptake in Zirconium Claddings Irradiated in Light
Water Reactors,” Zirconium in the Nuclear Industry: 17th International Symposium, STP 1543,
ABSTRACT
The objective of this paper is to summarize the results of the latest observations
performed at Paul Scherrer Institut on irradiated fuel claddings, to characterize
their corrosion and hydrogen-uptake behavior. Two categories of studies have been
performed. (1) A series of destructive tests were achieved on the fuel rods
Manuscript received January 7, 2013; accepted for publication September 14, 2013; published online
September 19, 2014.
1
Laboratory for Nuclear Materials, Nuclear Fuels Group, NES, Paul Scherrer Institut, 5232 Villigen PSI,
Switzerland (Corresponding author), e-mail: sousan.abolhassani@psi.ch
2
Laboratory for Nuclear Materials, Nuclear Fuels Group, NES, Paul Scherrer Institut, 5232 Villigen PSI,
Switzerland.
3
Institut für Angewandte Materialen, Karlsruher Institut für Technologie, DE-76021 Karlsruhe, Germany.
4
Westinghouse Electric Sweden AB, SE-72163 Västerås, Sweden.
5
Kernkraftwerk Leibstadt AG, CH-5325 Leibstadt, Switzerland.
6
CEA Grenoble, 17 rue de Martyrs, 38054 Grenoble Cedex 9, France.
7
Hot Laboratory Division, NES, Paul Scherrer Institut, 5232 Villigen PSI, Switzerland.
8
Hyderabad, India.
Copyright V
ABOLHASSANI ET AL., DOI 10.1520/STP154320130007 541
irradiated in a boiling-water reactor (BWR), including hydrogen concentration by

hot-gas extraction. These results provided the hydrogen content of the cladding at
different stages of irradiation, at different elevations along the rod. (2) Another
series of examinations using a correlative microscopy method, i.e., using different
techniques, including transmission electron microscopy (TEM), electron probe
microanalysis (EPMA), and secondary ion mass spectrometry (SIMS), on the same
material and in the same region of the metal–oxide interface have provided useful
data regarding the oxide layer combining the signals from oxides and from
hydrides. Furthermore, the effect of the type of alloying element has been
examined for in-reactor oxidation. These studies are subsequently combined with
the findings from out-of-pile studies, using techniques, such as neutron
radiography, to confirm the in-reactor observations. Results have shown that: (i)
the hydrogen pickup fraction varies at different conditions and could even decrease
as the oxide thickness increases; (ii) the distribution of hydrogen in the cladding is
usually inhomogeneous; (iii) the most determining parameter for hydrogen uptake
seems to be the microstructure of the oxide, and the nature of the alloying element
will influence to a certain extent this parameter; (iv) furthermore, the stress in the
oxide layer can modify the crack distribution in the latter, cracks will in turn shorten
the route for the hydrogen to access the metal. These results will be discussed as a
contribution to the available knowledge about hydrogen uptake and will provide a
global support for the models of the uptake phenomenon.
Keywords
hydrogen content, oxidation, irradiation effects, high burnup, TEM, EPMA, SIMS
Introduction
In the last two decades, a great amount of effort has been put by the fuel suppliers
and utilities to improve the performance of fuel in light water reactors (LWRs).
Such studies usually necessitate long preliminary investigations and both in-pile
and out-of-pile experiments are required to achieve any fuel development. In col-
laboration with nuclear power plants in Switzerland and certain fuel suppliers, Paul
Scherrer Institut (PSI) has contributed to several such investigations to characterize
the fuel and to determine its properties as a function of the fluence and the resi-
dence time in the reactor. The objective of these projects has been to improve the
reliability of the fuel, to increase the residence time and to reduce the production of
nuclear waste. These projects have been described in different publications [1–3]
and, in the case of boiling water reactor (BWR) studies, materials from these proj-
ects have been used in several international research programs worldwide. In the
case of BWR studies, details of irradiation, pool-side inspections, and a large
amount of data from destructive experiments (DT) are already reported in previous
papers [1,2,4]. Nevertheless, a detailed description of each part of these studies still
needs to be presented and discussed; this paper will attempt to report and examine
a part of the available data on the fuel cladding to the scientific community. It is
worth noting that the data is presented in view of understanding the mechanistic
phenomena, in an attempt to capture variability and/or dependencies that have not

been identified in earlier work, and are not necessarily presenting the specification
of the modern claddings used in LWRs.
The paper will be divided into three parts: the first part will report on the clad-
dings irradiated in a BWR, the second part will report on a study in a pressurized
water reactor (PWR), and the last part of the paper will discuss the results in view
of understanding the mechanisms behind hydrogen uptake, making use of exam-
ples both from in-pile data and out-of-pile experiments performed at PSI.
Materials and Methods

The chemical compositions of the claddings used for BWR are provided in Table 1.
Similar data for the PWR alloys are reported in Table 2. The history of each material
including the number of cycles of stay in the reactor, the burnup, and the elevation
of segments studied are provided in Tables 3 and 4. All the claddings were trans-
ported to the PSI hot cells and examined by non-destructive methods (NDT). Sub-
sequently, a number of segments were selected for further examination using
different destructive methods.
HYDROGEN HOT-GAS EXTRACTION

The inert-gas fusion (IGF) method using an instrument supplied by LECO was
employed for the hydrogen hot-gas extraction. PSI has developed a procedure that
provides the possibility to measure the hydrogen content of the metal and that of ox-
ide separately. This procedure has been described in a previous publication [5]. It has
been demonstrated that the low-temperature hydrogen release originates from the
hydrogen present in the oxide, and the high-temperature release originates from the
metal. In certain studies, to distinguish the difference in hydrogen content of the clad-
ding around the circumference of a ring, the ring is cut into four quarters and each
quarter is measured individually. Such measurements can confirm the results
obtained from the metallography and the fact that the distribution of hydrogen
around the circumference of the segment could be inhomogeneous in some alloys.
In this study, the total hydrogen content (i.e., total of low-temperature and
high-temperature release) is provided, as well as the hydrogen content in the metal.
This can illustrate the variations in the evaluation of the hydrogen content because
of the contribution of hydrogen present in the oxide part of the sample (see for
example Fig. 1, elevation ¼ 1700 mm, where the total hydrogen content is much
higher than the metal content). It must be noted that the contribution from the ox-
ide (the release at low temperature) is calculated with respect to the mass of the
sample. It is important to show this contribution with respect to the mass of the
sample, as it can indirectly give an indication of the oxide thickness; in this manner,
the difference that the hydrogen present in the oxide can introduce to the quantifi-
cation, if the total hydrogen is considered, can be estimated. The larger the low-
temperature release, the thicker would be the oxide layer. From adjacent segments
TABLE 1 The chemical composition of standard Zircaloy-2 claddings and the nominal composition of materials for BWR [A ¼ R ti exp(Q/RTi), where i is the ith heat treat-
ment and (Q ¼ 63 000 cal/mol)] [13]. The designation /L indicates the presence of an inner liner in the cladding.
Comp.
(wt. %):
Specimen Sn Fe Cr Ni Fe þ Cr þ Ni O ppm Si ppm log A
ABOLHASSANI ET AL., DOI 10.1520/STP154320130007

Standard 1.2–1.7 0.07–0.2 0.05–0.15 0.03–0.08 0.18–0.38 1000–1400 Maximum
Zircaloy-2 120
(ASTM B353)
LK3/L 1.34 0.18 0.11 0.05 0.34 1320 70 14.2
LK2/L 1.50 0.13 0.10 0.05 0.28 1090 80 16.0
543
544
TABLE 2 The chemical composition of alloys used and the final heat treatment of the materials for PWR.
SPP
Sn Fe Cr Nb Si C O H Mean
Material (wt. %) (wt. %) (wt. %) (wt. %) (ppm) (ppm) (ppm) (ppm) HT Size (m)
Low-tin Zircaloy-4 1.20 0.22 0.107 – – 140 1730 8 504 C SRA 190
Zr 2.5 % Nb – 0.07 – 2.5 60 180 1170 10 500 C na
PRX
HT, heat treatment; SRA, stress relieve annealed; PRX, partially recrystallized condition; na, not available.
TABLE 3 Irradiation data for the BWR Zircaloy-2 with LK3/L cladding grade for adjacent segments where both oxide thickness and hydrogen measurement are determined destructively.
No. of Cycles/ Segment Elevation Segment Elevation

Segment Rod Identity Peak BU mm (for Oxide Maximum Minimum Mean Total mm (for Hydrogen
Elevation and Segment (MWd/kgU)/ Measurement Oxidea Oxidea Oxidea H2 Measurement
(mm) Number Enrichment Sample) thickness (lm) thickness (lm) thickness (lm) (ppm) Sample)
Mid-span AEB070—E4/K 3/44.6/4.46 % 2005 6 4 4.4 6 (0.7) 44 2002

peak burnup AEB069—E4/N 5/59.6/4.46 % 2005 24 4 12 90 2002
AEB068—E4/O 6/66.9/4.46 % 2005 29 4 14 202 2002
AEB067—E4 7/67.4/4.46 % 1998 25 14 19 6 (1.4) 261 2004
AEB071—E4 7/67.6/4.46 % 1995 22 19 21 6 (0.9) 339 2000
AEB072—E4 7/73.3/4.46 % 1998 24 16 19 6 (2.3) 328 2004
AEB072—J9 7/73.0/3.71 % 1998 24 10 19 6 (2.3) 224 1995
AGB108-G6 9 / 78.7 / 4.46 % 2039.5d 50 40 46 6 2.5 595 2045
ABOLHASSANI ET AL., DOI 10.1520/STP154320130007

AGB108-J8 9 / 77.2 / 4.46 % 678 1998
AGB108-J10 9/70.7/3.71 % 695 1998
3045 AEB070—E4/O 3/44.6/4.46 % 3045 6 4 4.4 6 (0.6) 46 3051
661 AEB072—J9 7/73.0/3.71 % 661 25 14 19 6 (0.9) 189 666
3rd Spacer AEB070—E4/F 3/44.6/4.46 % 1770 70 –b 46 6 (17) 70 1767
AEB069—E4/F 5/59.6/4.46 % 1769 180 10c 52 144 1775
AEB068—E4/F 6/66.9/4.46 % 1769 71 9c 32 180 1775
AEB072—E4 7/73.3/4.46 % 1767 136 60 –b 145 1773
LK2/L seriese AEB070—E3 3/34.7/4.46 % 2005 3 8 561 45 2002
AEB068—E3 6/67.3/4.90 % 2005 14 46 26 6 - 222 2002
AEB072—J7 7/73.0/4.07 % 2004 67 186 115 6 2 469 2009
a
The oxide value is obtained from the metallographic measurements at the elevation.
b
Not reported because of uncertainty.
c
The minimum value in the spacer is not representative of the real thickness as it represents the spalling of the oxide.
d
Oxide thickness obtained from EPMA measurements.
e
The last three rows of data points are from LK2/L cladding for comparison.
545
546
TABLE 4 Irradiation data, oxide thickness, and hydrogen content for the PWR material.
Oxide Versus Hydrogen Content Data
No. of Rod Maximum Rod Available TEM TEM Sample

Cycles/ Oxide Data for Sample Oxide Rod Oxide
Burnup Thickness Hydrogen Elevation Thickness Elevation Thickness Hydrogen Pick-up
Material (MWd/kgU) (lm) Content (ppm) (mm) (lm) (MM) (lm) (ppm) Fraction
Zr 2.5 % Nb 3 /41.4 16 70a 869 6.9

Low-tin Zircaloy-4 4 / 51.1 59 52 to 460 1468 25
353 7.5 52.2 6 5 15 %
1561 25 209 6 19 19 %
2525 44 367 6 31 19 %
3120 56 460 6 44 18 %
a
One value available for this rod.
FIG. 1 Hydrogen content of an LK3/L fuel cladding (AEB072-J9) irradiated for seven
cycles in the BWR. The distributions of hydrogen in the metal and in the oxide
are presented separately. The total hydrogen content of the cladding shows the
sum of hydrogen in the metal and in the oxide.
of metallography and hydrogen measurement, the oxide thickness from metallogra-

phy has been used to calculate the hydrogen content in the oxide layer. The concen-
tration of hydrogen in the oxide has been calculated for this family of samples to be
in the range of 930 to 1050 ppm. As this value does not change considerably, the
value provided in the graphs will represent the contribution of oxide to the hydro-
gen evaluation, and will imply a change of oxide thickness for the samples with
larger low-temperature release.
TRANSMISSION ELECTRON MICROSCOPY (TEM)

A JEOL 2010 transmission electron microscope was used for the examination of the
claddings; the instrument was equipped with an Oxford Instruments EDS detector
and INCA software. The details of the sample preparations and the results of the pre-
vious TEM studies have been provided in previous publications [6,7]. In this paper,
after a short analysis of the metal–oxide interface of materials irradiated in BWR and
PWR, previous results will be used in light of the correlative studies performed, com-
paring the TEM results with those of other analytical techniques. A focused ion beam
was used for the preparation of TEM foils for metal–oxide analysis.
ELECTRON PROBE MICROANALYSIS (EPMA)

PSI has recently installed a new JEOL 8500F EPMA instrument dedicated to radio-
active materials in the hot laboratory. The instrument is equipped with a field
emission gun, allowing high resolution for image acquisition. It is shielded and
equipped with four wavelength dispersive spectrometers (WDS) in a lead-shielded
cabin in the hot laboratory. The details of the instrumentation and the setup are
provided elsewhere [8]. The acquisition conditions are provided for each
measurement.
SECONDARY ION MASS SPECTROSCOPY (SIMS)

The SIMS results have been obtained with a fully shielded SIMS ATOMIKA 4000
instrument, dedicated to the isotopic characterization of solid-state materials [9]. The
instrument is equipped with three different types of primary ion (PI) sources, namely
surface ionization source (133Csþ), field ionization source (69Gaþ), and plasma ion
source, usually used with (16O2)þ. To achieve high vacuum for hydrogen measure-
ments, the samples were used as received and were not embedded in any resin.
NEUTRON RADIOGRAPHY
The neutron imaging investigations were performed at the ICON facility at the
Swiss neutron source SINQ (PSI). The non-destructive character of neutron imag-
ing allows in situ investigations of hydrogen uptake and relocation processes in zir-
conium alloys. The strong contrast between hydrogen (very high total neutron
cross section) and zirconium (very low total neutron cross section) provides the
possibility to detect even small changes in the hydrogen concentrations.
For the in situ investigations the so-called “midi setup” was applied. An illumi-
nation time of 117 s results in a frame repetition time of 2 min. The spatial resolu-
tion was about 0.1 mm. For the mechanical loading during the annealing in the
INRRO furnace, a special sample load device was constructed. The experiments
comprise annealing at 350 C in argon/hydrogen atmosphere for hydrogen loading
of the samples up to a certain local hydrogen concentration, followed by an anneal-
ing in inert atmosphere at the same temperature. The field of view was limited by
the INRRO furnace windows with a diameter of 40 mm.
HYDROGEN CONTENT OF THE CLADDING IN BWR
To have an overview of the data available and the role of different parameters on
the hydrogen content of the cladding, materials irradiated up to different number
of cycles in the reactor are presented in the following.
DISTRIBUTION OF HYDROGEN ALONG THE ROD ELEVATION

The distribution of hydrogen has been examined along an LK3/L cladding after
seven cycles in the reactor for the same rod. Figure 1 presents the results of hydrogen
measurement at different elevations of this rod (AEB072-J9). The hydrogen present
in the oxide is indicated separately and the total amount of hydrogen in the seg-
ment is also reported. As can be observed, the hydrogen distribution in this rod
seems to be relatively homogeneous, if the separation of hydrogen in the metal part

of the material is considered. However, the distribution in the oxide seems to be in-
homogeneous. The hydrogen content in the metal part of this rod, in the upper and
lower elevations of the rod, is inferior to the rest of the rod.
RESULTS FROM DIFFERENT RODS AT SIMILAR ELEVATIONS

To compare the distribution of hydrogen in different rods from different assemblies after
the same number of cycles, hydrogen content of three other rods with the same cladding
grade has been examined. The hydrogen content of those claddings is higher than the
rod shown in Fig. 1, as it can be observed from Fig. 2. It is worth noting that, the maxi-
mum hydrogen content of this cladding grade, in the metal is below 300 ppm after seven
cycles in the reactor, in most measurements available and presented in this figure.
COMPARISON OF DIFFERENT CLADDING GRADES

To compare this cladding grade with other grades studied in the same project, an
LK2/L cladding, examined after seven cycles, is presented (Figs. 3 and 4). The distri-
bution of hydrogen in the latter, along the rod elevation, is different from that of
the LK3/L rod presented above. As can be observed from these hydrogen-content
measurements, the analysis of the hydrogen content in the LK2/L cladding selected
for this study shows higher hydrogen concentrations, in general, and the distribu-
tion along the elevation of the rod is totally different in comparison with that of the
LK3/L cladding examined. The hydrogen content of the LK2/L cladding has
FIG. 2 Hydrogen content of different rods with LK3/L fuel cladding irradiated for seven
cycles in the BWR. The distribution of hydrogen in the metal is presented. The
arrow indicates a spacer region.
FIG. 3 Comparison of the hydrogen content (in the metal) of an LK3/L cladding and an
LK2/L cladding, irradiated for seven cycles in the BWR. The positions of the two
rods in the fuel assemblies could be considered as comparable.
FIG. 4 Comparison of the hydrogen content (total) of an LK3/L and an LK2/L cladding
(as shown in Fig. 3), irradiated for seven cycles in the BWR. The positions of the
two rods could be considered as comparable.
increased in the lower elevations and in the upper elevations of the rod. In this rod,
again, the variation of the total hydrogen content of the cladding does not reflect
exactly the same trend as that of the hydrogen content of the metal (in particular,
for the point at the elevation of 3120 mm).
MAXIMUM RESIDENCE TIME

In the case of the LK3/L material, a further examination has been performed and
few assemblies have been irradiated to nine cycles. These materials gave rise to a
large series of post-irradiation examinations (PIE) and research projects reported in
the literature [2]. The hydrogen contents of three rods are provided in Fig. 5, and
compared with the results from the claddings irradiated for seven cycles. It can be
noted that the hydrogen content of these claddings is practically twice as high as
those with seven cycles of irradiation.
ROLE OF IRRADIATION AND OXIDE THICKNESS ON HYDROGEN UPTAKE

To determine the hydrogen uptake and the pickup fraction for each material, the
exact oxide thickness of the cladding at the elevation of interest would be needed.
The non-destructive (NDT) data for such measurements is available for some of
the claddings; however, the lower accuracy of the NDT oxide thickness values could
induce a large error in such calculations. The comparison of oxide thickness and
hydrogen content has, therefore, been limited only to the data points where a de-
structive value for the oxide thickness is available, from metallography or from
EPMA studies (Table 3). These values are represented both for LK3/L and for LK2/L
FIG. 5 Comparison of the distribution of hydrogen in the fuel rods, with seven and nine
cycles of irradiation, for LK3/L cladding grade. The distribution of hydrogen in
the metal is presented.
in Fig. 6. As can be observed from these results, the oxide thickness of the LK3/L
cladding grade is much lower than that of LK2/L cladding. It can be concluded that
at higher fluences, the oxidation of LK2/L proceeds at a higher rate, and the hydro-
gen pickup fraction is much lower in comparison with the LK3/L grade.
It must be mentioned that the data presented in Fig. 6 for the two cladding
grades concerns the midspan values (from the elevation of 2000 6 5 mm). Measure-
ments performed on the third spacer of the LK3/L (from the elevation of
1770 6 5 mm) show that the trend is not the same in the spacer area. To compare
the data with the midspan results, the data of the spacer region for LK3/L material
are provided as a function of oxide thickness. As it can be observed the hydrogen
uptake of the cladding in the spacer region is much lower, whereas the oxide thick-
ness is much higher (Fig. 7). In the case of the seven-cycle spacer segment (indicated
by the arrow), oxide spalling has been observed in the sample. For this reason, the
value of mean oxide thickness is confirmed from the values of circumferential NDT
measurements, as well as wall-thickness measurements and oxide-thickness meas-
urements from the DT segment. The value provided can be considered as the mini-
mum value for the mean oxide thickness.
FIG. 6 Correlation of total hydrogen content and oxide thickness, for LK3/L and LK2/L
claddings grades as a function of irradiation time. The theoretical uptake for
different pickup fractions is provided for reference. Data presented in the graph
is for midspan peak burnup elevation (approximately 2000 mm).
FIG. 7 Correlation of total hydrogen content and oxide thickness, for LK3/L at two
elevations (midspan: 2000 mm, and third spacer: 1770 mm). Arrow indicates the
data for AEB072-E4 spacer position, please refer to the text for details of this
point.
The data in above-mentioned Figures can also be presented as a function of flu-

ence or burnup of the studied segments (Table 3), not presented here. The results
indicate similar and cumulative effects of residence time and irradiation.
These data confirm that hydrogen pickup fraction is not always a fixed
parameter even within the same cladding grade, and in the same rod, at different
elevations. To verify if this behavior is specific to BWR material, one example is
provided from a PWR cladding.
DISTRIBUTION OF HYDROGEN AS A FUNCTION OF ELEVATION IN A PWR

CLADDING
In the case of PWR, less information is available for publication; nevertheless, one
example is provided for low-tin Zircaloy-4 (Fig. 8). As can be observed, the distribution
of hydrogen in this cladding is also inhomogeneous; furthermore, the oxide thickness
increases from the bottom of the rod toward the upper parts, and the oxide-thickness
values indicate a pickup fraction, which is homogeneous along the rod, in the limits of
data points available, apart from the lowest elevation of the material (Table 4). In com-
parison with this cladding grade, the hydrogen content of Zr2.5 %Nb is much lower,
both in the lower elevations (elevation 869 mm as provided in Table 4) and in the
higher elevations (around 3200 mm); previous studies have shown that the values are
in the range of 100 ppm for that elevation (details not shown).
FIG. 8 (a) Hydrogen content (total) and (b) oxide thickness of low-tin Zircaloy-4 fuel
cladding irradiated for 51.1 MWd/kgU in the PWR, as a function of rod elevation.
TEM ANALYSIS OF METAL–OXIDE INTERFACE OF IRRADIATED CLADDINGS

The analyses of the microstructure of the metal part of the cladding have been
reported in previous publications for different types of reactors. For references,
please see as an example, Refs 4 and 10–14. It is commonly agreed that the irradia-
tion modifies the microstructure, induces the dissolution of precipitates, and, in
some cases, the amorphization of the latter [15]. The absence of precipitates acceler-
ates the oxidation of the cladding; furthermore, the dissolution of the alloying ele-
ments in the cladding causes the modification of the chemistry of the interface, as
mentioned in Ref 16. One aspect of the dissolution that has not been sufficiently
stressed is the level of alloying-element content of the precipitates remaining in the
cladding. The results of studies in this laboratory have shown that the dissolution
rate of precipitates depends on the cladding grade and the composition of precipi-
tates; however, the alloying-element content in the SPP reduces with residence time
for all types of precipitates. At high burnups, this point should be taken into consid-
eration at the same time as the reduction of the volume fraction of precipitates,
when the limits of residence time of the cladding are defined.
On the other hand, the metal–oxide interface of different cladding grades irra-
diated in PWR reactors has been examined [6,7,17,18], and it has been demon-
strated that the morphology of metal–oxide interface plays a role on the oxidation
behavior of the material. The TEM samples for such studies have been prepared
using focused ion beam method. The details of this sample-preparation method are
published in previous publications [17]. In this study, after the analysis of the
metal–oxide interface of Zircaloy-2 irradiated for seven cycles, results of a correla-
tive study by TEM, EPMA, and SIMS will be reported.
TEM OF ZIRCALOY-2 AFTER SEVEN CYCLES

The material selected for this study was an LK3/L cladding grade irradiated for
seven cycles (AEB072-E4), and examined by non-destructive (NDT) and destruc-
tive tests (DT) as reported in Refs 1 and 4. The hydrogen data of this cladding are
reported in the present study. Several segments from this rod have been provided
for international programs. The elevation of the segment for this observation is
1850 mm, the oxide thickness measured by SEM is 16 lm. Figure 9(a) represents
the metal–oxide interface of this cladding. The EDS analysis of the microstructure
of this sample has shown no remaining precipitates in the metal or in the oxide
sides of the interface. Figure 9(b) provides the TEM dark field contrast of the metal–-
oxide interface of this material and hydride lenses can be observed in the metal side
of the interface. In this micrograph, in the oxide side of the interface, a feature is
marked with arrows indicating the presence of fine linear features leading to the
interface. The details of the fine feature are better observed in Fig. 9(c). No alloying
element has been detected in these features, till now.
Figure 10 shows in more detail the hydrides present in the metal side of the
interface. The hydrides are both radial and circumferential and they have a specific
orientation with respect to the matrix zirconium. In the oxide side of this material a
fine fibrous structure can be observed [also seen in Fig. 9(b)], which shows exactly
the same contrast as the hydrides. These features could be attributed to the contrast
FIG. 9 TEM bright field (a) and dark field (b) contrast of the metal–oxide interface of
LK3/L cladding grade irradiated for seven cycles in BWR. Arrows indicate the
features leading to the metal–oxide interface, the FIB sample being prepared
parallel to the metal–oxide interface, the artifacts of milling can be observed in
the direction perpendicular to the columnar growth of oxide, and (c) TEM bright
field micrograph of the LK3/L 7 c sample, for better observation of the fine
features (arrow indicates the feature in the metal).
FIG. 10 TEM dark field contrast of the metal–oxide interface of LK3/L cladding grade
irradiated for seven cycles in the BWR, indicating the hydrides in the metal side
of the interface (white arrows).
because of the crystal orientation of the oxide grains; however, as they are present
only in specific conditions, work is in progress to verify this point.
It is interesting to note that these features are also observed in materials studied
in PWR; one example of very similar feature of the metal–oxide interface of a low-
tin Zircaloy-4 is provided in Fig. 11(a). As can be observed, these features are very
similar in the two materials. In both cases, these features are linked to hydride
lenses and could imply that the hydrides themselves will play a role in the hydrogen
FIG. 11 (a) TEM dark field contrast of the metal–oxide interface of low-tin Zircaloy-4
irradiated for four cycles in the PWR, arrow indicates the feature penetrating in
the metal side of the interface (arrows). (b) Dark field contrast of metal–oxide
interface of a Zr 1 % Ni oxidized in an autoclave, indicating a similar feature at the
interface (arrow).
uptake at high-burnup stages. The role of hydrides in acceleration of oxidation has

already been studied; see, for example, Ref. 19. The metal–oxide interface of an un-
irradiated binary Zr 1 %Ni alloy studied previously [20] shows a similar feature
[Fig. 11(b)]. Both fine features are at the metal–oxide interface and are penetrating
into the metal side of the material. It is not clear if these two features represent the
same phenomenon, or are two different phenomena.
CORRELATIVE STUDY OF THE METAL–OXIDE INTERFACE OF PWR

CLADDINGS WITH TEM-EPMA-SIMS
In our previous studies, a comparative examination of the morphology of
the interface has been reported by TEM analyses [7]. As an example, the metal–-
oxide interface of two cladding grades irradiated in PWR showing considerably dif-
ferent oxidation rates have been examined in view of correlating the morphology of
the interface to this drastic difference of oxidation [7]. It has been shown that the
cladding with better resistance to oxidation has a much more inhomogeneous inter-
face in comparison to the cladding showing a faster oxidation rate.
To confirm that this finding is representative of the overall structure of the ma-
terial and the TEM analyses are reproducible by other methods, these TEM results
have been further explored by complementary analytical methods, namely EPMA
and SIMS.
Figure 12 provides the TEM results as studied in previous publications [7],
where the metal–oxide interface of Zr 2.5 % Nb has been compared with a low-tin
Zircaloy-4. The rod elevation and oxide thickness values for these two samples are
provided in Table 4. The metal–oxide interface of the Nb-containing alloy is much
more inhomogeneous, and this alloy has a much better resistance to oxidation. To
have a perfect correlation, the interface of these two materials, exactly at the same
elevation, has been examined by EPMA and, subsequently, by SIMS. In other terms,
the same specimen was used for all studies.
The EPMA maps of the two materials are not presented, however, it can be
described as follows: the presence of alloying elements in the oxide can be observed
in both materials. Hydrides can be observed in the SEM image and, in the case of
low-tin Zircaloy-4, these are more abundant; knowing the hydrogen content of
these two cladding grades (Table 4), this observation is not surprising.
Figure 13 provides the qualitative line scans of oxygen from the metal–
oxide interface of these two alloys. The oxygen profile of the low-tin Zircaloy-4 is
much more homogeneous. The profile in the case of Zr 2.5 % Nb is much more
variable and a decrease in oxygen concentration can be observed. The average val-
ues of several measurements presented on the same Figure, have also shown that
the diffusion distance of the oxygen into the metal side of the interface is longer in
low-tin Zircaloy-4. These results are in agreement with our TEM data, where we
have an inhomogeneous oxygen signal in the oxide for Zr 2.5 % Nb; and the diffu-
sion zone in the metal is larger in the case of the low-tin Zircaloy-4 [6]. Knowing
FIG. 12 (a) TEM bright field contrast of metal–oxide interface of the Zr 2.5 % Nb alloy
after three cycles of irradiation. Arrows indicate the limits of the interface. The
interface of this alloy has a very different profile compared to the Zircaoly-4
low-tin alloy. The metal–oxide interface shows regions that are rich in Nb (one
such region is indicated by the letter A). Cracks can be observed in this alloy, in
the oxide region. (b) Metal–oxide interface of the low-tin Zircaloy-4 in the
similar magnification for comparison. The TEM-EDS map of inset in (a) is
presented as overlay, indicating the presence of Nb in the oxide (red: Nb, green:
O, blue: Zr).
that the spatial resolution of EPMA is much lower than the TEM, however, the data
is from a much larger surface area of the specimen.
The same two samples in a very similar region have been studied by SIMS and
the results are presented in Fig. 14, where 3D maps of these two materials are
acquired and in this figure, a profile is produced at a given position for each mate-
rial to show that the metal–oxide interface of Zr 2.5 %Nb is indeed much more
irregular than that of low-tin Zircaloy-4. Please see Appendix 1 for the details of
sputtering, mapping, and averaging directions. Furthermore, these data are inte-
grated to obtain an average intensity for individual planes. Results are presented for
hydrogen, oxygen, and zirconium in Fig. 15.
Here, again, the distribution of oxygen in the case of Zr 2.5 % Nb cladding is
much more inhomogeneous; and the diffusion region in the oxide being substoi-
chiometric, this distance is considerably larger than in the case of low-tin Zircaloy-
4. A larger diffusion distance implies a better protection against penetration of dif-
ferent species into the metal. In the case of SIMS studies presented here, the analysis
FIG. 13 EPMA qualitative line scans of the metal–oxide interface of irradiated alloys,
typical transition zones. Average values from more than 10 line scans are
provided on the right-hand side of the figures.
FIG. 14 SIMS analyses of metal–oxide interface of irradiated claddings. One example of

2D profile for each material, extracted from the 3D maps of Zr. This method,
also called “checkerboard,” has been performed under identical conditions for
the two materials, as examined in Figs. 12 and 13 (total sputtered area: X, 10 lm;
Y, 10 lm; and Z, 0.5 lm). The depth resolution in this analysis is in the range of
10 nm. The interface in the case of Zr 2.5 % Nb (b) is much more irregular.
FIG. 15 SIMS analyses of the metal–oxide interface (M/O) of irradiated claddings.

Average intensities created from the 3D maps. (a) Across the M/O of Zircaloy-
4, and (b) across the M/O interface of Zr 2.5 % Nb. Average M/O transition area:
2 lm for Zircaloy-4 and 2.5–3 lm for Zr 2.5 % Nb. Each data point represents a
signal coming from a volume of 3 lm3 (10 lm Y 0.5 lm Z 0.6 lm X).
focuses on the oxide side of the interface, as the sputtering yield in the metal side is
very low.
Discussion
The measurement of oxidation and hydrogen uptake of different claddings irradi-
ated for a different number of cycles in the reactor is a difficult practice, is expen-
sive, and is time consuming. It is clear that such data is not very abundant. The
procedure starts with the selection of irradiated fuel rods after an NDT inspection
on the reactor site. It is followed by the transport of the selected rods to the hot lab-
oratory and the examination of the rods in the hot cells by NDT. Only after this
procedure and after the selection of certain positions on each rod, cutting and ex-
amination by DT of the selected segments are undertaken. The correlation of the
results to certain phenomenological behaviors is more difficult if the results
obtained from several different reactors are used. The reason for this difficulty is
that, in practice, every single data point will have a different history of irradiation,
oxidation and hydriding and, thus, the data cannot be perfectly comparable. Fur-
thermore, the water chemistry of different reactors is usually not identical and again
this parameter adds an extra variable to the results obtained.
In this respect, the data obtained from a single reactor has the advantage that,
provided the same time frame is considered, results can be at least comparable, as
the overall environment of the different materials would be similar. It is clear that
even in the same reactor, the water chemistry can change when a long period of
irradiation is considered, as it has been reported in the past by several studies; this
parameter, therefore, should also be taken into consideration when mechanistic
analyses are to be made.
All these facts can explain, to a great extent, the difficulties in analyzing the
results obtained from the reactors. For these reasons, the results presented in this
study should have a useful application in the analysis of the behavior of materials
under irradiation. Not only are they from the same reactor (i.e., a BWR and a
PWR), but they are mostly either from the same fuel rod, at different elevations, or
from different fuel rods made of the same cladding grade and irradiated at the same
elevation, and even at the same position in the rod assembly [1,2]. This would
mean that the variables in the data points available are minimized to a great extent
and the results could be directly correlated to the phenomenological behaviors.
Another difficulty in the interpretation of results may arise from the methodol-
ogies used in the quantification of different data points, such as the oxide thickness
and the hydrogen content. As an example, the oxide-thickness values can be
obtained by NDT or DT. The studies performed at PSI have shown that, although
great progress is made in the determination of the oxide layer by NDT, and the
methods developed to account for the magnetic CRUD present on the claddings, in
the case of studies performed by MAGNACROX [1,2], the NDT methods, in most
cases, may overestimate the oxide-thickness value and, therefore, the use of these
values could introduce a large error in the interpretation of results. Thus, in this
study, no NDT data points were considered and the more-reliable DT values were
used for oxide thickness.
The variation of oxide thickness around the circumferences of the segments
examined by DT has also to be mentioned, however, the average values used can
be, to a great extent, considered as representative.
Regarding the hydrogen data, the method used in the present study could also
introduce a certain error [21], however, in the case of irradiated materials, no other
technique was available at the time of the experiments and the data can be consid-
ered unique. The reproducibility of these results has been tested in an inter-
laboratory campaign and is reported elsewhere [1].
Based on these facts, the results presented in this study will be interpreted in
the following, knowing that, although a large amount of data presented is originat-
ing from BWR, the phenomenological conclusions can also be applicable to PWR,
taking into account the differences between the two types of reactors.
SEPARATION OF HYDROGEN CONTENT OF METAL AND OXIDE

The possibility to separate the hydrogen content of the metal and the oxide part of
the segments by hot-gas extraction allows a better estimation of the real hydrogen
content of the metal part of the cladding at each stage. The contribution of hydro-
gen content in the oxide in this study is usually not very high, however, in some
cases, it could be rather high, and, in that case it will bias considerably the measure-
ments. As an example, in Fig. 1, at one of the points studied (at the elevation of
1700 mm), the contribution of the oxide part of the segment is much higher than
the average value for the latter at that point; the total hydrogen content seems to be
much higher than the other data points available, whereas the metal part of the
segment has hydrogen content that matches with the trend observed in this clad-
ding. This point has an intermediate position, it is not very far from the spacer
region (but it is nevertheless outside the spacer region). Such discrepancies between
the metal and the total content of a segment have been observed on other rods as
well. Therefore, the possibility to separate the hydrogen content of the oxide and
the metal is useful for a better characterization of the material. Furthermore, it
must be noted that although in the present study, at least for one elevation (mid-
span elevation around 2000 mm) for each rod a hydrogen and a metallography (de-
structive) test were carried out and the two samples were selected adjacent to each
other, so that the oxide layer could be measured to compare with the hydrogen
sample’s oxide thickness; for many hydrogen samples, the direct metallography
data to measure accurately the oxide thickness are not available. The low-
temperature release in the above method could provide an indirect indication of the
oxide thickness of the segment analyzed. Spalling of the oxide, however, cannot be
detected.
VARIATION OF HYDROGEN CONTENT ALONG THE ROD AND AT DIFFERENT

POSITIONS IN THE REACTOR
The fact that the hydrogen content varies in a single fuel rod from one point to
another and also in different rods, implies that the distribution of hydrogen in dif-
ferent claddings, and at different elevations of the same cladding, is not necessarily
reflected from the measurement of a single point by destructive tests, in particular,
if a precise overview of the hydrogen content is required. In case a mechanistic
understanding of the hydrogen uptake is envisaged, it is interesting to know that,
even in the same rod at two different elevations, different values of hydrogen con-
tent can be seen; and, even at the same elevation, the hydrogen content is different
from one azimuth to another. Such observations have been made at PSI and con-
firmed by hot-gas extraction of the different quarters of the same segment. This var-
iability can also be observed from the metallographic results of the adjacent
segments. As an example, the rod AEB072-J9 (LK3/L), at the elevation of 1700 mm,
hydrogen contents of 137 ppm and 242 ppm were shown for two adjacent quarter
segments in the metal. However, in general, the LK3/L cladding shows very little az-
imuthal variation of hydrogen content.
This fact implies that averaging the data in view of depicting a trend
might totally falsify the mechanisms that are involved in hydrogen uptake if this
exercise is not performed with sufficient knowledge and is not done carefully.
Furthermore, the history of the segment used to obtain a data point could influ-
ence the value of uptake and, in the same family of material, in the same reactor,
and in the same axial position, and for the same position of the rod in the assembly,
the hydrogen content from one rod to another varies. As an example, if the typical
hydrogen content of Zircaloy-2 cladding grades is needed for a residence time of
seven cycles, making reference to the data presented in Figs. 2 and 3 will show that a
large database will be more reliable. This point can be well illustrated in Fig. 2, for
the data points at the elevation of 2000 mm for the rods with E4 designation in the
assembly. At this elevation, three rods at the position of E4 in three different rod
assemblies are tested and they do not have the same hydrogen content. It could be
argued that this scatter is within the accuracy of the hydrogen-measurement tech-
nique used, however, between the lowest and the highest values measured there is
100 ppm of difference and this can be considered as a clear variation.
In the present study, no data are available to examine for the same family of
fuel from two different rods, if the trend of increase and decrease of hydrogen con-
tent as a function of elevation, presented in LK3/L rod at the position of J9, is repro-
ducible. It must be borne in mind that the rod in question had a fuel with lower
enrichment compared to the other three rods examined, and this difference could
play a role together with its different position in the assembly on the nature of
uptake.
However, the data presented in Fig. 3 shows that the distribution along the ele-
vation could also vary from one rod to another. In the case of this comparison, it is
clear that the cladding grade of the two materials is not the same; regarding their
positions in the assembly, however, they can be considered as comparable.
Although they contain the same amount of Ni (0.05 %), their Sn and Fe contents
are different. Nevertheless, it could be concluded that the trend of increase and
decrease of content of hydrogen along the rod, cannot be extrapolated systemati-
cally from one series of results.
THE ROLE OF TIME OF RESIDENCE ON THE HYDROGEN PICKUP FRACTION

The maximum time of residence examined till present for BWR is nine cycles [2],
and the comparison of hydrogen uptake of the claddings shows that the uptake is
considerably higher in the last two cycles, in the case of LK3/L (Fig. 5). These data
can be examined in more detail, comparing the oxide thickness and the hydrogen
content of the segments. In this study, the oxide thickness of segments adjacent to
the hydrogen sample is available for certain elevations, and the examination of the
results, in view of observing the pickup fraction as a function of oxidation time, can
be seen in Fig. 6. As can be observed from this figure, for the elevation of 2000 mm
for different rods, the LK3/L cladding grade shows a different pickup fraction as
opposed to LK2/L. It must be noted that in the case of this Figure, every data point
comes from a different fuel rod and the time of extraction of the rod from the reac-
tor will not necessarily be the same as the number of cycles are also different.
Therefore, it is interesting to see that, despite the fact that the data are absolutely in-
dependent from one another, a clear trend can be observed. The graph can, there-
fore, be considered as very reliable. Although, at this stage it is difficult to interpret
this result in light of the composition of the two cladding grades, it is noteworthy
that both materials have the same Ni content. The presence of Ni has been [16]
considered as a major parameter for higher hydrogen pickup mainly in the case of
channel materials in BWR. In that study, the amount of Ni is recommended to be
in the range of 0.03 %–0.05 % to keep the accelerated pickup at the high burnups
limited. This recommended value is respected in the cladding grades, being pres-
ently the cladding grade for BWR. However, the aim of the present study being a
contribution to develop a mechanistic understanding, this data can bring a light on
the role of other alloying elements present in the cladding, and the impact of Ni in
parallel. It is further mentioned in Ref 16 that the role of Fe has also to be consid-
ered in this respect. In a different study, it had been shown that the addition of Fe
reduces the hydrogen uptake of the Zircaloy-2 family [22].
The present results demonstrates that the pickup fraction in the same cladding
grade is different at different elevations, and the data presented in Fig. 7 shows that
the oxide thickness of the cladding in the spacer region is much higher than in the
midspan, whereas in the example presented here, the pickup fraction is much lower.
It must be noted that the oxidation condition in the spacer is considered to be com-
pletely different from the midspan position, there, the oxide thickness is much
higher and the contact points with the spacers oxidize much more, so the oxygen
and hydrogen potential could also be different. However, this observation is useful,
as it is important to know that even along the same cladding rod, the material
shows different hydrogen content and a different pickup fraction.
The hydrogen distribution in the case of low-tin Zircaloy-4 irradiated in the
PWR, shows a variation along the rod, however, the variation of the oxide thickness
follows the same trend. This would imply a pickup fraction between 15 % and 19 %
for this material. The data is not abundant for the hydrogen measurements; further
microstructural information can better elucidate this part.
ANALYSIS OF METAL–OXIDE INTERFACE

In this part of the paper, the aim is to correlate the microstructural and
macrostructural behavior in view of a better understanding of the mechanism of ox-
idation and hydrogen uptake. The analysis of the microstructure of the metal–oxide
interface by TEM has been performed by several researchers.
In a previous study, it had been demonstrated by analytical TEM that the oxi-
dation of Zr-based cladding is among other parameters dependant on the intrinsic
diffusion properties of the material [6]. Furthermore, it had been discussed that the
barrier layer can be considered as the region at the metal–oxide interface where the
reactions take place and it can be described as a composite diffusion layer consisting
of a fine layer in the oxide and a diffusion layer in the metal side of the interface
[6,7].
In the present study, this aspect is further examined using the correlative
microscopy. The materials studied by TEM [7] have been re-examined by
EPMA and SIMS, and it has been confirmed that, in the case of material
showing a better resistance to oxidation, the morphology and the composition
of the metal–oxide interface is very inhomogeneous, this can be verified, in
particular, from the SIMS results. These observations further confirm that, in
the case of this material, the oxide part of the interface is more substoichio-
metric in comparison with the low-tin Zircaloy-4. The oxide in the case of
low-tin Zircaloy-4 has a homogeneous composition. This point can be verified

from the oxygen profiles from EPMA and SIMS. The fact that the metal side
of interface is richer in oxygen in the case of low-tin Zircaloy-4, i.e., for the
alloy with a faster oxidation rate, also confirms the observation in TEM. It is
interesting to note that although these three methods have different scopes
and limitations, regarding the detection limit, the resolution and, in the case
of SIMS, the sputtering yield (which also leads to a certain limitation), the
results are complementary and coherent. An inhomogeneous interface could
be defined from these observations as an interface, where the diffusion layer
is extended on a longer distance, in particular, in the oxide side of the inter-
face. On the other hand, deeper penetration of oxygen in the metal would be
a sign of a higher diffusion coefficient of oxygen into the metal.
Furthermore, the presence of micropores at the metal–oxide interface and the
fact that they can be responsible for the hydrogen uptake, independent from the
composition of the cladding, has been proposed by previous researchers [23]. In
this study, examples of metal–oxide interfaces of different cladding grades oxidized
in BWR, in PWR, or in an autoclave, are provided. Some features have been
observed that could have a similarity with such pores. Although the chemical analy-
sis of these features reveals no alloying-element segregation, these observations
need to be further studied to verify their nature, and their role on the extent of oxi-
dation and uptake of the material.
In this respect, the examination of hydrogen uptake of an un-irradiated clad-
ding in the presence of stress, reveals the role of stress on the uptake, which leads to
a heterogeneous hydrogen distribution. This point is further discussed in the next
section [24].
FACTORS INFLUENCING THE HETEROGENEITY OF HYDROGEN UPTAKE

As we have seen from the present results, the hydrogen content of the cladding is
not always homogeneously distributed. It is important to note that the hydrogen
uptake and oxidation behavior of the material remain related, however, the fact that
hydrogen can penetrate into the material independent of the oxidation phenomena,
and that the hydrogen distribution can be extremely inhomogeneous, should not be
neglected. This parameter could modify the hydrogen content of the material even
in the process of oxidation. The laboratory experiments of hydrogenation can well
demonstrate the influence of local temperature gradients on the extent of uptake of
hydrogen and its inhomogeneous distribution (not shown here).
Besides the temperature, stress distribution is another source for the heteroge-
neity of the hydrogen distribution. This aspect can be well demonstrated in the in
situ study carried out under a stress field in the case of a notched Zircaloy-4 tensile-
test sample having a fatigue crack.
In this experiment, the hydrogen uptake is monitored on line and a local
increase in concentration of hydrogen can be observed at the crack tip (Fig. 16).
This study demonstrates that the stress may enhance hydrogen uptake and result in
FIG. 16 Neutron radiograph sequences of a Zircaloy-4 taken during hydrogen loading

in Ar/H2 atmosphere until 18 000 s and inert annealing of a mechanical stressed
specimen until 57 000 s at 350 C.
higher local hydrogen concentrations. In this study, the hydrogen uptake starts after
a certain period of time, or a “delay” time. This behavior has been related to the
oxide-absorption phase; however, further examination of the material is necessary
to confirm this statement.
The above results imply that the heterogeneity of the distribution of hydrogen
in a cladding could originate from different microstructural conditions, and one
possible cause could be the tensile stress state at a crack tip.
One of the aspects of oxidation that has not been sufficiently explored is
the nature of the oxide layer over the whole surface of the material and along
the whole thickness of the oxide, including the porous oxide considered as a
non-effective barrier for penetration of oxidizing species and hydriding species.
In some cases, it has been observed that thinner oxide layers have more flaws
and many more cracks in them. This aspect could lead to easier access of the
hydrogen species to the interface. The present results illustrate that necessarily
a low oxidation rate will not lead to a lower hydrogen pickup and the hydro-
gen pickup fraction could have, in certain cases, an inverse relation with the
oxide thickness unlike what is usually assumed.
Conclusions
The aim of this paper has been to examine the available data on the hydrogen con-
tent of different cladding grades in the LWR, to bring new hints in the understand-
ing of the mechanism of oxidation and hydrogen uptake of the cladding. The
distribution of hydrogen has been examined along the elevation of the fuel rods, for
claddings with different levels of iron and identical levels of nickel, in case of the
BWR materials, and for a low-tin Zircaloy-4 cladding, in the case of PWR
materials.
The results have confirmed that the distribution of hydrogen for a given fuel
rod could be different and the pickup fraction of the cladding could be different at
different elevations of the same rod. This finding has led to the conclusion that the
composition of the cladding is not the only parameter responsible for a specific
behavior, but that the water chemistry and the history of the material will also influ-
ence its uptake behavior.
Furthermore, the microstructure of the metal–oxide interface of different
alloys has been examined, and, in the case of the alloys irradiated in the reac-
tor, traces of alloying-element segregation could not yet be observed in
the samples studied, in particular, for Zircaloy-2 and Zircaloy-4 near the
metal–oxide interface. In the case of Zr 2.5 % Nb, the Nb-rich second phase
is present, which has been determined to be b Nb. The metal–oxide interface
of this cladding is very inhomogeneous. The microstructural studies further
showed that, in the case of Zircaloy-2 and Zircaloy-4 and autoclaved Zr 1 %
Ni, linear features perpendicular to the interface leading to the metal are pres-
ent, and, in most cases, these features are in the vicinity of hydrides. It is,
therefore, suggested that in the search for parameters responsible for the
enhanced pickup fraction of the cladding, the origin and the nature of these
features and their numbers could be evaluated and their role on the uptake
could be further searched. The examination of the interface of two different
alloys by correlative analysis confirmed the differences between the interfaces
as observed by TEM. Furthermore, the role of stress has been demonstrated,
using neutron radiography.
From these results, the parameters that can influence hydrogen uptake could be
classified as follows:
(i) the composition and the microstructure of the cladding and the resist-
ance to irradiation dissolution of the precipitates,
(ii) the nature of oxide layer and its resistance to crack formation,
(iii) the microstructure and morphology of the metal–oxide interface and the
presence or absence of routes for the access of hydrogen and/or oxidizing
species to the metal,
(iv) the presence of localized flaws in the oxide and in the metal,
(v) stress distribution, and
(vi) the water chemistry and the amount of hydrogen present in the reactor
coolant
ACKNOWLEDGMENTS
The writers wish to thank Dr. S. Yagnik for the supply of hydrogen data for low-tin
Zircaloy-4. Mr. Andreas Urech is acknowledged for oxide-thickness measurements on
the destructive samples. Mrs. J. Krbanjevic is acknowledged for the preparation of one
FIB-TEM sample.
Appendix
Integration of the SIMS signal intensity along the Y and Z axes to obtain an average
intensity value for each plane, as used to calculate the data points for Fig. 15. Z indi-
cates the direction of sputtering; the data is averaged in Z and Y to provide the
variations along the X axis. The value of 0.6 lm in the X direction designates the step
size for the data points.
References
[1] Ledergerber, G., Abolhassani, S., Limback, M., Lundmark, R., and Magnusson, K. A.,
“Characterization of High Burnup Fuel for Safety Related Fuel Testing,” J. Nucl. Sci.
Technol., Vol. 43, 2006, pp. 1006–1014.
[2] Ledergerber, G., Valizadeh, S., Wright, J., Hallstadius, L., Gavillet, D., and Abolhassani, S.,
“Fuel Performance Beyond Design – Exploring the Limits,” Proceedings of 2010 LWR
Fuel Performance/Top Fuel/WRFRM, American Nuclear Society, Orlando, FL, Sept
26–29, 2010, Paper 0044.
[3] Bart, G., Blank, H., Garzarolli, F., Gebhardt, O., Hermann, A., and Ray, I., “Gösgen
Project—Post-Irradiation Characterization,” Electric Power Research Institute, Palo Alto,
CA, 1996.
[4] Valizadeh, S., Ledergerber, G., Abolhassani, S., Jädernäs, D., Dahlbäck, M., Mader, E. V. G.,
on the In-Reactor Performance of BWR Cladding,” Zirconium in the Nuclear Industry: 16th
Symposium, ASTM STP 1529, ASTM International, West Conshohocken, PA, 2011, p. 729.
[5] Hermann, A., Wiese, H., Bühner, R., Steinemann, M., and Bart, G., “Hydrogen Distribution
Between Fuel Cladding Metal and Overlying Corrosion Layers,” Proceedings of the ANS
International Topical Meeting on LWR Fuel Performance, Park City, Utah, American Nu-
clear Society, LaGrange Park, IL, 2000, pp. 372–384.
[6] Abolhassani, S., Bart, G., and Jakob, A., “Examination of the Chemical Composition of
Irradiated Zirconium Based Fuel Claddings at the Metal/Oxide Interface by TEM,” J.
Nucl. Mater., Vol. 399, 2010, pp. 1–12.
[7] Abolhassani, S., Restani, R., Rebac, T., Groeschel, F., Hoffelner, W., Bart, G., Goll, W., and
Aeschbach, F., “TEM Examinations of the Metal-Oxide Interface of Zirconium Based
Alloys Irradiated in a Pressurized Water Reactor,” J. ASTM Int., Vol. 2, 2005, p. 467.
[8] Restani, R. and Wälchli, A., “Shielded Field Emission EPMA for Microanalysis of Radioac-
tive Materials,” IOP Sci., Vol. 32, 2012, p. 1.
[9] Gebhardt, O., “SIMS Depth Profiling, Line Scanning and Imaging Analyses of the Oxide
Layer on In-Reactor Corroded Cladding Specimens With High Lateral Resolution,” J.
Anal. Chem., Vol. 365, 1999, p. 117.
[10] Griffiths, M., De Carlan, Y., Lefebvre, F., and Lemaignan, C., “A TEM Study of the Stability of Inter-
metallic Precipitates in Zircaloy Nuclear Reactor Components,” Micron, Vol. 26, 1995, p. 551.
[11] Pecheur, D., Lefebvre, F., Motta, A. T., Lemaignan, C., and Charquet, D., “Oxidation of
Intermetallic Precipitates in Zircaloy-4: Impact of Radiation,” STP 1245, ASTM Interna-
tional, West Conshohocken, PA, 1994, p. 687.
[12] Pecheur, D., Lefebvre, F., Motta, A. T., Lemaignan, C., and Charquet, D., “Effect of Irradia-
tion on the Precipitate Stability in Zr Alloys,” J. Nucl. Mater., Vol. 205, 1993, p. 445.
[13] Abolhassani, S., Gavillet, D., Groeschel, F., Jourdain, P., and Zwicky, H. U., “Recent Obser-
vations on the Evolution of the Secondary Phase Particles in Zircaloy-2 Under Irradiation
in BWR up to a High Burn-up. In Light-Water-Fuel-Performance,” American Nuclear So-
ciety, Park City, Utah, 2000, pp. 470–484.
[14] Garzarolli, F., Goll, W., Seibold, A., and Ray, I., “Effect of In-PWR Irradiation on Size,
Structure, and Composition of Intermetallic Precipitates of Zr Alloys,” STP 1295, ASTM
International, West Conshohocken, PA, 1996, p. 541.
[15] Motta, A. T. and Lemaignan, C., “A Ballistic Mixing Model for the Amorphization of Pre-
cipitates in Zircaloy Under Neutron Irradiation,” J. Nucl. Mater., Vol. 195, 1992, p. 277.
[16] Garzarolli, F., Cox, B., and Rudling, P., “Optimization for Zry-2 for High Burnups,” J. ASTM
Int., Vol. 7, 2010, Paper ID 102955.
[17] Abolhassani, S. and Gasser, P., “Preparation of TEM Samples of Metal-Oxide Interface by
the Focused Ion Beam Technique,” J. Microsc., Vol. 223, 2006, p. 73.
[18] Bossis, P., Lelievre, G., Barberis, P., Iltis, X., and Lefebvre, F., “Multi-Scale Characterization
of the Metal-Oxide Interface of Zirconium Alloys,” STP 1354, ASTM International, West
Conshohocken, PA, 2000, p. 918.
[19] Bisor-Melloul, C., Tupin, M., Bossis, P., Chêne, J., and Jomard, F., “Influence of Zirconium
Hydrides on Zircaloy-4 Corrosion in PWR Simulated Conditions in Laboratory,” 14th
International Conference on Environmental Degradation of Materials in Nuclear Power
Systems Water Reactors 2009, 2009, pp. 1434–1445.
[20] Proff, C., Abolhassani, S., and Lemaignan, C., “Oxidation Behaviour of Zirconium Alloys
and Their Precipitates – A Mechanistic Study,” J. Nucl. Mater., Vol. 432, 2013, p. 222.
Activation Analysis, A Non-Destructive Technique for Hydrogen Level Assessment in Zir-
conium Alloys,” J. Nucl. Mater., Vol. 425, 2012, p. 211.
[22] Kakiuchi, K., Itagaki, N., Furuya, T., Miyazaki, A., Ishii, Y., Suzuki, S., Terai, T., Yamawaki, M.,
Barberis, P., Kapoor, K., Kim, Y. S., Motta, A., Cox, B., Hallstadius, L., and Ramasubramanian,
N., “Role of Iron for Hydrogen Absorption Mechanism in Zirconium Alloys,”ASTM Special
Technical Publication, ASTM International, West Conshohocken, PA, 2005, pp. 349–366.
[23] Cox, B., “A Mechanism for the Hydrogen Uptake Process in Zirconium Alloys,” J. Nucl.
Mater., Vol. 264, 1999, p. 283.
[24] Grosse, M., Steinbrueck, M., Stuckert, J., Kastner, A., and Schillinger, B., “Application of
Neutron Radiography to Study Material Processes During Hypothetical Severe Accidents
in Nuclear Reactors,”J. Mater. Sci., 2012, pp. 6505–6512.
DISCUSSION
Q1:—The hydrogen content of LK3 at 9 cycles is according to your slide 600

wt ppm, which is similar to that of LK2 at 7 cycles. LK2 shows a big acceleration of
corrosion between cycles 5 and 7. Is the same acceleration seen in LK3 between 7
and 9?
Authors’ Response:—The oxidation rate of LK3/L has also changed between the
7 and 9 cycles. However, the change is not as high as that for LK2/L. The average
oxide thickness of the cladding at 9 cycles is more than double the value for 7
cycles, as it can be observed from Fig. 6. It is worth noting on the same figure that
the pick-up fraction of LK2/L decreased from 6 to 7 cycles.
Q2:—Has the oxide undergone any oxide spallation that could artificially
increase the hydrogen pickup fraction?
Authors’ Response:—The samples selected for hydrogen content measurements

are selected after non destructive examinations (NDT). Usually, if there is any spal-
ling, it can be observed in the NDT stage. Usually in the cladding grades reported
in this paper, the spalling has occurred at the spacer region. The hydrogen content
of the cladding at these points can be much lower than the midspan values (Fig. 7),
but in some cases it can be much higher than the surrounding hydrogen content
(not shown in this study).
One of the aims of this paper is to show that with the measurement of the
hydrogen content at one elevation, it is not possible to create a correct image of the
overall hydrogen content of a rod.
Another point is that in practice, the ratio of hydrogen uptake due to spal-
ling can not be distinguished, from that of pickup fraction, due to oxidation.
The results of the LK2/L would imply that the fraction can reduce with long
term oxidation.
Questions from K. Kapoor, NFC, Hyderabad, India
Q1:—How was hydrogen content estimated for the oxide and metal separately?
Authors’ Response:—This method is described in the reference [5] by A. Her-

mann et al. Briefly speaking the hydrogen content of the oxide is released at a lower
temperature.
Q2:—How was this estimation verified (i.e., about the location of measurement
oxide/metal)?
Authors’ Response:—To verify that the hydrogen in the oxide is released at low
temperature, two adjacent samples (of 3 mm rings) were measured, and the oxide
layer of one of the samples was removed by careful machining. In the case of the
sample with the oxide layer removed, the low temperature peak was not present.
This point is well described in the above mentioned reference.
Questions from Philippe Bossis, CEA
Q1:—What is the evidence for variation of hydrogen pick-up fraction along the
axial elevation for Zircaloy-4 cladding in PWR?
Authors’ Response:—According to the data presented, we could conclude that

the pick-up fraction is different in the lowest elevation, and it does not vary that
much in the higher elevations. Of course, as the statistics in this paper are not high
for PWR material, further examinations are necessary to confirm this conclusion.
However, the important message of the paper is that one hydrogen content mea-
surement per rod does not reflect the overall hydrogen content of the cladding.
This is clearly visible on Fig. 8.
Q2:—Are the EPMA results (oxygen quantification) the same if normalized to 100%?
Authors’ Response:—The EPMA line scans have shown that the slope of oxygen
profile is steeper towards the metal-oxide interface, in the case of Zr2.5%Nb,
because, the oxide is more sub-stoichiometric in this material, in comparison to the
low-tin Zircaloy-4. The observations from several measurements confirm this
statement.
Question from Hans-Olof Andrén, Chalmers University of Tech:
Q1:—Our present understanding of the oxide is that it is nanocrystalline, 20-30

nm grain width, some equiaxed, and most grains columnar. These oxide grains
have many different orientations, so if you make a dark field TEM micrograph
using a diffracted beam from zirconium hydride, you will most certainly also have a
number of oxide grains in contrast simultaneously. Therefore, any structure that
you might observe in the oxide must be considered as incidental unless there are
other observations that support the observation.
Authors’ Response:—We have considered this comment in the text. One point
that should be added is that such features are not present in all materials and under
all oxidation conditions, etc.
Question from David Schrire, Vattenfall, Sweden:
Q1:—What was the highest delta-H pick-up fraction for the BWR cladding,
e.g., LK2 from cycle 6 to cycle 7, or LK3 from cycle 7 to cycle 8 to cycle 9?
Authors’ Response:—Without going to mathematical calculations, if we consider

only the difference between the values of pick-up fraction measured between two differ-
ent rods at two different burnups, as it can be seen from Fig. 6, the highest difference for
LK3/L is seen from 5 cycles to 6 cycles where the pickup fraction increases from about
15% to 35%. For LK2/L in the present results, the largest change is a decrease that occurs
between 6 and 7 cycles. However, it must be noted that such a statement cannot be taken
as a characteristic of the cladding grade. For that information, a much larger number of
measurements is needed.
Question from B. K. Shah, BARC Mumbai:
Q1:—Can you comment on the role of hydrogen uptake on nodular corrosion

of BWR clad?
Authors’ Response:—The topic of nodular corrosion of BWR cladding and its

role on hydrogen uptake has not been examined in this study. In other studies on
inactive material, nodules have been observed on the site of SPPs. Please see as an
example the reference [21].
Question from Mirco Grosse, Karlsruhe Inst. Of Technology:
Q1:—What role do external parameters like local erosions, scratches, or plugs

play on the hydrogen uptake?
Authors’ Response:—As it is shown in the paper, in the section about in situ neutron
radiography, as soon as the metal is exposed to the outside environment, in the absence of
the oxide layer, the uptake is very rapid. Local erosions and scratches, etc. may remove the
oxide layer or reduce the thickness of the oxide, and this will increase the rate of uptake.
Therefore, in the regions of the sample where friction is present, the removal of oxide layer
is possible and the increase in H uptake is to be expected. See also reference: M. Grosse, S.
Valance, J. Stuckert, M. Steinbrueck, M. Walter, A. Kaestner, S. Hartmann and J. Santiste-
ban: Materials Research Society Symposium Proceedings (2013), P. 1528.
Question from K. Somasekhar Reddy, NFC Hyderabad:
Q1:—Discuss experimental details of neutron radiography to measure qualita-

tively H-pickup.
Authors’ Response:—The calibration of the neutron radiography data is not dis-

cussed in this paper. A comprehensive reference for the calibration of neutron radi-
ography for qualitative hydrogen content is provided in: Grosse, M., Kuehne, G.,
Steinbrueck, M., Lehmann, E., Stuckert, J., Vontobel, P.; Journal of Physics Con-
densed Matter 20 (2008), p. 104263.
To give a brief description of the procedure, 20 mm long cladding tube seg-

ments have been annealed in flowing Ar/H2 at temperatures ranging from 800 C to
1200 C at different hydrogen partial pressures and subsequently cooled in inert
atmosphere. According to Sieverts’ law various hydrogen concentrations in the
samples were produced. The hydrogen concentrations were determined by meas-
uring the mass gains. The treatments were controlled by mass spectrometer analysis
of the off-gas. It proves that the furnace atmospheres were clean and the mass gains
are not influenced by oxygen uptake. A linear dependence of the total neutron cross
section determined from the neutron transmission on the H/Zr atomic ratio was
found. This linear dependence is predicted by the theory.
Question from Susan Ortner, National Nuclear Laboratory, U.K.:
Q1:—You showed a TEM image of (what was probably) a hydride crossing

from metal to oxide. This suggests that the hydride was present when oxidation of
the underlying metal occurred, i.e., at the oxidation temperature. What was the
bulk H content in the metal of that sample?
Authors’ Response:—This material had 7 cycles of irradiation. The feature is a

hydride, at the interface and as it can be seen, it is oxidized, in the oxide side of the
interface. The hydrogen content of that segment has not been measured. To have
an idea about the level of hydrogen that might be expected in this segment, the clos-
est segment analyzed for hydrogen content would be the sample at 2000 mm eleva-
tion. The hydrogen content in the metal of the 2000 mm segment for this particular
rod (72-E4, 7 cycles) is 290 ppm (see Fig. 2 and Fig. 5).
IN REACTOR PERFORMANCE
J. M. Garcı́a-Infanta,1 M. Aulló,1 D. Schrire,2 F. Culebras,3

and A. M. Garde4
Oxidation and Hydrogen Uptake

of ZIRLO Structural Components
Irradiated to High Burn-Up
Reference
Garcı́a-Infanta, J. M., Aulló, M., Schrire, D., Culebras, F., and Garde, A. M., “Oxidation and
Hydrogen Uptake of ZIRLO Structural Components Irradiated to High Burn-Up,” Zirconium in
the Nuclear Industry: 17th International Symposium, STP 1543, Robert Comstock and Pierre
ABSTRACT
Good structural performance of the fuel assembly during irradiation is an
indispensable requirement. Extension of licensed burnups demands continuous
improvements, and more precisely on the design and processing of components
made of zirconium alloys. Experience feedback on the assembly behaviour is
necessary and continuous surveillance of the assemblies’ performance is maintained
through on site inspections and post irradiation examinations (PIE). For that purpose,
two research programs have recently been performed which included PIE on selected
pressurised water reactor (PWR) assembly components made of ZIRLO. In the first
program, a 15 by 15 fuel assembly irradiated for four annual cycles in Ringhals 2 NPP
was selected for PIE. Samples extracted from grid strap vanes, guide thimble, and
guide thimble end to top nozzle joints were subjected to visual examinations and
characterizations such as oxide layer thickness, orientation, and distribution of hydride
precipitates and hydrogen content. In the second program, as extension of the
irradiated material evaluation of 17 by 17 lead test assemblies (LTA) irradiated in
Manuscript received December 3, 2012; accepted for publication April 18, 2014; published online September
19, 2014.
1
ENUSA Industrias Avanzadas C/ Santiago Rusiñol 12, 28040 Madrid, Spain.
2
Vattenfall Nuclear Fuel AB, SE-16287 Stockholm, Sweden.
3
Associació Nuclear Ascó-Vandellòs II Edificio Sede L’Hospitalet de l’Infant, 43890 Tarragona, Spain.
4
Westinghouse Electric Company, Columbia, SC 29061.
5
Hyderabad, India.
Copyright V
Vandellós II NPP, outer grid strap vanes were removed from a normal operation three-
cycle assembly and from a four-cycle LTA and sent to the hot cell laboratory for
destructive examinations. One objective of this work was to analyse the behaviour
of skeleton key parts at the end of their irradiation life. Special attention was paid
to the performance of the guide thimble end to top nozzle joint. Another objective
was to study the effect of an additional irradiation cycle on the oxide thickness,
hydride precipitates distribution, hydrogen concentration, and hydrogen pickup
fraction of ZIRLO grids. Furthermore, an analysis of the oxidation and hydrogen
uptake contribution on ZIRLO grids growth was performed. The hot cell
examination results are presented and evaluated in the paper.
Keywords
ZIRLO, oxidation, guide tube, grid, hydrogen induced growth, irradiation
growth, grid growth
Introduction
The structure of a pressurised water reactor (PWR) fuel assembly is its skeleton,
composed of top and bottom nozzles, guide thimbles and grids. Oxidation, hydro-
gen uptake and irradiation of the skeleton zirconium components affect the dimen-
sions and structural behaviour of the entire skeleton. Zirconium skeleton
components are processed differently and the final microstructures vary depending
on the processing parameters. For instance, the final rolling microstructure and
crystallographic texture in grid straps is somewhat different to those obtained in the
guide thimbles during the pilgering process [1]. Furthermore, inner and outer grid
straps are assembled by welds, in which and in the heat affected zones (HAZ) the
microstructure and texture differ to those of the grid straps’ base metal. Therefore,
characterization and analysis of the behaviour of different skeleton components is
necessary to understand and to predict the performance of the skeleton during
operation. Moreover, extension of licensed end-of-life exposures requires changes
in the fuel designs to improve the structural behaviour. For design purposes, experi-
ence feedback on the assembly behaviour is necessary and continuous surveillance
is maintained through on-site inspections and post-irradiation examinations (PIE).
For that purpose, two research programs have been completed which included
PIE on site and in hot cells laboratories of selected skeleton components. The
inspections included oxide layer thickness and hydrogen content measurements,
metallographic studies of the hydride precipitates distribution and grid growth
measurements. In this report, results obtained in the two PIEs are analysed.
In the first program, ZIRLO6 skeleton parts from a 15 by 15 PWR fuel assembly
irradiated for four annual exposure cycles at Ringhals 2 plant were selected for PIE
6
ZIRLO is a trademark or registered trademark of Westinghouse Electric Company LLC, its affiliates and/or
its subsidiaries in the United States of America and may be registered in other countries throughout the
world. All rights reserved. Unauthorized use is strictly prohibited. Other names may be trademarks of their
respective owners. ZIRLO nominal chemical composition: 1 wt. % Nb, 1 wt. % Sn, 0.1 wt. % Fe, and balance Zr.
GARCÍA-INFANTA ET AL., DOI 10.1520/STP154320120191 579
in hot cell laboratories [2]. The assembly was irradiated to usual end-of-life expo-
sures. Special attention was paid to the performance of the guide thimble end to top
nozzle joint. In the second program, ZIRLO outer grid strap vanes were removed
from a three eighteen-month exposure cycles 17 by 17 PWR fuel assembly and
from a four eighteen-month exposure cycles high burnup lead test assembly (LTA)
irradiated in Vandellós II NPP and sent for PIE in hot cell laboratories [3].
One objective of the present work was the analysis of the behaviour of key parts
of the skeleton as guide tubes and grids after normal licensed burnups. Another
objective was to study the effect of the LTA additional irradiation cycle to extended
burnups on the grid oxide thickness, hydride precipitates distribution, hydrogen
concentration and hydrogen pickup fraction (HPF).
GRID GROWTH
Permanent strain of zirconium alloys during operation inside a nuclear reactor is
usually taken as the result of various additive not necessarily independent contribu-
tions [1], which are summarized in the following equation:
(1) e ¼ ecr þ eirr þ eH ¼ ecr-th þ ecr-irr þ ecr-ox þ eirr þ eH
The total permanent strain is designated as e, which has three contributions: creep
strain (ecr), irradiation growth (eirr), and strain caused by volume expansion due to
zirconium hydrides precipitation (eH) [1–4]. Creep strain is usually understood as
the result of thermal creep (ecr-th) and irradiation creep (ecr-irr) contributions. More-
over, contribution of the stresses generated in the metal by the oxide layer was iden-
tified to have a strong effect on the creep strain (ecr-ox) [5,6].
In the present work, an attempt to quantify the different contributions to grid
growth on the LTA is discussed.
RINGHALS 2
A 15 by 15 fuel assembly with ZIRLO skeleton (guide tubes and intermediate struc-
tural grids), identified herein as assembly A, operated during 4 twelve-month irradi-
ation cycles achieving an end of life fast fluence of 10.3 1025 n/m2 (E > 1 MeV).
See Table 1 for design and operation data. Pieces from different skeleton locations
were cut and sent to Studsvik Nuclear AB hot cell laboratories. The different sam-
ples selected for PIE are indicated in the skeleton schematics shown in Fig. 1. ZIRLO
guide tubes are attached to the top nozzle by a bulge joint between the guide tube
and external stainless steel sleeve. The ZIRLO guide tube is attached to the grid by a
bulge joint to a ZIRLO sleeve that is welded to the grid. One piece containing a
bulge joint between ZIRLO guide tube and the external stainless steel sleeve was
sectioned as illustrated in Fig. 1(g) a) cross section (T2) between bulges: The result-
ing piece between T2 and the right extreme of the piece was embedded in epoxy
resin with T2 surface as the observation plane; b) cross section (T1) across one
TABLE 1 Fuel assembly design and operation data.
Fluence
Assembly Structure Irradiation ( 1025 n/m2)
Id Plant Design Material Cycles (E > 1 MeV)
A Ringhals 2 15 by 15 ZIRLO 4 / annual 10.3

B Vandellós II 17 by 17 RFA ZIRLO 3 / eighteen 10.0
months
C Vandellós II 17 by 17 RFA ZIRLO 4 / eighteen 13.6
(LTA) months
bulge: the resulting piece between T1 and T2 was embedded in epoxy resin with T1
surface as the observation plane, and c) longitudinal section (L1): along the rest of
the piece. One of the halves was embedded in epoxy resin with L1 surface as the ob-
servation plane. Detailed information about the samples and the performed inspec-
tions is given in Table 2.
Metallography was performed to investigate the oxide layer integrity and thick-
ness. The samples were embedded in epoxy resin and put under vacuum for half an
hour to minimize the number of pores in the metallographic mount. The sample was
then left to harden for 10–16 h. Sample preparations were performed with Struers
grinding and polishing equipment. The first step of the preparation includes grinding
using silicon carbide papers with decreasing particle size 500#, 1200#, polishing with
diamond paste up to a particle size of 1 lm and a final polishing with a 0.04 lm SiO2
suspension. A Leica MeF4, a fully remote controlled inverted light optical microscope
was used for conventional light optical microscopy (LOM) imaging.
The integrity and thickness of the oxide layer on the ZIRLO components were
examined on unetched metallographic samples using LOM. Several images were
taken on the two sides of the analyzed pieces, i.e., inner and outer diameters of
guide tube and the two sides of grid straps. The oxide layer thickness was measured
on the obtained pictures using image analysis.
Distributions of zirconium hydride precipitates in the microstructure of the
ZIRLO samples were characterized on etched samples using LOM and backscatter
scanning electron microscopy (BSE–SEM) on as-polished samples.
Hydrogen concentration was measured using an inert gas fusion (IGF) instru-
ment ELTRA OH-900 in some samples and image analysis on as-polished BSE–SEM
images in the rest of samples. Hydride concentrations measured by BSE–SEM are
known to be in good agreement with hydrogen concentrations measured directly
by hot vacuum extraction (HVE) over a wide range of hydrogen concentrations in
unirradiated and irradiated zirconium alloys [7]. In the present work, hydrogen
measured by the IGF technique included hydrogen contained in both the metal and
oxide layer. In order to compare IGF and SEM measurements, IGF measurements
were corrected for hydrogen present in the oxide layer [H]ox. More details on IGF
data correction by [H]ox are in HPF Calculation section.
FIG. 1 Scheme of assembly A showing the skeleton pieces selected for PIE.
GARCÍA-INFANTA ET AL., DOI 10.1520/STP154320120191

581
TABLE 2 Assembly A samples source and performed inspections.
Microscopy (Oxide
Thickness Measurement, Hydrogen
Hydride Distribution and Content
Piece Samples Orientation) Measurement
Guide tube a) Guide tube at three different H H

regions: grid 6 elevation, þ20 mm
above and 30 mm below grid 6.
b) Guide tube section between IFM3
and grid 6 (Fig. 1(e)). Longitudinal
section of the guide tube.
Bulge joint Two cross sections (T1 and T2) and H H
between the one longitudinal section (L1)
guide tube and (Fig. 1(g))
top nozzle stain-
less steel sleeve
(Fig. 1(b))
Bulge joint Longitudinal section through the H —
between the bulges
guide tube and
grid 6 ZIRLO
sleeve (Fig. 1(d))
Grid 6 (Fig. 1(c)) Extracted from outer and inner straps, H H (for outer strap
inner strap dimples, springs, welds simples only)
and HAZ
Grid 2 (Fig. 1(f)) Samples extracted from outer straps H H
VANDELLÓS II
Two assemblies irradiated in Vandellós II were selected for this study. The first
assembly, identified herein as assembly B, was irradiated during three eighteen-
month cycles to a fluence of 10.0 1025 n/m2 (E > 1 MeV). The second assembly,
identified herein as assembly C, is one of the High Burnup LTAs [3], which were irra-
diated up to four eighteen-month cycles to a fast fluence of 13.6 1025 n/m2 (E > 1
MeV), above the licensed limit for structural components (11.0 1025 n/m2 (E > 1
MeV). See Table 1 for design and operation data.
The skeleton material (guide tubes and intermediate structural grids) in both
assemblies was ZIRLO. Selected vanes were removed from outer straps from the
upper structural ZIRLO intermediate grids (i.e., grid 7 from assembly B and grid 5,
6, and 7 from assembly C) for oxide and hydrogen uptake analysis. Detail of the
outer strap showing a scheme of the grid vanes along with a representative image of
one removed grid vane are also shown in Fig. 2. A total of three vanes were removed
from each grid: one was used for metallography, the second one for hydrogen con-
tent measurement and the third one was held as a reserve specimen for additional
evaluation, if necessary. All grid vanes were shipped to the hot cell facility at the
FIG. 2 Grid vanes. (a) Outer strap scheme showing the location of the grid vanes and
(b) removed grid vane representative image showing the section plane for
metallographic preparation.
Westinghouse Electric Company Churchill Site. Metallographic mounts were pre-

pared to examine the cross-section of the vane for oxide thickness and hydride dis-
tribution and orientation. The vanes were mounted on edge, ground to the
centreline and polished. The location of the polished cross-section is illustrated in
Fig. 2. Micrographs of the oxide were taken at numerous locations on both sides of
the strap. Care was taken to include only micrographs that provided a valid repre-
sentation of the total oxide thickness. Areas with large gaps between the oxide and
mounting material or where the oxide was severely cracked were not included. After
taking microphotographs for measurement of oxide thickness, the metallographic
mounts were chemically etched to reveal the distribution and orientation of hydride
precipitates.
Hydrogen concentration was measured on one of the removed grid vanes using
an IGF based hydrogen analyzer LECO RH-404. Each grid vane was cut in half to
provide two specimens for analysis. IGF data have been corrected by subtracting
the hydrogen content present in the oxide layer [H]ox using the method explained
in the Hydrogen Pickup Fraction (HPF) Calculation section. HPF calculation was
performed as described in that section using Eq 2.
Assembly C grids 5, 6, and 7 widths were measured in Vandellós II storage
pool using a LVDT based system called SICOM–DIM. This equipment uses twelve
linear variable differential transformers (LVDT), three on each face of the grids.
The LVDTs are placed on a support plate that is displaced along the fuel assembly
stopping on all the fuel assembly grids. The assembly is seated on the equipment base
and secured by the crane with negligible force. More details about SICOM–DIM may
be found in Ref. [3]. Grid growth (percentage) is calculated against a characterized
value of non-irradiated grid width.
HPF CALCULATION
HPF is the fraction of hydrogen absorbed by ZIRLO components by the total
amount of hydrogen generated in the oxidation reaction. HPF was calculated from
hydrogen content measurements according to Eq 2:
½HZr ½H0
(2) HPFð%Þ ¼ 100
½HT
where:
[H]Zr ¼ the hydrogen content in the metal,
[H]0 ¼ the initial hydrogen content in the metal (usually considered as
[H]0 ¼ 20 ppm), and
[H]T ¼ the total hydrogen content generated in the oxidation reaction.
In the case of IGF measurements, [H]Zr is obtained by subtracting the amount
of hydrogen contained in the oxide layer to the measured hydrogen content.
According to Kammenzind et al. [8], [H]ox for irradiated Zircaloy-4 is higher than
the hydrogen content in the metal, ranging from 500 to 2200 ppm with no simple
dependency on oxide thickness, fluence, or flux. In the present work, assuming sim-
ilar hydrogen contents in the oxide layer for ZIRLO alloy, an intermediate value of
[H]ox ¼ 1200 ppm based on the results shown in Ref. [8] has been subtracted to IGF
measurements, using the following expression based in Ref. [8]:
½HZr tm qm þ ½Hox tox qox
(3) ½H ¼
tm qm þ tox qox
where:
[H] ¼ as-measured hydrogen concentration,
tm ¼ the base metal thickness,
tox ¼ the total oxide layer thickness (two sides),
qm ¼ the density of ZIRLO (6.545 g/cm3), and
qox ¼ the density of the oxide layer (5.65 g/cm3).
Results
RINGHALS 2
Oxidation
The average oxide thickness values for each sample are given in Table 3. In the fol-
lowing paragraphs, the oxide thickness results are discussed for each skeleton piece
studied in the present work. Table 3 contains the oxide thickness average values (cal-
culated from 8 to 28 measurements depending on the sample) of the guide tube at
various locations around grid 6. Taking into account the standard deviations, there
are no significant differences in oxide thickness in the analysed axial elevation inter-
val of about 200 mm, being a representative value 20 lm.
Special evaluation was conducted for the guide tube end piece. The goal here
was to study the possible effect of the room temperature plastic strain imparted
TABLE 3 Assembly A oxide thickness, hydrogen concentration and hydrogen pick-up fraction
results.
Average Oxide Std.

Skeleton Thicknessb Deviation [H]Zrc HPFc
Piece Samplea (lm) (lm) (ppm) (%)
Guide tube Grid 6þ20 mm (L) 22.5 2.1 571 (IGF) 19

Grid 6 elevation (L) 18.5 5.7 — —
Grid 6–30 mm (L) 22.5 1.4 486 (IGF) 16
Between IFM3 and grid 19.6 1.5 559 (BSE) 21
6 (L)
Bulge joint T1 10.5 1.7 150 (BSE) 11
between the T2 9.6 1.4 156 (BSE) 13
guide tube L1 9.6 2.0 176 (BSE) 14
and top noz-
zle stainless
steel sleeve
Bulge joint Guide tube 14.8 1.9 — —
between the ZIRLO Sleeve 24.5 1.5 — —
guide tube
and grid 6
ZIRLO sleeve
Grid 6 Outer strap 31.3 1.3 414 (IGF) 19
Inner strap 30.3 4.2 — —
Inner strap spring 35.5 3.5 — —
Inner strap dimple 34.5 0.7 — —
Weld zones / HAZ 49.8 6.4 — —
Grid 2 Face 1 (Outer strap) 11.5 3.0 87 (IGF) 11
Face 3 (Outer strap) 10.8 2.1 91 (IGF) 12
a
T ¼ transversal; L¼ Longitudinal; HAZ¼ heat affected zone. L1, T1 and T2 sections schematics in
Fig. 1(g).
b
Taking into account the standard deviations, in samples where oxide thickness was measured in
two sides, there are no significant differences between oxide thicknesses measured at the inner and
outer diameters of the guide tube and between both sides of the grid strap samples.
c
[H]Zr values are differenced depending upon they were obtained by IGF or image analysis from
BSE-SEM images. The amount of hydrogen present in the oxide was subtracted from [H] measure-
ments obtained by IGF. Therefore, HPF calculation only takes into account hydrogen absorption by
the Zr metal in all cases.
during manufacturing bulge expansion process on the oxidation behaviour of the

bulges by comparison with non-deformed material. The oxide layer thickness was
measured at several locations in two cross sections (T1 and T2 in Fig. 1(g)) and one
axial section (L1 in Fig. 1(g)). The results are represented in Fig. 3. For the transverse
sections, the average oxide thickness (calculated from 3 to 6 measurements depend-
ing on the sample) around the cross sections of the guide tube end piece have been
represented with standard deviation bars against azimuthal position in Fig. 3(a)
(T1) and Fig. 3(b) (T2). For the longitudinal section, the average oxide thickness (of
586
FIG. 3 Assembly A bulge joint sections. The axial positions where oxide layer was measured is indicated in the macroscopic images. At the right side of
each image there is a representation of the oxide thickness against the measurement position. (a) Guide tube to top nozzle joint section T1; (b)
guide tube to top nozzle joint section T2; (c) guide tube to top nozzle bulge joint L1; (d) guide tube to grid 6 sleeve bulge joint axial section.
6–12 measurements) along the axial section of the guide tube end piece (L1) has
been represented with standard deviation bars against the axial position in Fig. 3(c).
Based on the results and the standard deviations, a dependence of the oxide thick-
ness on the azimuthal position is not observed. In the axial section (Fig. 3(c)), clear
differences between the bulged regions (positions 1 and 2) and the undeformed
guide tube (positions 3 and 4) are not observed. The slightly higher oxide thickness
observed in Fig. 3(c) graph at positions 1 and 2 could be interpreted as acceleration
of the oxidation rates on the bulges. However, based on the standard deviation of
the results, the differences between the undeformed guide tube and the bulge
regions are considered as being not significant. Moreover, oxide thickness values
measured in section T1 (deformed section) are not considered significantly higher
than the undeformed guide tube oxide thickness values. In summary, the oxide
thickness measured in the guide tube end piece is low (10 lm) in comparison
with the values obtained in the guide tube at the grid 6 axial elevation (20 lm).
The oxide thickness was characterized at one bulge joint between a guide tube
and its corresponding ZIRLO sleeve welded at grid 6 elevation. The analysed sam-
ple was cut axially and one half was mounted in epoxy as shown in Fig. 3(d). The
ZIRLO guide tube and sleeve oxide thicknesses were measured at the right and left
sides of the mounted sample. The results are given graphically in Fig. 3(d), where
one-side average values on each location for the guide tube and the sleeve are given.
The measurements were taken in four different axial locations of the sample, two of
them on areas affected by plastic deformation of the bulge, and other two unde-
formed locations of the guide tube. The sleeve oxide thickness (25 lm) was signif-
icantly higher than the result obtained in the guide tube at this elevation (15 lm).
It is important to note that the oxide on the inner diameter of the guide tube
detached during sample preparation and only the oxide on the outer diameter of
the guide tube was measured.
Grid 6 locations selected for oxide thickness characterisation were outer strap,
inner strap, inner strap spring, and dimple and inner strap welds (by laser welding
at the intersection of two inner grid straps) and HAZ. These samples have not
undergone exactly the same process during manufacturing. For instance, inner and
outer straps have different thicknesses after rolling, inner strap springs and dimples
have undergone extra cold plastic deformation during stamping process, respec-
tively, with respect to the base rolling plate, or even weld zones and HAZ have dif-
ferent microstructural state than the base strap metal. The goal was to compare the
oxide thicknesses of these samples oxidation and establish possible different oxida-
tion performance. According to the grid 6 results shown in Table 3, the inner strap
spring and dimple oxide thicknesses (35 lm) are slightly higher than those
obtained on the inner strap sample (30 lm). This difference, however, is consid-
ered to be too small for a stamping deformation effect. In weld zones and HAZ, the
average thickness is 50 lm, which is higher than the inner strap base metal aver-
age thickness. Weld zones and HAZ have different microstructure than the base
metal and the oxidation performance is not optimized due to higher temperatures
attained during welding process. In grid 2, measurements were taken in two outer
strap locations, and the obtained results (Table 3) are lower than those obtained in
grid 6, being 11 lm. Lower oxidation in grid 2 straps was expected because it is
placed at a lower axial elevation than grid 6 and so grid 2 operated at a lower cool-
ant temperature.
Hydrogen Uptake
Average [H]Zr and HPF values obtained from the measurements are given in Table 3.
Hydrogen concentration in grid 2 and in guide tube at top nozzle bulge joint eleva-
tion is lower than 200 ppm. A maximum [H]Zr equal to 571 ppm was measured in
the guide tube at grid 6 elevation þ20 mm. A maximum HPF equal to 21 % was
measured in the guide tube piece between intermediate flow mixing (IFM) grid 3
and grid 6.
Figures 4–6 show representative images of the hydride precipitates distributions
in the studied samples. As it is indicated on each case, some of the images were
FIG. 4 Zirconium hydride precipitates distribution (LOM images) in (a) assembly A grid
2 ([H]Zr ¼ 91 ppm) and (b) assembly A grid 6 ([H]Zr ¼ 414 ppm).
FIG. 5 Zirconium hydride precipitate distribution in assembly A guide tube. (a) Grid
6 þ 20 mm elevation (LOM, [H]Zr ¼ 571 ppm); (b) grid 6 þ 30 mm elevation (LOM,
([H]Zr ¼ 486 ppm), and (c) IFM3-grid 6 elevation (SEM, [H]Zr ¼ 559 ppm).
obtained by LOM on etched surfaces and the rest were obtained by BSE–SEM.
Nevertheless, for qualitative analysis of the hydride precipitates distribution eval-
uated by both techniques are valid [7].
Representative images of the hydride precipitates distribution in the grid 2 and
grid 6 outer straps are given in Fig. 4. Figure 4(a) corresponds to grid 2 and Fig. 4(b)
corresponds to grid 6 and, in both cases, images obtained at the extremes and
centre of the straps section are shown. All the images were obtained in a strap sec-
tion perpendicular to the rolling direction. In both cases, grid 2 and 6, the hydrides
lay mostly parallel to the rolling plane. This is expected and is a consequence of the
strong texture of the sample with most of basal planes almost parallel to the rolling
plane. Comparison of grid 2 microstructure (Fig. 4(a)) with that of grid 6 (Fig. 4(b))
reveals higher amount of precipitates in grid 6, confirming the differences between
[H]Zr results given in Table 3.
Figure 5 shows representative images of the guide tube approximately at grid 6
elevation. Figure 5(a) corresponds to sample at grid 6 elevation þ20 mm; Fig. 5(b)
corresponds to grid 6 elevation—30 mm and Fig. 5(c) corresponds to guide tube
FIG. 6 Zirconium hydride precipitates distribution (longitudinal section, SEM) in

assembly A guide tube at top nozzle bulge joint. (a) between two bulges; (b) in
one bulge and (c) far from the bulges.
section between IFM3 and grid 6. Figures 5(a) and 5(b) were obtained in the cross
section plane of the guide tube, and Fig. 5(c) in the longitudinal section plane. In all
cases, the hydrides lay mostly parallel to the tube surface. As it is observed in grid
strap samples (Fig. 4), this is a consequence of the strong texture of the tube, with
most of basal planes almost parallel to the tube surface. The three images show an
amount of hydrides as expected based on the results given in Table 3.
Representative images of the hydride precipitates distribution in a longitudinal
section of the guide tube at the top nozzle bulge joint are given in Fig. 6. Figure 6(a)
corresponds to the hydride precipitates distribution between two bulges; Fig. 6(b)
corresponds to the hydride precipitates distribution in one bulge, and Fig. 6(c) cor-
responds to the hydride precipitates distribution far from the bulges. In locations
far from the deformed areas of the guide tube (Fig. 6(c)), the hydrides are oriented
mainly parallel to the tube walls. However, in locations closer to the deformed
regions, radial hydrides (Fig. 6(a)) or even hydrides laying at 45 with the tube walls
(Fig. 6(b)) are observed. These hydride distributions are probably related to the
residual stresses at the bulges. The bulges are cold worked after the final heat treat-
ment of the guide tubes. Therefore, when the assemblies are introduced in the reac-
tor, the bulges are in different stress state to that of the rest of the tube. It is well
known that zirconium hydride precipitation is sensitive to the stresses in zirconium
alloys [9]. The small amounts of radial and 45 hydrides observed in the bulges can
be related with this phenomenon. Nevertheless, due to the low concentration of the
hydrides, a mechanical effect on the bulge joint behavior is unlikely. In fact, no per-
formance issues during operation or handling have been reported to date due to
guide tube bulges.
VANDELLÓS II
Oxidation
The oxide thickness average values with standard deviations (calculated from of
9–14 measurements) on each grid are given in Table 4. Comparing the grid 7 results
of the two assemblies, assembly C shows higher oxide thickness (51.4 lm) than
assembly B (34.3 lm), as expected since assembly C operated eighteen months more
than assembly B. On the other hand, comparing the results obtained from assembly
C grids 5, 6, and 7, the higher grid elevation, the higher the oxide thickness, which
can be attributed to the increase of coolant temperature with increasing elevation.
Hydrogen Uptake
The hydrogen concentration measurements and average value for each grid are
given in Table 4. Comparing the grid 7 results of the two assemblies, assembly C
shows higher [H]Zr (877 ppm) than assembly B (442 ppm), as it is expected, since
assembly C operated one more cycle than assembly B. On the other hand,
TABLE 4 Vandellós II grid vanes oxide thickness, [H]Zr, HPF and grid growth measurements.
Average One-Sided Std. [H]Zr Grid

Assembly Oxide Thicknessa Deviation Averageb HPFb Growth
Id. Grid (lm) (lm) (ppm) (%) (%)
C 7 51.4 3.1 877 (IGF) 18,4 0.70

6 42.5 1.7 656 (IGF) 16,9 0.55
5 35.1 2.2 339 (IGF) 10,5 0.37
B 7 34.3 0.8 442 (IGF) 14,2 —
a
Taking into account the standard deviations, there are no significant differences between oxide
thicknesses measured at both sides of the grid vanes samples.
b
All measurements were obtained by IGF technique. The amount of hydrogen present in the oxide
was subtracted from [H] measurements. Therefore, HPF only takes into account hydrogen absorp-
tion by the Zr metal.
comparing the results obtained from assembly C grids 5, 6, and 7, the higher grid
elevation, the higher is [H]Zr, which can be attributed to the effect of the coolant
temperature on the oxidation rate.
Representative images of the hydride precipitate distribution for each grid are
shown in Fig. 7. In all micrographs, the white areas represent the zirconium base
metal, while the black strings, which were revealed by the chemical etching, repre-
sent the hydride precipitates. In all cases, the hydride precipitates are mainly dis-
tributed parallel to the rolling plane. Figures 7(a), 7(b), and 7(c) represent the hydride
distribution in assembly C grids 5, 6, and 7, respectively. An increasing amount of
hydrides with grid elevation is observed on these three images, which is consistent
with the results given in Table 4.
Grid Growth
Grid growth (%) calculated against the unirradiated envelope value is shown in
Table 4 for assembly C grids 5, 6, and 7. Based on these results, the highest grid
growth (0.70 %) is measured in the top ZIRLO grid (grid 7).
FIG. 7 Hydride precipitate distribution in the cross-section plane of the vane removed
from (a) assembly C grid 5 ([H]Zr ¼ 339 ppm); (b) assembly C grid 6
([H]Zr ¼ 656 ppm); (c) assembly C grid 7 ([H]Zr ¼ 877 ppm); and (d) assembly B
grid 7 ([H]Zr ¼ 442 ppm).
Discussion
OXIDATION
The oxide thickness results are compared in Fig. 8 to previous data from other post
irradiation examination performed on a ZIRLO structure assembly subjected to
8.8 1025 n/m2 (E > 1 MeV) in VC Summer [1,4]. In the graph in Fig. 8, oxide
thickness results are represented against the axial elevation of the samples. All the
results obtained in assembly A and B are displayed. Assembly C results have not
been included in Fig. 8 because this assembly has significantly higher end of life fast
fluences and therefore, oxide thickness values are not comparable to results from
previous experiments and assembly A and B results at lower burnups. In all cases,
oxide layer is thicker with increasing axial elevation, which is related to the increase
of the coolant and structure temperature. All the results are consistent with data
from previous PIEs.
Regarding the origin of the analysed samples, grid straps show thicker oxide
layers than guide tubes at the same elevation (grid 6) and, in particular, the grid
welds show the highest values of oxide thickness. As it has been already com-
mented, oxide thicknesses on grid 6 inner strap dimples and springs are slightly
higher (5 lm higher) than on the inner strap samples (base metal). Since straps
dimples and springs underwent cold plastic deformation during stamping opera-
tion, different oxidation rates on them could be expected. However, taking into
account the variability of the measurements and of the previous database at these
FIG. 8 Comparison of assemblies A and B oxide thickness and data from previous PIEs
[1,4].
elevations, different oxidation performance of the dimples and springs compared

with the base inner strap cannot be definitively inferred.
Oxidation resistance of zirconium alloys is strongly related to the stoichiome-
try, size, and distribution of the second phase precipitates (SPP), which are directly
related to the thermalmechanical history of the component. Manufacturing heat
treatments parameters are adjusted to achieve optimized SPP size and distribution
for oxidation resistance. According to Sabol et al. [10], ZIRLO shows two principal
types of precipitates after the final annealing: b–Nb particles and Zr–Nb–Fe precipi-
tates, most likely Laves phase Zr(NbFe)2, based on Toffolon et al. [11] and Shishov
et al. [12,13]. b–Nb particles are almost stable during irradiation; however,
Zr(NbFe)2 particles decompose into b–Nb and the Fe content in the alpha Zr
increases around the former Zr–Nb–Fe precipitate locations. ZIRLO components
despite having the same chemical composition, undergo different manufacturing
processes, which could give place to microstructure differences. In addition to the
possible differences in the SPP size distribution in the starting as-fabricated micro-
structures, the radiation induced evolution of SPP size distribution and SPP compo-
sition (specifically the extent of Fe migration from Laves phase Zr(NbFe)2 SPP to
the surrounding matrix) are also expected to be different due to the differences in
irradiation temperature and neutron fluence. A recent paper on Alloy E6357 (alloy
with composition similar to ZIRLO) components [13], confirms that the Laves
phase particles in Alloy E635 decompose with neutron damage for VVER coolant
chemistry conditions. The extent of decomposition depend on several variables,
local coolant temperature, level of neutron fluence, alloy composition, burnup etc.
It is believed that these Alloy E635 results are applicable to ZIRLO irradiated in
western PWRs with coordinated boron-lithium water chemistry. These possible
anticipated differences in SPP size/composition evolution in different ZIRLO com-
ponents could imply different oxidation resistances. The best example observed in
the present study is grid welds and HAZ, which are performed by laser welding
after straps final heat treatment. Laser welding implies local material fusion and
rapid solidification through the two-phase region. After welding process, no addi-
tional heat treatment is performed to the grids. As a consequence, transformed
b–Zr microstructures with various levels of partitioning of alloying elements and
extent of SPP precipitation and non optimized SPP distributions in the welds and
HAZ are formed which contribute to lower oxidation resistance in comparison to
the straps base metal.
Oxidation of ZIRLO guide tube is significantly lower (10 lm) in the guide
tube end piece, than at grid 6 elevation (20 lm) despite a similar operation tem-
perature. The microstructure in this part of the guide tube is not so affected by irra-
diation as at grid 6 elevation because it is out of the active fuel column. A high
density of irradiation induced c-dislocation was found in similar ZrNbSnFe alloys
7
Alloy E635 chemical composition: 1 wt. % Nb, 1.2 wt. % Sn, 0.35 wt. % Fe and balance Zr.
specially close to the irradiation altered former Zr(NbFe)2 SPP in materials that
exhibit a significant acceleration of irradiation induced growth at high neutron flu-
ences [15]. The extent of Fe migration from the Laves phases to the surrounding
matrix depends on both the irradiation temperature and the extent of microstruc-
tural damage due to neutron damage [15]. Such iron migration from the SPP is
expected to be different for the guide tube at grid 6 and at the top end piece eleva-
tions. This indicates that the irradiation induced dissolution of Zr(NbFe)2 SPP, and
subsequent increase of Fe in solid solution in the Zr matrix, could be one effect that
causes late increased corrosion. In the present work, irradiation induced dissolution
of the ZrNbFe SPP in the guide tube end piece would be minimal in comparison
with the guide tube inside the active fuel column. Therefore, resistance to oxidation
in the guide tube end piece is preserved despite of being operated during the same
time and higher temperature than the rest of the guide tube length.
Regarding assembly A guide tube to grid 6 bulge joint, guide tube oxide thick-
ness at this location resulted in 15 lm, lower than the experience and other results
obtained from this guide tube around this elevation. However, special care is neces-
sary to analyse this result because the oxide on the inner diameter of the guide tube
detached during the sample preparation and only the oxide thickness correspond-
ing to the outer diameter of the guide tube is available. This result should not be
compared to the other guide tube results at grid 6 elevation because the outer diam-
eter surface could have undergone a protective effect against oxidation by the
ZIRLO sleeve. It is possible that the different coolant circulation on this surface
compared with the non-sleeved guide tube locations could have an effect of decreas-
ing the oxidation reaction kinetics. On the other hand, the ZIRLO sleeve shows
oxide thickness results higher than the guide tube at grid 6 elevation. Taking into
account that ZIRLO sleeves final heat treatment (stress relief annealing, SRA), is
different to the final treatment of the guide tube (recrystallization annealing, RXA),
there could be differences in SPP size/composition and evolution during irradiation
as discussed previously. Therefore, it would be reasonable that the oxidation per-
formance of the sleeve was different than that of the guide tube.
Figure 9, where one-side oxide thickness is represented against elevation,
focuses on comparison of assembly A, B, and C grid results. Regarding A lower and
upper ZIRLO grids and B upper ZIRLO grid, the results agree with data from previ-
ous experiments. Concerning assembly C grids, oxide thickness results have been
represented in Fig. 9 in order to estimate possible acceleration of the oxidation reac-
tion during the last cycle. Comparing the oxide thickness results at equal elevation,
assembly C results are 50 % higher than those of B, which operated three cycles in
the same plant. This significant oxide thickness increase during the fourth cycle
indicates that oxidation reaction accelerated during the fourth cycle.
Hydrogen Uptake
Hydrogen concentration results and data from previous PIEs in guide tube and grid
are represented against axial elevation in Figs. 10 and 11, respectively. No figure
FIG. 9 Assemblies A, B, and C grid oxide results along with the previous results [1,4].
representing all the [H]Zr results as Fig. 8 (for oxide thickness) has been included
because [H]Zr depends on the samples wall thickness. Since guide tubes and grid
straps have different as-fabricated wall thicknesses, it is not appropriate to show
both data sets on the same hydrogen level graph. As observed for oxide thickness,
FIG. 10 Assembly A guide tube hydrogen uptake results along with the previous results
[1,4].
FIG. 11 A, B, and C grid hydrogen uptake results along with the previous results [1,4].
[H]Zr in guide tube and grids is higher with increasing elevation, which is an effect
of the coolant and structure temperatures.
Figure 10 focuses on the assembly A guide tube results around grid 6 and top
nozzle bulge joint elevations (see Table 3 for reference). [H]Zr is slightly higher than
the previous results, probably due to higher end of life fast fluence of assembly A,
10.0 1025 n/m2 (E > 1 MeV), compared to the past data, 8.8 to 11 1025 n/m2
(E > 1 MeV). On the other hand, [H]Zr of ZIRLO guide tube is significantly lower
in the guide tube end piece (150 ppm) than that of measured at grid 6 elevation
(570 ppm) despite a similar operation temperature. The reason is the same as for
oxidation, the microstructure in this part of the guide tube is not so affected by irra-
diation as at grid 6 elevation because it is out of the active fuel column.
Figure 11 focuses on the grid results at assembly A lower and upper ZIRLO grids,
assembly B upper ZIRLO grid and assembly C three upper ZIRLO grids (see Table 3
and Table 4 for reference). Regarding assembly A and B lower and upper ZIRLO
grids, [H]Zr results agree with the results from previous PIEs. Concerning assembly
C grids, [H]Zr results are presented in Fig. 11 in order to estimate any possible accel-
eration of hydrogen pick up during the last cycle. Comparing the results at the
highest grid elevation, assembly C (grid 7) results are 80 % higher than A (grid 6)
or B (grid 7) results, which indicates that hydrogen pick up accelerated during the
last cycle. This effect, along with the oxidation reaction acceleration, is well known
and occurs after a certain burnup depending on the alloy, and it is due to micro-
structure degradation by irradiation damage [14]. Acceleration of corrosion rates at
usual commercial PWR fluences is not clear for ZrNbSnFe alloys. Shishov et al. [15]
showed that the iron depletion from Laves phases towards the Zr matrix has an
effect on the onset of irradiation growth acceleration in a E635 type alloy having
less iron (0.15 wt. %) than the nominal composition (0.35 wt. %); however, no cor-
rosion or hydriding results were presented in this study for low iron E635 alloy. It
is important to note that ZIRLO alloy has lower iron content (0.10 wt. %) than
E635 and the observed acceleration of corrosion and hydrogen pick up of ZIRLO
irradiated to high burnup analysed in this study could be associated with a similar
iron transfer from the Laves phase particles towards the Zr matrix.
Hydrogen pick up fraction (HPF) results given in Tables 3 and 4 are compared
in Fig. 12 to ZIRLO structural components HPF database. HPF is represented
against the average temperature of the grids during operation. The database is com-
posed mostly of data obtained from assemblies irradiated to fast fluences
8.8–11.0 1025 n/m2 (E > 1 MeV), however, some data of this database (marked as
white diamonds in Fig. 12) were obtained from the three upper grids of a high
burnup LTA irradiated in a US plant beyond 11.0 1025 n/m2 (E > 1 MeV). The
following is observed in the graph:
• Assembly B grid 7, assembly C grid 5 and assembly A grid 6 HPF values
(14.2 %, 10.5 % and 19.0 %, respectively) lie inside the HPF database; however,
assembly C grids 6 and 7 results are slightly above the higher HPF database
values at the corresponding average temperatures.
• The two upper grids of the US LTA (i.e., white square symbol in Fig. 12) also
showed HPF values higher than most of the HPF data obtained in lower
burnup assemblies with similar temperatures.
• Comparison of the values obtained from assembly B grid 7 (14.2 %) and
assembly C grid 7 (18.4 %) reveals an increase of HPF during the additional
cycle of assembly C.
FIG. 12 HPF against operation temperature.

• Assembly A grid 2 HPF (11 %–12 %) are is higher than the temperature-
equivalent data (7 %). However, comparison of grid 2 with grid 6 (19 %), is
also evidence of the temperature dependence of HPF.
These observations suggest an acceleration of the hydrogen uptake during
assembly C fourth cycle compared to the previous assemblies’ history. Nevertheless,
the maximum HPF value in the highest case (grid 7) keeps below 20 %, which is
considered normal for ZIRLO components under normal operation conditions
according to Fig. 12.
Grid Growth
A non-rigorous attempt has been made to quantify each contribution shown in Eq 1
to Vandellós II assembly C grid 7 growth. The following assumptions are made:
• eirr is approximately equal at all the grids. This assumption is based on the
usual uniform fast fluence versus axial elevation profiles [1].
• For [H]Zr ¼ 0 ppm, no corrosion and hydrogen uptake effects on grid growth
are assumed. Therefore, For [H]Zr ¼ 0 ppm, ecr-ox and eH are considered equal
to zero.
• No structural stresses due to external loads acting along the grid straps are
considered. This assumption cannot be made on guide tubes and fuel cladding,
but stresses along the strap bands due to external loads during operation are
considered negligible. Therefore, as a consequence, ecr-th and ecr-irr are
assumed to be equal to zero.
Estimation of eirr may be inferred from the assembly C grids growth and [H]Zr
data. Growth data has been represented against their corresponding [H]Zr measured
FIG. 13 Assembly C grids 5, 6, and 7 growth versus [H]Zr.

values in Fig. 13. In a first step of the analysis, eirr may be inferred to be approxi-
mately equal to 0.16 % for all grids performing an extrapolation to [H]Zr ¼ 0 ppm
as shown in Fig. 13. This value is higher than the value at 0-ppm for ZIRLO grid
growth given in Ref. [1] for a ZIRLO skeleton subjected to lower fast fluence.
Regarding the hydride contribution eH, according to experimental results
obtained from autoclave tested ZIRLO and Zircaloy 4 components [1], eH 0.20 %
/ 1000 ppm. Therefore, having eH 0.20 % is considered acceptable for assembly C
grid 7, which resulted in [H]Zr ¼ 877 ppm.
Taking into account these estimations, assembly C grid 7 irradiation growth is
eirr ¼ 0.16 %. The remaining strain is distributed to oxidation creep and hydride
growth; the distribution changes with elevation because the HPF changes, see
Fig. 12.
Conclusions
Assembly A, irradiated during four annual cycles in Ringhals 2 to a fast fluence of
10.3 1025 n/m2 (E > 1 MeV); assembly B, irradiated during three eighteen month
cycles to a fast fluence of 10.0 1025 n/m2 (E > 1 MeV) in Vandellós II, and assem-
bly C, one high burnup LTA irradiated up to four eighteen month cycles to a fast
fluence of 13.6 1025 n/m2 (E > 1 MeV) also in Vandellós II, were selected for post
irradiation examination in hot cell laboratories. The analyzed samples were
extracted from the ZIRLO skeleton of the selected assemblies. The examinations
comprised oxide thickness measurements, hydride precipitates metallographies,
hydrogen concentration measurements, hydrogen pickup fraction calculation, and
grid growth measurements. The main conclusions of the study are:
1. Metallographic study reveals that the hydride precipitates distribution is
closely related to the texture of the samples, in particular to the orientation of
the basal planes. The hydride precipitates are located almost parallel to the
surface of the guide tubes and to the rolling plane of the grid straps.
2. For ZIRLO grids or guide tube locations inside the inside the active fuel col-
umn, oxide thickness, hydrogen concentration, and hydrogen pick up fraction
depend on the axial elevation of the piece as an effect of the coolant tempera-
ture and fluence. The higher the axial elevation is, the higher the coolant
temperature and, therefore, the higher the oxide thickness, hydrogen concen-
tration, and hydrogen pick up fraction for identical materials.
3. Oxide thickness, hydrogen concentration, and pickup fraction in assemblies A
and B are in accordance with results from previous PIEs.
4. Results from assembly C show that during the fourth cycle the oxide thickness
is increased by about 50 % and the hydrogen concentration increased by about
80 %. Moreover, assembly C hydrogen pickup fraction is higher than that of
assembly B and of existing results from other assemblies irradiated up to the
licensing burnup. These results indicate an acceleration of oxidation and
hydrogen pickup, attributed to irradiation damage of the microstructure.
5. The analysis of the guide tube end piece shows good performance (assembly
A). Oxidation and hydrogen uptake of the ZIRLO guide thimble end part is
low in comparison with the ZIRLO guide thimble at a lower elevation within
the active fuel column. This is a consequence of less irradiation-induced
microstructure degradation, since it is located out of the active length.
6. Estimation of the different contributions to the growth of assembly C upper
ZIRLO grid reveals that corrosion of the grid straps has a major effect on the
grid growth. Around 80 % of the total grid growth would be due to oxidation
and hydrogen uptake effects.
ACKNOWLEDGMENTS
The writers acknowledge the Spanish utilities Associació Nuclear Ascó-Vandellòs II
(ANAV) and Centrales Nucleares Almaraz-Trillo (CNAT) for financial support of the
projects through the “Coordinated Research Program” (PIC). The efforts of the
numerous engineers, handling operators, technicians and support staff at Vandellós
II, Ringhals 2, Studsvik Nuclear AB, Westinghouse Electric Company Churchill Site
and ENUSA are gratefully acknowledged.
References
[1] King, S. J., Kesterson, R. L., Yueh, K. H., Comstock, R. J., Herwig, W. M., and Ferguson, S. D.,
“Impact of Hydrogen on Dimensional Stability of ZIRLO Fuel Assemblies,” Zirconium in
the Nuclear Industry: Thirteenth International Symposium, ASTM STP 1423, G. D. Moan,
and P. Rudling, Eds., ASTM International, West Conshohocken, PA, 2002, pp. 471–489.
[2] Garcı́a-Infanta, J. M., Canencia, R., Petersson, B., Schrire, D., and Claesson, B., “Post
Irradiation Examination of the Skeleton of a 15x15 PWR Fuel Assembly,” Proceedings
of the 2010 LWR Fuel Performance/TopFuel/WRFPM, Orlando, FL, Sept 26–29, 2010.
[3] Aulló, M., Garcı́a-Infanta, J. M., and Chapin, D., “Post Irradiation Examination of High
Burnup Assemblies in Vandellós II,” Proceedings of the 2010 LWR Fuel Performance/
TopFuel/WRFPM, Orlando, FL, Sept 26–29, 2010.
[4] Kesterson, R. L., King, S. J., and Comstock, R. J., “Impact of Hydrogen on Dimensional
Stability of Fuel Assemblies,” International Topical Meeting on Light Water Reactor Fuel
Performance, American Nuclear Society, April 2000, Park City, UT.
[5] Donaldson, A. T., “Growth in Zircaloy-4 Fuel Clad Arising from Oxidation at Tempera-
tures in the Range 623 to 723 K,” Zirconium in the Nuclear Industry: Ninth International
Symposium, C. M. Eucken, and A. M. Garde, Eds., ASTM International, Philadelphia, PA,
1991, pp. 177–197.
[6] Barberis, P., Rebeyrolle, V., Vermoyal, J. J., Chabretou, V., and Vassault, J. P., “CASTA
DIVA: Experiments and Modeling of Oxide-Induced Deformation in Nuclear
Components,” J. ASTM Int., Vol. 5, No. 5, 2008, p. 101124.
[7] Schrire, D. I. and Pearce, J. H., “Scanning Electron Microscope Techniques for Studying
Zircaloy Corrosion and Hydriding,” Zirconium in the Nuclear Industry: Tenth International
Symposium, ASTM STP 1245, A. M. Garde, and E. R. Bradley, Eds., ASTM International,
Redistribution in Alpha-Annealed Zircaloy-4,” Zirconium in the Nuclear Industry:
Eleventh International Symposium, ASTM STP 1295, E. R. Bradley, and G. P. Sabol, Eds.,
[9] Lloret, M., and Quecedo, M., “Results of Thermal Creep Test on Highly Irradiated ZIRLO,”
Proceedings of the 2008 Water Reactor Fuel Performance Meeting, Seoul, Korea, Oct
19–23, 2008.
[10] Sabol, G. P., Comstock, R. J., Weiner, R. A., Larouere, P., and Stanutz, R. N., “In Reactor
Corrosion Performance of ZIRLO and Zircaloy-4,” Zirconium in the Nuclear Industry:
Tenth International Symposium, ASTM STP 1245, A. M. Garde, and E. R. Bradley, Eds.,
ASTM International, Philadelphia, PA, 1994, pp. 724–744.
[11] Toffolon, C., Brachet, J-C., Servant, C., Legras, L., Charquet, D., Barberis, P., and Mardon,
J.-P., “Experimental Study and Preliminary Thermodynamic Calculations of the Pseudo
Ternary Zr-Nb-Fe-(O,Sn) System,” Zirconium in the Nuclear Industry: Thirteenth Interna-
tional Symposium, ASTM STP 1423, G. D. Moan, and P. Rudling, Eds., ASTM International,
[12] Shishov, V. N., Peregud, A. V., Nikulina, A. V., Pimenov, Y. V., Kobylyansky, G. P., Novoselov,
A. E., Ostrovsky, Z. E., and Obukhov, A.V., “Influence of Structure-Phase State of Nb Con-
taining Zr Alloys on Irradiation Induced Growth,” Zirconium in the Nuclear Industry:
Fourteenth International Symposium, ASTM STP 1467, P. Rudling, and B. Kammenzind,
[13] Shishov, V. N., Peregud, M. M., Yu. Shevyakov, A., Markelov, V. A., Nikulina, A. V., Novikov,
V. V., Volkova, I. N., Novoselov, A. E., Kobylyansky, G. P., and Obukhov, A. V., “Corrosion,
Dimensional Stability and Microstructure of VVER-1000E635 Alloy FA Components at
Burnups Up to 72 MWday/KgU,” Zirconium in the Nuclear Industry: Seventeenth Interna-
tional Symposium, ASTM STP 1543, R. Comstock, and P. Barbieris, Eds., ASTM Interna-
tional, West Conshohocken, PA, 2015, pp. 628–650, STP in progress.
[15] Shishov, V. N., Peregud, M. M., Nikulina, A. V., Shebaldov, P. V., Tselishchev, A. V., Novoselov,
A. E., Kobylyansky, G. P., Ostrovsky, Z. E., and Shamardin, V. K., “Influence of Zirconium
Alloy Chemical Composition on Microstructure Formation and Irradiation Induced
Growth,” Zirconium in the Nuclear Industry: Thirteenth International Symposium, ASTM
STP 1423, ASTM International, West Conshohocken, PA, 2002, pp. 758–779.
DISCUSSION
Question from Antoine Ambard, EDF R&D:—The metallography shows that
hydrides are along the rolling plane in the grid and in the vane. How do you explain
such a distribution: texture?
Authors’ Response:—Hydride orientation in zirconium alloys is strongly related

to texture, in particular to the fraction of basal poles in axial, tangential and radial
directions. This is due to the fact that precipitation habit planes for zirconium
hydrides are closely parallel to the basal plane. Guide tubes and grid straps have
strong textures with most of the basal planes oriented almost parallel to the tube
surfaces (guide tubes) and to the rolling plane (grid straps). Therefore, most of
hydrides exhibit a preferential orientation nearly parallel to the axial direction of
the guide tubes and rolling plane of the grid straps.
Question from Jean-Christophe Brachet, CEA, France:—To estimate the contri-

bution of hydrides to the observed growth of the grid materials, did you take into
account the crystallographic texture influence? Because the hydrides seem to be
strongly oriented in one direction on your micrograph, their contribution should be
significantly anisotropic.
Authors’ Response:—Estimation of impact of hydride volume expansion on

growth is based on results obtained in non-irradiated samples subjected to acceler-
ated autoclave test in lithiated water (Ref. 1). Samples studied in Ref. 1 were
extracted from guide tubes and grid straps having similar processing, heat treat-
ments (RXA) and final texture as irradiated samples studied in the present work.
Ref. 1 samples were hydrogenated up to [H] 1500 ppm, and a linear relationship
between growth and hydrogen concentration of 0.20% strain per 1000 ppm
hydrogen was derived. Assuming that the contribution (on growth) of volume
expansion due to hydride precipitation is insensitive to irradiation damage of the
samples, the authors consider that estimation given in Ref. 1 is reasonable and that,
by similarity of the former and current samples, this estimation already takes into
account possible anisotropy effects due to the texture of the samples.
Question from Mirco Grosse, Karlsruhe Inst. Of Technology:—Do you find some
reduction of the oxide layer thickness or a complete removal of the oxide scale dur-
ing usage due to friction wear?
Authors’ Response:—Guide tubes are fixed to the grids by bulge joints and there
is no possibility of wear. Assembly B and LTA grid vanes have no contact with
other assembly component (or with components of neighboring assemblies). There-
fore, in the present work, the only samples which could be affected by fretting wear
are assembly A grid 6 springs and dimples due to their contact with the fuel rods.
According to results given in Table 2, oxide thickness values obtained in grid 6

inner strap spring and dimple are 35 lm, being the oxide thickness on non-fuel-
rod-contact flat areas of the stamped strap 30 lm. If fretting wear would have
affected springs or dimples, oxide layer would be thinner or even detached from the
base metal. However, the oxide layer kept its integrity at these locations and the ox-
ide layer is even slightly thicker than in non-stamped flat areas. Therefore, in the
present work, no effects of fretting wear have been observed in the studied samples.
Question from Arthur Motta, Penn State University:—Do you think that there
could be some hydrogen redistribution within the grid strap, perhaps due to a tem-
perature gradient caused by differential heat transfer?
Authors’ Response:—Former experiments (Ref. 1) and current results show that,

unlike in fuel tubes, hydride precipitation is uniform across the guide tube wall and
the grid straps. Normally, guide tubes and grid straps are considered to have the
same temperature at inner and outer surfaces. Therefore, no temperature gradients
are considered across the guide tube wall or the grid straps. Former experiments
(Ref. 1) and current results seem to confirm that there are no temperature differen-
ces between inner and outer surfaces or, at least, the difference is small and it has
no significant effect on the hydrides distribution across the wall thickness of the
samples.
However, there could be particular locations in contact with fuel rods (whose
temperature is higher than in structural components) where a temperature gradient
would be higher than in the rest of the skeleton. Among all the samples analyzed in
this work, springs and dimples are the only samples in contact with fuel rods.
Hydride distribution micrographs corresponding to dimples and springs were dou-
ble checked at the time of answering this question and no hydride distribution gra-
dient was observed. Therefore, possible temperature gradient generated at dimples
and springs is still insufficient to create a non uniform hydride distribution across
the wall thickness.
Question from Suresh Yagnik, EPRI
Q1:—Is there a significant stress level on grids to cause a significant creep

strain?
Authors’ Response:—There are no significant loads during operation on grids.

The only significant stresses capable of causing strap creep are stresses generated by
volume expansion of the zirconium oxide layer during its formation on the metal
surface. According to simulations performed on fully recrystallized Zircaloy-4 grid
in Ref. 6, maximum stress in the metal due to oxide volume expansion could reach
50MPa. This stress could be enough to induce metal creep and should be taken into
account among the contributors to growth.
Comment:—The growth versus H data is not directly comparable to growth

from pre-hydrided samples. In the grids, the hydrogen ingress occurs simultane-
ously with the stress free irradiation growth.
Authors’ Response:—The authors agree with this comment. The non-rigorous

analysis of grid growth contributors performed in the manuscript precisely tries
to isolate free irradiation growth from other contributors. Some assumptions
were made for this analysis, which are explained in the text.
Questions from S. Anantharaman, BARC Mumbai
Q1:—What was the expected concentration of hydrogen? How does the meas-
ured value compare with the expected value?
Authors’ Response:—The expected values were very similar to the obtained val-
ues. The reason is that the expected values were based on the previous Westing-
house experiments obtained from irradiated material, Ref. 1.
Q2:—How were the specimens extracted from the fuel assembly?
Authors’ Response:—Assembly A: Figure 1 gives an idea regarding extraction of

the samples. All work was performed under water. The first step was transfer of all
the fuel rods from the target skeleton to a fresh skeleton. Once the skeleton was
empty, one corner of grid 2 and 6 including guide tube portions were sectioned.
Assembly B: while the fuel assembly was suspended (immersed in water) with
the handling tool, the grid vanes were latched and the tool bent the small vanes
back and forth until it was fractured from the grid perimeter strap and captured by
the tool.
Q3:—Did you carry out any mechanical tests to arrive at EOL mechanical prop-
erties? What is the value assumed by the designers?
Authors’ Response:—The goal of the present work was microstructure charac-

terization. Therefore, no mechanical testing was performed. Nevertheless, the EOL
mechanical properties assumed by the designers depend on the studied case, and
they are always selected to add conservatism to the calculations.
Q4:—Provide the method used for measuring hydrogen content.
Authors’ Response:—Methods used for hydrogen content measurement were

inert gas fusion and a scanning electron microscopy image analysis method
described in Ref. 7. More details about these methods are given in the manuscript.
Question from K. Kapoor, NFC:—What is the maximum discharge (end of life)

hydrogen content allowed (as per design) in the fuel tube.
Authors’ Response:—The current paper deals with structural components and

not cladding. The end-of-life cladding hydrogen levels at the 100 micrometer water-
side oxide thickness limit depend on the fuel power duty and the reactor coolant
temperature.
Questions from D. Kaczorowski, AREVA
Q1:—Oxide thickness was given for one side. Is it a mean value between inside
and outside diameter of guide tube and inner and outer grid strap?
Authors’ Response:—Yes, it is. An average 1-sided oxide thickness value was

used. Inner and outer oxide thickness results were compared in this study but no
significant thickness differences between them were observed. Therefore, average
1-sided oxide thickness is considered a representative value for the studied samples.
Q2:—Are hydrogen values for one or two sides?
Authors’ Response:—Unlike fuel rods, guide tubes and grid straps absorb hydro-
gen on two surfaces. This is an important point if guide tube and grid hydrogen
pick-up results are compared with fuel rod hydrogen pick-up results. Hydrogen
concentrations given in the present work are for two sides.
G. Pan,1 A. M. Garde,1,* and A. R. Atwood1
Performance and Property

Evaluation of High-Burnup
Optimized ZIRLOTM Cladding
Reference
Pan, G., Garde, A. M., and Atwood, A. R., “Performance and Property Evaluation
of High-Burnup Optimized ZIRLOTM Cladding,” Zirconium in the Nuclear Industry: 17th
ABSTRACT
Optimized ZIRLOTM cladding has been exposed to burnup levels exceeding 70
GWd/MTU, significantly higher than the licensed burnup in the United States of
62 GWd/MTU, in an effort to evaluate the performance margins of Optimized
ZIRLO cladding during irradiation to high burnups. Extensive poolside and hot-
cell post-irradiation examination (PIE) have been performed on the high-burnup
rods. Based on poolside PIE, Optimized ZIRLO-clad fuel rods have demonstrated
R
40 % lower in-reactor corrosion compared to ZIRLOV-clad fuel rods, at burnups
over 70 GWd/MTU. The oxide thickness of Optimized ZIRLO cladding is less than
70 lm for modified fuel duty index (MFDI) over 1000 and burnups over 70 GWd/
MTU. The Optimized ZIRLO-clad fuel rods have shown excellent in-reactor
dimensional stability, and similar diametral creep to ZIRLO-clad fuel rods.
Detailed examinations were conducted on the high-burnup Optimized ZIRLO-
clad rods in the hot cell, including length measurement, eddy current (EC) oxide-
thickness measurement, profilometry, visual inspection, metallography, scanning
electron microscopy (SEM), transmission electron microscopy (TEM), ring tensile
testing, axial tensile testing, and hot vacuum extraction hydrogen measurements.
Hot-cell examinations confirmed the low corrosion and low growth and also
revealed the high ductility of these high-burnup Optimized ZIRLO-clad rods. TEM
Manuscript received January 7, 2013; accepted for publication November 3, 2013; published online
October 6, 2014.
*ASTM Fellow.
1
Westinghouse Electric Company, Hopkins, SC 29061, United States of America.
2
Hyderabad, India.
Copyright V
examination on samples taken from the deformed region of the axial tensile test
samples relates dislocation channeling behavior to the ductility of Optimized
ZIRLO alloy. Metallography and SEM examinations reveal the layered structure of
the oxide layer. Poolside and hot-cell PIE results from the high-burnup Optimized
ZIRLO-clad rods are discussed in the paper.
Keywords
Optimized ZIRLO cladding, high burnup, hot-cell PIE, poolside PIE, in-PWR cor-
rosion, in-PWR irradiation growth, post-irradiation ductility, hydrogen pickup,
in-PWR fuel performance, dislocation channeling
Introduction
Optimized ZIRLO cladding material is a zirconium-niobium-tin based alloy, devel-
oped from and based on ZIRLO alloy, with a significant improvement in corrosion
resistance compared to ZIRLO alloy. Optimized ZIRLO cladding tubes have been
irradiated in more than 30 pressurized water reactors (PWRs) worldwide. The
improved corrosion performance of Optimized ZIRLO cladding and the excellent
dimensional stability, enable an increase in the fuel discharge burnup. The improved
in-reactor performance of Optimized ZIRLO cladding provides additional margin to
current Westinghouse fuel designs. The chemical composition of Optimized ZIRLO
cladding is tabulated in Table 1 with comparisons to ZIRLO cladding. The initial de-
velopment effort of an improved alloy based on ZIRLO cladding started with the
manufacture and testing of a 0.75 %-tin, low-tin ZIRLO alloy. Post-irradiation exami-
nation (PIE) of fuel rods fabricated with these tubes showed significant corrosion-
resistance improvement from the relatively modest tin reduction [1].
Further development of the alloy was undertaken in the subsequent years fol-
lowing the promising results from the initial in-PWR irradiation tests. This
included a further reduction of the tin level to a target value of 0.67 % and lowering
of the temperature of one intermediate annealing step to further improve corrosion
properties. As has been documented in a number of studies [1–4], reduction of the
tin content in zirconium-based alloys results in a reduction of uniform cladding
corrosion. However, a minimum tin content was needed to avoid accelerated corro-
sion in abnormal chemistry conditions for Zr-Nb alloys, such as elevated lithium
levels [1]. The reduction in the tin content inherently increases the creep rate of the
TABLE 1 Optimized ZIRLO and ZIRLO-cladding-alloy chemical compositions (nominal wt. %) and
microstructure from final heat treatment.
Alloy Microstructure Nb Sn Fe O Zr
Optimized pRXA 1 0.67 0.1 0.125 Bal.

ZIRLO
ZIRLO SRA 1 1 0.1 0.125 Bal.
PAN ET AL., DOI 10.1520/STP154320130008 609
final tubing. To compensate for this effect, Optimized ZIRLO is produced with a
final anneal at a slightly higher temperature than ZIRLO cladding. The higher final
annealing temperature results in a partially re-crystallized microstructure (PRXA).
Hence, Optimized ZIRLO cladding refers to a material with a target 0.67 % tin, a
specific intermediate anneal temperature, and a PRXA final anneal. Low Tin
ZIRLOTM historically refers to a cladding material with a target 0.75 % tin and a
stress-relief-annealed (SRA) final anneal, is currently being called “Low Tin ZIRLO-
SRA,” and will not be discussed in this paper (this material may be implemented in
the future in fuel assembly structures, but not in fuel-rod cladding). The nominal tar-
get value for niobium has remained at 1.0 wt. % for Optimized ZIRLO cladding.
An extensive out-reactor characterization program has been conducted on
Optimized ZIRLO cladding, which has been described in the Optimized ZIRLO
topical report [5].
Optimized ZIRLO fuel cladding use has expanded rapidly in recent years. As of
early 2013, there are over 1900 assemblies and over 400 000 Optimized ZIRLO fuel
rods currently in operation or discharged. The Optimized ZIRLO tubes that were
irradiated have been fabricated at different facilities and from multiple ingots and
tube lots, with no performance impact observed.
Optimized ZIRLO fuel irradiation performance in PWRs has been reported at
the 2010 Topfuel Conference [6]. Since then, more irradiation experience and per-
formance data have been obtained for Optimized ZIRLO cladding. In particular,
Optimized ZIRLO cladding has achieved high-burnup experience exceeding 70
GWd/MTU in the U.S. PWR plant A, significantly higher than the licensed burnup
limit in the United States of 62 GWd/MTU. In an effort to evaluate the perform-
ance margins of Optimized ZIRLO cladding during irradiation to high burnups,
extensive poolside and hot-cell PIEs have been performed on those high-burnup
rods. The focus of this paper is the high-burnup performance and properties of
Optimized ZIRLO cladding based on the poolside and hot-cell PIE results.
Description of the High-Burnup Fuel Rods

One lead test assembly with Optimized ZIRLO cladding was irradiated in the U.S.
PWR reactor plant A for three cycles and discharged in 2009. Poolside inspection
has been performed on these rods after each cycle of operation. After the last cycle,
14 Optimized ZIRLO rods were inspected with rod-average burnups in the range
from 65 GWd/MTU to 74 GWd/MTU. The inspection included visual inspection,
eddy current waterside oxide measurements, rod-length measurements, as well as
diametral profilometry. From the 14 inspected rods, three rods were sent to a hot
cell for further destructive examination. The selection criteria of rods shipped to the
hot cell cover the following aspects: internal/peripheral position within the fuel as-
sembly, different pellet type, highest/lowest burnup, and range of waterside oxide
thicknesses. All three rods have the same composition and heat treatment. The rods
were cut into four segments at the hot cell prior to further examination. Length
measurement and visual inspections were conducted on all four segments of the
rod. Eddy current (EC) oxide-thickness measurements, profilometry, were per-
formed on selected segments. The rod segments chosen to be extensively evaluated
in the hot cell were taken from the upper part of the rod where most features of in-
terest are present. Samples were cut from the chosen segments for destructive tests,
including metallography, scanning electron microscopy (SEM), transmission elec-
tron microscopy (TEM), ring tensile testing (RTT), axial tensile testing (ATT), and
hot vacuum extraction (HVE) hydrogen measurements. The primary segments
examined and the list of tests each segment was subjected to are listed in Table 2.
Additional tensile-test samples and hydrogen-measurement samples are also taken
from a second segment at lower elevation. The tensile-tests specimen, the
hydrogen-analysis sample, the metallography, and the SEM samples have been
selected to be adjacent to each other in the rod axial location. The most extensively
studied area for the examined rods is about 3 m from the bottom of the rod.

IN-REACTOR CORROSION PERFORMANCE
Post-irradiation examination results of Optimized ZIRLO cladding have shown
very good in-reactor corrosion performance based on a large database. The eddy
TABLE 2 Description of Optimized ZIRLO rods examined at hot cell.
Primary Segment
Measurement Examined (Segment
Rod Burnup Growth Peak Oxide Numbering from
ID (MWd/MTU) (%) (lm) (63 lm) Bottom of the Rod) List of Tests
1 73 359 0.22 41 Segment 3 Fission gas release,

(from 2.2 m to 3.4 m rod internal pressure
from rod bottom) on the rod before
cutting;
Length measurement
and visual inspection
on the entire rod;
EC oxide, profilome-
try on the entire
selected segment;
2 71 030 0.33 56 Segment 4 Metallography, SEM,
(from 2.8 m to top of RTT, ATT, and HVE
the rod) hydrogen measure-
ments on samples
from examined
segment
3 70 182 0.43 62 None Length measurement
on the entire rod
PAN ET AL., DOI 10.1520/STP154320130008 611
FIG. 1 Measured peak oxide thickness for Optimized ZIRLO and ZIRLO rods as a
function of burnup.
current measured oxide thicknesses from poolside measurements of different plants

are shown in Fig. 1 as a function of burnup and in Fig. 2 as a function of the modi-
fied fuel duty index (MFDI) [7,8]. The measurements represent peak oxide thick-
ness. The observed scatter in oxide thickness for the high-burnup rods from plant
FIG. 2 Measured peak oxide thickness for Optimized ZIRLO and ZIRLO-clad fuel rods
as a function of modified fuel duty index (MFDI). It should be noted that only
part of the Optimized ZIRLO oxide database has MFDI calculated.
A is commonly observed and likely a result of local operating environments on the

rods, which are not possible to quantify. The measurements indicate that oxide data
for Optimized ZIRLO cladding are similar to those for ZIRLO rods at low burnup
of about 20–30 GWd/MTU. At higher burnups, however, the advantage of the
Optimized ZIRLO alloy over ZIRLO alloy becomes significant. The oxide-thickness
values from the high-burnup rods from plant A are less than 70 lm, which are sig-
nificantly lower than the maximum value of 110 lm of ZIRLO rods at similar burn-
ups. The eddy current oxide measurements of the high-burnup rods from plant A
at the hot cell are consistent with the measurements from the poolside examination.
Corrosion data as a function of MFDI more accurately capture the corrosion per-
formance difference of the rods experiencing different fuel duties at diverse plant
operating conditions [7,8]. The fuel duty index (FDI) was initially developed as an
alternative to presenting corrosion data analysis as a function of burnup [7]. The
FDI (considering time and temperature) more accurately captures the dominant
corrosion behavior, which is primarily a function of time at temperature and only a
weak function of burnup. During further development, a boiling term was added to
improve the FDI. Consequently, the MFDI was developed to account for the effects
of boiling on corrosion [8]. The data show that at intermediate to high MFDI val-
ues, the oxide-thickness values of Optimized ZIRLO alloy are much lower than
those of ZIRLO alloy. The oxide thickness for Optimized ZIRLO alloy is less than
70 lm at MFDI over 1000 and burnup of about 70 GWd/MTU. Overall, at high
burnups, Optimized ZIRLO-clad fuel rods have demonstrated 40 % lower in-
reactor corrosion compared to ZIRLO-clad fuel rods.
ROD GROWTH
The Optimized ZIRLO-cladding fuel rod-growth database is composed of measure-
ments from a number of plants including both U.S. and European reactors, as
shown in the plot in Fig. 3, where the rod-growth data is plotted as a function of flu-
ence. The fluence is calculated from burnup using the fast fluence conversion factor,
which is a function of burnup. The rod-growth database is still very limited, espe-
cially at high fluence, and there is a large scatter among different reactors. Rod-
growth data obtained from gap measurement and single rod-length measurement
are both included, which is one of the contributing factors for the scatter. For gap
measurement, the rod length is obtained by subtracting the length of the gap
between the top of the rod and the bottom of the top nozzle from the distance
between the bottom and the top nozzle. The gap measurement method is based on
the assumption that there is no space between the bottom of the rod and the bot-
tom nozzle. Therefore, the rod length obtained from gap measurement will be
larger than the actual values if the rods are not dropped to the bottom nozzle. Single
rod-length measurement was done on rods pulled out of the assembly. The rod-
growth results of the high-burnup rods were obtained using single rod measure-
ment during poolside inspections. The rod-length measurements made in the hot
cell on three of these high-burnup rods from plant A confirmed the rod-growth
PAN ET AL., DOI 10.1520/STP154320130008 613
FIG. 3 Optimized ZIRLO and ZIRLO fuel-rod-growth database as a function of fast

neutron fluence.
results from poolside inspections. Based upon the available data, it can be seen that
the growth of Optimized ZIRLO rods are similar or lower than the growth of
ZIRLO rods and that the Optimized ZIRLO rods do not exhibit accelerated rod
growth at high burnup. Based on the high-burnup data from plant A at fluence
over 12 1021 n/cm2, E > 1 MeV (burnup over 70 GWd/MTU), Optimized ZIRLO
rods exhibit lower growth than ZIRLO rods. The specific reason for the low growth
of those rods cannot be identified as fuel rod growth is dependent on many compo-
nents that include irradiation growth, pellet column extension, contact between fuel
cladding and pellet, and cladding hydrogen pick up at higher burnups. Because
Optimized ZIRLO cladding has similar creep properties [9] and lower hydrogen
pickup (see the section Hydrogen Analysis below) than ZIRLO cladding, the rod
growth for Optimized ZIRLO rods is expected to be bounded by the growth of
ZIRLO rods as demonstrated in Fig. 3.
OXIDE-LAYER CHARACTERIZATION
High resolution visual inspection of the high-burnup Optimized ZIRLO rods has
revealed a minor degree of oxide surface peeling (OSP) at several locations in the
upper part of the rods. Oxide surface peeling, delamination of thin waterside oxide,
is the new process that has been recently observed on Optimized ZIRLO cladding
and other high-corrosion-resistant alloys [10]. This process is different from oxide
spalling, which has been seen on low–corrosion-resistance cladding such as
Zircaloy-4, where a major thickness fraction of the thick oxide was removed. The
key differences between oxide surface peeling and oxide spalling are that OSP
occurs on thin oxide and will not lead to hydride localization under the peeled
region and enhanced embrittlement, whereas oxide spalling occurs on thick oxide
and will lead to hydride localization under the spalled region and enhanced embrit-
tlement. The observations of OSP on Optimized ZIRLO cladding and other high-
corrosion-resistant alloys as well as the impact of OSP on fuel performance are dis-
cussed in Ref 10. Metallography samples were prepared for polished surfaces to
cover both peeled and non-peeled adjacent oxide so they could be examined using
light optical microscopy (LOM). Metallographic examination revealed the periodic
nature of the oxide layers formed for high-burnup Optimized ZIRLO cladding, sim-
ilar to non-irradiated autoclaved Zircaloy specimens [11]. Figure 4 shows one repre-
sentative image of the outer oxide profiles at 50 orientation from a longitudinal
metallographic sample at 3 m from the bottom of rod No. 1. The outer oxide thick-
ness is about 48 lm. The oxide has a layered structure throughout the oxide thick-
ness with a layer thickness of about 3 lm.
Short cracks can be observed between layers at the corrosion rate transition
point separating cyclic processes of oxide growth. In some cases, the cracking at the
transition zone has grown into long circumferential cracks. When the long circum-
ferential cracks meet the radial cracks that initiated on the outer surface of the ox-
ide, delamination of oxide layers and possible OSP may result. The extent of OSP
appears to be alloy composition dependent, which controls the corrosion resistance
of the alloy. For Optimized ZIRLO alloy, OSP removes only a few outer layers of
the oxide, but the radial cracking of the inner oxide layer was very limited. Also,
OSP is less frequently observed on Optimized ZIRLO cladding than other higher
corrosion-resistant alloys with very thin oxide, as shown in Ref 10.
Hydride localization was not associated with OSP, as shown in the hydride
micrographs in the following section. Therefore, it is not expected to lead to irradi-
ated alloy ductility reduction, as confirmed by the tensile tests results discussed
below.
FIG. 4 Light optical micrograph of oxide layers on high-burnup Optimized ZIRLO rods.
Longitudinal sample at a rod elevation of 3 m of No. 1 rod at 50 orientation.
PAN ET AL., DOI 10.1520/STP154320130008 615
HYDROGEN ANALYSIS
Hydrogen measurements were performed by the HVE method. A 1-mm-cladding
ring was cut at mid-pellet positions and the analyzed hydrogen represents the aver-
age content of the 1-mm ring (including both the metal and oxide layer). The
hydrogen measurement was taken at different rod elevations to investigate the
hydrogen content at different oxide thicknesses.
The hydrogen uptake ratio is calculated as the fraction of hydrogen absorbed
by the metal relative to the amount of hydrogen generated by the waterside corro-
sion reaction. Instead of 100 % theoretically dense oxide, a uniform layer of oxide
with slightly lower density was used in the calculation to account for voids and
cracking of the oxide. OSP was not present on the hydrogen analysis specimens
tested. The possible impact of oxide surface peeling on hydrogen pickup by the
metal was not considered. Such an effect, if present, is expected to be small because
of a small degree of OSP for Optimized ZIRLO cladding.
To calculate the amount of hydrogen absorbed by the metal, the hydrogen con-
tent was measured by HVE from a 1-mm-cladding ring (including metal and ox-
ide). The hydrogen content in the metal was obtained by subtracting the estimated
hydrogen content in the oxide from the measured hydrogen. Hydrogen content in
the oxide was estimated from published values of hydrogen content in the oxide
formed on both irradiated cladding [12] and autoclaved corrosion specimens [13].
Based upon those results, hydrogen in the oxide was estimated to be in the range of
1000 to 1100 ppm. After accounting for hydrogen in the oxide, the hydrogen con-
tent in the metal for the high-burnup Optimized ZIRLO-cladding samples ranged
from 140 ppm to 520 ppm. The calculated hydrogen uptake ratio of Optimized
ZIRLO cladding is similar to that of ZIRLO cladding, averaging about 15 %. How-
ever, because the Optimized ZIRLO cladding has lower oxide thickness, the overall
hydrogen level in Optimized ZIRLO cladding is less than in ZIRLO cladding.
The HVE samples were taken adjacent to the metallography samples. The
hydride distribution on the mounted irradiated samples was revealed by etching the
polished metallographic specimen with an HNO3/HF solution. The visual appear-
ance of the hydride needle density confirms the hydrogen content values obtained
from the HVE analysis. A hydride micrograph is shown in Fig. 5. This is a trans-
verse cladding cross section taken from rod No. 1 at an axial position of 3 m from
the bottom of the rod. It can be observed that most of the zirconium hydrides were
circumferentially oriented with a greater concentration near the tube outer surface
(metal waterside corrosion oxide interface) because of its lower temperature. The
hydride concentration was relatively uniform around the circumference from the
mid-wall thickness to the tube inner surface. Figure 5 shows two representative
images around the circumference. Under the peeled oxide location, the near outer-
surface hydrides were more concentrated than in surrounding areas (Fig. 5(a)).
However, the higher concentration in this area was similar to higher concentrations
measured in other areas where no peeling was observed (Fig. 5(b)). Overall, the near
outer-surface hydride localization below the oxide peeled location was generally
much lower as compared with that below spalled oxide on Zircaloy-4 rod shown in
the literature [14]. Quantitative evaluation of the extent of hydride localization
under the oxide peeled region [10] showed insignificant hydride localization
because of OSP. Based on Fig. 5, there seems to be more short hydride and less ra-
dial hydrides under the oxide peeled region. But further analyses are needed to ver-
ify this speculation.
MECHANICAL PROPERTIES
About 90-mm-long pieces were cut from the fuel rod, defueled, and then split axi-
ally into two equal halves. The final “dog-bone”-type tensile test samples were
machined from the cladding halves using a milling machine. The cladding axial ten-
sile tests were performed at room temperature and at 385 C in air on an Instron
1271 servo-hydraulic machine with a cross-head speed of 0.56 mm/min. The
reduced gage length is 33 mm. The engineering stress–strain diagrams were
obtained based on the recorded load-displacement data. The calculations are based
on initial adjusted gage length and cross-sectional area. The cross-sectional area is
FIG. 5 Etched appearance of zirconium hydride concentration and orientation at (a)

oxide peeling locations, and (b) non-oxide peeling regions. The estimated
hydrogen content in the metal in this region is 340 ppm.
PAN ET AL., DOI 10.1520/STP154320130008 617
calculated from the nominal wall thickness and the circumferentially averaged
width. The elongation is measured from crosshead displacement.
The analysis of the stress–strain curves gives the values of yield stress (YS,
0.2 % offset yield stress), the ultimate tensile stress (UTS), the uniform plastic
strain, and the total plastic strain. Those terms are noted on the stress–strain curve
for the axial tensile test in Fig. 6. This specimen was taken from the upper part of
the rod with maximum oxide thickness and tested at 385 C. The stress–strain curve
shows that the high-fluence Optimized ZIRLO alloy specimen has higher ductility
compared to Zircaloy-4, or even ZIRLO with 1.8 % uniform plastic elongation and
10.5 % total plastic elongation. The results of axial tensile tests at 350 C for ZIRLO
and Zircaloy-4 specimens show that the ZIRLO specimen (with a burnup level over
70 GWd/MTU) has a total plastic strain range of 1.3 % to 6.8 %, and Zircaloy-4
specimen (with a burnup level of about 55 GWd/MTU) has a total plastic strain
range of 0.1 % to 1.9 % [15]. The hydrogen level for the ZIRLO and Zircaloy-4
specimens tested range from 330 ppm to 937 ppm. The mechanical test data of all
Optimized ZIRLO specimens including specimen 1-A4 are summarized in Table 3.
The fast fluences associated with the tensile tests specimens were calculated from
rod burnups to be 15 1021 n/cm2 (E > 1 MeV) for the specimens from rod No.
1 (1-A1, to 1-A4) and 12 1021 n/cm2 (E > 1 MeV) for specimens from rod No.
2 (2-A1 and 2-A2). The reduction of area (RA) at the fracture point was also eval-
uated from the sample dimensions as a measure of ductility. The hydrogen analysis
results on the samples adjacent to the tensile test samples are included in Table 3 to
give an estimate of hydrogen values in these tensile test samples. It can be seen
from Table 3 that specimen 2-A1 and 2-A2 are both from areas with relatively high
hydrogen content and both specimens show relatively low ductility. Especially,
FIG. 6 Engineering stress–strain diagram for sample 1-A4, taken from position 3 m from
bottom of the rod and tested at 385 C.
TABLE 3 The axial tensile test results of Optimized ZIRLO cladding (dog-bone-shaped sample)
from high-burnup rods.
Sample Estimated
Position Hydrogen
(Meter, from Test Uniform Total Content Reduction
Sample Bottom of Temperature YS UTS Plastic Plastic in Metal of Area
IDa Rod) ( C) (MPa) (MPa) Strain (%) Strain (%) (ppm) (%)
1-A1 1.5 RT 749 937 3.5 5.6 140 21

1-A3 2.9 RT 641 832 3.6 6.6 340 27
2-A1 3.1 RT 626 801 2.2 2.3 520 21
1-A2 1.5 385 461 556 1.6 9.2 140 50
1-A4 2.9 385 436 508 1.8 10.5 340 47
2-A2 3.1 385 423 511 2.3 6.3 520 43
a
The first number of the sample ID is the rod ID.
specimen 2-A1 fractured at the pellet–pellet interface with a fracture surface per-
pendicular to the tubes tensile axis. All other specimens listed in Table 3 are frac-
tured in a ductile fashion with fracture surfaces at 45 to the tube tensile axis, as
shown in Fig. 7 for specimen 1-A4. The photograph of specimen 1-A4 after the
FIG. 7 Inward curvature surface of the axial tensile test sample 1-A4 after testing. The
TEM sample locations are shown as dashed white circles. (a) Overview of the
fractured tensile test sample 1-A4. The close-up view of the fracture surface (b)
shows macroscopic evidence of deformation bands (multiple bands) on the
specimen surface and the 45 fracture surface to the tube tensile axis.
PAN ET AL., DOI 10.1520/STP154320130008 619
tensile tests is presented in Fig. 7. The picture was taken from the inside surface of
the tube, where signs of the former pellet-to-pellet interface are visible at regularly
spaced transverse lines (Fig. 7(a)). It shows that the fracture surface is at the mid-
pellet location and oriented about 45 to the tube tensile axis. There are numerous
fine deformation bands at 45 as observed on ZIRLO tensile test specimens [15].
There are no localized deformation bands as seen in Zircaloy-4 tensile test speci-
mens [15].
MICROSTRUCTURE OF DEFORMED SPECIMEN
A correlation has been proposed between the macroscopic deformation bands

observed on the deformed surface of fractured-irradiated-material specimens and
the dislocation channeling observed at the microscopic scale in the individual grains
[15]. Dislocation channeling, the formation of channels in which irradiation loops
are annihilated by dislocation gliding, has been observed on irradiated materials
[16–18]. It was proposed that the multiple fine deformation bands could be a mac-
roscopic manifestation of the nucleation of large numbers of dislocation channels
as a result of niobium-induced work hardening preventing deformation concentra-
tion in a limited number of dislocation channels. The relatively higher ductility of
ZIRLO and Optimized ZIRLO alloys compared to Zircaloy-4 alloy could be because
of this mechanism forming more dislocation channels within the grains, so that the
deformation within the individual dislocation channels is limited. In an attempt to
verify the above proposed mechanism, TEM specimens were prepared from the
high-strain region of an axial tensile test sample of high-burnup Optimized ZIRLO
cladding. Thin foils were prepared from the gauge section of the specimen 1-A4,
next to the fracture surface, noted by the white circles in Fig. 7. Based on the area
reduction calculation, the local strains in these thin foils are close to 47 %.
Figure 8 showed the highly deformed microstructure of the tensile test specimen
1-A4. The subgrains, between 50 to 500 nm wide, are elongated in the loading
direction. The mis-orientation between adjacent subgrains is small. There are also
smaller grains in the microstructure appearing more equiaxed in shape. Further,
the defects and dislocations in the microstructure seem to be unevenly distributed.
However, dislocation-free zones (dislocation channels) are not observed. The for-
mation of new grain boundaries and the elongated subgrain structure as well as the
high dislocation densities complicate the analysis and make it difficult to elucidate
which of the deformation mechanisms (slip systems and even twinning) have been
activated. For high-fluence-irradiated zirconium alloys subject to axial tensile tests,
previous observations [16–18] of dislocation channeling has been mainly at low
strain regions, i.e., at strains below 1 % plastic strain, when the low density of dislo-
cations allows the analysis of activated slip systems to be carried out. It has been
reported at low strains that: (a) for axial tensile tests, dislocation channeling occurs
on prism and pyramidal planes; (b) such channeling cannot easily sweep loop hai
irradiation defects introduced by neutron damage into zirconium alloys as two out
FIG. 8 TEM micrograph of deformed tensile test sample 1-A4 showing the elongated
laminar structure in the direction of the large double arrow, approximately
parallel to the loading direction. Small light arrows show locations with more
equiaxed grains and small dark arrows show regions with high dislocation
density.
of three dislocation jogs generated by channel interaction with the irradiation-

induced hai loops are sessile; and (c) as a result, dislocation channeling may not cre-
ate totally defect-free bands in the axial tensile test specimens. It was also reported
that for high strains of a transverse tensile specimen, all three types of dislocation
channels (basal, prism, and pyramidal) are observed, whereas at lower strains, only
basal channels are observed. Therefore, there is a possibility that all three types of
channels may be forming at high strains for an axial tensile test specimen also.
These past observations are consistent with the current TEM observations on Opti-
mized ZIRLO axial tensile test sample regarding the absence of dislocation-free
channels. The elongated subgrains observed can be thought to be an effect of the
dislocation channeling. The absence of defect-free dislocation channels at high
strains implies that the deformation is not localized and new dislocations generated
at high strains within the channels nucleated at low strains provide work hardening
(deformation resistance) to improve the irradiated-material ductility. The niobium-
induced work hardening at high strains may be limiting the dislocation sweep dis-
tance in a channel to such an extent that the channel eventually loses its dislocation-
free band status. The dislocation-free channel generated at low strain gets filled with
new dislocations, resulting in the microstructure observed here in Fig. 8.
PAN ET AL., DOI 10.1520/STP154320130008 621
It is proposed that the dislocation channeling observations involve the follow-

ing variable: (a) level of irradiation damage: at low fluence, only hai loops are gener-
ated, whereas, at high fluence, hci loops are generated; (b) type of deformation test:
internal pressure, ring tensile, axial tensile, and transverse tensile tests generate dif-
ferent types of dislocation channels; (c) temperature of tensile test, which affects the
mobility of irradiation defects; (d) level of strain: at low strains, only one type of
channels are dominant, whereas, at high strains, multiple types of channels are
likely; (e) alloy composition: as the levels and types of alloying elements determine
the binding energy drag force opposing the dislocation motion; and (f) orientation
of the TEM foil, compared to the specimen stress direction applied during the test-
ing. The TEM data collected so far are not sufficient to evaluate all the above listed
factors. Additional work is needed to evaluate the impact of all listed parameters.
Conclusions
Optimized ZIRLO cladding is an improved version of ZIRLO cladding with the aim
of retaining the main properties of ZIRLO cladding in application as PWR cladding
material and significantly improving the corrosion properties. The Optimized
ZIRLO cladding has been irradiated in a multitude of plants around the world and has
demonstrated excellent performance. Optimized ZIRLO cladding has achieved high
burnup over 70 GWd/MTU. The high-burnup behavior of Optimized ZIRLO cladding
has been studied and summarized below:
1. At high burnups, Optimized ZIRLO cladding demonstrated a corrosion per-
formance 40 % better than that of ZIRLO alloy, providing a significant margin
in high fuel duty operating environments.
2. Optimized ZIRLO cladding exhibits excellent dimensional stability over 70
GWd/MTU, and there is no accelerated rod growth.
3. The Optimized ZIRLO cladding retains relatively high ductility after high-
burnup irradiation.
4. The hydrogen pickup fraction of Optimized ZIRLO cladding is similar to
ZIRLO cladding, averaging about 15 %; however, the overall hydrogen content
of high-burnup Optimized ZIRLO cladding is lower than ZIRLO cladding
because of lower oxide thickness.
5. The waterside oxide of high-burnup Optimized ZIRLO cladding has a layered
structure with a layer period thickness of about 3 lm.
6. A minor degree of OSP, delamination of thin waterside oxide, has been
observed on high-burnup Optimized ZIRLO cladding. Optimized ZIRLO clad-
ding has a lower extent of OSP than other higher-corrosion-resistant zirco-
nium alloys and does not exhibit oxide spalling, like the low-corrosion-
resistant alloys (such as Zircaloy-4).
7. The oxide surface peeling does not lead to hydrogen localization and, there-
fore, is not expected to reduce irradiated alloy ductility.
8. The highly deformed tensile test sample has an elongated laminar microstruc-
ture with smaller equiaxed grains and unevenly distributed dislocations. How-
ever, dislocation-free zones (dislocation channels) are not observed.
9. The microstructure of the highly deformed sample is probably the result of

new dislocations filling the dislocation-free channel generated at low strain
and the interaction of different deformation mechanisms.
ACKNOWLEDGMENTS
The writers acknowledge with thanks the efforts of personnel at Studsvik, Sweden, for
performing hot-cell examinations. Particularly, we thank Daniel Jädernäs for his input
on TEM discussions and Rikard Källström for team management to collect the hot-
cell PIE data.
References
[1] Yueh, H. K., Kesterson, R. L., Comstock, R. J., Shah, H. H., Colburn, D. J., Dahlback, M.,
and Hallstadius, L., “Improved ZIRLOTM Cladding Performance Through Chemistry and
Process Modifications,” J. ASTM Int., Vol. 2, No. 6, 2005, Paper ID JAI12344.
[2] Comstock, R. J., Schoenberger, G., and Sabol, G. P., “Influence of Processing Variables
and Alloy Chemistry on the Corrosion Behavior of ZIRLO Nuclear Fuel Cladding,” Zirco-
nium in the Nuclear Industry: Eleventh International Symposium, ASTM STP 1295, E. R.
Bradley and G. P. Sabol, Eds., ASTM International, West Conshohocken, PA, 1996, pp.
710–725.
[3] Sabol, G. P., Comstock, R. J., Weiner, R. A., Larouere, P., and Stanutz, R. N., “In-Reactor
Corrosion Performance of ZIRLO and Zircaloy-4,” Zirconium in the Nuclear Industry:
Tenth International Symposium, ASTM STP 1245, A. M. Garde and E. R. Bradley, Eds.,
[4] Garde, A. M., Pati, S. R., Krammen, M. A., Smith, G. P., and Endter, R. K., “Corrosion
Behavior of Zircaloy-4 Cladding With Varying Tin Content in High-Temperature Pres-
sured Water Reactors,” Zirconium in the Nuclear Industry: Tenth International Sympo-
sium, ASTM STP 1245, A. M. Garde and E. R. Bradley, Eds., ASTM International, West
[5] “Optimized ZIRLOTM,” Westinghouse Report WCAP-14342-A and CENPD-404-NP-A,

Addendum 1-A, ADAMS Public Documents, U.S. Nuclear Regulatory Commission (NRC),
[6] Mitchell, D., Garde, A., and Davis, D., “Optimized ZIRLO Fuel Performance in Westing-
house PWRs,” Proceedings of Top Fuel 2010, Orlando, FL, Sept 26–29, American Nuclear
Society, La Grange Park, IL, 2010.
[7] Kaiser, R. S., Leech, W. J., and Casadei, A. L., “The Fuel Duty Index (FDI)—A New Mea-
sure of Fuel Rod Cladding Performance,” An International Topical Meeting on Light
Water Reactor Fuel Performance, Park City, UT, April 10–13, American Nuclear Society,
La Grange Park, IL, 2000.
[8] Leech, W. J. and Yueh, K., “The Fuel Duty Index, a Method to Assess Fuel Performance,”
Top Fuel Conference 2001, Paper P2-16, Stockholm, Sweden, May 27–30, European Nu-
clear Society, Brussels, Belgium, 2001.
PAN ET AL., DOI 10.1520/STP154320130008 623
[9] Chapin, D. L., Wikmark, G., Maury, C., Therache, B., Claeys, M., Gutierrez, M. Q., and Ruiz,
C. M.-R., “Optimized ZIRLO Qualification Program for EDF Reactors,” Top Fuel Confer-
ence 2009, Paper 2040, Paris, France, Sept 6–10, European Nuclear Society, Brussels,
Belgium, 2009.
[10] Garde, A. M., Pan, G., Mueller, A. J., and Hallstadius, L., “Oxide Surface Peeling of
Advanced Zirconium Alloy Cladding After High Burnup Irradiation in Pressurized
Water Reactors,” Zirconium in the Nuclear Industry: Seventeenth International
Symposium, R. Comstock and P. Barberis, Ed., ASTM STP 1543, ASTM International, West
Conshohocken, PA, 2015, pp. 673–692, STP in progress.
[11] Motta, A. T., Yilmazbayhan, A. M., Comstock, R. J., Partezana, J., Sabol, G. P., Lai, B.,
and Cai, Z., “Microstructure and Growth Mechanism of Oxide Layers Formed on Zr
Alloys Studied with Micro-Beam Synchrotron Radiation,”Zirconium in the Nuclear
Industry: Fourteenth International Symposium, P. Rudling and B. Kammenzind,
Eds., ASTM STP 1467, ASTM International, West Conshohocken, PA, 2005, pp.
205–232.
[12] Hermann, A., Wiese, H., Buehner, R., Steinemann, M., and Bond, G., “Hydrogen Distribu-
tion between Fuel Cladding Metal and Overlying Corrosion Layers,” An International
Topical Meeting on Light Water Reactor Fuel Performance, Park City, UT, April 10-13,
2000, American Nuclear Society, La Grange Park, IL.
Redistribution in Alpha-Annealed Zircaloy-4,” Zirconium in the Nuclear Industry: Elev-
[14] Garde, A. M., Smith, G. P., and Pirek, R. C., “Effects of Hydride Precipitate Localization
and Neutron Fluence on the Ductility of Irradiated Zircaloy-4,” Zirconium in the
Nuclear Industry: Eleventh International Symposium, ASTM STP 1295, E. Ross Bradley
and G. P. Sabol, Eds., ASTM International, West Conshohocken, PA, 1996, pp.
407–446.
V R
[15] Garde, A. and Mitchell, D., “Comparison of Ductility of Irradiated ZIRLO and Zircaloy-4,”
Topfuel 2012 Reactor Fuel Performance Conference, Paper 2012-A0149, Manchester, UK,
Sept 2–6, European Nuclear Society, Brussels, Belgium, 2012.
[16] Fregonese, M., Regnard, C., Rouillon, L., Magnin, T., Lefebvre, F., and Lemaignan,
C., “Failure Mechanisms of Irradiated Zr Alloys Related to PCI: Activated Slip Sys-
tem, Localized Strains, and Iodine-Induced Stress Corrosion Cracking,” Zirconium
in the Nuclear Industry: Twelfth International Symposium, ASTM STP 1354, G. P.
Sabol and G. D. Moan, Eds., ASTM International, West Conshohocken, PA, 2000,
pp. 337–398.
[17] Onimus, F., Bechade, J. L., Prioul, C., Pilvin, P., Monnet, I., Doriot, S., Verhaeghe, B., Gil-
bon, D., Robert, L., Legras, L., and Mardon, J. P., “Plastic Deformation of Irradiated Zirco-
nium Alloys: TEM Investigation and Micro-Mechanical Modeling,” J. ASTM Int., Vol. 2, No.
8, 2005, Paper ID JAI12424.
[18] Onimus, F., Monnet, I., Bechade, J. L., Prioul, C., and Pilvin, P., “A Statistical TEM Investi-
gation of Dislocation Channeling Mechanism in Neutron Irradiated Zirconium Alloys,” J.
DISCUSSION
Q1:—Why do you promote to have more than 30 lm of oxide to prevent wear?

Have you a problem of fretting in your assemblies?
Authors’ Response:—Westinghouse does not have a fretting wear problem with

fuel with Optimized ZIRLO cladding. The wear data analysis shows that a 30 lm of
cladding oxide thickness provides sufficient margin to prevent wear failures due to
debris or grid to rod fretting (GTRF). Based on the fact that alloys with very thin
oxide are less resistant to oxide surface peeling and sometimes the entire oxide layer
may peel off to bare metal, we propose that the OSP may reduce the fretting wear
margin for alloys with high corrosion resistance. Oxide surface peeling (especially
to bare metal) is the concern that may lead to a reduction in wear margin (when
other factors promoting wear are present). Cladding with an intermediate oxide
thickness of 30 lm or greater is more resistant to oxide surface peeling and, there-
fore, reduces the susceptibility of fretting wear.
Q2:—Do you pre-oxidize the rods?
Authors’ Response:—We pre-oxidize the bottom part of the rods in selected fuel
designs.
Q3:—Fretting wear occurs in the bottom of the rods where 30 lm is never

reached. How do make you the connection between oxide thickness and wear
resistance? What is the limit?
Authors’ Response:—Although there is connection between cladding oxide thick-

ness and cladding wear resistance, oxide thickness is not the only parameter controlling
wear. Other factors such as vibratory contact, area of contact, contact pressure, relative
hardness of contacting surfaces are also important. As a result, there is no single limit
on oxide thickness. Alloys with very thin maximum oxide (high corrosion resistance)
are more susceptible to oxide surface peeling where a portion of the thin oxide might
peel off and expose an area with little or no oxide. When that area is in contact with a
piece of debris or grid, debris fretting or grid-to-rod fretting (GTRF) may occur. It is
necessary to implement other measures (such as reduced contact loads and decreased
vibrations) to achieve adequate wear resistance margin for high corrosion resistance
cladding alloys that are likely to experience OSP.
Q1:—What is the yield stress and ductility of the unirradiated material? What
was the change with irradiation?
PAN ET AL., DOI 10.1520/STP154320130008 625
Authors’ Response:—The room temperature yield stress and total elongation of

the unirradiated Optimized ZIRLO cladding are about 500 MPa and 25%, respec-
tively. Tensile properties following irradiation are provided in Table 3. The room
temperature yield stress increases to about 670 MPa while the total elongation
decreases to about 4.8%.
Q2:—Why does deformation localization not occur in your alloy? Was there
macroscopic localization (you reported no microscopic localization) during
deformation?
Authors’ Response:—For deformation localization to occur, both microscopic

and macroscopic deformation localizations are necessary with a direct connection
between the two with deformation bands extending to multiple grains. For
deformed irradiated Optimized ZIRLO specimens, limited macroscopic deforma-
tion localization was seen as short linear surface markings of multiple orientations
within individual grains. However, the markings in the neighboring grains were not
connected and aligned as a continuous deformation band extending to multiple
grains. Microscopic deformation localization was not observed at high strains in
deformed specimens as no dislocation channeling (bands without dislocations)
were observed in TEM. Most likely, the absence of deformation localization is
related to the work hardening associated with the interaction between dislocation
motion and the stable defect complexes (irradiation induced point defects com-
bined with niobium and oxygen atoms in the alloy) generated in the irradiated
Optimized ZIRLO material.
Q3:—What was the overall hydrogen content in the sample where you show
delamination but no hydride lens formation?
Authors’ Response:—Hydrogen measurements were made on specimens that

were adjacent to the metallography specimens. The estimated hydrogen content
in the metal of the specimen that exhibited oxide surface peeling shown in Fig-
ure 5 was 340 ppm.
Question from Yang-Pi Lin, Global Nuclear Fuel:—Can you provide informa-
tion on the power of the high burnup Optimized ZIRLO rods and discuss if this
power has an effect on lack of hydride lens at peeling locations?
Authors’ Response:—The linear heat rate of the Optimized ZIRLO rods varied
with irradiation time and reactor. The linear heat rate for the high burnup rods cov-
ered the range of roughly 15–26 kW/meter. There was no correlation between rod
power and lack of hydride lens at the peeling location. Hydride localization is
related to the local cladding thermal gradients, either circumferential or axial. The
lack of correlation with rod power is likely due to the thin oxides associated with
OSP and the resulting small thermal gradients.
Q1:—What was the method used for hydrogen measurement?
Authors’ Response:—The hydrogen content in the samples was determined by

the Hot Vacuum Extraction (HVE) method. This method is based on the fact that
the amount of hydrogen released from the sample upon heating leads to a propor-
tional increase in the pressure in a closed volume.
Q2:—Is the TEM located within a hot cell?
Authors’ Response:—The TEM is located outside the hotcell. The TEM foil
specimen was prepared in a hotcell.
Q3:—I believe the effect of hydrides may not manifest at tension test tempera-
tures at 385 C as this temperature is above the ductile-to-brittle transition tempera-
ture of Zr-hydrides. Please comment.
Authors’ Response:—We are dealing with the mechanical properties of a com-

posite made of metallic ligaments and hydride precipitates. The mechanical proper-
ties of the composite depend on the mechanical properties of the two constituents
and geometrical arrangement of the two constituents. Ductile-to-brittle transition
temperature (DBTT) of the composite also depends on the transition temperature
of the two components. Transition temperature of the two components depends on
the effect of the radiation damage in addition to chemical composition of each com-
ponent. At 385 C, some hydrides may dissolve reducing the volume fraction of
hydrides. A lower volume fraction of hydrides may decrease the DBTT of the
composite.
Q4:—Was the fuel removed before carrying out metallography of the cladding?
Authors’ Response:—The fuel was not removed from the metallgraphic

specimen.
Q5:—What was the effect of irradiation to the higher burnup on the fuel-clad-
ding gap?
Authors’ Response:—There is no gap between fuel pellet and cladding based on

the metallography of the high burnup rods. The gap was closed at lower burnup.
Available information is not sufficient to determine the burnup when the gap
closed.
PAN ET AL., DOI 10.1520/STP154320130008 627
Questions from Philippe Bossis, CEA
Q1:—How many cycles are needed to reach "intermediate" oxide thickness

(> 30 lm) in the lower spans of the assembly?
Authors’ Response:—For Optimized ZIRLO cladding, the oxide thickness will

not reach 30 lm at lower spans even after three cycle of operation with burnup over
70 GWd/MTU. All OSP observations were located in the upper elevations of the
rod where the extent of nucleate boiling in the coolant and the measured cladding
waterside oxide thicknesses were highest. Alloys that are more susceptible to OSP
have very thin “maximum” oxide. OSP resistant alloys with intermediate
“maximum” oxide thickness could have very thin oxide thickness in the lower
spans of the assembly.
Q2:—Please provide tin content of Optimized ZIRLO.
Authors’ Response:—The tin content for Optimized ZIRLO cladding is 0.6% -

0.8%.
Question from B. K. Shah, BARC Mumbai:—What is the maximum burnup

level that the US NRC has cleared the use of Optimized ZIRLO clad?
Authors’ Response:—The licensed burnup level for Optimized ZIRLO clad is 62

GWd/MTU (peak rod) in the US.
V. N. Shishov,1 V. A. Markelov,2 A. V. Nikulina,2

V. V. Novikov,2 M. M. Peregud,2 A. Yu. Shevyakov,2
I. N. Volkova,3 G. P. Kobylyansky,3 A. E. Novoselov,3
and A. V. Obukhov3
Corrosion, Dimensional Stability

and Microstructure of VVER-1000
E635 Alloy FA Components
at Burnups up to 72 MWday/kgU
Reference
Shishov, V. N., Markelov, V. A., Nikulina, A. V., Novikov, V. V., Peregud, M. M., Shevyakov, A. Yu.,
Volkova, I. N., Kobylyansky, G. P., Novoselov, A. E., and Obukhov, A. V., “Corrosion, Dimensional
Stability and Microstructure of VVER-1000 E635 Alloy FA Components at Burnups up to 72
MWday/kgU,” Zirconium in the Nuclear Industry: 17th International Symposium, STP 1543,
ABSTRACT
The use of irradiation-resistant E635 alloy for fuel rod cladding and skeleton
components of the VVER-1000 fuel assemblies advanced (FAA) has ensured the
stability of the fuel assembly (FA) geometrical dimensions, minimized their
bending and distortion, and increased the resistance of fuel cladding to shape
changes at burnups to 72 MWday/kgU after 6 years of operation. Post-irradiation
investigations of the VVER-1000 FAA components (fuel rod cladding, guide
thimbles, central tube, and rigid angles) of E635 alloy show that, in terms of their
major operational characteristics, additional margin to the design limits remain
following six 1-year cycles. The geometrical parameters, oxidation, hydrogen
absorption, tensile properties, and microstructural state of the components did
not reach values that would inhibit their further performance. The oxide film
Manuscript received November 8, 2012; accepted for publication April 18, 2014; published online June 17, 2014.
1
JSC, VNIINM, Moscow, Russia (Corresponding author), e-mail: shishovv@bochvar.ru
2
JSC, VNIINM, Moscow, Russia.
3
JSC, NIIAR, Dimitrovgrad, Russia.
4
Hyderabad, India.
Copyright V
SHISHOV ET AL., DOI 10.1520/STP154320120146 629
thickness and the hydrogen content of E635 alloy are correlated along the core
length, increase with burnup, and are a function of both temperature and
neutron fluence. The oxide film thickness increases up to 80 lm at an assembly
height of 3100 mm for cladding and is less than 42 lm on the outer surface of
the guide thimbles and rigid angles. Hydrogen content in the cladding is less
than 0.03 % and reaches 0.06 % around the bend of the rigid angles.
Transmission electron microscopy (TEM) studies of dislocation structure, and
chemical and phase composition of E635 alloy components revealed
microstructural characteristics influenced by temperature and neutron
irradiation. The Laves phase precipitates and the matrix composition change
caused by depletion of iron from SPPs in the irradiated fuel cladding, guide
thimbles, central tube, and rigid angles as a result of changes in temperature and
neutron fluence, and determine their mechanical properties and shape changes
resistance.
Keywords
neutron irradiation, E635 alloy, fuel rod cladding, fuel assemblies, oxide film
thickness, hydrogen, microstructure, precipitates, dislocations, electron
microscopy, growth, examination
Introduction
Today, the strategy of improving the safety and efficiency of power reactor cores is
most urgent for evolution of nuclear power in Russia. Increased fuel burnups and
operating times are limited by properties of zirconium materials. Oxidation, hydro-
gen absorption, and dimensional changes (growth, creep, bending) limit the service
life of the fuel assembly (FA). The problem is being solved by both changes in FA
designs and materials. In VVER-1000, several FA designs have been employed [FA,
FA-2, FA-2M, fuel assemblies advanced (FAA)] since 1995. The structural compo-
nents [guide thimbles (GT), central tube (CT), and rigid angles (RA) of the FAA
skeleton] and fuel rod cladding are produced from an irradiation-resistant E635
alloy that features high strength and high resistance to shape changes and
corrosion.
E635 alloy components operate in more than 687 FA-2 of VVER-1000 in four
units of Balakovo NPP [1] and in more than 2500 FAA of VVER-1000 in 17 units
of NPP in Russia, Ukraine, and Bulgaria [2]. For FAA, 450 have operated for 4 years,
whereas 32 FAA have operated for 5 years and longer to reach the maximum
burnup of 72 MWday/kgU per fuel rod [3–5]. The FA mechanical strength and
rigidity are determined by the properties of the skeleton. The skeleton shape change
results from irradiation-induced growth of the guide tubes and irradiation creep
during operation. Hence, the properties of FA component material play quite an
important role in their safe performance and reliability. The use of E635 alloy as an
assembly skeleton material gives significant improvement to reducing FA bowing
(the bending decreased by more than four times). This paper characterizes the cor-
rosion performance, tensile properties, and structural phase state of E635 used as
the material for the FAA VVER-1000 structural components and fuel cladding in the
Kalinin NPP for a 6-year operation with the maximum burnup 72 MWday/kgU
[5]. The E635 components with heat flux (cladding) and without heat flux (GT, RA,
CT, welded joints) were investigated.
Materials and Experiments

The skeleton is formed by six RAs that form the vertices of the hexagonal array, 15
SGs, 18 GTs, and one CT. The SG connection with the RAs is made by contact-spot
weld. GTs are not strictly linked to other elements of the frame and can slide in the
SG cells. A sketch of the FAA design is shown in Fig. 1. Fuel cladding, guide thim-
bles, the central tube and rigid angles are made from E635 (Zr-1 % Nb-1.2 %
Sn-0.35 % Fe) alloy, whereas the spacer grids are made from the E110 (Zr-1 % Nb)
alloy. GT3 and RA1 are located near fuel rod 21 with a maximum burnup of
72.4 MWday/kgU and GT9 and RA4 are near the fuel rod 302 with a minimum
burnup 54.3 MWday/kgU. Fuel rod 138 has an intermediate burnup of
67.2 MWday/kgU. Burnup distribution along the fuel rod length is characterized by
a plateau where the burnup is approximately constant over the axial elevation or
height from 500 mm to 3000 mm (Fig. 2).
The measurements indicate a high dimensional stability of FAA structural
components. The maximum bending value is 5.9 mm at axial elevation 2570 mm.
For FAA fuel rods with the E110 alloy cladding after 6 years of operation, the
assembly bending was 6.6 mm. CT is free of mechanical stress, so its dimensional
change is a result of irradiation growth. FAA guide tubes elongation during opera-
tion is 0.11 %–0.12 %, which is almost two times smaller than that of CT. This
means that the GTs were under compressive stress resulting in irradiation creep.
The cladding surface over the entire FAA height has a uniform grey–white
color. The oxide film peeling is not observed. The surface color becomes lighter
with increasing axial elevation. The rods (21, 138, and 302) were selected from a
fuel assembly in the direction of the maximum burnup gradient (from 48 to
72 MWday/kg U) and specimens were taken at four height levels with different
neutron fluence and temperature. The oxide film thickness was measured on the
fuel rod cladding by the eddy current method. The measurements were performed
along the length of fuel rods in increments of 3 mm in two opposing azimuthal ori-
entations. Metallographic studies were performed on transverse and longitudinal
sections.
The mechanical properties of E635 components were determined by testing
ring tensile specimens 3.0 mm in height at both room and operating (350 C) tem-
perature. The tensile strength (rB), yield strength (r0.2), total (do), and uniform (dp)
elongation were determined.
The hydrogen content in the fuel rod cladding was determined with an ELTRA
OH 900. Traditionally, the hydrogen content in fuel cladding was determined on
samples containing an oxide film. To compare results of a quantitative gas analysis
FIG. 1 Schematic diagram of FAA (a); and distribution of fuel rod enrichment, guide tube and central tube location, and faces (b). The fuel rods 21, 138,
and 302 are shown with arrows.
SHISHOV ET AL., DOI 10.1520/STP154320120146

631
FIG. 2 The fuel rods 21, 138, and 302 burnup height distribution.
with the degree of hydrogenation of the samples, as determined by metallographic

images, hydrogen must be determined in the samples with the oxide film removed.
Our results confirm the data on the higher hydrogen content in the samples with
the oxide film [6–11].
For transmission electron microscopy (TEM) studies, 3-mm-diameter discs
were cut out, mechanically polished to a thickness of 0.1 mm, and twin jet
thinned using a TENUPOL-5 at a temperature of ( 12 C to 15 C) in an electro-
lyte of 35 % C6H14O2, 5 % HClO4, and 60 % CH3OH. Thin foils and extraction car-
bon replicas were studied in an electron microscope (JEM-2000FXII) using an
EDAX energy dispersive X-ray spectrometer.
Results
Elongation of rods 302, 138, and 21 with fuel burnup 54.3, 67.2, and 72.4 MWday/kgU,
respectively, is shown in Fig. 3. “Fuel-cladding” gap of the fuel rods is reduced with
increasing burnup. A small gap exists in a cold state even in rods with a maximum
fuel burnup (Fig. 4).
The oxide film thickness on the outer cladding surface, GT3, GT9, and CT
increases with height up to 2800–3000 mm, and then decreases for cladding (Fig. 5).
The minimum oxide thickness is observed at the plenum level (3530–3750 mm).
There is a good agreement between the oxide film thickness determined by eddy
current and metallography.
Hydrogen absorption by the cladding, GTs, CT, and rigid angles occurred dur-
ing operation. Hydride precipitations of predominantly tangential orientation are
unevenly distributed over the cross section of the cladding; in fuel rod 21 with max-
imum burnup in areas 500–3300 mm, hydrides are concentrated to a greater extent
along the outer surface (Fig. 6). Weight fraction of hydrogen in the claddings does
not exceed 0.03 % and increases with height, correlating with the oxide film
thickness.
FIG. 3 Elongation of rods 21, 138, and 302, depending on burnup.
FIG. 4 Dependence of the “fuel-cladding” gap on the maximum fuel burnup.
FIG. 5 The oxide film thickness of the fuel rod 138, GT3, and GT9 and CT height
distribution.
FIG. 6 Microstructure of the fuel rod claddings cross section (height altitude
3000 mm) with burnup: (a) 54.3 MWday/kgU (rod 302); (b) 67.2 MWday/kgU
(rod 138); and (c) 72.4 MWday/kgU (rod 21).
TEM investigation of cladding microstructure of fuel rod 21 has shown that the
second phase precipitates (SPPs) (Fig. 7) are uniformly distributed over the grains.
The precipitate composition depends on their size (Figs. 7(c) and 7(d)). The smaller
the particle, the less iron it contains, the Laves phase SPPs are transformed into pol-
ycrystalline particles of b-Nb type. The size and number of these particles depends
on the height altitude coordinates of the sample.
The maximum polycrystalline particles of b-Nb size are detected in a sample
from the midwall of the cladding (Fig. 7(f)). In the plenum region, SPPs contain
traces of stacking faults (Fig. 7(e)), which indicate the absence of transformation. A
small number of fine (5 nm) radiation-induced precipitates (RIPs) are discovered.
Their alignment is not observed (Fig. 8).
The fuel rod 21 cladding dislocation structure is characterized by the presence
of dislocation loops a-type, and c-dislocations near particles (Fig. 9(d)). In the ple-
num, c-dislocations were not detected (Fig. 9(b)) and a-loop alignment is weak
(Fig. 9(a)). In all samples, loops are arranged in rows along the direction [01.0]
under the reflecting vector [01.1] (Fig. 9(c)).
The tensile strength (rB), yield strength (r0,2), total (dtot), and uniform (duni)
elongation of cladding are high at room and operating temperatures (Table 1). The
material retains reserves of plasticity.
Investigation of the fracture surface of ring samples from the fuel rod claddings
after tensile tests showed mostly ductile fracture mode. Near the inner surface, the
brittle transgranular fracture mode dominates. Cracks can be formed at locations of
tangentially oriented hydrides (Fig. 10).
The results of TEM studies of the structure GT, CT, angles, and weld joints
near the maximum burnup of fuel elements are shown in Figs. 11–16. The precipitate
composition dependence on their size shows that SPPs in GT contain less Fe com-
pared to the fuel cladding, and at elevation 900 to 3000 mm (SG4 to SG12) Fe is
FIG. 7 The Laves phase precipitates in the fuel rod 21 cladding at the plenum level (a, c,
e) and 3000 mm (b, d, f). (c, d) The particle composition dependence on the
size; (f) blocks b-Nb; and (g, h) the matrix composition dependence (Nb, Fe)
with elevation in rods 21 and 302.
almost completely absent, even in large particles. The presence of dark rings
around the particles (Fig. 11(a)) may be caused by changes in the matrix composi-
tion as a result of segregation of impurities on c-loops. At the level 3500 mm,
c-dislocations are practically absent. Irradiation-induced single fine precipitates
FIG. 8 Radiation-induced precipitates (RIP) in the fuel rod 21 cladding at the level of
plenum (a), and 3000 mm (b).
have been discovered. CT microstructure and composition studies have shown

that a-loops are arranged in rows and there is a minimum of c-dislocations den-
sity detected at 3000 mm level (Fig. 12). The Fe in the particles is almost entirely
absent.
FIG. 9 Fuel rod 21 cladding dislocation structure at the plenum level (a, b), and
3000 mm (c, d): (a, c) a-loops; (b, d) c-dislocations and precipitates.
TABLE 1 Tensile properties of E635 alloy cladding in the transverse direction.
Time (Years) Ttest ( C) rB (MPa) r0,2 (MPa) dtot (%) duni (%)
3 20 600–700 525–630 7.4–11 2.8–3.9

380 445–465 400–410 14–21 3.2–3.9
6 20 720–790 680–760 4.9–9.1 2.0–3.8
380 530–580 500–540 8–15 2.4–4.2
Rigid angle microstructure and particles/matrix composition have shown that,

at the level 3500 mm, most of Fe is in Laves phase particles, and Fe is almost absent
in the SPPs at the level 3000 mm and lower. In the center of large and medium-
sized particles at the level 3500 mm, the crystal lattice has remained unchanged.
Individual RIPs are observed in all samples except the sample at the plenum level.
Dislocation structure is characterized by a-dislocation loops. The ordering of loops
at the plenum is weak. A few c-type loops were found near the particles at elevation
3000 mm.
The results of the analysis of Sn, Nb, and Fe distribution in welded joints SGs–RAs
cut out at the level 2000 mm are shown in Fig. 14. SGs and RAs are made of different
alloys: SGs are an E110 alloy, and RAs are an E635 alloy. The element-distribution
investigation was carried out across the sample (from SG through the weld joint to
the rigid angle).
The analysis shows that tin from E635 alloy does not diffuse into E110 alloy
and iron from E635 alloy extends to a greater distance into E110 alloy. TEM study
of welded joints (Fig. 15) revealed a martensitic structure, typical of rapid cooling.
SPPs in the material of claddings, GT and CT, were dissolved by heating during
welding. Highly dispersed RIPs were found (Fig. 15(b)). The dislocation structure of
FIG. 10 Fuel rod 21 cladding structure at elevation 2000 mm: (a) near the inner
surface—brittle fracture; and (b) near the outer surface—mostly ductile fracture
mode.
FIG. 11 (a) SPP in GT3 at elevation 900 mm; and (b) particle size elemental
composition.
all samples is characterized by ordered a-type dislocation loops and c-dislocations

(Fig. 16).
Discussion
IRRADIATION SHAPE CHANGES
The effect of irradiation-induced growth (anisotropic dimension change at a con-
stant volume of material) is observed in zirconium alloys under irradiation in the
absence of stress. An irradiation creep process develops when load is applied to the
material under irradiation. During the reactor performance, only CT is free of
mechanical stress, so its dimensional change is a result of irradiation growth and is
in good agreement with irradiation-induced growth of E635 alloy samples
FIG. 12 CT dislocation structure at elevation 3000 mm: (a) a-loops; and (b)
c-dislocations.
FIG. 13 Content of Nb and Fe in the matrix at height altitude: (a) GT3; (b) GT9; and (c) CT.
SHISHOV ET AL., DOI 10.1520/STP154320120146

639
FIG. 14 Distribution of the elements across the welded joint core (SG E110 and RA
E635).
irradiated in BOR-60 (Fig. 17), and corresponds to the fluence range

2.1–2.6 1026 m 2 (E > 0.1 MeV) in the VVER-1000 (17–20 dpa).
Analysis of the VVER-1000 fuel assemblies element geometric parameters
shows that, at the average fuel burnup 55–65 MWday/kgU (six fuel cycles), elonga-
tion of rods with E635 alloy cladding is less than that of the fuel rod cladding from
E110 alloy (Fig. 18).
The fuel rod outer diameter with E110 alloy cladding because of creep under
the influence of differential pressure “coolant–fuel rod” decreases up to the burnup
55 MWday/kgU; at higher burn-ups, a change occurs in the opposite direction—
the effect of the so-called “inverse” deformation [1,2]. Figure 19 presents data on the
cladding diameter change as a function of burnup. The graph shows that at the
same fuel burnup the gap between the E635 alloy cladding and fuel must be main-
tained even at very high burnup. Because the gap “fuel cladding” in the E635 fuel
rods is preserved for a longer time than with E110 cladding, then, because of a high
FIG. 15 Microstructure of the weld core at level 2000 mm (a), and fine RIPs at 900 mm
(b).
FIG. 16 a-loops (a), and c-dislocations (b) in the welded joint at the level 3000 mm.
strength of the E635 alloy, the effect of “inverse” strain is hardly possible even for
the burnup 80 MWday/kgU.
Elongation of rods (mainly because of the process of anisotropic cladding irra-
diation creep), increasing with the fuel burnup, is about two times higher than the
CT and rigid angles elongation. In accordance with the concepts of irradiation
creep of anisotropic (because of texture) materials under the influence of external
coolant pressure and irradiation, the diameter of fuel rod cladding decreases,
whereas their length increases. The FAA guide thimbles during operation elongated
to 0.10 %–0.14 %, which is almost two times smaller than the CT and rigid angles.
FIG. 17 Irradiation-induced growth IIG of E635 alloy versus fluence (black line) under
BOR-60 irradiation at 320 C (the fluence is given taking into account the
VVER-1000 reactor spectrum) and the values of the elongation of CT of FAA as
a result of operation in the VVER-1000 reactor (filled diamonds).
FIG. 18 Elongations of fuel rods with E635 and E110 alloy cladding as a function of mean
burnup.
This means that the GTs were under the influence of compressive stresses. After
6 years of operation, the GT length because of irradiation growth was greater than
after 4 years of operation [8].
OXIDATION AND HYDROGEN UPTAKE

The temperature of the GT, CT, and RA corresponds to the temperature of coolant
and ranged from 286 C to 320 C from the bottom to the top of the FAA. Fuel clad-
ding temperature at the “metal-oxide” surface could be about 350 C. Estimates of
the height distribution of the surface temperature of the cladding and GT are shown
in Fig. 20. The oxide film thickness in the material increases with elevation to a
height of 2900–3100 mm with its subsequent decrease (Fig. 5). In E635 alloy clad-
dings, there is a noticeable increase in the oxide film thickness with increasing
height altitude with a subsequent decrease in the plenum. There are no signs of
FIG. 19 Dependence of the outer diameter of fuel rods with E110 and E635 alloys
cladding on burnup [2].
FIG. 20 Representative height distribution of the cladding (filled squares) and guide
thimbles GT (filled diamonds) surface temperature in VVER-1000 reactor.
E635 nodular oxidation, so the corrosion state of products from this alloy is easily
predictable for a given operating time.
The analysis of the corrosion condition of cladding, GT, CT, and angles, taking
into account results of studies of other assemblies with E635 alloy, shows that the
kinetics of oxidation can be approximated by linear dependence, which is influ-
enced by differences in conditions of the test of fuel rod claddings, GT, CT, and
RA, as shown by the changes in the oxide thickness of cladding and structural ele-
ments of FAA (Fig. 21). The experimental data can be used in advanced models of
zirconium materials corrosion.
The oxidation of FAA components was accompanied by the absorption of
hydrogen, which was dissolved in the alloy material, and precipitated in the form of
FIG. 21 The maximum oxide thickness versus time of operation of E635 alloy
components as irradiated in VVER-1000: filled circles, fuel rod claddings;
squares, GT; filled triangles, rigid angles.
FIG. 22 Temperature dependence of hydrogen content in the E635 GT (filled squares

and diamonds) and CT (filled triangles).
zirconium hydride during cooling. The hydrogen content, in the first approxima-
tion, was linearly dependent on the oxide film thickness in the E635 alloy and
reached a maximum in the rigid angles of the frame, which takes into account the
oxidation of both sides. Hydrogen content in GT and CT is plotted in Fig. 22 as a
function of operating temperature.
Weight fraction of hydrogen in guide thimbles did not exceed 0.03 %, taking
into account the bilateral oxide film of 80 lm. In the rigid angles near the bend,
hydrogen content reached 0.06 % at thickness of the bilateral oxide about 100 lm
after 6 years of operation (Fig. 23(c)).
Figure 24 shows results of hydrogen content calculation through the hydrides
number density in the metallographic images and the measured values of hydrogen
content in E635 alloy FAA elements and E110 alloy spacer grids.
Corrosion phenomena occurring in zirconium materials in the reactor core
represent a complex process that depends on a combination of many factors.
FIG. 23 Hydrides in a rigid angle on the bend after 3(a), 4(b), and 6(c) years of
operation.
FIG. 24 Dependence of hydrogen content on the oxide film thickness in the E635 alloy
structural elements (GT, CT, and angles taking into account bilateral oxidation).
Measurements: triange, rod; diamond, GT; circle, CT; square, angle; calculated
from micrograph: filled triangle, rod; filled diamond, GT; filled circle, CT; filled
square, angle.
Investigations of the FAA corrosion state showed that the degree of oxidation and
hydrogenation after six fuel cycles is not a limiting factor for further longer-term
performance of E635 products. Weight fraction of hydrogen in the claddings does
not exceed 0.03 % and correlates with the oxide film thickness (Fig. 24).
MICROSTRUCTURE
TEM studies have shown that the E635 alloy microstructure as part of FAA
depends on irradiation conditions (burnup, height-altitude fluence distribution,
and temperature). The size of a-type loops is about 20 nm and, the number density
slightly varies with elevation and product type. The increase in loop size is corre-
lated with the temperature of the material and increases slightly in height. Determi-
nation of the structural component phase (a-matrix, b-Nb, and Laves phase) and
chemical composition (the content of Zr, Nb, Fe, Sn) is of increased interest as it is
directly linked with exposure to neutron irradiation and temperature on formation
of oxide film and hydrogen absorption. Analysis of the matrix elemental composi-
tion showed (Fig. 7) that the top of the fuel rod cladding contains less Fe (up to
0.2 wt. %) than the middle and lower parts, where it reaches the level of 0.3–0.4 wt.
%. The maximum iron content is observed along most of the fuel element length,
up to elevation 3000 mm. Fe content in the GT and CT matrix remains unchanged
in height, but Nb content changes. It is clear that levels of these elements in the
matrix are because of the processes of redistribution between the second phase par-
ticles and solid solution during irradiation. E635 alloy in the initial state contains
the Laves phase Zr(Nb, Fe)2 with composition of roughly equal proportions of Zr,
Nb, and Fe [6,7]. The Laves phase during irradiation undergoes changes with deple-
tion of iron, which goes to the matrix. When iron content in a particle is reduced to
a certain value, the lattice transformation occurs. That is, the Laves phase (hcp) is
transformed into b-Nb (bcc). Because the surface layers loose iron more rapidly
than the center, the transformation begins earlier at the surface. With the growth of
the area depleted in iron, b-Nb crystallites grow in size and occupy the entire vol-
ume of the particle and the transformation is over. This fact was noted earlier; how-
ever, data on the structure in the early stages of transformation, taking temperature
into account, were not discovered. b-Nb particles in E110 alloy during irradiation
also change: their number density decreases, Nb content in SPPs decreases, and the
matrix oversaturates with niobium [6,7]. This is accompanied by release of fine
irrradiation–induced niobium–rich particles (presumably b-Nb) from solid solution
and subsequent niobium depletion of the matrix. In irradiated E635 alloy, the solid
solution becomes enriched by the iron from the Laves phase. The solid solution
contains a larger number of alloying elements (tin, niobium, and iron); perhaps,
therefore, the E635 alloy contains much fewer radiation-induced fine particles than
E110. Their average size is about 5 nm and, an ordering of their location was not
observed. The difference in the matrix composition is correlated with the precipi-
tates composition. The more that iron is contained in the particles, the less it is in
the matrix and vice versa. The temperature in the mid-section of the cladding
(350 C) is higher than in CT, GT, and SG (coolant temperature) and increases
with elevation. The matrix and SPP composition change under neutron irradiation
and is defined by the temperature parameters and the magnitude of the neutron flu-
ence (or burnup or damage dose). Dependence on particle size can also be observed
and is more pronounced for small particles up to 100 nm. The greatest amount of
iron stored in the particles is observed at 3500 mm, where the fluence is reduced,
but the temperature and the oxide film thickness are maximum. The maximum
amount of iron (30 %) is in particles at the plenum level with the lowest fluence.
The particle elemental composition in the fuel cladding, GT, and CT is very differ-
ent with iron content. The particles in the fuel cladding contain up to 10–15 at. %
Fe, whereas in GT and SG, Fe content reaches only a few %. This is, apparently,
because of the difference in temperature. GT, SG, and CT are irradiated at lower
temperature than the fuel cladding, so iron atoms cannot go back to the particle. In
the fuel cladding under irradiation at higher temperature, diffusion mobility of
atoms prevails over dynamic processes of collisions, and it is sufficient to reverse
the return by diffusion of most of the iron atoms. A marked depletion occurs only
in small particles.
High resistance to irradiation shape changes of the FAA elements made of
E635 alloy correlates with high mechanical properties. Guide tube strength charac-
teristics (r0,2 and rB) are at a high level, exceeding strength properties of fuel rod
cladding at room temperature. This is because of incomplete recrystallization of GT
and CT unlike the fully recrystallized cladding and higher iron content in solid
solution. The tensile strength (rB) and yield strength (r0.2) of cladding tubes, GTs,
and CT correlate with iron enrichment of the matrix after irradiation. Retaining of
the FAA structural elements ductility is of great importance. The fuel claddings and
GTs have a uniform elongation of more than 4.9 % even at room temperature.
Taking into account that these elements for 6 years of operation have significantly
oxidized and hydrogenised, and that there have been some conservative estimations
of ductility characteristics during the tensile ring test (in tension to straighten the
ring are also subjected to bending), this reserve of ductility allows continued utiliza-
tion of the material.
Conclusion
E635 alloy has confirmed high reliability as VVER-1000 assembly material for a
6-year-long performance to fuel burnup 72 MWday/kgU. Geometrical parame-
ters, corrosion conditions, and mechanical properties did not reach the values that
prevent their further exploitation:
• closing gap “fuel claddings” of fuel rods with claddings from E635 does not
occur;
• fuel rod elongation does not exceed 0.4 %; CT and rigid angle elongation is
within 0.18 % to 0.21 %; GT elongation does not exceed 0.11 % to 0.12 %;
• microstructural studies, determination of hydrogen content, and mechanical
properties of the cladding have shown that E635alloy retained sufficient corro-
sion and irradiation resistance; and
• the maximum thickness of cladding oxide film is 80 lm.
Hydrogen content in the claddings is not more than 0.03 %; it varies with eleva-
tion and correlates with the oxide film thickness and temperature. Weld joints
ensure operability of the FAA skeleton. Oxidation and hydrogenation of GT and
angles are not a limiting factor for long-term operation.
Strength characteristics of guide thimbles are practically unchanged with the
elevation and exceed the level of fuel rod cladding properties; their total elongation
exceeds 4.9 %.
The oxide film thickness and hydrogen content distribution in elements FAA at
height altitude are determined, first of all, by thermal irradiation effects (tempera-
ture and neutron fluence). Radiation damage of the structure and redistribution of
alloying elements during irradiation influence mainly resistance to shape changes
and mechanical properties.
References
[1] Vasilchenko, I. N. and Molchanov, V. L., “The Recent Advances and Achievements in
WWER-1000 Fuel Design Performance and Operation,” 8th International Conference on
VVER Fuel Performance, Modeling and Experimental Support, Helena Resort, Bulgaria,
Sept 27–Oct 2, 2009, pp. 181–185.
[2] Samoilov, O. B., Kaidalov, V. B., Falkov, A. A., Romanov, A. I., and Shishkin, A. A., “Results
of TVSA Fuel Assembly Development and 10-Year Operation in VVER-1000 Reactor
Cores. Development Trends,” 8th International Conference on VVER Fuel Performance,
Modeling and Experimental Support, Helena Resort, Bulgaria, Sept 27–Oct 2, 2009, pp.
191–200.
[3] Molchanov, V. L., “Nuclear Fuel for VVER Reactors: Current Status and Prospects,” 6th
International Conference on WWER Fuel Performance, Modeling and Experimental Sup-
port, Albena Congress Center, Albena, Bulgaria, Sept 19–23, 2005, pp. 28–39.
[4] Vasilchenko, I. N., Ryzhov, S. B., Dragunov, U. G., Kobelev, S. N., and Medvedev, V. S.,
“Trial Operation of TVS-2 at Balakovo NPP: Analysis of Results and Further Modern-
ization,” 6th International Conference on WWER Fuel Performance, Modeling and Exper-
imental Support, Albena Congress Center, Albena, Bulgaria, Sept 19–23, 2005, pp.
98–105.
[5] Volkova, I. N., Novoselov, A. E., Kobylyansky, G. P., Kostyuchenko, A. N., Kon’kov, V. F., Nov-
ikov, V. V., and Peregud, M. M., “E635 Alloy Corrosion in a VVER-1000 Reactors,” Proceed-
ings of the 19th International Conference on the Physics of Radiation Phenomena and
Radiation Material Science, Alushta, Crimea, Sept 6–11, 2010, p. 123.
[6] Shishov, V. N., Peregud, M. M., Nikulina, A. V., Pimenov, Yu. V., Kobylyansky, G. P., Novoselov,
A. E., Ostrovsky, Z. E., and Obukhov, A. V., “Influence of Structure–Phase State of Nb
Containing Zr Alloys on Irradiation Induced Growth,” 14th International Symposium on
Zirconium in the Nuclear Industry, STP 1467, ASTM International, West Conshohocken,
PA, 2006, pp. 666–685.
[7] Shishov, V. N., “The Evolution of Microstructure and Deformation Stability in Zr-Nb-
(Sn,Fe) Alloys Under Neutron Irradiation,” 16th International Symposium on Zirconium in
Nuclear Industry, STP 1529, ASTM International, West Conshohocken, PA, pp. 37–66.
[8] Kobylyansky, G. P., Novoselov, A. E., Obukhov, A. V., Ostrovsky, Z. E., Shishov, V. N., Nikulina,
A. V., and Markelov, V. A., “Radiation Damage E635 Alloy in Structural Elements of
VVER-1000 Fuel Assembly,” Problems of Atomic Science and Technology. Ser.: Phys.
Radiat. Damage Phys. Radiat Mater. Sci., Vol. 2, No. 93, 2009, pp. 57–68.
[9] Ramasubramanian, N., Perovic, V., and Leger, M., “Hydrogen Transport in Oxide and
Hydrogen Pickup by the Metal during out- and in-Reactor Corrosion of Zr-2.5Nb Pres-
sure Tube Material,” Zirconium in the Nuclear Industry, 13th International Symposium,
[10] Hermann, A., Wiese, H., Buhner, R., Steinemann, M., and Bart, G., “Hydrogen Distribution
between Fuel Cladding Metal and Overlying Corrosion Layers,” Proceedings of the ANS
International Topical Meeting on LWR Fuel Performance, Vol. 1, Park City, UT, April 2000,
pp. 372–384.
[11] Holt, R. A. and Ibrahim, E. F., “Factors Affecting the Anisotropy of Irradiation Creep and
Growth of Zirconium Alloys,” Acta Metall., Vol. 126, No. 8, 1978, pp. 1319–1328.
DISCUSSION
Question from Zoltán Hózer, Hungarian Academy of Sciences Centre for Energy
Research, Fuel and Reactor Materials Department, Hungary:—The E635 alloy seems
to have less corrosion resistance compared to E110. Do you think that the
improvement of mechanical behavior can compensate this effect?
Authors’ Response:—Alloy E635 really has less resistance to uniform corrosion

than alloy E110, but significantly exceeds E110 alloy in resistance to nodular corro-
sion. At the same time, E635 alloy has significant advantage over E110 alloy in
creep resistance and irradiation growth.
Questions from Srikumar Banerjee, BARC Mumbai Q1:—Have you noticed any
change in the volume fraction of the second phase particles (b phase and Laves
phase) as the composition of the second phase changes with irradiation?
Authors’ Response:—Under the influence of radiation, concentration of b-Nb

particles, including those formed during the transformation of Laves phase under
irradiation, slightly reduces.
Q2:—In case of volume fraction change of second phase particles, does it con-
tribute towards the dimensional change?
Authors’ Response:—The average size of the b-Nb precipitates increases in

general.
Question from Jean-Christophe Brachet, CEA, France:—You said that even for
the higher BU (72MWd/kgU), there was a residual gap. Does it mean that there is
no fuel bonding? Please compare with E110.
Authors’ Response:—There is no gap closure when using cladding from E635

alloy even with BU 72 MWd/kgU. Interaction of swelling fuel pellet with cladding
from E110 alloy occurs with BU 45 - 50 MWd/kgU.
Questions from Juan Garcia de la Infanta, ENSUA, Spain Q1:—Is the central
tube growth mainly due to irradiation growth. Have you considered a contribution
of hydrides on the central tube or guide tube growth?
Authors’ Response:—Hydride contribution to the deformation of irradiation

growth was not considered. Data from the elongation of the central tube and data
from irradiation growth of E635 alloy specimens which were irradiated in BOR-60
reactor correlate well. That indicates that there was no influence of hydrides.
Q2:—What is the hydrogen pickup fraction value of E110 and E635?
Authors’ Response:—Hydrogen pickup fraction value is 10 - 12 % for E110 and

6 - 8 % for E635 alloy.
Question from Antoine Ambard, EDF:—The gap is closed very late compared to
our PWR experience. Is it due to a difference between fuel design (pressure differ-
ence between primary coolant and inside rod), difference between pellet design,
larger initial gap (~ 100 lm in PWR rod), or more creep resistance of the alloy?
Authors’ Response:—Experience shows that the gap between the fuel pellet and
the cladding from E635 alloy in fuel rods is maintained through high creep resist-
ance of the alloy.
Question from David Schrire, Vattenfall, Sweden:—You showed the elemental

profile across the E110-E635 weld. Was this element distribution different after
irradiation compared to as-manufactured?
Authors’ Response:—Profile of the elements distribution across the welding core

was not considered before the irradiation.
S. K. Sinha1 and R. K. Sinha2
Corrosion and Hydriding Model

for Zircaloy-2 Pressure Tubes
of Indian Pressurised Heavy
Water Reactors
Reference
Sinha, S. K. and Sinha, R. K., “Corrosion and Hydriding Model for Zircaloy-2 Pressure Tubes
of Indian Pressurised Heavy Water Reactors,” Zirconium in the Nuclear Industry: 17th
ABSTRACT
Early generation of Indian pressurised heavy water reactor (PHWR) units—
MAPS-1and 2, NAPS-1 and 2, and KAPS-1 had used Zircaloy-2 pressure tubes.
Corrosion of the zirconium alloy pressure tube in the high temperature
(250 C–300 C) heavy water coolant flowing through it results in formation of an
oxide layer on its inside surface and evolution of deuterium (for its chemical
similarity with hydrogen, it will be described as hydrogen). A part of this
hydrogen is absorbed by the pressure tube material. Gradual build-up of
hydrogen causes degradation in the structural integrity of the pressure tube with
manifestations of either one or a combination of the nucleation and growth of
hydride blisters, hydride embrittlement at service induced flaw tip, and lowering
of fracture toughness of the material. Safety assessment of the operating
pressure tubes against these hydride induced degradation mechanisms requires
a conservative estimate of hydrogen concentration in each of these pressure
tubes. Although hydrogen ingress into a pressure tube during service may be
estimated from the material samples taken out from the inside surface of the
tube by sliver scrape sampling technique, such exercise is not feasible to be
Manuscript received November 26, 2012; accepted for publication April 18, 2014; published online
September 15, 2014.
1
BARC, Trombay, Mumbai 400085, India, e-mail: sunilks@barc.gov.in
2
BARC, Trombay, Mumbai 400085, India.
3
Hyderabad, India.
Copyright V
carried out on a large number of pressure tubes. Alternatively, the numerical

model for corrosion and hydrogen pickup developed using the database created by
the hydrogen measured in the bulk samples from the pressure tubes removed from
the different reactor units for material surveillance purposes can be used for
conservatively estimating the hydrogen pickup. The present paper describes the
methodology adopted for developing a numerical model for in-reactor corrosion
and hydriding of Zircaloy-2 material using data on oxide thickness and hydrogen
pickup generated from the pressure tubes removed from the operating Indian units.
Keywords
PHWR, Zircaloy-2, corrosion, pressure tube
Introduction
Early generation of Indian pressurised heavy water reactor (PHWR) units had used
Zircaloy-2 pressure tubes manufactured by either cold pilgering or cold drawing of
hot extruded Zircaloy-2 hollow bars. The cold drawn pressure tubes had been used
in units 1 and 2 of Rajsthan Atomic Power Station (RAPS) whereas the cold pil-
gered pressure tubes had been used in units 1 and 2 of Madras Atomic Power Sta-
tion (MAPS) and Narora Atomic Power Station (NAPS), and unit 1 of Kakrapar
Atomic Power Station (KAPS). Brief descriptions about these units are given in the
Ref. [1]. Units built after KAPS-1 use hot extruded and cold pilgered Zr-2.5 %Nb
pressure tubes. Corrosion of the zirconium alloy pressure tube with the high tem-
perature (250 C–300 C) heavy water coolant flowing through it results in forma-
tion of an oxide layer on its inside surface and evolution of deuterium (for its
chemical similarity with hydrogen, it will be described as hydrogen in the rest of the
paper). A part of this hydrogen is absorbed by the pressure tube material. Gradual
build-up of hydrogen causes degradation in the structural integrity of the pressure
tube as a result of manifestations of one or a combination of hydride blisters,
embrittlement at crack-tip, and lowering of fracture toughness of the material.
Zircaloy-2 pressure tubes in all the pressure tube type heavy water cooled and mod-
erated reactors world wide have been replaced with the stronger Zr–2.5 %Nb pres-
sure tubes which also happen to have lower hydrogen pickup. The last one where
the Zircaloy-2 pressure tubes are being replaced with Zr–2.5 %Nb pressure tubes is
the Indian PHWR unit KAPS-1.
Out reactor high temperature (300 C) aqueous corrosion of Zircaloy-2 material
has been discussed extensively by Refs. [2–7]. The corrosion kinetics has been con-
firmed to have three stages: pre-transition with cubic rate, transitory period with
alternate cycles of cubic rates and transitory periods, and a post-transition linear
rate [3,8]. The first transition takes place around 2–4 lm [2–9]. The transitory stage
extends for a period of nearly five years for the maximum temperature encountered
in Indian PHWRs [3]. During this period, the oxide weight gain has been found to
be 80–85 mg/dm2 [3]. Based on short term autoclave test data, Hillner [2] proposed
two stage corrosion kinetics comprising of cubic pre-transition and a linear
SINHA AND SINHA, DOI 10.1520/STP154320120183 653
post-transition for Zircaloy-2 material, and later on, based on the long term auto-
clave test data, proposed a second transition in the linear post-transition stage for
the oxide thickness greater than 30 lm [9]. In-pile corrosion of Zircaloy-2 material
has been dealt with by Johnson et al. [10,11], Hillner [12], and Urbainc et al. [8].
Johnson’s observations have been that the corrosion rate responds to changing
water chemistry until a certain oxide weight gain (200–250 mg/dm2) is reached.
Hillner’s [12] analysis of Shippingport Atomic Power Station fuel rod clad corro-
sion has revealed that transition from pre-transition cubic rate takes place at 20 mg/
dm2 of oxide weight gain after 200 days of in-reactor exposure and the linear post-
transition corrosion rate is observed after 1700 days (4.7 years) of exposure.
This indicates that the intervening transitory period could be as long as 1500 days
(4 years). Furthermore, oxide weight gain build-up beyond the transition weight
gain of 20 mg/dm2 has not been observed during the transitory period. Urbanic
analysis of corrosion behaviour of Zircaloy-2 pressure tubes of Pickering units
revealed existence of critical thickness of oxide (15–20 lm) for corrosion and
hydriding to occur at a much higher rate compared to one existing during transi-
tory stage. Localised water chemistry existing inside the pores of the thick oxide
layer becoming independent of external water chemistry has been cited as the rea-
son for this high rate in corrosion and hydriding.
The pressure tubes used in Indian PHWRs or CANDU4 reactors are steam
autoclaved to develop 1–2 lm of a black oxide layer to provide corrosion resistance
under in-reactor conditions. It should also be noted that removal of the first pres-
sure tube happened nearly after 5 years of operation. Oxide thickness data gener-
ated from this tube and the tubes removed subsequently represented corrosion data
pertaining to transitory stage and beyond. The transition from cubic rate which
might have occurred when the pressure tubes were in service was, therefore, not
seen in the data generated.
Monitoring of hydrogen ingress into pressure tubes as a result of aqueous cor-
rosion is done on regular basis so that any loss in structural integrity of the pressure
tube due to the hydrogen induced degradation mechanisms can be pre-empted.
This is accomplished by removing sliver scrape samples from the inner surface of a
selected group of pressure tubes by sliver scrape sampling tool and later analysing
them for the hydrogen content. In addition, a reasonably conservative analytical
model simulating corrosion and hydriding of the zirconium alloy under in-reactor
high temperature aqueous medium has been developed to help identification of the
vulnerable pressure tubes. The predicting capability of the analytical model has
been progressively improved based on the feedback from the measurements.
The first analytical model, HYCON-95 [13], developed for estimating the
hydrogen pick up in Zircaloy-2 pressure tubes of Indian PHWRs was based on
oxide thickness and hydrogen pickup data obtained from the five high flux pressure
4
CANDU is the trade mark of Atomic Energy of Canada Ltd.
tubes removed from CANDU units at Pickering site [8]. Pickering pressure tubes
data had been taken as the basis for the development of this model due to unavail-
ability of similar data for Indian Zircaloy-2 pressure tubes. HYCON-95 estimated
best-fit values of hydrogen pickup were multiplied with a factor “1.5” to have a con-
servative assessment. When hydrogen pickup data from the bulk samples/sliver
samples of the Indian pressure tubes became available in due course, this conserva-
tive multiplier was changed to 1.33 and then to 0.98.
Subsequently, data on oxide thickness and hydrogen pickup were generated
from the Zircaloy-2 pressure tubes removed at different points of time from the
operating units for post irradiation examination (PIE). Using these data, a correla-
tion following semi-empirical approach was developed in 1999 and put in the form
of computer code HYCON-99 [14] for day-to-day application. Although the devel-
oped correlation had the advantage of directly computing the hydrogen pickup
using the inputs such as operating history and channel specific temperature and
flux, the estimated hydrogen pickup at the maximum H-pickup location was found
to be highly sensitive to the coolant temperature. Evidently, this model would give
incorrect estimation of hydrogen concentration at the maximum H-pickup loca-
tions in the pressure tubes of the then operating three units (NAPS-1 and 2 and
KAPS-1) where the pressure tubes installed in the high flux locations might have
been undergoing corrosion in the linear post-transition domain. This observation
set the requirement of a better predicting tool in order to assess the safety of the
pressure tubes in these units in a reasonably conservative manner. The work being
described here was carried out as a part of life management activities pertaining to
Zircaloy-2 pressure tubes of Indian PHWRs.
This paper gives an overview of the development of the model and its suitability
to assess the hydrogen pickup in Zircaloy-2 pressure tubes of the concerned units.
Sampling Plan for Bulk and Sliver Samples

The pressure tube in a 220 MWe Indian PHWR is nearly 5 m long. Its ends are fur-
ther extended by 2 m long stainless steel end fittings. Hot coolant enters into the
pressure tube through one end fitting and leaves it through the other end fitting.
Hydrogen concentration build-up at an axial location in the pressure tube after
operating for a given period of time can be measured by analysing either the sliver
sample if the tube is still in service or the bulk sample if the tube is removed from
service. On the contrary, the oxide thickness data is generated from the bulk sam-
ples only. The bulk sample at a given axial location is collected from the 3 o’clock
position so as to obtain average value of the oxide thickness and the hydrogen con-
centration at that axial location. The sliver sample at a given axial location in an
operating pressure tube is collected from the 12 o’clock position due to the con-
straints in designing sliver sampling tool and its in situ sample collection facility for
the underwater operation. With respect to the pressure tube axis, 12 o’clock is
located at the top and 3 o’clock is oriented at 90 from the 12 o’clock position.
The weight of the bulk sample and the sliver sample for hydrogen measurement
by Differential Scanning Calorimetry (DSC) technique is usually kept between
100–150 mg. H-concentration measurements by Hot Vacuum Extraction Quadru-
pole Mass Spectrometry (HVEQMS) technique requires 20–50 mg of sample. Axial
locations along the length of pressure tube from where the bulk samples and the
sliver samples are obtained are given in Table 1.
Hydrogen Concentration Measurement

Techniques
Techniques employed to measure hydrogen concentration in the irradiated samples
are DSC and HVEQMS [15–17]. DSC gives (H) and (D) collectively whereas
HVEQMS provides (H) and (D) separately. The relative standard deviation for
DSC is 61 % whereas that for HVEQMS is 63 %. Both the instruments are cali-
brated using low concentration (10–15 ppm of hydrogen) primary zirconium stand-
ards and high concentration (25–100 ppm hydrogen) secondary zirconium
standards. Both the instruments are also calibrated with respect to deuterium
charged secondary zirconium standards (1–25 ppm) as well. The results of analysis
of the standards indicate that there is no significant difference between the mea-
surement and the recommended values in the certified standards. However,
HVEQMS analysis of zirconium alloy samples charged with hydrogen and deute-
rium of varying composition revealed that the results were within 10 % of composi-
tion of the gases employed in charging. This accuracy of measurement is derived
without giving due consideration to kinetics of absorption of (H) and (D) in zirco-
nium alloy [16]. The accuracy of measurement in DSC technique is expected to be
within 5 % as the calibration curve (Terminal Solid Solubility or TSS curve) used for
determining hydrogen concentration from the given TSS temperature has been gen-
erated using secondary zirconium alloy standards containing known amount of
hydrogen determined by the inert gas fusion technique with an accuracy of 3 % [17].
Oxide Thickness Measurement

Oxide thickness is measured in the bulk samples mounted in cold-setting resin and
polished to 1 lm surface finish. Ten measurements are carried out at different loca-
tions in the same sample and average value is reported for that location.
TABLE 1 Axial locations of the bulk samples and the sliver samples.
Sample Locations (mm) as Measured from Inlet

Type End of the Pressure Tube
Bulk 500, 1000, 1500, 2000, 2500, 3000, 3500, 4000,

4300, 4800, 5100
Sliver 800, 1200, 2000, 3000, 3800, 4800
Oxide Thickness and Hydrogen Pick-Up

Data for Indian Pressure Tubes
Oxide thickness and deuterium pickup data generated during PIE of eight Zircaloy-
2 pressure tubes removed from different units of Indian PHWRs after operating for
a varying length of time were used in the development of model. Five of the eight
pressure tubes had shown post-transition corrosion and hydriding behaviour
towards the outlet end. These five pressure tubes belonged to RAPS-2. The transi-
tory stage oxide thicknesses did not exhibit any significant variation with respect to
the operating periods of these pressure tubes and the axial locations along the
length of a pressure tube.
Table 2 gives range of values for oxide thickness and total H-conc. as observed
in these pressure tubes. The oxide thickness measured at different axial locations
along the length of the pressure tubes are given in Table 3.
Methodology for the Development

of Correlation
The oxide thickness data obtained from the removed Zircaloy-2 pressure tubes of
Indian PHWRs pertain to the transitory stage and post-transition linear stage. The
data indicates that transitory stage continues till the 8–12 lm (Figs. 2–5) of oxide
thickness. This range is similar to those observed by Peters [3] in long term auto-
clave tests and by Johnson et al. [11] in advanced test reactor (ATR) experiments.
In his model, Peters [3] neglected pre-transition part and modelled the transitory
TABLE 2 Summary of post irradiation examination data available on Indian Zircaloy-2 pressure
tubes.
Full Power Years Hin (ppm) Numberb Oxide Total Hydrogen Conc.
Tube (FPYs) at which Tube HVEQMS Value (Avg. Thickness Range Range Measured
Numbera was Removed along Length) (lm) (ppm)
M2N10 4.84 12 6–7 10–20

M1P13 6.24 12 7–13 19–25
M1O02 7 8.5 6–9 10–20
R2K07 8.25 18 5–15 20–69
R2J07 8.5 23.5 6–19 33–92
R2J10 8.5 19.2 6–23 22–66
R2E10 8.5 21 6–16 34–104
R2K12 8.5 21 5–26 40–91
a
In Tube No. format TUVMN, T stands for Reactor (M for MAPS, R for RAPS); U stands for unit
No. (1 for unit No. 1, 2 for unit No. 2); VMN stands for the tube location in the reactor core.
b
Initial hydrogen (Hin) was not available in the manufacturing record. It was measured by hot vac-
uum extraction quadruple mass spectrometry (HVEQMS) technique in the samples removed from
different locations along the length of pressure tube and then subsequently averaged.
TABLE 3 Details of the oxide thickness data (lm) at the different axial locations in the examined pressure tubes.
Axial Location (mm) for Oxide Measurement
Tube
Number Years 500 1000 1500 2000 2500 3000 3500 4000 4300 4800 5100
M2N10 4.84 6 6 6 7 5 7 6
M1P13 6.24 7 9 10 10 8 7.5 7.8 7 7 13 10
SINHA AND SINHA, DOI 10.1520/STP154320120183

M1O02 7.0 7.2 6 7 6 7 7 8 7 9
R2K07 8.25 6 9 5 8 10 15
R2J07 8.5 6 7 19 15 17 13
R2J10 8.5 6 7 13 23 21 9
R2E10 8.5 8 6 6 4 5 12 12 16 7
R2K12 8.5 6 5 8 8 7 13 17 18 26 17 8
657
stage using a rate constant which linearly increases with oxide thickness and finally
becomes equal to rate constant of post-transition linear corrosion. In the present
paper, it is proposed to simulate the pre-transition and the extended transitory
stage by cubic rate constant and the post-transition stage by linear rate constant.
The basis for using the cubic correlation for the pre-transition and the extended
transitory stage of corrosion is the fact that the cubic correlation can efficiently sim-
ulate the increasing corrosion rate existing during pre-transition and gradually
decreasing rate encountered during extended transitory stage. The characteristics of
a cubic correlation are shown in Fig. 1.
The characteristic equations for the transitory and the post-transition stages of
corrosion in autoclave conditions are given below.
(1) S3t ¼ Kt t
(2) Kt ¼ At ExpðQt =RTÞ
(3) Spost ¼ Kpost t
(4) Kpost ¼ Apost ExpðQpost =RTÞ
where:
St ¼ transitory stage oxide thickness (lm),
Kt ¼ transitory stage cubic rate constant,
FIG. 1 Characteristics of cubic correlation.

At ¼ constant in transitory stage rate constant,

Qt ¼ transitory stage activation energy (J/mole),
Spost ¼ post-transition oxide thickness (lm),
Kpost ¼ post-transition linear rate constant,
Apost ¼ constant in post-transition rate constant,
Qpost ¼ post-transition activation energy (J/mole),
R ¼ universal gas constant (J/mole-K),
T ¼ temperature (K), and
t ¼ time in years.
As the measured values of transitory stage oxide thickness in the pressure tubes
examined during post-irradiation investigation do not vary any significantly from
tube to tube and along the length of tubes, the effect of neutron flux on corrosion
rate enhancement is accounted for by evaluating the transitory stage rate constant
from the measured oxide data. The synergistic effect of the neutron fast flux on the
out of pile post-transition linear corrosion rate [2,18] is accounted for (in the model
developed) by a term simulating an increase in post-transition rate constant as a
linear function of neutron flux (Eq 5).
(5) DKpost ¼ C1 /ExpðQ=RTÞ
where:
DKpost ¼ increase in rate constant
C1 ¼ constant, and
/ (n/cm2 – s) ¼ neutron fast flux (> 1 MeV).
Combining Eqs 4 and 5 gives the expression for the post transition rate con-
stant (K0post) for in-reactor corrosion.
0
(6a) Kpost ¼ Kpost þ DKpost
0
(6b) Kpost ¼ ðApost þ C1 /Þ ExpðQpost =RTÞ
The correlations for in-reactor corrosion of the Indian Zircaloy-2 pressure

tubes were thus given by Eqs 1 and 2 combined (transitory stage) and the Eqs 3 and
6b combined (post transition).
(7) S3t ¼ At ExpðQt =RTÞt

(8) Spost ¼ ðApost þ C1 /ÞExpðQpost =RTÞt
Parameters A, Q, and C1 for the transitory stage and post- transition stage cor-
rosion regimes had been evaluated from oxide thickness data obtained from the
irradiated Zircaloy-2 pressure tubes removed from Indian PHWRs.
The analyses of data and the methodology used for the determination of these
constants are described in the subsequent paragraphs.
FIG. 2 Variation of observed oxide thickness with time.
Analysis of Oxide Thickness Data

OXIDE THICKNESS REQUIRED FOR TRANSITION OF CORROSION
CHARACTERISTIC FROM TRANSITORY STAGE TO LINEAR POST-TRANSITION
IN REACTOR ENVIRONMENT
The plot of oxide thickness measured at different locations along the length of the
pressure tubes (Table 3) against their in-reactor residence time (Fig. 2) indicated that
the linear post-transition corrosion rate would have been established after the oxide
thickness reached in range of 8–12 lm. Variation of oxide thickness data from dif-
ferent pressure tubes at each of the axial locations5 of 4800, 4300, and 4000 mm,
respectively, with the in-reactor residence time of the pressure tube (Figs. 3–5) con-
firmed the transition oxide thickness to be about 10 lm. Based on the above, the
oxide thickness data was divided in two groups—one belonging to the transitory
stage (2 lm < oxide thickness < ¼ 10.0 lm) and the other belonging to linear post-
transition stage (oxide thickness > 10 lm).
DETERMINATION OF RATE CONSTANTS AND ACTIVATION ENERGY

Transitory Stage of Corrosion
The oxide thickness (< ¼ 10 lm) data obtained from different pressure tubes were
sub-grouped with respect to the neutron flux corresponding to the axial location to
5
Distance measured from inlet end of pressure tube.
FIG. 3 Variation of observed oxide thickness with time at 4800 mm as measured from
the PT inlet end.
the PT inlet end.
the PT inlet end.
which the oxide data belonged. Data belonging to a particular group then had tem-
perature as the only independent operating parameter. The slope and the intercept
of the linear-fit to the plot of the logarithm of the cube rates (S3/t) of oxide against
the reciprocal of temperature gave Qt and At, respectively.
Qt and At were evaluated for each flux group. Since the Qt and At have their
parallel in the pre-transition cubic rate correlation, similar data derived from the
information on aqueous corrosion of zirconium alloys available in the open litera-
ture have been compiled for comparison purpose. Both of these data sets are pre-
sented in Tables 4 and 5, respectively.
TABLE 4 Evaluated At and Qt for irradiated Indian Zircaloy-2 pressure tubes.
Average Flux (n/cm2 s) At (lm3/year) Qt (J/mole)
12 9
7.87 10 2.86 10 8.79 104
13 11
1.02 10 2.35 10 1.09 105
13 6
1.25 10 2.20 10 5.35 104
13 9
1.72 10 2.86 10 8.74 104
13 2
1.49 10 7.81 10 1.94 104
13 2
2.12 10 3.11 10 1.63 104
TABLE 5 Compilation of values of Qpre and Apre derived from Zirconium alloy aqueous corrosion
test data published in open literature [2,5].
Qpre Aapre
J/mole lm3/year References Remarks
1.13 105 8.46 1010 [2] Zircaloy-2/pre-transition (autoclave data)

4
8.73 10 4.77 108 [5] Arc melted iodide processed zirconium/
pre-transition
9.06 103 7.26 100 [5] Zircaloy-2/pre-transition
a 3
Converted reported value to units of lm /year.
Table 5 indicates that the values of the pre-transition corrosion constant and the
activation energy derived from the corrosion tests carried out in autoclave have a quite
wide range. The values of Qpre range from 9.06 103 J/mole to 1.13 105 J/mole and
the corresponding values of Apre range from 7.26 lm3/year to 8.46 1010 lm3/year.
Values of equivalent parameters derived from the transitory stage oxide thickness
data of the irradiated Indian Zircaloy- 2 pressure tubes (Table 4: 1.63–1.09 105 J/
mole for Qt and 3.11 102 lm3/year–2.35 1011 lm3/year for At) also exhibited
similar variation.
In order to compare the in-reactor corrosion rates of Indian Zircaloy-2 pressure
tubes with the in-autoclave rate observed by Hillner [2], logarithm of cubic rates
(S3/t) for all the flux groups are plotted against the reciprocal of temperature (K)
along with those observed by Hillner (Fig. 6). Cubic rates derived from in-reactor
thin film corrosion oxide data of the N reactor pressure tubes [19] are also plotted
for comparison in the same figure.
The comparison indicates that the corrosion rate constants for the Indian
Zircaloy-2 pressure tubes appear to be in the range of 2.5–25 times the pre-
transition rate constant derived by Hillner in autoclave tests. Rate constants for the
assumed cubic corrosion kinetics for the thin film oxide data of the N-reactor pres-
sure tubes appear to be nearly 100–1000 times higher than the pre-transition rate
constant of Hillner. Lanning et al. [19] modelled the in-reactor oxidation of
N-reactor pressure tubes using bi-linear correlations—the post-transition regime,
also called thin film regime (for oxide thickness in the range of 3–20 lm) and the
accelerated corrosion regime (for oxide thickness > 20 lm). Compared to Hillner
[2] post-transition linear oxidation rate, N-reactor thin film rates are 20–60 times
higher. Difference in fabrication route, material chemistry, and operating environ-
ment of the N-reactor pressure tubes as compared to Indian pressure tubes could
be the reason for the observed comparatively greater oxidation rate in these tubes.
Based on the above discussion, the Qt and At for the Indian PHWR pressure
tubes corresponding to the upper bound rate constant (25 times the Hillner
in-autoclave rate constant) were chosen so that predictions in all the circumstances
FIG. 6 Comparison of transitory stage corrosion rate of irradiated Indian Zircaloy-2

pressure tubes with the one given by Hillner [2] pre-transition correlation and
the derived cubic rates from the thin film oxide data of N-reactor pressure tubes.
could be conservative with respect to measurement. These values were 113 370 J/mole
and 2.0 1012 lm3/year, respectively.
Post Transition
The post-transition corrosion was observed in the five pressure tubes removed from
RAPS-2. These tubes had seen 8.5 FPYs (10.0 Hot Operating Years or HOYs). In
order to estimate the constants (A, C) and the activation energy (Q) for the post
transition conditions, oxide thickness values higher than 10 lm had been consid-
ered. Autoclave thickness and the oxide thickness required for transition from tran-
sitory stage (10 lm) were subtracted from the total thickness to find out the oxide
thickness developed during the post-transition period. Assuming that Qpost is equal
to Qt [20], (Apost þ C1/) was evaluated for each oxide data point after normalising
it with respect to temperature. Apost and C1 were evaluated from the intercept and
the slope of the plot of (Apost þ C1/) versus neutron flux (/) (Table 6).
Estimation of Hydrogen Pickup Fractions

The hydrogen pickup fraction in the zirconium alloy pressure tube material is
defined as the ratio of hydrogen absorbed to the total hydrogen evolved during the
corrosion reaction. Oxide thickness and the measured hydrogen pickup at the same
TABLE 6 Values of constant and activation energy derived for post-transition corrosion.
Constant/Activation Energy Values Units
10
Apost 1.5 10 (lm/year)
C1 0.0243 (lm-cm2-s/neutron-year)
Qpost 113 370 (J/mole)
axial location from a number of pressure tubes were used to evaluate hydrogen
pickup fractions both during the transitory and post-transition periods. The pickup
fractions in all the pressure tubes were then plotted against the fast neutron flux
existing at the respective locations for both the transitory and post-transition
regimes of corrosion (Figs. 7 and 8).
The H pickup fraction data as plotted in the Figs. 7 and 8 showed large scatter
with conspicuously decreasing trend with respect to neutron flux, particularly dur-
ing transitory stage. The linear-fit corresponding to 95 % prediction limit was con-
sidered for estimating hydrogen pickup from the oxide thickness expected to form
during transitory period.
FIG. 7 Variation of hydrogen pickup fraction with neutron flux during pre-transition
corrosion for the pressure tubes removed from RAPS and MAPS units.
FIG. 8 Variation of hydrogen pickup fraction with neutron flux during post-transition
corrosion for the pressure tubes removed from RAPS units.
The best-fit line for the post-transition H pickup fraction data had indicated
marginal variation (0.5–0.7) with respect to neutron flux. The maximum observed
value was 0.85 and the mean of all the data was 0.6.
Comparison of Model Estimated Hydrogen

Pick-Up at the Slivered Locations
in the Pressure Tubes of RAPS-1 And NAPS-1
Sliver scrape samplings were obtained at nearly 200 locations in eighty pressure
tubes of RAPS-1 unit and at 36 locations in six pressure tubes of NAPS-1 unit at
10.6 HOYs6 and 9.8 HOYs, respectively. The samples of RAPS-1 were analysed for
H-pickup by DSC and HVEQMS techniques and those of NAPS-1 were analysed
by DSC technique. Choosing only DSC technique for measuring H-pickup in
NAPS-1 sliver samples was based on observation that both the techniques give
same results within the measurement accuracy [17].
The coolant channel assembly designs in these reactor units are different in
respect of the design of annulus between the pressure tube and the calandria tube.
6
HOYs stands for Hot Operating Years.
The annulus in the former is open and vault air flows through it whereas in the lat-
ter the annulus is closed and the CO2 is in re-circulation. Assuming that annulus
design and condition is not playing any role in hydrogen pickup in Zircaloy-2 pres-
sure tube from the inside surface, the developed correlations were applied to esti-
mate the hydrogen pickup at the sliver sample locations. Comparisons of the
estimated values with the measurement in respect of the both reactor units are
shown in the Figs. 9 and 10. The post-transition H-pickup estimations for NAPS-1
pressure tubes have been made using the both maximum observed pickup fraction
of 0.85 (Fig. 10(a)) and mean average pickup fraction of 0.6 (Fig. 10(b)).
Discussion
VARIABILITY IN OXIDATION RATE AND H-PICKUP
Oxidation and hydrogen pickup as a result of in-reactor long-term aqueous corro-
sion in Zircaloy-2 pressure tubes of Pickering units (removed after 3620 EFPDs)
has been dealt by Urbanic et al. [8]. Large variability in oxidation rate and H-
pickup rate had been observed in the examined pressure tubes which had been fab-
ricated by similar process and operated for the same period under nearly similar in-
reactor operating conditions. The ratio of the maximum to the minimum observed
FIG. 9 Comparison of estimated and measured hydrogen pickup for RAPS-1 pressure
tubes.
FIG. 10 Comparison of estimated and measured hydrogen pickup in NAPS-1 pressure

tubes for (a) post-transition pickup fraction of 0.85 (b) post-transition pickup
fraction of 0.6.
values of both the oxide thickness and the H-pickup, respectively, at the location of
maximum oxidation and H-pickup was approximately 1.7. This ratio for the loca-
tions where the oxide thickness had not reached the critical thickness (>15 lm as
reported for Pickering tubes) was approximately 3.0.
Large scatter in hydrogen pickup fractions as observed in Zircaloy-2 pressure
tubes removed from Indian reactor units also reflects variability in the oxidation
rate and the H-pickup rate. This variability in oxidation and H-pickup rates may be
due to the following:
• Pressure tubes of RAPS-2 and MAPS-1 were obtained from two different
sources.
• The pressure tubes from the same source might be from different production
lots.
• Chemical composition of the alloy and its metallurgical properties might be
varying from tube to tube.
• Different fabrication routes were adopted in manufacturing of these pressure
tubes.
It can be observed from Fig. 8 that the scatter in transitory stage H-pickup frac-
tion is quite large in RAPS-2 pressure tubes than in MAPS tubes. It should also be
noted that these pressure tubes belonged to the first generation of PHWR units
commissioned in early 1980 s. Many of the currently known in-reactor degradation
mechanisms for zirconium alloy pressure tube material and their dependence on
chemical composition and metallurgical parameters were not known to the design-
ers and the material scientists at that time.
Variability in corrosion rate and H-pickup in zirconium alloy under the reactor
operating condition has also been discussed by Lanning et al. [19]. They attributed this
variability to the re-distribution of hydrogen as a result of azimuthal temperature varia-
tion and variability in chemical and metallurgical parameters from one lot to another.
The effect of neutron irradiation on corrosion of zirconium alloys has been

investigated by Johnson et al. [10,11] and Harrop et al. [21]. While Johnson cited
irradiation induced increased oxygen transport or the early stages of transition to
unprotective oxide as the possible reasons for increased corrosion and H-pickup in
the neutron environment, Harrop et al. [21] investigated embrittlement of oxide in
the presence of neutron flux. Both these works have indicated that deterioration of
oxide integrity in the neutron environment could be the reason of increased corro-
sion and high H-pickup in zirconium alloys. Deterioration of oxide is expected to
be more at lower temperature due to increased radiation hardening of the substrate
and therefore lesser accommodating capacity for the strains during oxide formation
[21].
The scatter in the post-transition pickup fractions is expected as it depends
upon the local condition of oxide and also the local water chemistry built inside the
oxide pores and cracks. Nevertheless, the H-pickup fraction values being less than
or equal to 1.0 indicate that the ingressed hydrogen has evolved during the aqueous
corrosion of Zircaloy-2 pressure tubes with the hot coolant.
COMPARISON OF H-PICKUP ESTIMATED USING THE MODEL

WITH THE H-PICKUP MEASURED IN SLIVER SAMPLES
The comparison of H-pickup estimated using the correlation developed at the sliver
sample locations in the pressure tubes of RAPS-1 (Fig. 9) and NAPS-1 (Fig. 10) with
the H-pickup measured in the sliver samples removed from those locations estab-
lishes the upper bound predicting capability of the correlation developed. The
hydrogen pickup behaviour in the pressure tubes of RAPS-1 and NAPS-1 reactors
during transitory stage of corrosion appears to be similar.
The sliver samples locations in the RAPS-1 pressure tubes were in the portion
of pressure tube length undergoing transitory stage corrosion whereas those in the
NAPS-1 pressure tubes were spread along the length of the pressure tubes such that
the corrosion and the H-pickup corresponding to both the transitory and the post-
transition stages could be covered. Post-transition H-pickup estimated using the
post-transition H-pickup fraction of 0.85 (Fig. 10(a)) is more conservative than the
post-transition H- pickup estimated using post-transition H-pickup fraction of 0.6
(Fig. 10(b)).
Conclusions
Estimations of hydrogen pickup in the pressure tubes of RAPS-1 and NAPS-1 using
the correlations developed for transitory and post transition stages of corrosion and
hydriding together with the correlations defining transitory stage pickup fraction
variation with neutron flux and post-transition pickup fraction of 0.85 and 0.6 are
reasonably conservative. The lower value of post-transition pickup fraction should
be considered for more reasonable estimates of hydrogen pickup. The H-pickup
fraction during the transitory stage appears to be lower in the pressure tubes used
in MAPS units than those in RAPS-2 (Fig. 7). This observation could be unique to
the pressure tubes of these units as the transitory stage H-pickup in pressure tubes
of RAPS-1and NAPS-1 as observed from the hydrogen measurement carried out in
sliver scrape samples did not reveal the same. Variability in processing parameters
and the chemical compositions of the ingot within the acceptable band could be the
reasons for this difference in hydrogen pickup behaviour. Similar comparison for
H-pickup behaviour during post-transitory stage corrosion between the pressure
tubes of RAPS-2 unit and the other units could not be established due to unavail-
ability of data. However, the post-transitory stage predictions made for NAPS-1
pressure tubes using the maximum (0.85) and the mean values (0.6) of post-
transitory stage pickup fractions as observed in the RAPS-2 pressure tubes indicate
that predictions using pickup fraction value of 0.6 are just overlapping the measure-
ment and those with pickup fraction of 0.85 are quite conservative.
ACKNOWLEDGMENTS
The authors would like to sincerely thank their colleagues in Nuclear Power Corpora-
tion of India Limited, and Post-Irradiation Examination Division (PIED) and Reactor
Engineering Division (RED) of Bhabha Atomic Research Centre, Mumbai for their
active support in collecting the sliver samples and generating the data on oxide thick-
ness and hydrogen pickup in the bulk and sliver samples.
References
[1] IAEA, “Heavy Water Reactors: Status and Projected Development,” Technical Report
Series No. 407, IAEA, Vienna, Austria, 2002.
[2] Hillner, E., “Corrosion of Zirconium-Base Alloys—An Overview,” Zirconium in the Nuclear
Industry, ASTM STP 633, A. L. Lowe, Jr. and G. W. Parry, Eds., ASTM International, Phila-
delphia, PA, 1977, pp. 211–235.
[3] Peters, H. R., “Improved Characterisation of Aqueous Corrosion Kinetics of Zircaloy-4,”

Zirconium in Nuclear Industry: Sixth International Symposium, ASTM 824, D. G. Franklin
and R. B. Adamson, Eds., ASTM International, Philadelphia, PA, 1984, pp. 507–518.
[4] Bryner, J. S., “The Cyclic Nature of Corrosion of Zircaloy-4 in 633 K Water,” J. Nucl.
Mater., Vol. 82, No. 1, 1979, pp. 84–101.
[5] Thomas, D. E., “Aqueous Corrosion of Zirconium and its Alloys at Elevated Temper-
ature,” Proceedings of the First UN International Conference on Peaceful Uses of Atomic
Energy, Paper No. 537, Geneva, Switzerland, Sept 1–13, 1958.
[6] Stanley, K.,“Corrosion of Prefilmed Zircaloy,” Corrosion, Vol. 23, No. 12, 1967, pp.
374–378.
Alloys,” J. Nucl. Mater., Vol. 336, Nos. 2–3, 2005, pp. 331–368.
[8] Urbanic, V. F. and Cox, B., “Long Term Corrosion and Deuterium Uptake in CANDU-PHW
Pressure Tubes,” Zirconium in Nuclear Industry: Seventh International Symposium, ASTM
STP 939, R. B. Adamson and L. F. P. Van Swam, Eds., ASTM International, Philadelphia,
PA, 1987, pp. 189–205.
[9] Hillner, E., Franklin, D. G., and Smee, J. D., “Long-Term Corrosion of Zircaloy Before and
After Irradiation,” J. Nucl. Mater., Vol. 278, Nos. 2–3, 2000, pp. 334–345.
[10] Johnson, A. B., “Effects of Nuclear Radiation on Corrosion, Hydriding, and Oxide Proper-
ties of Six Zirconium Alloys,” Applications Related Phenomena for Zirconium and its
Alloys, ASTM STP 458, ASTM International, Philadelphia, PA, 1969, pp. 301–324.
[11] Johnson, A. B., Jr., LeSurf, J. E., and Proebstle, R. A., “Study of Zirconinum Alloy Corro-
sion Parameters in the Advanced Test Reactor,” Zirconium in Nuclear Applications,
ASTM STP 551, ASTM International, Philadelphia, PA, 1974, pp. 495–513.
[12] Hillner, E., “Long-Term In-Reactor Corrosion and Hydriding of Zircaloy-2 Tubing,” Zirco-
nium in Nuclear Industry: Fifth Conference, ASTM STP 754, D. G. Franklin. Ed., ASTM
International, Philadelphia, PA, 1982, pp. 450–478.
[13] Madhusoodanan, K. and Sinha, R. K., “Modelling of Oxidation and Hydrogen Pick-Up in
Zircaloy-2 Pressure Tubes,” Proceedings of the IAEA Consulant’s Meeting on Pressure
tube Integrity, Vienna, Austria, July 25–29, 1994.
[14] Madhusoodanan, K., Sinha, S. K., and Sinha, R. K., “A Computer Code for Estimation of
Hydrogen Pick-Up and Oxide Thickness in Zircaloy-2 Pressure Tubes,” Proceedings of
the International Symposium on Materials Ageing and Life Management, Kalpakkam,
India, Oct 3–6, 2000.
[15] Sesha Sayi, Y., Ramakumar, K. L., Prasad, R., Yadav, C. S., Shankaran, P. S., Chhapru, G. C.,
and Jain, H. C., “Determination of H2 and D2 Content in Metals and Alloys Using hot Vac-
uum Extraction,” J. Radioanal. Nucl. Chem., Vol. 230, Nos. 1–2, 1998, pp. 5–9.
[16] Komal Chandra, P. S., Ramanjaneyulu, C. S., Yadav, A. S., Kulkarni, Y. S., Sesha Sayi, Y.,
and Ramakumar, K. L., “Determination of Deuterium Pick-Up in Zr–Nb Alloy by Hot Vac-
uum Extraction–Quadrupole Mass Spectrometry,” Anal. Lett., Vol. 45, No. 15, 2012, pp.
2136–2147.
[17] IAEA, “Inter Comparison of Techniques for Inspection and Diagnostics of Heavy Water
Reactor Pressure Tubes,” IAEA-TECDOC-1609, IAEA, Vienna, Austria, 2009.
[18] Field, G. J., Dunn, J. T., and Cheadle, B. A., “Analysis of the Pressure Tube Failure at Pick-
ering NGS “A” Unit 2,” Can. Metal Q., Vol. 24, No. 3, 1985, pp. 181–188.
[19] Lanning, D. D., Johnson, A. B., Trimble, D. J., and Boyd, S. M., “Corrosion and Hydriding
of N Reactor Pressure Tubes,” ASTM-STP 1023, ASTM International, Philadelphia, PA,
1989, pp. 3–19.
[20] Billot, P. and Giordano, A., “Comparison of Zircaloy Corrosion Models from the Evalua-
tion of In-Reactor and Out-of-Pile Loop Performance,” ASTM STP 1132, ASTM Interna-
tional, Philadelphia, PA, 1991, pp. 539–565.
[21] Harrop, P. J. and Wanklyn, J. N., “The Embrittlement of Oxide Films on Zirconium by
Neutron Irradiation,” J. Nucl. Mater., Vol. 21, No. 3, 1967, pp. 310–316.
DISCUSSION
Questions from Ted Darby, Rolls-Royce
Q1:—You observed an increase in oxidation rate constant of about 25 times the
out-of-pile value. How did this in-pile rate constant vary with temperature (which
varies along the pressure tube)?
Authors’ Response:—The rate constant has an Arrhenius relation with tempera-
ture given by the expression A*Exp (-Q/RT)1. In the transitory corrosion regime
which is modelled by cubic kinetics, the cubic rate constants at the inlet end
(250 C) and the outlet end (300 C) of the pressure tube are 9 lm3/year and
91 lm3/year, respectively.
Q2:—What is (or will be) the longest operational life of a Zircaloy-2 tube in the
PHWR? Are there plans to do corrosion and hydrogen measurements at these con-
ditions to confirm that the model extrapolation is conservative?
Authors’ Response:—The longest operating period for which a Zircaloy-2 pres-
sure tube has remained in service is 12.0 years. These pressure tubes belonged to
Indian PHWR unit - KAPS unit -1. Estimation of H-pickup in an operating pres-
sure tube when corrosion and H-pickup is in the accelerated regime, involves lot of
un-certainty. In such case, the estimation becomes highly conservative. Threshold
stress for hydride re-orientation being nearly close to the operating stress, the pre-
cipitated hydrides are always radial. As a result, fracture toughness of the pressure
tube material even at the operating temperature degrades substantially and there-
fore ‘Leak before Break’ criteria will not be satisfied during normal operation.
A few pressure tubes removed after end of life were investigated during post
irradiation examination. Estimated values of hydrogen pick-up and oxide thickness
were found to be conservative with respect to the measured values.
Questions from N. Ramasubramanian, ECCATEC Inc. Canada:—Can you com-
ment on the percent pick-up variation along the channel length? How different is
it at 4.8 m compared to at 3.0 m?
Authors’ Response:—Percent pickup during transitory period of corrosion re-
gime along the length of channel has been found to be following ‘A – K sin(p/L*X)2’
variation. It is maximum at the inlet and the outlet ends and minimum at the
centre. It will be less at 3.0 meters than at 4.8 meters.
Question from David Schrire, Vattenfall:—Is there any difference or bias
between the DSC and the HVEQMS measurements?
Authors’ Response:—No, there is no bias between the DSC and the HVEQMS
measurements. It has been established that both measurement techniques give val-
ues within the limits of measurement accuracy.
1
Where A is constant; Q is Activation energy; R is universal gas constant and T is temperature.
2
Where A and K are constants; L is the length of pressure tube and X is the distance measurement from the
pressure tube inlet end.
A. M. Garde,1,* G. Pan,1 A. J. Mueller,2 and L. Hallstadius3
Oxide Surface Peeling of

Advanced Zirconium Alloy
Cladding after High Burnup
Irradiation in Pressurized Water
Reactors
Reference
Garde, A. M., Pan, G., Mueller, A. J., and Hallstadius, L., “Oxide Surface Peeling of Advanced
Zirconium Alloy Cladding after High Burnup Irradiation in Pressurized Water Reactors,”
and Pierre Barberis, Eds., pp. 673–692, doi:10.1520/STP154320130005, ASTM International,
West Conshohocken, PA 2015.4
ABSTRACT
Microscopic examinations of advanced zirconium alloy cladding irradiated to
burnups over 70 GWd/MTU have revealed de-lamination of surface layers of the
thin oxide. The new observation is termed “oxide surface peeling” or OSP.
Examinations have revealed the layered structure of the oxide. Metallographic
examination revealed that the waterside oxide on different irradiated Zr alloy
cladding had a layered structure similar to the autoclaved corrosion specimens
examined earlier. However, the OSP observations discussed here apply only to
irradiated cladding. OSP is not observed in autoclave corrosion. A featureless intact
oxide sub-layer was present in the interior oxide at the metal/oxide interface for all
alloys. On top of this featureless layer, there were additional sub-layers with fine
circumferential fissures believed to be associated with the cyclic corrosion rate
transitions. The number of sub-layers depended on the corrosion resistance of the
Manuscript received January 4, 2013; accepted for publication August 15, 2013; published online
September 22, 2014.
*ASTM Fellow.
1
Westinghouse Electric Company, Hopkins, SC 29061, United States of America.
2
Westinghouse Electric Company, Pittsburgh, PA 15235, United States of America.
3
Westinghouse Electric Sweden, Fuel Engineering, Vasteras SE-72163, Sweden.
4
Hyderabad, India.
Copyright V
alloy. Intermediate corrosion resistance alloy had many sub-layers forming an

overall intermediate thickness oxide. Higher corrosion resistance alloys had fewer
sub-layers on top of the barrier oxide layer. In some cases, small patches of the top
surface layers of the thin oxide were peeled off because of radial cracks generated
in the oxide layer caused by tensile stresses created by a hard pellet contact.
Metallography of the underlying cladding showed that hydride localization was not
associated with the oxide surface peeling; in contrast to previous experience on the
low corrosion resistance older claddings, such as Zircaloy-4, where a major
thickness fraction of the thick oxide extending to the underlying metal was
removed (“spalled”). The oxide surface layer peeling does not lead to irradiated
alloy ductility reduction or enhanced embrittlement. The impact of oxide surface
peeling observations on fuel performance is discussed in the paper. For fuel
designs with low margin against grid-to-rod fretting wear, OSP may reduce this
margin further.
Keywords
zirconium alloys, high corrosion resistance, oxide delamination, peeling, spalling,
hydride localization, ductility, wear resistance
Introduction
The zirconium alloy development programs in the nuclear industry for the past three
decades have been primarily focused on improvement of the waterside corrosion resist-
ance of the zirconium alloys. As the corrosion resistance was improved, the thickness of
waterside oxide decreased. Traditionally, very thin, particularly, “pre-transition” oxides
have a shiny black visual appearance, whereas the thick “post-transition” oxides have a
white gray appearance. At intermediate oxide thicknesses, the oxide has a mottled
appearance of patches of gray oxide in black shiny background. At very large oxide
thicknesses the oxide is entirely lighter gray. It is thought that black shiny oxide is pro-
tective, whereas gray oxide contains porosity and defects and the outer part of the oxide
is non-protective. In addition to the change in the oxide color visual appearance dis-
cussed above, a partial de-lamination of the thin oxide layer has been recently observed
when high corrosion resistance zirconium alloys are irradiated as fuel cladding in light
water reactors. This new process of de-lamination of thin waterside oxide is termed ox-
ide surface peeling (OSP). The objective of this paper is to describe the OSP observations
and evaluate the impact of OSP on nuclear fuel rod performance. The available data are
limited and statistical analysis of OSP is not yet possible. The purpose of the paper is to
document the phenomenon of OSP and to identify trends based upon the limited data.
Description of Examined Fuel Rods and

Techniques of Specimen Examination
Lead test assemblies (LTA) using Optimized ZIRLOTM alloy and AXIOM alloy
cladding tubes were irradiated in different pressurized water reactors (PWRs) for
GARDE ET AL., DOI 10.1520/STP154320130005 675
different cycles of irradiation. Compositions of different alloys are provided in

Table 1 [1,2]. Post-irradiation examinations (PIE) were performed on selected rods
from the LTAs after each cycle at the spent fuel poolside. The poolside PIEs include
visual inspection of the assemblies and removal of single rods for visual inspection
using high-definition camera, length measurement, profilometry for rod diameter
measurements, and eddy-current measurements of local waterside oxide layer
thickness. Based on the poolside PIE results, 10 Optimized ZIRLOTM and AXIOM
rods irradiated at U.S. PWR A for three cycles with burnup over 70 GWD/MTU
were sent to the hot cell for further destructive examination. All of the 10 rods
selected show representative oxide surface peeling indications of the particular
alloy. The selection criteria of rods shipped to the hot cell also cover other aspects
including internal/peripheral position within the fuel assembly, different pellet type,
highest/lowest burnup and range of waterside oxide thicknesses. The rods were cut
into four segments at the hot cell prior to further examination. All the non-
destructive tests performed at poolside were conducted on selected segments. The
rod segments chosen to be evaluated in the hot cell were taken from the upper part
of the rod where most features of interest are present. Samples were then taken
from the selected locations for destructive tests including metallography, SEM
(scanning electron microscopy), TEM (transmission electron microscopy), RTT
(ring tensile testing), ATT (axial tensile testing), and HVE (hot vacuum extraction)
hydrogen measurements. The most extensively studied area for the examined rods
is about 3 ms from the bottom of the rod. The tensile tests specimen, the hydrogen
analysis sample, the metallography and SEM samples have been selected to be adja-
cent to each other for the selected rod axial elevation.
OSP Observations
Oxide surface peeling (OSP) has been observed by Westinghouse during the post-
irradiation poolside examination of several advanced zirconium fuel cladding alloys
as discussed below. Initially, OSP was not recognized partially because of
TABLE 1 AXIOM and Optimized ZIRLOTM alloy nominal chemical compositions in weight % and
microstructure from final heat treatment.
Alloy Microstructure Nb Sn Fe Cr Cu V Ni Zr
AXIOM X1 pRXA 0.7–1 0.3 0.05 0.12 0.2 Bal.

AXIOM X2 RXA 1 0.06 Bal.
AXIOM X4 pRXA 1 0.06 0.25 0.08 Bal.
AXIOM X5 pRXA 0.7 0.3 0.35 0.25 0.05 Bal.
Optimized pRXA 1 0.67 0.1 Bal.
ZIRLOTM
Note: Bal. ¼ Balance.

insufficient resolution of the poolside camera photography. However, when OSP was
first recognized for the improved corrosion-resistant AXIOM alloys with high-
resolution photography, subsequent re-examination of Optimized ZIRLOTM poolside
records confirmed observation of minor degrees of OSP for Optimized ZIRLOTM
cladding. Westinghouse OSP observations are discussed below. As discussed below,
the appearance of OSP can be different depending on the degree of OSP. A minor
degree of OSP may be possibly seen in the form of fine superficial surface cracks or
“crocodile” skin and a higher degree of OSP exhibits oxide surface de-lamination.
OSP has been observed for the Westinghouse advanced AXIOM alloys, Duplex
[3] alloys (Zircaloy-4 cladding tube with an outer layer of corrosion-resistant alloy),
and infrequently for Optimized ZIRLOTM alloy and never for ZIRLO alloy. OSP was
first observed in 2009 for one of the AXIOM alloys irradiated in PWR A for a burnup
of about 50 GWd/MTU when the local waterside oxide thickness was only about
16 lm. Subsequently, after three-cycle irradiation in PWR A to burnups greater than
70 GWd/MTU, all AXIOM alloys in the assembly (oxide thickness range 12 to 37 lm)
and some of the Optimized ZIRLOTM (oxide thickness range 39 to 69 lm) showed
OSP. All OSP observations were located in the upper elevations of the rod where the
extent of nucleate boiling in the coolant and the measured cladding waterside oxide
thicknesses were highest. AXIOM rods after three-cycle irradiation in PWR B also
showed OSP for burnups around 73 GWd/MTU and oxide thickness range of 17 to
46 lm. Poolside visual examination in PWR C of fuel rods with Optimized ZIRLOTM
and duplex cladding after a burnup of about 46 GWd/MTU showed OSP. Duplex clad-
ding irradiated to burnups of about 45 GWd/MTU in PWR D also showed OSP. (The
date of poolside examination of duplex cladding was 1999. However, it was not recog-
nized that it is an OSP observation until 2009). OSP was also observed for Optimized
ZIRLOTM cladding irradiated in PWR E to burnups of about 60 GWd/MTU.
Poolside and Hot-Cell Visual Observations

OSP OBSERVATIONS
The presence of OSP is apparent as a non-uniform visual appearance on the clad-
ding outer surface during the poolside visual examinations during the refueling out-
ages. All AXIOM clad rods showed evidence of OSP. The extent of OSP varied
from rod to rod. As a typical example, see Fig. 1 for AXIOM X5 cladding rod irradi-
ated to about 55 GWd/MTU in PWR A. Although crud deposition (dark areas)
sometimes made it difficult to detect OSP, removal of loose crud by brushing
revealed OSP more clearly. The possible impact of crud removal in enhancing OSP
is possible but could not be positively confirmed. Figure 1 appears to indicate partial
removal of 2–3 thin layers of the oxide in different parts of the fuel rod surface.
Although OSP may be visible on the peripheral rod surface during the fuel assembly
surface visual examination, the extent of OSP is more discernable at higher magnifi-
cation fuel rod examination after removal of individual rod from the assembly for
single rod examination.
FIG. 1 OSP of AXIOM X5 fuel rod irradiated in PWR A to a burnup of about 55 GWd/
MTU. Three OSP spots of different sizes are seen aligned vertically in center
bottom of the figure.
Loose non-adherent crud was removed by brushing in Fig. 1. It appears that

OSP is associated with the crud deposition and some crud may still be present in
the deepest craters of OSP.
Typical examples of OSP for duplex cladding (base tube of high-tin Zircaloy-4
with an outer layer of 0.5 % tin alloy) from PWR D is shown in Fig. 2. The composi-
tion of duplex cladding is previously provided [3].
FIG. 2 OSP for duplex cladding irradiated in PWR D to burnup of about 55 GWd/MTU.
One OSP spot seen in the lower part of the fuel rod at the extreme right and two
OSP spots are close to the center of the third fuel rod from the right.
Typical examples of OSP for Optimized ZIRLOTM are provided in Fig. 3 after
irradiation to burnup of about 60 GWd/MTU in PWR E. The non-uniform visual
appearance of oxide is not associated with contact with grid support features. All
OSP photographs in this paper are from the mid-span locations away from the
grids. Light contact marks were observed at the grid contact points without signifi-
cant OSP. As shown in Fig. 3, shapes and sizes of the OSP features vary. They may
have straight or curved boundaries.
OTHER FORMS OF OXIDE NON-UNIFORMITY

Some Optimized ZIRLOTM rods show fine surface cracking of the oxide without
de-lamination as shown in Fig. 4. The fine surface cracking may be located at mid-
pellet locations, or at the pellet/pellet interface perpendicular to the fuel rod axis.
Such surface cracking was detected by high-resolution camera during post-
irradiation poolside examinations. It is suspected that this oxide surface cracking at
the pellet/pellet interface is caused by the tensile stress imposed on the oxide
because of volumetric expansion associated with the hydride precipitation at the
pellet/pellet interface as a result of slightly lower cladding temperature at the pellet/
FIG. 3 Typical examples of OSP for Optimized ZIRLOTM cladding irradiated to 60

GWd/MTU in PWR E (The diameter of the rods are 9.5 mm).
FIG. 4 Fine surface cracking of the oxide of Optimized ZIRLOTM clad fuel rod irradiated
in PWR E to a burnup of 65 GWd/MTU at the pellet–pellet interface.
pellet interface. Some of the AXIOM alloys showed a “crocodile” skin as shown in
Fig. 5. Crocodile skin was not detected at the poolside examination because of the
magnification and resolution needed to detect it. It was noticed only in a hot-cell
examination. Both fine surface cracking and crocodile skin are potentially related to
oxide surface peeling.
To characterize both the oxide layer undergone OSP and the cladding metal
underneath the oxide with OSP, fuel rods were sent to a hot cell for destructive eval-
uation. The hot-cell evaluation results are discussed in the next section.
FIG. 5 Crocodile skin for one of the AXIOM alloys.

HOT-CELL CERAMOGRAPHY AND METALLOGRAPHY

Fuel rods with the poolside examination evidence of OSP were sent to a hot cell for
destructive examination of both the oxide layer and the underlying cladding metal.
This was the first instance of conducting metallography on thin oxide on corrosion-
resistant zirconium alloys with known damage to the oxide surface. Although pre-
cautions were taken during specimen mounting and polishing to avoid additional
damage to the oxide layer during metallographic specimen preparation, there is
some possibility that thermal expansion/contraction difference between the metal
and the mount material may have contributed to some extent to the observed oxide
layer surface delamination. For this reason, the metallographic observations of OSP
reported in the current work need to be confirmed in the future with additional
extensive specimen edge protection precautions. This comment is particularly rele-
vant to the observation of peeling of the innermost oxide observed in the highest
corrosion-resistant alloys. Both longitudinal and transverse sections were mounted,
mechanically polished with care to preserve edges, and metallographically exam-
ined. Mechanically polished specimens (without chemical etching) were used to
evaluate the oxide layer thickness and oxide structure. Precautions were taken to
avoid damage to the thin oxide layer during the specimen sectioning and mechani-
cal polishing. Specimens were subsequently chemically etched with a mixture of
30 % water, 30 % nitric acid, 30 % sulfuric acid, and 10 % hydrofluoric acid to reveal
hydride precipitates in the underlying cladding metal.
The un-etched oxide layer microstructure of four alloys (burnup about 72
GWd/MTU) is presented in Fig. 6.
A 20 - or 100 -lm magnification bar is included in each image. A multilayered
oxide is observed for all alloys, similar to the layered structure reported earlier for
autoclaved specimens [4]. The thickness of individual layers in the autoclave corro-
sion and in-pile corrosion appear to be different. Thinner layer is seen in the auto-
clave corrosion. The possible reasons for the difference are (a) lower corrosion
resistance of past-evaluated autoclaved alloys, and (b) stress difference in the oxide
layer formed in autoclave corrosion and in PWR corrosion. The innermost layer of
the oxide next to the metal is featureless and appears darker in some cases. OSP
appears to be associated with the peeling of one to several layers of the thin oxide
layers. In some cases, even the innermost oxide layer may participate in the peeling
process as seen in Fig. 6(b) and 6(d). There are longitudinal cracks (parallel to the
metal/oxide interface) between oxide sub-layers and radial cracks (in the oxide
thickness direction) penetrating through different sub-layers. The severity of OSP
was alloy composition dependent. For alloys such as Optimized ZIRLOTM and
AXIOM X1 with an intermediate thickness oxide, cracking of the inner oxide layers
was very limited and OSP removed only a small fraction of the total oxide thickness
by peeling of few outer most layers of the oxide. For high-corrosion resistance
AXIOM alloys with a thin oxide, the oxide cracking was significant and the extent
of OSP was high. This possible difference of degree of OSP between different alloys
may be because of possible differences in the mechanical properties of the oxide
FIG. 6 OSP of fuel rod cladding irradiated to a burnup of 72 GWd/MTU: (a) AXIOM X1,
(b) AXIOM X5, (c) Optimized ZIRLOTM, and (d) AXIOM X2. The dark region at
the top of each figure is epoxy mount material, the light region at the bottom is
the zirconium alloy metal, and the grey layers in the center are the sub-layers of
the peeling waterside oxide.
developed on alloys of different compositions. Impact of different alloying elements

on the extent of OSP and on the oxide mechanical properties is not yet identified.
The metallographic specimens were etched to reveal hydriding of cladding
under the peeled oxide. Typical examples are shown in Fig. 7.
Hydride metallographic evaluation under the peeled oxide for the other
AXIOM alloys gave similar results. None of the alloys that experienced OSP showed
evidence of significant hydride localization under the peeled oxide. This observation
FIG. 7 Hydride appearance of fuel cladding under OSP regions: (a) Optimized
ZIRLOTM, and (b) AXIOM X4.
implies that peeling of thin oxide does not create a cold spot in the cladding like
thick oxide spallation does [5].
To estimate the temperature gradient created by OSP between peeled and adja-
cent non-peeled region of fuel cladding, a heat transfer calculation was performed
assuming a 23-lm oxide in the non-peeled region and a 0-lm oxide layer in the
peeled region (an extreme case of OSP). For the typical PWR power conditions and
fuel geometry, the temperature difference was estimated to be of the order of 5 K.
This is small compared to 10 K temperature differences estimated between the pellet
axial center and the pellet/pellet interface. The simplified heat transfer analysis and
hydride metallography under peeled oxide confirm that a significant cold spot is not
generated by OSP. Results of a similar calculation for a spalled thick oxide would
depend on the assumed power level of the rod, thickness of the unspalled oxide, and
fractional thickness spalled. For a 100 -lm total oxide thickness, complete spalling
can generate a temperature difference in the range of 20 to 35 K. These rough esti-
mates clearly show that the cold spot generated by peeling is much weaker than a
cold spot generated by spalling. The hydride metallographic photographs of the
peeled region were subjected to quantitative metallographic analysis to further evalu-

ate the localization of hydrides as discussed in the next section.
Hydride Volume Fraction and Orientation Image

Analysis of Optimized ZIRLOTM Cladding OSP
Area
Nine micrographs with hydride precipitates from a high burnup Optimized ZIR-
LOTM fuel rod transverse cross section at an elevation where oxide peeling was
observed (299 cm from the bottom of the fuel rod) were analyzed for hydride vol-
ume fraction and hydride orientation. OSP was observed over a small portion of
the circumference, 100 to 125 segment.
Eight other micrographs from circumferential locations without oxide peeling
at 45 increments around the circumference at the same axial elevation of the same
fuel rod were quantitatively analyzed for comparison. Image analysis software,
ImageJ and analySIS version 3.2, were used to collect and analyze the data. The
cladding wall thickness was divided into 50 -lm thick segments from the outer sur-
face to the inner surface. Hydride orientation 630 from the radial direction was
designated as radial hydrides. Hydrides with a length of 3 lm or greater length were
counted as individual hydrides. Hydride volume fractions were estimated for clad-
ding underneath peeled oxide as well as away from the peeled regions. Results of
the image analysis are presented in Table 2. It should be noted that all of the area
fraction estimates obtained through this analysis could be used for comparison pur-
poses only. As the hydride etching and image analysis settings can make a hydride
appear 10 to 15 times wider than the actual width, all measured area fractions sig-
nificantly overestimated the true hydride area fraction.
TABLE 2 Hydride volume fraction and radial hydride fraction under peeled and non-peeled oxide
of an Optimized ZIRLOTM fuel rod, burnup 72 GWd/MTU, and axial elevation 3 meter.
Azimuthal Orientation Total Relative Relative Hydride

of Photo, (Degrees), Hydride Area Fraction (%) Area Fraction Under Oxide
Peeled (P)/ (Over the Entire Clad (%) (200 lm % Hydrides With
Non-Peeled (NP) Wall Thickness) Outer Region) Radial Orientation
0, NP 16.8 32.5 12.9

45, NP 12.8 23.6 10.5
90, NP 7.3 9.1 5.6
135, NP 10.0 14.8 10.4
180, NP 13.9 23.1 7.8
225, NP 16.1 27.2 11.8
270, NP 9.4 15.5 4.7
315, NP 12.9 23.1 10.3
100–125, P 12.3 27.1 9.8
Because the hydride area fraction under the peeled oxide is within the nor-
mal variation of hydride area fraction under a non-peeled region in Table 2, it
is concluded that OSP does not lead to hydride localization under the peeled
oxide. Comparing the radial hydride fraction data for the peeled and non-
peeled regions in Table 2, it is concluded that OSP does not change radial ori-
entation of the hydrides.
Evaluation of Impact of OSP on Mechanical

Properties of Zirconium Alloys
Axial tensile tests were conducted on high burnup (burnup about 72 GWd/MTU)
Optimized ZIRLOTM and AXIOM clad fuel cladding sections removed from differ-
ent elevations of the fuel rod. Fuel rod sections taken at the 3 -m elevation were
close to the OSP region, whereas sections taken close to 1.5 -m elevation were far
away from the observed OSP elevations. Tensile tests were conducted both at the
room and elevated temperatures, 385 C. The tensile test data are presented in Fig. 8.
As shown in Fig. 8(a), there is no significant difference in the mechanical proper-
ties (yield strength, ultimate tensile strength, uniform elongation, and total elongation)
FIG. 8 Room (RT) and elevated (HT) temperature tensile results of (a) Optimized
ZIRLOTM, and (b) AXIOM fuel cladding irradiated to a burnup of approximately
72 GWd/MTU.
of Optimized ZIRLOTM specimens taken from the fuel rod elevations of 1.5 and
3.0 ms. All irradiated specimens have high ductility. This observation is consistent with
the earlier result of no hydride localization because of OSP. In Fig. 8(b), the ductility of
several AXIOM alloys (X1, X4, and X5) for specimens taken from the 3 m elevation is
high indicating the absence of hydride localization because of OSP in these alloys.
Mechanism of OSP
Based on the limited available OSP data covering ceramographic observations of
layers, circumferential and radial cracks in the oxide layer, and de-lamination of layers,
the following scenario is proposed as a hypothesis for the most likely sequence of
events leading to OSP. At the start of the waterside corrosion process, the first oxide
layer on the metal surface is protective and is adherent to the metal surface. It is defect
free and featureless. Because the density of oxide is lower than that of the metal, as the
corrosion proceeds, tensile stress develops in the metal and compressive stress devel-
ops in the oxide layer. The stresses increase as the oxide thickness increases. Eventu-
ally, the developing high stress breaks the protective nature of the oxide and defects
are generated in the oxide whereby the oxide loses its protective characteristics. These
defects appear as circumferential marks in the oxide. This leads to an increased corro-
sion rate frequently termed as “corrosion rate” transition. This cycle of development
of a protective oxide followed by eventual breakdown of the protective oxide of critical
thickness leading to the corrosion rate transition is repeated leading to multiple oxide
sub-layers separated by circumferentially marked defect lines like cracks.
At the start of irradiation of a fresh fuel rod, there is a designed gap between the
cladding tube inside surface and the pellet outer surface. At this point, the cladding
outer surface is under compressive stress as the coolant pressure is higher than the
fuel rod as-fabricated fill gas pressure. As irradiation continues, the clad creeps inward
because of higher external coolant pressure compared to the lower rod internal pres-
sure. At the same time, the pellet diameter eventually starts increasing because of pellet
swelling from fission product generation. Sometime near the end of the first irradia-
tion cycle or at the beginning of the second irradiation cycle, a hard contact is estab-
lished between cladding tube inside surface and the pellet outer surface. The
micrograph of the tube inside surface region shown in Fig. 9 (which was typical of all
high burnup specimens investigated in the current paper and showing OSP at the tube
outer surface) confirms complete closure of the pellet/cladding gap and formation of
tight interaction layer or layers at the pellet/cladding interface. The nature of interface
interaction layer depends on the cladding alloy composition and the pellet surface
composition. Figure 9 shows Optimized ZIRLOTM cladding with integrated fuel burn-
able absorber (IFBA) UO2 pellets. After the hard pellet/cladding contact, a tensile
stress develops in the outer layer of the oxide, which nucleates a radial crack at the
outer surface of the zirconium oxide corrosion layer. With the increasing tensile stress
associated with the development of the thicker oxide, the radial crack propagates
inward in the oxide layer. When such radial cracks meet the circumferentially oriented
FIG. 9 High-magnification view of the pellet cladding interface of high burnup

Optimized ZIRLOTM clad fuel rod inside surface showing complete closure of the
as-fabricated pellet cladding gap and formation of multi-layers of interaction
phases at the interface.
defect region (linear discontinuous features observed parallel to the tube surface in the
oxide micrographs shown in Fig. 6) associated with the corrosion rate transition, crack
propagation continues in the circumferential direction until it encounters a new radial
crack. The meeting of a circumferential crack and a new radial crack results in the
observed oxide peeling. Because only one thin sub-layer or a limited number of thin
sub-layers are removed, the thermal gradient developed between the peeled and non-
peeled regions is not significant. The development of radial cracks in the thin oxide
outer layer because of pellet cladding hard contact is a critical step in the removal of
few oxide sub-layers in the oxide peeling process. If the oxide is thick, a large number
of sub-layers are removed (as in the case of Zircaloy-4), which is the case in oxide spal-
lation. The critical difference between peeling and spallation is the absence of hydride
localization under peeled oxide (because of insignificant thermal gradients in the
peeled region of a thin oxide) and hydride localization under spalled oxide (because of
significant thermal gradients in the spalled region of a thick oxide). A comparison of
the conditions for oxide peeling and oxide spallation is presented in Table 3.
Although OSP observations so far have been from PWR cladding, in principle,
OSP is possible for BWR high-corrosion-resistant cladding provided similar tensile
stress on oxide and oxide thickness time sequence is encountered.
Why OSP ss Observed in LWR Irradiation and

Not during Autoclave Corrosion
The first step of OSP is the nucleation of radial crack on the outer surface of the
outermost sub-layer of waterside oxide. Circumferential tensile stress in the oxide is
TABLE 3 Difference between oxide spalling and oxide peeling.
Spalling Peeling
Base alloy corrosion resistance Low High

Total waterside oxide thickness >60 lm <45 lm
Hydride localization under dam- Yes No
aged oxide
Clad ductility degradation Significant None
Impact on wear resistance Possibly significant Significant
necessary for such radial crack nucleation. Such tensile stress is generated during
fuel rod irradiation to intermediate burnups because of (a) fuel pellet swelling after
establishment of hard pellet/cladding contact following cladding creep down, and
(b) the presence of heat flux, which creates a cooler cladding outer surface. Irradia-
tion damage to the oxide further enhances crack nucleation in the oxide. In auto-
clave testing, such tensile stress is absent under isothermal conditions without the
tube internal pressure. The second step of OSP is propagation of crack between ox-
ide sub-layers in a direction parallel to the metal surface. The concentration of OSP
at the location of maximum sub-cooled boiling in PWR implies that such boiling
probably promotes this longitudinal crack propagation. In autoclave corrosion test-
ing, such boiling is absent and this could be another reason for not encountering
OSP in autoclave testing.
Impact of OSP on Cladding Wear Resistance

A reduction in cladding wear resistance decreases the margin to the two fuel leakage
mechanisms: grid-to-rod-fretting (GTRF) and fretting because of foreign debris.
Because the waterside oxide layer acts as a protective barrier to both of these leakage
phenomena, OSP may reduce the margin for these two mechanisms. It is necessary
to implement other measures (such as reduced contact loads and decreased vibra-
tions) to achieve adequate wear resistance margins for high corrosion resistance
cladding alloys that are likely to experience OSP.
Possible Parameters to Measure the Degree of

OSP from Oxide Ceramography
OSP is a newly discovered observation. The degree of OSP for a specific cladding
alloy can be estimated from the post-irradiation poolside visual examination or
hot-cell ceramography by evaluation of the following items:
(1) The number of radial cracks on the oxide outer surface.
(2) The number and individual lengths of circumferential cracks at the boundary
between different oxide sub-layers.
(3) Depth of penetration of peeled area through the oxide thickness.

(4) Surface area fraction of the peeled oxide.
(5) Whether peeling removes the most interior protective oxide layer or not.
On the basis of all of the above-listed parameters, the higher corrosion resist-
ance alloys are more prone to OSP than lower corrosion resistance alloys. Conse-
quently, Optimized ZIRLOTM shows a similar susceptibility to OSP as duplex
cladding and AXIOM X1, but less than the other AXIOM alloys.
Impact of OSP on Fuel Performance Parameters

Although quantitative evaluation of the impact of OSP on important fuel performance
parameters would require additional experimental data, the following qualitative state-
ments are possible at this time based on the available information:
(1) OSP has negligible impact on the corrosion resistance of high-corrosion-re-
sistant zirconium alloys as new protective oxide forms rapidly in the region
where OSP has removed a part or most of the oxide layer.
(2) Because OSP does not lead to hydride localization, the impact of OSP on irra-
diated clad ductility is insignificant.
(3) OSP may degrade the grid-to-rod-fretting margin in GTRF-prone fuel
designs. In such cases, a robust modified grid design with improved GTRF
margin would be necessary.
(4) The amount of zirconium oxide material released into the coolant is expected to
be similar for both the low- and high-corrosion-resistant alloys.
Conclusions
The following conclusions are derived from the poolside and hot-cell examination of
several high-corrosion resistance zirconium alloys irradiated to high burnups:
(1) Some fraction of the thin waterside oxide is removed when tensile stress is
applied to the oxide because of pellet/cladding hard contact at high burnups.
This phenomenon is called oxide surface peeling (OSP). The crocodile skin
or fine superficial oxide surface cracks are potentially related to OSP.
(2) OSP is different from the thick oxide spallation previously observed for Zir-
caloys several decades ago.
(3) OSP is not associated with hydride localization and, therefore, has insignifi-
cant impact on the irradiated ductility of the alloys.
(4) OSP has no significant impact on fuel performance except potentially for the
wear resistance of the alloy. Other design modifications would be necessary
for adequate margin against grid-to-rod-fretting and debris fretting.
(5) High corrosion resistance of zirconium alloy cladding increases their suscep-
tibility to OSP.
References
[1] Pan, G., Long, C. J., Garde, A. M., Atwood, A. R., Foster, J. P., Comstock, R. J., Hallstadius,
L., Nuhfer L., and Baranwal, R., “Advanced Material for PWR Application: AXIOM
Cladding,” Proceedings of the 2010 LWR Fuel Performance/TopFuel/WRFPM, Paper

0074, Orlando, FL, Sept 26–29, 2010.
[2] Pan, G., Garde, A. M., and Atwood, A. R., “Performance and Property Evaluation of High
Burnup Optimized ZIRLOTM Cladding,” 17th International Symposium on Zirconium in
the Nuclear Industry, ASTM STP 1543, Feb 3–7, 2013, Hyderabad, India (to be published).
[3] Arborelius, J., Andersson, S., Hallstadius, L., Limback, M., Dahlback, M., Andersson, T.,
Lisdat, R., Hahn, M., and Toscano, E. H., “Duplex Claddings Performance at High Burn-
up,” Jahrestagung Conference, Kerntechnik Düsseldorf, May 25–27, 2004.
[4] Motta, A. T., Yilmazbayhan, A., Comstock, R. J., Partezana, J., Sabol, G., Lai, B., and Cai,
Z., “Microstructure and Growth Mechanism of Oxide Layers Formed on Zr Alloys Studied
With Micro-Beam Synchrotron Radiation,” J. ASTM Int., Vol. 2, No. 5, 2005, Paper ID JAI
12375.
[5] Garde, A. M., Smith, G. P., and Pirek, R. C., “Effect of Hydride Precipitate Localization and
Neutron Fluence on the Ductility of Irradiated Zircaloy-4,” 14th International Symposium
on Zirconium in the Nuclear Industry, ASTM STP 1295, ASTM International, West Consho-
hocken, PA, 1996, pp. 407–430.
DISCUSSION
Q1:—Are there differences between OSP and spalling in terms of dimensions:

diameter of the affected zone, remaining/removed oxide thickness?
Authors’ Response:—The data on irradiated oxide peeled cladding are currently

limited. As shown in Figure 3, the peeled regions are irregularly shaped with dimen-
sions being less than 5 mm. Due to wide scatter and the limited extent of the data-
base, it is difficult to assign an average size. In general, the size of the peeled oxide
regions is smaller than that of spalled oxide. Regarding removed oxide, the depth of
the peeled oxide (several micrometers) is significantly smaller than the depth of the
spalled oxide (several tens of micrometers). Thickness of remaining oxide at the
time of peeling/spalling is difficult to measure as in-reactor oxidation continues af-
ter both peeling and spalling events.
Q2:—The OSP mechanism implies tensile stresses in the oxide. Is there some
OSP during the tensile tests performed to assess the ductility? If not, why should
tensile stresses be biaxial (pellet contact for example)?
Authors’ Response:—The OSP process deals with de-lamination of the thin ox-
ide layer while the in-reactor corrosion is occurring. Tensile stress generated by
hard pellet/cladding contact is sufficient to generate radial cracks (perpendicular to
the metal/oxide, M/O interface) necessary for OSP. Weak zones in the oxide aligned
parallel to the M/O interface (generated by corrosion rate transitions) are needed
for the OSP crack propagation. If an external tensile load is applied to the oxide (as
in a tensile test) de-lamination of oxide similar to OSP can occur. For OSP initia-
tion, a tensile stress in the oxide is needed, but it does not need to be biaxial. Biaxial
tensile stress may extend the peeled (or spalled) oxide area over a greater surface
area due to ease of de-lamination propagation in two orthogonal directions.
Questions from David Schrire, Vattenfall
Q1:—Microcracking appearance at the pellet-pellet interface (Figure 4) looks

similar to features seen previously in more heavily corroding cladding, with locally
thicker oxide and increased hydriding at the pellet-pellet interface. Did you observe
any local increase in oxide at such locations in the Optimized ZIRLOTM? Did you
perform any longitudinal metallography through such a feature?
Authors’ Response:—Longitudinal metallography was performed on Optimized

ZIRLOTM rod covering multiple pellet-pellet interfaces. Somewhat higher cladding
hydrogen content was associated with the pellet-pellet interfaces due to minor axial
cold spots generated at the pellet-pellet interface. The hydride localization at the
pellet-pellet interface was not as strong as was previously observed for low corro-
sion resistance alloys due to lower oxide thickness for Optimized ZIRLOTM. Con-
sistent with the higher hydrogen levels at the pellet-pellet interfaces, the waterside
oxide thickness at the pellet-pellet interface was somewhat higher than that at the
mid-pellet elevations. The lower corrosion resistance of zirconium hydride leads to
thicker oxide at the pellet/pellet interface despite the slightly lower temperature at
the pellet/pellet interface.
Question from N. Ramasubramanian, ECCATEC Inc. Canada:
Q1:—Have you examined the fracture face of the oxide where oxide peeling
had occurred? Peeling implies a tear unlike a brittle fracture.
Authors’ Response:—The fracture surface of the oxide where oxide peeling

occurred was not examined in the hot cell. The non-conducting properties of the
oxide make such examination difficult. The fracture surface of the oxide is normal
to the metal/oxide interface (and not at 45 orientation). From such oxide fracture
surface orientation, it is concluded that oxide fracture mode is brittle and not a duc-
tile tear.
Questions from Antoine Ambard, EDF R&D
Q1:—The main difference between OSP and spallation occurring in-reactor

seems to be the lack of hydride concentration beneath metal/oxide interface. It can
be interpreted as the result of a very small temperature gradient between OSP zone
and virgin-zone. It suggests that the OSP could have occurred when the rod is
removed from the fuel assembly. You have shown:
Visual features on rods suggesting a lack of oxide but it is really impossible

to assert any lack of thickness or increase of thickness from this simple ob-
servation (not at pellet-pellet interface)
Micrographs showing a few micrometers removed from the surface of

specimens extracted from the fuel assembly may be from defueling and
preparation.
How do you relate the visual features shown on rods and metallograhic exami-
nations? How do you assert that OSP has occurred in-reactor?
Authors’ Response:—Evidence of OSP was clear in poolside examinations prior

to the removal of fuel rods from the fuel assembly for shipment to the hot cell.
Therefore, it is clear that fuel rod removal was not the primary cause of OSP. Fuel
rod removal may have slightly altered the extent of OSP. The oxide thicknesses
reported in the paper are not based on the visual examination results but more de-
finitive non-destructive examination by eddy current testing by poolside examina-
tion and destructive examination by metallography at the hot cell. Metallography
was conducted without defueling. Although special precautions of specimen edge
preservation were taken for the thin oxide layer specimens, some damage to the
thin oxide during specimen metallographic preparation cannot be completely ruled
out. The axial elevation of the fuel rod was matched between poolside evidence of
OSP and hot cell examination of oxide de-lamination. After establishing correspon-
dence between poolside visual examination features with cladding oxide de-lamina-
tion, it is concluded that OSP had occurred in-reactor.
Q2:—You have shown so-called OSP located at two locations: a) along the axial
scratch formed during rod loading and b) at the pellet-pellet interface. Is the cause
of OSP at the two locations the same? What is the basis for your conclusion?
Authors’ Response:—The primary cause of the peeling is the tensile stress

applied to the oxide layer. The tensile stress at the inter-pellet region is the cladding
volume expansion at the pellet-pellet interface caused by hydride localization at
that interface due to axial temperature gradient. The association of OSP with the
axial fuel rod loading scratch is due to preferential site for radial crack initiation in
the oxide at the axial scratch. Tensile stress in the oxide is the common cause. How-
ever the predominant cause for the generation of tensile stress is different for the
two locations. Pellet swelling is the dominant cause for the tensile stress at mid-pel-
let locations at different axial locations of the rod along the loading scratch.
Q3:—At the pellet-pellet interface, depending on power history, you can find
ridges especially at high burnup. Have you measured the ridges prior to metallogra-
phy? Have you considered the possibility that these ridges can generate the radial/
circumferential cracks seen at the surface of the rods?
Authors’ Response:—Ridges are seen at the pellet-pellet interfaces. Ridges are

more easily visible in the presence of crud. Ridges were also detected during diame-
tral profilometry along the length of the high burnup fuel rod. Such ridges can con-
tribute to the radial/circumferential cracks seen on the rod surface.
L. Walters,1 G. A. Bickel,2 and M. Griffiths2
The Effects of Microstructure and

Operating Conditions on
Irradiation Creep of Zr-2.5Nb
Pressure Tubing
Reference
Walters, L., Bickel, G. A., and Griffiths, M., “The Effects of Microstructure and Operating
Conditions on Irradiation Creep of Zr-2.5Nb Pressure Tubing,” Zirconium in the Nuclear Industry:
17th International Symposium, STP 1543, Robert Comstock and Pierre Barberis, Eds., pp. 693–
725, doi:10.1520/STP154320120158, ASTM International, West Conshohocken, PA 2015.3
ABSTRACT
Creep experiments have been performed on biaxially stressed 10 mm diameter
Zr-2.5Nb capsules. As the pressurized capsules were obtained from micro-
pressure tubes, which were fabricated by the same process as CANDU power
reactor pressure tubes, they have a similar microstructure to that of the full-size
tubes. The experiments were performed in the OSIRIS test reactor at nominal
operating temperatures ranging from 553 and 613 K in fast neutron fluxes up to
2 1018 nm–2s–1 (E > 1 MeV). Diametral and axial strains are reported as
functions of fluence for specimens internally pressurized to hoop stresses from 0
to 160 MPa and irradiated to 26.5 dpa. The effects of microstructure,
temperature, and cold work on irradiation creep are shown. The analysis of
OSIRIS data combined with data from in-service CANDU tubes has revealed
some significant observations regarding pressure tube deformation: (i) that
irradiation creep anisotropy varies with temperature, (ii) texture appears to have
a more significant effect on axial creep than on diametral creep, (iii) diametral
strain appears to be strongly dependent on grain size and aspect ratio, and (iv)
September 24, 2014.
1
Atomic Energy of Canada Limited, Chalk River Laboratories, Chalk River, ON, Canada, K0J 1J0, Canada
(Corresponding author) e-mail: waltersl@aecl.ca
2
Atomic Energy of Canada Limited, Chalk River Laboratories, Chalk River, ON, Canada, K0J 1J0, Canada.
3
Hyderabad, India.
Copyright V
694 STP 1543 on Zirconium in the Nuclear Industry
that whereas cold-work correlates with the axial creep of the capsules, there
appears to be no statistically significant dependence of diametral creep on cold-
work.
Keywords
Zr-2.5Nb alloy, pressure tube, neutron irradiation, irradiation creep, microstruc-
ture, in-reactor deformation
Introduction
In a CANDU4 reactor, the fuel bundles and primary coolant are contained within
Zr-2.5Nb pressure tubes that are approximately 6.3 m in length, have an internal di-
ameter of 104 mm and a wall thickness of 4.2 mm. The CANDU 6 design has 380
horizontal fuel channels producing a total electrical power output of about
700 MW. During service, the pressure tubes operate at temperatures between 538 K
and 583 K, and with inlet coolant pressures up to 11.5 MPa corresponding to initial
hoop stresses of 140 MPa. Pressure tube lifetimes can reach 30 years. The maximum
flux of fast neutrons from the fuel is about 4 1017 nm–2s–1 (E > 1 MeV).
In-reactor deformation of the Zr-2.5Nb pressure tubes results from the com-
plex interactions affecting the material including the tube’s microstructure, reactor
operating conditions, and geometrical constraints. The consequent steady-state
pressure tube deformation has previously been described in terms of three separa-
ble, additive components: thermal creep, irradiation creep, and irradiation growth
[1]. To a large extent, the irradiation-induced components have been considered to
be anisotropic and contribute to length as well as diameter changes, while thermal
creep has been considered to be isotropic for the ranges of stresses being considered
and does not contribute significantly to length changes [2].
The diametral expansion and elongation of the pressure tubes are important
behaviours that limit their useful life and the maximum power level for reactor opera-
tion. To further the understanding of the in-reactor deformation behaviour of pressure
tube material, a series of accelerated irradiation deformation tests were conducted using
bi-axially stressed creep capsules in the OSIRIS high flux reactor located at the Com-
missariat a l’Energie Atomique (CEA) laboratories in Saclay, France. Early results from
these studies [2,3] provided input to the development of an equation to predict pressure
tube deformation [4]. These early results have now been supplemented with more
recent data extending the available range of temperature, crystallographic texture, and
cold work data. The effects of these variables on diametral and axial strain rates have
been analyzed with the larger dataset, providing further insights into the dependence of
irradiation creep on temperature and microstructure.
In addition to the experimental program, deformation measurements from
hundreds of pressure tubes operating in reactors for the past three decades have
4
CANDU (CANada Deuterium Uranium) is a registered trademark of Atomic Energy of Canada Limited
WALTERS ET AL., DOI 10.1520/STP154320120158 695
been analyzed. A combination of the experimental and in-service observations is

providing a consistent description of the deformation process that is leading to a
better understanding of the effect of microstructure. This has provided the basis for
developing new models that explicitly include microstructure variables, such as
grain dimensions (size and shape) as part of the parametric input.
This paper describes and compares the results of the experimental program
and the in-service pressure tube observations. All results from OSIRIS (some previ-
ously published [2,3]) are described and compiled. Examples from in-service data
are used to support the conclusions drawn from the OSIRIS results, but more
importantly, demonstrate strong linkages between tube microstructure and diame-
tral strain. In particular, the effect of grain size/shape on diametral strain lends sup-
port to the use of a diffusional mass-transport model that predicts the diametral
strain based on interstitial/vacancy diffusion towards grain boundary sinks.
Creep Capsule Irradiation Experiment at OSIRIS

CREEP CAPSULE FABRICATION
Creep capsules used in the OSIRIS experiments were obtained from sections of Zr-
2.5Nb micro-pressure tubes. These 1.5 m long tubes were fabricated using a manu-
facturing route that was similar to that used to produce most of the full size pres-
sure tubes installed in CANDU reactors as shown in Fig. 1. Each capsule used in this
experiment had a total length of 46.4 mm including end-caps, outside diameter of
FIG. 1 Manufacturing process flow outline for Zr-2.5Nb pressure tubes.

10.0 mm, and a wall thickness of 0.45 mm. End-caps, machined from Zr-2.5Nb
rods, were electron-beam welded to the tubes, and the specimens were heated for a
total of 48 h at 673 K to relieve residual stresses. The capsules were pressurized
using high-purity helium gas and sealed by tungsten inert gas welding. The operat-
ing hoop stresses for each tube were calculated from the internal pressure at room
temperature assuming that the helium behaves as an ideal gas.
Microstructure information for each creep capsule (including crystallographic
texture [5] and the hai and hci component-type dislocation densities estimated
from broadening of the prism and basal X-ray diffraction peaks [6]) are summar-
ized in Table 1. The creep capsules’ microstructure consisted of a-Zr grains flattened
in the radial direction and elongated in the transverse and axial directions with
mostly thin b-phase ligaments separating the a-phase grains. Although most creep
capsules had a similar crystallographic texture to the CANDU pressure tubes (with
a majority of the resolved fractions of basal plane normals in the transverse direc-
tion), capsules with texture similar to fuel sheathing (strong radial texture) were
also included in the experiment in an effort to further extend the texture range.
EXPERIMENTAL PROCEDURES
The irradiation of the creep capsules in the OSIRIS reactor was carried out using a
stainless steel insert. This device was a sealed tubular heating furnace containing
the specimens mounted on a holder with NaK used as the heat transfer medium
and external heaters used to provide a uniform temperature distribution along the
length of the insert and to compensate for changes in reactor power. The inserts
consisted of seven tiers, with each tier containing one creep capsule and six growth
specimens. The temperature of each tier was monitored by two type-K
thermocouples.
Table 2 summarizes the nominal operating conditions in the OSIRIS creep cap-
sule experiments. The fast neutron fluences were determined from flux monitors at
various locations in each insert. The time-averaged neutron flux was determined by
dividing fluence by the time at power. An example of the vertical fluence variation
in the irradiation insert is shown in Fig. 2. CEA provided an APPOLLO flux spec-
trum for the test sites used in these experiments. Using this data, the Zr-2.5Nb
creep capsule damage rate per fast neutron flux (E > 1 MeV) was determined to be
6.56 1024 nm–2. The displacement rate was calculated using the cross sections
from the ENDF/B-VI database [7]. If it is assumed that the same dpa produces the
same amount of strain, then the neutron fluxes above 1 MeV from OSIRIS must be
reduced by a factor of 0.907 before being used in comparison with CANDU pres-
sure tubes.
STRAIN MEASUREMENTS
Measurements of the creep capsule dimensions after each phase of the irradiations
were taken at CEA using the underwater linear variable differential transformer
(LVDT) system in the OSIRIS reactor pool at a nominal temperature of 313 K. This
TABLE 1 Dislocation density and crystallographic texture.
Dislocation Densities (1 1014 m–2) Texture
hai component hci component

Capsule 1010
1120 f0002g fR fT
Trillium-2 Experiment
K (Fuel Sheath 2.3 — 0.6 0.56 0.37
Texture)
M18 4.01 1.62 0.47 0.342 0.564
M03 4.01 1.62 0.47 0.334 0.596
M17 4.01 1.62 0.47 0.342 0.564
M14 4.01 1.62 0.47 0.342 0.564
M13 4.01 1.62 0.47 0.342 0.564
B (Fuel Sheath 2.3 — 0.6 0.56 0.37
Texture)
L17 (Fuel 2.3 — 0.6 0.56 0.37
Sheath
Texture)
M15 4.01 1.62 0.47 0.342 0.564
M04 4.01 1.62 0.47 0.334 0.596
M02 4.01 1.62 0.47 0.334 0.596
M16 4.01 1.62 0.47 0.342 0.564
M01 4.01 1.62 0.47 0.334 0.596
L16 (Fuel 2.3 — 0.6 0.56 0.37
Sheath
Texture)
U8 (26 % CW) 3.4 1.64 0.36 0.472 0.486
U2 (6 % CW) 2.15 1.33 0.24 0.444 0.505
U1 (6 % CW) 2.15 1.33 0.24 0.444 0.505
U4 (12 % CW) 2.33 1.46 0.30 0.457 0.495
U7 (26 % CW) 3.4 1.64 0.36 0.472 0.486
U3 (12 % CW) 2.33 1.46 0.30 0.457 0.495
U5 (18 % CW) 2.22 1.46 0.29 0.469 0.486
Cardinal-1 Experiment
B11 4.21 3.73 1.17 0.421 0.545
B1 2.28 1.27 0.29 0.395 0.567
B8 2.31 1.15 0.29 0.4 0.56
B7 2.31 1.15 0.29 0.4 0.56
B14 4.21 3.73 1.17 0.421 0.545
B12 4.21 3.73 1.17 0.421 0.545
B5 2.31 1.15 0.29 0.4 0.56
TABLE 1 (Continued)
Dislocation Densities (1 1014 m–2) Texture
hai component hci component

Capsule 1010
1120 f0002g fR fT
B21 2.37 0.69 0.16 0.423 0.527
B10 2.31 1.15 0.29 0.387 0.572
B20 2.39 1.15 0.29 0.390 0.570
B22 2.37 0.69 0.16 0.423 0.527
B23 4.21 3.73 1.17 0.421 0.545
B9 2.31 1.15 0.29 0.387 0.572
B19 2.39 1.15 0.29 0.390 0.570
CANDU Pressure Tubes

Front 3.0 2.0 0.5 0.29860.029 0.64260.029
Back 3.0 2.0 0.5 0.3560.029 0.60560.029
system was used to measure creep capsules from Trillium-2, 3, 5, and the first phase
of the Cardinal-1 experiments, at which point it was retired due to mechanical
wear-out. For measurements of the capsules in the remaining phases of the
Cardinal-1 and in the Cardinal-2 experiments, the CEA SEMI (Service d’Études des
Materiaux Irradies) Laboratory non-contact optical silhouette system (Hommel)
was used. Tables 3–7 summarize the diametral and axial strain measurements for
each phase of the experiments.
Using the underwater LVDT bench method, diameter readings were made at
six axial positions along the capsules, each measurement being the average of 200
readings around the circumference of the capsule. The diameter values were deter-
mined from an average of measurements taken from three axial locations, at the
axial centerline and 5mm on either side of the centerline. The length was obtained
by averaging 200 end-cap to end-cap measurements. Since the end-caps are not
subjected to a biaxial stress state, this introduces a bias into the axial strain estimate.
No attempt was made to correct this bias.
Using the optical method, the diameter was measured at 17 axial locations. The
diameter values used in this analysis were taken from an average of the middle nine
measurements. The length was measured from end-cap to end-cap of the
specimens.
A standard specimen was measured before and after each irradiated specimen,
and the readings normalized to 313 K. This measuring procedure was repeated two
times on each specimen. The error in the measurement with the LVDT system of
the diameter and length of the capsules was estimated (95 % confidence level) as
TABLE 2 Operating conditions and strain.
Ave Ave Flux Diametral Creep Axial Creep

Stress Temp (E > 1 MeV) Strain Rate Strain Rate
Capsule MPa K 1 1018 n m–2 s–1 1x1028 m2 n–1 1 10–28 m2 n–1
K (Fuel sheath 147.0 553 1.75 2.24 0.14
texture)
M18 123.2 559 1.93 0.59 0.98
M03 0 540 2.02 –0.45 0.41
M17 160.0 554 2.02 0.89 1.23
M14 82.5 556 1.93 0.28 0.71
M13 43.3 560 1.79 –0.10 0.55
B (Fuel sheath 0 549 1.68 –0.14 0.32
texture)
L17 (Fuel 155.0 587 1.77 3.98 0.16
sheath texture)
M15 120.4 585 1.93 1.01 1.07
M04 0 593 2.02 –0.28 0.24
M02 159.0 579 2.02 1.52 1.47
M16 81.7 595 1.94 0.45 0.75
M01 40.2 586 1.8 0.03 0.48
L16 (Fuel 0 586 1.59 –0.11 0.15
sheath texture)
U8 (26 % CW) 140.8 586 2.06 2.25 0.63
U2 (6 % CW) 139.2 580 2.30 1.64 0.39
U1 (6 % CW) 139.7 582 2.43 1.53 0.33
U4 (12 % CW) 139.5 581 2.46 1.53 0.44
U7 (26 % CW) 142.2 592 2.39 1.92 0.60
U3 (12 % CW) 139.9 583 2.22 1.60 0.42
U5 (18 % CW) 140.0 583 1.95 2.53 0.53
B11 99.9 592 1.64 1.50 0.46
B1 0 592 1.84 –0.05 –0.02
B8 125.7 591 1.96 1.73 0.63
B7 75.0 593 1.98 1.08 0.48
B14 0 570 1.92 0.02 0.11
B12 124.8 591 1.78 1.62 0.63
B5 100.5 594 1.58 1.66 0.44
B21 0 611 1.68 0.18 –0.25
B10 124.3 608 1.83 2.96 0.69
TABLE 2 (Continued)
Ave Ave Flux Diametral Creep Axial Creep

Stress Temp (E > 1 MeV) Strain Rate Strain Rate
Capsule MPa K 1 1018 n m–2 s–1 1x1028 m2 n–1 1 10–28 m2 n–1
B20 98.2 599 1.92 1.49 0.63

B22 74.7 612 1.92 1.29 0.27
B23 0 591 1.87 0.06 0.12
B9 125.3 612 1.72 2.26 0.59
B19 100.1 614 1.54 1.69 0.52
62 lm and 65 lm, respectively. For the optical measurements, the uncertainties

are 61 lm and 63 lm for the diameter and length measurements, respectively.
Results of Creep Capsule Irradiation in OSIRIS

The observed creep capsule strain is affected by the operating conditions (fluence,
hoop stress, and temperature) and microstructure. Assuming the dependence on
fluence and hoop stress is linear, and expressing the results in terms of the creep
compliance (strain per unit fluence per unit stress), the remaining variables (tem-
perature and microstructure) can then be studied with respect to the creep
compliance.
FIG. 2 Variation of the fast fluence in the OSIRIS irradiation insert for the Cardinal-1
experiment.
TABLE 3 Diametral and axial strain in the Trillium-2 experiment.
Fluence 1 1025 nm–2 Diametral Axial

(E > 1 MeV) Strain Strain
Capsule K
0.95 0.00236 0.00033
2.17 0.00533 0.00028
4.04 0.00942 0.00032
6.81 0.01584 0.00043
9.91 0.02240 0.00092
12.44 0.02809 0.00143
14.86 0.03363 0.00205
16.96 0.03853 0.00256
Capsule M18
2.01 0.00146 0.00151
5.0 0.00364 0.00419
8.37 0.00566 0.00738
11.23 0.00732 0.01014
13.9 0.00870 0.01292
16.14 0.00983 0.01531
Capsule M03
2.11 –0.00089 0.00047
5.21 –0.00207 0.00151
8.71 –0.00347 0.00289
11.74 –0.00480 0.00408
14.53 –0.00628 0.00535
16.83 –0.00749 0.00646
Capsule M17
2.12 0.00238 0.00207
5.21 0.00549 0.00561
8.71 0.00867 0.00987
11.75 0.01128 0.01358
14.53 0.01362 0.01713
16.83 0.01552 0.02019
Capsule M14
2.02 0.00098 0.00109
5.0 0.00218 0.00291
8.37 0.00320 0.00532
11.27 0.00390 0.00741
13.93 0.00455 0.00940
16.16 0.00505 0.01111
TABLE 3 (Continued)

Capsule M13
1.88 0.00017 0.00063
4.65 0.00027 0.00180
7.77 –0.00003 0.00359
10.4 –0.00042 0.00510
12.84 –0.00070 0.00651
14.93 –0.00100 0.00764
Capsule B
0.84 –0.00012 0.00073
1.91 –0.00056 0.00122
3.61 –0.00054 0.00160
6.08 –0.00088 0.00222
8.86 –0.00123 0.00316
11.11 –0.00158 0.00399
13.23 –0.00192 0.00479
15.13 –0.00224 0.00541
FLUENCE DEPENDENCE
Diametral and axial strain were found to increase approximately linearly with flu-
ence up to 1.7 1026 nm–2. Figures 3–6 show the diametral and axial strain against
fluence for creep capsules in the Cardinal-1&2 irradiation. The primary creep tran-
sient was very small, almost negligible, in all cases. Steady-state strain rates (per
unit fluence) for each capsule summarized in Table 2 were obtained by taking the
slope of strain versus fast neutron fluence (E > 1 MeV).
CALCULATION OF CREEP COMPLIANCE

Figures 7 and 8 show diametral and axial strain rates versus hoop stress. The stress
dependence of irradiation creep on pressure tube materials up to 160 MPa is
approximately linear, i.e., e_ / rn where the stress exponent n1 and the strain rate
is expressed per unit fluence (nm–2). However, the capsules still deform at zero
stress values due to irradiation growth. Growth strain of an hcp crystal is character-
ized by shrinkage along the c-axis (diametral direction) and expansion along the
a-axis (axial direction). The growth strain rate is known to be affected by texture
as well as temperature [1]. The growth strain rate in the diametral direction
increases as the transverse basal texture component decreases (fT–fR decreases) and
as the temperature increases. Growth strain rate in the axial direction increases as
the axial/longitudinal basal texture component (fL) decreases and as the tempera-
ture decreases. The diametral growth strain rate of the capsules follows the expected
trend [1] so the OSIRIS zero stress capsules were fitted as a function of texture

Capsule L17
1.27 0.00458 0.00051
3.17 0.01023 0.00064
5.99 0.02139 0.00096
8.53 0.03145 0.00133
10.84 0.04179 0.00178
14.5 0.05646 0.00267
Capsule M15
1.37 0.00179 0.00131
3.43 0.00355 0.00283
6.51 0.00674 0.00576
9.3 0.00953 0.00866
11.85 0.01236 0.01176
15.86 0.01624 0.01680
Capsule M04
1.43 –0.00012 0.00014
3.58 –0.00048 0.00034
6.78 –0.00112 0.00075
9.7 –0.00185 0.00138
12.37 –0.00270 0.00220
16.57 –0.00436 0.00378
M02
1.43 0.00252 0.00180
3.58 0.00519 0.00400
6.76 0.01075 0.00824
9.67 0.01533 0.01243
12.34 0.01963 0.01691
16.58 0.02496 0.02400
Capsule M16
1.37 0.00104 0.00099
3.44 0.00187 0.00213
6.48 0.00340 0.00415
9.27 0.00460 0.00618
11.82 0.00576 0.00836
15.96 0.00752 0.01188
Capsule M01
1.28 0.00042 0.00055
3.2 0.00055 0.00117
TABLE 4 (Continued)

5.98 0.00072 0.00236

8.55 0.00073 0.00359
10.87 0.00087 0.00489
14.76 0.00077 0.00688
Capsule L16
1.14 –0.00004 0.00053
2.85 –0.00029 0.00077
5.27 –0.00051 0.00103
7.52 –0.00075 0.00133
9.53 –0.00093 0.00177
13.04 –0.00147 0.00237
(fT–fR) and temperature to allow prediction of the diametral growth contribution

for any individual stressed creep capsule. However, fL did not change significantly,
and the axial trend with decreasing fL was inconsistent with the expected trend [1];
as such the axial growth strain rates were fitted as a function of temperature only to
establish the predicted axial growth strain rate. To determine a creep strain rate per
unit stress (the creep compliance) for individual creep capsules, the predicted
growth strain rate was first subtracted from the creep strain rate before dividing by
the stress.
EFFECTS OF TEMPERATURE ON IRRADIATION CREEP

Early results from the OSIRIS creep capsule experiments where temperature ranged
from 553 to 583 K led to the assumption that the irradiation creep anisotropy pa-
rameter is independent of temperature [1–4]. However, a re-assessment of the phe-
nomena using data corresponding with a larger range of temperatures indicates
that irradiation creep anisotropy varies with temperature.
It was recognized that the creep capsules were not randomized with respect to
microstructure (namely texture). Since temperature and texture are independent
from each other, i.e., heat was applied to the capsules from external and nuclear
heating and microstructure was determined during the fabrication process, a multi-
variate analysis was performed by fitting creep compliance as a linear function of
temperature and texture to extract the unbiased temperature dependence and the
unbiased texture dependence. For this work, the radial texture parameter [5], fR,
was sufficient to quantify the texture variation from capsule to capsule, i.e., fL did
not change significantly so either fR or fT could be used alone to describe the texture
variation. Projections of the data on the fitted surface at a fixed temperature or at a
fixed texture are used in this paper to illustrate the unbiased trends. Figure 9 shows

Capsule U8 (26 % CW)

1.69 0.00635 0.00091
4.06 0.01203 0.00230
6.69 0.01763 0.00407
Capsule U2 (6 % CW)
1.88 0.00306 0.00080
4.52 0.00787 0.00176
7.46 0.01225 0.00299
Capsule U1 (6 % CW)
1.99 0.00304 0.00072
4.79 0.00742 0.00156
7.90 0.01211 0.00266

2.01 0.00301 0.00089
4.85 0.00772 0.00203
8.01 0.01223 0.00350

1.95 0.00404 0.00109
4.70 0.00944 0.00261
7.78 0.01526 0.00456

1.81 0.00285 0.00082
4.36 0.00705 0.00180
7.23 0.01150 0.00310

1.59 0.00335 0.00086
3.82 0.00892 0.00190
6.34 0.01537 0.00338
the effect of temperature on creep compliance at a fixed texture (fR ¼ 0.4). From
this figure, it appears that over the ranges of 553 to 613 K the diametral creep strain
rate has a significant positive temperature dependence, whereas the axial creep
strain rate is more weakly dependent on temperature. The higher temperature data
show greater variance from capsule to capsule. This is expected as the magnitude of
the variance depends on strain rate. Temperature does not contribute to axial creep
strain to the same degree that it does for diametral creep strain, i.e., the effect of
TABLE 6 Diametral and axial strain in the Cardinal-1 experiment.
Fluence 1x1025 nm–2 Diametral Axial

Capsule B11
1.4 0.00233 0.00069
2.94 0.00478 0.00151
4.49 0.00698 0.00212
Capsule B1
1.56 –0.00016 –0.00002
3.34 –0.00018 0.00021
5.03 –0.00032 –0.00008
Capsule B8
1.66 0.00265 0.00106
3.6 0.00603 0.00238
5.36 0.00905 0.00338
Capsule B7
1.68 0.00181 0.00065
3.65 0.00413 0.00172
5.41 0.00582 0.00244
Capsule B14
1.63 –0.00028 0.00030
3.53 –0.00022 0.00067
5.24 –0.00036 0.00070
Capsule B12
1.52 0.00265 0.00103
3.26 0.00541 0.00219
4.86 0.00805 0.00314
Capsule B5
1.35 0.00185 0.00069
2.84 0.00438 0.00157
4.31 0.00677 0.00200
temperature on irradiation creep compliance at stresses below 160 MPa appears to

be directionally dependent.
EFFECT OF COLD WORK AND DISLOCATION DENSITY ON IRRADIATION

CREEP
The Trillium-5 experiment was designed to evaluate the in-reactor deformation
rate as a function of the amount of cold work introduced after the tubes had been
extruded at high temperature. The Trillium-5 capsules with cold work ranging
TABLE 7 Diametral and axial strain in the Cardinal-2 experiment.

Capsule B21
1.38 0.00004 0.00029
3.27 0.00070 0.00074
5.01 0.00070 0.00118
Capsule B10
1.56 0.00551 0.00114
3.61 0.01218 0.00269
5.47 0.01706 0.00383
Capsule B20
1.67 0.00320 0.00059
3.81 0.00701 0.00222
5.72 0.00920 0.00314
Capsule B22
1.69 0.00249 0.00050
3.83 0.00569 0.00103
5.74 0.00768 0.00160
Capsule B23
1.65 –0.00008 –0.00022
3.71 0.00045 –0.00042
5.57 0.00016 0.00028
Capsule B9
1.49 0.00347 0.00082
3.39 0.00840 0.00195
5.14 0.01169 0.00298
Capsule B19
1.34 0.00243 0.00080
3.01 0.00599 0.00173
4.61 0.00795 0.00251
from 6 to 26 % were tested at one nominal stress value (140 MPa) and one nominal
temperature (583 K) up to an accumulated fluence of 8 1025 nm–2 (E > 1 MeV)
corresponding to 12 dpa.
Pressure tubes in current CANDU reactors are cold drawn after extrusion
(nominal cold work is 27 %). These operations produce the microstructure that
affects many of their properties. Although the dominant effect is from the extru-
sion, the amount of cold work is also thought to affect properties as it produces a
dislocation substructure which increases the strength. Table 1 shows that in the
FIG. 3 Diametral strain versus fluence for creep capsules irradiated in the Cardinal-1
experiment.
Trillium-5 experiment both the hai and hci component dislocations increased with
cold work.
The Trillium-5 creep capsules, with varying amounts of cold work, were not
randomized with respect to microstructure and temperature. In fact, the deviations
from the nominal design temperature were significant. Figure 10 shows the effects of
a fourfold increase in cold work on the creep compliances normalized for
FIG. 4 Axial strain versus fluence for creep capsules irradiated in the Cardinal-1
experiment.
FIG. 5 Diametral strain versus fluence for creep capsules irradiated in the Cardinal-2
experiment.
temperature and texture. Although axial strain rate is strongly dependent on cold
work, there is no clear evidence to indicate that the diametral strain rate is depend-
ent on cold work, i.e., dislocation density.
EFFECTS OF TEXTURE AND GRAIN SIZE ON IRRADIATION CREEP

The microstructure and crystallographic texture of Zr-2.5Nb present in operating
pressure tubes originates from the extrusion process during manufacture. The end
FIG. 6 Axial strain versus fluence for creep capsules irradiated in the Cardinal-2
experiment.
FIG. 7 Measured diametral strain rate (m2n–1) versus hoop stress (MPa) for micro-
pressure tube and fuel sheath (FS) textured capsules.
of the pressure tube which emerges first from the extrusion press and is the first to
enter the die during cold drawing operations is referred to as the front end. The
grain size is generally smaller at the back end of the pressure tube in comparison
with the front end due to cooling that occurs in the billet during extrusion. The
concentration of basal plane normals distributed in the radial/transverse direction
is predominantly oriented in the transverse direction. The basal texture parameter
FIG. 8 Measured axial strain rate (m2n1) versus hoop stress (MPa) for micro-pressure
tube and fuel sheath (FS) textured capsules.
FIG. 9 Creep compliance normalized to fR ¼ 0.4 versus temperature. Dashed lines are
the 95 % confidence interval on the mean value.
FIG. 10 Creep compliance normalized to temperature 585 K and texture fR ¼ 0.4 versus
% cold work for the Trillium-5 creep capsules. The slope for diametral strain
dependence is not significant, p ¼ 0.64. The slope for axial dependence on cold
work is significant, p < 0.001.
then tends to be higher at the front ends of the pressure tube. During service, the
back-end achieves significantly higher diametral strain than the front-end under
identical fluence, stress, and temperature conditions [8]. Therefore, texture and/or
grain size are anticipated to affect diametral strain. Furthermore, in-service pressure
tube elongation is strongly dependent on the average radial or transverse texture
along the length of a pressure tube [9].
Early analysis [2,3] of the Trillium-2&3 experiments, where the strain rates
from fuel sheath (FS) textured capsules (fR > fT) were compared against those of
micro-pressure tube (MPT) capsules (fT > fR), suggested that both diametral and
axial strain rate are dependent on the crystallographic texture. However, it should
be noted that there were significant differences between the fabrication methods for
the FS and MPT capsules [2,3], which resulted in different grain sizes and micro-
structures. Notably, the grain size aspect ratio (radial grain thickness/transverse
grain width) of the FS capsules (0.7) is approximately an order of magnitude larger
than that from the Trillium-2&3 MPT capsules (0.08) [2,3].
Figure 11 shows the effect of texture on the creep compliance normalized for
temperature (585 K). The data indicates that axial strain rate is dependent on tex-
ture and consistent with the observations from in-service pressure tube elongation
[9]. The effect of texture on the diametral strain rate in Fig. 11 has a high degree of
uncertainty. Were it not for the FS textured capsules, a dependence on texture
would not be evident for diametral strain.
FIG. 11 Creep compliance normalized to 585 K versus texture parameter fR. Excluding
the fuel sheath textured capsules (FS; having fR 0.56), the slope for diametral
strain dependence is not significant, p ¼ 0.27 and the slope for axial strain
dependence on fR is significant, p ¼ 0.002.
Grain Size and Texture Effects for In-Service

Pressure Tubes
Recent work [10] utilizing a multivariate approach that considered both texture and
grain size of in-service pressure tubes as drivers of diametral strain suggested that
grain size cannot be ignored. The wealth of data from operating CANDU reactors
can help resolve the effects of texture and grain structure on diametral strain for in-
service pressure tubes and the OSIRIS creep capsules. Periodic inspections of pres-
sure tubes have provided diametral gauging data for hundreds of pressure tubes.
Within the population of pressure tubes, subsets can be found that differ in diame-
tral strain behaviour and differ in microstructure. One such pair of subsets was
manufactured under the label of H-series and HM-series pressure tubes. The H-
series tubes were manufactured following the route shown in Fig. 1. The production
route for the HM-series was modified by moving the water b-quench step after tre-
panning the log section into a hollow billet.
Pressure tubes from the H- and HM- manufacturing series were shared
between a number of reactor units and the HM-series tubes always exhibit a higher
diametral strain than the H-series tubes. Figure 12 shows the diametral strain for the
H- and HM-series tubes found in four CANDU units. The data plotted in Fig. 12 are
for the strain observed at 1 m inboard from the inlet. All of these pressure tubes ex-
perience the same coolant temperature (542 6 2 K) and the same coolant pressure
(10.7 6 0.2 MPa) at this axial location with only fast flux being different between
the tubes. Plotting the strain per unit fluence for the same axial location then allows
a direct comparison of the relative strain rate behaviour of the two populations of
pressure tubes independent of the operating conditions. The strain differences
between the tubes are then attributable to variability in the material properties.
FIG. 12 Measured diametral strain for H- and HM-series pressure tube populations at 1 m
from inlet (front end), coolant temperature ¼ 542 K, coolant pressure ¼ 10.7 MPa.
Figure 13 shows the texture parameters for offcut material randomly sampled
from H- and HM-series tubes. This offcut material comes from the front end of the
tube, at the inlet, and therefore is considered to be a reasonable surrogate for the
texture expected 1 m away at the location of the strain measurements in Fig. 12.
Figure 14 shows distributions for the dislocation density of these two populations
measured by X-ray diffraction [6] from the same offcuts. The two populations show
no substantive differences in texture parameters or dislocation density yet the dia-
metral strain for the two populations of tubes remains distinct.
The remaining microstructure variable of interest is grain size/shape. Grain
structure images by transmission electron microscopy (TEM) have been obtained
for some of these tubes. Two of the more extreme cases are shown in Fig. 15. Tube
H0081M (HM-series) exhibits 50 % higher diametral strain than Tube H1848 (H-
series). Again, these images are of material from the nearby front end offcuts. While
the radial grain thickness is similar in both cases, the transverse grain width is
much reduced in the HM-series tube. The major (transverse) and minor (radial)
axes of the grains were measured from these images and are also plotted in Fig. 15.
This quantitative representation of the aspect ratio distributions clearly supports
the visual impression from the TEM images. Figure 15 shows that the mean radial
grain thickness is slightly larger for H0081M compared to H1848, while the mean
transverse grain width for H0081M is about a factor of two narrower than H1848.
The diametral strain can then be shown to be strongly correlated with the mean
grain aspect ratio in Fig. 16. The strain is normalized to the mean strain of the H-
series tubes (solid line in Fig. 12) and plotted as relative diametral strain in Fig. 16.
The relative strain is provided in Table 8 along with the measured mean aspect ratios
and the texture of the nearby offcuts. Although there is a slight univariate correla-
tion between texture and aspect ratio evident in Table 8, the two variables are not
co-linear when viewed as a multivariate system (strain as a function of both aspect
ratio and texture) and diametral strain is dominated by aspect ratio and dependence
on texture is not evident.
A Rate Theory Model for Diametral Creep

The strong dependence on grain aspect ratio and size observed for diametral creep
of in-service pressure tubes suggests that a rate theory model [11] is an appropriate
descriptor of the diametral creep behaviour. The neutron irradiation creates inter-
stitials and vacancies that migrate to sinks such as dislocations, grain boundaries,
and in some cases, cavities (neutral sinks). The net flux to a sink of a given orienta-
tion is determined by the difference in flow of interstitials and vacancies to those
sinks, which can be calculated using
Jm ¼ ðk2i Þm Di Ci ðk2v Þm Dv Cv
where m is the given type of sink and the subscripts i and v designate interstitials or
vacancies. The square of the average distance that a point defect migrates per
FIG. 13 Texture parameters of front end offcut material from H and HM-series pressure
tube populations.
FIG. 14 Dislocation density of front end offcut material from H- and HM-series pressure
tube populations measured by X-ray diffraction.
FIG. 15 TEM micrographs of front end offcut material for pressure tubes H0081M (HM-
series) and H1848 (H-series). The image is a cross section in the radial/
transverse plane. Inset on each image is a plot of the major and minor axes
dimensions (in microns) for the individual grains within that image.
second is given by D, and the average concentration in the matrix is C. The rate at
which point defects migrate is given by the product, DC. The probability that a
migrating point defect encounters a sink and therefore produces strain (positive for
interstitials and negative for vacancies) is determined by the sink strength, k2.
The sink strengths are relative and can be thought of as probabilities of encoun-
tering sinks of any one type when a point defect is migrating in a given direction.
The strength is determined by the density and orientation of sinks and also, in the
case of dislocations, the elastic interaction between the strain field around the sink
and the point defects. As one is dealing with balance equations, it is only necessary
to be able to compute the relative strengths of the different sinks and these can be
FIG. 16 Mean measured aspect ratio and relative diametral strain for the H- and HM-
series pressure tubes. The solid line is a linear trend with R2 ¼ 0.91.
incorporated into the model as bias factors that account for the probability that a
given sink has a propensity for absorbing interstitial, as opposed to vacancy, point
defects. This enables the building of simple models to explore the interplay between
various sinks and point defect properties. For example, one can represent the net
flux to sinks resulting in strain in the radial (R), transverse (T), and longitudinal (L)
directions of a pressure tube by the following expressions:
JR ¼ ½ð1 þ ð1 fR Þ pÞ Di Ci Dv Cv ðGBR þ qR Þ
JT ¼ ½ð1 þ fR p þ 2sÞ Di Ci Dv Cv ðGBT þ qT Þ
TABLE 8 Relative strain for gauged and imaged H- and HM-series pressure tubes.
Serial Number Relative Diame- Grain Aspect fR fT fL

tral Strain (%) Ratio
H0022M 33.4 0.393 0.301 0.653 0.046

H0059M 22.9 0.401 0.324 0.626 0.050
H0081M 35.2 0.461 0.305 0.636 0.060
H0179M 30.7 0.417 0.323 0.625 0.052
H0259M 6.7 0.351 0.280 0.671 0.049
H0788 –1.7 0.299 0.288 0.667 0.045
H0835 14.9 0.372 0.305 0.647 0.048
H0956 14.4 0.376 0.313 0.642 0.045
H1588 25.6 0.395 0.307 0.644 0.048
H1848 –17.2 0.273 0.283 0.665 0.051
H1852 0.8 0.335 0.305 0.651 0.043
JL ¼ ½ð1 þ p þ sÞ Di Ci Dv Cv ðGBL þ qL Þ
where
/
Di Ci ¼
ð1 þ ð1 fR Þ pÞ ðGBR þ qR Þ þ ð1 þ fR p þ 2sÞ ðGBT þ qT Þ
þ ð1 þ p þ sÞ ðGBL þ qL Þ
/
D v Cv ¼
ðGBR þ qR Þ þ ðGBT þ qT Þ þ ðGBL þ qL Þ
The grain boundary and dislocation sink densities corresponding with each direc-
tion R, T, and L are given by GBm and qm, where the orientation, m ¼ R, T, and L.
In this model, the interstitial bias factor [12], p, is assumed to be a function of sink
orientation and determined by interstitial diffusional anisotropy. The bias parame-
ter is therefore modified by the basal pole orientation parameter, fR, in order to cap-
ture the effect of the diffusional anisotropy difference of interstitial point defects
along the a and c-axes [13]. The bias factor due to stress, s, is applied to interstitial
diffusion based on the concept of the influence of stress on diffusion [14], and
developed as the “elastodiffusion” model for creep by Woo [15]. The sink densities,
GBm and qm, can be separated with appropriate changes of bias factors to account
for strain-field interactions if necessary.
FIG. 17 Illustrative calculation showing effect of varying relative radial and transverse
grain thickness at constant aspect ratio as a function of radial grain thickness on
diametral strain rate. The dislocation sink strength corresponding with a
dislocation density of 4 1014 m–2 and equivalent to a grain dimension of 0.5 lm
is shown as the vertical dashed line. The strain rate is calculated assuming a
displacement rate of 3 10 8 dpas–1 and a cascade efficiency of 3 %.
TABLE 9 Comparison of relative measured diametral strain and rate theory calculation with fixed fR
(0.305) and fixed dislocation density (4 1014 m–2).
Aspect Ratio Relative Strain (%) Rate Theory Rank
0.26 –17 Lowest

0.36 12 Middle
0.46 41 Highest
It is easy to see that : JR þ JT þ JL ¼ 0
One can then investigate how varying the grain boundary sizes and shapes affects
the diametral creep while holding the texture and dislocation density constant for
each grain.
Figure 17 shows the effect of varying the relative radial and transverse grain
dimensions as a function of the inverse sink density for three aspect ratio cases and
for fixed values of texture and dislocation density. The cases encompass those in
Table 8 and Fig. 16. The radial grain thickness of the H- and HM-series pressure
tubes ranges from 0.25 to 0.4 lm and the relative diametral strain can be read from
Fig. 17 for any combination of radial grain thickness and aspect ratio. The calculated
diametral strain can be compared to the observed diametral strain taken from the
trend line in Fig. 16. The calculated diametral strains from the rate theory expres-
sions rank exactly the same as the observed diametral strains (Table 9), showing
consistency between the rate theory formalism and the observations. For this exam-
ple, the model has been used to provide only qualitative insights on the effect of
grain aspect ratio, and no attempt has been made to fit the model parameters to
match the observed strain rates.
The rate theory formalism also sheds light on assumptions made [2,3] about the
diametral strain dependence on texture. Figure 11 includes the FS textured capsules
(high fR) and the high diametral strain suggests that texture was the explanatory vari-
able. The grain shape and size of these capsules was also quite different than typical;
mean radial width ¼ 0.8 lm, mean transverse width ¼ 1.1 lm, and an aspect ratio of
0.7 [3]. The rate theory formalism calculation shows (qualitatively) that grain size and
shape of the FS capsule would cause a higher diametral strain rate in comparison to the
Trillium-2&3 MPT capsules (mean radial width ¼ 0.4 lm, mean transverse
width ¼ 5 lm and an aspect ratio of 0.08 [3]). The high diametral strain of the FS tex-
tured capsules shown in Fig. 11 is, therefore, not explained by texture necessarily but
consistent with migration of interstitials and vacancies to grain boundary sinks.
Conclusions
Early results from the biaxially stressed creep capsule experiments performed in
OSIRIS have been supplemented with more recent data extending the available
range of temperature, crystallographic texture, and cold work data. The effects of
these variables on diametral and axial creep compliance have been analyzed with
the larger dataset and with in-service CANDU pressure tube data providing signifi-
cant insights into the dependence of irradiation creep on temperature and
microstructure:
1. There exists a strong temperature dependence on the diametral creep strain
rate and a weaker temperature dependence on the axial creep strain rate, i.e.,
irradiation creep anisotropy varies with temperature.
2. Although the creep capsules have a majority of the resolved fraction of basal
plane normals in the transverse/radial direction and only a small fraction in
the axial direction, texture appears to have a more significant effect on the
axial creep strain rate than on the diametral creep strain rate. In fact, micro-
structural parameters other than texture appear to have a stronger effect on dia-
metral creep strain rate. An analysis on in-service pressure tube data shows that
there is a strong diametral dependence on mean grain aspect ratio (radial grain
thickness/transverse grain width).
3. Based on the results from the Trillium-5 experiment, there is no statistically signif-
icant correlation between diametral creep strain rate and cold work in the range
of 6–26 %.
ACKNOWLEDGMENTS
Although numerous past and present members of the Deformation Technology
Branch at Chalk River Laboratories were involved with the OSIRIS experiments over
the past two decades, the authors would like to thank A. Buyers, W. Li, and A. I. Fluke
for their contributions to this paper and to S. Donohue for his guidance and insights
regarding in-reactor testing. The writers would also like to thank N. van den Brekel
from Ontario Power Generation for supplying data from CANDU reactor pressure
tubes and for useful discussions concerning the effect of manufacturing on
performance.
References
[1] Holt, R. A., “In-Reactor Deformation of Cold-Worked Zr-2.5Nb Pressure Tubes,” J. Nucl.
Mater., Vol. 372, Nos. 2–3, 2008, pp. 182–214.
[2] Causey, A. R., Holt, R. A., Christodoulou, N., and Ho, E. T. C., “Irradiation-Enhanced Defor-
mation of Zr-2.5Nb Tubes at High Neutron Fluences,” Proceedings of the Zirconium in
the Nuclear Industry, 12th International Symposium, ASTM STP 1354, G. P. Sabol and G. D.
Moan, Eds., ASTM International, West Conshohocken, PA, 2000, p. 74–85.
[3] Causey, A. R., Elder, J. E., Holt, R. A., and Fleck, R.G., “On the Anisotropy of In-
Reactor Creep of Zr-2.5Nb Tubes,” Proceedings of the Zirconium in the Nuclear
Industry, 10th International Symposium, ASTM STP 1245, A. M. Garde and E. R. Brad-
ley, Eds., ASTM International, Philadelphia, PA., 1994, pp. 202–220. Also AECL Report
AECL-10863.
[4] Christodoulou, N., Causey, A. R., Holt, R. A., Tome, C. N., Badie, N., Klassen, R. J., Sauve,
R., and Woo, C. H., “Modeling In-Reactor Deformation of Zr-2.5Nb Pressure Tubes in
CANDU Power Reactors,” Proceedings of the Zirconium in the Nuclear Industry, 11th
International Symposium, ASTM STP 1295, E. R. Bradley and G. P. Sabol, Eds., ASTM Inter-
national, Philadelphia, PA, 1996, pp. 518–537.
[5] Kearns, J. J., “Thermal Expansion and Preferred Orientation in Zircaloy,” Report No.
WAPD-TM-472, Bettis Atomic Power Laboratory, Pittsburgh PA, 1965.
[6] Griffiths, M., Sage, D., Holt, R. A., and Tome, C. N., “Determination of Dislocation Den-
sities in HCP Metals from X-ray Diffraction Line-Broadening Analysis,” Metall. Mater.
Trans. A, Vol. 33A, 2002, pp 859–865.
[7] Los Alamos National Laboratory, 2011, “ENDF/B-VI Nuclear Library,” http://t2.lanl.gov
[8] Bickel, G. A., Walters, L., and Griffiths, M., “Improved Zr-2.5Nb Pressure Tubing for Future
HW Reactors,” Proceedings of the International Conference on the Future of Heavy
Water Reactors (HWR-Future), Ottawa, Ontario, Canada, Oct 2–5, Canadian Nuclear So-
ciety, Toronto, Canada, 2011.
[9] Bickel, G. A., and Griffiths, M., “Manufacturing Variability, Microstructure and Deforma-
tion of Zr-2.5Nb Pressure Tubes,” J. ASTM Int., Vol. 4, No. 10, 2007, JAI 101126.
[10] Bickel, G. A., Griffiths, M., Douchant, A., Douglas, S., Woo, O. T., and Buyers, A.,
“Improved Zr-2.5Nb Pressure Tubes for Reduced Diametral Strain in Advanced CANDU
Reactors,” J. ASTM Int., Vol. 8, No. 2, 2011, JAI 103521.
[11] Heald, P. T., and Speight, M. V., “Point Defect Behaviour in Irradiated Materials,” Acta.
Metall., Vol. 23, 1975, pp. 1389–1399.
[12] Olander, D. R., “Fundamental Aspects of Nuclear Reactor Fuel Elements,” Report No.
TID-26711-P1, National Technical Information Service, U.S. Dept. of Commerce, Spring-
field, VA, April 1976.
[13] Woo, C. H., and Gosele, U., “Dislocation Bias in an Anisotropic Diffusive Medium and Irra-
diation Growth,” J. Nucl. Mater., Vol. 119, 1983, pp. 219–228.
[14] Dederichs, P. H. and Schroeder, K., “Anisotropic Diffusion in Stress Fields,” Phys. Rev. B,
Vol. 17, 1978, pp. 2524–2536.
[15] Woo, C. H., “Irradiation Creep due to Elastodiffusion,” J. Nucl. Mater. Vol. 120, No. 1, 1984,
pp. 55–64.
DISCUSSION
Questions from B. K. Shah, BARC Mumba:—Pressure tubes are used in cold-
worked condition. Since higher cold work leads to higher creep, have you consid-
ered reducing the cold work while still maintaining the strength by optimization of
chemical composition?
Authors’ Response:—The Trillium-5 experiment was designed to evaluate the

in-reactor deformation rate as a function of the amount of cold work. All variables
in the experiment were constant except the amount of cold work which ranged
from 6 to 26 %. The experiment showed that for this range of cold work the Zr-
2.5Nb creep capsules showed no statistically significant correlation between diame-
tral creep strain rate and cold work. Therefore, having a lower degree of cold work
or dislocation density is not considered effective as a means of lowering diametral
creep.
Questions from Jean-Christophe Brachet, CEA-Sailey Nuclear Materials Dept.

France
Q1:—Is there any correlation between crystallographic and morphological tex-

tures? How can you deconvolute these two effects?
Authors’ Response:—Yes, the crystallographic texture and the grain shapes are
reasonably well matched; the platelet grains tend to have their c-axes perpendicular
to the longitudinal direction of the tube and in the plane of the grain. When model-
ing, the two components are inter-linked; the crystallography dictating the intrinsic
diffusion of interstitial point defects in the material and the grain boundaries dictat-
ing the sink distribution. The mass transport is then a function of the tendency for
point defects to be moving in a particular direction (dictated by the crystallography)
and the tendency for the point defects to annihilate at sinks (dictated by the mor-
phology). In the case of a statistical model the synergy between the two effects can
be isolated by assuming simple linear cross-terms.
Q2:—Are there any differences between the spatial distribution and/or volume
fraction of bZr (metastable) phases?
Authors’ Response:—Such differences are not apparent in regular pressure tub-

ing. If the tubes were heated to decompose the beta-phase then spatial inhomogene-
ities do occur where smaller local volume fractions correspond with higher Nb
concentrations as one would expect.
Questions from Srikumar Banerjee, BARC Mumbai:—Diametral creep of Zr-2.5

Nb is of safety concern at high power operation. Since several factors such as
crystallographic texture, morphological texture, cold work, grain size, and several
phenomena such as reconstitution of b phase, relocation of b phase, irradiation
enhanced (dislocation and/or diffusional) creep are involved, can you summarize
the influence of these factors in controlling diametral creep?
Authors’ Response:—In this paper the effects of crystallographic texture,

grain size and cold work on diametral creep strain rate were examined. It was
found that mean grain aspect ratio (radial grain thickness/transverse grain
width) rather than texture appears to have a stronger effect on diametral creep
strain rate. The Trillium-5 experiment showed that there is no statistically sig-
nificant correlation between diametral creep strain rate and cold work in the
range of 6 to 26 %. In previous studies the state of the beta-phase was assessed
as part of a broad statistical study. No dependence on the beta-phase state was
observed.
Questions from N. Saibaba, Nuclear Fuel Complex, Hyderabad, India
Q1:—What is the effect of extrusion ratio on the variability of properties from

leading end to trailing end?
Authors’ Response:—The effect of extrusion ratio was not investigated in this

study.
Q2:—It is understood from literature that Fe has influence on axial creep and is
controlled well when used in the range of 1200 ppm to 1400 ppm. Any comments
on this?
Authors’ Response:—The effect of Fe was not investigated in this work. How-

ever, although Fe may affect the microstructure, and therefore the creep properties
of the pressure tubing, the effect of Fe as a causal variable is often confounded by
other variables.
Question from Javier Romero, Westinghouse Electric Co. :—Do you have any
data with compressive hoop stress?
Authors’ Response:—No, this paper focused on the results of internally pressur-

ized creep capsule experiments.
Q1:—How do data on creep from the small capsule samples compare with the
data on the actual pressure tubes?
Authors’ Response:—In terms of microstructural effects the small capsules

appear to behave in a similar manner with full-size pressure tubes in terms of tex-
ture dependence. However, variation in grain structure was not investigated for the
creep capsules and variation in dislocation density was not investigated for the full
size pressure tubes. In addition diametral strain rate is non-linear with fast neutron
flux. The OSIRIS experiments were performed at fluxes which are an order of mag-
nitude greater than CANDU power reactor fluxes and the creep rates cannot be eas-
ily normalized for flux because of the non-linearity.
Q2:—Was the effect of texture on diametric creep from the small capsule sam-
ples or on the actual pressure tube.
Authors’ Response:—The effect of texture on diametral creep were shown for

both creep capsules and for in-service CANDU pressure tubes. From the creep cap-
sule data, texture was shown to have a more significant effect on the axial creep
strain rate than on the diametral creep strain rate. An analysis on in-service pres-
sure tube data was used to demonstrate that the grain aspect ratio (radial grain
thickness/transverse grain width) rather than texture appears to have a stronger
effect on diametral creep strain rate.
IRRADIATION
AND HYDROGEN EFFECTS
Stanislav I. Golubov,1 Alexander V. Barashev,1,2

Roger E. Stoller,1 and Bachu N. Singh3
Breakthrough in Understanding
Radiation Growth of Zirconium
Reference
Golubov, Stanislav I., Barashev, Alexander V., Stoller, Roger E., and Singh, Bachu N.,
“Breakthrough in Understanding Radiation Growth of Zirconium,” Zirconium in the Nuclear
Industry: 17th International Symposium, STP 1543, Robert Comstock and Pierre Barberis, Eds., pp.
729–758, doi:10.1520/STP154320130043, ASTM International, West Conshohocken, PA 2015.4
ABSTRACT
The efforts of many scientists for more than a half of a century have resulted in a
substantial understanding of the response of Zr-based materials to irradiation.
However, the models of radiation growth proposed to date have not played a
decisive role in creating radiation-resistant materials and cannot predict strain
rates at high irradiation doses. The main reason for this is the common
assumption that, regardless of the incident particle mass and energy, the
primary damage consists of single vacancies and self-interstitial atoms (SIAs),
both diffusing three-dimensionally. Thus, the models ignore the distinguishing
features of the damage production in displacement cascades during fast-
particle, e.g., neutron, irradiation; namely, the intra-cascade clustering of
vacancies and SIAs and one-dimensional diffusion of SIA clusters. Over the last
twenty years or so, the production bias model (PBM) has been developed, which
accounts for these features and explains many observations in cubic crystals.
The cascades in hcp crystals are found to be similar to those in cubic crystals;
hence one can expect that the PBM will provide a realistic framework for the hcp
Manuscript received March 14, 2013; accepted for publication June 6, 2014; published online September 22,
2014.
1
Materials Science and Technology Division, ORNL, Oak Ridge, TN 37831-6138 (Corresponding author), e-mail:
golubovsi@ornl.gov
2
Center for Materials Processing, Department of Materials Science and Engineering, Univ. of Tennessee, East
Stadium Hall, Knoxville, TN 37996-0750.
3
Materials Research Department, Risø National Laboratory, Technical Univ. of Denmark, DK-4000 Roskilde,
Denmark.
4
Hyderabad, India.
Copyright V
metals as well. It is shown in this paper that it reproduces all the growth stages
observed in annealed materials under neutron irradiation, such as the high strain
rate at low, strain saturation at intermediate, and breakaway growth at relatively
high doses. It accounts for the striking observations of negative strains in
prismatic directions and co-existence of vacancy- and interstitial-type prismatic
loops, which have never been explained before. It reveals the role of cold work in
the radiation growth behavior and the reasons for the alignment of basal
vacancy-type loops along the basal planes. The critical parameters determining
the high-dose behavior are revealed and the maximum growth rate is estimated.
Keywords
neutron irradiation, zirconium, radiation growth, theory
Introduction
The radiation growth (RG) of Zr-based materials is one of the main concerns
regarding the safe operation of thermal nuclear reactors. Experiments have demon-
strated that deformation of these alloys at temperatures below 300 C is driven by
the evolution of dislocation structure, which includes nucleation and growth of dis-
location loops on both the prismatic and basal planes. The growth strain in the c
direction is always negative, and the basal-plane loops are always of the vacancy
type. The strains in prismatic directions are positive in the majority of cases, but
may also be negative. Another striking observation is that the prismatic loops of
both vacancy and interstitial type may be formed at the same time.
It is commonly accepted that RG occurs due to the asymmetry of the capture effi-
ciencies of a and c dislocations and dislocation loops for single vacancy and interstitial
atoms. However, the conventional concept of dislocation bias suggests that, in this case,
the strains must have opposite signs to those generally observed, i.e., positive/expansion
in c and negative/contraction in a directions. This is because the Burgers vector of c dis-
locations is larger than that of a dislocations, which creates larger bias of c dislocations
for self-interstitial atoms (SIAs). Several models have been proposed to resolve this con-
tradiction since the first model by Buckley [1] 50 years ago (see, e.g., Ref. [2] for a
review), all based on the dislocation bias approach, but none has resolved the issue.
A qualitatively new step in understanding the RG phenomenon was made by
Woo and Gösele [3,4] by introducing anisotropic diffusion of SIAs on the hcp lat-
tice. In the diffusion anisotropy difference (DAD) model [4], it was suggested that
the vacancy diffusion is isotropic, whereas the SIAs migrate preferentially along the
basal planes. This allowed explaining the contraction of c axes and the crucial role
of c loops in developing breakaway stage in annealed Zr crystals.
Nevertheless, the DAD model does not describe correctly the RG in neutron-
irradiated materials. This is because it assumes that the primary damage consists of
point defects, i.e., single vacancies and SIAs, only. Experiments [5] and molecular
dynamics (MD) simulations [6] have shown that under neutron irradiation, a large
20–50 % fraction of the defects form clusters. The SIA clusters migrate one-
GOLUBOV ET AL., DOI 10.1520/STP154320130043 731
dimensionally (1D), which results in the mixture of the second (for the point
defects) and third (for the SIA clusters) order reaction kinetics in neutron-
irradiated solids, rather than just second order, as in the DAD model. Holt et al. [7]
generalized the DAD model by accounting for the cascade production of SIA clus-
ters, but assumed the clusters to be immobile, which was wrong. In fact, the MD
simulations show that in all crystals including Zr [6] the SIA clusters are highly
mobile, diffusing 1D along close-packed directions. The authors of Ref. [7] made
another assumption that “…the higher reaction cross section with the primary clusters
of the a-dislocations than of c-component dislocations, due to the higher mobility, by
glide, of the former,” which is unphysical, since there is no reason for dislocation glide
without external stress. In the calculations presented in Ref. [7], the DAD bias factor
was taken to be equal to 200 %, which requires high anisotropy of single SIA diffusion,
Da =Dc 102 , which is not supported by ab initio calculations [8,9].
Another recent paper devoted to RG in Zr was published by Christien and
Barbu [10]; however, it did not provide any new insight into RG. The description of
the breakaway stage in Ref. [10] is the same as in the DAD model. The only innova-
tion is the assumption that the RG strain at low doses is due to accumulation of sin-
gle vacancies. The vacancy concentration required to reproduce observations is
104, i.e., close to the thermal-equilibrium value at the melting temperature,
which is unrealistic. In addition, the mutual recombination of point defects at such
a vacancy concentration would suppress the damage accumulation, so that the
breakaway stage would never take place. Moreover, single vacancies have to con-
tribute to strains in c as well as a directions, while the assumption “that vacancy
relaxation is anisotropic and is fully oriented along the c-axis” with reference to
papers published in nineteen eighties is not supported.
The negative a strain and coexistence of vacancy- and interstitial-type pris-
matic loops are the most intriguing parts of the RG phenomenon. Since the c
strain is always negative, the negative a strain violates the basic property of the
growth phenomenon: the volume conservation. The coexistence of the loops
violates the well-known loop property: vacancy- and interstitial-type loops of
large enough size have almost the same efficiencies for absorption of point
defects; hence, cannot grow at the same time. This is the reason why their
coexistence is never observed in cubic crystals. One may conclude that the neg-
ative a strains and coexistence of vacancy and interstitial a loops are funda-
mental features, specific to hcp crystals.
There were several publications devoted to explaining the coexistence of va-
cancy and interstitial prismatic loops (see, e.g., Refs. [11–13]); however, all were
unsuccessful. The main reason was that the models considered Frenkel pairs only,
ignoring the true nature of the primary damage in cascades. In addition, the models
[11–13] assumed that both vacancies and SIAs execute 3D random walk, which
ignores the DAD. Note also that the assumption in Ref. [13] that the vacancy dilata-
tion volume is larger than that of SIAs needed to explain the coexistence is not sup-
ported by ab initio calculations [14].
The current status of the theory may be summarized by the following citations
from two recent reviews: “… reliable mechanistic models to predict the deformation
of even a pure Zr single crystal are not known … We therefore still rely on a phe-
nomenological approach” [15], and “… understanding of the basic creep mecha-
nisms in anisotropic materials like zirconium alloys is still not strong enough to be
truly predictive… Today, most models are empirical in nature …” [16].
The situation described is similar to that of void swelling in the bcc- and fcc
metals 20 years ago, the time when the research directions of RG in hcp crystals
and void swelling in cubic crystals deviated from each other. Since then, the theory
of void swelling has made significant progress in accounting for observations and
obtaining consistency between experiments and modeling results. The main suc-
cesses of a new model, the Production Bias Model (PBM), came from including the
cascade-production and 1D migration of the SIA clusters. The PBM explains many
striking observations, e.g., the recoil–energy effect, the grain boundary and grain
size effects in void swelling, and the void lattice formation, which have been
reviewed by Singh et al. [17] already more than a decade ago. A recent review can
be found in Ref. [18].
The displacement cascades and properties of the SIA clusters in hcp Zr are
found to be similar to those in cubic materials [19–21]. In addition, the alignment
of vacancy loops [22] and voids [23,24] observed along the basal planes in irradi-
ated hcp crystals is qualitatively similar to void ordering in cubic metals. Hence, the
PBM may also provide a realistic framework for damage accumulation in the hcp
metals. The aim of the present work is to develop such a model for the RG in Zr.
The paper is organized as follows. In the second section, the problem is
described. In the third section, the model assumptions are listed and the rate equa-
tions are formulated. In the fourth section, the model predictions are described.
Estimates of the maximum strain rate are made in the fifth section. Applications of
the model for calculations of dose dependence of RG strain are presented in the
sixth section. A summary is given in the final section.
Problem Characterization
In annealed Zr crystals, the RG is characterized by expansion along a axes and con-
traction along c axis. The typical strain behavior consists of three distinct stages [2].
Stage I exhibits a high strain rate and lasts for 0.1–1.0 dpa (displacements per
atom, NRT standard [25]). Stage II demonstrates a very low strain rate, often inter-
preted as strain saturation, and proceeds up to 3 dpa. At higher doses, usually
referred to as the breakaway growth stage (III), the strain rates increase with
increasing dose and reach values as high as 103dpa1. The dose dependence of
these rates is debated. Transmission electron microscopy (TEM) examination of
irradiated samples revealed formation of interstitial-type prismatic loops with the
ð1=3Þh1120i Burgers vectors during stages I and II and vacancy-type c loops during
stage III.
In cold-worked materials, the strain rates are relatively high from the very be-
ginning and no strain saturation occurs (see Fig. 6 in Ref. [26]). In some cases, both
a and c strains have been found to be negative. Moreover, vacancy- and interstitial-
type prismatic loops of similar densities and sizes may coexist. To our knowledge,
these two observations: the negative a strains and coexistence of the vacancy- and
interstitial-type prismatic loops have never been explained. More generally, no
theory has been published which self-consistently explains all the observations.
The model presented here gives a self-consistent explanation of the RG phe-
nomenon in Zr single crystals and provides a framework capable of describing all
the observations quantitatively.
New Model
BASIC FRAMEWORK
The simple yet realistic model of a single zirconium crystal considered here takes
into account straight edge dislocations and dislocation loops. Similar to the DAD
model [4], we consider two types of dislocation defects: those lying on the basal
plane that have a c component to their Burgers vector and those lying on the pris-
matic planes with the a Burgers vector. For the sake of simplicity, in the following
discussion they are called c and a dislocations/loops, respectively.
The framework of the model, a preliminary version of which was formulated
by Golubov et al. in [27], is as follows:
• The primary radiation damage consists of point defects and SIA clusters.
• Single vacancies and SIAs migrate 3D.
• The SIA clusters migrate 1D along h11 20i close-packed directions.
• Interactions of SIA clusters with both a dislocations that have Burgers vectors
non-parallel to that of the clusters and c dislocations are neglected.
• The difference in absorption properties of dislocation loops and edge disloca-
tions for mobile point defects and SIA clusters is neglected.
• The dislocation bias due to interaction of point defects with dislocations/loops
and possible anisotropy of single point defects migration is neglected.
Note that the population of edge dislocations and dislocation loops will be
characterized by a single parameter for each Burgers vector qi ði ¼ a1 ; a2 ; a3 ; cÞ, the
total length per unit volume, since the difference in the absorption properties of
these two defect types is neglected. We assume that, in general, these densities may
be different from each other.
The above framework is essentially the PBM [18] adjusted to the hexagonal
symmetry of the crystal lattice. The only new assumption is the neglect of interac-
tions between SIA clusters with c dislocations and a dislocations that have Burgers
vectors non-parallel to that of the cluster. This originates from the dislocation na-
ture of the interactions, which is qualitatively different from those involving point
defects. The interactions depend on mutual orientation of the cluster and disloca-
tion Burgers vectors, for which the interaction energies are shown in Figs. 1–3 (see
the Appendix for calculations). For basal and prismatic dislocations with non-
parallel Burgers vectors, the interactions are significantly weaker than for a disloca-
tions with parallel Burgers vectors. The corresponding trapping zones, associated
with the cross-sections of absorption reactions, are significantly smaller. Note that
the assumption in question does not affect the results for isotropic distribution of
prismatic dislocation Burgers vectors. This is due to symmetry of the cluster pro-
duction and partitioning in this case: 1/3 part of the SIA clusters are absorbed by a
dislocations of each particular Burgers vector for any interaction scenario. It does
affect the results for non-isotropic distribution of prismatic dislocation Burgers vec-
tors, but the effect must be small for the above-mentioned reason. Accounting for
the ignored interactions is straightforward but would lead to a loss of clarity due to
a more complicated diffusion-reaction scenario. Finally, note that screw disloca-
tions are not considered in the present model.
MAIN EQUATIONS
The equations for concentrations of mobile defects, single vacancies (subscript v),
single SIAs (i), and SIA clusters (cl) in the framework of the model are as follows:
FIG. 1 Interaction energy between a dislocation with the Burgers vector parallel to that
of a 10-SIA cluster in Zr at 573 K. The dislocation line is at the coordinate origin
and perpendicular to the (x, y) plane. x is the distance from dislocation extra
plane, along the dislocation Burgers vector (shown in the figure as ba). y is the
distance from the dislocation line, along the direction perpendicular to the
dislocation Burgers vector. The cluster Burgers vector is shown in the figure
as bcl.
FIG. 2 Same as in Fig. 1 but for a dislocation with the Burgers vector at p/3 angle to that
of the cluster.
FIG. 3 Interaction energy between a c dislocation with the Burgers vector

perpendicular to that of a 10-SIA cluster in Zr at 573 K. The dislocation line is at
the coordinate origin and perpendicular to the (z, y) plane. z is the distance from
dislocation extra plane, along the dislocation Burgers vector (shown in the figure
as bc). y is the distance from the dislocation line, along the direction
perpendicular to the dislocation Burgers vector. The cluster Burgers vector is
shown in the figure as bcl.
dCv X
(1) ¼ GNRT ð1 er Þ Dv Cv qj ; ðj ¼ a1 ; a2 ; a3 ; cÞ
dt j
dCi g
X
(2) ¼ GNRT ð1 er Þð1 ei Þ Di Ci qj ; ðj ¼ a1 ; a2 ; a3 ; cÞ
dt j
g
dCclm ð1 er Þei
(3) ¼ GNRT Dcl Cclm k2m ; ðm ¼ a1 ; a2 ; a3 Þ;
dt 3n
where:
GNRT ¼ the NRT standard value for the defect production rate,
er ¼ the fraction of defects recombining during the cooling-down phase of a
cascade,
g
ei and n ¼ the fraction of SIAs survived intra-cascade recombination in the
form of clusters and the mean number of SIAs in a cluster, respectively,
Dv;i and Dcl ¼ the diffusion coefficients of point defects and SIA clusters,
respectively,
qj ¼ the densities of prismatic dislocations with the Burgers vectors along
a1 ; a2 ; a3, and basal dislocations with the Burgers vector along the c direction, and
k2m ¼ the sink strength for the SIA clusters migrating along m direction.
The factor 1/3 on the right-hand side (RHS) of Eq 3 accounts for the equality
of SIA cluster production rates in a1 , a2 , and a3 directions. The first terms on the
RHSs of Eqs 1–3 stand for the production of defects, while the second terms for
their loss at dislocations. The sink strength k2m in Eq 3 is given by
p2 r02 q2m
(4) k2m ¼ ; ðm ¼ a1 ; a2 ; a3 Þ
2
where r0 is the cluster capture radius of prismatic dislocations with the Burgers vec-
tors parallel to that of SIA clusters.
Note that the 3D migrating point defects are described by the second-order
reaction kinetics, so that the sink strengths in Eqs. 1 and 2 are proportional to the
P
total dislocation density, q ¼ j qj . In contrast, the 1D migrating SIA clusters are
described by third-order reaction kinetics, where the sink strength, k2m , is propor-
tional to the square of dislocation density [18]. In addition, the sink strength for
SIA clusters with a given Burgers vector direction, either a1 , a2 , or a3 , depends on
the density of dislocations of the same Burgers vector only. As a result, if the density
of a1 dislocations is larger than that of a2 and a3 , the absorption rate of a1 disloca-
tions for point defects will be larger. In contrast, the absorption rates of a1 , a2 , and
a3 dislocations for SIA clusters remain the same for any distribution of prismatic
dislocations, namely 1/3 of the clusters generated is captured by each type of pris-
matic dislocations. This is the key difference between 3D (and preferentially 2D)
diffusing point defects and 1D diffusing SIA clusters. All the predictions of the
model described below follow from this difference.
The steady-state defect fluxes are found by equating the time derivatives in
Eqs. 1–3 to zero:
GNRT ð1 er Þ
(5) D v Cv ¼
q
g
GNRT ð1 er Þ 1 ei
(6) D i Ci ¼
q
g
2 GNRT ð1 er Þei
(7) Dcl Cclm ¼ ; m ¼ a1 ; a2 ; a3
3n p2 r02 q2m
Note that the cluster flux to c dislocations is equal to zero.

The dislocation climb velocities, Vj, are defined through the net fluxes of vacan-
cies, single SIAs and SIA clusters to dislocations as
8 j 2
>
> n Dcl Ccl kj Dv Cv Di Ci
>
> ; j ¼ a1 ; a2 ; a3
< bj qj bj
(8) Vj ¼
>
> 1
>
>
: ðDv Cv Di Ci Þ; j¼c
bj
where bj is the Burgers vector of j-type dislocations. The strain rate in a particular
prismatic direction a, dea =d/, due to the climb of prismatic dislocations, is calcu-
lated by summing contributions from dislocations with different Burgers vectors
ðm ¼ a1 ; a2 ; a3 Þ:
dea X
¼ qm Vm bm cos2 um
dt m
(9) X
¼ nDcl Cclm k2m qm ðDv Cv Di Ci Þ cos2 um
m
where um is the angle between the vectors a and bm . The strain rate in c direction is
given by
dec
(10) ¼ qc Vc bc ¼ qc ðDv Cv Di Ci Þ
dt
By substituting Eqs 5–7 into Eqs. 9 and 10, one finally obtains
dea X 1 qm
(11) ¼v cos2 um
d/ m
3 q
dec q
(12) ¼ v c
d/ q
where / ¼ GNRT t is the irradiation dose and

g
(13) v ¼ ð1 er Þei
is the fraction of SIAs at the end of the cooling-down phase of cascades in the clus-
tered form. Note that v ¼ 0 for non-cascade conditions, e.g., for irradiation with 1
MeV electrons, since the model neglects dislocation bias for point defects.
In a Cartesian coordinate system where the x axis is along a1 , y along a2 a3 ,
and z along c, Eqs 11 and 12 take the following form:

dex 1 q
(14) ¼v x
d/ 2 q

dey 1 qy
(15) ¼v
d/ 2 q
dez q
(16) ¼ v z
d/ q
where

(17) qx ¼ qa1 þ qa2 þ qa3 cos2 ðp=3Þ

(18) qy ¼ qa2 þ qa3 cos2 ðp=6Þ
(19) qz ¼ qc
(20) q ¼ qx þ qy þ qz qa1 þ qa2 þ qa3 þ qc
Note that Eqs 14–16 can be presented by a diagonal matrix equation as

dei;j 1 q
(21) ¼v 1 di;3 i di;j ; ði; j ¼ 1; 2; 3Þ
d/ 2 q
where the indexes 1, 2, and 3 stand for x, y, and z, respectively, and di;j is the Kro-
necker delta. The equations above satisfy the volume conservation
dex dey dez

(22) þ þ ¼0
d/ d/ d/
as it has to be in the absence of swelling.

Equations 14–16 describe the RG rates as a function of the effective densities of
prismatic and basal edge dislocations (including loops), qx ; qy ; qz , and the parame-
ter v, which is the fraction of SIAs produced in cascades in the form of 1-D migrat-
ing SIA clusters.
Model Predictions
Equations 14–16 demonstrate that the strain rates are determined by the fractions
of a and c dislocation densities, qx =q, qy =q and qz =q, rather than by their absolute
values. This explains similar strain rates observed at low irradiation doses in
annealed and cold-worked materials. In accordance with Eq 16, the strain rate in c
direction is always negative. In contrast, the strain rates in a directions may be
either positive or negative depending on the distribution of the prismatic disloca-

tion Burgers vectors. The strain rates in prismatic directions are positive for iso-
tropic distribution of prismatic dislocations, i.e., when qx ¼ qy . In contrast, the
strain rate in the x direction is negative when qx =q > 1=2 (see Eq 14), i.e., when
the distribution of a dislocation Burgers vectors is non-isotropic. The inequality is
only valid for qx =qy > 1 at non-zero c dislocation density. The a strain rates as a
function of qx =qy are shown in Fig. 4 for the case qz =qx ¼ 1=5. The prismatic strain
rates, dex =d/; dey =d/, are positive and equal to each other for an isotropic distribu-
tion when qx =qy ¼ 1, whereas the x strain rate is equal to zero at qx =qy ¼ 1:25 and
it is negative and of similar magnitude to the c direction for qx =qy 2. Thus, the a
strain rates are sensitive even to small deviations from isotropic distribution of a
dislocation Burgers vectors. The analysis suggests that the conventional description
of RG in Zr as an expansion along a directions and contraction along c direction
should be changed to an expansion along at least one of the a directions and con-
traction along c direction. Note also that it follows from Fig. 4 that the absolute val-
ues of a strain rates increase strongly with increasing qx =qy ratio, while the c strain
depends only weakly on it. More details on the role of non-uniform distribution of
prismatic dislocations on RG are given in the Negative a Strains subsection below.
The upper bounds of the strain rates are fully determined by the parameter v,
the fraction of SIAs produced in cascades in the clustered form, as described further
in the Estimates of Strain Rates section.
FIG. 4 Effect of non-uniform distribution of a dislocation Burgers vectors on strain

rates for the case when density of c dislocations is five times smaller than a
dislocations. Note that x strain becomes negative at qx =qy > 1:25 and reaches
the strain rate of c dislocations at qx =qy ¼ 2:5.
STAGE I: INITIAL GROWTH IN PRE-ANNEALED MATERIALS

If the initial densities of a dislocations are low, on the order of 1012–1013 m2, and
the distribution of their Burges vectors is isotropic: q0x q0y , then q 2q0x
þ q0z 2q0x , since the density of c dislocation is normally 5–10 times smaller than
that of prismatic dislocations. In this case, it follows from Eqs 14–16 that the initial
strain rates are

dex dey 1 q0 v q0
(23) ¼ v 0 x 0 ¼ z0
d/ d/ 2 2qx þ qz 4 qx
dez vq
(24) ¼ 0z
d/ 2 qx
For q0z =q0x 0:2, the strain rate in prismatic direction is equal to v=20. To make
g
numerical estimates of the parameter v for Zr, one needs values for er and ei , which
are not available due to the lack of systematic studies of cascades in Zr. Since the cas-
cade damage is not drastically sensitive to the type of the lattice, in the following we use
the data for neutron-irradiated fcc copper derived within the same framework: er ¼ 0.9
g
and ei ¼ 0.2 [28], for which v ¼ 2 102. With this value, the strain rates in a direc-
tions are equal to 103 dpa1, which is in a good agreement with observations. Indeed,
at this strain rate, the saturation strain, which is found to be of the order 104
(Fig. 3(b) in Ref. [29]), is reached at a dose of 0.1 dpa, which is close to experiments.
STAGE II: STRAIN SATURATION IN PRE-ANNEALED MATERIALS

To understand the reasons for the strain saturation, one needs to take into account
that the total density of a dislocations is increased with increasing irradiation dose,
due to nucleation and growth of a-dislocation loops:
(25) qx;y ¼ q0x;y þ 2pRx;y Nx;y
where Rx;y and Nx;y are the loop radius and number density. Assuming the distribu-
tion of a-dislocation loops to be isotropic: Rx Ry ; Nx ¼ Ny , one can find from
Eqs 23 and 24 that the strain rates decrease strongly with the development of loop
population. When 2pRx;y Nx;y q0x;y ,
dex dey v q0z

(26) ¼
d/ d/ 8p Rx Nx
dez v q0z
(27)
d/ 4p Rx Nx
For values typically observed at the saturation stage: Rx;y ¼ 5 nm and

Nx;y ¼ 1022 m3 , the sink strength of dislocation loops is equal to 3 1014 m2.
Thus, for q0z ¼1012 m2, the strain rates given by Eqs 26 and 27 drop down by
about 300 times, as compared to those in the beginning of irradiation. With such
low strain rates, this stage can be considered as strain saturation, which continues
up to 3 dpa, until the c loops start nucleating. In other words, the saturation stage
corresponds to a very small but nonzero strain rate.
STAGE III: BREAKAWAY GROWTH

The nucleation and growth of c loops leads to an increase of the total sink strength
of c dislocations:
(28) qz ¼ q0z þ 2pRz Nz
hence to an increase of the qz =qx ratio and the strain rates. In the case when
2pRz Nz q0z , the strain rates start to increase as
dex dey v q0z þ 2pRz Nz v Rz Nz

(29) ¼
d/ d/ 8p Rx Nx 4 Rx Nx
dez v q0z þ 2pRz Nz v Rz Nz
(30) ¼
d/ 4p Rx Nx 2 Rx Nx
Thus, the breakaway strain rates are determined by the ratio Rz Nz =Rx Nx , which
increases with increasing density and size of c loops in agreement with observations
(see, e.g., Fig. 4 in Ref. [29]).
THE EFFECT OF COLD WORK

The initial dislocation density in cold-worked materials is high so that the nuclea-
tion and growth of a loops may not affect the strain rates as strongly as in annealed
materials. In the case when q0x;y are significantly smaller than 1014 m2, the satu-
ration stage may still be identified but less pronounced because an increase of the
total sink strength of a dislocations due to nucleation and growth of a loops is mod-
erate. At higher dislocation densities, q0x;y > 1014 m2 , an increase of the total sink
strength of a dislocations due to a loops becomes so small that saturation of strain
does not take place; the initial high strain rate will be maintained, in agreement
with observations [29].
NEGATIVE a STRAINS AND COEXISTENCE

OF VACANCY- AND INTERSTITIAL-TYPE a LOOPS
The models proposed previously were all based on only Frenkel-pair production
and assumed isotropic distributions of a-dislocation Burgers vectors. In such mod-
els, the driving force for radiation growth is the difference in the efficiencies of va-
cancy and SIA absorption by a and c dislocations. The net defect fluxes to a and c
dislocations depend on their total densities and are independent of the distribution
of a-dislocation (and a-loop) Burgers vectors. Thus, if a Frenkel-pair-based model
were generalized to account for anisotropy of the distribution, it would give differ-
ent but nevertheless positive values of a strains in different prismatic directions. If
an a strain were negative in such a model, it would be negative for all a directions;
then negative c strain would violate the volume conservation. In our view, the basic
assumption in these models, that Frenkel pairs are produced by irradiation,
prevented understanding the origin of the negative a strain phenomenon. For the
same reason, the coexistence of vacancy- and interstitial-type prismatic loops can-
not be explained by such models. For an isotropic distribution of prismatic disloca-
tions, our model also predicts positive strains in all a directions. However, as we
show below, the model predicts that any anisotropy in the dislocations can provide
an explanation for the observed negative a strains and coexistence of vacancy- and
interstitial-type a loops. Although it does not appear that past experimental investi-
gations have looked for anisotropy, we hope our results will motivate further work
in this direction.
To explain the mechanism, let us consider a limiting case. Assume the density
of a dislocations with the Burgers vectors along one of the prismatic directions, say
x, is much larger than the others: qx qy , and the density of c dislocations is
much smaller than a dislocations, qz qx ; qy . In this case, the partitioning of va-
cancy- and interstitial-type defects is as follows. The SIA clusters are absorbed
equally by both x and y dislocations; whereas the majority of point defects by x dis-
locations since qx qy . Because of significant clustering of SIAs, the production
rate of single vacancies is higher than single SIAs. As a result, the point defects pro-
duce an excess vacancy flux to all dislocations, with the vacancy flux to x disloca-
tions larger than to y dislocations. Due to the equality of SIA cluster absorption by
x and y dislocations, the net vacancy flux is positive to x and negative to y disloca-
tions, resulting in positive and negative strains in y and x directions, respectively.
The coexistence of vacancy- and interstitial-type a loops occurs for the same
reason as the negative a strain. This is because the defect absorption properties of loops
are similar to those of dislocations. As stated in Eqs 25 and 28, the loop nucleation and
growth just increase the effective dislocation density. In the example considered above,
the net vacancy flux to x dislocations causes nucleation and growth of x vacancy loops,
while the net SIA flux to y dislocations causes nucleation and growth of y interstitial
loops. Thus, the vacancy-type x and interstitial-type y loops coexist because they have
non-parallel Burgers vectors. Their contributions to strains are not additive for the
same reason. This explains why the contraction in the c direction takes place even
when the size and density of vacancy loops are similar to or even larger than those of
interstitial loops. The negative strains in one of the a directions and c direction are
compensated by a corresponding positive strain in another a direction.
Due to hexagonal symmetry, different orientations of the vacancy and interstitial
loops should not be obvious from an ordinary examination of the microstructure
since TEM observation of an arbitrarily-orientated sample would show both vacancy-
and interstitial-type loops at the same time. These may be revealed for special orienta-
tions of the sample. For example if vacancy loop Burgers vectors are all parallel to the
a1 direction, they will be invisible for orientations perpendicular to a1.
It should be emphasized that the explanation of the negative a strain proposed
is based on a non-isotropic distribution of edge a dislocations. Since the significance
of this has never been emphasized before, it does not appear to have been the sub-
ject of experimental measurement. For example, in the paper by Zee et al. [26],
where the negative a strain was observed in a single Zr crystal pre-strained along an
axis close to one of a directions, only the total density of dislocations was reported.
Moreover, the fractions of edge and screw dislocations are unknown. So, a compari-
son of the model with the experiment is not possible.
The dislocation structure in deformed Zr crystals is quite complicated [30] and
the information needed to compare with the new model was not anticipated. For
example, estimates of the Schmidt factors for experimental conditions in Ref. [26]
show that most favorable slip planes for prismatic dislocations will be in the direc-
tions non-parallel to the loading direction. This does not favor the production of
dislocations with the Burgers vector parallel to the loading direction required in the
present model. However, such an analysis does not take into account evolution of
screw dislocation structure during tensile straining and their effect on the distribu-
tion of edge dislocations. In addition, the dislocation density measured after defor-
mation in Ref. [26] is quite low (2 1012 m2), which is of the same order as in a
non-deformed crystal. This makes it difficult to speculate on the real structure of
edge a dislocations after deformation.
ALIGNMENT OF VACANCY-TYPE DEFECTS

The planar alignment of both vacancy-type loops and voids observed in c planes in
neutron-irradiated hcp metals [22] is analogous to void ordering in cubic metals
and may have a similar origin, which is the interaction of voids and loops with SIA
clusters diffusing 1D along close-packed crystallographic directions (see the referen-
ces in Ref. [17]). At high temperature, the void alignment in hcp metals could be
driven by the SIA cluster-void interactions. At low temperatures, the alignment of
c-type vacancy loops could be due to interaction with the SIA clusters, but the exact
ordering mechanism is not yet clear. There are at least two possibilities: (a) via com-
plete annihilation of unaligned loops, or (b) by the loop repulsion by moving SIA
clusters. Further modeling by MD may clarify the mechanism.
Note that the planar alignment of the vacancy-type defects was also found in
Zr under 1 MeV electron irradiation [24], i.e., in the absence of displacement cas-
cades. It seems reasonable to associate it with the preferential migration of single
SIAs in the basal plane, as in the DAD model. However, anisotropic vacancy diffu-
sion revealed in ab initio calculations [14] jeopardizes this explanation. The 1D dif-
fusion of small, kinetically formed clusters may be another possibility. Detailed
analysis of this case, however, is beyond the scope of the present study. Atomistic
simulations are required to clarify the issue.
Estimates of Strain Rates

ABSOLUTE MAXIMUM
It follows from Eqs 14–16 that one of the limiting cases for the strain rates may be
achieved when the density of c dislocations is very high: qz =q ! 1 and qx;y =q ! 0.
In this case

dex dey v
(31)
d/ max d/ max 2

dez
(32) v
d/ max
High rates may also be realized in the opposite case, when the density of c disloca-
tions is relatively small: qz qx ; qy , but qx and qy are very different, e.g., qx qy .
In this case

dex dey v
(33)
d/ max d/ max 2

dez
(34) 0
d/
In both cases, the a-strain rates are fully determined by the properties of cascades.
Taking the value of v equal to 2 102 (see the Stage I: Initial Growth section for ex-
planation), one may estimate the a-strain rate to be of the order of 102 dpa1 (1 %/
dpa), which is the same as the maximum swelling rate in fcc metals and austenitic
stainless steels. Such a coincidence could not be accidental, but must reflect the similar-
ity of the cascades in the metallic materials, particularly with fcc and hcp structures,
and mechanisms operating under cascade damage conditions. However, the first case
considered above is not realistic for Zr-based materials because the density of c disloca-
tions is normally significantly smaller than that of a dislocations. The second case is
more realistic, but to our knowledge such a high strain rate has not been reported yet.
ISOTROPIC DISTRIBUTION OF DISLOCATIONS

For an isotropic distribution of dislocation Burgers vectors: qx ¼ qy ¼ qz , Eqs
14–16 give

dex dey v
(35) ¼ ¼
d/ d/ 6

dez v
(36) ¼
d/ 3
The strain rates are three times smaller than the maximum values given by Eqs 31
and 32, but still about three times higher than 103 dpa1 observed. Realistic
strain rates are predicted by the model when the difference between densities of c
and a dislocations is taken into consideration and qz < qx ; qy , which reduces the c
and a strains according to Eq 16. For example, in the case when
qz ¼ qx þ qy =5 and qy ¼ qx , which is typical for Zr alloys, the a-strain rates are
103 dpa1 which fits well the observations.
Dose Dependence of Growth Strains

The analysis presented in the previous two sections is based on Eqs 14 and 16 for
the instantaneous strain rates in a crystal with a given dislocation structure. With
increasing dose, the total dislocation densities change due to nucleation and growth
of dislocation loops:
(37) qx;y;z ð/Þ ¼ q0x;y;z þ 2pRx;y;z ð/ÞNx;y;z ð/Þ
where q0 are the initial dislocation densities, and R and N are the radii and den-
sities of a and c loops. To calculate the dose dependence of strains, one needs to
know the dose dependences of the loop radii and densities.
The loop radii are described by equations similar to Eq 8 for the dislocation
climb velocities. For loops with particular Burgers vectors:
j
dRaj 2
n Dcl Ccl kj Dv Cv Di Ci
¼ ; j ¼ a1 ; a2 ; a3
dt bj qj bj
(38)
dRc 1
¼ ðDv Cv Di Ci Þ
dt bc
The equations for the effective radii of a loops in the Cartesian coordinate system,
Rx;y ð/Þ, can be obtained from Eq 38 using equations similar to Eqs 17 and 18,
which connect the effective dislocation densities with those in particular crystallo-
graphic directions.
The situation for the nucleation of loops is more difficult because not much is
known. Available experimental data can be summarized as follows: (a) a loops nu-
cleate from the very beginning of irradiation and reach the density of 1022 m3 af-
ter several dpa, and (b) c loops start nucleating at a dose of 3 dpa and reach an
order of magnitude smaller density, 1021 m3. Determining the nucleation mech-
anisms of a and c loops is beyond the scope of the present work. In this paper, we
use the loop nucleation scenario shown in Fig. 5, which has been derived from
observations (see the next section). Based on Eq 38, the dose dependence of the
growth strain has been calculated using a computer code RIMD-ZR.V1 (radiation-
induced microstructure and deformation of Zr, Version 1) developed by the authors
[31]. Selected calculations are presented below.
RADIATION GROWTH IN ANNEALED MATERIALS

As a first step, we fit the nucleation scenario (Fig. 6) to the experimental data from
[29] for low-dose neutron irradiation of annealed Zr crystals, using v ¼ 2 102
and 1012 m2 for the initial densities of a dislocations of all three prismatic direc-
tions, and two times smaller value for c dislocation. The best-fit results together
with the experimental data are presented in Fig. 6 and demonstrate excellent agree-
ment. Then the calculations have been continued to larger doses, for which the
results are shown on Fig. 7. The model reproduces all three stages observed: fast ini-
tial growth, strain saturation, and the breakaway growth.
EFFECT OF COLD WORK

The effect of cold work has been investigated by changing the initial dislocation
density by three orders of magnitude: from 3 1012 to 3 1015 m2, with the
FIG. 5 Vacancy and interstitial loop nucleation scenario.
description of the loop nucleation shown on Fig. 5. The results are presented in
Fig. 8. An increase of dislocation density leads to a qualitative change in the strain
behavior, with a high strain rate 103 dpa1 maintained from the very beginning
of irradiation. This is in a good agreement with experimental data [29].
FIG. 6 Best-fit calculations and experimental measurements from Ref. [29].

FIG. 7 Growth strain in a wider dose range.
RG AT HIGH DOSES
The results shown in Figs. 7 and 8 have been obtained for relatively low doses, <10
dpa, which allows comparing with observations available. A more important issue
is the dose dependence in the breakaway stage at doses beyond existing databases.
FIG. 8 Effect of cold work on growth strain behavior.

An analytical study [31] predicts constant strain rates at high irradiation doses,
determined by the number densities of a and c loops. The results of calculations are
presented in Fig. 9 and demonstrate the sensitivity to the ratio of the a- and c-loop
densities. The higher this ratio, the smaller the strain rate because it is proportional
to ðNic =Nia Þ1=2 . It follows from the model that the growth strain at high doses can
be predicted from the microstructure at end of the loop nucleation period.
COEXISTENCE OF VACANCY AND INTERSTITIAL A LOOPS

The calculated results presented above have been obtained for an isotropic distribu-
tion of a-dislocation Burgers vectors. In this case, the usual RG behavior is exhib-
ited, i.e., expansion in a directions and contraction in the c direction. The results
presented in Fig. 10 have been obtained for the case when the distribution of a dislo-
cations is not isotropic, the density of dislocations with the Burgers vectors parallel
to a1 is five times higher than those with the Burgers vectors along a2 and a3. The
nucleation of a loops is described using a commonly accepted scheme, where the
vacancy loops are nucleated for positive and interstitial loops for negative net va-
cancy flux to the loop embryos. The model predicts that the strain rate in the x
direction (parallel to a1) is negative, i.e., in the direction of the highest dislocation
density, in accordance with the discussion in the Negative a Strains section above.
The a strain is negative until 4 dpa when the distribution of a-dislocation Burgers
vectors becomes more isotropic and the strain becomes positive. The change of the
negative a strain to positive with increasing dose has been observed [2] so that our
calculations agree with experimental observations. Finally, we note that negative a
strains have to be quite common within certain dose ranges because the probability
FIG. 9 Growth strain behavior at very high doses.

FIG. 10 Growth strain behavior for anisotropic distribution of a-dislocation Burgers

vectors.
of isotropic distribution of a-dislocation Burgers vectors in cold-worked samples is

quite small.
Summary
A model of radiation growth of Zr single crystals under neutron irradiation has
been developed which takes into account the true nature of the primary damage
due to atomic displacement cascades, and the diffusion properties of SIA clusters.
The model contains one parameter only, the fraction of SIAs produced in cascades
in the form of clusters, which has been estimated from MD results and experiments.
The model explains all the major observations of RG in Zr including strain satura-
tion, breakaway growth, the effect of cold work, negative a strain, and coexistence
of vacancy and interstitial prismatic loops.
The main model predictions can be summarized as follows:
• The strains in prismatic directions are positive for an isotropic distribution of
prismatic dislocation Burgers vectors.
• The maximum strain rate in this case is estimated to be 103 dpa1, in ac-
cordance with experiment.
• For the first time, the linear dose dependence of the growth strain in the
breakaway stage is predicted at high doses. The corresponding rate can be cal-
culated from the microstructure at intermediate doses.
• It has been shown for the first time that an anisotropy in the distribution of
prismatic dislocations is an important factor determining strain behavior.
• Observations of negative a strain and co-existence of vacancy and interstitial
prismatic loops are both explained for the first time. It is shown that both of
these phenomena originate from anisotropy of prismatic dislocation Burgers

vectors.
• The absolute maximum of the strain rate is estimated to be 102 dpa1. It
may be realized if there is a significant anisotropy of prismatic dislocation Bur-
gers vectors and relatively small density of c dislocations.
It should be mentioned that the absolute maximum growth strain rate pre-
dicted by the model is of the same order as the maximum swelling rate observed in
cubic metals and predicted by the PBM (see the references in Ref. [18]). This indi-
cates that the mechanisms governing damage accumulation in cubic and hcp crys-
tals are similar. Thus, the PBM developed for cubic metals, may provide a general
framework for the description of radiation effects in metallic materials.
ACKNOWLEDGMENTS
This research was supported by the Consortium for Advanced Simulation of Light
Water Reactors (http://www.casl.gov), an Energy Innovation Hub (http://www.energy.
gov/hubs) for Modeling and Simulation of Nuclear Reactors under U.S. Department
of Energy Contract No. DE-AC05-00OR2272.
Appendix
The interaction between a dislocation loop and an edge dislocation depends strongly
on their mutual orientation. This was analyzed by Makin [32] using the infinitesimal
loop approximation in the framework of the isotropic elasticity theory. The corre-
sponding interaction energy can readily be obtained with the aid of Eq 1 in Ref. [32]
for the components of the stress tensor, rij , as
X
(A1) E¼A rij bi nj
i;j
where:
bi ¼ the component of the cluster Burgers vector on i direction, and
Anj ¼ the area of the loop resolved onto a plane perpendicular to the j direction.
We use this approach for SIA clusters. The result is as follows:
Consider a Cartesian coordinate system where the x axis is along a1 , y along
a2 a3 , and z along c, and the 13 ½2110 ð0110Þ straight edge dislocation with its line
along z direction (line sense) and the Burgers vector along x. If the Burgers vector of a
1
3 h2110i SIA cluster is perpendicular to dislocation line and makes an angle a (which
is equal to either zero or p/3) with the dislocation Burgers vector, the interaction
energy is defined by the following equation:
E0 bn 2 2
2 2 2
2 2
2
Ea ¼ 2 y 3x þ y cos a þ 2x x y cos a sin a þ y x y sin a
ðx 2þy Þ2
(A2)
where:
E0 ¼ lX=2pð1 Þ,
l ¼ the shear modulus,
¼ the Poisson ratio,
X ¼ the atomic volume, and
n ¼ the number of SIAs in the cluster, which enters via the relationship Ab ¼ Xn.
When the Burgers vectors of the SIA cluster and a dislocation are parallel to each
other, a ¼ 0, Eq A2 is reduced to the following equation:
yð3x2 þ y2 Þ
(A3) Ea ða ¼ 0Þ ¼ E0 bn
ðx 2 þ y 2 Þ2
When the Burgers vectors of the SIA cluster and a dislocation are non-parallel, a ¼ p/3,
Eq A2 is reduced to
pffiffiffi pffiffiffi
p E0 bn y 2 2
3 3 2
(A4) Ea a ¼ ¼ 3x þ y þ x þ y x y2
3 ðx 2 þ y 2 Þ2 4 2 2
Consider the c-type ½0001 ð0110Þ straight edge dislocation with its line along x direc-
tion and the Burgers vector along z. If the Burgers vector of an SIA cluster is perpen-
dicular to the dislocation Burgers vector and at an arbitrary angle b to the dislocation
line, the interaction energy is given by
E0 bn 2 2
2 2 2
2
(A5) Ec ¼ 2 y z y sin b 2y z þ y cos b
ðz 2þy Þ2
For the case when the cluster Burgers vector is perpendicular to that of the dislocation
line, b ¼ p=2, Eq A5 is reduced to the following equation:
p y ðz 2 y 2 Þ
(A6) Ec b ¼ ¼ E0 bn
2 ðz 2 þ y2 Þ2
Equations A3, A4, and A6 allow calculating the cluster-dislocation interaction energy,
E, and corresponding trapping zones: the areas with the binding energy (–E) higher
than the thermal energy, kBT, where kB is the Boltzmann constant and T the absolute
temperature. Figures 1–3 show the interaction energy between an edge dislocation and
a 10-SIA cluster in Zr at 573 K for three cases: (1) an a dislocation with the Burgers
vector parallel to that of the cluster, (2) an a dislocation with the Burgers vector at p/3
angle to that of the cluster, and (3) a c dislocation with the Burgers vector perpendicu-
lar to that of the cluster. Each figure shows three regions: the capture zone (dark)
where E kB T, the repulsion zone (grey) where E > kB T, and an intermediate
region where E 0 (bright gray). The calculations were performed with l ¼ 66 GPa,
¼ 0.34, X ¼ 2.33 1029 m3, for which E0 ¼ 2.0 eV. As can be seen from the fig-
ures, the cross-section of the capture zone (perpendicular to the cluster motion direc-
tion along its Burgers vector) in the first case is the largest, 250 b. The other two
cases are characterized by significantly smaller interaction cross-sections and weaker
interaction. This is the reason for the simplifying assumption in the model, which
ignores relatively weak interactions of the clusters and a dislocations with non-parallel
Burgers vectors and c dislocations.
References
[1] Buckley, S. N., Properties of Reactor Materials and Effects of Radiation Damage, W. J.
Littler, Ed., Butterworths, London, 1962.
[2] Holt, R. A., “Mechanisms of Irradiation Growth of Alpha-Zirconium Alloys,” J. Nucl.

Mater., Vol. 159, 1988, pp. 310–338.
[3] Woo, C. H. and Gösele, U. M., “Dislocation Bias in an Anisotropic Diffusive Medium and
Irradiation Growth,” J. Nucl. Mater., Vol. 119, 1983, pp. 219–228.
[4] Woo, C. H., “Theory of Irradiation Deformation in Non-Cubic Metals: Effects of Aniso-
tropic Diffusion,” J. Nucl. Mater., Vol. 159, 1988, pp. 237–256.
[5] Guerard, B. V., Grasse, D., and Peisl, J., “Structure of Defect Cascades in Fast-Neutron-
Irradiated Aluminum by Diffuse X-Ray Scattering,” Phys. Rev. Lett., Vol. 44, 1980, pp.
262–265.
[6] Stoller, R. E., “Primary Radiation Damage Formation,” Comprehensive Nuclear Materials,
Vol. 1, R. J. M. Konings, Ed., Elsevier, Amsterdam, 2012, pp. 293–332.
[7] Holt, R. A., Woo, C. H., and Chow, C. K., “Production Bias—A Potential Driving Force for
Irradiation Growth,” J. Nucl. Mater., Vol. 205, 1993, pp. 293–300.
[8] Samolyuk, G. D., Golubov, S. I., Osetsky, Y. N., and Stoller, R. E., “First Principles Analysis
Revises Understanding of Interstitial Defects in Hcp Zr,” Philos. Mag. Lett.,Vol. 93, No. 2,
2013, pp. 93–100.
[9] Christensen, M., Wolf, W., Freeman, C., Wimmer, E., Adamson, R. B., Hallstadius, L., Canton-
wine, P., and Mader, E. V., “Effect of Hydrogen on Dimensional Changes of Zirconium and
the Influence of Alloying Elements: First-Principles and Classical Simulations of Point
Defects, Dislocation Loops, and Hydrides,” ASTM Conference Proceedings of 17th Interna-
tional Symposium on Zirconium in the Nuclear Industry, Hyderabad, India (in press).
[10] Christien, F. and Barbu, A., “Cluster Dynamics Modeling of Irradiation Growth of Zirco-
nium Single Crystals,” J. Nucl. Mater., Vol. 393, 2009, pp. 153–161.
[11] Wolfer, W. G. and Si-Ahmed, A., “The Effect of Nonlinear Elasticity on the Capture Effi-
ciency of Dislocation Loops,” Phys. Lett., Vol. 76A, 1980, pp. 341–343.
[12] Woo, C. H., “Intrinsic Bias Differential Between Vacancy Loops and Interstitial Loops,” J.
Nucl. Mater., Vol. 107, 1982, pp. 20–30.
[13] Dubinko, V. I., Abyzov, A. S., and Turkin, A. A., “Numerical Evaluation of the Dislocation
Loop Bias,” J. Nucl. Mater., Vol. 336, 2005, pp. 11–21.
[14] Vérité, G., Willaime, F., and Fu, C. C., “Anisotropy of the Vacancy Migration in Ti, Zr and
Hf Hexagonal Close-Packed Metals From First Principals,” Solid State Phenom., Vol. 129,
2007, pp. 75–81.
[15] Holt, R. A., “In-Reactor Deformation of Cold-Worked Zr–2.5Nb Pressure Tubes,” J. Nucl.
Mater., Vol. 372, 2008, pp. 182–214.
[16] Adamson, R., Garzarolli, F., and Patterson, C., In-Reactor Creep of Zirconium Alloys,
Advance Nuclear Technology International, Krongjutarvägen, Sweden, 2009.
[17] Singh, B. N., Trinkaus, H., and Golubov, S. I., “Radiation Damage in Crystalline Solids:
Theory,” Encyclopedia of Materials: Science and Technology, K. H. J. Buschow, R. W.
Cahn, M. C. Flemings, B. Ilschner, E. J. Kramer, and S. Mahajan, Eds., Pergamon, Tarry-
town, NY, Vol. 8, pp. 7957–7972.
[18] Golubov, S. I., Barashev, A. V., and Stoller, R. E., “Radiation Damage Theory,” Comprehen-
sive Nuclear Materials, Vol. 1, R. J. M. Konings, Ed., Elsevier, Amsterdam, 2012, pp.
357–391.
[19] Wooding, S. J., Howe, L. M., Gao, F., Calder, A. F., and Bacon, D. J., “A Molecular Dynam-
ics Study of High-Energy Displacement Cascades in a-Zirconium,” J. Nucl. Mater., Vol.
254, 1998, pp. 191–204.
[20] De Diego, N., Osetsky, Y. N., and Bacon, D. J., 2000, “Mobility of Interstitial Clusters in
HCP Zirconium,” MRS Proceedings, Vol. 650, p. R3.30.
[21] De Diego, N., Osetsky, Y. N., and Bacon, D. J., “Structure and Properties of Vacancy and
Interstitial Clusters in a-Zirconium,” J. Nucl. Mater., Vol. 374, 2008, pp. 87–94.
[22] Griffiths, M., Holt, R. A., and Rogerson, A., “Microstructural Aspects of Accelerated De-
formation of Zircaloy Nuclear Reactor Components During Service,” J. Nucl. Mater., Vol.
225, 1995, pp. 245–258.
[23] Risbet, R. and Levy, V., “Ordre de Cavites Dans le Magagnesium et L’aluminium Irradies
aux Neutrons Rapides,” J. Nucl. Mater., Vol. 50, 1974, pp. 116–118.
[24] de Carlan, Y., Regnard, C., Griffiths, M., Gilbon, D., and Lemaignan, C., “Influence of Iron
in the Nucleation of (c) Component Dislocation Loops in Irradiated Zircaloy-4,” Zirco-
nium in the Nuclear Industry: Eleventh International Symposium, ASTM STP 1295, ASTM
[25] Norgett, M. J., Robinson, M. T., and Torrens, I. M., “A Proposed Method of Calculating Dis-
placement Dose Rates,” Nucl. Eng. Des., Vol. 33, 1975, pp. 50–54.
[26] Zee, R. H., Carpenter, G. J. C., Rogerson, A., and Walters, J. F., “Irradiation Growth in
Deformed Zirconium,” J. Nucl. Mater., Vol. 150, 1987, pp. 319–330.
[27] Golubov, S. I., Barashev, A. V., and Stoller, R. E., “On the Origin of Radiation Growth
of HCP Crystals,” ORNL/TM-2011/473, Oak Ridge National Laboratory, Oak Ridge, TN,
2011.
[28] Golubov, S. I., Singh, B. N., and Trinkaus, H., “On Recoil-Energy-Dependent Defect Accu-
mulation in Pure Copper. Part II. Theoretical Treatment,” Philos. Mag., Vol. A81, 2001, pp.
2533–2552.
[29] Carpenter, G. J. C., Rogerson, A., and Zee, R. H., “Irradiation Growth of Zirconium Single
Crystals: A Review,” J. Nucl. Mater., Vol. 159, 1988, pp. 86–100.
[30] Akhtar, A. and Teghtsoonian, A., “Plastic Deformation of Zirconium Single Crystals,”
[31] Barashev, A. V., Golubov, S. I., and Stoller, R. E., “Theoretical Investigation of Microstruc-
ture Evolution and Deformation of Zirconium Under Cascade Damage Conditions,”
ORNL/TM -2012/225, Oak Ridge National Laboratory, Oak Ridge, TN, 2012.
[32] Makin, M. J., “The Long-Range Forces Between Dislocation Loops and Dislocations,”
Philos. Mag., Vol. 106, 1964, pp. 695–711.
DISCUSSION
Question from Srikumar Banerjee, BARC Mumbai:—The change in Nb concen-
tration of b phase (from 85%Nb to much lower Nb content) is observed in samples
exposed to creep deformation (both under irradiation and without irradiation).
Have you examined the process under metastable thermodynamic equilibrium in
which the phase separation tendency of the b-phase plays a major role?
Authors’ Response:—No, the effect of b-phase has not been examined. The topic
is beyond the present form of the RG model. The model has first to be generalized
to account for the multigrain structure of Zr materials and internal and external
stresses. After that, an effect of secondary phase particles on dimensional instability
may be properly studied.
Q1:—How is the effect of irradiation temperature on the observed growth

addressed in the model?.
Authors’ Response:—The RG model provides a description of RG strain rate as a

function of dislocation structure and contains one parameter only, which is the fraction
of self-interstitial atoms (SIAs) generated by cascades in the form of mobile SIA loops.
Thus, in accordance with MD simulations of cascades, it depends only weakly on tem-
perature. In contrast, the dislocation structure depends on irradiation dose via nuclea-
tion of a- and c- dislocation loops that is a temperature dependent process as any other
nucleation process. However, it is not possible yet to describe the temperature depend-
ence because the loop nucleation in HCP metals, and particular in Zr, has not been stud-
ied yet. Note also that the RG model in its present form is valid in a limited temperature
interval only: the irradiation temperature has to be higher than that corresponding to
the annealing stage III, i.e., when vacancies become mobile, and lower than that of the
swelling regime, when voids form. The description of RG in the swelling regime has al-
ready been developed in the same framework and will be published soon.
Q2:—I observe that the data referred to for the formulation of the model is based
on fast reactor irradiations. How will it relate to the observations in thermal reactors,
where the damage rates are much lower than that experienced in a fast reactor?
Authors’ Response:—Indeed, the experimental observations in cubic materials

irradiated in fast reactors, together with those for ion and electron irradiations,
have been used to formulate the RG model. However, they are not just “the data ...
based on fast reactor irradiations”, they characterize the processes taking place in
metallic materials under neutron irradiation, such as the primary damage and prop-
erties of the SIA clusters, which are similar in cubic and HCP crystals. There is no
damage-rate dependence in the RG model in its present form, so the model is fully
applicable to thermal reactor conditions.
Q3:—What could be the effect of grain size and grain shape on irradiation
growth? How is this addressed in the proposed model?
Authors’ Response:—The model in its present form describes RG in a single crys-

tal. In multigrain materials, the situation is more complicated: finite size grains provide
(a) additional sinks for mobile defects, hence may affect RG and (b) radiation creep
induced by RG is unavoidable even in the absence of an external stress. The effect of
(a), according to our estimates, may be important at small grain sizes, less than about
several microns, whereas (b) could be important at larger grains as well. Note that the
RG model gives dislocation climb rates, hence provides a way for describing radiation
creep in the framework of the so-called climb-induced glide mechanism; this work is
currently in progress. Surprisingly such a task has never been attempted before.
Q4:—Is the model specific to Zr? Can this be used for example to predict irradi-
ation growth of U?
Authors’ Response:—The model has been developed for Zr based materials but
it may be valid for other HCP metals as well, e.g. for those with the c/a ratio smaller
than the ideal one. For example, void ordering along the basal planes found in Mg
[23], which is similar to that found in Zr, shows that the main feature of our model,
migration of SIA clusters along the basal planes, has to be valid for Mg as well.
Models of RG for metals with the c/a ratio higher than the ideal and for metals with
other crystal structures, e. g. uranium, require a detailed knowledge on the primary
damage and properties of SIAs and their clusters.
Questions from Ron Adamson, Zircaloy Plus
Q1:—In your model, is there a physical explanation that <c> strain is

always negative?
Authors’ Response:—Yes, and this is the central point of the model. Accounting
for the cascade production of SIA loops and their one-dimensional diffusion leads
to a qualitative change in the description of the interaction of dislocations with mo-
bile defects, hence their absorption ability, as compared to previous RG models
based on Frenkel pair production only. Indeed, the interaction of SIA loops with
dislocations depends on mutual orientation of their Burgers vectors, and is negligi-
ble when they are perpendicular to each other. Since the SIA loop Burgers vectors
are along prismatic directions, the loops are almost not absorbed by c dislocations,
which Burger vectors are perpendicular to that of SIA loops. This contrasts to the
case of single SIAs which are absorbed by c dislocations with practically the same
efficiency as a dislocations.
Q2:—What is the size range of the SIA clusters and can they be observed in
modern TEMs?
Authors’ Response:—According to MD simulations, the clusters vary from 2 to

several tens, with an average value of about 10 SIAs. The high mobility and small
size of the clusters (the diameter of a 10 SIA cluster is about 1 nm) make the TEM
observation difficult. In principle, they can be observed by X-ray as it was shown in
[5] for Cu irradiated at He temperatures.
Q3:—Most models today assume (or calculate) that migration of individual SIA
is highly biased in the <a> directions. Do you disagree with this?
Authors’ Response:—We disagree. Our recent ab-initio calculations have shown

that the migration barriers for in- and out- of c-plane diffusion of a single SIA are
very close to each other [8], thus the migration of individual SIAs is not highly bi-
ased in the a directions.
Question from P. Barberis, CEZUS Research Center:—From a mathematical point

of view, is your model stable? In other terms, do small variations in initial condi-
tions (or nucleation conditions) produce small variations in the resulting growth?
That is important regarding small variations in processing or operating conditions.
Authors’ Response:—The model is stable from a mathematical point of view.
Q1:—What experimental observations of growth does your model explain that

C.H. Woo’s DAD model does not?
Authors’ Response:—There are several such observations. The most important are
as follows: (a) Our model explains and quantitatively describes the coexistence of va-
cancy and SIA prismatic loops of comparable sizes and number densities in the bulk, in
contrast to the DAD model which suggested one type of the loop in the bulk and the
other near grain boundaries (to our knowledge this has never been observed). (b) Our
model explains and quantitatively describes the negative a-strain effect which has never
been explained by other RG models including the DAD. In addition, our model reveals
that there is a common origin for both observations. Note that there is no way to
explain the negative a-strain in the framework of Frenkel pair production, like DAD.
This is due to independence of interactions of prismatic dislocations with point defects
on the orientation of the dislocation Burgers vector. If negative a strain were observed
in this case, then all a and c strains would be negative, hence the total volume of crystal
would decrease, which is absurd.
Q2:—The Horak and Rhude paper from 1961 (J. A. Horak and H. V. Rhude,
Journal of Nuclear Materials, 3 (1961) 111-112). actually predicted a growth rate of
104/dpa, from what I recall. The damage in that case is very high (created by fis-
sion products) but the growth rate was similar to that in cladding, according to R.
Holt (400% strain with 40000 dpa).
Authors’ Response:—In the model, the maximum strain rate is determined by

the fraction of SIAs produced in the clustered form, which is 102 dpa and
depends only weakly on the PKA energy due to sub-cascade production at high val-
ues of PKA energy (see details in [6]). In Zr materials, where the density of c dislo-
cations is 5–10 times smaller than that of a dislocations, the maximum RG rate is
predicted to be of the order of 103/dpa. So, the value 104/dpa observed by
Horak and Rhude should correspond to a larger difference between c and a disloca-
tion densities (including contributions from corresponding dislocation loops).
Q3:—In your model you use a fraction of clusters derived from MD studies in
cubic materials. Why do you believe their results are valid for hcp Zr? Do you plan
to conduct MD simulations of cascade in Zr, or maybe use Bacon’s work?
Authors’ Response:—There are two reasons for using the value of the fraction
for cubic materials, namely from MD simulations and experimental data for Cu.
First, as can be seen on Fig. 8 in [19], the fractions of clustered SIAs in Zr and Cu
cascades found in MD are very close to each other. Second, the cluster formation
takes place at high PKA energies due to the shock waves interaction, which is prac-
tically independent on the details of interatomic interactions (see details in Calder
et al., Phil. Mag. (2010) 863).
Q4:—How does your model account for low temperature (<77K) growth under
irradiation?
Authors’ Response:—The RG model is not valid at these temperatures; as it has

already been pointed out above it is valid at temperatures higher than that of the
annealing stage III which is much higher than 77K.
Question from Igor Evdokimov, SRC RF Triniti, Russia:—What can you anticipate,
in the frame of your model, about the hydrogen effect on irradiation growth of Zr-based
alloys? What is more important: hydrogen in solid solution or in the hydride phase?
Authors’ Response:—The hydrogen effect is not included yet in the RG model.

S. Doriot,1 B. Verhaeghe,2 J.-L. Béchade,1 D. Menut,1

D. Gilbon,3 J.-P. Mardon,4 J.-M. Cloué,4 A. Miquet,5
and L. Legras6
Microstructural Evolution of
M5TM7 Alloy Irradiated in PWRs
up to High Fluences—Comparison
With Other Zr-Based Alloys
Reference
Doriot, S., Verhaeghe, B., Béchade, J.-L., Menut, D., Gilbon, D., Mardon, J.-P., Cloué, J.-M.,
Miquet, A.,and Legras, L., “Microstructural Evolution of M5TM Alloy Irradiated in PWRs up to
High Fluences—Comparison With Other Zr-Based Alloys,” Zirconium in the Nuclear Industry:
17th International Symposium, STP 1543, Robert Comstock and Pierre Barberis, Eds., pp. 759–
799, doi:10.1520/STP154320120179, ASTM International, West Conshohocken, PA 2015.8
ABSTRACT
This paper focuses on the microstructural evolution of the M5 alloy under
irradiation for fast neutron fluences up to 17.1 1025 n/m2 (E > 1 MeV) in pressurized
water reactors (PWRs). The precipitates and especially the radiation-enhanced
particles were studied first with analytical transmission electron microscopy
(ATEM) and transmission x-ray diffraction (T-XRD) analyses at the SOLEIL
synchrotron facility. Then ATEM was used to study the evolution versus the fast
neutron fluence of hai- and hci-component loops in M5 alloy and RXA Zy-4.
Original T-XRD permitted us to measure the composition and the lattice
Manuscript received November 23, 2012; accepted for publication July 21, 2013; published online September
15, 2014.
1
CEA-DEN, Section for Applied Metallurgy Research, CEA/Saclay, 91191 Gif-sur-Yvette Cedex, France.
2
CEA-DEN, Section for Research on Irradiated Materials, CEA/Saclay, 91191 Gif-sur-Yvette Cedex, France.
3
CEA-DEN, Nuclear Material Dept., CEA/Saclay, 91191 Gif-sur-Yvette Cedex, France.
4
AREVA, AREVA NP, Fuel Business Unit, 10 rue Juliette Récamier, 69456 Lyon Cedex 06, France.
5
Electricité de France-DIN Septen, 12-14 Ave. Dutriévoz, 69628 Villeurbanne Cedex, France.
6
Electricité de France, R&D Division, Materials and Mechanics of Components, Les Renardières, 77818 Moret
sur Loing, Cedex, France.
7
M5TM is a trademark of AREVA NP registered in the USA and in other countries.
8
Hyderabad, India.
Copyright V
parameters of the radiation-enhanced precipitation of nanometric needle-like

particles. Accurate evolutions of microstructural features such as hci-component
loops, hai-loops, natives, and radiation-enhanced precipitates have also been
determined as a function of irradiation doses up to 20 1025 n/m2. The precipitate
study permitted us to propose an evaluation of the niobium content in the matrix
during irradiation. All these results confirm the noteworthy microstructural and
microchemical stability of the M5 alloy during irradiation, especially at doses of at
least 8 1025 n/m2. The bNb native particles present in M5 alloy remain fully
crystalline after irradiation. After a fluence of 11 1025 n/m2, native particles reach
the “equilibrium” composition of 55 % Nb under irradiation. The irradiation-
enhanced precipitation leads to a noticeable decrease in Nb content in the matrix,
but no further evolution in size, density, or niobium content seems to occur for
these particles after a dose of 8 1025 n/m2. Finally, the density of hci-component
loops observed in this work for M5 cladding tubes remained moderate relative to
that for Zy-4 alloy even for the highest dose of irradiation. We can also underline
that the iron content in M5 in Zr1 %NbOA and Zr1 %NbOB alloys does not seem to
influence either hci-component loop linear density or needle-like particle size or
density in the range of 100 to 650 ppm.
Keywords
irradiation, dislocation loops, microstructure, Zr1Nb alloys, precipitation, growth
Introduction
The current challenge in the nuclear energy industry is to increase pressurized water
reactor (PWR) fuel cladding tube performance, especially for corrosion resistance and
dimensional stability (irradiation creep and growth). In this framework, for cladding-
tube applications, the fully recrystallized (RXA) M5 alloy is replacing the stress-
relieved (SRA) Zircaloy-4. M5 alloy has been used extensively in a wide range of duty
PWR environments at high burnups of up to 80 GWd/t (seven annual cycles) and has
demonstrated superior in-reactor behavior [1] relative to low-tin Zircaloy-4 [2,3].
Previous studies have detailed the microstructural evolution of Zr alloys under
irradiation and the strong influence of this evolution on corrosion and growth [4–6].
For M5 alloy, previous analytical transmission electronic microscopy examinations
[6] underlined the noteworthy microstructural and microchemical stability of this
alloy under irradiation. The native bNb particles contained in this alloy remain fully
crystalline even after six PWR annual cycles of irradiation (70 GWd/t), but it was
observed that a decrease of niobium content in these particles from 90 to 60 wt.
% correlated to a slight increase in size, as was previously shown for a Zr-1 %Nb alloy
[7–12]. The most important microstructural change for the M5 alloy under irradia-
tion seems to be irradiation-enhanced needle-like precipitation. These niobium-rich
particles were found to be about 4 to 6 nm in length and had a number density of
1.5 1022 m3. They were suspected of leading to a noticeable decrease of the nio-
bium content in the matrix, in agreement with the literature [8]. But because niobium
content in the matrix seems to have a strong influence on non-irradiated Zr–Nb alloy
DORIOT ET AL., DOI 10.1520/STP154320120179 761
thermal creep behavior [13], we completed and refined microstructural examinations

and data on these particles and on native bNb phases. Another important micro-
structural change arising during irradiation is the occurrence of hci-component loops.
Even if the number density of hci-component loops seems particularly low in M5 rel-
ative to RXA Zy-4, it is relevant to measure precisely the evolution of these irradiation
defects in order to predict further evolution under irradiation.
This paper focuses on the microstructural evolution of M5 alloy under irradia-
tion in PWRs for fast neutron fluences up to 17.1 1025 n/m2 (E > 1 MeV). The
first part of this paper discusses precipitates and especially the radiation-enhanced
needle-like particles, which were studied with two complementary techniques, ana-
lytical transmission electron microscopy (ATEM) and transmission x-ray diffrac-
tion (T-XRD), at the Multi-Analyses on Radioactive Samples (MARS) beam line of
the SOLEIL synchrotron facility. In the second part of the paper, we discuss ATEM
studies of the evolution versus the fast neutron fluence of hai- and hci-component
loops in M5 alloy and RXA Zy-4.
Materials
CHEMICAL COMPOSITION AND MICROSTRUCTURE BEFORE IRRADIATION
The chemical composition of the alloys studied in the present work is given in
Table 1. RXA Zy-4 alloy was manufactured in roughly the same way as a standard
SRA Zircaloy-4, with the exception of the final annealing temperature, which was
optimized to produce a fully recrystallized microstructure [1,2]. The M5 alloy pro-
duced using a “low temperature” process [14] was studied in three different chemi-
cal composition grades: M5A, M5D, and M5G.
The initial microstructure of M5 is characterized by a homogeneous, highly
refined dispersion of native bNb precipitates in fully recrystallized grains. Very few
Zr(Fe,Nb)2 Laves phases are observed in this alloy. Zr1 %NbOA, referred as to Zr-1
%NbO in Ref 6, is characterized by alignments of particles crossing the recrystal-
lized grains and composed of fine bNb and coarser bZr. Zr1 %NbOB exhibits
TABLE 1 Chemical composition of studied alloys (wt. % or ppm).
Alloy Sn ppm, % Fe ppm, % Cr ppm, % Nb, % O, % Zr, %
M5A <30 237 32 0.97 0.115 Bal.

M5D 31 146 23 1.01 0.132 Bal.
M5G 35 106 19 1.01 0.126 Bal.
Zr1%NbOA <30 570 76 1.03 0.105 Bal.
Zr1%NbOB <25 643 50 1.01 0.123 Bal.
RXA Zy-4 (1) 1.30% 0.21% 0.11% <0.010 0.125 Bal.
RXA Zy-4 (2) 1.4% 0.21% 0.11% <0.010 0.135 Bal.
FIG. 1 Microstructure of the studied materials: (a) M5; (b) Zr1%NbOA; (c) Zy-4 in the
recrystallized condition (RXA).
alignments of fine precipitates in a fully recrystallized microstructure. Zy-4 shows

the standard RXA Zy-4 microstructure with hexagonal Zr(Fe,Cr)2 Laves phases
evenly dispersed throughout the matrix [4].
Figure 1 illustrates the typical microstructures of M5, Zr1 %NbOA, and RXA Zy-
4 alloys.
IRRADIATION AND MECHANICAL TEST CONDITIONS

All these alloys were irradiated in PWR conditions. The fast fluence (E > 1 MeV)
was deduced from the local fuel rod burnups. The temperatures were calculated
cycle by cycle, taking into account the average and local linear power, the local tem-
perature of water, the thickness of the oxide, and the position of the thin foil in the
thickness of the tube (in the vicinity of the inner surface). The irradiation condi-
tions are given in Tables 2 and 3. In the following, we call the temperature of
TABLE 2 Irradiation condition in PWR 1 reactor and post-irradiation test conditions for M5A
material (cladding tubes).
Fluence, 1025 n/m2 Number of Test Conditions

(E > 1 MeV) PWR Cycles Tirr, C (Internal Pressure)
14.6 7 337 None

13.6 6 305 None
13.6 6 335 Creep, 400 C, 250 MPa, 380 h
13.5 6 320 Creep, 350 C, 300 MPa, 500 h
12.4 6 340 None
11 5 330 None
11 5 330 Creep, 400 C, 130 MPa, 240 h
9.3 4 305–325a Creep, 350 C, 300 MPa, 500 h
9 4 335–345a Creep, 400 C, 250 MPa, 83 h
8.7 4 440 None
7 4 340 None
7.2 4 340 Creep, 350 C, 450 MPa, 4 h
3.4 2 345 Stress relaxationb at e ¼ 0.8%,
350 C, 3 106 s1
2.6 1 350 Stress relaxation at e ¼ 0.8%,
395 C, 3 106 s1
2 1 325 Stress relaxation at e ¼ 0.8%,
350 C, 3 105 s1
2 1 325 Stress relaxation at e ¼ 0.7%,
380 C, 3 105 s1
2 1 350 Creep, 400 C, 130 MPa, 240 h
2 1 330 Creep, 350 C, 300 MPa, 500 h
a
The irradiation temperature was estimated via comparison with samples irradiated in similar
conditions.
b
e ¼ plastic strain level.
TABLE 3 Irradiation condition in PWRs for M5G, M5D, Zr1%NbOA, Zr1%NbOB, and Zy-4.
Fluence, Number of
Alloy PWR 1025 n/m2 (E > 1 MeV) PWR Cycles
M5G free-growth test tube PWR 2 13.4 4 ( 18 months)

M5G free-growth test tube PWR 2 17.1 5 ( 18 months)
M5D cladding tube PWR 3 14 5 ( 18 months)
Zr1%NbOA cladding tube PWR 4 6.5 3
Zr1%NbOA cladding tube PWR 5 12.4 6
Zr1%NbOB test tube PWR 6 19.5
Zy-4 (1) free-growth test tube PWR 2 13.4 4 ( 18 months)
Zy-4 (1) free-growth test tube PWR 2 21 6 ( 18 months)
Zy-4 (2) guide tube PWR 7 9.5 5
irradiation (Tirr) the maximum T nth temperature, where T nth represents the tem-
perature of the sample during the nth irradiation cycle.
When no as-irradiated sample was available, material tested under internal
pressure stress relaxation test conditions was used as the untested sample (see Table
2). Some samples that underwent an internal thermal pressurized creep test after
irradiation were also studied in order to have more data than that of the as-
irradiated ones only. The test conditions are given in Table 2.
ANALYTICAL TRANSMISSION ELECTRON MICROSCOPY
The transmission electron microscopy (TEM) examinations and microanalyses
were conducted using a Philips EM430 microscope operating at 300 kV, a 200-kV
Jeol 2010-F Field-Emission Gun microscope, and a 200-kV Jeol 2100. These exami-
nations were focused on the microstructural evolution of M5 fuel rod claddings
irradiated for between one and seven cycles. Test tube samples of M5, Zr1 %NbOB,
and Zy-4 alloys for irradiation doses up to 20 1025 n/m2 (E > 1 MeV) were also
examined. The aim of this study was mainly to analyze the number density of hci-
component loops and the distribution of needle-like irradiation-enhanced precipi-
tates as a function of irradiation conditions.
A prism-plane foil orientation and a (0002) diffracting vector were used to
image the hci-component loops and the needle-like radiation-enhanced precipitates
(s 0). The needle-like particle distribution was counted on several micrographs at
a magnification of 480 000 using the Noesis Visilog software. The hci-component
loops were measured on several micrographs at a magnification of 127 500. The
thin foil thickness was either measured via stereoscopy or assumed to be 150 nm.
For one, two, four, five, and seven cycles of M5A, some electron energy loss spec-
troscopy (EELS) measurements were conducted in order to have more accurate
thickness values. The (10-11) diffracting vector was used for micrographs at a mag-
nification of 340 000, for hai-loop imaging in prism foil orientation.
The microchemical composition was measured by a scanning and energy dis-
persive device (STEM/X-EDS) and was computed on an OXFORD system by proc-
essing the experimental x-ray spectra (Jeol microscope with a 2-nm probe diameter).
Usually, this technique does not allow us to measure the exact composition of
second-phase particles in thin foils because of the matrix contribution. For this rea-
son, we plotted Nb ¼ f(Zr) (or Fe ¼ f(Zr)) collected on numerous second particles of
the same kind. On the straight-line figure obtained, the lower the percentage of Nb
(or Fe), the higher the matrix contribution.
Because these secondary-phase particles are very tiny, determining their chemi-
cal composition or crystallographic structure in ultrathin specimens via TEM is of-
ten quite difficult because of the matrix contribution to the energy-dispersive
spectroscopy spectrum and the corresponding electron diffraction patterns. Precipi-
tate volume fractions in Zr alloys also are very small. Therefore, one can obtain a
good statistic in TEM analysis only by performing many examinations that are
quite time consuming. Thus, the use of synchrotron-based x-ray techniques is of
utmost interest, with a bigger volume fraction of the bulk materials simultaneously
analyzed that is statistically representative.
SYNCHROTRON X-RAY DIFFRACTION ANALYSES

MARS is the x-ray beam line of the French synchrotron facility SOLEIL dedicated
to the study of radioactive matter. Several papers [15–17] describe the infrastructure
that was defined for the construction of the beam line and the analytical capabilities
offered by the two commissioned end stations. The MARS beam line aims at
extending the possibilities of synchrotron-based x-ray characterizations toward a
wider variety of radioactive elements (a, b, c, and neutron emitters). Thus, its spe-
cific and innovative infrastructure has been optimized to carry out analyses on radi-
oactive materials with activities of up to 18.5 GBq (2 GBq c and neutron emitters)
per sample. This beam line, which was built thanks to a close partnership with and
support from the Commissariat à l’Energie Atomique, has been designed to provide
x-rays in the energy range of 3.5 keV to 35 keV.
The XRD analyses were performed in transmission mode at 17.038 keV (equiv-
alent to a wavelength of 0.72778 Å) with a collimated beam size of 100 lm2 by
100 lm2 (horizontal by vertical dimensions). The sample is positioned on an eight-
axis (six translation and two rotation stages) motorized stage manipulator, and dif-
fraction data are collected with a two-dimensionnal imaging MAR345 detector
mounted perpendicular to the incoming x-ray beam and centered on it.
Final XRD patterns were obtained by averaging 10 acquisition points in order to
increase the statistic without saturating the detector with the diffraction peaks of the
aZr matrix (acquisition time set at 240 s). Measurement parameters, including the
instrumental peak shape and resolution, were determined from the diffraction patterns
of a reference sample of LaB6 standard powder (NIST SRM 660 a). After corrections,
the Debye–Scherrer diffraction rings are unwrapped and integrated versus the azimuth
angle and produce a normal one-dimensional diffraction pattern. Fullproof software
[18] and TOPAS P software (Bruker AXS SAS) were used to perform whole-pattern
profile matching analyses on the diffraction patterns. Values for the mean size of the
coherent diffracting domains were evaluated using the Scherrer function [19]. This
variable (the mean size of coherent diffracting domains) is usually correlated with the
particle size for precipitates without microstrains or faults. Figure 2 and Table 4 give as
an example the whole-pattern profile matching analyses of the non-irradiated M5A ref-
erence sample plotted as the logarithm of intensity.
For the Zr(Fe, Nb)2 Laves phase, XRD whole-pattern profile matching was per-
formed only on the (200) diffraction line related to the hai lattice parameter. This
may explain the relatively high Bragg-R factor and R factor deduced for this phase
matching.
For the bNb phase, the XRD profile must be deconvoluted from different con-
tributions. Before irradiation, surface oxidation and hydrides should be taken into
FIG. 2 Whole-pattern profile matching analyses of the non-irradiated M5A reference

sample plotted as the logarithm of intensity. Solid diamonds, triangles, and
circles indicate the Bragg peak positions for aZr, native bNb, and Zr(Fe,Nb)2
Laves phases, respectively. The fine line at the bottom is the difference between
the observed intensity (dotted line) and the calculated value (solid line).
TABLE 4 Results of the whole-pattern profile matching analyses performed on a non-irradiated

M5A sample with agreement factors between the observed and the calculated data: chi-
square function (v2), the profile factor or R factor (RF), the Bragg-R factor (Bragg-RF),
and the goodness of fit (GoF).
Phase Space Lattice Bragg-RF RF v2 GoF

Group Parameters, Å
aZr P63mc a: 3.2312 0.2 103 0.1 103

(60.0001)
c: 5.1493
(60.0002)
c: 120
bNb Im-3 m a: 3.3258 0.5 102 0.1 101 89.1 9.4
(60.0012)
Zr(Fe, Nb)2 P63mc a: 5.2664
(60.0194)
c: 8.896 0.2 101 0.38 101
c: 120
FIG. 3 TEM micrographs of bNb in M5A irradiated for six cycles.
account. After irradiation, whole-pattern profile matching analyses gave rise to two
body-centered cubic (bcc) phases that agreed with bNb particles with enlarged lat-
tice parameters. Then, via comparison with what was observed in TEM analyses
[7], we made the assumption that the first ones, showing broadened diffraction
peaks, could be attributed to the radiation-enhanced needle-like bNb particles
(nanostructured particles), whereas the second ones, showing sharper diffraction
peaks, could be attributed to the native bNb particles after irradiation. T-XRD anal-
yses were performed on M5A samples irradiated for one, two, and five cycles in
PWR 1.
Results
NATIVE bNB PARTICLE EVOLUTION VERSUS FAST NEUTRON FLUENCE
As expected, all the bNb particles found via TEM were fully crystalline, even for the
highest dose, 19.5 1025 n/m2, but with a granitic aspect as shown in Fig. 3(b).
T-XRD measurement diffraction lines related to the native bNb particles con-
firmed TEM observations; mainly, the precipitates remained crystalline with the
bcc crystallographic structure. But as early as the first irradiation PWR cycle, the
diffraction lines shifted toward smaller angles of diffraction. This line shift
increased linearly through up to five cycles of PWR irradiation, with the lattice pa-
rameter also increasing (Table 5). The Nb content in the native bNb particles could
be deduced from the lattice parameter using a Vegard-type law [20], and data are
reported in Table 5 and Fig. 4.
The Nb content (wt. %) reported in Tables 6 and 7 and in Fig. 4 (ATEM data)
corresponds to the maximum values measured by ATEM of the Nb ¼ f(Zr) curve in
M5A and Zr1 %NbOA alloys irradiated at different fluences. T-XRD and ATEM
TABLE 5 Evolution of native bNb particles versus fast neutron fluence for M5A irradiated in PWR 1,
from T-XRD analyses (uncertainty is determined from line profile fitting errors).
Mean Size of the

Number of Lattice Nb Content, Integral Breadth, Coherent Diffracting
Cycles Parameter, Å wt. % rad Domains, nm
0 3.3258 (60.0012) 91.1 (60.4) 0.002179 (60.000163) 33 (62)

1 3.3577 (60.0031) 80.0 (61.1) 0.008200 (60.000912) 9 (61)
2 3.3703 (60.0022) 75.6 (60.7) 0.006230 (60.000851) 12 (62)
5 3.4267 (60.0013) 56.1 (60.4) 0.006697 (60.000407) 11 (61)
results are in complete agreement with each other, despite the matrix contribution
in ATEM,8 and both show a decrease of Nb content in native bNb particles versus
irradiation fluence until an “equilibrium value” of 55 % under neutron irradiation
conditions was reached, as seen before [6–12]. The Nb content in M5A and in Zr1
%NbOA alloys seems to undergo a similar change versus fluence. All these results
agree with published data [6,9,21–23]. The measurements made via ATEM on the
400 C creep-tested M5A alloy after six PWR cycles seem to show a slight decrease
in the niobium content of the precipitates during the creep test, but this phenom-
enon could be due to the matrix contribution [7] (see Fig. 4).
The particle size distribution obtained via TEM for M5A alloy irradiated in
PWR 1 is reported in Table 6. In M5A samples, the diameter of native bNb particles
increased until about the second cycle of irradiation, from 35 nm to between 45 and
FIG. 4 Nb content (wt. %) of native bNb in M5A and Zr1%NbOA versus fast neutron
fluence.
8
Because of the matrix contribution, the effective niobium content is usually greater than analyzed when
ATEM on thin foil is used.
TABLE 6 Distribution of native bNb particles versus fast neutron fluence for M5A irradiated in PWR
1, from TEM analyses.
Fluence, Number
Number of 1025 n/m2 /min, /mean, /max, Density, Nb, Volume
Cycles (E > 1 MeV) nm nm nm 1020 m3 wt. % Fraction, %
0 0 6 35 150 1.1 75 0.5

1 2 9 38 98 2.2 0.5
1 2.6 9 39.7 105 1 0.5
2 3.4 18 49 145 0.9 0.9
4 7 20 47.4 140 — —
4 8.7 12 45 145 0.6 0.6
5 11 18 52 130 1 61 0.7
5 (creep test, 11 6 41.7 122 1.1 0.9
400 C)
6 12.4 20 47.4 110 0.5 57 0.4
6 13.6 9 51.8 145 0.65 0.7
6 (creep test, 13.6 15 49 116 0.4 45 —
400 C)
7 14.6 12 44 101 0.6 57 0.4
55 nm, and then stabilized for the following irradiation cycles (Fig. 5). In T-XRD
analyses, the structural line broadening increased by a factor of 3 after one cycle of
PWR irradiation and did not change after up to five cycles of PWR irradiation (see
“Integral Breadth” in Table 5). Then, although the mean size of the native bNb par-
ticles increased from 35 nm to 50 nm after irradiation, the mean size of the coherent
diffracting domains of these particles dropped from 33 6 2 nm to 11 6 1 nm
(Table 5). For the non-irradiated M5A sample, TEM and T-XRD data are in com-
plete agreement and both show bNb as particles with a 35-nm diameter.
The number density of particles measured via TEM for M5A alloy irradiated in
PWR 1 was about 1 1020 m3 and exhibited a slight decrease9 (Table 6 and
10
Fig. 6 ) as the Nb content decreased. After a creep test at 400 C, the number
density and the average diameter of particles did not change significantly.
Considering the volume fraction of the particles (Table 6) and their Nb content
extrapolated from Fig. 4, we can plot in the evolution versus fluence of the whole
native particle Nb % content (Fig. 7). It appears that, as we previously concluded
9
It seems there is a slight coalescence of particles with a slight decrease in density and an increase in parti-
cle diameter. But the decrease in particle density is not obvious considering the scattering in the results
because of the difficulty of thickness measurement and the poor contrast of micrographs in irradiated
materials.
10
“Stereo” in Fig. 6 means that for the particle density calculation, the thickness was measured via stereos-
copy, and “150 nm” means that the thickness was supposed to be 150 nm in all the grains.
TABLE 7 Nb weight percent in native bNb versus fast neutron fluence for Zr1%NbOA from TEM
analyses.
Number of Cycles Fluence, 1025 n/m2 (E > 1 MeV) Nb, %
0 0 82
3 6.5 62.5
6 12.4 60.5
[6], the bNb second-phase particles contain about half of the total Nb content of
the alloy before and after irradiation.
LAVES PHASES
Very few Laves phases of Zr(Fe,Nb)2 were detected in TEM microanalyses on M5
alloy before irradiation, and none after irradiation [6], except for rare microcrystal-
lized and highly faulted particles appearing after irradiation (Fig. 8).
For the T-XRD technique, the (200) line profile was adjusted in the non-
irradiated sample despite its weak intensity. The mean size of the coherent diffract-
ing domains was about 57 6 23 nm, which is similar to the diameters observed via
TEM for these particles. As early as the first PWR cycle of neutron irradiation, the
Laves phase vanished in the diffuse background of the T-XRD pattern.
RADIATION-ENHANCED NEEDLE-LIKE PARTICLES

In agreement with previous studies [6–8,10,21,24,25], new phases were detected for
Zr1Nb alloys after neutron irradiation in TEM analyses. These precipitates appear
as early as the first cycle. They are supposed to be bNb particles for their thermody-
namic stability [6,24,26,27]. Their distribution along the basal planes (20 to 30
FIG. 5 Average diameter of native bNb in M5A versus fast neutron fluence.
FIG. 6 Number density of native bNb in M5A versus fast neutron fluence.
deviation), clearly related to the orientation-crystallographic correspondence of the

bcc and hcp lattices, is also typical of bNb particles [12]. Moreover, previous TEM
experiments revealed systematic niobium enrichment of these precipitates [6]. The
small size of these precipitates, however, makes it difficult to clearly characterize
them [12]. These authors suppose that these particles are due to irradiation-
enhanced diffusion controlled by the migration of non-equilibrium irradiation-
induced point defects [26,28–30]. As seen in Fig. 9(a), before irradiation, if we
FIG. 7 Total niobium content versus fast neutron fluence in the native bNb particles in
M5A.
FIG. 8 Second-phase particle assumed to be a prior-Laves phase particle in terms of its

morphology in M5A irradiated for seven cycles (14.6 1025 n/m2).
exclude minor oxide and hydride contributions, the only contribution to the XRD
profile of the bNb phase is the native bNb particles with the expected lattice param-
eter. After irradiation, as discussed under “Experimental Procedures,” whole-pattern
profile matching analyses from the XRD patterns confirmed the appearance of a new
bcc phase that agreed with the bNb particles with enlarged lattice parameters
[Fig. 9(b)]. Figure 10 shows the further evolution of the bNb (200) diffraction line ver-
sus fast neutron fluence of M5A, after two cycles and five cycles.
On irradiated TEM samples, the nanometer-size particles can only be observed
on the prism-plane foil orientation with the (0002) diffracting vector (and s 0) in
order to annihilate the contrast coming from hai-defects. In these diffracting condi-
tions they appear as sharp, dark needles, and for this reason they are commonly
called needle-like particles. These nanometer-size particles are illustrated in Fig. 11.
If the hai-defects were annealed, we could observe them in a near basal-plane orien-
tation and see that the “needles” were platelets and were seen on the edge side with
the (0002) diffracting conditions (Fig. 12). From the T-XRD point of view, the bcc
structure of the radiation-enhanced bNb particles observed with our experimental
geometry and conditions unfortunately did not allow us to suggest a model of the
particle’s shape.
FIG. 9 Comparison of the whole pattern profile matching decomposition of the bNb
(200) diffraction line of M5A: (a) in non-irradiated sample; (b) in irradiated
sample (after one cycle).
FIG. 10 Whole pattern profile matching decomposition of the bNb (200) diffraction line
versus fast neutron fluence of M5A: (a) after two cycles; (b) after five cycles.
FIG. 11 Micrographs of radiation-enhanced needle-like particles in an annealed sample

with the 0002 diffracting vector in M5A alloy after six cycles of PWR irradiation.
FIG. 12 Micrographs of radiation-enhanced particles in an annealed sample with a near

basal-plane orientation in M5A alloy after six cycles of PWR irradiation.
Lattice Parameter and Nb Content

The lattice parameter and the Nb content of such small particles measured via T-
XRD, which are unattainable with ATEM analyses on thin foils, are summarized in
Table 8 and Fig. 13. T-XRD measurements show that the Nb content deduced from
the lattice parameter of the radiation-enhanced bNb particles appears to be about
60 % for up to two PWR irradiation cycles (Fig. 13). But the Nb content determined
in needle-like particles in the sample irradiated for five PWR cycles seems greater
TABLE 8 Evolution of radiation-enhanced bNb particles versus fast neutron fluence for M5A irradi-
ated in PWR 1, from T-XRD analyses (uncertainty determined from line profile fitting
errors).
Mean Size of the

Number of Lattice Nb Content, Integral Breadth, Coherent Diffracting
Cycles Parameter, Å wt. % rad Domains, nm
1 3.4059 (60.0096) 63 (63) 0.010829 7 (62)

(60.003497)
2 3.4150 (60.0031) 60 (61) 0.006722 11 (62)
(60.001113)
5 3.3728 (60.0029) 75 (61) 0.008073 9 (61)
(60.001185)
than the contents in samples irradiated for one or two cycles and for the native
bNb particles for the same dose. This point is discussed in the last paragraph and
needs to be confirmed with measurements at higher fluence.
Size
Figure 14 shows that the Nb-enriched particles look almost circular (/ 3 nm) for
low doses, typically of 2 1025 n/m2 (one cycle). At higher fluence these particles
become elongated in the direction close to the basal plane trace with prism-plane
foil orientation. This shape is typically observed in these radiation-enhanced par-
ticles [12,24]. It could be explained by the attempt to minimize the stress field due
to matrix/precipitate misfit and by the anisotropy in diffusivity of irradiation-
induced point defects [28].
FIG. 13 Nb content (wt. %) versus fast neutron fluence of radiation-enhanced bNb

deduced from the lattice parameter for M5A alloy.
FIG. 14 Size of the needle-like particles in M5 and in Zr1%NbOB versus fast neutron
fluence.
A measurement of the width and length of these particles is plotted in Fig. 14.
Both parameters increased rapidly during the first irradiation cycle (2 1025 n/m2)
to reach saturation for higher fluence. The average length was about 3.5 nm after
the first cycle and saturated at about 7 nm after four or five irradiation cycles. The
average width saturated more quickly at a value of about 3.5 nm. This particle size
enhancement during irradiation was previously observed by SANS [6]. No differ-
ence was noticed for either parameter before and after creep tests.
Figure 15 shows the width and length distribution histograms of the needle-like
particles based on a huge amount of particles examined (between 300 and 600).
Both distributions are very narrow for the first irradiation cycle and seem to indi-
cate that we are still in the nucleation process or at the end of this process. The
rapid broadening of the histogram combined with a displacement toward higher
values between one and four annual cycles implies that the nucleation state is over
and has been replaced by the growth one. From four to seven annual cycles, the
FIG. 15 Histograms of width and length of radiation-enhanced needle-like particles of M5A

material: (a), (b) one cycle (2 1025 n/m2); (c), (d) four cycles (8.7 1025 n/m2);
(e), (f) seven cycles (14.6 1025 n/m2).
growth continues, but slowly. This fully agrees with Figs. 14 and 16, showing a rapid
growth of the needles between one and four cycles while the number density
remains at about 1.5 1025 n/m2.
Considering T-XRD analyses, the mean size of the coherent diffracting
domains is about 10 6 2 nm and remains constant with increasing fluence (Table 8).
If we consider that the size of the coherent diffracting domains can be related to the
size of the studied particles, the size of the needle particles is similar for both TEM
FIG. 16 Number density of needle-like precipitation in M5 and in Zr1%NbOB versus fast

neutron fluence as determined by TEM studies.
and T-XRD techniques. But T-XRD does not allow us to see the particle-shape evo-
lution during irradiation (from spherical to needle-like particles as the irradiation
dose increases).
Number Density
The evolution of the particle number density versus fast neutron fluence is shown
in Fig. 16. Despite the scattering of the data (probably due to thin-foil thickness
measurement uncertainty), it seems clear that the particle density is constant and
equal to 1.5 1022 m3. The particle density does not seem to be affected by creep
tests.
Volume Fraction
The volume fraction of radiation-enhanced particles was calculated from TEM
measurements with the hypothesis that these particles are disk-shaped (for one
second-phase particle, v ¼ pL2W/4).11 The evolution of the volume fraction is plot-
ted in Fig. 17 for as-irradiated samples. It appears that the volume fraction of the
needles grew rapidly between zero and four annual cycles (8 1025 n/m2) and then
saturated at a value of about 0.2 %.12 This result agrees with previous SANS meas-
urements [6] showing the same relative evolution of the volume fraction between
one and three cycles, even if the absolute values were higher for SANS than for
TEM. SANS was obviously giving an overestimated value because of the critical and
numerous deconvolutions [6]. Because of crystallographic texture effects, even for
11
W ¼ width, L ¼ length, v ¼ volume.
12
In order to calculate the volume fraction of radiation-enhanced particles, we assimilate them into discs.
Had we taken them as ellipsoids, we would have obtained a doubled volume fraction. For this reason, the
volume fraction of 0.2 % observed after four PWR cycles must be considered as a rough estimate and not
as an absolute value.
FIG. 17 Volume fraction of needle-like precipitation in M5 and in Zr1%NbOB versus fast

neutron fluence as determined by TEM studies.
bNb precipitates (especially radiation-enhanced needle-like particles), the volume

fraction of the different phases was not quantified from the XRD line.
Spatial Distribution
As we showed previously [6] for M5A cladding tubes irradiated in PWR 1, the
radiation-enhanced particles are randomly distributed after the first cycle of irradia-
tion. For higher fluence, they gradually tend to be located within about 50-nm
spaced layers parallel to the basal plane. We can see this phenomenon in Figs. 18(a)
and 18(b). It is not observed for free-growth test tubes, as seen in Fig. 18(c).
The density of needles is not observed to be higher in the vicinity of native bNb
particles. This observation is consistent with the lack of significant Nb redissolution
out of the native particles (the Nb diffusion rate in Zr is very low [11]).
hAi-LOOPS
A gray contrast due to the hai-defect contribution (black dots and hai-loops) usu-
ally appears on TEM micrographs of irradiated Zr alloys. In high-magnification
micrographs, circular defects corresponding to irradiation-induced hai-loops are
observed (as pointed out in Fig. 19).
The number density and diameter of hai-loops grow very quickly during the
first irradiation cycle. Then we observe a stabilization of their density and average
diameter with the fluence (Fig. 20). The histograms in Fig. 21 show the same tend-
ency with a very similar distribution of hai-loops after four and six cycles and a sim-
ilar density of smaller loops after one cycle. From 7 1025 n/m2 to 20 1025 n/m2,
the density of hai-loops is about 2 1022 m3 with an average diameter of about
10 nm. A rough estimation of the dislocation density due to hai-type loops is
6 1014 m/m3. There seems to be a slight influence of the span level of irradiation
(temperature of irradiation), with a higher density of smaller loops at lower
FIG. 18 Radiation-enhanced particles in M5A after irradiation in PWR 1: (a) one cycle
(2 1025 n/m2); (b) six cycles (13 1025 n/m2); (c) in M5G after four cycles of
irradiation in PWR 2 (13.4 1025 n/m2).
FIG. 19 hai-loops in M5A after seven cycles of irradiation (14.6 1025 n/m2).
irradiation temperatures (lower span level), as seen in Fig. 22. After creep tests there
was an increase in the hai-loops’ diameter and a decrease in their density (Table 9).
BASAL IRRADIATION-INDUCED hCi-COMPONENT LOOPS

TEM exams showed that there were no hci-component loops in M5A after one irra-
diation cycle in PWR 1 (2 1025 n/m2), and very few after two PWR irradiation
cycles (3.5 1025 n/m2). Only a few heterogeneously dispersed hci-component
loops were observed after four and six cycles (7 1025 to 13.5 1025 n/m2). For
highly irradiated Zr1 %NbOB in PWR 6 (19.5 1025 n/m2), hci-component loops
appeared more numerous and less heterogeneously dispersed [Fig. 23(d)] but were
obviously much less numerous than in RXA Zy-4 irradiated for six cycles in PWR 2
(21 1025 n/m2), as shown in Fig. 24(b). For fast fluence as low as 9.5 1025 n/m2
(five cycles in PWR 7), there are numerous and uniformly distributed hci-
component loops in RXA Zy-4 that can be compared to the rare hci-component
loops observed in M5A irradiated for four cycles in PWR 1 [Figs. 23(b) and 24(a)].
Contrary to what is observed in RXA Zy-4 [31], basal hci-component loops are
not more numerous in the vicinity of precipitates in M5 alloy.
13
Figure 25 shows the evolution of hci-loop linear density versus fast neutron
fluence in M5 alloy, Zr1 %NbOA, Zr1 %NbOB, and RXA Zy-4 for irradiation doses
13
In Fig. 25, the foil thickness is assumed to be 150 nm for each micrograph, except for fluences of
2.6 1025 n/m2 and 11 1025 n/m2 (for these two samples, the thickness was measured via EELS).
FIG. 20 Evolution of hai-loop distribution in M5 and in Zr1%NbOB versus fast neutron

fluence: (a) number density; (b) average diameter (data from TEM studies).
up to 21 1025 n/m2. For M5 alloy, hci-component loops appear at a threshold

value of 3 1025 n/m2, and then their linear density increases slowly versus the fast
neutron fluence. For RXA Zy-4 alloy, after irradiation in PWR conditions, basal
hci-component loops appear for a dose threshold value quoted by many authors
[32,33]. The irradiation doses investigated here for RXA Zy-4 alloys were not low
enough to confirm this dose threshold. We observed a ratio of 5 between RXA Zy-4
and M5 alloy for these basal defects.
Remark
Thanks to the needle-like particle number density (1.5 1022 m3), we can
“measure” or check the sample thickness in an original way, by counting radiation
FIG. 21 Comparison of the hai-loops diameter histograms in M5A for one cycle
(2 1025 n/m2), four cycles (8.7 1025 n/m2), and six cycles (12.4 1025 n/m2) in
PWR 1.
needle-like particles and by adjusting the thickness in order to obtain the right
number density. In that way we can obtain a good estimate of the hci-component
loop linear density without measuring the thickness of the sample (Fig. 26). Figure
26(a) is a part of Fig. 25, including M5 cladding tubes irradiated in PWR 1 and with
a thickness assumed to be 150 nm for each micrograph. Figure 26(b) was obtained
from Fig. 26(a) by carrying out a new calculation of the linear densities of hci-
component loops on the same micrographs, with the thickness adjusted to obtain
the expected density of needles.
FIG. 22 Comparison of the hai-loop diameter histograms in M5A irradiated for six cycles
(13 1025 n/m2) on the second and the sixth span levels.
TABLE 9 Evolution of hai-loop distribution versus fast neutron fluence in M5A, from TEM analyses.
Fluence (E > 1 MeV), /m, nm /max, nm Number Density,

Number of Cycles 1025 n/m2 1022 m3
7 14.6 9.7 27 1.8

6 13.6 8.8 21 1.8
6 (creep test, 13.6 — — Very few
400 C, 250 MPa,
380 h)
6 (creep test, 13.5 9.2 27 1.3
350 C, 300 MPa,
500 h)
6 12.4 9.8 24 1.1
5 (creep test, 11 16 36 0.4
400 C, 130 MPa,
240 h)
4 (creep test, 9.3 12 33 0.5
350 C, 300 MPa,
500 h)
4 (creep test, 9 30 87 0.05
400 C, 250 MPa,
83 h)
4 8.7 8.9 33 2.2
4 7 9.6 27 2
4 (creep test, 7.2 12.5 30 0.4
350 C, 450 MPa,
8 h)
1 2 7.5 21 2.4
1 (creep test, 2 19 42 0.3
400 C, 130 MPa,
240 h)
1 (creep test, 2 8.8 37 1.5
350 C, 300 MPa,
500 h)
0 0 0 0 0
Discussion
GENERAL COMMENT ON IRRADIATED M5 ALLOY
Based on our results for M5 alloy, we can say that concerning hai-loops, native bNb
particles, irradiation-enhanced particles, and Laves phase particles (and conse-
quently niobium and iron contents in the matrix), everything seems to take place
before 8 1025 n/m2 (four PWR annual cycles). Except for radiation-induced hci-
component loops, no further evolution seems to occur after a dose of 8 1025 n/m2
(four PWR cycles) in this material.
FIG. 23 Micrographs of hci-component loops in M5A irradiated in PWR 1: (a) one cycle
(2 1025 n/m2); (b) four cycles (8.7 1025 n/m2); (c) six cycles (12.4 1025 n/m2);
and (d) in Zr1%NbOB irradiated in PWR 6 (19.5 1025 n/m2).
The post-irradiation creep tests affect mostly the hai-loop defects. After creep
tests, the enlargement of the hai-loops’ diameter and their decrease in density are
due to thermal annealing and dislocation-glide sweeping [34]. The decrease in
number density seems to be due to both thermal annealing and dislocation-gliding,
FIG. 24 Micrographs of hci-component loops in Zy-4 irradiated for (a) five cycles in
PWR 7 (9.5 1025 n/m2) and (b) six cycles in PWR 2 (21 1025 n/m2).
FIG. 25 Evolution of hci-loop linear density versus fast neutron fluence for M5,
Zr1%NbOA, Zr1%NbOB, and RXA Zy-4 alloys by TEM.
whereas the diameter enlargement seems to be mostly a consequence of thermal

annealing—there is almost no diameter enlargement after 350 C creep tests. An
increase in test temperature, creep stress, or test time increases this phenomenon.
Based on the data in Table 9, creep tests at either a stress level higher than 300 MPa
or a temperature higher than 350 C are needed in order for one to observe a signifi-
cant effect of creep tests on the hai-loop distribution.
hCi-COMPONENT LOOP EVOLUTION VERSUS FAST NEUTRON FLUENCE

Several authors associate the growth breakaway phenomenon with the hci-
component loop occurrence [4–6,21,31]. The correlation was clearly established by
these authors for RXA Zy-4 alloys [4] with a growth breakaway dose of about
10 1025 n/m2 in PWR conditions (E > 1 MeV) [6]. For this fast neutron fluence,
the measured hci-component loop linear density for RXA Zy-4 in this study was as
high as 10 1013 m/m3 (Fig. 25). In a very similar way, an increase in the hci-type
dislocation density of more than 1 1014 m/m3 was observed by Shishov in Nb-
containing alloys [12] when they underwent significant irradiation growth.
For M5 alloy cladding tubes in PWR conditions, no growth breakaway was
seen for doses up to 14 1025 n/m2 (E > 1 MeV), despite the observation of hci-
component loops before this dose [6,35]. This absence of growth breakaway was
attributed to the very low content of hci-component loops in this material, but no
relevant measurements were made at that time [6]. Our additional data confirm
that for a fluence of 14 1025 n/m2, the linear density of hci-component loops
(about 3 1013 m/m3) is far from the linear density level of 10 1013 m/m3 corre-
sponding to the growth breakaway regime in Zy-4. In addition, we can notice that
for fluences up to 17.1 1025 n/m2, the linear density of hci-loops in M5 alloy clad-
ding tubes slightly increases but remains about five times lower than that of
FIG. 26 hci-loop linear density with thickness (a) assumed to be 150 nm and (b)
measured thanks to “needle” density.
recrystallized Zy-4, and that the free growth value of M5 alloy is lower than that of
RXA Zy-4 in the same proportion [31].
Concerning the influence of precipitation on hci-component loops, if in Zy-4
alloys these basal defects appear clearly correlated to iron redistribution into the
matrix [4,36,37], the correlation is not obvious in M5 alloy. Even if there is less iron
in M5 alloy and fewer hci-component loops than in RXA Zy-4, hci-component
loops are not more numerous in the vicinity of prior Laves-phase precipitates in
M5 alloy, as in other Zr–Nb-based alloys [7–9,12], and the hci-loops’ linear density
keeps increasing even after all the iron is rejected from the precipitates. In contrast,
in highly irradiated (Zr–Nb–Sn–Fe)-type alloys (E635), hci-component loops are
more numerous in the vicinity of prior Laves-phase precipitates [7–9,12]. We can
also underline that although the iron content of M5, Zr1 %NbOA, and Zr1 %NbOB
alloys varies from about 100 to about 650 ppm, it seems to have no significant effect
on the linear density of hci-component loops in that range of composition.
POTENTIAL EFFECT OF STRESS CONDITIONS ON RADIATION-ENHANCED

PRECIPITATION
Concerning the radiation-enhanced precipitation spatial distribution, we can com-
pare the M5A cladding tubes (PWR 1) with the M5G free-growth test tubes (PWR
2). After one irradiation cycle, the needle-like particles are randomly distributed in
the cladding tubes. As early as the second cycle of irradiation, they gradually start
to localize within layers parallel to the basal plane in the cladding tubes, while they
are randomly dispersed in the free-growth test tubes at all fluence levels observed in
this study.
This concentration of needle-like particles within parallel layers was observed
not only in M5A cladding tubes irradiated in PWR 1, but also in other Zr-Nb clad-
ding tubes (Zr1 %NbOA in PWR 4, for instance [6]). Other authors [8,12] observed,
as we did, alignments of needle-like irradiation-enhanced particles in rows parallel
to (0002) planes. For these authors, the alignments seemed particularly noticeable
on tubes irradiated under an internal gas pressure (creep test under irradiation),
and for this reason this particular distribution of particles observed by Shishov et al.
[8] can be related to the stress conditions undergone by the test tube during irradia-
tion. In M5A cladding tubes irradiated in PWR 1, the contact between the fuel pellet
and the claddings occurs after about two PWR irradiation cycles [6]. We can also
infer a relationship between this distribution and the stress/deformation state
imposed by the fuel pellet on the cladding tubes after two PWR cycles.
This alignment can be compared to the well-known “rafting” phenomenon
observed in Ni super alloys [38–41] and on Tungsten after neutron irradiation [42].
According to various authors, the rafting phenomenon is assumed to result from
the preferential drift of free interstitials to dislocations. For Ni super alloys, the
morphology of the rafts depends on several parameters such as the direction of the
stress and the sign of the misfit between the matrix and the precipitates [38,39].
Shishov [12] determined that not only stresses in materials, but also an alloying
element such as iron can lead to this preferential distribution of radiation-enhanced
precipitation. We did not notice this aspect because we were not working with the
same range of iron content.
NB IN NATIVE bNB AND IN RADIATION-ENHANCED PRECIPITATION

For the non-irradiated sample, the lattice parameter (0.332 nm) and Nb content
(91 wt. %) of native bNb particles measured via T-XRD completely agreed with
expected values and with data from other authors [8]. In this way T-XRD appears
to be a very accurate way to measure the right composition of native bNb particles
after irradiation, whereas ATEM measurements could be doubtful because of the
matrix contribution and activation spectrum. Considering that with T-XRD we
measured about 55 wt. % Nb for the sample irradiated for five PWR cycles, we can
assume that the “equilibrium” Nb content of bNb in irradiation conditions is very

close to the 55 % value estimated via ATEM on thin foils in this study and in previ-
ous works on extraction replicas [7,8]. The comparison between ATEM on extrac-
tion replicas and T-XRD underlines the good accuracy of T-XRD analyses. This
technique enables us to confirm and be accurate about the Nb content in bNb irra-
diated particles with a weight fraction less than 1 wt. %, without extraction replicas,
despite the matrix contribution and activation.
T-XRD analyses are less accurate for needle-like particles because of their small
volume fraction and their nanometric size, which made the lines broader. When we
compare the data plotted in Figs. 13 and 4, it appears that for irradiation doses for
up to two PWR irradiation cycles, these needle particles formed under irradiation
with a Nb content (and a lattice parameter) corresponding to the “equilibrium”
value of native bNb particles under irradiation conditions. According to previous
studies [8] with TEM observations on extraction replica, this “equilibrium value” is
supposed to be between 55 % and 60 %. This observation agrees well with what was
expected and confirms that these particles are bNb type [6–8,31]. Such information
cannot be obtained via ATEM on thin foils. Undoubtedly, diffraction analyses using
synchrotron radiation sources are very powerful techniques for studying the struc-
ture and identifying crystalline compounds forming secondary phases with small
volume fractions (about 0.2 % to 0.4 %) in zirconium alloys before and after neu-
tron irradiation.
But for five PWR cycles, the conclusions are less clear. For this fluence level,
the Nb content determined in needle-like particles seems higher than for one and
two cycles and higher than in the native bNb particles for the same dose, whereas
we had expected to have no change in the needle-like composition versus fluence
and a convergence in the composition of the two kinds of particles toward the equi-
librium value of 55 % to 60 %. This can be explained by the valuation we made for
the line profile deconvolution: we assumed there were only the contributions of
needle-like and native particles, with both of them showing a specific shape and in-
tensity. If we take into account the oxide contribution, which is easy to observe
before irradiation [see Fig. 9(a)] and no more observable after irradiation (convo-
luted with Nb peaks thanks to cell parameter evolution after irradiation), the Nb
content measured via T-XRD in both radiation-induced and native particles is in
the range 55 % to 60 % for the sample irradiated for five PWR cycles, as expected.
But this analysis needs to be confirmed with further studies at higher doses (e.g., six
and seven PWR cycles) in order to show whether the needle-like particle composi-
tion is stable from two PWR cycles.
This study gave accurate radiation-enhanced particle density (with EELS thick-
ness measurements), particle size (with higher TEM magnification), and particle
shape (with annealed samples). Firstly we can underline that the iron content of
M5 and Zr1 %NbOB alloys (in the range of 100 to 650 ppm) seems to have no no-
ticeable influence on needle-like particle size and density. Then, together with the
accuracy of the T-XRD Nb content analysis, we were able to plot the total niobium
FIG. 27 Total Nb taken out of the matrix in M5 and in Zr1%NbOB during irradiation by
irradiation-enhanced particles.
taken out of the matrix by the needles (Fig. 27). If we consider that there is no Nb
redistribution into the matrix by the native particles during irradiation, as was said
previously [6,8], it appears (Fig. 27) that the niobium content of the matrix
decreases by more than 0.1 % between one and four PWR cycles (8 1025 n/m2)
and stays at the same level for a longer irradiation time. These conclusions agree
with those of previous authors who have measured less Nb in the matrix after irra-
diation than before [7,8]. For a non-irradiated Zr1 %Nb alloy, such a decrease in
the niobium content in the matrix is known to influence the thermal creep proper-
ties [13]. Fortunately, it seems that this decrease in Nb content does not go further
during a creep test: the Nb content in precipitates does not seem higher after creep
tests than before in irradiated material.
TRANSFORMATION OF LAVES PHASES INTO bNB PARTICLES

Thanks to previous studies [7,8,21], a hypothesis could be expressed to explain the
vanishing of the Laves phases in the diffuse background of the T-XRD pattern as
early as the first PWR irradiation cycle: the Laves phases lose their iron very quickly
during irradiation and become Zr–Nb microcrystallized highly faulted particles
similar to those observed in Fig. 8, and then they transform into bNb particles. The
vanishing of the T-XRD line corresponding to Laves phases is consistent with the
transformation of the Laves phases (their Fe content decreases as a function of irra-
diation) into microcrystallized bNb particles described in Zr–Sn–Nb–Fe and in
Zr–1Nb(Fe) alloys by Shishov et al. [7–12]. For these authors, the transformation of
Laves phases in bNb particles occurs at a dose of about 10 dpa [12], corresponding
to about three PWR cycles, whereas the vanishing of the Laves phases in the diffuse
background of the T-XRD pattern is observed as early as 3 dpa. This could be
explained by considering that for such very scarce particles, the broadening of the
line due to the microcrystallization is probably sufficient to make it disappear as

early as 3 dpa. Shishov underlined that this microcrystallization corroborates the
important role played by the Nb and Fe combination in the resistance of phases
under irradiation [12].
In this way, prior Laves phases could add a third contribution to the XRD pro-
file of the bNb phase, along with prior bNb phase and irradiation-enhanced par-
ticles. We can ask ourselves whether they contribute to the prior bNb or to the
radiation-enhanced particle profile. In both cases, their contribution should be min-
imal because of the very low quantity of these particles.
SIZE OF THE COHERENT DIFFRACTING DOMAINS IN bNB

In XRD measurements, the coherent diffracting domains are usually correlated
with the particle size.
For the non-irradiated sample, the size of the coherent diffracting domains
measured via T-XRD and the size of the bNb particles observed in TEM studies are
almost the same. This is probably because bNb particles have no twins or stacking
faults (as expected in bcc second-phase particles).
After irradiation, the mean size of the coherent diffracting domains of the
native bNb particles drops from 33 6 2 nm to 11 6 1 nm, while TEM analyses mea-
sure an average diameter of about 50 nm for these particles after irradiation (and
35 nm before irradiation). It seems that the contribution of microcrystallized prior
Laves phases is not enough to explain why the size of the coherent diffracting
domains, as measured via T-XRD, is much smaller than the real size of the particles.
This phenomenon seems more likely to be due to irradiation defects and is consist-
ent with the granitic aspect of the precipitates observed in TEM. Moreover, we have
to keep in mind that coherent diffracting domains are obtained with XRD analyses
from a very simple model (Scherrer) neglecting microstrains and defaults, which is
strongly questionable, especially for irradiated samples.
Conclusion
The present study focused on the microstructural evolution of irradiated M5 clad-
dings. Analytical TEM examinations on second-phase particles were completed and
confirmed by original T-XRD analyses. T-XRD measurement allowed us to confirm
and be more precise about the Nb content in native bNb irradiated particles.
Thanks to these analyses, we were able to measure for the first time the composition
and the lattice parameters of radiation-enhanced nanometric needle-like precipi-
tates. Also in this study, accurate evolutions of microstructural features such as hci-
component loops, hai-loops, and native and radiation-enhanced particles have been
plotted as a function of irradiation doses up to 20 1025 n/m2. The precipitate
study permitted us to propose an evaluation of the niobium content in the matrix
during irradiation. All these results confirm the noteworthy microstructural and
microchemical stability of M5 and Zr1 %NbO alloys during irradiation. The native
bNb particles remain fully crystalline even after irradiation up to 19.5 1025 n/m2.
After a dose of 11 1025 n/m2, native particles reach the equilibrium composition
of 55 wt. % Nb under irradiation conditions. The irradiation-enhanced needle-like
precipitation leads to a noticeable decrease in Nb content in the matrix. But no fur-
ther evolution in size, number density, or niobium content seems to occur for these
particles after a dose of 8 1025 n/m2 or after creep tests. Finally, the linear density
of hai-component loops observed in this work for M5 cladding tubes remains mod-
erate relative to that in RXA Zy-4 alloy mainly for the highest dose of irradiation.
We can also underline that the iron content in M5, Zr1 %NbOA, and Zr1 %NbOB
alloys does not seem to influence either hai-component loop linear density or
needle-like particle size or density, in the range of 100 to 650 ppm, although Laves
phases seem to lose their iron very quickly and transform into microcrystallized
bNb particles, as also shown by previous works [7].
ACKNOWLEDGMENTS
The writers thank B. Sitaud and S. Schlutig for their help at the MARS beamline of the
French Synchrotron Radiation source SOLEIL. The writers are grateful to the staff
(CEA-DEN-DMN) for their assistance in the area of TEM examination: Th. Vanden-
berghe and B. Arnal. M5 is a trademark of AREVA NP registered in the United States
and other countries.
References
[1] Mardon, J. P., Garner, G., Beslu, P., and Charquet, D., “Update on the Development of
Advanced Zirconium Alloys for PWR Fuel Rod Claddings,” Proceedings of the 1997 Inter-
national Topical Meeting on LWR Fuel Performance, Portland, OR, March 2–6, 1997, pp.
405–412.
[2] Mardon, J. P., Charquet, D., and Senevat, J., “Optimization of PWR Behaviour of Stress-
Relieved Zircaloy-4 Cladding Tubes by Upgrading the Manufacturing and Inspection
Process,” Zirconium in the Nuclear Industry: Tenth International Symposium, ASTM STP
1245, ASTM International, West Conshohocken, PA, 1994, pp. 329–348.
[3] Ravier, G., et al., “Framatome and FCF Recent Operating Experience and Advanced
Features to Increase Performance Reliability,” Proceedings of the 1997 International
Topical Meeting on LWR Fuel Performance, Portland, OR, March 2–6, 1997, pp.
31–38.
[4] Gilbon, D. and Simonot, C., “Effect of Irradiation on the Microstructure of Zircaloy-4,”
Zirconium in the Nuclear Industry: Tenth International Symposium, ASTM STP 1245,
[5] Gilbon, D., Soniak, A., Doriot, S., and Mardon, J. P., “Irradiation Creep and Growth Behav-
iour, and Microstructural Evolution of Advanced Zr-Base Alloys,” Zirconium in the Nu-
clear Industry: Twelfth International Symposium, ASTM STP 1354, ASTM International,
[6] Doriot, S., Gilbon, D., Béchade, J. L., Mathon, M. H., Legras, L., and Mardon J. P.,
“Microstructural Stability of M5 Alloy Irradiated up to High Neutron Fluences,” Zirconium
in the Nuclear Industry: 14th International Symposium, ASTM STP 1467, ASTM Interna-
[7] Shishov, V. N., Peregud, M. M., Nikulina, A. V., Pimenov, Yu. V., Kobylyansky, G. P., Novose-
lov, A. E., Ostrovsky, Z. E., and Obukhov, A. V., “Influence of Structure-Phase State of Nb
Containing Zr Alloys on Irradiation-induced Growth,” Zirconium in the Nuclear Industry:
PA, 2005, pp. 666–685.
[8] Shishov, V. N., Peregud, M. M., Nikulina, A. V., Kon’kov, V. F., Novikov, V. V., Markelov, V. A.,
Khokhunova, T. N., Kobylyansky, G. P., Novoselov, A. E., Ostrovky, Z. E., and Obukhov, A.
V., “Structure-Phase State, Corrosion and Irradiation Properties of Zr-Nb-Fe-Sn System
Alloys,” Zirconium in the Nuclear Industry: 15th International Symposium, ASTM STP
[9] Nikulina, A. V., Shishov, V. N., Peregud, M. M., Tselischev, A. V., Shamardin, V. K., and
Kobylyansky, G. P., “Irradiation Induced Growth and Microstructure Evolution of Zr-1.2
Sn-1 Nb-0.4 Fe Under Neutron Irradiation to High Doses,” Effect of Radiation on Materi-
als: 18th International Symposium, ASTM STP 1325, ASTM International, West Consho-
hocken, PA, 1999, pp. 1045–1061.
[10] Shishov, V. N., Peregud, M. M., Nikulina, A. V., Sheballdov, P. V., Tselischev, A. V., Koby-
lyansky, G. P., Ostrovsky, Z. E., and Shamardin, V. K., “Influence of Zirconium Alloy Chem-
ical Composition on Microstructure Formation and Irradiation Induced Growth,”
Zirconium in the Nuclear Industry: 13th International Symposium, ASTM STP 1423, ASTM
[11] Averin, S. A., Panchenko, V. L., Kozlov, A. V., Sinelnikov, L. P., Shishov, V. N., and Nikulina,
A. V., “Evolution of Dislocation and Precipitate Structure in Zr Alloys Under Long-Term
Irradiation,” Zirconium in the Nuclear Industry: Twelfth International Symposium, ASTM
STP 1354, ASTM International, West Conshohocken, PA, 2000, pp. 105–121.
[12] Shishov, V. N., “The Evolution of Microstructure and Deformation Stability in Zr-Nb-(Sn,
Fe) Alloys Under Neutron Irradiation,” Zirconium in the Nuclear Industry: Sixteenth Inter-
national Symposium, ASTM STP 1529, ASTM International, West Coonshohocken, PA,
2012, pp. 37–66.
[13] Brenner, R., Béchade, J. L., Castelnau, O., and Bacroix, B., “Thermal Creep of Zr-Nb1 %-O
Alloys: Experimental Analysis and Micromechanical Modeling,” J. Nucl. Mater., Vol. 305,
Nos. 2–3, 2002, pp. 175–186.
[14] Mardon, J. P., Charquet, D., and Senevat, J., “Influence of Composition and Process on
Out-of-pile and In-pile Properties of M5 Alloy,” Zirconium in the Nuclear Industry:
Twelfth International Symposium, ASTM STP 1354, ASTM International, West Consho-
hocken, PA, 2000, pp. 505–524.
[15] Sitaud, B., Solaris, S., Schlutig, P. L., Llorenz, I., and Hermange, H., “Characterization of
Radioactive Material Using the MARS Beamline at the Synchrotron SOLEIL,” J. Nucl.
Mater., Vol. 425, 2012, pp. 238–243.
[16] Solari, P. L., Schlutig, S., Hermange, H., and Sitaud, B., “MARS, a New Beamline for Radio-
active Matter Studies at SOLEIL,” J. Phys.: Conf. Ser., Vol. 190, No. 1, 2009, pp. 012–042.
[17] Schlutig, S., Solari, P. L., Hermange, H., and Sitaud, B., “MARS, a New Facility for X-ray
Diffraction and X-ray Absorption for Radioactive Matter Studies,” Mater. Res. Soc. Symp.
Proc., Vol. 1264, 2010, pp. 131–136.
[18] Rodriguez-Carjaval, J., “FULLPROF : A Program for Rietveld Refinement and Pattern
Matching Analysis,” Abstracts of the Satellite Meeting on Powder Diffraction of the XVth
IUCR Congress, Toulouse, France, 1990, Vol. 127.
[19] Scherrer, P., “Bestimmung der Grösse und der inneren Struktur von Kolloiteilchen mittels
Röntgenstrahlen,” Nachrichten von der Gesellschaft der Wissenshaften zu Gottingen,
Mathematisch Physikalische, Vol. 1–2, 1918, p. 96–100.
[20] Cuello, G. J., Fernandez Guillermet, A., Grad, G. B., Mayer, R. E., and Granada, J. R.,
“Structural-Properties and Stability of the BCC-Phases and Omega-Phases in the Zr-Nb
System.1.Neutron-Diffraction Study of a Quenched and Aged Zr-10 WT Percent Nb
Alloy,” J. Nucl. Mater., Vol. 218, 1995, p. 236.
[21] Shishov, V. N., Nikulina, A. V., Markelov, V. A., Peregud, M. M., Kozlov, A. V., Averin,
S. A., Kolbenkov, S. A., and Novoselov, A. E., “Influence of Neutron Irradiation and
Dislocation Structure and Phase Composition of Zr-Base Alloys,” Eleventh Interna-
tional Symposium, ASTM STP 1295, ASTM International, West Conshohocken, PA,
1996, pp. 603–622.
[22] Griffiths, M., Mecke, J. F., and Winegar, J. E., “Evolution of Microstructure in Zirconium Alloys
During Irradiation,” Zirconium in the Nuclear Industry: Eleventh International Symposium,
[23] Parker, J. D., Perovic, V., Leger, M., and Fleck, R. G., “Microstructural Effects on the Irradiation
Growth of Zr-2,5Nb,” Zirconium in the Nuclear Industry: Seventh International Symposium,
[24] Coleman, C. E., Gilbert, R. W., Carpenter, G. J. C., and Weatherly, G. C., “Precipitation in
Zr-2.5 wt % Nb During Neutron Irradiation,” Phase Stability During Irradiation Sympo-
sium, Pittsburgh, PA, Oct 5–9, 1980, The Metallurgical Society of AIME, Englewood, CO,
pp. 587–599.
[25] Perovic, V., Perovic, A., Weatherly, G. C., Brown, L. M., Purdy, G. R., Fleck, R. G., and Holt,
R. A., “Microstructural and Microchemical Studies of Zr-2.5Nb Pressure Tube Alloy,” J.
Nucl. Mater., Vol. 205, 1993, pp. 251–257.
[26] Brebec, G., “Diffusion and Precipitation Under Irradiation,” Diffusion in Materials, Nato
Advanced Science Institute Series, Series E, Applied Sciences, Vol. 179, 1990, pp.
339–336.
[27] Davies, P. H., Hosbons, R. R., Griffiths, M., and Chow, C. K., “Correlation between Irradi-
ated and Unirradiated Fracture Toughness of Zr-2.5Nb Pressure Tubes,” Zirconium in the
Nuclear Industry: Tenth International Symposium, ASTM STP 1245, ASTM International,
[28] Sarce, A., “Stability of Precipitates in the Anisotropic a-Zr Matrix Under Irradiation,” J.
Nucl. Mater., Vol. 185, No. 2, 1991, pp. 214–223.
[29] Turkin, A. A., Buts, A. V., and Bakai, A. S., “Construction of Radiation-Modified Phase Dia-
grams Under Cascade-Producing Irradiation,” J. Nucl. Mater., Vol. 305, Nos. 2–3, 2002,
pp. 134–152.
[30] Potter, D. I. and Wiedersich, H., “Mechanisms and Kinetics of Precipitate Restructuring
During Irradiation,” J. Nucl. Mater., Vol. 273, 1999, pp. 208–213.
[31] Bossis, P., Verhaeghe, B., Doriot, S., Gilbon, D., Chabretou, V., Dalmais, A., Mardon, J. P.,
Blat, M., and Miquet, A., “In PWR Comprehension Study of High Burn-up Corrosion and
Growth Behavior of M5 and Recrystallised Low-tin Zircaloy 4,” Zirconium in the Nuclear
Industry: Fifteenth International Symposium, ASTM STP 1505, ASTM International, West
[32] Garzarolli, F., Dewes, P., Maussner, G., and Basso, H. H., “Effects of High Neutron Fluence
on Microstructure and Growth of Zircaloy-4,” Zirconium in the Nuclear Industry: Eighth
International Symposium, ASTM STP 1023, ASTM International, West Conshohocken, PA,
1989, pp. 641–657.
[33] Griffiths, M. and Gilbert, R. W., “The Formation of c-component Defect in Zircaloys Dur-
ing Irradiation,” J. Nucl. Mater., Vol. 150, No. 2, 1987, pp. 169–181.
[34] Ribis, J., Bechade, J. L., Onimus, F., Doriot, S., Lemaignan, C., Cappelaere, C., and
Rabouille, O., “Experimental and Modelling Approach of Irradiation Defects Recovery in
Zirconium Alloys, Impact of an Applied Stress,” Zirconium in the Nuclear Industry: Fif-
teenth International Symposium, ASTM STP 1505, ASTM International, West Consho-
hocken, PA, 2009, pp. 674–695.
[35] Mardon, J. P., Garner, C. L., and Hoffmann, P. B., “M5 a Breakthrough in Zr Alloy,” Pro-
ceedings of the 2010 LWR Fuel Performance Meeting Top Fuel WRFPM, Sept 26–29,
2010 Orlando, FL.
[36] Griffiths, M., “A Review of Microstructure Evolution in Zirconium Alloys During Irradi-
[37] de Carlan, Y., Regnard, C., Griffiths, M., Gilbon, D., and Lemaignan, C., “Influence of Iron
in the Nucleation of the Component Dislocation Loops in Irradiated Zircaloy-4,” Zirco-
nium in the Nuclear Industry: Eleventh International Symposium, ASTM STP 1295, ASTM
[38] Zhou, L., Li, S. X., Chen, C. R., Wang, Y. C., Zang, Q. S., and Lu, K., “Three-Dimensional Fi-
nite Element Analysis of Stresses and Energy Density Distribution around g’ Before
Coarsening Loaded in the [110]-Direction in Ni-Based Superalloy,” Mater. Sci. Eng., A,
Vol. 352, Nos. 1–2, 2003, pp. 300–307.
[39] Murakamo, T., Kobayashi, T., Koizumi, Y., and Harada, H., “Creep Behaviour of Ni-base
Single-Crystal Superalloy With Various g’ Volume Fraction,” Acta Mater., Vol. 52, No. 12,
2004, pp. 3737–3744.
[40] Henderson, P., Berglin, L., and Jansson, C., “On Rafting in a Single Crystal Nickel-base
Superalloy After High and Low Temperature Creep,” Scr. Mater., Vol. 40, No. 2, 1998, pp.
229–234.
[41] Ichitsubo, T. and Tanaka, K., “Interpretation in Elastic Regime for Rafting of Ni-base
Superalloy Based on the External-Stress-Free Dimensional Change due to Internal-
Stress Equilibration,” Acta Mater., Vol. 53, No. 17, 2005, pp. 4497–4504.
[42] Sikka, V. K. and Moteff, J., “Rafting in Neutron Irradiated Tungsten,” J. Nucl. Mater., Vol.
46, 1973, pp. 207–219.
DISCUSSION
Questions from D. Srivastava, BARC Mumbai:—Can you comment why Nb
content of bNb decreases from 90 % Nb to 60 % Nb during irradiation? What is the
driving force for going from equilibrium bNb (90 % Nb) to non-equilibrium bNb
(60 % Nb) during irradiation?
Authors’ Response:—Ballistic exchanges are probably responsible for this

“equilibrium” Nb content under irradiation which is different from the thermody-
namic equilibrium. We suppose that Zr atoms go inside the native particles.
Questions from Jean-Christophe Brachet, CEA, France:—Comment: The fine

bNb precipitation is enhanced by irradiation. There is a thermodynamic driving
force for that, because the as-received conditions correspond to more or less
0.5 wt % Nb remaining in solid solution which corresponds to the equilibrium solu-
bility limit at 580 C. The irradiation is performed at 380 C and at this temper-
ature, the solubility limit is lower. From “CALPHAD” calculations using
“Thermocalc” þ CEA “Zircobase” thermodynamic database, at 350-400 C, the Nb
solubility is ranging somewhere between 0.2–0.4 wt % which is in quite good agree-
ment with the tendency observed here.
Authors’ Response (if any) :—
Questions from K. Kapoor, NFC:—Could you explain the origin of the needle
shaped particles? Comment: Your observation on needle shaped particles does not
match with earlier observations by Dr. Shishov.
Authors’ Response:—This shape is typically observed for these radiation-

enhanced particles [12,14]. It could be explained by the attempt to minimize the
stress field due to matrix/precipitate misfit [12] and by the anisotropy in diffusivity
of irradiation-induced point defects [14].
On irradiated TEM samples the nanometric size particles can only be observed
on prism-plane foil orientation with the (0002) diffracting vector (and s>>0) in
order to annihilate the contrast coming from <a>-defects. In these diffracting con-
ditions and for irradiation fluences higher than 2x1025n/m2, they appear as sharp
dark needles and for this reason are commonly called “needle-like particles”. If the
<a> defects were annealed we can observe them on near basal-plane orientation
and realize that the “needles” were platelets and were seen on the edge side with the
(0002) diffracting conditions.
Questions from Ron Adamson, Zircology Plus

Q1:—How did you measure the “density” of <c> loops, and how did you take
into account of the fact that many loops are larger than the foil thickness?
Authors’ Response:—It is not relevant to measure the density and the mean di-
ameter of <c>-loops if the size of the “box” we used to count them is smaller than
the <c>-loops diameter. It is why we used the “linear density”, which is the sum
up of the apparent diameter of <c>-loops intersecting the thin foil surfaces. These
measurements must be done in the same conditions for all the samples and with
similar thin foil thicknesses (about 150 nm) to be comparable.
Q2:—A number of years ago it was reported that irradiation growth of M5 at

medium fluence decreased due to a relative density effect of the irradiation pro-
duced b SPPs. Can this be confirmed?
Authors’ Response:—In 2000 we have shown a slight decrease of irradiation

growth in M5TM at medium fluence (Gilbon D. et al. ASTM STP 1354 pp. 51-73).
More recently we calculated the volume reduction of M5TM due to the precipitation
of bNb under irradiation. This volume reduction was in the same range that the
observed decrease in the growth at medium fluence.
Questions from V. Shishov, Bochvar VNIINM, Moscow, Russia:—Can you

explain the influence of "needle-like" precipitates on the material properties?
Authors’ Response:—In the case of this irradiated material the probable soften-
ing effect of the lowering of the matrix Nb content is likely balanced by a likely
hardening effect of such an intense nanometric precipitation. But the resulting
effect cannot be easily evidenced because of the large and dominating hardening
effect by <a>-loops in irradiated materials.
Questions from Javier Romero, Westinghouse Electric Co.
Q1:—Regarding the appearance of irradiation-enhanced b Nb particles, is it

gradual with fluence or does it accelerate at some point?
Authors’ Response:—There is no acceleration at some point: these particles

appear as early as 1 cycle (2x1025 n/m2), but their density remains constant all along
the irradiation time; their size grows slowly until it reaches a saturation at the flu-
ence of about 8x1025 n/m2 in PWR conditions.
Q2:—What is the driving force of precipitation of b Nb particles in-reactor, the

irradiation damage or temperature/time?
Authors’ Response:—The temperature of irradiation is lower than the tempera-

ture of final heat treatment. For this reason the solubility limit of niobium is lower
under irradiation than during elaboration. Irradiation accelerates diffusing process
and permits a precipitation that will not take place without irradiation.
Questions from Hans-Olof Andrén, Chalmers University of Tech:—I want to

report an observation that we have made by APT (Atom Probe Tomography) of
small (3–4 nm) spherical Nb precipitates in un-irradiated ZIRLO. These precipitates
were found in autoclave tested material in the heavily deformed zone just under the
oxide, and they resemble very much the small precipitates that you found in irradi-
ated M5.
Authors’ Response:—It is a very interesting observation.
Questions from Alexandre Legris, Lille 1 University, France:—It seems that Fe

has no effect on <c> dislocation loop nucleation. This seems to be in contradiction
with previous observations and interpretation of Yann de Carlan??? a couple of dec-
ades ago. Can you comment on that?
Authors’ Response:—If in Zircaloy-4 type alloys, these basal <c>-component

loops appear clearly correlated to iron re-distribution from the precipitates into the
matrix [36,37], the correlation is not obvious in M5TM alloy and more generally in
Nb-containing Zr-based alloys. <c>-component loops are not more numerous in
the vicinity of prior Laves phase precipitates in M5TM alloy, as in other Zr-Nb based
alloys [7,8,9,12].
G. A. Bickel,1 M. Griffiths,2 H. Chaput,2 A. Buyers,2

and C. E. Coleman2
Modeling Irradiation Damage in

Zr-2.5Nb and Its Effects on Delayed
Hydride Cracking Growth Rate
Reference
Bickel, G. A., Griffiths, M., Chaput, H., Buyers, A., and Coleman, C. E., “Modeling Irradiation
Damage in Zr-2.5Nb and Its Effects on Delayed Hydride Cracking Growth Rate,” Zirconium in the
Nuclear Industry: 17th International Symposium, STP 1543, Robert Comstock and Pierre Barberis, Eds.,
pp. 800–829, doi:10.1520/STP154320120157, ASTM International, West Conshohocken, PA 2015.3
ABSTRACT
Zr-2.5Nb is a dual-phase alloy consisting of an hcp (a) phase containing up to
1 wt. % Nb and a bcc (b) phase containing about 20 wt. % Nb. The a phase
constitutes the majority of the material volume. For in-service Zr-2.5Nb CANDU
pressure tubes, the structures of both the a and b phases evolve as a result of
the effects of irradiation and operating temperature: dislocation loop formation
in the a phase and decomposition or reconstitution of the b phase. X-ray
diffraction data are used to study the irradiation damage (represented by the
integral breadth of hcp diffraction peaks and the lattice parameter of the b
phase). This evolution of the microstructure must be modeled as a function of
operating conditions so that the state of the microstructure of in-service
pressure tubes can be predicted. Delayed hydride cracking (DHC) growth rates
in Zr-2.5Nb CANDU pressure tube material also depends on the state of the
microstructure. In this paper, it is shown that the majority of the DHC growth
rate changes can be ascribed to thermal and irradiation effects on the
microstructure.
22, 2014.
1
AECL–Chalk River Laboratories, Chalk River, Ontario K0J 1J0, Canada (Corresponding author),
e-mail: bickelg@aecl.ca
2
AECL–Chalk River Laboratories, Chalk River, Ontario K0J 1J0, Canada.
3
Hyderabad, India.
Copyright V
BICKEL ET AL., DOI 10.1520/STP154320120157 801
Keywords
Zr-2.5Nb, neutron irradiation, delayed hydride cracking, pressure tubes, micro-
structure, X-ray diffraction, line broadening, beta-phase
Introduction
Pressure tubes in CANDU4 reactors are made from Zr-2.5Nb alloy. The tubes are
fabricated by a two-stage forging process to convert cast ingots into logs that are
then hollowed to produce billets suitable for extrusion. The billets are extruded in
the a þ b phase and cold worked (nominally 27 %) to produce tubes of appropriate
dimensions (6 m long, 104-mm inner diameter, 4-mm wall thickness). These tubes
are stress relieved prior to installation in the reactor. This process provides tubes
with the appropriate microstructure to meet the requirements for mechanical
strength and fracture toughness. During service these pressure tubes operate with
coolant temperatures between about 250 C and 310 C. The maximum flux of fast
neutrons in the pressure tubes from the fuel is about 4 1017 nm2s1. Such oper-
ating conditions cause the microstructure to evolve in ways that may affect
pressure-tube performance. Figure 1 shows a graphical representation of the rela-
tionships necessary to model and predict the mechanical properties of in-service
pressure tubes. Central to the model is the state of the microstructure that deter-
mines many of the mechanical properties of the pressure tube. The relationships
between the mechanical properties and the microstructure are established through
various testing programs. Unfortunately, with currently available technologies, the
state of the evolving microstructure can only be determined by destructive analysis.
Therefore, ways to predict the state of the evolving microstructure based on the
operating conditions and the manufacturing process need to be developed for in-
service pressure tubes. The predicted state of the microstructure can then be used to
predict the fitness of these pressure tubes with respect to their mechanical proper-
ties as the microstructure evolves during service.
X-ray diffraction (XRD) techniques are an inexpensive way to assess the state
of the microstructure in both irradiated ex-service pressure tubes and unirradiated
archive material trimmed from the pressure tube during installation (called off-
cuts). XRD provides crystallographic lattice parameters and dislocations densities,
both of which evolve depending on the irradiation conditions.
The purpose of this paper is to establish the relationships between irradiation
conditions and the resultant microstructure for all axial locations of the pressure
tube. For these relationships to be useful, the role that the evolving microstructure
plays in changing various mechanical properties must also be well understood.
Delayed hydride cracking growth rate provides an example where all aspects of the
model shown in Fig. 1 can be realized.
4
CANDU (CANada Deuterium Uranium) is a registered trademark of Atomic Energy of Canada Limited.
FIG. 1 Simplified directed graph model showing relationships for predicting mechanical
properties from process variables and operating conditions. Arrows point from
explanatory variables to the response variables.
Evolution of Microstructure in Zr-2.5Nb Under

Irradiation
Zr-2.5Nb is a cold-worked, dual-phase alloy consisting of an hcp a phase contain-
ing up to 1 wt. % Nb and interspersed with filaments of bcc b phase containing ini-
tially about 20 wt. % Nb. The microstructure of finished Zr-2.5Nb pressure tubes
are described in greater detail in previous publications [1–3]. The a phase consti-
tutes the majority of the material volume and contains dislocations that were mostly
introduced in the final stages of fabrication (about 27 % cold-draw followed by a
stress-relief, autoclaving treatment at 400 C for 24 h). The main effect of the
cold-draw is to create dislocations in the a phase and the main effect of the stress-
relief heat treatment is to decompose partially the metastable b-phase filaments
into a mixture of Zr-rich x phase surrounded by residual Nb-rich b phase. The
structure of these three phases further evolves as a result of the effects of irradiation
and operating temperature. This evolution can be monitored by the appropriate x-
ray diffraction (XRD) techniques [4,5]. All phases have a pronounced texture, but it
is the a-phase texture that imparts most of the anisotropic properties to the pres-
sure tubes. The a-phase texture is such that many of the a grains are oriented with
their c axes in the hoop (transverse) direction of the tube and this dictates the ori-
entations of samples to be examined by x-ray diffraction (Fig. 2(a)). Diffraction
from specific planes at appropriate specimen orientations provides a measure of the
dislocation structure consisting of dislocations that have Burgers’ vectors that are: c,
c þ a, 1/2 c þ p, and a, which in Miller-Bravais’ notation are: [0001], 1/3 h1123i, 1/6
h2023i, and 1/3 h1120i, respectively (Fig. 2(b)). For the purposes of this paper the
focus is on measurement of the effect of irradiation on the dislocation structure (as
measured by x-ray line broadening) and the state of the b phase (as measured by
the shift in the b-phase lattice parameter). The XRD experimental methods
employed for these analyses are detailed in Refs 1, 4, and 5.
FIG. 2 (a) Typical specimen orientations for x-ray diffraction; and (b) relevant vectors
and planes for the hcp a phase in Zr-2.5Nb pressure tubing. Important planes
and directions are shown (c corresponds with [0001], a with h11
20i, and p with 1/
3 h1010i).
XRD LINE BROADENING IN THE a PHASE

The XRD integral breadth can be used to show trends in line-broadening behavior
that represent changes in dislocation densities [4,5]. Irradiation of hcp grains in Zr
alloys results in the formation of a high density of a-type dislocation loops [1]. In
general, the a-component of the dislocation density increases rapidly at low fluences
(<1 1025 n/m2) followed by a very slow linear increase in the a-component of the
dislocations for higher fluences [1,2,6,7]. The c-component dislocation structure does
not exhibit a fast transient increase but evolves at a slow linear rate similar to the a-
component behavior at all but the lowest fluence values [2,6,7]. The contribution to
the broadening of the x-ray diffraction lines is complicated by intergranular stress dis-
tributions [5] introducing an unquantified uncertainty. Because the irradiation tends
to relax the internal stresses (sharpening the x-ray diffraction lines) a relative increase
in the observed linewidth is considered a reasonable measure of the increase in dislo-

cation density. Also, whereas the a-type dislocations also produce some distortion of
the basal planes and vice versa for c-type dislocations, the broadening from radiation
damage for a given plane has been shown to be largely caused by dislocations with
Burgers’ vectors that are inclined to that plane; the type-I f1010g and type-II f1120g
prism planes for a-type dislocations and the basal (0001) planes for c-type disloca-
tions. For dislocations with a mix of a and c vectors, c þ a (1/3 h1123i) dislocations,
and 1/2 c þ p (1/6 h2023i) dislocation loops, there are contributions to the distortion
of both the basal and prism planes. In this paper, the first-order f1010g prism-plane
diffraction-peak profiles (hereafter denoted prism-1) are used to represent the dislo-
cation evolution. The use of other prism-plane diffraction-peak profiles provides no
additional information about the a-type dislocations. Apart from the initial transient,
the a-type dislocations can also represent the trends that are observed for the c-type
dislocations and the long-term fluence dependence (>1 1025 n/m2) of the prism
line broadening may result from c-component loop evolution [7].
b-PHASE TRANSFORMATION UNDER IRRADIATION AND TEMPERATURE

An almost continuous network of b-phase filaments (a single bcc phase metastable
structure containing 20 wt. % Nb) surrounds the a-phase grains after extrusion of
Zr-2.5Nb pressure tubes [2]. Finishing the tubes with a stress-relief treatment in an
autoclave at 400 C for 24 h causes decomposition of the b-phase structure relative
to the as-extruded state. During the autoclave treatment, an hcp x phase is formed
that is depleted in Nb along with an enrichment of the remaining bcc b phase
(20–50 wt. % Nb). The metastable bcc b phase will continue to transform with time
and temperature below 600 C. It is a diffusional process; Zr diffuses to the x par-
ticles causing growth of the x particle leaving the surrounding b-phase matrix
enriched in Nb. The surrounding b-phase remnants may become highly enriched
in Nb toward the equilibrium phase concentration (>90 wt. %) given high enough
temperature and long enough time. The original b-phase filaments would, there-
fore, evolve with time into a discontinuous a/b interface between the a grains [3,8].
During reactor operation, the irradiation of the material arrests, or even
reverses this thermal decomposition of the b phase [1]. One explanation is that the
fast neutrons cause a disordering (or dissolution) of the x-particle surface as a
result of atomic displacement within collision cascades removing atoms from
within the x particle and leaving them in the surrounding b-phase matrix [9]. This
dissolution rate is related to the damage rate (proportional to fast flux). In the reac-
tor environment, new equilibrium conditions are created that depend on the rela-
tive sizes of the thermal and irradiation enhanced processes. That is, for a given
temperature and flux, the size of the x particles is maintained in equilibrium by the
competing thermal decomposition and radiation-induced mixing processes. When
the equilibrium condition leaves more or larger x-phase particles and Nb enriched
b phase in the matrix, it is termed “decomposition of the b phase.” When the x-
phase particles are shrinking and the b phase in the matrix begins to enrich in Zr, it
is termed “reconstitution of the b phase.” There is a transient period while the ma-
terial attains this equilibrium condition. In and near the out-of-flux ends of the
pressure tube, where the thermal diffusion dominates, the decomposition continues
to increase over time until the b-phase filaments are fully decomposed.
One indicator of the degree of b-phase decomposition would be the measure-
ment of Nb concentration in the b phase, which can be quantified with x-ray dif-
fraction from a bcc lattice parameter [1] because it is linearly related to the
concentration of Nb. Although the actual Nb concentration of the b phase is
unique, the Nb values are derived from either RN (200) or LN (110) diffraction
peaks, which may not yield identical results. The differences are caused by internal
stresses causing XRD line shifts in opposite directions, but these Nb concentration
values tend to converge as the decomposition proceeds. Again, whereas these errors
caused by internal stresses have not been quantified, relative changes in the derived
Nb concentration are reasonable indicators of the b-phase decomposition.
Statistical Modeling of Irradiation Damage as a

Function of Operating Conditions
The irradiation damage of Zr-2.5Nb material, embodied in the XRD line broaden-
ing and the Nb concentration in the b phase, can be modeled through correlations
with the appropriate operational factors. The key service variables that affect the
irradiation damage are irradiation temperature, neutron dose rate (flux), and neu-
tron dose (fluence).
5
Figure 3 combines the integral breadth for experimental data (ERABLE irradia-
tion [6]) and for Zr-2.5Nb pressure tubes removed from service. All the materials
were irradiated at about 250 C but the material for the ERABLE experiment was
exposed to a neutron flux6 between 10 and 18 1017 nm 2s1, whereas the tubes
removed from service were exposed to much lower fluxes of 0.1 to
4 1017 nm2s1. Because the ERABLE experimental data and the in-service
results are not distinguishable, dose rate is not considered to be an important vari-
able affecting dislocation density for fast neutron fluxes >0.1 1017 nm2s1. Fig-
ure 3 confirms that after an early transient, the a-component dislocation density
increases at a steady rate with increasing neutron fluence.
Figure 4 shows the prism-1 line-broadening data for tube 1790 removed from
service after 20 calendar years of operation. It is clear from Fig. 4(a) that the axial
dependence is related to the temperature, as well as the neutron fluence. The axial
temperature and flux profiles (circumferential averages) are shown in Fig. 4(b). Also
evident in Fig. 4(a) is differing line-broadening values for different clock positions at
each axial location. The temperature and fast neutron flux not only varies along the
5
The ERABLE irradiation was conducted in the OSIRIS reactor, CEA, Saclay, France.
6
Having different neutron energy spectra with different dpa per unit fast flux (>1 MeV), the OSIRIS fast flux
is multiplied by 0.908 to convert it to an equivalent fast flux in a CANDU reactor.
FIG. 3 The prism-1 integral breadth for ERABLE experimental data [6] and CANDU Zr-
2.5Nb pressure tubes removed from service. All points shown are from material
irradiated between 250 C and 260 C.
length of the tube but also around the circumference of the tube. The fuel bundle
rests on the bottom of the tube (denoted as the 6 o’clock position) leaving a small
gap between the fuel bundle and the top of the tube (denoted as 12 o’clock) where
some coolant bypass can take place (Fig. 5). This 12 o’clock gap increases as diame-
tral creep proceeds throughout the reactor life and a significant temperature gradi-
ent around the tube circumference results with the lowest temperature associated
with the largest gap. Diametral creep follows the temperature and flux profile so the
diametral strain increases from inlet to outlet. Similarly, a small circumferential var-
iation of the fast flux tracks the proximity of the tube wall to the fuel bundle in the
crept channel. The temperatures and flux can be calculated from industry standard
thermal hydraulics codes [10] for any crept configuration. This range of conditions
also affords the opportunity to increase the coverage of operating variables when
modeling the fluence and temperature dependence at a few axial locations. The in-
tegral breadth is plotted in three dimensions as a function of operating temperature
and fluence (Fig. 6). It can be seen in Fig. 6 that throughout the body of the tube, the
dependence on fluence and temperature can be described by a simple surface that
increases smoothly as fluence increases and as temperature decreases. A fitted sur-
face with the circumferential corrected operating variables in Fig. 6 can be used to
estimate more accurately the line broadening for all spatial locations in the body of
a pressure tube. Alternatively, the model can be applied in retrospective assessments
where an unknown temperature history can be established from the observed line
broadening. For example, the unknown azimuthal orientation of a removed pres-
sure tube can be inferred from the circumferential variation of line broadening.
FIG. 4 (a) The prism-1 integral breadth as a function of axial location and clock position
for tube 1790. Pre-irradiation off-cut values are shown with dotted line
suggesting the axial variation for the pre-service integral breadth. For clarity,
these data points are averages from multiple specimens and some data points
are shifted from the actual axial location to separate overlapping points. (b) Flux
and coolant temperature profiles for lattice site of reactor where tube 1790
resided.
The ERABLE data for Nb concentration as a function of fluence is reproduced

from [6] in Fig. 7. The b-phase transformation evolves over time before reaching a
steady-state condition at a fluence of about 2.5 1025 nm2. Figure 8 shows the
steady-state Nb concentration as a function of flux (fixed temperature) for experi-
mental irradiations and for pressure tubes irradiated and removed from service af-
ter exceeding a fluence of 2.5 1025 nm2. These data suggest that the variation of
the value of Nb concentration in the steady-state condition is caused by this dose
FIG. 5 Schematic diagram showing change in dimensions as a pressure tube creeps

with time. The increased gap at 12 o’clock allows for coolant to bypass the fuel
bundle, which produces a variation in temperature around the circumference of
the tube.
rate (flux) dependence. Figure 9 illustrates the steady-state Nb concentration in the

b phase for tube 1790. The Nb concentration correlates inversely with the flux pro-
file but is skewed with a positive dependence on temperature (compare Fig. 9 to Fig.
4(b)). The measured off-cut values shown in Fig. 9 represent those of the starting
microstructure (partially decomposed b phase) after the autoclave treatment. On
the inlet region of the pressure tube (250 C–255 C) the initial b-phase decomposi-
tion is reversed (reconstituted) by the irradiation flux (as evidenced by the b-phase
lattice parameter shift). This change is interpreted in terms of dissolution of the
original x phase during irradiation at these temperatures [1,9]. At the outlet region
of the tube (290 C–300 C), the thermal decomposition of the b phase overtakes the
flux effect and the Nb concentration exceeds that of the starting material (growth of
the original x phase). Also note that for the points closest to the inlet, the 6 o’clock
data points are lower than the 12 o’clock data points because the fast flux is 5 %
higher at the 6 o’clock position but the temperature differences remain small
(<1 C) because of minimal diametral creep at the inlet. The response at the clock
positions reverses at the outlet when the decomposition is dominated by tempera-
ture and the 6 o’clock position is much hotter than the 12 o’clock position (a differ-
ence of 17 C) because of the extent of diametral creep at the outlet. The effect of
FIG. 6 A 3D rendering of the prism-1 integral breadth as a function of irradiation

temperature and fluence for tube 1790. The inset shows a surface that fits these
data.
clock position is also illustrated in Fig. 10 showing that the difference in Nb concen-
tration and the temperature difference parallel each other. For prediction purposes
within the body of tube, the steady-state Nb concentration can be simply modeled
as a quadratic polynomial with terms involving neutron flux and irradiation tem-
perature. As with the line broadening model, the Nb concentration can also help es-
tablish the temperature and flux history in retrospective assessments.
FIG. 7 The Nb concentration in the b phase from the ERABLE dataset [6].
FIG. 8 Steady-state Nb concentration in the b phase as a function of flux for

experimental irradiations and for pressure tubes irradiated and removed from
service after exceeding a fluence of 2.5 1025 nm2. All points shown are for a
fixed irradiation temperature range between 250 C and 260 C.
Effects in Very Low Flux Regions of Pressure

Tubes
As the neutron flux decreases rapidly at the ends of the fuel channels the effect of
the irradiation on the b-phase state and dislocation loop density differs. In the
FIG. 9 Nb concentration in the b phase as a function of axial and clock position for tube
1790. Pre-irradiation off-cut values are shown with dotted line suggesting the
axial variation of the initial Nb concentration. For clarity, these data points are
averages from multiple specimens and some data points are shifted from the
actual axial location to separate overlapping points.
FIG. 10 Nb concentration difference between the 12 o’clock (top) and the 6 o’clock
(bottom) circumferential locations. The temperature difference between the
top and bottom are scaled to roughly overlay with the Nb concentration
difference.
main body of the tube the evolution of the dislocation loop structure is initially very
rapid with increasing neutron fluence and the b-phase state changes much more
slowly (Figs. 3 and 7). In contrast to the main body the fast neutron flux at the ends
of the tubes, that are still within the reactor core and subjected to concerns over
DHC, drops to values that are in the range of about 0.1 1017 nm2s1 to about
0.3 1017 nm2s1. Over this flux range the b-phase state is not dominated by the
radiation damage and thermal decomposition can progress given sufficient time
and temperature [1,3]. The a-component dislocation structure evolves subject to
the dose (fluence) achieved. For fluences >1 1025 nm2 the dislocation structure
can be said to be close to a steady-state condition after the initial rapid transient
and for lower fluences one can expect lower dislocation densities, less line broaden-
ing and lower integral breadths (Fig. 3). At the very edge of the reactor core and
close to the end of the fuel channel the line broadening behaves in a non-standard
manner with respect to the accumulated dose. Figure 11 shows the distribution in
prism-1 line broadening and the % Nb obtained from LN sections at the 6 o’clock
location of pressure tube E0218. The data extend up to the edge of the rolled joint
to avoid confusion with the effects of rolling that change the initial starting micro-
structure compared with the off-cuts. The trend lines through the line-broadening
and b-phase data shown in Fig. 11 are fitted to those data corresponding with an
accumulated dose >1 1025 nm2 (bounded by line A and line B) to show the
general trend in the main body of the tube as the material approaches steady-state
conditions. Also shown are straight trend lines for Nb and integral breadth showing
the interpolation between the corresponding off-cut values at each end of the pres-
sure tube. These almost horizontal trend lines give some idea of the initial state of
FIG. 11 Variation in prism-1 line broadening and b-phase state for pressure tube E0218
measured at 6 o’clock circumferential location. Fitted lines illustrate the main
body-of-tube (between A and B) axial dependence. The values of the irradiated
material are compared with the unirradiated state given by the approximately
horizontal lines interpolated from measurements of unirradiated off-cuts.
these parameters. As with tube 1790, the prism-1 line-broadening reaches a maxi-
mum and the b-phase a minimum at about 1 to 2 m from the inlet. There are fur-
ther peaks in the prism-1 line-broadening close to the channel ends that are clearly
dependent on factors other than accumulated dose, the flux falling continuously as
one progresses away from the centre locations (see Fig. 4(b)). At the inlet there is a
sharp rise followed by a drop (outboard of line A in Fig. 11). The drop in line-
broadening outboard of line A is consistent with a corresponding decrease in neu-
tron flux but the rise is difficult to explain in terms of an evolving microstructure
because the flux is decreasing as the broadening is increasing (see Fig. 4(b)). Prelimi-
nary TEM imaging of regions near the inlet provide little support for an increased
dislocation density as the source of the rise in line broadening: no visible irradiation
damage (loop formation) could be seen in the region corresponding to the rise in
line broadening. At the outlet (see line B in Fig. 11) both the line broadening and b
phase are increasing with increasing distance from the core centre, i.e., with
decreasing neutron flux and almost constant temperature (see Fig. 4(b)). The
increase in the b-phase numbers are consistent with a lower flux, especially as the
temperature is almost constant in this region but the increased line broadening is
difficult to explain. A drop in temperature (because of conduction through the end-
fitting) would be one possibility but this decrease in temperature would be reflected
in a corresponding change in the state of the b phase, i.e., if temperature was a con-
trolling factor then a higher integral breadth would coincide with a lower Nb value.
No such corresponding trend exists. Preliminary TEM imaging of the outlet region
revealed the loop formation because of irradiation but qualitatively, the loop density
appeared to be constant as the line broadening increased.
A more detailed plot of the microstructure in the regions at the edge of the core
for tube E0218 are shown for the inlet in Fig. 12, and for the outlet in Fig. 13. Also
plotted in these figures are data from hydrogen measurements and retrospective do-
simetry obtained from punches taken at specific locations. It is apparent from the
combined profiles at the outlet (Fig. 13) that the one factor that correlates well with
the peak in line broadening approaching the burnish mark (where the tube is rolled
into the end-fitting) is the hydrogen equivalent7 concentration. At the hydrogen
equivalent concentration shown, all hydrogen would have precipitated as hydrides
after the tube had been removed from service and the tube cooled to room tempera-
ture. The increased line broadening could, in this case, be related to hydride
formation.
At the inlet the data are complicated by the fact that the fast neutron flux is
lower compared to the equivalent position relative to the burnish mark at the outlet.
This lower flux inboard of the rolled-joint at the inlet comes about because the fuel
bundles are shifted upstream with respect to the channel ends because of the cool-
ant flow toward the outlet. This shift in flux profile across the fuel channel is exacer-
bated as the fuel channel elongates during service. Whereas the line broadening just
inboard of the burnish mark at the inlet could be accounted for based on the com-
bined effects of neutron flux and hydrogen equivalent concentration in this region,
the drop in Nb concentration with lower flux is inconsistent with the trend in the
main body of tube. As the Nb concentration appears to be lower than the off-cut
values in this region it seems likely that factors other than low flux and time at tem-
perature are affecting this parameter. As noted earlier residual stresses are factors
that influence lattice parameters as well as chemistry. It is conceivable that residual
stresses inboard of the rolled joint still exist at the inlet, even after many years of
operation, because stress relaxation has not progressed to the same extent as the
outlet and this difference, in turn, is because both the temperature and flux are
lower. Evidence that the measured Nb concentration has been affected by residual
stresses and, therefore, an unreliable indicator of b-phase decomposition at the inlet
region is provided in Fig. 14. As expected, the volume fraction of x phase (whose
measurement is insensitive to residual stresses) continues to increase as the flux
decreases (Fig. 14) confirming that the b phase does continue to decompose. At the
higher temperature outlet, the observed volume fraction of x phase rises to a maxi-
mum before dropping off within the end-fitting with the drop in flux. This drop at
the outlet can be attributed to the eventual transformation of the x phase to a phase
after long time exposures [3].
7
Because both isotopes of hydrogen are present in the pressure tube, the hydrogen equivalent concentra-
tion is the wt. ppm of protium plus half the wt. ppm of deuterium.
FIG. 12 Variation in prism-1 line-broadening and b-phase state for pressure tube E0218
at the inlet. The vertical dotted line represents the edge of the burnish mark
where the pressure tube is rolled into the end-fitting. The region to the right of
this line may be subject to DHC. The values of the irradiated material are
compared with the unirradiated state given by the horizontal lines interpolated
from measurements of unirradiated off-cuts.
FIG. 13 Variation in prism-1 line-broadening and b-phase state for pressure tube E0218
at the outlet. The vertical dotted line represents the edge of the burnish mark
where the pressure tube is rolled into the end-fitting. The region to the left of
this line is subject to DHC. The values of the irradiated material are compared
with the unirradiated state given by the horizontal lines interpolated from
measurements of unirradiated off-cuts.
FIG. 14 Comparison of the measured Nb concentration in the b phase with the growth
of the x phase at the inlet and outlet regions of tube E0218.
There are a number of factors affecting both Nb concentration and the line
broadening near to the rolled joints and it is important to understand these factors
because of their effect on DHC and also fracture toughness. The reason why the
higher line-broadening would be related to hydrogen is not known. Possible explan-
ations include: (i) creation of dislocations when hydrides precipitate; (ii) a decrease
in the a-phase coherent diffraction domain size because of dispersion of precipitates
throughout the material; and (iii) a strain distribution caused by the hydrides
directly. More detailed work, including more examinations with transmission elec-
tron microscopy, is required to resolve this issue.
Hydrogen effects aside, taking Figs. 11–13 together, it is clear that the integral
breadth (dislocation density) is close to the unirradiated state near the outlet (see
vertical line B in Fig. 11), whereas the b phase is close to the unirradiated state near
the inlet (see vertical line A in Fig. 11). At these locations, therefore, the confounding
effect of a correlation between the b phase and dislocation density can be broken.
The Effect of Irradiation Damage on Delayed

Hydride Crack Growth Rates
A key utility of the statistical models relating irradiation damage to the operating
conditions is to predict the state of the microstructure in operating pressure tubes
because it cannot be otherwise determined without destructive examination. Know-
ing the state of the microstructure enables subsequent predictions of the mechanical
behavior of these critical reactor components. For example, linkages between
delayed hydride cracking (DHC) growth rate and the irradiation conditions can,
therefore, be made.
In one model for DHC, hydrogen diffuses up the stress gradient at a stress
raiser, such as a crack in tension and, if sufficient hydrogen accumulates to exceed
the solubility limit for precipitation a hydride forms [11]. If the stress is large
enough to break the hydride, the crack will extend and the process can be repeated
[11–13]. The fact that the DHC growth rate depends on temperature and the irradi-
ated material state of the pressure tube, i.e., the microstructure in the vicinity of the
crack tip, is illustrated in Fig. 15. Apart from the test temperature dependence, Fig. 15
shows that irradiated material has higher growth rates than unirradiated material
and those regions with different temperature history and different flux, and, there-
fore, fluence, have different DHC growth rates.
A summary description of DHC and the detailed methodologies employed in
this work to measure the DHC growth rates can be found in [14].
FIG. 15 Temperature dependence of the axial DHC growth rate measured in tube 1790
when unirradiated and after irradiation at two locations. The calculated values
for the unirradiated material are based on Eq 1.
CORRELATING a-PHASE INTEGRAL BREADTH AND b-PHASE NB

CONCENTRATION WITH DHC GROWTH RATE
In previous irradiation experiments (e.g., ERABLE [6]) the relationship between the
DHC growth rates and irradiation damage could be inferred although, these experi-
ments did not involve paired tests and a direct measurement of the relationship
between the DHC growth rates and irradiation damage has so far not been
reported. DHC measurements have been carried out on the ex-service tube 1790.
Several locations were characterized by XRD analysis and DHC specimens were
taken immediately adjacent to the characterized regions. Figures 16 and 17 show the
relationship between DHC growth rates (axial and radial) at 150 C and integral
breadth or Nb concentration. When assuming a linear relationship, the coefficient
of determination, R2, is >0.9, strongly suggesting that either or both independent
microstructural variables are linked to the explanatory variables that control the
DHC growth rate. That is, confirming that DHC growth rate is strongly affected by
a-phase dislocation density or b-phase decomposition or both. Decomposition of
FIG. 16 Axial DHC growth rates as a function of the prism-1 integral breadth (upper)
and as a function of the Nb concentration in the b phase (lower) for tube 1790.
The DHC test temperature is 150 C.
FIG. 17 Radial DHC growth rates as a function of the prism-1integral breadth (upper)
and as a function of the Nb concentration in the b phase (lower) for tube 1790.
The DHC test temperature is 150 C.
the b phase can shut down the short circuit for hydrogen migration to the crack tip
[8,12,15]. Decomposing this b phase or transforming the filaments into a heteroge-
neous structure filled with x particles interspersed with Nb-enriched b-phase pre-
cipitates should impede hydrogen diffusion and, assuming that potential hardening
of the filaments does not dominate; therefore, should have the effect of lowering
DHC growth rates. Reconstituting the b phase should have the opposite effect and
increase DHC growth rates. The dislocation loops created in the a phase will
strengthen the matrix and potentially raise the crack growth rate [16]. Also appa-
rent in Figs. 4, 9, 16, and 17 is that, at least for the body of tube, a-phase dislocation
density and b-phase decomposition are completely correlated making it impossible
to discern the relative contribution of either mechanism (matrix strength or hydro-
gen diffusion through the b phase) to the DHC growth rate. The correlation
between a-phase dislocation density and b-phase decomposition can be broken at
the tube ends as described above but to date, insufficient DHC growth rate
FIG. 18 The radial DHC growth rate from the ERABLE data and other CANDU ex-
service tubes. The DHC test temperature is 240 C for all of the data shown.
Tube H737 is the source of the material for the ERABLE irradiation.
measurements have been carried out in that region. The ERABLE data affords the
chance to isolate the two effects at early irradiation times where the Nb concentration
is still in transition and the integral breadth has already reached steady state (see
Figs. 3 and 7). The radial DHC growth rate from the ERABLE data and other
CANDU ex-service tubes (all tested at 240 C) is shown in Fig. 18 and exhibits a simi-
lar transient as a function of fluence when compared with the Nb concentration
(Fig. 7) suggesting that the DHC growth rate mostly depends on the b-phase decom-
position, while being only weakly dependent on the hardening because of radiation
damage. This contrasts with other Zr-alloys where neutron irradiation increases the
strength and DHC rate but contain no b phase to complicate the interpretation [17].
COMPARING MECHANISTIC MODELS WITH DHC GROWTH RATE

Insights into the relative effects of a-phase dislocation density and b-phase decom-
position can be found in appropriately constructed mechanistic models. In a model
derived from both a hydrogen concentration gradient and a hydrostatic stress gra-
dient, providing a gradient in chemical potential, the rate of cracking, c, was shown
[11] to depend on:
• the amount of hydrogen in solution in the bulk material, Cb; this quantity
depends on the solubility limit, Cp, and the temperature history of the
specimen;
• the rate of arrival of hydrogen to the crack tip; this quantity depends on the
diffusivity of hydrogen in the zirconium alloy, D;
• the amplification of the hydrogen concentration at the stress raiser; this quan-
tity depends on the partial molar volume of hydrogen in the zirconium alloy,
VH, and the hydrostatic stress, rH, which can be evaluated from the yield
stress, ry.
The value of vc is related to the variables through:

rH VH
(1) vc ¼ W D Cb Cp exp
RT
where:
W ¼ empirical constant,
R ¼ gas constant, and
T ¼ temperature (in K).
The variables D, Cp, and rH each depend on temperature, microstructure, and
irradiation so each will affect c. Evidence that this model captures the DHC growth
rate behavior is shown in Fig. 15 where good agreement is found for the unirradiated
case [18], using Cp from [19], the values of ry measured from the source material
and D with activation energy of 37 kJ/mol [20]. Unlike zircaloy [21], unpublished
observations suggest that the values of the solubility limit in Zr-2.5Nb are either not
affected or slightly decreased by irradiation.
For Zr-2.5Nb the effect of irradiation conditions leading to the state of the b
phase (Fig. 9), and its effect on diffusivity, appears to be large compared with the
effect of the increase in yield strength. When the values of yield stress for the irradi-
ated material are applied in Eq 1, a close correspondence is observed for the mate-
rial taken from close to the outlet end (Fig. 19). The values of D (and Cp) are not
changed from those used for Fig. 15 and the 20 % to 30 % increase in strength is suf-
ficient to describe the crack growth rate. When the same calculation is performed
FIG. 19 Measurement and calculation of axial DHC growth rate for Zr-2.5Nb material
close to the outlet end of tube 1790. The values of CP and D were the same as in
Fig. 15 for unirradiated material but the yield stress was that of the irradiated
tube.
on the material from the center of the reactor the agreement with measurements of
crack growth rate is poor (Fig. 20(a)). The increase in yield strength is insufficient to
raise the amplification factor, exp(rHVH/RT), so that the calculated values of vc are
not similar to those measured. Figure 9 shows that the b phase is being reconstituted
in the central regions of the fuel channel. This change in the b phase will allow the
hydrogen to diffuse faster than if it were not altered, as described previously. This
increased diffusivity can be modeled by retaining the same activation energy but
increasing the pre-exponential factor, D, by 2.3 times. Now the combination of the
increased yield strength and diffusivity describe the measured values (Fig. 20(b)).
FIG. 20 Measurement and calculation of axial DHC growth rate for Zr-2.5Nb material
close to the centre of tube 1790. For (a) the values of CP and D were the same
as in Fig. 15 for unirradiated material but the yield stress was that of the
irradiated tube. For (b) the values of CP was the same as in Fig. 15 for
unirradiated material, the yield stress was that of the irradiated tube and the
value of D was increased by a factor of 2.3.
Discussion
The results show that, to a large extent, the DHC growth rate behavior of Zr-2.5Nb
pressure tubing can be related to the state of the microstructure and is linked to the
decomposition and reconstitution of the b phase that controls hydrogen diffusion
to the crack tip [20,22], and the dislocation loop density, causing hardening of the
a-phase matrix.
The relative importance of these two factors is captured in the mechanistic
model described above, which shows that the DHC growth rate can be accounted
for by the hydrogen diffusion rate and the strength of the matrix. To ascertain the
relative significance of the microstructural changes affecting hydrogen diffusion (b-
phase state) and the matrix strength (dislocation density) in irradiated material one
can take advantage of the differences in the rate of change of these parameters. In
the case of the ERABLE irradiation these changes lead to a narrow range of fluence
where the dislocation loop density had increased significantly and the b-phase state
had changed little (see Figs. 3 and 7). Consequently, for the ERABLE irradiation,
one dominant factor leading to increased DHC rates could be identified, i.e., the b-
phase reconstitution. Other than for the ERABLE data at low fluences (about
0.5 1025 nm2) there are no instances where the change in line broadening can
be separated from the changes in the state of the b phase.
In regions of the tube removed from the power reactor, where the b-phase
decomposition dominates (as opposed to reconstitution), e.g., at the tube outlet
(low flux, high temperature), one expects to see the opposite effect of irradiation on
the b phase compared with that at the center of the reactor (high flux, intermediate
temperature). High DHC rates at the outlet (low flux, high temperature) are then
dependent on the yield strength. Given the evidence that the DHC rate is, if any-
thing, only weakly related to line-broadening and given that it is relatively insensi-
tive to ry in the mechanistic model, it is likely that the b-phase state, and, therefore,
diffusion, is the most important aspect of DHC in irradiated Zr-2.5Nb pressure tub-
ing. This conclusion is consistent with those of Jovanovic et al. [15] examining the
effect of heat treatment on DHC growth rates.
Further work is necessary to separate out the effects of strength and diffusivity.
It should be possible to isolate the effects of both parameters by examining the
effect on DHC rates at the locations with the same low flux but differing tempera-
tures denoted A and B in Fig. 11. Comparisons of DHC data from Zr alloys (lacking
the b phase) may also shed light on the role of hydrogen diffusion through the Zr-
2.5Nb b phase on DHC growth rates.
It remains to be seen whether the increased line broadening at the very ends
of the tube attributed to hydride precipitation is irreversible, i.e., whether the
broadening is the result of dislocations generated by hydride formation [11], thus
changing the microstructure permanently and irrespective of whether the hydrides
re-dissolve as a fuel channel is taken back up to temperature after a shutdown
[13,16].
Conclusions
1. The microstructure in Zr-2.5Nb pressure tubing (dislocation density and state
of the b phase) evolves during service and approaches a pseudo-steady-state
condition after a fluence of about 0.5 1025 nm2 and about 3 1025 nm2,
respectively.
2. The dislocation density is insensitive to displacement damage rate for fast neu-
tron fluxes >1 1016 nm2s1. The state of the b phase is sensitive to the
fast neutron flux.
3. The dislocation density and the state of the b phase are both sensitive to tem-
perature and this dependence is easily seen around the circumference of the
pressure tube where the temperature only varies by 10 C to 20 C.
4. The change in the x-ray diffraction line broadening and b-phase lattice param-
eter that are used as measures of dislocation density and b-phase state are
complex in the regions within 200 mm of the rolled joints. These changes are
attributed to a combination of elevated hydrogen concentration and residual
stresses that may exist in these regions (primarily at the inlet).
5. Irradiation increases the rate of DHC in Zr-2.5Nb pressure tubing in operating
CANDU reactors and this increase is strongly correlated with changes in
microstructure measured by x-ray diffraction.
6. When the b phase is largely decomposed, the small increase in DHC growth rate
is fully explained by the increase in yield stress. This explanation is insufficient
when the b phase is reconstituted; now DHC growth rates are dominated by dif-
fusivity. The observations are consistent with the view that decomposition acts to
slow DHC growth rates, and reconstitution acts to increase DHC growth rates.
ACKNOWLEDGMENTS
The writers wish to acknowledge James Charbonneau for performing the DHC meas-
urements and Sterling St. Lawrence for helpful discussions. Portions of this work are
funded by the CANDU Owners Group.
References
[1] Griffiths, M., Mecke, J. F., and Winegar, J. E., “Evolution of Microstructure in Zirconium
Alloys During Irradiation,” Zirconium in the Nuclear Industry: Eleventh International Sym-
posium, ASTM STP 1295, E. R. Bradley and G. P. Sabol, Eds., ASTM International,
[2] Griffiths, M., Davies, P. H., Davies, W. G., and Sagat, S., “Predicting the In-Reactor Me-
chanical Behavior of Zr-2.5Nb Pressure Tubes from Postirradiation Microstructural Ex-
amination Data,” Zirconium in the Nuclear Industry: Thirteenth International Symposium,
ASTM STP 1423, G. D. Moan and P. Rudling, Eds., ASTM International, West Consho-
hocken, PA, 2002, pp. 507–523.
[3] Griffiths, M., Winegar, J. E., and Buyers, A., “The Transformation Behavior of the b-phase
in Zr-2.5Nb Pressure Tubes,” J. Nucl. Mater., Vol. 383, Nos. 1–2, 2008, pp. 28–33.
[4] Griffiths, M., Winegar, J. E., Mecke, J. F., and Holt, R. A., “Determination of Dislocation
Densities in HCP Metals Using X-Ray Diffraction and Transmission Electron Microscopy,”
Adv. X-Ray Anal., Vol. 35, 1992, pp. 593–599.
[5] Griffiths, M., Sage, D., Holt, R. A., and Tome, C. N., “Determination of Dislocation Den-
sities in HCP Metals from X-Ray Diffraction,” Metall. Mater. Trans. A, Vol 33, No. 3, 2002,
pp. 859–865.
[6] Pan, Z. L., St. Lawrence, S., Davies, P. H., Griffiths, M., and Sagat, S., “Effect of Irradiation
on the Fracture Properties of Zr-2.5Nb Pressure Tubes at the End of Design Life,” J.
ASTM Int., Vol. 2, No. 9, 2005, Paper ID JAI12436.
[7] Hosbons, R. R., Davies, P. H., Griffiths, M., Sagat, S., and Coleman, C. E., “Effect of Long-
Term Irradiation on the Fracture Properties of Zr-2.5Nb Pressure Tubes,” Zirconium in
the Nuclear Industry: Twelfth International Symposium, ASTM STP 1354, G. P. Sabol and
G. D. Moan, Eds., ASTM International, West Conshohocken, PA, 2000, pp. 122–138.
[8] Simpson, L. A. and Cann, C. D., “The Effect of Microstructure on Rates of Delayed
Hydride Cracking in Zr-2.5 % Nb Alloy,” J. Nucl. Mater., Vol. 126, No. 1, 1984, pp. 70–73.
[9] Nelson, R. S., Hudson, J. A., and Mazey, D. J., “The Stability of Precipitates in an Irradia-
tion Environment,” J. Nucl. Mater., Vol. 44, No. 3, 1972, pp. 318–330.
[10] Carver, M. B., Tahir, A., Kiteley, J. C., Banas, A. O., Rowe, D. S., and Midvidy, W. I., “Simulation
of the Distribution of Flow and Phases in Vertical and Horizontal Bundles Using the
ASSERT-4 Subchannel Code,” Nucl. Eng. Des., Vol. 122, Nos. 1–3, 1990, pp. 413–424.
[11] McRae, G. A., Coleman, C. E., and Leitch, B. W., “The First Step for Delayed Hydride
Cracking in Zirconium Alloys,” J. Nucl. Mater., Vol. 396, No. 1, 2010, pp. 130–143.
[12] Sagat, S., Coleman, C. E., Griffiths, M., and Wilkins, B. J. S., “The Effect of Fluence and
Irradiation Temperature on Delayed Hydride Cracking in Zr-2.5Nb,” Zirconium in the Nu-
clear Industry: Tenth International Symposium, ASTM STP 1245, A. M. Garde and E. R.
Bradley, Eds., ASTM International, West Conshohocken, PA, 1994, pp. 35–61.
[13] Cheadle, B. A., Coleman, C. E., and Ambler, J. F. R., “Prevention of Delayed Hydride
Cracking in Zirconium Alloys,” Zirconium in the Nuclear Industry: Seventh International
Symposium, ASTM STP 939, R. B. Adamson and L. F. P. Van Swam, Eds., ASTM Interna-
[14] “Delayed Hydride Cracking in Zirconium Alloys in Pressure Tube Nuclear Reactors,”
IAEA-TECDOC-1410, International Atomic Energy Agency, Vienna, Austria, 2004.
[15] Jovanovic, M. T., Eadie, R. L., Ma, Y., Anderson, M., Sagat, S., and Perovic, V., “The Effect
of Annealing on Hardness, Microstructure and Delayed Hydride Cracking in Zr-2.5Nb
Pressure Tube Material,” Mater. Charact., Vol. 47, Nos. 3–4, 2001, pp. 259–268.
[16] Shek, G. K. and Graham, D. B., “Effects of Loading and Thermal Maneuvers on Delayed
Hydride Cracking in Zr-2.5 Nb Alloys,” Zirconium in the Nuclear Industry: Eighth Interna-
tional Symposium, ASTM STP 1023, L. F. P. Van Swan and C. M. Eucken, Eds., ASTM Inter-
national, West Conshohocken, PA, 1989, pp. 89–110.
[17] Huang, F. H. and Mills, W. J., “Delayed Hydride Cracking Behavior for Zircaloy-2 Tubing,”
Met. Trans., Vol. 22A, 1991, pp. 2049–2060.
[18] Coleman, C. E. and Inozemtsev, V. V., “Measurement of Rates of Delayed Hydride Crack-
ing (DHC) in Zr-2.5Nb Alloys—An IAEA Coordinated Research Project,” J. ASTM Interna-
tional, Vol. 5, No. 2, 2008, Paper ID JAI101091.
[19] “Technical Requirements for In-Service Evaluation of Zirconium Alloy Pressure Tubes in
CANDU Reactors,” CSA N285.8-05, Canadian Standards Association, Toronto, 2005.
[20] Skinner, B. C. and Dutton, R., “Hydrogen Diffusivity in a-b Zirconium Alloys and Its Role
in Delayed Hydride Cracking,” Hydrogen Effects on Material Behavior, N. R. Moody and
A. W. Thompson, Eds., The Minerals, Metals and Materials Society, Warrendale, PA, 1990,
pp. 73–83.
in Zirconium Alloys,” Zirconium in the Nuclear Industry: Twelfth International Sympo-
sium, ASTM STP 1354, G. P. Sabol and G. D. Moan, Eds., ASTM International, West Con-
shohocken, PA, 2000, pp. 173–195.
[22] Sawatzky, A., Ledoux, G. A., Tough, R. L., and Cann, C. D., “Hydrogen Diffusion in
Zirconium-Niobium Alloys,” Proceedings of the International Symposium on Metal-
Hydrogen Systems, T. N. Veziroglu, Ed., Pergamon, Tarrytown, NY, 1981, pp. 109–120.
DISCUSSION
Questions from Jean-Christophe Brachet, CEA-Sailey Nuclear Materials Dept.
France
Q1:—Is there any evidence of bNb irradiation enhanced precipitation in the a

phase?
Authors’ Response:—Although not discussed in this paper, we have previously

reported bNb precipitates in cold-worked Zr-2.5Nb. (Pan, Z.L., et al, “Effect of irra-
diation on the fracture properties of Zr-2.5Nb pressure tubes at the end of design
life”, J.ASTM International, 2, (2005) Paper ID JAI12436). During irradiations at
temperatures higher than those in CANDU, these precipitates were confirmed to be
bNb from their crystal structure, lattice parameter and chemical composition and
were shown to be irradiation enhanced because they did not dissolve during post-
irradiation annealing (Coleman, C.E., Gilbert, R.W., Carpenter, G.J.C., Weatherly,
G.C., Precipitation in Zr-2.5 wt% Nb during neutron irradiation, in Phase Stability
during Irradiation, J.R. Holland, L.K. Mansur, D.I. Potter (Eds.,) Met. Soc. AIME,
New York (1981), 587-599.) Other studies (M. Griffiths and H. Muellejans, “A TEM
Study of a-phase stability in Zr-2.5Nb Pressure tubes following neutron irradiation,”
Micron, Vol. 26, 1995, p.555) show that they are stable and Nb-rich and are therefore
assumed to be bNb.
Q2:—I am not familiar with mechanism of decomposition of metastable 20%

Nb containing bZr phase. I understand that this decomposition occurs through
omega phase formation. From the equilibrium phase diagram, the final equilibrium
phases should be a þ bNb, so is it due too slow kinetics? Could we expect bNb pre-
cipitation at longer time and/or higher irradiation level than those experienced so
far?
Authors’ Response:—Yes, the kinetics are slow for the temperatures being con-
sidered. Decomposition to bNb is expected at long enough times (M. Griffiths, J. E.
Winegar, A. Buyers, “The Transformation Behavior of the b-phase in Zr-2.5Nb
Pressure tubes”, J. Nucl. Mater., vol. 383, 2008, pp. 28–33), except when hindered
by the effect of irradiation.
Questions from K. Somasekhar Reddy, Nuclear Fuel Complex, Hyderabad, India
Q1:—What is the range of threshold stress intensity at which DHC growth rate
is measured?
Authors’ Response:—DHC growth rate testing was performed in the applied KI

pffiffiffiffi
range of 15 to 30 Mpa m.
Q2:—What technique was employed to monitor the crack growth rates?
Authors’ Response:—Crack initiation was determined by acoustic emission. The

tests were performed in fixed displacement mode, which allowed prediction of the
amount of crack growth based on the load reduction due to cracking.
Q3:—Is there any effect on threshold stress intensity factor with b-

decomposition?
Authors’ Response:—No significant effects related to reactor operating parame-

ters on threshold stress intensity factor have been observed. Also, in cold-worked
and stress relieved Zircaloy, where there is no b-phase, KIH is similar to that in Zr-
2.5Nb.
Questions from Rishi Sharma, IIT Bombay, India:—Can you speculate about
the hydrogen absorption rate at the end fitting location? Is this rate constant over
the period of reactor operating time?
Authors’ Response:—The deuterium ingress rate is not measured directly.

Instead, the deuterium concentration in the pressure tube is measured at a
fixed point in time and this concentration is subject to both D-ingress and dif-
fusion resulting in a steep axial dependence for the deuterium concentration
near the rolled joint (e.g., see Figures 12 or 13). A sense of the ingress rate can
be seen from the rolled joint deuterium ingress profiles shown in Figures 12
and 13. This pressure tube had been in service for approximately 15 calendar
years when the deuterium concentration profiles were measured. Other unpub-
lished evidence suggests that the deuterium ingress rate decreases with reactor
operating time.
Q1:—What was the size and number of specimens analyzed to arrive at the Nb
concentration at a location in the irradiated pressure tube? Will this value represent
the cross section of the pressure tube at the location?
Authors’ Response:—The sample size has to be about 4mm x 10mm minimum

to have a reasonable XRD signal. The Nb concentration is determined at each loca-
tion in the irradiated pressure tube from 1 sample for the RN direction and 2 to 3
samples per location for the LN direction (see Figure 2 in the paper for the defini-
tion of the RN and LN directions). The RN represents the mid wall value for a spe-
cific axial and clock location, while the LN represents an average through wall
measurement at a specific axial and clock location.
Q2:—How did you separate the effect of damage due to irradiation from that
due to cold work at the roll joint taper region? What could be the effect of relaxa-
tion of stresses at the roll joint during the course of irradiation in the reactor?
Authors’ Response:—The broadening from radiation damage in the form of dis-

location loops is incremental to that from cold-work. The increased line-broadening
is attributed to radiation damage in general except the region approaching the bur-
nish mark near the rolled-joint. The peak in broadening is displaced inboard from
the burnish mark at the inlet and this fact leads us to suspect that increased hydrogen
is a contributor to the broadening in these regions. Relaxation of stresses is likely to
result in a reduction of line-broadening, thus reducing the potential contribution
from type-II stresses (“Determination of Dislocation Densities in HCP Metals from
XRD Line-Broadening Analysis”, M. Griffiths, D. Sage, R.A. Holt and C.N.Tome,
Proc. TMS 2001, Mater. and Metall. Trans. A, Vol. 33 (2002) p.859.).
Q3:—What could be the effect of globularization of b phase on DHC velocity?
Authors’ Response:—Globularization of the b-phase generally results from

excessively high annealing temperatures (N. Saibaba et al., “Study of microstructure,
texture and mechanical properties of Zr-2.5Nb alloy pressure tubes fabricated with
different processing routes”, J. Nucl. Mater., vol. 440, 2013, pp. 319–331) and is not
prevalent in CANDU pressure tubes. With time at temperature in the reactor the
b-phase may change chemical composition and lose continuity. Simultaneously the
strength of the alloy may change. We are currently investigating how to separate
the role of each of these effects.
Questions from B. K. Shah, BARC Mumbai:—Continuity of beta phase could be

responsible for higher delayed hydride crack growth rate. Why not break the conti-
nuity of beta phase by autoclaving of the pressure tube to a longer duration?
Authors’ Response:—Autoclaving the pressure tubes for a longer period will affect
both the continuity and composition of the b-phase, and does reduce the DHC growth
rate. The majority of the benefit is gained in the first 24 h at 400 C with diminishing
benefit at longer heating times. Also care has to be exercised that the strength is not
compromised by stress-relieving at long times, high temperatures or both.
Q1:—What is the objective of the autoclave exposure at the end of manufactur-

ing? What is the test condition for the autoclave exposure (water, steam...)?
Authors’ Response:—The objective of the treatment is stress-relief. The auto-

clave operates at 400 C with steam. Although sufficient stress relief is achieved
within 12 hours, for manufacturing convenience the heat-treatment lasts 24 hours.
Q2:—Do you autoclave all the components?
Authors’ Response:—No, just the pressure tubes.
Q3:—What is the thickness of the oxide formed during autoclaving?
Authors’ Response:—The oxide is 1 to 2 lm and black
Q4:—Is there any hydrogen ingress in Zr-2.5Nb close to the fitting with stain-
less steel?
Authors’ Response:—Yes, deuterium is produced by corrosion reactions on the

end fitting and enters the pressure tube by solid-state diffusion through the stainless
steel and into the Zr-2.5Nb at points of oxide-free contact that exist in the rolled
joint.
Q5:—Do you observe crevice corrosion?
Authors’ Response:—Crevice corrosion is a known and relatively common

occurrence in the pressure tube. It occurs in the occluded regions beneath fuel bun-
dle bearing pads (body of tube, not rolled joint). It does not occur at damaging
rates. Mainly because of this observation, crevice corrosion has been postulated to
occur in the rolled joint, but has not been actually demonstrated.
Javier Romero,1 Mats Dahlbäck,2 Lars Hallstadius,2

Maria Ivermark,2 and Guido Ledergerber3
Understanding the Drivers

of In-Reactor Growth
of b-Quenched Zircaloy-2
BWR Channels
Reference
Romero, Javier, Dahlbäck, Mats, Hallstadius, Lars, Ivermark, Maria, and Ledergerber, Guido,
“Understanding the Drivers of In-Reactor Growth of b-Quenched Zircaloy-2 BWR Channels,”
ABSTRACT
The performance of boiling water reactor (BWR) fuel channels of standard and
b-quenched Zircaloy-2 has been evaluated, aiming to determine the impact of
b-quenching on in-reactor growth of channels. An extensive database of BWR
fuel channel growth and bow from different reactors has been collected. Channel
bow, driven by differential growth between opposite channel sides, is a key
parameter for the safe performance of the fuel. Hot cell examinations have been
performed on channels with burn-up between 45 and 60 MWd/kgU. Oxide
thickness and hydrogen content measurements at different sides and elevations
have been conducted, as well as metallography and high resolution transmission
electron microscopy (TEM) work, including measurement of dislocation density
and characterization of size, crystal structure and chemistry of second phase
particles (SPPs). From the results of multiple inspection campaigns and hot cell
examinations, the effects of texture and hydrogen uptake on in-reactor growth
Manuscript received January 23, 2013; accepted for publication April 12, 2014; published online October 6,
2014.
1
Westinghouse Electric Company, Hopkins, SC 29061.
2
Westinghouse Electric Sweden AB, Västerås, SE 72163, Sweden.
3
Kernkraftwerk Leibstadt AG, CH 5325 Leibstadt, Switzerland.
4
Hyderabad, India.
Copyright V
ROMERO ET AL., DOI 10.1520/STP154320130020 831
of channels are differentiated. The b-quenching heat treatment of Zircaloy-2

sheets is effective in considerably weakening the texture of the material, which
in turn reduces bulk irradiation growth caused by anisotropic deformation of
grains. The growth of b-quenched Zircaloy-2 channels has been found to be
driven mainly by volume change caused by hydrogen uptake, which is essentially
independent of texture, with a known relationship between hydrogen uptake and
growth. Hydrogen uptake, and thereby in-reactor channel growth, accelerates in
b-quenched material at certain fluence levels, and is linked to evolution of SPPs
during irradiation. Zircaloy-2 is known to exhibit dissolution and amorphization
of Fe–Cr rich particles, coupled with slower dissolution of Fe–Ni particles. In this
study, the dissolution of SPPs of b-quenched Zircaloy-2 has been found to be
faster relative to non-b-quenched material, which is associated to the hydrogen
uptake late in life that drives the growth of channels with very weak texture.
Keywords
beta-quenching, irradiation growth, hydrogen pickup
Introduction
The nuclear industry is interested in dimensional and geometrical stability of boil-
ing water reactor (BWR) channels during service. Channel bow (loss of straight-
ness) is a significant safety issue, since excessive bow may close the gap between the
fuel assemblies and the control blades, potentially creating interference for insertion
if blades are needed for power control or emergency shutdown. Furthermore, chan-
nel bow has a significant negative impact on neutron economy and thereby fuel
cycle cost. Channel bow is attributed to differential elongation on opposite sides of
the channel. This differential elongation can be caused due to differences in neutron
fluence, operational history, or as it will be discussed here, on hydrogen pickup.
Irradiation-induced growth of BWR channels is in principle a well-known phe-
nomenon. In a single crystal scale, it is defined as a shape change with constant vol-
ume in the absence of applied stress during irradiation. It occurs due to unequal
distribution of interstitial and vacancy loops on different crystallographic planes of
the hexagonal zirconium structure. Irradiated annealed single crystals of zirconium
have been shown to expand in the direction parallel to the hai axis and contract in
the direction parallel to the hci axis [1]. In the presence of a strong crystallographic
texture, like the one developed by cold-rolling processing of sheet, the bulk material
undergoes anisotropic growth.
Some efforts to minimize the growth of BWR channels have focused on weakening
the crystallographic texture, performing final dimension b-quenching of the Zircaloy-2
sheets used to manufacture the channels [2]. This development demonstrated the effec-
tiveness of b-quenching due to the double phase transformation undergone by the ma-
terial, which exploits the multiple orientation variants available in the a!b and b!a
phase transformations [3]. Even in the presence of orientation variant selection, the tex-
ture of cold-rolled material is considerably weakened by b-quenching, which in turn
reduces the bulk irradiation-induced growth of the channels.
This paper presents a database of in-reactor performance of b-quenched chan-

nels, which clearly demonstrates the benefit of b-quenching. However, there are
cases where b-quenched channels exhibit a relative large growth, approaching the
average performance of standard (non b-quenched) material. This prompts a need
for better understanding of additional mechanisms, in addition to crystallographic
texture, that drive in-reactor channel growth and distortion.
For this purpose, hot cell examinations of irradiated materials at different burn
up levels have been performed. They encompassed oxide thickness and hydrogen
content measurements, metallography and high resolution transmission electron
microscopy (TEM), including measurement of dislocation density and characteriza-
tion of size, crystal structure, and chemistry of second phase particles (SPPs).
In-Reactor Performance of b-Quenched

Channels
OXIDE THICKNESS, GROWTH, AND BOW
Figures 1, 2, and 3 show comparisons of average of maximum oxide thickness,
growth and bow of standard and b-quenched Zircaloy-2 channels, as a function of
equivalent channel burn up. Data were obtained from different BWRs on channels
subjected to diverse water chemistries and irradiation histories with no or minimal
control blade presence. This data set is only included as a background to the chan-
nels examined in the hot cell and no further statistical evaluation has been made.
Oxide measurements on outer channel surfaces were performed using standard
eddy current (EC) techniques. Bow is defined as the maximum displacement from
vertical in the two directions perpendicular to the faces of the channel (x and y),
being positive towards the control blade and negative away from the control blade.
The fuel channel bow measurement equipment is further explained in Fig. 4. All the
points in the database correspond to the same Westinghouse SVEA-96 BWR fuel
channel design, which consists of an outer square channel welded to an internal
double-walled cruciform structure, called water cross, that forms a central canal
that accommodates non-boiling water (see Fig. 5). For reference of the growth and
bow reported, the length of as-fabricated channels is 4386 mm. Details of the proc-
essing of the sheets, the fabrication of the channels, and the measurement proce-
dures can be found elsewhere [2].
The average of maximum oxide thickness of the b-quenched channel is in gen-
eral higher than that of the standard Zircaloy-2 material. A detailed review of the
database reveals that for the b-quenched material, the average is dominated by data
from one plant where factors such as neighboring channels and water chemistry
potentially play an important role of thicker oxides, but any further investigation is
not presented here.
From Fig. 2, the effectiveness of b-quenching in final size to reduce the growth
of the channels is evident. This is due to the extremely weak texture produced by
FIG. 1 Average of the max oxide thickness of all four outer channel sides of standard
and b-quenched Zircaloy-2 BWR channels that have basically no control blade
history.
b-quenching. Figure 3 illustrates how the channel bow is very similar between the
two materials up to an equivalent burn-up level of approximately 40 MWd/kgU,
above which the scatter for the standard material is much more significant than
that of the b-quenched. Reduced channel bow in b-quenched channels is a conse-
quence of lower growth.
Despite the significant benefit obtained in the b-quenched channels, there are
points whose growth may be considered high (greater than 8 mm), approaching the
FIG. 2 Average growth of all four sides of standard and b-quenched Zircaloy-2 BWR
channels. Arrows pointing to data for investigated channel A and B, respectively.
FIG. 3 Maximum bow of standard and b-quenched Zircaloy-2 BWR channels in x,y
directions. Positive bow is towards the control blades and negative bow away
from the control blades. Arrows pointing to data for investigated channel A
(6 cycles) and channel B (7 cycles), respectively.
average performance of the standard material and indicated with an arrow in Fig. 2.
This level of growth may potentially produce unacceptable channel bow, even
though this particular case showed a maximum bow of 5 mm (see arrows in Fig. 3).
These points are of interest in order to understand which factors, in addition to tex-
ture, are important to reduce channel growth. The evolution of growth of the two
b-quenched channels that produced the points indicated in Fig. 2 is shown in Fig. 6,
as a function of fast fluence, and compared to the end of life (EOL) growth of stand-
ard Zircaloy-2 and Zircaloy-4 channels. The trends in Fig. 6 suggest accelerated
growth after two or three cycles for these two Zircaloy-2 b-quenched channels. The
indicated difference in growth for the two Zircaloy-2 b-quenched channels is within
the measurement uncertainty of approximately 0.5 mm. When growth is plotted as
a function of time (in 12 months cycle plants), the growth is close to linear (see the
reference b-quenched channel in Fig. 7).
HYDROGEN CONTENT OF IRRADIATED CHANNELS

Four BWR channels were selected for oxide thickness and hydrogen content
measurements, having been irradiated in the Leibstadt nuclear power station
(abbreviated KKL, Kernkraftwerk Leibstadt), including standard and b-quenched
Zircaloy-2, as well as one standard Zircaloy-4 channel. None of these channels had
experienced control blade presence. The subject channels are identified in Table 1.
Channels A, B, and D correspond to the growth data depicted in Fig. 6 and hydro-
gen data in Fig. 8, while channel C was selected to obtain data of a b-quenched
FIG. 4 Schematic of the channel bow measurement device. The device contains a set
of ultrasonic transducers that measure the bow at several axial levels and
calculates the max measured bow as indicated in the figure. The max bow
measured is defined as the max bow between the lowest and the uppermost
transducer. To calculate the total bow, which corresponds to the channel bow
away from the dashed line, the length of the channel above and below these
outer transducers must be corrected for, and depending on this length the
correction varies from one equipment to another and from reactor to reactor.
The bow referenced in this paper is the max measured bow plus the additional
bow, i.e., total bow.
channel at an intermediate burn-up. The nominal chemical composition of the

materials is shown in Table 2.
Samples were taken at different axial levels and from the different sides of the
channels. Some samples were taken from the water cross. Oxide measurements
were performed on the samples using eddy current, while the hydrogen content was
measured using the commercial ELTRA OH 900 instrument, that employs the gas
thermal conductivity to determine the amount of released hydrogen as the sample
is heated above the melting point. The hydrogen concentration is determined as
weight ppm, relative to the total specimen weight, including the oxide. Figure 8
shows the hydrogen content of all four channels, summarized in Table 1, at different
axial positions and sides. Figure 9 shows the hydrogen pickup fraction as a function
FIG. 5 Westinghouse SVEA-96 BWR channel. The convention for identification of the
sides of the channel is indicated with numbers.
of the sum of the average of inside and outside oxide thickness. The values of oxide
thickness reported are averages between positions where the oxide was intact, inside
(towards the fuel) and outside the channel. As mentioned earlier, none of the
sampled fuel channels had any control blade exposure; thus the influence of shadow
corrosion on the samples is insignificant.
Figures 8 and 9 suggest that for the b-quenched channels, an accelerated hydro-
gen uptake occurs earlier in life, compared to the standard material, which results
in a high hydrogen content (average 500 ppm) after 7 annual cycles. Standard
Zircaloy-2 maintains a low hydrogen content (average 150 ppm) even at high burn-
up and after 7 cycles. Standard Zircaloy-4 shows an intermediate hydrogen content
(average 300 ppm) with thicker oxides, i.e., with a significantly lower hydrogen
pickup fraction than for the two Zircaloy-2 materials.
SECOND PHASE PARTICLE EVOLUTION AND DISLOCATION DENSITY

TEM examinations of standard and b-quenched Zircaloy-2 channels were con-
ducted to characterize the microstructure regarding SPP size, distribution, and
chemistry, as well as the dislocation density. Channels A, B, and C in Table 1 were
FIG. 6 Evolution of growth of two Zircaloy-2 b-quenched channels as a function of fast

fluence, compared to standard Zircaloy-2 and Zircaloy-4. The equivalent
channel fast fluence is given as average fast fluence over the channel.
included in the TEM examinations, together with two additional Zircaloy-2 stand-
ard channels, listed in Table 3, which were irradiated in another European unit
under similar operating conditions to those irradiated in KKL.
A JEOL 2100F field emission TEM operated at 200 kV was used to study elec-
tropolished samples. The microscope is equipped with two bright field detectors,
FIG. 7 Evolution of growth of two Zircaloy-2 b-quenched channels as a function of

cycles of operation.
TABLE 1 Channels subjected to oxide thickness and hydrogen content measurements.
Burnup Annual Fast Fluence < 1 MeV

Channel Material (MWd/kgU) Cycles (1021 n/cm2)
A Zircaloy-2 standard 60.4 7 13.1

B Zircaloy-2 60.3 7 13.1
b-quenched
C Zircaloy-2 45.6 5 9.45
b-quenched
D Zircaloy-4 standard 46.2 5 9.45
high angle annular dark field (HAADF) detector, electron energy loss spectrometer
(EELS), and an energy dispersive X-ray detector (EDX). Unirradiated reference
materials of both types (standard and b-quenched) were also analyzed.
Table 4 shows the summary of SPP size obtained from standard and b-
quenched Zircaloy-2, unirradiated and irradiated for 5 and 7 cycles. Figure 10 shows
the evolution of the mean and median SPP diameter for both materials, while Fig. 11
details the evolution of size distributions. It was found that for both types of mate-
rial the SPP number density decreases as a function of burn-up. The results illus-
trate how the small particles are dissolved after the first 5 cycles for both materials,
but the dissolution is more pronounced for the b-quenched material. In the as-
fabricated condition, the SPPs of the b-quenched material are on average larger
FIG. 8 Hydrogen content in Zircaloy-2 and Zircaloy-4 channels. The channel side is
indicated as (1,4) facing control blade and (2,3) away from control blade.
TABLE 2 Nominal chemical composition of Zircaloy-2 (standard and b-quenched) and standard
Zircaloy-4 (wt. %).
Sn Fe Cr Ni O Zr
Zircaloy-2 1.35 0.17 0.11 0.07 0.12 Balance

Zircaloy-4 1.30 0.20 0.09 — 0.12 Balance
than those of the standard material, and as irradiation progresses the small SPPs in
the b-quenched material are dissolved faster, leaving a population with a relatively
large size, when compared to the standard material. The mean and median SPP size
remains relatively constant in standard Zircaloy-2, while b-quenched material
exhibited an increase in both average and median SPP diameter with irradiation.
The reason for faster dissolution of SPP in the as b-quenched material would be
explained by the fact that the SPPs in b-quenched material are precipitated mostly
at the boundaries of lamellae [4] while the standard Zircaloy-2 material has the SPP
located mainly in the matrix. Since the grain boundary diffusion is several orders of
magnitude higher than bulk diffusion [5] the SPP dissolution in the b-quenched
material should be expected to be faster than for the standard Zircaloy-2 material.
Small SPPs are normally Zr(Cr,Fe), while larger SPPs are normally Zr(Ni, Fe).
The increase in mean and median SPP size with irradiation for both materials is
only apparent, in that it is caused by dissolution of the smaller Zr(Cr,Fe) SPPs as a
consequence of initial Fe depletion (the Fe/Cr ratio decreases from 0.8 to less
FIG. 9 Hydrogen pickup fraction versus sum of the average inside and outside oxide
thickness.
TABLE 3 Additional channels subjected to TEM examinations.
Burnup Annual Fast Fluence > 1 MeV

Channel Material (MWd/kgU) Cycles (1021 n/cm2)
E Zircaloy-2 standard 50.9 5 11.0

F Zircaloy-2 standard 50.9 5 11.1
than 0.2 for both materials in the first 5 cycles), followed by amorphization and
eventual Cr depletion. The dissolution process of Fe–Cr particles appears to be
completed below a burn-up of 50 MWd/kgU and/or below a fast neutron fluence of
1 1022 n/cm2. The Zr(Ni,Fe) SPPs are more stable, remaining crystalline, but
with simultaneous depletion of Fe and Ni (the Fe/Ni ratio decreases from 1.3 to
0.8 after 5 cycles for both materials). Both standard and b-quenched Zircaloy-2
exhibit amorphization of Zr(Cr,Fe) SPPs after 5 cycles (see examples in Fig. 12).
All the irradiated samples were analyzed to determine the density of hai and
hci-type dislocations. Representative images are shown in Fig. 13. In the fast neutron
fluence range studied, no significant changes in hai-type dislocation density were
observed in either material between 5 and 7 cycles. The hci-type dislocation loop
density slightly increases with fluence for the b-quenched material, while remaining
relatively constant for the standard material (see Fig. 14). The apparent increase in hci-
type dislocation loop density should be evaluated taking into account potential corre-
lation to sampling direction. Through thickness samples may not be ideal for standard
Zircaloy-2 since the data statistics is generated from samples where the dominating
TABLE 4 SPP size summary.
SPP Diameter (nm)
Annual SPP Density

Material Cycles Channel Mean Median r (1019 m3)
Zircaloy-2 0 55 42 36 8.6 6 0.7

standard
Zircaloy-2 5 E 43 39 23 4.2 6 0.5
standard
Zircaloy-2 5 F 46 38 25 3.7 6 0.4
standard
Zircaloy-2 7 A 51 48 27 1.8 6 0.3
standard
Zircaloy-2 0 65 51 44 5.0 6 0.5
b-quenched
Zircaloy-2 5 C 85 78 40 1.7 6 0.2
b-quenched
Zircaloy-2 7 B 95 89 42 1.5 6 0.2
b-quenched
FIG. 10 Evolution of SPP diameter for standard and b-quenched Zircaloy-2.
texture is aligned parallel to the hai-axis and fewer hci-type dislocation loops may be
identified compared to the samples of randomized b-quenched Zircaloy-2.
Discussion
EFFECT OF HYDROGEN ON IN-REACTOR GROWTH
The irradiation-induced growth due to anisotropy and texture in b-quenched chan-
nels should be minimal, as illustrated by the significant difference in behavior with
respect to standard material, and also by the very small (in some cases zero) growth
FIG. 11 SPP number density as a function of equivalent diameter for standard (left) and
b-quenched (right) Zircaloy-2. The 5 cycles b-quenched Zircaloy-2 data is
generated from two sides of the same channel.
FIG. 12 Arrows pointing to amorphous Fe–Cr (Fe/Cr0.2) (left) and crystalline Fe–Ni
(Fe/Ni0.9) (right) SPPs from standard (top) and b-quenched (bottom)
Zircaloy-2 channels irradiated for 5 cycles. Inserts show diffraction patterns and
dark field images of the respective SPP.
observed in early stages of irradiation (cycles 1 to 3, burn-up < 35 MWd/kgU).

However, there are cases of relatively high in-reactor growth at intermediate and
high burn-up, suggesting that in-reactor growth of b-quenched channels at these
burn-up levels is driven by other factors.
FIG. 13 Representative images of hai (left) and hci (right) type dislocations used for
density investigations.
FIG. 14 hci-type dislocation density as a function of fast neutron fluence for standard
and b-quenched Zircaloy-2.
The hydrogen measurements performed illustrate dramatic differences in

hydrogen pickup between the materials. The Zircaloy-2 b-quenched channels
examined exhibit high hydrogen content at end of life, which according to the
results, was mostly absorbed in the last two cycles.
The hydrogen content and the texture of the materials studied facilitate a direct
interpretation of the in-reactor growth observed. The total growth is the aggregate
of the growth given by volume expansion due to hydrogen uptake and the
irradiation-induced growth, i.e., the elongation produced by anisotropic displace-
ment of defects. Each of these components can be estimated.
Zirconium alloys show an approximate 1 % volume increase per 1000 ppm of
hydrogen absorbed [6]. For the channel length and thickness of the Westinghouse
SVEA-channels this translates into approximately 4 mm growth per 300 ppm of
hydrogen absorbed. Assuming that the relationship between hydrogen content and
growth due to volume expansion is linear, the expected growth for certain hydrogen
content can be estimated.
On the other hand, the irradiation-induced growth of a polycrystal in any given
direction is related to its texture through an anisotropic growth factor g:
g ¼ 1 3f
where f is the resolved fraction of basal poles in that direction [7]. Thus, the larger
the fraction of basal poles in this direction, the smaller the irradiation-induced
growth. The fraction f for standard cold-rolled material in the rolling direction, i.e.,
the axial direction of the channel is small (<0.100), while in b-quenched material
this fraction is close to 0.300. It can be assumed that the relationship between
irradiation-induced growth (given by texture) and burn-up is linear. A review of
TABLE 5 In-reactor growth of BWR channels (refer to Fig. 6).
Channel Growth (mm)
Attributable to:
Burnup
Channel Material (MWD/kgU) fA[2] H (ppm) Measured H fAa
A Zircaloy-2 51.0 0.078 150 7.8 1.5 5.8

standard
B Zircaloy-2 b- 60.3 0.267 500 8.4 6.7 1.7
quenched
C Zircaloy-4 46.1 <0.100a 300 10.4 4.0 6.4
standard
a
Estimated.
in-reactor data has established an empirical approximation of 1 mm of growth per

10 MWd/kgU for standard cold-rolled Zircaloy-2.
Table 5 summarizes the measured and estimated growth attributable to hydro-
gen uptake and to texture, the latter illustrated as the fraction of basal poles in the
axial direction (fA Þ. For example, channel A (b-quenched) had a total growth of
8.4 mm measured, 6.7 mm of which can be attributed to volume change due to
hydrogen absorption, while the remaining 1.7 mm can be attributed to irradiation
induced growth, consistent with the relatively large fA . Channel B (standard) on the
other hand, had a total growth of 7.8 mm measured, but due to the relatively low
hydrogen content, only 1.5 mm can be attributed to hydrogen absorption, while the
remaining 5.8 mm can be attributed to irradiation-induced growth, consistent with
the small fA and the empirical approximation above. It is recognized that this inter-
pretation is a simplification of the multiple mechanisms governing in-reactor
growth of channels, which most likely are a combination or irradiation growth, irra-
diation creep, thermal anisotropy, relaxation of residual stresses, and synergetic
effects between hydrogen and irradiation-induced growth. Nevertheless, the num-
bers in the table serve to illustrate the point that in-reactor growth of b-quenched
channels is dominated by volume increase caused by hydrogen uptake.
The next question is why b-quenched material has accelerated hydrogen pickup
late in life. TEM studies on irradiated Zircaloy-2 cladding, with similar composition
to that employed for channels [8] have suggested a correlation between irradiation-
induced dissolution of SPPs and an increase in hydrogen pickup late in life. The
exact mechanisms linking SPP dissolution and hydrogen pickup are still not com-
pletely clear. The TEM results reported here demonstrate that the SPPs of
b-quenched Zircaloy-2 dissolve faster than those of standard material, at least in
the first 5 cycles, which support a similar synergetic mechanism of dissolution of
SPPs and increased hydrogen pickup as reported in cladding.
HYDROGEN-ENHANCED IRRADIATION-INDUCED GROWTH

In addition to the independent effect of hydrogen uptake and irradiation-induced
growth on the total in-reactor growth, it is important to consider the synergy
between the two mechanisms. This synergy has been alluded in experiments in the
BOR-60 test reactor under the Nuclear Fuel Industry Research (NFIR) program [9],
where Zircaloy-2 shows sensitivity to enhanced irradiation growth in the presence
of hydrogen, that is, larger irradiation induced strain (related to hci loop formation)
in hydrogen charged samples, for the same composition and heat treatment condi-
tions. This is not the case for other alloys. The mechanism could be termed
hydrogen-enhanced irradiation-induced growth. Although more work directed to
find a direct correlation between hydrogen content and hci loop density is required,
the evidence collected leads to postulate a scenario for enhanced (and delayed) in-
reactor channel growth and bow. This scenario is consistent with the discussion
here and is being studied using first principles [6]. In plants with long cycles (24
months), the fuel is subjected to significant early control blade operation, which can
increase the probability of early shadow corrosion and the subsequent hydrogen
pickup. In the presence of hydrogen picked up early in life, the irradiation-induced
strain rate would be accelerated later in life. This has been explained by interaction
of interstitial hydrogen with dislocations, retarding or preventing the collapse of va-
cancy loops [6]. For the b-quenched material this appears to occur only very late in
life, as indicated by the high hydrogen content after 7 cycles, and the increase of
hci-type dislocation loops. In any case, the weakening of the texture is effective not
only to minimize normal irradiation-induced growth, but also the hydrogen-
enhanced variant.
Conclusion
From the results of in-reactor follow-up programs and hot cell examinations, the
effects of texture and hydrogen uptake on in-reactor growth of BWR channels are
differentiated. Final dimension b-quenching has been effective in considerably
reducing bulk irradiation growth and reducing bow caused by anisotropic deforma-
tion of grains. The growth of b-quenched channels has been found to be driven
mainly by volume change caused by hydrogen uptake, which in some cases offsets
the benefit of the weak b-quenched texture.
Similar to the behavior of Zircaloy-2 cladding, hydrogen uptake, and thereby
in-reactor growth of channels, accelerates at certain fluence levels. This is associated
to evolution of SPPs during irradiation. Dissolution and amorphization of small
Fe–Cr particles was observed in standard and b-quenched Zircaloy-2 channels,
coupled with delayed dissolution of larger Fe–Ni particles. Dissolution of SPPs and
the associated acceleration of hydrogen pickup is more pronounced in b-quenched
material, which is attributed to the easier diffusion of alloying elements from SPPs
precipitated at boundaries of lamellae in the b-quenched structure.
In addition to the separate effect of hydrogen uptake and texture on the total
in-reactor channel growth, the synergy between the two appears to play an
important role. It has been observed elsewhere that irradiation-induced growth is

enhanced by the presence of hydrogen, and the results reported here do not pre-
clude the presence of this mechanism late in life in b-quenched Zircaloy-2 BWR
channels. However, the weakening of the texture controls not only the normal
irradiation-induced growth, but also any hydrogen-enhanced variant.
The knowledge of the importance of hydrogen content for in-reactor growth of
BWR channels provides a path forward on the development of materials. Minimizing
hydrogen uptake by optimizing SPP sizes and composition would reduce the bulk
volume expansion, as well as any hydrogen-enhanced irradiation induced growth.
References
[1] Christensen, M., Wolf, W., Wimmer, E., Adamson, R. B., Hallstadius, L., Cantonwine, P.,
“Effect of Hydrogen on Dimensional Changes of Zirconium and the Influence of Alloying
Elements: First Principles and Classical Simulations of Point Defects,” Zirconium in the
Nuclear Industry: 17th International Symposium, ASTM STP 1543, Robert Comstock and
Pierre Barbaris, Eds., ASTM International, West Conshohocken, PA, 2014, pp. 55–92, STP
in progress.
[2] Dahlbäck, M., Limbäck, M., Hallstadius, L., Barberis, P., Bunel, G., Simonot, C., Andersson,
T., Askeljung, P., Flygare, J., Lehtinen, B., and Massih, A. R., “The Effect of Beta-
Quenching in Final Dimension on the Irradiation Growth of Tubes and Channels,”
J. ASTM Int., Vol. 2, No. 6, 2005, JAI12337.
[3] Fidleris, V., “The Irradiation Creep and Growth Phenomena,” J. Nucl. Mater., Vol. 159,
1988, pp. 22–42.
[4] Tenckhoff, E., Deformation Mechanisms, Texture and Anisotropy in Zirconium and
Zircaloy, ASTM International, Philadelphia, PA, 1988.
[5] Valizadeh, S., Ledergerber, G., Abolhassani, S., Jädernäs, D., Dahlbäck, M., Mader, E.,
on the In-Reactor Performance of BWR Cladding,” J. ASTM Int., Vol. 8, No. 2, 2011,
JAI103025.
[6] Mader, E. V., Reitmeyer, M., Garcia Sedano, P., Morris, J., Cantonwine, P., Hallstadius, L.,
et al., “EPRI BWR Channel Distortion Program,” Proceedings of the Water Reactor Fuel
Performance Meeting, Chengdu, 2011.
[7] Romero, J., Preuss, M., Quinta da Fonseca, J., Comstock, R., Dahlbäck, M., and Hallsta-
dius, L., “Texture Evolution of Zircaloy-2 During Beta-Quenching: Effect of Process Varia-
bles,” J. ASTM Int., Vol. 7, No. 9, 2010, 103014.
[8] Sabol, G., “ZIRLO—An Alloy Development Success,” J. ASTM Int., Vol. 2, No. 2, 2005,
12942.
[9] Yang, W. Y., Tucker, R. P., Cheng, B., and Adamson, R. B., “Precipitates in Zircaloy: Identi-
fication and the Effects of Irradiation and Thermal Treatment,” J. Nucl. Mater., Vol. 138,
Nos. 2–3, 1986, pp. 185–195.
DISCUSSION
Q1:—How does the percent pick-up of hydrogen vary for the b-quenched
material when the acceleration in growth happens?
Authors’ Response:—At the acceleration in growth after three cycles, an increase

in hydrogen pick-up occurred. As shown in Figure 9, the pickup fraction for the
beta quenched material increased from 20% to approximately 40%.
Q2:—Does the dissolution of SPP accelerate the growth for non-quenched,

standard material?
Authors’ Response:—Yes, the acceleration in growth for Zircaloy-2 alpha and

b-quenched material is most likely related to SPP dissolution that results in an
increased hydrogen pick-up fraction.
Question from Antoine Ambard, EDF:—Could you detail the way you get a vol-
ume expansion of 1% for 1000 ppm? Is it an experimental value, a calculated value
and how does it compare to the results already published (Kesterson, R. et al.,
“Impact of hydrogen on dimensional stability of fuel assemblies,” Proceedings of
Light Water Reactor Fuel Performance Meeting, sponsored by ANS, Park City, UT,
10–13 April 2000.)
Authors’ Response:—The volume expansion of 1% for 1000 ppm H is calculated

and based upon the volume expansion of metal to hydride as described by Kester-
son et al.
Question from S. K. Jha, NFC, Hyderabad, India:—What is the effect of SPP

size on fuel clad performance? What stage of processing should b-quenching be
done for manufacturing of fuel clad tube?
Authors’ Response:—The effect of SPP size in cladding is similar to that in

channel material with small SPPs having faster dissolution and consequently, earlier
increased hydrogen pickup. (Tägtström et al., “Effects of Hydrogen Pickup and
Second-Phase Particle Dissolution on the In-Reactor Corrosion Behavior of BWR
Claddings,” Zirconium in the Nuclear Industry: Thirteenth International
Symposium, ASTM STP 1423, G. D. Moan and P. Rudling, Eds., ASTM Interna-
tional, West Conshohocken, PA, 2002, pp. 96–118.) Westinghouse BWR
standard cladding, designated LK3, is normally b-quenched at billet size before
extrusion.
Question from B. K. Shah, BARC, Mumbai:—You have reported higher in-

reactor growth in Zircaloy-2 BWR channels. What has been the experience with
respect to Zircaloy-4 BWR channels?
Authors’ Response:—Zircaloy-4 alpha BWR channels have higher elongation,

bow, and corrosion than Zircaloy-2. That was the reason for the development of
Zircaloy-2 instead of Zircaloy-4 for channel material.
Question from Juan M Garcia-Infanta, ENUSA, Spain:—The author has shown

that hydrogen pick-up in Zircaloy-2 b-quenched channels is much higher than in
non b-quenched (standard) Zircaloy-2 channels. Is there an effect on the shadow
corrosion induced bow of the channels?
Authors’ Response:—The channels did not have any significant impact from
shadow corrosion since all data were from 12-month cycles with no association
with control blades early in life.
Questions from Kumar Vaibhaw, NFC, Hyderabad:—From the manufacturing

point of view, can you describe how the b-quenching was performed (e.g., in vac-
uum or inert atmosphere followed by water or gas cooling?) If air-cooling, micro-
structure would not be completely uniform.
Authors’ Response:—The process for fabrication of the SVEA sheet material in

1.48 mm and 0.88 mm thickness is given in Reference 2.
Questions from S. Anantharaman, BARC, Mumbai
Q1:—What is the service temperature of the channels?
Authors’ Response:—The service temperature of channels is 286 C.
Q2:—What was the morphology in the oxide layer and its thickness at the
examined locations? Could the higher hydrogen content be the result of spalling of
the oxide at the location examined?
Authors’ Response:—The oxide morphology was not examined and the oxide
thickness was <60 lm; no spalling occurred.
Q3:—What was the distribution of hydrogen content along the channel height?
Authors’ Response:—The hydrogen content was mainly taken at three eleva-

tions: 1500, 2500 and 3500 mm from the bottom. Similar hydrogen contents were
measured at all three elevations.
Q4:—Was the density of the sample measured during the PIE?
Authors’ Response:—The density was not measured during the PIE.
Q5:—Did the location of the channel in the core affect the observed growth?
Authors’ Response:—Correlation of the observed growth to the core position

was not evaluated as only a single channel at high burn-up was examined.
Q6:—Why not change over to Zircaloy-4 instead of Zircaloy-2?
Authors’ Response:—The in-pile data for Zircaloy-4 alpha material has higher
corrosion, elongation and bow compared to Zircaloy-2 alpha material. The evalua-
tion of b-quenched Zircaloy-4 is still under irradiation in another plant.
Questions from Ron Adamson, Zircology Plus
Q1:—Could you provide details about the composition of the SPPs existing in
the “BQ” material after annealing? Also, provide details about the annealing time/
temperature to achieve the microstructure. The dissolution behavior of the SPPs
may depend on these details.
Authors’ Response:—The smaller SPPs (Zr(Fe,Cr)2) had an Fe/Cr ratio of 0.8

while the larger SPPs (Zr2(Fe,Ni)) had an Fe/Ni ratio of 1.4.
The cumulative effect of intermediate heat treatments after the b-quenching is

described by an annealing parameter, A, given by:
A ¼ Ri ti expðQ=RTi Þ
Where
ti ¼ annealing time in hours

Ti ¼ annealing temperature in K
Q ¼ activation energy = 63000 cal/mole
R ¼ gas constant
The heat treatments after the b-quenching gave an approximate log A ¼ 14,4.
Q2:—Was the texture measured after the post-quench anneal?
Authors’ Response:—The texture after alpha annealing was not changed since
no recrystallization occurred. The b-quenched structure remained. The only effect
of the alpha annealing was the growth of the SPP’s in the alpha lamella. All texture
measurements (Reference 2) are at final delivery condition, i.e., after b-quenching
and alpha annealing of the sheets. The texture is close to a Kearns factor of 0.3.
Questions from R. K. Chaube, NFC, Hyderabad
Q1:—Does in-reactor growth and bow also depend on the fabrication route of
square channels, like pilgered channels?
Authors’ Response:—The channel fabrication does not contain any major cold
work operations, only normal bending of channel ‘U-half’ is made and welded
together.
Q2:—What is the specification of Zircaloy-2 channels for bow as compared to

1.8 mm for Zircaloy-4 channels in as-fabricated finished square channels?
Authors’ Response:—The specified maximum of as fabricated bow is not de-

pendent on material (Zircaloy-2, Zircaloy-4, etc.).
Q3:—How is b-quenched structure retained after welding of sheets for square

channel?
Authors’ Response:—The b-quenched structure is only affected at the weld with

an additional local b-quenching. The impact on channel performance is negligible.
Q1:—Did you observe c-loops during the accelerated growth? Is there any dif-
ference between bQ and standard channels? Is there any correlation with the SPP
dissolution?
Authors’ Response:—c-loop formation seems to increase with increased fluence,

for both alpha and b-quenched material and the difference is that c-loop density
seems to be slightly higher for b-quenched material than alpha material. If this is an
artefact from measurement due to more randomly oriented grains for b-quenched
material or a real value is not finally concluded from this examination. Since c-loop
formation correlates to fluence and the accelerated growth is occurring late in life at
high fluence, it is not possible to separate the effects of SPP dissolution and c-loop
formation. However since the c-loop formation seems to be linear with fluence and
the SPP dissolution also has a fluence dependency, it is difficult with a few measure-
ments to identify the root cause for the acceleration growth and increased hydrogen
pick-up.
Q2:—You did not take into account the growth that is induced by the oxide
stresses. Their effect could be different from bQ to standard materials due to their
likely difference in mechanical (creep) properties. It could impact the conclusion
that H affects the growth.
Authors’ Response:—The oxide thickness was fairly low (20–45 lm and

20–60 lm at 45 and 60 MWd/kgU, respectively) for both b-quenched channels and
it is not possible to correlate oxide formation to the overall channel growth. How-
ever a good correlation to hydrogen content and channel growth was concluded.
Questions from Ted Darby, Rolls Royce:—You were able to account for the
observed growth strains by a single summation of effects due to irradiation growth
and hydrogen. Other data (e.g., King, S. J. et al., “Impact of Hydrogen on Dimen-
sional Stability of ZIRLO Fuel Assemblies,” Zirconium in the Nuclear Industry:
Thirteenth International Symposium, ASTM STP 1423, G. D. Moan and R. Rudling,
Eds., ASTM International, West Conshohocken, PA, 2002, pp. 471–489) have sug-
gested an unexplained component of strain is greater than the summation of the
two components. Please comment on these diverse observations. Is it necessary to
invoke a mechanistic synergy between irradiation growth and hydrogen?
Authors’ Response:—Significant hydrogen pickup early in life is known to

enhance later-life irradiation growth, i.e., a synergy between irradiation growth and
hydrogen. This is one hypothesis regarding the origin of shadow corrosion driven
channel bow. However, since no significant early hydrogen pick-up occurred for
the Zircaloy-2 alpha material (a total of only 150 ppm at end of life was measured),
the Zircaloy-2 alpha material is understood to have had no or negligible impact of
the irradiation assisted growth phenomenon that was observed by King et. al. The
b-quenched channel growth correlates well to the hydrogen content within the
measurement accuracy, leaving little room for any synergistic effects.
Questions from Jean-Christophe Brachet, CEA, France:—Measurements of

growth are performed at room temperature (RT). In fact, at in-service temperature,
part of hydrides will be dissolved and this will induce two opposite effects on the
volumetric expansion:
Decrease of volume due to dissolution of hydrides
Increase of volume due to hydrogen atoms going into solid solution which
will increase the aZr matrix lattice parameter.
So, the extrapolation of the volumetric expansion measured at RT to in-service

temperature has to take into account these two opposite effects, and the effect of
crystallographic texture when comparing standard and b treated materials.
Authors’ Response:—Since the hydride formation as well as hydrogen going

into solid solution results in an isotropic expansion of the hcp-lattice i.e. 1/3 in each
direction, consequently no difference in volume expansion between an alpha or b-
quenched material will occur. Furthermore, state-of-knowledge implies that dis-
solved hydrogen has the same impact on volume as does hydrogen in the form of
hydrides.
L. Tournadre,1 F. Onimus,2 J.-L. Béchade,1 D. Gilbon,1

J.-M. Cloué,3 J.-P. Mardon,3 and X. Feaugas4
Impact of Hydrogen Pick-Up and

Applied Stress on C-Component
Loops: Toward
a Better Understanding of the
Radiation Induced Growth of
Recrystallized Zirconium Alloys
Reference
Tournadre, L., Onimus, F., Béchade, J.-L., Gilbon, D., Cloué, J.-M., Mardon, J.-P., and Feaugas,
X., “Impact of Hydrogen Pick-Up and Applied Stress on C-Component Loops: Toward
a Better Understanding of the Radiation Induced Growth of Recrystallized Zirconium Alloys,”
ABSTRACT
Under neutron irradiation, recrystallized zirconium alloys, used as structural
materials for Pressurized Water Reactor (PWR) fuel assemblies, undergo stress-
free growth which accelerates for high irradiation doses. This acceleration is
correlated to the formation of c-component vacancy dislocation loops lying in
the basal plane. The growth behavior observed on some PWR fuel assemblies
suggests that a macroscopic stress applied under irradiation could affect the c-
loop microstructure and therefore influence the subsequent stress-free growth.
In addition, some feedbacks show that in-service hydrogen pickup could also
Manuscript received December 13, 2012; accepted for publication November 3, 2013; published online June
19, 2014.
1
CEA, DEN, Service de Recherches Métallurgiques Appliquées (SRMA), Gif-Sur-Yvette, France 91191.
2
CEA, DEN, Service de Recherches Métallurgiques Appliquées (SRMA) Gif-Sur-Yvette, France 91191
(Corresponding author), e-mail: fabien.onimus@cea.fr
3
AREVA AREVA NP Fuel BusinessUnit10, Rue Juliette Recamier, 69456 Lyon Cedex, France.
4
LaSIE, FRE CNRS 3474, Univ. de La Rochelle, 17042 La Rochelle Cedex 1, France.
5
Hyderabad, India.
Copyright V
influence the fuel assembly radiation-induced deformation. The impact of an

applied stress on c-loops has been studied on as-received recrystallized (RXA)
Zircaloy-4 irradiated with 300 keV and 600 keV Zr ions at 573 K. An original
device, specifically designed for the irradiation facility, applies a stress which
relaxes during the early stage of irradiation. In these conditions, observations by
Transmission Electron Microscopy (TEM) reveal no effect of the initial applied
stress on c-loops. However, when the stress is applied when c-loops are already
created, c-loop growth is affected. Moreover, 2 MeV proton irradiations
performed on as-received and pre-hydrided M5 (M5 is a trademark of AREVA NP
registered in the USA and in other countries.) and RXA Zircaloy-4 materials were
conducted at 623 K. For both grades, the c-component loop evolution with dose
is examined and compared to neutron irradiated microstructures. As for neutron
irradiated materials, the c-loop density in Zircaloy-4 is higher than in M5. For
pre-hydrided RXA Zircaloy-4 the effect of hydrogen is not significant since the
loops are already present in high density. On the other hand, TEM observations
on pre-hydrided M5 highlight that c-loop density is higher, far from hydrides,
than without pre-hydriding. Furthermore, a high c-loop density is observed in
the surrounding of hydrides partially or fully dissolved. To explain these
observations, the role played by hydrogen in solid solution and as precipitated
hydrides is discussed.
Keywords
zirconium, growth, c-loops, hydrogen, stress, ion irradiation
Introduction
Zirconium alloys are used as cladding and structural material for Pressurized Water Re-
actor (PWR) fuel assemblies. During in-reactor normal operation, the length of the
PWR fuel assembly evolves under irradiation. This dimensional change is the result of
three main different phenomena: the thermal creep, the irradiation creep and the stress
free growth [1–3]. The stress-free growth phenomenon corresponds to an elongation
along the basal plane of the hexagonal close packed structure and a shortening along
the c-axis, the volume remaining constant [4]. Due to the marked texture of zirconium
tubes [5], the growth of the hexagonal close packed grains, at the microscopic scale,
leads to a macroscopic elongation along the axial direction of the tubes. For recrystal-
lized zirconium alloys, at high fluence, typically after an incubation dose of
4.0 1025 n m2 (corresponding to 6 dpa, E > 1 MeV) [6], the growth rate increases.
This acceleration is often referred to as “breakaway growth.” Some feedbacks from
industry [7,8] show that external parameters like an applied stress or in-service hydro-
gen pickup could also increase the resultant deformation after the “breakaway growth.”
Since the growth acceleration can have a significant impact on the performance of the
fuel assembly, this phenomenon has to be well understood and predicted.
According to several authors, the breakaway growth is clearly correlated to the
nucleation at high doses of a specific microstructural irradiation defect: the
c-component dislocation loops [9]. C-component loops have already been observed
TOURNADRE ET AL., DOI 10.1520/STP154320120200 855
after neutron irradiations [10,11], but also after charged particle irradiations con-
ducted with electrons [12–14], heavy ions [15–17], and protons [15]. C-loops are of
vacancy type only and are located in the basal plane. When they are observed after
neutron irradiation by Transmission Electron Microscopy (TEM), they are faulted
and their Burgers vector, given by b ¼ 1=6h2023i, has a component along the c-
axis [11]. Whatever the irradiation conditions, these c-component loops, which
appear after an incubation dose, are always present in conjunction with more
numerous and finer a-loops, which appear from low irradiation dose. The c-
component loops are much larger than the a-loops, but their number density is
much lower. For instance, for recrystallized Zircaloy-2 and Zircaloy-4 (also referred
as RXA Zy-4) irradiated at 573 K, after 5.4 1025 n/m2, c-component loops are
found with a diameter of 120 nm and with a number density between 3 and
6 1020 m3 whereas the a-loops density saturates at a value of 3 1022 m3, from
a relatively low fluence of approximately 2 1024 n/m2 [3] with a diameter of about
7 nm. Due to the high density of small a-loops, the c-component loops can only be
observed edge-on by TEM by using the g ¼ 0002 diffraction vector, which leads to
invisible a-type defects. The c-loops thus appear as straight-line segments.
It is rather surprising that although the most stable loops in zirconium alloys
are the prismatic loops, in agreement with the Foll and Wilkens criterion
p
(c=a < 3) [18], basal loops are also observed. Moreover, these loops are of the va-
cancy character. According to the usual rate theory, vacancy loops should not grow
as a result of the bias of edge dislocation toward self-interstitial atoms (SIA). The
reason for the nucleation and growth of c-component loops in zirconium alloys has
been analyzed and discussed in great detail by Griffiths and co-workers [11,19,20],
but this phenomenon is still not well understood. It has been shown by Molecular
Dynamics computations for a-zirconium [21] that most of the small interstitial
clusters produced in the displacement cascade have the form of a dislocation loop
with Burgers vector ¼ 1=3h1120i. Small vacancy clusters are also found in the pris-
matic plane. On the other hand, vacancy clusters in the basal plane form a hexago-
nal loop enclosing an extrinsic (E) stacking fault with 1=2h0001i Burgers vector. As
reviewed by Hull and Bacon [22], condensation of vacancies in a single basal plane
results in two similar atomic layers coming into contact. This unstable situation of
high energy is avoided in one of two ways described by Berghezan et al. [23] imply-
ing the glide of partial dislocations in the platelet plane. The first process leads to
the creation of a sessile Frank partial with 1=2h0001i Burgers vector surrounding a
high energy extrinsic (E) stacking fault disk. The second process leads to the crea-
tion of a sessile Frank partial 1=6h2023i Burgers vector surrounding the type I1 low
energy intrinsic stacking fault. The energy of a faulted edge loop with radius R can
expressed as Eq 1 according to the elastic theory [24]. A more comprehensive treat-
ment accounting for the character of the loop can be derived easily.

lb2 4R
(1) Eloop ðRÞ ¼ Eline þ Efault ¼ 2pR ln 1 þ pR2 cfault
4pð1 Þ r0
where:
l ¼ shear modulus,
¼ Poisson’s ratio,
b ¼ Burgers vector,
r0 ¼ core radius, and
cfault ¼ stacking fault energy.
From this formula, it can be seen that, for a small loop, the energy of the dislo-
cation line dominates; however, as the loop grows, it is the energy of the stacking
fault which dominates. As suggested by Hull and Bacon [22], since the E-type c-
loop has a higher stacking fault energy than the I1-type c-loop
(cE 249 mJ=m2 > cI1 124 mJ=m2 [25]) as the loop grows, it is likely to trans-
form into the I1-type c-loop for a critical loop size, in a similar manner to the
unfaulting of loops in face-centered cubic metals. The critical loop size for the
transformation from E-type loop to I1-type loop is likely to be influenced by stress
or impurity content according to Hull and Bacon [22] since it can affect the stack-
ing fault energy.
There is indeed considerable evidence that the formation of c-component loops
is dependent on the purity of the zirconium used [11,14,20,26]. This can probably
be attributed to the effect of solute elements on the basal stacking-fault energy. It is
also possible that small impurity clusters, especially iron in the form of small basal
platelets, could act as nucleation sites for these loops [14,27].
However, according to Griffiths [11], this cannot account alone for the very
large vacancy c-component loops observed, since the growth of vacancy loops is
not favorable considering the elastic interaction difference between dislocation and
SIAs or vacancies. In order to understand the reason for the important growth of
the c-component loops, another mechanism must occur. As discussed by Woo [28],
the growth of c-component loops is well understood in the frame of the difference
in anisotropic diffusion model. Indeed, because of the higher mobility of SIAs in
the basal plane rather than along the c-axis (and the quasi-isotropic diffusion of
vacancies), dislocations parallel to the c-axis will absorb a net flux of SIAs whereas
dislocations in the basal plane will absorb a net flux of vacancies. This can therefore
explain why the basal vacancy loops can grow.
The origin of the incubation period before the appearance of c-component
loops is not clearly understood. Griffiths et al. [6] explain that the c-loop formation
is dependent on the volume of the matrix containing a critical interstitial solute
concentration. This volume increases as the interstitial impurity concentration is
gradually supplemented by the radiation-induced dissolution of elements such as
iron from intermetallic precipitates [14,19,29,30].
According to our current knowledge, there is no experimental evidence in the
literature showing any effect of the stress on c-loops in zirconium alloys. However,
there are many papers theoretically describing the effect of stress on loops, espe-
cially for Face-Centered Cubic metals despite the scarce experimental evidences, all
dating back from the 1970 s, of this phenomenon [31–33]. Two main mechanisms
are described in the literature to explain the effect of an applied external stress on
irradiation induced dislocation loops: the Stress Induced Preferential Nucleation
(SIPN) mechanism and the Stress Induced Preferential Absorption mechanism
(SIPA).
The SIPN mechanism is based on the assumption that the stress enhances
the loop nucleation depending on the loop orientation with respect to the applied
stress [34–36]. The other mechanism is the SIPA mechanism. According to this
mechanism, the stress influences the absorption bias between a point defect and
a cluster [37,38]. Garner et al. [31] have already studied the effect of an applied
stress on Frank loops observed after irradiation in austenitic steels. According to
these authors, the Frank loop densities are influenced by the deviatoric stress
component normal to the loop habit plane. Indeed, these authors show that
depending on their orientation with respect to the applied stress, biaxial stress in
their case, in some planes the interstitial Frank loop density is higher than in
unstressed material and in other planes the loop density is lower than in the
unstressed material.
It is known that in-reactor hydrogen pickup induces elongation of zirconium
alloys tubes [39]. However, as it is discussed in Ref. [40], due to the very low
hydrogen content usually picked up in-service in M5 tubes (around 100 wppm),
the amount of growth due to hydrogen can be neglected. In the case of M5, the
accelerated irradiation growth is therefore only attributed to c-component loops.
To our knowledge, there is yet no experimental evidence of any effect of hydro-
gen on c-loops. Only indirect evidence can be found in the literature concerning
the interaction between hydrogen and irradiation defects in zirconium alloys in
general. Indeed, several authors [41–43] have shown by using differential scan-
ning calorimetry (DSC) experiments that hydrogen solubility limit is higher for
neutron irradiated samples than for non-irradiated samples. Moreover, synchro-
tron X-ray diffraction measurements [44] were performed on neutron irradiated
Zircaloy-4 in order to determine the number of precipitated hydrides. A heat
treatment at 873 K has been done in order to anneal out the irradiation loops.
Diffraction measurements show that the hydrides density increases after post
irradiation annealing, suggesting that the hydrogen solubility limit decreases
when the loops are annealed out. These authors show, for instance, that at a
temperature around 573 K, the solubility limit increases from 70 to 180 wppm
and at 598 K this solubility limit increases from 100 to 210 wppm due to irradia-
tion. These results imply that hydrogen atoms are trapped on irradiation defects.
Especially, for high irradiation doses, the authors [41–43] assume an interaction
between hydrogen atoms and c-loops. This is also suggested by high temperature
post-irradiation annealing and synchrotron analysis [42] where c-component
loops are more stable and remain traps for hydrogen atoms after annealing. Fur-
thermore, Chung et al. [45] have observed the precipitation of micrometric delta-
hydrides aligned with c-loops in the basal planes. This observation suggests an
interaction between nano-hydrides and c-loops.
As for the influence of stress on c-loops, there is no experimental result in the

literature, to our knowledge, that shows an increase of c-loop density with the
hydrogen content in the material.
In order to investigate the role of an applied stress and the influence of hydro-
gen pickup on c-loops, charged particle irradiations have been performed. Previous
studies [15] have shown that 600 keV Zr ion irradiations conducted at 573 K and 2
MeV proton irradiations conducted at 623 K permit to obtain c-loop. Starting from
these conditions we have conducted several Zr ion irradiations under an applied
stress, using a specific device especially designed for this purpose, on RXA Zy-4 and
M5. Furthermore, 2 MeV proton irradiations have been conducted on pre-hydrided
and as-fabricated RXA Zy-4 and M5 samples [46].
In the first part of the paper, the materials, used in as-received and pre-
hydrided materials conditions, are described. Then the charged particles irradia-
tions and the device to apply a stress on the samples, under irradiation, are detailed.
Eventually, the optical microscopy technique and the TEM analysis method are
explained with the associated sample preparation. In the second part, the results
concerning the influence of an applied stress and the impact of hydrogen are
shown. The results are finally discussed in the last part of the paper, considering the
nucleation and growth process of point defect clusters in metals under irradiation.
AS-RECEIVED MATERIALS
The materials studied are industrial recrystallized zirconium alloys: Zircaloy-4
(RXA Zy-4) and M5. The chemical composition is given in Table 1. As in previous
studies [15,46], the specimens are taken from an intermediate product (in the form
of thick tube), which exhibits a basal plane transverse texture (Fig. 1). The Kearns
factors of both tubes are given in Table 2. The final manufacturing stages of TREX
include a classical pilgering and recrystallization annealing cycle. The last heat treat-
ment (about two hours at temperatures close to 853 K) produced a recrystallized
grain structure with a grain size of 6.2 6 0.6 lm for RXA Zy-4 and 5.3 6 0.4 lm for
M5. The microstructures obtained are rather similar to the microstructures of
recrystallized thin tubes. The second phase precipitates are the same as the ones
usually found in thin tubes made of M5 and RXA Zy-4. Chemical analyses of the
thick tubes have also shown a very good chemical homogeneity in the thickness of
the tube.
TABLE 1 Chemical composition of recrystallized Zy-4 and M5TM intermediate products (wt. %).
Alloy Sn % Fe % Cr % S% Nb % O%
RXA Zy-4 1.35 0.21 0.10 0.0020 – 0.13

M5 – 0.0370 – 0.0017 1.0 0.14
FIG. 1 (0002) pole figures (a) RXA Zy-4, (b) M5.
PRE-HYDRIDED MATERIALS
In order to study the effect of hydrogen on c-loops, RXA Zy-4 and M5 samples
have been hydrided up to various hydrogen contents. Typical hydrogen contents,
representative of in-reactor hydrogen pickup of 80 wppm for M5 [47] has been
considered here. Furthermore, in order to enhance the potential effect of hydrogen
on c-loops, two higher hydrogen contents (125 and 350 wppm) have also been
studied. Concerning RXA Zy-4, the samples have been pre-hydrided up to 185
wppm. Hydriding has been performed under Ar-5 %H gas at 673 K in several steps
for the highest hydrogen contents.
PROTON IRRADIATIONS AND SAMPLE PREPARATION

2 MeV proton irradiations were carried out on the Michigan Ion Beam Laboratory
(MIBL/University of Michigan) Tandem accelerator [48] at 623 K under high vac-
uum up to four different doses. The irradiation temperature of 623 K has been cho-
sen, in agreement with the experiment done by Zu et al. [49], to simulate the in-
reactor irradiation temperature of a guide tube, which is typically between 573 and
TABLE 2 Kearns factors of recrystallized Zy-4 and M5 intermediate products.
Alloy fL fT fR
RXA Zy-4 0.09 0.53 0.38

M5 0.09 0.49 0.42
593 K. This temperature shift aims at accounting for the increase in damage rate
during proton irradiation compared to neutron irradiation. The high deposited
power per unit surface by the proton beam (4.8 105 W/m2) requires the use of liq-
uid indium at the back of the samples, to extract the deposited heat during irradia-
tion. Furthermore, the temperature is controlled by a 2D thermal camera
throughout the experiment in addition to usual thermocouples welded on the sam-
ples outside of the beam.
In spite of damage rates higher than for neutrons, several days are required to
reach high irradiation doses. The TRIM calculation (Fig. 2) shows that the damage
peak is obtained between 25 and 30 lm depth, whereas the damage increases only
slowly between 5 up to 20 lm depth. The samples are in the form of bars of 20 mm
long, 2 mm wide, and 1.55 mm thick. The bars are taken in the axial direction
(AD)–transverse direction (TD) plane, the long direction of the bar being along the
AD direction. The surface (AD–TD plane) is polished up to a mirror finish before
irradiation. After irradiation, the bars are mechanically polished on the back side
down to a thickness of 0.1 mm. Then, 3 mm diameter disks are punched out of the
thin strip and usual two-side electro-polishing is performed to remove around 15
lm from the irradiated surface. This surface is then protected with a varnish and
the final electro-polishing is performed on the rear surface. Thanks to this specific
preparation method, a fully irradiated bulk material is observed by TEM.
All the thin foils were prepared by electro-polishing using a solution of 20 % 2-
Butoxyethanol and 10 % perchloric acid in ethanol at temperatures around 278 K.
Eight samples can be fully irradiated at the same time during one proton irradi-
ation, and since the irradiated part of the bar is 10 mm long, three thin foils can be
taken out of each bar.
FIG. 2 TRIM calculations for 2 MeV protons in Zr with a dose of 8:9 1023 protons/m2.
Zr displacement energy is considered equal to 40 eV. (a) Damage profile in dpa;
(b) implanted hydrogen atoms distribution.
It is worth pointing out that, as observed in the previous study [15], the bars af-
ter mechanical polishing from the rear surface (100 lm thin strip) systematically
exhibit a macroscopic bending, which is due to the irradiation induced growth
strain of the 30 lm thick irradiated layer, the bending increasing with irradiation
dose.
Proton irradiations were performed on as-received RXA Zy-4 and M5 samples
(control samples) and pre-hydrided samples. Irradiation conditions for control
samples are given in Table 3. Irradiation conditions for pre-hydrided samples are
given with the hydrogen content in Table 4.
During proton irradiations, hydrogen atoms are implanted into the zirconium
alloys. Considering the high diffusivity of hydrogen in zirconium at 623 K (diffu-
sion coefficient of 1:2 1010 m2 =s [50]) the hydrogen atoms are probably homo-
geneously distributed in the whole bar (volume of 62 mm3). Assuming that there is
no desorption, the amount of implanted hydrogen can be estimated for each irradi-
ation dose reached. These values are given in Tables 3 and 4.
ZR ION IRRADIATIONS AND DEVICE FOR APPLYING A STRESS UNDER

IRRADIATION
The 300 keV and 600 keV Zr ion irradiations were conducted, respectively, on the
IRMA and ARAMIS facilities at CSNSM/IN2P3-Orsay [51–53] at 573 K. At these
energies, the heavy ion flux provides a high damage rate in the irradiated layer, and
therefore, a high irradiation dose in only one day. The damage profiles created by
300 keV and 600 keV Zr ions are shown in Fig. 3. It can be seen that contrary to
proton irradiation, the irradiated layer is very thin (between 150 to 300 nm thick)
and only a thin part of the 6 lm grain is irradiated close to the free surface. The
irradiated material cannot be considered as bulk material in that case, and free
TABLE 3 Proton irradiation conditions conducted on as-received M5 and RXA Zy-4 samples (con-
trol samples).
Dose
Damage Flux Duration Estimated

Material dpa ions/cm2 T (K) Rate (ions/cm2/ (h) Implanted
(dpa/s) s) Hydrogen
wppm
M5 4.9 3.50 1019 623 1.50 105 1.07 1014 90 29

19 5
8.1 5.78 10 1.64 10 1.17 1014 150 46
12.5 8.93 1019 2.02 105 1.44 1014 200 79
20 5
19 1.36 10 1.80 10 1.28 1014 266 107
Zy4 8.1 5.78 1019 623 1.64 105 1.17 1014 150 46
19 5
12.5 8.93 10 2.02 10 1.44 1014 200 79
19 1.36 1020 1.80 105 1.28 1014 266 107
TABLE 4 Proton irradiation conditions conducted on pre-hydrided M5 and Zy-4 samples.
Dose
Initial Estimated
Hydrogen Implanted Damage Flux
Content Hydrogen Rate (ions/ Duration
2
Material dpa ions/cm (wppm) wppm T (K) (dpa/s) cm2/s) (h)
M5TM 4.9 3.50 1019 350 29 623 1.50 105 1.07 1014 90
8.1 5.78 1019 350 46 1.64 105 1.17 1014 150
12.5 8.93 1019 350 79 2.02 105 1.44 1014 200
19 1.36 1020 350 107 1.80 105 1.28 1014 266
19 1.36 1020 80 107 623 1.9 105 1.36 1014 266
125
350
Zy-4 19 1.36 1020 185 107 623 1.9 105 1.36 1014 266
surface effects, such as point defects elimination on the surface, are believed to be
significant.
In order to apply an external stress on the samples, an original device, adapted
for both facilities, has been designed (Fig. 4). The samples are in the form of thin
strips of 25 mm long, 1.9 mm wide, and 0.1 mm thick. A four-point bending load-
ing is applied on the samples (Fig. 4). This loading geometry induces a pure shear
stress state between the two central supporting points, inducing a pure tensile stress
state on the thin irradiated layer of the samples. The stress level is deduced from
the thickness of the sample h, its length L, the distance between the two central
points, l, and the imposed displacement, u, between outer loading points and the
inner loading points from Eq 2
FIG. 3 TRIM calculations (a) for 300 keV Zr ions in Zr 11.5 1014 ions/cm2, (b) for 600
keV Zr ions in Zr, 11.5 1014 ions/cm2 (Ed ¼ 40 eV).
FIG. 4 Original device designed to apply a four-point bending loading under Zr ion
irradiations. (a) Principle of the four bending loading, (b) picture of the device.
6uEh
(2) rirradiated area ¼
ðL l ÞðL þ 2l Þ
Sixteen samples can be irradiated at the same time, and various stress levels can be
applied on the samples, thanks to modular central blocks with various displace-
ments u. For each irradiation, several control samples are irradiated exactly in the
same conditions but without any applied stress. The results obtained on samples
irradiated under stress can be therefore always compared to control samples.
Despite the poor thermal contact between the heating stage and the samples, it
has been checked, using a thermocouple welded on a sample, that for a given heat-
ing power, the irradiation temperature on the samples is 573 K. Furthermore, due
to the very low power deposited per unit surface by the Zr ion beam (67 W/m2),
there is no additional heating due to the ion beam. Because of the limited power of
the heating stage and the poor thermal contact between the samples and the holder,
it was not possible to conduct the irradiation with a specimen temperature higher
than 573 K. Additional experiments are needed to identify the appropriate tempera-
ture shift to simulate PWR neutron irradiation.
The samples used here have been taken in the AD–TD plane, the long direction
of the strip being either along the TD or along the AD. Before irradiation, the thin
strips of 0.1 mm thick are electro-polished on one side either in a form of flat
electro-polishing or in a form of shallow dimple. After irradiation 3 mm disks are
punched out of the strips and the irradiated surface is protected by a varnish.
Electro-polishing on the rear side is then performed in order to obtain thin foils
that are then observed by TEM.
The original device developed for this study induces a bending of the sample,
with a constant displacement during the experiment. Therefore, the unirradiated
side of the strip (99.7 lm thick) subjected to a stress gradient relaxes the applied
stress through conventional thermally activated viscoplastic deformation mecha-
nisms. This thermally activated stress relaxation is observed after the experiment
thanks to a residual bending of the samples. Nevertheless, this stress relaxation in

the unirradiated part of the sample remains limited. It is estimated from residual
bending measurements that only 23 % of the initial applied stress is relaxed during
the experiment. This is in rather good agreement with the stress relaxation meas-
ured by Carassou et al. [54] using three-point bending experiments and thicker
samples where only 14 % of the applied stress is relaxed after 100 h at 588 K.
On the other hand, in the 0.3 lm thick irradiated part of the sample, rapid stress
relaxation occurs because of irradiation activated deformation mechanisms. It is shown
by Carrassou et al. [54] that the applied stress is nearly fully relaxed (85 % relaxed) after
a fluence of 5 1024 n/m2 irradiation in Materials Testing Reactor at 588 K. Assuming
that the same relaxation applies for the layer irradiated with Zr ions, despite the com-
plex two-layer structure of the sample, the stress is nearly fully relaxed after 1.25 dpa
using the conversion factor given in Ref. [55]. This shows that the device used here can
only apply a stress on the irradiated layer for a short irradiation period. Thanks to this
device two different loading histories have been studied.
Loading History I
During this experiment, the stress is applied from the beginning of irradiation. The
initial applied tensile stress on the irradiated layer is about 125 MPa. This applied
stress is assumed to be in the elastic domain, considering a yield stress of the order
of 150 MPa at 573 K. The stress in the irradiated layer is assumed to decrease rap-
idly during the first dpa of the Zr ion irradiation. The irradiation is conducted up to
4.8 dpa. Only M5 samples taken in the AD have been studied in this experiment.
Loading History II
During the second loading history, the samples are first irradiated without any applied
stress up to 4.1 dpa at 573 K. Then the samples are moved on the specimen holder, so
that a tensile stress is applied on the irradiated layer. The initial applied stress is consid-
ered to be in the plastic domain in that case (higher than 150 MPa). The stress then is
assumed to decrease rapidly between 4.1 to 5.1 dpa. The irradiation is conducted up to
7 dpa on overall. Only RXA Zy-4 samples taken in the AD and in the TD are studied
in this experiment. A similar study conducted on M5 sample is still under progress.
It has to be pointed out that for both loading histories the high stress applied
on the samples is not representative of the in-reactor applied axial stress on the
guide tubes which is usually of the order of 30 MPa.
The irradiation conditions are summarized in Table 5. Control samples have
also been irradiated in the same conditions. Furthermore, additional control sam-
ples irradiated up to various doses have also been studied. All the irradiation condi-
tions for control samples are given in Table 6.
OPTICAL MICROSCOPY AFTER PROTON IRRADIATION

Unirradiated hydrided samples, control samples and hydrided samples irradiated
with protons, have been polished and etched with fluoridric acid in order to reveal
TABLE 5 Irradiations performed to study the influence of an applied stress on c-loops.
Total Irradiation Dose Applied Stress
Loading Irradiation Material dpa ions/cm2 Stress Direction Damage Creation Flux (ions/cm2/s) Duration (h)
Rate (dpa/s)
I Zr2þ M5 4.8 9.0 1014 Elastic (stress AD 4.7 104 8.8 1010 2 h 50
TOURNADRE ET AL., DOI 10.1520/STP154320120200

300 keV 125 MPa)
573 K
II Zrþ Zy4 pre- 7 1.15 1015 plastic AD and TD 4.6 104 7.6 1010 4 h 10
600 keV irradiated at 4.1
573 K dpa
865
TABLE 6 Zr ion irradiation conditions conducted on control samples without any applied stress.
Dose
Damage Flux Duration

Facility Ion Energy dpa ions/cm2 T (K) Rate (dpa/ (ions/cm2/ (h)
s) s)
ARAMIS Zrþ 600 keV 4.1 6.7 1014 573 4.6 104 7.0 1010 2 h 30
5.5 9.0 1014 3 h 20
7 1.15 1015 4 h 10
IRMA Zr2þ 300 keV 4.8 9.0 1014 573 4.7 104 8.8 1010 2 h 50
6.2 1.16 1015 4.9 104 9.2 1010 3 h 30
the hydrides. Conventional optical microscopy has then been performed on the
samples.
TEM ANALYSIS METHOD AFTER CHARGED PARTICLE IRRADIATION

TEM observations have been performed on Jeol 2100 and Jeol 2010 transmission
electron microscopes, both operating at 200 kV with a LaB6 filament. For each of
the irradiation conditions, several thin foils have been studied and for each thin foil,
several grains were analyzed. Every thin foil has been taken in the AD–TD plane in
this study. Thanks to the texture of the material many grains have their c-axis close
to the TD. As a consequence, many grains can be easily tilted in the appropriate
conditions to observe c-loops, that is to say in the diffraction condition with
g ¼ 0002.
For each grain, several pictures are taken to cover most of the grain surface and
in the grain many loops are analyzed. The number of c-loops per unit volume NV
and their mean diameters hdi were measured using an interactive digital drawing
tablet connected to Visilog software. The grain thickness was measured using elec-
tron energy loss spectroscopy. When the c-loops are very large and, as a conse-
quence, are truncated by the thin foil surfaces, the linear density Lv, defined in Eq 3,
is used.
Total length of hci loop segments measured ðmÞ

(3) LV ðm2 Þ ¼
Total volume studied by TEM ðm3 Þ
The orientation of each grain was determined by using diffraction pattern indexing
and recording the tilt angles. The stereographic projection was then drawn for each
studied grain. The AD of the tube was always located on the thin foil, and the thin
foil was always orientated so that the AD is parallel to the primary tilt axis.
When a stress (r) was applied on the specimen under irradiation, the angle (h)
between the tensile direction and the c-axis is measured on the stereographic pro-
jection. Then, the component normal to the loop plane of the deviatoric stress
hci
tensor, rs , is then computed according the SIPA theory (Eq 4). Loop densities and
diameter are then plotted as a function of this component.

2 2 1 2
(4) rhci
s ¼ r cos h sin h
3 3
Results on the Effect of an Applied Stress Under

Zr Ion Irradiation
CONTROL SAMPLES AFTER ZR ION IRRADIATION
TEM observations performed on control samples irradiated with 600 keV Zr ions at
573 K on the ARAMIS facility first show that the c-loop microstructure obtained in
M5 and RXA Zy-4 are very similar (Fig. 5). It is observed that the mean loop diame-
ter increases from 24 to 36 nm from 4.1 to 7 dpa. The number density also increases
from 2.9 1020 to 1.0 1021 m3. Similar results are obtained with 300 keV Zr
ions, although a slight shift in number density can be noticed as shown on Table 7.
FIG. 5 c-loop microstructures observed after 600 keV Zr ions at 573 K on ARAMIS. (a)
RXA Zy-4 irradiated up to 4.1 dpa, (b) RXA Zy-4 irradiated up to 7 dpa, (c) M5
irradiated up to 4.1 dpa, (d) M5 irradiated up to 7 dpa.
TABLE 7 Mean loop diameter, number density, and linear density of c-loops after irradiations con-
ducted at 5 7 3 K on the IRMA and ARAMIS facilities respectively with 300 and 600 keV
Zr ions.
Dose Mean Number Lv Number of

Material Energy (dpa) Diameter (nm) Density (m3) (m m3) Loop
Analyzed
Zy-4 300 keV 6.2 35 1.6 1021 5.7 1013 525

Zy-4 600 keV 4.1 24 2.9 1020 6.6 1012 640
Zy-4 600 keV 5.5 26 6.7 1020 1.6 1013 1300
Zy-4 600 keV 7 36 9.1 1020 3.3 1013 1400
M5 300 keV 4.8 19 2.5 1021 5.0 1013 1860
M5 300 keV 6.2 43 1.9 1021 8.7 1013 475
M5 300 keV 5.5 30 1.1 1021 3.4 1013 1100
M5 300 keV 7 28 1.0 1021 2.8 1013 610
STRESS APPLIED AT THE BEGINNING OF ZR ION IRRADIATION AT 573 K

TEM observations performed on M5 specimens irradiated with 300 keV Zr ions
under an applied stress (Loading History I) up to 4.8 dpa show no influence of the
stress when compared with control samples. Indeed, as illustrated in Fig. 6, when
plotted as a function of the component normal to the loop plane (along the c-axis)
of the deviatoric stress tensor, no trend is observed concerning the loop number
density or the loop mean diameter. This proves that there is no effect of an applied
stress at the early stage of irradiation on c-loops.
Since M5 and Zy-4 control samples exhibit the same microstructures, it is
assumed that this result also applies for Zy-4 irradiated under stress.
FIG. 6 M5 irradiated by 300 keV Zr ions at 573 K on IRMA up to 4.8 dpa. (a) C-loop
mean diameter and (b) number density N hci as a function of rhci
s (deviatoric
stress component along the c-axis).
FIG. 7 c-loop microstructures observed in RXA Zy-4 irradiated by 600 keV Zr ions at
573 K after the second loading history in six grains. The stress direction is given
in the crystal frame on the inverse pole figure.
FIG. 8 RXA Zy-4 irradiated by 600 keV Zr ions at 573 K according to the second
loading history (up to 4.1 dpa without any stress, during 2.9 dpa more under
stress). C-loop linear density LV as a function of the deviatoric stress component
along the c-axis (rhci
s ). The cross on the y-axis corresponds to the mean value
obtained for control samples.
FIG. 9 RXA Zy-4 irradiated by 600 keV Zr ions at 573 K according to the second
loading history (up to 4.1 dpa without any stress, during 2.9 dpa more under
stress). (a) C-loop mean diameter and (b) number density as a function of rhci
s .
The cross on the y-axis corresponds to the mean value obtained for control
samples.
STRESS APPLIED AFTER 4.1 DPA ZR ION IRRADIATION AT 573 K

TEM observations were performed on Zy-4 samples which have undergone the sec-
ond loading history. TEM pictures are given in Fig. 7. The number density, mean
hci
loop diameter, and linear density are reported in Figs. 8 and 9 as a function of rs ,
the component normal to the loop plane (along the c-axis) of the deviatoric stress
tensor. The mean values measured on control samples irradiated up to 7 dpa in the
same conditions, but without any applied stress [56], are shown by a cross on Figs.
8 and 9.
It is shown in Fig. 8 that, when a positive stress is applied along the basal plane
normal to the c-axis, the c-loop density (number density NV and linear density LV )
is higher than when it is applied along the c-axis. On the other hand, the loop diam-
eter seems to be not sensitive to the applied stress.
Furthermore, TEM observations performed on control samples irradiated up to
4.1 dpa without applied stress show that c-loops are already present in the material
in low density. This proves that if a stress is applied when c-loops start to nucleate
and grow, their evolution is affected.
Results Concerning the Effect of Hydrogen

Using Proton Irradiation
OPTICAL MICROSCOPY AND THERMO-DESORPTION RESULTS
First of all, optical microscopy observations performed on M5 control sample irra-
diated with protons up to 19 dpa show (Fig. 10) that many small hydrides are pres-
ent and homogeneously distributed in the sample thickness. This is consistent with
the assumption of the rapid diffusion of the implanted hydrogen in the entire speci-
men. Nevertheless, a fine line located at 30 lm under the irradiated surface can be
noticed in Fig. 10. This “line” could be due to an enhanced hydride precipitation at
the implantation and/or damage peak (located between 25 and 30 lm depth). It is
also possible that preferential chemical etching occurs at damage peak because of
the very high density of irradiation defects.
Then, optical microscopy observations performed on pre-hydrided samples
before irradiation show that hydrides precipitation is homogeneous and consistent
with the hydrogen contents measured by thermo-desorption. The same thermo-
desorption experiments, this time performed on a pre-hydrided M5 sample
annealed during 200 h at 623 K under secondary vacuum, prove that hydrogen de-
sorption remains limited (less than 20 wppm).
However, after proton irradiation (260 h at 623 K), a significant gradient of pre-
cipitated hydrides in the thickness of the sample can be noticed (Fig. 11). Indeed,
numerous precipitated hydrides are observed in the back of the sample. This phe-
nomenon is probably due to the thermal gradient in the samples under protons
irradiation. Indeed, because of the high power deposited by the proton beam on the
irradiated surface, the temperature on the surface is 100 K higher than on the back
FIG. 10 Hydrides precipitated in “as-received” M5 after a 19 dpa proton irradiation at

623 K. Optical Microscopy performed in the TD-ND plane (where ND stands for
Normal or Radiale Direction).
FIG. 11 350 wppm pre-hydrided M5. Optical microscopy observations in the TD-ND
plane. (a) before irradiation, (b) after 2 MeV proton irradiation at 623 K up to 19
dpa.
of the sample which is in contact with the liquid indium. Since it is known that
hydrogen diffuses towards colder areas [57], the hydrogen content is higher on the
back of the sample than on the irradiated side.
Nevertheless, optical microscopy and TEM observations proved that hydrides
are still present in significant amount in the 30 lm thick irradiated layer where
TEM analysis are performed, but the local hydrogen content is not accurately
known and is presumably lower than the overall hydrogen content estimated by
adding the implanted hydrogen and the initial content.
TEM RESULTS: CONTROL SAMPLES AFTER PROTON IRRADIATION

TEM observations have been performed on control samples in M5 and RXA Zy-4
after proton irradiation. As described in a previous study [46], c-loops are observed
after proton irradiation (Fig. 12). The c-loop diameter and density are in rather good
agreement with the observations performed after neutron irradiation (Table 8). The
diameter is nevertheless slightly smaller than after neutron irradiation. Most
remarkably, as for neutron irradiation, it is observed that after proton irradiation
the c-loop density in M5 is much lower than in RXA Zy-4 (Table 8 and Fig. 20(a)).
TEM RESULTS: IMPACT OF THE HYDROGEN IN M5

TEM observations have been performed on pre-hydrided M5 samples irradiated
with protons up to 19 dpa. When conducting TEM observations, two distinct areas
of the grain have been analyzed successively: the surroundings of the hydrides and
the others areas of the grain, the so-called “matrix,” far from hydrides.
C-Loops Far From Hydrides in M5

Firstly, only the results obtained in the “matrix” (far from precipitated hydrides)
are reported. TEM observations were performed on M5 samples pre-hydrided up to
three different contents. These samples were irradiated by 2 MeV protons at 623 K
up to 19 dpa. Measurements of c-loop densities have been done in the matrix, far
from the precipitated hydrides (Fig. 12). The densities obtained are given in Fig. 13 as
a function of the hydrogen content. These results tend to show a linear relationship
between c-loop densities and the hydrogen content up to 350 wppm hydrogen.
Moreover, TEM observations have been conducted on 350 wppm pre-hydrided
M5 after the four irradiation doses. These observations (Fig. 12) show a higher c-
loop density and a more homogeneous c-loop microstructure in the pre-hydrided
samples than in the control samples. Indeed, in the control samples, c-loops are
mainly observed at the vicinity of some precipitates and at grain boundaries [46].
Even if these phenomena are still observed in the pre-hydrided samples, c-loops are
also present in the middle of the matrix, far away from the secondary phase precipi-
tates and grain boundaries. Most remarkably, c-loops are observed after 4.9 dpa in
the 350 wppm pre-hydrided samples whereas they are not present after the same
irradiation dose on control samples. This proves that the pre-hydriding induces a
shift toward lower doses of the threshold dose for c-loop nucleation. From these
FIG. 12 C-loop microstructures observed in M5 control samples: (a), (c), (e), and (g))
and 350 wppm pre-hydrided M5 (b), (d), (f), and (h)) after 2 MeV proton
irradiations at 623 K up to: (a) and (b) 4.9 dpa; (c) and d) 8.1 dpa; (e) and (f) 12.5
dpa; (g) and (h) 19 dpa.
TABLE 8 Mean loop diameter, number density, and linear density of c-loops after 2 MeV proton
irradiation conducted at 6 2 3 K on Zy-4 and M5 control samples. PWR results are also
given in the table.
Mean Number Number

Irradiation and Dose Diameter Density Lv of Loop
Temperature Material (dpa) (nm) (m3) (m m3) Analyzed
Protons, 623 K Zy-4 8,1 102 4.0 1020 4.4 1013 4133
Protons, 623 K Zy-4 12,5 123 1.0 1021 1.1 1014 350
Protons, 623 K Zy-4 19 111 1.1 1021 1.3 1014 1048
PWR, 583 K [26] Zy-4 14 >150a <1.2 1021 a 1.8 1014
PWR, 593 K [40] Zy-4 low 35 >150a <1.17 1021 a 1.75 1014
Sn RXA
Protons, 623 K M5 4,9 0 0 0 0
Protons, 623 K M5 8,1 55 1.0 1020 5.7 1012 202
Protons, 623 K M5 12,5 49 2.2 1020 1.1 1013 290
20
Protons, 623 K M5 19 71 2.5 10 2.0 1013 543
PWR, 593 K [40] M5 29,1 >150a <2.3 1020 a 3.4 1013
quantitative results (Fig. 14), a c-loop evolution law as a function of dose (D in dpa),
far from hydrides, could be proposed (Eq 5). The threshold dose (Dc ) is equal to
4.36 dpa for 350 wppm pre-hydrided samples whereas it is equal to 5.55 dpa for
control samples. The slope of the linear evolution (A) is equal to
3:92 1012 m2 dpa1 for 350 wppm pre-hydrided samples whereas it is equal to
2:63 1012 m2 dpa1 for control samples.

If D < Dc ; LVhci ¼ 0
(5)
If D Dc ; LVhci ¼ AðD Dc Þ
To finish, in the 350 wppm pre-hydrided M5 samples irradiated up to 19 dpa,

moiré fringes have been observed on c-loops (Fig. 15). This contrast results probably
from the presence of nano-particles precipitated on c-loops. Observation of these
particles by HR-TEM shows that they are coherent with the Zr matrix with a little
misfit. Such particles are not observed in the control samples coming from the
same irradiation. It can be assumed that these particles are nano-hydrides precipi-
tated under irradiation.
C-Loops in the Surroundings of “Hydrides” in M5

In addition to the increase of c-loop density far from hydrides, another peculiar fea-
ture has been studied by TEM. Some sort of c-loop bundles are observed, as shown
in Figs. 16–18. In Fig. 16, it is clearly seen that the c-loop bundle appears between two
delta hydrides (with an OR2 orientation with the matrix). In Figs. 16–18, it can be
noticed that the c-loop bundle exhibits a form very similar to hydrides, elongated
FIG. 13 C-loops linear density LV evolution as a function of the initial hydrogen content.
M5 irradiated by 2 MeV protons at 623 K up to 19 dpa.
FIG. 14 C-loops linear density LV in 350 wppm M5 and control samples as a function of
the irradiation dose (D in dpa).
FIG. 15 Nano-contrast observed on c-loops in 350 wppm M5 samples irradiated up to

19 dpa with 2 MeV protons at 673 K. (a) and (b) “moiré” fringes; (c) high
resolution TEM on a nano-particle.
along the basal plane. Furthermore, it is also observed that c-loop bundles are often
associated with the presence of hc þ ai dislocations. C-loop bundle formation has
been confirmed by numerous observations performed on various samples irradiated
different up to 19 dpa and pre-hydrided from 80 wppm to 350 wppm.
It is believed that these c-loop bundles are created close or at the location of
partially or fully dissolved hydrides. Indeed, hc þ ai dislocations could be the
remaining accommodation dislocations that may be created around hydrides due
to the high volume increase (DV=V ¼ 17%) during hydride precipitation. The
form of the bundles and the presence of c-loops between two partially dissolved
hydrides also support this hypothesis.
On the other hand, it has been noticed that when a hydride is fully precipitated,
that is to say not partially nor fully dissolved, no effect on c-loop is observed
(Fig. 19).
TEM RESULTS: IMPACT OF HYDROGEN ON ZY-4

TEM observations have also been performed on pre-hydrided RXA Zy-4. Many c-
loops are already observed in control sample in RXA Zy-4 (Fig. 20). In pre-
hydrided samples, no clear increase of c-loop density is observed (Fig. 20).
FIG. 16 350 wppm pre-hydrided M5 irradiated by 2 MeV protons at 623 K up to 19

dpa.(a) c-loop bundle at a vicinity of a partially dissolved hydride. (b)
Diffraction pattern of the precipitated hydride (OR2 orientation) for a g ¼ 0002
diffraction vector.
FIG. 17 350 wppm pre-hydrided M5 irradiated by 2 MeV protons at 623 K up to 19 dpa.

C-loop bundles (a) hcþai dislocations pointed by white arrows, (b) the same
picture at a higher magnification.
FIG. 18 80 wppm pre-hydrided M5 irradiated by 2 MeV protons at 623 K up to 19 dpa.

(a) C-loop bundle on a former hydride precipitated in the basal plane and now
completely dissolved, (b) former hydride now completely dissolved.
Discussion
Under irradiation, point defects are created in displacement cascades. These
point defects migrate and cluster together leading to the creation of point defect
clusters in the form of hai and c-loops in zirconium alloys [3]. The evolution of
hai loops and c-loop diameter and number density depend on the absorption
FIG. 19 M5 irradiated by 2 MeV protons at 623 K up to 19 dpa. Observation at the

vicinity of fully precipitated hydrides (a) in 80 wppm pre-hydrided M5, (b) in
350 wppm pre-hydrided M5.
FIG. 20 RXA Zy-4 irradiated by 2 MeV protons at 623 K up to 19 dpa. (a) control
sample, (b) 185 wppm pre-hydrided sample.
and emission of point defects by the loops and also by the other sinks present in
the material. The evolution of the microstructure under irradiation can be simu-
lated using cluster dynamics modeling [27,58]. In this framework, loop nuclea-
tion and growth is governed by the net flux of point defects on loop. A vacancy
loop can nucleate and grow if the flux of vacancies absorbed by the loop is
higher than the sum of the flux of self-interstitial atoms (SIA) absorbed and the
flux of vacancies emitted by the loop. Because of the high formation energy of
self-interstitial atoms, point defect clusters cannot emit SIA. Thus, the resulting
c-loop microstructure, and therefore the resulting free growth, depends on the
emission and capture rate of point defects by c-loops. The differences observed
between the c-loop microstructures obtained after various types of irradiation
can be explained by considering the balance between emission and capture rate
of point defects.
The influence of the applied stress or of the hydrogen content can also be inter-
preted by considering their potential effects on the absorption or capture rate of
point defects by c-loops.
DISCUSSION ON THE DIFFERENCES BETWEEN THE VARIOUS IRRADIATION

TYPES
For 300 keV or 600 keV Zr ion irradiation, displacement cascades are large, as
for PWR neutron irradiation [59] as shown by the weighted recoil spectra
given in Ref. [46]. However, for transferred energies higher than 10 keV, these
large cascades break into sub-cascades. On the other hand, proton irradiation
displacement cascades are very small, as shown by the weighted recoil spectra
given in Ref. [46]. This difference in the primary damage morphology could
have significant effects on the microstructure evolution under irradiation.
Indeed, in cascades, small point defect clusters can be formed which can consti-
tute nuclei for loops such as c-component loops. However, recombination of
point defects occurs [59], leading to a lower fraction of defects available to
migrate. As a consequence, for the same given dose in dpa, Zr ion irradiations
are less efficient than for proton irradiation. In this respect, Zr ion irradiations
should better simulate neutron irradiations than proton irradiations. Neverthe-
less, TEM observations have shown that the c-loop microstructure obtained
after proton irradiation is representative of the one obtained after neutron
irradiation. This proves that the primary damage morphology does not have a
significant effect on c-loop microstructure [15,46].
On the contrary, the damage rate seems to play a major role on c-loop
microstructure. Indeed, for Zr ion irradiation, the damage rate is high (5 10–4
dpa/s), leading to a high concentration of point defects in the material, consid-
ering that the steady state has been reached. Therefore the nucleation rate of
point defect cluster nuclei, such as di-vacancies clusters, is significantly higher
than the emission rate of vacancies by the nuclei. Indeed, the clustering rate of
two single vacancies is proportional to the square of the vacancy concentration.
A detailed derivation of the nucleation rate is reported by Was [59]. This
author also states that higher vacancy production rates promote greater vacancy
concentration and a higher cluster nucleation rate. This, therefore, explains that
the nucleation rate of c-loops is higher under Zr ion irradiation, resulting in a
higher density of small c-loops as observed by TEM after Zr ion irradiation. On
the other hand, for proton irradiation, the damage rate is lower (1.50 105
dpa/s), leading to a lower concentration of point defects in the material. This
induces a lower capture rate of point defects by cluster nuclei, the emission of
vacancies remaining the same, leading therefore to a lower nucleation rate. The
c-loop density is thus lower and these fewer loops absorb the point defects pres-
ent in the material, explaining the larger diameter of c-loops observed after pro-
ton irradiation.
It is difficult to explain, in this framework, the origin of the incubation
dose observed for the nucleation of c-component loops. It is proposed here
that the transition from E-type c-loop to I1-type c-loop could be one of the
origins of the incubation dose observed. Indeed E-type loop nuclei could ex-
hibit a very low growth rate; however, as soon as I1-type loops are formed
beyond a critical radius, they grow rapidly. The fact that the critical loop size
for the transformation from E-type loop to I1-type loop can be influenced by
impurity content, according to Ref. [22], through its effect on the stacking
fault energy, could also explain the strong sensitivity to alloying elements (Nb,
Sn, Fe) observed in this study when comparing c-loops in M5 and RXA Zy-4
irradiated with protons. Additional numerical work is, however, needed to
conclude on this point. It is also possible, as proposed in Refs. [14,27], that
because of diffusion coupling between vacancies and solute atoms, such as
iron, heterogeneous precipitation occurs on point defect clusters. This type of
phenomenon is observed in steels using a tomographic atom probe [60,61].
This could also play a significant role in the incubation dose of c-loops in zir-
conium alloys.
The fact that no effect of alloying elements is observed for Zr ion irradiation
could be explained by considering that during the 4 h Zr ion irradiation at 573 K,
alloying elements do not have the time to diffuse toward the loop and therefore do
not affect the loop nucleation and growth.
DISCUSSION ON THE EFFECT OF STRESS

According to the SIPA mechanism described in detail by Garner [31,32] and Wolfer
[33], the applied stress affects the capture efficiency of SIAs with respect to vacan-
cies, which is characterized by the bias (Zid ). Indeed, it is known that the interaction
energy between a straight dislocation and a SIA is higher than between a dislocation
and a vacancy (Elastic Interaction Difference, EID) leading to a bias higher than
one (Zid > 1, usually a value of Zid ¼ 1.1 is used for Zr [58]), the bias used for va-
cancy being equal to one (Zvd ¼ 1). For a dislocation loop, the bias also depends on
radius R of the loop Zil ðRÞ ¼ Zid f ðRÞ, where the function f ðRÞ can be found in Refs.
[33,58,62].
In presence of other unbiased sinks, the flux of SIAs toward dislocations will
therefore be higher than the vacancy flux. Because of the difference in anisotropic
diffusion of point defects (faster migration of SIAs along the basal plane than along
the c-axis) dislocations in the basal plane, such as c-component loops, absorb more
vacancies than SIAs. Following Woo’s approach [28], this phenomenon induces a
geometrical bias that multiplies the bias due to EID. In the case of straight disloca-
tion in the basal plane, the bias becomes lower than unity (Zid < 1). This is also true
for loops in the basal plane [27] leading to growth of vacancy c-loops.
When a stress is applied, the bias due to EID is modified according to Eq 6.
(6) Zil ðrÞ ¼ Z0 þ Crsij ni nj
where:
Z0 ¼ the EID bias without an applied stress,
C ¼ a constant which depends on the material and on the loop size,
r0S
ij ¼ the deviatoric stress tensor, and
ni ¼ the unit vector components which describe the vector normal to the loop
habit plane.
For a loop in the basal plane, Eq 6 can be rewritten as Eq 7.
(7) Zil ðrÞ ¼ Z0 þ Crhci

s
hci
where rs is the component of the deviatoric stress tensor along the c-axis.
hci
For tensile stress (r) applied along the c-axis, it can be seen that rs ¼ 2r=3
and the bias is equal to Eq 8.
2
(8) Zil ðrÞ ¼ Z0 þ Cr
3
From this formula, it can be seen that a tensile stress applied along the c-axis
increases the capture efficiency of SIAs compared to vacancies, therefore leading to
a reduced growth of vacancy c-loop.
On the other hand, when the stress is applied along the basal plane,
hci
rs ¼ r=3, and the bias is expressed as Eq 9.
1
(9) Zil ðrÞ ¼ Z0 Cr
3
In that case, the applied stress decreases the capture efficiency of SIAs compared to
vacancies, thus increasing the vacancy loop growth.
The results obtained on c-loops observed in several grains with various ori-
entations are well explained by the SIPA mechanism. Indeed it can be seen in
Fig. 8 that, when the deviatoric stress component along the c-axis is close to
2r=3 (tensile stress along c-axis), the number density of c-loops (NV ) is lower
than the number density measured without an applied stress. On the other hand,
when the deviatoric stress component along the c-axis is close to r=3 (tensile
stress along the basal plane), the number density of c-loop increases and
becomes significantly higher in some grains than the control samples without
applied stress. The effect on the linear density is also significant despite the dis-
persion of the results. The mean loop diameter appears to be only slightly
affected by the applied stress, in good agreement with the results and the analy-
sis given by Garner et al. [31] and Wolfer [33]. It has also been checked by lim-
iting the analysis to grains with c-axis close to the surface of the foil that the
observed effect is not due to differences in orientation of the c-axis with respect
to the foil surface. From these results, it can be deduced that when the stress is
applied when c-loops start to nucleate and grow, the applied stress affects the
loop growth thanks to the SIPA mechanism.
On the other hand, the experimental results show that, for the first loading history,
no effect of stress on c-loops is observed. In this case, the stress decreases rapidly in the
irradiated area (85 % relaxed after 1.25 dpa), the applied stress is close to zero when the
first c-loops are clearly observed by TEM (with diameter larger than 5 nm). This obser-
vation suggests that at the beginning of irradiation (when the stress is applied) no c-
loop nuclei are created or these nuclei are too small to be affected by the stress.
This phenomenon, observed here for the first time concerning c-loops, could
have a significant impact on the in-reactor deformation of PWR fuel assemblies
made of Zr alloys.
DISCUSSION ON THE EFFECT OF HYDROGEN

Before irradiation, the hydrogen content in the pre-hydrided M5 and RXA Zy-
4 specimens is completely precipitated as hydrides since hydrogen solubility
limit at room temperature is less than 20 wppm. At 623 K, the hydrogen solu-
bility limit is 130 wppm [42,63]. Thus, in the case of the 80 and 125 wppm
pre-hydrided M5 samples, during irradiation, all the hydrogen is in solid solu-

tion. These samples allow studying the role played by hydrogen in solid solu-
tion. On the other hand, for the 350 wppm pre-hydrided M5 and the 185
wppm pre-hydrided Zy-4, some hydrides remain during irradiation. These sam-
ples permit to observe both the effect of hydrogen in solid solution and the
effect of hydrides remaining precipitated during the irradiation.
In the following section, the role played on c-loops by hydrogen in solid solu-
tion will be first discussed. Then, the role on c-loops of hydrides precipitated before
or during the irradiation will be addressed.
Effect of Hydrogen in Solid Solution

Ab initio computations performed at atomic scale have shown that hydrogen atoms
decrease the basal stacking fault in a-zirconium [64]. Indeed, Domain et al. [64], in
studying the effect of hydrogen on dislocation glide, showed that when 25 % of tet-
rahedral sites on the stacking fault plane are filled by hydrogen atoms, the I2 basal
stacking fault excess energy decreases from 200 down to 80 mJ/m2. This shows that
segregation of hydrogen on stacking fault is likely to occur. Furthermore, when
more than 40 % tetrahedral sites are filled, the stacking fault excess energy becomes
negative. This suggests, according to Ref. [64], that high local hydrogen content can
even induce a stacking fault. Assuming that hydrogen also decreases the I1 basal
stacking fault energy, the c-loop energy given in Eq 1 is decreased by the presence
of hydrogen atoms.
In the framework of cluster dynamics modeling [58], the vacancy emission rate
by a vacancy loop, avnv , is function of the binding energy between a vacancy and a
B
loop (Env ) according to Eq 10.

EB
(10) avnv / exp nv
kT
where:
k ¼ the Boltzmann constant, and
T ¼ temperature.
This binding energy, which is a positive quantity, is defined by Eq 11.
B

(11) Env ¼ Evf Env Eðn1Þv
where:
f
Ev ¼ the vacancy formation energy,
Env ¼ energy of the loop containing n vacancies, and
Eðn1Þv ¼ the energy of the loop containing n-1 vacancies.
Using the loop energy based on elastic theory (Eq 1), a first order approxima-
tion [62] can be obtained for the binding energy (Eq 12).

B 1 Vat lb2 4R
(12) Env ffi Evf 2p ln þ 2pRcfault
2 Rpb 4pð1 Þ r0
where Vat is the atomic volume. From Eq 12, it can be seen that a decrease in the
stacking fault energy, because of the presence of hydrogen atoms, induces an increase
in the binding energy and therefore a decrease of the emission rate of vacancies by va-
cancy loops. As a consequence, c-loop nucleation and growth rate are increased by
the presence of hydrogen in solid solution. This phenomenon would be enhanced by
segregation of hydrogen atoms on the loop stacking fault disk.
It is also possible, according to some authors [65,66], that hydrogen atoms cre-
ates a Cottrell atmosphere around the dislocation line of the edge loop leading to a
screening of the stress field and therefore to a decrease of the line energy of the
loop. This phenomenon could also lead to an increased nucleation and growth of c-
loops.
The possible precipitation of nano-hydrides on c-loops in 350 wppm M5 irradi-
ated up to 19 dpa could be mentioned as an experimental evidence for hydrogen
trapping by c-loops.
These two mechanisms therefore explain the enhanced nucleation and growth
of c-loops far from hydrides. The trapping of hydrogen atoms around dislocation
lines (for both a-type and c-type loop) and on the stacking fault disks (only for c-
loops) could also explain the increased apparent solubility of hydrogen in irradiated
Zr alloys [41–43].
The linear evolution of the c-loop linear density with hydrogen content (Fig. 13)
above 130 wppm hydrogen is rather surprising since for contents above the solubil-
ity limit at 623 K, the solid solution content remains constant by definition. This
surprising phenomenon could be explained first by the increase of the apparent sol-
ubility under irradiation but also by the fact that the local hydrogen content in the
irradiated area is lower than the overall hydrogen content due to the temperature
gradient in the samples. Furthermore, one must also keep in mind that a significant
amount of hydrogen is introduced in the material by the proton beam. In the case
of the 350 wppm pre-hydrided samples irradiated up to 19 dpa, the overall hydro-
gen content after irradiation is of the order of 450 wppm. Considering that the
hydrogen gradient is linear and that the hydrogen content is roughly three times
higher in the cold side than in the irradiated side (qualitatively estimated from opti-
cal microscopy observations), the hydrogen content in the irradiated layer could be
approximately of the order of 230 wppm only. From the data given by Vizcaino
et al. [42], it can be estimated that at 623 K the hydrogen apparent solubility limit is
around 230 wppm after irradiation instead of 130 wppm. In that case, the local
hydrogen content in the irradiated layer would be close to the apparent solubility
limit. The observation of few partially dissolved hydrides suggests that the local
hydrogen is slightly higher than the apparent solubility limit. This can therefore
explain that the c-component loop density increases when the hydrogen content
during pre-hydriding is increased since the higher local hydrogen content is
believed to be only slightly higher than the solubility limit. Nevertheless, further
analysis is needed to have an accurate measurement of the local hydrogen content
in the irradiated layer and then provide a thorough understanding of the results.
Effect of Hydrogen in the Form of Hydrides During Irradiation

Because of the increase of the apparent hydrogen solubility limit due to hydrogen
trapping on irradiation defects and of hydrogen migration toward the cold area, the
remaining hydrides are dissolved progressively during the proton irradiation. The
hydrides act as source of hydrogen which migrate progressively and can be trapped
by c-loops nuclei enhancing their growth. This could therefore explain the high c-
loop density around dissolved or partially dissolved hydrides.
Another mechanism could also be proposed to explain some experimental
results. Indeed, hc þ ai dislocations have been observed on some c-loop bundles.
These hc þ ai dislocations are believed to be accommodation dislocations that are
created because of the large volume increase (DV=V ¼ 17 %) due to hydrides pre-
cipitation. Indeed, it is described by several authors [67,68] that a-type dislocations
nucleate during the precipitation of small hydrides. Growth of large hydrides could
also result in the nucleation of hc þ ai dislocations although further investigations
are needed to confirm this point.
As shown by Griffiths et al. [30,69] these hc þ ai dislocations could be nuclea-
tion sites for c-loops. Indeed, it is proposed here that a kink (of edge character) con-
tained in the basal plane on a hc þ ai dislocation (of screw character), with Burgers
vector 13 ½1123 , can dissociate into two partial dislocations with Burgers 16 ½0223 and
1
6 ½2023 (Eq 13) (Fig. 21). This creates a stacking fault ribbon of I1 type. Such dissoci-
ation has already been discussed concerning plasticity mechanisms in zinc and zir-
conium [70,71].
1 1 1
(13) ½1123 ! ½0223 þ ½2023
3 6 6
FIG. 21 Dissociation of a screw hcþai disslocation in two partial dislocations with a c-

type burgers vector separated by a stacking fault ribbon.
Under irradiation, one or both of these partial dislocations absorb more vacancies
than SIAs since they are in the basal plane. As a consequence, they climb and the
stacking fault ribbon widens resulting in the formation of a c-loop (Fig. 21).
The observation of c-loop bundles in 80 wppm M5 is an experimental evidence
of this mechanism. Indeed, in this material, all the hydrides are dissolved at irradia-
tion temperature and only the hc þ ai accommodation dislocations remain. Thus,
the bundle formation could be explained only by the fact that these hc þ ai disloca-
tions act as nucleation sites for c-loops.
It has been pointed out that around fully precipitated hydrides, there is no
increase in c-loop density nor c-loop bundle. This could be explained by consider-
ing that these hydrides have precipitated during the cooling down, after proton irra-
diation. It is also possible that some hydrides present during irradiation did not
dissolve and therefore did not affect the c-loop evolution.
The impact of hydrogen on c-loops observed here for the first time, could have
a significant impact on the in-reactor deformation of PWR fuel assemblies made of
Zr alloys.
Conclusions
This experimental study first shows the relevance of using ions, such as protons or
Zr ions, to simulate neutron irradiation and perform a thorough analytical study on
c-loops. Indeed, 2 MeV proton irradiations conducted at 623 K allow reproducing
qualitatively and quantitatively the c-loop microstructures obtained in PWR operat-
ing conditions for the two studied alloys: M5 and RXA Zy-4. In particular, some
phenomena are also observed after proton irradiation, such as the alloying effect (c-
loop densities are less numerous in M5 than in RXA Zy-4) or the preferential c-
loop nucleation around some precipitates. Furthermore, thanks to the hydrogen
atoms penetration depth, proton irradiations allow to obtain a bulk material fully
irradiated and to avoid the free surface effect. This therefore appears as a good tool
to study the role of hydrogen on c-loops.
On the other hand, Zr ion irradiations performed at 573 K also induce c-loops
in the material, but the microstructure is not representative of the neutron irradi-
ated material. However, Zr ion irradiation appears to be a good tool to study the
effect of stress on c-loops, one of the reasons being the low power deposited by the
ion beam.
The differences observed between the two types of irradiations are well
explained by the differences in damage rates. Thus, Zr ion irradiations would prob-
ably be more representative of neutron irradiations by using a higher irradiation
temperature.
The effect of stress on c-loops has been studied for the first time. These results
first show that a stress applied at the beginning of the irradiation has no impact on
the c-loop microstructures. On the contrary, an effect is observed when a stress is
applied when c-loops are already created. This phenomenon is well explained by
the SIPA mechanism which describes the effect of stress on irradiation
microstructures.
Then, the effect of pre-hydriding on c-loops has been studied for the first time.
A strong effect of the pre-hydriding is observed. In the matrix, far from precipitated
hydrides, the c-loop density is higher and the microstructure more homogeneous in
the pre-hydrided material than in the control samples. Moreover, c-loop “bundles”
are observed in the alignment of what seem to be former hydrides partially or com-
pletely dissolved.
These results can be explained by the fact that hydrogen could influence c-loop
microstructures by two different mechanisms:
• On the one hand, the growth of the c-loop nuclei could be enhanced by the
trapping of hydrogen atoms in solid solution on the defects, reducing the loop
energy. In this mechanism, the precipitated hydrides act as source of hydrogen
atoms.
• On the other hand, the remaining hc þ ai dislocations after the precipitated
hydrides dissolution could act as nucleation sites for c-loops.
Due to the low hydride density, the role of c-loop bundles created on former
hydrides is believed to remain minor. On the other hand, the increase of c-loop
density within all the grains of the material can have a significant consequence on
the accelerated growth of recrystallized zirconium alloys.
These experimental results prove that an applied stress can have an effect on c-
loops. Furthermore, these results also show that pre-hydriding leads to an increase
of the c-loop densities after proton irradiations. Since c-loops are responsible for
the growth acceleration of recrystallized Zr alloys, these two effects could have a sig-
nificant impact on the in-reactor deformation of the fuel assemblies.
ACKNOWLEDGMENTS
The writers would like to thank AREVA NP for the supply of the material and the fi-
nancial support of this study. For the hydriding and the hydrogen content measure-
ments, the authors thank B. Guerin from CEZUS-AREVA, and C. Berziou from LaSIE
(Université de La Rochelle). For the active discussion concerning c-loops the authors
thank S. Doriot. For specimen preparations, chemical analysis, grain size and texture
measurements, the authors thank R. Danguillaume, B. Arnal, S. Bosonnet, D. Hamon
S. Urvoy, T. Vandenberghe and E. Rouesnes. The writers also greatly acknowledge N.
Gharbi, T. Jourdan, C. Varvenne and E. Clouet for discussions on mechanisms at
atomic scale.
References
[1] Franklin, D. G., Lucas, G., and Bement, A., Creep of Zirconium Alloys in Nuclear Reactors,
STP 815, ASTM, 284, ASTM International, West Conshohocken, PA, 1983.
[2] Franklin, D. G. and Adamson, R. B., “Implications of Zircaloy Creep and Growth to Light
Water-Reactor Performance,” J. Nucl. Mater., Vol. 159, 1988, pp. 12–21.
[3] Onimus, F. and Bechade, J.-L., “Radiation Effects in Zirconium Alloys,” Comprehensive
Nuclear Materials, R. Konings, Ed., 2012, Elsevier Ltd., Oxford, UK.
[4] Carpenter, G. J. C., Zee, R. H., and Rogerson, A., “Irradiation Growth of Zirconium Single-
Crystals—A Review,” J Nucl. Mater. Vol. 159, 1988, pp. 86–100.
[5] Lemaignan, C. and Motta, A. T., “Zirconium Alloys in Nuclear Applications,” Materials Sci-
ence and Technology, R. T. Frost, Ed., VCH, New York, 1994, Vol. 10B, pp. 1–51.
[6] Griffiths, M., Gilbert, R. W., and Fidleris, V., “Accelerated Irradiation Growth of Zirconium
Alloys,” Zirconium in the Nuclear Industry (Eighth International Symposium), STP 1023,
1989, pp. 658–677.
[7] McGrath, M. A., Yagnik, S., and Jenssen, H., “Effects of Pre-Irradiation on Irradiation
Growth & Creep of Re-Crystallized Zircaloy-4,” Zirconium in the Nuclear Industry (16th
International Symposium), STP 1529, American Society for Testing and Materials, West
[8] Zbib, A. A., Blavius, D., Morris, J. C., Smith, M. H., Cole, S. E., Garner, N. L., and Moeckel,
A., “Recent Channel Bow Events and AREVA’s Integrated Solution,” Proceedings of the
2011 Water Reactor Fuel Performance Meeting, Sept 11–14, Chengdu, China, 2011.
[9] Holt, R. A. and Gilbert, R. W., “c-Component Dislocations in Annealed Zircaloy Irradiated
at About 570-K,” J. Nucl. Mater., Vol. 137, No. 3, 1986, pp. 185–189.
[10] Jostsons, A., Kelly, P. M., and Blake, R. G., “Nature of Dislocation Loops in Neutron-
Irradiated Zirconium,” J. Nucl. Mater., Vol. 66, No. 3, 1977, pp. 236–256.
[11] Griffiths, M., “A Review of Microstructure Evolution in Zirconium Alloys During Irradi-
[12] Griffiths, M., Loretto, M. H., and Smallman, R. E., “Anisotropic Distribution of Dislocation
Loops in Hvem-Irradiated Zr,” Philos. Mag., Vol. 49, No. 5, 1984, pp. 613–624.
[13] Griffiths, M., Styles, R. C., Woo, C. H., Phillipp, F., and Frank, W., “Study of Point-Defect
Mobilities in Zirconium During Electron-Irradiation in A High-Voltage Electron-Micro-
scope,” J. Nucl. Mater., Vol. 208, No. 3, 1994, pp. 324–334.
[14] De Carlan, Y., Régnard, C., Griffiths, M., Gilbon, D., and Lemaignan, C., “Influence of Iron
in the Nucleation of hci Component Disclocation Loops in Irradiated Zircaloy-4,” Zirco-
nium in the Nuclear Industry (Eleventh International Symposium), STP 1295, American
Society for Testing and Materials International, West Conshohocken, PA, 1996, pp.
638–653.
[15] Tournadre, L., Onimus, F., Béchade, J.-L., Gilbon, D., Cloué, J.-M., Mardon, J.-P.,
Feaugas, X., Toader, O., and Bachelet, C., “Experimental Study of the Nuclea-
tion and Growth of c-Component Loops Under Charged Particle Irradiations of
Recrystallized Zircaloy-4,” J. Nucl. Mater., Vol. 425, Nos. 1–3, 2012, pp. 76–82.
[16] Yamada, S. and Kameyama, T., “Observation of c-Component Dislocation Structures

Formed in Pure Zr and Zr-Base Alloy by Self-Ion Accelerator Irradiation,” J. Nucl. Mater.,
Vol. 422, Nos. 1–3, 2012, pp. 167–172.
[17] Hengstler-Eger, R.-M., Baldo, P, Beck, L, Dorner, J, Ertl, K., Hoffmann, P. B., Hugensch-
midt, C., Kirk, M. A., Petry, W., Pikart, P, and Rempel, A., “Heavy Ion Irradiation Induced
Dislocation Loops in AREVA’s M5 Alloy,” J. Nucl. Mater., Vol. 423, Nos. 1–3, 2012, pp.
170–182.
[18] Foll H. and Wilkens, M., “Transmission Electron Microscope Studies of Dislocation Loops
in Heavy-Ion Irradiated HCP Cobalt,” Phys. Stat. Sol. A, Vol. 39, No. 2, 1977, pp. 561–571.
[19] Griffiths, M. and Gilbert, R. W., “The Formation of C-Component Defects in Zirconium
Alloys During Neutron-Irradiation,” J. Nucl. Mater., Vol. 150, No. 2, 1987, pp. 169–181.
[20] Griffiths, M., Loretto, M. H., and Sallmann, R. E., “Electron Damage in Zirconium: II. Nucle-
ation and Growth of c-Component Loops,” J. Nucl. Mater., Vol. 115, Nos. 2–3, 1983, pp.
313–322.
[21] De Diego, N., Osetsky, Y. N., and Bacon, D. J., “Structure and Properties of Vacancy and
Interstitial Clusters in a-Zirconium,”J. Nucl. Mater., Vol. 374, Nos. 1–2, 2008, pp. 87–94.
[22] Hull, D., and Bacon, D. J., Introduction to Dislocations, Vol. 37, Pergamon Press, New
York, 1984.
[23] Berghezan, A., Fourdeux, A., and Amelinckx, S., “Transmission Electron Microscopy
Studies of Dislocations and Stacking Faults in a Hexagonal Metal: Zinc,” Acta Met-
all., Vol. 9, No. 5, 1961, pp. 464–490.
[24] Hirth, J. P. and Lothe, J., Theory of Dislocations, Wiley, New York, 1982.
[25] Domain, C., 2002, “Simulations Atomiques Ab Initio des Effets de L’hydrogène et de
L’iode dans le Zirconium,” Ph.D. thesis, Université des sciences et technologies de Lille,
Lille, France.
[26] Gilbon, D. and Simonot, C., “Effect of Irradiation on the Microstructure of Zircaloy-4,”
Zirconium in the Nuclear Industry (Tenth International Symposium), STP 1245, American
Society for Testing and Materials International, West Conshohocken, PA, 1994, pp.
521–548.
[27] Christien, F. and Barbu, A., “Cluster Dynamics Modelling of Irradiation Growth of Zirco-
nium Single Crystals,” J. Nucl. Mater., Vol. 393, No. 1, 2009, pp. 153–161.
[28] Woo, C. H., “Theory of Irradiation Deformation in Non-Cubic Metals: Effects of Aniso-
tropic Diffusion,” J. Nucl. Mater., Vol. 159, 1988, pp. 237–256.
[29] Adamson, R. B., Tucker, R. P., and Fidleris, V., “High-Temperature Irradiation Growth in
Zircaloy,” Zirconium in the Nuclear Industry (Fifth International Symposium), STP 754,
American Society for Testing and Materials International, West Conshohocken, PA, 1982,
pp. 208–234.
[30] Griffiths, M., “Microstructure Evolution in Zr Alloys During Irradiation: Dose, Dose Rate,
and Impurity Dependence,” Zirconium in the Nuclear Industry (Fifteenth International
Symposium), STP 1505, American Society for Testing and Materials International, West
[31] Garner, F. A., Wolfer, W. G., and Brager, H. R., “A Reassessment of the Role of Stress in
Development of Radiation-Induced Microstructure,” Effects of Radiation on Structural
Materials, ASTM International, West Conshohocken, PA, 1979, pp. 160–183.
[32] Garner, F. A. and Gelles, D. S., “Irradiation Creep Mechanisms: An Experimental

Perspective,” J. Nucl. Mater., Vol. 159, 1988, pp. 286–309.
[33] Wolfer, W. G., “Correlation of Radiation Creep Theory with Experimental Evidence,” J.
Nucl. Mater., Vol. 90, No. 1–3, 1980, pp. 175–192.
[34] Leteurtre. J., Pouchou, J. L., and Zuppiroli, L., “Experimental Method For Determining
Dislocation Loop Nucleus Size,” Philos. Mag., Vol. 27, No. 6, 1973, pp. 1323–1334.
[35] Matthews, J. R. and Finnis, M. W., “Irradiation Creep Models–An Overview,” J. Nucl.
Mater., Vol. 159, 1988, pp. 257–285.
[36] Brager, H. R., Garner, F. A., and Guthrie, G. L., “Effect of Stress on Microstructure of Neutron-
Irradiated Type-316 Stainless-Steel,” J. Nucl. Mater., Vol. 66, No. 3, 1977, pp. 301–321.
[37] Wolfer, W. G. and Ashkin, M., “Stress-Induced Diffusion of Point-Defects to Spherical

Sinks,” J. Appl. Phys., Vol. 46, No. 2, 1975, pp. 547–557.
[38] Heald, P. T. and Speight, M. V., “Point-Defect Behavior in Irradiated Materials,” Acta Met-
all., Vol. 23, No. 11, 1975, pp. 1389–1399.
[39] Hellouin de Menibus, A., Guilbert, T., Auzouxa, Q., Toffolon, C., Brachet, J.-C., and
Bechadeb, J.-L., “Hydrogen Contribution to the Thermal Expansion of Hydrided
Zircaloy-4 Cladding Tubes,” J. Nucl. Mater., Vol. 440, No. 1–3, pp. 169–177.
[40] Bossis, P., Verhaeghe, B., Doriot, S., Gilbon, D., Chabretou, V., Dalmais, A., Mardon, J. P.,
Blat, M., and Miquet, A., “PWR Comprehensive Study of High Burn-Up Corrosion and
Growth Behavior of M5R and Recrystallized Low-Tin Zircaloy-4,” Zirconium in the Nu-
clear Industry (Fifteenth International Symposium), STP 1505, American Society for Test-
ing and Materials International, West Conshohocken, PA, 2009, pp. 430–456.
in Zirconium Alloys,” Zirconium in the Nuclear Industry (Twelfth International Sympo-
sium), ASTM STP 1354, American Society for Testing and Materials International, West
[42] Vizcaino, P., Banchik, A. D., and Abriata, J. P., “Solubility of Hydrogen in Zircaloy-4: Irradi-
ation Induced Increase and Thermal Recovery,” J. Nucl. Mater., Vol. 304, No. 2–3, 2002,
pp. 96–106.
[43] Lewis, M. B., “Deuterium-Defect Trapping in Ion-Irradiated Zirconium,” J. Nucl. Mater.,

Vol. 125, No. 2, 1984, pp. 152–159.
[44] Vizcaino, P., Banchik, A. D., and Abriata, J. P., “Hydrogen in Zircaloy-4: Effects of the
Neutron Irradiation on the Hydride Formation,” J. Mater. Sci., Vol. 42, No. 16, 2007, pp.
6633–6637.
[45] Chung, H. M., Daum, R. S., Hiller, J. M., and Billone, M. C., “Characteristics of Hydride Pre-
cipitation and Reorientation in Spent-Fuel Cladding,” Zirconium in the Nuclear Industry
(Thirteenth International Symposium), ASTM STP 1423, American Society for Testing and
Materials International, West Conshohocken, PA, 2002, pp. 561–582.
[46] Tournadre, L., Onimus, F., Béchade, J.-L., Gilbon, D., Cloué, J.-M., Mardon, J.-P., and Feau-
gas, X., “Toward a Better Understanding of the Hydrogen Impact on the Radiation
Induced Growth of Zirconium Alloys,” J. Nucl. Mater., Vol. 441, No. 1–3, 2013, pp. 222–231.
[47] Mardon, J.-P., Charquet, D., and Senevat, J., Influence of Composition and Fabrication Process
on Out-Of-Pile and In-Pile Porperties of M5 Alloy, ASTM, West Conshohocken, PA, 2000.
[48] Damcott, D. L., Cookson, J. M., Rotberg, V. H., and Was, G. S., “A Radiation Effects Facil-
ity Using a 1.7 MV Tandem Accelerator,” Nucl. Instrum. Methods Phys. Res. B, Vol. 99, No.
1–4, 1995, pp. 780–783.
[49] Zu, X. T., Sun, K., Atzmon, M., Wang, L. M., You, L. P., Wan, F. R., Busby, J. T., Was, G. S.,
and Adamson, R. B., “Effect of Proton and Ne Irradiation on the Microstructure of Zirca-
loy 4,” Philos. Mag. A, Vol. 85, No. 4–7, 2005, pp. 649–659.
[50] Kearns, J. J., “Diffusion-Coefficient of Hydrogen in Alpha Zirconium, Zircaloy-2 and Zir-
caloy-4,” J. Nucl. Mater., Vol. 43, No. 3, 1972, pp. 330–338.
[51] Bernas, H., Chaumont, J., Cottereau, E., Meunier, R., Traverse, A., Clerc, C., Kaitasov, O.,
Lalu, F., Le Du, D., Moroy, G., and Salomé, M., “Progress Report on Aramis, the 2 MV Tan-
dem at Orsay,” Nucl. Instrum. Methods Phys. Res. B, Vol. 623, No. 3, 1992, pp. 416–420.
[52] Cottereau, E., Camplan, J., Chaumont, J., Meunier, R., and Bernas, H., “ARAMIS: An Ambi-
dextrous 2 MV Accelerator for IBA and MeV Implantation,” Nucl. Instrum. Methods Phys.
Res. B, Vol. 45, No. 1–4, 1990, pp. 293–295.
[53] Ruault, M.-O., Fortunaa1, F., Bernasa1, H., Chaumonta1, J., Kaı̈tasova1, O., and Borodina, V.
A., “In Situ Transmission Electron Microscopy Ion Irradiation Studies at Orsay,” J. Mater.
Res., Vol. 20, No. 7, 2005, pp. 1758–1768.
[54] Carassou, S., Duguay, C., Yvon, P., Rozenblum, F., Cloué, J. M., Chabretou, V., Bernaudat,
C., Levasseur, B., Maurice, A., Bouffioux, P., and Audic, K., “REFLET Experiment in OSIRIS:
Relaxation Under Flux as a Method for Determining Creep Behavior of Zircaloy Assem-
bly Components,” J. ASTM Int., Vol. 7, No. 8, 2010, JAI 103065.
[55] Shishov, V. N., Peregud, M. M., Nikulina, A. V., Pimenov Yu. V., Kobylyansky, G. P., Novose-
lov, A. E., Shebaldov, Ostrovsky, Z. E., Obukhov, A. V., “Influence of Neutron Irradiation
and Dislocation Structure and Phase Composition of Zr-Base Alloys,” Zirconium in the
Nuclear Industry (Fourteenth International Symposium), ASTM STP 1467, American Soci-
ety for Testing and Materials International, West Conshohocken, PA, 2006, pp. 666–685.
[56] Tournadre, L., Onimus, F., Béchade, J.-L., Gilbon, D., Cloué, J.-M., Mardon, J.-P., Feaugas,
X., Toader, O., and Bachelet, C., “Experimental Study of the Nucleation and Growth of C-
Component Loops Under Charged Particle Irradiations of Recrystallized Zircaloy-4,” J.
Nucl. Mater., Vol. 425, 2011, pp. 76–82.
[57] Sawatzky, A., “Hydrogen in Zircaloy-2: Its Distribution and Heat of Transport,” J. Nucl.
Mater., Vol. 2, No. 4, 1960, pp. 321–328.
[58] Christien, F. and Barbu, A., “Effect of Self-Interstitial Diffusion Anisotropy in Electron-
Irradiated Zirconium: A Cluster Dynamics Modeling,” J. Nucl. Mater., Vol. 346, No. 2–3,
2005, pp. 272–281.
[59] Was, G. S., Fundamentals of Radiation Materials Science, Springer-Verlag, Berlin- Heidel-
berg, 2007.
[60] Etienne, A., Radiguet, B., Pareige, P., Massoud, J.-P., and Pokor, A., “Tomographic Atom
Probe Characterization of the Microstructure of a Cold Worked 316 Austenitic Stainless
Steel After Neutron Irradiation,” J. Nucl. Mater., Vol. 382, 2008, pp. 64–69.
[61] Meslin, E., Lambrecht, M., Hernández-Mayoral, M., Bergner, F., Malerba, L., Pareige, P.,
Radiguet, B., Barbu, A., Gómez-Briceño, D., Ulbricht, A., and Almazouzi, A.,
“Characterization of Neutron-Irradiated Ferritic Model Alloys and a RPV Steel from Com-
bined APT, SANS, TEM and PAS Analyses,” J. Nucl. Mater., Vol. 406, 2010, pp. 73–83.
[62] Ribis, J., Onimus, F., Béchade, J.-L., Doriot, S., Barbu, A., Cappelaere, C., and
Lemaignand, C., “Experimental Study and Numerical Modelling of the Irradiation Dam-
age Recovery in Zirconium Alloys,” J. Nucl. Mater., Vol. 403, No. 1–3, 2010, pp. 135–146.
[63] Kearns, J.-J., “Dissolution Kinetics of Hydride Platelets in Zircaloy-4,” J. Nucl. Mater., Vol.
27, No. 1, 1968, pp. 64–72.
[64] Domain, C. and Legris, A., “Investigation of Glide Properties in Hexagonal Titanium and
Zirconium: An Ab Initio Atomic Scale Study,” IUTAM Symposium on Mesoscopic
Dynamics of Fracture Process and Materials Strength, Springer, Dordrecht, The
Netherlands, 2004, pp. 411–420.
[65] Girardin, G. and Delafosse, D., “Solute-Dislocation Interactions: Modelling and Experi-
ments in Hydrogenated Nickel and Nickel Base Alloys,” Mater. Sci. Eng. A, Vols.
387–389, 2004, pp. 51–54.
[66] Oudriss, A., Creus, J., Bouhattate, J., Savall, C., Peraudeau, B., and Feaugas, X., “The Dif-
fusion and Trapping of Hydrogen Along the Grain Boundaries in Polycrystalline Nickel,”
Scr. Mater., Vol. 66, 2012, pp. 37–40.
[67] Puls, M. P., The Effect of Hydrogen and Hydrides on the Integrity of Zirconium Alloy
Components, Springer-Verlag, London.
[68] Shinohara, Y., Hiroaki, A., Takeo, I., Naoto, S., Toshiya, K., Hiroyuki, Y., and Tomitsugu, T.,
“In Situ TEM Observation of Growth Process of Zirconium Hydride in Zircaloy-4 During
Hydrogen Ion Implantation,” J. Sci. Technol., Vol. 46, No. 6, 2009, pp. 564–571.
[69] Griffiths, M., Gilbon, D., Regnard, C., and Lemaignan, C., “HVEM Study of the Effects of
Alloying Elements and Impurities on Radiation Damage in Zr-Alloys,” J. Nucl. Mater., Vol.
205, 1993, pp. 273–283.
[70] Yoo, M. H. and Wei, C. T., “Application of Anisotropic Elasticity Theory to the Plastic De-
formation in Hexagonal Zinc,” Philos. Mag., Vol. 13, 1966, pp. 759–775.
[71] Yoo, M. H., “cþa Dislocation Reactions in H.C.P. Metals,” Scr. Metall. Vol. 2, 1968, pp.
537–540.
DISCUSSION
Question from Shiori Ishino, (Retired) Univ. of Tokyo:—The fact that dislocation
bundles remain after hydride dissolution is interesting. We have observed that the
same remaining dislocations serve as nucleation sites for hydride reprecipitation,
i.e., a sort of "memory effect". Have you observed similar phenomenon?
Authors’ Response:—It is known that <a> dislocations can be created during

precipitation of small hydrides. These <a> dislocations are believed to be responsi-
ble for the “memory effect” during reprecipitation. It is proposed here that <c þ a>
dislocations are also created during precipitation of large hydrides and that these
<c þ a> dislocations could be responsible for the creation of some sort of c-loops
bundles. These <c þ a> dislocations could indeed induce a “memory effect”. Fur-
ther investigations are however required to assess this point.
Question from Michael Preuss, University of Manchester:—Did you exceed the

yield point during the four-point bending experiment?
Authors’ Response:—In the four-point bending experiment, we observed an

effect of stress on c-loops when the stress was applied after 4.1 dpa irradiation. In
that case the applied stress exceeded the yield stress. Similar experiments are in pro-
gress with an applied stress below the yield stress. These experiments confirm the
results presented here.
Question from Ron Adamson, Zircology Plus:—Do I understand correctly that

both H in solid solution and H as hydrides increase <c> loop density and that the
influence of hydrides is to provide additional H to the matrix during irradiation?
Authors’ Response:—This is correct. It is observed that H in solid solution

increases the c-loop density. Progressive dissolution of hydrides also provides addi-
tional H to the matrix during irradiation. Moreover, sort of c-loops bundles are
observed. These c-loops bundles are believed to be due to remaining <c þ a> dislo-
cations after the dissolution of hydrides. Nevertheless, due to the low density of
these hydrides, the resulting macroscopic effect is believed to be low whereas the
increase of c-loops throughout all of the grains of the material must have a signifi-
cant macroscopic effect.
HIGH TEMPERATURE
TRANSIENT BEHAVIOR
M. Négyesi,1,2 J. Krejčı́,3,4 S. Linhart,2 L. Novotný,2

A. Přibyl,2 J. Burda,5 V. Klouček,6 J. Lorinčı́k,7,8
J. Sopoušek,9 J. Adámek,1 J. Siegl,1 and V. Vrtı́lková2
Contribution to the Study of the

Pseudobinary Zr1Nb–O Phase
Diagram and Its Application to
Numerical Modeling of the High-
Temperature Steam Oxidation of
Zr1Nb Fuel Cladding
Reference
Négyesi, M., Krejčı́, J., Linhart, S., Novotný, L., Přibyl, A., Burda, J., Klouček, V., Lorinčı́k, J.,
Sopoušek, J., Adámek, J., Siegl, J., and Vrtı́lková, V., “Contribution to the Study of the
Pseudobinary Zr1Nb–O Phase Diagram and Its Application to Numerical Modeling of the
High-Temperature Steam Oxidation of Zr1Nb Fuel Cladding,” Zirconium in the Nuclear
Industry: 17th International Symposium, STP 1543, Robert Comstock and Pierre Barberis,
Eds., pp. 897–931, doi:10.1520/STP154320120162, ASTM International, West Conshohocken,
PA 2015.10
September 15, 2014.
1
Dept. of Materials, Faculty of Nuclear Sciences and Physical Engineering, Czech Technical Univ. in Prague,
Trojanova 13, 120 00 Praha 2, Czech Republic.
2
UJP Praha a.s., Nad Kamı́nkou 1345, 156 10 Praha - Zbraslav, Czech Republic.
3
Dept. of Nuclear Reactors, Faculty of Nuclear Sciences and Physical Engineering, Czech Technical Univ. in
Prague, V Holešovičkách 2, 180 00 Praha 8, Czech Republic.
4
CHEMCOMEX Praha a.s., Elišky Přemyslovny 379, 156 10 Praha - Zbraslav, Czech Republic.
5 130, 250 68 Rež,
NRI Rez plc, Husinec-Rež Czech Republic.
6
UNIPETROL RPA, s.r.o., Zálužı́ 1, 436 70 Litvinov, Czech Republic.
7
Institute of Photonics and Electronics, Academy of Sciences of the Czech Republic, Chaberska 57, 182 51
Praha 8, Czech Republic.
8
Dept. of Physics of the Faculty of Science, J. E. Purkinje Univ., Ceske mladeze 8, 400 96 Usti nad Labem,
Czech Republic.
9
Ústav chemie, Masarykova univerzita, Kamenice 5, 625 00 Brno, Czech Republic.
10
Hyderabad, India.
Copyright V
ABSTRACT
The pseudobinary Zr1Nb–O phase diagram was recently estimated at the Czech
research institute UJP using a new experimental procedure. It is based on the
oxygen distribution measurement inside fuel claddings previously exposed to
high-temperature steam environments, assuming equilibrium conditions being
fulfilled at the phase boundaries. The experimental results agreed satisfactorily
with the CALPHAD approach, and a further experimental program followed to
affirm the phase diagram. This paper is concerned with additional measurements
of the oxygen concentrations using wavelength-dispersive spectrometry,
secondary ion mass spectrometry, and inert gas fusion techniques and
examination of the microstructure in the Zr1Nb cladding wall after high-
temperature oxidation. The main goal was to check the validity of the recently
proposed Zr1Nb–O phase diagram, or to refine it. The new results of the oxygen
concentration measurements confirmed the phase diagram. After that, a new
diffusion model, the Jakub Krejci oxidation model, was created using the
Zr1Nb–O phase diagram describing the double-sided high-temperature oxidation
of the Zr1Nb fuel cladding including the (a þ b)-Zr layer. The numerical
calculations were compared to the experimental results with satisfactory
agreement. It could be concluded that the proposed experimental procedure
provided a good estimation of the Zr1Nb–O phase diagram. Moreover, it can be
used for alloys containing a greater amount of hydrogen. The Zr1Nb–O phase
diagram may be applicable for oxygen diffusion models predicting the oxidation
behavior of Zr1Nb fuel claddings upon high-temperature oxidation.
Keywords
fuel cladding, oxidation, Zr1Nb–O, modeling, microstructure
Introduction
The microstructure evolution inside fuel claddings upon an loss of coolant accident
(LOCA) is of significant importance, because the fuel cladding integrity after the
thermal transient depends strongly on the microstructure. Generally, the micro-
structure consists of several reaction layers: the oxide layer evolving on the cladding
surface, the adjacent a-Zr(O) layer, and the innermost b-phase, which is marked as
the prior b-phase after cooling down to room temperature. In the Zr1Nb alloy, a-
phase incursions grow from the a-Zr(O) layer toward the b-phase [1–3]. Addition-
ally, a-Zr(O) grains can precipitate inside the b-phase upon both elevated tempera-
ture and cool-down when the oxygen solubility limit is exceeded [4–6]. The prior
b-phase is considered load-bearing because the oxide and a-Zr(O) layers are brittle
as a result of the high oxygen concentration. Depending on the oxygen concentra-
tion, the b-phase can be ductile or brittle [2,7,8]. Thus the fraction of the b-phase
and the oxygen contained in that phase are the most decisive factors regarding fuel
cladding embrittlement. The oxygen uptake in the b-phase is influenced mainly by
temperature and exposure time. The fuel claddings can also absorb a significant
NÉGYESI ET AL., DOI 10.1520/STP154320120162 899
amount of hydrogen during both the in-service corrosion and the LOCA [7,9–11].
The absorbed hydrogen elevates the oxygen solubility in the b-phase, which leads
to higher risk to the cladding embrittlement as well as oxygen [2,6,7]. Another im-
portant issue is cladding embrittlement because of hydrides precipitated during the
cooling-down process [12,13].
More precise prediction of the oxidation degree might lead to better assessment
of the oxidation criterion in LOCA safety analysis. Nowadays, the 17 % equivalent
cladding reacted (ECR) criterion tied to empirical oxidation correlations is
employed. However, the 17 % ECR criterion has shown itself to be insufficient in
several cases [7,10,14]. The current demands require a more accurate description of
the ongoing oxidation process inside the fuel cladding wall upon the LOCA. In
order to predict the oxidation behavior of the fuel rods via a computational tool,
one needs thermodynamic data such as the equilibrium oxygen concentrations at
phase boundaries. Several attempts to study the Zr1Nb–O system have been made
so far [2,8,15–17]. Recently, a Zr-rich part of the pseudobinary Zr1Nb–O phase dia-
gram was estimated at UJP (a Czech research institute) via a new experimental pro-
cedure. It is based on the oxygen distribution measurement inside fuel claddings
previously exposed to high-temperature steam environments and assumes that
equilibrium conditions have been fulfilled at the phase boundaries. The experimen-
tal results agreed satisfactorily with results from the CALPHAD approach, and a
further experimental program followed to affirm the Zr1Nb–O phase diagram. The
present paper is concerned with additional measurements of oxygen concentrations
using wavelength-dispersive spectrometry (WDS), secondary ion mass spectrome-
try (SIMS), and inert gas fusion techniques and examination of the microstructure
in the Zr1Nb cladding wall after high-temperature oxidation. The main goal was to
check the validity of the recently proposed Zr1Nb–O phase diagram or contribute
to its refinement. The following paragraphs summarize the achieved results and
present new experimental results. The phase diagram was also validated using an
in-house calculation tool employing the Jakub Krejci oxidation model (JKOX). The
influence of hydrogen is also discussed. The last paragraphs link the microstructure
to the mechanical properties. The recently proposed oxidation criterion Ob is
treated as well.
Experimental
EXPERIMENTAL MATERIAL
All samples examined in this study were fabricated from modified E110 alloy with
improved oxidation properties (partially fabricated from Zr sponge). The chemical
composition is presented in Table 1. The tested tubular specimens had an outer di-
ameter of 9.1 mm and a wall thickness of 686 lm. Non-irradiated segments 30 mm
in length were either as-received or corroded in steam at 425 C and 10.7 MPa to
form an oxide layer 2 (6 days), 10 (63 days), or 20 lm thick (189 days). The corro-
sion procedure was performed in static conditions in autoclaves of 4.5-dm3 volume.
TABLE 1 The chemical composition of the studied material.
Wt. % Nb O Fe Zr
Zr1Nb 1.0–1.1 0.08 0.02 Balanced
A large number of specimens (more than 100) were exposed at a time. After a given
period, the autoclave was always switched off, opened, and cooled down. The speci-
mens were withdrawn, rinsed with distilled water, dried out at 80 C, and then
weighed. After the corrosion procedure, the specimens were exposed to steam
(0.1 MPa) at high temperatures (850 C to 1300 C) for variable time intervals (0 to
480 min, depending on the temperature) in a resistance furnace. Tests were con-
ducted with average steam flow rates of 0.61 g/cm2/min (deduced from water con-
sumption). A single sample was exposed at a time. The oxidation was double-sided.
The sample temperature was measured by a thermocouple placed inside the tube.
After the high-temperature oxidation, the samples were quenched in ice water.
Both corrosion and oxidation procedures were carried out with no applied stress.
EXPERIMENTAL METHODS
After the thermal processing, the specimens were cut into several rings for further
experimental investigation. The hydrogen content was measured using vacuum
extraction on an Exhalograph EA–1. A NEOPHOT 21 light microscope, a JSM
5510LV scanning electron microscope, and a LUCIA G image analyzer were used
for the microstructure observation. Measurements were conducted using cross-
sections. For the determination of the mechanical properties, ring compression tests
(RCTs) followed. The RCT is not a standardized testing method for the evaluation
of the plastic deformation ability of the material under study for material embrittle-
ment. Nevertheless, it has been the most widely used ductility test for determining
cladding embrittlement after LOCA simulation because of its simplicity—it requires
minimum sample preparation and only a limited amount of material. The sample
(a short cladding ring approximately 10 mm long) is compressed between two flat
pistons. The loading force and the displacement of one of the pistons are recorded.
Generally, the residual ductility is employed to characterize the specimens with a
limit value of 1 % to 2 % for the ductile-to-brittle transition [18]. A limit value of
1 % was applied in this study. The RCT was carried out on an Instron 1185 tensile
testing machine with the Instron SFL temperature chamber at a temperature of
135 C. The output force was recorded versus time. The rate of the cross beam was
1 mm/min. Figure 1 depicts an example of the RCT working curve. The residual duc-
tility Ac was evaluated from the working curve. It was defined as the nominal plastic
hoop strain at failure of the ring specimen (see the definition in Fig. 1). Fracto-
graphic analyses of several fractured samples were then performed. The oxygen dis-
tribution was determined using the following experimental methods.
FIG. 1 An example of an RCT working curve.
X-ray Microanalysis
An INCA Wave 700 wavelength-dispersive spectrometer and a Johan-Johansson
fully focusing six-crystal spectrometer (Oxford Instruments) were used for the
determination of the solute concentration. An INCA Energy 350 energy-dispersive
spectrometer (spectral resolution: 133 eV) was used to determine the Zr concentra-
tion. The cross-sections (fixed in a conductive thermoset) were used for x-ray
microanalysis measurements. Air-formed surface oxide layers were removed from
the samples. The samples were polished with an OP-S suspension of a colloidal sili-
cate (Struers) and coated with carbon (15 nm) immediately after that. The carbon
coating was used to prevent sample charging. SiO2 as a standard and LSM80N with
a resolution of 17 eV as a crystal (full width at half-maximum) were used for the ox-
ygen analyses. The high oxygen standard was employed instead of a low oxygen
standard because the oxide layer, containing around 25 wt. % oxygen, was involved
in the wavelength dispersion analysis as well. However, those results are not pre-
sented here.
Secondary Ion Mass Spectrometry

An Atomika 3000 secondary ion mass spectrometer was used with the following pa-
rameter settings: Csþ primary beam; kinetic energy, 15 keV; spot size, 100 to
200 lm; impact angle, 45 ; ion current, 750 to 950 nA; raster area, (500 by 500)
lm2; acceleration voltage, 200 V; energy window, 10 eV; negative secondary ion
mode of operation; count time, 1 s; gating, 18.75 % (linear); vacuum pressure during
the measurement, 2 109 Torr. The depth scale was quantified using a stylus
profilometer. The sample surface was coated with a 10 nm gold thin film. The ox-
ygen concentration was calibrated using a set of nine implanted calibration stand-
ards of O18 (U ¼ 5 1014 at./cm2 at 360 keV) in Zr alloys of variable degrees of
oxidation, which allowed for the construction of a working curve. Tangential sec-
tions were used for SIMS analyses. A detailed description of the experimental set-
ting was already published elsewhere [19].
Inert Gas Fusion

The determination of oxygen concentration using the inert gas fusion technique
consists of thermal decomposition of samples in a graphite crucible in a flowing
stream of inert gas (He). A detailed description can be found in Ref 19. In this
work, a LECO TC500C Nitrogen/Oxygen Determinator with a thermal conductivity
cell/solid state infrared detector was used. An RM LECO 502-201 titanium pin was
employed as a calibration standard. The oxide and a-Zr(O) layers had to be
removed mechanically. After the machining, the sample was pickled in a 4 % HF
solution. Figure 2 shows a light micrograph of the sample after the machining. One
can easily observe the remaining a-Zr(O) phase on the outer surface. In such cases,
more thorough pickling had to be carried out to ensure that the a-Zr(O) phase was
removed. The samples were then rinsed with water, degreased in acetone, and dried
with warm air. The sample weights were approximately 0.1 g.
Microhardness and Nanohardness Measurements

Microhardness measurements were carried out on an OPL microhardness tester
with an automatic-shift tip and a Vickers indenter. The force was induced by a
100 -g weight so that the diagonals of projected surfaces of the indents had a length
of tens of micrometers.
FIG. 2 Light micrograph of a sample cross-section after machining and before pickling.
The nanohardness measurements were carried out using a Nano Indenter XP

(MTS System Corp. USA) with an automatic shift-tip and a Berkovich indenter.
The force was induced by a 0.8 -g weight so that the projection of the indentation
surface was about 0.5 to 2 lm2 and the indentation depth was less than 500 nm.
The polished cross-sections were used for both microhardness and nanohardness
measurements.
In order to predict the oxidation behavior (the evolution of the reaction layers, the
kinetics of the oxygen pickup in the b-phase, etc.) of the fuel claddings upon the
LOCA via a calculation tool, one must have experimental data such as the equilib-
rium oxygen concentrations and the oxygen diffusivities. The zirconium alloys,
which are used as material for the fabrication of the fuel rods, are low alloy. That is
why the equilibrium oxygen concentrations from the Zr–O phase diagram [20] and
oxygen diffusivities in pure Zr (hexagonal a-Zr and cubic b-Zr) or in ZrO2 (mono-
clinic, tetragonal, or cubic phase) [21] can be used instead. Figure 3 presents a typi-
cal microstructure with a schematic drawing of the oxygen distribution inside the
wall of the Zr1Nb fuel cladding exposed to a high-temperature steam environment.
The resulting layers and the oxygen concentrations at the phase boundaries are
highlighted. In previous works [6,22], the WDS results confirmed that the oxygen
concentrations measured in the oxide (Cox/a) and at the oxide–metal phase bound-
ary in a-Zr(O) (Ca/ox) are approximately equal to the equilibrium concentrations
coming from the Zr–O phase diagram. A value of 6.7 wt. %, which can be consid-
ered as the ceiling of the oxygen concentration in a-Zr for a temperature range of
950 C to 1300 C [20], was measured at the oxide–metal interface in most of the
samples with experimental scatter of less than 10 %. The weak dependence of the
Ca/ox oxygen concentration on the temperature was not measured. A value of
25 wt. % was measured in most the oxide layer, which is also close to the equilib-
rium value [20]. Again, the dependence of the equilibrium oxygen concentration on
the temperature is negligible in the temperature range of 950 C to 1300 C and,
thus, was not measured. Concerning the Ca/aþb and Cb/aþb oxygen concentrations,
the influence of the main solutes (Nb or Sn) is not negligible. Zr-rich pseudobinary
phase diagrams of Zr-alloy-oxygen must be determined (from 0 up to 5 wt. % of ox-
ygen, approximately). The pseudobinary Zry-4-oxygen phase diagram convenient
for predicting the oxidation behavior upon LOCA was already published elsewhere
[23]. As a consequence, the numerical modeling of the oxidation behavior of Zry-4
fuel cladding is well developed nowadays. Several works have been devoted to the
assessment of the Zr1Nb–O phase diagram [2,8,15–17]. The first estimation of the
pseudobinary Zr1Nb–O phase diagram, which might be used for modeling high-
temperature oxidation, was recently made [6]. It was derived based on the oxygen
concentration measurements inside the fuel cladding wall after high-temperature
steam oxidation assuming equilibrium conditions at the phase boundaries during
FIG. 3 A typical microstructure (bottom) with a schematic drawing of the oxygen

distribution (top) inside the wall of Zr1Nb fuel cladding exposed to steam at
1100 C for 15 min (double-sided oxidation).
the oxidation. All samples were quenched from the experimental temperature by a
sudden fall into ice water directly from the heated-up furnace, so that the oxygen
redistribution upon cool-down was considered negligible. Consequently, the meas-
ured oxygen concentrations at phase boundaries of the quenched sample sections
could be used for estimating the phase diagram, assuming that the oxygen concen-
trations at the phase boundaries were close to the equilibrium ones upon high-
temperature oxidation at elevated temperatures. Additionally, because of the
improved oxidation properties (the adherent oxide layer and low hydrogen uptake),
the influence of hydrogen could be excluded. On the other hand, using pre-
corroded samples containing a greater hydrogen content, the influence of hydrogen
on the equilibrium oxygen concentrations can be investigated, too.
OXYGEN DISTRIBUTION
Ca/a1b Oxygen Concentration
The measurements made by the wavelength-dispersive spectrometer microprobe
were realized in several lines perpendicular to the oxide–metal interface. Figure 4(a)
shows an example of the measured oxygen concentration profile of a sample oxi-
dized at 1300 C for 3 min. The lines lead across the a-Zr(O) layer (showing an
approximately linear decrease of measured values) and the (a þ b)-Zr region
(showing an experimental scatter increase due to the occurrence of two phases with
FIG. 4 The results of the oxygen and niobium concentration measurements: (a) the
sample oxidized at 1300 C for 3 min; (b) SEM image of the sample surface—the
location of the measured profiles.
different oxygen solubilities and a decrease in the slope) toward the middle of the
cladding wall (the prior b-phase region) with values independent of the distance.
The slight decrease in the oxygen concentration in the prior b-phase region toward
the wall center is not so apparent. This is probably due to the oxygen saturation in
the b-phase. The measured values also display substantial experimental scatter in
the region, probably because of solute redistribution. Figure 4(b) presents a scanning
electron microscopy (SEM) image of the location of the WDS analyses. As can be
seen, the prior b-phase region contains dark gray areas, which are assumed to be
the a-Zr(O) phase. Those areas should be rich in oxygen, but the oxygen level in
the vicinity of the areas is lower than the matrix oxygen content. That microstruc-
ture is typical for samples containing a higher amount of hydrogen [24]. The meas-
ured hydrogen content in this case was 840 wppm. However, the hydrides are not
visible on the SEM image. The samples were not etched to make them visible, and,
additionally, the hydrides are supposed to be tiny because of the rapid cooling.
Figure 4(a) also shows the concentration profile of niobium. Niobium is a b-stabi-
lizer [25], and thus it diffuses into the b-phase region (see Fig. 5, in which an SEM
image of a sample surface and the surface analysis of the niobium content are
depicted together with a line scan of Nb). There are areas within the a-Zr(O) layer
and the (a þ b)-Zr region where the measured niobium concentration is far greater
than the nominal concentration of Nb in the Zr1Nb alloy. The presence of such
high Nb concentrations in (a þ b)-Zr is expected because that region contains the
prior b-phase. Though the a-Zr(O) layer should not contain any b-phase, there are
tiny spots of the phase, probably because of the slower diffusion of Nb relative to
oxygen. Those spots are highly rich in niobium. One can easily indicate the
FIG. 5 (a) SEM image of a sample surface; dark gray areas ¼ a-Zr(O) phase, light gray
areas ¼ prior b-phase, black area on right-hand side ¼ oxide. (b) Surface analysis
of niobium content together with a line scan of Nb for the sample oxidized at
950 C for 60 min.
positions of the a–a þ b and a þ b–b phase boundaries from the graph. The
a–a þ b interface is situated around 50 lm and the a þ b–b approximately 130 lm
from the oxide–metal phase boundary (according to the WDS results in the area of
the measured profile).
The determination of the exact value of the oxygen concentration at the
a–a þ b phase boundary is not so obvious in some cases because there is no steep
decrease in all measured values like in the case of Zry-4 alloy [22,26] due to the
occurrence of a-Zr(O) incursions. Thus the following procedure, described in five
steps, was adopted to determine the Ca/aþb oxygen concentration:
• Checking of the slope—there might be a change in the slope of the oxygen
concentration at the interface.
• Checking of the scatter—greater scatter in the measured oxygen concentration
indicates a two-phase (a þ b)-Zr region.
• Checking of the niobium profile—there may be locations containing more
than 1 wt. % (a nominal content) of niobium in front of the a-Zr(O) layer
[Fig. 4(a)]. However, those locations can also appear inside the a-Zr(O) layer,
which might be confusing.
• Checking of the metallographic examination—the metallographic results of
the a-Zr(O) layer thickness measurements can be used for the determination
of the Ca/aþb oxygen concentration, too.
• Checking of the values measured after the estimated location of the a–a þ b
interface—oxygen concentrations at the a–a þ b phase boundary must not be
lower than the values measured in the (a þ b)-Zr and prior b-phase regions.
The presented procedure can be demonstrated step by step using the WDS
results plotted in Fig. 4(a). There is a visible change in the slope of the measured
oxygen concentration at a 55-lm distance from the oxide–metal interface (it is
highlighted by the linear regression lines). Behind this distance there is an
increase in the experimental scatter regarding the oxygen concentration. However,
no oxygen concentrations close to those measured in the b-phase occur in the vi-
cinity of the 55-lm distance. The explanation is that the double-phase region of
(a þ b)-Zr is preferentially occupied by the a-Zr(O) phase in this case, as can be
seen from Fig. 4(b). Additionally, only one profile was measured in this sample
(within the a-Zr(O) layer). More results (more parallel profiles) would be needed
for more accurate analysis. There is an increase in Nb at a 40-lm distance from
the oxide–metal interface and then directly behind the 55-lm distance. Using
the image of the location of the measured profile along with the results of the
metallography measurement, one can easily observe that the a–a þ b interface is
located rather close to the 55-lm point. Metallography measurements employing
two different methods gave 54 lm when using light microscopy (LM) and 65 lm
when using SEM micrographs (see below). By averaging all of the presented esti-
mations of the interface location, one can determine the mean value of the
a–a þ b location. After that, the oxygen concentration at the a–a þ b phase
boundary can be assessed using linear regression via the equation depicted in the
plot in Fig. 4(a).
The nanohardness measurements were carried out much like the WDS meas-
urements, that is, in lines perpendicular to the oxide–metal interface leading via a-
Zr(O) and (a þ b)-Zr layers into the prior b-phase region. Figure 6(a) shows an
example of nanohardness results plotted versus the distance from the oxide–metal
phase boundary. A tendency similar to that in the case of the oxygen concentration
can be seen. The nanohardness decreases linearly in a-Zr(O). The decrease is lower
and the experimental scatter is higher in the (a þ b)-Zr region. The nanohardness
FIG. 6 (a) The results of the nanohardness measurement of the sample oxidized at
1150 C for 9 min. (b) Image of the indentation rows.
values in the prior b-phase region are constant (independent of the distance from
the oxide–metal interface). Again, one can easily distinguish the locations of both
interfaces. The a–a þ b interface is close to 50 lm (65 lm) from the oxide–metal
phase boundary, and the a þ b–b interface is around 100 lm (65 lm) from the
oxide–metal phase boundary. Relative to the WDS measurements of the sample
exposed at 1300 C for 3 min, many more measurements of the nanohardness of the
sample treated at 1150 C for 9 min were conducted—20 parallel rows. One can eas-
ily observe the substantial experimental scatter. That is why plenty of measurements
need to be performed in order to get more accurate and statistically meaningful
results. The nanohardness also depends on the quality of the sample surface to be
analyzed. Many cracks close to the oxide–metal phase boundary can cause a
decrease in measured nanohardness values. This must be taken into consideration,
and those points must be extracted from any further analysis. On the other hand,
the elevated nanohardness values measured in the prior b-phase region correspond
to the precipitated a-Zr(O), which generally contains more oxygen than the matrix.
It can be concluded that the nanohardness depends strongly on the oxygen
concentration. This conclusion was already drawn in a previous work [27]. For the
determination of the values at the a–a þ b interface, the same rules were used as in
the case of the oxygen concentration (except using the niobium concentration pro-
file). The approximate values of the nanohardness at the a–a þ b interface were
plotted versus the oxygen concentration measured at the interface. The relationship
between the two magnitudes was assessed. The new WDS results confirmed the
relation and refined it slightly relative to the old results published in Ref 27. Figure 7
plots the oxygen concentration at the phase boundary against the nanohardness.
The trend is fitted by the linear regression. The plot also shows the equation of the
regression line. The graph contains results for two alloys, Zry-4 and Zr1Nb. It can
be concluded that the relation is equal for both alloys. The highest value of the
nanohardness and the oxygen concentration corresponding to the metal–oxide
phase boundary value (in the metal) are also plotted in the graph and taken into
consideration. After the relation between the nanohardness and the oxygen concen-
tration at the a–a þ b interface had been acquired, a comprehensive series of nano-
hardness measurements followed, with the advantage of lower cost and higher
throughput relative to the WDS measurement. The nanohardness values at the
a–a þ b interface in the a-Zr(O) layer were determined for specimens exposed in
the temperature range of 950 C to 1200 C for different time intervals. The mean
values for the constant temperatures were then assessed. Eventually, the oxygen
concentrations at the a–a þ b interface in the a-Zr(O) layer were calculated from
those values using the equation depicted in Fig. 7.
Cb/a1b Oxygen Concentration

The WDS setting used within the framework of the previous experimental program
was not convenient for the determination of the Cb/aþb oxygen concentration
because of the method’s inaccuracy [6]. However, new WDS analyses give
FIG. 7 The oxygen concentration at the phase boundary versus the nanohardness. The
points were fitted via linear regression.
reasonable results for higher temperatures (i.e., higher oxygen contents in the prior
b-phase). The results for 1250 C and 1300 C are plotted in Figs. 10(a) and 10(b).
The experimental values of the Cb/aþb oxygen concentration (coming from the
WDS analysis) were determined as the mean values of all the measurements con-
ducted within the prior b-phase. The oxygen solubility limit was considered to have
been achieved in the samples [see Fig. 4(a)]. Another procedure for the Cb/aþb oxy-
gen concentration measurement was proposed for the rest temperatures [6]. The
oxygen solubility limit in the b-Zr phase is achieved and exceeded after a certain ex-
posure time upon high-temperature oxidation, and it is assumed that it proceeds in
the whole b-phase region at approximately the same time because of the small
thickness and the double-sided oxidation. Microhardness measurements were used
for the determination of the oxygen saturation time in the b-phase instead of the
oxygen concentration measurement, with the advantages of lower cost and higher
throughput. Those measurements were carried out at random places in the prior b-
Zr region. The averages of at least 10 indentations were plotted against the exposure
time for constant temperature and constant hydrogen content (see an example in
Fig. 8). After that, the oxygen concentration was measured. Oxygen is a light ele-
ment, so it is not easy to determine its exact content in Zr, which is a quite heavy
element relative to oxygen. Additionally, only a small amount of oxygen (up to
1 wt. % or rather less) diffuses into the b-phase during the high-temperature oxida-
tion. Thus, two experimental techniques (SIMS and inert gas fusion) were used for
determining the oxygen concentration. These two methods analyze a much larger
FIG. 8 The results of SIMS, inert gas fusion, and microhardness measurements of as-
received samples exposed to steam at 1100 C plotted versus the exposure time.
The time dependences are highlighted by the curves and the plateau region is
outlined by an ellipse for clarity.
volume than WDS—with SIMS, the scan size is (500 by 500) lm2 with a depth of 2
to 5 lm, and inert gas fusion analyzes approximately the whole b-phase region. The
oxygen concentration was plotted together with the microhardness results into
graphs as a function of the exposure time (Fig. 8). The results of both SIMS and
inert gas fusion analyses were in satisfactory agreement. The results (for both
microhardness and oxygen concentration) showed a similar tendency for all tem-
peratures. At first, both values increase with increasing exposure time. Then there is
a plateau, and after a certain time the values begin to increase again (in the case of
the oxygen concentration, the tendency is not so apparent because of the smaller
amount of experimental data; that is why the microhardness measurements were
carried out instead). Eventually, one can obtain an approximate oxygen saturation
time, which is at the beginning of the plateau region. Consequently, using the
results of SIMS and inert gas fusion, one can estimate the approximate ceiling of
the oxygen concentration in the b-phase Cb/aþb. In order to get more accurate and
statistically meaningful results, all measured values within the plateau region were
considered. The resulting ceiling of the oxygen concentration in the b-phase was
then determined as the mean value.
Light micrographs of the cross-sections (Fig. 9) confirmed the above conclu-
sion. The a-Zr(O) grains precipitated in the prior b-phase region because the oxy-
gen solubility limit was exceeded (at 15 and 30 min). The microhardness also
depends on the hydrogen content due to hydride precipitation. Nevertheless, the
FIG. 9 LM images of cross-sections: (a) sample exposed to steam at 1100 C for 9 min
showing no precipitation of a-Zr(O) grains in the prior b-phase region; (b), (c)
samples exposed to steam at 1100 C for 15 or 30 min, respectively, showing
precipitation of a-Zr(O) grains in the prior b-phase region.
hydride effect on the microhardness can be neglected in the case of small amounts
of hydrogen (most samples in the presented study) or a constant amount of hydro-
gen at the same temperature (e.g., samples pre-oxidized to a 20 -lm-thick corrosion
layer contain approximately 600 wppm of hydrogen even after the high-
temperature oxidation).
Zr1Nb–O PHASE DIAGRAM

CALPHAD Calculations
Sopoušek and Svobodová [28] have calculated the Zr1Nb–O phase diagram via the
CALPHAD approach using their own thermodynamic database, Zr_BASE (involv-
ing Zr, Nb, O, and H), which involves thermodynamic parameters of the phases
from publicly accessible sources. The resulting database was verified based on
known binary (Zr–O, Zr–Nb, Zr–H, and Nb–O) and ternary (Zr–Nb–O and
Zr–O–H) subsystems. There were neither fittings nor optimization due to currently
carried out experiments. The Zr1Nb–O phase diagrams for 1000, 2000, and 3000
wppm hydrogen were then calculated. Figure 10 shows the calculated pseudobinary
Zr1Nb–O phase diagram for a temperature range of 400 C to 1400 C and an oxy-
gen concentration range of 0 to 6 wt. % for 0 and 1000 wppm of hydrogen.
Comparison of the Experimental Data With the CALPHAD Calculations

Concerning the previously published experimental results [6], very good agreement
was achieved between the CALPHAD calculations and the experimental results,
especially for the oxygen concentrations Cb/aþb. However, more results were
required for more precise analysis, mainly for the oxygen concentration Ca/aþb.
Figure 10(a) shows the new experimental results for the oxygen concentrations
Ca/aþb coming from the WDS measurements. The new results were obtained for
temperatures of 1000 C, 1100 C, 1250 C, and 1300 C. They are in satisfactory
FIG. 10 Pseudobinary Zr1Nb–O phase diagram for temperatures from 400 C to 1400 C
and oxygen concentrations from 0 to 6 wt. % estimated from the WDS and
nanohardness measurements: (a) only WDS analyses; (b) WDS analyses along
with nanohardness analyses. The lines indicate calculated a–a þ b and a þ b–b
phase interfaces.
agreement with the CALPHAD calculations. The oxygen concentrations Cb/aþb

stayed unchanged. For the temperatures 1250 C and 1300 C, two new estimations
of the Cb/aþb oxygen concentrations were added employing the WDS analysis.
They are rather close to the calculated equilibrium values for 1000 wppm of hydro-
gen. Taking into account that the samples can pick up more hydrogen upon high-
temperature oxidation (because of the higher temperatures), this is a very good
result.
The Ca/aþb oxygen concentrations come from only one sample exposed at a
given temperature. However, several different definitions of the location of the
a–a þ b interface were used (see above), because this is considered to be the most
decisive factor affecting the estimated value of Ca/aþb. Then the oxygen concentra-
tion at the phase boundary was determined via linear regression. That is why the
oxygen concentrations Ca/aþb at other temperatures were also shifted relative to the
previously published results (due to the new metallographic results). There is still
high experimental scatter in several cases. More experimental results are still needed
for more accurate analysis. For this purpose, the nanohardness measurements were
conducted with the advantages of lower cost and higher throughput. Figure 10(b)
presents the mean values of the Ca/aþb oxygen concentrations determined from the
nanohardness measured at the a–a þ b interface compared to the pseudobinary
Zr1Nb–O phase diagram. The error bars quantify the experimental scatter. The
lowest scatter was quantified at 1100 C (only 1.6 %), and the highest at 1000 C
(13 %). The nanohardness measurements of specimens exposed at 850 C, 900 C,

1250 C, and 1300 C have not been carried out yet. It can be concluded that the
Ca/aþb oxygen concentrations determined using the nanohardness measurements
are all in the vicinity of the calculated equilibrium oxygen concentrations. The ex-
perimental scatter here is lower than in the plot presented in Fig. 10(a), where the
oxygen concentrations Ca/aþb come from only one measured sample. Thus
the previous experimental data were refined using the new experimental data. Con-
sequently, better agreement between the experimental results and the calculated
phase diagram has been obtained, and it can be concluded that the new experimen-
tal results have confirmed the calculated phase diagram.
The results of the Cb/aþb measurements were also compared to the results of
other authors [15] with satisfactory agreement. In contrast, the results of Ref 8 are
systematically shifted to lower temperatures for given oxygen concentrations (con-
cerning Cb/aþb). Figure 10 also presents the results of Ref 8 for the Ca/aþb oxygen
concentrations, but only for low values (up to 0.5 wt. %). These results seem to be
systematically shifted to higher temperatures for given oxygen concentrations. This
shift (for both Ta/aþb and Taþb/b temperatures) relative to the experimental results
and the calculations of the present study is probably caused by the definitions of the
phase transformation temperatures. The values presented in Ref 8 were determined
via dilatometry upon cooling, whereas the Taþb/b temperature corresponded to
10 % of precipitated a-Zr grains and the Ta/aþb temperature corresponded to
90 % of precipitated a-Zr grains. Hence, the Taþb/b temperature seems to be lower
and the Ta/aþb temperature seems to be higher relative to our results. This is prob-
ably the main difference between the results of the present study and those of Ref 8.
Numerical Calculations
DESCRIPTION OF THE PROBLEM
Modeling of the oxygen diffusion into the cladding wall upon high-temperature oxida-
tion is a complex problem. The exact solution requires using a model based on the
multicomponent diffusion of all solutes in three dimensions including diffusion along
the grain boundaries. However, simplified solutions, which give reasonable results, are
sufficient. Many calculation tools have been created so far regarding the oxygen diffu-
sion into the fuel cladding upon high-temperature oxidation (see, e.g., Refs 26 and 29).
Most of them are based on the solution of the second Fick law (Eq 1) together with the
mass balance equation, which is satisfied on each interface (Eq 3). This is also the case
for the presented model JKOX. The diffusion equation can be expressed as
@Cðr; tÞ 1 @
(1) ¼ ðrJr Þ
@t r @r
where
@Cðr; tÞ
(2) Jr ¼ DðTÞ
@r
The mass balance equation can then be written as

ds Ji ðr; tÞ Jj ðr; t Þ
(3) ¼
dt Ci C j
where:
J ¼ oxygen mass fluence, kg/m2/s,
D ¼ oxygen diffusivity, m2/s,
C ¼ oxygen concentration, kg/m3, and
i, j ¼ ox, a, b.
DESCRIPTION OF THE JAKUB KREJCI OXIDATION MODEL

The calculation is treated employing the one-dimensional finite difference method
(with the implicit solution) using the Gaussian elimination process. The mesh stays
unchanged during the calculation. It is recalculated in each time step to fit the
motion of the interfaces. The concentrations are then determined using parabolic
interpolation. The diffusion in the axial and circumferential directions is neglected,
as well as the grain boundary diffusion. It calculates in cylindrical coordinates. First
the motion of the interfaces is calculated, and then the oxygen concentration pro-
files are determined. The diffusion of niobium is also considered negligible,
although this is not the correct approach, as the diffusion of Nb together with the
diffusion along the grain boundaries probably results in the creation of the a-Zr(O)
incursions into the b-phase region. In this case, the process is simplified so that the
a-Zr(O) incursions are treated as an additional layer, which is called the (a þ b)-Zr
layer (see Fig. 11). The oxygen concentrations at the phase boundaries within the
(a þ b)-Zr phase and the oxygen diffusivity are chosen so that the (a þ b)-Zr layer
thickness fits the experimental results. The JKOX code predicts the double-sided
oxidation. The problem is considered symmetrical (i.e., only half of the cladding
wall is treated). The oxygen concentration in the middle of the cladding wall is
determined using the condition of constant flux. For the time being, it can be used
only for temperatures equal to or higher than 1100 C, at which the kinetics of the
oxide layer is parabolic (up to the oxygen saturation in the b-phase).
The oxygen concentrations at the phase boundaries are extracted from the
Zr–O phase diagram and the calculated Zr1Nb–O phase diagram. The oxygen con-
centrations in the (a þ b)-Zr phase at the interfaces were determined by multiplying
the equilibrium oxygen concentrations coming from the pseudobinary Zr1Nb–O
phase diagram. The influence of hydrogen is not included in the code. No pre-
corrosion has been modeled so far. The initial conditions are always chosen in order
to fit the experimental data. The following assumptions are made:
• The temperature is considered as independent of the distance and time.
• The diffusion of oxygen is rate controlling. The Fick laws are applied in each
phase separately.
• The diffusivities depend only on the temperature.
• The mass balance at the phase boundaries indicates the motion of the
interfaces.
FIG. 11 The computed oxygen distribution inside the wall of the Zr1Nb fuel cladding
exposed to double-sided high-temperature oxidation.
OXYGEN DIFFUSIVITIES
Other experimental data required in order for the calculations to be run are the dif-
fusion coefficients. The oxygen diffusivities used in the presented model are the fol-
lowing. For predicting the oxidation behavior above 1100 C, the oxide phase can be
considered tetragonal. Hence, an expression from Ref 21 is employed.
(4) Dox ¼ 0:127 e35140=kT
For the metal phases, the following expressions were used, also from Ref 21:
(5) Da ¼ 3:923 e51000=kT
(6) Db ¼ 0:0263 e28200=kT
where:
k ¼ 1.987 cal/mol/K, and
T ¼ thermodynamic temperature, K.
The oxygen diffusion coefficient in the (a þ b)-Zr layer is a function of the dif-
fusivities Da and Db. The function was designed so that the thickness of the
(a þ b)-Zr layer fits the experimental data for all temperatures considered.
Discussion
This section compares the experimental results to the numerical calculations of the
JKOX code. The first part contains an analysis of the kinetics of the reaction layers.
The second part compares the calculated and measured kinetics of the oxygen
uptake in the b-phase. The last paragraphs are devoted to the connection between
the oxygen distribution inside the fuel cladding wall and the mechanical properties.
The recently proposed oxidation criterion Ob is also treated.
THE KINETICS OF THE REACTION LAYERS

The kinetics of all three layers [oxide, a-Zr(O), and (a þ b)-Zr] was already exam-
ined in previous work [3]. Because of the existence of the (a þ b)-Zr region between
the a-Zr(O) layer and the prior b-phase [the a-Zr(O) incursions leading toward the
prior b-phase region], it is not easy to measure the thicknesses of the reaction layers
(except the oxide layer) in Zr1Nb alloy. That is why two different experimental pro-
cedures were employed. Plenty of measurements have been performed so far. The
statistical approach was then adopted. First, light micrographs were employed for
the assessment of the kinetics. The definition of all layers considered is depicted in
Fig. 12. The measurement of the oxide layer thickness is quite apparent. The a-
Zr(O) layer was defined as a straight layer without areas of the prior b-phase. The
prior b-phase region was defined as an area without a-Zr(O) incursions. However,
the a-Zr(O) grains precipitated within the prior b-phase were included. The
remaining region between the a-Zr(O) layer and the prior b-phase was defined as
the (a þ b)-Zr region. The border between that phase and the prior b-phase is a zig-
zag line, in contrast to the rather straight oxide–a-Zr(O) and a-Zr(O)–(a þ b)-Zr
boundaries. Secondly, a comprehensive SEM investigation was carried out within
this study for the assessment of the kinetics of the reaction layers [especially
a-Zr(O)]. Figure 13 presents the experimental procedure employed using the SEM
images. The phase fraction was measured a certain distance from the oxide–metal
interface. The results were then plotted as a function of the distance. This procedure
FIG. 12 Definitions of the oxide layer, the a-Zr(O) layer, the (a þ b)-Zr region, and the
prior b-Zr region. The lines mark the regions of the phases.
FIG. 13 (a) The results of the a-Zr(O) fraction measurements of the sample exposed to
steam at 1150 C for 6 min. (b) An SEM image of the sample after the
thresholding. The lines indicate the location of the phase boundaries.
has been used by other authors investigating the M5 alloy (Zr1Nb) microstructure
[30]. The a-Zr(O) region is defined as the area where the a-Zr(O) phase fraction is
equal to or greater than 99 %. This definition of the examined layer slightly differs
from that proposed by the authors of Ref 30. So, different thicknesses can be
expected for samples being oxidized under similar conditions, although both alloys
(M5 and E110) are expected to have similar properties. At least five images were
always taken and analyzed because of the irregularity of the interfaces. Thus the an-
alyzed region is much more extensive than the figure shows. Therefore, it may seem
that the phase boundaries are slightly shifted (mainly the a þ b–b phase boundary).
The a-Zr(O) fraction is smaller close to the oxide–metal interface because of the
large amount of cracks in this region. This also must be taken into consideration
when evaluating the measured results.
Figure 14 presents a comparison between the measured and calculated thick-
nesses of the reaction layers for a temperature of 1100 C. Good agreement has been
obtained between the calculated and experimental values. The heating-up phase is
not involved in the code. At the moment, only the isothermal conditions can be
predicted by the code. However, modeling of the thermal transients is one of the
future demands. For the time being, the code starts at the elevated temperature, and
the effect of the heating-up phase is included in the initial conditions (they always
fit the experimental results). That is why the reaction layers have nonzero thick-
nesses at the beginning of the exposure. The calculation shows that the a-Zr(O) and
(a þ b)-Zr layers first increase, and after a certain time they start to decrease
because of the finite thickness of the cladding wall, causing oxygen saturation in the
b-phase (the layers from the outer and inner sides join). The oxide layer is still
increasing in the given range of exposure times. The b-phase is still decreasing
because of the enlargement of the oxide, a-Zr(O), and (a þ b)-Zr layers. After join-
ing the a-Zr(O) and (a þ b)-Zr layers, the decreasing of the b-phase slows down.
Figure 15 shows the comparison between the measured and calculated thick-
nesses of the reaction layers for all temperatures considered in the present study.
Good agreement has been achieved for all reaction layers. The measured oxide
FIG. 14 Comparison of the measured and calculated reaction thicknesses for the
specimens oxidized at 1100 C.
FIG. 15 Comparison of the measured and calculated reaction thicknesses: (a) the oxide
layer; (b) the a-Zr(O) layer; (c) the a-Zr(O) þ (a þ b)-Zr layer; (d) the prior b-
phase layer.
layers are slightly underpredicted, mainly for higher thicknesses. The decline might
be a consequence of ongoing changes in the oxide properties (e.g., the oxide den-
sity) during the oxide growth. The greatest experimental scatter was obtained for
the a-Zr(O) layer and the (a þ b)-Zr layer. This is assumed to be a consequence of
the difficult definition of the interfaces between the layers. Figure 15(b) presents the
results of two different methods that were employed in this study. There is no dif-
ference between the two experimental procedures [upon measuring the thickness of
the a-Zr(O) layer].
OXYGEN CONTENT IN b-PHASE

The oxygen pickup in the b-phase is also of high importance, as the residual ductil-
ity of the fuel rods after the LOCA depends mainly on the oxygen level in that
phase. That is why it is necessary to know the kinetics of the oxygen uptake in the
b-phase. Figure 16 shows the comparison between the oxygen concentrations meas-
ured in the prior b-phase and the calculated ones. They come just from the one
value calculated in the center of the b-phase. That is why the model rather under-
predicts the experimental results. Another reason for the underprediction is the use
of the equilibrium oxygen concentrations determined via the CALPHAD approach
in the JKOX code. Those are systematically shifted to lower values relative to the ex-
perimental concentrations (see Fig. 10). On the other hand, the experimental
results—the oxygen concentrations measured in the prior b-phase—might be ele-
vated by the innermost a-Zr(O) incursions, which were not removed by the pick-
ling. Nevertheless, satisfactory agreement has been achieved for all temperatures
considered in this paper.
After the oxygen solubility limit in the b-phase was exceeded, the a-Zr(O)
grains began to precipitate in the area, and consequently the oxygen uptake contin-
ued to increase in the b-phase. However, this phenomenon is not included in the
code, and this causes greater misfit for higher exposure times. The overall underpre-
diction (for all temperatures considered in this study) can be seen in Fig. 17. The
underprediction is more apparent for higher values of oxygen concentration [i.e.,
for the exposure times when the oxygen solubility limit in the b-phase had already
been exceeded and the a-Zr(O) grains had precipitated inside].
FIG. 16 Comparison of the measured and calculated oxygen concentrations in the b-

phase for the specimens oxidized at 1100 C.
FIG. 17 Comparison of the measured and calculated oxygen concentrations in the b-

phase.
THE CONNECTION BETWEEN THE MICROSTRUCTURE AND THE MECHANICAL

PROPERTIES
The alteration of the microstructure after the LOCA affects the residual ductility of
the fuel claddings substantially. That is why the link between the microstructure
and the mechanical properties of the fuel cladding is of high importance. The oxi-
dation criterion included in the LOCA safety analyses is based on such a link. How-
ever, the current 17 % ECR criterion has shown itself to be insufficient in several
cases [7,10,14]. The criterion (connected with the 1204 C limit) is based on the
retention of ductility at the temperature during re-flooding (135 C) in Zry-4 alloy
[31], and for 686-lm-thick cladding wall and double-sided oxidation it can be
expressed as
(7) t ¼ 5:345 104 e20070=T
where:
T ¼ temperature, K, and
t ¼ time, s.
Other oxidation criteria based on the cladding residual ductility have been pro-
posed instead of the 17 % ECR criterion [32–34]. Recently at UJP a new oxidation
criterion Ob, based on the link between the residual ductility and the oxygen con-
centration level in the b-phase, was suggested. An RCT along with fractographic
analysis was employed to get the ductile-brittle threshold. The following paragraphs
summarize the obtained results. The criterion Ob is then calculated by the JKOX
code.
Fractographic Analyses
The fractographic analysis unambiguously confirmed three basic failure mecha-
nisms of specimens during RCTs:
• Failure of the surface layers of the specimens [the oxide layer and the a-Zr(O)
phase]. Failure of these areas is always under way as a result of either cleavage
mechanisms or mechanisms of intergranular decohesion [i.e., processes with-
out recognizable plastic deformation; see Fig. 18(a)].
• Failure of the a-Zr(O) phase grains occurring inside the prior b-phase (pre-
cipitated in the b-phase upon either elevated temperature or cool-down as a
consequence of exceeding the oxygen solubility limit). These grains fail via
mechanisms of intergranular decohesion [Fig. 18(a)].
• Failure of the prior b-phase. This failure occurs by means of transgranular
ductile fracture. The degree of local plastic deformation determining the
micromorphology of the corresponding fracture areas is influenced by both
oxygen and hydrogen [Fig. 18(b)].
The fractographic analysis confirmed that only the prior b-phase could be
treated as load bearing. The resulting failure mode is a combination of all three
aforementioned failure modes. The proportion of individual failure processes is pre-
determined by the microstructure of the fuel cladding. The link between the resid-
ual ductility and the fracture surface of failure samples was already discussed
elsewhere [8]. It could be concluded that the microscopic failure mode corre-
sponded to the macroscopic RCT ductile-to-brittle threshold.
FIG. 18 (a) Fracture micromorphology corresponding to the failure of the prior b-

phase, a-Zr(O) phase, and oxide layer on the surface of a specimen. (b) A detail
of the prior b-phase (the residual ductility Ac ¼ 1.2 %).
Oxygen Concentration Threshold of the b-Phase Embrittlement

Figure 19 presents the residual ductility (determined via RCT) plotted versus the ox-
ygen concentration measured in the prior b-phase employing SIMS and inert gas
fusion. The graph includes all temperatures considered in this study (850 C
to 1300 C) and also all corrosion levels up to a 20-lm-thick oxide layer (the sam-
ples containing 600 wppm of hydrogen or more). Figure 19(a) shows the variation in
the temperature, and Fig. 19(b) shows the variation in the hydrogen level. No corre-
lation with the temperature or with hydrogen was detected. However, the samples
oxidized at higher temperatures and the samples with more hydrogen had higher
oxygen uptakes in the prior b-phase and, consequently, lower residual ductility
than the samples oxidized at lower temperatures or with less hydrogen uptake. The
plot shows a steep decrease in the measured results close to the oxygen concentra-
tion equal to 0.38 wt. %. This value can be considered as the oxygen concentration
threshold of the b-phase embrittlement. After this oxygen content is exceeded, the
tubes may become brittle. The threshold has been compared with the results of
other authors. Brachet et al. [2] determined the oxygen concentration threshold of
the b-phase embrittlement using x-ray microanalysis and fractographic analyses.
They concluded that the prior b-phase had to contain at least 0.3 to 0.4 wt. % oxy-
gen in order for cleavage to occur (for Zry-4 and M5 as well). The two results are in
excellent agreement. Stern et al. [8] estimated (using tensile testing) that the room
temperature macroscopic ductile-to-brittle transient occurs at a critical oxygen con-
centration close to 0.5 wt. % (again for both alloys). However, this still leads to
satisfactory agreement among all results considered. One must also take into
account that the samples in this study were all water quenched directly from the ex-
perimental temperature.
FIG. 19 The results of RCTs depending on the oxygen concentration in the prior b-
phase: (a) for different temperatures; (b) for different hydrogen levels. The solid
line indicates the 0.38 wt. % oxygen concentration threshold.
There are many samples containing more than 0.38 wt. % oxygen in the prior
b-phase that are not brittle (Fig. 19). It does not seem that this is only a consequence
of the experimental scatter coming from the nature of both the experimental mate-
rial and methods employed in this study. This means that the decrease in the resid-
ual ductility was not caused by the oxygen content in the prior b-phase alone.
Generally, the residual ductility of fuel rods after a thermal transient such as an
LOCA depends mainly on two factors: the hydrogen content and the oxygen distri-
bution inside the cladding wall. These are not independent; hydrogen influences the
oxygen distribution—it elevates the oxygen solubility limit in the b-phase [7]. Sec-
ondly, because of the low hydrogen solubility limit in Zr alloys at low temperatures,
especially at room temperature, the hydrides precipitate inside the fuel claddings.
They are far more brittle than the Zr-alloy matrix and cause substantial material
degradation. The effect is known as the intrinsic effect of hydrogen [12,13]. Besides
the oxygen concentration level in the b-phase, the fraction of this phase within the
cladding wall is also an important factor. So, two other major factors, the effect of
hydrides (the intrinsic effect of hydrogen) and the fraction of the b-phase, have to
be considered as well. However, the simplification was made, and those two effects
are not discussed further in this paper.
Ob Criterion
After the oxygen content threshold of the b-phase embrittlement has been acquired,
it can be put into the kinetics of the oxygen pickup in the b-phase. The kinetics was
already determined both experimentally and theoretically (using the JKOX code).
For the temperature range of 950 C to 1200 C, the following expression has been
obtained experimentally:
(8) Cb ¼ 128 e10875=T t 1=3
where:
Cb ¼ oxygen concentration in the b-phase, wt. %,
t ¼ time, s.
Equation 8 was derived using only samples with low hydrogen contents (<100
wppm). In contrast, the determination of the oxygen content threshold of the prior
b-phase embrittlement also involved samples containing more than 100 wppm of
hydrogen. If we substitute an oxygen threshold of 0.38 wt. % into the previous rela-
tion, a new oxidation criterion can be expressed as Eq 9, which is graphically
depicted in Fig. 20.
(9) t ¼ 2:6 108 e32625=T
where:
t ¼ time, s.
FIG. 20 Comparison of the oxidation criteria and the RCT experimental results.
Figure 20 compares the experimentally determined criterion Ob with the Ob cal-

culated by the JKOX code. The plot also contains the current 17 % ECR criterion
and the empirical criterion K (see, e.g., Ref 13) for comparison, as well as the exper-
imental results of the RCT. The criterion K is based on the results of the RCT. The
results (ductile versus brittle samples) were plotted in coordinates log(t) and 1/T.
The boundary line was drawn between ductile and brittle samples, which had resid-
ual ductility of less than 1 %. Based on this equation, the K criterion was proposed.
(10) t ¼ 6:9 108 e28340=T
where:
t ¼ time, s.
The calculated Ob agrees satisfactorily with the experimentally determined Ob
criterion and is valid for the tubes pre-corroded to a 10-lm-thick oxide layer con-
taining up to 150 wppm of hydrogen. However, the Ob criterion based on the nu-
merical calculation is slightly shifted to longer exposure times relative to the
experimental one. That is a consequence of employing the equilibrium oxygen con-
centrations Cb/aþb in the JKOX code, because they are systematically lower than the
experimental values. Nevertheless, the criterion is still valid for all samples consid-
ered in the present study (up to a hydrogen content of 150 wppm), but only for
temperatures equal to or higher than 1100 C (because of the JKOX restriction).
Extrapolation toward the lower temperatures would be possible via an extension of
the slope. It can be concluded that the recently proposed oxidation criterion Ob was
verified using the calculation tool JKOX with a positive result.
Comparing the Ob criterion with other criteria considered, one can say that the
K criterion is more conservative, mainly due to the fact that it involves more alloys
(including older E110 alloy), as opposed to the Ob criterion, which involves only
the modified E110 alloy and, additionally, mainly as-received samples and only
some pre-corroded samples containing greater amounts of hydrogen. That is why
there are still several samples (containing more hydrogen) that are brittle prior to
the Ob threshold. More experimental data are needed (especially on pre-corroded
samples with high hydrogen contents) to modify this criterion for tubes with higher
levels of fuel burnup. The criterion is now valid only for tubes pre-corroded to a
10-lm-thick oxide layer and containing up to 150 wppm of hydrogen. How-
ever, this is assumed to be a sufficient margin, as the maximal hydrogen pickup
from in-service reactor corrosion is considered to be approximately 100 wppm for
that alloy. The 17 % ECR oxidation criterion is not valid for specimens exposed at
higher temperatures with high hydrogen contents. The 17 % ECR was derived using
only as-received Zry-4 fuel claddings. Thus the influence of hydrogen is not
included in the criterion. The Ob criterion fits the experimental results better than
the 17 % ECR criterion. However, one has to realize that the presented experimental
setting does not correspond to a real LOCA (different temperature histories, no
inner pressure applied, no irradiation, etc.).
Conclusions
The presented paper was concerned with the measurement of oxygen concentra-
tions and examination of the microstructure in a Zr1Nb cladding wall after high-
temperature oxidation. The main goal was to check the validity of the recently pro-
posed Zr1Nb–O phase diagram. The following conclusions can be drawn:
• The new results of the oxygen concentration measurements confirmed the cal-
culated phase diagram. The experimentally determined oxygen concentrations
Ca/aþb have been slightly refined using the new experimental results.
• The new diffusion model JKOX has been created using the Zr1Nb–O phase
diagram describing the double-sided oxidation of the Zr1Nb fuel cladding
including the (a þ b)-Zr layer.
• The numerical calculations were compared to the experimental results with satis-
factory agreement. The oxide layer is slightly underpredicted, mainly for higher
exposure times. The oxygen uptake in the b-phase is also underpredicted.
• The proposed experimental procedure provided a good estimation of the
Zr1Nb–O phase diagram. Moreover, it could also be used for other Zr alloys,
and even for alloys containing higher amounts of hydrogen. Consequently, the
Zr1Nb–O phase diagram may be applicable for oxygen diffusion models pre-
dicting the oxidation behavior of Zr1Nb fuel claddings upon high-temperature
oxidation.
• The link between the oxygen distribution and the mechanical response during
RCTs was investigated. The newly proposed oxidation criterion Ob has been
calculated using the JKOX code and compared to the experimental results,
with satisfactory agreement.
Further investigation requires including the influence of hydrogen and pre-
oxidation simulating the oxidation behavior of fuel claddings with high burnup.
The extension of the JKOX model for lower exposure temperatures is also among
the future demands.
ACKNOWLEDGMENTS
The writers thank J. Šustr for metallographic sections and O. Bláhová for nanohard-
ness measurements. Financial support for this research through Grant No. SGS13/
222/OHK4/3T/14, Grant No. TA02011025, the ALFA–TAčR program, research center
grant LC06041, and CEITEC MU CZ.1.05/1.1.00/02.0068 is gratefully acknowledged.
References
[1] Yegorova, L., Lioutov, K., Jouravkova, N., Konobeev, A., Smirnov, V., Chesanov, V., and
Goryachev, A., “Experimental Study of Embrittlement of Zr-1 %Nb VVER Cladding Under
LOCA-Relevant Conditions,” Report No. NUREG/IA-0211, IRSN 2005-194, NSI RRC KI
3188, U. S. NRC, Washington, D.C., 2005.
[2] Brachet, J.-C., Vandenberghe-Maillot, V., Portier, L., Gilbon, D., Lesbros, A., Waeckel, N.,
and Mardon, J.-P., “Hydrogen Content, Pre-oxidation and Cooling Scenario Influences on
Post-quench Mechanical Properties of Zy-4 and M5(TM) Alloys in LOCA Conditions—
Relationship with the Post-quench Microstructure,” J. ASTM Int., Vol. 5, No. 5, 2008, pp.
91–118.
[3] Négyesi, M., Bláhová, O., Adámek, J., Pøibyl, A., and Vrtı́lková, V., “Microstructure Evolu-
tion in Zr1Nb Fuel Cladding During High-temperature Oxidation,” J. Nucl. Mater., Vol.
416, 2011, pp. 298–302.
[4] Sawatzky, A., Ledoux, G. A., and Jones, S., “Oxidation of Zirconium During a High-
temperature Transient,” Zirconium in the Nuclear Industry, G. W. Parry, Ed., ASTM STP
[5] Brachet, J. C., Portier, L., Forgeron, T., Hlvroz, J., Hamon, D., Gullbert, T., Bredel, T., Yvon,
P., Mardon, J.-P., Jacques, P., “Influence of Hydrogen Content on the a-b phase Transfor-
mation Temperatures and on the Thermal-Mechanical Behavior of Zry-4, M4 and M5
Alloys During the First Phase of LOCA Transient,” Zirconium in the Nuclear Industry: Thir-
teenth International Symposium, Annecy, France, June 10–14, ASTM International, West
[6] Négyesi, M., Burda, J., Klouček, V., Lorinčı́k, J., Sopoušek, J., Kabátová, J., Novotný, L.,
Linhart, S., Chmela, T., Siegl, J., and Vrtı́lková, V., “Contribution to the Study of the Pseu-
dobinary Zr1Nb-Oxygen Phase Diagram by Local Oxygen Measurements of Zr1Nb Fuel
Cladding after High Temperature Oxidation,” J. Nucl. Mater., Vol. 420, 2012, pp. 314–319.
[7] Chung, H. M., “Fuel Behavior under Loss-of-coolant Accident Situations,” Nucl. Eng.
Technol., Vol. 37, 2005, pp. 327–362.
[8] Stern, A., Brachet, J.-C., Maillot, V., Hamon, D., Barcelo, F., Poissonnet, S., Pineau, A., Mar-
don, J.-P., and Lesbros, A., “Investigations of the Microstructure and Mechanical Proper-
ties of Prior-beta Structure as Function of the Oxygen Content (0.9 wt %) in Two
Zirconium Alloys,” J. ASTM Int., Vol. 5, No. 4, 2008.
[9] Billone, M., Yan, Y., Burtseva, T., and Daum, R., “Cladding Embrittlement during Postu-
lated Loss-of-Coolant Accidents,” Report No. NUREG/CR-6967, Argonne National Labo-
ratory, Argonne, IL, 2008, p. 336.
[10] Portier, L., Bredel, T., Brachet, J. C., Maillot, V., Mardon, J. P., and Lesbros, A., “Influence
of Long Service Exposures on the Thermal-Mechanical Behavior of Zry-4 and M5TM
Alloys in LOCA Conditions,” J. ASTM Int., Vol. 2, 2005, pp. 103–126.
[11] Nagase, F. and Fuketa, T., “Results From Studies on High Burnup Fuel Behavior Under
LOCA Conditions,” Proceedings of the Nuclear Safety Research Conference, Washington,
D.C., Oct 25–27, 2004, U. S. NRC, Washington, D.C.
[12] Hsu, H. H., “An Evaluation of Hydrided Zircaloy-4 Cladding Fracture Behavior by X-
specimen Test,” J. Alloys Compd., Vol. 426, 2006, pp. 256–262.
[13] Vrtilkova, V., Novotny, L., Kolencik, J., Chmela, T., and Stech, S., “Analysis of Revised 17 %
ECRC-P Criterion Using Oxidation Database UJP,” Proceedings of the 7th Plenary Meet-
ing of SEGFSM and Ad-hoc Meeting on Status of LOCA Tests for Burn-up Dependent
LOCA Criteria, Paris, June 26, OECD NEA, Paris, 92130 Issy-les-Moulineaux, 2006.
[14] Vrtilkova, V., Novotny, L., Doucha, R., and Vesely, J., “An Approach to the Alternative
LOCA Embrittlement Criterion,” Proceedings of the Meeting of OECD Nuclear Energy
Agency Committee on the Safety of Nuclear Installations Special Expert Group on Fuel
Safety Margins (SEGFSM), Argonne, IL, May 25–26, OECD NEA, Paris, 92130 Issy-les-
Moulineaux, 2004.
[15] Hunt, C. E. L. and Niessen, P., “The Continuous Cooling Transformation Behavior of Zirco-
nium-Niobium-Oxygen Alloys,” J. Nucl. Mater., Vol. 38, 1971, pp. 17–25.
[16] Toffolon, C., Brachet, J.-C., Servant, C., Legras, L., Charquet, D., Barberis, P., Mardon, J.-P.,
“Experimental Study and Preliminary Thermodynamic Calculations of the Pseudo-
ternary Zr-Nb-Fe-(O) System,” Zirconium in Nuclear Industry: 13th International Sympo-
sium, ASTM STP 1423, ASTM International, West Conshohocken, PA, 2002, p. 361.
[17] Jerlerud Perez, R. and Massih, A. R., “Thermodynamic Evaluation of the Nb-O-Zr Sys-
tem,” J. Nucl. Mater., Vol. 360, 2007, pp. 242–254.
[18] Hózer, Z., Györi, C., Matus, L., and Horváth, M., “Ductile-to-Brittle Transition of Oxidised
Zircaloy-4 and E110 Claddings,” J. Nucl. Mater., Vol. 373, 2008, pp. 415–423.
[19] Lorincik, J., Kloucek, V., Négyesi, M., Kabátová, J., Novotný, L., and Vrtilkova, V., “SIMS
and Thermal Evolution Analysis of Oxygen in Zr-1 %Nb Alloy After High-Temperature
Transitions,” Surf. Interface Anal., Vol. 43, 2011, pp. 618–620.
[20] Abriata, J. P., Garcés, J., and Versaci, R., “The O-Zr (Oxygen-Zirconium) System,” Bull.
Alloy Phase Diagrams, Vol. 7, 1986, pp. 116–124.
[21] Pawel, R. E., “Oxygen Diffusion in the Oxide and Alpha Phases During Reaction of
Zircaloy-4 With Steam from 1000 C to 1500 C,” J. Electrochem. Soc., Vol. 126, 1979, pp.
1111–1118.
[22] Négyesi, M., Burda, J., Bláhová, O., Linhart, S., and Vrtı́lková, V., “The Influence of Hydro-
gen on Oxygen Distribution Inside Zry-4 Fuel Cladding,” J. Nucl. Mater., Vol. 416, 2011,
pp. 288–292.
[23] Chung, H. M. and Kassner, T. F., “Pseudobinary Zircaloy-Oxygen Phase Diagram,” J. Nucl.
Mater., Vol. 84, 1979, pp. 327–339.
[24] Brachet, J. C., Hamon, D., Béchade, J. L., Forget, P., Toffolon-Masclet, C., Raepsaet, C.,
Mardon, J. P., Sebbari, B., “Quantification of the Chemical Elements Partitioning Within
Pre-hydrided Zircaloy-4 after High Temperature Steam Oxidation as a Function of the
Final Cooling Scenario (LOCA Conditions) and Consequences on the (Local) Materials
Hardening,” Proceedings of the IAEA Technical Meeting on Fuel Behavior and Modeling
under Severe Transient and LOCA Conditions, Japan, Oct 18–21, 2011.
[25] Okamoto, H., “Nb-Zr (Niobium-Zirconium),” J. Phase Equilib., Vol. 13, 1992, p. 577.
[26] Corvalán-Moya, C., Desgranges, C., Toffolon-Masclet, C., Servant, C., and Brachet, J. C.,
“Numerical Modeling of Oxygen Diffusion in the Wall Thickness of Low-tin Zircaloy-4
Fuel Cladding Tube During High Temperature (1100-1250 C) Steam Oxidation,” J. Nucl.
Mater., Vol. 400, 2010, pp. 196–204.
[27] Négyesi, M., Bláhová, O., Burda, J., and Vrtı́lková, V., “Experimental Background for Diffu-
sion Models of Zr1Nb-O System,” Chem. Listy, Vol. 104, 2010, pp. 356–359.
[28] Sopoušek, J. and Svobodová, M., “Thermodynamic Prediction of Zr-Nb-O-H Phase Dia-
gram Sections,” Proceedings of Solid-Solid Phase Transformations in Inorganic Materials
(PTM 2010), Avignon, France, June 6–11, FFC, 28, rue Saint-Dominique, Paris, 2010,
p. 652.
[29] Chung, H. M. and Kassner, T. F., “Embrittlement Criteria for Zircaloy Fuel Cladding Appli-
cable to Accident Situations in Light-water Reactors,” Report No. NUREG/CR-1344,
ANL-79-48, Argonne National Laboratory, Argonne, IL, 1980.
[30] Brachet, J. C., Portier, L., Hamon, D., Trouslard, Ph., Urvoy, S., Rabeau, V., Lesbros, A.,
Mardon, J. P., “Quantification of the a(O) and Prior-b Phase Fractions and Their Oxygen
Contents in High Temperature (HT) Oxidised Zr Alloys (Zy-4, M5),” Proceedings of
SEGFSM Topical Meeting on LOCA, Chicago, May 25–27, 2004, Argonne National Labo-
ratory, Argonne, IL.
[31] Hagrman, D. L., Reymann, G. A., Mason, R. E., MATPRO, Version 11: A Handbook of Materi-
als Properties for Use in the Analysis of Light Water Reactor Fuel Rod Behavior, 1979, pp.
450–457.
[32] Hobson, D. O., “Ductile-Brittle Behavior of Zircaloy Fuel Cladding,” Proceedings of the
ANS Topical Meeting on Water Reactor Safety, Salt Lake City, UT, March 26, 1973, pp.
274–288.
[33] Pawel, R. E., “Oxygen Diffusion in Beta Zircaloy During Steam Oxidation,” J. Nucl. Mater.,
Vol. 50, 1974, pp. 247–258.
[34] Sawatzky, A., “Proposed Criterion for the Oxygen Embrittlement of Zircaloy-4 Fuel
Cladding,” Proceedings of the 4th Symposium on Zirconium in the Nuclear Industry,
Stratford-on-Avon, UK, June 27–29, 1978.
DISCUSSION
Question from Marc Tupin, CEA - Saclay:—Did you test the remaining ductility
of your sample after oxygen saturation in the ex-b matrix?
Authors’ Response:—Yes. We used RCT (ring compression test). While not included
in the presentation, the threshold of the prior b embrittlement is discussed in the paper.
Question from Suresh Yagnik, EPRI:—Could you comment on how your data
will be affected if the corrosion was performed in H2O/H2 mixture rather than pure
steam?
Authors’ Response:—It can be assumed that the oxidation kinetics and the
hydrogen uptake will change in H2O/H2 mixture. The oxidation tests in H2O/H2
mixture are needed to be performed to discuss the effect thoroughly.
Question from Duriez Christian, IRSN, France:—How do you take into account
the surface oxidation, after sample preparation, for the EPMA and SIMS oxygen
concentration measurements?
Authors’ Response:—No special technique was used to take into account the
surface contamination after the sample preparation.
Question from Javier Romero, Westinghouse Electric Co.:—How much does the
assumption that oxygen diffusivity is the same between Zircaloy-4 and Zr-1Nb
affect your results?
Authors’ Response:—Sensitivity analyses were performed regarding the oxygen

diffusivities. However, employing the calculated phase diagram Zr1Nb-O and the
Zircaloy-4 oxygen diffusivities, good agreement was achieved between the calcula-
tions and the experimental data. Hence it could be concluded that there was no
need to change the oxygen diffusivities.
Questions from P. Barberis, CEZUS Research Center:—The so-called a þ b diffu-

sion coefficient is likely a weighted mean of the coefficients in a and b phase, taking
into account the a needle fraction in the a þ b layer. How does it evolve with the oxi-
dation temperature? Did you try to deduce from this coefficient the a needle fraction?
Authors’ Response:—The same weighted mean for all temperatures considered was
employed, so that the dependence of lnDaþb vs. 1/T is linear. First, the fraction a/b was
measured. Afterwards, the diffusion coefficient was estimated based on the fraction a/b.
Zoltán Hózer,1 Erzsébet Perez-Feró,1 Tamás Novotny,1

Imre Nagy,1 Márta Horváth,1 Anna Pintér-Csordás,1
András Vimi,1 Mihály Kunstár,1 and Tamás Kemény2
Experimental Comparison of the

Behavior of E110 and E110G
Claddings at High Temperature
Reference
Hózer, Zoltán, Perez-Feró, Erzsébet, Novotny, Tamás, Nagy, Imre, Horváth, Márta, Pintér-
Csordás, Anna, Vimi, András, Kunstár, Mihály, and Kemény, Tamás, “Experimental Comparison of
the Behavior of E110 and E110G Claddings at High Temperature,” Zirconium in the Nuclear
Industry: 17th International Symposium, STP 1543, Robert Comstock and Pierre Barberis, Eds.,
pp. 932–951, doi:10.1520/STP154320120165, ASTM International, West Conshohocken, PA 2015.3
ABSTRACT
A new experimental program was carried out in order to compare the behavior of
E110 and E110G type alloys at high temperature. The program included differential
scanning calorimetry studies on the phase transition process, ballooning and burst
tests, oxidation in steam and hydrogen rich steam atmospheres, ring compression
tests of oxidized samples and posttest examination with optical and scanning
electron microscopes.The two alloys showed very similar behavior in the non-
oxidized state. The phase transitions took place practically in the same range of
temperature and the cladding burst due to ballooning also happened under similar
conditions. The oxidation caused significant differences. The breakaway oxidation
typical for E110 could not be observed with E110G samples. The E110G had much
better load bearing capabilities in oxidized state and did not pick up as much
hydrogen as the other alloy did.
Keywords
Zirconium oxidation, cladding embrittlement, ballooning and burst
Manuscript received November 19, 2012; accepted for publication November 3, 2013; published online June
17, 2014.
1
Centre for Energy Research, Hungarian Academy of Sciences, Budapest, Hungary H-1525.
2
Wigner Research Centre for Physics, Hungarian Academy of Sciences, Budapest, Hungary H-1525.
3
Hyderabad, India.
Copyright V
HÓZER ET AL., DOI 10.1520/STP154320120165 933
Introduction
The E110 type cladding has been used in VVER reactors for many decades
in different countries [1–3]. The E110 cladding has outstanding corrosion
resistance during normal operation in the reactors. After four cycles and up to 65
MWd/kgU burnup, the thickness of an oxide scale on the external surface of the
cladding does not exceed 10 lm and the hydrogen content remains below 120 ppm
[4–6]. The behavior of E110 at high temperature (especially in the range of
900 C–1000 C) is characterized by breakaway oxidation, high hydrogen uptake,
and embrittlement at low degree of oxidation compared to other zirconium alloys,
as, e.g., Zircaloy-4 [7–11].
The E110 cladding is traditionally produced on electrolytic basis. The Russian
fuel supplier intends to introduce the production of cladding tubes from sponge
material [12]. The new alloy named E110G is characterized by the same composi-
tion as E110, but some preliminary tests indicated significant differences in the
behavior of the two alloys at high temperatures [7].
In order to cover a wide range of parameters for loss-of-coolant-accident
(LOCA) conditions, a systematic experimental program has been carried out in
Hungary. The main objectives of the tests were to compare the behavior of the two
alloys at high temperatures and to produce E110G specific data for the
development of numerical models that can be built into transient fuel behavior
codes and used in safety analyses. The program included the following elements:
• determination of the composition of the alloys,
• phase transition studies,
• ballooning and burst of tube samples with inner pressurization,
• oxidation in steam atmosphere,
• oxidation in hydrogen rich steam atmosphere,
• investigation of breakaway effect,
• mechanical testing of oxidized cladding samples,
• determination of the hydrogen content of oxidized samples,
• post-test examination of samples by optical and scanning electron
microscopy.
COMPOSITION
The composition of the investigated E110 and E110G cladding tubes was deter-
mined with an AEI MS702/R type spark source mass spectrometer (SSMS) with
Mattauch–Herzog geometry. 10 mm long and 1 mm wide electrodes were produced
from the cladding tubes. The samples were degreased with acetone and alcohol
before testing. High voltage (20 kV) power with 100 pulse/s frequency and 100 ls
length was applied to ionize the samples. The spectra were recorded on traditional
photo plates (Ilford Q2) and evaluated using a Zeiss MD-100 microdensitometer.
There were no available zirconium standards to determine the correct Relative Sen-
sitivity Coefficient (RSC) values for the analyzed elements. RSC values were taken
from earlier studies and 93Nb (1 % content) and 96

Zr (2.8 % abundance) isotopes
were used as internal standards.
PHASE TRANSITION
The phase transition process was investigated by Differential Scanning Calorimetry
(DSC) method. Small pieces of E110G and E110 tubes were placed in a sample
holder with Al2O3 powder. For calibration purposes, a few mg of Ag was added to
the reference sample holder. The measurements were carried out at four different
heating rates: 2.5, 5, 10, and 20 C/min. Control measurement was performed with
pure Zr. The DSC measurements were carried out at the Material Physics Dept.,
Eötvös University, Budapest.
OXIDATION
The oxidation of E110G cladding was investigated in a high temperature furnace. A
steam generator with controllable power supply provided steam for the experi-
ments. The tube type furnace was placed in horizontal position and it had three
heating sections with microprocessor control. The inlet junction of the furnace was
connected to the steam generator. The remaining steam fraction of the outlet gas
was condensed in the condenser unit. A quartz tube with 19 mm diameter was used
inside of the furnace. Additional electrical heaters were installed in the inlet and
outlet connecting tubes to prevent condensation in the facility (Fig. 1).
In order to investigate breakaway phenomena, the experimental facility
was connected to a thermoconductometric detector (TCD) of a gas-
chromatograph. The signal of the TCD was proportional to the concentration
of the hydrogen in the outlet gas and thus it indicated the intensity of oxida-
tion process.
FIG. 1 Scheme of the high temperature furnace used for cladding oxidation.
The tests were carried out with 8 mm long fuel cladding samples cut from the
original cladding tubes. The external diameters of the E110 and E110G tubes were
9.14 and 9.10 mm, respectively. The cladding thickness was 0.65 mm for both mate-
rials. In most of the experiments, two samples were placed in the furnace: one E11G
and one E110 sample, in order to have the same oxidizing conditions and to allow
the direct comparison of the two alloys. The vertically positioned samples were
standing one after another in the horizontal sample holder, which had contact with
the bottom rim of the samples. The samples did not touch each other and both
samples were in the uniform temperature section of the furnace. Acetone was
applied to degrease the samples before testing. The masses of samples before and af-
ter oxidation were carefully measured. The extent of cladding oxidation was
expressed as measured Equivalent Cladding Reacted (ECR) and it was calculated on
AZr Dm
the basis of mass gain: ECR ¼ MO2 mi 100, where AZr ¼ molar mass of zirconium,
MO2 ¼ molar mass of oxygen, mi ¼ mass of sample before oxidation, and
Dm ¼ mass gain during oxidation.
At the beginning of the tests, the furnace was heated up to the target
temperature without the samples. After stabilization of thermal conditions, the
samples were moved from the unheated zone of the quartz tube into the cen-
ter of furnace.
Isothermal tests in steam atmosphere were performed between 600 C–1200 C.
The steam was mixed with 12 v/v% high purity argon carrier gas. This argon con-
tent does not influence the oxidation process, but it is needed to stabilize gas flow
in the outlet section of the furnace. The steam supply was constant during the tests
and the flow rate was high enough to prevent steam starvation inside of the furnace.
The typical steam flow was 4–4.5 mg/cm2s, while the starvation threshold even at
1200 C remained below 3.5 mg/cm2s for the given sample geometry.
The breakaway oxidation test series covered the temperature range
of 800 C–1200 C and only single samples were used in order to avoid any potential
interference between the signals originated from the two different alloys and to
characterize the E110G and E110 samples separately. The typical oxidation time
was 2700 s.
Additional tests series were conducted to investigate the effect of high hydrogen
content in the steam atmosphere. These tests were carried out in the temperature
range of 900 C–1100 C with 65 v/v% hydrogen content in the steam. This hydro-
gen content was selected in order to have comparable conditions to the earlier
experiment with E110 cladding [11].
BALLOONING AND BURST

The plastic deformation of cladding tubes at high temperature was investigated in
ballooning tests with inner pressurization of cladding tubes in a vertical tube fur-
nace. The experiments were carried out under isothermal conditions. The samples
were placed in the center of the furnace where the axial temperature profile was
uniform.
FIG. 2 Drawing of the sample used in ballooning tests.
The samples were produced from original cladding tubes. Both ends of the
50 mm long tubes were closed by welded Zr plugs, but one of them had an internal
hole for the connecting tube (Fig. 2). The pressurization of the samples was per-
formed through this small diameter Zr connecting tube. The pressurization of the
samples was driven by a control unit using capillary tubes, needle valves, and a step-
per motor. The linear pressurization was continued until the burst of the sample
(Fig. 3). The temperature and pressure histories were recorded during the tests.
The basic series of ballooning tests was performed with non-oxidized samples
in argon atmosphere. In order to investigate the effect of oxidation, some samples
were pre-oxidized before the ballooning test in the oxidation facility (Fig. 1) in steam
atmosphere at 900 C. The isothermal tests were conducted between 700 C–1200 C.
The pressurization rates covered three orders of magnitude from 0.007 bars/s to 7
bars/s.
FIG. 3 Pressure and temperature histories in ballooning test at 800 C with

pressurization rate 0.007 bars/s.
After the tests, post-test examination of the samples was carried out and the av-
erage and maximum values of deformation were determined.
MECHANICAL TESTING
Mechanical testing of oxidized E110G and E110 ring samples was carried out with
an INSTRON 1195 type tensile testing machine. A special heater module can be
mounted to the machine for testing at high temperature.
Ring compression tests were performed at room temperature and at 135 C. In
case of high temperature testing, the temperature in the machine was stabilized
before starting the mechanical loading. The speed of the cross head was 0.5 mm/
min. The load-displacement curves were recorded during the ring compression
tests.
POST-TEST EXAMINATION
Metallography examination of oxidized samples was carried out with a Reichert
type optical microscope. A Philips SEM 505 type scanning electron microscope and
a JEOL Superprobe 733 type microprobe were also used to investigate the cladding
microstructure.
The hydrogen content of the oxidized Zr samples was determined by hot
extraction method at 1150 C using a CHROMPACK 438 A type gas cromatograph.

COMPOSITION
The main alloying element in both E110 and E110G is Nb with 1 %. The SSMS
measurements indicated that impurities are higher in E110G than in E110, except
Ni and Hf (Table 1). Significant differences were found for hafnium and iron. The
hafnium content was intentionally decreased in the new alloy from 100 ppm to
10 ppm in order to reduce neutron absorption by the cladding. The iron content
was increased in the new cladding from 50 ppm to 500 ppm value. According to
TABLE 1 Impurities present in E110 and E110G alloys.
Element E110 E110G
Mg 0.5 ppm 1.5 ppm

Al 0.5 ppm 10 ppm
Si 1 ppm 35 ppm
Cr 10 ppm 30 ppm
Mn 0.1 ppm 5 ppm
Fe 45 ppm 500 ppm
Ni 15 ppm 15 ppm
Cu 0.5 ppm 5 ppm
Hf 100 ppm 10 ppm
Nikulin et al. [12], the sum of the following impurities (C, Si, Ni, P, Cl, N, F, and
Al) was about 140 ppm in E110 and about 70 ppm for E110G. Further analysis of
impurities present in E110 and E110G is required to more clearly establish the dif-
ference between the materials.
PHASE TRANSITION
The DSC measurements indicated that phase transition took place for E110G
cladding between 800 and 930 C. The higher heating rate resulted in higher tem-
perature for both onset and end of phase transition (Fig. 4). The measured data
showed 100 C difference between the onset and end of phase transition, which
is in good agreement with the results of previous studies [13].
OXIDATION
The comparison of mass gains of E110G and E110 during oxidation in steam indi-
cated two types of oxidation kinetics (Fig. 5).
• At 600, 700, and 800 C the E110G reached a higher degree of oxidation in all
measured points. However, this difference is rather small, e.g., the oxidation at
800 C for 2 h resulted in 3.7 % ECR for E110G and 2.9 % ECR for E110. So
the relative difference for this point is only 27 %. At 1100 and 1200 C, the
E110G oxidized faster, too. Oxidation at 1100 C for 1700 s created 20.7 %
ECR for E110G and 16.6 % for E110. The relative increase in this case was
25 %. The comparison with weight gain calculated using the Cathcart–Pawel
correlation indicates that the oxidation kinetics was slower at 600, 700, and
800 C for both E110 and E110G alloys. It is not surprising, since the correla-
tion is valid above 1000 C. At 1100 and 1200 C, the correlation is in very
FIG. 4 Onset and end temperatures of E110 and E110G phase transition (DSC data).
FIG. 5 Results of oxidation of E110G and E110 samples in steam at 800 and 1000 C.
good agreement with the E110G oxidation data and slightly overestimates the
E110 oxidation kinetics.
• At 900 and 1000 C, the oxidation of E110 was much more intense than that of
E110G. At 900 C, the oxidation of E110G is much slower than that of E110
and also much less than predicted by the Cathcart–Pawel correlation (Fig. 5).
• The oxidation at 1000 C for 1 h lead to 9.5 % ECR for E110G, while it pro-
duced 26.9 % ECR for E110. The relative difference in this case was 283 %.
The Cathcart–Pawel correlation gives 18.1 % ECR for this oxidation time at
1000 C. The start of breakaway oxidation can be observed on the E110 curve
after 800 s, where the oxidation kinetics significantly accelerates compared
to E110G data and to the Cathcart–Pawel correlation.
The most significant difference between the oxidation behavior of E110 and
E110G alloys was found at 900 and 1000 C, where the E110 cladding is character-
ized by intense breakaway oxidation. The oxidation process could be described by
parabolic functions of time with the exception of E110 oxidation at 900 and
1000 C.
The oxidation in hydrogen rich steam showed that the hydrogen content in the
atmosphere does not change the oxidation kinetics. The mass gains of samples oxi-
dized in steam with and without hydrogen were very close to each other for both
E110 and E110G. The difference between the ECRs of two alloys was practically the
same for the cases with and without hydrogen in the atmosphere (Fig. 6).
BREAKAWAY EFFECT
The visual observation of ring samples after oxidation in pure steam and in hydro-
gen rich steam indicated significant differences. The oxide scale on the E110G sam-
ples was always compact, while on the surface of the E110 samples in many cases—
especially after long oxidation at 900 and 1000 C—spalling oxide scales could be
observed (Fig. 7).
The breakaway effect was detected in the tests with on-line H release measure-
ments, too. The first peak of the TCD signal (Fig. 8) indicated the intense start of
oxidation process. After the peak, the signal for E110G started to monotonously
decrease since the developed compact oxide layer played a protective role and
slowed down the oxidation process. In case of E110 at 1000 C the TCD signal did
FIG. 6 Results of oxidation of E110G and E110 samples in hydrogen rich steam at 900 C.
not decrease monotonously, but it was characterized by several local peaks and local
minima. The acceleration of hydrogen production after each local minimum
showed that the oxide layer spalled and the direct contact between the metal surface
and the atmosphere resulted in intense oxidation. A development of the compact
oxide layer continued until the next local minimum. For E110G, breakaway oxida-
tion could not be observed in the investigated range of temperatures and oxidation
times, even in case of 7200 s oxidation at 1000 C.
The reason for the different breakaway behavior of Zr claddings was analyzed
by several authors in the past. Chung [14] pointed out that some impurities (e.g.,
fluorine) are deleterious, and can lead to increased susceptibility to nodular oxida-
tion. He emphasized that Zr metal produced by electrolysis contains F, while Kroll
FIG. 7 View of E110G (left) and E110 (right) cladding samples after 5000 s oxidation at
900 C in steam.
FIG. 8 Comparison of hydrogen release in case of E110 and E110G cladding at

temperature 1000 C.
fabrication process prevents F pickup. Yan et al. [9] carried out experiments with
polished E110 samples. Smooth, black oxide was formed on the polished surface
during steam oxidation, while cracked, delaminated oxide layer could be observed
on the as-received E110 surface. Yegorova et al. [15] examined several advanced
types of El10 claddings manufactured on the basis of sponge Zr to evaluate the sen-
sitivity of the oxidation phenomena to the bulk chemistry of the cladding material.
They concluded that the best way for the improvement of the corrosion resistance
of E110 cladding could be the fabrication of cladding material from sponge Zr and
the application of polishing to the internal and external surfaces of the tubes. The
E110 and E110G cladding tubes used in the present study were produced by the
same technology in the fuel factory. The only difference was that sponge ingot was
used for E110G instead of traditional electrolytic ingot. It means that the surface of
the two tested materials were the same (they were produced in the same factory by
the same production line), while the content of some impurities were different, and
it can lead to the differences observed at high temperature oxidation.
BALLOONING AND BURST

The ballooning tests showed that cladding burst pressure decreased with the increase
of temperature. The 800 C sample pressurized with 2.6 bars/s rate burst at 87.8 bars,
while the sample at 1200 C with the same pressurization rate failed at 15.4 bars.
The pressurization rate had a very important effect on burst pressure.
For example, the cladding tube tested at 800 C with 0.007 bars/s load lost its integ-
rity at 23.7 bars, while the pressurization rate of 7.1 bars/s resulted in burst at 115
bars. Similar tendency could be observed for other temperatures, too (Fig. 9).
FIG. 9 Burst pressure of E110G tubes as function of pressurization rate.
The deformation of the tube due to ballooning reached about 100 % at 700 and
800 C. At higher temperatures the deformations were smaller, between 1000 and
1200 C the typical values were 10 %–40 % (Fig. 10). Furthermore, the faster pressur-
ization lead to larger deformations.
The comparison of new E110G data with previously measured E110 data [13]
indicated that failure of E110G cladding took place at slightly higher pressure than
E110. Figure 11 shows some measurement points with 0.08 bars/s pressurization rate.
The burst pressure for the E110G is higher for all samples. The figure shows con-
sistent small differences between the two materials. Similar behavior was observed
at other pressurization rates for most of the measured points.
Earlier tests indicated that the initial oxide layer can increase the mechanical
strength of the E110 cladding [11,13] up to some degree of oxidation. After reach-
ing a maximum value, the further oxidation leads to strong embrittlement and fail-
ure at lower loads. This effect was observed in the current test series with E110G
cladding, too. The maximum values for both E110 and E110G corresponded to
FIG. 10 Ballooned E110G samples (left: 700 C and 0007 bars/s, center: 900 C and 0,6
bars/s, right: 1200 C and 6 bars/s).
FIG. 11 Comparison of the burst pressures of E110 and E110G claddings at 0.08 bars/s
pressurization rate.
20 lm oxide scale (Fig. 12). It must be mentioned that the pre-oxidation was per-
formed at 900 C, where the E110 alloy shows a short time to breakaway. The maxi-
mum burst pressure in Fig. 12 with 20 lm oxide scale corresponds to 400 s
FIG. 12 Burst pressure of oxidized E110G and E110 sample tested at 800 C with 0.025
bars/s pressurization rate.
oxidation time for E110 cladding and 5150 s for E110G. As such, in accident condi-
tions, significant differences exists in the burst behavior of the oxidized claddings.
MECHANICAL TESTING OF OXIDIZED SAMPLES

The ring compression tests produced similar load-displacement curves for E110
and E110G samples with low degree of oxidation. The increase of ECR, however,
proved that the new E110G cladding had much better load bearing capability com-
pared to E110. The most significant differences were found for the samples oxidized
at 900 and 1000 C. This is the same interval, where the oxidation of E110 was
much more intense than that of E110G. Figure 13 shows the load-displacement
curves of two samples that were oxidized parallel at 1000 C for 1 h. The E110G
could be loaded in the tensile test machine up to 550 N and a curve showed ductile
character. The E110 sample failed at 120 N and the curve showed the typical form
of brittle material. Similar effect was seen for many other pairs of samples were oxi-
dized not only in steam but in hydrogen rich steam atmosphere as well.
On the basis of ring compression tests, the ductile-to-brittle transition of oxi-
dized samples was determined. The analyses of the load-displacement curves
allowed us to distinguish between ductile and brittle samples. In case of ductile
samples the elastic deformation is followed by a plastic plateau, while in case of brit-
tle material, the breakdown of the curve indicates the brittle failure of the ring after
the elastic deformation. In most cases, the peak of the curves of brittle materials
does not reach as high values as the ductile ones do. Furthermore, the integral of
FIG. 13 Load-displacement curves of E110G and E110 ring samples oxidized at 1000 C
for 3600 s.
the curves of up to the first breakdown (indicating the accumulated energy to fail-
ure) has much lower values compared to ductile samples with similar geometry.
In Fig. 14, each oxidized sample is marked as ductile or brittle. The transition
line, representing the ductile-to-brittle transition limit, was drawn to separate the
points in a conservative way: all the brittle points and a few ductile points are above
the line, and the remaining ductile points are below the line. This approach guaran-
tees that the cladding has some ductility under the oxidation conditions below the
line. These lines in Fig. 14 can be described by an Arrhenius type equation: the time
of oxidation to reach brittle state is proportional to the exponent of the reciprocal
of absolute temperature. The application of such functions facilitates the integration
of the degree of embrittlement during transient conditions and can be used in safety
analyses [8]. The difference in time between the two materials is only apparent at
900 and 1000 C, where E110 shows high oxidation accumulation (breakaway
effect).
Comparing the two transition lines it can be seen (Fig. 14) that the E110G tran-
sition is shifted to higher temperatures and longer oxidation times. At a given oxi-
dation time this shift means 130 C, or at a fixed temperature four times longer
oxidation. Testing of samples oxidized at 135 C proved that transition to brittle
failure of oxidized Zr rings took place at higher degree of oxidation than at room
temperature.
POST-TEST EXAMINATIONS
The post-test examinations pointed out some microstructural differences between
the two alloys in oxidized state. Figure 15 shows that the oxide scale on E110 alloy
had a layered structure that is typical for breakaway oxidation. The micrograph
shown is typical only when there is breakaway oxidation for E110. On the surface
of the E110G sample oxidized together with the E110 ring, a compact oxide layer
was formed. The average thickness of the E110G sample was 24 lm, while the E110
cladding had several 10–20 lm layers of oxide with the total thickness of 89 lm.
FIG. 14 Ductile-to-brittle transition of oxidized E110 and E110G samples based on ring
compression tests at room temperature.
FIG. 15 Metallography pictures of E110 (left) and E110G (right) after oxidation at 1000 C
for 3600 s.
The secondary electron images (SEI) taken during SEM analyses indicated that
the E110 sample oxidized at 1000 C for 3600 s failed due to brittle fracture in the
ring compression test, for the microstructure was characterized by large plates (Fig.
16). The fracture surface of the counterpart E110G sample showed small scale hon-
eycomb structure that is typical for ductile deformation.
The oxide scale thicknesses were determined during metallography examina-
tion. The two alloys had similar thicknesses at the same ECRs (Fig. 17), but it must
be emphasized that the same ECRs were reached in different oxidation times. Figure
17 includes data from different temperatures between 900 and 1200 C.
The hydrogen content in selected samples showed also significant differences
between the two alloys. The H uptake of the examined E110G samples was below
the detection limit (100 ppm), while some oxidized E110 samples contained several
thousand ppm hydrogen. For example, in case the samples oxidized at 1000 C for
3600 s (Figs. 13, 15, and 16) the hydrogen content in E110G was less than 100 ppm,
FIG. 16 Secondary electron images of the fracture surfaces of E110 (left) E110G (right)
samples after oxidation at 1000 C for 3600 s and ring compression testing.
FIG. 17 Oxide layer thickness as function of ECR for E110 and E110G alloys oxidized in
steam.
but the E110 ring had 3990 ppm hydrogen. The data points in Fig. 17 are character-
ized with hydrogen content below 100 ppm for all the points, except E110 samples
above 15 % ECR, which contained 3530–5720 ppm hydrogen.
Conclusions
The high temperature behavior of E110 and E110G type alloys was compared
through a series of experiments. The experimental program covered most of the
cladding related phenomena that can take place during a LOCA event in a nuclear
reactor.
The experimental program included differential scanning calorimetry and ther-
momechanical studies on the phase transition process. The plastic deformation and
the burst of cladding was investigated in isothermal ballooning tests in the tempera-
ture range of 600 C–1200 C with linear pressurization rates of 0.007–7 bars/s. Oxi-
dation of samples was carried out in steam atmosphere between 600 C–1200 C and
in hydrogen rich steam atmosphere between 900 C–1100 C. Special studies were
devoted to the investigation of breakaway oxidation in steam atmosphere with on-
line hydrogen detection. The mechanical load bearing capability of the oxidized
samples was investigated through ring compression tests at room temperature and
at 135 C. Optical and scanning electron microscopy examinations were performed
and the hydrogen content was determined for selected oxidized samples.
The results of phase transition studies and ballooning tests showed that, with-
out oxidation, the behavior of the two alloys was well comparable in the
investigated range of parameters. The oxidation, however, resulted in significant

differences, especially at 900 and 1000 C. The E110 samples had typically layered
oxide scale due to breakaway phenomena. In case of E110G, the breakaway process
did not take place and compact oxide layers could be observed on all samples.
According to the mechanical tests, the ductile-to-brittle transition for the oxidized
E110G alloy takes place at much favorable conditions (i.e., at higher temperature
and at longer oxidation times) than in case of E110, and the E110G cladding picks
up much less hydrogen during oxidation in steam than the E110 alloy.
ACKNOWLEDGMENTS
The performed test series was supported by the Paks Nuclear Power Plant, Hungary.
The writers are grateful to Alajos Ö. Kovács (Material Physics Department, University,
Budapest) for the TMA and DSC measurements.
References
[1] Gerasimov, V. V., and Monakhov, A. S., 1982, Materials for Nuclear Technology, Energoa-
tomizdat, Moscow (in Russian).
[2] IAEA, “Design and Performance of WWER Fuel,” Technical Report Series No. 379, Inter-
national Atomic Energy Agency, Vienna, 1996.
[3] Shebaldov, P. V., Peregud, M. M., Nikulina, A. V., Bibilashvili, Y. K., Lositski A. F., Kuz’-
menko, N. V., Belov, V. I., and Novoselov, A. Y., “E110 Alloy Cladding Tube Properties and
Their Interrelation With Alloy Structure-Phase Condition and Impurity Content,” Pro-
ceedings of the 12th International Symposium on Zirconium in the Nuclear Industry, To-
ronto, ON, ASTM STP 1354, ASTM International, West Conshohocken, PA, 2000, pp.
545–558.
[4] Smirnov, A. V., Markov, D. V., Smirnov, V. P., Polenok, V. S., Ivashchenko, A. A., and Stroz-
huk, A. V., “Results of Post-Irradiation Examination of VVER Fuel Assembly Structural
Components Made of E110 and E635 Alloys,” Proceedings of the 6th International Con-
ference on WWER Fuel Performance, Modelling and Experimental Support, Albena, Bul-
garia, Sept 19–23, 2005, pp. 231–243.
[5] Novikov, V. V., Markelov, V. A., Shishov, V. N., Tselishchev, A. V., and Balashov, A. A.,
“Results of Post-Irradiation Examinations (PIE) of E110 Claddings and Alloy Upgrading
for VVER,” Transactions of TOPFUEL 2006, Poster Session II, http://www.euronuclear.
org/events/topfuel/transactions/Topfuel-Poster%20Session%20II.pdf
[6] Smirnov, V. P., Markov, D. V., Smirnov, A. V., Polenok, V. S., Perepelkin, S. O., and Ivash-
chenko, A. A., “VVER Fuel: Results of Post Irradiation Examination,” Proceedings of the
2005 Water Reactor Fuel Performance Conference, Kyoto, Japan, Oct 2–6, 2005, pp.
217–226.
[7] Yegorova, L., Lioutov, K., Jouravkova, N., Konobeev, A., Smirnov, V., Chesanov, V., and Gor-
yachev, A., “Experimental Study of Embrittlement of Zr-1 %Nb VVER Cladding Under
LOCA-Relevant Conditions,” NUREG/IA-0211, 2005, http://pbadupws.nrc.gov/
docs/ML0511/ML051100343.pdf
[8] Hózer, Z., Gyo†ri, C., Matus, L., and Horváth, M., “Ductile-to-Brittle Transition of Oxidised
Zircaloy-4 and E110 Claddings,” J. Nucl. Mater., Vol. 373, 2008, pp. 415–423.
[9] Yan, Y., Burtseva, T. A., and Billone, M. C., “High-Temperature Steam-Oxidation Behavior
of Zr-1Nb Cladding Alloy E110,” J. Nucl. Mater., Vol. 393, 2009, pp. 433–448.
[10] Steinbrück, M., Birchley, J., Boldyrev, A. V., Goryachev, A. V., Grosse, M., Haste, T. J.,
Hózer, Z., Kisselev, A. E., Nalivaev, V. I., Semishkin, V. P., Sepold, L., Stuckert, J., Vér, N.,
and Veshchunov, M. S., “High-Temperature Oxidation and Quench Behaviour of Zircaloy-
4 and E110 Cladding Alloys,” Prog. Nucl. Energy, Vol. 52, 2010, pp. 19–36.
[11] Perez-Feró, E., Gyo†ri, C., Matus, L., Vasáros, L., Hózer, Z., Windberg, P., Maróti, L., Hor-
váth, M., Nagy, I., Pintér-Csordás, A., and Novotny, T., “Experimental Database of E110
Claddings Exposed to Accident Conditions,” J. Nucl. Mater., Vol. 397, 2010, pp. 48–54.
[12] Nikulin, S. A., Rozhnov, A. B., Belov, V. A., Li, E. V., and Glazkina, V. S., “Influence of Chem-
ical Composition of Zirconium Alloy E110 on Embrittlement Under LOCA Conditions—
Part 1: Oxidation Kinetics and Macrocharacteristics of Structure and Fracture,” J. Nucl.
Mater., Vol. 418, 2011, pp. 1–7.
[13] Hózer, Z., Gyo†ri, C., Horváth, M., Nagy, I., Maróti, L., Matus, L., Windberg, P., and Frecska,
J., “Ballooning Experiments With VVER Cladding,” Nucl. Technol., Vol. 152, 2005, pp.
273–285.
[14] Chung, H., “The Effects of Aliovalent Elements on Nodular Oxidation of Zr-Base Alloys,”
Proceedings of the 2003 Nuclear Safety Research Conference, Oct 20–22, 2003,
NUREG/CP-0185, pp. 283–298.
[15] Yegorova, L., Lioutov, K., Smirnov, V., Goryachev, A., and Chesanov, V., “LOCA Behavior
of E110 Alloy,” Proceedings of the 2003 Nuclear Safety Research Conference, Oct
20–22, 2003, NUREG/CP-0185, pp. 123–140.
DISCUSSION
Question from Robert J. Comstock, Westinghouse Electric Co.:—What is the cur-
rent understanding regarding why sponge-based E110G shows a dramatic improve-
ment in breakaway oxidation performance?
Authors’ Response:—There are two reasons mentioned in the literature to

explain the potential reasons for breakaway oxidation:
a) The presence of some impurities (e.g. halogens) can be deleterious, and can
lead to increased susceptibility to nodular oxidation.
b) Experiments with polished E110 samples showed that surface treatment can
be used to avoid breakaway oxidation.
In the present study the two cladding tubes had the same surface roughness,
but the content of impurities was different. Probably the higher content of halogens
in the E110 could cause the observed breakaway phenomena.
Question from K. Kapoor, NFC:—Please provide individual concentrations of

Si, C, Ni, Cl, N, F, etc., for the two alloys?
Authors’ Response:—The F content was 30 ppm in E110 and 10 ppm in E110G,

while the Cl content was 3 ppm in E110 and 1 ppm in E110G. The concentration of
other above listed elements has not been identified yet.
Question from Mirco Grosse, Karlsruhe Inst. Of Technology:—The improvement

of the oxidation and hydrogen uptake is impressive. What is the time scale for
licensing and application of this material?
Authors’ Response:—The new alloy has already been tested in Russia in VVER-
440 reactors. The change in the composition of the alloy does not need necessarily
new licensing procedures. Taking into account that the behavior of the new alloy is
very similar to the currently used one in normal operational conditions, and its pa-
rameters are much better for accident situations, the new alloy can be quickly intro-
duced in the countries where the E110 is used.
Question from Martin Steinbrück, Karlsruhe Inst. Of Technology:—Did you

compare your results on E110G to M5?
Authors’ Response:—The E110G and M5 alloys have not been directly com-
pared in the present test series. The comparison with M5 oxidation data published
by AREVA shows that E110G and M5 oxidation kinetics at high temperature were
very close to each other. The direct comparison would be desirable in the future to
compare the behavior of two Nb containing Zr alloys produced in different factories

with very similar compositions. Our research center could carry out such compara-
tive tests if M5 alloy was accessible to us.
Question from Ron Adamson, Zircology Plus:—Were the surface conditioning

processes the same for both alloys? For instance, was one belt polished and the
other etched, etc.?
Authors’ Response:—According to the Russian fuel supplier the surface condi-

tioning processes was the same for both alloys.
Questions from S. Anantharaman, BARC Mumbai Q1:—What is the reason for

choosing ring compression tests instead of ring tension test?
Authors’ Response:—Both ring compression and ring tension tests can be used
to determine the ductile-to-brittle transition of oxidized cladding, since the load-
displacement curves from both tests show the characteristic behavior of ductile or
brittle material. The ring compression tests can be carried out with regular ring
shaped tube samples. For tension testing additional manufacturing may be needed
to produce a reduced section in the ring. Our laboratory traditionally applies ring
compression tests, and in order to compare the new results with the data of previ-
ous studies we continue to use this technique.
Q2:—What should be the limiting value of the hydrogen content of the fuel
cladding in your view?
Authors’ Response:—According to our measured data the Zr cladding shows

brittle behavior above 500 ppm H content.
Q3:—What is the effect of cladding hydrogen content on the % ECR?
Authors’ Response:—Our experiments with H pre-charged Zr cladding tubes

showed that H content up to 600 ppm does not change the oxidation kinetics in
high temperature steam. However, the mechanical testing of oxidized Zr tubes indi-
cated that the H pre-charged samples became brittle at lower ECRs, compared to as
received cladding.
Questions from P. Barberis, CEZUS Research Center:—E110G contains much

lower Hf than E110. Does it mean that the Zr-Hf separation process was changed
when going to Zr sponge?
Authors’ Response:—It was the intention of the fuel supplier to reduce the Hf
content of the alloy, and obviously they had some changes in the separation process.
S. Guilbert,1 P. Lacote,1 G. Montigny,1 C. Duriez,1

J. Desquines,1 and C. Grandjean1
Effect of Pre-Oxide on Zircaloy-4

High-Temperature Steam
Oxidation and Post- Quench
Mechanical Properties
Reference
Guilbert, S., Lacote, P., Montigny, G., Duriez, C., Desquines, J., and Grandjean, C., “Effect of
Pre-Oxide on Zircaloy-4 High-Temperature Steam Oxidation and Post- Quench Mechanical
Properties,” Zirconium in the Nuclear Industry: 17th International Symposium, STP 1543,
Robert Comstock and Pierre Barberis, Eds., pp. 952–978, doi:10.1520/STP154320120166, ASTM
ABSTRACT
During a loss-of-coolant accident (LOCA), the oxide layer grown during normal
operations is expected to have a protective effect against high-temperature
steam oxidation. However, with respect to post-quench ductility, such a barrier
effect may be counterbalanced by a partial dissolution of the pre-oxide layer and
associated diffusion of oxygen toward the metal. An extensive study has been
carried out on the high-temperature oxidation behavior and post-quench
mechanical properties of pre-oxidized Zircaloy-4. Cladding tubes have been pre-
oxidized in oxygen environment at 425 C for increasing durations (up to 550
days), leading to pre-oxide layer thicknesses ranging from 7 to 65 lm. All
samples have been then subjected to high-temperature steam oxidation at
900 C, 1000 C, or 1200 C followed by water quenching. Metallographic
examinations, micro-hardness measurements, mean hydrogen content
measurements, ring compression tests, and EPMA oxygen profiles have then
been performed. At 900 C and 1000 C, referring to weight gain, the protective
Manuscript received November 19, 2012; accepted for publication September 11, 2013; published online
September 22, 2014.
1
Institut de Radioprotection et de Sûreté Nucléaire (IRSN), PSN-RES, Saint Paul-lez-Durance, France.
2
Hyderabad, India.
Copyright V
GUILBERT ET AL., DOI 10.1520/STP154320120166 953
efficiency of the pre-oxide layer grown at 425 C appears to depend on its

thickness. The protective effect is significant for 20 - to 40 -lm pre-oxide
thicknesses. Above 40 lm, a partial loss of protectiveness is observed, associated
to a significant hydrogen pickup. Thus, mechanical properties of such pre-
oxidized claddings are highly degraded compared to an initially bare cladding
undergoing a LOCA transient for which post-test ductility is maintained. At
1200 C, at similar oxidation durations, weight gains decrease continuously as the
pre-oxide layer thickness increases. Thus, the pre-oxide can be viewed as partially
protective; however, post-quench ductility is not improved compared to initially
bare claddings.
Keywords
LOCA, oxygen diffusion, corrosion layer, steam oxidation, high temperature
Introduction
Under a loss-of-coolant accident (LOCA), the fuel cladding is submitted to a rapid
temperature increase (possibly up to 1200 C) in a steam environment, which indu-
ces an accelerated oxidation until quenching occurs, because of reflooding by the
emergency core cooling system operation (ECCS). Acceptance criteria defined in
1973 by the US-AEC limit the cladding temperature to 1204 C and the oxidation
amount to 17 % ECR to prevent excessive embrittlement of the fuel cladding and to
maintain a coolable geometry of the reactor core. These criteria were derived from
mechanical testing of as-received cladding tubes after oxidation in steam atmos-
phere. However, the influence of the oxide layer formed on the outer cladding sur-
face during the normal reactor operation (the so-called corrosion layer) has also to
be taken into account.
Fuketa [1] reports steam oxidation tests in the temperature range 1000 C
1200 C and time representative of a LOCA transient on samples irradiated at
48 GWd t1. A slight protective effect of the corrosion layer was observed regarding
the weight gain at the lowest LOCA transient temperatures for the shortest oxidation
time, while this protective effect was clearly reduced as LOCA transient temperature
and time were increased. Chuto et al. [2] also observed that a 50 -lm-thick corrosion
layer delays the growth of the high-temperature (HT) oxide on irradiated Zircaloy-2
(Zr2) cladding. At 1000 C and 1100 C, the high-temperature oxidation initiates first
beneath the radially oriented cracks of the corrosion layer and grows uniformly after
longer oxidation durations or at higher temperature (1200 C).
For steam oxidation at 900 C–1200 C with duration up to 2400 s, Baek and
Jeong [3] also observed lower weight gains compared to as-received samples for
unirradiated samples pre-oxidized at 450 C in steam and having a pre-oxide layer
thickness of 6 lm. A decrease of the weight gain for samples having a thin pre-
oxide layer (<15 lm, obtained in the CEA Reggae corrosion loop at 350 C) after
steam oxidation at 1200 C for 55 s is also observed by Brachet et al. [4]. However,
the post-quench mechanical behavior of the pre-oxidized samples was similar to
that of as-received samples oxidized at high temperature in the same conditions.

These results were attributed to the fact that the overall quantity of oxygen atoms
which diffuse into the metallic layers of the cladding is nearly the same with and
without a pre-oxide layer for short-term oxidation, whatever the source of oxygen
(the pre-oxide or the steam atmosphere under LOCA transient). Recently, Le Saux
et al. [5] obtained comparable results for samples pre-oxidized in autoclave at
340 C–360 C up to a pre-oxide layer thickness ranging from 10 lm to 35 lm and
subsequently oxidized in steam at 1000 C–1200 C for short oxidation durations.
Leistikow and et al. [6] performed steam oxidation for a 5-min exposure at
1000 C–1200 C on samples pre-oxidized in steam and having several pre-oxide
layer thickness ranging from a few lm up to 50 lm. They observed a protective
effect of pre-oxide layers in terms of weight gain for a thickness up to 10 lm. For
thicker pre-oxide layers, weight gains were similar or even higher than for as-
received samples oxidized under the same high-temperature conditions. Leistikow
et al. attributed the loss of the protective effect for thicker pre-oxide layers to
stresses between metal and oxide, leading to the spread of cracks and delamination
of the pre-oxide layer.
Experiments performed by Vrtilkova [7] showed a lower ductility and a higher
micro-hardness for cladding having a 10 - to 50 -lm pre-oxide layer (obtained at
425 C in steam), compared to as-received samples, after steam oxidation at
900 C–950 C. A degradation of the mechanical properties for pre-oxidized clad-
ding was also observed by Guilbert et al. [8] after steam oxidation at 900 C for long
duration (100 min). However, these results were questioned as to a possible impair-
ing influence of the pre-oxidation temperature (500 C).
To summarize, data published in the literature indicate that, referring to the weight
gain, a pre-oxide layer acts as a protective barrier against high-temperature oxidation.
This protective effect delays the formation of the HT oxide. It is less effective at 1200 C
or after long oxidation durations. But referring to post-quench mechanical properties
of the cladding, the effect of a pre-oxide layer is less obvious. It also has to be pointed
out that in most of the published data, the pre-oxide layers were obtained by oxidation
in steam or in water, involving possibly significant hydrogen pickup during pre-
oxidation, which might have influenced the post-quench embrittlement. Unfortu-
nately, the hydrogen pickup during the pre-oxidation was not provided for most of the
related experimental data.
To get a clearer view on the effect of corrosion layer on weight gain and on
post-quench mechanical properties, a parametric study on the high-temperature
oxidation of cladding specimens pre-oxidized in pure oxygen has been launched
and results are presented in this paper. Pre-oxidation conditions were chosen to
avoid any uncontrolled hydrogen loading. This allows to separate the influence of
the hydrogen content from the influence of the corrosion layer. Zircaloy-4 (Zr4)
cladding samples were pre-oxidized for increasing durations to obtain a pre-oxide
layer thickness ranging from 10 to 70 lm then oxidized in steam at 900 C, 1000 C
or 1200 C. For each temperature, two oxidation durations were chosen: one long
duration to better understand the involved phenomena, the other one being shorter
to be more representative of a LOCA transient. Particular experimental attention
was paid to manage the hydrogen pickup during the pre-oxidation phase (corrosion
layer formation) and during the high-temperature steam oxidation. Hydrogen con-
tents were measured before and after HT oxidation. A few complementary tests
were also performed on autoclaved samples provided by EDF and on one sample
pre-hydrided under controlled conditions prior to pre-oxidation.
Experimental
MATERIAL AND EQUIPMENTS
The specimens used in this study were cut from low tin stress-relieved annealed
(SRA) Zircaloy-4 PWR 17 17 industrial fuel cladding tubes, provided by AREVA-
NP. Composition of the alloy is given in Table 1. Outer diameter and wall thickness
of the tubes are 9.5 mm and 570 lm, respectively. Open samples were used, leading
thus to two-sided oxidations.
For pre-oxidation, 100-mm-long cladding samples were heated at 425 C in a
tubular resistive furnace in a pure oxygen flow. 20-mm-long samples were then cut
and oxidized at high temperature in a vertical resistive furnace in a steam and argon
flow (with a heating rate of 50 C s1), then directly water quenched by dropping
into a water bath. This direct quench may be the cause of loss of pre-oxide frag-
ments observed in few cases, when cohesion with the underlying HT oxide was
poor (specifically for tests at 1200 C). Regarding the influence of a direct quench
on ductility, Brachet et al. [4] have shown that slow cooling before the final water
quenching induces a significant oxygen redistribution. Depending on the final
quenching temperature, mechanical testing revealed either no significant effect of
this redistribution or a post-quench ductility restoration effect. Direct quenching
can thus be considered as conservative with regards to safety.
The argon and steam flow rates were, respectively, 2 Nl min1 and 100 g h1
at 900 C and 1000 C (50 % vol. Ar and 50 % vol. steam). At 1200 C, the argon and
steam flow were increased to 10 Nl min1 and 500 g h1 to avoid any occurrence
of steam starvation.
Weight and length of the samples were measured with an accuracy of 0.1 mg
and 0.01 mm, respectively, before and after pre-oxidation and HT oxidation.
After each step of the preparation protocol, metallographic cuts were examined
by optical microscopy. Hydrogen contents were measured by the hot extraction tech-
nique (analyzer system JUWE ON/H-mat 286). For each sample, measurements are
TABLE 1 Chemical composition of the Zr-4 batch used for pre-oxidation in oxygen.
Sn (wt. %) Fe (wt. %) Cr (wt. %) O (wt. %) H (wppm)
1.3 0.21 0.10 0.13 10

repeated at least three times on small pieces of few tens of milligrams on two different
rings after sandblasting the oxide layer. The hydrogen pickup during HT oxidation is
calculated as the difference of hydrogen content measured after steam oxidation and
the one measured after pre-oxidation.
After the high-temperature oxidation, radial oxygen concentration profiles and
distribution maps were obtained by electron probe microanalysis (CAMECA SX-
100). The oxygen Ka line was measured with a W/Si multi-layer synthetic crystal.
Most of the profiles were performed along 200 -lm-long lines with a 1 -lm stage
displacement step. Maps were obtained from 100 200 lm2 or 100 300 lm2
scanning with a stage displacement step of 1 lm.
Microhardness measurements were performed using a BUEHLER apparatus
equipped with a Vickers diamond tip. A 100 -g load was used. On each metallo-
graphic cut, four profiles were measured across sample radius in the metal with
100 -lm spacing between indentation marks. For a given sample, a mean prior-b
micro-hardness value is calculated from all the data measured in the prior-b region
(about 10 to 20 data by sample).
Ring compression tests (RCT) were performed at room temperature on
10-mm-long samples using an INSTRON 5566 machine, with a 1 mm min1 dis-
placement rate until fracture. Maximum load, deformation at failure and plastic de-
formation at failure were derived from the load-displacement curve. The failure of
the ring is considered to occur at the first significant load drop in the curve. The
“offset strain” parameter [1] is defined as the ratio of the plastic displacement at
failure (d) to the outer initial diameter of the sample [9]:
d
(1) offset strain % ¼ 100
diameter
PRE-OXIDATION
Specimens were pre-oxidized in pure oxygen at 425 C for duration ranging from 64
to 552 days, leading to oxide layer mean thickness of 7.1 6 1.6, 12.2 6 1.1,
17.9 6 1.1, 32.9 6 2.0, 41.1 6 2.5, 53.2 6 2.5, and 64.5 6 2.4 lm, respectively. Here,
each pre-oxide layer thickness and standard deviation is determined from a set of
16 measurements performed on two metallographic radial cuts. Cladding pre-
oxidized in autoclave at 360 C (in pressure and water chemistry conditions close to
PWR operation) were provided by EDF with 18.1 6 0.6 lm and 28.3 6 0.7 -lm pre-
oxide layer thicknesses. At last, a pre-hydrided and successively oxygen pre-
oxidized sample has been also tested (437 6 39 wt. ppm hydrogen and 42.8 6 2.7 -
lm pre-oxide layer thickness). Note that for both pre-oxidation modes (oxygen and
autoclave), the inner and outer oxide layer thicknesses are similar. Examples of met-
allographs are presented in Figs. 1(b), 1(c), and 1(d) and compared with metallographs
of irradiated cladding (Fig. 1(a)).
Clear differences between the different pre-oxidation conditions are evidenced.
Autoclaved pre-oxide layers appear quite compact compared to pre-oxide layer
FIG. 1 Comparison of oxide layers formed in oxygen, in autoclave and in PWR [12]: (a)
irradiated in PWR, 61 MWd/t, 5 cycles, 3rd span, oxide thickness ¼ 30 lm, BSE
image; (b) oxide layer formed in autoclave at 350 C, thickness ¼ 28 þ/ 1 lm,
BSE image; (c) oxide layer formed in oxygen at 425 C, thickness ¼ 33 þ/ 2 lm,
BSE image; and (d) oxide layer formed in oxygen at 425 C, thickness ¼ 65 þ/
2 lm, BSE image.
formed at 425 C or to irradiated cladding. The typical stratified microstructure of

zirconium alloy corrosion [10–12] is seen for all pre-oxidation conditions, but for
oxides formed in autoclave, the circumferential cracks associated with the stratifica-
tion are barely seen, whereas these cracks are clearly more opened for oxidation
conditions at 425 C with oxygen. With regards to this feature, oxidation in oxygen
at 425 C seems to better correspond to in reactor corrosion than oxidation in auto-
clave. A 2 -lm crack spacing in the radial direction is reported for irradiated clad-
ding [12], which is consistent with the metallographic examinations on our
samples. However, for irradiated cladding, the crack size seems to decrease beyond
20–25 lm from the outer surface for thick oxide layers, whereas for pre-oxide
grown in oxygen, longer and wider cracks can be observed close to the metal/oxide
interface for a pre-oxide layer thickness equal or above about 40 lm. Radially ori-
ented veins free of cracks are also visible for oxygen and autoclave pre-oxidation, as
for irradiated cladding [12]. At 425 C, both the metal/oxide interface and the strata
are wavy, whereas much more regular interface and strata are observed for auto-
claved samples. From that point of view, the autoclaved samples show similarities
with irradiated cladding. Finally, at 425 C, for pre-oxide layer thickness exceeding
30 lm, a network of regularly spaced radial cracks is observed, giving to the sample
surface a snake skin aspect. Such cracks have also been observed on irradiated clad-
dings but only for thick corrosion layers (80–100 lm) [12]. It has been suggested
that these cracks result from the metal creep when subjected to tensile stresses
induced by the growing oxide [13]; cracks would initiate above a critical oxide layer
thickness, depending on the temperature, in the most external oxide sub-layer
where compressive stresses have been converted into tensile stresses as a conse-
quence of metal creep. Because the newly formed oxide sub-layer is always sub-
jected to compressive stress, the cracks never reach the metal, as far as no external
mechanical load is applied.
Vickers micro-hardness measurements for pre-oxidized samples indicated flat
profiles in the metal. For samples pre-oxidized at 425 C, mean values appears sig-
nificantly lower than the one measured for the as-received alloy as a result of the
long-term treatment at 425 C. However, it remains higher than in fully recrystal-
lized Zircaloy, indicating that recrystallization at 425 C is only partial, and increases
with time. Micro-hardness for the autoclave pre-oxidized samples is slightly higher
than for oxygen pre-oxidized samples, because of the lower pre-oxidation tempera-
ture and thus lower defect annealing and recrystallization (see Fig. 2).
FIG. 2 Micro-hardness measurements on oxide layers formed in oxygen or in autoclave.

Each data point is the average over 40 measurements and error bars give the
standard deviation over these 40 measurements.
Hydrogen analysis shows that the oxygen pre-oxidation process induces low
but non-zero hydrogen pickup (<75 wt. ppm). Autoclaved sample have higher
hydrogen contents (305 6 10 wt. ppm for 18 lm, 670 6 35 wt. ppm for 28 lm).
HIGH-TEMPERATURE STEAM OXIDATION ON BARE CLADDING SAMPLES

For initially bare samples, after HT steam oxidation, metallographs show the pres-
ence of a dense and uniform oxide on both the inner and the outer side. The oxide/
metal interface is rather smooth for all the tested conditions and the morphology of
the oxide is typical of a pre-breakaway oxidation regime. Hydrogen pickup during
the steam oxidation is low (5 wppm), which is also consistent with a pre-
breakaway regime. The metal hardness is homogeneous in the central prior-b
region, and higher than before oxidation. This is consistent with expected oxygen
diffusion and saturation of the b layer. At 900 C and 1000 C, the mean micro-
hardness in the prior-b region increases only moderately and no significant drop in
the RCT load-displacement curve is evidenced, indicating that the samples remain
ductile. At 1200 C, where the prior-b region hardness increase is significant
(>þ200 Hv0.1) compared to the as-received alloy, the offset strain value is close to
or lower than the 2 % limit, indicating a brittle behavior [9].
HIGH-TEMPERATURE STEAM OXIDATION ON PRE-OXIDIZED CLADDING

SAMPLES
Regarding weight gain and oxide morphology, for the thinnest oxygen pre-oxide
layer (7 lm), the sample color, which was light brown after the pre-oxidation, has
turned to a mix of black and beige areas. Proportions of black and beige depend on
temperature and oxidation duration. For thicker oxygen grown pre-oxide layers,
the black color is replaced progressively by a beige network of cracks as the pre-
oxide layer thickness increases. For the thickest pre-oxide layer, the beige color cov-
ers the entire sample surface, whatever the conditions for the high-temperature oxi-
dation. The black color at the surface of the samples indicates that the pre-oxide
layer has become sub-stoichiometric up to its surface, whereas beige regions corre-
spond to area where the oxide is stoichiometric, at least at the surface. Chemical
reduction of the pre-oxide layer is also observed for the pre-hydrided and pre-
oxidized sample, as well as for the samples pre-oxidized in autoclave. For the latter,
sample color has turned to black with small beige islands interspersed with cracks.
Regarding the weight gained during the HT oxidation at 900 C–1000 C (see
Fig. 3), the general trend for oxygen pre-oxidized samples is a decrease as the thick-
ness of the pre-oxide layer increases, for pre-oxide thickness up to 40 lm. This
indicates that, for the investigated durations, there is a protective effect of these pre-
oxide scales. However, above 40 lm pre-oxide thickness, weight gains show that the
protective effect is partly lost. For autoclaved samples, as well as for the pre-
hydrided and pre-oxidized sample, weight gains caused by HT steam oxidation are
low and comparable to the weight gains measured in the same high-temperature
FIG. 3 Weight gain and hydrogen pickup during steam oxidation versus pre-oxide
thickness.
conditions for samples pre-oxidized in oxygen and having similar pre-oxide layer
thickness. At 1200 C (Fig. 3), the observed trend regarding weight gain as a function
of pre-oxide thickness is also a decrease, but unlike at 900 C–1000 C, the protective
effect efficiency still increases beyond 40 lm pre-oxide layer thickness.
Hydrogen pickups at 900 C and 1000 C appear to be non-zero for all types of
pre-oxidized samples, when no hydrogen pickup was measured for the bare alloy in
the same oxidation conditions (Fig. 3). Hydrogen incorporation remains low (but
significant) for pre-oxide layer thicknesses lower than 40 lm, but can be very high
for the thickest pre-oxide layers investigated.
At 1200 C, hydrogen pickup is low up to 40 lm pre-oxide thickness, however a
large data scatter is observed for the samples having a 10–20 lm pre-oxide layer
thickness, revealing that the hydrogen charging during the HT oxidation is spatially
non homogeneous within a given sample. Hydrogen pickup then increases signifi-
cantly above 40 lm pre-oxide layer thickness, as it was the case at 900 C and
1000 C.
For the thinnest oxygen pre-oxide layer (7 lm), metallographs always show af-
ter high-temperature oxidation the presence of a high-temperature oxide, which
has grown beneath the pre-oxide, both on the outer and the inner side (Figs. 4 to 6).
At 900 C–1000 C, this HT oxide is barely observed for pre-oxide thicknesses up to
40 lm. Observation at high magnification of the metal/oxide interfacial region
reveals that part of the pre-oxide layer has turned back into metal (dissolution of
oxygen). Above 40 lm, metallographs show again the presence of the HT oxide on
the outer side and seldom on the inner side. Its thickness and lateral spreading
depend on the oxidation duration. For short oxidation durations, it is worthwhile
noticing that the HT oxide does not form where pores or voids are present at the
metal/oxide interface. In other terms, the HT oxide forms only where the pre-oxide
is in close contact with the metal, giving rise to nodular growth of the HT oxide;
thus accentuating the waviness of the metal/oxide interface. At 1200 C, metallo-
graphs show the presence of the HT oxide both on inner and outer sides, with a flat
metal/oxide interface, whatever the pre-oxide thickness (Fig. 6). Thickness of the
HT oxide layer decreases as the pre-oxide layer thickness increases. It is also inter-
esting to note the presence of circumferential cracks at the pre-oxide/HT oxide
interface. These cracks become longer and wider as the pre-oxide layer thickness
increases. For the thicker pre-oxide layer, cracks even get connected and the pre-
oxide layer delaminates. This phenomenon of growth of cracks and coalescence
beneath the pre-oxide layer is also observed at 900 C and 1000 C, even if less pro-
nounced. Increase of the hydrogen pickup observed above 40 lm for all tempera-
ture may be linked to this phenomenon.
For autoclaved samples and the pre-hydrided and pre-oxidized sample, metal-
lographs show that HT oxide growth initiates locally at cracks in the pre-oxide. As
for oxygen pre-oxidized samples, observation of the metal/oxide interfacial region
reveals that part of the pre-oxide layer has turned back into metal (Fig. 7).
For post-test ductility and microhardness, the mean micro-hardness in the cen-
tral prior-b region is determined after the high-temperature oxidations from at least
16 indentation measurements per sample. At 900 C and 1000 C, for oxygen pre-
oxidized samples, results depend on the pre-oxide layer thickness: up to about 30 -
lm thickness, hardness of the prior-b region does not differ much from the ones
measured on initially bare cladding for the same oxidation temperature and dura-
tion. Above 30 -lm pre-oxide layer thickness, again at 900 C and 1000 C, a signifi-
cant increase of the micro-hardness of the prior-b phase is observed, which may be
linked to the high hydrogen content of these samples after the HT oxidations.
Indeed, it is well established that hydrogen dissolved in the metal increases the solu-
bility of oxygen, both in the a and b phase of zirconium [4,14]. Oxygen concentra-
tion profiles determined from EPMA scans after HT oxidations do confirm that the
oxygen concentration in the prior-b central region is high for samples having a
FIG. 4 Micrographs on oxygen pre-oxidized samples after steam oxidation at 900 C.



FIG. 7 Micrographs on pre-hydrided and oxygen pre-oxidized samples and autoclaved

samples after steam oxidation.
high pre-oxide layer thickness (see Fig. 8(a)). Autoclaved samples, having significant
amount of hydrogen incorporated during the pre-oxidation, also show a rather
high mean hardness of the prior-b region after HT oxidation, and correlatively a
high oxygen concentration. Same trends are observed for the pre-hydrided sample.
RCTs show loads at failure and offset-strains that are strongly correlated to
hydrogen pickup at 900 –1000 C for oxygen pre-oxidized samples. Where the pre-
oxide induces a high hydrogen pickup, cladding embrittlement is observed (see Fig.
9). However, for autoclaved samples and the pre-hydrided and pre-oxidized sample,
although the hydrogen pickup during steam oxidation remains low, the mechanical
properties are degraded compared to as-received samples. In that case, the degrada-
tion of the mechanical properties are linked to the initial hydrogen contents, which
increases the solubility limit of oxygen in the a- and b-Zr phases. Indeed, for
these samples, higher micro-hardness and oxygen concentrations are measured
compared to oxygen pre-oxidized samples (see Fig. 8(a)). At 1200 C, oxygen con-
tent in the metal is higher than at 900 C–1000 C and all samples are brittle (see
Fig. 8(b)).
FIG. 8 Experimental oxygen profiles measured by EPMA: (a) comparison between

oxygen, autoclave, and pre-hydrided pre-oxidized samples after oxidation at
1000 C for 15 min, and (b) comparison between oxygen pre-oxidized samples
after oxidation at 1000 C for 15 min and at 1200 C for 200 s.
FIG. 9 Offset-strain from ring compression tests at RT plotted versus pre-oxide

thickness.
For all temperatures and pre-oxide thickness above 10–20 lm, the HT oxide on
the inner side is always thinner than on the outer side, unlike as-received samples
where the HT oxide thickness is the same on the inner and the outer sides. It is
likely because of compressive stresses which develop in the internal oxide sub-layer
and give a slightly higher protective influence to the inner pre-oxide.
Discussion
Zircaloy-4 cladding samples pre-oxidized in oxygen at 425 C or in autoclave at
360 C in pressurized water have been oxidized in flowing steam at high tempera-
ture (900 C, 1000 C, and 1200 C) for different durations. Comparing weight gains
to the ones obtained with bare samples in the same oxidation conditions, it appears
that the pre-oxide has a protective effect against high-temperature oxidation, for
both types of pre-oxide. This protective effect is less pronounced at 1200 C than at
900 C and 1000 C, however, it is still effective. It is also less pronounced for a pre-
oxide layer thickness beyond 40 lm (in the present study pre-oxide thicknesses up
to 65 lm have been investigated for samples pre-oxidized in oxygen).
It is also worthwhile noticing the very good agreement between the data
obtained with the 425 C pre-oxide and those obtained with the autoclave
pre-oxide. Metallographic examinations of the samples after the high-temperature
oxidations corroborate these weight gain measurements, showing that the amount
of high-temperature oxide that has grown beneath the pre-oxide layers is lower
(and even much lower at 900 C and 1000 C) than the one observed on metallogra-
phies of initially bare samples. These results are in agreement with the ones
obtained in other laboratories on irradiated and non-irradiated cladding
[2,5,15,16]. It has to be pointed out that only rather thin pre-oxide thicknesses were
investigated in these previous studies.
Hydrogen pickup during the high-temperature oxidation has also been investi-
gated, by systematically measuring the samples hydrogen contents before and after
the HT oxidation phase. It was evidenced that the presence of a pre-oxide layer, for
all pre-oxide layer thickness tested, promotes hydrogen pickup during the high-
temperature steam oxidation phase. Comparatively, no hydrogen pickup is
observed in the same oxidation conditions using initially bare cladding samples. At
1200 C, below 40 -lm pre-oxide thickness, despite significant weight gains, hydro-
gen pickup remains negligible, as for bare Zircaloy-4. At 900 C and 1000 C,
however, moderate but significant hydrogen pickup is measured. A strong degrada-
tion in the hydrogen pickup at the three considered temperatures is observed for
pre-oxide layers thicknesses above 40 lm. At 900 C and 1000 C, the hydrogen
absorption can reach 100 % of the hydrogen released by the oxidation reaction
(Fig. 10).
As a consequence of the hydrogen absorption, greater embrittlement for pre-
oxidized samples than for the bare cladding is observed, as evidenced by ring com-
pression tests. The protective effect of the pre-oxide against the oxidation in steam
atmosphere does not prevent from oxygen dissolution in the metal, because the
pre-oxide is itself a source of oxygen. The chemical reduction of the pre-oxide layer
even in steam environment, revealed by its color change from beige to black, proves
that oxygen release from the zirconia pre-oxide layer is faster than its replacement
by oxygen from the steam.
FIG. 10 Hydrogen pickup fraction during HT steam oxidation versus pre-oxide layer
thickness. Error bars correspond to 6 one standard deviation on a given set of
measurements. A high value of the standard deviation is associated to
inhomogeneous hydrogen content on a given sample rather to actual
uncertainty of the measurement method. Dotted lines are trends for
explanation purposes.
High hydrogen pickups were also observed by Nagase on Zircaloy-4 samples

having a 50 -lm-thick pre-oxidation layer grown in oxygen and further oxidized in
steam at 1100 C, when bare Zr-4 cladding oxidized in comparable conditions
showed no hydrogen absorption [17]. It is worthwhile noticing that present results
confirm the one previously obtained with pre-oxide layers grown at 500 C, with
also higher weight gain and hydrogen absorption above 40 -lm pre-oxide layer
thickness [8].
Using in situ neutron radiography, Grobe [18] has shown that high hydrogen
absorption rates during high-temperature steam oxidation are observed only at the
beginning of the exposure, that is when the metal is bare or covered with only a
thin oxide layer. For pre-oxidized cladding, it is likely that significant hydrogen
absorption can be obtained only if hydrogen is released by the oxidation reaction in
the vicinity of the metal surface. In other terms, it means that H2O gas-phase diffu-
sion through a network of cracks and porosities connected to the external
atmosphere should contribute to the oxygen supply to the metal. The presence of
pores close to or at the metal/oxide interface would favor accumulation of hydrogen
close to the metal (if H2O dissociation occurs there) and, thus, its absorption. Of
course, O2- solid-state diffusion through the pre-oxide can also contribute to the ox-
ygen supply. A high hydrogen pickup fraction means a high contribution of the
gas-phase transport mechanism. Decrease of the hydrogen pickup fraction as the
temperature increases (see Fig. 10) can be explained by the relative increase of the
contribution of the solid state diffusion mechanisms. Indeed, the oxygen solid state
diffusion rate increases exponentially with temperature, whereas the gas-phase dif-
fusion rate is a less thermally activated process.
The microstructure of the pre-oxide (the distribution and size of cracks, the
degree of percolation) thus undoubtedly determines the possibility and relative im-
portance of the gas-phase transport mechanism.
It is worthwhile noticing that the gas phase transport mechanisms are likely
slowed down in autoclaved samples, because the corrosion layer is denser.
There is potentially a contribution of the network of radially oriented incipient
cracks on the oxidation mechanism. Such periodic set of cracks is observed when
the pre-oxidation duration at 425 C exceeds 297 days, that is, when the pre-oxide
thickness exceeds 30 lm. It could promote faster access of steam to the metal and
then induce the partial loss of protectiveness of the pre-oxide and the high hydro-
gen pickups observed above 40 -lm-thick pre-oxide layers. Nagase and co-workers
have performed oxidation tests at 1100 C with Zircaloy-4 cladding samples pre-
oxidized in oxygen, and have indeed observed that the high-temperature oxide pref-
erentially forms beneath each radial cracks and progressively spreads to the entire
surface [17]. Such a direct effect of the radial cracks is not evidenced on the metallo-
graphic data supporting the present experimental program, even at 1200 C (the ox-
idation behavior in steam is generally similar at 1100 C and 1200 C). The
nucleation of theses radial cracks is related to the metal creep during the low-
temperature corrosion process [19,20]. The thickness of the un-cracked pre-oxide
layer depends on the metal creep rate and on the corrosion rate, that is, for a given
alloy, on the pre-oxidation temperature. Nagase’s results may be explained by a
high pre-oxidation temperature, resulting in deeper incipient cracks at similar pre-
oxide layer thickness [17]. In the present study, using 425 C corrosion temperature,
measurements of the crack depth show that the unaffected pre-oxide thickness
varies between 20 and 35 lm. Consequently, the radial crack network cannot be the
only cause for the gas-phase transport down to the metal, even if it may favor it.
Undoubtedly, the circumferentially oriented crack network within the pre-oxide
plays a role.
It is worthwhile noticing that after the high-temperature oxidation, metallo-
graphic images often reveal the presence of widely connected circumferential
cracks, located either at the pre-oxide/high-temperature oxide interface when high-
temperature oxide have formed (Fig. 12(b)), or otherwise at the metal/oxide interface
(Fig. 12(a)). At 1200 C, these strongly connected porosities even lead to
FIG. 11 Example of post-test metallography illustrating the reduction of the pre-oxide

at the metal/oxide interface (pre-hydrided and pre-oxidized sample after steam
oxidation at 1000 C for 15 min).
delamination of the remaining pre-oxide layer (Fig. 12(c)). It is suggested that this
porosity layer forms as a result of the pre-oxide reduction process, which occurs at
beginning of the high-temperature phase by dissolution of oxygen from the pre-
oxide into the metal, when oxygen from the steam atmosphere has not still reached
the metal.
Experimentally, dissolution of the pre-oxide can be evidenced by post-test sam-
ple examination: for the thinnest pre-oxide thicknesses, it is observed that the sam-
ple’s color has turned to black after the high-temperature phase (it was light brown
before), which undoubtedly corresponds to a pre-oxide chemical reduction. For
some samples, it can be even observed a conversion of a thin layer of pre-oxide into
metal by metallographic examination of the metal/oxide interface at high magnifi-
cation (see Fig. 11).
The process of formation of the porosity layer just above the metal is shown
schematically on Figs. 13(a), 13(b), and 13(c): because of the outward progression of
the metal/oxide interface resulting from the pre-oxide dissolution, cracks which
were initially inside the pre-oxide come at the interface, and then, outward progres-
sion of the interface stops. Further dissolution of the pre-oxide proceeds by dissolu-
tion of the oxide bridges between cracks; still in contact with the metal. It
progressively leads to cracks lateral spreading and coalescence. The higher the pre-
oxide thickness is, the longer is the dissolution period before oxygen from the out-
side reach the metal. At 1200 C, pre-oxide dissolution is so fast that it can go up to
delamination of the pre-oxide layer. When oxygen from the steam atmosphere
becomes available—either through solid-state diffusion or through gas-phase trans-
port of steam and further dissociation—a high-temperature oxidation process ini-
tiates, forming first oxide nodules (Figs. 12(a) and 13(d)), then as a continuous layer
if the oxidation duration is long enough (Figs. 13(e), 12(b) and 12(c)). If oxygen
reaches the metal/oxide interface as H2O molecules (gas-phase transport), the
hydrogen released by the H2O dissociation will accumulate in the free volume
formed between the remaining corrosion layer and the high-temperature oxide
layer connected to the metal, favoring its incorporation in the metal.
FIG. 12 Typical post-test metallographs illustrating the formation of a porous layer in

the pre-oxide by lateral spreading and coalescence of circumferential cracks.
Zircaloy-4 samples pre-oxidized in oxygen during 552 d (65 -lm pre-oxide
thickness) and oxidized in steam at 1000 C for 15 min (a), at 1000 C for 30 min
(b), and at 1200 C for 10 min (c).
FIG. 13 Schematic of the different steps involved in the high-temperature oxidation of

Zircaloy-4 pre-oxidized cladding.

An extensive study has been carried out on the protective effect of corrosion layer
with regards to the high-temperature oxidation, which happens during a LOCA
transient. Zircaloy-4 cladding samples pre-oxidized in oxygen at 425 C or in auto-
clave at 350 C–360 C in pressurized water have been oxidized in flowing steam at
high temperature (900 C, 1000 C, and 1200 C) for different durations. For all
tested temperatures and for both types of pre-oxide, the weight gains measured for
pre-oxidized samples are lower than those obtained for bare cladding, indicating
that the pre-oxide has a protective effect against extended oxidation at high temper-
ature. Metallographic examinations of the samples after the high-temperature oxi-
dation confirm the protective effect of the pre-oxide layer, showing lower amounts
of high-temperature oxide formed when a pre-oxide layer is present. However,
above a 40 -lm corrosion layer, a partial loss of protectiveness with regards to
embrittlment is observed for oxygen pre-oxidized samples, associated to a signifi-
cant hydrogen pickup.
As a consequence of the hydrogen absorption, greater embrittlement for pre-

oxidized samples than for the bare cladding is observed. Hydrogen absorption can
occur only if at least part of the oxygen is supplied to the metal as H2O molecules, that
is, via gas phase transport through the pre-oxide scale. The radial crack network
observed on oxygen pre-oxidized samples for pre-oxide thicknesses exceeding 30 lm
probably facilitates this gas-phase transport. However, it cannot be the only cause for.
The circumferentially oriented crack network within the pre-oxide plays also a role.
Samples pre-oxidized in autoclave have been investigated for comparison, how-
ever with a pre-oxide thickness range limited to 20–30 lm. As for the oxygen pre-
oxidized samples having comparable pre-oxide layer thickness, the hydrogen
absorption inside these samples during the high-temperature phase is moderate.
Nevertheless, autoclaved samples are brittle after HT oxidation, whereas initially
bare cladding or oxygen pre-oxidized cladding keeps significant ductility. Such a
degradation of the mechanical properties is linked to the high hydrogen loading
during the autoclave treatment.
This work has shown that the corrosion layer formed during normal reactor
operation may have a strong influence on the cladding mechanical behavior after
HT transients. A better understanding of the various parameters influencing, for a
given alloy, this microstructure would be needed. Specifically, the role played by
mechanical coupling between the metal and the oxide, and by the hydrogen absorp-
tion and mobility during the corrosion process, has to be considered.
ACKNOWLEDGMENTS
Authors are grateful to EDF for its financial support and EDF/CEA/AREVA R&D
partners for providing autoclaved samples.
References
[1] Fuketa, T., “JAERI Experimental Basis on RIA and LOCA,” OECD/CSNI/SEGFSM Meeting,
OECD Headquarters, April 25–26, Paris, France, 2005.
[2] Chuto, T., “Oxidation of High Burnup Fuel Cladding in LOCA Conditions,” Fuel Safety
Research Meeting, May 19–20, Tokai, Japan, 2010.
[3] Baek, J. H. and Jeong, Y. H., “Steam Oxidation of Zr-1.5Nb-0.4Sn-0.2Fe-0.1Cr and

Zircaloy-4 at 900–1200 C,” J. Nucl. Mater., Vol. 361, 2007, pp. 30–40.
[4] Brachet, J. C., Vandenberghe-Maillot, V., Portier, L., Gilbon, D., Lesbros, A., Waeckel, N.,
and Mardon, J.-P., “Hydrogen Content, Preoxidation, and Cooling Scenario Effects on
Post-Quench Microstructure and Mechanical Properties of Zircaloy-4 and M5 Alloys in
LOCA Conditions,” Zirconium in Nuclear Industry: 15th International Symposium, ASTM
STP 1505, Sunriver, OR, ASTM International, West Conshohocken, PA, 2008.
[5] Le Saux, M., Brachet, J. C., Vandenberghe, V., Gilbon, D., Mardon, J. P., and Sebbari, B.,
“Influence of Pre-Transient Oxide on LOCA High Temperature Steam Oxidation and
Post-Quench Mechanical Properties of Zircaloy-4 and M5 Cladding,” 2011 Water Reactor

Fuel Performance Meeting, Sept 11–14, Chengdu, China, 2011.
[6] Leistikow, S., Schanz, G., and Berg, H. V., “Kinetik and Morphologie der Isothermen
Dampf-Oxidation von Zircaloy 4 bei 700–1300 C [Cinetics and mechanisms of isotherm
steam oxidation of Zircaloy-4 in the range 700–1300 C],” Karlsruhe Institute of Technol-
ogy, Karlsr̈he, Germany, 1978 (in German).
[7] Vrtilkova, V., “Review of Recent work at UJP PRAHA on the LOCA Embrittlement
Criterion,” 6th Plenary Meeting of the OECD/CSNI/SEGFSM, April 25–26, Paris, France,
2005.
[8] Guilbert, S., Duriez, C., and Grandjean, C., “Influence of a Pre-Oxide Layer on Oxygen Dif-
fusion and on Post-Quench Mechanical Properties of Zircaloy-4 after Steam Oxidation
at 900 C,” LWR Fuel Performance Meeting/TOPFUEL, Sept 26–29, Orlando, FL, 2010.
[9] Billone, M., Yan, Y., Burtseva, T., and Daum, R., “Cladding Embrittlement during Postu-
lated Loss-of-Coolant Accidents,” NUREG-CR-6967, Nuclear Regulatory Commission,
[10] Maroto, A. J. G., Bordoni, R., Villegas, M., Olmedo, A. M., Blesa, M. A., Iglesias, A., and
Koenig, P., “Growth and Characterization of Oxide Layers on Zirconium Alloys,” J. Nucl.
Mater., Vol. 229, 1996, pp. 79–92.
No. 3, 2006, pp. 265–281.
[12] Bossis, P., Pêcheur, D., Hanifi, K., Thomazet, J., and Blat, M., “Comparison of the High
Burn-Up Corrosion on M5 and Low Tin Zircaloy-4,” Zirconium in the Nuclear Industry:
14th International Symposium, ASTM STP 1467, Stockholm, Sweden, ASTM International,
[13] Busser, V., Desquines, J., Fouquet, S., Baietto, M.-C., and Mardon, J.-P., “Modelling of Cor-
rosion Induced Stresses during Zircaloy-4 Oxidation in Air,” Mater. Sci. Forum, Vol.
595–598, 2008, pp. 419–427.
[14] Duriez, C., Guilbert, S., Stern, A., Grandjean, C., Belovsky, L., and Desquines, J.,
“Characterization of Oxygen Distribution in LOCA Situations,” J. ASTM Int., Vol. 8, No. 2,
2011, Paper ID JAI103156.
[15] Nagase, F., Otomo, T., Tanimoto, M., and Uetsuka, H., “Experiments on High Burnup Fuel
Behavior under LOCA Conditions at JAERI,” Light Water Reactor Fuel Performance, April
10–13, American Nuclear Society, Park City, UT, 2000.
[16] Ozawa, M., Takahashi, T., Homma, T., and Goto, K., “Behavior of Irradiated Zircaloy-4
Fuel Cladding under Simulated LOCA Conditions,” Zirconium in the Nuclear Industry:
12th International Symposium, ASTM STP 1354, Toronto, Canada, ASTM International,
[17] Nagase, F., “Recent Results on High Burnup fuel Behavior under LOCA Conditions,” Fuel
Safety Research Meeting, Oct 7, Kyoto, Japan, 2005.
[18] Große, M., “Neutron Radiography: A Powerful Tool for Fast, Quantitative and Non-
Destructive Determination of Hydrogen Concentration and Distribution in Zirconium
Alloys,” Zirconium in the Nuclear Industry: 16th International Symposium, ASTM STP
1529, Chengdu, Sichuan Province, China, ASTM International, West Conshohocken, PA,
2011.
[19] Barberis, P., Rebeyrolle, V., Vermoyal, J. J., Chabretou, V., and Vassault, J. P., “CASTA
DIVA: Experiments and Modeling of Oxide-Induced Deformation in Nuclear
Components,” J. ASTM Int., Vol. 5, No. 5, 2008, pp. 612–630.
[20] Busser, V., 2009, “Mécanismes d’endommagement de la couche d’oxyde des gaines de
crayons combustible en situation accidentelle de type RIA [Determination and analysis
of the damage mechanisms affecting the outer oxide layer using laboratory oxidized
samples],” Ph.D. thesis, Institut National des Sciences Appliquées, Villeurbanne, France
(in French).
DISCUSSION
Questions from Arthur Motta, Penn State University:—Have you verified your
model of initial reaction of the pre-existing oxide with the underlying metal by
heating a pre-existing oxide in vacuum to isolate that effect? If so, is the morphol-
ogy of the reaction similar to that which you observe?
Authors’ Response:—Chemical reduction of a pre-oxide formed at 500 C was

indeed observed after annealing in vacuum at 900 C for a long duration (100 min),
see [1]. Vacuum annealing was also performed on a 425 C pre-oxidized sample
heated at 1200 C during 200s: macroscopically, the sample color turned from light
brown to dark bronze. The pre-oxide thickness was decreased from 17.5 6 1.5 lm
to 4.3 6 0.9 lm and a stratified metallic region was observed underneath the oxide,
indicating that part of the pre-oxide had converted into metal.
[1] Characterization of Oxygen distribution in LOCA Situations, Duriez C.,

Guilbert S., Stern A., Grandjean C., Bělovský L., Desquines J., Journal of
ASTM International, vol. 8, No.2, 2010.
Questions from Valerie Vandenberghe, CEA, France
Q1:—Usually, such high levels of H pick up are linked to breakaway oxidation.

Do you have any sign of macroscopic breakaway oxidation on your samples
(spalled oxide, black and white oxide)?
Authors’ Response:—For the rather thick pre-oxide layers investigated in this

study, macroscopic post-test observation of the sample’s surface gives information
mainly on the pre-oxide and to a lesser extent on the high temperature oxide. Oxide
spallation was only observed for pre-oxide layers at 1200 C. Spallation of the high
temperature oxide layer was never observed, indicating that breakaway oxidation is
very unlikely.
Q2:—The waviness of the interface of your air preoxidized samples is higher

than for autoclaved or inreactor samples. Waviness of the interface is usually con-
sidered as a precursor of breakaway oxidation at high temperature. Don’t you think
that your air preoxidized samples would short cut part of the incubation time for
breakaway at high temperature, and so explain your very high H pick-up?
Authors’ Response:—The pre-oxide layer is first subjected to a preliminary

reduction, followed by significant high temperature oxidation and in some situa-
tions, a final breakaway period. This potential breakaway takes place on a high tem-
perature oxide layer having a rather straight oxide-metal interface. The pre-oxide
layer is most probably not affected by the above mentioned effect.
Questions from Itai Panas, Chalmers University of Technology:—Is it crucial to

explain your data that H2O (g) attacks the metal-oxide interface? What proof of
such a mechanism do you have? An alternative is decomposition of water into
hydroxides, which propagate as effective “H2O” by consecutive hydrolysis.
Authors’ Response:—There is no direct proof of the mechanism, but there is a

set of consistent conclusions. In some situations the outer oxide layer shows inter-
spersed nets of axially-radially and circumferentially-radially oriented cracks giving
a rectangular-shaped crack pattern. This crack pattern appears to be layered (see
Fig. 5(b) and (e), as an illustration) by a bright aspect showing that the oxide layer
is close to stoichiometry in the vicinity of these cracks, whereas, far from these
cracks the oxide layer is dark consistently with sub-stoichiometric oxide. This phe-
nomenon is consistent with localized flow of oxidation gas species along the cracks.
This oxidation is attributed to steam. It is considered that in such situation the gas
steam directly accesses beneath the remaining pre-oxide layer and oxidizes the high
temperature layer or directly the metal when there is no high temperature oxide
protecting the metal. The laboratory has also some experiments showing that with-
out sufficient gas flow in the furnace, steam starving conditions can occur, leading
to significant hydrogen pickup during high temperature oxidation for pre-oxide
layer thickness below 40 micrometer. The steam starvation beneath the remaining
pre-oxide layer and the high temperature oxide layer is expected to concentrate gas-
eous hydrogen that cannot be removed by the argon gas flow due to the shielding
effect of the pre-oxide layer.
Research, Fuel and reactor Materials Department, Hungary:—Have you observed
the development of thick a-layer on pre-oxidized samples?
Authors’ Response:—a-Zr(O) layers, for given oxidation temperature and dura-

tion, are thicker when the pre-oxide is fully protective, because in this case, the
metal/oxide interface does not move inwards.
Questions from Prerna Mishra, BARC Mumbai
Q1:—What was the minimum thickness of the oxide layer in the samples?
Authors’ Response:—Regarding the pre-oxide, the minimum thickness investi-

gated was 7.0 6 1.3 lm. In all cases where the pre-oxide is fully protective, no high
temperature oxide is observed under the pre-oxide.
Q2:—Have you carried out the oxygen concentration profile measurement in

the samples?
Authors’ Response:—Although not shown in the presentation, oxygen profiles

were systematically measured by EPMA on each cross-sectional metallography (see
the full length article which includes experimental oxygen profiles).
Question from Gargi Choudhuri, BARC Mumbai:—Can you comment on why

425 C oxide shows more waviness at the oxide-metal interface?
Authors’ Response:—Low-temperature interface waviness is thought to result from

a coupled mechanical-chemical interface instability, developing under large compres-
sion stresses applied to a bi-material (oxide and metal in the present study) interface.
In this situation, to minimize the total energy, the optimal compromise consists in
increasing the interface energy (more wavy interface) and decrease the mechanical
energy, and consequently the local stresses on average basis [2]. The oxidation process
alone (without hydriding) induces a compression stress of about 1200 to 1400 MPa in
the direction tangent to oxide-metal interface. However, hydrogen pick-up during the
pre-oxidation, induces a two component metal swelling (see [3]). One component is an
increase of the lattice parameter of the zirconium hcp crystal induced by dissolved
hydrogen and the second component is metal swelling induced by hydride precipitates
at larger concentrations. This hydrogen induced swelling of the metal reduces the tan-
gential stress at the oxide metal interface and consequently limits the waviness of the
oxide layer. This explanation was confirmed by direct comparison between air and
steam oxidations of Zircaloy-4 at 470 C (see [3]). The air-formed oxide was found to
be wavy and the steam formed oxide had a rather straight interface. The O2 formed
pre-oxide supporting this study was wavy due to the low hydrogen pickup during pre-
oxidation. In contrast, the autoclave pre-oxide interface was straight and associated
with higher hydrogen content.
[2] Thermodynamic approach of surface instability under irradiation, J. Colin,

M. Grinfeld and J. Grilhe, Acta Mater. 49 (2001) 3711–3718.
[3] The combined influence of hydrogen and oxidation on fuel cladding me-
chanical behaviour during simulated normal operation, B. Krebs et Al.,
TopFuel 2013, 2–6 Sept. 2012, Manchester, UK.
Question from K. Kapoor, NFC:—What is the reason for higher hydrogen pick-
up fraction for lower temperature (900 C) than at higher temperature (1100 C/
1200 C)?
Authors’ Response:—The hydrogen pickup fraction is expected to depend on both

oxidation temperature and duration. Because test durations are different, direct com-
parison of hydrogen pickup at the three test temperatures is not appropriate. However,
the hydrogen pickup at 1200 C was considered to be surprisingly low by the authors.
No explanation is yet available to understand this phenomenon.
Martin Steinbrück1 and Mirco Grosse2
Deviations From Parabolic

Kinetics During Oxidation
of Zirconium Alloys
Reference
Steinbrück, Martin and Grosse, Mirco, “Deviations From Parabolic Kinetics During Oxidation
of Zirconium Alloys,” Zirconium in the Nuclear Industry: 17th International Symposium, STP
1543, Robert Comstock and Pierre Barberis, Eds., pp. 979–1001, doi:10.1520/
ABSTRACT
The oxidation of zirconium cladding alloys used in nuclear reactors was
investigated under the conditions of loss-of-coolant and severe accidents, i.e., at
temperatures between 600 and 1600 C and in various atmospheres. The kinetics
were parabolic or sub-parabolic as long as the superficially formed oxide scale
remained intact and protective. More or less linear kinetics were found after
degradation of the oxide layer due to phase transitions connected with volume
changes and formation of cracks and pores. The presence of nitrogen further
accelerated oxidation rates by formation of zirconium nitride at the metal-oxide
boundary and its re-oxidation with the progressing reaction. This paper
summarizes extensive experimental work on high-temperature oxidation of
zirconium alloys performed at KIT focusing on effects causing non-parabolic
oxidation kinetics.
Keywords
zirconium alloy, oxidation, kinetics, nuclear safety research
Manuscript received January 25, 2013; accepted for publication January 19, 2014; published online
September 19, 2014.
1
Karlsruhe Institute of Technology, Institute for Applied Materials IAM-AWP, P. O. Box 3640, 76021 Karlsruhe,
Germany, e-mail: martin.steinbrueck@kit.edu
2
Karlsruhe Institute of Technology, Institute for Applied Materials IAM-AWP, P. O. Box 3640, 76021 Karlsruhe,
Germany.
3
Hyderabad, India.
Copyright V
Introduction
Zirconium alloys are used worldwide as cladding materials in nuclear reactors.
Their oxidation and reaction behavior at high temperatures, relevant for nuclear
design-basis and beyond design-basis accident scenarios, has been under investiga-
tion at Karlsruhe Institute of Technology for many years. Recently, these activities
have been intensified again in the frame of the research on advanced cladding
alloys, new LOCA (Loss Of Coolant Accident) embrittlement criteria, and air
ingress scenarios. Results of experiments in steam [1,2], oxygen [3], and nitrogen
containing atmospheres [4] were already published in detail elsewhere.
The oxidation of zirconium alloys causes degradation of mechanical properties
of claddings and thus may impair the barrier effect against the release of volatile fis-
sion products. During oxidation in steam, the most common atmosphere during
hypothetical accidents, hydrogen is formed. It is either released to the atmos-
phere—hence jeopardizing the containment—or absorbed by the remaining Zr
metal [5,6], and with this further weakening the mechanical properties of the clad-
ding. Furthermore, most oxidation reactions are exothermic, i.e., considerable
energy is released especially by the formation of zirconia. The chemical energy
released can be in the order of or even higher than the residual nuclear decay heat
and may have a strong impact on the course of nuclear accidents [7].
The kinetics of the oxidation reactions are determined by the condition of the
growing oxide scale and can be described, e.g., by Eq 1.
Dm
(1) ¼ km t n
S
where:
Dm ¼ the mass of oxygen absorbed by the specimen,
S ¼ the sample surface,
km ¼ the rate constant, and
t ¼ the oxidation time at constant temperature.
The exponent n equals 1=2 for the parabolic oxidation kinetics, applicable to a
protective oxide scale at temperatures above 1000 C. In this case, the oxidation rate is
determined by the diffusion of oxygen through the oxide scale. At lower temperatures,
cubic or sub-cubic oxidation kinetics (n ¼ 1/3 and n < 1/3, respectively) are observed,
as, e.g., described and explained by Evans [8]. When the oxide scale loses its protective
effect, oxygen may be transported much faster through pores and cracks to the metal
surface, thus accelerating the oxidation kinetics to linear (n ¼ 1) or even faster.
Nitrogen, as a component of air and, e.g., used for inertization of BWR (boiling
water reactor) containments and pressurization of emergency core cooling systems,
strongly affects reaction kinetics at high temperatures; it does not behave like an
inert gas under the conditions of a nuclear accident. Quite the contrary, it acceler-
ates oxidation by the formation of zirconium nitride with much higher density
compared to the oxide and its re-oxidation with progressing reaction [4].
STEINBRÜCK AND GROSSE, DOI 10.1520/STP154320130022 981
Despite the complex oxidation mechanisms, high-temperature oxidation of zir-

conium alloys in steam, oxygen, and air is usually described by parabolic kinetics,
and parabolic correlations and models are applied in almost all computer codes for
simulation of LOCA and severe accidents beyond LOCA.
This paper presents some highlights of the zirconium oxidation research at
KIT, compares the various atmospheres and discusses the mechanisms leading to
quite different oxidation kinetics. It will be shown that parabolic kinetics can only
be applied under certain conditions and that effects like breakaway oxidation and
the influence of nitrogen have to be taken into account for many scenarios.
Experimentals
TEST RIGS AND TEST CONDUCT
Most experiments presented in this paper were conducted in a commercial thermo-
gravimetric (TG) device (NETZSCH STA-409) with a vertical tube furnace and bal-
ance below the furnace, which was coupled to a quadrupole mass spectrometer
(NETZSCH Aeolos) by a capillary. The gases (Ar, O2, N2, air) were supplied via
Bronkhorst flow controllers in the lower part of the vertical tube furnace. Experi-
ments with steam containing atmospheres were conducted with a special furnace
coupled with a steam generator (NETZSCH). Argon flowed through the balance
containment into the furnace. The reaction gases were injected directly into the
reaction tube to prevent contamination of the balance and to ensure a well-defined
gas mixture in the furnace. All gases used were highly pure with nominal less than
1 or 10 vppm impurities, respectively. In this paper, TG results are presented unless
otherwise noted.
Some tests with steam at high temperatures were run in the so-called BOX rig,
a horizontal tube furnace coupled with a supply system for steam and gases at the
inlet and to a mass spectrometer (Balzers GAM300) at the outlet. Details of both
test setups are given in Ref. [9]. Experiments with final quench phase were con-
ducted in the inductively heated vertical QUENCH-SR (single rod) rig which allows
reflood of the specimens by a rising cylinder filled with water. The cooling rates
during quenching were in the range of 100 K/s. Details of this facility are described
in Ref. [10].
All isothermal experiments were conducted in flowing gas mixtures using argon
as reference gas for off-gas analyses by mass spectrometry. Inert atmosphere, i.e.,
pure argon, was applied during heatup and cooldown with heating rates between 10
and 30 K/min and cooling rates determined by the thermal capacity of the furnace.
Typical flow rates were in the range of 10 l/h during the experiments in the thermal
balance and between 100 and 200 l/h in the BOX and QUENCH-SR experiments.
SPECIMENS
Various alloys currently used in light water reactors (LWRs) were investigated. The
composition of the main alloys is summarized in Table 1. Two centimeter long tube
TABLE 1 Composition of the zirconium alloys investigated (main alloying elements in wt. %, rest
Zr). D4 is a ca. 100 lm external layer on Zircaloy-4 base tube applied in AREVA Duplex
claddings.
ZIRLOTM
R
V
Element Zircaloy-4 D4 M5 E110
Nb — — 1.0 1.0 1,0

Sn 1.5 0.5 1.0 0.01 —
Fe 0.2 0.5 0.11 0.05 0.008
Cr 0.1 0.2 < 0.01 0.015 0.002
segments (approximately 10 mm outer diameter and 0.5–0.7 mm wall thickness)

were cut from original cladding tubes, deburred, ground at both ends, and cleaned
in an ultrasonic bath of acetone. All claddings were supplied with standard surface
finish as final cladding products. After cutting into segments they were not further
surface treated (polished or etched), except for the cleaning procedure.
The samples were positioned vertically in the furnace of the TG and
QUENCH-SR apparatuses and horizontally in the BOX rig. The specimens were
open at the ends as a result of which, the tube segments were oxidized on both sides
(except for the single-rod QUENCH experiments with 15 cm long closed tubes
filled with ZrO2 pellets). The analysis of the mass gain results is based on the oxida-
tion of inner and outer surfaces excluding the base areas. This procedure may have
led to inaccuracies in case of differing oxidation inside and outside.
POST-TEST EXAMINATIONS
Macro photos were taken of all specimens after the tests. Then, the specimens were
embedded in epoxy resin, cut, ground, and polished for metallographic examina-
tions by optical microscopy. Some specimens were investigated by SEM/EDX and
microprobe.
Oxidation of Zirconium Alloys in Steam

INTACT OXIDE SCALE
Steam is the most common and prototypical atmosphere expected to be available
during nuclear accident scenarios. The oxidation of zirconium alloys in steam is
described by parabolic correlations in all computer codes simulating nuclear acci-
dents. Parabolic kinetics (n ¼ 0.5 in Eq 1) are found at temperatures above 1000 C
and as long as the superficial oxide scale is protective. Hence the reaction is deter-
mined by the diffusion of oxygen (vacancies) through the growing oxide layer. Fur-
thermore, parabolic kinetics could be only applied strictly for limited oxidation times
as long as the metal phase is available and no steam starvation takes place [11].
Typical examples of cross sections through oxidized cladding walls with protec-
tive oxide scales are given in Fig. 1.
FIG. 1 Micrographs of Zircaloy-4 claddings oxidized in steam and quenched in water in

the QUENCH-SR rig.
The left micrograph shows the characteristic layer structure with external ZrO2
scale, oxygen-stabilized a-Zr(O), and formerly b-Zr. According to the phase dia-
gram Zr-O [12] the hexagonal a-Zr(O) phase can dissolve up to 30 at. % (7 wt. %)
oxygen. Complete consumption of the cubic b-Zr phase is shown in the right
micrograph after oxidation at 1600 C. Furthermore, metallic precipitates are seen
in the oxide layer which are formed during cooling of the sub-stoichiometric cubic
ZrO2 by its decomposition into the (less sub-stoichiometric) tetragonal oxide and
a-Zr(O). Such precipitations are a clear indication for temperatures beyond
1500 C. Figure 1 also reveals the brittle character of both oxide and a-phase. The
only ductile layer is b-Zr with lower oxygen concentration which keeps the me-
chanical stability of the cladding during reflood and quenching.
At temperatures below 1000 C, cubic or sub-cubic oxidation kinetics (n ¼ 1/3
and n < 1/3, respectively) are observed [8]. As an example, Fig. 2 presents the mass
gain during oxidation of various zirconium alloys at 700 C in a log-log diagram. In
the initial part of all curves the kinetics is rather cubic than parabolic. All alloys
except M5 experience a transition to more or less linear kinetics (n ¼ 1 in Eq 1).
The mechanism for this transition to the so-called breakaway oxidation is described
in the next section; critical times and oxide scale thicknesses are given in Table 2.
The deviation from the parabolic to cubic or even slower kinetics has been dis-
cussed by various authors [8,13–15]. Evans [8] discusses the buildup of stresses in
the oxide scale and their relief with growing oxide scale which affect the concentra-
tion of vacancies and with this the oxygen diffusion through the oxide, leading to
slower than parabolic kinetics.
BREAKAWAY OXIDATION
Breakaway oxidation results in loss of the protective properties of the oxide scale
due to its mechanical failure. It is associated to the phase change from the initially
formed meta-stable tetragonal phase, stabilized by lattice defects, small grain sizes
FIG. 2 Mass gain during oxidation of zirconium alloys in steam at 700 C. The slope of
the auxiliary lines corresponds to the exponent n in Eq 1. The units for Dm/S and
Time are g/cm2 and s, respectively.
and compressive stresses, to the thermodynamically stable monoclinic oxide phase

[16–23]. During progressing oxidation the effects stabilizing the tetragonal phase
decline, i.e., the grains grow, the compressive stresses are relieved during oxide
growth at the convex surface and the oxygen content increases to more stoichio-
metric composition and hence cause the transition to the monoclinic phase at a cer-
tain distance from the metal-oxide interface. This transformation goes along with a
5 % volume increase resulting in formation of cracks and oxide spalling. Two exam-
ples of breakaway oxide scales are given in Fig. 3. The critical times and oxide thick-
nesses for the transition to breakaway oxidation are strongly depending on
TABLE 2 Transition times in hours from protecting to non-protecting oxide scales (breakaway) and
calculated oxide scale thickness at transition in lm (italic) during oxidation of zirconium
alloys in steam.
T, C ZIRLOTM
R
V
Zircaloy-4 Duplex-D4 M5 E110
a
600 8.2 (7) 7.7 (3) 6.3 (8) nt nt
700 2.2 (9) 2.2 (9) 1.4 (9) nt 13 (21)
800 7.4 (41) 7.3 (35) 5.2 (32) nt 0.9 (12)
900 1.3 (40) 1.4 (38) 2.1 (40) nt 0.8 (24)
1000 0.59 (89) 0.93 (105) 0.58 (76) nt 0.64 (48)
1100 nt nt nt nt nt
a
nt ¼ no transition.
FIG. 3 Typical oxide scale morphology after breakaway oxidation of zirconium alloys.
temperature and vary with the type of alloy and experimental conditions which is
obvious, e.g., from Figs. 2 and 4 as well as Table 2.
In most cases, breakaway occurs after hours and should be of limited impor-
tance for fast reactor transients. Even the shortest transition times at around
1000 C amount to more than half an hour. However, breakaway has to be taken
into account during slow transients, e.g., during accidents in spent fuel pools or
cleaning tanks. Generally, the ZrO2 oxide scale thickness at transition increases
with temperature due to the rising ductility of the involved phases, i.e., of the subja-
cent metal, but probably also of the oxide itself. Thus, it is known that creep and
R
plasticity of ceramics increase above 0.5 Tmelt. M5V shows (almost) no breakaway
oxidation at all; only slight deviations from ideal oxidation kinetics were observed
in these experiments as can be seen in Fig. 4.
FIG. 4 Mass gain during oxidation of Zircaloy-4 (left diagram) and (R) (right diagram)
in steam in dependence on temperature. The slope of the curves and auxiliary
corresponds to the exponent n in Eq 1. The units for Dm/S and Time are g/cm2
and s, respectively.
The late transitions of E110 at 700 C and of the tin-bearing alloys at 800 C
(Table 2) are remarkable and may be correlated to the properties of the metal
phase. It is known that Zr alloys have a maximum ductility at the beginning of the
a–b transition which is for Zircaloy-4 at around 850 C and for niobium-bearing
alloys at slightly lower temperatures [24]. Oxygen, as an a-phase stabilizer, and
hydrogen, as an b-phase-stabilizing element, influence this transition temperature;
however, there is probably a correlation between the high ductility of the metal phase
and the delay in transition to accelerated oxidation which is dependent on alloy
composition.
HYDROGEN ABSORPTION
The hydrogen produced by oxidation of zirconium in steam may be released to the
atmosphere or absorbed by the remaining metal. According to the phase diagram
Zr-H [25], the b-Zr phase is stabilized by hydrogen and can absorb up to more
than 50 at. % (10 000 wppm) hydrogen. The terminal hydrogen solubility of zirco-
nium is proportional to the square root of the hydrogen partial pressure as
described by Sieverts’ law [26]. Recent results obtained by in-situ neutron radiogra-
phy [27] show that hydrogen absorption may be very fast immediately after initia-
tion of steam injection when no or only very thin oxide scale is existing and when
the oxidation kinetics are the highest. As the reaction progresses the hydrogen con-
tent in the metal slowly decreases due to the decreasing hydrogen partial pressure
in the surrounding atmosphere. However, significant amounts of hydrogen are
absorbed (>40 at. %, 7000 wppm) with the transition to breakaway due to the
“hydrogen pump” effect [5,28].
The close correlation between oxide morphology and hydrogen absorption,
discussed in Ref. [5], is clearly confirmed by Fig. 5. Local enrichment of hydrogen
in pores and cracks of the oxide layer leads to significantly higher hydrogen par-
tial pressures at the metal/oxide interface than in the gas atmosphere and, there-
fore, according to Sieverts’ law, to higher hydrogen concentrations in the metal
phase. The hydrogen contents in the remaining metal phase were determined
post-test by neutron radiography [29]. High concentrations of hydrogen in the
metal have been found in specimens with pronounced breakaway oxide scales,
with the highest values for specimens with early transition times to breakaway
R
(compare also with data of Table 2). The hydrogen concentration in M5V alloy is
monotonically increasing with temperature, indicating that here only the hydro-
gen partial pressure in the gas bulk played a role and confirming the absence of
cracks parallel to the metal-oxide interface seen in the metallographic
examination.
COMPARISON BETWEEN CURRENTLY USED ALLOYS

In the previous sections, it was already shown that the high-temperature oxidation
behavior of the various cladding alloys is quite different especially with respect to
their susceptibility to breakaway. Figure 6 proves this with mass gain diagrams at
FIG. 5 Hydrogen absorbed during oxidation of zirconium alloys in steam in

dependence on temperature and type of alloy. The micrographs illustrate the
correlation between hydrogen uptake and oxide scale morphology. Annealing
times were 3 h at 1100 and 1000 C, 24 h at 900, 800, and 700 C and 48 h at
600 C.
selected temperatures. Differences in oxidation rates up to 800 % were observed at

temperatures up to 1000 C, i.e., within the breakaway region. Smaller, but still sig-
nificant variations (about 30 %) were observed at higher temperatures [1]. The
rapid initial mass gain for the 1100 C samples is caused by a temporary tempera-
ture increase after switching on the steam flow due to chemical heat release. Gener-
R
ally, the advanced cladding alloys, especially the niobium bearing alloy M5V, exhibit
a more favorable behavior also at high temperatures typical of reactor accident
scenarios.
At some temperatures, a significant difference between the two Zr-Nb alloys
RV
M5 and E110 was observed which is caused by different fabrication procedures
resulting in different impurities [30]. Meanwhile, an advanced E110 alloy has been
developed by Chepetsky Mechanical Plant in Russia based on zirconium sponge
R
(E110G) exhibiting similar properties as M5V [31]. However, this material was not
yet available for own measurements.
Oxidation of Zirconium Alloys in Oxygen

Oxygen is sometimes used as a substitute for steam in high-temperature simulation
experiments because it is easier to handle. There are two obvious differences
FIG. 6 Mass gain curves of zirconium alloys during oxidation in steam at 700 and
1100 C.
between the oxidation in steam and oxygen: (1) the heat of reaction in oxygen is
about twice as much as for steam and (2) during oxidation in oxygen no hydrogen
is produced and absorbed by the metal. However, use of oxygen as simulant is justi-
fied because the oxidation kinetics and resulting oxide scales are quite similar for
both gases as can be seen in Fig. 7. There is no significant variation of the oxidation
rate constants obtained for the pre-transition phase in oxygen and water steam
within the scatter band of the two alloys selected in this diagram (the rate constants
have been calculated by linear fitting of the mass gain versus square root of time
curves, excluding the very initial part after switching on the oxidizing atmosphere
which resulted in a temporary temperature increase). Slightly different activation
energies (slopes in Fig. 7) were observed for temperatures above and below
FIG. 7 Parabolic rate constants (if applicable for the pre-transition phase) for the
R
oxidation of Zircaloy-4 and M5V in steam, oxygen, and air, respectively. For
comparison the correlation by Cathcart and Pawel [32] is shown.
approximately 1050 C due to the phase transition from monoclinic to tetragonal

zirconia. The results obtained in this study by thermogravimetry well correlate with
literature data, as is seen by comparison with the correlation by Cathcart and Pawel
[32]. These data were proposed as best fitted correlation for temperatures below
1800 K in the state-of-art paper by Schanz [11]. Furthermore, the TG data are very
similar to results obtained earlier by one of the authors in the BOX test rig [2].
The transition from protective to non-protective oxide scale behavior is also
observed in oxygen with comparable critical times and oxide scales, see Fig. 8 and
Table 3. Remarkable differences between steam and oxygen were observed at 900 C,
which may be related to the stabilizing effect of the b-Zr phase by absorbed hydro-
gen and the corresponding shift of the a to b phase transition in the metal. It should
be mentioned that transition times depend on the experimental setup (sample size,
sample orientation, temperature and gas flow gradients, etc.). Therefore, the values
given in Tables 2 and 3 can be used for comparison of the behavior of the different
alloys and may not be identical for other experimental setups and reactor
conditions.
Effect of Nitrogen
Nitrogen is the main component of air; furthermore, it is used, e.g., for inertization
of boiling water reactor (BWR) containments and for pressurization of emergency
core cooling water systems. Hence, there is a certain probability that nitrogen
comes into contact with the cladding during severe accidents.
R
FIG. 8 Mass gain during oxidation of Zircaloy-4 (left diagram) and M5V (right diagram)
in oxygen in dependence on temperature. The slope of the curves and auxiliary
lines corresponds to the exponent n in Eq 1. The units for Dm/S and Time are
g/cm2 and s, respectively.
TABLE 3 Transition times in hours from protecting to non-protecting oxide scales (breakaway) and
calculated oxide scale thickness at transition in lm (italic) during oxidation of zirconium
alloys in oxygen.
T, C
R
V
Zircaloy-4 Duplex-D4 M5 E110
600 10.6 (6) 10.8 (6) naa na

700 2.9 (9) 5.0 (11) 12.2 (29) 4.4 (14)
800 6.9 (33) 11.1 (39) na na
900 7.8 (50) 11.9 (55) na 0.4 (19)
1000 1.1 (69) 1.2 (74) nt 0.3 (30)
1100 ntb nt nt nt
a
na ¼ not analyzable.
b
nt ¼ no transition.
REACTION OF ZIRCONIUM ALLOYS IN PURE NITROGEN

Nitrogen reacts very slowly with as-received zirconium (alloys) due to the forma-
tion of a very thin protective zirconium nitride scale and, compared to oxygen, by
2–3 orders of magnitude lower diffusion coefficient of nitrogen in a-Zr [33,34]. The
reaction kinetics are of parabolic type, but two orders of magnitude slower in com-
parison with oxygen. However, if the metal contains oxygen the reaction of the
oxygen-stabilized a-Zr(O) is much faster and leads to the formation of a homoge-
neously distributed two-phase mixture ZrO2-ZrN over the cladding wall after short
times [35]. Figure 9 illustrates this behavior. At 1200 C, saturation of the nitrogen
FIG. 9 Mass gain during reaction of Zircaloy-4 with and w/o oxygen dissolved with
nitrogen at 1200 C and micrographs of corresponding specimens.
uptake by a-Zr(O) is already observed within 1000–2000 s. Obviously, no protective

nitride scale is formed under these conditions.
The reaction of the oxygen-stabilized a-Zr(O) with nitrogen is the key reaction
during the oxidation of zirconium alloys in air as it is explained in the next section.
OXIDATION OF ZIRCONIUM ALLOYS IN AIR

Air oxidation of zirconium alloys is of special interest because it leads to (1) signifi-
cant heat release causing temperature runaway from lower temperatures compared
to steam, (2) strong degradation of the cladding causing early loss-of-barrier effect,
and (3) high oxygen activity in the core influencing chemistry and transport of fis-
sion products [4,36]. Here, only the second point will be discussed.
During reaction of zirconium (alloys) in air, the reaction with oxygen is pre-
ferred compared to nitrogen due to the higher enthalpy of reaction and first leads
to the formation of zirconium oxide. Air may diffuse through imperfections in this
initially formed oxide layer produced by mechanical failure or oxygen starvation,
and at the metal-oxide interface, oxygen is consumed first. Only after wide con-
sumption of oxygen resulting in a very low oxygen partial pressure, does the
remaining nitrogen react with the oxygen saturated zirconium available at the
metal-oxide interface and form the golden-colored ZrN. This nitride moves out-
wards relative to metal-oxide interface and is re-oxidized with proceeding reaction
when the oxygen partial pressure exceeds a critical value. This reaction is associated
with release of nitrogen as well as with a volume increase by 48 %. Hence, a very
FIG. 10 Micrograph after oxidation of Zircaloy-4 at 1000 C in air: (1) initially formed
dense ZrO2, (2) porous oxide after oxidation of ZrN, (3) ZrO2-ZrN mixture, (4)
a-Zr(O).
porous oxide is formed which is not protective at all and linear or even accelerating
kinetics are established. Figure 10 provides a typical micrograph of an air-oxidized
Zircaloy-4 showing the layers of initially formed dense ZrO2, porous oxide after ox-
idation of ZrN, ZrO2-ZrN mixture, and remaining a-Zr(O).
The formation of such strongly degraded regions usually starts locally and then
spreads out to finally form layer-like structures. Figure 11 demonstrates the strongly
enhanced degradation of Zircaloy-4 cladding by oxidation in air in comparison
with steam due to the mechanisms described above. Nevertheless, the oxidation
rates in air before the transition are comparable to the ones in oxygen and steam as
is shown in Fig. 7. Only at temperatures above 1100 C the pre-transition kinetics in
air tend to be higher than in oxygen and steam. Of course, the oxygen supply is lim-
ited in air in comparison to pure oxygen, which may cause oxygen starvation condi-
tions earlier than in oxygen, especially during the initial oxidation period at high
temperatures (see Fig. 13).
OXIDATION OF ZIRCONIUM ALLOYS IN PROTOTYPIC ATMOSPHERES

During a nuclear accident, pure air or nitrogen will never have contact with as-
received zirconium alloys. Prototypically, (1) the cladding will be first oxidized in
steam and then react with air or nitrogen and (2) atmosphere will most likely not
be dry, i.e., mixtures between steam and air/nitrogen will be available.
The effect of pre-oxidation in oxygen and steam was described in Ref. [4]: an
intact oxide scale, even a thin one, formed before air ingress effectively protects the
cladding from nitrogen attack and the associated degradation. During subsequent
annealing in air, the scale continues growing as before in oxygen as long as the
FIG. 11 Zircaloy-4 cladding segments after 1 h oxidation at 1200 C in steam and air,
respectively, showing enhanced degradation and loss-of-barrier effect by
oxidation in air.
FIG. 12 Zircaloy-4 cladding cross sections after 1 h oxidation at 1000 C in steam (left)
and in a 50/50 steam-nitrogen mixture (right).
access of nitrogen to the metal is blocked and as long as oxygen/steam is available.

Global oxygen starvation, i.e., the absence of oxygen or other oxidizing gases in the
atmosphere, changes the situation significantly. In experiments with pre-oxidized
zircaloy specimens in nitrogen, large amounts of nitrides were formed externally in
the oxide scale independently of the degree of pre-oxidation. Stuckert [37] reports
FIG. 13 Reaction of Zircaloy-4 in various atmospheres at 1200 C: Mass gain curves and
related micrographs (PO in the legend means pre-oxidation in oxygen. Nitrogen
after PO started at time 0 s; air after PO started at 150 s, see arrow).
that oxygen diffusion from the oxide to the metal occurs under steam starvation
conditions, leading to the formation of hypostoichiometric oxide, a-Zr(O) precipi-
tations in the oxide, and even external a-Zr(O) layers on the oxide. Although the
oxide is thermodynamically much more stable than the nitride, nitrogen may react
with hypostoichiometric oxide and the a-Zr(O) phase, as was shown above.
A considerable nitrogen attack, i.e., extremely porous oxide layers with nitride pre-
cipitations, was also observed in air-steam and nitrogen-steam mixtures of a large com-
position range [38]. Degradation increased with increasing temperature (at least up to
1300 C) and an increasing content of air in the mixture. The maximum degradation of
the Zircaloy-4 cladding in nitrogen-steam was found for 50/50 mixtures. Figure 12 com-
pares micrographs after oxidation in pure steam and in a 50/50 steam-nitrogen mix-
ture. Even though breakaway oxidation has already started most of the sample is still
metallic after 1 h oxidation at 1000 C in steam as can be seen in the left picture. On the
other hand, addition of nitrogen caused complete oxidation of the sample under identi-
cal boundary conditions. The oxidation of about ten times more metal of course also
caused the production of ten times more hydrogen during the same duration.

This paper revealed the complex oxidation behavior of zirconium alloys used for
nuclear fuel cladding depending on temperature, atmospheres, and type of alloy
which cannot be always described by parabolic kinetics. Figure 13 gives a kind of
summary of the results. On the one hand, dense and protective oxide/nitride layers
are connected with (sub-)parabolic reaction kinetics determined by the diffusion of
oxygen/nitrogen through growing scales. On the other hand, porous superficial
layers produced by phase transitions connected with changes in density lead to
more or less linear kinetics defined by gaseous diffusion of the oxidants to the metal
through cracks and pores. Even strongly accelerating global kinetics are found
when the nitrogen attack starts locally and spreads out over the cladding surface.
The reaction of zirconium alloys with nitrogen is very slow (invisible in the diagram
of Fig. 13). Pre-oxidized zirconium alloys react ten times faster with nitrogen, and pre-
oxidized and homogenized cladding, i.e., the oxygen stabilized a-Zr(O), reacts almost
two orders of magnitude faster. Generally, zirconium nitride is formed when oxygen is
absent in the gas phase and, at the same time, present in the metal phase. These condi-
tions are locally established at the metal-oxide interface or can be globally existent at oxy-
gen starvation conditions. The significantly different densities of the involved phases (Zr:
6.5 g/cm3, ZrO2: 5.6 g/cm3, ZrN: 7.1 g/cm3) result in the formation of very porous layers.
The oxidation in oxygen and steam is of parabolic kinetics at temperatures
beyond 1050 C and at lower temperatures initially sub-parabolic before the transition
to breakaway. A mixture of steam or oxygen with nitrogen accelerates degradation
due to the formation and re-oxidation of zirconium nitride resulting in porous and
non-protecting oxide/nitride. Porous oxide and steam in the atmosphere results in
strong uptake of hydrogen by the remaining metal (not to be seen in Fig. 13).
Generally, (sub-)parabolic oxidation kinetics can be applied for fast transients,

like LOCA scenarios, when the breakaway temperature region is passed rapidly
before transition to linear kinetics and at temperatures above 1050 C as long as no
starvation of oxygen/steam occurs and until significant metal consumption. Break-
away oxidation connected with rather linear kinetics has to be taken into account
for long-term scenarios at intermediate temperatures.
Furthermore, the paper has clearly shown that nitrogen, which is used for iner-
tization of BWR containments, does not behave like an inert gas at the conditions
of a severe nuclear accident, but rather, it acts like a catalyst accelerating oxidation.
ACKNOWLEDGMENTS
This work was sponsored by the HGF Program NUKLEAR at the Karlsruhe Institute of
Technology and partially done within the framework of the EC supported SARNET-2
Program (contract FP7-231747). The writers are very grateful to P. Severloh and U.
Stegmaier (KIT) for sample preparation and microscopic examinations, and to N. Vér,
C. Vorpahl, M. Böttcher, M. Jung, and S. Prestel for performing a number of TG tests
mainly in the framework of seminar papers. The Zr1Nb (E110) was provided by Rus-
R
sian institutions in the context of the EU program ISTC 1648.2. M5V and Duplex-D4
rod cladding were supplied by AREVA, ZIRLO by Westinghouse.
References
[1] Steinbrück, M., Vér, N., and Grosse, M., “Oxidation of Advanced Zirconium Cladding Alloys
in Steam at Temperatures in the Range of 600-1200 C,” Oxid. Metals, Vol. 76, Nos. 3–4,
2011, pp. 215–232.
[2] Grosse, M., “Comparison of the High-Temperature Steam Oxidation Kinetics of Advanced
Cladding Materials,” Nucl. Trans., Vol. 170, 2010, pp. 272–279.
[3] Steinbrück, M., “Oxidation of Zirconium Alloys in Oxygen at High Temperatures up to

1600 C,” Oxid. Metals, Vol. 70, Nos. 5–6, 2008, pp. 317–329.
[4] Steinbrück, M., “Prototypical Experiments Relating to Air Oxidation of Zircaloy-4 at High
Temperatures,” J. Nucl. Mater., Vol. 392, No. 3, 2009, pp. 531–544.
[5] Große, M., Steinbrück, M., Lehmann, E., and Vontobel, P., “Kinetics of Hydrogen Absorp-
tion and Release in Zirconium Alloys During Steam Oxidation,” Oxid. Metals, Vol. 70, Nos.
3–4, 2008, pp. 149–162.
[6] Große, M., Lehmann, E., Steinbrück, M., Kühne, G., and Stuckert, J., “Influence of Oxide Layer
Morphology on Hydrogen Concentration in Tin and Niobium Containing Zirconium Alloys Af-
ter High Temperature Steam Oxidation,” J. Nucl. Mater., Vol. 385, No. 2, 2009, pp. 339–345.
[7] Steinbrück, M., Große, M., Sepold, L., and Stuckert, J., “Synopsis and Outcome of the
QUENCH Experimental Program,” Nucl. Eng. Des., Vol. 240, No. 7, 2010, pp. 1714–1727.
[8] Evans, H. E., Norfolk, D. J., and Swan, T., “Perturbation of Parabolic Kinetics Resulting
from the Accumulation of Stress in Protective Oxide Layers,” J. Electrochem. Soc., Vol.
125, No. 7, 1978, pp. 1180–1185.
[9] Steinbrück, M., Stegmaier, U., and Ziegler, T., “Prototypical Experiments on Air Oxidation
of Zircaloy-4 at High Temperatures,” Report FZKA 7257, Forschungszentrum Karlsruhe,
Germany, 2007.
[10] Hofmann, P., Miassoedov, A., Steinbock, L., Steinbrück, M., Berdyshev, A. V., Boldyrev, A.
V., Palagin, A. V., Shestak, V. E., and Veshchunov, M. S., “Quench Behavior of Zircaloy Fuel
Cladding Tubes. Small-Scale Experiments and Modeling of the Quench Phenomena,”
Report FZKA 6208, Forschungszentrum Karlsruhe, 1999.
[11] Schanz, G., Adroguer, B., and Volchek, A., “Advanced Treatment of Zircaloy Cladding High-
Temperature Oxidation in Sever Accident Code Calculations. Part I. Experimental Data Base
and Basic Modeling,” Nucl. Eng. Des., Vol. 232, No. 1, 2004, pp. 75–84.
[12] Abriata, J. P., Garcés, J., and Versaci, R., “The Zr-O (Zirconium-Oxygen) System,” Bull.
Alloy Phase Diag., Vol. 7, No. 2, 1986, pp. 116–124.
[13] Porte, H. A., Schnizlein, J. G., Vogel, R. C., and Fischer, D. F., “Oxidation of Zirconium and
Zirconium Alloys,” J. Electrochem. Soc., Vol. 107, No. 6, 1960, pp. 506–515.
[14] Dawson, J. K., Long, G., Seddon, W. E., and White, J. F., “The Kinetics and Mechanism of the
Oxidation of Zircaloy-2 at 350–500 C,” J. Nucl. Mater., Vol. 25, No. 2, 1968, pp. 179–200.
[15] Arima, T., Moriyama, K., Gaja, N., Furuya, H., Idemitsu, K., and Inagaki, Y., “Oxidation
Kinetics of Zircaloy-2 Between 450 C and 600 C in Oxidizing Atmosphere,” J. Nucl.
Mater., Vol. 257, No. 1, 1998, pp. 67–77.
[16] Baek, H. J. and Jeong, Y. H., “Breakaway Phenomenon of Zr-Based Alloys During a High-
Temperature Oxidation,” J. Nucl. Mater., Vol. 372, Nos. 2–3, 2008, pp. 152–159.
[17] Leistikow, S. and Schanz, G., 1985, “The Oxidation Behavior of Zircaloy-4 in Steam
Between 600 and 1600 C,” Werkstoffe Korrosion, Vol. 36, No. 3, 1985, pp. 105–116.
[18] Tupin, M., Pijolat, M., Valdivieso, F., Soustelle, M., Frichet, A., and Barberis, P., “Differences in
Reactivity of Oxide Growth During the Oxidation of Zircaloy-4 in Water Vapour Before and
After the Kinetic Transition,” J. Nucl. Mater., Vol. 317, Nos. 2–3, 2003, pp. 130–144.
[19] Park, D. J., Park, J. Y., Jeong, Y. H., and Lee, J. Y., “Microstructural Characterization of ZrO2
Layers Formed During the Transition to Breakaway Oxidation,” J. Nucl. Mater., Vol. 399,
Nos. 2–3, 2010, pp. 208–211.
[20] Yan, Y., Burtseva, T. A., and Billone, M. C., “High-Temperature Steam-Oxidation Behavior
of Zr-1Nb Cladding Alloys,” J. Nucl. Mater., Vol. 393, No. 3, 2009, pp. 433–448.
[21] Idarraga, I., Mermoux, M., Duriez, C., Crisci, A., and Mardon, J. P., “Raman Investigations of
Pre- and Postbreakaway Oxide Scales Formed on Zircaloy-4 and M5 in Air at High Tem-
perature,” J. Nucl. Mater., Vol. 421, Nos. 1–3, 2012, pp. 160–171.
[22] Polatidis, E., Frankel, P., Wei, J., Klaus, M., Comstock, R. J., Ambard, A., Lyon, S., Cottis, R.
A., and Preuss, M., “Residual Stresses and Tetragonal Phase Fraction Characterisation of
Corrosion Tested Zircaloy-4 Using Energy Dispersive Synchrotron X-Ray Diffraction,” J.
[23] Pétigny, N., Barberis, P., Lemaignan, C., Valot, C., and Lallemant, M., “In Situ XRD Analysis
of the Oxide Layers Formed by Oxidation at 743 K on Zircaloy 4 and Zr–1NbO,” J. Nucl.
Mater., Vol. 280, No. 3, 2000, pp. 318–330.
[24] Forgeron, T., Brachet, J. C., Barcelo, F., Castaing, A., Hivroz, J., Mardon, J. P., and Bernaudat,
C., “Experiment and Modeling of Advanced Fuel Rod Cladding Behavior Under LOCA Con-
ditions: Alpha-Beta Phase Transition Kinetics and EDGAR Methodology,” Zirconium in the
Nuclear Industry: Twelfth International Symposium, G.P. Sabol and G.D. Moan, Eds., ASTM
[25] Zusek, E., Abriata, J. P., and San-Martin, A., “The H-Zr (Hydrogen-Zirconium) System,”
Bull. Alloy Phase Diagr., Vol. 11, No. 4, 1990, pp. 385–395.
[26] Steinbrück, M., “Hydrogen Absorption by Zirconium Alloys at High Temperatures,” J.

Nucl. Mater., Vol. 334, No. 1, 2004, pp. 58–64.
[27] Grosse, M., “Neutron Radiography: A Powerful Tool for Fast, Quantitative and Non-
Destructive Determination of the Hydrogen Concentration and Distribution in Zirconium
Alloys,” J. ASTM Int., Vol. 8, 2011, JAI 103251.
[28] Grosse, M., “High Temperature Oxidation in Nuclear Reactor Systems,” Nuclear Corrosion Sci-
ence and Engineering, D. Feron, Ed., Woodhead Publishing Limited, Cambridge, UK, 2012.
[29] Grosse, M., Lehmann, E., Vontobel, P., and Steinbrueck, M., “Quantitative Determination
of Absorbed Hydrogen in Oxidized Zircaloy by Means of Neutron Radiography,” Nucl.
Instrum. Methods Phys. A, Vol. 566, No. 2, 2006, pp. 739–745.
[30] Chung, H. M., “Fuel Behavior Under Loss-of-Coolant-Accident Situations,” Nucl. Eng.
Des., Vol. 37, 2005, pp. 327–362.
[31] Perez-Feró, E., “High-Temperature Behaviour of E110G Cladding,” Proceedings of the 17th
International QUENCH Workshop, Karlsruhe, Germany, Nov 22–24, 2011.
[32] Pawel, R. E., Cathcart, J. V., and McKee, R. A., “The Kinetics of Oxidation of Zircaloy-
4 in Steam at High Temperatures,” J. Electrochem. Soc., Vol. 126, No. 7, 1979, pp.
1105–1111.
[33] Anttila, A., Räisänen, J., and Keinonen, J., “Diffusion of Nitrogen in a-Zr and a-Hf,” J. Less
[34] Perkins, R. A., “The Diffusion of Oxygen on Oxygen Stabilized a-Zirconium and Zircaloy-
4,” J. Nucl. Mater., Vol. 73, No. 1, 1978, pp. 20–29.
[35] Steinbrück, M. and Jung, M., “High-Temperature Reaction of a-Zr(O) with Nitrogen,” Pro-
ceedings of the International Congress on Advances in Nuclear Power Plants ICAPP2011,
Nice, France, May 2–5, 2011, pp. 948–954.
[36] Steinbrück, M. and Böttcher, M., “Air Oxidation of Zircaloy4, M5 and ZIRLO Clad-
ding Alloys at High Temperatures,” J. Nucl. Mater., Vol. 414, No. 2, 2011, pp.
276–285.
[37] Stuckert, J. and Veshchunov, M. V., “Behaviour of Oxide Layer of Zirconium-Based Fuel
Rod Cladding Under Steam Starvation Conditions,” Report FZKA 7373, Forschungszen-
trum Karlsruhe, Germany, 2008.
[38] Steinbrück, M. and Ver, N., “High-Temperature Oxidation of Zircaloy-4 in Mixed Steam-
Air and Steam-Nitrogen Atmospheres,” Proceedings of the International Congress on
Advances in Nuclear Power Plants ICAPP2010, San Diego, CA, June 13–17, 2010, pp.
1051–1061.
DISCUSSION
Questions from Gargi Choudhuri, BARC Mumbai:—Why does M5 alloy show
no transition compared to E110 and does Fe have any effect on this? What will be
the corresponding H-intake behavior of the two alloys?
Authors’ Response:—Although the composition of the two Zr1Nb alloys is

very similar, their oxidation behavior especially with respect to breakaway and
hydrogen uptake is very different. One reason for these differences between the
alloys seems to be a different fabrication procedure resulting in different impur-
ities (Chung, 2005 [30]). The zircon ore decomposition, hafnium purification
processes, and electrolytic zirconium reduction for E110 raw material are based
on fluorine chemistry. The presence of fluorine is known to be deleterious, i.e.
increasing the susceptibility to breakaway oxidation. On the other hand, M5V R
fabrication is based on processes leading to Ca, Mg, and Al impurities which are
beneficial for the formation of oxide scales less susceptible to breakaway (Chung,
2005 [30]). Surface roughness can be another reason for the formation of layered
oxide scales.
As shown in Fig. 7 of the paper, the influence of the oxide scale morphology on
the hydrogen uptake is significant. Local enrichment of hydrogen in cracks and
pores near the oxide-metal interface causes higher uptake of hydrogen after transi-
tion to breakaway, see also ref. [6].
The advanced E110G alloy based on sponge zirconium (like M5V R ) shows
better properties compared to the classical one especially with respect to

breakaway.
Questions from Marc Tupin, CEA Saclay
Q1:—Which reaction does nitrogen catalyze exactly?
Nitride process
Oxidation process
Authors’ Response:—Catalysts accelerate chemical reaction without being con-

sumed. The authors call the effect of nitrogen "catalytic" because nitrogen acceler-
ates the oxidation kinetics via temporary formation of ZrN. At the end of the
oxidation in air, ZrN is completely converted into oxide. Hence, it is not a catalytic
reaction in the strict chemical sense.
Q2:—Which step of the corrosion mechanism does nitrogen impact?
Authors’ Response:—The mechanism of air oxidation including the formation

of very porous oxides scales via the formation and oxidation of the ZrN phase is
explained in the paper. The formation of a large amount of cracks and pores in the
oxide layer accelerates the transport of oxygen and nitrogen to the oxide/nitride –
metal interface where the reaction takes place.
Questions from Robert J. Comstock, Westinghouse Electric Co.
Q1:—The detrimental effect of air appears to be the presence of nitrogen.

What is the threshold level of nitrogen contamination in steam to cause the degra-
dation of the oxide and increase in kinetics?
Authors’ Response:—The threshold level in steam-nitrogen mixtures has not yet

explicitly determined. But, for instance in the oxygen-nitrogen system 1-99% of
nitrogen in the gas mixture causes a significant increase of the reaction rates in
comparison to the pure gases at 800 C (Recent, not yet published results of KIT).
1% nitrogen in oxygen increased the mass gain after 6h at 800 C by 120%; 10%
nitrogen by 600% compared to pure oxygen.
Q2:—Do you expect any difference in oxidation in steam and steam plus
oxygen?
Authors’ Response:—The oxidation kinetics in a steam-oxygen mixture should

be very similar to the kinetics in the pure gases. No hydrogen is released as long as
oxygen is in the mixture.
Question from Arthur Motta, Penn State University:—What is the physical rea-
son, in your view, of the nearly cubic kinetics you observe at 700 C? If you invoke
the effect of stress on migration energy, have you evaluated quantitatively how
much the diffusion coefficient of oxygen can change by stress, say at the yield stress
of the oxide?
Authors’ Response:—No investigations by us on the effect of stresses in the ox-

ide scale on oxygen diffusion coefficients have been performed. Cubic, and, more
generally, sub-parabolic kinetics have been observed in the given temperature
range. Possible explanations of such behavior have been taken from the literature
[8, 13-15].
Question from Valerie Vandenberghe, CEA:—You are performing double sided

oxidation, but on your picture, when in air, the oxidation does not appear to be
symmetric. Can you explain why and how do you take that into account when you
calculate the weight gain?
Authors’ Response:— In the TG experiments the tube segments are supported

by a sample holder in the form of a plate. The gas flow from the bottom to the top
of the vertical reaction tube causes different thermal hydraulic boundary conditions
for the external (steady flow) and internal (dead volume) surface.
Especially oxidation in air is strongly dependent on the boundary conditions

like flow rates and partial pressures of the involved gases. This may cause different
oxidation of the outer and inner surfaces.
The mass gain is always referred to the curved surface area of the hollow cylindrical
samples and the non-symmetrical behavior is only taken into account qualitatively.
Question from S. Anantharaman, BARC Mumbai:—What is the basis for the

time scales of the experiments especially at 1200 C and below?
Authors’ Response:—The duration of most of the basic tests is chosen in a way

that a results in a partial oxidation of the cladding segments in order to extract in-
formation on the oxidation mechanism by post-test examinations. (A completely
oxidized sample would only consist of ZrO2.)
Tests specially related to LOCA and beyond design basis accidents are planned
according to the conditions of the respective scenarios.
Question from Antoine Ambard, EDF:—What is the range of variation of the

breakaway time: If you have specimens coming from the same batch of material in
the same conditions of oxidation, can you give an estimate of the dispersion of the
breakaway time?
Authors’ Response:—We did not perform special reproducibility tests for all
conditions. But, if one performed tests with specimens from the same batch in the
same conditions the transitions time could vary by ca. 25%. A standard deviation of
24% of the transition time at 1000̄C in steam was determined from a test series
experiments in the BOX rig.
Nevertheless, any change in sample batch, sample preparation, and experimen-

tal boundary conditions may result in larger variations.
Questions from Yang-Pi Lin, Global Nuclear Fuel:—With respect to steam oxi-
dation of Zircaloy-4, specifically the time to breakaway oxidation, your results
appear to be different in some case to previous data. The work of Leistikow and
Schanz had similar time to breakaway at 1000 C as you report. But there are differ-
ences at other temperature. For example, 900 C, L-S showed very long time to
breakaway, but your results show much shorter time. Can you comment on the
differences?
Authors’ Response:—As already discussed, breakaway times are strongly de-

pendent on many conditions like sample batch, sample preparation, thermal hy-
draulic and other boundary conditions etc. For instance, in our tests in oxygen the
transition time at 900 C resulted also in a considerable later transition in compari-
son with the test in steam.
The conditions of the TG tests described in this paper and the experiments by
Leistikow and Schanz differed with respect to many parameters like sample size,
flow rates, furnace type and geometry, gas and steam flow rates etc. Hence it is
hardly possible to directly compare the data especially with respect to breakaway.
M. Le Saux,1 V. Vandenberghe,2 P. Crébier,3 J. C. Brachet,2

D. Gilbon,4 J. P. Mardon,5 P. Jacques,6 and A. Cabrera6
Influence of Steam Pressure on

the High Temperature Oxidation
and Post-Cooling
Mechanical Properties of
Zircaloy-4 and M5 Cladding
(LOCA Conditions)
Reference
Le Saux, M., Vandenberghe, V., Crébier, P., Brachet, J. C., Gilbon, D., Mardon, J. P., Jacques, P.,
and Cabrera, A., “Influence of Steam Pressure on the High Temperature Oxidation and Post-
Cooling Mechanical Properties of Zircaloy-4 and M5 Cladding (LOCA Conditions),” Zirconium
ABSTRACT
Studies that deal with small and intermediate break loss of coolant accident
(LOCA) conditions, for which steam pressure remains relatively high in the
reactor primary system, are scarce, even though it has been sometimes observed
that steam pressure could have a significant effect on the high temperature (HT)
oxidation kinetics of zirconium alloys. Thus, in order to study the effect of steam
Manuscript received December 7, 2012; accepted for publication October 6, 2013; published online
September 19, 2014.
1
CEA, DEN, Section for Applied Metallurgy Research, 91191 Gif-sur-Yvette Cedex, France (Corresponding
author), e-mail: matthieu.lesaux@cea.fr
2
CEA, DEN, Section for Applied Metallurgy Research, 91191 Gif-sur-Yvette Cedex, France.
3
CEA, DEN, Section for Technologies of Industrial Reactors, 38054 Grenoble Cedex 9, France.
4
CEA, DEN, Nuclear Material Department, 91191 Gif-sur-Yvette Cedex, France.
5
AREVA, AREVA NP SAS, Fuel Business Unit, 69546 Lyon Cedex 06, France.
6
EDF-SEPTEN, 69628 Villeurbanne Cedex, France.
7
Hyderabad, India.
Copyright V
LE SAUX ET AL., DOI 10.1520/STP154320120196 1003
pressure on the oxidation of low-tin Zircaloy-4 and M58 alloys, cladding samples
were oxidized at temperatures between 750 and 1200 C in flowing steam at
pressures ranging from 1 to 80 bars. It is shown that the oxidation kinetics of as-
fabricated Zircaloy-4 is enhanced under high steam pressure within the 750 to
1000 C oxidation temperature range. The effect is quite low at such
temperatures for as-fabricated M5, and at 1100 and 1200 C for both alloys. In
order to examine the influence of burn up effects, oxidation tests were
performed on Zircaloy-4 and M5 samples pre-hydrided at approximately their
respective end-of-life hydrogen expected contents for high burn up claddings
and on autoclaved specimens pre-oxidized (pre-existing oxide layer thicknesses
from 3 up to 8 lm) at 360 C with typical pressurized water reactor (PWR)
primary water chemistry. For both materials, it is shown that pre-hydriding (by
gaseous charging) does not significantly modify the oxidation kinetics at 850
and 1000 C (temperatures at which the steam pressure effect is maximal for the
fresh materials) under steam pressures of 1 and 80 bars. For the same HT
oxidation conditions, the weight gains are lower for the pre-oxidized samples
and the differences between sensitivities of Zircaloy-4 and M5 to a pressurized
steam environment are significantly reduced after pre-oxidation. No significant
hydrogen uptake was observed in the case of M5, whatever the investigated
oxidation conditions. Metallurgical observations revealed that the enhanced
oxidation kinetics of Zircaloy-4 under high pressure is associated with a
significant but limited hydrogen uptake and a thickening of the oxide layers
formed at HT. However, it is observed that the diffusion of the oxygen (coming
from the cladding oxidation process) through the metallic substrate is not
modified. The results show that, for Zircaloy-4 alloy, the steam pressure effect is
associated with changes in the oxide microstructure (crystallites morphology,
porosity, and cracks). These changes may be, for example, induced by a
destabilization of the tetragonal phase of zirconia, so that oxygen and hydrogen
transport through the oxide is easier. Ring compression tests performed at room
temperature and at 135 C showed that the effect of steam pressure on the post-
cooling mechanical properties of fresh, pre-hydrided, or pre-oxidized Zircaloy-4
and M5 samples after transient oxidation at HT is, for the conditions investigated,
generally limited, although significant in a few cases for Zircaloy-4. Finally, it was
determined that the impact of steam pressure observed for low-tin Zircaloy-4 is
very low or negligible on the equivalent cladding reacted (ECR) and the hydrogen
pickup of the cladding during typical in-reactor transients and has hence no
significant consequence on the verification of LOCA cladding safety criteria.
Keywords
low-tin Zircaloy-4, M5, high temperature oxidation, hydrogen pickup, steam
pressure, pre-hydriding, pre-transient oxide, post-cooling mechanical proper-
ties, oxide morphology
8
M5 is a trademark of AREVA NP registered in the USA and in other countries.
Introduction
During a loss of coolant accident (LOCA) in a light water reactor (LWR), zirco-
nium alloy fuel claddings may be exposed to steam at High Temperature (HT, up
to 1200 C) until they are quenched by emergency core cooling water. Most studies
performed on zirconium alloys under LOCA conditions focus on the large break
conditions, and thus have been performed at nearly atmospheric pressure [1–8].
The small and intermediate break LOCA conditions, for which steam pressure
remains relatively high in the reactor primary system, have been far less investi-
gated, even though it has been sometimes observed that steam pressure could
have a significant effect on the HT steam oxidation kinetics of the zirconium
alloys.
For example, an increase in the rate of oxidation with increasing the steam
pressure was observed by Pawel et al. [9] for Zircaloy-4 (Zy-4) oxidized at 900 C,
whereas no significant effect was noticed at 1100 C for steam pressures up to 103
bars. Bramwell et al. [10] reported a significant effect of steam pressure on the oxi-
dation rate of Zy-4 at 800 and 900 C, with a saturation of the effect at 800 C above
117 bars, and a less important impact at 1000 C. An exponential increase of the ox-
idation rate of Zy-4 with increasing the steam pressure up to 150 bars was observed
by Park et al. [11–15] for oxidation temperatures of 700, 750, 800, 850, and 900 C,
the effect being the more significant at 750 and 800 C. The authors deduced from
oxidation tests performed under a mixture of steam and argon that the oxidation
rate depends on the steam partial pressure rather than on the total pressure of the
atmosphere. The steam pressure effect under these conditions is higher for conven-
tional Zy-4 (1.5 %Sn) than for low tin Zy-4 (1.35 %Sn) [15]. A far less significant
impact of steam pressure (up to 150 bars) was observed for ZIRLO9 for oxidation
temperatures ranging from 700 to 900 C [13,14]. A delay in the oxidation rate was
observed at 750 and 850 C in the presence of a 20 or 50 lm thick pre-existing oxide
layer (formed under steam at 500 C) [12] or at 800 C in the presence of a 8 lm
thick pre-transient oxide [14], either for low or high steam pressures (up to 100
bars). However, the impact of steam pressure on the oxidation kinetics is the same
with or without a pre-existing oxide layer. Park et al. [15] recently observed that the
oxidation kinetics of low-tin Zy-4 at 800 C (during 1500 s) under a steam pressure
between 30 and 100 bars is faster when the material is previously hydrogen charged
up to 800 wt. ppm, whereas the oxidation kinetics of fresh and pre-hydrided speci-
mens are similar at atmospheric pressure. Vrtilkova et al. [16] observed a significant
effect of steam pressure on the oxidation kinetics of Zircaloy-4 (Sandvik co.) at
750 C (quasi-linear increase of the oxidation rate with increasing the steam pres-
sure up to 100 bars). However, contrary to Park et al.’s results, the effect is low at
850 C. At 750 C, the steam pressure effect is shown to be less important for E11010
9
ZIRLO is a trademark of Westinghouse Electric.
10
E110 is a trademark of A.A. Bochvar VNIINM.
than for Zy-4. At 850 C, the increase in weight gain of E110 with increasing steam
pressure is very large between 1 and 20 bars and very small between 20 and 100
bars.
Although some scatter does exist from one study to another, presumably due to
differences in material processing or experimental procedure associated with the
HT oxidation test, or both, this short review nonetheless illustrates that the effect of
steam pressure on the HT oxidation of zirconium alloys is dependent upon both
alloy and oxidation temperature. Furthermore, an additional consideration is that
during in-service operation, an oxide layer appears on the outer surface of the clad-
ding due to corrosion by the water environment and a fraction of the released
hydrogen is absorbed by the cladding. These changes may have an influence on the
steam pressure effect at HT, which thus indicates that the effect of steam pressure
will likely be a function of burnup.
The present paper investigates the influence of steam pressure (from 1 to 80
bars) on the oxidation of stress-relieved annealed low-tin Zircaloy-4 and M5 clad-
ding materials between 750 and 1200 C. For alloy M5, the data presented herein
are the first available in the literature about the effect of steam pressure at HT. The
impact of pre-hydriding (at approximately the end-of-life hydrogen contents
expected for high burn up claddings) and pre-oxidation (thicknesses of pre-
transient oxide layers comprised between about 3 and 8 lm) on the steam pressure
effect are examined for both materials. From these tests, the oxidation kinetics, the
hydrogen pickup at HT and the post-cooling mechanical properties at 135 C are
discussed. Results of metallurgical investigations are also presented and potential
mechanisms responsible for the steam pressure effect are discussed. Finally, the
impact of the steam pressure effect during representative in-reactor transients is
examined.
Materials and Experimental Procedure

MATERIALS
The materials studied were stress-relieved annealed (SRA) low tin Zircaloy-4 and
fully recrystallized M5 [17] cladding tube specimens from AREVA NP. The nomi-
nal tube outer diameter and thickness are about 9.5 and 0.57 mm, respectively. The
chemical compositions of the tested materials are given in Table 1.
TABLE 1 Nominal chemical compositions (in wt. %) of the studied materials.
Material Sn Fe Cr Nb O Zr
Zircaloy-4 1.3 0.2 0.1 — 0.13 Bal.

M5TM — 0.04 — 1.0 0.14 Bal.
Pre-Hydriding
In order to study the potential effect of pre-hydriding, some specimens were hydro-
gen pre-charged by CEZUS Paimboeuf (France) by gaseous charging in an argon/
hydrogen mixture at 400 C. A hydrogen content of about 500–600 wt. ppm was
achieved for SRA Zircaloy-4 samples, which is approximately the end-of-life hydro-
gen content expected for high burnup Zircaloy-4 cladding [18]. For alloy M5, the
190–230 wt. ppm hydrogen content obtained is, however, about two times higher
than the maximal hydrogen content expected for high burn up M5 cladding, which
is typically less than 100 wt. ppm [19].
Pre-Oxidation
The investigation of the influence of a pre-existing oxide layer was performed on
specimens pre-oxidized on both the tube interior and external surfaces (two-side
oxidation) by EDF R&D at 360 C (during 2400 and 7200 h) in a static autoclave
with a typical pressurized water reactor (PWR) primary water chemistry (1000 ppm
boron, 2.2 ppm lithium and 30 cc/kg hydrogen). These conditions were chosen in
order to be as close as possible to the in-reactor conditions at the outer surface of
fuel claddings. The thicknesses of the pre-existing oxide layers, which are expected
to be the same on outer and inner surfaces, were 3 and 8 lm for Zircaloy-4 samples
and 3 and 5 lm for M5. The outer surface of the pre-oxidized samples was homoge-
neously smooth and black/dark grey. During the pre-oxidation process, although
hydrogen pickup occurs, the hydrogen content remains lower than 100 wt. ppm
even for the thickest pre-oxide layers investigated.
HIGH TEMPERATURE OXIDATION AND COOLING

The 20 mm long samples were two-side oxidized in a flowing steam environment
(generated from demineralized water) in the CINOG HP facility of CEA in
France. The oxidation tests were performed at 750, 850, 900, 1000, 1100, or
1200 C with steam pressures of 1, 10, 20, 40, 60, and 80 bars. Depending upon
the temperature and pressure conditions, oxidation times ranged from 200, 400,
600, 1800, 3600, 7200, 9000, and 18 000 s. To obtain the desired temperatures, the
specimens were induction-heated. The specimens were shorter than the induction
coil in order to the minimize temperature gradient along the specimens. Maximal
variations of transient oxide layer thickness and temperature are expected to be
lower than 10 % and 10 C, respectively, over the 5 mm long central region of the
samples.
Steam flow was directed down through a quartz-tube chamber. The steam flow
rate is about 200 mg/s (or 140 mg/(cm2 s) if normalized to the cross-sectional area
of the steam chamber), which is sufficient to avoid steam starvation according to
the results reported in [3,4,20]. The outer temperature of the specimens, obtained
in the region near their mid-length, was measured using two optical pyrometers.
One pyrometer was used to monitor the temperature (feedback to the furnace
power) and the other one was used to check the temperature.
The specimens are heated at 20 C/s up to the set-point oxidation temperature
without any temperature overshoot. The oxidation time commenced when the oxi-
dation temperature was reached. After the HT oxidation, the specimens are cooled-
down by shutting off the induction heating. The achieved cooling is quite fast (cool-
ing rate between 100 and 150 C/s down to 500 C) but significantly slower than
during a direct water quench (1000 C/s).
Results of two-side oxidation tests previously performed in the CINOG BP facility
(CEA, France) [21,22] on SRA Zircaloy-4 and M5 under steam at atmospheric pres-
sure between 700 and 1400 C are reported in this paper for comparison. Heating
mode and temperature measurement are similar in CINOG BP and CINOG HP facili-
ties, but the CINOG BP facility operated under atmospheric pressure only. In addition,
some results of one-side oxidation tests performed in steam at atmospheric pressure
between 1000 and 1250 C in DEZIROX 1 [7,8,23], another facility of CEA, which used
a vertical resistive furnace to heat the 150 mm long samples, are also reported.
POST-TEST HYDROGEN CONTENT ANALYSIS

Post-tests hydrogen content measurements were performed at CEZUS Paimboeuf
by an inert gas fusion thermal conductivity technique on samples (including oxide
layers) extracted from a select number of samples oxidized at HT.
POST-COOLING MECHANICAL TESTING

In order to evaluate the post-cooling mechanical properties of the materials oxi-
dized at HT under various steam pressures, ring compression tests (RCTs) were
performed on 10 mm long specimens cut from the middle of the CINOG HP
specimens. Testing was conducted on a universal electromechanical tensile
machine at temperatures of 20 and 135 6 2 C (saturation temperature during
reflood) with a cross-head displacement rate of 0.5 mm/min (up to a maximum
displacement of 6 mm and a maximum load of 2 kN). Failure of RCT samples was
considered to occur at the first significant load drop of more than 30 %, which
indicates the formation of at least one through-wall crack. Finite element simula-
tions and interrupted tests have been performed to confirm the validity of this
procedure [24].
METALLURGICAL OBSERVATIONS
Optical and Scanning Electron Microscopy
Depending on the oxidation temperature, several layers can be observed after cool-
ing in the samples oxidized at HT: a ZrO2 oxide, an aZr(O) (oxygen stabilized aZr)
phase layer, a two phases (aZr þ bZr) layer and a so-called prior-bZr layer (bZr phase
during oxidation but aZr with a typical Widmanstätten or parallel laths structure
morphology after cooling in the absence of a stabilizing element). Measurements of
the different layers’ thicknesses were performed via metallographic examinations
on polished transverse cross-sections obtained from the middle part of selected
samples that had been oxidized at HT in the CINOG HP facility.
Optical microscopy was also used to measure the oxide thickness (measure-
ment uncertainty of a few microns). When the thickness of the oxide layer was
expected to be low, a nickel coating was applied before imbedding and polishing
the sample on its inner and outer surfaces in order to stabilize the oxide layer and
minimize the end effects during polishing. Scanning electron microscopy (SEM)
(mainly in back scattered electron mode) and a specific image analysis procedure
were used to measure the aZr(O) and prior-bZr phase fractions across the cladding
wall thickness. In the following, the so-called aZr(O) phase layer corresponds to the
“continuous” aZr(O) phase layer containing more than 90 % of aZr(O) (more details
about the procedure are given in Refs. [8,25]).
Electron Probe Micro Analysis (EPMA)

Oxygen diffusion profiles across the cladding wall thickness were quantified by
EPMA using a CAMECA SX100 electron microprobe. A typical accuracy of
61000 wt. ppm is expected for oxygen concentration measurements due to surface
contamination.
Field Emission Gun Scanning Electron Microscopy (FEG–SEM)

In order to examine the morphology of the oxide layers, the fracture surface of
some samples failed during RCT after oxidation and cooling in the CINOG HP fa-
cility was observed by FEG-SEM.
Micro Laser Induced Breakdown Spectroscopy (lLIBS)

Some lLIBS experiments were performed at CEA on some Zy-4 samples oxidized
at 850 C for which hydrogen uptake was observed during oxidation at HT, in order
to perform semi-quantitative mapping of the hydrogen distribution within the me-
tallic matrix with a spatial resolution close to 1 lm3 (lLIBS is a new spectroscopic
analysis technique that uses the light emitted from a laser-generated micro-plasma
to determine the composition of a sample based on elemental and molecular emis-
sion intensities [26]). EPMA X-ray mapping of the main alloying elements (Fe, Cr,
Sn, and O) was also performed on these same samples.
OXIDATION KINETICS AND HYDROGEN PICKUP

Oxidation Kinetics of the Fresh Materials at Atmospheric Pressure
First of all, the consistency of the results obtained in the CINOG HP facility under
steam at atmospheric pressure with data previously obtained under similar conditions
in the CINOG BP and DEZIROX 1 facilities was checked for both studied alloys. The
oxidation kinetics can be described by the following relationship between the weight
gain Wg and the oxidation time t at a constant temperature T:
(1) Wg ðT; t; 1Þ ¼ K ðT; 1Þt 1=nðT;1Þ
where:
K (T, 1) and n (T, 1) ¼ respectively, the rate constant and rate exponent at
atmospheric pressure (1 bar).
The oxidation times longer than or equal to 5000 s at 1000 C were not included
so that the breakaway oxidation phenomenon is not taken into account. The break-
away oxidation phenomenon of Zy-4 and M5 is characterized, in particular, by a
rapid increase of the oxidation rate and a significant hydrogen uptake after about
5000 s at 1000 C under steam at atmospheric pressure [6,7,22,23]. The results of
the CINOG HP experiments at atmospheric pressure are compared in Fig. 1 to the
oxidation kinetics deduced from previous CINOG BP and DEZIROX 1 data and
described by Eq 1. The results are in good agreement for oxidation temperatures of
750, 850, 900, 1000, 1100, and 1200 C. The oxidation kinetics of fresh Zy-4 under
steam at atmospheric pressure is faster than that of fresh M5 for temperatures
below 1100 C. The kinetics of fresh Zy-4 and M5 are nearly the same at higher tem-
peratures. For both materials, whatever the temperature, the weight gains are lower
than those predicted by the Baker–Just (BJ) parabolic rate correlation [1], which is
commonly considered to be conservative (at least for oxidation temperatures above
about 900 C). The measured weight gains are slightly underpredicted by the
Cathcart–Pawel (CP) parabolic rate correlation [2] for temperatures exceeding
1000 C in the case of Zy-4 and 1100 C for M5. For temperatures less that the iden-
tified threshold, the measured results were slightly overpredicted.
According to CINOG BP, DEZIROX 1, and CINOG HP data, the oxidation
kinetics follows a sub-parabolic law (n 2.6 on average in Eq 1, with values ranging
from about 2.5 to 3) between 700 and 900 C and is nearly parabolic (n 2, with
values ranging from 1.8 to 2.5) within the 1000–1400 C temperature range. These
FIG. 1 Oxidation kinetics obtained in the CINOG HP facility under steam at atmospheric
pressure for fresh (a) Zircaloy-4 and (b) M5, and comparison with oxidation
kinetics deduced from previous oxidation tests performed in the CINOG BP and
DEZIROX 1 facilities.
FIG. 2 Oxidation kinetics of fresh Zircaloy-4 and M5 under steam at (a) 750, 850, and
900 C and (b) 1000, 1100, and 1200 C for various steam pressures. Comparison
with the weight gains estimated from the BJ and the CP parabolic rate
correlations.
oxidation rate exponents are fairly consistent with those determined from tests per-
formed under steam at nearly atmospheric pressure on Zy-4 or ZIRLO [27–29].
EFFECT OF STEAM PRESSURE ON THE FRESH MATERIALS

Oxidation Kinetics
The results of oxidation tests performed under steam at 750, 850, 900, 1000, 1100,
and 1200 C on fresh Zy-4 and M5 are shown in Fig. 2 for various steam pressures
up to 80 bars. In this figure, the weight gains estimated from the Baker–Just [1] and
the Cathcart–Pawel [2] parabolic rate correlations obtained under steam at atmos-
pheric pressure are reported for comparison. The evolution of weight gain as a
function of steam pressure for various oxidation time/temperature conditions is
plotted in Figs. 3 and 4 for fresh Zy-4 and M5, respectively. No transition of the
kinetics (e.g. “breakaway” oxidation) was observed during the oxidation for the
conditions discussed here (oxidation times lower than 7200 s). For all tested
FIG. 3 Evolution as a function of steam pressure for various oxidation times of

the weight gain of fresh Zircaloy-4 oxidized at 750, 850, 900, 1000, 1100,
and 1200 C. Comparison with the weight gains estimated from the BJ
correlation.
FIG. 4 Evolution as a function of steam pressure for various oxidation times of

the weight gain of fresh M5 oxidized at 750, 850, 900, 1000, 1100,
and 1200 C. Comparison with the weight gains estimated from the BJ
correlation.
conditions, inner and outer oxide layers are homogeneously black or grey (more or
less dark depending on the oxidation condition), without any spalling.
At 750 C, the weight gains of fresh Zy-4 increase with increasing steam pres-
sure, from at least 200 s after the beginning of oxidation, while, at this temperature,
the effect of steam pressure is low for M5.
At 850 C, the weight gains of Zy-4 increase slightly with increasing steam pres-
sure between 1 and 40 bars, from 200 s after the beginning of oxidation. The steam
pressure effect is significant for steam pressures as low as 10 bars compared to the
results obtained at nearly atmospheric pressure. A more significant steam pressure
effect is observed from a “threshold” steam pressure close to 40 bars. The increase
of weight gain as a function of steam pressure is nearly proportional between 40,
60, and 80 bars. The weight gains measured after oxidation under a steam pressure
above 60 bars are underpredicted by the Baker–Just correlation. The weight gain of
M5 at 850 C progressively increases with increasing the steam pressure from 1 to
80 bars, without an acceleration between 40 and 80 bars as observed for Zy-4. In
addition, as observed in Fig. 2 for M5, the measured weight gains are overpredicted
by the Baker–Just correlation for all cases.
At 900 C, the weight gain of fresh Zy-4 gradually increases with increasing
steam pressure. The weight gains obtained for fresh Zy-4 under a 80 bars steam
pressure are slightly higher than those estimated from the Baker–Just parabolic cor-
relation, whereas for M5, the weight gains slightly decrease with increasing steam
pressure at this temperature.
At 1000 C, the weight gain of fresh Zy-4 also increases with increasing steam
pressure. This increase, which is observed less than 200 s after the oxidation tem-
perature is reached, is relatively low between 1 and 40 bars and is more significant
between 40 and 80 bars. The weight gains obtained under 80 bars are underpre-
dicted by the Baker–Just correlation. The weight gains of fresh M5 decrease
between 1 and 40 bars, whereas the steam pressure effect is insignificant between 40
and 80 bars.
At 1100 and 1200 C, the impact of steam pressure between 1 and 80 bars on
the oxidation of fresh Zy-4 and M5 is low or negligible.
As a first approximation, the dependency of the weight gain as a function of
steam pressure can be fairly well described by an exponential function, as already
proposed by Park et al. [12,13,15]
Wg ðT; t; PÞ
(2) ¼ expðaðT; t Þ ðP 1ÞÞ
Wg ðT; t; 1Þ
where:
Wg (T, t, P) and Wg (T, t, 1) ¼ the weight gains at steam pressures of P and
1 bar, respectively, for an oxidation time t at a constant temperature T, and a ¼ a
parameter in bars–1 that depicts the steam pressure effect on weight gain.
The normalized values of the steam pressure effect parameter a deduced from
data fitting with Eq 2 for each oxidation time and temperature couple (and
FIG. 5 Evolution as a function of the oxidation time of the steam pressure effect
parameter (normalized values of a in Eq 2) for oxidation tests performed at 750,
850, 900, 1000, 1100, and 1200 C on fresh (a) Zircaloy-4 and (b) M5.
normalized by the same factor for all data for illustration purposes) are illustrated in
Fig. 5. The steam pressure effect increase with increasing the oxidation time, fairly sig-
nificantly at the beginning of oxidation and at a slower rate afterward (there is a kind
of saturation of the steam pressure effect as a function of the oxidation time, i.e., as
the oxide growths). The dependency of the steam pressure effect parameter to the ox-
idation time seems to be relatively similar for both alloys and to depend slightly on
the oxidation temperature. In the case of fresh Zy-4, the steam pressure effect is maxi-
mal at 850 C, but does not vary so much at the other temperatures within the
750–1000 C temperature range. The effect is negligible (i.e. within the expected ex-
perimental uncertainty range) at 1100 and 1200 C. For fresh M5, the steam pressure
effect is positive but nearly four times lower than that of Zy-4 at 850 C, negative but
quite low at 900, 1000, and 1100 C and is negligible at 750 and 1200 C.
The steam pressure does not have any significant effect on the exponent n of
the general oxidation kinetics model (Eq 1): for both materials, whatever the steam
pressure, the oxidation kinetics follows a sub-parabolic law (n 2.6 6 0.2) at 750,
850, and 900 C and is nearly parabolic (n 2 6 0.3) at 1000, 1100, and 1200 C.
Hydrogen Pickup
The hydrogen contents of both fresh Zy-4 and M5 samples were measured after ox-
idation at HT. No significant hydrogen uptake was observed in the case of M5,
whatever the tested oxidation conditions. Figure 6 shows the hydrogen contents of
Zy-4 samples as a function of the oxidation time and the equivalent cladding
reacted (ECR, in %) determined from weight gain Wg (in mg/cm2) using the follow-
ing equation (established for two-side oxidation):
(3) ECR ¼ 0:0878Wg=t0

FIG. 6 Evolution as a function of (a) the oxidation time and (b) the ECR determined
from measured weight gain of the hydrogen content of fresh Zircaloy-4 samples
oxidized under various steam pressures at 750, 850, 900, 1000, 1100, and
1200 C.
where:
t0 ¼ the initial specimen thickness (in cm).
The ECR, which measures total oxidation of the cladding, is the fraction of the
cladding thickness that would be converted to oxide assuming that all the absorbed
oxygen reacts with zirconium and forms stoichiometric ZrO2. On the one hand, it
is shown that at 750, 850, 900, and 1000 C, the increase in weight gain of Zy-4 with
increasing steam pressure is associated with a significant but limited hydrogen
pickup. Hydrogen uptake increases during oxidation. The higher the steam pres-
sure, the higher the hydrogen uptake. No significant hydrogen pickup is observed at
atmospheric pressure for the conditions investigated. For a steam pressure of 80
bars, the hydrogen uptake is about 200 wt. ppm after 3600 s at 850 C and 300 wt.
ppm after 1800 s at 1000 C. On the other hand, at these temperatures, no signifi-
cant hydrogen uptake is observed in the case of fresh M5, whose oxidation kinetics
depend only little on steam pressure. After 400 s at 1100 and 1200 C, temperatures
at which the steam effect on the oxidation kinetics is low, a slight hydrogen pickup
(about 40 wt. ppm) is observed for Zy-4 samples oxidized under 80 bars, whereas
no hydrogen uptake is noticed for M5, as expected under these conditions [30].
If the samples oxidized 600 and 3600 s at 850 C and 1800 s at 1000 C, for
which hydrogen pickup is above the general trend, are not taken into account, Fig. 6
suggests that: (i) for a given steam pressure, the hydrogen uptake (and the increase
in weight gain) of Zy-4 is nearly parabolic as a function of oxidation time, more or
less independently of the oxidation temperature. In other words, for given steam
pressure and oxidation time, the hydrogen uptake depends little on the oxidation
temperature; (ii) for a given oxidation temperature, the hydrogen uptake of fresh
Zy-4 increases nearly linearly with increasing ECR, almost independently of the
steam pressure when it is comprised between 40 and 80 bars.
The hydrogen pickup fraction absorbed by the material during the oxidation at
HT is the ratio between the mass of hydrogen absorbed by the sample and the total
mass of hydrogen released during oxidation
DmHa
(4) fHa ¼
DmHt
where:
DmHa ¼ DCHm ¼ the mass of hydrogen absorbed by the sample at HT,
DCH ¼ the measured hydrogen pickup,
m ¼ the mass of the sample after oxidation, and
DmHt ¼ the total mass of hydrogen released during oxidation.
As a first approximation, this last quantity can be estimated from the measured
weight gain, that makes it possible to evaluate the total quantity of oxygen having
reacted with the material and consequently—by considering the decomposition of
the water molecule—the total quantity of hydrogen released
MH
(5) DmHt ¼ 2Sðdm DCH mÞ
MO
where:
S ¼ the surface of the sample exposed to steam,
MH and MO ¼ the respective molar masses of hydrogen and oxygen, and
dm ¼ the difference between the sample weight gain measured after oxidation
and the weight gain at which hydrogen pickup starts.
As shown in Fig. 6, the hydrogen pickup observed for Zy-4 oxidized under
steam pressures of 40, 60, or 80 bars occurs relatively early. As a first approxima-
tion, in order to evaluate the hydrogen pickup fraction absorbed at HT, and in the
absence of more accurate information, hydrogen pickup is considered to start at the
very beginning of oxidation. Although they are approximate and constitute a lower
bound of the actual fraction of hydrogen absorbed, the results represented in Fig. 7
suggest that: (i) for given steam pressure and oxidation time, the higher the oxida-
tion temperature, the lower the hydrogen pickup fraction; (ii) for given steam pres-
sure and oxidation temperature, the hydrogen pickup fraction seems to decrease as
a function of the oxidation time; (iii) the hydrogen pickup fraction remains rela-
tively low (lower than 10 %).
EFFECT OF PRE-HYDRIDING
The weight gains obtained after 600 s at 850 C and 200 and 400 s at 1000 C under
atmospheric and 80 bars steam pressure on pre-hydrided Zy-4 and M5 (at about
600 wt. ppm for Zy-4 and 150 wt. ppm for M5) are compared in Fig. 8 to those
acquired under similar conditions on fresh materials. The 850 and 1000 C
FIG. 7 Evolution as a function of the oxidation time of the hydrogen pickup fraction of
fresh Zircaloy-4 samples oxidized under various steam pressures at 750, 850,
900, 1000, 1100, and 1200 C.
temperatures were selected to assess the effect of pre-hydriding due to the fact that
at these temperatures, the steam pressure effect was significant for fresh Zy-4.
The weight gains of pre-hydrided materials oxidized at atmospheric pressure
are similar to those of fresh materials for both alloys. This is consistent with previ-
ous results obtained for oxidation temperatures of 1000, 1100, and 1200 C [8]. The
FIG. 8 Evolution as a function of steam pressure of the weight gain of fresh and pre-
hydrided (a) Zircaloy-4 and (b) M5 oxidized during 600 s at 850 C and 200 and
400 s at 1000 C. Comparison with the weight gains estimated from the BJ and
the CP parabolic rate correlations.
FIG. 9 Evolutions as a function of (a) steam pressure and (b) weight gain of the
hydrogen pickup of fresh, pre-hydrided and pre-oxidized Zircaloy-4 samples
oxidized at 850 and 1000 C (error bars are due to uncertainties on the actual
hydrogen contents before HT oxidation, since the hydrogen analyses were done
at the ends of 200 mm long sections, from which the 20 mm long samples were
cut).
weight gains of fresh and pre-hydrided materials are also very similar under a steam
pressure of 80 bars, i.e., significantly higher than those measured at atmospheric
pressure in the case of Zy-4 and not very different in the case of M5.
For both Zy-4 and M5, fresh or pre-hydrided, no significant hydrogen uptake
under 1 bar steam pressure was observed. Under an 80 bars steam pressure, the
hydrogen uptake is significant but limited for pre-hydrided Zy-4 as for fresh Zy-4
(Fig. 9), where for fresh and pre-hydrided M5 there was no significant hydrogen
uptake at this steam pressure.
These results show that the oxidation kinetics and the hydrogen uptake of Zy-4
and M5 and the effect of steam pressure at 850 and 1000 C are not significantly
impacted by pre-hydriding (by gaseous charging) up to end-of-life hydrogen
contents: for the investigated test conditions, the effect of steam pressure is important
for Zy-4 and much less significant for M5, whether the material was pre-hydrided
or not.
These results are different from those obtained by Park et al. [15], who
observed that at 800 C under a steam pressure between 30 and 100 bars, Zy-4 sam-
ples pre-hydrided at 800 wt. ppm exhibit a faster oxidation kinetics than the as-
received material, whereas the oxidation kinetics are similar at atmospheric pres-
sure. This may result from different hydrogen charging processes or HT oxidation
conditions, or both. Furthermore, as discussed in Ref. [30], the HT oxidation pro-
cess may be influenced by the initial sample surface conditioning or the steam flow
for instance.
FIG. 10 Evolution as a function of steam pressure of the weight gain of fresh and pre-
oxidized (pre-existing oxide layers thicknesses of 3 and 8 lm for Zircaloy-4 and
3 and 5 lm for M5) Zircaloy-4 and M5 oxidized during (a) 600 s at 850 C and
(b) 400 s at 1000 C. Comparison with the weight gains estimated from the BJ
and the CP parabolic rate correlations.
EFFECT OF A PRE-EXISTING OXIDE LAYER

The weight gains of pre-oxidized Zy-4 and M5 samples (pre-transient oxide layers
thicknesses of about 3 and 8 lm for Zy-4 and 3 and 5 lm for M5) oxidized 600 s at
850 C and 400 s at 1000 C under steam at 1 and 80 bars are compared in Fig. 10 to
those obtained on fresh and pre-hydrided materials.
For the same oxidation conditions, the weight gains are lower for the pre-
oxidized samples, for both investigated steam pressures. This is consistent with the
observations noted for short oxidation times at temperatures between 900 and
1200 C under steam at atmospheric pressure [23,31]. Indeed, it was shown that the
formation of a “fresh” transient oxide at HT at the metal/oxide interface is delayed
in the presence of a pre-existing oxide layer (the delay roughly corresponds to the
time necessary for oxygen anions coming from the outer steam to diffuse across the
pre-existing oxide layer) [23].
In the case of pre-oxidized materials, for a given pre-existing oxide layer thick-
ness, the weight gains of M5 samples are close to those of Zy-4, whereas they are
lower (at a given oxidation time at 850 or 1000 C) in the case of non-pre-oxidized
specimens. In other words, the decrease in weight gain due to the presence of a pre-
oxide layer is lower for M5 than for Zy-4 for the conditions investigated. A similar
tendency was formerly highlighted during oxidation tests performed at 1000 C
under steam at atmospheric pressure [23].
As shown in Fig. 11, the effect of steam pressure on the oxidation of Zy-4 at 850
or 1000 C is less important for coupons that had been pre-oxidized: the weight
gains are still higher under a steam pressure of 80 bars than at atmospheric pres-
sure; however, the increase in weight gain is, for both investigated pre-existing ox-
ide layer thicknesses (3 and 8 lm), nearly two times lower than for non-pre-
oxidized samples. This tendency is not rigorously similar to that observed by Park
et al. who did not noticed any significant impact of pre-oxidation on the effect of
steam pressure for oxidation tests performed at 750 and 850 C under steam pres-
sures up to 100 bars on Zy-4 specimens with 20 and 50 lm thick pre-existing oxide
layers (formed under steam at 500 C) [12] or for tests performed at 800 C on Zy-4
samples with 8 lm thick pre-oxide [14]. This difference in the protectiveness of the
pre-films may be due to the characteristics of the pre-oxide layers, which are prob-
ably different from those of the specimens tested here, pre-oxidized in water at
360 C.
For M5, the effect of steam pressure at 850 and 1000 C remains low after pre-
oxidation (relative deviation between weight gains obtained under steam pressures
of 80 bars and 1 bar lower than 20 %). The steam pressure effects on the oxidation
FIG. 11 Evolution as a function of the pre-existing oxide layer thickness of the steam
pressure effect parameter (normalized values of a in Eq 2) for oxidation tests
performed at 850 C (600 s) and 1000 C (400 s) on Zircaloy-4 and M5.
of Zy-4 and M5 appear to be close after pre-oxidation, while the effect is much
higher for Zy-4 in the case of the non-pre-oxidized materials. The differences
between sensitivities of Zy-4 and M5 to steam pressure are significantly smaller af-
ter pre-oxidation.
Due to both the delay effect imparted by the prefilmed oxide layer and the asso-
ciated reduction of the steam pressure effect, the weight gains of pre-oxidized Zy-4
at 850 and 1000 C are substantially lower than those predicted by the Cathcart–-
Pawel and Baker–Just correlations, even for a steam pressure as high as 80 bars.
Similar behavior is noted in the case of M5.
At 850 and 1000 C, the hydrogen uptake of Zy-4 with 3 lm thick pre-existing
oxide layer is negligible at atmospheric pressure and significant but low (lower
than 50 wt. ppm) under 80 bars steam pressure (Fig. 9(a)). After 600 s at 850 C or
400 s at 1000 C, the hydrogen pickup of the pre-oxidized samples are lower than
those of the fresh or pre-hydrided material. This is correlated with the lower
weight gains measured for the pre-oxidized material for the same oxidation condi-
tions (Fig. 9(b)). As for the fresh or pre-hydrided material, no significant hydrogen
uptake is observed in the case of M5 oxidized at 850 or 1000 C under 1 or 80
bars.
These results show that thin pre-existing oxide layers (3 and 8 lm thick) are pro-
tective in terms of weight gains (at least for short oxidation times) at 850 and 1000 C
for both alloys (the protective effect is nevertheless more significant in the case of Zy-
4), whether at atmospheric pressure or at 80 bars. At these temperatures, the steam
pressure effect on the oxidation of Zy-4 is reduced in the presence of a pre-existing
oxide layer, so that it becomes as low as that of M5.
METALLURGICAL OBSERVATIONS
Oxide and Metallic Layers Thicknesses
Metallurgical observations were performed on some fresh Zy-4 and M5 samples
oxidized at 850, 1000, and 1200 C under various steam pressures and cooled (typi-
cal cooling rate of about 100 C/s) in the CINOG HP facility.
Examples of optical micrographs of fresh Zy-4 and M5 samples oxidized under 1
and 80 bars steam pressures are given in Figs. 12 and 13. It is notable that the metal/ox-
ide interface of the samples oxidized during 1800 s at 1000 C under 1 or 80 bars is
scalloped. Although less significant, it is also the case for the samples oxidized during
3600 s at 850 C. The amplitude of the undulations at the metal/oxide interface is
higher when the oxidation is performed under high steam pressure, consistent with
the observations reported in Refs. [10,15] for oxidation tests performed at 900 C. One
should note that the transition from a smooth metal/oxide interface to a scalloped
interface is generally considered as a precursor to the breakaway oxidation transition
that occurs around 1000 C under steam at atmospheric pressure in zirconium alloys
[4,6,22]. The amplitude of the undulations is higher and the oxide is thicker for Zy-4
than for M5. On the other hand, the metal/oxide interface of Zy-4 and M5 samples
oxidized at 1200 C at 1 or 80 bars is smooth and without undulations.
FIG. 12 Optical micrographs of transverse cross-sections of fresh Zircaloy-4 samples

two-side oxidized at 850, 1000, and 1200 C in the CINOG HP facility under
steam at atmospheric pressure and at 80 bars.
The measured thicknesses of the oxide layers and the “continuous” aZr(O)
phase layers are depicted as a function of steam pressure in Figs. 14 and 15, where
the error bars represent the standard deviations on the measured thicknesses. The
thicknesses estimated from the results of oxidation tests performed earlier by CEA
at atmospheric pressure in the CINOG BP and DEZIROX 1 facilities are also
included for comparison. It is to be noted that the oxide layers thicknesses are uni-
form circumferentially, close at the samples inner and outer surfaces and in good
FIG. 13 Optical micrographs of transverse cross-sections of fresh M5 samples two-side

oxidized at 850, 1000, and 1200 C in the CINOG HP facility under steam at
atmospheric pressure and at 80 bars.
agreement with those acquired earlier at atmospheric pressure by CEA with two
different facilities. The variation of the oxide layer thickness within a given sample
is generally lower than 10 % from the average value. The standard deviation is
slightly higher (between 10 and 30 %) for the samples oxidized at 1000 C, especially
under 80 bars. The more significant scatter may, in part, be due to the scalloped
metal/oxide interface. It is known that under these test conditions, breakaway oxi-
dation (e.g., at 1000 C and atmospheric steam pressure) occurs and that oxidation
is less homogeneous due to the scalloped metal/oxide interface.
FIG. 14 Evolution as a function of steam pressure of the inner and outer oxide layer
thicknesses of fresh (a) Zircaloy-4 and (b) M5 oxidized at 850, 1000, and
1200 C. Comparison with thicknesses estimated from correlations deduced
from previous oxidation tests performed in the CINOG BP and DEZIROX 1
facilities.
The oxide thicknesses measured after oxidation at 850 and 1000 C are lower
for M5 than for Zy-4, which is consistent with the differences in weight gains. The
oxide layer thicknesses formed after 3600 s at 850 C or 1800 s at 1000 C on M5 are
not significantly modified by the increase of steam pressure up to 80 bars (the rela-
tive variation of the oxide layer thicknesses is lower than 30 %). In the case of Zy-4,
in consistency with the measured weight gains, the oxide layer thicknesses depend
much more on steam pressure: for example, the oxide layers formed under 80 bars
after 3600 s at 850 C or 1800 s at 1000 C are 70 to 80 % thicker than those pro-
duced at atmospheric pressure. For both Zy-4 and M5, the steam pressure effect
(between 1 and 80 bars) on the oxide thickness formed after 400 s at 1200 C is in-
significant (relative variation lower than 5 %). The increase of the oxide layer thick-
ness as a function of the oxidation time and temperature or the steam pressure is
well correlated to the increase of the measured weight gains. It is also consistent
with the tendencies deduced from previous tests performed under steam at atmos-
pheric pressure in the CINOG BP facility.
As illustrated by Fig. 15, the aZr(O) phase layers thicknesses of the samples oxi-
dized 1800 s at 1000 C are not modified by an increase of the steam pressure, either
for Zy-4 or M5. The oxygen diffusion profiles across the metallic substrate of fresh
Zy-4 samples oxidized 3600 s at 850 C are the same after oxidation under steam
pressures of 1 and 80 bars (Fig. 16). Thus, the quantity of oxygen that diffuses within
the metallic substrate is not intrinsically modified under high steam pressures. In
other words, the increase in weight gain observed for Zy-4 when increasing the
FIG. 15 Evolution as a function of steam pressure of the inner and outer aZr(O) phase
layers of fresh (a) Zircaloy-4 and (b) M5 oxidized at 1000 C. Comparison with
thicknesses estimated from correlations deduced from previous oxidation tests
performed in the CINOG BP and DEZIROX 1 facilities.
FIG. 16 Oxygen concentration profiles (EPMA) normalized by the oxygen

concentration just beneath the metal/oxide interface (6 wt. %) (the measured
raw data overestimate the actual oxygen content values due to surface
contamination which was not corrected here) across the half thickness (from
the initial outer surface of the cladding, i.e., by taking into account the
Pilling–Bedworth ratio of 1.56) of fresh Zircaloy-4 samples oxidized 3600 s at
850 C under steam pressures of 1 and 80 bars.
FIG. 17 FEG–SEM micrographs of the fracture surfaces of fresh Zy-4 samples failed
during RCT after oxidization at 850 C (3600 s) and 1000 C (1800 s) under
steam at atmospheric pressure and at 80 bars and cooling in the CINOG HP
facility.
steam pressure is associated with an increase of the oxide layer, without additional
diffusion of oxygen into the metallic substrate.
MORPHOLOGY OF THE OXIDE LAYERS

In order to examine the morphology of the oxide layer and its evolution as a func-
tion of the oxidation conditions, FEG–SEM observations were made on the fracture
surface of some fresh Zy-4 and M5 RCT samples that were tested at 20 or 135 C af-
ter oxidation and cooling in the CINOG HP facility. It was verified that the oxide
morphology was not impacted by the temperature of mechanical testing. Some
examples of micrographs focusing on zirconia layer (brittle fracture mainly inter-
granular) are given in Figs. 17 and 18 for Zy-4 and M5, respectively.
All the observed oxide layers formed between 750 and 1200 C either under
atmospheric pressure or 80 bars show a columnar structure. The crystallites are
mainly oriented perpendicularly to the metal/oxide interface. The development of
such a structure is driven by the minimization of the compressive stresses within
the oxide [32] or more secondarily via the effect of stress on the effective oxygen
FIG. 18 FEG–SEM micrographs of the fracture surfaces of fresh M5 samples failed

during RCT after oxidization at 850 C (3600 s), 1000 C (1800 s), and 1200 C
(400 s) under steam at atmospheric pressure and at 80 bars and cooling in the
CINOG HP facility.
vacancy volume, which enhances the diffusion of oxygen through the oxide [33].
The higher the oxidation temperature, the wider and longer the oxide crystallites.
On the one hand, the oxide crystallites formed on fresh Zy-4 in the 750 C
1000 C temperature range (in which the steam pressure has a significant effect on
the oxidation kinetics and the hydrogen pickup) are, on average, for given oxidation
time or weight gain, smaller and more disorganized and disoriented after oxidation
under high steam pressure. In addition the oxide grain boundaries are not as well
defined as in the oxide layers formed at atmospheric pressure (Fig. 17). Furthermore,
the oxides formed under high steam pressure appear less structured and exhibit
more numerous nano-pores and cracks (mainly circumferential, sometimes radial).
As confirmed by the differences noticed between the microstructures of oxides with
similar thicknesses formed at 1000 C under steam pressures of 1 and 80 bars (dif-
ferent oxidation times) on fresh Zy-4, the oxide does not grow in the same manner
under high steam pressure—at least for these specific conditions.
On the other hand, in the case of fresh M5, there is no significant difference of
morphology, porosity or damage between the oxide layers formed between 750 and
1200 C under 1 and 80 bars steam pressures (Fig. 18). In that case, only a few pores
and micro-cracks can be discerned and the oxide morphology is close to that
observed on Zy-4 samples oxidized at atmospheric pressure. Some nano-sized pre-
cipitates and cavities with similar dimensions (maybe due to nucleation by
precipitates-matrix decohesion) are observed on Zy-4 and to a lesser extent on M5
samples oxidized at 1000 C.
The oxide morphology does not depend on steam pressure between 1 and 80
bars at 1200 C. This is consistent with the insignificant effect of steam pressure on
the growth kinetics observed at this temperature. The crystallites are columnar, rel-
atively wide and long and well aligned perpendicularly to the metal/oxide interface,
with a few precipitates and pores (smaller than 100 nm).
Finally, one should keep in mind that the observations were made after cooling
and mechanical testing, during which some of the pores and cracks that have been
observed may have formed. Furthermore, some small pores and cracks may not be
easily detectable because of the appearance of the oxide microstructure, which is
sometimes “disordered.”
POST-COOLING MECHANICAL BEHAVIOR

RCTs were performed at 20 and 135 C on fresh, pre-hydrided and pre-oxidized
Zy-4 and M5 samples oxidized in the CINOG HP facility at 750, 850, 900, 1000,
1100, and 1200 C under steam pressures of 1, 40, and 80 bars. The results of this
testing are shown in Figs. 19 and 20 and are described in terms of maximal plastic
displacement (determined from load-displacement curves) normalized by a single
factor for illustration purposes. Maximal plastic displacement gives qualitative in-
formation about residual ductility. When the test was stopped before sample failure
(due to displacement and load limits), the reported mechanical properties are lower
than the actual material properties at failure (underestimation represented by
upwards arrows in Figs. 19 and 20). For comparison, the results of RCTs previously
performed after one-side oxidation at 1000, 1100, and 1200 C at atmospheric pres-
sure and direct quenching in water down to room temperature in the DEZIROX 1
facility [7,8,25] are also reported in those both figures. One should keep in mind
that the samples oxidized in the CINOG HP facility were not quenched (cooling
rate close to 1000 C/s expected) but swiftly cooled (cooling rate of about 100 C/s).
It has been noticed that the cooling scenario after oxidation at HT may have a
FIG. 19 Evolution as a function of the oxidation time and the ECR determined from
measured weight gain of the normalized maximal plastic displacement
measured during RCT at 20 and 135 C for fresh (a) Zircaloy-4 and (b) M5
samples oxidized under various steam pressures at 750, 850, 900, 1000, 1100,
and 1200 C and cooled in the CINOG HP facility. Comparison with results
previously obtained for the fresh materials after one-side oxidation at 1000,
1100, and 1200 C under atmospheric pressure and direct quenching in the
DEZIROX 1 (DEZ1) facility [7,8,25].
significant impact on the post-cooling/quench mechanical properties [34–37].

Therefore, the comparison between the results obtained in DEZIROX 1 (direct
water quenching) and those obtained in CINOG HP (fast cooling) facilities must be
done with caution. The hydrogen analyses previously discussed were not always
performed on the same samples that have been mechanically tested. Although prob-
ably very close, the hydrogen contents represented in Fig. 6 are thus not rigorously
those of the samples tested in compression.
FIG. 20 Evolution as a function of the oxidation time and the ECR determined from
measured weight gain at HT of the normalized maximal plastic displacement
measured during RCT at 135 C for fresh, pre-hydrided (600 wt. ppm for
Zircaloy-4 and 200 wt. ppm for M5) and pre-oxidized (3 lm thick pre-
existing oxide layer) (a) Zircaloy-4 and (b) M5 samples oxidized under various
steam pressures at 850 and 1000 C and cooled in the CINOG HP facility.
Comparison with results previously obtained for the pre-hydrided materials
(600 wt. ppm for Zircaloy-4 and 150 wt. ppm for M5) after one-side
oxidation at 1000 C under atmospheric pressure and direct quenching in the
DEZIROX 1 (DEZ1) facility [7,8].
FRESH MATERIALS
The results show that the impact of steam pressure on the post-cooling mechanical
properties evaluated from RCT at 135 C for fresh Zy-4 and M5 oxidized at 750 C
(18 000 s under 1 bar, 9000 s under 40 bars), 850 C (600 and 3600 s under 1 and 80
bars), 900 C (600 s under 1 and 80 bars) and 1200 C (400 s under 1 and 80 bars) is
limited (Fig. 19): for a given oxidation time and temperature, the mechanical
properties at 135 C after oxidation under steam pressures of 1, 40, or 80 bars are
relatively similar. For these specific oxidation conditions, the additional ECR (i.e.
thicker oxide layer, as oxygen diffusion through the metal is not modified) and the
hydrogen uptake that result from the oxidation under high steam pressure have
only a minor effect on the residual mechanical properties. For samples oxidized at
750, 850, and 900 C ductile fracture behavior is observed at the 135 C test
temperature.
The post-cooling ductility at 135 C of fresh Zy-4 oxidized 1800 s at 1000 C is
lower when the oxidation was performed under a steam pressure of 80 bars in com-
parison with an oxidation carried out at atmospheric pressure. This decrease in
ductility is, however, consistent with the higher ECR (thicker oxide layer) and the
hydrogen content of the sample oxidized under an 80 bars steam pressure (Fig. 6).
In the case of fresh M5, no significant effect of steam pressure on the post-cooling
residual ductility is observed. The residual ductility remains high. Regarding the
ECR, the post-cooling mechanical properties of fresh Zy-4 and M5 oxidized at
1000 C and tested by compression at 135 C are globally in fairly good agreement
with the results obtained after one-side oxidation under atmospheric steam pressure
and direct quenching in the DEZIROX 1 facility.
Fresh Zy-4 and M5 samples oxidized 400 s at 1200 C under steam pressures of
1 or 80 bars exhibit poor ductility and are nearly brittle under compression at
135 C due to their high ECR. These low values of residual ductility are consistent
with the tendencies deduced from RCT performed after one-side oxidation and
direct quenching in the DEZIROX 1 facility. The differences in ductility between
these samples are due to slightly different ECR, which is high in any case. After
400 s at 1100 C under a 80 bars steam pressure, fresh Zy-4 is quite ductile at 135 C,
as after oxidation at atmospheric pressure and direct quenching. Thus, the post-
cooling mechanical properties of fresh Zy-4 and M5 oxidized at 1100 and 1200 C
are not really affected by a higher steam pressure during the oxidation. This is con-
sistent with the low or even negligible effect of steam pressure on both oxygen diffu-
sion and hydrogen pickup at these temperatures.
The post-cooling ductility of the fresh samples oxidized 3600 s at 850 C is
lower at room temperature than at 135 C. This result is consistent with the effect of
the post-cooling testing temperature already observed after one-side oxidation at
atmospheric pressure between 1000 and 1250 C and direct quenching in the
DEZIROX 1 facility [7,8,25]. At room temperature, the post-cooling ductility of
fresh Zy-4 oxidized 3600 s at 850 C under a steam pressure of 80 bars is lower than
after oxidation under steam pressures of 1 and 40 bars. However, this decrease in
ductility, which is not observed for M5, is limited, since the maximal plastic dis-
placement remains quite high after oxidation at 80 bars. Nevertheless, it seems that
this drop in ductility is too significant to be only due to the additional ECR (thicker
oxide layers) resulting from the oxidation under high steam pressure because, as al-
ready discussed, the oxygen diffusion profile within the sub-oxide metallic layer is
the same for both steam pressure. Some metallurgical observations were hence
FIG. 21 SEM micrographs of fracture surfaces (secondary electrons) and transverse

cross-sections near the fracture surfaces (back-scattered electrons), EPMA X-
ray maps of chromium in the same region near the fracture surfaces and lLIBS
maps of hydrogen within the metallic layer (which was two-phase (aZr þ bZr) at
the oxidation temperature) of fresh Zircaloy-4 samples failed during RCT at
room temperature after oxidation during 3600 s at 850 C under steam
pressures of (a) 40 and (b) 80 bars.
carried out in order to investigate this decrease in ductility. It was observed that the
sub-oxide metallic part of Zy-4 samples oxidized 3600 s at 850 C is mostly in the
aZr phase with a rather continuous network of prior-bZr phase at aZr phase grain
boundaries (Fig. 21).
FIG. 22 Schematic of the partitioning of Fe, Cr, H, O and Sn between aZr and bZr grains
in Zircaloy-4 at 850 C (see Ref. [40]).
As iron and chromium are b-stabilizing chemical elements and thus concen-
trate into the bZr phase, the mean surface or volume fraction of the prior-bZr phase
around the aZr phase grain boundaries can be evaluated from EPMA Fe or Cr X-
ray mapping. A prior-bZr phase mean volume fraction of about 30 % was found.
This volume fraction is quite consistent with the equilibrium bZr phase fraction pre-
dicted for as-received low-tin Zy-4 by using the “Zircobase” thermodynamic data-
base (Thermo-Calc11 formalism) [38]. Numerous hydrides are observed in Zy-4
samples oxidized 3600 s at 850 C under steam pressures of 40 or 80 bars. This is
consistent with the previously discussed significant hydrogen uptake that occurs
during oxidation under these conditions (naturally, the samples oxidized at atmos-
pheric pressure do not show any hydride because of the absence of significant
hydrogen pickup). Hydrides are systematically localized within the prior-bZr phase
areas, due to the fact that hydrogen (as iron and chromium) is a b-stabilizing ele-
ment and it concentrates within the bZr phase at 850 C [39,40] (Fig. 22).
Thus, although it is quite high, the cooling rate after oxidation in the CINOG
HP facility is sufficiently low to allow the precipitation of hydrogen in the form of
coarse hydrides, which can be observed by optical or scanning electron microscopy.
Note that after direct quenching from HT down to room temperature, it is not pos-
sible to observe hydrides at these scales, since hydrogen may be retained in supersa-
turated solid-solution in the aZr phase or may precipitate as very fine hydrides, or
both [8]. The comparison of hydrogen and iron or chromium distributions mapped
by lLIBS and EPMA, respectively, confirms that hydrogen is located in the prior-
bZr phase of Zy-4 samples oxidized 3600 s at 850 C under steam pressures of 40
11
Thermo-Calc is a registered trademark of Thermo-Calc Software AB.
and 80 bars (Fig. 21). Examination of the fracture surfaces of the samples failed dur-
ing RCT at room temperature after oxidation during 3600 s at 850 C reveals a duc-
tile failure mode with fracture surfaces covered by transgranular dimples in the case
of the samples oxidized under 1 or 40 bars (Fig. 21). In contrast, a mixed brittle/duc-
tile failure mode with the presence of secondary cracks is highlighted after oxida-
tion under 80 bars. It was noticed that for this last sample, the main crack path
follows the prior-bZr skeleton with some obvious indications of secondary cracks—
probably nucleated at hydrides/matrix interfaces—located within this phase under
the failure surface. In that case, the rather continuous prior-bZr phase network
where hydrogen is preferentially located forms a quite continuous brittle path for
crack initiation and propagation. Indeed, if it is considered that all the hydrogen is
concentrated within the prior-bZr phase, and given the 30 % prior-bZr phase vol-
ume fraction and the 180 wt. ppm total hydrogen content measured for Zy-4
after oxidization during 3600 s at 850 C under 80 bars, local hydrogen contents
higher than 500 wt. ppm are expected within the prior-bZr phase regions.
This explains why these regions are more or less brittle at room temperature. In the
case of Zy-4 oxidized under 40 bars steam pressure, hydrides are too scarce to
form a preferential brittle path for cracking. This embrittlement mode highlighted
here for samples tested under compression at room temperature and specific to
materials treated into the two-phase (aZr þ bZr) temperature domain and contain-
ing hydrogen is quite different from the hydrogen embrittlement mechanisms
observed in materials oxidized at higher temperature in the bZr phase domain [8].
This embrittlement is, however, limited and was not observed for RCT performed
at 135 C.
PRE-HYDRIDED MATERIALS
For the same oxidation conditions, the post-cooling ductility of Zy-4 and M5 sam-
ples pre-hydrided at about 600 and 200 wt. ppm, respectively, and tested at 135 C is
lower than that of the fresh materials; as a result of the well-known embrittlement
effect of hydrogen (Fig. 20). However, Zy-4 and M5 samples pre-hydrided at about
600 and 200 wt. ppm, respectively, oxidized 600 s at 850 C or 200 and 400 s at
1000 C under steam pressures of 1 and 80 bars are ductile at 135 C. For similar ox-
idation conditions, M5 pre-hydrided at 200 wt. ppm is naturally more ductile than
Zy-4 pre-hydrided at 600 wt. ppm due to the lower initial hydrogen content and the
thinner oxide layers formed at 850 and 1000 C under a steam pressure of 80 bars.
There is no significant effect of steam pressure on the post-cooling mechanical
properties of the pre-hydrided materials.
For the same ECR, the post-cooling ductility of the samples two-side oxidized
in the CINOG HP facility seems to be slightly higher than the ductility of samples
with similar hydrogen contents one-side oxidized and directly quenched in the
DEZIROX 1 facility [7,8]. These small deviations may be due to slightly higher
hydrogen contents of the samples oxidized in DEZIROX 1 or a slower cooling dur-
ing the CINOG HP experiments, or both.
PRE-OXIDIZED MATERIALS
As shown in Fig. 20, Zy-4 and M5 samples with 3 lm thick pre-existing oxide layers
oxidized 600 s at 850 C and 400 s at 1000 C under steam pressures of 1 and 80 bars
and tested at 135 C are quite ductile. For the same oxidation time at a given tem-
perature, the post-cooling mechanical properties of the pre-oxidized samples are
not very different from those of the fresh materials, despite the lower weight gains
they have reached at HT. As already observed for oxidation tests performed at 1000
and 1200 C at atmospheric pressure on specimens with various pre-existing oxide
layers thicknesses between 10 and 35 lm, this is due to the fact that the overall
quantity of oxygen atoms, which diffuse into the sub-oxide metallic layer at HT,
does not depend on the “source of oxygen” (partial dissolution of the pre-oxide or
formation of a “fresh” oxide at HT) [23].
Discussion
MECHANISMS RESPONSIBLE FOR THE STEAM PRESSURE EFFECT
The results discussed in this paper confirm that the steam pressure effect depends
on the oxidation temperature and the alloy: the steam pressure has a strong effect
on the oxidation of fresh and pre-hydrided Zy-4 between 750 and 1000 C but only
a slight impact on the oxidation of M5 within this temperature range or at 1100 or
1200 C for both materials. Furthermore, the steam pressure effect on the oxidation
of Zy-4 is reduced in the presence of a pre-existing oxide layer (even as thin as a
few micrometer) and the differences between the sensitivities of Zy-4 and M5 to
steam pressure for the investigated conditions seem to decrease when the materials
are pre-oxidized. The mechanisms responsible for the steam pressure effect at HT
are not clearly identified so far. Most studies reported in the literature that deal
with the effect of steam pressure on the oxidation of zirconium alloys were carried
out at temperatures lower than 500 C. In this temperature range, the characteristics
(phase, precipitates…) of the metallic substrate and the oxide do not evolve so
much and are quite well-known. However, in the temperature domain of interest
for LOCA related studies, significant changes occur in both the metallic substrate
(phase transformation, intermetallic precipitates dissolution in Zy-4, Nb-enriched
bZr phase apparition in M5…) and the oxide layer (phase transformation…). As a
consequence, the mechanisms at the origin of the steam pressure effect maybe not
the same as at lower temperatures.
Main Oxidation Mechanisms

The steam pressure effect on the oxidation kinetics and the associated hydrogen
pickup seem to appear at the beginning of oxidation (less than 200 s at the oxida-
tion temperature), whereas the breakaway oxidation phenomenon occurs relatively
suddenly during the oxidation (e.g. after about 5000 s at 1000 C at atmospheric
pressure for Zy-4 and M5 [6,7,22]). The mechanisms inherent to the steam pressure
effect thus take place at the early stage of oxidation and keep up during further
oxidation.
It was shown that the rate exponent of the oxidation kinetics is not modified
under high steam pressure: the oxidation kinetics remain parabolic or sub-
parabolic depending on the temperature. Thus the oxidation mechanisms are not
fundamentally different under high steam pressure and atmospheric pressure (apart
from the post-breakaway oxidation regime), i.e. anion diffusion from the environ-
ment to the metal/oxide interface where the chemical reaction takes place. In addi-
tion, it was determined that the faster thickening of the oxide layer under high
steam pressure observed for Zy-4 is not associated with the additional diffusion of
oxygen through the metallic substrate (at 850 and 1000 C at least). Steam pressure
clearly has an influence on the reaction at the outer oxide/steam interface or on the
oxygen transport through the existing oxide, or both, but not on the oxygen diffu-
sion beneath the inner metal/oxide interface.
It was shown that both the steam pressure effect on the HT oxidation of Zy-4
and the differences of sensitivities to steam pressure of Zy-4 and M5 are reduced in
the presence of a pre-existing oxide (the thicker the pre-oxide, the stronger the
reduction). This suggests that steam pressure has an alloy dependent influence on
the microstructure (e.g., porosity and grain morphology) of the oxide formed at HT
(and thus on the oxygen diffusion through it) but not really on the oxygen transport
rate through the pre-existing scale. One can assume that the mechanisms responsi-
ble for the steam pressure effect (e.g., tetragonal to monoclinic zirconia phase trans-
formation in the oxide formed at HT), which presumably take place at the early
stage of oxidation, are inhibited (or delayed) in the presence of a pre-existing oxide
layer.
The hydrogen pickup fraction (fraction of hydrogen liberated during the oxida-
tion, which has penetrated and diffused through the oxide layer and has then been
dissolved into the metal) observed between 750 and 1000 C under high steam pres-
sure in the case of Zy-4 is relatively low, of the same order of magnitude as the hydro-
gen pickup fraction measured for Zy-4 under nominal conditions in PWR, i.e. about
10–20 %. Therefore, the oxide layer formed under high steam pressure in the temper-
ature range in which a hydrogen uptake is observed is still partially protective. If it
were not the case, the hydrogen pickup fraction would have been significantly higher
and the oxidation kinetics would not have been (sub)parabolic anymore, as in the
case of the post-breakaway oxidation transition regime for example during which the
hydrogen pickup fraction is of several tens of percents and the oxidation kinetics is
nearly linear.
Oxide Morphology, Porosity, and Cracking

The oxide layers formed on Zy-4 between 750 and 1000 C under high steam pres-
sure show in general more pores and cracks than the oxides formed at atmospheric
pressure (for the same oxidation time and temperature). These observations are
consistent with those of Pawel et al. [9] and Bramwell et al. [10]. According to the
theory proposed by Cox [41] for lower temperatures, the steam pressure effect at
HT is probably associated with a faster development of porosities and micro-cracks
within the oxides formed under high steam pressure. These pores and cracks would
make the access of oxygen (and hydrogen) atoms to the metallic substrate easier.
On the assumption that radial cracks are present within the oxide layer, steam pres-
sure may also have an influence on the penetration of steam along these cracks.
In addition, it was observed that the crystallites (mainly columnar) of the
oxides formed under high steam pressures on Zy-4 in the 750–1000 C temperature
range are, on average, smaller and more disorganized and disoriented with respect
to grain boundaries and not as well defined as in the oxide layers formed at atmos-
pheric pressure. Such a microstructure appears to promote the formation of pores
at grain boundaries. Porosity may induce local strains/stresses, which can lead to
the formation of cracks that would propagate along the grain boundaries. Oxygen
transport through the oxide occurs by both volume (i.e., bulk) and grain boundary
diffusion. However, diffusion along the grain boundaries is often expected to be
predominant, at least below 400 C The effective resistance to diffusion at the grain
boundaries and thus the corrosion resistance [42,43] may be reduced in the pres-
ence of fine grains and disoriented grains boundaries. It has already been observed
that steam pressure can have an effect on both zirconia grain growth rate and grain
size [44,45]. Under the effects of the compressive stresses that result from the
Pilling–Bedworth ratio and the volume increase induced by the potential tetragonal
to monoclinic phase transformation of zirconia, this faster grain growth under high
pressure may facilitate the decohesion of the crystallites. Decohesion would be eas-
ier when the crystallites are equiaxed rather than columnar [33].
Thus the results suggest that the oxide grains morphology and orientation are
modified under high steam pressure, so that steam diffusion and porosity formation
at grain boundaries and inter-granular cracking are facilitated.
Oxide Phases and Stoichiometry

It seems that the steam pressure effect is limited to temperatures lower than (or
close to) 1000 C, at which the tetragonal phase of zirconia is metastable [46]. Zirco-
nia is mainly tetragonal beyond 1000 C, while the monoclinic phase is thermody-
namically stable below this temperature. Nevertheless, the tetragonal phase can be
stabilized at lower temperatures in the presence of anion vacancies [47] or some
chemical or doping elements [48–50], high pressure [51,52], compressive stresses
[50,53] or small grains [48,54]. The tetragonal phase fraction of zirconia is generally
greatest near the metal/oxide interface, where the oxide grains are the smaller
[33,42,43] and the compressive stresses the higher [53,55,56], and diminishes dur-
ing the oxide growth [55,56]. Most existing references about stabilization of the zir-
conia tetragonal phase deal with oxidation temperatures lower than those
investigated in this paper, at which the metallic substrate is weaker. However, it is
to be noted that, contrary to the material oxidized at lower temperature, the metal-
lic substrate below the oxide layer formed at HT is constituted by aZr(O) phase
saturated with oxygen, which is known to exhibit a significant mechanical strength-

ening at HT in comparison to the metal with its nominal oxygen content [57]. Fur-
thermore, the higher temperatures investigated here are closer to the temperature
range in which the tetragonal zirconia phase is “naturally” stable (i.e. in the absence
of stabilizing factors), and lower compressive stress is needed to stabilize this phase.
On the basis of these results and as already postulated by Park et al. [11–15], one
can assume that steam pressure has an influence on the stability of the tetragonal
phase of zirconia or on the tetragonal to monoclinic phase transformation, or both.
The compressive stresses generated within the growing oxide layer due to the
high value (1.56) of the Pilling–Bedworth ratio between the molar volumes of zirco-
nia and zirconium are not well-known for HT oxidation, but one can expect that
they are quite high (several tens or hundreds of MPa) in comparison to the steam
pressures investigated in this paper (maximal steam pressure of 80 bars or 8 MPa).
These steam pressures are thus probably not sufficient to have a direct impact on
the stability of the zirconia phases. Furthermore, it has been shown that the oxida-
tion kinetics at HT depends on the partial steam pressure and not on the total pres-
sure [12]. Therefore, one can consider that the enhancement of the oxidation
kinetics under high steam pressure is due to physical and chemical changes rather
than mechanical (direct effect of stresses).
The HT oxidation and the pressure effect should not be related to the oxygen
chemical potential [58,59]. It was shown that the higher the steam pressure the
faster the tetragonal to monoclinic phase transformation of tetragonal zirconia
powder stabilized by Yttrium [60]. The potential destabilization of the tetragonal
phase of zirconia may result from a reduction of the surface energy (sensitive to
the chemical interactions with the atmosphere) induced by the high steam
pressures.
The tetragonal to monoclinic phase transformation of zirconia is associated
with strong lattice distortions and a significant volume expansion of about 3–4 %
[51], which can locally generate high stresses. Pure sub-stoichiometric (oxygen defi-
cient) zirconia can show superplastic properties at HT or during the monoclinic-
tetragonal phase transformation (martensitic like transition) [61], in which case no
significant damage of zirconia is expected. Thus, tetragonal sub-stoichiometric zir-
conia may be able to accommodate oxidation induced stresses. As a consequence,
one can assume that the tetragonal to monoclinic phase transformation occurs
quite naturally during the oxidation at atmospheric pressure, due in particular to
the progressive reduction of the compressive stresses in the oxide layer. This allows
the formation of a dense (i.e. low porosity) monoclinic and well-structured zirconia.
On the contrary, as postulated by Park et al. [14,15], if it is considered that the tet-
ragonal phase of zirconia is destabilized and the tetragonal to monoclinic phase
transformation is enhanced under high pressure, the tetragonal to monoclinic phase
transformation may occur more sharply under high steam pressure, with a poorer
stress accommodation. One can expect that in the case of Zy-4, (pseudo)plasticity
of the oxide layer is not sufficient to sustain the stresses induced by this sharp
transformation, which may lead to a less protective porous or micro-cracked mainly

monoclinic zirconia. Then, these structural changes generate additional stresses
which can induce a new phase transformation of the zirconia and the formation of
new pores (probably at grain boundaries) or even cracks within the oxide if its me-
chanical resistance is not sufficient.
Since the tetragonal to monoclinic transformation is partly responsible for the
formation of porosity within the oxide, the stronger sensitivity to steam pressure of
Zy-4 compared to M5 for the tested conditions may be due to a higher initial frac-
tion and a lower stability of the tetragonal phase of zirconia in the case of Zy-4, due
to the presence or absence of some alloying elements (tin and niobium in particu-
lar). In Zy-4, tin may tend to promote an increase of the oxygen vacancy concentra-
tion which favors the formation of tetragonal zirconia [48,62], whereas in M5,
niobium is expected to promote a decrease in oxygen vacancy concentration result-
ing in a lower tetragonal zirconia content [42,56,63]. It has already been suggested
that the lower the tetragonal phase fraction in zirconia, the better the corrosion re-
sistance of zirconium alloys under water or oxygen within the 360 C470 C tem-
perature range [42,48,56,63]. Baek et Jeong [64] observed that the lower the tin
content, the better the resistance to breakaway oxidation at 1000 C under steam at
atmospheric pressure and proposed that the effect was related to the destabilization
of the tetragonal phase in zirconia and a faster cracking of the oxide layer in the
presence of tin. The influence of these alloying elements on the stabilization of the
tetragonal phase of zirconia probably does not rely on a direct effect but rather on
the resulting oxygen vacancies concentration within the oxide [65,66].
While the oxide layers of most of the samples oxidized at HT within the present
study are black or dark grey, some Zy-4 samples show slightly different colors. For
example, in the case of fresh Zy-4 samples oxidized 3600 s at 850 C under a steam
pressure of 80 bars, the oxide exhibits a whitish/light grey appearance on the outer
surface. However, for the oxides formed at 1 bar for a similar oxidation time and tem-
perature the oxide appears black/dark grey. This color change, which affects the outer
surface of the oxide but not necessarily its bulk, is often associated with the oxide stoi-
chiometry: sub-stoichiometric (by oxygen deficiency) zirconia formed at HT is gener-
ally black while stoichiometric zirconia tends to be more whitish (or pinkish or
yellowish, depending on the alloy). Therefore, the steam pressure effect observed in
some specific conditions may be related to oxide stoichiometry. Indeed, the oxide
stoichiometry (oxygen vacancies) may influence the diffusion of atoms through the
oxide, the stability of the zirconia tetragonal phase or the oxide plasticity, or both.
Stresses in the Oxide Layer and Influence of the Metallic Substrate

If the assumption about the tetragonal to monoclinic transformation promoted
under high steam pressure is correct, higher compressive stresses are expected
within the oxide formed under high stream pressure due to the distortions and the
volume expansion associated with the phase transformation, but it is provided that
the oxide is able to sustain this transformation without significant damage. The
results show that the oxide formed on Zy-4 between 750 and 1000 C has more mo-
res and cracks when the oxidation was performed under high steam pressure,
resulting in a faster oxygen transport through the oxide and an accelerated oxida-
tion kinetics. In the case of M5, no additional porosity or cracking was observed
under high steam pressure, possibly due to a lower initial volume fraction of tetrag-
onal zirconia or a greater (visco)plasticity of the metallic substrate and the oxide
layer, or both. In that case, one can expect that oxygen diffusion through the oxide
layer and the resulting oxidation kinetics are not be significantly modified. It can
even be assumed that the higher compressive stresses expected under high steam
pressure would make the oxygen diffusion through the oxide harder and the oxida-
tion kinetics slower. Indeed it has been observed in yttria-stabilized zirconia [66] that
the O2 ion conductivity decreases with increasing compressive strain between 600
and 800 C. This decline in conductivity under compressive strain/stress conditions is
considered to be due to a decrease in oxygen vacancy mobility (independently of their
concentration), as a result of clustering of oxygen vacancies associated with changes
in the distance between ions. This may thus explain the slight decrease in weight gain
with increasing steam pressure observed at 900, 1000, and 1100 C for M5.
The level of compressive stress developed during the oxide growth depends on
the oxide plastic deformation but also on the creep of the metallic substrate (i.e. its
ability to accommodate the stresses). The weaker the creep resistance of the sub-
strate is, the lower the compressive stresses within the oxide. Thus, it cannot be
excluded that the steam pressure effect is related to the features of the metallic sub-
strate, in particular at the early beginning of the oxidation process.
The steam pressure effect appears to be the most important within the
750–1000 C temperature range, over which the metallic substrate is in aZr phase or
two-phase (aZr þ bZr) [39]. No significant effect of steam pressure was observed in
the bZr phase temperature domain (>1000 C). The aZr/bZr phase transition occurs
at lower equilibrium temperature in M5 than in Zy-4, so that the bZr phase fraction
at a given temperature is higher in M5. This may contribute to the lower impact of
steam pressure observed for M5. However, the equilibrium temperatures of the aZr/
bZr phase transition decrease with increasing the hydrogen content [39], so that for
example they are not very different for M5 with less than 200 wt. ppm hydrogen
and for Zy-4 with 600 wt. ppm hydrogen. It was determined that pre-hydriding
does not have any significant impact on the effect of steam pressure. Therefore, the
lower sensitivity to steam pressure of M5 should not be simply and solely related to
a different (aZr þ bZr) mix. For example, within the 750–1000 C temperature do-
main, dissolution of some intermetallic precipitates is expected to occur in Zy-4
and Nb-enriched bZr phase appear in M5. These evolutions of the characteristics
and the resulting properties of the metallic substrate presumably have an influence
on the properties of the oxide layer and the induced stresses.
The viscoplastic properties of the metallic substrate and therefore the stresses
within the oxide are probably not exactly the same in oxidized Zy-4 and M5 alloys.
For example, Kaddour et al. [67] determined that in the absence of oxidation effect,
Zr-1 %Nb-O alloy is a little less resistant to creep (faster creep rate) than Zy-4 in
the aZr and (aZr þ bZr) phases domains but more resistant in the bZr phase domain.
However, it is to be noted that this tendency may be different for higher oxygen
concentration within the metal. Furthermore, due to the slow thermal diffusion of
Nb, the aZr(O) phase layer beneath the oxide layer in M5 oxidized at HT shows a
lamellar (or layered) structure (“eutectoid-like”) composed of alternating layers of
aZr(O) phase depleted from niobium and bZr phase enriched with Nb (several wt.
%), The partition of oxygen is the opposite of the Nb distribution [25]. As suggested
by the results reported in Ref. [68], the mechanical properties of niobium-enriched/
oxygen-depleted bZr phase domains and niobium-depleted/oxygen-enriched zones
are probably not the same. This particular microstructure, which should have an
effect on stresses for example, may also contribute to the lower impact of steam
pressure on M5, in comparison to Zy-4 which shows a “continuous” aZr(O) phase
layer without bZr phase platelets.
A scalloped metal/oxide interface was observed in the samples oxidized at 850
and 1000 C. This undulation could result from heterogeneities of chemical compo-
sition, crystallographic orientation, phase, or stoichiometry. Furthermore, the
undulation of the metal/oxide interface may generate heterogeneous stresses within
the oxide scale [69], which may locally stabilize or destabilize one or the other phase
of zirconia. The undulation of the metal/oxide interface may also be associated with
variations of oxygen diffusion through the oxide and may be enhanced by all the
previously discussed phenomena.
Finally, it must be recalled that most of the mechanisms evoked in this discussion
were reported or postulated in the literature for oxidation temperatures lower than those
investigated in this study, for which the mechanisms may not be exactly the same.
STEAM PRESSURE EFFECT CONSEQUENCES DURING IN-REACTOR

TRANSIENTS: PRELIMINARY ANALYSIS
During an intermediate break LOCA (IB-LOCA), unlike during large break LOCA,
the core is uncovered at a relatively high primary pressure. The uncovering of the
core begins with a primary pressure that is lower than the secondary circuit equilib-
rium pressure, which is around 70 and 90 bars for the three loops 900 MWe and
the four loops 1300 MWe reactor fleets, respectively. Core reflood is achieved by ac-
cumulator progressive injection that begins at about 40 bar for all reactor types.
A significant hydrogen pickup was observed for low tin Zy-4 oxidized under
high steam pressure between 750 and 1000 C. However, the hydrogen pickup
remains limited for most tested conditions. A hydrogen uptake up to about 300 wt.
ppm was measured under 80 bars steam pressure but for oxidation times at HT far
longer than those expected during an IB-LOCA (a few tens or possibly a few hun-
dreds of seconds). For all of the investigated oxidation temperature and steam pres-
sure, it was shown that hydrogen pickup is always lower than 100 wt. ppm for
exposure times up to at least 400 s. As a consequence, a very low (a few tens of
wt. ppm) or insignificant hydrogen pickup is expected under IB-LOCA conditions,

so that it should not have consequence on cladding acceptance criteria. It was shown
that the oxidation rate of fresh or pre-hydrided low-tin Zy-4 oxidized within the
750 C 1000 C temperature range is enhanced under high steam pressures and may
become slightly faster than the prediction of the Baker–Just correlation. A prelimi-
nary analysis of the impact of the effect of steam pressure on the ECR actually
achieved during representative in-reactor transients was performed. Two IB-LOCA
transients corresponding to fuel with Zy-4 cladding were studied. The first case corres-
ponds to a 10 cm break with a delayed primary pump trip 9 min after the opening of
the break in a three loop 900 MWe PWR. The second case corresponds to a 10 cm
break with a delayed primary pump trip 13 min after the opening of the break in a
four loop 1300 MWe PWR. These two transients have been calculated with the deter-
ministic realistic methodology for IB-LOCA based on the CATHARE code. An older
and more conservative version of the model was used for the second transient, which
leads to a higher Peak Cladding Temperature (PCT ¼ 1115 C). Primary pressure and
hot spot cladding temperature values are given in Table 2 for both transients.
The faster oxidation kinetics within the 750–1000 C temperature range of low-
tin Zy-4 due to the high steam pressure reached during IB-LOCA transients was
taken into account by applying multiplicative factors to the Baker–Just correlation.
For each combination of steam pressure and temperature studied, the enhancement
factors were determined from the measured weight gains for the shortest oxidation
times investigated (which are the more representative of IB-LOCA transient condi-
tions) by linear interpolations for intermediate steam pressure and temperature val-
ues. The multiplicative factors were applied to the Baker–Just correlation when the
measured weight gains were underpredicted (i.e., the enhancement factors eval-
uated for a 50 bars steam pressure have been applied until the primary pressure has
been reduced to 40 bars), but not applied when the Baker–Just correlation overpre-
dicted the measured weight gains (i.e., for steam pressures lower than 40 bars or ox-
idation temperatures higher than 1100 C typically).
After cladding burst, the inner-side oxidation including the potential high
steam pressure-induced enhancement was taken into account. For the first tran-
sient, the Baker–Just weight gain rate was enhanced by 40 % up to 800 C and then
by 10 % up to 875 C. Beyond this last temperature, no enhancement factor was
applied (oxidation tests performed at 900 C under a 40 bars steam pressure have
TABLE 2 Primary pressure and hot spot cladding temperature values for the two reactor transients
considered for the preliminary analysis of the impact of the steam pressure effect.
Case 1: PCT ¼ 1020 C Case 2: PCT ¼ 1115 C

Transient Burst at 880 C Burst at 900 C
Primary pressure (bars) 50 40 30 70 60 50 40

Hot spot cladding temperature ( C) 660 880 1020 600 760 950 1110
FIG. 23 Additional weight gains resulting from the steam pressure effect in comparison
to the weight gains calculated from the Baker–Just correlation for two typical
in-reactor transients corresponding to fuel with Zy-4 cladding (Table 2).
shown that the Baker–Just correlation remains conservative). For the second tran-
sient, the Baker–Just weight gain rate was enhanced by about 45 % up to 800 C. A
decreasing enhancement factor with increasing temperature was then considered
up to 1050 C, above which no enhancement factor was applied (the Baker–Just cor-
relation remains conservative at 1100 C for steam pressures up to 80 bars).
For each transient, the maximum ECR value was calculated from the tabulated
time evolution (1 s time step) of the maximum cladding temperature. The ECR values
(taking burst strain into account) obtained with the unmodified Baker–Just correla-
tion are 5.9 % and 8.3 % for transients 1 and 2, respectively. The additional weight
gains resulting from the steam pressure effect in comparison to the weight gains cal-
culated from the Baker–Just correlation are shown in Fig. 23 for both in-reactor tran-
sients. The results show that the steam pressure effect on the ECR actually achieved is
very low or negligible: the final values of the additional ECR due to the steam pres-
sure effect are about 0.015 % and 0.12 % for transients 1 and 2, respectively. Indeed,
the cladding oxidation is low or negligible at temperatures for which the steam pres-
sure effect is significant and the cumulated penalty on ECR is quickly reduced
because of the nearly parabolic character of the oxidation kinetics.
Conclusions
The effect of steam pressure from 1 to 80 bars on the oxidation of stress-relieved
annealed low-tin Zircaloy-4 and M5 alloys in the 750–1200 C temperature range
was investigated. It was shown that the oxidation kinetics of as-fabricated Zircaloy-
4 is enhanced under high steam pressure between 750 and 1000 C, since the early
stages of oxidation (the dependency of weight gain on steam pressure is nearly ex-
ponential). The effect is quite low at those temperatures for as-fabricated M5, and
at 1100 and 1200 C for both alloys. Steam pressure potentially has an effect on the
oxidation rate, but does not substantially modify the sub-parabolic or parabolic oxi-
dation kinetics.
In the case of low-tin Zircaloy-4, a significant hydrogen uptake is associated
with the enhancement of the oxidation kinetics under high steam pressure observed
for temperatures between 750 and 1000 C. It was, however, limited for the investi-
gated conditions. The hydrogen pickup fraction is expected to be lower than 10 %.
No significant hydrogen pickup is noted for the M5 alloy, whatever the steam pres-
sure for the investigated conditions.
The oxidation kinetics, the hydrogen uptake, and the effect of steam pressure at
850 and 1000 C are not significantly modified by pre-hydriding (by gaseous charg-
ing) up to the end-of-life hydrogen content expected for each material.
For the same oxidation conditions at 850 and 1000 C, the oxidation weight
gains of pre-oxidized materials (dense pre-existing oxide layers with thin thick-
nesses ranging from 3 up to 8 lm formed in autoclave at 360 C with typical PWR
primary water chemistry) are lower than those of fresh or pre-hydrided samples.
The thicker the pre-oxide layer is, the greater the decrease in weight gain. The steam
pressure effect on the oxidation of Zircaloy-4 is reduced in the presence of a homo-
geneous pre-existing oxide layer (even thin, i.e., a few micrometers) and the differ-
ences between sensitivities of Zircaloy-4 and M5 to steam pressure for the tested
conditions are significantly reduced after pre-oxidation. These results would have
to be confirmed for thicker and less homogeneous pre-existing oxide layers.
Metallurgical observations have shown that the enhanced oxidation kinetics of
Zircaloy-4 under high steam pressure is correlated with a thickening of the oxide
layers formed at HT but is not associated with additional oxygen diffusion through
the metallic substrate. In the case of Zircaloy-4, steam pressure has an influence on
the microstructure of the oxide formed in the 750 C 1000 C temperature range:
the oxides formed under high steam pressure appear less structured and exhibit
more numerous pores and cracks. In addition, the columnar crystallites seem to be
smaller and more disorganized and disoriented with respect to grain boundaries rel-
ative to the oxide scales formed at atmospheric pressure.
The more disrupted oxide is probably responsible for faster oxygen (and hydro-
gen) transport through the oxide layer. This may result from a greater potential
destabilization of the tetragonal zirconia phase under high steam pressures. Thus,
the stronger sensitivity to steam pressure of Zircaloy-4 compared to M5 for the
investigated conditions may be related to the higher volume fraction and the lower
stability of the tetragonal zirconia phase in the oxide formed on Zircaloy-4 and/or
to a lower capacity of this oxide to accommodate without significant damage to the
stresses induced by the tetragonal to monoclinic phase transformation. The higher
tetragonal zirconia phase volume fraction is consistent with alloy composition and its
expected effect on the oxygen vacancy concentration in the oxide. However, further
investigation is needed to improve the understanding of the steam pressure effect and
its alloy dependence.
Ring compression tests performed at room temperature and 135 C showed
that the effect of steam pressure on the post-cooling mechanical properties of fresh,
pre-hydrided or pre-oxidized samples oxidized at HT is generally limited for the
conditions investigated. The additional oxidation and hydrogen uptake that result
in the case of Zircaloy-4 from the oxidation under high steam pressure have mostly
only a minor effect on the residual ductility, which remains significant for most of
the samples tested under compression. However, a specific embrittlement mode
was observed at room temperature for some Zircaloy-4 samples oxidized at 850 C.
At this temperature, the alloy microstructure is comprised of a two phase micro-
structure (aZr þ bZr) and allows from the absorption of significant amounts of
hydrogen at HT under high steam pressure, resulting in the formation of a nearly
continuous brittle hydrogen enriched prior-bZr phase network along which crack
initiation and propagation will preferentially occur. This embrittlement mode is,
however, limited and was not observed for RCT performed at 135 C.
Finally, it is of note that for low-tin Zircaloy-4, the impact of steam pressure on
the ECR and hydrogen pickup of the cladding during typical in-reactor transients is
very low or negligible and has hence no significant consequence on LOCA cladding
acceptance criteria.
ACKNOWLEDGMENTS
The writers would like to thank S. Paradowski, C. Cobac, J. L. Flament, E. Rouesne, S.
Urvoy, V. Rabeau, D. Hamon, P. Bonnaillie, K. Béranger, and J. L. Lacour from CEA for
their contribution to this work. The writers are also grateful to EDF R&D for perform-
ing the pre-oxidations and CEZUS Paimboeuf for hydrogen charging and analysis.
References
[1] Baker, L. and Just, L. C., “Studies of Metal-Water Reactions at High Temperatures. III. Ex-
perimental and Theoretical Studies of the Zirconium-Water Reaction,” ANL-6548,
Argonne National Laboratory, Lemont, IL, 1962.
[2] Cathcart, J. V., Pawel, R. E., McKee, R. A., Druschel, R. E., Yurek, G. J., Campbell, J. J., and
Jury, S. H., “Zirconium Metal-Water Oxidation Kinetics IV. Reaction Rate Studies,” ORNL/
NUREG-17, US Nuclear Regulatory Commission, Washington, D.C., 1977.
[3] Kawasaki, S., Furuta, T., and Suzuki, M., “Oxidation of Zircaloy-4 Under High Tempera-
ture Steam Atmosphere and its Effect on Ductility of Cladding,” J. Nucl. Sci. Technol.,
Vol. 15, 1978, pp. 589–596.
[4] Leistikow, S. and Schanz, G., “Oxidation Kinetics and Related Phenomena of Zircaloy-4
Fuel Cladding Exposed to High Temperature Steam and Hydrogen-Steam Mixtures
under PWR Accident Conditions,” Nucl. Eng. Des., Vol. 103, 1987, pp. 65–84.
[5] Schanz, G., Adroguer, B., and Volchek, A., “Advanced Treatment of Zircaloy Cladding
High-Temperature Oxidation in Severe Accident Code Calculations: Part I. Experimental
Database and Basic Modelling,” Nucl. Eng. Des., Vol. 232, 2004, pp. 75–84.
[6] Billone, M., Yan, Y., Burtseva, T., and Daum, R., “Cladding Embrittlement During Postulated
Loss-of-Coolant Accidents,” NUREG/CR-6967, Argonne National Laboratory, Lemont, IL, 2008.
[7] Portier, L., Bredel, T., Brachet, J. C., Maillot, V., Mardon, J. P., and Lesbros, A., “Influence
of Long Service Exposures on the Thermal-Mechanical Behaviour of Zy-4 and M5 Alloys
in LOCA Conditions,” J. ASTM Int., Vol. 2, No. 2, 2005, JAI12468.
[8] Brachet, J. C., Vandenberghe-Maillot, V., Portier, L., Gilbon, D., Lesbros, A., Waeckel, N.,
and Mardon, J. P., “Hydrogen Content, Preoxidation, and Cooling Scenario Effects on
LOCA Conditions,” J. ASTM Int., Vol. 5, 2008, JAI101116.
[9] Pawel, R. E., Cathcart, J. V., and Campbell, J. J., “The Oxidation of Zircaloy-4 at 900 and
1100 C in High Pressure Steam,” J. Nucl. Mater., Vol. 82, 1979, pp. 129–139.
[10] Bramwell, I., Haste, T., Worswick, D., and Parsons, P., “An Experimental Investigation into the
Oxidation of Zircaloy-4 at Elevated Pressures in the 750 to 1000 C Temperature Range,”
Proceedings of the 10th International Symposium on Zirconium in the Nuclear Industry,”
[11] Park, K., Kim, K. P., and Whang, J. H., “Pressure Effects on High Temperature Zircaloy-4
Oxidation in Steam,” Proceedings of the International Topical Meeting on Light Water
Reactor Fuel Performance, Park City, UT, April 10–13, 2000.
[12] Park, K., Kim, K. P., Yoo, T. G., and Kim, K. T., “Pressure Effects on High Temperature
Steam Oxidation of Zircaloy-4,” Metals Mater. Int., Vol. 7, 2001, pp. 367–373.
[13] Park, K., Yang, S., and Kim, K., “Nb Effect on Zr-alloy Oxidation Under High Pressure
Steam at High Temperatures,” Proceedings of the Water Reactor Fuel Performance
Meeting, Kyoto, Japan, Oct 2–6, 2005, pp. 811–826.
[14] Park, K., Yang, S., and Ho, K., “Steam Pressure Effects on Zr-alloy Oxidation at High Tem-
peratures,” Proceedings of the 2011 Water Reactor Fuel Performance Meeting, Chengdu,
P. R. C., Sept 11–14, 2011.
[15] Park, K., Yang, S., and Ho, K., “The Effect of High Pressure Steam on the Oxidation of
Low-Sn Zircaloy-4 at Temperatures Between 700 and 900 C,” J. Nucl. Mater., Vol. 420,
2012, pp. 39–48.
[16] Vrtilkova, V., Valach, M., and Molin, L., “Oxiding and Hydriding Properties of Zr-1Nb Clad-
ding Material in Comparison With Zircaloys,” Proceedings of the Technical Committee
Meeting on Influence of Water Chemistry on Fuel Cladding Behaviour, Rez, Czech
Republic, Oct 4–8, 1993, pp. 227–250.
[17] Mardon, J. P., Charquet, D., and Senevat, J., “Influence of Composition and Fabrication
Process on Out-of-Pile and In-Pile Properties of M5 Alloy,” Proceedings of the 12th Inter-
national Symposium on Zirconium in the Nuclear Industry, ASTM STP 1354, ASTM Inter-
national, West Conshohocken, PA, 2000, pp. 505–524.
[18] Bossis, P., Pêcheur, D., Hanifi, K., Thomazet, J., and Blat, M., “Comparison of the High Burn-
up Corrosion on M5 and Low Tin Zircaloy-4,” J. ASTM Int., Vol. 3, No. 1, 2005, 12404.
[19] Mardon, J. P., Garner, G., and Hoffmann, P. B., “M5 a Breakthrough in Zr Alloy,” Proceed-
ings of the International Topical Meeting on Light Water Reactor Fuel Performance, Or-
lando, FL, Sept 26–29, 2010, pp. 577–586.
[20] Aomi, M., Nakatsuka, M., Komura, S., Hirose, T., and Anegawa, T., “Behavior of BWR Fuel
Cladding Tubes Under Simulated LOCA Conditions,” Proceedings of the 7th Interna-
tional Conference on Nuclear Engineering, Tokyo, Japan, April 19–23, 1999.
[21] Grandjean, C., and Hache, G., “A State-of-the-Art Review of Past Programmes Devoted
to Fuel Behaviour Under Loss-of-Coolant Conditions. Part 3. Cladding Oxidation. Resist-
ance to Quench and Post-Quench Loads,” IRSN SEMCA 2008-093, IRSN, Saint Paul-lez-
Durance Cedex, France, 2008.
[22] Vandenberghe, V., Brachet, J. C., Le Saux, M., Gilbon, D., Mardon, J. P., and Sebbari, B.,
“Sensitivity to Chemical Composition Variations and Heating/Oxidation Mode of the
Breakaway Oxidation in M5 Cladding Steam Oxidized at 1000 C (LOCA Conditions),”
Proceedings of TopFuel 2012, Manchester, UK, Sept 2–6, 2012.
[23] Le Saux, M., Brachet, J. C., Vandenberghe, V., Gilbon, D., Mardon, J. P., and Sebbari, B.,
“Influence of Pre-Transient Oxide on LOCA High Temperature Steam Oxidation and
Post-Quench Mechanical Properties of Zircaloy-4 and M5 Cladding,” Proceedings of the
2011 Water Reactor Fuel Performance Meeting, Chengdu, P. R. C., Sept 11–14, 2011.
[24] Cabrera, A., Vandenberghe, V., Besson, J., Le Saux, M., Brachet, J. C., Mardon, J. P., and
Hafidi, B., “Finite Element Modeling of Ring Compression Tests on Post-Quenched Single
Side Oxidized Zirconium-Based Alloys (LOCA Conditions),” Proceedings of the Interna-
tional Topical Meeting on Light Water Reactor Fuel Performance, Orlando, FL, Sept
26–29, 2010, pp. 261–269.
[25] Brachet, J. C., Pelchat, J., Hamon, D., Maury, R., Jacques, P., and Mardon, J. P.,
“Mechanical Behavior at Room Temperature and Metallurgical study of Low-Tin Zy-4
and M5 (Zr-NbO) Alloys After Oxidation at 1100 C and Quenching,” Proceedings of the
Technical Committee Meeting on Fuel Behaviour Under Transient and LOCA Conditions,
IAEA-TECDOC-1320, Halden, Norway, Sept 10–14, 2001, pp. 139–158.
[26] Mauchien, P., Beranger, K., Caron, N., Gosmain, L., Lacour, J. L., Terlain, A., and Hamon,
D., “LIBS Microprobe for Elemental Mapping in Solids at the Micron Scale,” Proceedings
of the Sixth Laser Induced Breakdown Spectroscopy 2010 (LIBS 2010), Memphis TN,
Sept 13–17, 2010.
[27] Nagase, F., Otomo, T., and Uetsuka, H., “Oxidation Kinetics of Low-Sn Zircaloy-4 at the
Temperature Range from 773 to 1573 K,” J. Nucl. Sci. Technol., Vol. 40, 2003, pp. 213–219.
[28] Baek, J. H., Park, K. B., and Jeong, Y. H., “Oxidation Kinetics of Zircaloy-4 and Zr-1Nb-
1Sn-0.1Fe at Temperatures of 700–1200 C,” J. Nucl. Mater., Vol. 335, 2004, pp. 443–456.
[29] Kim, W. Y., Kang, C. G., and Lee, S. M., “Effect of Viscosity on Microstructure Characteris-
tic in Rheological Behaviour of Wrought Aluminium Alloys by Compression and Stirring
Process,” Vol. 26, No. 1, 2010, pp. 20–23.
[30] Brachet, J. C. and Vandenberghe, V., “Comments to Papers of J. H. Kim et al. [1] and M.
Große et al. [2] Recently Published in JNM ‘On the Hydrogen Uptake of Zircaloy-4 and
M5 Alloys Subjected to Steam Oxidation in the 1100-1250 C temperature range,’” J.
Nucl. Mater., Vol. 395, 2009, pp. 169–172.
[31] Baek, J. H. and Jeong, Y. H., “Steam Oxidation of Zr-1.5Nb-0.4Sn-0.2Fe-0.1Cr and

Zircaloy-4 at 900-1200 C,” J. Nucl. Mater., Vol. 361, 2007, pp. 30–40.
[32] Cox, B., Oxidation of Zirconium and its Alloys, Advances in Corrosion Science and Tech-
nology, M. Fontana and R. W. Staehle, Eds., Plenum Press, New York, 1976.
[33] IAEA, “Waterside Corrosion of Zirconium Alloys in Nuclear Power Plants,” IAEA-TEC-
DOC-996, IAEA, Vienna, Austria, 1998.
[34] Komatsu, K., Takada, Y., Mizuta, M., and Takahashi, S., “The Effects of Oxidation Temper-
ature and Slow-Cooldown on Ductile-Brittle Behavior of Zircaloy Fuel Cladding,” Pro-
ceedings of the Specialist Meeting on the Behavior of Water Reactor Fuel Elements
under Accident Conditions, Spatind, Norway, Sept 13–19, 1976.
[35] Chung, H. M. and Kassner, T. F., “Embrittlement Criteria for Zircaloy Fuel Cladding Appli-
cable to Accident Situations in Light-Water Reactors: Summary Report,” NUREG/CR-
1344, Argonne National Laboratory, Lemont, IL, 1980.
[36] Vandenberghe, V., Brachet, J. C., Le Saux, M., Gilbon, D., Mardon, J. P., and Hafidi, B.,
“Influence of the Cooling Scenario on the Post-Quench Mechanical Properties of Pre-
Hydrided Zircaloy-4 Fuel Claddings After High Temperature Steam Oxidation (LOCA
Conditions),” Proceedings of the International Topical Meeting on Light Water Reactor
Fuel Performance, Orlando, FL, Sept 26–29, 2010, pp. 270–277.
[37] Brachet, J. C., Hamon, D., Béchade, J. L., Forget, P., Toffolon-Masclet, C., Raepsaet, C., Mar-
don, J. P., and Sebbari, B., “Quantification of the Chemical Elements Partitioning Within
Pre-Hydrided Zircaloy-4 After High Temperature Steam Oxidation as a Function of the
Final Cooling Scenario (LOCA Conditions) and Consequences on the (Local) Materials
Hardening,” Proceedings of the IAEA Technical Meeting on Fuel Behaviour and Modelling
Under Severe Transient and LOCA Conditions, Mito-city Ibaraki-ken, Japan, Oct 19–21, 2011.
[38] Dupin, N., Ansara, I., Servant, C., Toffolon, C., Lemaignan, C. and Brachet, J. C., “A Ther-
modynamic Database for Zirconium Alloys,” J. Nucl. Mater., Vol. 275, 1999, pp. 287–295.
[39] Brachet, J. C., Portier, L., Forgeron, T., Hivroz, J., Hamon, D., Guilbert, T., Bredel, T., Yvon,
P., Mardon, J. P., and Jacques, P., “Influence of Hydrogen Content on the a/b Phase
Transformation Temperatures and on the Thermal-Mechanical Behavior of Zy-4, M4
(ZrSnFeV), and M5 (ZrNbO) Alloys During the First Phase of LOCA Transient,” J. ASTM
Int., 2002, STP11411S.
[40] Brachet, J. C., Toffolon-Masclet, C., Hamon, D., Guilbert, T., Trego, G., Jourdan, J., Stern,
A., and Raepsaet, C., “Oxygen, Hydrogen and Main Alloying Chemical Elements Parti-
tioning Upon Alpha-Beta Phase Transformation in Zirconium Alloys,” Solid State Phe-
nom., Vol. 172–174, 2011, pp. 753–759.
[41] Cox, B., “Accelerated Oxidation of Zircaloy-2 in Supercritical Steam,” Report AECL-4448,
Atomic Energy of Canada, Chalk River, ON, 1973.
[42] Park, J. Y., Kim, H. G., Jeong, Y. H., and Jung, Y. H., “Crystal Structure and Grain Size of Zr
Oxide Characterized by Synchrotron Radiation Microdiffraction,” J. Nucl. Mater., Vol.
335, 2004, pp. 433–442.
2006, pp. 265–281.
[44] Méthivier, A., 1992, “Etude Expérimentale et Théorique de l’Evolution Texturale et Struc-
turale de Poudres de Zircone Pures et Dopeées,” Ph.D. thesis, Institut National Polytech-
nique de Grenoble, France.
[45] Dali, Y., Tupin, M., Bossis, P., Pijolat, M., Wouters, Y., and Jomard, F., “Corrosion Kinetics
Under High Pressure of Steam of Pure Zirconium and Zirconium Alloys Followed by In
Situ Thermogravimetry,” J. Nucl. Mater., Vol. 426, 2012, pp. 148–159.
[46] Gosset, D., Le Saux, M., Simeone, D., and Gilbon, D., “New Insights in Structural Charac-
terisation of Zirconium Alloys Oxidation at High Temperature,” J. Nucl. Mater., Vol. 429,
2012, pp. 19–24.
[47] Lu, X., Liang, K., Gu, S., Zheng, Y., and Fang, H., “Effect of Oxygen Vacancies on Transfor-
mation of Zirconia at Low Temperatures,” J. Mater. Sci., Vol. 32, 1997, pp. 6653–6656.
During 400 C Autoclave Tests,” J. Nucl. Mater., Vol. 226, 1995, pp. 34–43.
[49] Harada, M., Kimpara, M., and Abe, K., “Effect of Alloying Elements on Uniform Corrosion
Resistance of Zirconium-based Alloys in 360 C Water and 400 C Steam,” Proceedings
of the 9th International Symposium on Zirconium in the Nuclear Industry, ASTM STP 1132,
ASTM International, West Conshohocken, PA, 1991.
[50] Beie, H., Mitwalsky, A., Garzarolli, F., Ruhmann, H., and Sell, H., “Examinations of the Cor-
rosion Mechanism of Zirconium Alloys,” Proceedings of the 10th International Sympo-
sium on Zirconium in the Nuclear Industry, ASTM STP 1245, ASTM International, West
Conshohocken, PA, 1994.
[51] Kisi, E. H. and Howard, C. J., “Crystal Structures of Zirconia Phases and Their Inter-
Relation,” Key Eng. Mater., Vol. 153–154, 1998, pp. 1–36.
[52] Bouvier, P., Godlewski, J., and Lucazeau, G., “A Raman Study of the Nanocrystallite Size
Effect on the Pressure–Temperature Phase Diagram of Zirconia Grown by Zirconium-
Based Alloys Oxidation,” J. Nucl. Mater., Vol. 300, 2002, pp. 118–126.
[53] Godlewski, J., Gros, J., Lambertin, M., Wadier, J., and Weidinger, H., “Raman Spec-
troscopy Study of the Tetragonal-to-monoclinic Transition in Zirconium Oxide
Scales and Determination of Overall Oxygen Diffusion by Nuclear Microanalysis of
O18,” Proceedings of the 9th International Symposium on Zirconium in the Nuclear
Industry, ASTM STP 1132, ASTM International, West Conshohocken, PA, 1991,
25520S.
[54] Simeone, D., Baldinozzi, G., Gosset, D., and Le Caër, S., “Phase Transition of Pure Zirconia
Under Irradiation: A Textbook Example,” Nucl. Instrum. Methods Phys. Res. B, Vol. 250,
2006, pp. 95–100.
[55] Godlewski, J., “How the Tetragonal Zirconia is Stabilized in the Oxide Scale that is
Formed on a Zirconium Alloy Corroded at 400 C in Steam,” Proceedings of the 10th
International Symposium on Zirconium in the Nuclear Industry, ASTM STP 1245, ASTM
[56] Pétigny, N., Barberis, P., Lemaignan, C., Valot, C., and Lallemant, M., “In Situ XRD Analysis
of the Oxide Layers Formed by Oxidation at 743 K on Zircaloy 4 and Zr-1NbO,” J. Nucl.
Mater., Vol. 280, 2000, pp. 318–330.
[57] Chow, C. K., Rosinger, H. E., and Bera, P. C., “Creep Behaviour of Oxidized Zircaloy-4
Fuel Sheathing,” Proceedings of Materials in Nuclear Energy: Proceedings of an Interna-
tional Conference, Huntsville, Canada, Sept 29–Oct 02, 1982, pp. 112–116.
[58] Uetsuka, H. and Hofmann, P., “High-Temperature Oxidation Kinetics of Zircaloy-4 in Oxy-
gen/Argon Mixtures,” J. Nucl. Mater., Vol. 168, 1989, pp. 47–57.
[59] Arima, T., Moriyama, K., Gaja, N., Furuya, H., Idemitsu, K., and Inagaki, Y., “Oxidation
Kinetics of Zircaloy-2 Between 450 C and 600 C in Oxidizing Atmosphere,” J. Nucl.
Mater., Vol. 257, 1998, pp. 67–77.
[60] Guo, X., “Property Degradation of Tetragonal Zirconia Induced by Low-Temperature

Defect Reaction With Water Molecules,” Chem. Mater., Vol. 16, 2004, pp. 3988–3994.
[61] Maehara, Y. and Langdon, T. G., “Review Superplasticity in Ceramics,” J. Mater. Sci., Vol.
25, 1990, pp. 2275–2286.
[62] Li, P., Chen, I. W., and Penner-Hahn, J. E., “Effect of Dopants on Zirconia Stabilization—
An X-Ray Absorption Study: I, Trivalent Dopants,” J. Am. Ceram. Soc., Vol. 77, 1994, pp.
118–128.
[63] Yilmazbayhana, A., Motta, A. T., Comstock, R. J., Sabol, G. P., Laid, B., and Cai, Z.,
“Structure of Zirconium Alloy Oxides Formed in Pure Water Studied With Synchrotron
Radiation and Optical Microscopy: Relation to Corrosion Rate,” J. Nucl. Mater., Vol. 324,
2004, pp. 6–22.
[64] Baek, J. H., and Jeong, Y. H., “Breakaway Phenomenon of Zr-Based Alloys During a
High-Temperature Oxidation,” J. Nucl. Mater., Vol. 372, 2008, pp. 152–159.
[65] Garde, A. M., Pati, S. R., Krammen, M. A., Smith, G. P., and Endter, R. K., “Corrosion
Behavior of Zircaloy-4 Cladding With Varying Tin Content in High-Temperature Pressur-
ized Water Reactors,” Proceedings of the 10th International Symposium on Zirconium in
the Nuclear Industry, ASTM STP 1245, ASTM International, West Conshohocken, PA,
1994, pp. 760–778.
[66] Sato, K., Suzuki, K., Narumi, R., Yashiro, K., Hashida, T., and Mizusaki, J., “Ionic Conductiv-
ity in Uniaxial Micro Strain/Stress Fields of Yttria-Stabilized Zirconia,” Jpn. J. Appl. Phys.,
Vol. 50, 2011, 055803.
[67] Kaddour, D., Frechinet, S., Gourgues, A. F., Brachet, J. C., Portier, L., and Pineau, A.,
“Experimental Determination of Creep Properties of Zirconium Alloys Together With
Phase Transformation,” Scr. Mater., Vol. 51, 2004, pp. 515–519.
[68] Trego, G., 2012, “Comportement en Fluage à Haute Température dans le Domaine
Biphasé (aþb) du M5,” Ph.D. thesis, Ecole des Mines de Paris, France.
[69] Parise, M., Sicardy, O., and Cailletaud, G., “Modelling of the Mechanical Behavior of the
Metal-Oxide System During Zr Alloy Oxidation,” J. Nucl. Mater., Vol. 256, 1998, pp.
35–46.
DISCUSSION
Research, Fuel and reactor Materials Department, Hungary:—In some accident con-
ditions (e.g., ATWS - Anticipated Transients Without Scram), the pressure can be
higher than the maximum pressure of your tests (80 bar). Do you think that the
enhancement of oxidation kinetics will continue (e.g., up to 150 bar)?
Authors’ Response:—Based on our data, the enhancement of the oxidation

kinetics of Zircaloy-4 for temperatures between 750 and 1000 C is rather continu-
ous with increasing steam pressure from 1 bar up to 80 bar. So, the evolution of
weight gain as a function of steam pressure can be fairly well described by an expo-
nential function. According to Park et al.’ results obtained for steam pressures up to
150 bar [1], one can expect that the enhancement of the oxidation kinetics of
Zircaloy-4 will continue for higher steam pressure.
For M5TM, a low or insignificant effect of steam pressure was observed for
steam pressure up to 80 bar. As far as the exact mechanisms responsible for the
steam pressure effect are not clearly identified, it is difficult to extrapolate the results
to higher steam pressure. However Park et al. results on Zr-1%Nb [2] suggest that
the effect of steam pressure on the oxidation kinetics of M5 would not be enhanced
for higher steam pressure, up to 150bar.
[1] Park, K., Kim, K.P., Yoo, T.G. and Kim, K.T., “Pressure Effects on High
Temperature Steam Oxidation of Zircaloy-4,” Metals and Materials International,
Vol. 7, 2001, pp. 367–373.
[2] Park, K., Yang, S. and Kim, K., “Nb Effect on Zr-alloy Oxidation Under
High Pressure Steam at High Temperatures,” Proceedings of the Water Reactor Fuel
Performance Meeting, Kyoto, Japan, October 2–6, 2005, pp. 811–826.
Question from Marc Tupin, CEA Saclay:—Did you perform X-ray diffraction
analysis or Raman spectroscopy to determine the volume fraction of quadratic (tet-
ragonal) zirconia? Did you see a difference between low and high pressure?
Comment:— It has been actually shown in the literature at lower temperature

than LOCA conditions that steam pressure accelerates the phase transformation
from quadratic (tetragonal) to monoclinic zirconia.
Authors’ Response:—The greater potential destabilization of the tetragonal zir-

conia phase under high steam pressure is the main assumption made in this paper
to explain the effect of steam pressure. The stronger sensitivity to steam pressure of
Zircaloy-4 compared to M5TM between 750 C and 1000 C may then be related to
the higher volume fraction and the lower stability of the tetragonal zirconia phase
in the oxide formed on Zircaloy-4 (and/or to a lower capacity of this oxide to
accommodate without significant damage the stresses induced by the tetragonal to
monoclinic phase transformation).
We have not yet performed X-ray diffraction or Raman spectroscopy analysis

on the tested samples, but we have shown that cooling after oxidation at high tem-
perature induces irreversible phase transitions within the oxide, so that the struc-
ture of the growing oxide cannot be observed post-facto, neither at room
temperature nor after reheating at the prior oxidation temperature [1]. As a conse-
quence, it would be necessary to perform in-situ structural analysis of the oxide, i.e.
during oxidation, to evaluate the actual volume fraction of the tetragonal zirconia
phase. This is a quite difficult task for the high temperatures and steam pressures
investigated here. Nevertheless, a comparative analysis performed after cooling may
provide tendencies on the volume fraction of the tetragonal zirconia phase as a
function of the alloy and the steam pressure.
[1] Gosset, D., Le Saux, M., Simeone, D. and Gilbon, D., “New Insights in
Structural Characterisation of Zirconium Alloys Oxidation at High Temperature,”
Journal of Nuclear Materials, Vol. 429, 2012, pp. 19-24.
Question from Arthur Motta, Penn State University:—You postulate an effect of

the steam pressure on the morphology of the oxide formed at 1 bar and 80 bar (0.1
MPa and 8 MPa), but these pressures appear to be too low to cause, for example, a
stress stabilization of a phase. Do you think that the observed effect could be an
indirect effect of steam pressure, for example, on the ability of steam to penetrate
into porosity and accelerate chemical reactions?
Authors’ Response:—The steam pressures investigated in this paper (0.1 MPa

up to 8 MPa) are indeed probably too low to cause direct stress stabilization of one
zirconia phase. It is possible that steam penetration into oxide porosity and chemi-
cal reactions are accelerated under high steam pressure, but it has also been
observed that the oxide formed on Zircaloy-4 between 750 C and 1000 C is more
disrupted and porous when grown under high steam pressure. It is assumed that
this microstructure is the result of a greater destabilization of the tetragonal zirconia
phase under high steam pressure. This microstructure probably contributes to a
faster oxygen and hydrogen transport through the oxide layer.
Questions from Martin Steinbrück, Karlsruhe Inst. of Technology
Comment:—At KIT, we performed test on dependence of oxidation kinetics on

partial pressure of steam and oxygen. We found no effect of steam partial pressure
as long as oxide scale was intact but a strong influence of steam partial pressure
during breakaway oxidation.
Q1:—Are you sure that high pressure affects oxide morphology and that the
dependency on pressure is not caused by different oxide morphologies of Zircaloy-
4 and M5?
Authors’ Response:—Some results reported in the literature have shown that the
oxidation rate depends on the steam partial pressure rather than on the total pres-
sure of the atmosphere mixture. On the one hand, no significant difference between
the microstructures of the oxides formed on Zircaloy-4 and M5TM under steam at
atmospheric pressure was observed. On the other hand, it has been observed that
steam pressure has an effect on the microstructure of the oxide formed on Zircaloy-
4 in the 750-1000 C temperature range: the oxide grown under high steam pressure
is more disrupted and exhibit more pores and cracks. The microstructure of the ox-
ide layers formed on M5TM is not significantly affected by steam pressure for the
tested conditions. No breakaway oxidation (transition from a (sub-)parabolic to a
nearly linear oxidation kinetics and large hydrogen pick-up fraction) was observed.
It has been determined that steam pressure has an effect since the early stages of ox-
idation on the oxidation rate but does not substantially modify the (sub-)parabolic
oxidation kinetics. The insignificant effect of steam partial pressure observed at KIT
(before the breakaway oxidation transition) is maybe due to steam partial pressures
that were lower than those investigated here.
M. Grosse,1 J. Stuckert,1 C. Roessger,1 M. Steinbrueck,1

M. Walter,1 and A. Kaestner2
Analysis of the Secondary

Cladding Hydrogenation
During the Quench–LOCA
Bundle Tests With Zircaloy-4
Claddings and its Influence
on the Cladding Embrittlement
Reference
Grosse, M., Stuckert, J., Roessger, C., Steinbrueck, M., Walter, M., and Kaestner, A., “Analysis of
the Secondary Cladding Hydrogenation During the Quench–LOCA Bundle Tests With
Zircaloy-4 Claddings and its Influence on the Cladding Embrittlement,” Zirconium in the
Nuclear Industry: 17th International Symposium, STP 1543, Robert Comstock and Pierre
ABSTRACT
LOCA simulation tests were performed in the QUENCH facility of KIT on fuel rod
bundle scale. The first two tests using Zircaloy-4 claddings, out of a series of six
tests with different cladding alloys, were already performed. The test conditions
and results are described. The secondary hydrogenation of the Zircaloy-4
cladding tubes was investigated by means of neutron imaging. In the cladding of
the inner rods, hydrogen enriched bended bands were found. They are non-
symmetrical to the tube axis. The bands are located at the position where
significant inner oxidation ends. X-ray diffractometry (XRD) measurements show
that the hydrogen remains at least partially in the zirconium lattice. The
formation of the hydrogen enriched bands results in an embrittlement of the
Manuscript received November 15, 2012; accepted for publication April 13, 2014; published online September
22, 2014.
1
Karlsruhe Institute of Technology, Institute for Applied Materials, P.O. Box 3640, Karlsruhe, D-76021 Germany.
2
Paul Scherrer Institute, Spallation Neutron Source Division, Villigen, CH-5232 Switzerland.
3
Hyderabad, India.
Copyright V
GROSSE ET AL., DOI 10.1520/STP154320120155 1055
cladding tubes and in changes of the fracture mode in tensile tests. The micro-
hardness of the cladding increases at the band position. The reasons of the
formation of these hydrogen enriched bands and the dependence of its
appearance on the temperature scenario are discussed.
Keywords
loss of coolant accidents, secondary hydrogenation, neutron imaging, hydrogen
embrittlement
Nomenclature
D4/Dx ¼ duplex cladding material produced by AREVA consisting of a Zry-4 bulk
and an outer layer with reduced tin content
ECR ¼ equivalent cladding reacted
HV ¼ Vickers hardness
ICON ¼ imaging beamline using cold neutrons
KIT ¼ Karlsruhe Institute of Technology
LOCA ¼ loss of coolant accident
M5 ¼ cladding alloy produced by AREVA
PSI ¼ Paul Scherrer Institut
PWR ¼ pressurized water reactor
SINQ ¼ Swiss neutron source
SPP ¼ second phase particles
TC ¼ thermocouple
TEM ¼ transition electron microscopy
VGB ¼ association of the German utilities (Verein der Großkraftwerksbetreiber)
XRD ¼ X-ray diffraction
ZIRLO ¼ cladding alloy produced by Westinghouse
Zry-4 ¼ cladding alloy Zircaloy-4
Introduction
Currently, an amending of the LOCA criteria (TPCT 1200 C and ECR 17 %) is
under discussion worldwide to consider the embrittlement effect of hydrogen absorbed
after cladding burst and oxidation of the inner cladding surface. In order to study this
effect known as secondary hydriding on fuel rod bundle scale, the QUENCH–LOCA
program was started at KIT in 2010. Two tests using simulation bundles with Zircaloy-4
(Zry-4) claddings were already performed until November 2012 [1–5]. Tests with
D4/Dx, M5TM, and ZIRLOTM claddings in the as-received state and pre-hydrided to
simulate high burn-up are in preparation and shall be performed until 2015.
The influence of absorbed hydrogen on the mechanical properties was investi-
gated for instance in Refs. [6–15]. Generally, a decrease of the ductility, impact
energy, and maximal load with increasing hydrogen content were observed. For
instance, the rupture stress and time decreases with increasing hydrogen
concentration [8]. The impact energy decreases strongly with the hydrogen concen-
tration [13]. Brittle cladding material behavior was found in ring compression tests
at ECR 12 % specimens containing 1000 ppm hydrogen [11], at ECR 3 %–10 %
with hydrogen concentrations of 300 wppm and above, at ECR 6 % in samples
containing 550 wppm hydrogen [14] or at ECR 5 % with 400 wppm hydrogen
[15]. It shows that the limit value of ECR ¼ 17 % defined in the rules is not conserv-
ative if a large amount of hydrogen was taken up.
The hydrogen distributions in the cladding tube after LOCA were determined
in Refs. [8,9,14] by means of hot extraction using segments of some millimeter
length. Whereas in the axial range of the burst opening no hydrogen was found,
broad maxima were found at both sides of the burst up to a distance of 100 mm.
The hydrogen distribution in the cladding tubes of the QUENCH–LOCA simu-
lation tests were determined by means of neutron imaging techniques. Neutron
radiography was applied for the quantitative determination of the hydrogen con-
centration in in-situ and ex-situ separate-effect tests [16–19].
Whereas in Refs. [1–5] mainly the results of the commissioning test
QUENCH-L0 are presented and discussed, this paper is focused on the hydrogen
distribution in the rods of the QUENCH-L1 reference test and on the qualitative
explanation of the hydrogen distribution. However, the post-test examinations of
the QUENCH-L1 bundle are not yet finished. Therefore, the results for these sam-
ples are preliminary and not complete.
The Quench-L0 and L1 Tests

In the QUENCH facility a bundle of 21–31 rod simulators consisting of the clad-
ding, pellets simulators made from stabilized zirconia and resistivity heaters is
tested in a flowing atmosphere of superheated steam and argon as a carrier gas. The
argon is needed for quantitative off-gas measurements by a state of the art mass
spectrometer [20]. The gases together with hydrogen produced in the zirconium-
steam reaction flow from the bundle outlet at the top through a water-cooled off-
gas pipe to the condenser where the steam is separated from the non-condensable
gases. The test section has a separate inlet at the bottom to inject water for reflood
(quenching). The system pressure in the test section was 3 bar, what corresponds to
the lowest pressure level in reactor after the break of a main coolant line and a fast
pressure loss from 155 bar during blowdown phase [21,22]. The system pressure is
only of secondary importance for LOCA simulations; more important is a differ-
ence between inner rod pressure and the system pressure.
The QUENCH-L0 and QUENCH-L1 test bundles are approximately 2.5 m
long and are made up of 21 fuel rod simulators. Figure 1 gives a scheme of the bun-
dle cross section. The fuel rod simulators are held in their positions by five grid
spacers, four of Zry-4, and one of Inconel 718 in the lower (colder) bundle zone.
This bundle design is applied with a pitch of 14.3 mm and corresponds to 16 by
16 PWR assembly design. The Zry-4 cladding of the fuel rod simulator has an o.d.
FIG. 1 Schemes of the cross sections of the QUENCH-L0 and -L1 bundles. The colors
indicate the initial inner pressure value in units of bar. The direction of the burst
opening is marked by the lines.
of 10.75 mm and a wall thickness of 0.725 mm. The heating is electric by tungsten
and tantalum heaters for QUENCH-L0 and QUENCH-L1 test, respectively, with a
diameter of 6 mm and a length of 1024 mm. They are installed in the rod center
between bundle elevations 0 and 1024 mm.
The heaters are surrounded by annular ZrO2 (yttria-stabilized) pellets, whose
heat capacity of 0.9 J*K1/pellet is comparable with the value of 1 J*K1/pellet for
UO2 pellets.
The test bundle is surrounded by a Zr702 shroud (inner diameter: 80 mm, outer
diameter: 86 mm). It acts as steam and gas guide tube and simulates an adiabatic
surrounding of the reactor core.
Special corner rods are inserted between bundle and shroud. They reduce addi-
tionally the coolant channel area to a more representative value (hydraulic diameter
of the test bundle coolant channel is 11.5 mm versus 9.5 mm the corresponding re-
actor area). These corner rods are either made of solid rods at the top and tubes at
the bottom to be used for thermocouple instrumentation or of solid Zry-4 rods of
6 mm diameter for withdrawal from the bundle during the test to check the degree
of bundle oxidation at specific times. The consideration of heated rod claddings,
corner rods, and shroud, manufactured from similar zirconium alloys, results in the
surface of 37 effective rod simulators.
For temperature measurements, the test bundle is equipped with NiCr/Ni ther-
mocouples (TCs). The 56 surface thermocouples were resistance spot-welded to the
Zry-4 cladding of the fuel rod simulators #02, #04, #07, #11, #15, #19 and distrib-
uted at 17 bundle elevations between z ¼ 250 to 1350 mm. Sixteen other thermo-
couples have no contact with steam: three thermocouples are installed inside the
corner rods and 13 TCs are located at the outer shroud surface. The shroud ther-
mocouples are mounted between 250 and 1250 mm.
Basis of the QUENCH-LOCA simulation tests is the LOCA scenario valid for
German Konvoi reactors. However, the rapid temperature increase in the first
seconds after pressure drop is not possible to simulate by the QUENCH facility.
The time scenarios of the maximal temperature measured at the rods #04 and #15
at the axial position z ¼ 950 mm are given in Fig. 2 for the two tests already per-
formed. For the commissioning test QUENCH-L0, the aspired heating rate could
not be achieved because of the limits of maximal current and maximal voltage. The
test was finished by rapid water quenching after reaching the maximal temperature.
Detailed information about the QUENCH-L0 test is given in Refs. [2,4].
The QUENCH-L1 test differs from the commissioning test QUENCH-L0 mainly
in the heater material and with it a larger heating rate (5.8 K/s for QUENCH-L1; 2.5 K/s
for QUENCH-L0) and a cooling phase before quenching (cooling rate of 3.8 K/s for
QUENCH-L1; quenching rate of 400 K/s for QUENCH-L0). In the QUENCHL1 test,
the tungsten heaters were replaced by tantalum heaters with a higher electrical resistiv-
ity. Due to the radial temperature gradient in the bundle, the outer, peripheral rods do
not reach the temperatures of the inner rods in both tests. This provides the possibility
to study the effect of temperature on the secondary hydrogenation.
The gas supply system for individual pressurization of rods consists of pressure
controller, 21 valves, 21 pressure transducers, and 21 justified compensation vol-
umes for setting of prototypic volume value of 31.5 cm3 (the compensation is
needed because of absence of empty plenums inside rod simulators). The gas supply
is connected by capillary tubes to each rod at its lower end with drilled copper elec-
trode. The gas gap under the cladding is: 0.15 mm in the region of Cu/Mo electro-
des and 0.075 mm in the region of W-heater/ZrO2-pellets. Before gas filling the
rods and gas supply system were evacuated.
FIG. 2 Temperature scenarios of the QUENCH-L0 and QUENCH-L1 test. The

temperatures were measured at the axial position of z ¼ 950 mm (hottest
position) at the rods #04 (inner rod) and #15 (outer rod).
During the initial phase of the experiment, the fuel rod simulators were back-
filled with Kr gas to 20 bars. Then, before the transient (which has been started at
the peak cladding temperature of 800 K), they were separately pressurized to the
target pressures of 35, 40, 45, 50, and 55 bars in the QUENCH-L0 test. One rod
(rod #15) of the QUENCH-L0 test was used as reference rod with the internal pres-
sure of 3 bars, equal to the system pressure. This rod was the only one that did not
burst during the two tests. Figure 3 gives the dependence of the burst time on the
inner pressure for the QUENCH-L0 test. It was found that the initial inner tube
pressure has only a minor effect on burst time and burst opening dimensions. The
burst time seems to be controlled by the temperature gradients. Therefore, in
QUENCH-L1 the same initial inner pressure of 50 bars was applied to all rods.
Usually, the inner rod pressure of the PWR fresh rod is between 70 and 80 bars at
800 K. However, these pressure values are very near to the established facility limit
(90 bars). Therefore, to exclude the possible leakages, it was decided to apply lower
pressures.
As mentioned before, the cladding tubes for the QUENCH-L0 and QUENCH-
L1 simulation tests were made of Zircaloy-4 (Sn-1.5, Fe-0.23, Cr-0.11, and
O-0.14 wt. %, Zr balance).
After the tests, the bundles were dismounted. The heaters and zirconia pellet
were removed from the rods. The cladding tubes were investigated separately. The
post-test investigations comprehend determination of the plastic strain induced
during the tests (profilometry), neutron imaging experiments for the determination
of the hydrogen distribution, metallographic examinations, XRD, and TEM investi-
gations of the absorbed hydrogen, micro-hardness measurements, and tensile tests
to determine the degradation of the mechanical properties.
FIG. 3 Plot of the burst time versus the inner pressure of the rod simulators.
Laser profilometry allowed measuring the degree of axial and circumferential

ballooning for each rod. On the basis of these data, the axial distribution of the
cooling channel blockage was calculated. The maximum blockage was measured as
23 % by QUENCH-L0 (at elevation of 995 mm) and as 24 % by QUENCH-L1 (at
elevation of 950 mm). The maxima of cladding ballooning were observed for the
QUENCH-L1 bundle at lower elevations compared to the QUENCH-L0 bundle
due to corresponding shift of highest bundle temperatures for QUENCH-L1 down-
wards about 50 mm.
Results of the Neutron Imaging Investigations

The neutron imaging investigations were performed at the ICON facility at the
Swiss Neutron Source SINQ at the Paul Scherrer Institut in Villigen (Switzerland)
[23]. Details to these measurements are given in Ref. [2].
For the QUENCH-L0 test the results are already published in Ref. [2]. There-
fore, only two representative examples are given in this paper. The rod #06 is an
example for the inner part of the bundle. Its neutron radiograph is given in
Fig. 4(a). Typical for these rods are the clearly visible darker hydrogen enriched
bended bands. Maximal hydrogen concentrations up to about 2700 wppm were
determined by means of neutron tomography. As an example for the outer bundle
part the neutron radiograph of rod #10 is given in Fig. 4(b). No hydrogen enrich-
ments were found in the most of the outer claddings.
Figure 5 gives an overview of the neutron radiographs of the QUENCH-L1
rods. Rod #01 was ruptured above and below the burst opening during the remov-
ing of heater and pellet simulators. As the radiography given in Fig. 5(a) shows, the
two fracture positions have a high hydrogen concentration (darker gray value).
Some of the rods were bended during the test. The curvatures of the rods are clearly
visible for the inner rods #03–#07, and for #17, #18, #20, and #21. Possible reasons
for the bending are residual stresses, radial, and tangential temperature gradients in
the bundle, friction between the ballooned oxidized rods and the spacer grids, and/
or inhomogeneous oxidation and hydrogenation. In contrast to the QUENCH-L0
test where the stiff tungsten heaters suppress the noticeable bending, the tantalum
heaters of the QUENCH-L1 bundle become soft at higher temperatures and allow
the significant deformation of the rods. Hydrogen enrichments were found only for
the inner rods #01–#09. The hydrogen enriched bands in the QUENCH-L1
FIG. 4 Neutron radiographs of the rods #06 and #13 of the QUENCH-L0 test.
FIG. 5 Neutron radiographs of the claddings of the QUENCH-L1 rod simulators. Rod
#01 was ruptured above and below the burst opening during the removing of
heater and pellet simulators.
specimens seem to be wider and have less contrast to the adjacent regions com-
pared to the bands in the QUENCH-L0 samples. The rods #02, #08, and #09 show
hydrogen enrichments in the burst opening region, too. In the rods of the
QUENCH-L0 test, hydrogen enrichments at this position could not be found. The
maximal hydrogen concentrations cannot yet be quantified exactly, because neu-
tron tomography investigations are not yet performed successfully at the
QUENCH-L1 specimens. A first glance on the radiography data and a comparison
with the radiography data of the QUENCH-L0 specimens give hints that the
maximal hydrogen concentrations should be about 25 %–35 % lower in the

QUENCH-L1 test than in the QUENCH-L0 test.
Investigation of Hydrogen Distribution Inside

the Metal
The inner oxide layer was studied by means of metallographic investigations. The
maximal oxide thicknesses in the burst region are in the order of 25 lm for both
tests (QUENCH-L0 rod #08: outer surface: 25 lm, inner surface: 22 lm;
QUENCH-L1 rod #01: outer surface 25 lm, inner surface: 28 lm). The maximal
oxide layer thicknesses of the peripheral rods do not reach 20 lm. Mean ECR values
of 2 % for inner and 1.2 % for peripheral rods were determined. At the burst open-
ings, the ECR values can reach 4 and 2.2 % for inner and outer rods, respectively.
The thicknesses decrease with increasing distance to the burst opening. The borders
between oxidized and non-oxidized region corresponds with the positions of the
hydrogen enriched bands. The oxide layer thicknesses measured could not be
expected by the temperature scenario applied. Maximal oxide layer thicknesses of
less than 15 lm could be expected for those scenarios. Local and temporal tempera-
ture escalations because of the chemical heat release during oxidation have to be
taken into account to explain the unexpected large oxide scales.
XRD analysis was applied to investigate the phases existing in the QUENCH-
L0 cladding rods including possibly precipitated hydrides. The measurements were
performed using a laboratory facility with Cu–Ka radiation. The sample from rod
#3 was arranged of four small longitudinal strips of about 0.725 mm by 1 mm by
15 mm that were axially cut out at the hydrogen enriched zones; the X-ray diffrac-
tion measurements were performed at the inner cladding surface. The same form
had the sample from as-received Zircaloy-4 cladding. The X-ray diffraction of rod
#08 was measured at the internal cladding region after removing of about 300 lm
of the surface by grinding and polishing. Figure 6 gives an overview of the whole dif-
fraction patterns measured at the three specimens. In the small diagram included,
the scattering angular region between 31 2H 39 is zoomed. No clear zirconium
hydride peaks were observed (positions of the expected ZrH2 Bragg peaks are
marked by the black and red lines). Only for sample #08, weak shoulders at the zir-
conium peaks at the hydride positions cannot be excluded. In addition, two small
peaks at about 28 and 34 were measured at the sample prepared from rod #08. On
the other hand, shifts of the zirconium peak positions are clearly visible for the two
specimens prepared from the hydrogen enriched bands compared to the as-
received Zry-4 specimen (see the diagram included). The shift in specimen #03 is
larger than in specimen #08.
The results of the XRD measurements indicate that most of the hydrogen
remains in solution after quenching and the amount, size or crystallinity of the
hydride precipitates is too small to result in Bragg peaks.
FIG. 6 XRD patterns of samples prepared from the hydrogen enriched bands of
QUENCH-L0 rods #03 and #08 and of a rod in the as-received state. The
included small diagram zooms the region between 31 and 39 .
To check the formation of very small hydrides, transmission electron micros-

copy (TEM) investigations were started using a FEI Tecnai 20F. Samples were pre-
pared from as-received Zry-4 and from the hydrogen enriched bands of rod #08. In
Fig. 7, TEM micrographs of both specimens are shown. In contrast to the as-
received state, small objects with a size of 3–7 nm were found in the hydrogen
enriched band of specimen #08. In fresh Zry-4 specimen, only precipitates of 50 to
300 nm size were detected. They could be identified as the well-known iron and
FIG. 7 TEM image of inhomogeneities found in the hydrogen enriched band of

QUENCH-L0 rod #08 (b) and SEM image of an as-received sample (a), for
comparison.
chromium containing second phase particles (SPP). One of these precipitates is

marked as an example in Fig. 6(a).
The 3–7 nm particles, detected in the hydrogen rich bands of the #08, can be
ZrH2 or other H- and Zr-phases. Due to the short period between onset of second-
ary hydriding and fast intensive reflood of the bundle no formation of macroscopi-
cally larger hydrides was possible. Taking into account the small size of presumable
hydrides, the missing of additional Bragg peaks in the XRD diagram (see Fig. 5) can
be understood. Small particle size of the hydrogen containing phase would result in
broad Bragg peaks with low maximal intensity which are not detectable in the XRD
pattern. However, meta-stable hydride phases like the f phase as found in Ref. [24]
should also be taken into account. This phase was found in a sample produced by
cathodic hydrogen charging, annealing at 430 C for 24 h and water quenching.
Even if the thermal treatment resulting in the formation of f hydrides is not compa-
rable with those of the QUENCH–LOCA tests, it cannot be excluded that such
nanosized precipitates are formed. However, the two non-identified small peaks
measured at rod #08 cannot be explained by the assumption of the formation of f
hydrides. Whereas the peak measured at about 34 is near the (004) peak expected
for f hydrides at 34.84 , no peak can be expected at about 28 for this phase.
Results of the Mechanical Testing

It is well known that hydrogen uptake degrades the mechanical properties of zirco-
nium alloys. Hydrogen absorption results in embrittlement of the metal and reduces
the thermo-shock stability. In order to examine the influence of the hydrogen uptake
that occurred during the QUENCH-L0 test, micro-hardness at the rods #03, #08 and
#17 and tensile tests at all rods excluding these three rods were performed.
Figure 8 plots the micro-hardness versus the axial position in the hydrogen
enriched band of rod #03. Values of about 220 HV were measured at positions far
away of the hydrogen enriched bands. At the hydrogenated band positions determined
by neutron radiography, maximal micro-hardness values of about 360 HV were deter-
mined for all three rods investigated. The maximal value found for QUENCH-L1- rod
#01 is 420 HV measured at the hydrogen enriched zone close to the fracture. All other
rods of this test do not show any hardness maxima at the hydrogen enriched bands.
The mean values of the hardness are for all rods about 250 HV.
The measurements allow the determination of the Young modulus, too. In the
hydrogen enriched bands, the Young moduli are significantly lower than outside of
these sample regions.
At room temperature, the fracture behavior of the QUENCH-L0 specimens in
the tensile test can be divided into three types:
1. Double rupture near to the burst opening due to hydrogen embrittlement,
2. Rupture across the burst opening mid position due to stress concentration at
burst crack edges, and
3. Rupture near end plugs after necking.
FIG. 8 Dependence of the micro-hardness on the axial position in the hydrogen

enriched band region of QUENCH-L0 rod #03.
Figure 9 gives examples for these behaviors. Most of the inner rods from the
QUENCH-L0 test failed within the hydrogen rich areas close to the burst opening,
whereas most of the outer rods failed after larger plastic deformation caused by
necking. Fracture across the burst opening, caused by stress concentrations result-
ing from discontinuities like buckles or a small cross-cracks, was observed for both
inner and outer claddings.
FIG. 9 Examples of the three rupture types seen during tensile tests at room
temperature.
In contrast to the QUENCH-L0 specimens, all tensile tested rods from the
QUENCH-L1 bundle showed rupture across the center of the corresponding burst
opening, independent of the position of a single tube within the bundle or of the
presence or absence of hydrogen enriched bands (excluding rod #01 which was rup-
tured above and below the burst opening during the removing of heater and pellet
simulators rod).
In general, it is observed that in particular the specimens from the center of the
bundles show lower strength and strain values at failure. In contrast, single clad-
dings from the outer bundle area partly reveal strong plastic deformations up to
fracture, even when they failed within the burst region.
For the QUENCH-L0 specimens, it was observed that ultimate tensile strength
and elongation at fracture are smaller for inner samples than for the outer rods.
The ultimate strength at room temperature is reduced from about 510 MPa in the
rods with no or less hydrogenation to values between 222 and 276 MPa in the sam-
ples breaking at the hydrogen enriched band positions detected by neutron radiog-
raphy as can be seen in Fig. 10. In some tests, double breaks occurred in both bands,
below and above the burst opening. The plastic elongation decreases dramatically
by the hydrogen absorption. Whereas some outer rods show a fracture elongation
of about 11 %, fracture elongations of the hydrogen embrittled inner rods of mostly
0.3..0.4 % are reached. The only exception is rod #02 with an ultimate strength of
408 MPa and a fracture elongation of nearly 1 %.
The differences between the inner and peripheral QUENCH-L1 rods are
smaller than in the QUENCH-L0 specimens. A significant reduction of the fracture
strain compared to the as-received state can be observed at the peripheral rods, too.
However, the embrittlement of the inner rods, shown in the reduction of fracture
strain and ultimative stress, is stronger in the inner rods (see Fig. 10).
FIG. 10 Examples of the results of tensile tests at QUENCH–LOCA rods.

Discussion of the Hydrogen Distribution

The results of the tensile test show that the failure behavior of QUENCH–LOCA
tested claddings may strongly depend on the hydrogen concentration and distribu-
tion. Therefore it is important to understand the mechanisms resulting in the
hydrogen distributions found. Figure 11 gives a scheme of the processes. After the
burst, at first the filling and fission gases are released and with it the pressure drops to
ambient pressure. This process needs about 40 s as the internal pressure measure-
ments have shown. When the total outflow of the fission gases becomes low or stops,
steam can penetrate the burst opening and diffuse into the ballooned region and the
gap between pellets and inner cladding surface (wide bright gray arrow in Fig. 9).
Simultaneously, the steam will be consumed by oxidation of the fresh metallic inner
cladding surface and by adsorption at the pellet surface. The steam oxidation pro-
duces free hydrogen (small arrows in Fig. 11). To answer the question why the hydro-
gen is concentrated in small bands, two models are discussed by the authors.
THE GAP EFFECT APPROACH

The basis of this model is the theoretical model of hydrogen uptake given in Refs.
[25,26]. The system hydrogen in the gas phase, in the oxide and in the metal tends
to equilibrium of the chemical activities. At temperatures below 1250 K occurring at
the beginning, the oxidation rate and with it the hydrogen source term are low. In
the ballooning region, the hydrogen produced by the steam oxidation meets a rela-
tively large gas volume. The hydrogen partial pressure in the gas is low and, hence,
its concentration in the remaining metal, according to Sievert’s law [27]. The
hydrogen diffuses much faster than steam into the gap between pellets and inner
surface. This hydrogen and the hydrogen produced in the gap and at the border
between ballooned and non-ballooned regions meet a small total gas volume in the
gap. The hydrogen enrichment is here much higher. The higher partial pressure
results in a higher uptake of hydrogen.
THE CRITICAL OXIDE LAYER APPROACH

The basis of this model is the existence of a sub-micrometer thick oxide layer at the
inner surface of the cladding tube produced during manufacturing of the cladding
FIG. 11 Scheme of the secondary hydriding after cladding burst during LOCA.
tube and its transport in air. As discussed in Ref. [28], a critical oxide layer thick-
ness exists above which the hydrogen transport is nearly suppressed. This critical
oxide layer thickness is assumed to increase with temperature and is for the tetrago-
nal oxide significantly higher than for the monoclinic structure. The initial oxide
layer hints the hydrogen produced by steam oxidation to be absorbed by the zirco-
nium. As a result of the steam oxidation, the oxide layer (middle gray in Fig. 11)
thickness grows. The increase of temperature has the consequence that the initial
oxide layer becomes penetrable for the hydrogen. On the other hand, the newly
formed oxide with a thickness of some micrometers keeps suppressing the hydro-
gen uptake. The uptake can occur rapidly at positions without newly formed oxide.
Because the highest hydrogen concentration is at the boundary between the oxi-
dized and non-oxidized region, the strongest uptake occurs here.
Both, the gap effect and the critical oxide layer approach can explain the exis-
tence of the hydrogen enriched bands. However, the two approaches do not exclude
each other, and perhaps a combination of the two effects results in the hydrogen
distributions observed.
A possible explanation of the difference in the hydrogen bands between the two
tests is the difference in the temperature scenarios. In QUENCH-L0, directly after
reaching the maximal temperature, the bundle was quenched. In QUENCH-L1, the
temperature increase was much faster. After reaching the maximum, a longer and
slower cooling phase followed. The duration of this phase was about 130 s. In addition,
the temperature increase due to switching off the cooling steam was registered at the
beginning of the quench phase. All together, the z ¼ 950 mm axial position had a tem-
perature above 800 K for nearly 200 s. During this time, hydrogen diffuses in the metal-
lic zirconium. On the basis of the results in Ref. [29], the diffusion length during 100 s
is in the order of magnitude of 1 mm (0.8 and 1.6 mm for 1000 and 1300 K, respec-
tively). It means that the diffusion is fast enough to equalize the hydrogen concentra-
tion in the cooldown phase significantly. The gradients in the hydrogen distribution
become lower and the hydrogen bands lose contrast to the adjacent regions of the tube.

Two LOCA simulation tests were performed in the QUENCH facility with bundles
consisting of 21 electrically heated rods with Zircaloy-4 claddings. Different post-
test investigations were performed to analyze the phenomena of the secondary
hydrogenation after the cladding burst. In order to determine the lateral hydrogen
distributions in the cladding tubes, neutron imaging experiments were performed.
This method reveals formation of axially non-symmetrical hydrogenated bands
near the burst openings for both bundle tests. The maximal hydrogen concentra-
tions found in rods of the QUENCH-L0 test is about 2700 wppm. The maximal
concentrations in rods of the QUENCH-L1 test seems to be lower but in the same
order of magnitude. The metallographic investigations showed that the maxima of
hydrogen concentration correspond to the boundary of the oxidized region at the
inner cladding surface around the burst openings. The relative long cooling phase
between reaching the maximal temperatures and quenching during the QUENCH-
L1 test should be the main reason for the observed increased blurring of hydrogen
maxima in the QUENCH-L1 claddings in comparison to the QUENCH-L0
claddings.
Furthermore, it can be assumed that the stronger bending of the QUENCH-L1
rods can result in (1) closing the gap between inner cladding surface and pellets; (2)
closing of some burst openings by neighbor rods, and therefore, hindering the
steam and hydrogen transport as seen in rod #08.
The mechanical investigations show that the micro hardness increases from
220 HV on periphery of the hydrogenated bands to the 360 HV (QUENCH-L0) or
420 HV (QUENCH-L1) in the band middle. Simultaneously, the Young module
decreased inside this region. The tensile tests with ballooned and burst cladding
segments revealed double rupture along hydrogenated bands for the QUENCH-L0
claddings with sharp outlined band boundaries and rupture in the middle of the
burst opening for the QUENCH-L1 claddings with blurred hydrogen bands. The
rupture stress and the ductility of the cladding tubes decrease strongly. The fracture
strain decreases from about 11 to less than 0.5 %.
The fact that all peripheral rods of the QUENCH-L1 test and nearly all periph-
eral rods of the QUENCH-L0 test do not show hydrogen enriched bands can be
explained by the assumption that these enrichments are formed only at tempera-
tures above 1250 K, at least in Zry-4. If this assumption could be confirmed, acci-
dent measures should be available to keep the temperature below this critical value.
ACKNOWLEDGMENTS
The QUENCH–LOCA experiments are supported and partly sponsored by the associ-
ation of the German utilities (VGB).
The neutron imaging investigations were performed at the ICON facility at SINQ
(PSI Villigen, Switzerland). The writers thank S. Hartmann for his support in the neu-
tron imaging measurements.
The broad support needed for preparation, execution, and evaluation of the
QUENCH-L0 experiment is gratefully acknowledged. In particular, the authors would
like to thank Mr. J. Moch for the assembly including instrumentation as well as disas-
sembly of the test bundle, Dr. H. Leiste for the XRD measurements, Dr. M. Klimenkov
for the TEM investigations and Mrs. U. Peters for the metallographic examinations
and the photographic documentation.
References
[1] J. Stuckert, M., Große, C., Rössger, M., Steinbrück, and Walter, M., “Experimental Results
of the Commissioning Bundle Test QUENCH-L0 Performed in the Framework of the
QUENCH-LOCA Program,” Proceedings of the ICAPP 2011, Nice, France, May 2–5, 2011,
Paper No. 11226.
[2] Grosse, M., Stuckert, J., Steinbrück, M., and Kaestner, A., “Secondary Hydriding During
LOCA–Results from the QUENCH-L0 Test,” J. Nucl. Mater., Vol. 420, Nos. 1–3, 2012, pp.
575–582.
[3] Stuckert, J., Große, M., Rössger, C., Klimenkov, M., Steinbrück, M., and Walter, M.,
“QUENCH-LOCA Program at KIT and Results of the QUENCH-L0 Bundle Test,” LOCA-
Halden Workshop, Lyon, France, May 29–30, 2011.
[4] Stuckert, J., Große, M., Rössger, C., Klimenkov, M., Steinbrück, M., and Walter. M.,
“QUENCH–LOCA Program at KIT on Secondary Hydriding and Results of the Commis-
sioning Bundle Test QUENCH-L0,” Nucl. Eng. Des., (accepted).
[5] Grosse, M., Rössger, C., Stuckert, J., Steinbrück, M., Walter, M., Klimenkov, M., and
Kaestner, A., “Analysis of the Absorbed Hydrogen in Cladding Tubes Applied in the
QUENCH-LOCA Tests,” Proceedings of the ICAPP’12, Chicago, IL, June 24–28, 2012,
Paper No. 12134.
[6] Uetsuka, H., Furuta, T., and Kawasaki, S., “Zircaloy-4 Cladding Embrittlement Due to
Inner Surface Oxidation Under Simulated Loss-of-Coolant Condition,” J. Nucl. Sci. Tech-
nol., Vol. 18, No. 9, 1981, pp. 705–717.
[7] Mardon, J. P., Lesbros, A., Bernaudat, C., and Waeckel, N., “Recent Data on M5 Alloy
under RIA and LOCA Conditions,” Proceedings of the 2004 International Meeting on
LWR Fuel Performance, Orlando, FL, Sept 19–22, 2004, Paper No. 1110.
[8] Nagase, F. and Fuketa, T., “Behavior of Pre-Hydrided Zircaloy-4 Cladding Under Simu-
lated LOCA Conditions,” J. Nucl. Sci. Technol., Vol. 42, No. 2, 2005, pp. 209–218.
[9] Nagase, F. and Fuketa, T., “Effect of Pre-Hydriding on Thermal Shock Resistance of
Zircaloy-4 Cladding Under Simulated Loss-of-Coolant Accident Conditions,” J. Nucl. Sci.
Technol., Vol. 41, No. 7, 2005, pp. 723–730.
[10] Hózer, S., Gyóri, C., Horváth, M., Nogy, I., Maróti, L., Matus, L., and Windberg P.,
“Ballooning Experiments with VVER Cladding,” Nucl. Technol., Vol. 152, 2005, pp.
273–285.
[11] Kim, J. H., Lee, M. H., Choi, B. K., and Jeong, Y. H., “Effects of Oxide and Hydrogen on the
Circumferential Mechanic Properties of Zircaloy-4 Cladding,” Nucl. Eng. Des., Vol. 236,
No. 18, 2006, pp. 1867–1873.
[12] Kim, J. H., Choi, B. K., Baek, J. H., and Jeong, Y. H., “Effects of Oxide and Hydrogen on
the Behavior of Zircaloy-4 Cladding During the Loss of the Coolant Accident (LOCA),”
Nucl. Eng. Des., Vol. 236, No. 22, 2006, pp. 2386–2393.
[13] Brachet, J.-C., Vandenberghe-Maillot, V., Portier, L., Gilbon, D., Lesbros, A., Waeckel, N.,
and Mardon, J.-P., “Hydrogen Content, Preoxidation, and Cooling Scenario Effects on
LOCA Conditions,” J. ASTM Int., Vol. 5, No. 5, 2008, p. 101116.
[14] Billone, M., Yan, Y., Burtseva, T., and Daum, R., “Cladding Embrittlement During Postulated
Loss-of-Coolant Accidents,” Report No. NUREG/CR-6967, NRC, Washington, DC, 2008.
[15] Grandjean, C. and Hache, G., “A State-of-the-Art Review of Past Programmes Devoted
to Fuel Behaviour Under Loss-of-Coolant Conditions. Part 3,” IRSN Report SEMCA 2008
093, IRSN, Saint Paul lez Durance Cedex, France, 2008.
[16] Grosse, M., Lehmann, E., Vontobel, P., and Steinbrueck, M., “Quantitative Determination
of Absorbed Hydrogen in Oxidised Zircaloy by Means of Neutron Radiography,” Nucl.
Instr. Methods Phys. Res. A, Vol. 566, No. 2, 2006, pp. 739–745.
[17] Grosse, M., Lehmann, E., Steinbrueck, M., Kühne, G., and Stuckert J., “Influence of Oxide
Layer Morphology on Hydrogen Concentration in Tin and Niobium Containing Zirconium
Alloys After High Temperature Steam Oxidation,” J. Nucl. Mater., Vol. 385, No. 2, 2009,
pp. 339–345.
[18] Grosse, M., “Neutron Radiography—A Powerful Tool for Fast, Quantitative and Non-
Destructive Determination of Hydrogen Concentration and Distribution in Zirconium
Alloys,” J. ASTM Int., Vol. 8, No. 4, 2011, pp. 575–591.
[19] Grosse, M., van den Berg, M., Goulet, C., Lehmann, E., and Schillinger, B., “In-Situ Neutron
Radiography Investigations of Hydrogen Diffusion and Absorption in Zirconium Alloys,”
Nucl. Instr. Methods Phys. Res. A, Vol. 651, No. 1, 2011, pp. 253–257.
[20] Steinbrück, M., Große, M., Sepold, L., and Stuckert, J., “Synopsis and Outcome of the
QUENCH Experimental Program,” Nucl. Eng. Des., Vol. 240, 2010, pp. 171–1727.
[21] Erbacher, F. J. and Leistikow, S., “A Review of Zircaloy Fuel Cladding Behavior in a Loss-
of Coolant Accident,” Scientific Report KFK-3973, Karlsruhe, Germany, 1985.
[22] Wiehr, K., “REBEKA-Bündelversuche Untersuchungen zur Wechselwirkung zwischen

aufblähenden Zircaloyhüllen und einsetzender Kernnotkühlung. Abschlußbericht,”
Scientific Report KFK-4407, Karlsruhe, Germany, 1988.
[23] Kaestner, A. P., Hartmann, S., Kuhne, G., Frei, G., Grunzweig, C., Josic, L., Schmid, F., and
Lehmann E. H., “The ICON Beamline—A Facility for Cold Neutron Imaging at SINQ,”
Nucl. Instr. Methods Phys. Res. A, Vol. 659, No. 1, 2011, pp. 387–393.
[24] Zhao, Z., Blat-Yrieix M., Morniroli J.-P., Legris A., Thuinet L., Kihn Y., Ambard A., and Leg-
ras, L., “Characterization of Zirconium Hydrides and Phase Field Approach to a
Mesoscopic-Scale Modeling of their Precipitation,” J. ASTM Int., Vol. 5, No. 3, p. 101161.
[25] Veshchunov, M. S. and Berdyshev, A. V., “Modelling of Hydrogen Absorption by Zirco-

nium Alloys During High Temperature Oxidation in Steam,” J. Nucl. Mater., Vol. 255, Nos.
2–3, 1998, pp. 250–282.
[26] Veschunov, M. S. and Shestak, V. E., “Models for Hydrogen Uptake and Release Kinetics
by Zirconium Alloys at High Temperatures,” Nucl. Eng. Des., Vol. 252, 2012, pp. 96–107.
[27] Steinbrück, M., “Hydrogen Absorption by Zirconium Alloys at High Temperatures,”

J. Nucl. Mater., Vol. 334, No. 1, 2004, pp. 58–64.
[28] Cox, B., “Mechanisms of Hydrogen Absorption by Zirconium Alloys,” MRS Fall Meeting,
Boston, MA, Nov. 26–30, 1984.
[29] Grosse, M., van den Berg, M., Goulet, C., and Kaestner, A., “In-Situ Investigation of
Hydrogen Diffusion in Zircaloy-4 by Means of Neutron Radiography,” IOP: Conf. Ser., Vol.
340, 2012, p. 012106.
DISCUSSION
Question from Suresh Yagnik, EPRI:—In your tests, the hydriding is occurring
from clad ID side due to ingress of steam and water (quench) after burst. You do
not find significant hydriding on either side of bursts. It would be useful to exam-
ine clad ID oxide thickness values to relate them to hydrogen pickup (from ID) to
assess how much hydrogen the clad ID may have been exposed to. Also due to
quenching (water), the hydriding from ID side becomes limited, anyway, as it is not
gaseous hydriding.
Authors’ Response:—For your first point I completely agree. The examination

of the real oxide thickness distribution at the inner surface of the cladding is diffi-
cult because the oxide thickness is very inhomogeneous in both the axial and hoop
directions. A raw analysis of the oxide layer thickness distribution gives hints that
nearly all hydrogen produced by the oxidation of the cladding inner surface is
absorbed. However, the modeling of the processes can answer, how much free
hydrogen is produced by the oxidation and how much is absorbed by the remaining
metal. Unfortunately, the actual version of the modeling code gives only correct val-
ues for the oxide thickness distribution and the shape and position of the hydrogen
enriched bands but not satisfying quantitative values of the hydrogen enrichments.
We hope that we can answer the question in the future following improvements to
the modeling code.
During quenching the temperatures are low and the time is very short. There-
fore, the distribution of this process to the hydrogen uptake can be assumed to be
not significant.
Question from Johannes Bertsch, Paul Scherer Institute:—As the pellets are
simulated fuel, they will not be bonded (at least partially) to the cladding, a gap is
partially present. Are there considerations how far this influences the inner hydro-
gen uptake?
Authors’ Response:—You are completely right. The gap between the inner clad-
ding tube surface and the pellets determines the ratio between steam transport and
consumption and with it the effective steam transport rate in the gap. It determines
the amount of steam consumed by the oxidation. The larger the gap, the greater is
the steam transport rate and with it the distance of the hydrogen enriched band to
the burst opening. In the QUENCH-LOCA tests the pellets are artificially aligned
by the heater rods. However, most of them are not completely centered in the clad-
ding tube. Because of the bending of the cladding tubes, a non-symmetrical align-
ment of the pellets in the tube can also be expected. Different shapes of the
hydrogen enriched bands are the consequence.
Questions from R. K. Chaube, NFC, Hyderabad
Q1:—What is the hydrogen content in the initial Zircaloy-4 cladding?
Authors’ Response:—The initial claddings were as-received with no pre-charging

of hydrogen. However, tests with pre-hydrided claddings are planed (QUENCH-L4
and –L5).
Q2:—Does the initial burst opening depend on initial hydrogen present?
Authors’ Response:—We have planned to study this question in the future

QUENCH tests mentioned.
Question from N. Ramasubramanian, ECCATEC Inc. Canada:—Would you

consider a non-uniform corrosion of the inner surface because of corrosion pro-
duced hydrogen resulting in a variation of oxidation environment along the
length of cladding, e.g., from pure steam to steam plus increasing concentrations
of hydrogen with elevation.
Authors’ Response:—You are right; particularly the hydrogen content in the gas
phase influences the oxidation kinetics. Our model does not yet consider this effect.
It can be the objective for the improvement of the model in future work. In the
actual version of the model, we consider gradients in the gas mixing in the gap
between inner cladding surface and pellet. The steam diffusion model clearly shows
that the steam transport is much slower than expected for pure gas mixing because
the steam is consumed by the oxidation until the oxide layer is thick enough that
the steam consumption by oxidation becomes lower than the steam supply by diffu-
sion in the gap. Then not all steam reacts with the zirconium. Hydrogen has a
higher diffusion coefficient in gases than steam. It was firstly expected that at higher
distances to the burst opening the atmosphere consist of an argon/hydrogen mix-
ture. The modeling indicates that hydrogen diffusion does not take place in a signif-
icant manner. It means that the free hydrogen produced is strongly absorbed in the
cladding and adsorbed at the cladding and pellet surfaces.
DEGRADATION
AND FAILURE MECHANISMS
Suresh K. Yagnik,1 Jen-Hung Chen,2 and

Roang-Ching Kuo2
Effect of Hydride Distribution on

the Mechanical Properties of
Zirconium-Alloy Fuel Cladding
and Guide Tubes
Reference
Yagnik, Suresh K., Chen, Jen-Hung, and Kuo, Roang-Ching, “Effect of Hydride Distribution on
the Mechanical Properties of Zirconium-Alloy Fuel Cladding and Guide Tubes,” Zirconium in
the Nuclear Industry: 17th International Symposium, STP 1543, Robert Comstock and Pierre
ABSTRACT
Localization of hydride precipitates exacerbates the hydrogen embrittlement
effects on the deformation and fracture properties of Zircaloy fuel cladding
materials. Thus, at comparable hydrogen concentration levels, localized hydride
precipitates are more detrimental from the standpoint of cladding integrity
during service. Indeed, the hydride precipitates are often non-homogeneously
distributed in fuel assembly components; for example, in irradiated fuel cladding,
the hydride rim is formed near the outer oxide–metal interface because of the
temperature gradient that exists during operation. With increasing fuel burnup,
this hydride rim not only becomes denser but might be accompanied by
gradients in local hydrogen and hydride concentrations through the rest of the
cladding wall thickness. Whereas the importance of hydride spacing and their
orientation, as well as the alloy matrix ligaments interspaced with the distributed
hydride has been recognized in the literature, little work has been reported on
Manuscript received December 4, 2012; accepted for publication October 1, 2013; published online June 17,
2014.
1
Electric Power Research Institute (EPRI), Palo Alto, CA 94304, United States of America (Corresponding
author), e-mail: syagnik@epri.com
2
Institute of Nuclear Energy Research (INER), Lungtan, Taiwan.
3
Hyderabad, India.
Copyright V
the effects of hydride precipitate distribution on the mechanical properties of

Zircaloy fuel assembly component materials. In this paper, we report on an
extensive mechanical test program on low-tin Zircaloy-4 specimens from stress-
relieved cladding and recrystallized guide tubes, charged with hydrogen to
obtain uniform, rimmed, and layered hydride distributions. The hydrogen
concentration (0–1200 ppm) and hydride rim thickness (10–90 lm) were also
varied. The strain rate was kept at 104/s to simulate in-service steady-state
conditions and the tests were conducted both at room temperature and 300 C.
All test specimens were of small-gauge-section, cut-outs from cladding, and
guide tubes. The loading configurations included slotted-arc test (SAT) on half-
ring-shaped specimens and uniaxial tension test (UTT) on dog-bone-shaped cut-
outs. Further, prompted by the finite-element analysis of the gauge-section
region, a unique geometry of internal slotted-arc specimens with parallel gauge
section (ISATP) was chosen. Detailed stress–strain curves for all tests were
measured, and post-test fractography and local hydrogen concentrations within
the gauge sections were measured by hot extractions. Comparative data on the
measured strengths and elongations for the three types of hydride distributions
(i.e., uniform, rimmed, and layered) are presented. Quantification and analyses of
these effects have provided a general constitutive stress–strain relationship for
assessing margins to cladding or guide tube failures.
Keywords
zirconium alloys, hydrides, mechanical properties
Introduction
Zirconium-alloy fuel cladding in light water reactors (LWRs) is exposed to an envi-
ronment of neutron irradiation in high-temperature, high-pressure water coolant.
As a result, the cladding suffers from embrittlement, not only from the neutron
damage in the alloy material, but also because of the hydrogen that is picked up by
the alloy from the coolant corrosion reaction. The effect of irradiation and hydro-
gen on the cladding ductility has been reported extensively in the literature.
Under in-service conditions, a part of the hydrogen remains in solid solution in
the alloy but the rest, exceeding local solubility limit, is precipitated as hydrides.
These hydride precipitates are often non-homogenously distributed in the cladding.
For example, a hydride rim is formed near the outer oxide–metal interface because
of the temperature gradients that exist during operation. With increasing fuel
burnup, the hydride rim not only becomes denser but is accompanied by gradients
in local hydrogen and hydride concentrations through the rest of the cladding wall
thickness. In addition, the hydride morphology predominantly depends on the alloy
processing (e.g., SRA or RXA). Because the hydrides are inherently brittle, this
raises concerns about cladding integrity, especially at high burnups. There are
many instances of severely hydride cladding failures under in-service conditions or
in post-operation handling of the fuel. But at low hydrogen levels, the unhydrided
alloy matrix ligaments can often support considerable loads.
YAGNIK ET AL., DOI 10.1520/STP154320120192 1079
Based on experimental programs of mechanical testing spanning over several

years, we have examined and published data on the (1) loss of ductility with uni-
form hydriding [1], (2) effect of radial hydrides and hydride spacing [2], and (3)
effect of very high strain rates [3] on irradiated and unirradiated zirconium alloy
specimens. A common emphasis in these investigations was the need to interpret
the mechanical test results based on the local hydrogen in the gauge sections of the
test specimens, and the orientation of hydrides with respect to loading direction.
Regarding inherent non-homogeniety of hydrides precipitates, we have previ-
ously reported [4] on the effects of local hydride blisters resulting from the localized
cooling because of oxide spallation. Tubular burst test specimens from cladding
without and with such hydride blister of different sizes were tested at low strain
rates. In this paper, mechanical tests performed on slotted arc test (SAT) specimens
and uniaxial tensile test (UTT) specimens with hydride localization in the form of
circumferential hydride rims is investigated. A comparison was pursued among
unirradiated specimens with three different hydride distributions, viz., uniform,
rimmed, and layered. The material was low-tin Zircaloy-4 stress-relieved anneal
(SRA) cladding.
In Ref 1, we did report mechanical properties of irradiated SRA fuel cladding.
Subsequently, in this paper, we report the results from UTT of irradiated SRA clad-
ding and RXA guide tubes. For comparable hydrogen contents in the gauge sec-
tions, a comparison between the SRA and RXA strength and elongation data in
UTT testing has been attempted at room temperature and at 300 C. All test speci-
mens are of small gauge section in this investigation: half-ring-shaped cut-outs
from cladding and guide tubes, emulating plane-strain loading configuration and
dog-bone tensile test specimen, emulating plane-stress configuration. The loading
configuration in the half-ring arc-test specimens of different designs were evaluated
by FEM and the one with minimum bending during testing was selected.
Experimental
MATERIALS
The RXA Zr-4 guide tube material was irradiated in Ringhals-2 to a fluence of
7 1025 n/cm2 (E > 1 MeV). It was manufactured by then Siemens KWU. Ref 1
provides alloy composition and initial tubing dimensions as well as a typical exam-
ple of hydrides in this irradiated material (reproduced here from Ref 1 as Fig. 1).
The reason for such high hydriding (up to 1800 ppm) of the guide tubes is not
fully clear [5] and is outside the scope of this investigation, which is focused on me-
chanical testing of materials with SRA and RXA heat treatments and on the effects
of hydride distribution.
The hydride rims, gradients, and layers were introduced and tested in unirradi-
ated low-tin, stress-relief annealed (SRA) Zircaloy-4 cladding specimens, lot
DBV81, fabricated by the Sandvik Special Metals Corporation (SSM). The lot com-
position and final anneal conditions are given in Table 1.
FIG. 1 Cross section of guide tube with uniform hydrogen. The initial guide tube wall
thickness is 0.43 mm.
The irradiated SRA cladding is of the same pedigree as was used in our previ-
ous investigation; the details of which are given in Ref 1.
HYDRIDING
The mechanical testing specimens for the guide tube were fabricated using
as-irradiated guide tube segments shipped from Ringhals-2 to the hot cell lab in
INER. No additional laboratory hydriding was performed. The experimental chal-
lenge was the precision machining of the semicircular arc test specimens with slots
in the gauge section in relatively thin walled (wall thickness 0.43 mm) guide tubes.
Another experimental challenge for the cladding material was to specifically
introduce at various total H contents, the desired hydride distributions described in
this paper as uniform, rimmed, and layered hydride distributions. Uniform hydrid-
ing was achieved with gas phase hydrogen charging in a tubing piece 16–18 cm
long, followed by temperature cycling between 300 C and 90 C. To develop hydride
distribution that is typical of that in a fuel cladding, a temperature gradient was
imposed on the uniformly hydrogen charged specimens of various hydrogen
TABLE 1 Cladding material specifications.
Element Sn Fe Cr O Ni Fe þ Cr Final
(wt. %) Annealing
DBV 81a 1.4 0.24 0.11 0.13 – 0.35 6 h at 507 C

a
ICP analysis.
concentrations. Alternatively, to develop a layered hydride distribution on

200 ppm uniformly charged specimens, a cathodic charging method was
employed. Further details on these techniques are given in Ref 2.
Figure 2 shows optical micrographs of the three different hydride distributions
at 500 ppm average hydrogen concentration.
SPECIMEN GEOMETRIES
The mechanical testing was performed in two different specimen geometries. First,
the uniaxial tension tests conducted on dog-bone-type miniature cut-outs from
cladding or guide tubes to emulate plane-stress conditions. In this case, the precipi-
tate planes of circumferential hydrides are in the loading direction. Second, the
semicircular arc tests with slots in its gauge section, where three different variations
in the ring geometry were evaluated. These ring tests emulated plane-strain condi-
tions, where the circumferential hydrides are in the direction of the loading. For
both specimen geometries, essentially similar methodology was applied for validat-
ing the experimental test procedure as well as for characterizations such as the
hydrogen hot extractions locally from the gauge section, metallography, and frac-
tography. These are elaborated further in the next two sub-sections only for the arc
tests.
FEASIBILITY STUDY FOR HALF-RING ARC TESTS

Each ring test was performed on a 1-cm-long semicircular specimen cut-out from
the guide tube or the cladding with the loading applied in the azimuthal direction
with respect to the tube. Three different types of specimen geometries were exam-
ined in a feasibility study and analyzed by finite element methods for assessing
bending effects during the test: (1) ISTAP (internal slotted arc tension with parallel
gauge section), (2) ESAT (external slotted arc tension), and (3) ISAT (internal slot-
ted arc tension). These are shown, respectively, as Fig. 3(a), 3(b), and 3(c).
In Ref 1, ESAT specimen geometry was used. When using the external slot in
the gauge section for ESAT geometry, the concomitant drawback for the fuel
FIG. 2 Optical micrographs for three distinct hydride distribution specimens with
500 ppm hydrogen level. The initial cladding wall thickness is 0.72 mm.
FIG. 3 Dimensions of plane-strain tension specimens of various designs (for tube

OD ¼ 10.75 6 0.05 mm).
cladding specimens is that the high hydrogen hydride rim needs to be machined
out. Obviously, this was not an issue for irradiated guide tubes with uniform
hydrides throughout the wall thickness reported previously [1] and in this paper.
Nevertheless, the comparative FEM analysis of the three geometries provided some
useful insights.
With reference to Fig. 4, the bending effect was assessed as follows:
jDri0 j jri r0 j
Bending Effect ¼ ¼
rmean ðri þ r0 Þ=2
Where ri is the hoop stress at the cladding inside surface, r0 is the hoop stress at
the cladding outside surface, and rmean is the mean value of (ri þ r0)/2.
Clearly, the specimen design that could possibly preserve or retain the outer
hydride rim (or layer) and minimize specimen bending effects would best serve the
purpose. However, no specimen design could meet both requirements at once. As
FIG. 4 Schematic representation of stress distribution though wall thickness.
shown in Fig. 5, based on the FEM analyses, among the specimen designs investigated,
the ESAT specimen exhibits negligible bending effects with increasing displacements,
albeit with the drawback, as previously mentioned, that more interesting hydride rim
at clad OD will be machined out. By comparison, ISAT specimens have the advantage
FIG. 5 A comparison of the bending effects associated with plane-strain tension

specimens of various designs.
of the specimen configuration to retain or preserve the outer hydride rim (or layer),
but suffer from very large bending effects at low displacements. A compromise is
struck in ISATP, where with small grinding at OD to provide a more flattened parallel
gauge section (see Fig. 3(a)), the bending effect is reduced by approximately a factor of
4 compared to ISAT. Figure 5 also shows that the bending effect disappears for all three
specimen geometries after 0.04 mm displacement. Thus, ISATP specimen geometry
was chosen for minimizing the bending effect for testing the half-ring-type cut-outs
while preserving most of outer hydrides.
Further, as Fig. 6 illustrates, 65 % of the gauge width in the middle of an
ISATP specimen is in the plane-strain loading state. Also shown in this figure are
the three principal strain directions (E11, E22, and E33) with reference to the gauge
section. The specimen is deformed in the E11–E22 plane, and only negligibly small
strains occur in the E33 direction.
SPECIMEN FABRICATION AND ASSOCIATED HYDRIDE CHARACTERIZATIONS

UTT and ESAT specimen fabrication has been described previously in Ref 1.
ISATP specimens were fabricated by a sequential process of strip grinding,
hydriding, and machining. Two parallel strips were milled to a maximum depth of
150 lm on the OD of fuel cladding on diametrically opposite sides to make flat gauge
sections. The milled cladding segments were uniformly hydrided, some of which were
further treated by either a cathodic-charging process or hydride redistribution process
[2] to prepare non-uniform specimens. The internal slot was cut by a diamond tipped
wheel on either half of a hydrided cladding section of 10 mm in length to make ISATP
specimens, the slot dimensions were periodically checked by a micrometer caliper
with a resolution of 10 lm.
FIG. 6 Strain distribution in the symmetrical quadrants of an ISATP specimen.

Shown in Fig. 7 are details of the steps involved in fabricating ISATP specimens
for mechanical testing, and their proximate pieces for hydrogen hot extractions, and
metallographic characterizations from an initial cladding or guide tube segment.
A similar scheme was used for test specimens requiring hydride redistrib-
ution starting with a single 180-mm-tube length. The longer tubing made it
FIG. 7 Steps involved in preparing ISATP cladding specimens and performing hydride
characterizations of hydride-rimmed specimen.
possible to apply the requisite temperature gradient in the boiling capsule to

redistribute the hydrides through the tubing wall. Each initial piece of tubing
provided up to 16 semicircular ISATP samples. The total hydrogen concentra-
tion (H) for each mechanical test specimen in the gauge section was deduced
from a number of hot extractions made axially along the tube, as shown in
Fig. 7. Five values of H were targeted (0, 200, 500, 800, and 1200 ppm). The
hydride distribution (uniform, rimmed, or layered, depending on the method
of hydriding applied) was confirmed by optical microscopy of etched samples.
Outlier specimens, whether in terms of target total H or hydride distribution,
were excluded from mechanical testing data collection based on such detailed
pre-characterizations. Fractographic examinations of the gauge section follow-
ing the mechanical testing were also performed.
VALIDATION OF THE TEST PROCEDURE

Instron model 5582 and model 4505 machines were used to apply the load on the
unirradiated and irradiated specimens at a low strain rate of 104 per second,
respectively. To minimize specimen bending and slippage, the mating surfaces of
the gripping device were grooved and machined to snugly fit the specimen during
the test. In addition, an anti-bending mechanism was added to the specimen grip
with its trough-shaped guides positioned on the opposite ends of the specimen slot.
As depicted in Fig. 8, the guide rails of sufficient stiffness and smoothness direct the
upper and lower grips to travel in a straight line to eliminate bending. For tests at
300 C, specimens were heated by a chamber-type resistance furnace with the fur-
nace temperature stability within 62 C.
A dual-head optical extensometer, with a resolution of 4 lm, was employed
to take the strain measurements of the specimen from the indentations made inside
and outside of its gauge section concurrently during testing, which could be com-
pared to indicate the extent of the bending effect associated with the specimen. The
strain measurements were then averaged to give a mean stress-strain curve repre-
sentative of the cladding deformation behavior, nullifying the counteracting bend-
ing effects on both sides of the specimen gauge section.
The strain measurements taken by the optical extensometer were verified and
calibrated by the strain gauges at room temperature. Two ISATP specimens were
machined from archive Zircaloy-4 cladding and tested. For one test, the strain
measurements were taken from both sides of the gauge section by the dual-head op-
tical extensometer and averaged to give a mean stress-strain curve. For the other,
two miniature strain gauges were glued on both sides of the gauge section of a sib-
ling ISATP specimen for strain measurement. As compared in Fig. 9, the mean
stress-strain curves obtained, respectively, by the dual-head optical extensometer and
the miniature strain gauge for the two specimens tested are in good agreement. These
observations validate the test procedure in that the mean stress-strain curve can be
obtained by means of a dual-head optical extensometer to characterize the mechanical
properties of Zircaloy cladding specimens under plane-strain loading conditions.
FIG. 8 A schematic sketch of tension test on an internally slotted arc tension specimen
with a parallel gauge section.

All results are presented pertain to ductility behavior at low strain rates for UTT
and ISATP specimen geometry at room temperature and 300 C. Comparisons of
total elongation (TE), uniform elongation (UE) as well as yield strength (YS), and
ultimate tensile strengths (UTS) at these two temperatures are made graphically.
FIG. 9 A comparison of mean stress–strain curves measured with miniature strain

gauges and a dual-head optical extensometer for ISATP specimens tested at
room temperature.
EFFECT OF HYDRIDE DISTRIBUTION ON ELONGATION

At a given value of hydrogen concentration, the elongations for UTT specimen geom-
etry for layered, rimmed, and uniform hydrogen distribution are compared in Figs. 10
and 11 at room temperature and in Figs. 12 and 13 for at 300 C. Similar comparisons
are made for ISATP specimen geometry in Figs. 14 through 17. Liner regression line fit-
ting was done for the elongation data corresponding to each specific hydride distribu-
tion (see Table 2). For several entries in Table 2, this coefficient is rather low, reflecting
large scatter in data and inherent difficulties in achieving exactly the intended hydride
distributions and experimental difficulties in fabricating small test specimens. The lin-
ear fits are provided to assist in qualitative interpretation of the results. In comparing
Figs. 12 and 13, it is noted that, whereas the data for TE follow a certain trend with
hydride distribution, it is not the case for UE. With each linear regression in Table 2, a
coefficient of determination (R2), is also provided. This fitting coefficient shows statis-
tically how well the regression line approximates real data points. An R2 value of 1
indicates that the regression line perfectly fits the data.
At both temperatures, as expected, the elongations (TE and UE) of the three
different specimens decrease with increasing hydrogen under both plane stress
(UTT) and plane strain (ISATP). The hydride-layered specimens show the most
severe reduction in elongations with the increasing hydrogen content, whereas the
uniform specimens show the least. The rimmed hydrides fall in between. Thus, in
terms of ductility at a hydrogen level, uniform hydriding is most ductile and layered
hydrides are least ductile.
FIG. 10 Total elongations in plane-stress UTT geometry at room temperature for three
distinct hydride distributions (equations for linear regression are given in Table 2).
FIG. 11 Uniform elongations in plane-stress UTTgeometry at room temperature for three

distinct hydride distributions (equations for linear regression fitting are given in Table 2).
FIG. 12 Total elongations in plane-stress UTT geometry at 300 C for three distinct
hydride distributions (equations for linear regression fitting are given in Table 2).
FIG. 13 Uniform elongations in plane-stress UTT geometry at 300 C for three distinct
hydrogen distributions (equations for linear regression fitting are given in Table 2).
FIG. 14 Total elongation in plane-strain ISATP geometry at room temperature for two distinct
FIG. 15 Uniform elongation in plane-strain ISATP geometry at room temperature for two distinct
FIG. 16 Total elongation in plane-strain ISATP geometry at 300 C for two distinct hydrogen
distributions (equations for linear regression fitting are given in Table 2).
FIG. 17 Total elongation in plane-strain ISATP geometry at 300 C for three distinct
TABLE 2 Linear regression analysis of ductility data of SRA Zr-4 for various hydrogen distributions.
UTT ISATP

Linear Regression RT 300 C RT 300 C
3 3 a
Uniformly Total TE (%) ¼18.525–9.091 10 CH TE (%) ¼15.876–1.378 10 CH NA NAa
2 2
hydrided elongation (R ¼ 0.92) (R ¼ 0.07)
Uniform UE (%) ¼ 7.983–1.692 103 CH UE (%) ¼ 5.283–2.988 104 CH NAa UE (%) ¼ 6.481–3.943 104 CH
2 2
elongation (R ¼ 0.51) (R ¼ 0.19) (R2 ¼ 0.01)
2 3 4
Hydride- Total TE (%) ¼ 19.502–1.498 10 CH TE (%) ¼ 16.268–4.686 10 CH TE (%) ¼ 8.982–8.896 10 CH TE (%) ¼ 14.691–2.858 103 CH
rimmed elongation (R2 ¼ 0.68) (R2 ¼ 0.26) (R2 ¼ 0.67) (R2 ¼ 0.78)
YAGNIK ET AL., DOI 10.1520/STP154320120192

Uniform UE (%) ¼ 8.040–2.909 103 CH UE (%) ¼ 5.139–5.624 104 CH UE (%) ¼ 7.536–8.276 104 CH UE (%) ¼ 6.030–7.035 104 CH
elongation (R2 ¼ 0.77) (R2 ¼ 0.33) (R2 ¼ 0.82) (R2 ¼ 0.39)
2 3 3
Hydride- Total TE (%) ¼ 18.173–1.609 10 CH TE (%) ¼ 17.358–7.589 10 CH TE (%) ¼ 8.664–3.476 10 CH TE (%) ¼ 16.754–5.538 103 CH
2 2 2
layered elongation (R ¼ 0.86) (R ¼ 0.63) (R ¼ 0.72) (R2 ¼ 0.85)
3 3 3
Uniform UE (%) ¼ 8.062–4.808 10 CH UE (%) ¼ 6.118–1.650 10 CH UE (%) ¼ 7.453–3.045 10 CH UE (%) ¼ 7.462–1.789 103 CH
2 2 2
elongation (R ¼ 0.86) (R ¼ 0.72) (R ¼ 0.71) (R2 ¼ 0.52)
Note: CH, hydrogen concentration (ppm); R2, coefficient of determination for linear regression.
a
Indicates lack of reliable data.
1093
FRACTOGRAPHY
The fracture surfaces of 800 ppm uniform hydride specimens are shown in Figs. 18(a)
and 18(b) following the ISATP tests at room temperature and 300 C, respectively. Similar
fractographs for 800 ppm layered hydride specimens are shown in Figs. 19(a) and 19(b).
FRACTURE MODE DIFFERENCES BETWEEN RIMMED AND LAYERED

HYDRIDES
UTT Specimens
Fracture modes for UTT specimens following room temperature tests are shown in
longitudinal section in Fig. 20 (hydride-rimmed specimens) and Fig. 21 (hydride-lay-
ered specimens). In both micrographs, a combined mode of tensile separation and
shear fracture is apparent.
Several cracks initiate on the OD hydride rim and a dominant one propagates
through the rim into the cladding matrix along the hydride platelets. The hydride
rim itself fractures by a tensile separation with its fracture surface roughly perpen-
dicular to the tensile axis. The cladding matrix of the hydride-rimmed specimen
beneath the rim fractures by a shear mode. At room temperature, the fracture sur-
face of the cladding matrix showed a shear lip near the cladding inner surface.
The hydride population beneath the rim failed by a mixed mode of brittle and
ductile fracture, as exemplified by the rugged features of intermingled shear and
tensile fracture surfaces in Fig. 21. By contrast, the nearly hydride-denuded zone of
the alloy at near the ID fractured by a pure shear process with a fracture surface ori-
ented at 45 with respected to the tensile axis.
ISATP Specimens
Fracture modes for ISATP specimens following room temperature tests are shown in
cross section in Fig. 22 (hydride-rimmed specimens) and Fig. 23 (hydride-layered speci-
mens). The hydride rim in Fig. 22 was produced by a cathodic charging process [2],
which resulted in more densely packed hydride precipitates relative to the hydride
layer in Fig. 23, which was produced by the hydride-redistribution process [2]. Both of
these specimens are also fractured by a combined mode of tensile separation and shear
fracture. Similar to UTT specimens, the hydride rim of a hydride-rimmed specimen
fractured by a tensile-separation process with the fracture surface roughly perpendicu-
lar to the loading axis.
Indeed the fracture modes were very similar after 300 C tests (not shown) for
both types of specimen geometries, except for one discernible difference in the frac-
ture process; the whole cladding matrix, including the gradient hydride layer and
the hydride-denuded zone, tended to fracture by a shear mode with a fracture sur-
face oriented at 45 with respect to the tensile axis at 300 C.
EFFECT OF HYDRIDE DISTRIBUTION ON STRENGTH

Despite discernible effects on measured elongations, the tensile strengths (UTS and
YS) of the three different hydride distributions in both UTT and ISATP geometries
FIG. 18 (a) Fractography following room temperature ISATP test on uniformly hydrided
SRA Zircaloy-4 specimen with average 719 ppm H concentration showing
brittle failure with only few microcracks. (b) Fractography following 300 C
ISATP test on uniformly hydrided SRA Zircaloy-4 cladding specimen with
average 842 ppm H concentration showing ductile failure.
FIG. 19 (a) Fractography following room temperature ISATP test on SRA Zircaloy-4
hydride-layered specimen with average 1020 ppm H concentration showing
brittle failure with many microcracks. (b) Fractography following 300 C ISATP
test on SRA Zircaloy-4 hydride-layered specimen with average 778 ppm H
concentration showing mixture of ductile dimples and brittle microcracks.
reveal similar relationships. Within experimental scatter band, UTS and YS do not
change with hydrogen concentrations. Figures 24 and 25 show the results for UTT
geometry at room temperature and 300 C, respectively. For ISATP geometry, at
300 C, the tensile strengths (UTS and YS) are shown in Fig. 26.
COMPARISON OF DUCTILITY OF IRRADIATED GUIDE TUBES AND

IRRADIATED CLADDING
The guide tubes utilized in this investigation were RXA processing, whereas the fuel
cladding were SRA. This provided an opportunity to compare their ductility behav-
iors at a fluence level of 7–8 1025 n/cm2 (E > 1 MeV). This comparison was pos-
sible only for as-irradiated hydride distributions, which were known to be rather
uniform for RXA guide tube samples (see Fig. 1), but for the SRA cladding the
hydride distribution was that of a typical PWR cladding with its built-up hydride
FIG. 20 Longitudinal-sections of SRA Zircaloy-4 hydride-rimmed cladding specimen

with 60 lm hydride rim and average 700 ppm H after room temperature UTT.
rim. That is, the comparison between RXA and SRA was possible for a given hydro-
gen content but with different hydride distributions. In Figs. 27 and 28, the room
temperature strengths and elongations from UTT are compared. The same compar-
ison at 300 C is shown in Figs. 29 and 30. In these figures, the hydrogen content of
the SRA cladding is limited to low hydrogen content compared to those in guide
tubes, recalling that all testing for these comparisons was done on specimens from
as-irradiated material without any hydrogen charging.
FIG. 21 Longitudinal-sections of SRA Zircaloy-4 hydride-layered cladding specimen

with 80 lm hydride rim and average 843 ppm H after room temperature UTT.
FIG. 22 Cross-section of SRA Zircaloy-4 hydride-rimmed cladding specimen with

40 lm hydride rim and average 2363 ppm H tested under plane-strain tension
Comparable tests for ESAT were also performed but not included in the paper.
The RXA versus SRA comparison in such plane-strain testing is less meaningful in
that the high hydrogen rim zone had to be machined off because of specimen
design, which would have had much greater impact on the SRA cladding (with its
FIG. 23 Cross-section of SRA Zircaloy-4 hydride-layered cladding specimen with

40 lm hydride rim and average 1020 ppm H tested under plane-strain tension
FIG. 24 Tensile strengths UTS and YS at room temperature for UTT tests on SRA
Zircaloy-4 cladding specimens of three distinct hydride distributions.
FIG. 25 Tensile strengths UTS and YS at 300 C for UTT tests on SRA Zircaloy-4
cladding specimens of three distinct hydride distributions.
FIG. 26 Tensile strengths UTS and YS at 300 C for ISATP tests on SRA Zircaloy-4
cladding specimens of three distinct hydride distributions.
FIG. 27 UTT measured strengths at room temperature for RXA and SRA irradiated
materials.
FIG. 28 UTT measured elongations at room temperature for RXA and SRA irradiated
Zr-4 materials.
FIG. 29 UTT measured strengths at 300 C for RXA and SRA irradiated Zr-4 materials.
FIG. 30 UTT measured elongations at 300 C for RXA and SRA irradiated Zr-4
materials.
hydride rim removed) than on RXA guide tube (with uniform hydrogen distribu-
tion through its wall thickness).
Conclusions
An extensive mechanical testing investigation on low-tin Zircaloy-4 cladding mate-
rial of SRA processing and guide tube material of RXA processing has been
completed.
The unirradiated cladding samples were hydrided by different procedures to
introduce three distinct hydride distributions. They were further tested at room
temperature and 300 C in two loading directions with respect to the cladding tube
and associated circumferential hydrides, namely axial loading (UTT) and azimuthal
loading (ISATP). The experimental methodology for both loading directions was
validated. The bending effects for the azimuthally loaded specimens in three differ-
ent geometries were evaluated by finite element analyses, based upon which ISATP
specimen geometry was optimized and adopted.
The effects of hydrides distributions on the ductility of the cladding material
can be summarized as follows:
1. The effective tensile strengths of uniformly hydrided, hydride-rimmed, and
hydride- layered specimens show little or no dependence on hydrogen concen-
tration and hydride rim thickness.
2. The elongations decrease with increase in bulk hydrogen concentration and

hydride rim thickness.
3. At comparable hydrogen concentrations, the order of embrittlement effect
ranks as hydride layered > hydride rimmed > uniformly hydrided material.
Uniformly hydrided material sustains the largest deformation and the hydride
layered the least.
4. Fracture of uniformly hydrided material occurs by a shear process. Hydride-
rimmed and hydride-layered specimens fail by a combined process of tensile
separation (rim and dense hydrides region) and shear fracture (alloy matrix
region).
5. The interaction of hydrogen concentration and hydride distribution governs
the ductility of the cladding.
Comparison of ductility behavior for irradiated cladding (SRA) and irradiated
guide tubes (RXA) materials was limited because of available hydrogen distribu-
tions in their as-irradiated states. As a result, this comparison could not reveal a
clear-cut effect of alloy processing.
This paper underscores the effect of hydride distribution (specially, localiza-
tion) on the mechanical properties of Zr-alloy fuel cladding and guide tubes. The
testing was performed using small machined specimens, complementing the burst
tests data of Ref 4, which utilized larger tubular specimens with local hydride accu-
mulations (i.e., blisters). Recognizing the loading configuration in the testing and
statistical significance of the experimental data, the modelers can subsequently de-
velop constitutive models using generalized stress-strain criterion. Such constitutive
models can then be used in fuel performance codes to quantify the effect of hydro-
gen in terms of both its distribution and concentration.
ACKNOWLEDGEMENTS
The work was performed under the auspices of the EPRI-led NFIR program. The
authors acknowledge technical input from the NFIR members and thank them for
permission to publish the results.
References
[1] Suresh, K., Yagnik, A. H., and Kuo, R.-C., “Ductility of Zircaloy Fuel Cladding at High
Burnup,” 14th International Symposium on Zirconium in the Nuclear Industry, Stockholm,
Sweden, ASTM STP-1467, J. ASTM Int., Vol. 2, No. 5, 2005, Paper ID12423.
[2] Yagnik, S. K., Kuo, R.-C., Rashid, Y. R., Machiels, A. J., and Yang, R. L., “Effect of Hydrides
on the Mechanical Properties of Zircaloy-4,” Proceedings of the 2004 International
Meeting on LWR Fuel Performance, Orlando, FL, September 19–22 2004, Paper 1089.
[3] Nakatsuka, M. and Yagnik, S., “Effect of Hydrogen on Mechanical Properties and Failure
Morphology of LWR Fuel Cladding Tubes under Rapid Deformation,” 16th International
Symposium on Zirconium in the Nuclear Industry, Chengdu, China, ASTM STP-1529, J.
ASTM Int., Vol. 8, No. 1, 2010, Paper ID102954.
[4] Hermann, A., Yagnik, S. K., and Gavillet, D., “Effect of Local Hydride Accumulations on
Zircaloy Cladding Mechanical Properties,” 15th International Symposium on Zirconium in
the Nuclear Industry, Sunriver, OR, ASTM-1505, J. ASTM Int., Vol. 4, No. 10, 2007, Paper
ID101143.
[5] Pettersson, H., Bengtson, B., Andersson, T., Sell, H.-J., Hoffmann, P.-B., and Garzarolli, F.,
“Investigation of Increased Hydriding of Guide Tubes in Righals 2 during Cycle Startup,”
Proceedings of the 2007 International LWR Fuel Performance Meeting, San Francisco,
CA, Sept 30–Oct 3, 2007, Paper 1082.
DISCUSSION
Question from S. Anantharaman, BARC Mumbai Q1:—The reported hydrogen
contents represent the bulk value or the values at the rim?
Authors’ Response:—The reported hydrogen contents represent the value meas-

ured locally near the fracture surface of a given test specimen. Each such value is
the average hydrogen content through the thickness of the gauge section at that
location of the specimen gauge section.
Q2:—Will the presence of hydrides at the rim affect the long term corrosion of
spent fuel elements?
Authors’ Response:—The effect of hydrides at the rim on the corrosion of spent

fuel elements was not the subject of our investigation and needs to be studied
separately.
Q3:—What will be the effect of uneven distribution of hydrides on a growing

crack front, i.e., from a hydride free zone into a hydrided zone.
Authors’ Response:—In general, cracks were observed to initiate at the outer

cladding at certain sites of the densely populated hydride zone. Subsequently, the
crack propagated into less dense hydride zone or hydride-free zone. The effect of
such distribution on measured ductility is reflected in the fact the crack is propagat-
ing from more brittle region of the specimen to less brittle region.
Question from Javier Romero, Westinghouse Electric Co.:—When the parallel

faces of the ISATP samples are machined, how much of the hydride rim is removed,
and how does that affect the correlations obtained?
Authors’ Response:—In our work, ISATP specimens were fabricated by a se-

quential process of strip grinding, hydriding and machining. That is, the parallel
faces of the ISATP specimens were machined before hydriding. So there was no
hydride rim removed by machining. However, for as-irradiated fuel cladding speci-
mens, some of the hydride rim present on the outer cladding will be removed when
the parallel face of an ISATP specimen is machined according to the dimensions
given in the paper.
Questions from R. K. Chaube, NFC Hyderabad:—Are the three different hydride

types (i.e., uniform, rimmed, and layered) also possible in unirradiated and irradi-
ated guide tubes?
Authors’ Response:—From the experimental perspective, it is possible to prepare

unirradiated guide tubes to have uniform, rimmed, and layered hydride distribu-
tions. It is also possible, in principle, to hydrogen- charge irradiated guide tubes
with such unique distributions in a radiation-shielded chamber. However, as-irradi-
ated guide tubes from normal in-service conditions do not exhibit significantly
rimmed and layered hydrides. All our testing on irradiated guide tubes in our work
was limited to uniform hydride distribution.
Kimberly Colas,1 Arthur Motta,2 Mark R. Daymond,3

and Jonathan Almer4
Mechanisms of Hydride
Reorientation in Zircaloy-4
Studied in Situ
Reference
Colas, Kimberly, Motta, Arthur, Daymond, Mark R., and Almer, Jonathan, “Mechanisms of
Hydride Reorientation in Zircaloy-4 Studied in Situ,” Zirconium in the Nuclear Industry: 17th
ABSTRACT
Zirconium hydride platelet reorientation in fuel cladding during dry storage and
transportation of spent nuclear fuel is an important technological issue. Using an
in situ x-ray synchrotron diffraction technique, the detailed kinetics of hydride
precipitation and reorientation can be directly determined while the specimen is
under stress and at temperature. Hydrided Zircaloy-4 dogbone sheet samples
were submitted to various thermo-mechanical schedules, while x-ray diffraction
data was continuously recorded. Post-test metallography showed that nearly full
hydride reorientation was achieved when the applied stress was above 210 MPa.
In general, repeated thermal cycling above the terminal solid solubility
temperature increased both the reoriented hydride fraction and the connectivity
of the reoriented hydrides. The dissolution and precipitation temperatures were
determined directly from the hydride diffraction signal. The diffraction signature
17, 2014.
1
America (Corresponding author), e-mail: kimberly.colas@cea.fr. (Currently at CEA, French Atomic Energy
Commission, DEN/DANS/DMN/SEMI/LM2E, Saclay, France, 91191.)
2
America.
3
Dept. of Mechanical and Materials Engineering, Queen’s Univ., Kingston, ON K7L3N6, Canada.
4
Advanced Photon Source, Argonne National Laboratory, Argonne, IL 60439, United States of America.
5
Hyderabad, India.
Copyright V
of reoriented hydrides is different than that of in-plane hydrides. During cooling

under stress, the precipitation of reoriented hydrides occurs at lower
temperatures than the precipitation of in-plane hydrides, suggesting that applied
stress suppresses the precipitation of in-plane hydrides. The analysis of the
elastic strains determined by the shift in position of hydride and zirconium
diffraction peaks allowed following of the early stages of hydride precipitation.
Hydride particles were observed to start to nucleate with highly compressive
strain. These compressive strains quickly relax to smaller compressive strains
within 30 C of the onset of precipitation. After about half of the overall hydride
volume fraction is precipitated, hydride strains follow the thermal contraction of
the zirconium matrix. In the case of hydrides precipitating under stress, the
strains in the hydrides are different in direction and trend. Analyses performed
on the broadening of hydride diffraction peaks yielded information on the
distribution of strains in hydride population during precipitation and cooldown.
These results are discussed in light of existing models and experiments on
hydride reorientation.
Keywords
zirconium hydrides, reorientation, synchrotron x-ray diffraction
Introduction
During service in light water reactors (LWRs) hydrogen atoms from the corrosion
reaction and water radiolysis enter the zirconium alloy nuclear fuel cladding, where
they precipitate as a brittle hydride phase [1]. For older alloys, such as Zircaloy-4, at
the high burnup rates used in today’s reactors, oxide layers as thick as 100 lm and
hydrogen content up to 700 wt. ppm can be observed in the cladding [2,3]. The
hydrogen precipitates as brittle hydride platelets of the d-hydride phase (face-cen-
tered cubic, ZrH1.66), which leads to a degradation of the cladding ductility [4,5].
Although other hydride phases, such as c- or e-hydrides have also been observed
under certain conditions, they are less typical in reactor cladding [6,7]. The d-
hydride platelets are composed of thin plates, the stacking of which results in the
observed orientation of the large macroscopic hydrides seen by metallography. This
orientation depends on the texture, grain morphology, and thermal history of the
material [8]. When precipitated under no applied stress in cladding material,
hydrides tend to precipitate in the circumferential-axial plane. However, stress can
significantly affect the orientation of these hydride platelets: when precipitated
under sufficient stress (such as can be found in used fuel cladding because of fission
gas pressure at temperature), hydride platelets can precipitate with a different orien-
tation in which the hydride platelet normal is parallel to the loading direction [9].
Reorientation of these hydrides can lead to a severe embrittlement of the material,
because hydride cracking can facilitate crack propagation through the cladding
thickness. The conditions of high stress and high temperature, which could lead to
hydride reorientation, could potentially be found during transportation and dry
COLAS ET AL., DOI 10.1520/STP154320120168 1109
storage of used nuclear fuel. Thus the understanding of the hydride reorientation
mechanism is of significant importance for used nuclear fuel storage and
disposition.
Previous studies on hydride reorientation performed with conventional metal-
lography techniques have provided essential information on the threshold stresses
for reorientation and influencing factors [9,10]. However, the study of the detailed
mechanisms of nucleation and growth of reoriented hydrides and the fine structure
of these particles is difficult with such techniques as they can only access the initial
and final states. Recent in situ synchrotron experiments have shown the potential
to study hydride dissolution and precipitation using synchrotron radiation in nor-
mal mechanical test specimen [11,12], and in samples with a stress concentration
[13,14]. The in situ x-ray diffraction technique is used in this article to study the
mechanisms of hydride nucleation and growth with and without applied stress. The
hydride volume fraction, phase and strain states are continuously followed during
dissolution and precipitation. Several samples with no reorientation and full reor-
ientation and different levels of applied stresses are studied.
MATERIAL AND SAMPLE PREPARATION
The material used in this study is cold-worked Zircaloy-4 sheet of 675-lm thickness
furnished by Teledyne Wah-Chang. The as-received sheet was stress relieved for 2 h
at 510 C under a vacuum of 103 Pa, resulting in a cold-worked stress-relieved
(CWSR) state. The microstructure, texture and mechanical properties of this mate-
rial are detailed in [5]. The zirconium grains are elongated in the rolling direction
with an average grain size of 6 lm in the rolling direction, 4.5 lm in the transverse
direction and 2.5 lm in the normal direction. The crystallographic texture of the
sheet is similar to that seen in CWSR zirconium cladding as illustrated in Fig. 1 [15].
The yield stress of this material tested at several temperatures is 5–10 % lower than
that of cladding tube.
The samples are then hydrided using a gaseous charging technique as described
in [11]. This technique starts by removal of the native oxide layer by acid etching
followed by deposition of a thin nickel layer, which will act as a window for hydro-
gen atoms and prevent oxidation. After this step, the samples are heated to 450 C
in a volume filled with a mixture of 88 % argon and 12 % hydrogen to introduce
the hydrogen and held for 1 h at temperature then furnace cooled. The hydrogen
content of the hydrided samples was systematically tested using hot vacuum extrac-
tion. The hydrogen contents of the samples in this study range from 150 to 300 wt.
ppm.
Once the samples are hydrided, they are machined into small dogbone-shaped
tensile specimens, the dimensions of which are presented in Fig. 2. It is important to
note that the transverse direction (TD) is along the sample loading direction. The
rolling direction (RD) is perpendicular to the gage section along the plane of the
FIG. 1 (a) Schematic representation of the texture directions in the sheet material used
in this study; (b) {00.2} pole figure of the sheet material used in this study [5];
(c) schematic representation of the texture directions in typical cladding
material; and (d) {00.2} pole figure of typical cladding material [30].
tensile specimen. The normal direction (ND) is perpendicular to the plane of the
tensile specimen as illustrated in Fig. 1(a).
METALLOGRAPHY
Metallography is performed on the samples to observe the hydride microstructure.
The samples are mounted in epoxy casts, polished to 1200 grit silicon carbide paper
then etched for a few seconds in a solution of one volumetric part hydro-fluoric
acid (HF), 10 parts nitric acid and 10 parts H2O.
FIG. 2 Dogbone specimen geometry (sample thickness was 0.6 mm).

A small piece of the grip section of each sample is analyzed by metallography

prior to any thermo-mechanical treatment, to determine the starting microstructure
of the samples. Figure 3 shows optical micrographs of the ND-TD cross section of a
typical starting hydride microstructure for different samples (labeled initial). The
hydrides are mostly in-plane (i.e., platelets in the rolling-transverse plane). This is
due both to the texture of the CWSR material, which presents many basal poles
along the normal direction, perpendicular to the rolling-transverse plane and to the
grain morphology. Hydride platelets typically precipitate in an orientation perpen-
dicular to the basal pole because of the d-hydride{111}// a-zirconium{0002} orien-
tation relationship [16] and to the greater availability of grain boundaries of
elongated grains in CWSR. After thermo-mechanical treatment at the Advanced
Photon Source (APS) Synchrotron, the final hydride microstructure is also observed
by metallography.
FIG. 3 Thermo-mechanical cycles and hydride microstructure before and after cycle:
(a) sample with 294 wt. ppm of hydrogen, no applied stress, no reorientation; (b)
sample with 246 wt. ppm of hydrogen, 160 MPa applied stress, no reorientation;
and (c) sample with 192 wt. ppm, 240 MPa applied stress, and reorientation.
Note: all micrographs are 500 lm in width.
THERMO-MECHANICAL TREATMENT
Three different thermo-mechanical treatments are performed in this study on three
different samples with similar levels of hydrogen. These thermo-mechanical treat-
ments are shown for each sample in Fig. 3. The resulting hydride microstructure for
each sample is also presented alongside the initial hydride microstructure in Fig. 3.
In the treatment shown in Fig. 3(a) the sample is heated to full dissolution (450 C
for 294 wt. ppm of hydrogen) and cooled without any applied load. In this case, the
resulting hydride precipitates are mostly in-plane because of the initial texture.
When a sample is heated to full dissolution and cooled with a low applied stress of
160 MPa, the final hydride microstructure is similar to the unstressed sample with
most of the hydride platelets in the in-plane direction as can be seen in Fig. 3(b).
Finally, when a sample is heated until all the hydrides are dissolved and cooled
under a high applied stress of 240 MPa, many hydrides reorient upon cooling, as
seen in Fig. 3(c). The radial hydride fraction (RHF) is calculated as a weighted aver-
age of hydride lengths where hydrides with an orientation between 0 and 40 to
the transverse direction have a weight of 0, hydrides with an orientation between
40 and 65 have a weight of 0.5, and hydrides with an orientation of 65 to 90
have a weight of 1. More details on the methods of determining the RHF are
described in [11]. Calculations based on image analysis of the hydrides as described
in [11] show that the radial hydride fraction is approximately 60 % after one thermo-
mechanical cycle. After two thermo-mechanical samples under 240 MPa, full reorien-
tation is achieved (RHF 100 %). This indicates that the threshold stress for hydride
reorientation for this CWSR Zircaloy-4 sample is between 160 and 230 MPa, which is
consistent with literature results [17].
SYNCHROTRON X-RAY DIFFRACTION

X-ray diffraction (XRD) experiments were performed at beamline 1-ID at the
Advanced Photon Source Synchrotron at Argonne National Laboratory. A detailed
description of the transmission in situ diffraction technique and data analysis can
be found in Refs 11 and 18. A 76-keV x-ray beam was used for these experiments,
which allows transmission of the x rays through the sample thickness. The diffrac-
tion rings obtained are recorded onto a two-dimensional (2D) amorphous silicon
detector allowing for some texture information to be measured. Diffraction data is
recorded every 30 s during heating and cooling of the sample, allowing for the full
hydride dissolution and precipitation kinetics to be studied.
The obtained diffraction rings are integrated along specific azimuthal directions
(65 ). Families of planes perpendicular to the TD (direction of the applied stress)
and perpendicular to the RD are studied in this experiment. Once integrated, the
diffraction peaks are then fitted using software called General Structure Analysis
System (GSAS) by iterative refinement of the peaks’ shapes, intensities, and widths
[19]. The peak shape chosen for these fits is a pseudo-Voigt shape, which is a con-
volution of Gaussian and Lorentzian shapes typically used for synchrotron diffrac-
tion peaks. The focus of this study is on the most intense hydride peak d{111} and
its neighboring zirconium a{10.0} peak. The fitting error for the zirconium a{10.0}
peak in a sample with 294 wt. ppm of hydrogen in the TD at room temperature is
less than 1 % for the peak intensity and 0.002 % for the d-spacing (which corre-
sponds to an error of about 0.01 millistrain for the zirconium elastic strain at room
temperature). For the d{111} hydride peak in the same sample at the same tempera-
ture, the error is less than 5 % for the peak intensity and 0.023 % for the d-spacing
(which corresponds to an error of approximately 0.22 millistrain for the hydride
elastic strain at room temperature). At 400 C where the hydride peak is much
smaller, the error for the hydride peak become close to 45 % for the peak intensity
and 0.36 % for the d-spacing (which corresponds to an error of approximately 3.5
millistrain for the hydride elastic strain at that temperature). The error for the fit-
ting of the zirconium peak is the same at 400 C than at room temperature. The
error bars were not represented for the zirconium and hydrides strain plots in Figs.
7–13, and for the Williamson-Hall plots in Fig. 16 for the sake of clarity, but the error
values mentioned above applicable to all observed samples. The value of the typical
error in the Williamson-Hall plots in Fig. 16 is equivalent to the scatter in the data.
Results
DISSOLUTION AND PRECIPITATION KINETICS
As described in Ref 11, the integrated intensity of the hydride peak d{111} can be
followed continuously during heating and cooling of the sample. The heating rate
for the samples observed in this study is 25 C/min and the cooling rate is 1 C/min.
As the sample is heated and hydrides dissolve, the hydride peak intensity decreases,
reaching zero when full dissolution is achieved, which allows determination of dis-
solution temperature, Td. When hydrides start to re-precipitate, the hydride peak
reappears at the precipitation temperature Tp, and increases as more and more
hydrides precipitate. After full re-precipitation is achieved, the hydride peak inten-
sity is similar to the initial value for sample cooled under no applied stress. The
determination of Tp and Td was performed for several samples of various hydrogen
contents, and the results compared with differential scanning calorimetry (DSC)
determination [11]. Figure 4 shows the experimental Td and Tp compared to Td and
Tp curves determined in Une and Ishimoto [20] using differential scanning calorim-
etry [11]. The measured Td and Tp for all samples correspond reasonably well with
the temperatures measured in Ref 20 using DSC. The temperature hysteresis
(Td Tp) is also similar to that observed in Ref 20. The temperature error bars in
this figure were determined by the heating/cooling rate and the time between
acquisitions of successive diffraction patterns. The good agreement between the
synchrotron XRD technique and the DSC techniques validates the use of the XRD
technique to study hydride dissolution and precipitation.
It is interesting to note that when hydrides are already present at the start of
cooling (full dissolution not achieved) the hysteresis disappears. This is illustrated
in Fig. 5, which shows the hydride diffraction signal from a sample containing
FIG. 4 Dissolution and precipitation temperatures measured from in situ XRD

compared to DSC [20].
FIG. 5 Evolution of the intensity of the hydride d{111} peak intensity with temperature
when full dissolution is not achieved (CWSR Zircaloy-4 sample with 246 wt. ppm
of hydrogen.
246 wt. ppm of hydrogen, which was subjected to a heat treatment at 400 C (Td for
246 wt. ppm of hydrogen is 430 C). Because not all the hydrides dissolve in these
conditions, hydrides are already present when cooldown starts. As shown in Fig. 5,
the diffracted intensity starts to increase as soon as the temperature is decreased
(without hysteresis) indicating that the hysteresis is linked to the initial hydride
precipitation.
To understand the mechanisms of hydride reorientation, it is important to
understand the effect of stress on hydride precipitation. Studies have shown that
whereas stress does not significantly affect the solubility of hydrogen in Zircaloy, it
has a strong effect on hydrogen mobility [21,22]. However, most of the studies on
the effect of stress on hydride precipitation have been performed on samples in
which hydride reorientation did not occur. The effect of stress and reorientation on
hydride precipitation kinetics is presented in this study. The precipitation tempera-
ture was measured for samples with different hydrogen content where the hydrides
were dissolved without applied stress then precipitated under different levels of
applied stress (no stress, low stress below threshold stress for reorientation, and
high stress above the threshold stress for reorientation). The ratio of the precipita-
tion temperature measured in situ by XRD over that expected for unstressed mate-
rial from DSC [20] is presented in Fig. 6. For samples precipitated under no applied
stress and under low applied stress (60–160 MPa), it can be seen that the ratio is
close to one which signifies the precipitation temperature remains well within the
values obtained from DSC [20], and thus no significant effect of stress on Tp is
observed for samples in which no hydride reorientation occurs. This is consistent
with the literature results by Kammenzind et al. [21]. The error bars in Fig. 6 were
determined by the heating/cooling rate and the time between acquisitions of succes-
sive diffraction patterns similarly to those presented in Fig. 4. The effect of stress on
the dissolution temperature is not presented here, because in most of our studies dis-
solution was performed under no applied stress (previous studies on the effect of
stress on the dissolution temperature are presented in Ref 11). The precipitation tem-
perature measured for samples precipitated under high applied stress (between 230
and 240 MPa), which showed some hydride reorientation are presented on the right-
hand side of Fig. 6. It can be seen that for samples in which hydride reorientation
occurs, the precipitation temperature is below the value measured from DSC. This
means that a greater degree of undercooling is required to precipitate hydrides in the
reoriented direction than to precipitate hydrides in their common in-plane
orientation.
EFFECT OF HYDROGEN IN SOLID SOLUTION ON ZIRCONIUM STRAINS

The study of the diffraction peak positions of the hydride d{111} and zirconium
a{10.0} gives information on the elastic strains of these two phases during hydride
dissolution and precipitation. The zirconium strain represented here is the uniaxial
strain calculated using the first data point at time zero (room temperature) in the
following manner:
FIG. 6 Ratio of the precipitation temperature measured in situ with XRD for several
CWSR Zircaloy-4 samples with hydrogen content ranging from 80 to 530 wt.
ppm precipitated under different levels of stress (no stress, low stress of
60–160 MPa and high stress between 230 and 240 MPa) over the predicted
precipitation temperature without stress from DSC [20]. The threshold stress for
hydride reorientation for this material is 200 MPa.
dðfhkl g; T Þ d ðfhkl g; 30 CÞ

(1) e¼
dðfhklg; 30 CÞ
The reference d-spacing at room temperature for the zirconium peaks comes
from the first diffraction pattern, which was recorded at room temperature (30 C)
before any thermo-mechanical treatment. The formula above is only valid for the
calculation of the zirconium strain; for the calculation of the hydride strain, the
same formula is used but the reference d-spacing is no longer taken at 30 C. Fur-
ther discussion on the reference d-spacing for the hydride strain calculation is pre-
sented in the next section.
By taking an unhydrided Zircaloy-4 sample through the temperature schedule
for dissolution and precipitation and measuring the d-spacing increase with tem-
perature, the coefficient of thermal expansion in the “a” direction for the hexagonal
structure was calculated to be about 6.2 106 C1 in the transverse direction and
FIG. 7 Schematic of the effect of hydrogen in solid solution on zirconium strains with
temperature in unhydrided and hydrided samples.
5.6 106 C1 in the rolling direction. Using the evolution of strain in the zirco-
nium a{00.2} peak with temperature, the coefficient of thermal expansion in the “c”
direction was found to be 8.2 106 C1 in the transverse direction (there are
almost no basal planes in the rolling direction, so the coefficient of thermal expan-
sion was only calculated in the TD). These values are in close agreement with the
previously reported values of the thermal expansion in the “a” direction, which is
5.8 106 C1, and that in the “c” direction, which is 7.6 106 C1 for CWSR
Zircaloy-4 [23].
The zirconium strain behavior of hydrided samples during heating and cool-
ing is slightly different than that of unhydrided samples. An additional contribu-
tion to the zirconium matrix strain is caused by hydrogen atoms in solid solution
as the hydrides dissolve and re-precipitate. The expected effect of hydrogen atoms
in solid solution on zirconium strains is schematically represented in Fig. 7. As the
temperature increases from room temperature, the zirconium a{10.0} d-spacing
increases proportionally as predicted by the thermal expansion coefficient of
6.2 106 C1 in the transverse direction. When hydrides start to dissolve, the
hydrogen atoms in solid solution cause an additional expansion of the zirconium
planes (red, right-most curve in Fig. 7). This additional expansion ends when all
hydrides are dissolved (Td) leaving again only thermal expansion strain. During
cooldown, the same process is seen in reverse as the hydrogen atoms are removed
from solid solution for T < Tp.
Thus, the change of slope in zirconium d-spacing can also be used to determine
the dissolution and precipitation temperatures. Figure 8 shows the measured strain
FIG. 8 Strain in TD in the zirconium a{10.0} peak in a sample with 129 wt. ppm of
hydrogen heated and cooled under no applied stress. The strain in unhydrided
material is represented as well.
in the TD for the zirconium a{10.0} peak in a sample with 129 wt. ppm of hydro-
gen. As the temperature increases, the strain initially follows the expected thermal
expansion behavior. At around 250 C, significant hydride dissolution starts to
occur. This process accelerates, until above 380 C (Td) the strain again follows the
thermal expansion curve (all hydrogen in solid solution). Upon cooling, a change of
slope of the zirconium strain occurs at about the hydride precipitation temperature,
as hydrides start precipitating and fewer atoms remain in solid solution in the ma-
trix, which reduces strain. Once most hydrides have precipitated, the zirconium
strain becomes again governed by thermal expansion only. The deviation from the
ideal d-spacing is more pronounced for samples with higher hydrogen content, as
can be seen in Fig. 9 where zirconium strains in a sample containing 294 wt. ppm of
hydrogen are shown. These effects of hydrogen on zirconium strains are coherent
with the schematic representation of Fig. 7.
The hydride phase elastic strains during hydride dissolution and precipitation
can also be studied by a manner similar as in the zirconium phase as presented in
the following paragraphs.
HYDRIDE STRAINS DURING NUCLEATION AND GROWTH

Thermo-Mechanical Cycle Without Applied Stress
The elastic strain in the hydride as measured using the d{111} planes in the TD in a
sample with 294 wt. ppm of hydrogen are presented in Fig. 10 as the hydrides are
dissolved and re-precipitated. As the temperature increases, the hydrides dissolve
and the hydride d-spacing follows the same trend as the zirconium d-spacing (as
FIG. 9 Strain in TD in the zirconium a{10.0} peak in a sample with 294 wt. ppm of
hydrogen heated and cooled under no applied stress.
seen in Fig. 9), rather than the thermal expansion coefficient of hydrides of
14.2 106 C1 [23], likely because their expansion is limited by being embedded
in the matrix. In the final stages of dissolution, the hydrogen in solid solution
increases the zirconium d-spacing (as discussed in the previous paragraph), whereas
the hydride d-spacing is very close to the unstressed d-spacing value from the
FIG. 10 Hydride strain behavior during cooling (calculated from d{111} peaks in the TD)
in a sample with 294 wt. ppm of hydrogen cooled without applied stress.
Specific locations noted that A, B, C, D, and E correspond to the schematic in
Fig. 17.
FIG. 11 Strain in RD in the hydride d{111} peak in a sample with 294 wt. ppm of hydrogen
heated and cooled under no applied stress.
literature [24] calculated at T ¼ 450 C, of 2.7672 Å. This value of the d-spacing was
found at the final stage of dissolution, approximately, for most samples, and there-
fore this final d-spacing value was taken as the reference value for unstressed
hydride lattice parameter.
Beyond 450 C, all diffraction signals disappear, so the hydrides are completely
dissolved (point A). As the temperature is decreased, precipitation starts to occur
when the temperature falls below the precipitation temperature of 398 C (point B).
At that stage, a small diffraction peak appears. Analysis of the peak full-width at
half-maximum (see section on Hydride Peak Broadening below) shows that down
to 354 C, size broadening is the dominant mechanism. The other feature is that the
d{111} d-spacing in the TD (Fig. 10) and in the RD (Fig. 11) is much lower than the
unstressed value, likely indicating that these particles are under compression at
about 10 millistrains, which corresponds to a stress of 660 MPa at their edges.6 The
average compressive strain of the hydride population decreases as the temperature
decreases to 375 C (point C). This could be because of a change of shape from
sphere to plate as the precipitate grows. In the case of platelet-shaped precipitates,
shear stresses can lead to formation of dislocations in the matrix (whereas for
spheres, only hydrostatic stresses are present, thus creating no shear stresses that
generate dislocations). It is also possible that “sympathetic nucleation,” by which a
new hydride nucleus precipitates within the strain field of neighboring hydrides
6
It is also possible that this high compressive strain, which is a diffraction peak shift, could be because of
the initially forming hydrides having a slightly different stoichiometry than the Zr/H ratio of 1.66 typical in
the d-hydride phase. Within the d-hydride phase, a maximum change in d-spacing because of stoichiometry
of 0.002 nm could lead to a pseudo-strain of 4 millistrain [26,27].
(thus requiring less strain to form), contributes to the decrease in average strain of
the hydride population during precipitate growth in cooldown. The available data
does not enable us to distinguish between these possible causes. In addition, the
phenomenon of high stress zirconium creep could induce local relaxation of the
hydride strains. However, the zirconium diffraction data is averaged over many
grains, a majority of which do not contain hydrides. Therefore, local effects on the
zirconium phase cannot be measured with this technique.
One might consider that the higher differential thermal expansion of hydrides
(14.2 106 C1) compared to that of zirconium (6 106 C1) [23] could
cause some relaxation upon cooldown as the hydride shrinks away from the ma-
trix. A simple calculation indicates that it would only account for 7 %–10 % of the
strain relaxation observed from 400 C to 375 C [on a temperature change of
50 C, ethermal difference ¼ 50 (14.2 106 6 106) ¼ 0.4 millistrain]. At 375 C,
half of the hydrides have precipitated and the strains start to follow the thermal
expansion of zirconium, i.e., decreasing with decreasing temperature (from point
C to points D and E). This suggests that below this temperature less nucleation
occurs and additional hydrogen precipitation occurs by growth of existing
hydrides. As a consequence, less relaxation is observed associated with a post-
nucleation change of shape and/or dislocation formation. The strain value at
room temperature after the thermal cycle suggests that hydrides are subjected to a
compressive stress of about 500–700 MPa, which is below the hydride yield
stress of about 800 MPa [26–28] (calculated using a multi-axial stress state) [18].
The observed strain behavior is the same in the rolling direction as in the trans-
verse direction as can be seen in Fig. 11. This is logical because, in both cases, the
planes observed are the side {111} planes of the hydride platelet, and the texture
and microstructure of the edges or the hydrides are expected to be similar in the
TD and the RD.
THERMO-MECHANICAL CYCLE WITH 160 MPA APPLIED STRESS: NO HYDRIDE

REORIENTATION
In the sample studied here, a stress of 160 MPa was applied, which was not suffi-
cient to induce any reorientation of hydrides as seen in Fig. 3. This allows us to
investigate the effects of stress in circumferential hydrides and reoriented hydrides.
Figure 12 shows the strains in the zirconium matrix, calculated using the zirconium
a{10.0} peaks, as a function of temperature in the TD and the RD. The 160 MPa
stress was applied in the TD during cooling. The tensile frame was operated in load
control, ensuring that a constant load was applied on the sample during cooling. As
the temperature increases, the d-spacing increases because of thermal expansion.
When hydrogen starts to dissolve into the matrix in significant quantities, the strain
deviates from thermal expansion as shown in Fig. 12. The change of slope of the zir-
conium matrix strain as hydrides dissolve and re-precipitate is similar to that
observed in Fig. 8. The load was applied at high temperature and removed at 150 C
as shown in Fig. 12. The applied load stretches the zirconium a{10.0} planes in the
FIG. 12 Strain in the zirconium a{10.0} peak in a sample with 124 wt. ppm of hydrogen
heated and cooled under 160 MPa tensile stress: (a) in the TD; and (b) in the RD.
TD and compresses them in the RD by Poisson’s effect, as seen in Fig. 12.7 When
the load is removed, the value of the Zr{10.0} d-spacing in the RD returns to its
expected d-spacing value; however, the TD planes remain a little stretched by
approximately 0.2 millistrain compared to the unstressed value in Fig. 8. This could
be because of relaxation of the residual compressive strains of the zirconium matrix
caused by manufacturing of the material (the effect is mitigated in the RD because
of a Poisson’s ratio of 0.32 [28]).
Figure 13 shows the elastic strains in the d{111} hydride peaks measured in the
TD and the RD for a sample with 245 wt. ppm of hydrogen when cooled under
FIG. 13 Strain in the hydride d{111} peak in a sample with 245 wt. ppm of hydrogen
heated and cooled under 160 MPa tensile stress: (a) in the TD; and (b) in the RD.
7
When a tensile stress is applied in the TD, Poisson’s ratio is defined as ¼ deðNDÞ=deðTDÞ
¼ deðRDÞ=deðTDÞ. For small deformations, DdðRDÞ=DdðTDÞ ¼ DdðNDÞ=DdðTDÞ [3].
160 MPa applied tensile stress. The evolution of hydride strains during hydride dis-
solution shown in Figs. 13(a) and 13(b) is similar to that observed in Fig. 10. This is
logical because hydrides were dissolved under no applied stress. During cooling, the
first regime of precipitation is similar to that of unstressed samples and starts at
about the same compressive strain of 10 millistrain as seen in Fig. 13(a). Even when
formed under stress, hydride particles start as highly compressed small precipitates,
then relax some of these compressive strains as they grow, change shape, and form
dislocations as discussed above. The change to the second regime of precipitation
occurs at 350 C, which is also when half of the hydrides are precipitated as can be
determined by the hydride volume fraction curve obtained from the hydride peak
intensity curve similarly to that obtained in Fig. 10, but not shown here for the sake
of clarity. In the case of the 294 wt. ppm sample, the change of precipitation regime
also occurred at a temperature when half of the hydride population was precipi-
tated. However, the hydride strains observed in this secondary regime at lower tem-
peratures are different from those observed in the unstressed sample. In the
stressed sample, hydride compressive strains depend less on temperature and are
more constant, even though the entire system is cooling. This could be explained by
the fact that under a far-field strain, the zirconium matrix deforms more easily and
transfers some of the tensile strain to the hydride. This load transfer gives enough
tensile stress to compensate for the compressive strains that would otherwise appear
because of cooling. This explanation has been advanced by Kerr et al. when they
studied room temperature deformation of hydrided samples [14]. Finally, this ex-
planation is also in agreement with the fact that when the applied load is removed
at 150 C, the hydrides planes in the RD are slightly less compressed (and those in
the TD are slightly more compressed) than initially at that temperature (seen from
the dissolution curve), which would indicate that some strain relaxation occurs in
hydrides grown under tensile stress.
THERMO-MECHANICAL CYCLE WITH 240 MPA APPLIED STRESS: HYDRIDE

REORIENTATION
In this section, we discuss a sample that has been cooled under an applied stress of
240 MPa, thus leading to hydride reorientation, as seen in Fig. 3. Figure 14 shows the
strain in the hydride d{111} peak for the planes oriented along the transverse and
the rolling directions. During heating (performed under no applied stress), the
strain curves in the TD and the RD are similar to those observed for unstressed
samples as would be expected. As shown in Fig. 14(a), when the temperature reaches
330 C during cooling, the hydrides initially precipitate in a highly compressed state.
As the temperature decreases further, the magnitude of the strain in the TD
decreases and then actually becomes positive. The hydride strain in the RD, shown
in Fig. 14(b), are also seen to be compressive at the onset of hydride precipitation,
and the compressive strain decreases as the temperature decreases, remaining how-
ever in the compressive regime throughout. Once the secondary regime kicks in,
FIG. 14 Strain of the hydride {111} peak in a sample with 192 wt. ppm of hydrogen cooled
under a 240 MPa applied stress applied in the TD, reorientation of hydrides: (a)
in the TD; and (b) the RD.
the strains remain approximately constant, except one is tensile and the other
compressive.
The measured hydride tensile strains in the TD could be because of the fact
that we are now observing reoriented hydride planes on the face of the hydride pla-
telet and no longer on the edge, as for circumferential hydrides and due to the
applied tensile stress in the TD, those reoriented hydride faces could be in tension.
The measured hydride compressive strains in the RD could be caused by the com-
pressive stress applied by the matrix because of the applied tensile stress in the TD
(compression by Poisson’s effect), or because the hydride planes in the RD are the
edges of the reoriented hydride platelets, in contrast to those planes observed in
the TD.
HYDRIDE PEAK BROADENING

The third parameter analyzed by the fitting of the diffraction peaks is the peak width or
full-width at half-maximum (FWHM). The broadening of these x-ray diffraction peaks
can be caused by several factors including instrumental broadening. The instrumental
broadening was measured using a powder standard and removed by quadratic subtrac-
tion [FWHMsample ¼ H(FWHMmeasured2 FWHMinstrumental2)]. Once instrumental
broadening has been removed, both strain broadening and size broadening need to be
considered. To quantitatively differentiate between strain and size broadening and to
better understand the general state of strain of the hydrides, an analysis was carried out
using Williamson–Hall plots [29]. The following relations allow us to distinguish
between the two types of broadening:
sin h
(2) FWHMsample / elattice strain broadening
cos h
0:9k
(3) FWHMsample / ¼ d size broadening
t cos h
FIG. 15 Schematic representation of a Williamson-Hall plot.
where FWHMsample is the measured Gaussian FWHM minus the instrumental broad-
ening (in radians 2h), h is the Bragg angle (in radians), elattice is the root-mean-square
strain, t is the sample particle size (in nm), and k is the x-ray beam wavelength (in
nm). A Williamson–Hall plot presents FWHM cos h- noted B- as a function of sin h
for different peaks of the same phase as represented schematically in Fig. 15. Strain
broadening varies with the 2h value of the peak under consideration, whereas size
broadening is independent of angle when plotted in the Williamson-Hall plots. There-
fore, in a Williamson-Hall plot, the slope of the curves plotted is proportional to strain,
and the y-intercept is proportional to the amount of size broadening.
This plot was done for the d-hydride phase using the {111} and {220} peaks
during cooling, the slopes and y-intercept were then obtained for the sample with
294 wt. ppm of hydrogen heated and cooled under no applied stress. The evolution
of the slope and intercept of these Williamson-Hall curves (WH) can then be fol-
lowed as a function of temperature as the hydrides are dissolved and re-
precipitated. The results for these fits are presented in Fig. 16. Fits of higher
FIG. 16 (a) Slope, and (b) intercept of the Williamson-Hall plots for a sample with
294 wt. ppm of hydrogen, heated and cooled under no applied stress.
temperature diffraction patterns are, of course, less accurate than those obtained at
lower temperatures because, at the onset of hydride precipitation, hydride diffrac-
tion peaks are quite small and thus difficult to fit with confidence. The scatter
observed in the determination of the slope is caused in part by the difficulty of fit-
ting small diffraction peaks. In addition, the WH fits in this figure are based on the
only two hydride diffraction peaks available, and thus would add another level of
uncertainty and increase scatter in the WH data. However, the overall trends of the
change of slope with temperature can be discerned in these WH plots.
Indeed, it can be seen in Fig. 16 that the slope of the WH plot decreases as the
temperature increases, approaching zero at the dissolution temperature. This is log-
ical as the strain should decrease as the particle size becomes smaller. In contrast
the y-intercept is small initially, increasing as the particle dissolves. This suggests
that above 375 C, there is mostly size broadening in our samples, whereas below
375 C, the peak broadening is dominated by strain. In Fig. 16(b), the size of the par-
ticles as calculated using Eq 3 is shown. It should be noted that after 325 C, only
strain broadening is detected, so the particle size may be changing without being
detected by the size-broadening technique. The initial diameter of the particles
forming at high temperatures, calculated using Eq 3, is about 25 Ås. When the par-
ticle size reaches approximately 300 Ås, the size-broadening effect becomes negligi-
ble compared to the strain-broadening effect.
Discussion
Two different aspects of hydride and matrix strains have been presented in the pre-
vious sections: (i) the evolution of elastic strains calculated from the shift in peak
positions, and (ii) the evolution of the FWHM of these diffraction peaks with tem-
perature. The first yields information on the average elastic strains in the diffracting
phase, and the second on non-uniform elastic strain and the distribution of elastic
strain in the sampled volume as well as information on the size of the diffraction
particles.
In samples cooled without applied stress, at any given temperature above the
solubility limit, the hydrogen in solid solution causes the zirconium lattice parame-
ter to be higher than that of zirconium without hydrogen in solid solution. The pre-
cipitation temperature then corresponds to the temperature at which the lattice
parameter variation with temperature changes slope. This allowed us to verify that
the precipitation temperature matched well with the precipitation temperature val-
ues obtained from DSC.
A simplified view of the precipitation mechanism can be considered as follows.
Hydride particles form as highly compressed, small precipitates as seen in Fig. 17(a)
(the hydride nuclei are represented in dark red because of their high compressive
strain state). The small precipitate size is estimated from the size broadening
observed in the diffraction signal, whereas the compressive strain state is estimated
by the shift in hydride peak position. As the temperature decreases, the hydride
FIG. 17 Schematic of hydride strain behavior during cooling without applied stress: (a)
all hydrides are dissolved, (b) first hydride nuclei appear, (c) previous hydride
nuclei grown and relaxed (dislocations in the matrix), new hydride nuclei form
less compressed (sympathetic nucleation), (d) all hydrides are precipitated, and
(e) precipitated hydrides in compression because of thermal contraction.
volume fraction increases as a result of nucleation of new precipitates and growth

of the previous ones. The average strain increases (Fig. 17(b)) indicating that the new
hydride precipitates are less compressed or that some strain is relaxed by plastic de-
formation. This continues until the hydride strain reaches 3.5 103 at which
point half of the hydride population is precipitated (Fig. 17(c)). From this point down
to room temperature, hydride strains follow the thermal contraction of the zirconium
matrix (thermal expansion coefficient of 6.2 106 C1) instead of the hydride ther-
mal contraction (thermal expansion coefficient of 14.2 106 C1). Although the
hydrides would like to contract further away from the matrix, they are forced to fol-
low thermal contraction of the matrix likely because the hydride particles are embed-
ded in the matrix (Fig. 17(d) and 17(e)). From 450 C to 360 C, the hydride d{111}
diffraction peaks show predominantly size broadening and the hydride size grows
from 2.5 to 30 nm. From 360 C to room temperature, the dominant broadening
mechanism is strain broadening. This strain broadening is constant, showing that the
strain distribution in the hydride population is constant from 360 C to room
temperature.
A similar behavior is observed in samples cooled under an applied stress below
the critical value to reorient hydrides. The zirconium matrix deforms slightly
because of the thermal cycle under applied stress. Hydride precipitation occurs at
the expected precipitation temperature for unstressed hydrides. Hydride precipita-

tion is also divided into two precipitation regimes. The first precipitation regime
occurring at high temperature is nearly identical to that of unstressed hydrides. The
onset of the second precipitation regime also occurs when half of the hydride popu-
lation is precipitated. Hydride strains during this second precipitation regime
evolve differently than in unstressed hydrides. The strains in the case of stressed
hydrides remain constant as the hydrides finish precipitating and cool down to
room temperature. This is likely because of the fact that in the presence of the far
field strain, the matrix deforms more easily and transfers load to the hydrides, thus
compensating for thermal contraction. Hydrides are still in compression after the
load is removed in both TD and RD although slightly less in the TD where the ten-
sile load had been previously applied.
In samples cooled under an applied stress sufficient for reorientation, hydride
precipitation occurs at a temperature below the value obtained from DSC under no
load. During the first precipitation regime, the measured hydrides strains are com-
pressive, but as the temperature decreases these strains become tensile in the direc-
tion of the applied stress (TD). During the second precipitation regime, hydride
strains in the direction of the load remain constant and tensile. These strains
remain tensile (although less) even after the load is removed. Hydride strains are
compressive in the direction perpendicular to the load (RD) during cooldown.
These strains also become compressive once the load is removed.
Conclusions
The main conclusions of the study of hydrides in uniformly stressed samples in
cold-worked stress-relieved Zircaloy-4 are presented here:
1. The hydride dissolution and precipitation temperatures were determined in
situ by synchrotron XRD, and validated by comparison to previous DSC deter-
mination finding good agreement in that the hysteresis observed between dis-
solution and precipitation temperatures corresponds to values measured
previously by DSC in the absence of stress. This means that the dissolution
and precipitation temperatures can be directly determined from each sample
examination.
2. For a stress above the threshold stress for reorientation, it is found that the
precipitation temperature is lower than that of unstressed samples. For a stress
below the threshold stress for reorientation, the precipitation temperature cor-
responds to the unstressed value.
3. The change in hydride d-spacing was measured during hydride dissolution
and precipitation by transmission XRD. Considering that most of the d-spac-
ing change is because of strain, when unstressed hydrides precipitate, two
strain regimes are observed. Hydrides first nucleate as highly compressed par-
ticles, then quickly relax some of these compressive strains by either change of
shape, sympathetic nucleation, or formation of dislocations in the matrix.
When half of the hydride population is precipitated, the hydride strain
behavior changes to follow the matrix thermal expansion all the way down to
room temperature.
4. When hydrides are precipitated under stress but not reoriented, the strain
behavior is different than that of unstressed hydrides. The first precipitation
regime with relaxation of highly compressed particles is similar to that of
unstressed hydrides. However, the average hydride strain during the second pre-
cipitation regime is constant. This could be because of a greater ease in deforming
the matrix because of the applied far-field stress.
5. When hydrides precipitate under stress and reorient, during the first precipita-
tion regime, the hydride strains become tensile in the direction perpendicular
to the hydride platelet face. During the second precipitation regime, these
strains remain constant in tension. This indicates a different hydride strain
state for reoriented hydrides than for circumferential hydrides. Neither cycling
under stress nor increasing cooling rate appear to significantly affect the strain
state of reoriented hydrides.
ACKNOWLEDGMENTS
This research is funded by the Materials World Network grant DMR-0710616 from the
National Science Foundation, with corresponding funding from NSERC for the Cana-
dian partners. The writers are grateful for their support. The research for this publica-
tion was supported by the Pennsylvania State University Materials Research Institute
Nano Fabrication Network and the National Science Foundation Cooperative Agree-
ment No. 0335765, National Nanotechnology Infrastructure Network with Cornell Uni-
versity. Use of the Advanced Photon Source is supported by the U.S. Department of
Energy, Office of Basic Energy Sciences under Contract No. DE-AC02-06CH11357.
References
[1] Lemaignan, C. and Motta, A. T., “Zirconium Alloys in Nuclear Applications,” Materials Sci-
ence and Technology, A Comprehensive Treatment, R. W. Cahn, P. Haasen, and E. J.
Kramer, Eds., VCH, New York, 1994, pp. 1–51.
[2] Bossis, P., Pêcheur, D., Hanifi, L., Thomazet, J., and Blat, M., “Comparison of the High
Burn-up Corrosion on M5 and Low Tin Zircaloy-4,” Zirconium in the Nuclear Industry:
14th International Symposium, ASTM STP 1467, P. Rudling and B. F. Kammenzind, Eds.,
2005, pp. 494–525.
[3] Schmitz, F. and Papin, J., “High Burnup Effects on Fuel Behaviour Under Accident Condi-
tions: The Tests CABRI REP-Na,” J. Nucl. Mater., Vol. 270, 1999, pp. 55–64.
[4] Daum, R. S., 2007, “Hydride-Induced Embrittlement of Zircaloy-4 Cladding under Plane-
Strain Tension,” Ph.D. thesis, Materials Science, The Pennsylvania State University, Uni-
versity Park, PA.
[5] Raynaud, P. A. C., Koss, D. A., and Motta, A. T., “Crack Growth in the through-Thickness
Direction of Hydrided Thin-Wall Zircaloy Sheet,” J. Nucl. Mater., Vol. 420, 2012, pp. 69–82.
[6] Bradbrook, J. S., Lorimer, G. W., and Ridley, N., “The Precipitation of Zirconium Hydride
in Zirconium and Zircaloy-2,” J. Nucl. Mater., Vol. 42, No. 2, 1972, pp. 142–160.
[7] Beck, R. L., “Zirconium-Hydrogen Phase System,” Trans. ASM, Vol. 55, 1962, pp. 542–555.
[8] Chung, H. M., Daum, R. S., Hiller, J. M., and Billone, M. C., “Characteristics of Hydride Pre-
cipitation in Spent-Fuel Cladding,” Zirconium in the Nuclear Industry: 13th International
Symposium, ASTM STP 918, 2002, pp. 78–101.
[9] Kearns, J. J. and Woods, C. R., “Effect of Texture, Grain Size, and Cold Work on the Pre-
cipitation of Oriented Hydrides in Zircaloy Tubing and Plate,” J. Nucl. Mater., Vol. 20,
No. 3, 1966, pp. 241–261.
[10] Fredette, J. C., Perovic, V., and Holt. R. A., “Orientation of Hydrides in Zr-2.5Nb Tubes
Under Biaxial Stress,” Zirconium in the Nuclear Industry: 15th International Symposium,
ASTM STP 1505, B. F. Kammenzind and M. Limback, Eds., June 25–28, Sunriver, OR,
2007.
[11] Colas, K. B., Motta, A. T., Almer, J. D., Daymond, M. R., Kerr, M., Banchik, A. D., Vizcaino,
P., and Santisteban, J. R., “In-Situ Study of Hydride Precipitation Kinetics and Re-
Orientation in Zircaloy Using Synchrotron Radiation,” Acta Mater., Vol. 58, 2010, pp.
6565–6583.
[12] Zanellato, O., Preuss, M., Buffiere, J.-Y., Ribeiro, F., Steuwer, A., Desquines, J., Andrieux,
J., and Krebs, B., “Synchrotron Diffraction Study of Dissolution and Precipitation Kinetics
of Hydrides in Zircaloy-4,” J. Nucl. Mater., Vol. 420, 2012, pp. 537–547.
[13] Kerr, M., 2009, “Mechanical Characterization of Zirconium Hydrides With High Energy X-
Ray Diffraction,” Ph.D. thesis, Mechanical and Materials Engineering, Queen’s University,
Kingston, ON, Canada.
[14] Kerr, M., Daymond, M. R., Holt, R. A., and Almer, J. D., “Strain Evolution of Zirconium
Hydride Embedded in a Zircaloy-2 Matrix,” J. Nucl. Mater., Vol. 380, Nos. 2–3, 2008, pp.
70–75.
[15] Link, T. M., Koss, D. A., and Motta, A. T., “Failure of Zircaloy Cladding under Transverse
Plane-Strain Deformation,” Nucl. Eng. Design, Vol. 3, 1998, pp. 379–394.
[16] Perovic, V., Weatherly, G. C., and Simpson, C. J., “Hydride Precipitation in a/b Zirconium
Alloys,” Acta Metall., Vol. 31, No. 9, 1983, pp. 1381–1391.
[17] Daum, R. S., Majumdar, S., Liu, Y., and Billone, M. C., “Mechanical Testing of High-Burnup
Zircaloy-4 Fuel Cladding under Conditions Relevant to Drying Operations and Dry-Cask
Storage,” Water Reactor Fuel Performance Meeting, Oct 3–6, Kyoto, Japan, 2005, pp.
498–531.
[18] Colas, K. B., Motta, A. T., Daymond, M. R., Kerr, M., and Almer, J. D., “Hydride Platelet
Reorientation in Zircaloy Studied With Synchrotron Radiation Diffraction,” J. ASTM Int.,
Vol. 88, No. 1, 2011.
[19] Larson, A. C. and Dreele, R. B. V., General Structure Analysis System (GSAS), Los Alamos
National Laboratory, Los Alamos, NM, 2000.
[20] Une, K. and Ishimoto, S., “Dissolution and Precipitation Behavior of Hydrides in Zircaloy-
2 and High Fe Zircaloy,” J. Nucl. Mater., Vol. 322, No. 1, 2003, pp. 66–72.
[21] Kammenzind, B. F., Berquist, B. M., Bajaj, R., Kreyns, P. H., and Franklin, D. G., “The Long-
Range Migration of Hydrogen through Zircaloy in Response to Tensile and Compressive
Stress Gradients,” Zirconium in the Nuclear Industry: 12th International Symposium,
ASTM STP 1354, G. P. Sabol and G. D. Moans, Eds., June 15–18, 2000, pp. 196–233.
[22] Eadie, R. L. and Coleman, C. E., “Effect of Stress on Hydride Precipitation in Zirconium-
2.5% Niobium and on Delayed Hydride Cracking,” Scripta Metall., Vol. 23, No. 11, 1989,
pp. 1865–1870.
[23] Douglass, D. L., “The Metallurgy of Zirconium,” Atomic Energy Review, Z. I. Turkov, Ed.,
International Atomic Energy Agency, Vienna, Austria, 1971.
[24] The Powder Diffraction File, International Center for Diffraction Data, Newtown Square,
PA, 2006.
[25] Zuzek, E., Abriata, J. P., San-Martin, A., and Manchester, F. D., “The H-Zr (Hydrogen-Zir-
conium) System,” Bull. of Alloy Phase Diag., Vol. 11, No. 4, 1990, pp. 385–395.
[26] Yamanaka, S., Yamada, K., Kurosaki, K., Uno, M., Takeda, K., Anada, H., Matsuda, T., and
Kobayashi, S., “Characteristics of Zirconium Hydride and Deuteride,” J. Alloys Compd.,
Vols. 330–332, 2002, pp. 99–104.
[27] Puls, M. P., Shi, S.-Q., and Rabier, J., “Experimental Studies of Mechanical Properties of
Solid Zirconium Hydrides,” J. Nucl. Mater., Vol. 336, No. 1, 2005, pp. 73–80.
[28] Yamanaka, S., Yoshioka, K., Uno, M., Katsura, M., Anada, H., Matsuda, T., and Kobayashi,
S., “Thermal and Mechanical Properties of Zirconium Hydride,” J. Alloys Compd., Vols.
293–295, 1999, pp. 23–29.
[29] Snyder, R. L., Fiala, J., and Bunge, H., “Defect and Microstructure Analysis by Diffraction,”
International Union of Crystallography Monographs on Crystallography, Oxford Univer-
sity Press, New York, 1999.
[30] Cook, C. S., Sabol, G. P., Sekera, K. R., and Randall, S. N., “Texture Control in Zircaloy Tub-
ing Through Processing,” Zirconium in the Nuclear Industry: 9th International Sympo-
sium, ASTM STP 1132, Nov. 5–8, Kobe, Japan, ASTM International, West Conshohocken,
PA, 1991, pp 80–95.
DISCUSSION
Question from Ron Adamson, Zircology Plus
Q1:—Do I interpret one of your curves to indicate that lattice strain caused by
H in solution is different than lattice strain caused by hydride? This is different
than commonly assumed and indicated by the modeling work of Wolf et al. in this
meeting.
Authors’ Response:—Hydrides generate both elastic and plastic strain while

hydrogen in solid solution generates only elastic strain. The strain measured here is
only elastic strain. The hydrides will have a small volume fraction around them
where the elastic strain field is present, but this is a relatively small overall volume
fraction of the whole material. The difference observed in the effect of hydrogen in
hydrides and in solid solution on zirconium matrix elastic strains is due to the fact
that the X-ray beam probes a large number of zirconium grains. Indeed, the in-situ
synchrotron diffraction technique used in this work has a beam size of 200 lm x
200 lm and the beam goes through the entire sample thickness which is around
600 lm. Therefore for our CWSR material, hundreds of zirconium grains are
sampled. When hydrogen is precipitated in hydrides, the strain induced by the
hydrides on the zirconium lattice is localized around the hydrides themselves,
therefore the diffraction signal from the zirconium comes from many grains with-
out hydrides and a few with hydrides, thus the strains due to the hydrides is not
measured when all hydrogen is precipitated. When all hydrogen comes into solid
solution it induces a uniform strain in all the zirconium grains measured thus it is
visible in the diffraction signal.
We note that when the temperature increases the lattice parameter change is
occurring as the hydrides are dissolving, so the increase in lattice parameter is prop-
erly attributed to hydrogen in solid solution.
Q2:—Are you convinced that you were observing d (delta) hydride?
Authors’ Response:—When observing the diffraction patterns at room tempera-

ture (an example of which can be found in [1]), the position of the hydride peaks
are consistent with those expected from d-hydrides (Powder Diffraction File num-
ber 00-034-0649). At very high temperature when the hydrides first form, it is pos-
sible that there could be some change in stoichiometry that could induce a peak
shift (see ‘Thermo-mechanical cycle without applied stress’ section in the paper).
Q1:—Does the hydride orientation have a memory effect?

Authors’ Response:—Given the fact that when several thermo-mechanical cycles

are applied, the radial hydride fraction increases, it does appear that there is a
hydride orientation memory effect. This effect is essentially the same as the hydride
memory effect observed for circumferential hydrides and is normally ascribed to
dislocations “nests” created during the nucleation of the original hydrides and in
which hydrides can re-form and grow. If the maximum temperature of the thermo-
mechanical cycle is not high enough to recover the material, the hydrides will likely
tend to preferentially re-precipitate in these nests.
Q2:—Why do you need multiple cycles to reorient all of the hydrides?
Authors’ Response:—The need to cycle a sample several times to reach a fully

reoriented microstructure could be due to the fact that our cooling rate of 1 C/min
is too fast to allow all hydrides to nucleate in the preferred out of plane orientation.
Several cycles will allow a greater percentage of hydrides to form in the preferred
direction. This cooling rate was a compromise chosen to be able to perform experi-
ments in-situ at the APS synchrotron that would finish in a reasonable time.
Q3:—Why did you select (111) d hydride peak for strain measurement?
Authors’ Response:—The (111) d hydride peak is the most intense hydride peak
that can be measured with our experimental geometry and can thus give the great-
est accuracy in its determination.
Question from Johannes Bertsch, Paul Scherer Institute:—Have you kept the
load until the complete cool down, so that the reorientated hydrides remained
under the influence of external stress? If yes, what would the stress of the hydrides
be after unloading?
Authors’ Response:—As can be seen in Figure 3(b) and 3(c) which represent
typical thermo-mechanical cycles performed at the APS, the applied tensile load
is kept constant until 150 C at which temperature all hydrides are precipitated.
Diffraction data are still acquired after the load is removed and can be seen in
Figure 13 after the ‘Load off’ step for the stressed but not reoriented sample and
in Figure 14 after the strain step at 150 C for the reoriented sample. As can be
seen from these figures, when the applied stress is removed, the hydride strains
in the TD and the RD for the non-reoriented sample remain both compressive.
However, the hydride strains in the TD for the reoriented samples remains posi-
tive even though the applied tensile load is removed, implying the reoriented
hydride faces are in tension even without applied stress. The reoriented hydride
edges measured in the RD remain in compression even after the load is
removed.
Questions from Michael Preuss, University of Manchester
Q1:—You have inferred strain/stress from the change of d-spacing in the

hydride. Have you also considered that the change of d-spacing could be due to a
change of stoichiometry? This seems far more likely when looking at your data.
Authors’ Response:—It is indeed possible that the high compressive strain meas-
ured in newly precipitated hydrides as a diffraction peak shift could be due to these
initially forming hydrides having a slightly different stoichiometry than the Zr/H
ratio of 1.66 typical in the d-hydride phase. Within the d-hydride phase, a maxi-
mum change in d-spacing due to stoichiometry of 0.002 nm could lead to a pseudo-
strain of 4000 microstrain [2,3]. This cannot account for the entirety of the 10 milli-
strains measured and thus should not be responsible for the entire peak shift,
although it could certainly be a contributing factor.
Q2:—It is very unusual that at high T during cooling, the hydrides are highly
stressed while the stresses are relieved at lower temperature. Can you please com-
ment on this?
Authors’ Response:—Hydrides first nucleate as platelets and large surface ener-

gies relative to the hydride volume could induce a larger d-spacing variation than
after the hydrides grow. When hydrides first nucleate, the difference in crystal
structure between the hydride phase and the zirconium matrix can induce signifi-
cant stresses in the nucleating hydrides. The stress relaxation observed as hydrides
nucleate and grow upon cool-down could be explained by the formation of plastic
dislocations in the matrix surrounding the hydrides therefore relaxing some the ini-
tial stresses. In addition, formation of new hydrides close to pre-existing hydrides
(sympathetic nucleation) take advantage of the strain fields around the pre-existing
hydrides and these newly precipitated hydrides could be forming with a less com-
pressed strain state thus lowering the overall hydride population compressive
strain.
Questions from Malcolm Griffiths, AECL
Q1:—For the hysteresis, how do you know that you are not seeing a kinetic
effect? Is there a difference in the dissolution and precipitation rates for hydriding?
Authors’ Response:—The hysteresis measured between the dissolution and pre-

cipitation temperature agrees well with previous literature determinations using dif-
ferential scanning calorimetry with varying heating/cooling rates. In addition, when
measuring dissolution and precipitation for various cooling rates no significant
effect was measured [4]. This leads us to believe that the kinetic effect, if it present,
is small.
Q2:—At the microscopic scale, what is the average spacing between all hydrides
— assuming that there are finer hydrides between the coarser hydrides?
Authors’ Response:—As you mention, the hydrides seen after etching are really
collections of microscopic hydrides. The very small distance between these can best
be investigated by TEM. Our limited TEM studies showed that the microscopic
hydride platelets are stacked on top of each other and don’t have measurable gaps
between them [5]. For the macroscopic hydrides the average spacing depends on
the hydrogen concentration
Question from Rishi Sharma, IIT Bombay, India:—You mentioned that the Zr
material shows different strain behavior with the presence of reoriented hydrides
than the Zr material with hydrides in as received material. Can you explain the
above phenomenon?
Authors’ Response:—The observed difference in strain is in the hydrides. The

overall zirconium matrix strain behavior during dissolution and precipitation of
hydrides is not significantly impacted by applied stress except for the stretching and
compressing of the zirconium planes in the TD and the RD respectively when the
tensile stress in the TD is applied. The hydride strain behavior is however very dif-
ferent between unstressed/not reoriented and stressed/reoriented samples. This sig-
nifies that reoriented hydrides have a different strain state than circumferential
hydrides.
Question from Ted Darby, Rolls-Royce:—Do you propose an explanation for the
downward shift in TSSP for reoriented hydrides? Could it simply be due to the
need for these hydrides to nucleate in the matrix without the assistance of pre-
existing hydrides?
Authors’ Response:—The downward shift in the precipitation temperature for

reoriented hydrides signifies a greater degree of undercooling is needed to precipi-
tate radial hydrides. This suggests that stress acts to suppress circumferential
hydride precipitation. In both unstressed and fully reoriented hydrides, full dissolu-
tion was achieved so no pre-existing hydrides are present when the precipitation
temperature is measured. Therefore, this downward shift in Tp could be due to the
difficulty of hydrides to nucleate in favorably oriented sites in the radial orientation
or to the small number of these sites available in the CWSR material used in this
study.
Question from R. N. Jayaraj, Dept. of Atomic Energy, India:—Hydrides get re-

oriented to "normal" direction under "tensile" stress. To what direction they get re-
oriented under compressive stress?
Authors’ Response:—No experiments presented in this work have been per-

formed under compression. However, literature results show that under sufficient
compressive stress, hydrides can reorient parallel to the applied compressive
stress [6].
Question from B. K. Shah, BARC Mumbai:—Different Zr alloys have different

threshold stress for hydride reorientation. Please comment on the factors which are
responsible for this difference.
Authors’ Response:—The factors influencing the threshold stress for hydride

reorientation were not investigated precisely in the work presented here. However,
since the threshold stress for recrystallized material typically tends to be lower than
that of cold-worked stress relieved material, influencing factors could include grain
microstructure [7]. A larger fraction of grain boundaries oriented in the out of
plane direction could increase hydride reorientation. Grain texture and in particular
orientation of the basal poles have also been found to have a strong influence of
hydride reorientation [8].
Questions from K. Somasekhar Reddy, Nuclear Fuel Complex, Hyderabad, India
Q1:—What is the necessity of more thermal cycles when all of the hydrides are
dissolved in the matrix during the first thermal cycle?
Authors’ Response:—As the material cools the hydrogen in solid solution is

called upon to precipitate and the applied load introduces a bias for precipitation in
the out of plane direction, so that depending on the cooling rate a fraction of the
hydrides will reorient. Presumably a very slow cool would require only one cycle
for complete reorientation.
Q2:—What is the effect of cooling rate on the reorientation of hydrides?
Authors’ Response:—See the above response. Additional results on the effect of

cooling rate on hydride reorientation can be found in [5].
Additional References
1. Colas, K.B., et al. Hydride Platelet Reorientation in Zircaloy Studied with Syn-
chrotron Radiation Diffraction. in 16th International Symposium on Zirco-
nium in the Nuclear Industry 2010. Chengdu, China: ASTM STP 1529, pp.
496–522.
2. Zuzek, E., et al., H-Zr (hydrogen-zirconium): phase diagrams of binary hydro-

gen alloys, A. International, Editor 2000: Ohio. p. 309–322.
3. Yamanaka, S., et al., Characteristics of zirconium hydride and deuteride. Jour-

nal of Alloys and Compounds, 2002. 330–332: p. 99–104.
4. Colas, K.B., et al., In-situ study of hydride precipitation kinetics and re-orien-
tation in Zircaloy using synchrotron radiation. Acta Materialia, 2010. 58: p.
6565–6583.
5. Colas, K.B., Fundamental Experiments on Hydride Reorientation in Zircaloy,

in Department of Nuclear Engineering, 2012, PhD Dissertation in Nuclear
Engineering, The Pennsylvania State University: University Park.
6. Louthan, M.R. and R.P. Marshall, Control of Hydride Orientation in Zircaloy.

Journal of Nuclear Materials, 1963. 9(2): p. 170–184.
7. Bai, J., et al., Hydride Embrittlement in Zircaloy-4 Plate, Part I Influence of

Microstructure on the Hydride Embrittlement on Zircaloy-4 at 20C and 150C
and Part II Interaction Between the Tensile Stress and the Hydride Morphol-
ogy. Metallurgical and Materials Transactions A, 1994. 25.
8. Hardie, D. and M.W. Shanahan, Stress reorientation of hydrides in zirco-

nium-2.5% niobium. Journal of Nuclear Materials, 1975. 55(1): p. 1–13.
Dan Lutz,1 Yang-Pi Lin,2 Randy Dunavant,2

Rob Schneider,2 Hartney Yeager,2 Aylin Kucuk,3
Bo Cheng,3 and Jim Lemons4
Hydriding Induced Corrosion

Failures in BWR Fuel
Reference
Lutz, Dan, Lin, Yang-Pi, Dunavant, Randy, Schneider, Rob, Yeager, Hartney, Kucuk, Aylin,
Cheng, Bo, and Lemons, Jim, “Hydriding Induced Corrosion Failures in BWR Fuel,” Zirconium
ABSTRACT
Fuel rods in 63 bundles failed starting in late 2001 during their second cycle of
operation in a U.S. boiling water reactor (BWR). Poolside and hot cell
examinations were performed on failed and non-failed bundles to understand
the failure mechanism and to gain insight into the failure root cause. Results
showed that the fuel cladding failed due to accelerated corrosion that resulted in
the formation of localized hydrides on the outer cladding surface prior to failure.
Primary cladding perforation then occurred due to cracking of the brittle hydride
lenses, rather than through-wall corrosion. The specific characteristics of these
nodular corrosion-related hydride failures present a new or previously
unrecognized variation of a BWR cladding corrosion failure mechanism.
Characteristics of the damaged rods suggested that the hydride localizations
formed under the action of local thermal gradients due to local variations in
oxide thickness. Finite element modeling of hydrogen diffusion under simplified
conditions indicated that it is a plausible explanation for their formation.
Manuscript received December 10, 2012; accepted for publication August 31, 2013; published online June 17,
2014.
1
Global Nuclear Fuel–Americas, Sunol, CA 94586, United States of America.
2
Global Nuclear Fuel–Americas, Wilmington, NC 28401, United States of America.
3
Electric Power Research Institute, Palo Alto, CA 94304, United States of America.
4
Tennessee Valley Authority, Chattanooga, TN 37402, United States of America.
5
Hyderabad, India.
Copyright V
LUTZ ET AL., DOI 10.1520/STP154320120198 1139
Investigations related to possible contributions of the BWR coolant chemistry on

the failures led to the discovery of Li in the damaged cladding oxide at a
concentration that has been associated with high corrosion rates. Based on SIMS
analyses of selected rods from the affected BWR and from other unaffected
BWRs, the possible role of Li in initiating elevated corrosion and in advancing
corrosion to a point of failure is discussed.
Keywords
BWR, Zircaloy-2, corrosion, hydriding, failure mechanism, secondary ion mass
spectrometry, hydrogen diffusion, finite element modeling
Introduction
Sixty-three boiling water reactor (BWR) nuclear fuel assemblies failed in Browns
Ferry 2 (BF2) during Cycle 12 by a corrosion-related hydriding failure mechanism.
Failures began in late 2001, about seven months after the cycle began, and then con-
tinued through the end of the cycle. The failures occurred only in the second cycle
Reload 10 fuel, which accounted for about 40 % of the core. The first failures
occurred at exposures of about 30–35 GWd/MTU; the failures exhibited a sensitiv-
ity to power such that the highest exposure bundles in the reload were the first to
fail, but power levels were not unusually high.
The damage was widespread, and affected fuel rods that used cladding
fabricated from multiple ingots and cladding lots. The affected bundles were dis-
tributed throughout the core. The affected fuel rods were fabricated from a type of
corrosion-resistant, heat-treated Zircaloy-2 barrier cladding that had been used
widely across the BWR fleet. The experience base for the type of affected cladding is
comprised of 3 106 fuel rods that operated in the BWR fleet over a ten year period
without a comparable incident. Highly common cladding fabrication lots that are
representative of the affected BF2 Reload 10 fuel operated successfully in 12 other
reactors. These cladding lots shared common ingots and were fabricated in the
same general time period. These observations in part indicate that the cladding ma-
terial was not the sole or primary failure root cause; however, the progression and
extent of corrosion damage was sensitive to cladding alloying content, primarily
iron and tin.
The failures and their cause were investigated from 2002 to the present through
a series of poolside and hotcell examinations; some descriptive characteristics have
been documented elsewhere [1,2] and are summarized here. Elevated, coalesced
nodular corrosion and oxide spallation were observed along the length of affected
rods, but the most excessive corrosion, spallation, and hydriding (Figs. 1–3), and the
primary failure perforations, occurred near 2413 mm (95 in.) and above. On the
other hand, corrosion was relatively low beneath fuel assembly spacers as shown in
Fig. 2. The surfaces of fuel rods also exhibited large azimuthal variation in corrosion
FIG. 1 Axial cladding eddy current lift-off profile for Browns Ferry 2 Reload 10 sound
YJS616 B8 with advanced corrosion damage.
FIG. 2 Visual appearance of Browns Ferry 2 Reload 10 sound rod YJS616 B8 with
advanced corrosion and spalling occurring at upper rod elevations but stopping
sharply beneath the fuel assembly spacer as shown.
LUTZ ET AL., DOI 10.1520/STP154320120198 1141
(visually and via measurement); surfaces that faced the bundle periphery or central
water rods had markedly lower corrosion and spalling. The observations of extreme
azimuthal variation and dramatically reduced corrosion beneath fuel assembly
spacers at various elevations could not be accounted for on the basis of variations in
the cladding composition or microstructure, suggesting other factors also contrib-
uted to the corrosion behavior. These other factors are not well understood, but are
thought to be related to a synergistic effect between the local coolant chemistry and
fuel rod power. Numerous localized hydride lenses were observed by neutron radi-
ography (Fig. 3) and metallography on the outer cladding surface starting near the
2159 mm elevation and above as shown in Figs. 4 and 5 for sound but heavily dam-
aged BF2 Reload 10 rod YJS616 B8. Moving up the rod, the localized hydrides
emerged just below the start of the upper elevation corrosion and spalling peak.
Cracks initiated on the cladding outer surface in the brittle hydride lenses when the
tensile stress became sufficient. Cracking of hydrides was the primary cladding
breach mechanism, rather than through-wall corrosion; fractography of the fracture
surfaces was not performed. A hydride band also formed at the top and bottom
transitions to the fifth spacer elevation where there is a large step change in oxide
thickness.
The failure investigation determined that material or power factors were not
the sole or primary causes, which led to the conclusion that a coolant water chemis-
try condition must have affected the Reload 10 BF2 fuel and contributed to the fail-
ure conditions. However, no severe upset conditions were detected by normal
FIG. 3 Axial cladding hydrogen content profile for Browns Ferry 2 Reload 10 sound rod
YJS616 B8 with advanced corrosion damage. Hydride damage indications from
neutron radiography correspond to locations with heavy spalling.
FIG. 4 Polished and etched fuel cladding cross-sections showing oxide thickness
(upper row) and zirconium hydrides (lower row) at two different elevations for
Browns Ferry 2 Reload 10 sound rod YJS616 B8 with advanced corrosion,
spalling, and hydriding damage.
coolant monitoring practices during the time Reload 10 was operating. Some small
sulfate spikes that occurred during Cycle 11 when the BF2 Reload 10 fuel was in its
first cycle were among the few, but mild, coolant water chemistry indications that
were identified in multiple reviews of the chemistry data by various parties. How-
ever, the coolant’s role in the failures was not readily identifiable in the absence of
prominent indications.
The present paper focuses on two different aspects of the corrosion-related fuel
failures. In the first aspect, potential coolant effects that went undetected by normal
means were investigated by analyzing cladding samples with secondary ion mass
spectrometry (SIMS). In the second aspect, the observations of hydride localizations
at discontinuities in oxide thickness suggested that corrosion-generated hydrogen
LUTZ ET AL., DOI 10.1520/STP154320120198 1143
FIG. 5 Etched fuel cladding cross-section showing hydrogen localization at the

2464 mm (97 in.) elevation on outer surface of Browns Ferry 2 Reload 10 sound
rod YJS616 B8 with advanced corrosion and hydriding damage; the arrows
highlight a variable thickness converging oxide layer that forms a relatively thin
oxide plateau where the hydrogen accumulates beneath.
localized under the action of local thermal gradients that were caused by differences
in oxide thickness. To help support this view, hydrogen diffusion and accumulation
were evaluated using a finite element model constructed with simplified conditions
meant to broadly simulate those encountered in the affected rods. The results of
both of these efforts are provided in this paper to advance industry knowledge
about this new BWR corrosion failure mechanism.
Secondary Ion Mass Spectrometry

SAMPLES FOR SIMS ANALYSES
Ten fuel cladding samples were analyzed using SIMS to investigate potential contri-
butions of the coolant water chemistry on the failures. The samples were extracted
from seven fuel rods from seven fuel assemblies that operated in four different
BWR power plants. Details of the operational time and exposure are provided in
Table 1. One of the examined rods was in the failed condition; the other six rods
were in the sound, unfailed condition. The various samples were selected to provide
a range of conditions to help judge the significance of differences. The YJP354 E8
samples (H2-1 and H2-2) from reactor H2 Reload 14 are particularly important
because their cladding originated from the same ingot and cladding lot as the BF2-4
and BF2-5 samples from the damaged BF2 Reload 10 YJS616 B8 rod. The reactor
H2 Reload 14 rod, YJP354 E8, operated with relatively high power conditions, to
end of life, without experiencing the damage that occurred in BF2 Reload 10.
The rods were sectioned perpendicular to the rod axis and metallographic sec-
tions were prepared and examined in the hotcell prior to SIMS analyses. The cir-
cumferentially averaged oxide thickness for each analyzed sample is given in Table 1.
Note that for sample BF2-5, there was substantial oxide spalling, so the remaining
oxide thickness measured in the hotcell is lower than that prior to spalling.
ANALYSES USING SIMS

Two sets of SIMS measurements were made using an ATOMIKA 4000 SIMS instru-
ment equipped with a quadrupole mass spectrometer. Both sets were conducted
using 69Gaþ as the primary ion with 25 keV beam energy. In the first measurement
set, the beam intensity used was 700 pA and line scans from samples BF2-1 through
BF2-3, as listed sequentially in Table 1, were obtained from pre-defined regions. The
isotopes and species analyzed were 6Li, 7Li, 10B, 11B, 12C, 19F, 31P, 24 Mg, 27Al, 28Si,
52
Cr, 56Fe, 58Ni, 64Zn, and 106ZrO. In the second set of measurements, performed
on samples BF2-4 through H2-2, as listed sequentially in Table 1, the beam intensity
used was 2000 pA and 512 by 512 pixel maps were acquired for each species of in-
terest. Line scans were then electronically extracted from the maps. The isotopes
and species analyzed in the second set of analyses were: 6Li, 7Li, 10B, 11B, 16O, 52Cr,
55
Mn, 56Fe, 58Ni, 63Cu, 64Zn, 69Ga, 90Zr and 106ZrO.
Quantification of data from the last six samples listed in Table 1 was performed
only for Li using a zirconium alloy cladding that had been oxidized in an autoclave
and implanted with 7Li to a concentration of 4.4 ppm. The relative sensitivity factor
(RSF) method was used, in which RSF is defined as
ðI std Þ
(1) RSF ¼ Zr
ðC std Þ
ðI std
Li
Þ Li
where:
ILi ¼ the peak lithium intensity,
IZr ¼ the zirconium intensity at the same depth,
CLi ¼ the corresponding implanted lithium concentration, and
superscript std ¼ the measurements made on the standard specimen.
Based on the implanted standard, an RSF value of 75.8 ppm was determined
and applied to the quantification of lithium in the oxide layer. The Li concentration
in the oxide samples is then given by
sample
sample ðI Li Þ
(2) ðC Li Þ ¼ RSF sample
ðI Zr Þ
where the superscript sample refers to measurements made on the sample

specimen.
TABLE 1 List of fuel cladding samples investigated using SIMS and associated rod and operational information.
Rod Average
Average Residence Sample Oxide
Exposure Insertion Time Elevation Thickness
Sample Assembly Rod Condition (GWd/MTU) Reactor Date (Days) Reload (mm/inch) (lm)
BF2-1 YJS734 H2 Failed 47.3 BF2 May 1999 1211 10 2286 / 90 n/a*
BF2-2 YJK363 B3 Sound 35.1 BF2 October 1997 1751 9 3023 / 119 21.1
L-1 YJ1380 D1 Sound 68.9 L July 1992 2708 4 2413 / 95 21.5
BF2-3 YJS614 G9 Sound 34.5 BF2 May 1999 1040 10 3023 / 119 21.9
BF2-4 724 / 28.5 28
YJS616 B8 Sound 41.1 BF2 May 1999 1040 10
LUTZ ET AL., DOI 10.1520/STP154320120198

BF2-5 2311 / 91 23*
BF3-1 762 / 30 18
YJN587 E9 Sound 17.0 BF3 October 1998 548 8
BF3-2 2362 / 93 17
H2-1 699 / 27.5 18
YJP354 E8 Sound 44.0 H2 November 1998 1498 14
H2-2 2286 / 90 10
*
Note: Severely spalled.
1145
RESULTS
The line scan data for samples BF2-1, BF2-2, L-1, and BF2-3 generally traversed the
oxide layer and revealed three distinct regions: metal side, oxide, and waterside
region. In the metal (i.e. cladding) region, the reference 106ZrO signal was generally
low and most species were near or below the detection limit. In the waterside
region, Cr, Fe, Ni, and Zn that comprise the crud layer typically showed higher
intensities than elsewhere. In the oxide, most species showed low, variable inten-
sities, with some occasional peaks that could be identified with cracks in the oxide.
The Cu intensities were insignificant in all cases and regions. The notable feature of
most interest was 7Li in varying amounts in the zirconium oxide of analyzed sam-
ples. There was no particular association of Li with the waterside crud layer. The
isotopes 6Li, 10B, and 11B were generally below the detection limit in the first set of
analysis, except for 6Li in sample BF2-1. Accordingly, the results discussed in this
paper will focus on Li.
In sample BF2-1, obtained near the upper elevation failure location of the failed
BF2 Reload 10 rod, high Li7 within the oxide was observed. In a scan across the ox-
ide thickness, Fig. 6, the highest Li concentration was observed in a plateau region
within the oxide and the Li concentration decreased just before the interface with
the metal is reached. Although not quantified through a standard, the Li concentra-
tion in the oxide was estimated to be near 550 ppm, based on previous experience
with examination of Zr-alloys exposed to lithiated water. The 6Li profile generally
followed the 7Li profile. In the plateau region where 7Li concentration was high, 6Li
was also detected, and the average 7Li to 6Li ratio was 450.
FIG. 6 SIMS profiles of Li (mass 6 and 7) and ZrO for sample BF2-1. Metal/oxide
interface is on the right hand side.
LUTZ ET AL., DOI 10.1520/STP154320120198 1147
FIG. 7 SIMS profiles of Li and ZrO for sample BF2-2. Metal/oxide interface is on the
right hand side.
Sample BF2-2 was from a fuel rod in an assembly that was loaded into the
same reactor (BF2) as sample BF2-1 but one cycle earlier (there were no corrosion
failures associated with this sample or its companion fuel assemblies in the same
reload). Sample L-1 was from a fuel assembly that operated for three cycles in one
reactor followed by a fourth cycle in a sister reactor—reactors that were unrelated
to the corrosion failures. Representative SIMS line scan profiles for samples BF2-2
and L-1 are shown Figs. 7 and 8, respectively. Both samples showed measurable but
FIG. 8 SIMS profiles of Li and ZrO for sample L-1. Metal/oxide interface is on the right
hand side.
low amounts of 7Li. Sample BF2-3 was from a sound rod in assembly YJS614 that
operated in the same reactor (BF2) as failed sample BF2-1’s rod in assembly
YJS734; assembly YJS614 was discharged earlier than assembly YJS734 due to a fail-
ure in assembly YJS614’s H2 lattice position. Figure 9 shows a representative line
scan for sample BF2-3. Variable amounts of 7Li were observed. The amount of Li in
sample BF2-3 is lower than in sample BF2-1 from a failed rod, but local 7Li reached
concentrations higher than in sample BF2-2 or L-1.
In the second set of analyses (last six samples listed in Table 1), samples taken at
two elevations (2300 and 700 mm) from three sound rods (from plants BF2,
BF3, and H2) were examined. The Reload 10 rod for samples BF2-4 and BF2-5 was
a heavily corroded but sound sibling (in symmetric position within the fuel assem-
bly) to a failed Reload 10 rod that started operation in BF2 at the same time as the
fuel assembly for Reload 10 sample BF2-3. One reason for choosing the rod was to
confirm the Li-related observations made on BF2 Reload 10 sample BF2-1. The rod
for samples BF3-1 and BF3-2 was from a Reload 8 fuel assembly that operated for
only one cycle in plant BF3, a sister plant to BF2 that had also experienced a limited
number of corrosion failures that had similar characteristics. The purpose for exam-
ining BF3 Reload 8 samples BF3-1 and BF3-2 was to seek similarities with BF2
Reload 10 samples BF2-4 and BF2-5 that might shed light on a potential common
cause for the corrosion. The plants share a common liquid radwaste processing sys-
tem, which provides one potential source for a common aggravating water chemis-
try factor. As noted above, for unaffected plant H2 Reload 14 samples H2-1 and
H2-2, the fuel rod was from the same manufacture lot as the rods for BF2 Reload
FIG. 9 SIMS profiles of Li and ZrO for sample BF2-3. Metal/oxide interface is on the
right hand side.
LUTZ ET AL., DOI 10.1520/STP154320120198 1149
10 samples BF2-1 and BF2-4/BF2-5 but operated normally in plant H2 without

developing accelerated corrosion or hydriding.
Representative scans for 7Li, 6Li, 11B, 10B, and 90Zr isotopes across the oxide in
samples BF2-4 through H2-2, as listed sequentially in Table 1, are shown in Fig. 10.
The figure also shows the maximum Li concentration for each sample following
quantification using the RSF method. Figure 10 shows that 7Li was detected in all
samples, and was the highest in sample BF2-5 from the BF2 Reload 10 rod at
FIG. 10 SIMS profiles of Li (mass 6 and 7), B (mass 10 and 11), and Zr for samples BF2-4
to H2-2. Metal/oxide interface is on the left hand side. The max Li concentration
for each case from RSF quantification is given below each set of profiles.
2311 mm elevation, reaching about 300 ppm. Sample BF3-2 for the BF3 Reload 8
rod at 2362 mm elevation, and samples BF2-4 and H2-1 (from 700 mm elevation
of the BF2 Reload 10 and H2 Reload 14 rods, respectively) had 2–3 ppm Li in the
oxide. Samples BF3-1 and H2-2 (for the BF3 Reload 8 rod at 762 mm and H2
Reload 14 rod at 2286 mm elevation, respectively) had low Li in oxide, both esti-
mated to be below 1 ppm.
Samples BF2-5 from BF2 Reload 10 and BF3-2 from BF3 Reload 8 were the
only samples in the second set of measurements where 6Li was above the detection
limit. It is therefore possible to derive the 7Li/6Li ratio for these two samples. The
ratio is of interest in that it could potentially provide insight into the timing of Li
introduction relative to the end of irradiation, but within the confines of assump-
tions about the likely source of Li. For sample BF2-5, the 7Li/6Li ratio was about
420, based on measurement positions where counting statistics were adequate, simi-
lar to 450 found for BF2 Reload 10 sample BF2-1. For sample BF3-2 (BF3 Reload
8 rod at 2362 mm), 6Li was just above detection level at most measurement posi-
tions. The estimated 7Li/6Li ratio for BF3 Reload 8 sample BF3-2 was about 20, sig-
nificantly lower than the ratio for BF2 Reload 10 sample BF2-5, but higher than the
natural ratio of 12. A similar 7Li content was found for sample BF2-4 (BF2 Reload
10 rod at 724 mm), sample H2-1 (H2 Reload 14 rod at 699 mm) and sample BF3-2
(BF3 Reload 8 rod at 2362 mm), but only BF3 Reload 8 sample BF3-2 had detecta-
ble level of 6Li, suggesting higher 6Li inventory.
In all specimens, 11B was detected in small amounts and 10B was low near the
detection limit for most specimens, but was below the detection limit in sample
BF2-5, the BF2 Reload 10 rod at 2311 mm elevation, where the highest 7Li was
observed. A spatial correlation of intensities in 7Li and 11B maps was evident for
samples BF2-5 and BF3-2, as shown in Fig. 11. The mapping capability provided vis-
ual assistance in recognizing a correlation.
Hydrogen Localization Modeling

The excessive corrosion exhibited by the affected rods resulted in high hydrogen
contents. The interesting observation is the localized nature of the hydride forma-
tions. The failures were caused by cracking of these localized, brittle hydrides that
originated from the unusually high corrosion. The cladding hydrogen pickup is
complex in relation to initiation and progression of the damaging corrosion; how-
ever, it is clear that at some time during operation of the Reload 10 fuel, the amount
of hydrogen became sufficient to damage and fail the fuel. It is postulated that
hydrogen was absorbed uniformly, in variable amounts along a rod, and then accu-
mulated in localized cold spots under the action of thermal gradients that were
caused by several different types of zirconium oxide discontinuities including severe
spalling (Fig. 2, Fig. 4), converging oxide fronts that created thin oxide plateaus (Fig.
5), and step changes in corrosion beneath the spacers (Fig. 2). In order to demon-
strate the plausibility of this mechanism, finite element modeling was first
LUTZ ET AL., DOI 10.1520/STP154320120198 1151
FIG. 11 SIMS maps of ZrO, Li, and B for samples BF2-5 and BF3-2. In each map, the
metal is on the left hand side. Field of view is 133 by 133 and 66 by 66 microns
for samples BF2-5 and BF3-2, respectively.
performed to estimate the thermal gradients that resulted from the fuel cladding
surface oxide discontinuities, and then to model the diffusion of hydrogen in the
presence of those thermal gradients as discussed below. The operating histories and
performance/damage varied considerably from rod to rod, and the damage pro-
gressed over time, so the parameters used for both the thermal and diffusion models
are meant to be exemplary of general trends.
THERMAL MODELING AND RESULTS

Steady-state thermal analysis was performed for an axisymmetric 2D (r–z) model
based on Fourier’s equation in cylindrical coordinates. The intent of the thermal
model is to estimate the thermal gradients that may be present during reactor oper-
ation for cladding affected by discontinuous oxide, where a thin oxide layer is
located adjacent to a thicker oxide layer for the purpose of simulating oxide spal-
ling, converging oxide fronts with thin oxide plateaus, or step changes in corrosion
beneath the spacers.
FIG. 12 Model geometry for thermal analysis.
The fuel cladding thermal analysis model is shown in Fig. 12. The model
assumes a uniform thicker oxide layer (denoted the reference oxide layer) over an
axial length of cladding with an adjacent thinner oxide layer (denoted as an oxide
layer discontinuity). Assumptions on the range of axial height and depth of oxide
discontinuities modeled are based on inspection and hot-cell examinations of the
upper elevations of the severely damaged and failed Browns Ferry fuel Reload 10
rods typical of samples BF2-1 and BF2-5. For the case of oxide spall, the axial height
of the discontinuity (i.e., spall region) is assumed to be significantly larger than in
the case of an oxide plateau. It is noted that due to the similarity in the severity of
the assumptions made relative to the size of the discontinuity, the case of oxide
spallation is analogous to oxide suppression underneath the upper elevation spacers
typical of that shown in Fig. 2. The heat flux at the inner and outer fuel cladding
surfaces is applied via specification of a linear heat generation rate (LHGR) typical
of nodal powers at the upper elevation of sound but heavily damaged Browns Ferry
fuel rod YJS616 B8. Note that crud is not included in the thermal model as it is
highly conductive relative to the oxide and thus has small impact on the determined
thermal gradients. Specific thermal model parameters used are described by the pa-
rameters defined in Table 2.
The resulting fuel cladding (r–z plane) thermal gradients due to oxide layer dis-
continuities from oxide spalling are shown in Fig. 13 and from an oxide plateau are
shown in Fig. 14. Two different reference oxide layers are assumed, i.e., 30 and
60 lm. Based on the assumed LHGR, for the 30 lm reference oxide layer, the refer-
ence cladding surface temperature (i.e., cladding surface temperature beneath the
reference oxide layer) is 306 C; and for the 60 lm reference oxide layer, the
LUTZ ET AL., DOI 10.1520/STP154320120198 1153
TABLE 2 Model parameters used to determine cladding thermal gradient.
Fuel Cladding Thickness, TH 0.7 mm

Fuel Cladding Section Axial 25 mm
Height Modeled, HT
Thick Oxide Thickness, OXTHICK 30 or 60 lm
Thin Oxide Thickness, OXTHIN 15 or 45 lm

Axial Height of Spall or Thin Ox- 2:5; Spall mm
ide Plateau, OXHT 0:25; Plateau
LHGR 250 W/cm
reference cladding surface temperature is 317 C. For an oxide spall, the thermal
gradient on the outer cladding surface occurring in the spall region is approxi-
mately 1.5 C/mm for both 30 and 60 lm reference oxide thicknesses that have a
15 lm oxide discontinuity, and 4 C/mm for a 60 lm reference oxide thickness that
has a 45 lm oxide discontinuity. For an oxide plateau like that shown in Fig. 5, the
thermal gradient on the outer cladding surface occurring in the region of the oxide
plateau is approximately 2 C/mm for both 30 and 60 lm reference oxide thick-
nesses that have a 15 lm oxide discontinuity, and 6 C/mm for a 60 lm reference
oxide thickness that has a 45 lm oxide discontinuity. The total temperature drop
for the case of an oxide spall across the analyzed axial height of the outer cladding
surface is 5 C for both 30 and 60 lm reference oxide thicknesses that have a 15 lm
oxide discontinuity, and 16 C for a 60 lm reference oxide thickness that has a
30 lm oxide discontinuity. The total temperature drop for the case of an oxide pla-
teau across the analyzed axial height of the outer cladding surface is 3 C for both
30 or 60 lm uniform thicker oxide layers that have a 15 lm oxide discontinuity,
and 10 C for a 60 lm uniform thicker oxide layer that has a 45 lm oxide
discontinuity.
As described, the axial height of the discontinuity for an oxide spall is assumed
to be significantly larger than an oxide plateau. Therefore, the affected (i.e. cooler)
cladding region is larger for the case of oxide spall, as evident in comparing Figs. 13
and 14. Additionally, from Figs. 13 and 14, it is noted that, for oxide spalling, the
magnitude of the thermal gradient present on the fuel cladding outer surface is
maintained virtually through the entire cladding thickness, while for an oxide pla-
teau, the magnitude of the thermal gradient decreases through the thickness of the
cladding.
HYDROGEN DIFFUSION MODELING AND RESULTS

A 1D finite element hydrogen diffusion model was developed in order to predict
the change in hydrogen distribution in fuel cladding over time due to the estimated
thermal gradients. The hydrogen diffusion model is based on the kinetics of the
thermal diffusion of hydrogen in Zircaloy-2 in the single (a-Zr) or two-phase (a-
FIG. 13 Predicted thermal gradient ( C) through thickness of Zircaloy-2 fuel cladding in

the presence of a discontinuous oxide layer for the reference conditions in Table
2 representing oxide spalling.
Zr þ d-hydride) region as described in Ref. [3]. The governing equations of hydro-

gen flux (J) and hydrogen diffusion (@N=@t) are described next.
The hydrogen flux in the single (a-Zr) or two-phase (a-Zr þ d-hydride) region
is given by the following relation:

Q N i
(3) J i ¼ Di rN i þ i 2 rT ; i ¼ aZr; d-hydride
RT
LUTZ ET AL., DOI 10.1520/STP154320120198 1155
FIG. 14 Predicted thermal gradient ( C) through thickness of Zircaloy-2 fuel cladding in

the presence of a discontinuous oxide layer for the reference conditions in Table
2 representing converging oxide fronts surrounding an oxide plateau.
where:
R ¼ the gas constant ð8:314 J=K molÞ,
ð0Þ
D ¼ the diffusivity of hydrogen ðDi ¼ Di eQi =RT Þ,
(0)
D ¼ the hydrogen diffusion constant,
Q ¼ the hydrogen diffusion activation energy,
Q ¼ the heat of transport,
N ¼ the hydrogen concentration, and

T ¼ the temperature.
The total hydrogen flux J and total hydrogen content N are given by the
following:
(4) J ¼ Ja þ Jd
(5) N ¼ Na þ Nd
The change of hydrogen concentration with time is given by application of the con-
tinuity equation as
@N
(6) ¼ r J
@t
Model properties used in Eq 3 are provided in Table 3. Other properties used in the
model include the terminal solid solubility for dissolution and terminal solid solu-
bility for precipitation (TSSP) of hydrogen in Zircaloy, which are also provided in
Table 3.
Benchmarking was performed in order to demonstrate the adequacy of the fi-
nite element model. Sawatzky’s experimental results for the thermal diffusion of
hydrogen where hydrogen exists in the a-Zr þ d-hydride phase [3] were modeled.
The results are presented in Fig. 15. The figure shows good agreement between the
benchmark cases and the finite element diffusion model.
The thermal analysis results described in the previous section were then applied
to the diffusion model. As discussed, the hydrogen diffusion model is 1D; therefore,
the thermal gradients applied correspond to the outer, cooler cladding surface. For
oxide spalling, thermal gradients at the outer cladding surface from Fig. 13 were
applied. For the case of an oxide plateau, thermal gradients at the outer cladding
surface from Fig. 14 were applied. The reference cladding surface temperature
TABLE 3 Model properties used in hydrogen diffusion model.
Da
ð0Þ cm2 Ref. [3]
2:17 103
s
Dd
ð0Þ cm2 Ref. [3]
1:09 103
s
Qa 4170 K Ref. [3]
R
Qd 5730 K Ref. [3]
R
Qa 3015 K Ref. [3]
R
Qd 653 K Ref. [3]
R
TSSD 1:28 105 e36540=RT ppm Ref. [4]
(
TSSP 1:07 104 e21028=RT ppm T 533K Ref. [4]
5:26 104 e28068=RT ppm T > 533 K
LUTZ ET AL., DOI 10.1520/STP154320120198 1157
FIG. 15 Hydrogen distribution in a 2.5 cm Zircaloy-2 rod after 816 h; experimental [3]
and finite element results. The initial hydrogen distribution is 130 ppm uniformly
distributed. End temperatures are at 133 C and 477 C.
applied is 306 C and 317 C for 30 and 60 lm reference oxide thicknesses, respec-
tively. The initial hydrogen concentration was assumed to be uniform. The magni-
tudes of initial hydrogen concentration used were one case below (100 ppm) and
one case above (200 ppm) TSSP. Cladding hydrogen contents in the heavily dam-
aged but sound YJS616 B8 rod are below TSSP in the lower half of the rod and
above TSSP in the upper half of the rod. An overall modeling time of 1000 h was
chosen to demonstrate hydride localization trends and to compare localization rates
based on thermal gradients.
The results of the diffusion modeling are shown in Fig. 16 for oxide spalling and
Fig. 17 for an oxide plateau. Modeling results for the oxide spalling case shown in
Fig. 16 demonstrate that significant hydride localizations are predicted in a short
time relative to a reactor cycle for oxide spalling when hydrogen concentrations are
above TSSP for the thermal gradients analyzed. Sensitivity to the magnitude of the
gradient is evident. For an oxide plateau, Fig. 17 demonstrates that significant
hydride localization is predicted for hydrogen concentrations above TSSP for the
case of 6 C/mm thermal gradient and 10 C temperature difference; however, for
the lower thermal gradients analyzed, the hydride localization is significantly
reduced. In all cases, for hydrogen concentrations below TSSP, no localization is
predicted to occur.
Evaluation of the effects of the clad surface average temperature (i.e., compar-
ing the 30 lm versus 60 lm reference oxide thickness cases with similar discontinu-
ities and thermal gradients) on localization indicates that the rate of hydride
FIG. 16 Predicted hydrogen diffusion in Zircaloy-2 in the presence of a thermal gradient

due to a discontinuous oxide layer representing oxide spalling. The initial
hydrogen concentrations evaluated are 100 ppm and 200 ppm assumed to be
uniformly distributed.
LUTZ ET AL., DOI 10.1520/STP154320120198 1159
FIG. 17 Predicted hydrogen diffusion in Zircaloy-2 in the presence of a thermal gradient

due to a discontinuous oxide layer representing converging oxide fronts
surrounding an oxide plateau. The initial hydrogen concentrations evaluated are
100 ppm and 200 ppm assumed to be uniformly distributed.
localization is increased with increased reference clad surface temperature. Com-

parison of the oxide spall results to the oxide plateau results suggests that at similar
thermal gradients, localizations associated with the oxide spalling cases are more
severe, indicating sensitivity to the size of the affected (cold spot) area. Additionally,
considering the difference in thermal gradients through the clad due to oxide spall
(Fig. 13) and the oxide plateau (Fig. 14), it can be reasoned that the depth of the
hydride localization for oxide spalling is expected to be more severe than the oxide
plateau, as is evident when comparing Figs. 4 and 5.
Discussion
ROLE OF LITHIUM IN FAILURES
Comments on Li Role as Event Initiator or Late-stage Aggravating Species: One
objective for the second set of SIMS analyses was to look for similarities between
the one-cycle BF3 Reload 8 rod (samples BF3-1 and BF3-2) and the two-cycle
Reload 10 BF2 rod (samples BF2-4 and BF2-5), as a way of identifying a possible
common water chemistry factor that could be attributed to causing the unusual cor-
rosion. A common water chemistry factor is a possibility because there is a com-
mon radwaste cleanup system that is shared between the BF2 and BF3 plants,
which returns some coolant to both reactors from the same source. If Li had caused
the elevated corrosion, then similar residual Li loading characteristics that have a
distinctive pattern might be expected to be present in affected rods from both the
BF2 and BF3 plants, and to be dissimilar from rods that operated in unaffected
plants like L and H2. The BF3 rod was discharged much earlier than the BF2 rod,
and before its corrosion condition was very far advanced. Under these circumstan-
ces, any similarity in Li loading characteristics that includes the early-discharged
BF3 rod would support Li’s role as an initiating cause of the corrosion condition.
The SIMS results showed similarities as well as differences among the affected
plants. The strongest similarity for both the affected BF2 and BF3 affected rods is
that the higher elevation samples had the highest amount of Li, whereas the unaf-
fected plant H2 three cycle rod (samples H2-1 and H2-2) showed the reverse trend.
However, the SIMS results, Fig. 10, showed that Li concentration in the BF3 rod is
low compared with the BF2 rod, which is not a strong similarity. Nevertheless, the
lower Li level in the BF3 rod could have been affected by Li leach out during stor-
age. The BF3 rod was discharged after one cycle of operation and spent over two
years longer time in the spent fuel pool prior to inspection/retrieval compared with
the two cycle BF2 rod. The extent to which the observed Li in the oxide has been
affected, if at all, by the difference in operation and post-operation history is there-
fore not known. As a result, the residual Li loading patterns in BF2 and BF3 are not
strong enough to conclude that Li was a common water-borne factor between the
failure events in the two plants. Likewise, with comparably low amounts of Li in the
affected BF3 samples, and the plant L and H2 samples, a conclusion as to whether
LUTZ ET AL., DOI 10.1520/STP154320120198 1161
Li initiated the events in both reactors cannot be drawn from this simple compari-
son either.
It is known that Li can cause accelerated corrosion in Zr-alloys if the concentra-
tion is high, if the heat flux is high, and/or enhanced by boiling through concentrat-
ing in the oxide (e.g. Refs. [5–8]. The role of Li (and B) under irradiation and in the
presence of heat flux and boiling has been reported [5]. This work showed that
increased corrosion of Zircaloy-4 tested out-of-pile under two-phase heat transfer
conditions occurred after 26 days with 10 ppm Li in the water with heat flux of
100 W/cm2 and with void fractions above 30 %. In high void fraction locations,
where corrosion was increased, there was an associated increase in Li in the oxide
and 100 ppm Li was measured in the oxide (60–70 lm thick). McDonald et al.
also observed that 100 ppm Li in the oxide was a critical concentration for acceler-
ated corrosion of Zircaloy-4 in autoclave tests with LiOH [6].
The role of heat flux has been found to be applicable under in-reactor condi-
tions and it has been reported [7] that high Sn Zircaloy-4 cladding can undergo ex-
cessive corrosion resulting in cladding failure when subjected to a high heat flux
condition under normal pressurized water reactor (PWR) water chemistry condi-
tions (2 ppm Li). There is apparent correlation between the Li pickup in the oxide
and the oxide thickness. It was reported [7] that Li concentration in the oxide of
PWR cladding was dependent on both heat flux as well as oxide thickness, that the
Li concentration was significantly increased when oxide thickness was above
40 lm, and that Li pickup in the oxide was generally low provided oxide thickness
was below 30 lm. In a review of available laboratory data in 1992 [8], it was
shown that for moderate concentrations of Li in water (7 to 70 ppm) a corrosion
enhancement occurred only after a weight gain of around 30 mg/dm2 (2 lm ox-
ide) was reached under non-heat flux conditions. Such a general trend is qualita-
tively consistent with the in-reactor information [7], where increased Li in oxide
was observed only when the corrosion oxide thickness was increased (40 lm or
more). The low levels of post-damage Li in samples BF2-4, BF3-1, and BF3-2, could
be attributed to a similar phenomenon, since the oxide thickness for these samples
were less than 30 lm. In the context of BWRs, heat flux and void fractions used in
Ref. [7] are common and applicable to the damaged BF2 fuel rods.
These reports [5–8] collectively suggest that under BWR operation, Zircaloy
can develop increased corrosion with Li in the coolant under high heat flux boiling
conditions that allow it to concentrate on the cladding surface, and it could remain
in the oxide and only be detected post-exposure in substantial amounts if the oxide
is sufficiently thick to retain it. Furthermore, concentrations of Li in the oxide above
100 ppm are associated with Li-enhanced corrosion.
The affected BF2 Reload 10 rods have more than 300 ppm Li at the most heav-
ily damaged elevations, and have a sufficiently thick oxide to retain it to some
extent in the oxide post-operation. An accelerating effect on corrosion is expected
once > 100 ppm Li is present in an oxide regardless of its buildup history or
whether it initiated the corrosion in the first place. In this sense, the results suggest
that Li contributed to the failures at least by aggravating the late stage corrosion of
the BF2 Reload 10 rods, regardless of any role it may have had in initiating the
corrosion.
Natural Source of6Li and Ratio of 7Li/6Li as Indicator of Event Timing: the
detection of 6Li in samples BF2-1, BF2-5, and BF3-2 implies a natural Li source, as
there is no significant mechanism for producing the 6Li isotope in-reactor. It was
considered then that the 7Li/6Li ratio might provide insight into the timing of an
assumed introduction of natural Li. Knowledge of the timing of Li uptake could
provide further insight into Li’s role as an early initiator of the failures rather than
solely as a late-stage aggravator.
The ratio of 7Li to 6Li based on natural abundance is 12.2. The 7Li/6Li ratio is
expected to increase with irradiation time based on the neutron capture cross sec-
tions for the removal of 7Li via the 7Li(n,c)8Li reaction and 6Li via the 6Li(n,c)7Li
and 6Li(n,a)3H reactions. The SIMS results showed an association of B with Li. The
generation of 7Li from 10B therefore needs to be considered also. The ratio of 11B to
10
B based on natural abundance is 4.05. 10B has a significantly larger cross-section
than 11B and results in the generation of 7Li. If this source is considered, the genera-
tion of 7Li from 10B will additionally enhance the 7Li/6Li ratio with irradiation time.
Since the irradiation time for the BF2 Reload 10 rods is greater than the BF3 Reload
8 rod, a greater amount of 7Li is expected from this source, and the greater 7Li/6Li
ratio (420) observed in the two-cycle BF2 Reload 10 rod, sample BF2-5, compared
with 20 for the one-cycle BF3 rod, sample BF3-2, is qualitatively consistent with
such a consideration. However, the concentration of B in the coolant throughout
the lifetime of the affected rods is not known in enough detail to accurately model
the expected 7Li contribution from 10B. Any estimation if 7Li from this B source is
therefore subject to large uncertainties, and thus it is not possible to use the meas-
ured 7Li/6Li ratios to assess the introduction timing of any Li sources, and for that
matter to help differentiate Li’s role as an initiator of the failures rather than only as
a late stage aggravator.
Hideout Return Study: The SIMS results showed presence of Li in the affected
oxide from BF2 and BF3 rods and therefore indicate presence of Li during opera-
tion. Unlike PWRs, Li is not intentionally added to the coolant in BWRs, and the
presence of Li in BWRs in sufficient quantities to affect cladding corrosion is there-
fore not expected. However, further evidence for low Li concentrations in the reac-
tor water that might have concentrated on the cladding surface at damaging levels
can be deduced from a hideout return study [9] that was performed as part of the
failure investigation. Hideout return studies are typically used (mostly in PWRs) to
evaluate the expected impurities “hidden” in water within crud deposits, i.e., in
close proximity to the cladding. Hideout is typically increased with the degree of
boiling. Hideout return studies are performed during the reactor shutdown process
as impurities can return from the hidden locations due to, for example, decreased
boiling as reactor power decreases or increased solubility as reactor temperature
decreases. In the hideout return study [9], reactor water samples from affected plant
LUTZ ET AL., DOI 10.1520/STP154320120198 1163
BF2 and unaffected plant H2 at their respective end of cycle (both in February
2003) were collected at various stages of the reactor shutdown, and analyzed for 19
species. Five of the species analyzed showed a significant difference between the two
plants, with BF2 showing the greater return in all five cases. The five species are sul-
fate, chloride, Ca, Li and B. Figure 18 shows the concentration variation for Li and B
as the reactor power is reduced from full power (right hand chart for a given spe-
cies) to zero power and then as the reactor temperature is reduced at zero power
(left hand chart). Li concentration in the reactor water during the study was up to
0.79 ppb for BF2, compared to values that are below the 0.1 ppb detection limit for
H2. Li reactor water concentrations were not measured at other times when the
BF2 Reload 10 fuel was operating. Assuming a 10,000X concentration factor, an av-
erage of about 5 ppm Li could be present on the BF2 cladding at the time of the
FIG. 18 Variation of Li (upper set) and B (lower set) during shutdown stages for BF2
and H2 at end of February 2003. During shutdown, the reactor power is
reduced at temperature from 100 to 0 % power (from the right side of the chart
on the right); the reactor temperature is then reduced (chart on left). LLD is the
Lower Level of Detection. From Ref. [9].
study. Similarly, B levels for BF2 during the study period were up to 33 ppb, which
were higher than for H2.
By taking into account factors such as reactor cooling system water mass and re-
actor water cleanup system flow rate, the concentration values from the hideout
return study can be used to estimate the total mass inventory returned for each spe-
cies. Based on the data in the hideout return study [9], the ratio of total return at BF2
relative to H2 was highest for Li at 7.3, followed by sulfate at 4.7 and Cl at 4.1.
Higher hideout returns of Li, Ca, B, chloride, and sulfate species at BF2 suggest con-
gruent differences in water chemistry during operation in the cycle immediately pre-
ceding the hideout study conducted in February 2003. Steady state water chemistry
data for that cycle as reported in the study [9] showed that chloride and sulfate levels
were higher at BF2 than at H2, consistent with the hideout return results.
Other than the measurements made for the hideout return study, no
other relevant reactor water Li measurements were made during the Reload 10 oper-
ation, and B measurements were infrequent during the Reload 10 operation. Since
the same concentration factor (concentration in deposit hideout/concentration
measured after shutdown) is typically expected for hideout species, the results from
the hideout return study provide a surrogate to more frequent, direct reactor water
measurements and suggest that Li and B reactor water concentrations were higher
at plant BF2 compared with plant H2 during the operating cycle that ended with
the February 2003 hideout study just as sulfate and chloride were higher. Although
the hideout study results point to an important water chemistry difference between
the affected and unaffected plants related to Li, they do not conclusively establish
the timing of Li ingress into the oxide, and for that matter a cause and effect rela-
tionship between Li and the failures.
Comments on B-generated Source of Li: B is commonly present in BWR coolant
across the fleet now, and also during the period that the affected BF2 Reload 10 fuel
operated. During the failure cycle 12 in BF2, only occasional measurements were
made, which showed B to vary between about 20 and 180 ppb (Fig. 19). Present-day
median B levels for 12 month rolling monitoring periods ending between December
2011 and March 2013 are not much different (Fig. 19). B is likely from a water-
borne source that may originate from cracks in control blade absorber tubes that
contain B4C. A control blade that contains cracks could provide a chronic source of
B, as well as acute releases when a failure first occurs or when a failure is aggravated
during operation.
One possible water-borne source of 7Li is likely to be from the transmutation of
the B that is in the coolant, even though such a source cannot account for the obser-
vation of 6Li in some samples (which may be indicative of more than one Li
source). Deviations in the 7Li/6Li ratio measured by SIMS indicate that this trans-
mutation has in fact occurred. The correlation between Li and B in the oxide
observed in the SIMS results supports the possibility that the Li and B are both
from a related water-borne source, rather than from some other source such as an
as-fabricated contaminant in the cladding metal. However, higher coolant B levels
LUTZ ET AL., DOI 10.1520/STP154320120198 1165
are common in other unaffected BWRs (Fig. 19), therefore, at least one other factor
would be needed for a B-generated Li source to impact corrosion as it did at BF2
but not in other BWRs like plant H2.
One of these other possible factors is that there might have been a difference in
the coolant ion filter exchange efficiencies for Li and B when the BF2 Reload 10 fuel
operated compared to various other plants that have operated with higher levels of
B in the coolant, and from cycle to cycle within BF2, i.e. the affected Reload 10
cycles versus the unaffected Reloads 9 and 11 cycles. With low efficiency filtration,
ppb levels of Li and B could exist in the coolant, acutely or chronically, and concen-
trate over time on the cladding surface under BWR high heat flux conditions. Dif-
ferent fuel reloads might be affected differently depending on the time that a
filtration system underperformed concurrent with periods that high Li might have
existed in the coolant. For example, if the BF2 filtration system was particularly
inefficient during Cycle 11 when the affected Reload 10 fuel was in its first cycle
and had a nascent oxide compared to the older, unaffected Reload 9 fuel that had a
more mature oxide, and when the unaffected Reload 11 fuel was not yet operating,
then the effect of Li exposure could be different for these three Reloads. Infrequent
measurements of B and non-existent measurements of Li in the coolant during the
time that the BF2 Reload 10 fuel operated makes it impossible to fully evaluate such
a condition. Under these circumstances it is not possible to conclusively establish a
causal relationship between B and Li and the failures.
FIG. 19 Comparison of present-day 12 month rolling median BWR fleet reactor water B
contents to 1999–2003 fleet reactor water B data applicable to operating
period of failed BF2 Reload 10 fuel and BF3 Reload 8 fuel. The availability of the
present-day rolling median values depends on the plant and range from
periods ending between December 2011 and March 2013.
CORROSION-RELATED HYDRIDE FAILURE MECHANISM

The failures had the external appearance of crud induced localized corrosion
(CILC) that was prevalent into the mid-1980s [10–12]. With the CILC failure
mechanism, nodular corrosion initiated in nodular corrosion susceptible materials,
and then progressed to autocatalytic corrosion when high concentrations of water-
borne copper infiltrated into the cladding oxide, which impeded heat transfer and
resulted in local, through-wall corrosion pits. Perforations located at corrosion pits
occurred at relatively low exposure when cladding hoop stresses are insignificant.
The 2001–2003 BF2 failures proved to be different than the past CILC experience
in that water-borne copper concentrations and copper infiltration into the oxide
were negligible, corrosion penetration was relatively shallow (maximum 127 lm ox-
ide), nodules were relatively short and squat rather than long and lenticular, pri-
mary cladding perforation occurred due to cracking of brittle hydride lenses that
formed on the outer surface prior to failure, and at higher exposures (20 GWd/
MTU for CILC versus 30–35 GWd/MTU for BF2). Garlick [12] recognized that a
hydride ring could form around a CILC oxide patch driven by thermal gradients,
which could crack and fail the cladding, but in the BF2 case no such thick, CILC-
like oxide patches exist. In contrast, at BF2, spalling, oxide plateaus, and step
changes in corrosion beneath the spacer created the thermal gradients that localized
hydrogen, which the finite element modeling demonstrated can readily form when
a modest amount of hydrogen is picked up. These differences establish a new varia-
tion, or previously unrecognized variation, of a BWR cladding nodular corrosion-
related, hydride failure mechanism.
CORRECTIVE ACTIONS
A lengthy investigation of the corrosion failures determined that they were
caused by a complex interaction of cladding material, power/duty, and water
chemistry. Although some important underlying aspect related to the initiation
of the failure event has not yet been determined, corrective measures were taken
to the extent possible to provide some margin against recurrence. Actions by the
fuel vendor led to the 2004 introduction of cladding that has tighter specification
limits for Zircaloy-2 alloying elements and a modified heat treatment. The clad-
ding is expected to provide additional resistance to nodular corrosion and pro-
vide more margin in demanding water chemistry environments. Actions were
also taken to implement improved water chemistry controls, which included con-
trolling feedwater zinc injection to <0.4 ppb, installation of higher efficiency filter
elements to reduce feedwater iron concentrations, installation of new filter con-
nection hardware to improve condensate demineralizer operations, and expanded
reactor water chemistry monitoring. These changes would reduce the degree of
crud formation or increase ion exchange efficiency, and hence diminish the like-
lihood of recurrence of corrosion failures. However, some risks will remain in
the BWR fleet since some of the corrosion initiating conditions remain to be
understood and defined better.
LUTZ ET AL., DOI 10.1520/STP154320120198 1167
Conclusions
In the present work, a significant BWR fuel failure event has been described. The
failure mechanism is a new variation, or previously unrecognized variation, of a
BWR corrosion failure mechanism. The failures occurred in cladding with corro-
sion resistance that was inadequate for the demanding environment that it operated
in, which led to elevated hydrogen pickup and severe spalling. Under these condi-
tions, a hydrogen diffusion finite element model showed that hydrogen above the
solubility limit can readily accumulate under the action of local thermal gradients
that are caused by the types of oxide discontinuities that developed in the affected
rods and the power levels that they operated with. Failure ultimately occurred
within these brittle hydride accumulations.
Cladding samples were examined by SIMS in an attempt to identify a water chem-
istry factor in the failure mechanism. One such possible factor is Li, which was detected
in the affected cladding. It was spatially correlated with B at surprisingly high levels for
a BWR, up to 550 ppm, but also at much lower, unremarkable levels in unaffected
samples from other BWRs. A plausible water-borne source for 7Li is leakage of B from
control blades and transmutation of 10B to 7Li. However, such a source cannot com-
pletely reconcile the limited observation of 6Li or that the presence of B in BWR cool-
ant is common. Comparisons of cladding samples with different operating histories,
deviations of the 7Li/Li6 ratio from natural abundance, and Li and B hideout return
studies were considered in the absence of adequate Li coolant concentration data to
determine if Li may have initiated and caused the corrosion failures. These efforts were
ultimately inconclusive regarding the possible role of Li as the initiator and primary
cause, and at this time, no other potential initiating conditions or primary causes have
been identified. However, greater than 100 ppm Li in the oxide is detrimental to corro-
sion, which the late-stage affected rods held. In this sense, it is likely that these high Li
levels in the oxide at least contributed to the failures secondarily by advancing corro-
sion in later stages of the failures, regardless of its inconclusive role as an initiator.
ACKNOWLEDGMENTS
The writers would like to express their appreciation to Dr. Stephane Portier, Dr. Mat-
thias Martin, and the Paul Scherrer Institut for providing “above and beyond,” world-
class SIMS measurements on the irradiated cladding materials.
References
[1] Schardt, J., Keys, T. A., Lemons, J. F., Ottenfeld, C., “Fuel Corrosion Failures in the Browns
Ferry Nuclear Plant,” Proceedings of the 2004 International Meeting on LWR Fuel Per-
formance, Orlando, FL, Sept. 19–22, 2004, Paper No. 1036.
[2] Lutz, D., Lin, Y. P., Schneider, R., Yeager, H., Kucuk, A., Cheng, B., Lemons, J., Nesmith, K.,
“Investigation of BWR Fuel Failures,” Proceedings of the European Nuclear Society Top-
Fuel 2012 Conference, Manchester, UK, Sept 2–6, 2012.
[3] Sawatzky, A., Vogt, E., “Mathematics of Thermal Diffusion of Hydrogen in Zircaloy-2,”
AECL-1411, Atomic Energy of Canada, Chalk River, ON, Oct. 1971.
[4] Une, K., and Ishimoto, S., “Dissolution and Precipitation Behavior of Hydrides in Zircaloy-
2 and High Fe Zircaloy,” J. Nucl. Mater., Vol. 322, 2003, pp. 66–72.
[5] Billot, P., Yagnik, S., Ramasubramanian, M., Peybernes, J., and Pêcheur, D., “The Role of
Lithium and Boron on the Corrosion of Zircaloy-4 Under Demanding PWR-Type Con-
ditions,” Proceedings of the Zirconium in the Nuclear Industry: Thirteenth International
Symposium, ASTM STP 1423, G. D. Moan and P. Rudling, eds., American Society for Test-
ing and Materials, West Conshohocken, PA, 2002, pp. 169–189.
[6] McDonald, S. G., Sabol, G. P., and Sheppard, K. D., “Effect of Lithium Hydroxide on the
Corrosion Behavior of Zircaloy-4,” Proceedings of the Zirconium in the Nuclear Industry:
Sixth International Symposium, ASTM STP 824, D. G. Franklin and R. B. Adamson, eds.,
American Society for Testing and Materials, West Conshohocken, PA, 1984, pp. 519–530.
[7] Garzarolli, F., Sell, H.-J., and Thomazet, J., “PWR Li Coolant Chemistry Performance and
Fuel Cladding Performance,” Annual Meeting on Nuclear Technology, (Jahrestagung
Kerntechnik), May 2002, Stuttgart, Germany.
[8] Polley, M. V., and Evans, H. E., “Review of Effect of Lithium on PWR Fuel Cladding
Corrosion,” Water Chemistry of Nuclear Reactor Systems 6, BNES, London, 1992.
[9] EPRI, Shutdown Hideout Return Chemistry at Browns Ferry-2 and Hatch-2, 1009448,
EPRI, Palo Alto, CA, 2004.
[10] Marlow, M. O., Armijo, J. S., Cheng, B., and Adamson, R. B., “Nuclear Fuel Cladding Local-
ized Corrosion,” American Nuclear Society International Topical Meeting on Light Water
Reactor Fuel Performance, April 21–24, 1985, Orlando, FL.
[11] Ogata, K., Matsuoka, H., Yamada, T., and Narama, T., “Post Irradiation Examination of the
Failed Fuel Rod in the Hamaoka Atomic Power Station Unit 1,” Proceedings of the Inter-
national Topical Meeting on Light Water Reactor Fuel Performance, West Palm Beach,
FL, April 17–21, 1994, ANS, La Grange Park, IL.
[12] Garlick, A., Sumerling, R., and Shires, G. L., “Crud Induced Overheating Defects in Water
Reactor Fuel Pins,” J. Br. Nucl. Soc., Vol. 16(1), 1977, pp. 77–80.
LUTZ ET AL., DOI 10.1520/STP154320120198 1169
DISCUSSION
Question from B. K. Shah, BARC Mumbai:—You have shown nodular corro-
sion on BWR clad. Can you comment on the characteristics of the local region of
the clad which led to nodular oxidation initiation at that local region?
Authors’ Response:—The clad characteristics, such as grain size, texture, com-

position and SPP size, are generally considered to be uniform in a given fuel rod ini-
tially; the SPP size can vary depending on the accumulated fluence on account of
dissolution under irradiation. These attributes were investigated by examining
archive materials and also in the plenum section of failed and non-failed fuel rods,
and no abnormalities were found. Observed local, large changes in corrosion behav-
ior, such as azimuthal variation and at spacer locations, are due to changes in the
exposed environment as opposed to variations in the clad.
Question from D. Kaczorowski, AREVA:—You measured by SIMS up to a 300

ppm of lithium. What was the thickness of oxide in this region?
Authors’ Response:—For the region where 300 ppm Li was measured, the
local thickness of oxide was about 55 lm.
Question from N. Ramasubramanian, ECCATEC Inc. Canada:—Implicating

lithium: 10B transmutes to 7Li. 10B is only at 10% natural abundance. The observa-
tion of 7Li as well as 6Li is a bit puzzling. If 7Li/6Li ratio is at normal abundance (6Li
undergoes transmutation), then it seems like chemistry had not been controlled
well.
Authors’ Response:—In the paper, the source of 7Li and 6Li is discussed, includ-
ing transmutation of 10B to 7Li and the removal of 7Li via the 7Li(n, c)8Li reaction
and 6Li via the 6Li(n, c)7Li and 6Li(n, a)3H reactions. As noted in the paper, 7Li/6Li
ratio is not at normal abundance ratio and presence of 6Li implies a natural source.
Efforts were made to use 7Li/6Li ratio to assess timing of an assumed natural Li
introduction. However, conclusions could not be reached due to uncertainties over
timing of B release.
Questions from Anand Garde, Westinghouse Electric Co.
Q1:—Was the crud deposition level for failed rods unusual or high?
Authors’ Response:—No, the crud deposition level in failed rods was typical of
BWR operations.
Q2:—What was the iron level in the coolant for the cycle in which failures
occurred?
Authors’ Response:—The coolant iron level at BF2 was between 1 - 2 ppb during
the operational times of interest; at BF3, the Fe level was between 0.5 - 2 ppb.
Questions from Philippe Bossis, CEA Q1:—Are the sides of “plateau corrosion”
corresponding to nodular corrosion?
Authors’ Response:—The plateau region is in between adjacent regions of nodu-

lar corrosion that have often developed or coalesced into sheet oxide with similar
thickness.
Q2:—What was the water chemistry during these failures? HWC (hydrogen
water chemistry), NMCA (noble metal chemical addition)?
Authors’ Response:—At BF3, some fuel failures occurred in the third cycle of
operation. The paper presented results from a one-cycle rod that showed elevated
corrosion was established during the first cycle of operation. The examined rod
with elevated corrosion operated in water chemistry without HWC and without
NMCA. However, the second and third cycles of the affected BF3 fuel operated
with HWC and NMCA. At BF2, fuel failures occurred in the second cycle of
affected fuel. Both cycles were operated with HWC and the second cycle was oper-
ated with NMCA.
Q3:—How did spacers protect against corrosion failure? Electrochemically?
Authors’ Response:—The precise reason for lack of severe corrosion at spacer

locations is not known. It is known that the thermal-hydraulic conditions at spacer
locations are different. There is a faster and more turbulent coolant flow at spacer
locations. At high axial elevations, where the void fraction is higher, the spacers
have the effect of increasing the amount of solid water at the surface of the
cladding.
Question from Itai Panas, Chalmers University of Tech:—For Li2 ZrO3, which
when treated by 5M HNO3 is converted into H2 ZrO3 (a proton conductor) (please
see A. Orera et al., Z. Anorg. All. Chem. 631, 1991-1993 (2005) and L. J. Enriquez
et al., Trans J Br Ceram Soc, 81:—, 17 (1982)). Li+ may cause H+ conduction paths
to stay open.
Authors’ Response:—Thank you for the information.
Question from Suresh Yagnik, EPRI:—You have measured high Li by SIMS in

the oxide layer. Together with the boiling condition present, this could readily cause
accelerated corrosion in Zr alloy cladding. Could you comment on this possibility?
LUTZ ET AL., DOI 10.1520/STP154320120198 1171
Authors’ Response:—Although the role of Li in initiating the elevated corrosion

cannot be conclusively proven, many of the observed attributes match published in-
formation on Li effect on Zircaloy corrosion. In the paper, the results from Refer-
ence 5 are discussed by noting that accelerated corrosion in Zr-alloys can occur if
the Li concentration is high, if the heat flux is high, and/or Li level is enhanced by
boiling through concentrating in the oxide. In the context of the present fuel fail-
ures in Boiling Water Reactor, elevated corrosion was most notable at upper eleva-
tions, where boiling void fraction is high. The presence of significant quantities of
Li is not expected in BWRs. The observation of high Li in the thick oxide near the
failure location is supporting evidence for associating Li with accelerated corrosion.
1173
Author Index
A Chaput, H., 800–829
Chen, J.-H., 1077–1106
Abolhassani, S., 540–573 Cheng, B., 1138–1171
Adámek, J., 897–931 Choudhuri, G., 95–117
Adamson, R. B., 55–92 Christensen, M., 55–92
Almer, J., 1107–1137 Cloué, J.-M., 759–799, 853–894
Ambard, A., 404–437, 438–478 Colas, K., 1107–1137
Andrén, H.-O., 373–403, 515–539 Coleman, C. E., 800–829
Atwood, A. R., 607–627 Comstock, R. J., 404–437, 479–514
Aubin, J. L., 184–224 Cottis, R. A., 404–437
Aulló, M., 577–606 Couet, A., 479–514
Crébier, P., 1002–1053
B Cuisinier, D., 438–478
Culebras, F., 577–606
Babu, C. Phani, 302–330
Banerjee, S., 23–51, 259–281 D
Barashev, A. V., 729–758
Barberis, P., 118–137, 159–183, Dahlbäck, M., 373–403, 830–852
225–256 Daymond, M. R., 1107–1137
Bart, G., 540–573 Desquines, J., 952–978
Bay, F., 331–345 Dey, G. K., 95–117, 138–158, 282–301
Béchade, J.-L., 759–799, 853–894 Doriot, S., 184–224, 759–799
Bertsch, J., 540–573 Dunavant, R., 1138–1171
Bickel, G. A., 693–725, 800–829 Duriez, C., 952–978
Blat, M., 438–478 Duthoo, D., 331–345
Blat-Yrieix, M., 404–437
Bossis, P., 438–478 F
Bourdiliau, B., 184–224
Brachet, J. C., 184–224, 1002–1053 Feaugas, X., 853–894
Burda, J., 897–931 Forsey, A., 404–437
Buyers, A., 800–829 Francis, E. M., 404–437
Frankel, P. G., 404–437
C Freeman, C., 55–92
Fremiot, P., 118–137
Cabrera, A., 1002–1053
Cantonwine, P., 55–92 G
Carcey-Collet, D., 331–345
Chakraborty, S., 95–117 Gaillac, A., 225–256, 331–345
Chakravartty, J. K., 259–281 Ganesha, G. N., 302–330
1174
Garcı́a-Infanta, J. M., 577–606 K

Garde, A. M., 577–606, 607–627,
673–692 Kaczorowski, D., 159–183, 438–478
Gilbon, D., 759–799, 853–894, Kaestner, A., 1054–1073
1002–1053 Kapoor, K., 302–330
Golubov, S. I., 729–758 Kapoor, R., 259–281
Grandjean, C., 952–978 Kearns, J. J., 3–22
Griffiths, M., 693–725, 800–829 Kemény, T., 932–951
Grosse, M., 540–573, 979–1001, Keskar, N., 282–301
1054–1073 Kim, H.-G., 346–369
Grovenor, C. R. M., 404–437 Kim, I.-H., 346–369
Guerin, P., 118–137 Klouček, V., 897–931
Guilbert, S., 952–978 Kobylyansky, G. P., 628–650
Guilbert, T., 184–224 Koo, Y.-H., 346–369
Krejčı́, J., 897–931
Kucuk, A., 1138–1171
H
Kumar, V., 259–281
Hallstadius, L., 55–92, 373–403, Kunstár, M., 932–951
404–437, 515–539, 540–573, 673–692, Kuo, R.-C., 1077–1106
830–852 Kuri, G., 540–573
Hamann, J., 438–478
Hamon, D., 184–224 L
Hermann, A., 540–573
Lacote, P., 952–978
Hoffmann, P. B., 159–183
Ledergerber, G., 540–573, 830–852
Horváth, M., 932–951
Legras, L., 759–799
Hózer, Z., 932–951
Lemaignan, C., 540–573
Lemons, J., 1138–1171
I Leo Prakash, D. G., 138–158
Le Saux, M., 1002–1053
Itisri, K., 302–330
Lin, Y.-P., 1138–1171
Ivermark, M., 373–403, 830–852
Lindgren, M., 515–539
Linhart, S., 897–931
J Lorinčı́k, J., 897–931
Lozano-Perez, S., 404–437
Jacques, P., 1002–1053 Lutz, D., 1138–1171
Jha, S. K., 259–281, 282–301, Lyon, S., 404–437
302–330
Jomard, F., 438–478 M
Josefsson, B., 373–403
Joubert, J. M., 184–224 Mader, E. V., 55–92
Jourdan, J., 184–224 Mani Krishna, K. V., 138–158, 282–301
1175
Mardon, J.-P., 159–183, 759–799, Prahlad, B., 302–330

853–894, 1002–1053 Preuss, M., 138–158, 404–437
Markelov, V. A., 628–650 Přibyl, A., 897–931
Martin, M., 540–573 Proff, C., 540–573
Mascaro, A., 184–224
Menut, D., 759–799 Q
Miquet, A., 438–478, 759–799
Mistry, R. K., 302–330 Quinta da Fonseca, J., 138–158
Moat, R., 404–437
Mocellin, K., 331–345
R
Montheillet, F., 225–256
Montigny, G., 952–978
Raepsaet, C., 184–224
Motta, A. T., 479–514, 1107–1137
Raizada, V., 282–301
Mueller, A. J., 673–692
Rao, S. V. Ramana, 302–330
Restani, R., 540–573
N
Roessger, C., 1054–1073
Romero, J., 830–852
Nagy, I., 932–951
Négyesi, M., 897–931
Nikulina, A. V., 628–650 S
Ni, N., 404–437
Novikov, V. V., 628–650 Saibaba, N., 138–158, 259–281,
Novoselov, A. E., 628–650 282–301, 302–330
Novotny, T., 932–951 Sarkar, A., 259–281
Novotný, L., 897–931 Schneider, R., 1138–1171
Schrire, D., 577–606
O Shah, B.K., 95–117
Shevyakov, A. Yu., 628–650
Obukhov, A. V., 628–650 Shishov, V. N., 628–650
Olier, P., 184–224 Siegl, J., 897–931
Onimus, F., 853–894 Singh, B. N., 729–758
Sinha, R. K., 651–672
P Sinha, S. K., 651–672
Sopoušek, J., 897–931
Pan, G., 607–627, 673–692 Srivastava, D., 95–117, 138–158,
Panas, I., 515–539 282–301
Park, J.-Y., 346–369 Steinbrueck, M., 979–1001,
Peregud, M. M., 628–650 1054–1073
Perez-Feró, E., 932–951 Stevens, J., 159–183
Pintér-Csordás, A., 932–951 Stoller, R. E., 729–758
Piot, D., 225–256 Stuckert, J., 1054–1073
Portier, S., 540–573 Sundell, G., 373–403, 515–539
1176
T Vauglin, C., 118–137, 331–345

Verhaeghe, B., 759–799
Tejland, P., 373–403 Vimi, A., 932–951
Thuvander, M., 373–403, 515–539 Volkova, I. N., 628–650
Toffolon-Masclet, C., 184–224 Vrtı́lková, V., 897–931
Tournadre, L., 853–894
Tran, M. T., 225–256 W
Tupin, M., 184–224, 438–478
Walter, M., 1054–1073
U Walters, L., 693–725
Wei, J., 404–437
Urvoy, S., 184–224 Wiese, H., 540–573
Wimmer, E., 55–92
V Wolf, W., 55–92
Vaibhaw, K., 282–301 Y

Valance, S., 540–573
Valizadeh, S., 540–573 Yagnik, S. K., 1077–1106
Vandenberghe, V., 184–224, 1002–1053 Yeager, H., 1138–1171
1177
Subject Index
A delayed hydride cracking, 800–829
density functional theory, 515–539
allotriomorph, 95–117 diffusion, 55–92, 438–478
alloy composition, 159–183 diffusion coefficients, 3–22
alloying elements, 479–514 dimensional properties, 159–183
anisotropy, 3–22 dislocation channeling, 607–627
atom probe tomography, 373–403, dislocation loops, 55–92, 759–799
515–539 dislocations, 628–650
atomistic simulation, 55–92 ductility, 673–692
AXIOM, 673–692
E
B
E635 alloy, 628–650
ballooning and burst, 932–951
EBSD, 225–256
basketweave, 225–256
electron microscopy, 628–650
beta-phase, 800–829
electronic conductivity, 479–514
beta-quenching, 830–852
embrittlement, 3–22
BWR, 1138–1171
EPMA, 540–573
erbium, 184–224
C
ERDA, 185–207
CALPHAD, 184–224 examination, 628–650
chemical potential, 118–137
cladding embrittlement, 932–951 F
c-loops, 853–894
CNPGAA, 479–514 failure mechanism, 1138–1171
coating, 346–369 finite element modeling, 282–301,
computer modeling, 55–92 1138–1171
cooling rate, 225–256 finite element, 331–345
corrosion layer, 952–978 fuel assemblies, 577–606, 628–650
corrosion mechanisms, 404–437 fuel cladding, 159–183, 628–650,
corrosion, 159–183, 438–478, 897–931
515–539, 651–672, 1138–1171 fuel rods, 331–345
creep, 159–183
crystallographic texture, 302–330 G
D grid, 577–606
growth, 577–606, 628–650, 759–799,
defects, 55–92 853–894
deformation mechanisms, 138–158 guide tube, 577–606
1178
H irradiation growth, 577–606, 729–758,

830–852
high burnup, 540–573, 607–627 in situ compression, 138–158
high corrosion resistance, 673–692
high temperature, 952–978 K
high-temperature oxidation, 346–369,
1002–1053 kinetics, 979–1001
hot deformation, 259–281
hot extrusion, 282–301 L
hot-cell PIE, 607–627
HPU, 515–539 lath, 95–117
HPUF, 515–539 Laves phase, 118–137
hydride localization, 673–692 line broadening, 800–829
hydrides, 55–92, 515–539, 1077–1137 loss of coolant accidents, 952–978,
hydriding, 1138–1171 1054–1073
hydrogen content, 540–573 low-tin Zircaloy-4, 1002–1053
hydrogen diffusion, 1138–1171
hydrogen effects, 55–92 M
hydrogen embrittlement, 1054–1073
hydrogen induced growth, 577–606 M5TM, 159–183, 1002–1053
hydrogen pickup, 479–514, 607–627, martensite, 23–51
830–852, 1002–1053 mechanical properties, 159–183,
hydrogen trapping, 184–224 1077–1106
hydrogen, 3–22, 118–137, 628–650, microstructure, 225–256, 628–650,
853–894 693–725, 759–799, 800–829, 897–931
miscibility gap, 118–137
I modeling, 225–256, 897–931
image analysis, 225–256 N

ingot breakdown, 259–281
interface, 95–117 neutron diffraction, 138–158
intergranular strains, 138–158 neutron imaging, 1054–1073
in-PWR corrosion, 607–627 neutron irradiation, 628–650,
in-PWR fuel performance, 607–627 693–725, 729–758, 800–829
in-PWR irradiation growth, 607–627 neutronic burnable poison, 184–224
iron, 118–137, 159–183 nuclear safety research, 979–1001
ion irradiation, 853–894 numerical modeling, 331–345
in-reactor deformation, 693–725
in situ oxidation, 404–437 O
irradiation, 438–478, 759–799
irradiation creep, 693–725 Optimized ZIRLOTM, 607–627,
irradiation effects, 540–573 673–692
1179
oxidation, 373–403, 540–573, simulation, 282–301

577–606, 897–931, 979–1001 solubility, 3–22, 55–92
oxide delamination, 673–692 spalling, 673–692
oxide film thickness, 628–650 steam oxidation, 952–978
oxide morphology, 1002–1053 steam pressure, 1002–1053
oxygen diffusion, 952–978 stress, 853–894
suboxide, 515–539
P supersaturation, 95–117
synchrotron x-ray diffraction,
peeling, 673–692 404–437, 1107–1137
phase field method, 95–117
phase transformation, 23–51, T
404–437
PHWR, 651–672 TEM, 373–403, 540–573
pilgering, 302–330 texture, 3–22, 302–330
platelets, 225–256 theory, 729–758
poolside PIE, 607–627 thermal-diffusion, 3–22
post-cooling mechanical properties, thermo-mechanical-treatment, 302–330
1002–1053 Thermocalc, 184–224
post-irradiation ductility, 607–627 thermodynamics, 118–137
pre-hydriding, 1002–1053 three-layer-clad, 184–224
pre-transient oxide, 1002–1053 twinning, 138–158
precipitation, 759–799
pressure tube, 282–301, 651–672, U
693–725, 800–829
up-hill diffusion, 118–137
Q
V
quenching, 225–256
variant selection, 225–256
R
W
reorientation, 1107–1137
wear resistance, 673–692
S weld, liner, 118–137
welding, 331–345
second-phase precipitates, 159–183,
373–403, 628–650 X
secondary hydrogenation, 1054–1073
secondary ion mass spectrometry, XANES, 479–514
1138–1171 X-ray diffraction, 404–437, 800–829,
SIMS, 540–573 1107–1137
1180
Z ZIRLO, 404–433, 577–606

Zr alloys, 259–281
Zircaloy, 55–92 Zr–Sn alloys, 138–158
Zircaloy-2, 373–403, 651–672, Zr-2.5b, 282–301, 302–330, 693–725,
1138–1171 800–829
Zircobase, 184–224 Zr1Nb alloys, 759–799
Zirconium, 23–51 Zr1Nb–O, 897–931
zirconium alloys, 331–345, 404–437, ZrO2, 404–437
479–514, 515–539, 673–692, Zirconium oxidation, 932–951
979–1001, 1077–1106
zirconium-erbium-hydrogen-oxygen
phase diagrams, 184–224
Comstock | Barbéris
Zirconium in the Nuclear Industry 17th International Symposium STP1543
ASTM INTERNATIONAL
Selected Technical Papers
Zirconium in
the Nuclear
Industry
17th International Symposium
ASTM INTERNATIONAL STP 1543
Helping our world work better Editors
Robert J. Comstock
Pierre Barbéris
ISBN: 978-0-8031-7529-7
Stock #: STP1543
www.astm.org

STP 1543-2015

Uploaded by

Copyright:

Available Formats

You might also like

STP 1543-2015

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

STP 1543-2015

Uploaded by

Copyright:

Available Formats

Comstock | Barbéris

Zirconium in the Nuclear Industry 17th International Symposium STP1543

Editors: Robert Comstock, Pierre Barbéris

Zirconium in the Nuclear

ASTM Stock #STP1543

Peer Review Policy

Printed in Bay Shore, NY

Reﬂections on the Development of the “f ” Texture Factors for Zirconium

Displacive and Diffusional Transformations of the Beta Phase in Zirconium Alloys 23

Schemel Award Paper

Effect of Hydrogen on Dimensional Changes of Zirconium and the Inﬂuence of

Basic Metallurgy and Alloying Effects

Phase Field Modeling of Microstructure Evolution in Zirconium Base Alloys 95

Thermodynamics of Zr Alloys: Application to Heterogeneous Materials 118

Inﬂuence of Sn on Deformation Mechanisms During Room Temperature

Impact of Iron in M5TM 159

Characterizing Quenched Microstructures in Relation to Processing 225

Identiﬁcation of Safe Hot-Working Conditions in Cast Zr–2.5Nb 259

A Numerical Study of the Effect of Extrusion Parameters on the Temperature

Study on Effect of Processing on Texture Development in Zirconium-2.5 %

Numerical Modeling of Fuel Rod Resistance Butt Welding 331

Application of Coating Technology on Zirconium-Based Alloy to Decrease

Corrosion and Hydrogen Pickup

Oxidation Mechanism in Zircaloy-2—The Effect of SPP Size Distribution 373

Effect of Sn on Corrosion Mechanisms in Advanced Zr-Cladding for Pressurised

Understanding of Corrosion Mechanisms of Zirconium Alloys after Irradiation:

Effect of Alloying Elements on Hydrogen Pickup in Zirconium Alloys 479

Corrosion and Hydrogen Uptake in Zirconium Claddings Irradiated in Light

Oxidation and Hydrogen Uptake of ZIRLO Structural Components Irradiated to

Performance and Property Evaluation of High-Burnup Optimized ZIRLOTM

Corrosion, Dimensional Stability and Microstructure of VVER-1000 E635 Alloy

The Effects of Microstructure and Operating Conditions on Irradiation Creep of

Irradiation and Hydrogen Effects

Breakthrough in Understanding Radiation Growth of Zirconium 729

Microstructural Evolution of M5TM7 Alloy Irradiated in PWRs up to High

Understanding the Drivers of In-Reactor Growth of β -Quenched Zircaloy-2

Impact of Hydrogen Pick-Up and Applied Stress on C-Component Loops: Toward

High Temperature Transient Behavior

Contribution to the Study of the Pseudobinary Zr1Nb–O Phase Diagram

Experimental Comparison of the Behavior of E110 and E110G Claddings at High

Effect of Pre-Oxide on Zircaloy-4 High-Temperature Steam Oxidation

Deviations From Parabolic Kinetics During Oxidation of Zirconium Alloys 979

Inﬂuence of Steam Pressure on the High Temperature Oxidation and Post-Cooling

Analysis of the Secondary Cladding Hydrogenation During the Quench–LOCA

Degradation and Failure Mechanisms

Effect of Hydride Distribution on the Mechanical Properties of Zirconium-Alloy

Hydriding Induced Corrosion Failures in BWR Fuel 1138

Author Index 1173

Subject Index 1177

STP 1543, 2015 / available online at www.astm.org / doi: 10.1520/STP154320120216

Reflections on the Development

Development of the “f” Texture Factor

(1) Pref ¼ P½0001 cos2 / þ P?½0001 ð1 cos2 /Þ

Because RVi is unity,

(4) f ¼ RVi cos2 /i

(5) Pref ¼ fP½0001 þ ð1 f ÞP?½0001

FIG. 1 Correlation of expansion and x-ray preferred orientation measurements.

If the intensity, I/, is expressed in times random units, the denominator of Eq 6 is

X-RAY METHOD FOR CALCULATION OF “F” FACTORS

(9) f0002 þ 2f1010 ¼ 1

(10) f0002 þ f1010 ¼ f1120 ¼ 1

(5a) DV=V0 100 ¼ 1:7 103 CH;sol

(5b) Dl=l0 100 ¼ 5:6 104 CH;sol

(8) DV=V0 100 ¼ 1:7 103 50 ¼ 0:085

DV=V0 100 ¼ 2095:2 ð1 e0:7452 XÞ