The Effect of Stress Concentration On Hydrogen Embrittlement of A Low Alloy Steel

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Corrosion Science, Vol. 38, No. 5, pp.

721-733, 1996
Pergamon Copyright 0 1996 Elsevier Science Ltd
Printed in Great Bridn. All rights reserved
0010-938X/96%15.00+0.00

0010-938X(96)00161-1

THE EFFECT OF STRESS CONCENTRATION ON HYDROGEN


EMBRITTLEMENT OF A LOW ALLOY STEEL*
D. HARDIE and SU’E LIU’
Department of Mechanical, Materials and Manufacturing Engineering, University of Newcastle upon Tyne,
Newcastle upon Tyne NE1 7RU, U.K.

Abstract-The hydrogen embrittlement of a quenched and tempered low alloy steel has been investigated by
straining notched specimens in 1 bar hydrogen atmosphere at 1.3 x 10m4 mm/s. The circumferentially notched
specimens exhibited a significant embrittlement when their mechanical behaviour in hydrogen was compared with
that in air. Although the effect of notch depth on fracture strength in air is negligible, an increase in the depth of
notch increases susceptibility to embrittlement when testing in gaseous hydrogen. Analysis of the effects is
complicated by the facts that: (i) the specimens show some degree of notch sensitivity even when strained in air. and
(ii) localised plastic deformation may occur with relatively shallow notches. Such effects are eliminated at high stress
concentration factors, where there is a systematic loss in fracture stress in hydrogen as the notch severity increases
from Kc = 2.6 to 5.7 (where a 87% reduction of fracture stress occurs), but a relatively stable value is then reached
even for very severe notching by fatigue pre-cracking. Whether or not the effect is due to increasing concentration of
hydrogen in the triaxial stress region ahead of the notch, there is no doubt that increasing the stress concentration
makes hydrogen more effective as an embrittling agent. On the other hand, it is also clear that severe embrittlement
can be introduced with relatively low nominal stress concentration factors by increasing the depth of relatively blunt
notches, and this is believed to be due to the greater volume of material then affected.

Keywords: low alloy steel, hydrogen embrittlement, SEM.

INTRODUCTION
It has long been recognised’ that hydrogen may produce a disastrous reduction in the
ductility and toughness of steel, particularly when the steel is in the quenched and tempered
condition. The susceptibility of a steel to embrittlement by hydrogen is influenced by the
strength leve1233and by the detail of the microstructure.475 The source of the hydrogen is
obviously of less importance, although acid pickling’ and cathodic charging6 have proved to
be potent sources of embrittlement, and a reasonable assessment of susceptibility to
embrittlement may be obtained by straining in an atmosphere of hydrogen.’ Although
factors such as stress induced diffusion’ and dislocation transport’ have figured in many
proposed mechanisms, there has not been a great deal of interest shown in the effect of stress
concentration on the degree of embrittlement. Van Ness et &loS1’ in their work on 4140
steel having a tempered martensite structure used notched cylindrical specimens to assess

Manuscript received 12 July 1995.


*This paper was presented at the International Conference on Environment-Sensitive Cracking: Mechanisms
and Industrial Applications, which was held at Guilin, China, &9 September 1994.
‘This work was carried out whilst Su’e Liu was a visiting scientist from the Institute of Corrosion and
Protection of Metals, Chinese Academy of Sciences, Shinyang 110015, supported by the British Council.

721
122 D. Hardie and Su’e Liu

the effect of straining in high pressure hydrogen and actually compared sharp and blunt
notches. ‘I They concluded that the effect of hydrogen increases with notch severity. Later
Walter and Chandler12 studied the effect of both hydrogen pressure and notch severity on
the embrittlement of low alloy and stainless steel, but the examination of stress
concentration effects was again strictly limited. More recently, Toribio and Elices13
carried out a certain amount of work on notched samples of a pearlitic steel with a view to
identifying the importance of localised strain rate in an embrittlement mechanism but,
although the notch radii covered a wide range, the variation in stress concentration factor
was much more restricted than in the earlier work.

EXPERIMENTAL METHOD
Material and specimens
The material used in the work was a commercial oil hardening steel containing (wt%)
nominally 0.35-0.45 carbon, 1.3-l .8 nickel, 0.45-0.70 manganese, 0.4-l .4 chromium and
0.2-0.35 molybdenum, in the form of 25 mm diameter bar. All of the tensile specimens
employed were machined from the bar after it had been cut into equal quadrants. Smooth
specimens were machined with a 11.35 mm gauge length of 3.15 mm diameter, whereas
notched specimens had a circumferential 60” V-notch machined at the midpoint of the
11.35 mm gauge section. The latter were divided into three groups according to D-the
outside diameter, d-the minimum diameter at the root of the notch and R-the root radius
of notch, so that the effect of notch depth and stress concentration factor on hydrogen
embrittlement could be investigated systematically:
A. Specimens having an outside diameter (D) varying between 3.4 and 4.0 mm and a
minimum diameter at the root of the notch (d,l of 3.17 f 0.05 mm, with the notch root radius
(R) of 0.212 0.02 mm kept constant.
B. Specimens having a minimum diameter at the root of the notch (6) varying between
3.2 and 4.0 mm and an outside diameter (D) of 3.95 +0.05 mm, with the notch root radius
(R) of 0.21 f 0.02 mm kept constant.
C. Specimens with a notch root radius (R) varying from 0.032 to 0.32 mm, whilst the
outside diameter (D) of 3.95 + 0.05 mm and the minimum diameter at the root of the notch
(d) of 3.17kO.05 mm were kept constant, i.e. d/D=O.80. In addition a specimen with
D=4.0 mm, d= 3.6 mm and R= 0.032 was pre-cracked in air by rotary fatigue-the
resulting fatigue crack was about 0.2 mm around the periphery.
The introduction of the notch produces a discontinuity in the stress field and causes a
stress concentration, the severity of which may be expressed by a theoretical stress
concentration factor (KJ that is defined as the ratio between the highest load and the
nominal stress, where the nominal stress is the (nominal) uniform stress on the minimum
section. The stress concentration factors calculated according to Peterson14 are strongly
dependent on the root radius of the notch. Due to the constancy of the root radius of the
notch for specimens in Groups A and B, Kt for these notched specimens changed only very
slightly from 2.1 to 2.6 and therefore they were mainly used to investigate the effect of notch
depth on the degree of embrittlement. The specimens in Group C, where Kt varied between 2
and 6 and included a pre-cracked specimen, could be employed in a more extensive study of
the influence of notch severity. The nominal stress concentration factors obtained in this
way should, of course, be used with caution because they are strictly only applicable to an
Hydrogen embrittlement of a low alloy steel 723

elastic rather than an elastic-plastic situation and the true localisation of stress will be
modified by any plastic component of the strain.
The Ni-Cr-Mo steel used is known to be very susceptible to embrittlement by a
hydrogen environment in certain tempered conditions and its susceptibility increases with
decreasing tempering temperature after an oil quench. In order to study the effect of stress
concentrations introduced by a notch, it is important to define a suitable heat treatment
procedure to apply to specimens in order to produce comparative data. Some pre-testing
was therefore done using smooth and notched specimens that had been given various heat
treatments and the following optimum test conditions were confirmed for the subsequent
study of the influence of notch severity.
Specimens were heat treated by austenitising at 840°C in argon for 1 h and quenching in
oil before tempering at 300°C in argon for 1 h, then air cooling. This produced a
microstructure of tempered martensite, an ultimate tensile strength (UTS) of 1720 MPa, a
fracture strength of 2500 MPa, a reduction in area at fracture of 50% and an elongation of
11%. The gauge length, including the surface inside the notch for notched specimens and the
transitional portions at the ends of the gauge length of machined specimens were carefully
polished on 1000 grade silicon carbide paper and degreased by acetone. Various
dimensional parameters D, A and R were then carefully measured by a measuring
microscope before straining.

Tensilestraining
The equipment used for tensile straining over a wide range of crosshead speeds has been
described elsewhere. ” The specimens and part of the straining rod could be sealed in a
pressure vessel connected to an oil diffusion pump and a rotary backing pump producing a
vacuum < 10v4 mPa. After the degreased specimens had been inserted in the grips of the
tensile machine, the vessel was sealed and evacuated then filled with purified hydrogen to
1 bar. The specimens were then strained to failure at a cross head speed of 1.3 x 10y4 mm/s
(giving an initial strain rate for the smooth specimens of 1.2 x 10F5/s) at ambient
temperature. The resultant load-displacement curves from an X-Y recorder could be
converted to stress-strain curves.
Various parameters were chosen to evaluate the strength and ductility of smooth
specimens. Reduction in area at fracture (RA%) and elongation (El%) were employed to
indicate ductility and the maximum load divided by the original cross sectional area
(ultimate tensile stress, omax = or,), and fracture load divided by the area at the fracture (the
true fracture stress, ar) were all employed to assessstrength. Since the plastic deformation of
the notched specimens is relatively small, the main parameters adopted for comparison were
the notched strengths UD and od calculated from the following equations:
P
gross section stress, on = max
7rD2/4

Pmax
net section stress, cd = -
rd=/4
where Pm,, was the maximum load from load-displacement curves.
The degree of embrittlement produced by the hydrogen may be evaluated by comparison
of the results for specimens tested in hydrogen with those in air. The percentage reduction of
124 D. Hardie and Su’e Liu

notched strength CD (RNS%) for specimens with different Kt may be calculated using the
following equation:
o, RNS = aD(air) - ED
0 x 100.
cD(air)
Smooth specimens strained in hydrogen showed very little drop in UTS (about 2%) but
both the reduction in area and the elongation fell to zero and there was a resultant decrease
of 34% in fracture stress.
The fracture surfaces of typical specimens were examined by scanning electron
microscope.

EXPERIMENTAL RESULTS
The effect of notch depth was examined using specimens from Groups A and B (Fig. 1).
The dependence of (TD upon d/D shows an obvious decrease in the notched stress [TDin 1 bar
hydrogen compared with that in air. Although the effect of d/D on fracture strength in air is
negligible, a decrease in the ratio d/D increases susceptibility to embrittlement when testing
in gaseous hydrogen and results for the two groups lie on a common curve. The dependence

z 1200.
0
i= ,b o HydrOW+
- -0-o
w”

: I I I
400
z 0.8 0.9 1.0
d d/D

+
Y 0.8 d/D
1 .o

Fig. 1. Variation of notched strength Q, and Q with 5 for notched specimens having either
constant d and varying D (Group A) or constant D and varying d (Group B) strained in air or 1 bar
hydrogen.
Hydrogen embrittlement of a low alloy steel 725

Air

% STRAIN

Hydrogen
d/D-l.0

04 0404
% STRAIN

Fig. 2. Typical stress strain curves for smooth and notched specimens with various6 values (shown)
strained in air and in 1 bar hydrogen at 1.3 x 10e4 mm/s.

of Crdupon d/D in hydrogen is slightly more complex because of the counteracting effect of
the increase in gd with decreasing d/D in air. Interpretation of the results is complicated by
variations in the individual stress strain curves (Fig. 2) because plastic deformation was
associated with the specimens strained in air, even after notching, at the high ratios of d/D
(> 0.8).
When specimens covering a wider range of stress concentration factor (KJ were strained
in air and in 1 bar hydrogen at 1.3 x 10-4mm/s the notched strength was found to be
approximately the same for all notched specimens with Kt between 2 and 6 strained in air,
and even for pre-cracked specimens, whereas the notched strength was considerably
reduced by 1 bar hydrogen (Fig. 3). The discontinuity in the plot above Kt = 2.2 for straining
in air arises because of the incorrect use of UTS rather than fracture stress where significant
plastic deformation is involved (Fig. 4).
If gD of notched and pre-cracked specimens in air is assumed to be a constant at
1430 MPa, the reduction of notch strength CrD(%) in 1 bar hydrogen compared with that in
air versus stress concentration factor may be calculated according to eqn (3). The degree of
hydrogen environment embrittlement obviously increases with increasing Kt between 1 and
6 but then reaches a relatively stable value (%RNS ~87%) that is not altered even by very
severe pre-cracking.
The fracture surfaces of typical smooth and notched specimens strained in air were
examined by scanning electron microscope and in both cases showed obvious dimpled
rupture. When straining was conducted in hydrogen, however, the fracture was very
different (Fig. 5) and clearly divided into two regions: A-an intergranular fracture zone
(b)
Air
A - .-.-A
2000
c?‘\ A
0’0 .A

\
2 1600
m I
!i co
B
::
w 1200- E 1200

5
g
g
P
i aoo- (I) a00
ii
L
2
z
is
" 400- 4oc

I I I I I
0 ’ 0 VA
1.0 3.0 5.0 1.0 3.0 5.0
STRESS CONCENTRATION FACTOR STRESS CONCENTRATION FACTOR

Fig. 3. Variation of notched strength or, and od with stress concentration factor K, for specimens having a constant $= 0.8 (Group C) and varying $(Groups A and B).
Hydrogen embrittlement of a low alloy steel 721

Fig, 4. True fracture strength and reduction in area at fracture for smooth and notched specimens
(Groups A and B) strained in air at 1.3 x lop4 mm/s.

and B-a region of mixed ductile and brittle fracture. Shear lips were present on specimens
fractured in air but were smaller on specimens fractured in 1 bar hydrogen, especially for
notched specimens (Fig. 6). This indicates that fracture in air initiated inside the specimen
and propagated to the surface by shear to form a shear lip. Fracture propagated from
initiation sites, such as scratches on the surface, when straining was conducted in hydrogen.
For pre-cracked specimens, the fractography of the fatigue pre-crack was transgranular
whereas the fracture in hydrogen propagated from the pre-crack in an intergranular fashion
(Fig. 7).

DISCUSSION
The stress strain curves (Fig. 2) indicate that notched specimens are more brittle than
smooth specimens, even when strained in air. There is an obvious restriction imposed upon
plastic deformation by the presence of a notch and, even if the notch region could deform in
the same way as the smooth bar itself before fracture, the total elongation would be much
smaller than that in the absence of a notch. On the other hand, in addition to the stress
concentration introduced by the notch, there is also a change in the state of stress and, even
when the stress is uniaxial throughout the remainder of the gauge length, there is a region of
triaxial stress in the vicinity of the notch where much higher stresses are consequently
required to produce yield. This is reflected in the increasing value of ffd with decrease in the
ratio d/D (Fig. 1) for straining in air. Since the additional material contributing to the gross
cross section is not completely redundant, it is also worth noting the associated variation in
QD where increasing the severity of the notch causes a steady, though small, decrease. The
results of straining in a hydrogen atmosphere show significant difference because of the
embrittling effect of hydrogen. There is a steady decrease in both gd and OD with notch depth
until a critical value of d/D =0.87 is reached, beyond which the gross section stress at
fracture becomes reasonably constant and, as a result, there is a continual increase in frdwith
decrease in d/D. All of the specimens tested in hydrogen show no measurable gross plastic
deformation and a completely linear (elastic) stress strain curve is obtained (Fig. 2). For the
728 D. Hardie and Su’e Liu

Fig. 5. Fracture surface of a smooth specimen strained in 1 bar hydrogen at 1.25 x 10b5is showing
intergranular failure (A), ductile rupture (B) and large shear lips.
Hydrogen embrittlement of a low alloy steel 729

Fig. 6. Frac :ture surface of a notched specimen strained in 1 bar hydrogen at 1.3 x 10K4mm/s
($ = 0.80, Kt = 4.9).
730 D. Hardie and Su’e Liu

Fig. 7. Fracture surface of a pre-cracked specimen strained in 1 bar hydrogen at 1.3 x 10W4mm/s,
showing the transition from transgranular fatigue to intergranular crack growth in hydrogen.

particular geometry of specimen employed here, the embrittlement by hydrogen appears to


reach a maximum value when the ratio d/D becomes ~0.87. This suggests that the
embrittlement is essentially concerned with the severity of the stress concentration rather
than the detailed geometry.
The effects of stress concentration may be assessed by employing the theoretical stress
concentration factors (KJ calculated by Peterson,‘* even though these are strictly only
applicable to a perfectly brittle material. Although the calculated stress concentration factor
varied over a very limited range (2.1-2.6) for specimens in Groups A and B, specimens in
Group C were designed (by careful machining of root radii) to extend the range to Kt = 5.7
and even to a specimen pre-cracked (by fatigue) and these results are particularly
illuminating (Fig. 3). It then becomes obvious that, although the fracture strength in air is
unaffected by change in Kt between 2 and 6, straining in hydrogen introduces a significant
embrittling effect that increases almost linearly with increasing stress concentration.
Although on and cd both exhibit the same effect of stress concentration (Fig. 3) i.e. a
steady increase in the embrittlement caused by the hydrogen environment as the stress
concentration becomes more severe above Kt = 2, there is a discontinuity in the various
curves because of the value employed for Kt = 1, i.e. smooth specimens. Relationships in the
region Kt = l-2.6 are however likely to involve localised plastic deformation (Fig. 4) and this
complicates the true assessment of stress concentration effects, as discussed above.
It appears that the presence of hydrogen, either absorbed on the steel surface14 or
transported into the metal by localised plastic deformation at the notch,” is sufficient to
induce crack initiation long before the stress reaches a high enough level for general ductile
collapse across the section, so creating a deeper “notch” and a higher stress concentration
leading to rapid failure. In this case the initiating crack is intergranular along the prior
austenite grain boundaries even though the final failure is still by dimpled ductile rupture.
The greater area of intergranular cracking on the fracture surface of notched specimens
(Fig. 6) compared with smooth (Fig. 5) probably represents the longer period available for
slow crack growth before the reduced section becomes overloaded when early crack
initiation occurs.
Hydrogen embrittlement of a low alloy steel 731

Intergranular brittleness of iron and iron-based alloys in the presence of hydrogen can
result from several factors, but the grain boundary composition is generally important.16
The segregation of certain impurities in the grain boundaries can alter markedly the
cohesive strength of the interface. In the highly stressed region ahead of the notch, after local
yielding occurs, a microcrack may be nucleated and propagate along the grain boundary.
When a specimen is strained in a hydrogen atmosphere, it is postulated that the
concentration of hydrogen that occurs in the triaxial stress region ahead of a notch or a
crack can lower the cohesive strength even more. Hirth” has also suggested that a particle/
matrix interface is often a deep trap for hydrogen. The thermodynamic argument of Li et
al.” supports the view that the hydrogen concentration C, in a stressed body varies
exponentially with the local hydrostatic tension (Thaccording to
ah VH
CH = COexp RT

where dh = (0, i + 022 + g&/3, VH is the partial molar volume of hydrogen in iron, and Co is
the average hydrogen concentration varying as the square root of the hydrogen pressure
according to Sievert’s law. In an elastic-plastic material with a pre-crack or a notch, gh may
be taken as roughly 3 times the yield strength. Since the triaxial stress region ahead of a
notch is often a shallow trap for hydrogen, the combination of a triaxial stress with the
segregation of impurities at grain boundaries has concentrated hydrogen ahead of the notch
by either dislocation transport or lattice diffusion. Cavities nucleate at the high hydrogen
accumulated region during the slow straining process and the rupture ductility is reduced.
So fracture occurs at stresses well below the value found in air (Figs 2 and 3). The extent of
the stress concentration increases with increasing stress concentration factor (KJ and the
hydrogen concentration in the triaxial stress region ahead of a notch or a pre-crack increases
simultaneously. As a result, the critical hydrogen concentration for nucleation of a
microcrack is reached earlier during straining for specimens having a severe stress
concentration. Therefore, there is a systematic loss in notched strength bo in hydrogen as
the notch severity increases to 5.7. When the specimen has a very sharp notch on even a pre-
crack, the maximum stress in the triaxial stress region is almost independent of the applied
concentration factor Kt or the stress intensity, and it is mainly the volume of the highly-
stressed region which increases with notch severity. So a relatively stable value of notched
strength (To in hydrogen is then reached which is not altered even by very severe notching by
fatigue pre-cracking.
All the results here are not explicable purely in terms of the stress concentration
introduced by the notch. There is a very significant discrepancy between the Group C
specimens and the rest in this respect (Fig. 3). Increasing the stress concentration factor of a
notch of constant depth by decreasing the notch radius (Group C) is much less effective in
increasing the effect of hydrogen than increasing the depth of a relatively blunt notch
(Groups A and B). This could be concerned with the greater volume of material influenced
by the stress intensification in the latter case. Some confirmation for such a hypothesis is
obtained from the result at Kt = 2.2 in Group C, which seems to fall in the wrong batch of
points. This could be because the specimen concerned had a blunter notch than the rest
($ = 0.08 1). There would appear to be two different, though related, factors contributing to
the effectiveness of embrittlement of this low alloy steel by hydrogen: the stress
concentration factor and some other parameter related to the depth of blunt notches. Both
of these could influence the volume of material affected by a particular stress concentration.
732 D. Hardie and Su’e Liu

Table 1. The range of notch geometries investigated in this and earlier work

Authors 4D RID 4

Steinman et al.” 0.50 0.015 and 0.102 4.2 and t2


Walter and ChandlerI 0.60 0.004,0.008 and 0.018 8.4, 5.8 and 3.8
Toribio and Elicesr3 0.80 0.03 and 0.36 3.2 and 1.4
0.22 0.05 and 0.40 1.7 and 1.12
Present work-A and B 0.94-0.80 0.055 2.10-2.61
C 0.80 0.008-0.081 5.7-2.2

Table 2. Effect of high pressure hydrogen on the notched strength of


quenched and tempered low alloy steels

Notched strength (MPa)


% Reduction
Steel Helium/air Hydrogen in strength

ASTM A302 1465+61 1188&30 18.8


ASTM A517 1589&23 1224+35 23.0

Although the present work has encompassed a wider spectrum of stress concentration
than some earlier work, 10*‘1,13it was concerned with relatively shallow grooves (Table 1)
and, despite some clarification of the general situation, has exposed a significant degree of
uncertainty surrounding the detailed effect of notch geometry. The danger associated with
reaching profound conclusions on the basis of a relatively few results may be demonstrated
by a careful re-examination of the results of Walter and Chandler’* for two quenched and
tempered low alloy steels. They suggested, from a comparison of tensile strength in high
pressure hydrogen gas with that in air or helium, that the embrittlement increased with stress
concentration factor, passing through a maximum for K=69. Reappraisal taking due
account of scatter indicates a 19-23% reduction in notch strength that is independent of K
(Table 2); varying the radius of their deep notch (Table 1) had a big effect on stress intensity
factor but caused little variation in embrittlement. The much greater variation in notch
geometry used by Toribio and Elices13 revealed a significant variation in embrittlement
upon which K seemed to have little effect, and the maximum occurred with a sharp deep
notch. It is worth noting that they were dealing with a pearlitic eutectoid steel that was
charged with hydrogen by cathodic polarization and the elastic phase of deformation in
their tests occupied only a small fraction of the loading process, in marked contrast to the
present work. Steinman et al.” achieved a 68% loss in the strength of a quenched and
tempered steel when testing a sharp-notched specimen in high pressure hydrogen.
Add to all this the observation that global strain rate has itself been found to influence
the results obtained,” and this has been very variable in past work, and it may be concluded
that a great deal of further work, paying particular attention to the important variables
identified herein, will be necessary before the correct picture is resolved.
Hydrogen embrittlement of a low alloy steel 733

CONCLUSIONS
1. Circumferentially notched specimens exhibit a significant embrittlement when
strained in hydrogen as compared with in air. Analysis of the effects is complicated by the
facts that: (i) the heat treated steel concerned shows some degree of notch sensitivity even
when strained in air, and (ii) the behaviour is influenced by the localised plastic deformation
that may occur with relatively shallow notches.
2. There is no doubt that, when such effects are eliminated at high stress concentration
factors, there is a systematic loss in fracture stress in hydrogen as the notch severity is
increased to 5.7 (where an 87% reduction occurs) although a relatively stable value is then
reached which is not altered even by the very severe notching introduced by fatigue pre-
cracking.
3. Due to increasing concentration of hydrogen in the triaxial stress region ahead of the
notch, increasing the stress concentration makes hydrogen more effective as an embrittling
agent. However, deepening a relatively blunt notch may have a greater effect on
embrittlement in certain circumstances because of its effect upon the volume of material
influenced by the stress concentration.

Acknowledgemenrs-This work was performed in the Department of Mechanical, Materials and Manufacturing
Engineering, University of Newcastle, U.K., as part of an academic link with the Institute of Corrosion and
Protection of Metals of The Chinese Academy of Science, China, supported by the British Council.

REFERENCES
1, W. H. Johnson, Proc. Roy. Sot. 23, 168-179 (1875).
2. K. Farrel and A. G. Quarrel, J. Iron and Steel Inst. 202, 1002-1011 (1964).
3. G. Sandoz, Met. Trans. 3, 1169-1176 (1972).
4. T. Boniszewski and F. Watkinson, Metals and Mafer. 7, 9G96 (1973).
5. D. Hardie and T. I. Murray, Metals Technol. 5, 145-149 (1978).
6. A. R. Troiano, Trans. Amer. Sot. Metals 52, 54-80 (1960).
7. P. Bowker and D. Hardie, Mefal Sci. 9, 432-436 (1975).
8. R. Dutton, Int. J. Hydrogen Energy 9, 147-155 (1984).
9. J. K. Tien, A. W. Thompson, I. M. Bernstein and R. J. Richards, Met. Trans. 7A, 821-829 (1976)
IO. R. H. Cavett and H. C. Van Ness, Welding J. Res., Suppl. 42, 316s-319s (1963).
11. J. B. Steinman, H. C. Van Ness and G. S. Ansell, Welding J. Res., Suppl. 44, 221s-224s (1965)
12. R. J. Walter and W. T. Chandler, Mater. Sci. and Eng. 8, 9G97 (1971).
13. J. Toribio and M. Elites, Corros. Sci. 33, 1387-1409 (1992).
14. R. E. Peterson, Sfress Concentration Facfors. J. Wiley and Sons, New York, p. 21 (1953).
15. J. R. Buckley and D. Hardie, Corros. Sci. 34, 93-107 (1993).
16. C. L. Briant and S. K. Banerji, Int. Metall. Rev. 23, 164-199 (1978).
17. D. Hardie, manuscript in preparation.
18. J. C. M. Li, R. A. Oriani and L. S. Darken, Zeitschr. fiir Physikalische Chem. 49, 271 (1966).

You might also like