Professional Documents
Culture Documents
Bulletin: U R:J'J
Bulletin: U R:J'J
-~--
USI~./iI~U\S ; j 6 ~J.AR
,.., ~Sl\
- -, n ., c a S. \
Tee 145 / OCTOBER ;1969
...l 1,\+c~'\1l1COes
Centr0 l' e ,. , )
BULLETIN
Welding Research Council Bulletins are published January, February, April, May, June, July, August, September, October and
November at 20th & Northampton Sts., Easton, Pa., by the WELDING RESEARCH COUNCIL, 345 East 47th St.. New York. N. Y.
10017. C. F. Larson, Editor. Subscription: $175.00 per year. Single copies: price as noted.
Second·class postase paid at Easton, Pa. C Copyright 1969 by the Weldlnl Research Council.
Interpretive Report on Effect of Hydrogen
in Pressure-Vessel Steels
FOREWORD
Many designers, fabricators and users of pres- ments can be classified quantitatively, such as
sure vessels have had only occasional and casual hydrogen at elevated temperature, the hydro-
contact with the problems of hydrogen. Welders genizing potential of corrosive environments can
were the only large group who had first-hand and usually be rated only qualitatively. Further,
frequent contact with the detrimental effects of the design methods commonly used for pressure
hydrogen and knew some of the means of combat- vessels generally ignore many properties of steels
ing them. Currently the increasing use of higher degraded by the effects of hydrogen. The quanti-
strength steels, as well as the more widespread use tative effects of hydrogen on load-carrying ability
of processes and operations capable of charging of the range of steels and geometries are not yet
steel with hydrogen, make detailed information adequately defined. This will be a broad field for
more necessary. This bulletin is published to future research. Nevertheless, the data in this
make available up-to-date summaries on the most bulletin should find ready application to many
important aspects of hydrogen damage, as well as problems in the design and use of pressure vessels
keys to the literature. and piping.
The information compiled is complete enough
to evaluate the need for special attention and is F. PRANGE, Chairman
sufficiently detailed so that many potential prob- PVRC Subcommittee on
lems can be avoided. While some of the environ- Hydrogen Eml)rittlement
Section I-Basic and Research Aspects
by C. G. Interrante
ABSTRACT. The detrimental effects of a few parts per mil- I ntrod uction
lion of hydrogen in steel at temperatures below about 400 F
have long been recognized. Common experience of such
Service failures, with their potential serious con-
effects include reduction of ductility and reduction of frac- sequences, have led to extensive research studies
ture strength. More recently there has been added some ; n the effects of hydrogen in steel. The work of
realization of accelerated subcritical crack propagation when
steel is exposed to hydrogen environments. All these ef-
the last two decades has led to an understanding
fects are important to the application of pressure vessels for of the effects of temperature, of pressure, of crystal
high-pressure hydrogen, and for vessels exposed to corrosive structure, and of compositional and steel pro-
media capable of charging steels with hydrogen. From the
materials properties standpoint, damage is much more
cessing variables on the solubility, on diffusivity
apt to occur as the hardness or the strength level incre3ses. and on permeability of hydrogen in iron and steel.
Thus, the problem of damage by hydrogen is particularly However, although it is known that the harmful
important in the application of higher strength steels, which effects of hydrogen in steel are governed by these
have the most favorable strength-to-weight ratio for pres-
sure vessels. basic behaviors, many gaps still remain and much
Although the detrimental effects of hydrogen are gener- more research work will be required to permit the
ally. recognized, the significance of many of the variables establishment of quantitative limitations of com-
has not been quantitatively established in terms that allow
direct application to pressure vessel design or use. Only position, strength level, microstructure, hydrogen
in a general way are the effect of notches or other stress content, and environmental features necessary to
raisers known. The primary objectives of research pro- prevent service failures. The high-strength steels
grams should be to furnish quantitative information on
limiting values for mechanical properties and enviromental used for the aerospace applications have been
factors so as to insure against low stress failures in environ- studied most extensively. Particularly needed is
ments capable of supplying hydrogen to the vessel steel. the applicable information for the lower strength
The interpretation of the literature and experiences ex-
, amined in the report suggests that the predominant mode of pressure-vessel steels.
failure in the higher strength steels is a relatively brittle, In the first and second parts of this section of the
quasi-cleavage fracture which results from propagation interpretive report on hydrogen in pressure-vessel
of microcracks, and that this mode may extend into the
realm of the lower strength steels also. Research should be steels, background information is presented on
aimed toward establishing limiting conditions with respect solubility, diffusivity, permeation, removal, and
to strength levels, hydrogen content or effective hydrogen sources of hydrogen. In the third and fourth
pressure, and temperature for the change over from this
quasi-cleavage fracture behavior to one of ductile rupture. parts, the effects of hydrogen on mechanical
properties are discussed, both for steels containing
hydrogen and for steels exposed to hydrogen en-
vironments. The fifth part reviews the mech-
anisms that are proposed to explain the experimen-
C. G. Interrante is Sr. Research Metallurgist, Applied Research Labora-
tal observations. Finally, the further research
tory, United States Steel Corporation, Monroeville, Pa. that is most needed to furnish pertinent data for
This report was prepared for the Subcommittee on Hydrogen Em-
brittlement of the Pressure Vessel Research Committee. pressure-vessel applications is outlined.
I
/
"'::Ea.
ity values for hydrogen in iron, as a function of I
I
/ 1200
"'
l-
the temperature and the partial pressure of hydro- / a + G
/
gen in the environment, are shown in the iron- I
/ SOO
/
hydrogen equilibrium diagram depicted in Fig. 1. 33 I
I
400
This diagram is based on measurements of the I
0.0012 0.003
absorption boundary for iron in equilibrium with WEIGHT PERCENTAGE HYDROGEN
a partial pressure of hydrogen of 1 atm. at tempera- I I I I I I I
0 5 10 15 20 25 30
tures above 750 0 F; the isobars for the other pres- CUBIC CENTIMETERS OF HYDROGEN PER 100 GRAMS OF IRON
sures of from 0.01 to 100 atm. represent calculated
values, based on the assumption that the solubil- Fig. I-Equilibrium diagram, iron·hydrogen system, by C. A.
Zapffe (Ref. 33)
ity is proportional to the square root of the par-
tial pressure of hydrogen in accordance with Sie-
vert's law:
S = K (PH')!;' (1) which are generally much higher than those which
Deviations from this behavior have been obser- would be predicted from an extrapolation of the
ved, * particularly at the lower temperatures, so results of solubility measurements at tempera-
that these isobars are not very accurate, but the tures above 750 0 F. This behavior is illustrated
significant influence of the pressure of the hydro- in Fig. 2, based on measurements by Hill and John-
gen environment on the solubility of hydrogen in son54 of the solubility of hydrogen at a partial
iron is evident from this diagram. pressure of hydrogen of 100 atm. in pure iron
This diagram shows that the solubility of hydro- specimens. The departure from the extrapolated
gen in liquid iron is much higher than in any of the values at temperatures below about 750 0 F is
solid phases, being above 25 parts per million by immediately evident. Furthermore, the discrep-
weight (1 ppm = 0.0001 %) at a pressure of 1 atm. ancy between the extrapolated value, which pre-
It is likewise apparent that the solid solubility in sumably represents the interstitial solubility, and
face-centered gamma iron is much higher than in the measured value, increases with decreasing
body-centered alpha iron, being above 4 ppm at temperature. This anomalous behavior has gen-
1 atm. pressure in gamma iron and below 3 ppm at eraJly been explained on the basis that, at the lower
1 atm. pressure in aJpha iron. The solubility in temperatures, the excess hydrogen, beyond that
alpha iron, furthermore, decreases very rapidly which is dissolved interstitially, is retained in sites
with decreasing temperature. At temperatures in the steel which are commonly referred to as
near room temperature, at which the deleterious "traps." The solubility value, or saturation con-
effects of hydrogen in steel are generally most pro- centration, as measured, includes both the inter-
nounced, the interstitial solubility is very low, stitially dissolved hydrogen and the "trapped"
being about 0.002 ppm at 1 atm. pressure. The hydrogen. *
difference between the interstitial solubility of The nature of the traps in which hydrogen is
hydrogen in gamma and that of alpha iron presum- retained has not, however, been fully established.
ably reflects primarily the difference in the size of Observations of the densities of dislocations and
the interstices of the two crystal structures. vacancies in steels have generally indicated that
they are several orders of magnitude smaller than
Saturation Hydrogen Concentration
the number of trapped hydrogen atoms,135 and it
At temperatures below about 750 0 F, and par- appears that such lattice imperfections could not
ticularly at temperatures below 400 0 F, measure- accommodate the amount of the "excess" hydrogen
ments of the solubility of hydrogen in iron and which is trapped. Hydrogen can also be trapped
steel indicate solubility values at a given pressure in voids, microcracks, microfissures, and larger
defects, where it may be present as chemisorbed
*In thermodynamic terms the solubility is proportional to the square root
of the fugacity, p;" but the more familiar term, pressure, is used throughout * The concept of hydrogen trapping was introduced by Darken and
this report, and the error involved is not large except at very high pressures.' Smith."
I 1 1
N
1 1 20 o
1.3 _ \ - - - GELLER AND SUN 731 0.3
(EXTRAPOLATED FROM I ATM I
- - HILL AND JOHNSON 54)
1.2 \ EXPERIMENTAL DATA FOR
1.0- \l
\ 100 ATM WITH 95 % CONFIDENCE
liMITS INDICATED ~ 100
-i~~'Oi
0.9
z
\
0.8- ~ 0.2
o
0.7f- '"
c
>-
::t:
.~
O.Gr-
0.5-
~ OA-
0.1
g
..J
0.3r-
\' "'J 2.0 ~
0.2r-
-0.3
\ ,
. "0.6
'\.
0.5
and on saturation hydrogen concentration of wire stock ('/2 in.
\
\,--EXTRAPOLATED VALUES" diameter green rod) immersed in N H 2S04 at 35 C
• \ FOR INTERSTITIAL "
-0.4r-
\ ' SOLUBILITY " " - OA
-0.5f-
.'\'
\\ '\. 0.3
-0.6f-
-0·71- , \ - 0.2
-0.8"__ ;7;-rr-+r*-,l,;--+-,....."l,~\~'*-+-~:---;l"'""""'i..,---,~
1.0 1.2 1,4 1.6 1.8 2.0 2.2 2A 2.6 2.8 3.0 3.2 3A 3.6 3.8 by extraction at 155 0 C (311 0 F), ranged from 2
TEMPERATURE. lOOO/T"K
cc/l00 g (1.8 ppm) to 30 cc/l00 g (27 ppm). The
Fig. 2-Solubility of hydrogen in alpha iron at 100 atmospheres measured values were 30 cc/l00 g for enameling
of hydrogen gas
steels, 20 cc/l00 g for rimmed steels, 15 cc/l00 g
for silicon steel, 8 cc/l00 g for bessemer steel, 5
cc/l00 g for aluminum-killed steels, and 2 cc/l00
g for AISI Type 430 stainless steel. These wide
variations presumably reflect predominantly dif-
hydrogen, or may recombine to form a molecular ferences in the number of available trapping sites.
compound. Hydrogen may also be absorbed in Cold working usually affects both the saturation
certain nonmetallic phases that are present in steel, concentration of hydrogen and the time required to
such as MnS, TiC, TiN, and ~ carbide. saturate steel with hydrogen. The effect on the
As indicated in Fig. 2, in pure iron at room tem- saturation concentration, as shown in Fig. 3, can
perature, most of the hydrogen is present as trapped be very large, with the saturation concentration
hydrogen. Factors such as cold working or increasing from less than 2 ppm in the absence of
the presence of nonmetallic inclusions, which tend cold work to more than 35 ppm after a cold reduc-
to increase the number of trapping sites, can thus tion of about 76%. Cold-worked steel has a
markedly increase the apparent solubility or the decreased density which results from the formation
saturation concentration of hydrogen. Although of voids at inclusion-steel interfaces during plastic
the effects of environmental factors, such as tem- flow. These voids are presumed to be the princi-
perature and pressure, on the saturation concen- pal source of the trapping sites for hydrogen in
tration will be similar to their effect on the inter- cold-worked steel, and they affect both the satu-
stitial solubility, the combined effects of these ration concentration and the time required for
factors and the factors which change the trapping saturation in a given environment because they
behavior can be very large, and can lead to very must be filled when the steel is saturated with
wide variations in the saturation concentration of hydrogen. Annealing of a cold-worked steel, how-
hydrogen in steels exposed to hydrogenizing envi- ever, reduces the saturation concentration of
roments. The magnitude of such variations can hydrogen without significantly changing the den-
be illustrated by the results of measurements by sity, although the saturation concentration is still
Hudson, et al.,71 of the saturation concentration of higher than that before cold working. Much of
hydrogen in various steels which had been charged the increased trapping associated with cold work-
by immersion in a 2N H 2S04 solution at 38 0 C ing thus cannot be completely explained as result-
(100 0 F). The hydrogen contents, as measured ing from void formation, since annealing only par-
I
'"
~" I,
methane formation has been observed 53 at tem-
peratures as low as 380° F (200° C). The forma-
tion of large amounts of methane at elevated tem-
peratures, commonly referred to as "hydrogen
I:
V attack," may lead to decarburization, grain-boun-
(I cc NTP per grom = 89.9 ppm H)
Diffusivity
tially heals the traps. However, nothing short of The movement of hydrogen in steel occurs by the
melting will restore the original low apparent solu- migration of atoms through the lattice, and thus,
bility of hydrogen of the cold-worked steel. The only the atomic form of hydrogen can diffuse.
significance of the voids as trapping sites is further Furthermore, the rates of diffusion of hydrogen in
confirmed by the observation 136 that neither the single crystals and in poly crystalline iron are
density nor the saturation concentration of very nearly identical, indicating that diffusion occurs
pure iron is affected by cold working. primarily through the lattice, without a preferen-
The effect of variations in the effective pressure tial diffusion along grain boundaries. The driving
of the hydrogen environment on the rate of absorp- force for the diffusion of hydrogen is an activity
tion and saturation concentration of hydrogen in a gradient which generally results from a gradient in
steel, in which the number of trapping sites is pre- the lattice hydrogen concentration, but can also
sumably constant, is illustrated in Fig. 4. The arise from a gradient in the hydrostatic component
variations in effective hydrogen pressure result of an elastic stress field. 94 Under such gradients,
from changes in the pH of the charging medium, hydrogen will diffuse from a region of high activ-
with a decreasing pH leading to an increasing ity to a region of low activity until the chemical
effective hydrogen pressure. This environmental potential of hydrogen is uniform throughout the
effect can obviously be very large, with the satura- steel. For example, when a concentration gra-
tion concentration of hydrogen increasing from dient exists in an unstressed body of uniform tem-
about 1 ppm to about 30 ppm when the pH of the perature, hydrogen atoms will diffuse from a region
charging medium decreases from 8 to 1.4. of higher interstitial concentration. This diffu-
Residual Hydrogen sion rate is thus related to the hydrogen concentra-
tions in the lattice, and the lattice diffusivity, D.
In addition to the hydrogen in interstitial solu- (D is defined as the amount of hydrogen per square
tion or contained in traps, hydrogen may also be centimeter which diffuses per second for a unit
present in steel in a comparatively immobile form, concentration gradient. The units of Dare cm 3 /
commonly referred to as residual hydrogen. This cm -second or cm 2 / second.) This diffusion will,
form of hydrogen cannot be readily extracted at in general, continue until the concentration gra-
temperatures below about 1470° F, presumably dient has been eliminated, with the limiting con-
because it is present as a compound which is stable dition being one in which the lattice hydrogen con-
below 1470° F. Johnson and Hill52 have suggested centration is equal to the equilibrium lattice solu-
that the compound may be methane. The rate of bility throughout. The lattice diffusivity, D, will
chemical fixing of residual hydrogen in steel has vary with temperature in the same general manner
been found to be maximum at about 1110° F and as the lattice solubility and can be expressed by the
4 Basic and Research Aspects
normal Arrhenius equation: iron and stainless steel by various investigators
D = Do e- Q
/
RT
(2) are ~hown in Fig. 6. The lattice diffusivity of
hydrogen in gamma iron is much lo~~r th~n in
A plot of log D versus l/Tabs for pure iron,
alpha iron, and is still lower in austemtIc staml~ss
based on measurements by Hill,60 is shown in Fig.
steel, presumably reflecting the relatively hIgh
5. The values of the measured diffusivity, D, are
alloy content of the latter material. .
in accord with eq (2). However, the activation
Concentration gradients which normally prOVIde
energies for diffusion, represented by the slopes of
the driving force for diffusion may be developed ~t
the two straight lines, are about 3000 cal/mol at
the surface of the steel exposed to hydrogen enVI-
temperatures above about 400 0 F, and about 8000
ronments or internally at local sites, as, for exam-
cal/mol at temperatures below 400 0 F, with the
ple, when hydrogen is released locally whe~. re-
diffusivities ranging from about 3 X 10- 3 to
tained austenite transforms to ferrite. In addItIOn,
3 X 10- 4 cm2/sec at the higher temperatures, and
stress gradients such as those produced by
notches, by oth~r sharp defects, or by bending
from about 10-5 to 10-7 cm2/sec at the lower tem-
peratures. The straight-line relationships indi- moments can also provide a driving force for
cate that the behavior in both temperature ranges localized' diffusion of hydrogen. This driving
is one of thermally activated interstitial diffusion. force results from the increased solubility of the
The low apparent diffusivity, D, values, measured hydrogen in the stretched lattice of the triaxially
at the low temperatures, and the much higher acti- stressed region which develops when such stress-
vation energy for diffusion at temperatures below
concentrating factors are present, with a resultant
about 400 0 F, are explained on the basis that the
flow of hydrogen atoms into this region of ~igher
interstitial diffusion is hindered at these lower solubility. The amount of hydrogen that diffuses
temperatures by the presence of traps which cap- into these regions of higher solubility will be pro-
ture and delay migrating hydrogen atoms.
portional to the gradient in the triaxial.c~mponent
Diffusivity values ranging from 10 -5 to 10- 9
of the stress in these regions. The dnvmg forces
cm2/sec have been reported for iron at room tem- provided by gradients in concentration or stress,
perature. The lower values are believed to have
however, act independently, and the concen~ra
presumably been influenced by surface trapping or
tion of hydrogen at local triaxially stressed reg~o~
trapping behaviors. The value of 10-5 cm 2/sec,
is a very important factor in the cha~a~tenstlC
which is in agreement with the value which would delayed-failure behaviors of steels contammg hy-
be predicted by an extrapolation from higher tem-
drogen.
peratures, is now generally accepted 94 as the most
probable value for the true diffusivity, D, in iron at
room temperature.
Similar Arrhenius plots depicting the results of
TEMPERATURE
diffusivity measurements on alpha iron, gamma 00 0 010 CD
888888
to CDV ... .... '"
00 0
0 OJ CD
0'"
0
2
"'-- - I
F
0000 000 0 0 C
0000
~=(7)f"-. "' ....... '"0
000 2
TEMPERATURE, F
3
1400
---, 800
i
~oo 300
i
200
i
IOu
i
50
i ~~ a IRON (GELLER AND SUN)73)
10-' 600
1
400
1
TEMPERATURE,
200
I
C
100
1
25 4
" \
"'~
" .... ,~ "-
\ '\
u I~ % Ni -C,-Mo STEEL
,,
\ ,2
" I x 10-5
"'.:?:. 5 , 0 (HOBSON)
E
u
a IRON (JOHNSON AND
10 10 HILL) 52)
,: I X 10-6
!:::
>
iii
"'6
:: ,\ \ \
\ 0
:> \ \ \
l::: I X 10-7 7 \ \ \
Ci \ \ \
\' "'-, ' .....
I X 10-8 8 \,
\
\
\
\
' .....
'0
\ \
AUS SS(l8%C,- \ \ /yIRON (GELLER AND
8.9%Ni)(GELLER \ \ SUN)73)
9 AND SUN)73) \ \
10-7
~ \
1 ~ ~ 1 ~ \ \
0.8 1.2 1.6 2.0 2.4 2.8 32 3.6 10 I , \
10' o I 3 4 5 6
TloKI RECIPROCAL TEMPERATUF1E, 1000/oK
Fig. 5-Diffusivity of hydrogen in iron (Ref. 60) Fig. 6-Diffusivity of hydrogen in iron and steel
a Room temperature.
30L-~ __~__L--l__~~
properties of steel is a decrease in ductility and true o 0.1 0.2 0.3 0.4 0.5 0.6
stress at fracture. This effect is manifested in ten- TRUE STRAIN
sion tests with smooth specimens as a decrease in LOGIO (Ao / A) = 2.303
reduction of area and elongation, in tension tests Fig. 7-True·stress-true-strain curves for a hydrogen-impreg-
with notched specimens as a decrease in notch nated 3Cr- Mo steel after standing in air for varying periods
tensile strength, and in bend tests as a decrease in (Ref. 80)
bend ductility. This effect of hydrogen, com-
monly referred to as hydrogen embrittlement, is
also generally associated with an increased propen-
sity for a cleavage or quasi-cleavage mode of frac-
ture. when they are put into service, or III services
The magnitude of these effects increases with which add hydrogen.
increasing hydrogen contents in the steel. How- The effects of high-temperature high-pressure
ever, it is also dependent on both material char- hydrogen on the properties and behaviors of steel
acteristics and external factors, with the principal will be described in a forthcoming section of this
material characteristics being crystal structure and report, where the damaging effects of hydrogen
strength level, and the principal external factors under these conditions of exposure, resulting prin-
being strain rate and temperature. Hydrogen has cipally from the methane reaction which leads to
little or no effect on the properties or behavior of decarburization and intergranular cracking, are
austenitic steels except at very high hydrogen con- discussed in detail.
tents, but all ferritic steels are affected in varying
Effects of Hydrogen on Ductility
degrees. The magnitude of the effects in ferritic
steels increases markedly with increasing strength Steels containing hydrogen characteristically
level. The effects of hydrogen may be observed fracture after the onset of necking but at lower
within the broad range of temperatures from normal strain values, as illustrated in Fig. 7 given
-300 to 400 0 F, but are usually most pronounced by Hobson and Sykes80 for smooth tensile bars of
at temperatures near room temperature and at an 0.26C - 3Cr - 0.4 Mo steel tempered to a strength
very slow strain rates. At very low temperatures, level of 51.8 tsi (116 ksi) and containing hydrogen
or at very fast strain rates, the properties and contents varying from 0.1 to 4.8 ppm. In this
behaviors of steels containing hydrogen do not dif- work the hydrogen content was decreased by aging
fer significantly from those of hydrogen-free steels. as indicated in Fig. 7. This plot shows that the
Under either of these extreme conditions the true stress at fracture decreases as the hydrogen
mechanism of hydrogen damage is inhibited concentration in the steel increases, but fracture
because hydrogen movement is greatly limited. occurs only after the apparent maximum stress
Because of the strain-rate effect, impact proper- has been reached, and necking is in progress.
ties are usually unaffected by the presence of hy- Thus, the presence of hydrogen does not signifi-
drogen in steel. On other hand, the magnitude of cantly affect the flow behavior, but fracture occurs
the effects of hydrogen generally increases mark- at decreasing levels of strain and true stress as the
edly with decreasing strain rate. In service, fail- hydrogen content increases.
ures of steels containing hydrogen are characteristi- In addition to the effect of the hydrogen content
cally delayed failures, which occur at relatively itself, the measured tensile properties are affected
low stress levels after the steel has been in service by the temperature and strain rate of the test, by
for relatively long periods of time. the stress concentration of the specimen, and by
Such delayed failures are aggravated by the the strength level of the steel and the amount of
presence of stress concentrators, such as defects and prior cold work which it has received. The effect
notches, which cause hydrogen to segregate at of strength level is most significant. In higher
local regions in members that are subject to ten- strength steels comparatively small amounts of
sile of bending stresses. Delayed failures can hydrogen lead to large changes in properties, but
occur either in steels which contain hydrogen, in steels with lower strength the influence of hydro-
..
c:
..
"
Q.
20 STRAIN RATE, in. / in.lmin
TEMPERATURE. F
<1
W
0::
<1
~ 80~---,----.---_.----~--~
-1.2
z
o DUCTILITY (IN aolat)
I-
g 60
a
w
0::
40
because of the increased strain localization. When subject to a sustained load, a steel con-
The effect of hydrogen content on notch strength taining hydrogen may fail at a stress level which is
is illustrated in Fig. 10, based on tests recently much below its tensile strength as measured in a
conducted by the British Welding Research Asso- sh ort-time test. This behavior is variously termed
ciation (BWRA)'·l on thermally charged notched delayed failure, hydrogen-stress failure , and de-
tensile specimens of 0.30C - Ni - Cr - Mo marten- layed, low-stress brittle failure, and is the most
sitic steel, which had been aged for various times to common type of failure encontered in service appli-
provide a range of hydrogen concentratio ns up to cations of steels containing hydrogen or exposed to
about 3.7 ml/ 100 g (3.3 ppm) . A function ( uo- hydrogen environments. It represents essen-
uu ) / Uu of the notch tensile strength (where Uo tially a manifestation of the decreased fradure
and UH are room-temperature notch strengths for stress or strain at fracture in the presence of hy-
the hydrogen-free and the hydrogen-ch arged speci- drogen, as exemplified by the results of the notch
mens, respectively) is plotted against the hydrogen tension tests just described, with the fracture ini-
content. The value of this function, which in- tiating in regions of highly localized stress in which
creases with decreasing values of UH, increases hydrogen is concentrated as a result of the aug-
linearly with increasing hydrogen content. T he mented diffusion of hydrogen into triaxially
behavior is similar to and reflects the straight- stressed sites.
line decrease in tensile ductility which is observed T he term "brittle," which is SOlnetunes used in
connection with this effect of hydrogen, is indica-
tive of the increase in the cleavage or quasi-cleav-
age mode of failure that is generally observed on
12~--~--T---'------'
the fracture surfaces of such failures. For exam-
ple, the fracture surfaces of notched-bar, charged
• • tensile samples generally have an outer ri.ng of
'0 • smooth fracture , the dark region shown in Fig. 11.
,
•
l>l'b
S' I z 6
:oj O
0~_1----L_-'--_ JJO 000 PS I 32 4 000 PS I
fRACTURE STIIESS
] HOURS 12 HOURS
o 2:3 4 BAKING TIME
175 , 000P51 200 . 000 P 51
APPliED STIlESS
TOTAL HYDROGEN CONCENTRATION (HI, mI/IOO;
Fig. lO-Plot of (UII - UH)/UH against total hydrogen content fa Fig. ll-Fracture surfaces of precharged, sharply notched ten·
steel samples tested at room temperature (Ref. 93) sile samples (Ref. 19)
a
0;/~1l90
~ ~
tion and crack propagation.
w
...J
a.
CRACK INITIATION
CURVE DEFINING
V0'O't-'V
(/, Q
/
Q
'V
FAILURE CURVE
~~-t-.:..J'''::-
a. Incubation Period for Crack Initiation
<l
I NCUBATION PER I 00 _ _ _ _ _ _--'----i
At the intermediate stress leveis an incubation
period, which in Fig. 12 is bounded by the crack-
REGION OF NO DAMAGE initiation curve, is presumably required to initiate
a crack because the average hydrogen concentra-
tion in the steel is lower than the hydrogen concen-
tration required for crack initiation at potential
LOG TIME - - - -
cracking sites. During the incubation period the
Fig. 12-Schematic drawing of characteristic failure behavior of distribution of hydrogen in the steel is changing.
notched tension specimens containing hydrogen
Hydrogen is diffusing ahead of the notch to local
highly stressed regions at which cracks initiate
after a critical hydrogen concentration has accu-
mulated. The kinetics of the crack-initiation pro-
Both transgranular and inter granular fractures cess are therefore dependent on the temperature
have been reported for failures occurring by the and the stress gradient as they relate to the diffu-
hydrogen-stress mechanism. Hydrogen-stress sion and concentration of hydrogen in the steel.
fractures are not readily distinguishable, with the The applied-stress level exerts only a minor influ-
unaided eye, from other low-stress failures, such ence on the incubation period, as is indicated in
as those from stress-corrosion cracking, since the Fig. 12. However, the effect of temperature on
fracture surfaces may appear quite similar. How- the incubation period is much greater, as shown in
ever, cracking caused by a hydrogen-stress mecha- an Arrhenius plot (Fig. 13) for similar notched,
nism generally does not exhibit the branching of precharged tension specimens that were tested at
the type that is often characteristic of stress-corro- an applied stress level of 150,000 psi at tempera-
sion cracking. In stress-corrosion cracking, sev- tures between room temperature and -50 F. 0
eral secondary cracks are usually produced in addi- The activation energy for crack initiation, as cal-
tion to the major crack, and small branching cracks culated from these data, is 7610 cal/mole, which is
are seen emanating from these larger cracks.
Fractographic techniques have, furthermore, been
reported 134, 139 for distinguishing stress-corrosion
from hydrogen-stress fractures.
Because of its importance, this delayed-failure 5,0',---,----,---.--TT--,---------,
behavior of precharged steels has been extensively
investigated in laboratory studies by many investi-
gators including Troiano and his associates. 22 In 2,0
the Troiano experiments, quenched and tempered
4340 steel specimens of various notch geometries
were cathodically charged, cadmium-plated, and
then baked at 300 F to give "homogeneously"
0
~,~.--
Q.
200,000
(J)
(J)
\" 800F....-
D~"7_
ILl
cr
I-
(J)
100,000 o "7500 F
t:r-
O~~ __~____~~~~~~~~____~
0.01 1000
TIME TO RUPTURE, hours
300pOOr------.------,------,,------r------,
.;;;
Q.
u;
(J) ~~~~-------.,-----------.--
ILl
cr
l-
(/) 100,000
Fig. 14-Static fatigue curves for various hydrogen concentra· Fig. 15-Stress vs time-to-rupture for cadmium-plated 4340
tions corresponding to different baking times at 300 F. Sharply steel austenitized at 1525 F, oil-quenched, and tempered as
notched specimens of 4340 steel at 230,000 psi strength level indicated. Stress concentration; Fig. a, K = 10, Fig. b, K =
(Ref. 9) 1 (Ref. 86)
~o
::> f-
~ ~ 7
restraint is shown to decrease markedly as the
'"'"
z .E
--'
u
hydrogen in the welding arc increases from zero to
..
u;
!oJ '"
U
6%. This corresponds to room-temperature dif-
fusible-hydrogen concentrations in the cold weld
U'"
'!;
_ _ 0 _____ _ deposit ranging from near zero to about 0.2 ppm
(0.22 cc hydrogen per 100 g of weld deposit).
Thus, even relatively low levels of hydrogen may
cause delayed cracking in susceptible steels such as
those with high HAZ hardness. The residual
PERCENT HYDROGEN IN WELDING ATMOSPHERE
stresses in weldments are accentuated by internal
microdefects in the steel and by external notches,
Fig. 17-Effect of hydrogen concentration in the welding atmo.
sphere on the cracking sensitivity of HY·80 steel (low chemistry)
such as discontinuities in the contour of welded
welded with an alloy electrode (Ref. 85) joints. Thus, local residual stresses, being at the
level of the yield strength in local sites, can be very
high in high-strength steels. However, much
higher hydrogen concentrations, as high as 10 ppm
or even higher, may be tolerable in some lower
stresses of welding. The fracture surfaces of weld- strength weldments, since ferrous materials vary
metal tension samples and of cold cracks will, in widely in their susceptibilities to cold cracking.
general, tend to exhibit a cleavage or quasi-cleav- The cold-cracking susceptibility of steel is
age mode of fracture, and they sometimes contain strongly affected by composition, since composi-
small bright areas, commonly referred to as "fish- tion affects both strength level and transformation
eyes," which are regions of cleavage fracture sur- temperature. Because weldments generally cool
rounding inclusions. Their presence indicates very rapidly, the hardness levels of both the weld
that excessive hydrogen was contained in the steel metal and the heat-affected zone will approach the
weld metal. As might be expected, impact speci- maximum attainable hardness level for a given
mens do not normally contain fisheyes because the composition. The microstructures of such rapidly
hydrogen-stress mechanism is not operable at the cooled areas are often those of self-tempered or
fast strain rates that are associated with impact untempered martensite and bainite, and these
testing. microstructures are most susceptible to hydrogen-
Cracking caused by hydrogen in weldments induced, delayed, brittle failure. However, the
usually occurs at or near room temperature and influence of composition on transformation tem-
generally after a delay of 24 hrs of less. 85 This perature is also important, since hydrogen out-
cracking phenomenon is therefore called delayed gassing from the weldment will decrease as the
cracking or hydrogen-induced cold cracking. temperature of transformation decreases.
This behavior is consistent with the overall The susceptibility to cold cracking can be empiri-
experience with delayed cracking in steels contain- cally related to the composition of steel. This
ing hydrogen. Hydrogen may lead to cold crack- relationship is often expressed in terms of a "carbon
ing in either the weld metal or the heat-affected equivalent" of the various alloying elements
zone (HAZ). It can occur in various orientations because carbon is particularly effective in promot-
and locations, such as longitudinal and trans-verse ing cold cracking in weldments. However, there
cracking, and root and underbead cracking. Its is no general agreement on the most appropriate
occurrence in weldments is governed by hydrogen factors for each of the alloying elements. One of
concentration, stress level, strength, composition, the commonly used carbon-equivalent formulas
cooling rate, transformation temperature, and is given below: 85
bead size.
The susceptibility to cold cracking increases, Carbon 01 C + % Mn + % Cr +
with a decrease in the threshold stress for cracking, equivalent /0 4 10
as the hydrogen concentration of the weld deposit
increases. This effect is shown in Fig. 17 taken % Ni % Cu %Mo %V
(11)
20 +-.w-
----
from work by Interrante and Stout85 for welded 50 10
u.
the susceptibility to flaking greatly increases, and 0
1
below which very little flaking occurs. This '"
.... I
I
Z I
100 I
ber of flakes found by macroetching and visually Ni-Mo-V I
I
l!.
sections taken from 12 by 12 by 24 in. test speci- I
./ P
I 6 - 1 0 8 0 STEEL
mens) from each of two steels that were cast at I
I
posed to hydrogen-charging environments may re- Fig. 19-Delayed-fai/ure behavior during cathodic charging of an
sult at low stress levels whenever the environment is SAE 4340 steel with an ultimate strength of 230,000 psi (Ref. 129).
so severe that it leads to surface and internal fissur- Battelle Charging Condition C: Electrolyte: 1/2% sodium hy-
droxide in water; current density: 125 mA/in.2; poisons: none.
ing of the steel. The effective pressure of certain Battelle Charging Condition D: Electrolyte: 1/2% sodium hy-
aqueous environments can be very high, sometimes droxide in water; current density: 500 mA/in.2; poisons: none.
reaching 10 4 to 106 atm., and these pressures may Battelle Charging Condition A: Electrolyte: 4% (by weight)
aqueous solution of sulfuric acid; current density: SmA/in.2;
be beyond the bursting strength of the steel. I t is poisons 5 drops per liter of poison containing 2 grams of phos-
believed that when steel is exposed to such envi- phorus in 40 ml of CS 2
200~-'------'-------'------'-------'---
CONTINUOUSLY CHARGED
" 150,OOO-PSI
x 125.000- PSI
APPLIED
APPLIED
STRESS
STRESS
•\230.000 ps; USING BATTELLE
CONDITION A (GIVEN IN
_ 100.000- PSI APPLIED STRESS FIGURE 19)
o 75,000 - PSI APPLIED STRESS .~ 150
"
\'='"
+ 50.000 - PSI APPLIED STRESS
o
o
o
(f)
W (f)
a: UJ
"~..
:::> a: 100
il: 10 f-
(f) \\:2,000 pSI 0
:::>
a: o
\~~SI
UJ
o ::i
..
f-
Cl.
PSI
LU Cl.
::;
i= 50
~
.-.~
~.
,,-------- --
-,150,000 0
!.".-.L----:\::----*.------;rfM.---I01~--5500"OOO
EQUIVALENT PERMEATION RATE FOR VARIOUS CHARGING
TIME TO RUPTURE, m;nutes
CONDITIONS, 10-3 ;n,3/in,2/m;n
Fig.20-Rupture time for SAE 4340 steel at the 190,000 psi Fig. 21-Delayed·failure behavior during cathodic charging of
strength level as' a function of cathodic charging condition unnotched tension samples of SAE 4340 steel heat-treated to
(Ref. 130) • • • •...
, t various strength levels (Ref. 36)
li-o
..- - - - - - 8"------.1
1
~
-
ELECTROLYTE
i
g
Q
ui
<n
~ 80
in
160
12
\ x
" "
o
~~--·~------~____~x_~
6 6 X 6-~X-
~RO~' ~
I HYCAR RUBBER
40
WITHOUT FAILURE
-S~ ~
=-IN=
r
~ ;8
~ ~g~CI~6
GASKET
BENDING
I
TIME TO RUPTURE, minutes
I I I I
0.01 0.1 0.5124
THICKNESS TIME TO RUPTURE, days
Fig. 22-Schematic arrangement of the electrolytic cell and Fig. 23-Stress-rupture characteristics of notched specimens of
specimen used in the bend tests of Figure 23 (Ref. 129). Charg- SAE 4340 steel heat-treated to an ultimate tensile strength of
ing Condition: Electrolyte: 4% sulfuric acid; current density: 230,000 psi and loaded in bending while being charged cathod-
33 mA/in.2; poison: presumably the same as that used for ically with hydrogen in a poisoned sulfuric acid electrolyte (Ref.
Battelle Charging Condition A given in Figure 19 129)
necessary condition for low -stress failure of steel cussed86 for electroplated unnotched samples of
in severe environments. In this study the envi- steel containing 2.5 ppm hydrogen (Fig. 15b), one
ronment was permitted to contact only the com- may conclude that the hydrogen concentrations in
pressively loaded side of a notched bending speci- the Battelle samples are much higher at the loca-
men, as shown in Fig. 22, while the tension side was tion where failures are presumed to have initiated,
exposed to air. This technique inhibits the forma- because failures occur at much lower stress levels.
tion of surface micro-fissures, because the surface However, the charging conditions found in most
exposed to the charging medium is loaded in com- service environments are normally not as severe as
pression. Permeation of hydrogen is not signifi- those of Charging Condition A. It follows that
cantly affected, and it is believed that the effective the hydrogen content of steel in most service envi-
hydrogen pressures that develop inside the speci- ronments is also generally much lower.
men are very high at steady state. The test speci-
mens were an SAE 4340 steel heat-treated to an
Hydrogen Sulfide Environments
ultimate tensile strength of 230,000 psi. Failures
at stress levels considerably lower than the tensile Exposure to hydrogen sulfide environments,
strength of the steel were observed, as indicated in commonly encountered in oil-field and refinery
Fig. 23, and none of these failures originated at the service, has frequently led to low-stress failures.
compressively loaded cathodically charged surface. Extensive laboratory studies have shown that these
It can be argued that in this instance, the action of failures are delayed failures which will occur on
the electrolyte at the specimen surface does not exposure to most aqueous solutions containing
influence the behavior by producing surface micro- hydrogen sulfide, because the effective hydrogen
fissures which may then propagate at low applied- pressure is markedly increased when hydrogen
stress levels, leading eventually to failure of the sulfide is present in most aqueous solutions. This
sample. However, if it is assumed that under includes aqueous solutions at any pH level from
these conditions the effective hydrogen pressure at zero to 9, as indicated by the data given in Fig. 24,
this cathodic surface is very high (it seems reason- after Hudgins and others. 40 In this illustration
able to assume that the effective pressure of this the severity of solutions with and without sulfides
solution is very high, perhaps in the range 10 4 to and at various pH levels is compared in tests with
105 atm.), then it follows that at steady state very notches "C" ring specimens of steel at a hardness
high internal pressures develop in the tensile region level of 33 Rockwell C. The samples were tested
beyond the neutral axis of this specimen, even under a sustained load at 115 % of the yield defor-
though the pressure approaches zero at the surface mation. The sulfide solutions are much more
of maximum tensile stress where the sample is severe, as the failure times in the sulfide solutions
exposed to the atmosphere. These authors right- are much shorter at aU pH levels. This is in agree-
fully concluded that low-stress failure in this in- ment with the findings of others, including those of
stance resulted from the combined effects of a low the Battelle investigators discussed earlier, who
uniaxial tensile stress (neglecting any effect from found the poisoned Charging Condition A to be
the notch) and a high hydrogen content generated severe compared with sulfide-free environments.
by the severe charging conditions. By comparing Accordingly, the acceptable strength level for
this charging condition with those previously dis- steels for service in sulfide environments will gener-
9 35
a: 30
u
7
en
6
...enz 25
;. 5 c
a:
~
4 J:
20
3
ally be much lower than that for steels for appli- where 0' is the applied tensile stress in kgjmm2
cation in sulfide-free environments. (1 kgjmm2 = 1.422 ksi) and HV10 is the Vickers
hardness number as measured with a 10-kg load.
Material Factors Affecting Behavior If this product of stress and hardness number is
less than 21,000 kgjmm2 (29,900 ksi) , infinite life is
Both laboratory tests and service performance predicted. These authors report that this equation
have indicated that, as with precharged steels, the was found to apply to all steels tested (carbon and
principal material factor affecting the behavior of stainless chromium steels), regardless of whether
steel exposed to environments containing hydrogen the hardness level was affected by quenching,
sulfide is the strength level. However, at the high tempering, isothermal transformation, or cold roll-
hydrogen contents resulting from the high effective ing. However, it should be noted that the behav-
pressure of these environments, even relatively low- ior of steel in H 2S environments has been shown by
strength steels may be significantly affected. For others to be sensitive to composition, micro-
steels for petrochemical service, strength level is structure, and processing history, and that very
commonly expressed in terms of hardness, which is small amounts of cold working may be deleterious.
more convenient to measure. Approximate rela- At a given strength level, tempered structures are
tionships between hardness and failure in hydrogen generally better than untempered structures, and
sulfide environments have been given by Hudgins structures of tempered maitensites and bainites
and others. 40 Harder steels fail in shorter time, as generally appear to exhibit better performance
indicated in Fig. 25 for samples at various hardness than ferritic or pearlitic structures. However, the
levels, tested in 5% NaCI solutions containing var- effects of heat treatment and microstructure are
ious concentrations of hydrogen sulfide. The hard- not entirely clear because the data are sparse and
ness level also governs the maximum stress level often confused by other variables, particularly by
that can be sustained by the steel in a sulfide envi- strength level.
ronment. This is illustrated in Fig. 26 for carbon It has been generally recognized that steels with
steels tested in a 5 % N aCI solution containing a hardness level of 22 to 25 Rockwell C or above
3000 ppm H 2S. Steels at lower hardness levels may be susceptible to hydrogen-stress failure in
sustain a yield-point stress level without failing, sulfide environments. Therefore, it is essential
whereas steels at higher hardness levels sustain that, for welded structures used in sulfide service,
only subyield stress levels. the hardness in the heat-affected zone (HAZ) be
These data indicate that the stress level that considered, in addition to that of the unwelded
may be sustained by steel in a given environment steel. Because of the rapid cooling rate in the
is related in a systematic manner to the hardness of heat-affected zone of weldments, the maximum
the steel. This interdependence of stress level and hardness in the as-welded heat-affected zone will
hardness has been expressed quantitatively by generally be much higher than the hardness of the
Naumann and Carius 126 , 143 for steels tested in an unaffected steel. Therefore, when the heat-affected
aqueous solution saturated (at 1 atm.) with pure zone hardness of the steels used in welded struc-
hydrogen sulfide. These authors give the ex- tures intended for hydrogen sulfide service is
pression unacceptably high, a stress-relief heat treatment
0'·HV10 constant ~ 21,000 kgjmm2 (12) must be used to reduce the hardness to acceptable
.\'0,
- ~20~
z
UJ
III
"-
o
~
\'
", "" , III
l5 ~. 0 ___ ----_@_
®
ffi ~ A. -
0:
UJ
III
~
151-
'-''0,
- g)
:::>
15-
.-.--yo
'-... ----0
--0-
:::> z
z
~ lO- -
~ leI- - t: KEY:
KEY:
t: SS 430
..J SS 420
..J
;:: III 7,000 PSI ~:::> ® 7,500 PSI 046,000 PSI
~
5- o 15,000 PSI d'52,ooo PSI -
• ~ 15,000 PSI
"
5 5'- " • 36,000 PS I -
o 36,000
40,000
PSI
PSI
.60,000 PSI
EXPOSURE TIME IN DAYS - SQUARE ROOT SCALE Fig. 28-Ductility of AISI Type 420 martensitic stianless steel
tested in air after exposure to hydrogen gas at room tempera-
Fig. 27-Ductility of AISI Type 430 ferritic stainless steel tested
ture (Ref. 28)
in air after exposure to hydrogen gas at room temperature
(Ref. 28)
pressure showed that the unnotched strength is not exposure to high-pressure hydrogen. For ex-
significantly affected, but the notched-bar ductility ample, tests of two steels of about 230,000-psi
and strength may be greatly impaired. 29 - 32 , 132 tensile strength 30 -32 showed that the strength of
Figs. 29 and 30 show results for Armco iron and notched specimens tested in hydrogen is below the
two carbon steels loaded immediately after hy- strength of unnotched specimens tested in air.
drogen at 2200 psi was introduced into the testing Notched specimens tested under high-pressure
chamber. As shown in Fig. 29, the reduction-of- hydrogen normally fail at low stress levels with-
area values are significantly reduced to about out a significant time delay, and the incubation
half of the values obtained for specimens tested period, characteristic of the behavior of precharged
in air. Figure 30 compares the strengths of these specimens, is normally either absent or very short.
steels in tests conducted with notched and One example of a delayed failure in high-pressure
unnotched specimens in hydrogen. The strength, hydrogen has, however, been reported by Hof-
as measured by the smooth-bar specimens, is gen- mann and Rauls. 32 In this instance, notched
erally unaffected by exposure to hydrogen at 2200 specimens of a concrete-reinforcing steel at a
psi, but the strength of the notched-bar specimens strength level of 233,000 psi tested under 100
is lower in hydrogen than in air. However, higher atm. of hydrogen failed at a stress level of 215,000
strength steels are more markedly impaired by psi; when smooth-bar specimens were tested under
similar conditions they failed at 200,000 psi. The
probability of delayed failures in steels exposed to
high-pressure hydrogen thus appears to be quite
remote.
1...J·:o[
w
;~1
~------~----~--~-----~I-----~I~-~
OJ
~ 80~--- r--- 11141JOO PSIL
80,--------------------------------------, , _ , IIOO,OOo!,~
0-
CARBON CONTENT IN %
B NOTCHED SPECIMEN
f/I
ILl
HYDROGEN, AT 150 ATM
RECALCULATED TO I ATM
Fig. 29-Ductility of Armco iron, CK 22 N (0.22% C - normalized)
steel, and C 45 N (0.45% C - normalized) steel tested in air at Fig. 30-Tension tests on notched and unnotched specimens of
1 atmosphere and under high·purity hydrogen at 150 atmo- Armco iron, CK 22 N (0.22% C - normalized) steel, and C 45 N
spheres. Deformation rate, 3% per minute (Ref. 32) (0.45% C - normalized) steel in air and iri hydrogen (Ref. 32)
!c::[ ~OQO~~o
]
Z
9
~O
20
o
z ~
o 0
000 '"
Q ; o ~ 15'~====~~~~==~========~==~======~
70r
~60~
t; <>. 20
§~ 0 I I I I I I
~ 50~ ~ ~ .:::--~
II II
100%~10~~1~~1~0-+1~-10~-2~~104-3~-1~~~4~-10~-~5~
..
AIR OXYGEN CONTENT IN HYDROGEN IN VOL. %
~ 40'___ """""------Ii 0-8_ ARMCO-IRON
o : _ _, ~~
Fig. 31-Effect of trace impurities in 100 atm (partial pressure) i= 30 -,___ _ ::-..,.... CK 22 N
.~.
hydrogen on the elongation and reduction of area of CK 22 N 15u • • ...-C45 N
(0.22% C - normalized) steel (Ref. 32) '" 20·~__~~~~~~~~~~wu~~~~~~~
~ I 5 10 20 50 100 200 500
HYDROGEN PRESSURE IN ATM
SUMMARY. The first section of this report discussed the decrease the rate of surface decarburization,
general effects of hydrogen in various forms, but largely
excluded the effects at elevated temperature. The addi- since the carbon in solution is supplied continually
tion of temperature aggravates the effects of high pressure from the carbides and the more stable carbides
hydrogen. This section of the report will discuss the effects give lower dissolved carbon content.
of hydrogen combined with temperature. The most sensi-
tive parameters are the partial pressure of hydrogen, the Actual values of pressure-temperature com-
temperature, and the material chemistry. While some of binations at which surface decarburization be-
the other effects that were of great significance at low tem- comes pronounced have not been extensively
perature, such as the stress and cold work, are still signifi-
cant, the effects of high temperature are less sensitive to these studied, but the limits defined by Naumann 3
or cannot be summarized so neatly. probably give the most accurate appraisal. Fig-
The areas to which this section applies are those in which ure 1, a graphical presentation of operating
hydrogen or hydrogen containing fluids are handled at high
temperatures (above 430 0 F) and high hydrogen pressures limits for steel in hydrogen service, includes data
(up to 13,000 psia). Under these conditions carbon steel points determined by Naumann. The mode of
will be unsatisfactory as a constructional material because damage is indicated by the shading. In addition
of both chemical and physical changes that occur.
Mechanisms for producing the chemical changes are dis- to his references, based on 100 hr tests, several
cussed together with alloying lequirements to prevent the additional long time references from industrial
damage. Additional sections are devoted to incubation service are included. These show a slightly lower
periods before chemical changes occur and to effects of hard-
ness, cold work and stress at high temperature. Methods level of susceptibility to decarburization than the
of preventing high temperature damage to pressure vessels 100 hr tests. Limits for surface decarburization
by specialized design or suitable alloying are also discussed. of Cr-Mo steels, as shown on Fig. 1, have been
extended only to 6000 psia. At pressures above
Surface Decarburization
this figure, the Cr-Mo steels must be further
When carbon steel is exposed to hydrogen at fortified with other carbide stabilizers such as
high temperature (above 1050° F and at pressures vanadium or tungsten to make them resistant to
below 200 psia) surface decarburization occurs surface decarburization.
with a resultant loss in strength and an increase
in ductility; in this respect being similar to steels Internal Decarburization (Hydrogen Attack)
decarburized by exposure to other gases, such as When carbon steel is exposed to hydrogen at
air, oxygen or carbon dioxide. A number of temperatures above ca. 430° F and at pressures
theories have been proposed to explain this above 200 psia, it becomes internally damaged;
phenomenon 1 ,2 but the presently accepted view it loses tensile strength and ductility, and some-
is based on a continuous migration of carbon to times cracks and blisters. The deterioration of
the surface, at which place it is removed as gaseous mechanical properties is caused by atomic hydro-
compounds of carbon~CH4 or CO (when carbon gen permeating the steel and reacting with other
dioxide is present). Moisture obviously hastens elements to form other gases. 4 For example,
the reaction. Carbon in solution must diffuse to it may react with carbon in solution to form
the surface, and the rate control ing mechanism methane. The methane thus formed cannot
is apparently diffusion of carbon. The addition diffuse out of the steel and accumulates princi-
of carbide stabilizing elements to the steel can pally at grain boundaries. High local stresses
The late G. A. Nelson was a metallurgical consultant in Berkeley, Calif.
eventually develop and become of such magnitude
and formerly with Shell Development Company, Emeryville, Calif. that the metal cracks or blisters. The formation
This report was prepared for the Subcommittee on Hydrogen Em-
brittIement of the Pressure Vessel Research COInmittee. of methane gas by exposure of steel to hot hydrogen
•**
COX. JOUR 'RON AND STEEL INST, SEPTEMBER, 1918
.00 I II
Hydroglin Attock.
OltCorbU"'Ol,on 181 )7 '(
.... •
.I>:l liX
~
:<Y
SARGENT AND MIDDLEHAM fR. CHEM ENG. CONGo LONDON, 1936
~~
1400 I ± I , t--+---t-l~~: ~~~ :~~~~~Ts~i:~~S~~~c~~~:~:EE ~UR"fY 1957. IN) t~~·~,.7c~~~.~6::'v~'o' sQh!~~~'ncQ."cc":'~~nbn?09.dJ:~~C,
22. DONAVAN AND GARDNER. TRANS AS~E, VOL 75. (1953). (P) Aile' lO 1ea'$ ser.'" Q fo'glng conTo'~'n9 D.lO"loC,
r--t t t i j l . AME~IC"'N Oil COMPANY, I9bO.. OHC'. 0"3"1oN, was ~naffecle(J.
1Fi-
24. (800.) .
::t::
~.
~
1200 I ··~r!.._ l··~*·-r··
-..
~ ~ --~,_~ t-=i
_ 1---.-
---
I Lb/in2 obs
M. HAS 4 TIMES THE RESISUICE OfCrTO HZlTTAtl
~~, ~t ECG~IV~~E:JT T~.l:dsfO:[f~S frM~~
-..
0.1 \.
( 6.0 Cr-0.5 Mo STEEL
1100 600
.
i
~
>:i
~
~IOOO
' -
1.'
.....J
.'.2!~ ¢I I I. T LL
I --"'-· .. ___1
I I I
I I I
TIT T T
r-....-"t··-~- __
---------- ~·+0.2:5\V
13+0.I~vt I ~
~
~
;= 10 "" ll"- 1 I 1.25 Cr-0.5 I
ci g 5 T~ ~'" I 0.5 Mo STEEL ' Mo STEEL 2.0 Cr-0.5 Mo STEEL 15 LJ...) 500 g
~ ~ 900 .ELOED 1 .... "'-- \7 ~
;:;: ~ OR ~ I -- I)Q ... 10 ~
R. mT lOT fElDEO 'J SII ..... "-' \ L " .... "
i"" _
-j).!'
...
I
26
I
DETAIL
FOR
1:;Fi~
--
......
~
)
.........
400
~ ELEIElTS\
I
_I ""-r-.- _. . . . ~
'" I
IJ 22
ci [7,,~
IL
0.5 Mo STEEL
"" ~ "l:J1
10 '--:"::...
.. 1.1, 05 1'0
CARBON STEEL 1\ )1
~I)p
I i I I I I I I I I I I I I I I I I I I I
500 1000 1500 2000
rI I I I I 1I I
2500
,~:Jo,j I 1'111 t" I to
3000 5000 7000 9000 11000 13000
IfI'OROGEN PARTIAL PRESSURE, L b/in' obs.
:~.;
HYDROGEN PARTIAL PRESSURE. Kg/cm 2 abs.
o 50 100 150
1000'-IrT-"~~~'-~Ir~--~--~----------~--'-------------'---+---------'
3000) 30E) 100)
A) AMERICAN OIL COMPANY (PRIVATE CONNUNICATlON), 1960
< 100.)
• B) CIUFFREDA AND ROWLAND PROC. API MID-YEAR IlEETING, NAY 1957
C) C. A. ZAPFFE, 'SOlLER ENBRITTLENENT: TRANS. ASME, FEBRUARY 1944 500
280E) 65 El 25El 0) R. E. ALLEN, ET AL, API, PREPRINT, MAY 1961
900 1--++--+--+-......::......----1~-----++ E) L. C. WEI NER, 'CORROSION', VOL 17, PP. 109-115, 1961
F) HUR, DEICHLER, AND WORRELL, OIL AND GAS JOURNAL, OCTOBER 29, 1960
G) F. K. NAUNANN, TECH. NITT. KRUPP. VOL I, NO. 12, 1938
a.> 10001
H) HASAGAYlA, TETSU TO HAGANE, VOL 46, 1960
J) T. C. EVANS, MECH. ENG., IIAY 1948
800 1--++-+-+-......-+--~-----1~+ K) AIR PRODUCTS, INC., (PRIVATE CONMUNICATlON), 1960
500
__~~~~__-=1f::::~:=~:;~1r~~::~~~:;~~I'~OO~0:~HO~U~RS~::::~~~<5~5~0~OG~I~(I~25~0~OP~S~IA~)~
t NO ATTACK
10,000 HOURS
Fig. 3-Time for incipient attack of carbon steel in hydrogen service (Note-The American Petroleum Institute, as one of its
continuing projects, will bring these curves up to date. It may be desirable to refer to current literature for best data)
< 100Cl4200PSIA'
~
::;
-
-<
--- - - --
= 800
~
-
'"
~ ,.. 200 HOURS
Al
"'l
, .'
•
.,
~-
,. ..
.",....
, ~~
, / ~/-
:--"--""'\.; .
• , .... ,-r:a -
••
}
" . •
•
~
~
£ .,. - --
Fig. 6-Characteristic shape of fissures in 5% cold worked SAE Fig. 7-Characteristic sha pe of fissure s in 39% cold worked SAE
1020 steel after 245 hours of exposure to 900 psi hydrogen pres- 1020 steel after 245 hours of exposure to 900 psi hydrogen pres·
sure at 10000 F (800x) sure at 1000' F (800x)
Effect of Hardness
100 Steels shown on Fig. 1 are considered to be in
98
the stress relieved condition. Heat treatments to
stabilize carbides increase the strength and en-
96
- - - - - - - - -_ _ 700 F
hance hydrogen resistance. Usual treatments are
94
-
------80OF
_ _ _ 1000F 39% STRAIN
normalizing followed by tempering, although
92 quenching followed by tempering has also been
performed. Caution should be exercised with
200 400 600 800 1000 1200 heat treatments to avoid excessiveJy high hard-
TIME, HOURS
ness. Long time service has been attained with
Fig. 8-Effect of temperatures and cold work on specific gravity low alloy steels, heat treated to a maximum hard-
of SAE 1020 steel exposed to hydrogen at 900 psi
ness of Rockwell C-23. However, experiences
have shown that brittle fractures can occur when
harder steels (above Rockwell C-30) are subjected
cold work after exposure to 900 psi hydrogen to high stresses, either residual or applied, while in
pressure at 700, 900, and 1000 F. For the an- 0
contact with high-temperature, high-pressure
nealed set of samples (0 % strain) the curves show hydrogen. Instances of brittle failures of equip-
incubation periods, followed by a decrease in ment while operating at high-temperatures and
specific gravity and a final diminishing rate of pressures are shown on Table 1. Cases 1, 2,
loss. The decrease in specific gravity indicates and 4 are border line for alloy content. However,
the presence of internal fissures. even though equipment had shown a slight surface
Steel with 5 % strain showed different character- effect of 0.02 in., when the hardness was below
istics than the annealed samples. The specific Rockwell C-23 no failures were reported. Harder
gravity of the 700 0 F specimen began to decrease materials of same analysis exposed in the same
at a shorter time and to a greater degree than the unit failed in a brittle manner at 1400 psi hydrogen
0
annealed sample. The same was true for the pressure and a temperature of 800 to 850 F.
800 0 F sample. The 1000 0 F sample, however, After some redesign and use of softer materials no
showed no incubation period and specific gravity further problems occurred.
decreased less than the annealed sample. This
apparent reversal of the extent of attack in the Effect of Hydrogen on Stress-Rupture
Strength at High Temperatures
5% sample was attributed to the development of
parallel longitudinal fissures in the cold rolled When steels are under strain at high tempera-
samples in contrast to the heterogeneous oriented tures, even where hydrogen attack may not
90
ORIGINAL
STRESS
80
RUPTURE
STRENGTH
RETAINED,
STRESS
PER CENT
RUPTURE
70
STRENGTH
RETAINED,
PER CENT
60
50
occur, stress-rupture strengths are lower in hydro- Fig. lO-Effect of pressure on stress·rupture strength of SAE
1020 steel exposed to hydrogen at 10000 F. Based on samples in
gen than in other atmospheres. This was demon- argon as 100%
strated by Vitovec and associates 8 - 1o • Figures
9-12 were prepared from these works to illustrate
decreases in stress-rupture strengths of various
steels when tested at elevated temperatures in
hydrogen. The curves demonstrate that the
presence of strains created by applied stress de- at a much steeper s]ope. I t is also to be noted
creases the stress-rupture strength of carbon and that in a short test at 50 psi and 1000° F the loss
low alloy steels. In fact, ]oss in stress-rupture in relative stress-rupture strength is the same as
strength of carbon steel is experienced at pressures when the pressure was 400 psi, indicating the
as low as 50 psia, whereas in the unstressed extreme sensivitity of strained metal to hydrogen
condition no attack by hydrogen can be detected. at any pressure. As previously mentioned, this
Improvement in stress-rupture strength occurs sample would be unaffected when in the unstrained
when elements such as molybdenum or chromium condition.
are added. Improvement in stress-rupture strength is ac-
Figure 9 compares stress-rupture strengths of complished by addition of strengthening elements
SAE 1020 steel tested in hydrogen at various such as molybdenum or chromium. Figure 11
pressures at 800° F in relation to its strength when compares relative stress-rupture strength of SAE
tested in argon. It is noted that all samples 1020 steel with low-alloy steels when tested in
immediately declined in relative strength upon hydrogen at 900 psi and 1000° F. An immediate
application of load and further decreased in improvement from alloying is noted but full
strength at the same high rate for approximately stress-rupture strength is not retained in these
50 hr. This time of rapid decline in stress-rupture short-time tests until the alloy steel grade reaches
strength corresponds approximately to that of 2 1/ 4 Cr-l Mo.
the incubation time for unstrained samples where Figure 12 compares the relative stress-rupture
no attack is occurring (see Fig. 8). Following strength of high purity iron at two levels of carbon
the steady rate decline, stress-rupture strength content when tested in hydrogen at 900 psi and
decreased in direct relation to the applied pres- 1000° F with that in argon at 1000° F. The
sure. higher carbon material was in the initial condition
Figure 10 compares relative stress-rupture of a vacuum annealed 3/8 in. thickness strip
strength of SAE 1020 stee] when tested in hydrogen which had been vacuum normalized at 1650° F
at various pressures and in argon at 1000° F. for 1 hr. The lower carbon material had been
Trends are the same as in the 800 ° F samples, further treated by heating in hydrogen at 1800° F
but initially the reduction in stress-rupture for 500 hr. During this exposure the carbon
strength of the samples exposed to hydrogen is content was reduced from 90 to 15 ppm. It is
90
100 - - DECARBU RIZED
HIGH PURITY IRON
STRESS
RUPTURE
80
STRENGTH 90
RETAINED,
STRESS
RUPTURE
PER CENT
70
STRENGTH
RETAINED, 80
PER CENT
HIGH PURITY IRON
60
150
50
TIME, HOURS
SU~fMAH.Y. The fi rst section of this interpretive report laid hydrogen that enters the steel often comes from
the theoretical foundation for the study of effects of hydro-
gen. The second discussed the effects of h ydrogen at corrosion reactions but can come from hydrogen
elevated temperatw·es. The purpose of t his section is to gas as well.
discuss the particai aspects of the lo wer temperature damage
by hydrogen. The selection of materials, then' use, pitfalls Failures
to be avoided, and some possible problems wi ll be reviewed.
Some examples of this type of failure will be
given. Figure 1 shows a moderately high strength
Introduction
tubing joint which completely parted shortly after
A brief summru:y of the types of damage along being lowered into a well producing sulfide fluids.
with some examples will serve to introduce the ExposUTe time of the failed joint was only a few
reader to the natme of problems which have been hours. The hardness was about RC 29, corre-
encountered. The examples have been largely sponding to approximately 110- 120,000 psi yield
drawn from the petroleum industry because of the strength . Numerous failures of valve trim parts
author's background. Another author with a (stems, seats, springs, etc. ) occurred when very
different background could cite different examples sour (up to 50 % H ,S) high preSSUTe gas wells were
of the same type of problems. Throughout the first being produced in Canada, East Texas and
paper one theme will keep appearing- bulk prop- France. The stresses were usually very low, but
erties and average nominal stresses do not ade- the hardnesses were quite high . A typical ex-
quately account for observed problems. Localized ample is shown in Fig. 2. Bourdon tubes made of
property differences at welds or coldworked areas high strength carbon steel or 400 series staiuless
may make them susceptible to hydrogen damage steels have failed when exposed to high pressure
while the bulk material remains nonsusceptible.
Stress concentrations at welds, threads, defects,
nozzles, etc., may render those areas susceptible to
damage. What subsequently propagates from
these locations is influenced by bulk properties.
Delayed fracture at low stress is perhaps the
most serious of t he types of problems because it can
result in complete and <I instantaneous" parting of
a material without warning after a period of time
has passed. The stresses can be well below the
yield strength, even as low as 10 % of the y ield.
Failure can occur within lninutes or 1110nths after
the stress is applied and the environment contacts
the material. This complete type of failure is
usually restricted to higher strength materials.
Transgranular cracking usually p redominates over
intergranular, with relatively little branching
being present in the majority of cases. The
Fi g. I -Cracking of N·80 grade oil well tubing. 2 ~/B in. 00 . The
C. 1\1. Hudgins is with the Conlinenlal Oil Cnmpony, POIICII City, cracking occu rred at a tong mark within a few minutes after
OklahomK.
The preparation of this report was sponsored by the SubcommIttee on contactin g well fl uids con tami nated wi th hydrogen sulfide.
Hydrogen Embrittlemeot of the Pressure Vesse l Researc h Committee. Hardness of the steel was Rockwell C·29
.411lbient TeI1l1)eJ'ature 43
ceptible to blistering but certainly is not immune.
Figure 4 shows blisters in fully killed, seamless
Grade B line pipe ('" x 13 in. ) caused by lamina-
tion defects due to inadequate cropping of the
ingot. Inclusions in electric-resistance-welded
pipe made from semi-killed steel led to the !issw'es
and blisters shown in Fig. 5.
Dissolved hydrogen greatly reduces the ability
of steel to deform. This is important even in
materials too soft to be susceptible to the delayed
failures discussed above. The force required to
unseat or unstick an oilfield downhole pump is
frequently large enough to exceed the yield
strength of the sucker rods. Each 25-ft rod will
normally stretch a foot or more past yield without
breaking, offering considerable "safety factor" to
the operator. In sour wells, however, the rods will
Fig. 2-Carbon steel relief valve spring after exposure to hy· part with very little permanent elongation. Bend-
drogen sulfide environment (Reprinted with permission of the
American Chemical Society from Ind . & Engrg. Chem ., 44 (10) ing or flattening of embrittled pipe will usually
1952, p. 2500) result in fractUl'e at much less strain than normal.
The wall of the blister shown in Fig. 4 has parted
with little evidence of yielding, whereas the new
pipe could be almost flattened without any separa-
pure hydrogen gas. I Cracking has occurred in the tion.
shells of transport trucks fabricated from high
strength A-517 steel within hoUl's after sour natural Preventive Measu res
gasoline was loaded. High strength tUl'bine com- There are numerous ways to combat hydrogen
pressor blades failed after three month service re- problems, the actual choice usually being con-
cycling H ,S contaminated hydrogen gas in a re- t.rolled by economic as well as technical limitations.
finery .' Cracking failures have occUl'red in hard Fortunately, hydrogen damage at atmospheric tem-
welds in softer, " non-susceptible" storage tank peratUl'es can often be largely circumvented by
materials for crude oil. ' FailUl'es of A-517F hydro- careful attention to material selection and fabrica-
gen storage tanks (5000 psi) were associated with tion practices. Moreover, the additional cost is
welds.! Some of the tanks were found to have had often relatively minor. Restriction of the maxi-
cracks before service so that this example might mum strength level is the single most important
better be considered an example of interaction of material consideration. This includes weldments,
high hardness, defects, and hydrogen gas. cold-rolled threads, etc., as well as bulk material.
Elimination of cold work, stress raisers, and hard
weld areas are critical in fabrication, especially for
Blistering stronger steels. A combination of these measw'es
has virtually eliminated value failw-es in the pro-
Hydrogen blistering results from hydrogen gas duction and handling of sulfide-containing oil, gas
collecting in defects. Hydrogen can enter and and salt water.' Often it is more desirable to
move rather freely t hrough many materials as long
as the hydrogen exists as an atom. If the hydro-
gen atoms combine to molecular hydrogen in a void
or defect in the material, hydrogen is rather ir-
reversibly trapped. Continued accumulation of
molecular hydrogen gas in the defect can lead to
very high preSSUl'es and the formation of blisters in
softer steels. The large blisters observed in
storage tanks containing soul' crude oil is perhaps
the most common example of this problem (Fig. 3).
However, blistering of heavy walled refinery
reactor vessels was a lTIore serious problelll because
of the pressure involved. The use of alloy linings
and multiwalled carbon steel vessels has minimized
this problem, but the inner liner must be vented to
prevent hydrogen accumulation which could Fig. 3-Cross section of a hydrogen blister in ~/u in. thick plate
collapse the linear. Piping is generally less sus- from a sour crude oil storage tank
A I1Ibieni 'Temperature
-
Fig. 4- Bli stering of Grad e B line pipe- 12 in. 00 x '/4 in. wa ll. Fig. 5-Magnetic particle ind ications of fissures associated with
Thi s had been exposed in the laboratory to 5% sodium chloride hydrogen blister formation in electric·resistance-welded line
solution sat urated with hydrogen sulf ide pipe in sour service. The blisters occurred at inclusions in steel
separate the structural material from the environ- strength less than approximately 90,000 psi
ment. This approach is used in many refinery will be resistant to cracking failures. ' This
applications where a thin alloy liner is used to corresponds roughly to Rockwell C22 hardness.
prevent damage to the thick carbon steel pressure This latter is mentioned because it has been used
shell.' A third approach involves ch anging the as a criteria in NACE , API and other specifica-
envirOlunent to reduce its aggressiveness. Con- tions," and is particularly useful for nondestructive
trol of pH in drilling muds' and refinery wash testing in the laboratory or in the field, even on
streams7 are instances which have met with con- complex shapes. It should be emphasized that
siderable success. the 90,000 psi yield presently represents a maxi-
Let us now exalnine these approaches in con- mum in the available useable strength for carbon
siderably more detail. It should be pointed out a nd low alloy steels in sulfide service. This in-
that most of the supporting data and considerable formation has led to the introduction of a special
additional information is included in Section I of grade, C-75, for oilfield tubing and casing for
this bulletin . The engineer faced with a specific hydrogen sulfide service. A major factor in this
problem would be well a dvised to stud y pertinent specification is the restriction of the yield strength
areas there also. to a range of 75,000- 90,000 psi. At higher
strengths, the threshold stress to failure actually
Materials
decreases by approximately 10,000 psi for each
The best technical solution to delayed fracture 10,000 psi increase in yield strength. This has
problems u sually is to select a material that is not been confirmed by a number of investigators. '"
susceptible to this mode of failure and fabricate For example, low alloy steels with a yield strength
and use it in such a lnanner so as not to induce of 150,000 p si can be expected to fail with only
susceptibility. A number of the factors involved 20,000- 30,000 psi applied stress in a corrosive
in this approach will now be discussed. Es- sulfide environment. Likewise, it is important to
sentially all investigators who have studied hydro- point out that welding or excessive cold work on
gen damage from any h ydrogenizing source have l1ol1susceptible lnaterials can increase theil' sus-
found that strength level of material is the most ceptibility. Figure 6 shows two failed B-7 bolts
significant single variable.' (RC 31- 34) removed from the floating head of a
CARBON AND LOW-ALLOY STEELS. For the rel- heat exchanger exposed to a sour gas. The failure
atively severe sulfide h ydrogenizing environment can be prevented by using softer (RC 22 max.)
it has been found that most steels with yield materials, if compatible with strength require-
JG Jbllbien( T CIIIIJeralu7"l'
to hydrogen cracking. 16 The fine carbide disper- Rockwell C units higher than in the adjoining base
sion usually associated with tempered martensite is metal. It should be emphasized that only one
considered advantageous while the coarser precipi- material did not fail in this series of tests-and
tates of carbides, such as pearlite, are less desir- that one was the A-285-C steel (which has a
able. 17 To some extent this superior performance specified minimum tensile strength of 55,000 psi).
is due to crack propagation resistance. The In other more severe environments outside of
maximum effect of the differences in microstructure above tests, A-285 steel, cold-worked by bending
appears to be approximately 15,000 psi increase in to less than 5 % strain and under yield-strength
useable strength as long as the untempered mar- loading, did fail in operating equipment. Mc-
tensite is excluded. Useful strengths to 107,000 Pherson and Cataldo I reported on a number of
psi were recently attained by a double tempering failures of A-517-F steels in high pressure hydrogen
process whereby newly formed martensite was gas service. Essentially all of the failures were
tempered in the second treatment. IS associated with welds, either in nozzle area or in
longitudinal and circumferential internal welds.
Cold Work
While many of the failures were associated with
The adverse effect of cold work has been ob- poor quality welds the authors concluded that
served by almost every investigator who has under present fabrication techniques high strength
studied hydrogen embrittlement. 19 There is little steel is not satisfactory for gaseous hydrogen ser-
doubt that more than about 5 % cold work will vice.
result in a marked increase in the susceptibility to
hydrogen damage and cracking failures. Below Quality Control
5%, however, it appears that the significance of the
The importance of quality control throughout
damage will depend on the original properties of
the welding process cannot be overemphasized.
material. For example, materials that are barely
Defective welds provide stress concentrations, laps,
safe in the non-cold worked condition might be-
porosity, etc., which can greatly aggravate hydro-
come dangerously susceptible with a small incre-
gen damage. Likewise, high hardnesses result in
ment of cold work, while soft materials would not
greater susceptibility to hydrogen-induced delayed
be significantly affected by the same amount of
failure. An important step in minimizing high
cold work. One source of damage that should
weld area hardness is to specify base metal and
generally be restricted is cold straightening of
welding electrodes that do not have unnecessarily
tubular goods during processing without subse-
high carbon and alloy contents. This usually
quent stress relieving. Another example where
means restricting the carbon, manganese, nickel,
problems can result is the cold bending of tubular
and chrome contents to the minimum value con-
goods during installation, for example the installa-
sistent with the properties of the base metal. A
tion of pipelines in hilly terrain. Cold formed
suitable preheat practice may be required to mini-
pressure vessel heads have been found to be un-
mize quench cooling in the heat affected zones as
usually susceptible to hydrogen damage. Like-
well as insure a dry material so that minimum
wise, cold formed tray support rings welded to the
amounts of hydrogen will be introduced during the
inside of a vessel have been observed to crack when
welding process. Note that specifications for
exposed to sour environments, that is, refinery
welding electrodes do not always control the high
1luids containing H 2S and cyanides.
side of mechanical properties.
Welding Good design and welding techniques should be
used at all times. Rounded corners, a minimum
I t is no coincidence that delayed cracking failures
number and limitation on severity of stress raisers,
of pressure vessels are frequently initiated in and
full penetration of welds, minimum porosity,
often propagate along the weld area (including both
smooth changes in cross section, etc., are quite
the weld metal and the heat affected zone). These
important. Random arc strikes on a vessel wall
failures generally re1lect higher hardness in the
nearly always leave hard spots and should be
weld metal or in the heat affected zone. For
rigorously avoided. Inspection of critical welds
example, Bates 3 reported on a study of moderate
should include X-ray for determination of porosity
to high strength steels exposed to sour crude oil
and also magnetic-particle or dye-penetrant ex-
environments. One of his most important con-
amination to find cracks.
clusions was that welding of the higher strength
steels greatly increased the probability of cracking
Heat Treatment
failure. Metallographic examination revealed
that cracks in the welded specimens usually oc- Heat treatments after fabrication are commonly
curred in the weld zones. Of the 72 welded used to reduce residual stresses and hardnesses to
specimens that failed, only one specimen cracked acceptable levels and minimize the effect of cold
in a spot remote from the weld. Hardnesses in work. The importance of these operations in the
the heat affected zone were generally 10 to 20 fabrication of pressure vessels, containers, pipe
Ambient Temperature 47
values, etc., has been proven time and time again. cracks originating at such areas will not propagate
Specific practices are included in many specifica- into failure of the vessel, pipe or unit. Sufficient
tions and codes. These operations are especially notch ductility in the base metal to accept such
important in a low temperature hydrogenizing small cracks is vital when equipment is to be used
environment. For example, the introduction of in hydrogenizing service.
the NACE specification (1-F-1-66) for valves for
sour service' virtually eliminated cracking failures Field Repairs
in this severe hydrogenizing environment. A Repair of equipment in the field poses even more
major factor in the success has been the required severe problems than the original construdion.
heat treatment at 1150 0 F after all welding repair, There is nearly always an urgency to get the sys-
overlay, cold work, etc., has been completed. tem back on stream. It often is almost completely
At temperatures in this range residual stresses are impractical to heat treat a pressure vessel in the
reduced to relatively low values because most field after welding repairs. Very careful control
steels have lost much of the strength. Perhaps must be exercised in localized heat treatments to
more important, however, is t he tempering that avoid creating still more problems. Selection of
will occur. Tempering at 1150 0 F will soften welding electrodes is very important, especially if a
nearly all carbon or low alloy steels to below RC different base metal is to be used for a modification.
30 and most to RC 20- 25, thereby greatly reducing Pads or patches added to a vessel can introduce
the probability of a cracking failure. Selection of residual stresses as well as heat affected zones
specmc treatments must necessarily take wall which enhance susceptibility to delayed cracking
thickness into account as well as type of steel. failure. The low-hydrogen rods ease weld crack-
The effect of Lhe treatment of other properties ing, but do little for the hardness of the heat-
(e.g. , impact strength and corrosion resistance) affected zone . A very common source of high
must also be considered. Nevertheless, hard weld hardness weld metal is high manganese welding
areas constitute an inunediate threat to the integ- electrodes or flux.
rity of the vessel whenever hydrogenizing condi-
tions are present. Hardnesses in these areas Blistering
should be kept as low as feasible, even at some Another problem which may seriously affect
sacrifice of other properties. Hydrogen introduced pressure vessels exposed to hydrogen charging
during the weld operation can be removed during media is that of blistering, which occurs most com-
a post-fabrication heat treatment. monly in hydrogen sulfide or hydrofluoric acid
Service Pro blems
Ambient Temperature 49
particularly important when the external layers because of metal loss. Sprayed aluminum coatings
are high strength steel and the primary concern is have been used very successfully to protect relief
cracking rather than blistering. Considerable value springs.
care must be exercised in the fabrication procedures,
especially where various internal support members Nonmetallic Coatings
are connected to the main structural body. Par- Nonmetallic coatings also have a definite role
ticular care must also be exercised so that vibration in preventing hydrogen damage, especially where
and thermal fluctuations will not result in fatigue corrosion is the source of the hydrogen. When
cracking of the protective lining. Since the higher the right coating is properly applied, a very high
alloy content usually results in less corrosion degree of protection can be obtained. Properties
generated hydrogen and lower permeation, the of these materials have been improving over the
rate of accumulation of hydrogen in the annulus last few years. Temperature limits have been
will be reduced, resulting in a corresponding re- raised and materials are now available for above
duction in the probability of a collapse failure of 400 0 F in fairly severe chemical environments.
the inner liner. Reasonably good information should be available
Fabrication of multi-layered vessels is more com- on the use characteristics of the coating before it is
plicated than single layered vessels. Details are relied on the protect high strength steel. In
beyond the scope of this paper, but several im- addition, it is highly desirable to have some sort of
portant items should be pointed out. First of all, monitoring system available to detect the first
there should be a minimum number of welds which stages of deterioration of the coating. This would
connect an inner layer of carbon steel to the outer prevent a premature coating failure from resulting
layer. Whenever there is a continuous metallic in a catastrophic failure of the pressure containing
bond between the two layers, hydrogen can diffuse wall. The bonding characteristics of the material
through the bonds without having to go through are also quite important. It may be that minor
the gaseous phase. This, in effect, reduces the defects can be tolerated for steels that are only
effectiveness of the multi-layer construction. moderately sensitive to embrittlement if the re-
Secondly, even more rigid precautions and specifi- mainder of the coating will remain bonded. This
cations must be used in welding the seams on the is so because the hydrogen which contacts the steel
inner liner. If this liner should crack, a high at a point defect will be able to diffuse away from
strength external layer could be immediately the defect in hemispherical permeation geometry,
exposed to the full impact of the hydrogen environ- resulting in relatively low concentrations in es-
ment with an increased possibility of a catastrophic sentially all of the metal. Rapid depressurization
failure. Provision must be made so that a rapid can be a serious problem to coatings. Gases ab-
pressure buildup in the annulus will trigger shut- sorbed into the plastic should have time to diffuse
down procedures and slow depressurization of the from the plastic without disintegrating it or dis-
vessel. Precautions must be taken in welding bonding it from the base metal. Plastic coatings
alloy liners to the outer layers to minimize hard have the distinct advantage of being able to be
spot formation and residual stresses that can result applied after the rest of the vessel is completely
from the welding of dissimilar metals of this type. fabricated.
There are a number of process vessels which are
Metallic Overlays glass lined for corrosion protection from very
severe corrosive environments. Many of the
Recent developments in metallic coatings show same criteria discussed for the plastic coatings, also
considerable promise for improving the economics apply here. The complete vessel wall must be
and the effectiveness of control of corrosion and covered. To date most of these lining techniques
permeation of hydrogen. The integral liners can involve thermal treatments of the entire vessel
be weld overlays or the more recently introduced which must be performed in the fabricators shop.
explosive clad type. Actual operating experience Care must be taken in subsequent transportation
with the latter type of materials is still rather and installation to prevent damage to the coating.
limited. Some significant successes, however,
have been observed in the petroleum industry with
aluminized steels for various refinery applications. Alternation of Environment
While the primary purpose of the coating is usually
to resist moderately high temperature sulfide As mentioned in a number of cases through the
corrosion, it may also minimize the permeation of report, the amount of hydrogen actually entering
hydrogen into the underlying steel. It would be the steel is greatly influenced by the presence of
very desirable, however, for any linings such as certain contaminants such as hydrogen sulfide,
this to be 100% complete. Pin holes can lead to hydrogen cyanide, arsenic compounds, phosphorus
hydrogen entry into the structural material and compounds, etc. One approach to minimizing and
also to undercutting of the surrounding coating perhaps eliminating the problem in some circum-
50 Ambient Temperature
stances is to remove or inactivate the so-called plication. If so, some material compatibility
poisons. Perhaps the most well known example of tests will certainly be in order, as the present state
this is the practice of removing H 2S from natural of knowledge does not allow us to anticipate these
gas down to very low levels of a few ppm. While problems.
this serves many purposes, the most important is Considerable success has been encountered in
to minimize subsequent corrosion problems in the many fields using corrosion inhibitors such as
gas pipeline network. filming amines to control weight loss corrosion.
Another example is the practice of adding small There is little doubt that such inhibitors have a
amounts of oxygen or ammonium polysulfide to place in reducing hydrogen embrittlement from
the refinery process streams which are contami- corrosion generated hydrogen. A number of
nated by H 2S or cyanide. 7 The practice appar- factors should be considered, however, before
ently converts the H 2S to polysulfide and the relying on this approach to protect susceptible
cyanide to thiocyanate ions, neither of which cata- high strength steels. First of all, only a small
lyze hydrogen entry. Close pH control is required amount of hydrogen will crack highly susceptible
in this practice to avoid severe weight loss corro- materials. This means that levels of inhibition
sion. Hydrogen sulfide can also be oxidized to must be very high, approaching 100%. Secondly,
sulfur or sulfate ion. Chlorine has been used in a the time required for the cracking to occur may be
number of oil field applications to provide a com- only a matter of minutes for a highly susceptible
bination of hydrogen sulfide oxidation and bacte- steel in a severe environment. This means that a
rial control in brines. Control of pH itself can fail-safe continuous inhibition system must be
sometimes result in significantly less hydrogen used. Both of these requirements strongly suggest
entry. Drilling muds can be maintained at as that high concentrations of highly persistent
high a pH value as practical (pH 10-12) in order to inhibitor will be required. This approach may be
minimize sulfide cracking. 6 However, particular suitable for closed systems where the aggressive
care should be exercised in the pH control when- fluid is at a fixed volume; however, for a "once-
ever there is a possibility that the solution will be through" stream the economics and disposal
evaporated towards dryness and where the carbon- considerations may make this approach of ques-
ate or sulfide content of the solution is high. tionable value. The intermediate condition where
Under these circumstances one can get into a a safe or perhaps a moderately susceptible steel
caustic embrittlement problem which, in part, is a is exposed to corrosive environment probably
result of hydrogen embrittlement. 20 represents the best possibility for using this ap-
Corrosion is one of the most serious sources of proach. This is also essentially the conclusion
hydrogen in plant service, since any aqueous that Bates" reached from his test data mentioned
corrosion process is a potential source. Whether earlier. Failures of the less susceptible nonwelded
hydrogen damage actually occurs depends on specimens were almost eliminated by adding
whether or not hydrogen is liberated and whether inhibitor. Failure fraction of the welded speci-
that hydrogen goes into the steel. To minimize mens was reduced to 13% from 39%, but this is
the hydrogen damage one can thus interfere with hardly satisfactory for engineering use. If the
either one or both of these processes. In many application involves one side attack (hydrogen
situations the problem can be drastically reduced, diffuses through the material and out the back
if not completely eliminated, by maintaining the side) then a reduction of 80 to 90 % in the hydrogen
aqueous dew point of a process stream at a low generation rate may keep the concentration of
temperature. In most systems corrosion will not hydrogen in the steel down to a tolerable level.
proceed (or will proceed at an extremely slow rate) Conversely, if the exposure is for a short time only,
unless a conductive liquid phase, which is required inhibition may prevent failure for the required
for the electrochemical corrosion mechanism, is time.
present. In the majority of cases this phase would A widespread successful practice is the use of
be water, with or without contaminating electro- inhibitors on wirelines used in downhole operations
lytes. However, we should hasten to point out in oil wells. The lines are very susceptible (232,-
that there are many nonaqueous solvents with 000 psi tensile strength) but inhibitors allow
adequate conductivity to allow corrosion to pro- successful use in mild environments for a few
ceed electrochemically. If these solvents will also hours. In moderately high H 2S environments,
furnish hydrogen for the cathode reaction exposure stainless steel (170,000 psi tensile strength) plus
to them may lead to hydrogen embrittlement prob- inhibitors is required. In severe environments it is
lems. The anhydrous hydrogen fluoride used in usually necessary to fill the tubing with an in-
petroleum alkylation processes is perhaps the most hibited hydrocarbon fluid to prevent failure of the
important commercial example. The increasing wire line. Unfortunately, there is no adequate
attention being given nonaqueous solvents in both quantitative data to determine the hydrogen per-
academic and industrial research laboratories may meation rate or concentration required to produce
be a prelude to more widespread commercial ap- cracking in specific steels. The use of corrosion
Ambient Temperature 51
inhibitors ordinarily should be considered as a good Applications of new materials for these must be
insurance technique rather than a primary control made with great caution. In conclusion, it is
procedure to prevent hydrogen embrittlement truly amazing that one of the elements of water
from hydrogen generated by corrosion. can exert so much influence on the choice of pres-
There are also some subtle sources of hydrogen. sure vessel materials and design.
A common means of cleaning process equipment is
to clean it with acids. If iron sulfide is present, References
copious quantities of hydrogen sulfide can be 1. McPherson, W. B., and Cataldo, C. E., "Recent Experience in High
Pressure Gaseous Hydrogen Equipment Operated at Room Temperature,"
released. This hydrogen sulfide dissolved in presented at the 1968 Materials Engineering Exposition and Congress,
Oct. 14-17. ASM Technical Report No. D·8-14.1.
water and forming condensed droplets or dissolved 2. Kohut, G. B., and McGuire, W. J., "Materials for Centrifugal Com-
in the acidic solution is a potent cracking agent. pressors Operating in Refinery Environments Containing Hydrogen Sulfide,"
Materials Protection, 7,17 (June 1968).
Some metal loss inhibitors may aggravate the 3. Bates, J. F., "Sulfide Stress Cracking of High Yield Strength Steels
in Sour Crude Oils, "Materials Protection, 8, 33 (January 1969).
entry of hydrogen into the steel. Electroplating, 4. Preliminary report of NACE Committee T-1B presented at Western
Canadian Regional Conference, February 7-9, 1962, at Calgary, Alberta.
electrogalvanizing and cathodic cleaning are also Reported in Materials Protection, 2, 89 (March 1963). Also NACE pub-
sources of hydrogen leading to cracking and some- lication 1F1-66 published in Materials Protection, Ii, 81, (September 1966). .._
5. McCabe, J. S., and Hickey, K. J., "Field Assembly of High Pressure
times to formation of small blisters. On the very Reactors Comes of Age," Oil Gas Jnl., 113 (May 19, 1969).
6. Hudgina, C. M., McGlasson, R. L., Mehdizadeh P., and Rosborough,
high strength steels even rusting can introduce W. M., "Hydrogen Cracking of Carbon and Alloy Steels," Corrosion, 22, 238
enough hydrogen to crack the steel if high stresses (August 1966).
7. Wilson, J. G., Bonner, W. A., Burham, H. D., and Skei, T., "Occlusion
are present from imposed loads or from residual of Hydrogen in Steel," Proc. 4th World Petrol. Congr., Section VIlIC Paper 3,
1955.
stress. If welds are present, the heat-affected 8. Groeneveld, E. Fletcher, E., and Elsea, A. R., "Special Report on
zones give both the high hardness and the residual Review of Literature on Hydrogen Embrittlement," NASA Report CR-
74034, STARF Accession No. N 66-23505, January 1966.
stresses. 9. Treseder, R. S. and Swanson, T. M., "Factors in Sulfide Corrosion
Cracking of High Strength Steels," Corrosion, 24,31 (February 1968).
10. Warren, D., Beckman, G. W., "Sulfide Corrosion Cracking of High
Conclusions Strength Bolting Material," Corrosion, 13, (10), 63lt (October 1957).
11. Fletcher, E. E. Berry, W. E., and Elsea, A. R., "Stress Corrosion
Cracking and Hydrogen-Stress Cracking of High-Strength Steel," DMIC
The examples used in this report demonstrate Report 232,Defense Metals Information Center, Battelle Memorial Institute,
Columbus, Ohio, July 29, 1966.
the complexities of designing for hydrogenizing 12. Smialowski, M., "Hydrogen in Steel," Addison-Wesley Publishing
environments. Pitfalls are many, the most serious Co., Inc., Reading, Mass., 1962, p. 102.
13. Truman, J. E., Perry, R., and Chapman, G. N., "Stress-Corrosion
being those associated with welding of the more Cracking of Martensitic Stainless Steels, Jnl. Iron Steel Inst., 745 (September
1964).
hardenable steels. Such steels need not always be 14. Whiteman M. B. and Troiano, A. R. "Hydrogen Embrittlement of
exotic, since damage to A-515 steels is frequent Austenitic Stainless Steel," Corrosion, 21, 53 (1965).
15. Vennett R. M. and Ansell, G. S. "The Effect of High Pressure Hydro-
enough. There is no steel in common use which is gen Upon the Tensile Properties and Fracture Behavior of 304L Stainless
Steel," Trans. ASM, 60, 242 (1967).
not sensitive to the most severe of exposures. 16. Baldy, M. F., and Bowden, R. C., Jr., "The Effect of Martensite on
Sulfide Stress Corrosion Cracking," Corrosion, 11 (10), 417t (October 1955).
Fortunately the most common environments are 17. Snape, E., "Roles of Composition and Microstructure in Sulfide
indeed compatible with the ordinary carbon steels. Cracking of Steel," Corrosion, 24, (9) 261 (September 1968).
18. Snape, E., Schaller, F. W., and Forbes Jones, R. M., "A Method for
Oil refinery and sour streams, oil well fluids, hydro- Improving Sulfide Cracking Resistance of Low Alloy Steels," presented at
the 25th Annual Conference of the National Association of Corrosion Engi-
fluoric acid, hydrogen sulfide, cyanides, high pres- neers, Houston, Texas, March 10, 1969.
sure hydrogen, and cathodic currents at high cur- 19. Bowers, C. N., McGuire, W. J., and Wiehe, A. E., "Stress Corrosion
Cracking of Steel Under Sulfide Conditions," Corrosion, 8, 333 (1952).
rent density all require specialized attention. 20. Smialowski, M. op. cit., p. 416.