You are on page 1of 32

ANNUAL

REVIEWS Further
Ann. Rev. Mater. Sci. 1978. 8: 327-57 Quick links to online content
Copyright © 1978 by Annual Reviews Inc. All rights reserved

HYDROGEN EMBRITTLEMENT x8622

OF STEELS
R. A. Oriani
us Steel Corporation Research Laboratory, Monroeville, Pennsylvania 15146
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

INTRODUCTION

Hydrogen is able to lower the load-bearing ability and/or the energy­


absorption ability of many metallic alloys such as those of iron, nickel
by Cornell University on 06/09/12. For personal use only.

(1, 2), aluminum (3), titanium (4), zirconium (5), tantalum (6), hafnium (7),
niobium (8), vanadium (9), tungsten (10), molybdenum ( 1 1), and uranium
( 12). Despite the great technologic importance of hydrogen embrittlement
phenomena in all these systems and the undoubted scientific value of
exploring the similarities and dissimilarities among them, because of space
limitations this review considers only the embrittlement of steels by
hydrogen. Even in this narrower field, the l iterature is vast. This review
does not consider cyclic loading or the high-temperature attack of steels
by hydrogen where, for example, methane is generated. It concerns itself
with phenomena such as the reduction of the load-bearing capability
of high-strength steels, and the reduction of the ductility of low-strength
steels.
The emphasis throughout is on developing understanding, not on
cataloguing phenomena or on engineering information. We begin with a
brief presentation of the major theoretical points of view that have been
advanced. We proceed to describe the important properties and effects
of hydrogen in iron and steel that appear to be relevant to understanding
the embrittlement phenomena. We then discuss experiments on the
fracture of steels and the understanding that has been generated. We
conclude with an assessment of current understanding and of fruitful
avenues of further investigation.

BRIEF SURVEY OF IMPORTANT IDEAS


CONCERNING HYDROGEN EMBRITTLEMENT
IN STEELS

Early observations of the production of blisters by hydrogen put into


steel by gaseous charging at high temperatures followed by cooling, by
327
0084-6600/78/0801-0327$01 .00
328 ORIANI

cathodic charging, or by corrosion reactions led Zapffe & Sims ( 13) to


develop the pressure-expansion theory of hydrogen embrittlement. The
idea, further developed by others ( 14--1 7), is that very large hydrogen gas
pressures generated in internal microcracks and voids force their expan­
sion either by plastic deformation or cleavage, leading to the coalescence
of microcracks or microvoids. Since the large internal gas pressures
constitute an attempt by the system to achieve thermodynamic equili­
brium with the large hydrogen activity developed by the method of
charging (see below), it is clear that the observed ( 1 8) cracking of higher­
strength steels by gaseous hydrogen at 1 atm or less cannot be explained
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

by the thermodynamically motivated generation of large internal pres­


sures. Other ideas must be considered. Kinetic variants of the internal
pressure theory are presented and discussed later.
Petch & Stables (19,20) pointed out that the surface free energy, y,
by Cornell University on 06/09/12. For personal use only.

of an atomically clean metal is lowered by adsorption of hydrogen.


Therefore the fracture stress, as given by Griffith's thermodynamic
criterion, is lowered as yt. Only recently have data (2 1 ) for the adsorption
of hydrogen upon iron become available to make possible a testing of the
Petch-Stables model. The result did not verify the model (22). Such an
exercise is made difficult by the fact that the Griffith equation cannot
be rigorously modified to apply to a non ideally brittle material (23).
The idea that adsorption lowers the energy required to create new surface
is not necessarily linked to any one mechanism for the generation of new
crack surface, as is true for any thermodynamic reasoning. It would be as
applicable to a mechanism involving the breaking of interatomic bonds
as to one in which the new surface is produced by plastic tearing.
The decohesion model (24--26) for hydrogen embrittlement is more
mechanistic. It supposes that dissolved hydrogen at high concentrations
lowers the maximum cohesive force, Fm, between the atoms of the alloy
in the iron-alloy lattice, at grain boundaries and at interfaces. Bond
breaking results when the local stress, (J�, equals the hydrogen-lowered
Fm (each in equivalent units). The tensile stress serves to concentrate
the hydrogen by several orders of magnitude through the thermodynamic
relation between the chemical potential of an interstitial solute and the
stress state (27), and also to sever atomic bonds weakened by the
hydrogen. The necessary, very large, non-Hookean stresses may exist
directly at a crack tip in a material exhibiting some plasticity, as shown
by Rice & Thomson (28), as well as at dislocation pile-ups, intersections
of slip bands, and regions of plastic incompatibility; these latter are
consequences of inhomogeneous plastic deformation.
Development of an explicit formulation for the decohesion model is
made difficult by the current ignorance of how to express the relevant
HYDROGEN EMBRITTLEMENT 329
local stresses in terms of macroscopic parameters, and by the total
ignorance of how Fm varies with hydrogen concentration or chemical
potential, and with alloy concentration and interface structure. The de­
cohesion model can be reformulated to take advantage of experimental
information on the variation of interfacial free energy, y, with the chemical
potential, jJ., of the hydrogen. The relation (29, 22) is
2 [dr(b,jJ.)/db]om(/l) dy
1.
n roJjJ.) - ro,(jJ.) djJ.'
where r i s the excess interfacial concentration o f the adsorbed hydrogen
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

at chemical potential jJ., 15 is the interplanar distance, boo is infinite


separation (i.e. two well-separated surfaces), be is the equilibrium stress­
free interplanar distance, 15m is the 15 at which the cohesive force equals
Fm (the maximum cohesive force per atom), and n is the number of alloy
by Cornell University on 06/09/12. For personal use only.

atoms per unit area of interface. However, the advantage of this formula­
tion is more illusory than real because of the necessity of using an
ad hoc assumption for dr/db at 15m•

Relation between Dissolved Hydrogen and Dislocations


Beachem (30) has asserted that the essential action of hydrogen is to
diffuse into the lattice ahead of the crack tip and to aid whatever plastic
deformation processes the system displays. The argument, quite the
opposite of the usual interpretation, seems to be that hydrogen causes
cracking by increasing the plasticity. This is discussed at greater length
in a subsequent section.
Bastien & Azou (31) suggested that dissolved hydrogen can be carried
along by moving dislocations and precipitated elsewhere. Whatever may
be the importance of such transport of hydrogen to the kinetics of
hydrogen-assisted crack propagation, it should be noted that the sug­
gestion does not address the problem of the mechanism by which
hydrogen causes embrittlement.

RELEVANT PROPERTIES AND EFFECTS OF


HYDROGEN IN IRON AND STEEL

Equilibrium Properties
Hydrogen dissolved in the lattice of iron exists predominantly in the
dissociated (monatomic) form and certainly occupies interstitial positions
(32); which of the possible interstices is occupied is still subject to
reasonable doubt [for a discussion see (33)]. The partial molal volume,
VR, of lattice-dissolved hydrogen is 2.0 em3/mole H in bee iron (34-36),
and probably not much different from this in fcc iron. The solubility of
330 ORIANI

hydrogen in o:-iron in the absence of extraneous attractive interactions


is very small and may be expressed (32) as c 3.7p� exp (-6500/RT),
=

where c is in cm3 H2 (STP)/cm3 Fe, p in atm, and T in degrees Kelvin.


The solubility in y-iron is somewhat larger, and under 1 atm of Hz gas,
between 1 500 and 1 673K, is given (37) by 10glO c = 0.788 - 1 563/T in the
same units.
Because the intrinsic, lattice-dissolved content is so small, by far the
major portion of the hydrogen content of ferritic iron and steels is due
to attractive interactions at chemical and structural features in the metal.
With dissolved carbon and nitrogen atoms in o:-iron, hydrogen interaction
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

enthalpies of -3.3 kJ/mole H and -12.5 kJ/mole H, respectively, have


been estimated (38) from magnetic relaxation studies. The same workers
have deduced an enthalpy change of about -4.2 kJ for the formation
of a di-hydrogen complex from two mono-interstitial hydrogens. Blythe &
by Cornell University on 06/09/12. For personal use only.

Kronmuller (39) have inferred the existence of H-vacancy complexes that


anneal out at about 1 00°C.
Extended structural features in iron and steel have stronger interactions
with hydrogen. Using tritium autoradiography, Lacombe and colla­
borators (40) have observed accumulations of tritium at dislocations,
internal interfaces, and grain boundaries, the latter exhibiting obvious
dependence of the tritium accumulation on their character. Asaoka et al
(41) have measured the hydrogen gas evolution with continuous elevation
of the temperature from a previously charged specimen of Fe-0. 1 5% Ti,
and have shown in a semi-quantitative fashion that the enthalpy of binding
of hydrogen ranges from about - 45 to about - 1 20 kJ/mole H, the
larger values being attributable to sites at precipitate-matrix interfaces
and the smaller values to grain boundaries and dislocations. This
qualitatively agrees with equilibrium isotherm determinations of
Podgurski & Oriani (42), who found in addition that the interaction
of hydrogen with dislocations is lessened when nitrogen is first adsorbed
upon them, and that the enthalpy of adsorption of hydrogen upon the
o:Fe-AlN interface is about - 65 kJ/mole H as referred to lattice-dissolved
hydrogen.
Many determinations (43-46) of the cold-work internal friction peak
have been made in hydrogenated iron, and the value of -26.5 kJ/mole
H can be taken (47) as the best value for the interaction of H and a
dislocation core in o:-iron. Zielinski et al (46) have inferred that the cold­
work peak due to hydrogen can be strongly influenced by interactions
from other interstitial solutes. Recent work by Atrens et al (48) gives
- 1 3. 5 kJ/mole H as the corresponding quantity for fcc Fe- 1 8'X,Cr-Ni
alloys. Whiteman & Troiano (49) measured a diminution of the stacking
fault energy in 25%Cr-20%Ni stainless steel from about 0.030 to about
2
0.0 1 6 J/m by hydrogen.
HYDROGEN EMBRITTLEMENT 331
Hydrogen adsorbs strongly on atomically clean surfaces of iron ; the
measured values of the enthalpy of adsorption [i.e. for the reaction
H2 (gas)--+2H (ads.)] range from -84 to - 142 kJ/mole H2 at low
coverages on evaporated films of undetermined surface roughness (50-52).
Recent work by Chornet & Coughlin (2 1 ) by the flash-filament technique
on pure iron wires gives an enthalpy of adsorption of -85 kJ/mole H 2 .
These workers measured the equilibrium amount of hydrogen adsorbed
as a function of H2 gas pressure at room temperature ; it is notable
that essentially complete monolayer coverage is produced by very small
gas pressures of the order of 1 mPa. One may confidently predict that
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

alloying of iron with other metallic elements (except possibly gold) will
not change this situation appreciably. However, the prior adsorption of
nonmetallic elements may markedly change the adsorbability of
hydrogen ; reliable data on these effects are apparently not available.
by Cornell University on 06/09/12. For personal use only.

Recent work of Lunarska et al (53) showed that the shear modulus


of iX-iron is decreased by dissolved hydrogen at a rate of about 8% for
each 1 at.% H. Baranowski (54) was not able to find evidence for the
formation of a hydride of iron (i.e. of a phase that is not coterminous
with the terminal solid solution) at H2 gas pressures up to 25,000 atm.
Ni and Mn, common alloying elements in steels, form hydrides when the
pure metals are exposed to high-pressure hydrogen (6000-8000 atm for
Ni, and 1 8,000 atm for Cr) (54); in the case of nickel, at least, alloying
into iron quickly raises the formation pressure of the hydride into a
range inaccessible to the high-pressure apparatus. By high-fugacity
cathodic charging (see below), Smialowski et al (55) were able to form a
hydride with the approximate composition NiHo 7 in Ni-Fe alloys with
.

an iron content up to about 25 at.%. Szummer & Janko (56) found


evidence by X-ray diffraction at - 1 70°C that high-fugacity cathodic
charging of25%Cr-20%Ni austenitic stainless steel at room temperature
produces a fcc hydride phase as well as hexagonal a-martensite, in agree­
ment with prior work (57, 58).

Kinetic and Transport Phenomena


INGRESS OF HYDROGEN The overall process of the entry of hydrogen
from its source in the environing phase into the iron or steel is complicated
irrespective of the nature of the source, and full understanding is not
available for any case. For pure H2 gas in contact with pure iron, the
sequence of necessary steps must include at least the impingement of the
H 2 molecule, its dissociation into adsorbed atoms, and the transition
from the adsorbed to the dissolved state. Shanabarger (59) infers the
existence of an intermediate molecular adsorbed state. Which of these
steps is rate-limiting probably depends critically on the chemical state
of the metal surface, its local atomic configuration, and the temperature
332 ORIANI

From the isotope-exchange work of Cavalier & Chornet (60) it is reason­


able to infer that at 294K, for the purity of iron used by them, the
adsorbed H --+ dissolved H step is rate-limiting. A comparison of the
'
temperature dependence of crack propagation in 4 1 30 steel under
molecular hydrogen (6 1 ) gas to t�at under partially dissociated hydrogen
gas (62) may be interpreted as showing that H2 dissociation in the former
case is rate-controlling in the higher-temperature range but oot in the
lower range. Wheo either HzS gas or HzO vapor is used as the environing
phase, formation of a layer of reaction product with iron and/or the
alloying elements may complicate matters further since the reaction
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

kinetics and the entry of the hydrogen thereby made available may be
complicated functions of the chemistry and morphology of the product
layer. It is known that both the ingress ( 1 8) and the exit (63) of hydrogen
from iron are seriously impeded by oxygen; it appears probable that the
by Cornell University on 06/09/12. For personal use only.

kinetics of hydrogen availability to a cracking process depends on how


the cracking itself mechanically disrupts the film of reaction product.
When the environing phase is aqueous, the kinetics of the entry of
hydrogen depends critically on many factors whose interplay is poorly
understood. For cathodic charging of hydrogen upon a flat surface of iron
or steel (i.e. as distinguished from sharply recessed geometries such as
those of crevices or cracks), the pH of the solution, the applied potential,
the chemical composition of the aqueous phase in the near vicinity of the
metal, and the compositions of the metal and of the inclusions in it
control the rate of deposition of atomic H upon the interface, the rate
of recombination of the adsorbed H atoms to form H2 gas, and the
rates of the H (ads.) � H (dissolved) transitions (64, 65). The net resultant
of these steps determines the rate at which dissolved hydrogen appears
at the first, subsurface layer of metal atoms. This rate determines the
hydrogen activity, ab which sets the boundary condition for the diffusion
of hydrogen into the bulk of the metal. It is evident that in this case,
unlike the case where Hz gas is the environing phase, Ql cannot be
calculated from thermodynamics even for the situation of isoactivity of
hydrogen within the metal.
When the metal-aqueous solution interface has crevices, pits, or cracks,
the situation is further complicated. Neither the chemistry (66, 67) of the
aqueous phase within the crevice, nor the relevant electrical potential (68)
across the double layer at the metal at the root of the crevice, is related
in a simple way to the parameters that can be controlled and measured
outside the crevice. Indeed, Pickering & Frankenthal (69) demonstrated
cathodic evolution of hydrogen at the base of pits when the external
surface was under anodic potential. It follows that when corrosion
reactions produce pits and crevices, hydrogen may be generated within
HYDROGEN EMBRITTLEMENT 333
them at the same time that anodic dissolution is itself continuing the
pitting (70), and the a1 of the hydrogen there produced cannot be related
easily to the controllable parameters.
Cathodic charging and corrosion reactions are the techniques com­
monly used for presenting hydrogen to specimens being tested for
resistance to hydrogen embrittlement; it is easy thereby to introduce
hydrogen just below the first atomic layer of metal atoms at very large
thermodynamic activity, at. which may be thought of as a measure of the
driving force with which the dissolved hydrogen can enter into chemical
or physical reactions. It is important therefore to have even an approxi­
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

mate measure of a1 for any particular combination of steel, electrolyte,


temperature, and electrical potential. This may be obtained (32, 71, 72)
by measuring the steady-state permeation, J00' produced by the desired
electrolyte, electrical potential, and temperature upon the desired steel
by Cornell University on 06/09/12. For personal use only.

in the form of a thin sheet. The measurement may be carried out con­
veniently by the double-cell electrochemical technique of Devanathan &
Stachurski (73), in which the output surface of the steel sheet is palladized
to decrease impediments to the exit of hydrogen. The hydrogen at the
input surface may be deposited potentiostatically or galvanostatically.
The input surface of the permeation sheet should be as similar as possible
in all characteristics to the surface of the specimen for hydrogen­
embrittlement studies.
The sub�urface (i.e. below the first atomic plane) concentration, C1> of
lattice-dissolved hydrogen may be calculated from JooLjDLA, where L
and A are the thickness and permeation area of the sheet, respectively,
and DL is the diffusivity of hydrogen at the relevant temperature and in
the absence of any energetic sites at which hydrogen may be trapped.
The subsurface concentration C1 is proportional to the thermodynamic
activity at. the constant of proportionality depending on the arbitrarily
chosen reference state. The subsurface concentration is related to the
fugacity, 1H2, of gaseous hydrogen with which the subsurface layer would
be in equilibrium through a Sievert's law parameter. This is therefore
the effective fugacity, Ie, of the hydrogen-producing medium. An alterna­
tive method of calculation is let JooL/cPA, where ¢ is the permeability
=

of the particular steel extrapolated to the temperature in question from


the temperature range where reliable measurements are available.
These approaches are directly applicable to charging with a dilute
atomic hydrogen gas, produced for example by a hot-wire technique at
a specified hot-wire temperature, geometrical arrangement, steel surface,
and gas pressure and composition. They are also applicable to the
determination of some average al over the entire surface of a sheet
specimen when hydrogen is generated by corrosion. Of course, such a
334 ORIANI

Table 1 Fugacity-pressure relationship for


gaseous hydrogen at 25°C (54)

Fugacity (atm) Pressure (atm)

28.1 27.7
247 217
896 609
1860 987
3060 1315
6880 1955
1.25 x 104 25 10
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

1.13 X 105 5045


1.67 X 106 8990
1.01 X 107 1.20 x 104
1.66 X lOB 1.73 X 104
by Cornell University on 06/09/12. For personal use only.

1.22 X 109 2.14 X J04

determination will not be easily related to the local ai at the base of a pit
or a crack.
As described above, al is rdatable to the effective input fugacity of
molecular hydrogen. To know the maximum hydrostatic pressure that
can be generated by H-atom recombination in a void within a steel
exposed to a given value of fe in the absence of any relaxation processes,
one must know the pressure-fugacity relation for hydrogen (54, 74). This
is given in Table 1 for 25°C, We note that fH2 � PH,' The H2 gas pressure
within a cavity or microcrack in the metal cannot be larger than the
PH2 that corresponds to ai'
INTERNAL MOBILITY OF HYDROGEN In the absence of chemical and
structural features in a-iron with which dissolved hydrogen interacts
attractively, the diffusivity (7 1 , 75), about 5 x 1 0-9 m 2 jsec at 25°C, is so
large and the temperature dependence so small that the usual interpreta­
tion of the latter as an activation energy of 8.0 kJ/mole H (7 1 ) with
which an energy barrier is surmounted becomes suspect. Flynn &
Stoneham (76) have considered alternative conceptual formulations for
the motion of light interstitials in metals.
Attractive interactions between various chemical and structural
singularities and dissolved hydrogen slow down its overall migration in
iron alloys, as Darken & Smith (77) first pointed out. Treatments of
this effect, to various degrees of approximation and generality, have been
carried out by them, by McNabb & Foster (78), Oriani (7 1 ), Ellerbrock
et al (79), Koiwa (80), Allen-Booth et al (81), and Pressouyre & Bernstein
(82). Because steels have many singularities with which hydrogen can
HYDROGEN EMBRITTLEMENT 335

interact and which can interact with each other, the analysis of a diffusion
experiment to give information on the important interacting features is
extremely difficult.
In 1951 Bastien & Azou (31) proposed that hydrogen may be trans­
ported via entrainment by dislocations moving in response to applied
stresses, and that hydrogen dragged in this manner would move faster
and over longer distances than it could be diffusion motivated solely
by concentration gradients. Experiments by Frank (83) with mild steel, by
Broudeur et al (84) with 1 8-8 stainless steel, and by Louthan et al (85)
with iron, 304L stainless steel, and nickel, leave no doubt that hydrogen
entrainment by moving dislocations can indeed occur.
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

The emergence of dislocations at a surface, or their initiation at the


surface and progression into the interior, may also have the effect of
removing surface impediments to the ingress or exit of hydrogen. This
by Cornell University on 06/09/12. For personal use only.

is what appears to have occurred in the experiments of Fricke et al (86),


who deformed a mild steel bar under 1 50 atm of hydrogen gas. No
uptake of hydrogen occurred without ongoing plastic deformation, but
the absorption with plastic deformation could be well accounted for
solely by ordinary diffusion. Why the dislocation-drag effect was not
observed needs investigation.

PRESSURE BUILD-UPS BY ANNIHILATION OF HYDROGEN-LOADED DISLOCA­

TIONS The Bastien-Azou suggestion (3 1 ) that moving dislocations can


drag hydrogen and upon their annihilation deposit it to produce sub­
stantial hydrogen concentrations has been developed recently (85, 87-89)
as a basis for the generation of very large H2 gas pressures within
microvoids. Such pressures, far larger than that which would thermo­
dynamically correspond to the input activity at. would aid the growth
of preexisting microvoids and thereby contribute to hydrogen embrittle­
ment. Tien et al (89) have attempted a quantative evaluation of the
dynamically generated pressure from hydrogen dumped by a dislocation
into a microvoid. Johnson & Hirth (90) showed that the k inetic super­
saturation that is generated by the annihilation of hydrogen-charged
dislocations, and that is mitigated by the diffusing away of the hydrogen,
is very small in all realistic situations. It follows that the internal pressure
produced when hydrogen-charged dislocations are annihilated through
interception by a microvoid is negligible. Tien et al (89) consider that
the charged dislocations are intercepted by an inclusion of large volume
Vi to which a microvoid of volume v" is attached. In this way they obtain
a nominal amplification factor of Vi/v" over the situation in the absence
of the inclusion. Whether or not this amplification factor is an effective
one depends critically on the kinetics of diffusion of hydrogen along an
336 ORIANI

internal solid-solid interface compared with the very rapid diffusion away
through the matrix surrounding the inclusion. The dynamically generated
internal pressure does not seem important as a contributor to the
phenomena of hydrogen embrittlement.
Yu & Li (9 1 ) developed the idea that a stress-induced, large con­
centration of dissolved hydrogeh becomes supersaturated when the stress
field is suddenly relaxed by the formation of a separation at an inclusion­
matrix interface. The supersaturated, dissolved hydrogen enters the micro­
void formed and exerts a large pressure, causing the microvoid to grow.
The diffusion and recombination to form molecular hydrogen must
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

compete with diffusion away from the locale of the supersaturated


hydrogen; considerations such as those of Johnson & Hirth (90) would
show that the sequence suggested by Yu & Li probably makes a negligible
contribution to hydrogen emb rittlement.
by Cornell University on 06/09/12. For personal use only.

Physical Effects of Hydrogen in Iron and Steels


HYDROGEN CHARGED IN THE ABSENCE OF EXTERNALLY APPLIED STRESS
Raczynski cathodically charged hydrogen into high purity, zone-refined
iron wire, and followed the length change thereby induced (M. Smialowski,
personal communication, 1 975). He found that upon discharging the
hydrogen from the iron the original length was regained even when the
effective input fugacity was about 106 atm. Furthermore, the partial molal
volume calculated from the measured length change and hydrogen con­
centration, on the assumption of isotropic expansion of the wire, was
2.0±O.05 cm3/mole H, in agreement with earlier measurements (3�36).
This work demonstrates that very pure iron can dissolve hydrogen at
extremely high concentrations (relative to that characteristic of iron under
1 atm of H2 gas) without producing either dislocations or internal voids.
However, decreasing the purity of the iron may cause irreversible effects.
Kufukadis & Raczynski (92) used the deflection of a metal ribbon charged
with high-fugacity hydrogen at only one side to demonstrate that with
very pure zone-refined iron the deflection caused by the charging was
completely removed by the discharging of the hydrogen. However, with
Johnson-Matthey iron the deflection caused by charging was not removed
by discharging, and with Armco iron the irreversible deflection caused
by the charging was even greater.
Another effect of decreasing the purity of the iron was found by
Cornet et al (93), who charged in hydrogen cathodically, and after
removing the hydrogen by degassing at 1 00°C proceeded to strain the
specimens to failure. Pure, zone-refined iron broke at a constant value of
strain, a value independent of the input fugacity during the original
charging, which reached over 107 atm. However, the hydrogen-freed
HYDROGEN EMBRITTLEMENT 337
Armco iron brokc at a strain that decreased as the fugacity of the
original charging was increased beyond an undetermined value ; at lower
fugacities the strain at failure of the degassed iron was not affected. Thus,
we see that permanent damage caused by hydrogen occurs only beyond
some threshold value of effective input fugacity, and that this value
becomes unattainable when the iron is sufficiently pure. It would seem
that the impurity responsible for causing Armco iron (40 ppm C, 890 ppm
0, 260 ppm S) to retain after degassing a residual effect of the hydrogen
charging is not the carbon, since Cornet (94) found that zone-refined
iron with 630 ppm of additional carbon did not show any residual effect
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

of prior hydrogen charging. Cornet (94) also found that an iron showed
a residual effect of hydrogen (strained after removal of the hydrogen) only
when the permeation transient determined on that iron exhibited
anomalies.
by Cornell University on 06/09/12. For personal use only.

In the case of zone-refined iron even the plastic deformation involved


in the region of the failure is not enough (93) to cause microvoids to
be formed by high-fugacity hydrogen charging. However, a more thorough
study of this point by Raczynski (95) has shown that there is a threshold
value of deformation above which microvoids will be formed by subse­
quent hydrogen charging, and that this threshold value depends on the
effective hydrogen fugacity employed and on the purity of the iron.
Raczynski found that microvoids are formed in zone-refined iron de­
formed by 89% and then charged at Ie � 4 X 106 atm, whereas micro­
voids do not form in such iron when deformed to 75% even with
subsequent charging at Ie> 107 atm. On the other hand, Armco iron
cold-worked to a similar extent and then hydrogen charged exhibits
microvoids at Ie :=::; 1700 atm, and similarly, mild steel at Ie :=::; 3800 atm.
Beck et al (96) also observed threshold fugacities of about 2000 atm,
above which hydrogen permeation experiments display anomalous tran­
sient effects that were attributed to hydrogen-produced internal damage.
Bernstein (97) showed the strong effect of the degree of purity and of
the metallurgical heat treatment of the iron upon the frequency and
location (trans- or inter-granular) of microvoids1 produced by cathodic­
ally charged hydrogen. Using Ie values of the order of 106 atm, he found
that the number of trans granular microvoids produced per unit area with
continued charging time reached plateau values that increase in magnitude
with decreasing levels of interstitial solutes. Rath & Bernstein (98) found
that with very-high-fugacity hydrogen in irons of various interstitial solute
contents, grain boundaries exhibited large differences in the resistance
1 In this review the term microvoid does not carry an implication of a specific shape

or configuration. A micro void may be lenticular, crack-like, etc, depending on the manner
of its origin.
338 ORlAN I

to formation of intergranular microvoids according to the orientation


and type of boundary. Boundaries with angular orientations of up to
about 20° did not develop microvoids, irrespective of the boundary type
or of the common axis of rotation. The fact that hydrogen introduced
by cathodic charging or by corrosion can produce voids and blisters
has been known for a long time (99), and provided the impetus for the
planar pressure theory (13) of embrittlement. The existence of concurrent
plastic deformation has been studied in some instances (100, IOn
Several important lessons may be drawn from this section. Certain
impurities are usually needed in iron for the formation of microvoids
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

with even very high-fugacity hydrogen. Grain boundaries and/or cold­


work are not of themselves sufficient, but must interact somehow with
impurities to permit microvoids to form. In the absence of prior cold­
work, high-fugacity hydrogen can cause the growth of preexisting,
by Cornell University on 06/09/12. For personal use only.

extremely small microvoids by the action of reassociated molecular


hydrogen at a pressure consistent with the fugacity ; the growth of these
causes plastic deformation. In the case of prior cold-work it appears
likely that microvoid nucleation could, upon charging with hydrogen,
occur also via the decohesioning effect of dissolved hydrogen at points
of large stress produced by dislocation interactions and by plastic in­
compatibilities at second-phase particles and at regions of enriched solute
concentrations. Much more work is needed to understand the roles of
second-phase particles, of impurities and their mutual interactions, and
of their interactions with grain boundaries. Prior hydrogen charging
seriously decreases the subsequent strain at fracture in a hydrogen-free
condition if the charging had produced microvoids. We agree with Asano
& Otsuka (102) that charging damage is a serious source of inconsistency
and confusion in much of the existing literature on hydrogen embrittle­
ment.

EFFECTS OF HYDROGEN ON THE PLASTIC PROPERTIES OF IRON AND STEEL


One of the identifying characteristics of the hydrogen embrittlement of
steels is that in the presence of hydrogen the stress-strain curve is
terminated by fracture at a total strain smaller than that attained in the
absence of hydrogen. This has been observed repeatedly for many kinds
of steels. For example, Toh & Baldwin (103) found the fracture strain,
Sf, of SAE 1020-steel to be markedly reduced by hydrogen with a non­
monotonic dependence on temperature, and Grant & Lunsford (104)
found that sf for the same steel, but cold-worked, was increasingly reduced
as the amount of charged-in hydrogen was increased. The same effect
of increasing hydrogen content on 6f was found by Hobson & Sykes ( 105)
for 3% Cr-Mo steel. Frohmberg et al (106) found a marked diminution
caused by hydrogen in 4340 steel of 1860 M N/m2 yield strength.
HYDROGEN EMBRITTLEMENT 339
The same phenomenon is observed in irons of varying purity. For
example, 8f was found to be reduced by hydrogen for Armco iron (94, 107),
electrolytic iron and decarburized mild steel (102), coarsely polycrystalline
iron of about 30 ppm nonmetallic impurities ( 108), and coarsely poly­
crystalline iron of extremely high purity, about 0.3 ppm carbon (109).
It is important to note that the experiments recounted above, as
examples from a large literature, were all done with cathodically charged
hydrogen under conditions that we estimate to have produced input
fugacities of 106 atm or more. We have discussed the internal damage
that charging at such fugacities generates, except in the rare instances
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

indicated. Therefore it is of interest to note that Sf can be markedly


diminished by low-fugacity hydrogen also. Hoffmann & Rauls ( 1 10)
found that 150 atm of gaseous hydrogen decreased the reduction in area
(RA) at fracture of 0.22% C normalized steel of tensile strength 530 MN/m 2
by Cornell University on 06/09/12. For personal use only.

as unnotched specimens, and also that of various austenitic stainless steels.


Walter & Chandler (111) observed a decrease in RA at fracture under
68 atm of gaseous hydrogen of unnotched bars of ASTM A-302 steel,
and even 1 atm was enough to show an effect on 8f. Indeed, the reduction
in RA was found to be proportional to the half-power of the hydrogen
gas pressure. Bernstein ( 1 1 2) found that 0.7 atm of H2 gas diminished
the plastic strain at fracture of purified Fe-3%Si alloy from 26% to 1 6%.
Kamdar ( 1 13) obs.erved a linear relationship between the fracture stress
and the reciprocal of the square root of the grain size for Fe-3%Si alloy
tested in 0.7 atm of gaseous hydrogen. Even a material as soft as Armco
iron suffers a decrease ( 1 10) of I:f under Hz gas pressures as low as
2 atm. Because the internal pressures generated, either thermodynamically
or kinetically, under such low external fugacities are completely negligible,
it is clear that the internal pressure due to H 2 -molecule formation at
high external fugacities is not of itself the essential factor in the hydrogen
embrittlement of either high- or low-strength steels, although it furnishes
an additional factor that must be considered where high-fugacity hydrogen
is employed.
The effect of charged-in hydrogen or of a hydrogen environment on
the deformation resistance of iron and steels is very complex, and the
results of experiments often seem mutually contradictory. For example,
Seabrook et al (1 1 4) observed that heavily Charged 1020-steel exhibits
a true stress-true strain curve when tested after charging that lies at higher
stress values than that for a lightly charged sample. However, Grant &
Lunsford ( 104) found that prior cold-work reversed the effect of hydrogen
for the same steel. Schultz & Robertson ( 107) observed a lowering of
the stress-strain curve, relative to the hydrogen-free specimen, of Armco
iron held for 1 hr in an aqueous solution of H 2 S and strained while
therein immersed. However, increasing the time of prior immersion to
340 ORlANl

24 hr raised the stress-strain curve. The yield strength of 304L stainless


steel is lowered by high-pressure hydrogen ( 1 1 5). Beachem (30) found
that for 1020-steel considerably less applied torque was needed to achieve
the same strain when the steel tubes were twisted during high-fugacity
charging than when twisted in air. Asano & Otsuka ( 102) noted that
an increase in flow stress was produced in a pure iron single crystal,
in a decarburized mild steel (5 ppm C+N), and in a ferritic 1 8% Cr
steel upon initiating charging with high-fugacity hydrogen after 8% plastic
strain in the absence of hydrogen. They also found that the increase in
flow stress itself increased with increasing charging current, which is to
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

say, increasing input fugacity. Kusch ( 1 1 6) also found larger flow stresses
when iron single crystals ( 1 0 to 20 ppm C, N < 5 ppm, 0 < 5 ppm,
Si ;:;;; 100 ppm) were strained under high-fugacity hydrogen. On the other
hand, Lee et al ( 1 1 7) observed that hydrogen caused a lowering of the
by Cornell University on 06/09/12. For personal use only.

stress-strain curve of spheroidized 1090-steel, as well as the production


of larger plastic zones in bent specimens.
These diverse results on the effect of hydrogen on the resistance to
deformation of irons and steels were all obtained using hydrogen of very
high fugacity, the value of which was not measured by the investigators
but which we estimate to have been 106 atm or higher. It is instructive
to consider experiments in which either the input fugacity was so low
that an internal pressurization effect was negligible or an especially high
purity precluded the formation of internal cavities by high-fugacity
hydrogen. For example, the stress-strain curve was lowered for high­
purity Fe-3 %Si alloy under 0.7 atm H2 gas (112). Matsui et al (109)
observed a large lowering of the stress-strain curve of extremely pure
(0.3 ppm C) polycrystalline iron when strained under very-high-fugacity
hydrogen (probably above lOb atm) at temperatures between 200 and
294K. The softening caused by hydrogen disappeared when the hydrogen
was removed, and reappeared when it was reapplied. This reversibility
is evidence that internal cavities or microcracks, which would reduce the
effective cross-section, were not formed. The same investigators found
that the same charging conditions upon a less pure iron produced
hardening at 277 and 273K, but softening was obtained between 250 and
200K. It may be significant that for the less-pure iron blistering was
observed, but unfortunately no information was given about a possible
correlation between a hardening effect in the stress-strain curve and the
generation of microvoids by the high-fugacity hydrogen. The softening
produced by the hydrogen was more marked as the purity of the iron
increased. The authors (109) attributed the softening to a lowering of the
frictional stress of a dislocation by the large concentration of hydrogen
at its core.
HYDROGEN EMBRITTLEMENT 341
Softening effects were also observed by Lunarska (108) on an iron
with about 30 ppm of nonmetallic impurities when charged with very
high-fugacity hydrogen but without evidence of microvoid generation.
The charging-in of hydrogen during a stress-relaxation test sharply
increased the rate of load drop compared to that prior to hydrogen
input. Analysis of the data according to the method of Gupta & Li (118)
yielded the result that the hydrogen reduced the internal stress, (fi, as well
as the dislocation velocity exponent, m*. The magnitudes of these reduc­
tions increased with increasing input-hydrogen fugacity for specimens
initially strained to necking. With increasing initial strain, eo, the
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

hydrogen-induced reductions in both (fi, and m* at first increased sharply


and then stayed constant to C;o 0.2.
=

A decrease of the resistance to d islocation motion within grain interiors


produced by hydrogen was also found by Bernstein (119), using both
by Cornell University on 06/09/12. For personal use only.

decarburized Ferrovac-E iron (30 ppm C, 50 ppm 0, 10 ppm N) and


Fe-O.15%Ti in which the titanium served to remove the interstitial solutes
from solid solution. Bernstein measured the Hall-Petch parameters before
and after hydrogen charging at very high fugacity without, however, any
evidence of microvoid formation. The parameter (fo was lowered by
hydrogen, whereas the slope of the Hall-Petch curve was raised in the
decarburized iron and was not changed in the Fe-0. 15%Ti alloy. For
the former material, the results indicate that hydrogen can decrease the
overall resistance of the lattice while increasing the grain-boundary
resistance. Bernstein's results contrast strongly with those of earlier work
by Adair (120) on zone-refined iron (10 ppm C, 1 ppm N, 22 ppm 0).
Charging with very-high-fugacity hydrogen raised the (fo parameter and
lowered the slope of the Hall-Petch curve, leading to conclusions opposite
to those of Bernstein. However, Adair observed that surface blistering
and some internal fissuring resulted from the hydrogen charging, and it
may be surmised that the microvoids thus produced affected the Hall­
Petch parameters independently of the action of hydrogen per se. It may
be significant in this connection that the Hall-Petch curves of hydrogen­
charged iron and of iron that had been hydrogen charged and subsequently
discharged prior to testing were essentially identical (120). How micro­
voids per se can affect the Hall-Petch parameters is a matter to be
investigated.
The effect of hydrogen upon the plastic properties of iron has been
most extensively investigated at Vitry (93, 94, 121-123). Using zone­
refined iron (7 ppm C, 13 ppm N, 13 ppm 0), both mono- and poly­
crystalline, as well as Armco iron, the Vitry group has established the
complex nature of the effect 'of hydrogen. In the domain of low plastic
extension (10 � 2%), hydrogen of very high fugacity lowers both the flow
342 ORIANI

stress and the rate of work hardening, do/de, of zone-refined iron. How­
ever, in the domain of larger plastic extensions, hydrogen increases both
the flow stress and do/de. Working with polycrystals, Gourmelon ( 1 2 1 ,
1 22) decomposed the flow stress into the thermal and athermal com­
ponents by a procedure that implies identity between the effective stress
(that which is effective in causing motion of dislocations) and the thermal
component. The result of this analysis is that for low e the thermal
component is raised by hydrogen but the athermalls strongly decreased,
whereas at larger values of e, where multiple slip becomes important,
both components are raised by hydrogen. These studies were comple­
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

mented by investigations ( 1 23) with transmission electron microscopy of


the change in dislocation arrangements induced by hydrogen. Ih general,
the presence of high-fugacity hydrogen produces a marked increase of
interactions among the dislocations. Helical dislocations and dislocation
by Cornell University on 06/09/12. For personal use only.

tangles elongated along the (100) appear at e 2% in the presence of


=

hydrogen but are totally absent in the absence of hydrogen. Strains of


1 7% generate dense dislocation tangles in the presence of hydrogen but
not in its.absence.
One interpretation (123) of these results is that the stacking-fault energy
of bcc iron is lowered by hydrogen, as is known to occur for fcc iron­
alloys (49). Therefore cross slip of screw dislocations is reduced, with a
consequent lowering of the mean velocity of the dislocations and a greater
mutual entanglement in the domain of multiple slip. The authors ( 123)
point out that the lowering of the stacking-fault energy by hydrogen is
consistent with the hypothesis that hydrogen lowers the attractive inter­
action between iron atoms (106) because the hydrogen about a dislocation
would favor the displacement of atoms at its core. However, it is difficult
to believe that hydrogen can produce a lowering of the reputedly very
large (no measurements exist) stacking-fault energy of bcc iron sufficient
to perceptibly reduce the ease of cross slip. [See the negative results
for the dissociation of screw dislocations by N in tantalum ( 1 24).] The
same lowering of the stacking-fault energy by hydrogen is advanced ( 125)
as the basic reason for the softening produced by hydrogen in the easy­
slip regime. The argument is that hydrogen blocks sources of dislocations
on some of the slip planes. The suppression of the number of active slip
planes leads to less interference with dislocation motion on the remaining
slip planes. It must be admitted that the physical basis for this argument
is somewhat obscure. The production of helical dislocations in the
presence of hydrogen is thought (123) to be due to a stabilization of
vacancies by hydrogen so that the climb of the dislocations is favored.
The material outlined in this section furnishes ample evidence that
hydrogen markedly affects the plastic response of iron and steels in various
HYDROGEN EMBRITTLEMENT 343
and contrary ways. The only common phenomenon is that hydrogen
decreases the strain at fracture. It is clear that the elucidation of the
phenomena will require very carefully controlled addition of solutes, both
singly and in combinations, the distribution of which among micro­
structural features must bc determined. In addition, experiments must
be done over ranges of grain size, of strain, of strain rate, and of controlled
and known hydrogen fugacity, care being taken both to avoid complica­
tions due to kinetic impediments to hydrogen entry and to separate out
effects due to void growth by internal pressurization.

Hydrogen-Assisted Fracture
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

EXAMINATION OF THE FRACTURE SURFACE The technique most frequently


applied to the study of hydrogen-assisted fracture is fractography.
Bernstein & Thompson (126) have emphasized the observed diversity
by Cornell University on 06/09/12. For personal use only.

of fractographic responses to hydrogen; indeed, it is very difficult to


deduce the occurrence of hydrogen embrittlement from fractographic
appearance ( 127). For very-high-strength steels one can probably assert
that hydrogen produces a decrease in the prevalence of ductile processes
in comparison to fracture without hydrogen. This seems to be a conse­
quence of the lowered stress intensity parameter, K, needed to produce
cracking in the presence of hydrogen, since it has been shown that in
hydrogen-assisted fracture of 4340-steels (30, 128) and maraging steels
( 129), decreasing K is accompanied by a transition from dimpled to
cleavage to intergranular mode. Gerberich et al ( 130) also found an
increasing amount of ductile component with increasing K and con­
comitant crack velocity in the fracture of cathodically charged 4340-steel,
and Oriani & Josephic (26) made the same observation for increasing K
at constant crack velocity. Thompson (131) noted that high-fugacity
charging of PH 13-8%Mo steel (yield strength 1500 M N/m2) increased
the areal extent of the brittle-appearing fracture surface to 60% from
the 40% characteristic of the steel broken in air. However, Gangloff &
Wei ( 132) did not observe any marked effect from increasing K for the
hydrogen-assisted cracking of 18 % Ni maraging steels. Nevertheless, there
is a general trend that hydrogen reduces the ductile contribution by
reducing the K needed for fracture.
In softer steels many instances have been observed in which hydrogen
does not change the ductile-dimple character of the fracture, despite an
accompanying significant decrease in the RA at fracture. Thompson &
Brooks ( 133) found an inverse correlation between the loss in RA at
fracture produced by moderate-fugacity hydrogen on a series of modified
A-286 stainless steels (austenitic) and the reduction in mean dimple size.
The greater loss of ductility at the diminished dimple size was interpreted
344 ORIANI

as due to enhanced nucleation of voids at inclusions, which in turn was


related to increased misfit between the austenite matrix and the y' preci­
pitate. Garber et al (1 34) found that the fracture of near-eutectoid,
spheroidized steel charged with high-fugacity hydrogen was characterized
by larger ductile dimples along with large loss in ductility at fracture.
These results were interpreted as showing that hydrogen aided the growth
of voids at inclusions, probably by the large internal gas pressure
generated by the high charging fugacity.
Thompson ( 1 35) has plotted much of the available data of this sort
in terms of RA-loss vs the ratio, r, of mean dimple size for fracture with
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

hydrogen to that without hydrogen. The plot has two branches. One
branch, with r ranging from unity to 0.2, represents data for fcc iron
and nickel alloys with hydrogen at moderate values of supersaturation.
by Cornell University on 06/09/12. For personal use only.

The other branch, from unity to 1.6, has only two data points, one from
(127) with very-high-fugacity hydrogen, and one for Ni-2Th0 2 from un­
published work by Thompson. The branch for r < 1 can be un­
ambiguously ascribed to a nucleation enhancement of microvoids by
hydrogen, presumably by hydrogen's decreasing the force needed to
rupture bonds across the inclusion-matrix interface. The branch for
r > 1 must be ascribed to an enhancement of void growth by hydrogen,
probably by the internal pressurization effect referred to by Garber et al
( 134). It would clearly be important, because unexpected, to find cases
where r > 1 in the absence of any possibility of an internal pressurization
effect.
Small amounts of impurities can change the path of hydrogen-assisted
cracking. Cornet (94) found that carbon added to very pure zone-refined
iron under externally applied stress and with cathodically charged
hydrogen changes the crack path from intergranular to transgranular.
The same behavior has been found for nitrogen ( 122). Bernstein (97)
found a similar effect of carbon upon the crack path of purified irons
subjected to high-fugacity charging without an externally applied
mechanical load. He also found that the crack path could be varied
reversibly from inter- to transgranular by suitable variation of the thermal
history or of the concentration of interstitial solutes, and that for any
one specimen the crack path under hydrogen charging was the same as
that observed after low-temperature tensile straining in the absence of
hydrogen. Addition of silicon to an Fe-C alloy inverts the intergranular­
transgranular crack-path transition caused by carbon (127). In a high­
strength Cr-Mo alloy (1 36), increasing the amount of dissolved nitrogen
changes the predominant mode of failure under tensile loading with
cathodically charged hydrogen from trans- to intergranular.
It is often possible by suitable thermal treatments to produce cracking
HYDROGEN EMBRITTLEMENT 345
paths that are the same both for low-temperature straining or impact
testing without hydrogen and for hydrogen-assisted cracking at room
temperature. Bernstein et al (127) used the same heat treatments on Sb­
doped Fe-C alloy as those used on it by Rellick & McMahon ( 1 37)
and found that high-fugacity hydrogen, without externally applied load,
produced cracks along the same loci, different for each heat treatment,
as those produced by low-temperature tension ( 137). Similarly, Yoshino
& McMahon ( 1 38) found that the heat treatment that causes HY 1 30
steel to embrittle and fail intergranularly by impact in the absence of
hydrogen causes that steel to undergo a change from trans- to inter­
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

granular when cracked in high-fugacity hydrogen. Briant et al ( 139) applied


Auger electron spectroscopy to the fracture surfaces of impact specimens
of HY 1 30 steel broken in air. They found that temper-embrittled
specimens exhibited segregation of Si, P, N, and Sn upon the inter­
by Cornell University on 06/09/12. For personal use only.

granular fracture surface, and that such temper-embrittled steel cracked


intergranularly at low stress intensities under about 2 atm of gaseous
hydrogen. The inference is therefore strong that the segregated impurities
that cause low impact resistance are also responsible for facilitating
hydrogen-induced cracking.
Similar work was carried out by Banerji et al (140) for 4340-type steels,
using both commercial and high-purity steels. They demonstrated by
Auger electron spectroscopy that the 500°F embrittlement of commercial
4340 (one-step temper embrittlement, or OSTE) is associated with segrega ­
tion of small amounts of P and of N to prior austenite grain boundaries
along which the OSTE fracture occurs. Because commercial 4340-steel
subjected to about 1 atm of gaseous hydrogen always (i.e. independently
ofOSTE) cracks intergranularly along prior austenite boundaries, whereas
high-purity N iCrMoC alloy (simulating the basic composition of 4340-
steel) cracks by cleavage and rupture under hydrogen and only at much
higher stress intensities and does not exhibit OSTE, the authors ( 140)
concluded that segregation of a small amount of P, and in some cases
also of N, was responsible for the great susceptibility of 4340-steel to
gaseous hydrogen.
Clearly, it would be highly desirable to apply Auger spectroscopy to
the fracture surfaces produced by hydrogen-assisted cracking. Wei &
Simmons (141) have shown the feasibility of this, and in this and in
later work (142) no segregation of solutes to prior austenite boundaries
was detected in 4340-steel. This is not necessarily in conflict with the
work of Banerji et aI, since Simmons et al did not temper-embrittle
the specimens prior to hydrogen cracking. Wei & Simmons (141) did
detect segregation of the principal alloying elements at the hydrogen­
induced fracture along prior austenite boundaries in 18 % Ni maraging
346 ORIANI

steel. Viswanathan & Hudak (128) also applied Auger electron spectro­
scopy to impact-fracture surfaces of 4340-type steel that had been temper
embrittled ; they inferred that the greater susceptibility to gaseous H 2 S
cracking of such steel was due to segregation of P and Ni.
Fractographic work has demonstrated that in many cases, especially
in lower-strength steels, hydrogen can have marked effects on macro­
scopic ductility without producing a qualitative change in the ductile
appearance of the fracture surface. The quantitative changes therein must
be investigated further. Impurities acting singly or synergistically can
change the crack path in iron and iron alloys exposed to high-fugacity
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

hydrogen. It is beginning to seem probable that th<;: same impurities


that produce temper embrittlement also aggravate the susceptibility of
internal boundaries to hydrogen, probably through an augmentation of
the decohesioning tendency of the internal boundary ( 143).
by Cornell University on 06/09/12. For personal use only.

KINETIC ASPECTS OF HYDROGEN-ASSISTED CRACKING From an experiment


involving the measurement of time it is in general very difficult to obtain
information on the mechanism of the essential embrittling action of
hydrogen. The many reasons for this difficulty are described in this
section.
A common measurement is that of the incubation time for the onset
of extension of a crack, put into a specimen by machining or by fatiguing
in air, after immersion in a hydrogen-yielding medium. Three factors
are involved in the observed incubation time : the kinetics of breakdown
of the air-formed film on the metal at the crack tip, the kinetics of the
permeation of this film by hydrogen, and the local crack geometry. The
commonly measured time-to-failure vs applied load usually contains a
large contribution from the incubation time. When the specimen has been
precharged either cathodically or by high-temperature exposure to
hydrogen gas, the time-to-failure may be determined by diffusion and
trapping phenomena ; the thermodynamic activity of the hydrogen at
the locus of critical action is therefore not known. Such experiments
demonstrate the existence of, but do not quantitatively characterize, the
threshold hydrogen activities below which a crack will not propagate
( 1 44). When one attempts to determine the value of mode-I, plane-strain
stress-intensity factor below which the growth of a machined-in or
fatigue-induced crack will not initiate in a hydrogen environment (a
quantity usually called KId, the result is a function not only of kinetics
but also of the character of the artificially developed crack, which in
general is different from that of a crack growing under the action of
hydrogen. The second variable may be avoided by a rising-load test (145)
in a hydrogen-yielding environment, provided that the load is not raised
HYDROGEN EMBRITTLEMENT 347
too rapidly ; but the activity of the hydrogen where the hydrogen is effective
is still controlled by kinetics and has never been determined in such
experiments.
Measurements of the crack-propagation velocity as a function of stress­
intensity parameter in hydrogen-bearing environments are valuable not
only for engineering purposes but also in the pursuit of insight into
which portions of the overall transport process determine the kinetics.
For the latter application the thermodynamic activity of the hydrogen
must be known and controlled over the ranges of tcmperature and
crack velocity investigated. This condition cannot easily be met with
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

cracked specimens in contact with aqueous systems, and it has not been
met with precharged specimens.
The most scientifically useful experiments are those done with gaseous
hydrogen, especially where great care has been taken with the purity of
by Cornell University on 06/09/12. For personal use only.

the gas. The work of Williams & Nelson (61) with fully hardened 4 1 30-
steel and Hz gas in the region of about 1 atm pressure demonstrated
for the first time that the logarithm of crack velocity, v, vs reciprocal
temperature, at constant K and Hz gas pressure, p, is an inverted U­
curve, and that the function v(p) has different forms in the various
temperature ranges of the U-shaped curve. This complex behavior has
been corroborated by others ( 1 32, 1 46-148), and was explained (61) in
terms of chemisorption kinetics. Although the specific model used by
these authors has been shown (149, 1 50) not to agree with the experimental
results, the idea that the rate-controlling step is at the gas-metal interface
is in all probability correct (24), and is greatly strengthened by the later
work of Nelson et al (62). These investigators found that with partially
dissociated hydrogen gas the log v(1/T) curve is a straight line continuing
to rise in the temperature domain in which with molecular hydrogen
there is a downturn in the log v(IIT) curve (61). Clearly, the dissociation
of the H2 molecule must be at least one of the kinetically controlling
steps in the higher-temperature branch of the U-shaped curve. The kinetic
details of the gas-to-subsurface process have yet to be experimentally
elucidated. The alloy composition will probably be found to affect the
various kinetic steps.
Simmons et al (142) found that the activation energy for crack growth
in 4340-steel, in which the environment in contact with the crack surfaces
is pure water, is the same as that for the reaction of water vapor with
the steel to form oxide, as followed by Auger electron spectroscopy. This
suggests that the oxidation reaction that produces the hydrogen respon­
sible for the embrittlement is the rate-limiting process in this case.
At constant environmental conditions the log v(K) curve is generally
found to be composed of three stages ( 1 39, 1 46, 1 5 1-1 54). At low K there
348 ORIANI

exists a threshold value of K below which the crack is immobile (i.e.


velocity is below a detectable value) and above which v increases very
rapidly with increasing K (Stage I). At larger values of K the cracking
velocity is much less dependent on K (Stage II), and in some cases is
virtually independent of K. At still larger values, the cracking velocity
again increases rapidly with K as unstable cracking in the absence of
hydrogen is approached (Stage III). As discussed earlier, the character of
the fracture surface of high-strength steels usually changes with increasing
K towards a larger contribution from ductile tearing. The bimodal manner
of crack propagation is another important reason that the theoretical
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

calculation of cracking velocity will be a formidable problem.


When cracking is observed with sufficient sensitivity, evidence is found
of intermittent crack propagation. Briant et al ( 1 39) observed dis­
continuous variation of crack length with time in HY 1 30 steel, but a
by Cornell University on 06/09/12. For personal use only.

smooth variation in high-strength 4340-steel was observed by Banerji


et al ( 1 40). Sonic emissions have been noted by many investigators ( 1 55,
1 56). Oriani & Josephic (22) found that the hydrogen-induced cracking
of high-strength 4340-steel was accompanied by sonic events whose
frequency varied approximately as vK6 They argue that sonic emissions
are a consequence of the ductile component of the bimodal cracking,
and that they do not carry information relative to the essential action
of hydrogen.
The kinetically determined threshold value of K, below which crack
growth is imperceptible with the technique employed, was found ( 1 39)
to decrease dramatically for HY 1 30 steel with increasing concentration
of impurities in the grain boundaries. Similarly, high-purity NiCrMoC
steel was found (140) to have a much larger threshold value of K than do
commercial 4340-steels when exposed to the same pressure of gaseous
hydrogen. Gerberich & Chen (1 57) have assembled kinetically determined
threshold Ks for a variety of steels, and have shown that despite the large
variety of methods of charging and of hydrogen fugacities there is a
definite trend for threshold K values to increase with decreasing yield
strength of the steel. For a more restricted set of 4340-steels of varying
heat treatment, the same workers (1 57) found a similar trend for cathodic­
ally precharged specimens, in which, however, the hydrogen activity was
not the same in all specimens. Loginow & Phelps (1 54) measured the
K (threshold) vs hydrogen pressure from 2 1 to 105 MPa of steels with a
variety of yield strengths. They found that the results for steels with
yield strengths between 586 and 793 M N/m 2 fell on one curve, and those
with yield strengths between 861 and 1 069 M N/m 2 fell on another ; each
plot of log PH2 vs K (threshold) was nearly linear. Gerberich et al ( 1 58)
found the same behavior for a 4340-steel of yield strength 1 3 50 MN/m 2 .
HYDROGEN EMBRITTLEMENT 349
From the foregoing we see that the evidence for the existence of a
threshold K for hydrogen-aided crack propagation is persuasive. We
observe that measurements of cracking velocities are not useful for eluci­
dating the basic mechanism of the action of hydrogen. They may be
usefully related to the kinetically controlling step in the delivery of
hydrogen to the region of action, but only if independent physico­
chemical measurements are also carried out. This has been accomplished
in only one instance (142). Kinetic measurements are also useful as a
measure of the effects of impurities and of alloying elements on hydrogen
embrittiement, but only if Ie is kept at a known constant value. An
explication of the role of impurities in sensitizing to hydrogen will require
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

further careful work. Finally, the attempt to calculate a cracking velocity


from a model appears futile at present.
by Cornell University on 06/09/12. For personal use only.

EQUILIBRIUM ASPECTS OF H YDROGEN-ASSISTED FRACTURE AND THE DE­

COHESION MODEL An immediate deduction from the decohesion model


( 1 , 24, 25) is that any one stress state should have associated with it a
value of hydrogen pressure below which it is impossible to have hydrogcn­
assisted cracking. Thus for a given cracked specimen, and characterizing
the complex inhomogeneous stress field by K, there should exist a set of
PH, -K points that forms the boundary between the domain of crack
mobility and that of crack immobility, that constitutes a curve of unstable
equilibrium character, and that is therefore treatable thermodynamically.
The predicted threshold curves were established in a preliminary way
for 1 8-Ni maraging steel of 250 ksi yield strength (1 59) and more com­
pletely for 4340-steel of 250 ksi yield strength with both hydrogen and
deuterium gases (26). The difference between the curves for the two
isotopes is ascribable to the differences between the partial molal volumes
of solution of the isotopes in the steel and to differences in the inter­
action with the metal.
The mathematical formulation of the model (26) may be closely fitted
to the experimental threshold-equilibrium p(K) curve. From this fitting,
numerical values may be obtained for the effective hydrostatic component
of stress, for the ratio of stress-enhanced hydrogen concentration to
normal lattice solubility, and for the ratio of hydrogen-reduced maximum
cohesive force to the maximum cohesive force in the absence of hydrogen.
All these quantities are in reasonable semi-quantitative agreement with
the premises of the model and vary with K in a reasonable manner.
The analysis has recently been improved (22) to consider the physically
necessary variation of effective crack-tip radius, p, with the applied K,
which is due to the plastic deformation field. The temperature dependence
of the equilibrium p(K) curve was also determined (22) in preliminary
3 50 ORIANI

measurements, and was found to agree roughly with that which may be
deduced from the theory. The temperature dependence is in qualitative
agreement with the finding of Gangloff & Wei (1 32) that a small increase
of temperature causes a crack in 1 8-Ni maraging steel to become static
and a small decrease in temperature causes it to propagate.
That there should be some unique relationship between the macro­
scopically measurable stress intensity parameter K and the maximum
stress in the region of the crack tip for a given steel is physically
reasonable for a slowly moving crack, as long as K is either constant
or is monotonically increased. The relationship will be different, however,
if the applied K (or load) is decreased, or if the crack has been put in
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

by machining or by fatiguing. An example of the former is the hysteresis


effect (22) : A small unloading makes necessary a hydrogen gas pressure
for resumption of crack propagation much higher than that required in
by Cornell University on 06/09/12. For personal use only.

monotonic loading. The necessary pressure can be predicted within about


10% from the decohesion theory (22) coupled with the assumption that
the effective crack-tip radius, p, characteristic of the higher K is not
changed by a small de<,:rease in K. This is tantamount to asserting that
the plastic deformation field established at the higher K is not changed
upon lowering K.
A specimen with a smoothly rounded notch will have a stress field
about the notch very different from that about an ongoing crack at the
same K. Since the value of K attained by monotonically increasing the
load should determine the stress state in a reproducible fashion (provided
there are no preexisting microcracks in the region of the notch), there
should also exist a threshold-equilibrium PH2(K) relation for crack initia­
tion in a steel, as may be inferred from the decohesion theory, although
it should be different from that for the ongoing crack in the same steel.
Oriani & Josephic (160) attempted to establish such a relationship for
4340-steel of 250 ksi yield strength. Crack initiation under about 1 atm
of hydrogen was found at nominal stresses above and below the gross
yield strength but with much scatter. The lack of success in finding the
expected p(K) functional for crack initiation in this high-strength steel
was ascribed ( 1 60) to the existence of microcracks generated by processes
prior to testing. It may be expected that such an experiment would be
successful in a lower-strength steel in which the high stresses required
for the decohesion are more reproducible from specimen to specimen
because they are generated by plastic phenomena that concurrently
mitigate the effect of preexisting microcracks.
The weakest portion of the formulation of the decohesion model is
the connection between the macroscopically defined quantity, K, and the
maximum stress at a very localized region. This connection was made
HYDROGEN EMBRITTLEMENT 351
(24) by means of Gilman's (161) modification of an Inglis-type expression
for a crack tip blunted by plastic deformation to a radius of curvature,
p, with the plastic response of the material and the stress history
controlling the magnitude of p. The success (22) of this formulation in
describing the unstable equilibrium p(K) curve and the hysteresis pheno­
menon may indicate tha,t in a very-high-strength steel the decohesion
does predominantly occur within atomic distances of the crack-tip surface
at regions of very large non-Hookean stresses (Elliott regions). The
theoretical work of Thomson (162) lends additional support to this point
of view. However, with weaker steels it is certain that the decohesion
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

predominantly occurs deep within the plastic enclave and that the
necessary non-Hookean stresses are generated via dislocation interactions.
How these large, highly localized stresses may be related to macro­
scopically measured parameters such as K is an unsolved problem in
by Cornell University on 06/09/12. For personal use only.

mechanical metallurgy.
Another weakness of the current expression of the decohesion theory
(22) is its inability to explicitly consider the role of interfaces in the
steel that are structurally weaker than the matrix, where hydrogen
aggravates an already vulnerable situation. The role of interfaces was
appreciated in the early formulation of the model (24), and their im­
portance has been seen in some of the experimental work described in
earlier sections of this review. However, an explicit, operationally useful
formulation of the role of the interfaces has yet to be incorporated into
the decohesion model.

GENERAL DISCUSSION AND CONCLUSIONS

The variety and complexity of the actions of hydrogen are responsible


for causing the history of the investigation of the hydrogen embrittlement
of steels to resemble the fable of the blind men and the elephant.
Investigators have tended to perceive only single aspects of the problem
and to design experiments in which important variables were either not
appreciated, not controlled, or not measured. Endeavoring to avoid a
similar myopia, I here assess the state of the field and try to indicate
likely fruitful avenues of future research.
That decohesion is the predominant mechanism by which hydrogen
embrittles high-strength steels seems at present to be most probably
correct, but the details of the roles of interfaces and of inhomogeneous
plastic deformation are obscure. To study the former, equilibrium PH -K
,
measurements on model steels as a function of temperature and of known
impurity content and distribution, coupled with Auger electron spectro�
scopy upon the surface of the hydrogen-induced fracture, would be helpful.
352 ORIANI

Crack initiation on smooth specimens, in which load is kept constant


and the hydrogen fugacity is very slowly increased, should be measured
as a function of impurity content and distribution ; coupled to these
experiments should be advanced acoustic techniques to locate the origin
ofthe crack, Auger spectroscopy, and metallography. In all cases, fracture­
mechanics design should be used, the input fugacity of the hydrogen
should be known and controlled, and sources of kinetic artifacts should
be avoided.
Cracking velocities in hydrogen-yielding gaseous environments should
be determined as a function of temperature, stress intensity parameter,
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

and hydrogen fugacity for a variety of high-strength steels and model


alloys, but they should be associated with measurements of the kinetics
of the various steps involved in the decomposition or reaction of the
hydrogen-bearing gaseous molecule and the formation of the metal­
by Cornell University on 06/09/12. For personal use only.

dissolved hydrogen. It would also be extremely valuable to employ some


form of surface analysis of very fine spatial resolution to investigate the
role of impurities in each of the modes of fracture, cleavage, and plastic
tearing, the ratio of which varies with magnitude of K for a given steel
and with yield strength for a given K.
To put the decohesion theory on a firm basis it would be highly
desirable to devise an experimental means for demonstrating and for
measuring the lowering by hydrogen of the maximum cohesive force
between atoms of iron and/or its alloying elements. This is a very
difficult task ; I have suggested one way in which this might be done
(25 ).
In lower-strength steels the role of decohesion as the predominant
mechanism by which hydrogen decreases the ductility is not so well
established. Arguing from the lessons learned with high-strength steels,
and applying the philosophical principle of Occam's razor, the working
hypothesis here recommended is that in soft steels hydrogen-induced
decohesion occurs at locations where very high stresses are developed
by a variety of mechanisms associated with inhomogeneous plastic
deformation ; the microcracks thereby generated grow by plastic mechan­
isms that may be affected by hydrogen and by hydrogen-assisted de­
cohesion, until plastic instabilities develop between the voids when their
mean separation reaches some critical value. Thompson ( 1 63) has expressed
a similar point of view. In addition, if Ie is large enough, internal
pressurization aids the growth of microvoids simply by adding to the
local stress.
In probing such a working hypothesis for the hydrogen embrittlement
of the lower-strength steels, one encounters the unsolved problems of
mechanical metallurgy in general : the magnitude of stresses built up at
HYDROGEN EMBRITTLEMENT 353

inclusion interfaces and grain boundaries because of plastic incompati­


bility across the interface, the description of these stresses in terms of
macroscopically measurable parameters, dislocation dynamics at second­
phase particles, the measurement of interface strength as a function of
its structure and of solute concentration at the interface, and the generation
of plastic instability. It may be feasible to evaluate interfacial strength
by the sophisticated use of hydrogen of known and controlled fugacity
in the absence of externally applied stress, and by applying metallographic
techniques along the lines of the work of Rath & Bernstein (98).
In order to understand the effect of hydrogen on embrittlement, it
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

seems essential to study the effect of hydrogen on each of the relevant


phenomena in isolation. Just how to do this is at present obscure. The
situation is so complex that all sources of additional complication should
be assiduously avoided in experimehts on hydrogen embrittlement. For
by Cornell University on 06/09/12. For personal use only.

example, unless it is the specific phenomenon under study, the internal


pressurization effect associated with hydrogen of high input fugacity
should be avoided. In general, j� should be controlled and measured ;
this requirement eliminates the use of pre cracked specimens in aqueous
environments until a reliable way is found to measure the input fugacity
at crack surfaces.
It would be helpful to develop experimental approaches in which
kinetic aspects (dislocation dynamics and/or hydrogen transport) are
minimized in order to increase one's knowledge of the local stress state
and local hydrogen activity. Because inhomogeneous plastic deformation
causes high stresses to develop, it is necessary to investigate the effect of
hydrogen on each of the various aspects of dislocation dynamics : disloca­
tion emission and absorption, kink formation and motion, glide, cross
slip, slip band formation, tangle formation, etc. The effect of hydrogen
must be studied in concert with the effects of other solutes. This is
obviously an extension of the current work on pure iron with controlled
solute additions described in an earlier section of this review.
Of value in the study of the gamut of steels would be the measurement
of the hydrogen population adsorbed at stress-free internal interfaces
of characterized structure and impurity-content. Such measurement would
require systems having predominantly one kind of interface and would
necessitate the use of equilibrium adsorption isotherm techniques or of
permeation transient measurements analyzed by a sophisticated trapping
theory. The latter would be useful also in the eventual attainment of a
very difficult goal : the understandin,g of the kinetics of crack propagation.
In this connection, the kinetically controlling steps of the processes at
the interface between the metal and the source of hydrogen must be
measured and understood.
3 54 ORIANI

Hydrogen embrittlement of steels is not only an important technological


challenge but also a most difficult theoretical problem. Its solution will
require the resources of many disciplines and must carry some of them
beyond their current states of understanding.

Literature Cited

1 . Blanchard, P., Troiano, A. R. 1960. Mem. Atomistic Mechanism of Fracture,


Sci. Rev. Metall. 57 : 409-22 Swampscott, 1 959, pp. 49g-523. New
2. Latanision, R . M., Operhauser, H. Jr. York : Wiley
1974. In Hydrogen in Metals, ed. I. M. 24. Oriani, R. A. 1972. Ber. Bunsenges.
Bernstein, A. W. Thompson, pp. 539- Phys. Chem. 76 : 848-57
42. Boston : Am. Soc. Met. 25. Oriani, R. A. 1977. Proc. Int. Can! on
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

3. Speidel, M. O. 1974. See Ref. 2, pp. Stress Corrosion Cracking and Hydroyen
249-7 1 Embrittlement of Iron�Based Alloys.
4. Patton, N. E., Williams, 1. C. 1974. See U nieux-Firminy, 1 973. Houston : Natl.
Ref. 2, pp. 409-3 1 Assoc. Corros. Eng. pp. 35 1-58 .
5 . Birnbaum, H . K., Grossbeck, M . 1 9 74. 26. Oriani, R. A., Josephic, P. H. 1974.
See Ref. 2, pp. 303--2 3 Acta Metall. 22 : 1065-74
by Cornell University on 06/09/12. For personal use only.

6. Ingram, A. G., Bartlett, E. S., Ogden, 27. Li, J. C. M., Oriani, R. A., Darken,
H. R. 1963. Trans. Metal/. Soc. A/ME L. S. 1966. Z. Phys. Chem. Frankfurt
227 : 13 1-36 49 : 27 1-90
7. Seelinger, S., Stoloff, N. S. 1 97 1 . Metall. 28. Rice, 1. R., Thompson, R. 1974. Phi/os.
Trans. 2 : 1481-84 Mag. 29 : 73--9 7
8. Gahr, S., Grossbeck, M. L., Birnbaum, 29. R ice, 1. R. 1 976. See Ref. 10, pp. 455--66
H. K. 1977. Acta Metal/. 25 : 1 2 5-34 ; 30. Beachem, C D. 1972. Metall. Trans.
Grossbeck, M. L., Birnbaum, H. K. 3 : 437-5 1
1977. Acta Metall. 25 : 1 3 5-48 31. Bastien, P., Azou, P. 1 9 5 1. C. R. Acad.
9. Van Fossen, R. H. Jr., Scott, T. E., Sci. Paris 232 : 1 845--48
Carlson, O. N. 1965. J. Less-Common 32. Oriani, R. A. 1969. Proc. Conf Funda­
Met. 9 : 437-5 1 mental Aspects of Stress Corrosion
10. Magnani, N. J. 1976. In Effect of Cracking, Ohio State University, 1 967,
Hydrogen on Behavior of Materials, ed. pp. 32--49. Houston : Natl. Assoc. Corros.
A. W. Thomson, I. M. Bernstein, pp. Eng.
189-99. New York : Metall. Soc. AI ME 33. da Silva, J. R. G., Stafford, S. W.,
1 1 . Birnbaum, H. K., Wadley, H. 1975. McLellan, R. B. 1976. J. Less-Common
Scripta Metall. 9 : 1 1 13-- 1 6 Met. 49 : 407-20
1 2 . Johnson, H. R., Dini, J . W . 1974. See 34. Oriani, R. 1966. Trans. Metall. Soc.
Ref. 2, pp. 325--43 AIME 236 : 1 368-69
1 3 . Zapffe, C, Sims, C 1941 . Trans. AI ME 35. Bockris, J. O'M., Beck, W., Genshaw,
145 : 225..59 M. A., Subramanyan, P. K., Williams,
14. de Kazinsky, F. J. 1954. J. Iron Steel F. S. 1 9 7 1. Acta Metall. 1 9 : 1209-18
/ nst. 177 : 85-92 36. Wagenblast, H., Wriedt, H. A. 197 1 .
1 5. Garofalo, F., Chou, Y. T., Ambegaokar, Metall. Trans. 2 : 1 393--97
R. 1960. Acta Metall. 8 : 504-- 1 2 37. Schenck, H., Lange, K. W. 1966.
16. Bilby, B . A., Hewitt, J . 1962. Acta Arch. E isenhuttenw. 9 : 739-48
Metall. I O : 58 7-600 38. Au, 1. 1., Birnbaum, H. K. 1978. Acta
1 7. Tetelman, A. S., Robertson, W. D. 1963. Metall. In press
Acta Metall. 1 1 : 415-26 39. Blythe, H. J., Kronmuller, H. 1971.
1 8 . Hancock, G. G., Johnson, H. H. 1966. Phys. Status Solidi 4A : 603-18
Trans. Metal/. Soc. A/ME 236 : 5 13-- 1 6 40. Laurent, J. P., Lapasset, G., Aucouturier,
19. Petch, N . J., Stables, P . 1 952. Nature M., Lacommbe, P. 1974. See Ref. 2, pp.
1 69 : 842-43 559-73
20. Petch, N. J. 1956. Phi/os. Mag. 1 41. Asaoka, T., Dagbert, C, Aucouturier,
(Ser. 8) : 331-37 M., Galland, J. 1977. Scripta Metall.
2 1 . Chornet, E., Coughlin, R. W. 1972. 1 1 : 467-72
J. Catalysis 27 : 246-65 42. Podgurski, H. H., Oriani, R. A. 1972.
22. Oriani, R. A., Josephic, P. H. 1977. M etall. Trans. 3 : 2055-63
Acta Metall. 25 : 979-88 43. Gibala, R. 1967. Trans. Metal/. Soc.
23. Friedel, R. A. 1 959. Proc. Int. Can! AIME 239 : 1574--8 5
HYDROGEN EMBRITTLEMENT 355
44. Asano, S., Hara, K., Nakai, Y., Ohtani, 70. Ateya, H. G., Pickering, H . W. 1975.
N. 1974. J. Jpn. lnst. Met. Sendai J . Electrochem. Soc. 122 : 1 0 1 8-2 6
38 : 626-32 7 1 . Oriani, R. A. 1 9 70. Acta Metall. 1 8 :
45. Kikuta, Y., Sugimoto, K., Ochiai, S., 1 47-57
Iwata, K. 1975. Trans. Iron Steel lnst. 72. Kumnick, A. J., Johnson, H. H. 1 975.
Jpn. 1 5 : 87-94 M etall. Trans. 6A : 1087-9 1
46. Zielinski, A., Lunarska, E., Smialowski, 73. Devanathan, M. A. V., Stachurski, Z.
M. 1 977. Acta Metall. 25 : 55 1-56 1964. J . Electrochem. Soc. 1 1 1 : 6 1 9-23
47. Gibala, R. 1 977. Hydrogen-Defect Inter­ 74. Bockris, J. O' M ., Subramanyan, P. K.
actions in Metals. Presented at Symp. 1 9 7 1 . Electrochim. Acta 1 6 : 2 1 69-79
Hydrogen Embrittlement and Corrosion 75. Kumnick, A. J., Johnson, H. H. 1974.
of Metals, Jablonna, Poland. Metall. Trans. 5 : 1 199-1 206
48. Atrens, A., Fiore, N. F., Miura, K. 1977. 76. Flynn, C. P., Stoneham, A. M. 1970.
J. Appl. Phys. 48 : 4247-5 1 Phys. Rev. I B : 3966-78
49. Whiteman, M. B., Troiano, A. R. 1964. 77. Darken, L. S., Smith, R. P. 1949.
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

Phys. Status Solidi 7 : KI09-12 Corrosion 5 : 1 - 1 6


50. Beeck, 0., 1 950. Adv. Catalysis 2: 1 5 1-95 7 8 . McNabb, A . , Foster, P. K . 1 9 6 3 . Trans.
5 1 . Bagg, J., Tompkins, F. C. 1955. Trans. Metall. Soc. AIME 227 : 61 8-27
Faraday Soc. 5 1 : 107 1-80 79. Ellerbrock, H. G., Vibrans, G., Stiiwe,
52. Porter, A . S., Tompkins, F. C. 1 953. H. P. 1972. Acta Metall. 20 : 5 3-60
by Cornell University on 06/09/12. For personal use only.

Proc. R . Soc. London A2 1 7 : 529-44 80. Koiwa, M. 1 974. Acta Metall. 22 : 1259-
53. Lunarska, E., Zielinski, A., Smialowski, 68
M. 1977. Acta Met all . 25 : 305-8 8 1 . Allen-Booth, D. M ., Atkinson, c., Bilby,
54. Baranowski, B. 1972. See Ref. 24, pp. B. A. 1975. Acta Metall. 23 : 37 1-76
7 1 4-24 82. Pressouyre, G. M., Bernstein, I. M. 1 978.
55. Smialowski, M., Szklarska-Smialowska, Corros. Sci. In press
Z., Janko, A. 1966. Omagiu Acad. Prof 83. Frank, R. C. 1 959. In Internal Stresses
Raluca Ripan (Roumania), pp. 541-46 and Fatigue in Metals, ed. G. M .
56. Szummer, A., Janko, A. 1 977. Low Rassweiler, W. L. Grube, pp. 4 1 1-24.
Temperature FCC Hydride Phase in New York : Elsevier
Stainless Steel. Presented at Symp. 84. Broudeur, R., Fidelle, J. P., Auchere, H .
Hydrogen Embrittlement and Corrosion 1972. Proc. Int. eongr. Hydrogen i n
of Metals, Jablonna, Poland. Met., Paris, 1 972, pp. 1 06-7. Paris :
57. Holzworth, M. L., Louthan, M. R. Jr . Editions Science et Industrie
1968. Corrosion 24: 1 1 0-2 4 85. Louthan, M. R. Jr., Caskey, G. R. Jr.,
58. Okada, H., Hosai, Y., Abe, S. 1 970. Donovan, J. A., Rawl, D. E. Jr . 1 972.
Corrosion 26 : 1 8 3--86 Mater. Sci. Eng. 1 0 : 357-68
59. Shanabarger, M. R. 1975. Surf Sci. 52 : 86. Fricke, E., Stuwe, H. P., Vibrans, G.
689-96 1 9 7 1 . Metall. Trans. 2 : 2697-2700
60. Cavalier, 1. c., Chornet, E. 1 976. Surf 87. Donovan, .T. A. 1 976. Metall. Trans.
Sci. 60 : 125-46 7A : 1 677-83
6 1 . Williams, D. P., Nelson, H . G. 1970. 88. Thompson, A. W. 1 9 74. Metall. Trans.
Metall. Trans. 1 : 63-68 5 : 1 855-61
62. Nelson, H. G., Williams, D. P., 89. Tien, J. K., Thompson, A. W., Bernstein,
Tetelman, A. S. 1 9 7 1 . Metall. Trans. I. M., Richards, R. J. 1 976. Metall.
2 : 953-59 Trans. 7A : 82 1-29
63. Ewing, V. c., Ubbe\ohde, A. R. 1 9 5 5 . 90. Johnson, H. H., H irth, J. P. 1 97 6 .
Proc. R . Soc. London A 2 30 : 30 1-1 1 Metall. Trans. 7A : 1 543--48
64. McBreen, J., Genshaw, M. A. 1969. 9 1 . Yu, H.-Y., Li, J. C. M. 1 976. In Proc.
See Ref. 32, pp. 5 1-62 Conf Computer Simulation for Materials
65. Zakroczymski, T., Szklarska-Smialow­ Applications, Gaithersburg, 1 976, ed.
ska, Z., Smialowski, M. 1 975. Werkst. R. J. Arsenault, J. R. Beeler Jr., J. A.
Karras. 2 6 : 6 1 7-24 Simmons, pp. 872-81
66. Brown, B. F., Fujii, C. T., Dahlberg, E. P. 92. Raczynski, W., Kufudakis, A. 1 976.
1969. J. Electrochem. Soc. 1 1 6 : 2 1 8- 1 9 Czech. J. Phys. B26 : 1 36(}--65
6 7 . Szklar�ka-Smialowska, Z. 1 9 7 1 . Cor­ 93. Cornet, M ., Raczynski, W., Talbot­
rosion 2 7 : 223-33 Besnard, S. 1972. Mem. Sci. Rev. Metall.
68. Hersleb, G., Engell, H. J. 196 1 . Z. 69 : 27-32
Elektrochem. 65 : 8 8 1 -87 94. Cornet, M. 1973. Etude de la Fragilisa­
69. Pickering, H. W., Frankenthal, R. P. tiondu Fer par I 'Hydrogene. D.Sc. thesis,
1 972. J . Electrochem. Soc. 1 1 9 : 1 297- Univ. Paris. 1 12 pp.
1 304 95. Raczynski, W. 1 977. Critical Con centra-
3 56 ORlAN I

tion of Hydrogen in Iron. Presented at 1 1 9. Bernstein, I. M. 1 974. Scripta Metall.


Symp. Hydrogen Embrittlement and the 8 : 343-50
Corrosion of Metals, Jablonna, Poland. 1 20. Adair, A. M. 1 966. Trans. Metal/. Soc.
96. Beck, W., Bockris, J. O'M., McBreen, AIME 2 3 6 : 1 6 1 3- 1 5
J., Nanis, L. 1 966. Proc. R. Soc. London 1 2 1 . Gourmelon, A. 1974. Deformation
A290 ; 220-35 Plastique et Rupture du Fer de Haute
97. Bernstein, I. M . 1 970. Metal/. Trans. Purete. D.lng. thesis, Univ. Paris.
1 : 3 1 43-50 208 pp.
98. Rath, B. B., Bernstein, I. M. 1 9 7 1 . 122. Gourmeion, A. 1 975. Mem. Sci. Rev.
Metall. Trans. 2 ; 2845-51 Metall. 72 : 475-89
99. Smialowski, M. 1962. Hydrogen in 1 2 3 . Cornet, M., Trichet, M. F., Talbot­
Steel. Oxford : Pergamon. 452 pp. Besnard, S. 1977. Mem. Sci. Rev. Metall.
100. Tetelman, A. S., Wagner, C. N. J., 74 : 307- 1 6
Robertson, W. D. 1 9 6 1 . Acta Metall. 124. Welsch, G., Gibala, R., Mitchell, T . E.
9 : 205- 1 5 1 975. Acta Metall. 2 3 : 1 4 6 1 -68
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

1 0 1 . Tetelman, A. S . , Robertson, W. D. 1962. 125. Talbot-Besnard , S. 1 977. Idees Actuelles


Trans. Metall. Soc. AIME 224: 775-83 sur les M ecanismes de la Fragilisation
1 02. Asano, S., Otsuka, R. 1976. Scripta par I'Hydrogene due Fer et des Alliges
Metall. 1 0 : 1 0 1 5-20 Ferreux. Presented at the Second Int.
103. Toh, T., Baldwin, W. M. Jr. 1 956. In Congr. on Hydrogen in Met., Paris. To
by Cornell University on 06/09/12. For personal use only.

Stress Corrosion Cracking and Em­ appear in Proceedings


brittlement, ed. W. D. Robertson, pp. 126. Bernstein, I. M., Thompson, A. W.
1 76-186. New York : Wiley 1977. An Eva luation of Hydrogen Em­
104. Grant, N. J., Lunsford, J. L. 1 955. Iron brittlement Mechanisms. Presented at
Age 1 7 5 : 92-94 Can! on Mech. Environ. Embrittlement,
105. Hobson, J. D., Sykes, C. 1 9 5 1 . J. Iron Guildford, Enqland. To appear in Pro­
Steel 1 nst. 1 69 : 209-20 ceedings
1 06. Frohmberg, R. P., Rarnett, W. J., 1 27. M., Garber, R . , Pressou yre,
Bernstein, I.
Troiano, A. R. 1 955. Trans. Am. Soc. G. M . 1976. See Ref. 10, pp. 37-57
Met. 47 : 892-925 128. Viswanathan, R., Hudak, S. J. Jr. 1977.
107. Schultz, A. E., Robertson, W. D. 1957. Metall. Trans. 8A : 1 633-37
Corrosion 1 3 : 437t-58 t 129. Dautovich, D. P., Floreen, S. 1973.
108. Lunarska, E. 1 977. Scripta Metal/. 1 1 : MetaII. Trans. 4 : 2627-30
2 8 3-87 1 30. Gerberich, W. W., Chen, Y. T., St. John,
109. Matsui, H., Moriya, S., Kimura, H . C. 1 975. Metall. Trans. 6A : 1 485-98
1 976. Proc. 4 t h Int. Can! Strength of 1 3 1 . Thompson, A. W. 1977. See Ref. 1 1 7,
Metals and Alluys, Nancy, vol. 1 , pp. pp. 2 3 7-42
2 9 1 -95. Nancy : Laboratoire de 1 32. Gangloff, R. P., Wei, R. P. 1977.
Physique du Solide Metall. Trans. 8A : 1043-53
1 10. Hoffmann, W., Rauls, W. 1 965. Weld. J. 1 33. Thompson, A. W., Brooks, J. A. 1975.
44: 225S-30S Metall. Trans. 6A : 1 4 3 1 -42
1 1 1 . Walter, R. J., Chandler, W. T. 1 9 7 1 . 1 34. Garber, R., Bernstein, I. M., Thompson,
Mater. Sci. Eng. 8 : 90-9 7 A. W. 1976. Scripta Metall. 1 0 : 34 1 -45
1 1 2. Bernstein, I. M. 1969. U.S. Steel Corp., 1 35. Thompson, A. W. 1 976. See Ref. 10,
E. C. Bain Lab. Rep. No. 1515. pp. 467-77
Monroeville, Pa. : U.S. Steel 1 36. Sugino, K., Miyamoto, K. Nagumo, M.
1 1 3. Kamdar, M. H. 1974. See Ref. 2, pp. 1976. See Ref. 109, vol. 2, pp. 550- 5 4
107- 1 1 2 1 37. Rellick, J. R., McMahon, C. J. Jr. 1974.
1 1 4. Seabrook, 1 . B., Grant, N . J., Carney, Metall. Trans. 5 : 2439-50
D. 1 950. Trans. Metall. Soc. AIME 1 38. Yoshino, K., McMahon, C. J. Jr. 1 974.
1 8 8 : 1 3 1 7-2 1 Metall. Trans. 5 : 363-70
1 1 5. Vennett, R. A., Ansell, G. S. 1967. 1 39. Briant, C. L., Feng, H. c., McMahon,
Trans. Am. Soc. Met. 60 : 242-5 1 C. 1. Jr. 1978. Metall. Trans. In press
1 1 6. Kusch, H. G. 1977. Influence of H on 1 40. Bancrji, S. K., McMahon, C. J. Jr.,
the Plastic Deformation of Pure Iron. Feng, H. C. 1978. Metall. Trans. 9A :
Presented at Symp. Hydrogen Em­ 237-47
brittlement and the Corrosion of Metals, 1 4 1 . Wei, R. P., Simmons, G. W. 1 976.
Jablonna, Poland Scripta Metall. 1 0 : 1 5 3-57
1 1 7. Lee, T. D., Goldenberg, T., H irth, J. P. 142. Simmons, G. W., Pao, P. S., Wei, R. P.
1977. Proc. Fracture 1 977, Waterloo, 1978. Metall. Trans. In press.
Canada. 2 : 243-48 1 43. McMahon, C. J. Jr., Briant, C. L.,
1 1 8. Gupta, I., Li, J. C. M. 1970. Metall. Banerji, S. K. 1977. See Ref. 1 1 7, 1 :
Trans. I : 232 3-30 363-S5
HYDROGEN EMBRITTLEMENT 3 57
144. Johnson, H. H., Moriet, J. G., Troiano, 1 969. See Ref. 32, pp. 420-38
A. R. 1958. Trans. Metall. Soc. AIME 1 56. Dunegan, H. L., Tete1man, A. S. 1 9 7 1 .
2 1 2 : 528-36 Eng. Fracture Mech. 2 : 387-402
1 45. McIntyre, P., Priest, A . H. 1 972. Brit. 1 57. Gerberich, W. W., Chen, Y. T. 1 9 75 .
Steel Corp. Rep. MG/31/72 Metall. Trans. 6A : 2 7 1 -78
146. Hudak, S. J. Jr. Wei, R. P. 1 976. Metall. 1 58 . Gerberich, W. W., Garry, J., Lessar,
Trans. 7A : 235-41 J. F. 1976. See Ref. 10, pp. 70- 8 1
147. Gangloff, R. P., Wei, R . P. 1 974. Scripta 1 59. Oriani, R. A . , Josephic, P. H . '1972.
Metall. 8 : 661-68 Scripta Metall. 6 : 68 1 -8 8
1 48. Pao, P. S., Wei. R . P. 1 977. Scripta 1 60. Oriani, R. A., Josephic, P. H. 1 977.
Metall. 1 1 : 5 1 5-20 Hydrogen-Assisted Crack Initiation in a
149. Oriani, R. A. 1970. Metall. Trans. 1 : High-Strength Steel. Presented at Ann.
2346-47 Meet. Metall. Soc. AI ME, Chicago. To
1 50. Sawicki, V., Johnson, H. H. 1 9 7 1 . appear in Proceedings
M eta II. Trans. 2 : 3496-97 161. Gilman, J. J. 1960. In Plasticity, Proc.
1 5 1 . Carter, C. S. 1 9 7 1 . Corrosion 27 : 2nd Symp. Naval Structural Mech.,
Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

47 1 -77 pp. 43-96. Oxford : Pergamon


1 52. Parrish, P. A., Das, K. B., Chen, C. M., 162. Thomson, R. 1 978. J. Mater. Sci. 1 3 :
Verink, E. D. Jr. 1976. See Ref. 1 0, pp. 1 28-42
169-80 163. Thompson, A. W. 1977. Micro­
153. Landes, J. D., Wei, R . P. 1973. Int. J. structural Factors in Hydrogen
by Cornell University on 06/09/12. For personal use only.

Fracture 9: 27 7-9 3 Embrittlement and Stress Corrosion


1 54. Loginow, A. W., Phelps, E. H. 1975. Cracking. Presented at Conf Environ.
Corrosion 31 : 404- 1 2 Degradat ion Eng. Mater., Blacksburg,
1 5 5. Gerberich, W . W., Hartbower, C . E . Virginia. To appear in Proceedings
ANNUAL
REVIEWS Further
Quick links to online content

Annual Review of Materials Science


Volume 8,1978

CONTENTS

PREFATORY CHAPTER
RESOURCES AND THE QUALITY OF LIFE IN 2000, Harvey Brooks

EXPERIMENTAL AND THEORETICAL METHODS


Annu. Rev. Mater. Sci. 1978.8:327-357. Downloaded from www.annualreviews.org

MODERN EXPERIMENTAL METHODS FOR SURFACE AND THIN-FILM CHEMICAL


ANALYSIS, Charles A. Evans, Jr. and Richard J. Blattner 181
ELECTROCHEMICAL METHODS FOR DETERMINING KINETIC PROPERTIES OF
SOLIDS, Werner Weppner and Robert A. Huggins 269
by Cornell University on 06/09/12. For personal use only.

PREPARATION, PROCESSING, AND STRUCTURAL CHANGES


PREPARATION OF METAL POWDERS, Alan Lawley 49
VPE GROWTH OF IIIjV SEMICONDUCTORS, G. B. Stringfellow 73

PROPERTIES AND PHENOMENA


CHEMICAL PROPERTIES OF AMORPHOUS METALS, Tsuyoshi Masumoto and
Koji Hashimoto 215
FRACTURE AT HIGH TEMPERATURES UNDER CYCLIC LOADING, D. M. R.
Taplin and A. L. W Collins 235
SOME PHYSICAL PROPERTIES OF ORGANOSILICON LADDER POLYMERS, K. A.
Andrianov, A. A. Zhdanov, and V. Yu. Levin 313
HYDROGEN EMBRITTLEMENT OF STEELS, R. A. Oriani 327
MAGNETIC PROPERTIES OF AMORPHOUS ALLOYS, C. D. Graham, Jr. and T. Egami 423
RELIABILITY AND FAILURE MECHANISMS OF ELECTRONIC MATERIALS, A. T.
English and C. M. Melliar-Smith 459

SPECIAL MATERIALS
NEW HIGH-PERFORMANCE PERMANENT MAGNETS BASED ON RARE EARTH-
TRANSITION METAL COMPOUNDS, A. Menth, H. Nagel, and R. S. Perkins 21
SEMICONDUCTORS FOR PHOTOELECTROLYSIS, L. A. Harris and R. H. Wilson 99
MATERIALS FOR INTEGRATED OPTIcs: GAAs, Esther M. Conwell and Robert D.
Burnham 135

DATA SOURCES
DATA SOURCES FOR MATERIALS SCIENTISTS AND ENGINEERS, J. H. Westbrook
and J. D. Desai 359

INDEXES
Author Index 497
Subj ect Index 510
Cumulative Index of Contributing Authors, Volumes 4-8 521
Cumulative Index of Chapter Titles, Volumes 4-8 522

You might also like