The Role of Aging Reactions in the Hydrogen Embrittlement Susceptibility of an HSLA Steel M. E STEVENS and I. M. BERNSTEIN Hydrogen embrittlement susceptibility, as measured from room temperature precharged tensile specimens, indicates that the type, extent, and morphology of carbide precipitation are all important in determining the degree and mode of degradation. At equivalent charging conditions, embrittlement is virtually eliminated by aging to produce fine scale clustering of Ti(C, N), even when concurrent with cementite precipitation. High temperature aging (> 500 ~ results in exclusive precipitation of the alloy carbide, but also in a total loss of ductility due to a fracture mode transition to intergranular. This is shown to be associated with metalloid (P, S) segregation to grain boundaries accompanying depletion of Ti in solution. Intermediate behavior is observed in microstructures produced by high temperature quenching or aging at temperatures ( - 4 0 0 ~ where only cementite precipitation is observed.

I.

INTRODUCTION

RECENTLY, much of the research on the hydrogen embrittlement susceptibility of metals and alloys has centered on the role of metallurgical variables. 1-4 This emphasis is an outgrowth of the better emerging picture of the basic physical parameters which govern the manifestation of hydrogen embrittlement in many systems. Although some mechanistic models have been suggested, 3'5 no single mechanism is yet satisfactory in explaining the range of hydrogen effects, particularly in ferrous alloys. As stated above, there have been several recurring phenomenological features of hydrogen embrittlement identified, such as temperature dependent ductility minima, 6 an inverse strain rate effect on ductility, and three stage in(v) v s K~ hydrogen induced crack growth dependence. These and other observations have led to the general premise that the embrittlement step is preceded by localized transport of hydrogen to incipient failure initiation sites such as reentrant crack tips, inclusions, and grain boundaries. 7'8 This step may be particularly detrimental in bcc metals such as iron, due to the relatively high diffusivity (approximately 10 -6 c m 2 per second at 25 ~ and low equilibrium solubility (0.01 ppm at 25 ~ in the lattice. 9 Previous measurements of these parameters have been complicated by the effects of hydrogen "trapping" at microstructural features such as grain boundaries, dislocations, second phase particles, voids, and interstitial solute atoms. 3'5,~~ Attempts have been made to correlate the susceptibility of alloys to hydrogen embrittlement with independent determinations of such trapping parameters. Although certain of the abovementioned traps wilt act as failure initiation sites (e.g~ ,grain boundaries), it has also been suggested that the detrimental effects of internal hydrogen may be ameliorated by the proper control of a benign trapping population. 1~ Such a trapping density would either preclude the accumulation of critical hydrogen concentrations at incipient failure initiation sites by maintaining a distributed hydrogen profile, or M. F. STEVENS is Staff Member, Los Alamos National Laboratory, Los Alamos, NM 87545. I.M. BERNSTEIN is Professor and Head, Metal: lurgical Engineering and Materials Science Department, Carnegie-Mellon University, Pittsburgh, PA 15213. Manuscript submitted January 11, 1985.

METALLURGICAL TRANSACTIONS A

it could delay the onset of hydrogen embrittlement to higher bulk concentrations at longer times. Specifically, Pressouyre and Bernstein 12'13'~4have studied the hydrogen trapping and embrittlement susceptibility of a number of Fe-Ti-C alloys with varying Ti:C ratios. Through the use of electrochemical permeation techniques and analyses, these authors characterized Ti substitutional atoms as low occupancy reversible trap sites for H with an interaction energy of 0.27 eV (-- 30 kJ/mol). Incoherent TiC particles, on the other hand, were found to be larger occupancy, irreversible trap sites with an interaction energy of 0.98 eV ( - 100 kJ/mol). Embrittlement susceptibility determinations led to the conclusions that an alloy consisting of fine, homogeneously distributed, irreversible traps (TIC) possessed the greatest resistance under dynamic and static testing conditions whereas reversible traps (Ti substitutional atoms) were found to give detrimental effects under dynamic conditions by acting as an additional internal hydrogen source and beneficial effects under static conditions, by acting as an additional hydrogen trap. In a previous paper, ~5 the authors reported on the correlation of embrittlement susceptibility with trapping in an HSLA steel of nominally identical composition to that of the present paper. It, was indeed found that deep trapping of hydrogen was associated with the precipitation of fine (20 to 40 A) TiC particles. Combined with the strong, ductile properties of the highly dislocated, fine ferrite lath structure, microstructures containing these precipitates displayed enhanced ductility under conditions where internal hydrogen was present during uniaxial loading. Furthermore, these traps were found to be saturable under the given charging conditions, returning the microstructure to a highIy susceptible condition after overcharging. The highly reactive nature of Ti has concurrently prompted studies to evaluate its use to offset the temper embrittlement of steels due to segregation of phosphorus to grain boundaries. ~6-~9 Since titanium is known to have a highly attractive interaction energy with P and forms stable phosphides, a beneficial scavenging effect is predicted. However, extended period exposure to temperatures -> 500 ~ may lead to formation of the more stable carbide phase, release of the phosphorus, and perhaps subsequent embrittlement. In such a case and if the precipitation of the VOLUME 16A, OCTOBER 1985-- 1879

matrix was characterized by a high dislocation density and a mean ferrite lath width of 0.5 /xm. These two factors are commonly associated with the high ductility, smooth yielding, and moderately high strength common in these steels. 21 No evidence for cementite or alloy carbide phases was found in these as-quenched microstructures as a result of autotempering. Aging at 400 ~ led to the nucleation of cementite at lath boundaries, as shown in Figure 2. The associated loss of carbon from the matrix resulted in a decrease in hardness, as is common during the precipitation of carbide in iron-carbon martensites.Z2 The progress of thisprecipitation can be seen in Figure 3 which summarizes the time dependence of the isothermal aging response at 400 ~ 500 ~ and 550 ~ A slight increase in hardness can be observed after aging at 400 ~ for aging times greater than 10,000 seconds. This most likely is associated with Ti(C, N) clustering, although no direct evidence for this was obtained.

alloy carbide phase results in segregation of P, then the presence of temper embrittlement could offset or even cancel any beneficial effects of deep hydrogen trapping at the carbide interfaces. However, it has been suggested 18 that if phosphorus can be effectively trapped at TiC particle interfaces in analogy with the hydrogen trapping case, then controlled precipitation of this phase should result in enhanced hydrogen and temper embrittlement resistance simultaneously. In a recent study, 2~ part of which is the subject of the present paper, a titanium bearing HSLA steel was chosen in order to study the effect of aging condition on hydrogen trapping and embrittlement susceptibility. In this case, since the stoichiometry was fixed, the type and extent of carbide precipitation was varied by using different aging temperatures. It will be seen that the mechanical and trapping response of the steel to moderate hydrogen charging was markedly dependent on the extent of TiC precipitation for a variety of reasons. II. M I C R O S T R U C T U R E S AND A G I N G R E A C T I O N S All materials for this study were taken from a single heat of experimental HSLA steel prepared as a 45 kg ingot, hot-rolled from a temperature of 1260 ~ to a thickness of 25 mm. The chemical composition of this heat is given in Table I. Standard tensile specimens from the as-received plate were subsequently reheated to 1200 ~ for 1 hour, water quenched, and aged for 1 hour at temperatures between 400 ~ and 900 ~ The microstructure produced as a result of the quench from 1200 ~ is shown in Figure 1. This acicular ferrite

(a)

(b) Fig. 2 - - G r a i n boundary precipitation of cementite after aging 1 h at 400 ~ (a) bright field and (b) centered dark field.

Fig. 1 - - T E M micrograph of as-quenched microstructure of HSLA steel. Note fine lath size and heavily dislocated substructure.

Table I.

Chemical Composition of HSLA Steel (Wt Pct)

C

Mn

P

S

Si

A1

V

Ti

N

0.050

1.30

0.005

0.028

0.31

0.003

<0.002

0.22

0.001

1880--VOLUME 16A, OCTOBER 1985

METALLURGICAL TRANSACTIONS A

Fig. 3 - - A g i n g reaction sequences at intermediate temperatures.

Precipitation of cementite and "clustering" of TiC was, however, observed during aging at 500 ~ As opposed to the 400 ~ cementite morphology, which suggested that nucleation occurred at the lath boundaries and subsequent growth proceeded into one of the adjacent grains, the cementite phase found after direct aging at 500 ~ appeared to grow along the boundary, or intergranularly, as shown in Figure 4. Evidence for TiC clustering was obtained by use of a field ion microscope,* using neon as the imaging gas. *We are indebted to Dr. S.S. Brenner of the University of Pittsburgh who while at the U.S. Steel Research Laboratory permitted us to use his atom probe field ion microscopy facility.

Examples of this clustering are shown in Figure 5, supporting the previous results of Cuddy e t a l . 23 in similar Fe-Ti-C compositions at 500 ~ A transient strengthening effect was observed by them, attributed to the formation of stable clusters of Ti with C and N under the stress field of both stationary and moving dislocations. This phenomenon was considered distinct from the nucleation of the stable Ti(C, N) phase, since an overaging effect was not observed at this temperature. This argument is supported by the results of the isothermal aging response at 500 ~ in Figure 3. Note that the softening associated with cementite formation is delayed to longer times and that a more significant secondary hard-

Fig. 4--Interlath cementite precipitation following 1 h aging at 500 ~ Contrast with Fig. 2.

METALLURGICAL TRANSACTIONS A

Fig. 5 - - Field ion micrograph showing Ti(C, N) cluster after aging 25 rain at 500 ~ Magnification ~ 1,000,000 times.

ening is observed at longer times. The delayed softening may be attributed to either the competition for carbon between cementite and the Ti clusters or to a more direct inhibition of the cementite growth by the silicon or titanium present. In fact, the effect of silicon on cementite growth is well known 24 and attributed to its negligible solubility in cementite and the effect of lowering of the local carbon activity. Titanium may cause the same result, by either entering into the growing cementite particles or by segregation to the interfaces. Either diminished cementite growth rate or the aforementioned strengthening due to Ti(C, N) clusters would thus explain the delayed softening. At longer times, the continued nucleation of Ti(C, N) clusters apparently prevails, producing a substantial hardening effect. It is generally accepted that alloy carbides do not preferentia!ly form in ~lloy steels until temperatures in the range 500 C to 600 ~ are attained. 25 This is most probably due to the more sluggish diffusivity of substitutional alloying elements such as titanium, in comparison with interstitial elements such as carbon and nitrogen. Referring again to Figure 3, it can be seen from the 550 ~ isothermal aging behavior that a very rapid secondary hardening effect takes place. Microstructurally, this is associated with the "direct" precipitation of alloy carbide and an overall bypassing of the cementite phase. This direct precipitation of Ti(C, N) is quite evident from Figure 6, which is a centered dark-field TEM image of fine Ti(C, N) precipitates formed by aging one hour at 600 ~ Note the very fine size and apparent heterogeneous nucleation at lath boundaries and dislocations. Isochronal aging at 700 ~ yields more pronounced coarsening of the TiC phase (Figure 7). However, due to the increased mean size of these precipitates, they are less effective in pinning the motion of subgrain and lath boundaries with the result that the microstructure is partially recrystallized, as shown in Figure 8, and the strength drops sharply. 2~ VOLUME 16A, OCTOBER 1985--1881

III.

Fig. 6 - - C e n t e r e d dark-field micrograph showing fine Ti(C, N) precipitation along dislocations following aging for 1 h at 600 ~

MECHANICAL PROPERTIES

The effect of aging and particularly the transition from primary cementite to primary TiC formation on the uniaxial tensile properties of this alloy were determined through the use of 6.4 mm diameter, 44.4 mm gage length cylindrical tensile specimens. All tests were conducted at strain rates of 2 x 10_4 s -~. Figure 9 summarizes the tensile properties in both the longitudinal and transverse orientation. It may be concluded that the macroscopic yield strength in the longitudinal direction is not greatly affected by aging, but that the transverse strength is somewhat raised by aging. It has previously been established 26 that Ti additions to steel may induce strong (I 12)[ 110] type textures due to inhibited recrystallization during processing. This preferred crystallographic texturing may be responsible for the observed effect, although no evidence was obtained to support this suggestion. The ultimate tensile strength (UTS), however, appears to be influenced more significantly by aging reactions in both orientations. The as-quenched structure possesses a reasonably high UTS probably due to quenched-in stresses and a high density of mobile dislocations. At intermediate aging temperatures (400 to 500 ~ it is found that the work hardening capacity is lowered; recovery of the dislocation structure during aging may account for this. At higher aging temperatures (600 ~ increased work hardening capacity is once again observed. This is most likely due to precipitate induced cross-slip resulting from the high density of fine TiC precipitates. Identical tensile specimens were used to determine the effect of internal hydrogen on tensile ductility. These specimens were polished to a 600 grit surface and chemically polished in a 47 pct H20-47 pctH202-6 pct HF solution for 10 seconds. Cathodic charging was conducted in a solution of 1 N H2SO4 with additions of CS2 and As203. Platinum wire was used as an anode and the current density was held at 20 ~ A / c m 2 for 5 hours. Figure 10 summarizes the

Fig. 7 - - C e n t e r e d dark-field micrograph of coarsened Ti(C, N) precipitates following aging for 1 h at 700 ~ Faceting suggests a role of N on surface energy.

Fig. 8--Partially recrystallized microstructure produced after 1 h at 700 ~ aging.

1882--VOLUME 16A, OCTOBER 1985

Fig. 9--Longitudinal and transverse strength levels following isochronal 1 h agings at various temperatures.

METALLURGICAL TRANSACTIONS A

VII3 LONGITUDINAL

TENSILE

DUCTILITY

9 UNCHARGED o CHARGED

I00-

80i 60%RA

#

-

/

40-

20-

oi

, A0

,!.

~ 400

b 500

AGING TEMPERATURE

a

0 600

O------700

*C

Fig. 10--Uniaxial tensile ductility for charged and uncharged tests at various aging conditions.

ductility response of this material for both charged and uncharged conditions. In the uncharged condition, this material displays highly ductile tensile behavior and there is no apparent dependence on aging condition. In the charged condition, however, it is readily observed that the response is strongly microstructurally dependent. In particular, it may be noted that the overall response to internal hydrogen may be separated into two categories: aging treatments of 500 ~ and less where some ductility was observed and aging at temperatures above 500 ~ where no measurable reduction in area occurred. The fractographic features of the failure surfaces were reflective of the overall ductility observed. In the uncharged condition, all specimens exhibited the expected microvoid coalescence failure mode for alloys of this strength level. This general fracture mode was also evident on charge specimen fracture surfaces for aging treatments -< 500 ~ with the noteworthy exception of the presence of isolated "facets" of low ductility, as illustrated in Figure 11. It is

evident that the fracture mode is quite different within this isolated region which appears to fit the general definition of TTS or "tearing topography surface" as discussed by Thompson and Chesnutt,27 In addition, each of the observed TTS facets was associated with an inclusion or inclusion cluster as shown plainly in Figure 12. Energy dispersive X-ray spectroscopic analysis revealed no unique chemical composition of these inclusions, although A1 and Ti were most prevalent. The drastic drop in bulk ductility for aging treatments at temperatures > 500 ~ was accompanied by a transition to intergranular fracture, as shown in Figure 13. This fracture mode persisted through aging treatments at temperatures of 700 ~ for 1-hour periods, followed by water quenching. It should be noted that this severe fracture mode transition coincides quite closely with the transition to primary precipitation of TiC upon aging. The "clean" appearance of many of the grain boundary facets is reminiscent of temper embrittled steels, and since this fracture transition occurred so sharply with aging temperature, it was considered and will

(a)

(b) Fig. 1 1 - Scanning electron fractograph of low ductility "facet" located near edge of charged tensile specimen.

METALLURGICAL TRANSACTIONS A

Fig. 12--High magnification images of "facet" center from Fig. 11. (a) Aluminum and (b) titanium X-ray maps showing apparent Ti precipitation about preexisting A1 rich inclusion.

VOLUME 16A, OCTOBER 1985-- 1883

Fig. 13--Scanning electron fractograph of predominantly intergranular fracture observed in specimens aged at temperatures >500 ~ and charged.

be discussed in the next section that a similar cause (i.e., metalloid segregation) might be responsible for predisposing the grain boundaries to fracture when the additional embrittling effect of hydrogen was introduced. In order to examine the prior austenite grain boundary chemistry of this material, it was decided to take advantage of the hydrogen induced grain boundary brittleness in order to produce "clean" fracture facets in an ultra-high vacuum environment for surface analysis. Notched pin specimens were heat-treated, charged, and fractured in situ under a high vacuum (10 -9 torr) in a high resolution scanning Auger spectrometer. Auger spectra from the grain boundary facets typically revealed the presence of P and occasionally S, with average peak height ratios of Ap//AFe703 ~ 0.06 and As//AFe703 ~ 0.07. The significance of this segregation and possible causes will also be explored in the next section. IV.

DISCUSSION

It is evident that the strength and ductility of this alloy are directly affected by the sequence and extent of carbide precipitation, both with and without the effect of internal hydrogen. Additionally, it appears that the division of response, for this steel at aging temperatures near 500 ~ demonstrates the two basic forms of embrittlement that can be exhibited by iron and steel. The first and possibly most intriguing aspect is the apparent intrinsic effect of hydrogen on localized ductile deformation, as demonstrated by the fracture mode appearance within the TTS facets. This and independent evidence from other studies z8'29 indicate that hydrogen may alter the mode and distribution of slip on a local (microscopic) level with macroscopic property consequences. The supporting evidence includes the fact that hydrogen may induce softening in sufficiently pure iron at room temperature. These results may apply to steel microstructures on a localized level but where complicating factors in slip (solutes, dislocation structures, grain boundaries, etc.) obscure a macroscopic influence on flow. Experiments on poly and single crystals of hydrogen precharged iron strained in tension have resulted in cracking 1884--VOLUME 16A, OCTOBER 1985

along {110} type planes. 37 More recently, Birnbaum and co-workers 3~and Takeyama et al.3~ have demonstrated using in situ electron microscopy that the surfaces of hydrogen induced cracks and blisters are associated with {110} and {112} type slip planes. Additionally, earlier experiments have suggested that enhanced transport of hydrogen by glissile dislocations occurs in iron, 3233 and this has now been directly confirmed. 34 These as well as other observations have led to the recent suggestion by many researchers that the intrinsic nature of hydrogen embrittlement in iron as well as other metals is usually ductile and can involve significant plasticity. 2'3'~5-38 Most often, the typical fracture mode encountered for iron-based alloys containing internal hydrogen is also ductile but on a much finer scale, that is significantly different from cleavage and dimple fracture modes. The appearance of the TTS fracture facets in the present study closely resembles that of other recent studies where hydrogen affected fracture was demonstrated. 39In both cases, and others, hydrogen acceleration is closely associated with a low fracture energy ductile process. This localization of hydrogen influenced ductile separation about inclusions is a manifestation of another wellknown characteristic of hydrogen in iron, i.e., the trapping at heterogeneities. While it is likely that some hydrogen was trapped at these sites prior to the application of stress, the localized stress concentrations (especially the hydrostatic components normal to the applied stress) during plastic flow no doubt created the requisite conditions for enhanced hydrogen flux to create the supersaturations to drive the hydrogen affected fracture mode. Once accumulated at these interfaces, the supersaturated hydrogen apparently causes slip to concentrate along distinct bands, with the result that fracture occurs by an apparent shear localization, as has been observed by Hirth and c o - w o r k e r s . 4~ It is not presently known what effect hydrogen has on the iron lattice in order to bring about this altered dislocation distribution. It has been suggested 42'43 that hydrogen in solution may alter the interatomic potential of the iron lattice and hence enhance the mobility of screw dislocations, which ordinarily control plastic flow at room temperature. At the same time, cross-slip of screws may be inhibited, 28 resulting in more intense planar slip. In turn, hydrogen transported and trapped along these slip bands would contribute to a failure mode resulting from a shear instability. The appearance of the tear ridges within the TTS facets about inclusions, as shown in Figure 12, supports such a shear localization scenario. The only significance of the inclusions, then, would be that they provide macroscopic stress concentrations inducing the necessary hydrogen supersaturations necessary for the observed effect. When the facets separate, the intervening hydrogen-lean regions of the specimen fail by overload, resulting in a net loss of bulk ductility. The transition to predominantly intergranular fracture, after aging at temperatures > 500 ~ is strong evidence of a significant change in the nature of the aging reactions in this temperature regime, most probably associated with the direct precipitation of Ti(C, N) and likely a synergistic interaction between hydrogen and temper embrittlement. Previous studies of temper embrittlement in alloy steels have revealed that a key precursor is the accumulation of metalloid elements (Sn, Sb, S, P) at high angle boundaries (prior 3') during tempering. 44'45There are, in addition, complex changes in the interactions of such metalloids with METALLURGICALTRANSACTIONS A

other alloying elements in solution. Recent work by Erhart et al. 18has revealed that Ti present in solid solution in iron interacts quite strongly with P. In quaternary Fe-Ti-C-P systems, however, this interaction is diminished due to Ti-C clustering and eventual nucleation and growth of TiC phase. It should be noted that this behavior is well predicted by a ternary equilibrium segregation model, as proposed by Guttman, 46 to rationalize temper embrittlement susceptibility in low alloy steels. The relative susceptibility can be deduced by the relative strength of the alloying elementmetalloid interaction coefficient. When this interaction is sufficiently strong (--~ 50,000 cal/g-atom or - 210 kJ/ g-atom), clustering or precipitation is predicted, as is the case for Mo, Ti, and Zr. The suggested interactions are as follows: after a solutionization treatment, a significant quantity of Ti(C, N) phase has been dissolved, leaving Ti in solution but most probably attracting interstitial elements such as C, N, O, and P in the form of clusters. It is important that these clusters not be considered a distinct second phase; the mobility of the element in the matrix is simply diminished, much in the same way that hydrogen is "trapped" at Ti atoms in solutions. At elevated aging temperatures, however, a stronger reaction with C and N takes place; P is rejected from this phase and, coupled with greater mobility at elevated temperature, is free to segregate to the boundaries. Ustinovshchikov 19 has confirmed these trends in alloy steels containing Cr, V, Mo, and deliberate P additions. In each of these alloy systems, the transition to intergranular fracture (enhanced segregation) was associated with secondary hardening during tempering and the associated loss of alloy carbide formers from solution in analogy with the present case. In support, it has been noted by McMahon and co-workers47 that V additions made to Cr-Ni-Mo steels stabilize against Mo losses during service at high temperature. It seems evident that this concurrent temper embrittlement is responsible for the precipitous fall in the hydrogen charged ductility of this alloy. In the temper embrittled condition, as has been generally observed for other steels, the prior y boundaries become the preferred crack path when both P and H reside at these sites. As regards the authors' previous results on the nominally same alloy, 15 it was found through heat chemistry analyses that significant variations in metalloid concentration existed. In the prior study, the combined P and S levels were generally < 0.015 wt pct whereas the same element concentrations in the present study were - 0.035 wt pct. It must be concluded that within this range of metalloid concentration, a critical grain boundary concentration of metalloids is attained. Although it iS evident that both metalloid and H must coexist in sufficient concentrations at the grain boundaries, it is still unknown whether these segregants act in an additive or more complex synergistic way. The sine q u a n o n of these results is that despite the high density of irreversible, fine (and therefore presumably innocuous) traps present in this microstructure, the heightened susceptibility of the prior y boundaries takes primary importance and a beneficial trapping effect is not observed.

V.

CONCLUSIONS

1. The hydrogen embrittlement susceptibility of the steel in this study was highly dependent on the aging condition METALLURGICALTRANSACTIONSA

insofar as this latter property determined the degree of TiC precipitation and the amount of free Ti in solid solution. 2. Microstructures aged at 500 ~ or less displayed reasonable resistance to hydrogen compared to the as-quenched condition. Fracture surface studies revealed that failure usually initiated about large (10 to 70/xm) inclusions; the fracture mode emanating from these features was a ductile "tearing topography surface" or TTS, suggesting a hydrogen associated plastic separation. 3. Microstructures aged at temperatures greater than 500 ~ exhibited catastrophic intergranular failure with no apparent ductility. This behavior was correlated with the transition from cementite to alloy (TIC) carbide precipitation in this aging temperature regime and a concomitant loss of scavenging effect of Ti on P and to a lesser extent S, resulting in segregation and enrichment of these elements along high energy prior y boundaries. In the combined presence of hydrogen, these boundaries then become the preferred low energy crack paths. 4. The presence of fine, irreversible trapping is not sufficient, in and of itself, to ameliorate embrittlement if during aging to produce such a distribution other previously innocuous traps become easy crack initiation or propagation sites.

ACKNOWLEDGMENTS We are indebted to the many insightful discussions with Professor A. W. Thompson and Dr. C. Hwang and to the experimental assistance of P. Kullen. Dr. C. Briant of G,E. Research Lab kindly performed some preliminary Auger studies. The generous support of the Office of Naval Research is gratefully acknowledged.

REFERENCES 1. A.W. Thompson and I.M. Bernstein: "The Role of Metallurgical Variables in Hydrogen Assisted Environmental Fracture", in Advances in Corrosion Science and Technology, M.G. Fontana and R.W. Staehle, eds., Plenum Publishing Corp., 1980, vol. 7, pp. 53-175. 2. H.K. Birnbaum: "Hydrogen Related Failure Mechanisms in Metals", in Environment Sensitive Fracture of Engineering Materials, Z.A. Foroulis, ed., TMS-AIME, Warrendale, PA, 1979, pp. 326-60. 3. J.P. Hirth: Metall Trans. A, 1980, vol. llA, pp. 861-90. 4. I.M. Bernstein and A. W. Thompson: Int. Met. Rev. (Rev 212), 1976, vol. 21, pp. 269-87, 5. I.M. Bernstein: Mater. Sci. Eng., 1970, vol. 6, pp. 1-19. 6. H.K. Birnbaum: "Hydrogen Effects on the Fracture of BCC Metals", in Mechanical Properties of BCC Metals, M. Meshii, ed., TMSAIME, Warrendale, PA, 1982, pp. 153-72. 7. A. WI Thompson and i. M. Bernstein: "Stress Corrosion Cracking and Hydrogen Embrittlement", in Metallurgical Treatises, J. K. Tien and J. E Elliott, eds., TMS-AIME, Warrendale, PA, 1981, pp. 589-601. 8. Y. Kikuta: in Hydrogen Effects in Metals, I. M. Bernstein and Anthony W. Thompson, eds., TMS-AIME, Warrendale, PA, 1981, pp. 755-65. 9. K. Kiuchi and R.B. McClellan: Acta Metall., 1983, vol. 31, pp. 961-84. 10. R. Gibala: in Stress Corrosion Cracking and Hydrogen Embrittlement oflron Base Alloys, NACE-5, Houston, TX, 1977, p. 244. 11. G.M. Pressouyre and I.M. Bernstein: Metall. Trans. A, 1981, vol. 12A, pp. 835-44. 12. G.M. Pressouyre and I.M. Bernstein: Metall. Trans. A, 1978, vol. 9A, pp. 1571-80. 13. G.M. Pressouyre and I.M. Bernstein: Corrosion Science, 1978, vol. 18, pp. 819-33. 14. G.M. Pressouyre and I. M. Bernstein: Acta Metall., 1979, vol. 27, pp. 89-100. VOLUME 16A, OCTOBER 1985--1885

15. M. E Stevens, I.M. Bemstein, and W.A. Mclnteer: in Hydrogen Effects in Metals, I. M. Bemstein and Anthony W. Thompson, eds., TMS-AIME, Warrendale, PA, 1981, pp. 795-802. 16. S.M. Meyers, D. M. Follstaedt, and H. J. Rack: J. Appl. Phys., 1978, vol. 33, pp. 396-98. 17. J. Pillar, M. K. Miller, and S. S. Brenner: Trans. 29th Field Emission Syrup., Gotheburg, Sweden, 1982, pp. 473-80. 18. H. Erhart, H.J. Grabke, and R. Moiler: Arch. Eisenhiittenwesen, 1981, vol. 54, no. 7, pp. 285-89. 19. J.I. Ustinovshchikov: Acta Metall., 1983, vol. 31, pp. 355-64. 20. M.F. Stevens: Ph.D. Thesis, Carnegie-Mellon University, Pittsburgh, PA, 1984. 21. W.C. Leslie: The Physical Metallurgy of Steels, McGraw-Hill Book Co,, New York, NY, t981, p. 20t. 22. G.R. Speich: Trans. AIME, 1969, vol. 245, p. 2553. 23. L.J. Cuddy, H. E. Knechtel, and W. C. Leslie: Metall. Trans., 1974, vol. 5, pp. 1979-2003. 24. W.S. Owen: Trans. Am. Soc. Met., 1954, vol. 46, p. 812. 25. R. W. K. Honeycombe: Steels--Microstructure and Properties, Edward Arnold and ASM, Metals Park, OH, 1981, p. 152, 26. L. Meyer, F. Heisterkamp, and D. Lauterborn: in Processing and Properties of Low Carbon Steel, TMS-AIME, 1973, pp. 279-320. 27. Anthony W. Thompson and J.C. Chesnutt: Metall. Trans. A, 1979, vol. 10A, pp. 1193-96. 28. M. Comet and S. Talbot-Besnard: in "Hydrogen in Metals", Proc. 2nd JIM Int. Symp., Suppl. to Trans. Japan Inst. Met., 1980, p. 545. 29. K. Yoshino and C.J. McMahon, Jr.: Metall. Trans., 1974, vol. 5, p. 363. 30. T. Tabata and H.K. Birnbaum: Scripta Met., 1984, vol. 18, pp. 231-36. 31. T. Takeyama and H. Takahashi: "Nucleation of Crack and Microstructure Induced by Hydrogen", in Mechanical Properties of BCC Metals, M. Meshii, ed., TMS-AIME, Warrendale, PA, 1982,

1886--VOLUME 16A, OCTOBER 1985

pp. 153-72. 32. R.C. Frank: in Internal Stresses and Fatigue in Metals, G.M. Rossweiler and W.L. Grube, eds., Elsevier, New York, NY, 1959, pp. 411,424. 33. J.A. Donovan: Metall. Trans. A, 1976, vol. 7A, pp. 145-49. 34. C. Hwang: Ph.D. Thesis, Carnegie-Mellon University, Pittsburgh, PA, 1984. 35. Y. Takeda and C. J. McMahon, Jr.: Metall. Trans. A, 1981, vol. 12A, pp. 1255-66. 36. I.M. Bernstein: Metall. Trans., 1970, vol. 1, p. 3143. 37. E Nakasato and I.M. Bernstein: Metall. Trans. A, 1978, vol. 9A, pp. 1317-26. 38. H. Kimura, H. Matsui, S. Takaki, A. Kimura, and K. Oguri: in Mechanical Properties of BCC Metals, M. Meshii, ed., TMS-AIME, Warrendale, PA, 1982, pp. 125-33. 39. B. Craig and G. Krauss: in Hydrogen Effects in Metals, I.M. Bernstein and Anthony W. Thompson, eds., TMS-AIME, Warrendale, PA, 1981, pp. 795-802. 40. T.D. Lee, T. Goldenberg, and J.P. Hirth: Metall. Trans., 1974, vol. 10, pp. 439-48. 41. T. Goidenberg, T.D. Lee, and J.P. Hirth: Metall. Trans. A, 1978, vol. 9A, pp. 1663-71. 42. H. Matsui, S. Moriya, and H. Kimura: in "Strength of Metals and Alloys", Proc. 4th Int. Conf., 1976, vol. 1, p. 291. 43. H. Kimura, H. Matsui, and S. Moriya: Scripta Met., 1977, vol. 11, p. 473. 44. W. Steven and K. Balajiva: J. Iron Steel Inst., 1959, vol. 178, p. 141. 45. C.J. McMahon: Proc. 4th Bolton Landing Conf., J.L. Walter, J. H. Westbrook, and D. A. Woodfrod, eds., Claitor's Pub. Division, 1975, pp. 525-52. 46. M. Guttman: Met. Sci., 1976, vol. 10, pp. 337-41. 47. Z. Qu and C.J. McMahon, Jr.: Metall. Trans. A, 1983, vol. 14A, pp. 1101-08.

METALLURGICAL TRANSACTIONS A