Hydrogen embrittlement of α titanium: In situ tem studies

Hydrogen embrittlement of α titanium: In situ tem studies

Acre me:&. Vol. 36, No. I. pp. 111-124. 1988 Printed in Great Britain. All rights reserved HYDROGEN Department Copyright c owl-6160188 $3.00 + 0.0...

2MB Sizes 1 Downloads 50 Views

Acre me:&. Vol. 36, No. I. pp. 111-124. 1988 Printed in Great Britain. All rights reserved

HYDROGEN

Department

Copyright c

owl-6160188 $3.00 + 0.00 1988 Pergamon Journals Ltd

EMBRITTLEMENT OF a TITANIUM: IN SlTU TEM STUDIES

D. S. SHEIt, I. M, ROBERTSON and H. K. BIRNRAUM of Materials Science and Materials Research Laboratory, University of Illinois at Urbana-Champaign, 1304 W. Green Street, Urbana, IL 61801, U.S.A. (Received 13 August 1986; in revisedfarm

9 April 1987)

Abstr&--The effect of hydrogen on fracture in the h.c.p. a ‘K-Q wt % Al alloy and the role of titanium hydride in the fracture process have been studied by deforming samples in situ in a high-voIta~ electron microscope equipped with an en~ronment~ cell. Two fracture m~hanisms have been observed in a gaseous hydrogen environment at room temperature; one is fracture by localized plastic deformation enhanced by the presence of hydrogen, the other is a brittle fracture of the stress-induced titanium hydride which precipitates at elastic singularities. The local stress intensity determines which mechanism predominates. At high stress intensities the crack propagates by the process of hydrogen enhanced localized plasticity, while at low stress intensities titanium hydrides form in the vicinity of crack tips and crack propagation proceeds though the hydrides. R&sum&-Nous avons itudit l’effet de l’hydrogine sur la rupture dans l’alliage hc Tia $ 4% en poids d’aluminium, ainsi que le rhle de l’hydrure de titane dans le mkanisme de rupture, en dtformant des &.zhantillons in situ dam un microscope tlectronique & haute tension &quip&d’une chambre d’atmosph&e contrblable. Nous avons observC deux mCcanismes de rupture dans un environnement gazeux d’hydrogene $ la temgrature ambiante: (1) rupture par dkformation plastique local& activee par la prtsence d’hydrogine; (2) rupture fragile de I’hydrure de titane que est induit par la contrainte et qui pr+cipite ii des points singuliers elastiques. L’intensith des contraintes locales determine quel est le m&anisme preponderant. Four les intensitbs de contrainte &lev&s, les fissures se propagent gtice au m~~nisme de plasticid localis&e au&l&e par l’hydrogene, alors que pour les intensitks de contrainte faibles, les hydrures de titane se forment au voisinage des &es de fissures et les fissures se propagent & travers les hydrures. Zusammenfassung-Der EinfluD des Wasserstoffs auf den Bruch in der hex. Legierung a-Ti-4 Gew. - % Al und die Rolle des Titanhydrids beim Bruchablauf wurden untersucht, indem Proben in einem H~hs~nnung~iektronenmikroskop in einer Umgebungsk~mer in-s&u verformt wurden. Zwei B~chm~h~ismen, die in g~~~igern Wasserstoff bei Raumtem~ratur ablaufen, wurden beobachtet. Der eine ist der Bruch durch Iokalisierte plastische Verformung in Gegenwart von Wasserstoff, der andere ist ein Spriidbruch des spannungsinduzierten Titanhydrids, der sich an elastischen SingularitIten ausscheidet. Die lokale Spannungsintensitit bestimmt, welcher Mechanismus iiberwiegt. Bei hoher Spannungsintensitit breitet sich der Ril3 ilber den ProzeD der Wasserstoff-verstlrkten lokalisierten Plastizitit aus, bei niedrigen Spannunsintensitlten bildet sich dagegen Titanhydrid in der Nlhe der RiBspitzen und die Risse breiten sich durch die Hydride aus.

1. INTRODUCTION

The mechanical properties of a material can be dramatically affected by the presence of hydrogen either in solid solution or in the local environment. This change often manifests itself as a reduction in strain to failure and may cause the fracture mode to change from one of ductile microvoid coalescence to a “brittle” intergranular. Embrittlement 6y hydrogen has been reported for a number of materials but the mechanisms by which hydrogen causes these effects are not completely understood. For materials stressed in low hydrogen fugacity environments “hydrogenembrittled”’ materials can be divided into two categories: (a) non-hydride and (b) hydride forming systems.

TPresent address: GE Aircraft Engines, Mail Drop M-89, 1 Neumann Way, Cincinnati, OH 45215, U.S.A.

In non-hydride forming systems, there is considerable controversy over the hydrogen embrittlement mechanism. One school of thought is that hydrogen causes a decrease in the cohesive strength of the atomic bond [l-5]. Several decohesion models have been developed f&7] but there is no supporting experimental evidence. An alternative mechanism, based on hydrogen enhancing plastic processes around the crack tip [S], has received direct experimental support. Deformation experiments, in situ in an electron microscope equipped with an environmental cell, have demonstrated that in Ni 19, IO], Fe 111, 121, Al and Al alloys 1131. and In 903 1141, the mobility of dislocations was enhanced when hydrogen was introduced into the sample. This enhancement and a concomitant localization of the slip to a narrow deformation zone ahead of the crack tip, allows crack advance to occur at lower stress levels in hydrogen than in inert environments [g-14]. These microscopic observations are supported by obser111

112

SHIH et al.: HYDROGEN EMBRITTLEMENT OF tl TITANIUM

vations of localized plastic deformation at fracture surfaces of macroscopic specimens [ 15-201. In hydride forming systems embrittlement occurs by hydride formation at stress concentrations followed by cleavage of the hydride [21-241. Hydride formation within the non-uniform tensile stress region at elastic singularities is enhanced as the hydrogen solute chemical potential and the free energy for hydride formation are reduced (25-271. Since the hydride phase is generally brittle, the crack propagates in a cleavage-like fashion through the hydride phase or along the hydride-matrix interface. A transition from a cleavage to a ductile fracture mode occurs as the strain rate is increased or the temperature is reduced [27] as a consequence of the inability to form hydrides ahead of a rapidly propagating crack. Post-fracture evidence suggests that hydrogen embrittlement in Ti occurs by the stress-induced hydride formation mechanism [28]. The situation in Ti-based systems is complicated by the existence of different phases, a and fl, which have different susceptibilities to hydrogen attack; the h.c.p. a phase being more susceptible to embrittlement and having a lower hydrogen solubility and diffusivity than the b.c.c. /l phase [28,29]. Alloys containing a mixture of the a and fi phases often fracture at the interface. of the a-/? phases [30]. In addition to the above complexities, three types of hydrides can form in the titanium matrix: 6, L and y hydrides. The 6 hydride (f.c.c., CaF, structure) forms between TiH,,, and TiH,,W [31-341 while the f.c.t. c hydride with c/a < 1 is formed below 310K near TiH, [32,35]. In the h.c.p. a phase matrix, at low hydrogen concentrations, the hydride formed is the metastable y f.c.t. hydride having a c/a = 1.09 [36,371 or c/a = 1.12 [38,39]. Additionally, a strain induced b.c.c. hydride has been reported [40] to form on the (lO’f0) slip plane during deformation of a solid solutions. In alloys with the a and fl phase the 6 hydride forms [41,42]. In pure titanium, hydrides form as lenticular plates with habit planes of {lOTO}, (lOT1) and {0225}, with the {lOTO} being the most common [36,43,44]. In Ti alloys, hydride habit planes have been reported to be {lOTO}, {OOOl}, {lOTl}, {lO’f2} and (1121) [40,4-l. In Ti-AI alloys, for example, the hydride habit plane is the {lOTO}for low and {OOOl}for high Al concentrations [44]. This correlates with the predominance of slip on { lOTO}at low Al concentrations and an increased tendency for slip on (0001) at high Al concentrations. This correlation may be a consequence of the 18% volume increase on forming the hydride from the a solid solution, as this is largely accommodated by plastic deformation [36]. In the Ti-4 wt % Al alloys used in the present study, both the {OOOl}and the (lOi0) hydride habit planes have been reported. In the present paper we direct our attention towards the general mechanisms of hydrogen em-

brittlement of the h.c.p. a Ti4 wt% Al alloy with the effects of hydrogen on the properties of the b.c.c. /? Ti-30 wt% MO discussed in a subsequent publication [48). The emphasis of this work is on microscopic studies of the crack tip phenomena carried out in an environmental cell in an HVEM. 2. EXPERIMENTAL METHODS The material used in this study was an equiaxed h.c.p a phase Ti-4 wt% Al alloy that contained 0.053 wt% Fe, 0.06 wt% 0 and 0.11 wt% N and had a basal texture [49]. Rectangular tensile specimens, having approximate dimensions of 10 x 3 x 0.1 mm, were cut parallel to either the basal or prism plane from a bulk plate. Prior to electropolishing, the samples were annealed in a vacuum of 6 x lo-’ Pa at 1173 K for 1 h. The central portion of the tensile sample was jet electropolished to perforation in a solution of 1 part (by volume) of perchloric acid, 6 parts n-butanol and 10 parts methanol at 250 K. The electron microscopy was performed in the Argonne National Laboratory’s high-voltage electron microscope operating at a voltage of 500 keV. Using this accelerating voltage allowed relatively thick samples to be used without producing significant levels of radiation damage; indeed, no evidence of radiation damage was found. An environmental cell which was differentially pumped and incorporated windowless apertures was installed in the objective lens of the HVEM and allowed up to 26 kPa of gaseous hydrogen to be maintained in the specimen region. The hydrogen fugacity at the sample is, however, between 30 and 710MPa because of dissociation and ionization of molecular hydrogen by the electron beam [50]. Straining of the sample was accomplished by a single-axis-tilt straining stage that stresses the samples in Mode I. The stage design allows for straining at a constant displacement rate, a fixed stage displacement, or maintenance of a constant load. The inability to achieve several different diffraction conditions due to use of a single-tilt straining stage, limited stage tilt due to the environmental cell, and preclusion of high-angle diffraction data due to fixed apertures in the environmental cell prevent diffraction contrast analysis experiments from being performed. Diffraction conditions were therefore selected to maximize the contrast of the features being investigated. Diffraction information on the hydrides formed during the in siru deformation experiments was obtained by post-fracture examination of samples in a Philips EM 420. Since these experiments address the dynamic behavior of deformation and fracture, much of the information was recorded on video tape and later analyzed on a frame-by-frame basis. These dynamic results can only be summarized and illustrated by successions of images reproduced from the video tape using image digitization and enhancement techniques.

SHIH et al.:

HYDROGEN

EMBRI~LEME~

3. RESULTS AND DISCUSSION 3.1 Nucleation and growth of hydrides

Growth of hydrides was observed in specimens which were plastically strained and then held at constant stress or at constant displacement in Hz gas at about 13 kPa. While hydrides formed preferentially in regions close to the crack tip and other elastic singularities, they were not contined to these regions. Their habit planes were the (0001) and the { 1070) as expected for the 6 hydride found in the Ti-4 wt% Al alloy. A field of hydrides, having several variants, which formed in the stress field of a crack is shown in Fig. 1. Lenticular hydrides (some examples are arrowed) can clearly be seen in the bright field micrograph in Fig. l(a) and in the dark field image, Fig. l(b), which was taken using the hydride diffraction spot arrowed in the diffraction pattern shown in the inset. Some hydrides are imaged as bright lines and internal structure can be seen. Other hydrides, belonging to

OF a TITANIUM

di&rent variants, may berendered visible by using the other hydride diffraction spots. The hydrides have an average length of 1.2 pm, although in some cases they grew to a length in excess of 4 pm. Development of large hydrides is achieved, not by the growth of one single hydride, but by a process of discrete nucleation, growth and coalescence of several hy drides on adjacent planes as shown in Fig. 2. The first hydride (labelled A) was formed ahead of a stressed, but stationary crack tip after 14 min in 16 kPa of gaseous hydrogen; the crack was initially propagated in hydrogen to ensure that it was under stress and to produce fresh metal surfaces to facilitate hydrogen entry. A second hydride formed and grew ahead of the first one which was now growing at a reduced rate. Coalescence of the second hydride (Iabelled B) with the first one is shown in Fig. 2(b). The formation and growth of a third hydride (labelled C) can be seen in Figs 2(c) and (d). Under a constant applied stress, the hydride growth rate was about 3 nm s-‘. Growth of the hydride plates, nucleation of new plates close

Fig. 1. A field of hydrides having several variants formed in the stress field of the crack. (a) Bright field and (b) dark field images. AM.%!I-”

113

114

SHIH et al.:

HYDROGEN

~B~~LEMENT

OF iz TITANIUM

Pig. 2. Hydride nucleation, growth and coalescence processes in front of the crack tip in 16 kPa H,. (a) Time, 15:49:39; (b) time, I5:56:30; (c) time, 15:59:20; (d) time 15:59:57. The stress axis is vertical.

to the plate tip and coalescence of the plates continued under stress provided the hydrogen environment was maintained. As the hydrides grew in length they also became thicker. Hydride formation is enhanced in the inhomogeneously stressed crack tip region as a result of the 24.2% vol increase on forming the d hydride from the a solid solution [43]. The soivus under stress, CfWa, at 300K is [51,52] rS0lVus.o

when the local stress is taken as nYS(uYS= yield stress, 524 MPa [49]). Hydride fo~ation in the stress field is autocatalytic. The stress field at the tip of a growing hydride may be described by the stress field of an obiate ei~psoid~ precipitate with the moiai volume change relative to the surrounding solid solution of 3cT, where cf is the stress-free transformation strain. As shown by Fsheiby f53] for an isotropic medium and later shown to be valid for anisotropic solids [54], the stress is uniform within an ellipsoidal inclusion and is given by (Fig. 3)

& (unrelaxed) = - 16,480 MPa

(3af

C& (relaxed) = - 524 MPa.

(3b)

The stress in the solid solution adjacent to the broad faces of the ellipsoidal hydride plate is also compressive. The stress field at the edge of a hydride plate may be modeled as that of a wedge loaded Mode I crack. Due to the expansion which accompanies the transformation to the hydride, the hydride is a self-stressed elastic inhomogeneity with the stress applied to the interface between the matrix and the hydride given by equation (2). The stress in front of the hydride plate lies between two limits as shown in Fig. 3. The upper bound is given by the case of pureiy elastic accommodation of the hydride for which the stress intensity is Kjlydtidc (upper) = a& (unrelaxed) [n ahydridc]‘/’ (4) where a&d,, is the half length of the hydride plate. The stress in front of the plate is given by

C$ (unrelaxed) = -B 6 u& (relaxed) < - uY,.

In the “relaxed” case the volume change on forming the hydride is accommodated partially by plastic deformation. The parameter B is the bulk modulus and v is Poisson’s ratio. Using the moduli for OLTi (B = 103 GPa) and the appropriate transformation strain, cT = 0.08, the values for the hydrides are

(2b)

az (upper) = [(ICWe tu~r))~t2~~)“‘l~te)

(5)

SHIH et ui.: HYDROGEN EMBRUTLEMENT

where r is the radial distance from the hydride plate edge and f (6) is a function of the polar coordinate, 8, measured from the plane of the hydride plate. The lower bound is determined by plastic accommodation of the volume change of the hydride plate for which the *‘effective COD’” is 6, = c’t,,* where thydridcis the thickness of the hydride plate. In this case, the hydride is treated as a Dugdale crack [55] for which the stress intensity is K:yd”‘d” (lower) = (6Tt,,ydti*uvSE)‘12

(6)

and the stress in front of the hydride is then given by a, (lower) = [KFydri*(lower)/(Znr)“*f g(8)

(7)

where g(0) is a function of the polar-coordinate, 6. For typical values of the parameters observed in the HVEM observations; ahydride = 2 pm thydridc = 0.2 pm, cT = 0.08, ayf= 524 MPa [49] and E = 103 GPa (for hydrides having a (~2) habit plane), the stress intensity of the hydride is bounded by ICiYtih (upper) = 41.3 MPa(m)‘p

(8a)

K:ydride(lower) = 0.93 MPa(m)‘/*.

(gb)

The values of a, calculated from equations (5) and (7) are shown in Fig. 3. Once the hydrides are nucleated, their associated volume change creates a stress field in front of the growing hydride plate which is similar to that of a Mode I crack, even in the absence of an external stress. Nucleation and growth of new hydrides in front of the growing plate (Fig. 2) follows directly from this self stress field of the plate. Once nucleated, growth of a hydride is expected to continue, even in the absence of an external stress, by this autocatalytic process provided the supply of hydrogen is sufficient

=22

t

OF 01TITANIUM

11.5

and the free energy of the system is decreased by hydride growth. Accommodation of the volume change associated with hydride formation can be elastic, by plastic deformation of the matrix or by a combination of the two, depending on the magnitude of the volume change and the morphology of the hydride [36, .56-58J. Numakura and Koiwa [36] suggested that the accommodation mode in Ti is dependent on the hydride habit plane; hydrides on the prism plane are elastically accommodated with the principal strain in the [lOfO], while hydrides with a (0225) habit plan are plastically accommodated by punching our shear dislocation loops on the basal plane. Elastic accommodation of hydrides in Ti-Al-H systems is however unlikely to be significant since a large thermal hysteresis exists between hydride formation and reversion suggesting that the dominant a~omm~ation mode is plastic [SS]. Hydrides growing in thin TEM specimens show both elastic and plastic accommodation. Figure 4(a) shows a bend contour emanating from the tip of the hydride plate, indicative of elastic strain. Much of the a~ommodation is, however, plastic. Dislocations emitted on a plane inclined to the growing hydride habit plane are shown in Fig. 4(b) while in Fig. 4(c) the dislocations are moving on the same plane as the hydride. The black-white lobe contrast of the dislocations in Fig. 4(b) are typical of pure screw dislocations lying parallel to the surface normal. Reversion of the hydrides was not observed either when the stress was reduced or when hydrogen was removed from the environmental cell. These hydrides remained even on storage in a dessicator and examination in the Philips 420 at a pressure < 1.3 x 10v4 Pa. The lack of reversion may be attributed to the plastic a~ommodation of the hydrides and to the surface oxide reducing permeation of hydrogen from the specimen. 3.2. Hydrogen effects on the fracture process in u titanium

16,CKXJEqn.20 -17,om

-

Fig. 3. Elastic stress field associated with a hydride. The parameters ia the calculation were: B = 103 GPa; f f = 0.08; hydride half length, a,,* = 2 pm; hydride thickness, lhydridc = 0.2 rum; yield strength, uyt = 524 MPa. -Elastic solution, equations (2a) and (5); ---Plastic accommodation of hydride AVjV,, equations (2b) and (7).

Two hydrogen-assisted fracture mechanisms were observed to operate in the a phase Ti4 wt% Al alloy with the operative mechanism determined by the crack-tip stress intensity. At low stress intensities, hydrogen embrittlcment occurred by repeated formation and cleavage fracture of hydrides. At high stress intensities, a transition occurred blinding to crack propagation rates which exceeded those at which hydrides could form in front of the crack. At these high stress intensities hydrogen embrittlement occurs by the hydrogen enhanced localized plasticity process. 3.2. I Fracture by the stress induced hydride cleauage mec~un~m. As described earlier, a field of hydride variants develops around a stationary, stressed crack. At a constant applied stress the hydrides grow but the crack remains stable. On increasing the applied stress, rapid crack advance proceeded through the hydride

116

SHIH et af.:

HYDROGEN

EMBRIlTLEMENT

Fig. 4. (a) Elastic and (b) and (c) plastic accommodation precipitation.

at rates which were too rapid to be recorded on video tape. Post fracture ex~ination of the hydride revealed that it had been cleaved and that the crack had propagated through the hydride phase (Figs 5 and 6). In Fig. S(b), the hydride can be seen on both sides of the cleavage crack indicating that the crack propagated through the hydride rather than along the hydride-matrix interface; the arrows indicate the position of the hydride-matrix interface. The original crack and hydrides (labelled I) are shown in Fig. 5(a). Scanning electron micro~phs show that the fracture surface of the hydride is a featureless cleavage surface, indicating a brittle fracture with little accompanying plasticity as is consistent with the lack of deformation seen in TEM images of the hydride and the limited strain to failure reported for the hydride [46]. Crack propagation in the presence of hydrides is similar to fracture of other systems in which stressinduced phase transformations occur fS9,60]. As previously discussed, the volume expansion which accompanies hydride plate formation results in a stress field at the hydride tip which is similar to, and of the same sign as, the stress field of the crack. As

OF a TITANIUM

of the volume expansion during hydride

a result, the stress field between the crack tip and the hydride plate in front of the crack is increased and either the hydride grows towards the crack tip or the crack advances into the hydride. At that point the local stress intensity at the crack tip is reduced f59] due to the compressive stress in the hydride which surrounds the crack tip. Prior to the formation of a hydride at the crack tip the maximum normal stress, (122,may be limited to a value of - 2.5 bvSnear the crack tip where it is limited by plastic flow or may exceed 2.5 eY3 due to the limited number of slip systems available in the Ti alloys. When the crack tip is within the hydride the local stress intensity is reduced by [59]:

with a co~s~nding reduction of uu. The actual maximum cr22will depend on the extent of plastic deformation at the crack tip, the extent of plastic accommodation of the hydride volume change, and whether the crack tip may be described by plane stress or plane strain conditions.

SHIH ef al.: HYDROGEN EMB~~LEME~

OF

K

TITANIUM

117

A

Fig. 5. (a) Hydride plates forming ahead of the crack tip, (b) brittle crack advancing through the hydride plate [labelled 1 in Fig. 5(a)] rather than along the hydride-matrix interfaces. Based upon the above, the fracture m~h~ism is as shown schematically in Fig. 6. At K, < K& for the solid solution, Fig. 6(a), stress-induced hydride growth occurs in front of the crack because of the influx of hydrogen to the crack tip region. Once nucleated, these hydride plates grow, largely influenced by their own stress fields. If the hydride plates do not start at the crack tip, the increased stress due to the combination of the stress fields of the crack and the hydride results in either crack advance into the hydride or hydride growth to the crack. At

WK%

Fig. 6. Schematic diagram of the stress-induced hydride fracture mechanism. (a) Flux of hydrogen to the crack tip region due to the decrease of the hydrogen chemical potential in solid solution under the applied stress. (b) Growth of a hydride at the crack tip. The volume expansion during hydride formation reduces the local stress intensity and the crack remains stable. (c) Increase in the applied tensile stress resulting in cleavage fracture of hydride plates.

that point, Fig. 6(b), the local stress intensity at the crack tip, Kid, is reduced and may become negative due to the elastic stress field of the hydride behind the crack tip. The crack remains stable at constant applied stress while the hydride continues to grow [Fig. 6(b)]. Crack propagation occurs if the applied stress is increased to the point where K,, > KhI$‘“’ [Fig. 6(c)]. At that stress, the crack propagates through the hydride plate until it reaches the a solid solution. Since K,“,> Kgw”, the crack stops and the above process is repeated. The stress intensity is increased because the crack length is increased by the length of the hydride and the external stress is increased to reinitiate crack motion in the hydride. This stress-induced hydride cleavage mechanism will occur repeatedly unless: (a) the supply of hydrogen is limited so that no further hydrides can form, or (b) the stress intensity of the crack when it reaches the hyd~delsolid solution interface exceeds the value required to initiate fracture by an alternative mechanism. While the cleavage plane’s indices were not specifically determined in the present experiments, the observation that the cleavage crack traversed the hydride in front of the crack is consistent with near-basal or near-prism cleavage as these were the hydride habit planes. Reported deviations from basal or prism plane fracture in hydrogen embrittlement or S.C.C. [61-63] may result from the repeated hydride formation and cleavage processes and from the fact

118

SHIH et al.:

HYDROGEN

EMBRITTLEMENT

OF a TITANIUM

SHIH er al.:

HYDROGEN

EMBRITTLEMENT

that a field of hydrides forms in front of the cra6k tip. The loading conditions may be expected to infiuence how the hydrides are distributed and which are selected for crack propagation; small deviations may thereafter be expected for the macroscopic fracture plane. 3.2.2. Fracture by the hydrogen enhanced localized plasticity mechanism. At high crack advance rates, where repeated hydride formation is not possible, the fracture beyond the hydride field is by the hydrogen enhanced localized plasticity mechanism similar to that previously reported for Ni 19, lo], Fe [1 1, 121and Al [13]. An example of this transition from a brittle (hydride cleavage) to a hydrogen enhanced localized ductility crack advance mechanism is shown in Fig. 7. In this mechanism, the effect of hydrogen is to reduce the stress required to cause dislocation motion and to increase the mobility and density of the mobile dislocations at a constant stress. Since these effects reflect the local hydrogen concentration, which is highest at elastic singularities and at points where hydrogen can most easily enter from the atmosphere (e.g. notches), hydrogen enhanced plasticity is localized at these points and thereby results in slip localization, constrained plasticity, and localized plastic fracture processes. As these occur at stresses below those required for general yielding, the net material

OF a TITANIUM

119

behavior is a “brittle” response to stress. In the in situ HVEM deformation experiments this process is manifested by a localized deformation zone which develops ahead of the crack. The crack propagates along slip planes within the confines of this zone by the direct emission of dislocations from the crack tip. The thinning can be quite extensive, making regions of samples that were initially opaque transparent to the electron beam. An example illustrating the extent of the deformation zone is shown in Fig. 8. Figure 9 is a sequence of micrographs taken from the video tapes showing a crack advancing as a result of adding hydrogen to the environmental cell with the specimen displacement constant. The effect of hydrogen was to cause dislocations which were immobile in vacuum to begin to move and new dislocations to be emitted from the crack tip and nearby sources. Detailed examination of the crack tip, at high magnification, did not show any evidence of hydride formation. Crack advance was due to the rapid motion of dislocations near the crack tip which started and stopped as hydrogen was added and removed from the environmental cell. It was not possible to capture the crack tip activity on video tape due to the speed at which it occurred, but the formation of a dense dislocation structure in front of the crack tip is shown in Figs 9(a) and (b). Crack

Fig. 8. Crack tip region for fracture in 16 kPa H, gas. Plastic thinning and hole formation are shown in front of the crack tip.

120

SHIH ei crl.: HYDROGEN EMBRITI’LEMENT OF a TITANIUM

Fig. 9. Crack tip advance as a result of adding H, to the environmental cell. Arrow “1” indicates points of intense dislocation activity. Arrows “2” and “3” indicate a t&d point and the crack, respectively. (a) Time, f7:OP:39; (b) time, 17:OP:51; (c) time, 17:10:01, (d) time, 17:10:03.

advance was by a locally plastic behavior in which plastic deformation occurred at the crack tip at stresses lower than that required for deformation in vacuum. Removal of hydrogen from the environmental cell caused the dislocation motion to cease and crack advance to stop. This process could be repeated with the crack advancing when hydrogen was added and stopping when it was removed from the environmental cell. The fracture generally occurred in a tmnsgranul~ manner along slip planes although instances of intergranular fracture were noted. As shown in Fig. 10, intergranular fracture occurred in the region adjacent to the grain boundary along slip planes rather than in the grain boundary itself; similar observations have been made in other systems [lo]. The mechanism of fracture described above, hydrogen enhanced localized plasticity, is similar to the mechanism of crack advance which operates in inert environments but with the important difference that the stresses needed to move dislocations and therefore propagate cracks are significantly lower when hydrogen is present. This decrease of the flow stress is localized to the volume-near the crack tip where the hydrogen concentration is highest. This localization of slip and the plastic fracture leads to the impression that the fracture surface is “brittle” when viewed macroscopically. Enhancement of the dislocation mobility by hydro-

gen can be seen qualitatively in the micrographs presented in Fig. 11 which were taken from the video tape. On stressing at a constant displacement rate the dislocations moved with an average velocity of 5 x lo-*fims-’ in vacuum [Figs 11(a) and (b)]. The stage displacement was then stopped, the sample heid at constant stress, and hydrogen introduced into the environmental cell; the dislocation velocity increased to an average value of 0.12 #rn s-’ with 13 kPa of hydrogen in the cell [Figs 1 l(c) and (d)]. On removal of the H, gas from the environmental cell the dislocation velocity decreased significantly and eventually stopped; compare Figs 1l(d), (e) and ( f). The dislocations could be made to move again at constant stress by reintroducing hydrogen to the environmental cell; the average dislocation velocity was 7.6 x 10e2 pm s-i at 13 kPa of Hz [Figs 11(g) and (h)]. Dislocations were observed to move in both directions. The dislocations labelled “0” in Fig. I l(h) were emitted from a source, indicated by arrow “A” and then moved in opposite directions. The effect of hydrogen on dislocation velocity is shown quantitatively in Fig. 12 where the ratio of dislocation velocity in hydrogen to that in vacuum, (V,/v,) is shown as a function of the external hydrogen pressure. The velocity increase when hydrogen was introduced to the cell for the first time, (Curve I), is greater than the velocity increase after hydrogen has been removed and then re-introduced

SHIH et OZ.: HYDROGEN

Fig. 10. “Intergranular”

EMBRITTLEMENT

OF a TITANIUM

fnrcture in 16 kPa H,. The crack advances by slip band cracking in the deformation band adjacent to the grain boundary.

into the cell (Curve II). This difference is probably due to the stress having relaxed after the initial introduction of hydrogen. The process by which hydrogen reduces the stress required to move disl~ations is still not fully understood. A reduction in the Peierls’ barrier can be eliminated since the effect occurs in a number of f.c.c., h.c.p. and b.c.c. materials. It is unlikely that it is an intrinsic mechanism, since the magnitude of the effect is often greater in materials containing a concentration of impurities which act as barriers to dislocation motion. A possible explanation, which is currently being investigated, is that the association of hydrogen with dislocations reduces the dislocation stress field, thereby reducing the elastic interaction energy between dislocations and obstacles. This reduction in the interaction energy would result in an increased dislocation mobility and a reduction in the stress to unpin dislocations from the solute pinning points.

3.3 Relation of present results to macroscopic behavior Considerable evidence for the association of hydrides with the fracture of a and near a Ti alloys has been obtained for failure under aqueous stress corro-

sion cracking and in gaseous hydrogen atmospheres as well as for alloys which contain solute hydrogen [28,64-&i]. The role of hydrides has not, however, been definitely established although a fracture mechanism based on hydride film formation has been proposed. The difficulty has been that while hydrides have been associated with crack tips [56] and with the fracture surfaces under some conditions (28,6466], the fracture surface either corresponds to the a-8 interfaces in the Widmanstatten structure [30,671 or to a plane which is about 15” from the basal plane 168,691. The lack of agreement of the fracture surface with the hydride habit and cleavage plane may reflect the hydride distribution and the presence of steps along the fracture surface. In the a+ Widmansdtten alloy the /3 phase has a much higher H diffusivity than the a [70] and hence acts as the primary path for H transport. Since under the fracture conditions hydrides are stable in the a phase [42] we may expect the hydrides to form along the a+ interface. When subjected to an external stress this hydride may cleave along its (0001) although the macroscopic fracture will follow the hydride distribution, i.e. the a-/l interface. Similarly the near basal plane transgranular fracture may actually consist of basal plane sections connected by the

122

SHIH et al.: HYDROGEN EMBRITTLEMENT OF a TLTANHJM

Fig. 11. Dislocation motion in the uniform material ahead of a crack tip. Arrows indicate fixed reference points. (a) Time, 16:55:37 and (b) time, 16:55:42 vacuum; dislocation velocity is 5 x 10e2 pms-r; (c) time, 16:55: 57 and (d) time, 16:56:05 13 kPa H,; dislocation velocity is 1.25 x 10-r pm s-l; (e) time, 16: 56: 32 and ( f) time, 16: 56: 37 vacuum; dislocation velocity is effectively zero; (g) time, 16: 56: 52 and (h) time, 16: 56:58 13 kPa H,; dislocation velocity is 7.6 x lo-* ym s-r.

cleavage steps to produce an average fracture plane which deviates from the (0001) as repeated hydride plates are nucleated and then cleaved. In the present experiments slow crack propagation in the a phase is clearly consistent with the hydride formation and cleavage mechanism. The observation of a field of hydrides in the various variants of the

(0001) and { lOTO}habit planes in front of the cracktip (Fig. 1) in the thin specimens is typical of what may be expected in thick specimen cross sections. In addition, hydride variants with the other reported habit planes [36,40,43,44] may form in the thick section, crack tip stress field despite their lack of observation in the thin HVEM specimens. In the

SHIH et of.: HYDROGEN EMBRITTLEMENT OF a TITANIUM

;,1, , , , , , , , , ,I Hi

F&d

10 In 6EmAmwrtol

12 14 16 16 Cell, 103 Pascal

Fig. 12. The effect of hydrogen on dislocation velocity.

presence of a complex hydride field the crack can propagate along the cleavage planes of the various hydride variants and produce a relatively complex fracture surface, as is observed in a titanium alloys in the presence of hydrogen [66]. Each fracture facet may be a cleavage plane and yet the average fracture plane may deviate from the (0001). As the crack is forced to propagate at higher velocities the present results show that the mechanism changes to the hydrogen enhanced localized plasticity mechanism. In a macroscopic specimen this transition would be expected to occur as the stress intensity and hence the macroscopic crack velocity increased and exceeded the rate at which the hydride stress field can form. The effect of hydrogen enhanced dislocation velocities is to produce. a localized softening of the region just in front of the crack tip where the hydrogen concentration is the greatest. This localized softening leads to highly ductile fracture but only in the region along the slip planes in front of the crack tip [71]. It is this hydrogen enhanced softening and slip localization which distinguishes this type of fracture from the more general ductile behavior which occurs at still higher K, (higher crack velocity) and leads to a ductile microvoid coalescence type of fracture surface. Evidence for this fracture mode, intermediate between the low stress intensity hydride related fracture and the high stress intensity general ductility microvoid coalescence failure, has generally not been sought in macroscopic fracture experiments. 4. SUMMARY AND EXPERIMENTS Hydrogen-assisted fracture mechanisms in the h.c.p. a Ti-4 wt% Al alloy have been established with the use of in situ environmental cell HVEM techniques. The nucleation and growth of hydrides at elastic singularities has been observed directly and the accommodation of the stress field around the hydride has been shown to be partially elastic and partially plastic. The stress field at the hydride tip was discussed and shown to result autocatalytic growth of the hydride plates. Two hydrogen-assists fracture mechanisms have been shown to be operative in a Ti. At low stress

123

intensities, corresponding to small crack velocities, the fracture procee& by ‘;a stress-induced hydride formation and cleavage mechanism. Consideration of the stress at the crack tip in the presence of hydrides has shown that hydride fo~ation decreases the local stress intensity at the crack tip and as a result continued fracture requires an increase in the external stress. Once the local stress intensity is increased to exceed the critical stress intensity for hydride fracture, the crack can proceed by repeated hydride nucleation, growth and cleavage. At high stress intensities, corresponding to crack velocities higher than can be a~ommodat~ by hydride cleavage, failure occurs by hydrogen-assisted localized plasticity at the crack tip. The failure is by a constrained plastic mechanism in which the stress for plastic flow is locally reduced by a high hydrogen concentration. Direct obseNatiOnS of dislocations moving under constant stress have shown that their velocities are enhanced by hydrogen in solid solution in eq~lib~~ with the gaseous atmosphere. Acknowledgemen&-The authors wish to acknowledge the support of the Department of Energy through contract DE AC02-76ER01198 and through its support of the Center for the Microanalysis of Materials at the University of Illinois and the HVEM Center at the Argonne National Iaboratory. Also, the assistance of Mr E. Ryan and Mr A, Philippedes at the HVEM Center is greatly appreciated. The authors would like to thank Professor J. C. Williams of Carnegie-Mellon University for supplying the bulk Ti-4Al alloy.

REFERENCES 1. E. A. Steigerwald, F. W. Schaller and A. R. Troiano, Tmns. meta&. Sot., A.Z.M.E. 21% 832 (1960). 2. A. R. Troiano, BZSXA, The iron and Steel Inst., Horrogate Court, p. 1 (1961). 3. R. A. Griani, Proc. Znti. Co& on Stress Corrosion Cracking and Hydrogen Embr&lemenr of Iron Based Alloys, Unie~-Finns. June (19731. 4. R. A. Griani and P. HrJosephic, A& metall. 22, 1065 (1974). 5. R. A. Oriani and P. H. Josepbic, Acta metall. 25, 979 (1977). 6. J. R. Rice and R. Thompson, Phil. Mug. 29,73 (1974). 7. R. P. Messmer and C. I. Briant, Acta met& 30, 457 (1982). 8. C. D. Beachem, Metall. Trans. 3, 437 (1972). 9. T. Matsumoto, J. Eastman and H. IL Birnbaum, Srr@a mefall. 15, 1033 (1981). 10. I. M. Robertson and H. K. Bimbaum, Rcru metalt. 34, 353 (1986). Il. T. Tabata and H. K. Birnbaum, Scripta metall. 17,947 (1983). 12. T. Tabata and H. K. Bimbaum. Scr&ra met& I& 231 (1984). 13. G. M. Bond, I. M. Robertson and H. K. Bimbaum, Acta meraN. 35. 2289 (1987). 14. 1. M. Robertson and HI K. Bimaum, unpublished work. 15. J. Eastman, T. Matsumoto. N. Narita, F. Heubaum and H. K. Bimbaum, Hydrogen Effczs in Met& (edited by I. M. Bernstein and A. W. Thompson), p. 397. A.I.M.E. New York (1980). 16. W. Wei, PhD. thesis, University of Illinois, Urbana, Ill. (1983). 17. F. Heubaum, M.S. Thesis, University of Illinois, Urbana, Ill. (1981).

124

SHIH et al.:

HYDROGEN

EMBRITTLEMENT

18. F. Zeides and H. K. Birnbaum, to be published. 19. s. P. Lynch, scripro merall.13, 1051 (1979). 20. S. P. Lynch, Fracture (edited by D. M. R. Tap&), Vol. 2, p. 859 (1979). 21. S. Takano and T. Suzuki, Acre mezall.22, 265 (1974). 22. S. Gahc, M. L. Grossbeck and H. K. Birnbaum, Acta med. 36, 125 (1977). 23. R. Dutton, N. Nuttall, M. P. Pub and L. A. Simpson, Metall. Truns. 8A, 1553 (1977). 24. R. D&ton, C. H. Woo, K. Nuttall, L. A. Simpson and M. P. Pub, ti Inf. Conf. on ffytir~er~ in Mehzh, Pergamon Press, Oxford, paper 3C6 (1977). 25. D. G. Westlake, Trans. Am. Sot. Metals 62, 1000 (1969). 26. H. K. Birnbaum, M. L. Grossbeck and S. Gahr, Hydrogen in Metals (edited by I. M. Bernstein and A. W. Thompson). n. 203. Am. Sot. Metals, Metals Park, Ohio (i974$’ 27. M. L. Grossbeck and H. K. Birnbaum, Acta metalL 25, 135 (1977). 28. N. E. Paton and J. C. Williams, Hydrogen in Metals (edited by I. M. Bernstein and A. W. Thompson), p. 409. Am. Sot. Metals, Metals Park, Ohio (1974). 29. R. J. Wasilewski and G. L. Kehl, Metullurgkt 50, 225 (1954). 30. D. S. Shih, Ph.D. thesis, Georgia Institute of Technology, Atlanta, Ga (1983). 31. S. S. Sidhu, L. Heaton and D. D. Zauberis, Acta Crystallogr. 9, 607 (1956). 32. H. L. Yakel Jr, Acta Cryst. 11, 46 (1958). 33. P. E. Irving and C. J. Beevers, Metall. Trans. 2, 613 j1971). 34. P. Hirsch and M. Givon, J. ~~~o~on Metals 87, 179 (1982). 35. Z. M. Azarkh and P. I. Gavriiov, Soviet Phys. Crysrallogr. 15, 231 (1970). 36. H. Numakura and M. Koiwa, Acta me&l. 32, 1799 (1984). 37. H. Numakura, M. Koiwa, H. Asano, H. Murata and F. Izumi, Scr@ta rnetalf. 28,213 (1986). 38. 0. T. Woo, G. C. Weatherby, C. E. Coleman and R. W. Gilbert, Acta mera#. 33, I897 (1985). 39. 0. T. Woo and G. J. C. Carpenter, Script0 metall. 19, 931 (1985). 40. J. D. Boyd, Trans. Am. Sot. Metalls 62, 977 (1969). 41. M. J. Trzeciak, D. F. Dilthey and M. W. Mallet& USAEC Reoort BMI-1112. Battelle Memorial Institute (1956). = 42. D. S. Shih and K. H. Bimbaum, Scrinta metalf. 20.1261 (1986). 43. T. S. Liu and M. A. Steinberg, Tram Am. Sot. Metals 50. 455 (1958). 44. N..E. Paton and R. A. Spurling, Metall. Truns. 7A, 1769 (1976).

OF a TITANIUM

45. P. E. Irving and C. J. Beevers, J. Mater. Sci. 7, 23 (1972). 46. I. W. Hall, Metal, Trans. 9A, 815 (1978). 47. C. Hammond, R. A. Spurling and N. E. Paton, Metall. 7’bans.HA, 813 (1984). 48. D. S. Shih and H. K. Bimbaum, to be published. 49. H. M. Kim, Ph.D. thesis, Carnegie-Mellon University, Pittsburgh, Pa (1981). 50. 0. M. Bond, I. M. Robertson and H. K. Bimbaum, Scrivta metall. 20. 653 11986). 51. J. d. M. Li, R. A. Or&i aid L. S. Darken, Z. Phys. Chem. Neue Folge 49,271 (1966). 52. T. B. Flannagan, N. B. Mason and H. K. Bimbaum, Scripra metall. 15, 109 (1981). 53. J. D. Eshelby, Proc. R. Sot. 241, 376 (1957). 54. L. J. Walpole, Proc. R. Sot. A300, 276 (1967). 55. D. S. Duadale. J. Me&. Pkvs. Solids 8. 100 119601. 56. N, E. Pat&, B: S. Hickman &d D. H. Leslie, Leroy. Trusts.2, 2791 (1971). 57. H. K. Bimbaum, M. L. Grossbeck and M. Amano, J. less-common Metals 49, 357 (1976). 58. B. J. Makenas and H. K. Bimbaum, Acta metall. 28, 979 (1980). 59. R. M. McMeeking and A. G. Evans, J. Am. Ceramic SW. 65. 242 (1982). 60. J. C. M. Li and S: C. Sanday, Scripta metall. 19, 935 (1985). 61. M. J. Blackbum and J. C. Williams, Fu~~&tnenrd Aspects of Stress Corrosion Cracking (edited by R. W. Staehle, A. J. Forty and D. van Rooyen), p. 620. National Ass. of Corrosion Engineers (1869). _ 62. J. C. Laziou. J. less-common Metals 46. 251 fl9711. 63. D. A. Meyn,‘~et~l. Trans. 3, 2302 (197i). \ ’ 64. G. H. Koch, A. J. Bursle and E. M. PI@, Metall. Trans. IZA, 1833 (1981). 65. W. J. Pardee and N. E. Paton, Metall. Trans. llA, 1391 (1980). 66. A. J. Bursle and E. N. Pugh, Environmental-Sensitive Fracture of Engineering Materials (edited by 2. A. Forouiis). D. 18. TM& Warrendale. Pa (19791. 67. G. H. Koch, A. J. B&e and E. N. &r&P&. &dZnt, Congress on Hydrogenin Metals. paper 3D4. Pergamon Press, New York (1977). 68. M. J. Blackbum and J. C. Williams, Fumabmental Aspects of Stress Corrosion Cracking, p. 684 NACE, Houston, (1969). 69. D. A. Meyn and G. Sandoz. Trans. T.M.S.-A.I.M.E. 245, 1253 (1969). 70. W, R. Holman, Trans. T.M.S.-A.I.M.E. 233, 1836 (1965). 71. P. Sofronis, H. K. Birnbaum and R. M. McMeeking, to be published.