Previously published in METALLOGRAPHY 5:235-250 (1972)
Fracture Modes in a Binary Titanium Alloy
V. CHAA-DRASEKARAh-, R. TAGGART, ANDD. H. POLONIS Co/legeof Engineering,FB–Io Unic,evsit>of Washington,Seattle, Washington,98195
The scanning electron microscope has been used to int,estigatethe fracture surfaces of Ti–Cr alloys, and special emphasis has been placed on the effects of the omega phase on the modes of fracture. It has been found thatan alloy containing the athermalomega phase exhibits a differentmode of fracturefrom thatof an alloy containing the aged omega phase. The difference has been attributed to the presence of the strain-induced martensite phase in regions adjacent to the fracture in alloys containing the athermal omega phase. The fracturemode has also been investigatedfor alloys in which the omega phase was reverted b~.appropriatethermal treatments.
Introduction In order to achieve high strengths and ductility in titanium alloys, it is necessary to avoid the formation of the transition omega phase either by making a suitable choice of alloying elements or by adopting appropriate thermal treatments. In certain alloys with less than a critical composition, the omega phase can form during quenching, no matter how fast the quench, in which case it is referred to as the athermal or diffuse omega [1–3]. When the alloy content is increased above a critical value, the metastable beta phase is retained  and by aging at temperatures over the range of 100–500°C the metastable beta decomposes into beta plus omega. Both the athermal and the aged forms of the omega phase are coherent with the beta matrix . The hexagonal structure of the omega phase [6, 7] can be reverted to a bcc structure  by holding for a short time at a temperature above its range of stability. The reversion of the omega phase leaves solute lean regions in an enriched beta matrix ; the solute lean regions are coherent with the matrix and contribute to the matrix strain, the magnitude of which depends on the reversion time. The matrix strains are only slightly diminished after holding for a few minutes at the reversion temperature. In the present work the scanning electron microscope has been used to examine the effect of the microstructure on the fracture mode of Ti–Cr alloys containing
183 MATERIALS CHARACTERIZATION 39:183-198 0997) @ElsevierScienceInc.,1998 655Avenue of the Americas,New York,NY1OO1O
1044-5803/97/$17.00 PII S1044-58O3(97)OO131-9
V. Chandrasekaran,R. Taggart, and D. H. Polonis
the beta phase and mixtures of the beta and omega phases that have been modified by aging and reversion treatments. An important aspect of this investigation is the superior mechanical properties of Ti–Cr alloys in which the omega phase has been reverted by thermal treatments. The role assumed by strain-induced martensite has been examined in relation to the ductility and fracture mode of as-quenched alloys containing the athermal omega phase. Experimental Procedure Ti–Cr binary alloys containing 10 l/o Cr and 15 a/o Cr were prepared in an atmosphere of dry He by a standard levitation melting technique. Tensile tests were performed with an Instron testing machine on flat tensile specimens 0.40-mm thick and having gage lengths of 2.0-3.0 cm; all the tests were conducted at room temperature. The fractured surfaces were examined in a Cambridge Stereoscan scanning electror, microscope operated in the secondary mode, using an accelerating potential of 20 Kv. Prior to any heat treatment, the specimens were encapsulated in vycor tubes to a vacuum of 5 x 10–6torr, and the solution heat treatments were performed at 950°C for 45 minutes. The heat treatment schedule shown in Table I was employed to produce the desired microstructure. TABLE I Schedule of Heat Treatments Alloy Composition
1. Ti–15 a/o Cr 2. Ti–15 a/o Cr
SHT 950°C/45 rein, WQ SHT 950”C/45 rein, WQ aged 350 ‘C/2.5 hours SHT 950°C/45 rein, WQ aged 350‘C/2.5 hours +450°C15 min SHT 950°C’/45 rein, WQ
3. Ti–15 a/o Cr
4. Ti–10 a/o Cr
Microstructure All beta Beta+ omega (aged) Betal~an+betae~riCb,d
Beta+ athermal omega
Results The Ti–15 a/o Cr alloy contained only the beta phase after solution heat treatment and quenching to room temperature. The tensile-test results showed a 0.2./0 yield Strength of 129,000 psi and ductility corresponding to 17~0 elongation. All of the as-quenched
specimens exhibited a ductile mode of fracture as
shown in Figs. 1 and 2. Figure 1 indicates the mixed nature of the fracture
IT. ChandrasekaYan, R. Taggart, and D. H. Polonis
mode, while the elongated dimples revealed in Fig. 2 show that there has been considerable local plastic deformation. The Ti–15~& Cr alloy aged at 350”C to produce a mixture of the beta phase and the aged omega phase, exhibited a fracture stress of 156,000 psi and only 2~& elongation. Such brittle behavior is consistent with that reported in other titanium alloys containing the aged omega phase [9–1 1]. Figure 3(a) is an example of
FIG. 3a. .Appearanceof the fracture surface in a Ti–15% Cr alloy containing the omega phase after aging at 350”C for 2.5 hours. Mug x 2000. the appearance of the corresponding fracture surface at high magnification. It has been reported previously  that it is difficult to correlate the network of fine dimples on the fracture face with the size and spacing of the aged omega particles in the matrix, and this is also true for the present alloy. Figure 3(Zr) shows the slip lines that appeared only in the grains adjacent to the fractured surface, indicating that localized plastic deformation has occurred prior to fracture. In some grains the slip lines were straight, in others wavy slip occurred and some evidence of intersecting slip lines was also observed. The omega phase produced by aging a Ti–15°/0 Cr alloy at 350”C for 2.5 hours was reverted by heating to 450°C and holding for a period of five minutes. Following a heat treatment of this type it has been possible to obtain a 0.2% yield strength of 166,000 psi and a ductility of 6~0. Figure 4 characterizes the
V. Chandrasekaran,R. Taggart, and D. H. Polonis
fracture topography of this alloy when the omega phase has been reverted, and it is seen that the fracture surface consists of a network of dimples. A comparison of Figs. 3(a) and 4 show that the dimple size in the two cases is not noticeably different. The Ti–10 a/oCr alloy contains both the metastable beta phase and the athermal omega phase in the quenched condition following solution heat treatment. The metastable beta phase in this alloy undergoes a strain-induced martensite transformation during tensile deformation. Tensile tests on the asquenched 10~0 Cr alloy revealed a 0.2~0 yield strength of 116,000 psi and a ductility of 13’7”. The strain-induced martensite was not confined to the area immediately adjacent to the fracture but was distributed uniformly over the gage section. Scanning electron micrographs taken remote from the fracture surface are shown in Figs. 5(a) and (b). The strain induced martensite is characterized by
FIG. 5a. Ti–10~o Cr alloy in the as quenched condition, exhibiting strain-induced martensite. The martensiteis characterizedby internal twins and initiation of secondary plates. Mug x 775. internal twinning of the plates and by the initiation of secondary martensite as shown in Fig. 5(a); the propagationof the martensiteplates across a grain
boundaryis shownin Fig. S(b).
Fracture Modes in a Binay TitaniumAlloy
FIG. 5b. Same alloy as in Fig. 5a, showing the propagation of the strain-induced martensiteacross a grain boundary. Mug x 775. A total view of the fracture surface is presented in Fig. 6. Starting from the right-hand corner of the fracture face”there are relatively flat regions interspersed with a low density of dimples. Toward the center of the fracture surface the concentration of dimples increases considerably. The dimples appear to be elongated in the direction corresponding to the specimen thickness, while the dimples appear to be smaller and less dense in the foreground of the fractured face. The fracture surface on the left-hand side exhibits a very angular topography. Figure 7 shows an enlarged view of the dimple network in the central fracture area marked (X) in Fig. 6, and it is seen that the dimples are fringed by cusped ridges. Figure (8a) which is a closeup of the fracture region marked (Y) in Fig. 6, shows no sign of dimples but only a series of coarse ridges at different levels traversed in some cases by secondary bands that maybe attributed to the presence of the strain induced martensite. It is evident from Figs. 9 and 10 that the fracture process has undergone a transition from ductile to brittle behavior. In Fig. 9 there is evidence of intergranular fracture and both Figs. 9 and 10 indicate that the fracture path has been influenced by the plastic deformation of the grains adjacent to the fracture surface.
V. ChandYasekaran, R. Taggart, and D. H. Polonis
Composite view of the fracture surface in a Ti-10 a/o Cr alloy in the led condition. l!lag x 25. quenck
Fracture Modes in a Binary TitaniumAlloy
FIG. 7. Enlarged view of the region marked X in Fig. 6. The size and depth of the dimples are seen to vary from the foreground to the background. Magx 775. Discussion
Ti-15~o Cr Allo>’ The formation of cavities in the all beta Ti-15~o Cr alloy can be explained on the basis of stress concentrations resulting from the dislocation substructure within the material. The surface of the Ti–15‘jJo Cr alloy tensile specimen in regions adjacent to the fracture zone showed a high density of slip lines. Over the gage length, some grains exhibited one set of parallel slip lines, indicating that the deformation occurred on a single-slip system, while other grains revealed intersecting slip lines characteristic of duplex-slip processes. In other grains there was evidence of wavy slip. Secondary cracks at grain boundary triple. points and along grain boundaries were discernible on the specimen surface close to the fractured edge. The termination of slip lines at the grain boundaries lends credibility to a model in which fracture is initiated by dislocation pile-ups occurring at the boundaries, according to the model suggested by Stroh .
V. Chandvasekavan, R. Taggart, and D. H. Polonis
In the presentstudy,transmissionelectronmicroscopyhasbeenusedto show that the slip systemin the beta phaseof the Ti-15~~ Cr alloyis the (110)- type. Interactionsbetweena)2  dislocationsmovingon intersecting(110) planesand resultingin the formationof a networkof dislocationsof the type a[OOl]have been identifiedin the all beta commercialalloy Ti–13V-–llCr–3Al . This dislocationreactionhas been used by Cottrell  as the basis for a crack nucleation model, and would also be applicable in the present alloy. The appearance of dimples on the fracture surface of a material that fails in a brittle fashion is not expected unless there is localized plastic flow accompanying the fracture process. The Ti–15 Cr alloy aged at 350°C for 2.5 hours to produce a mixture of the beta and omega phases failed in tension with only 2’3~elongation. The corresponding fracture surface exhibited the dimpled texture shown in Fig. 3(a), which is in close agreement with the observations reported by Williams [9, 15] and Feeney  for Ti-11.6 w/o Mo, Ti-8 w/ohlo and Beta-III containing a mixture of the beta and omega phases. Feeney and Blackburn  have reported that slip is confined to the locality immediately adjacent to the fracture surfaces in the commercial &III alloy. This type of deformation could intensify the local stress level in the vicinity of the omega phase and Williams et al.  have cited that the dimpled fracture observed in titanium alloys containing the aged omega phase could be due to the accumulation of a critical local stress at the beta–omega interfaces. Transmission electron microscopy evidence was presented to demonstrate that dislocations do not penetrate the omega-phase particles but tend to accumulate in the betaphase matrix. In the aged Ti–15~& Cr alloy the localized nature of the slip lines indicates that concentrated plastic flow has accompanied the process of fracture, as shown in Fig. 3. The evidence of localized plastic flow and the confinement of dislocation interactions to the beta matrix provides a mechanism to account for the nucleation and growth of voids close to groups of omega-phase particles during the fracture of alloys containing mixtures of the beta and omega phases. The omega phase is coherent with the beta matrix  but it has a larger specific volume than the matrix. Colonies of the omega phase can form with a common variant as evidenced by the fact that prolonged aging of a Zr–Nb alloy produces large omega particles due to the coalescence of smaller particles . Accordingly there will be compressive residual stresses generated within the omega phase and tensile residual stresses in the surrounding matrix. Under an applied tensile load, voids can nucleate if the local resolved shear stresses exceed the fracture stress at the outer periphery of colonies of omega particles having a common variant. Such voids are therefore nucleated in the beta phase according to the model proposed by Cottrell  and grow into voids, the size of which is dictated by the distribution of the colonies of the omega precipitate. There is no apparent relation between the size of the aged omega particles and the size of the dimples observed on the fractured face as shown in Fig. 3a.
Fracture Modes in a Binary TitaniumAlloy
Ratherthanassociatingeachomegaparticlewitha dimple,it is moreappropriate to considera clusterof omega-phaseparticlesof one variantin close proximity to a groupof anothervariantas being potentialsites for the initiationof voids that growintocavities.The joiningof thesecavitiesthroughinternalneckingcan lead to a fracture surface having a dimpled topography. In the present work, it has been shown that the strength properties of an aged Ti–15°/0 Cr alloy can be improved significantly by reheating to 450”C in
orderto revertthe omegaphase.The revertedalloyexhibiteda 0.2°/0yieldstress of 166,000psi comparedto a fracture stress of 156,000psi when the omega phasewaspresentas a result of agingat 350”C. The propertychangessuggest that an optimumyieldstrengthand ductilityshouldbe achievedby controlling the sizeof the omegaparticlesandby varyingthe timeof the reversiontreatment. Althoughthe hexagonalstructureof the omegaphaseis changedto bcc during reversion,a longtimeat the reversiontemperaturewouldbe necessaryto eliminate the transformationstrainsassociatedwiththe remnantfluctuationsin solute composition.These fluctuationsand their associatedstrain fields can act as effectivebarriersto dislocationmotionandprovidea meansof nucleatingcavities in the solute-richbeta matrix.Whilelongerreversiontimesmighteliminatethe residualstrainsin the revertedalloy,it is importantto recognizethatthe initiation of a competitiveprocesssuchas the nucleationof the alphaphasewill also influence the strength and ductility. The fact that cavities are nucleated in the reverted alloys suggests a fracture mechanism similar to that already proposed for the alloys containing the omega phase. In this case the cavities nucleate at the interface between the parent beta phase and groups of the remnant zones. Ti-lO~o Cr Alloy Figure 6 is a composite view of the fractured surface of a Ti–lO~o Cr alloy heat treated to form the diffuse omega in a beta matrix. The appearance of the fractured surface varies considerably from one region to another across the width of the specimen. The central portion of the fractured surface is characterized by a network of large dimples while regions to the right are relatively flat and regions to the left have a peaked and angular topography. A closer look at the flat areas at high magnification reveals a network of extremely fine dimples as shown in Fig. 11. There is no sharp demarcation between the regions of large dimples and those containing the fine dimples. The regions to the left are peaked and angular but exhibit surface markings of the kind shown in Figs. 8a and 8b. In accounting for this fracture appearance, it is necessary to consider the stress distribution in the necked region of the material, the orientation of the grains in the necked region, and the possible influence of the strain-induced martesnite. The stress distribution in the necked portion of a cylindrical tensile specimen is triaxial, whereas the necked region of a flat tensile specimen has a predominantly
V. Chandrasekaran, R. Taggavt, and D. H. Polonis
FIG. 11. Network of extremelyfine dimples seen in the relativelyflat areasat the righthand side of Fig. 6. Magx 7400. biaxial stress condition at the center of the specimen, giving rise to an octahedral shear-stress component which varies along the width of the specimen and has a maximum at the center of the width in the necked region . The cavities or voids form in the region of maximum distortion energy and the propagation of the central crack occurs by failure in this region where there is a heavy cavity density and a maximum value of the axial tensile stress . The network of large dimples occurs in the center of the fractured face as shown in Fig. 6, in accordance with the behavior expected for the stress distribution in the necked region. Once a central crack has been established and spreads outward, crack propagation is controlled by the platic flow-of the metal at the advancing tip of the crack . Thus the presence of a network of extremely fine dimples as in Fig. 11 can be attributed to a change in the state of stress and also to the interruption of continuous crack propagation by grain boundaries lying in the path of an advancing crack. The fracture surface markings in Figs. 8a and 8fssuggest that plastic deformation still accompanies the process of separation in the final stages of fracture. It is difficult to characterize this mode of separation by the conventional termin-
Fracture Modes in a Binary TitaniumAlloy
ology used in the field of fracture studies. Crussard et al.  have found that “shear rupture surfaces often show, besides regions covered with elongated dimples, rather flat area, more or less extended.” The flat areas have been referred to by Crussard  as “decohesion along glide planes” or “ductile cleavage” and have been attributed to “fracture that occurs in slip planes that have been weakened by deformation.” Flat areas of the kind just described, further accentuated by step-like markings, have been observed in the Ti–lOO~ Cr alloy; this can be attributed more logically to separation along the matrix-martensite interfaces which constitute planes that are susceptible to crack propagation. The fracture mode in the Ti–lO~O Cr alloy appears to be controlled by the presence of the strain-induced martensite together with the change in the biaxial state of the internal stress as a crack nucleates in the center of the specimen and propagates outward. Conclusions (1) The Ti-15~o Cr alloy containing the aged omega phase failed with a tensile elongation of only 2°/0but the fractured surface showed very fine dimples which are attributed to the capability of the beta matrix to accommodate limited plastic flow prior to fracture. (2) In contrast, the Ti-lO~o Cr alloy containing difiuse omega, exhibited 13~o tensile elongation but the fractured surface exhibited ductile dimples together with ductile cleavage facets. The strain induced martensite observed in the Ti– IOYOCr alloy is believed to control the mode of fracture in the final stages of separation by promoting failure along the matrix–martensite interfaces. (3) Reversion of the omega phase in the Ti-15~0 Cr alloy improved both the yield strength and the ductility. The corresponding fracture surface exhibited ductile dimples. The reversion treatment is a promising means of achieving high yield strength with reasonable ductility. (4) The nucleation and growth of cavities are considered to be responsible for the dimpled structures observed on the fracture face. The manner in which the cavities are nucleated in the beta phase is consistent with the dislocation crack nucleation model put forth by Cottrell. The authorswish to thankMr. R. R. BoyeYfor helpful comments.The Ti-15$Yo Cr alloy used in this study was provided by Mr. H. Kato of the United States Bureau of Mines, Albany, Oregon. This researchwas spomoredby AEC’ Contract AT(45-1)-2225-T13 Report No. RLO-2225-T13-9. References 1. Yu. A. Bagaryatskiy,G. I. Nosova, and T. V. Tagunova, Soviet Phvs. Dokl. 3, 1014 (1959).
V. Chandrasekaran,R. Taggart, and D. H. Polonis
2. Y. F. Bychov, Irsshener,Fig. Zbur. Akad. Nauk. Belorus SSSR 3, 95 (1960). 3. Yu. A. Bagaryatskiy,and G. I. Nosova, Phys. Metals and Metallography 13, No. 3, 92 (1962). 4. M. K. McQuillan, Met. Rev. 8, 41 (1963). 5. M. J. Blackburn, and J. C. Williams, Trans. AIME 242, 2461 (1968). 6. Yu. A. Bagaryatskiy,G. I. Nosova, and T. V. Tagunova, Doklady Akad. SSSR 105, 1225 (1955). 7. J. M. Silcock, M. H. Davies, and H. K. Hardy, The Mechanismsof Phase Tratssfo~jormations in Solids, Inst. of Metals, London, p. 93. (1956). 8. T. S. Luhman, R. Taggart, and D. H. Polonis, .%ripta Met. 5, 81 (1971). 9. J. C. Williams, R. R. Boyer, and M. J. Blackburn, ASTM-STP 453, 215 (1969). 10. J. A. Feeney, and M. J. Blackburn, Met. Trans. 1, 3309 (1970). 11. M. K. Koul, and J. F. Breedis, Met. Tram. 1, 1451 (1970). 12. A. M. Stroh, Advats. Phys. 6, 418 (1957). 13. G. Hari Narayanan,Ph.D. Thesis, Div. Met. Eng., University of Washington (1971). 14. A. H. Cottrell, Fracture, Wiley, New York (1959), p. 20. 15. J. C. Williams, B. S. Hickman, and H. L. Marcus, Met. Tram. 2, 1913 (1971). 16. S. L. Narasimhan, R. Taggart and D. H. Polonis, unpublished research. 17. J. Aronofsky, ~. Appl. Mech., 75 (March 1951). 18. H. C. Rogers, Tram. AIME 218,498 (1960). 19. J. D. Meakin, and N. J. Petch, Fracture of Solids, AIME Conference (1962), p. 393. 20. C. Crussard, J. Plateau, R. Tamhankar, G. Henry, and D. Lajennesse, Fracture, Wiley, New York (1959), p. 524.
.4ccepted October 12, 1971