You are on page 1of 15

Investigation on the Fracture Behavior of Shape Memory

Alloy NiTi
J.H. CHEN, W. SUN, and G.Z. WANG

This work investigates the fracture behavior of shape memory alloy NiTi (50.7 at. pct Ni) at room
temperature. Macroscopic mechanical tests, microscopic in situ observations of tensile fracture processes
by scanning electron microscopy (SEM), and detailed analyses of fracture surfaces were carried out.
The results reveal that specimens with different thicknesses show various shape memory effects and
superelasticities. The main crack with a quasi-cleavage mode that combines cleavage with ductile
tearing is initiated at the notch tip and is stress-control-propagated in line with the direction of the
maximum normal stress. The microstructure has little effect on the direction of crack propagation,
but coarser substructures show lower resistance to the crack propagation. In specimens with various
types of notches, various notch acuities present different effects on the crack initiation and propagation
and result in different fracture behaviors.

I. INTRODUCTION forged, and rolled sheets were then heated to 800 °C for 15
minutes, ice water quenched, and aged at 500 °C for 25 min-
IN addition to shape memory effect, NiTi alloy has utes (1.5- and 2.0-mm sheets) and 30 minutes (4-mm sheet).
properties such as repeatability of mechanic behaviors, wear The microstructures of the aged NiTi sheets were examined
resistance, corrosion resistance, and biocompatibility. These as B2 and shown in Figure 1(a), with their X-ray diffraction
superior properties make NiTi alloys one of the most com- (XRD) diagram shown in Figure 1(b). The sheets were uniform
mercially successful shape memory alloys used for throughout, regardless of the thickness. No texture microstruc-
nonvascular and vascular stents in biomedical applications tures were found in the specimens. However, the lamellae-like
and for the fixation of damaged bone.[1,2,3] However, as an substructures in the grains looked like martensite lamellae. The
intermetallic compound, it is rather brittle and rare accidents XRD diagram (Figure 1(b)) showed a small amount of marten-
have been reported of broken NiTi parts in the human body. site (B19), which may have been produced during specimen
Therefore, the fracture behavior is a subject to be investigated preparation. The differential scanning calorimetry (DSC)
in order to improve this alloy. In past decades, most work curves for the three sheets are shown in Figure 2. The DSC
for this alloy had been devoted to the study of the martensitic heating rate for all specimens was 10 °C/min. The measured
transformation, the shape memory effect, and the super- data of the critical temperatures of phase transformations for
elasticity of these alloys.[4–10] Many articles describe the sheets are summarized in Table II. The difference in the critical
mechanical behaviors of NiTi alloys mainly concerning temperatures of phase transformations for the specimens with
plastic deformation.[11,12] Few studies were found to be various thicknesses was considered to be caused by minor
devoted to the fracture behavior of NiTi alloy such as that variation in heat treatment. The variation of heat treatments
reported by Gall et al.[13] In that work, the mechanism of resulted in various amounts of Ti3Ni4 precipitate, which
fracture was identified as being due to (1) nucleation, growth, affected the composition of the B2 phase and then the phase
and coalescence of voids from Ti3Ni4 precipitates; and (2) transformation temperature.
cleavage fracture on {100}. Cleavage fracture was found
to act in conjunction with ductile tearing.
B. Specimens
Based on observations of in-situ tensile tests and detailed
analysis of fracture surfaces, this work aims at the fracture Figure 3 shows the dimensions of specimens for in-situ
mechanism and the factors affecting the fracture process of observation of tensile tests via scanning electron microscopy
NiTi alloy. (SEM). The detailed geometric parameters are shown in
Table III. Figure 4 shows the dimensions of specimens for
macroscopic tensile tests. Figure 5 shows the dimensions of
II. EXPERIMENTAL specimens with narrow slit notch (II type) for three-point
A. Materials bending (3 PB) tests and with single or double V notches
for four-point bending (4 PB) tests.
A commercial binary NiTi alloy with 50.7 at. pct Ni was
used, the compositions of which are shown in Table I. Speci- C. Macroscopic Tensile Tests
mens were 1.5, 2, and 4 mm in thickness. The vacuum-melted,
Tensile tests were carried out with a SHIMADZU-10TA
(Japan) universal tester at a crosshead speed of 0.1 mm/min at
room temperature. Applied loads and displacements were
J.H. CHEN and G.Z. WANG, Professors, and W. SUN, Postgraduate recorded and the charts converted to engineering stress and
Student, are with the Key State Laboratory for New Materials of Non-
Ferrous Alloys, Lanzhou University of Technology, Lanzhou, Gansu, strain curves.
730050, P.R. China. Contact e-mail: zchen@lut.cn In loading-unloading tests, specimens were loaded to the
Manuscript submitted August 11, 2004. desired applied loads, and then unloaded at the same rate

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 36A, APRIL 2005—941

the fracture facets could be related in a one-to-one fashion to the sections of surface crack. Fig.63 37.07 30. APRIL 2005 METALLURGICAL AND MATERIALS TRANSACTIONS A . Composition of NiTi Alloy (Weight Percent) Table II. by SEM at various loading steps. and Af the temperature of fin- ishing anstenitic transformation in heating.97 18. 2—DSC curves for three sheets. Rf the temperature of finishing R-phase transformation in cooling.03 12. E. The processes of crack initiation and propagation were recorded in conjunction with applied loads.05 0.63 12. D. 3—Nominal dimensions of tensile specimen for in-situ observation via SEM.47 17. The loads and displacements were recorded during both processes and the measured values were converted to stress and strain data. as the loading processes.8 0.0 21.08 balance 1.81 29. 3 PB and 4 PB Tests The 3 PB and 4 PB tests were carried out using a SHIMADZU-10TA universal tester at a crosshead speed of 0.0 28.92 38. Mf the temperature of finishing martensitic transformation in cooling.45 Rs is the temperature of starting R-phase transformation in cooling.61 21. Fracture surfaces were observed.5 26.08 0.98 2.96 43. Table I. Fracture Surface Observations Fracture surface observations were carried out along the path of the surface crack extension that was recorded in the in-situ observation.96 27. Applied loads and displacements were recorded. By measuring the distances from the notch root to the designated sections of cracks on the sur- face of an in-situ tensile specimen.29 4. and measuring the dis- tance from the notch root to designated fracture facets on the fracture surfaces of the same specimen. 942—VOLUME 36A.10 34. In-Situ Tensile Tests via SEM In-situ tensile tests were performed in a vacuum using a calibrated load-controlled stage installed in a scanning elec- tron microscope (SEM) S-520. The specimens were slowly (b) step-loaded manually and the crack patterns were recorded Fig. Ms the tem- perature of starting martensitic transformation in cooling. 1—(a) Microstructure of NiTi alloy and (b) XRD diagram. As the temperature of starting austenitic transformation in heating.3 26.85 21.1 mm/min. (a) Fig.33 23. F. Critical Temperatures of Phase Transformation (°C) Ni C O H Ti Thickness Rs Rf Ms Mf As Af 55.

and detwinning of stress-induced martensite and plas- (a) 3 PB bending and with (b) single or (c) double V notches for 4 PB. shows a higher stress plateau (around for specimens with three thicknesses.0 mm.5 mm during repeated loading and unloading processes.901 38. t thickness of specimen.98 °C (temperature of finish METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 36A.96 1. 1.98 406 II 87 2. (3) the postplateau deformation of oriented martensite (until 8 pct. EXPERIMENT RESULTS AND ANALYSES its lower Ms (starting temperature of martensite transfor- mation in cooling) and Mf (temperature of finishing marten- A. 4—Dimensions of specimen for macroscopic tensile tests.0 0. The transformation and the reorientation of stress-induced marten- site increase the straining-capability of NiTi alloy. caused by combined effects (c) due to further transformation of residual austenite. and line 3: 4 mm.5 mm.501 V 250 2.96 1.95 1.48 5. Tensile Tests site transformation in cooling).98 203 V 240 2. Fig. (2) the stress-induced Figure 7 shows a perfect superelasticity of specimen of martensitic deformation (a pseudoelastic-plateau deforma. line 1: 1. APRIL 2005—943 . Similar to those 300 MPa) for inducing martensitic transformation and has described in Reference 12. 7—Stress-strain curve for a specimen of 1.428 38.0 0.87 5. 6—Stress-strain curves of tensile tests on specimens with various thick- ness.5 mm. Due to III.95 II is straight blank notch. V v-type notch. tic deformation). reorien- Fig.97 1. (a) (b) Fig. and (4) the plastic deformation. A area of ligament of specimen. and L length of specimen. which is Figure 6 shows the stress-strain curves of tensile test results 1. Parameters of Specimens for In-Situ Observation of Tensile Tests via SEM Number Notch Type  (m) d (mm) t (mm) w (mm) A (mm2) L (mm) 401 II 87 2.94 3.906 38.48 5.  radius of notch root.0 0.5 mm in thickness. tion until around 5 pct).485 38. w width of specimen. 5—Dimensions of specimens with narrow straight notch (II type) for tation. Table III.375 5. resenting (1) the elastic deformation. a lower plastic strain at fracture. all curves have four sections rep. in which the Af of 23. the specimen. d depth of notch.0 0. line 2: 2. Fig.

10—(a) Applied load and (b) stress plotted against crack length during loading and unloading processes.0 mm during the loading and unloading processes. large.29 °C) is appreciably higher than room tempera. the recovered strain is less.austenite transformation in heating) is reached by the room load drops observed during the reloading procedure need temperature and the strain up to 1. the stress-strain curve of specimens of following events are visible. 8—Stress-strain curve for a specimen of 2. of loading and unloading operations. the applied load needed for Figure 8 shows the stress-strain curve for the three times further propagation decreases with increasing the crack length. men 401 with a narrow slit notch of 87 m in width.45 °C) is higher. showing the shape memory effect.0 mm during loading and unloading process shows a The root of the original slit notch is shown in Figure 11(a). (a) (b) Fig. terns during in-situ tensile tests. it is shown in Figure 11(b). The subsequent curve but the average stress increases although the increment is not sections follow smoothly one by one and finally form a pro. APRIL 2005 METALLURGICAL AND MATERIALS TRANSACTIONS A . similar pattern to that of the 2-mm specimens. the strain accompanying the stress-induced martensitic and stress-crack length are obtained and shown in Fig- transformation is not fully recovered after the stress is ures 10(a) and (b) for specimen 401 of Table III. stress-induced martensitic transformation is fully recovered after the stress is released. This means that the processes happened in crack patterns observed during in-situ tensile test of speci- unloading are rebuilt in the reloading operations. ment. 4. The propagated to a short distance. As shown in the figures. curves of load-crack length ture. file similar to that carried out by the single operation shown Figures 11(a) through (h) show the consequent series of in Figure 7.0 mm during repeatedly Fig.5 pct accompanying the to be investigated further. and trans-grain- the Af (30. Here. The retaining deformation will be fully recovered stress shown is the average stress of the remaining liga- in the heating process. Because A crack is initiated at the notch root. The In Figure 9. Fig. In-Situ Observations of Tensile Tests in SEM Figure 8 shows the stress-strain curve of specimens of 2. Because From the applied load data and the pictures of crack pat- the Af (27. 944—VOLUME 36A. B. in-situ observation of a tensile test of specimen 4 mm. 9—Stress-strain curve for a specimen of 4.0 mm during repeated loading and unloading processes. the released.

74N. (a) (b) (c) (d) Fig.02N. (b) 278. (d) 270. APRIL 2005—945 . METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 36A. (g) 242. 11—A series of crack patterns observed in in-situ tensile test of specimen 401 with narrow straight notch 87 m in root radius: (a) 249. (c) 297.42N.45N. (f) 247. (h) 242. (e) 260.96N.33N.74N. and (j) 158.67N.48N.82N. (i) 241.

(d) 270.67N. (e) (f) (g) (h) Fig.82N. 946—VOLUME 36A.02N. (e) 260. and ( j) 158. 11—(Continued).74N. (c) 297. (f) 247.96N.45N. (b) 278. APRIL 2005 METALLURGICAL AND MATERIALS TRANSACTIONS A . (h) 242.42N. A series of crack patterns observed in in-situ tensile test of specimen 401 with narrow straight notch 87 m in root radius: (a) 249.48N.74N. (g) 242. (i) 241.33N.

This is con- tiated in a band of cavities left by erased second-phase par. The higher age one.5 mm and remains in imen 406 is appreciably wider than that of specimen 401. Sometimes the crack propagates unstably that initiated the crack is lower than 200 MPa  1. sidered to be the reflection of Luderslike deformation ticles and finally connects with the main crack. Figure 11(i) shows a crack initiated at a junc.48N. which extends to around 2.74N. The produced pseu- on the way. 11—(Continued). (b) 278. which lower than the maximum tensile stress produced by stress make the fracture a quasi-cleavage pattern rather than a cleav- intensification at a distance ahead of the notch tip. out the remaining ligament is triggered at a crack length and grain boundaries (Figures 11(g) and (h)) show little around 3. line with the maximum normal stress. The isotropic An analysis of the fracture mechanism is as follows. Figure 11(j) shows a general view of crack tip. Figure 12 shows two short white strips developing at the The direction of lamellae seems to have little effect on the crack tip in directions of 45 deg inclining toward the direc- propagation. (i) 241. (f) 247.5 mm. the higher the tensile stress in a notched been transformed from the existing phase. The crack propagates in the martensite. Figures 11(e) and (f) show that a crack is ini. The martensite transfor- a strong stress-controlling mode of crack propagation. (g) 242.67N. tion of maximum normal stress in specimen 406. is trans-lamellar-propagated through a grain (Figure 11(d)). This process makes specimen. A series of crack patterns observed in in-situ tensile test of specimen 401 with narrow straight notch 87 m in root radius: (a) 249.74N. It is mation induced by the high stress and the strain ahead of well known that the tensile stress at the notch tip is much the crack tip produced appreciable additional strains. The behavior of the NiTi alloy is unusual for intermetallic com- main crack is initiated at the notch tip and propagates in line pounds and is considered to result from the dominant effect with the direction of the maximum normal stress showing of the martensitic transformation. The fact that main crack is initiated just at the the originally existing microstructure insignificant to crack notch root (instead of at a distance ahead of the notch tip) propagation.82N.45N. (d) 270. and (j) 158. METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 36A. Due to the high stress and heavy strain developed in through the grain boundary. martensite is induced the maximum normal stress even though some twists occur before the crack propagated further. The width of the blunted main crack tip in spec- a crack.42N. The crack is propagated in line with the maximum normal and at a low applied load means the local fracture stress stress and bypasses the white second phase (Ti3Ni4) particles of NiTi alloy is very low. (i) ( j) Fig. A fracture through. (h) 242.96N. which has the applied load. The crack remains in line with front of an appreciably blunted crack tip. the stress (Figure 11(c)). APRIL 2005—947 .33N. Figures 11(g) associated with stress-induced martensitic transformation in and (h) show that the crack does not prefer to propagate NiTi. effect on the direction of crack propagation. From Figure 10(b). (e) 260. doelastic plateaus in Figure 6 create extra strain to blunt the tion of three grains.02N. (c) 297. The crack 230 MPa  250 MPa.15  through a region with coarse microsubstructure. lamellae orientation (Figure 11(d)). The grain orientation.

The specimen is suddenly fractured after the main crack behaviors are considered to be caused by different acuities propagates a very short distance. In the narrow slit-notched specimen. at which the accumulated elastic energy is insuf- one crack is initiated and propagated close to the main crack. the values (Figure 11(c)). 948—VOLUME 36A. 12—Luderslike deformation developed in front of a crack tip: (a) 442.22N and (b) 442. The crack is initiated at lower a short extension of the crack. It implies that a crack in-situ tensile test of a specimen of 2 mm with V notch of is produced at a low applied load and propagates to a long 270 m in root radius. but in some cases.5 m. and the 4 PB test of double V-notch specimen. the higher Figure 14(a) shows a similar case for a specimen of stress at the notch root is sufficient to initiate a crack at a low 1. For the specimen ures 11(e) and (f)) with a narrow slit notch. For specimens with V-notch. notch specimen. In a narrow slit notch specimen. yet a 45 deg V-notch of 250 m in radius is too tiated and propagates a short distance at higher applied loads blunt to evaluate the fracture toughness of NiTi alloy. In V-notched specimens. in the notch root. propagates step-by-step. applied loads. the mulated energy is sufficient to cause an overall fracture after stress intensification is higher. the notch root and propagated directly from it. microcracks can be initiated at the junc. instead of many microcracks. and is controlled by It seems that a narrow slit notch of 87 m in width is stress (Figure 10). The second-phase particles (Ti3Ni4) have higher resistance Specimens are cut from 4-mm NiTi sheet. As analyzed previously. C. Due to the different to crack propagation and make detours for the main crack spans between the points of loading and reactions. the different fracture root. the specimens are considered to be caused by different acuities crack is initiated at a higher applied load at which the accu- of the notch root. An and higher average stress. the trends of load-displacement curves look like that shown many microcracks are produced in the vicinity of the notch in Figure 13(c). (a) (b) Fig. Bending Tests tion of three grains (Figure 11(i)) where the incompatibil. the 4 PB test of V- stress triaxiality. Figure 15 shows load-displacement curves for the 3 PB test ity of deformation causes higher stress concentration and of a specimen with a narrow slit notch. only applied load. The similar trend is observed in bending tests.22N. 0 deg and a root radius between 250 and 43. In a few cases. In this specimen. In a V-notch specimen with a wider root. too acute. The main crack is also initiated at distance before final fracture. However. microcracks can be of applied loads are incomparable. the trend of load-displacement curve Figure 13 shows the crack pattern produced during the looks like that shown in Figure 10(a). alloy should have a notch with a blank angle between 45 and mens. APRIL 2005 METALLURGICAL AND MATERIALS TRANSACTIONS A . but the curves show a initiated under the effect of second-phase particles (Fig. the crack is ini.5 mm. which are sufficient to trigger fracture over all speci. similar trend as in the in-situ tensile tests. ficient to cause an overall fracture and the crack is propagated The different cracking behaviors presented in the preceding step by step. as shown in Figures 14(c) and ideal specimen for evaluating the fracture toughness of NiTi (d).

the frac- Figure 16(c).to etched for observation. and (c) load. morphology are surrounded by tear ridges (white strips in By measuring the distances from the notch root. Many ture facets corresponding to designated crack sections with parts. where the surface is polished and revealed that in the region of crack length from 200. Fracture Surfaces strip ridges. APRIL 2005—949 . which are produced by the polishing tensile-tested at room temperature. Facets Due to the high Af. it is ures 16(a) and 17(b)). the stress-induced martensite cannot be of cleavage (Figures 16(b) and (c)) mixed with featureless recovered at room temperature. 13—The crack pattern of in-situ tensile test of V-notched specimen of (a) and (b) 2 mm. prevails throughout the fracture surface (Figure 16(a)). The quasi-cleavage pattern procedure done on the specimen’s surface before the test.and (d ) stress-crack length curves. especially on the left side of the fracture surface (Fig. D. (a) (b) (c) (d) Fig. show lamellar facets surrounded by 450-m crack propagation is more difficult with increments METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 36A. the same distances can be found. From Figure 10. The lamellar substructure is considered to be Figure 16 shows typical fracture surfaces of NiTi alloy the martensitic layers. which also show the brittle pattern).

compounds. 15—Loads-displacement curve for the 3 PB test of narrow straight The quasi-cleavage prevails throughout the fracture process notch specimen line 1. 14—The crack pattern of in-situ tensile test of V-notched specimen of (a) 1. It is considered to result from the extra plas- 950—VOLUME 36A. This means that loading process factors. Fig. Therefore. In general. and 4 PB and shows that the NiTi is not as brittle as TiAl intermetallic test of double V-notch specimen of 4 mm line 3. The cor- responding fracture surface shows appreciably coarser facets (Figure 18(b)). can improve the alloy’s toughness. (b) (a) (c) Fig. (a) 383. APRIL 2005 METALLURGICAL AND MATERIALS TRANSACTIONS A . Similar cases are observed in the crack length region around 800 and 2400 m. The corresponding fracture surface shows finer fracture facets (Figure 17(b)). Figure 19 shows the fracture surfaces of bending speci- mens. which cause the induced martensite layers to be finer. of stress more than 300 MPa.and (c) stress-crack length curves. 4 PB test of V-notch specimen line 2. although the microstructure has little effect on the direction of crack prop- agation.5 mm and (b) load. In the region of crack length around 1800 m. coarser substructures show lower resistance to the crack propagation. the fracture surfaces of bending specimens show similar patterns as that of in-situ tensile specimens.31N. the crack propagated unstably with an increment of 500 m at a maintained stress and caused a remarkable decrease of applied load.

(a) (b) (c) Fig. APRIL 2005—951 . METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 36A. 16—Typical fracture surfaces of NiTi alloy tensile tested at room temperature.

952—VOLUME 36A. 18—Crack pattern and corresponding fracture surface at crack length around 1800 m. APRIL 2005 METALLURGICAL AND MATERIALS TRANSACTIONS A . (a) (b) Fig. (a) (b) Fig. 17—Crack pattern and corresponding fracture surface at crack length around 300 m.

IV. martensite constituent. The fracture surface (Figures 20(b) 4. and additional extra plastic strain. conditions. At room temperature. Materials Cooperation for offering the NiTi alloy and Miss METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 36A. APRIL 2005—953 . Facets of cleavage mixed with featureless mor- Figure 20 shows a fracture surface of the smooth tensile. the crack propagates at a high rate. however. Stress (strain) induced martensitic transformations play an important role in improving the toughness of NiTi alloy. The pro. This extra ture behaviors of NiTi alloy at room temperature can be plasticity offers a wider blunted crack tip and makes the summarized. 3. This means that NiTi alloy is sensitive to stress direction of maximum normal stress. (a) (b) Fig. Finally. 6. the frac. and then the accompanying plastic 2. The fracture is initiated from one ridges. The different fracture and at inclusions or second-phase particles. A quasi-cleavage prevails in the fracture process of NiTi agated by tearing. NiTi shows either a shape mem- tic strain produced by crack blunting is low. Much finer facets resistance to the crack propagation. The plas. SUMMARY ACKNOWLEDGMENTS Based on macroscopic mechanical tests and microscopic in-situ observations of tensile fracture processes via SEM The authors express their gratitude to the Xi Mai New in connection with analyses of the fracture surfaces. alloy. featureless-morphology facets in Figure 16(c). but coarser substructures show lower those observed in notched specimens. 1. Notch acuity has an appreciable effect on the frac- side of the specimen (bottom side in Figure 20(a)) and shows ture pattern. the crack is prop. with the featureless and even dimplelike facets are 5. The crack shows the cleavage pattern. stress showing a strong stress-controlling mode of crack duced plastic strain induces more martensite transformation propagation. In a agated in line with the direction of the maximum normal fine grain region. clear Chevron strips. this still needs further investigation. The initiated patterns shown by smooth specimens and by notched and extended crack is finally connected with the main specimens are apparently caused by different stress triaxiality crack and keeps the crack propagation in line with the values. 19—Fracture surface of (a) 3 PB testing specimen and (b) 4 PB testing specimen. In a coarse grain region. Microcracks can be initiated at the junction of three grains surrounded by much denser ridges. ticity caused by induced martensite formation. The microstructure has little effect on the direction of and (c)) shows a fracture pattern appreciably tougher than crack propagation. the crack is appreciably blunted. the induced ory effect (T  Af) or superelasticity (T Af). phology produced by tearing are surrounded by tear tested specimen of 4 mm. The main crack is initiated at the notch root and is prop- strain is low.

APRIL 2005 METALLURGICAL AND MATERIALS TRANSACTIONS A . 20—Fracture surface of tensile tested specimen. 954—VOLUME 36A. (a) (b) (c) Fig.

225-37. METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 36A. Y. Zhu: Mater. Miyazaki and K. 1988. T. V. vol. Shimizu: Scripta Metall. 50. pp. Stachowiak and P. pp. J.R. pp. Sittner. United Kingdom.C. Butler. M. 109 pp.M. 2000.G. Hanlon: Metall. pp.. 5. on Shape Memory 10. and F. pp.. pp. Chumlyakov: Int. Saunders: Scripta Mater. Phys. Saburi: Shape Memory Materials. 6.M. S. 115-22.. and K. vol. and J. Mccormick: Acta Metall.. 137-47. 229-38. Tan. and K. Mag. 1994. Fract. Otsuka: J. G. pp. Wayman. K.B. Chu. S.. Cambridge University Press. 287-93. Salvitori: JVIR. vol. 1. Ohmi. and Superelastic Technology. Waitz. Conf. K. Trans. 435-40. Hess: Proc. vol. Tan and W. Gall. 1972. 50. Crone: Scripta Mater. W. Orsi. J. Dai. REFERENCES Cambridge. Special 6. Wasilewski. 327-28 and 11. Sci. 193-98. p. Bezzi. 13. F. 7. 9. Bramfitt and R.I. 393- 55-62. 43 (C4). H. vol. Liu. and M.. 2. 2004. Yang. G. 291-96. Miyazaki. 12. S. 1982. pp.Cao Rui for her help in manuscript preparation. and Y. and M.P. Sehitoglu. vol. 189-207. Forum. N. vol. Otsuka: Phil. 2001. thanks to Miss Martha Ello for her help in literal revision. L. 1971. 1994. P. 50. 2004. 36. A. vol. 77. Karnthaler: Acta Mater. APRIL 2005—955 . T. 1st Int. 1998. 4.J. pp. pp. vol.L. C.E. J. Y. 819-23. 1984. 408. 5. Y. pp. 8.. and H. 2004. 3. Kazykhanov.E. Cornelis. 2.. CA.