Reverse transformations in CuAlNiMnTi alloy at elevated temperatures

Reverse transformations in CuAlNiMnTi alloy at elevated temperatures

Pergamon REVERSE [email protected])00233-2 Acta mater. Vol. 44, No. 3, pp. 1189-l 199, 1996 Elsevier ScienceLtd Copyright 0 1996Acta MetallurgicaInc. Prin...

978KB Sizes 2 Downloads 34 Views

Pergamon

REVERSE

[email protected])00233-2

Acta mater. Vol. 44, No. 3, pp. 1189-l 199, 1996 Elsevier ScienceLtd Copyright 0 1996Acta MetallurgicaInc. Printed in Great Britain. All rights reserved 1359-6454/96 $15.00+ 0.00

TRANSFORMATIONS IN CuAlNiMnTi AT ELEVATED TEMPERATURES

ALLOY

Z. G. WEI’?, H. Y. PENG’, D. Z. YANG’, C. Y. CHUNG2 and J. K. L. LA12 ‘Department of Materials Engineering, Dalian University of Technology, Dalian 116024, P.R. China and ZDepartment of Physics and Materials Science, City University of Hong Kong, Hong Kong (Received 2 January 1995; in revised form 31 March 1995)

Abstract-The reverse transformation sequences in an as-quenched Cu-11.88Al-5.06Ni-1.65Mn-0.96Ti (wt%) shape memory alloy during heating or isochronical annealing up to 650°C have been studied by means of differential scanning calorimetry, high temperature X-ray diffraction as well as transmission electron microscopy and optical observations. The results show that at least seven stages of structural evolution, depending upon heating rate, can be distinguished: the reverse martensitic transformation and the martensite stabilization; atomic reordering in parent phase; two phase separation in DO, and the reversion; atomic disordering in matrix; bainitic transformation; DO, = B2 transition; and the precipitation of equilibrium phases. The different dependence of the evolution stages on heating rate is analyzed and

INTRODUCTION In comparison with CuZnAl alloys, CuAlNi shape memory alloys have a better thermal stability and higher operating temperature. However, the practical applications of CuAlNi alloys are highly restricted because of their poor workability and susceptibility to brittle intergranular cracks. Apart from the high elastic anisotropy, the precipitation of ~~(c+,Al~) phase arising from hypereutectoid alloy composition and the grain coarsening when the alloys are solutiontreated at high temperatures in the j? phase region are the main causes of the impaired ductility and workability of the alloys. As a potential commercial Cu-based shape memory alloy for higher temperature performance, a Cu-12AlLSNi-2Mn-1Ti alloy, also named CANTiM, has been developed by Sugimoto et al. [l]. In the new alloy, the Al content is reduced to 12% into a hypo-eutectoid composition range to prevent precipitation; the adding of Mn and the alteration of Ni content can change the electron:atom ratio to extend the single /l phase region and modify the martensitic transformation temperatures, and the adding of Ti can refine the grains by forming X-phase particles ((Cu, Ni),TiAl) which can hinder grain growth during annealing [l]. Recent investigations have shown that the ductility and other mechanical properties of the alloys can be significantly improved by adding these alloying elements [l-6]. It is well known that the Cu-based shape memory alloys are susceptible to postquench aging effects. The tTo whom all correspondence

should be addressed.

aging effects and the transformation sequence in ternary CuAlNi alloys at elevated temperatures have been extensively investigated [7-l 11. Recently, the aging effects in the pentatomic alloys including CuAlNiMnTi, CuAlNiMnB and CuAlNiTiB alloys have attracted some attention [5,6, 12-151, however, both the results and the interpretation of the aging effects remain controversial. In order to overcome the difficulties resulting from the undesired aging effects and to further improve the performance of the alloys, it is necessary to clarify the microstructural changes which occur in the alloys during aging. The changes in transformation temperatures and the microstructural evolutions in CuAlNiMnTi alloy during isothermal aging processes at various intermediate temperatures have been characterized by our recent preliminary work [ 161.In the present work, the reverse transformation sequence in the as-quenched alloy during nonisothermal annealing at temperatures up to 650°C was investigated systematically by differential scanning calorimetry (DSC), hightemperature X-ray diffraction, transmission electron microscopy (TEM) and optical observations. EXPERIMENTAL The chemical composition of the alloy used for this work was Cu-1 1.88Al-5.06Ni-1.65Mn-0.96Ti (wt%). The alloy was prepared by induction melting the commercial pure metals in a graphite crucible and casting into a cylinder 30 mm in diameter. The ingot was hot-rolled to 2 mm thick strips and all the samples used were cut from the strips. The samples were homogenized at 950°C for 30min and then water quenched.

1189

WE1 et al.:

REVERSE TRANSFORMATIONS

I

I

I

I

I

100

200

300

400

500

Temperature

(“C)

Fig. 1. DSC curves showing the transitions that occurred during heating and the dependence of the peak temperature on heating rate. (a) SO”C/min; (b) 40”C/min; (c) 30”C/min; (d) 2OYJmin; (e) lO”C/min.

The DSC measurements were carried out using a Perkin-Elmer DSC-7 at heating rates of lo-50°C per min. The samples of 3 mm in diameter and 0.3 mm in thickness were heated up to 700°C under a helium atmosphere. The in situ X-ray diffraction measurements were performed using a Rigaku D/Max-rA diffractometer with copper radiation. The X-ray profiles were measured at various temperatures when heating the samples at the rates of 20 and SO”C/min. To reduce the influence of aging during profile measurements, the counter scanning was limited to the ranges of interest. The scanning rate was 0.02” 261s. Microstructural observations were performed by optical and transmission electron microscopy and the latter was carried out using a Philips CM-12 electron microscope operated at 120 kV. Table Heating (“C) Peak Peak Peak Peak Peak

1 2 3 4 5

1. Peak temperatures

RESULTS AND DISCUSSION

Figure 1 shows the heating DSC curves of the as-quenched alloy at the heating rate of 10,20, 30,40 and SO”C/min, respectively. From the curves at least five transitions, i.e. four endothermic peaks and one exothermic peak, can be distinguished, as designated in the figure. The peak temperatures for the curves are listed in Table 1. Obviously, the peak positions have different dependence on heating rate. The endothermic peak 2 and peak 3, and the exothermic peak 4 shift to higher temperatures with increasing heating rate, indicating that the processes corresponding to the peaks are typically diffusioncontrolled transitions. The endothermic peak 1 which is the reverse martensitic transformation, shifts to

in the DSC curves under different

heating

rates

rate IOWmin

2O”Cimin

30”Cimin

172.2 328.6 354.8 492.0

156.8 252.3 344.5 382.2 487.2

154.9 369.2 397.5 483.0

4OWmin 157.5 268.9 380.9 408.1 472.4

SO”C/min 161.5 275.6 405.4 417.2 460.8

WE1 et al.:

REVERSE TRANSFORMATIONS

1191

Fig. 2. B2 and DO9 type antiphase domains observed using centred dark field imaging. (a) B2 APDs obtained using the superlattice spot 020 of 18R martensite; (b) DO, APDs obtained using the superlattice spot 019 of 18R martensite.

lower temperature at first and then shifts slightly upwards with increasing heating rate. The unusual dependence on heating rate can be ascribed to the well-known martensite stabilization phenomenon found at a lower heating rate which means longer aging duration in the martensite state. As we will show later, the CnAlNi alloys with Mn and Ti

additions are more susceptible to martensite stabilization than ternary CuAlNi alloys. The endothermic peak 5 shifts to a lower temperature with increasing heating rate, suggesting that it’s not a normal diffusional transition process. Accordingly, among the five peaks in Fig. 1, peaks 1 and 5 are thermodynamically controlled transitions whereas peaks 2-4 are

Fig. 3. Optical micrographs showing the typical microstructures in the as-received alloy quenched from various temperatures. (a) Martensite in as-quenched state; (b) martensite obtained after heating to 300°C; (c) bainite formed at 4OOC; (d) precipitates of equilibrium phases at 550°C. The white particles are X-phase.

1192

WEI et al.: REVERSE TRANSFORMATIONS

kinetically dominated processes. As well known, the martensitic transformations have a diffusionless character so it is very easy to understand that the forward and the reverse transformations are thermodynamically controlled processes. In the Cu-based alloys, other transitions which may be of thermodynamic character are order/disorder transitions. Recently, it has been established that on quenching, as observed in CuZnAl alloys, both CuAlNi and CuAlMn system alloys also undergo the nearest neighbor B2 ordering before ordering into a next nearest neighbor DO, or L2, ordering structure [17-191. Both the B2 and DO, type antiphase domains (APDs) were observed by TEM, using centred dark field imaging, as shown in Fig. 2. It is well known that in the Cu-based alloys, the A2 = B2 transition is of second order and hence thermodynamically controlled. Recently, neutron diffraction measurements have revealed that the B2 = DO,(L2,) transition also possesses some features of second order transition [20,21], though it is more kinetically dominated. Hence, the peak 5 in Fig. 1 can be attributed to either B2 = A2 disordering transition or to DO3= B2 disordering transition, remaining to be further clarified by in situ X-ray diffraction measurements. Generally, the formation of martensite, bainite, atomic ordering, precipitation of equilibrium phases are exothermic processes whereas the reverse martensitic transformation, dissolution of bainite and atomic disordering are endothermic processes. Consequently, it’s easy to determine the peak 4 to be due to the formation of bainite. Figure 3 shows the optical microstructures of the as-received alloy when heated to different temperatures and then water quenched. The typical morphology of bainite plates is evident after the alloy is heated to 400°C. TEM observation also confirmed the presence of the bainite plates with a 9R structure. By means of EDS analysis, we also found that the bainite plates are poor in solute content while the retained matrix is enriched with solute content. No structural changes prior to the formation of bainite can be detected by TEM and optical observations. Accordingly, the endothermic peaks 2 and 3 must be related to the changes in the state of atomic order in the long range ordered parent phase. To further evidence the structural evolutions, high temperature X-ray diffraction measurements were carried out at the heating rate of 20 and SO”C/min, respectively. Figure 4 shows the in situ X-ray diffraction measurement results by using a sample at the heating rate of ZO”C/min. From the profiles we can find that on the slower heating the M18R martensite is highly stable and some of the martensite can be retained to as high as 450-500°C. However, two factors must be considered when interpreting the results: first, the X-ray diffraction measurements process is in fact an isochronical annealing process because of the duration at a given temperature to measure the profile, rather than the nonisothermal

continuous heating as performed in the DSC measurements; second, since the bainite formed in the Cu-based alloys has the 9R or 18R structure [l 1, 16,22,23], which has similar or even identical X-ray diffraction profile to that of the stabilized martensite, it is unlikely, in the case of slow heating, to distinguish bainite from stabilized martensite by

L)

24

I

I

I

I

I

I

,

,

,

25

26

27

28

39

30

31

32

33

Two-theta Fig. 4(a). Caption opposite.

WE1 et al.:

1193

REVERSE TRANSFORMATIONS

(b)

650°C

250°C

160°C

RT I 38

I

I

I

I

I

I

39

40

41

42

43

44

I

I

I

I

I

I

45

46

47

48

49

50

51

Two-theta Figure 4(b). Fig. 4(a, b). The XRD profiles in the ranges of interest for the as-quenched sample at various temperatures under a heating rate of ZO”C/min, using one sample.

1194

WE1 et al.: REVERSE TRANSFORMATIONS

X-ray diffraction. However, bainite can be identified from the X-ray profiles in the case of rapid heating. Figure 5 shows the diffraction profiles measured at the heating rate of SO”C/min. Here, Fig. 5(a) was obtained by in situ measurements using one sample in order to compare the changes in the intensity of the superlattice peaks, whereas, Fig. 5(b) was obtained by using several samples which were heated at the rate of SO”C/min to different temperatures, respectively, to measure the diffraction profiles at the temperatures. Comparing Fig. 5 with Fig. 4, it is evident that at a higher heating rate the stabilization of martensite can be lessened and most of the martensite will transform into parent phase above 250°C. When heated to 4OO”C,the diffraction profile of 9R bainite appeared. Similar profiles can also be obtained in the prolonged isothermal duration at the temperatures between 250-4OO”C, if only the samples were rapidly heated to the temperature to make sure no martensite is retained. The intensity of the superlattice peaks 111 and 019 for the stabilized martensite show insignificant changes during the slow heating from room temperature to 250°C as shown in Fig. 4(a). Within the experimental error, at least, no increase in the intensity of the superlattice peaks with increasing temperature can be found before obvious parent phase can be detected. Contrarily, in the case of rapid heating, the intensities of the peaks show a significant increase, as shown in Fig. 5(a). Calculation of the degree of DO, long range order and analysis of the variation of the order parameter with temperature seem unreasonable because the intensity of the peaks may be contributed from stabilized martensite and parent phase or even bainite whose relative fractions are changing on heating. It seems that the parent phase has a higher superlattice peak intensity than the corresponding martensite, and the higher the fraction of parent phase, the higher the superlattice peaks. Several samples were heated to temperatures between 250-400°C at the rate of SO”C/min and then rapidly cooled to room temperature to measure and compare the intensity of the peaks for the martensite before and after the treatment, it was found that the intensities of the superlattice peaks increase with increasing temperature. The foregoing facts strongly indicate that in the parent phase state, DO, atomic reordering, i.e. the increase in the degree of long range order, takes place on heating or annealing. However, once the alloy is quenched into martensite state, even though the next nearest neighbor DO3 ordering is incomplete, the normal atomic reordering can not proceed as in the parent phase state, instead, atomic disordering may take place on subsequent aging. This result evidences that it is the atomic disordering rather than DO, atomic reordering that should be responsible for the martensite stabilization, just as we recently discussed [20]. The amount of splitting of some pairs of X-ray diffraction peaks such as I22 and 202, 1210 and 2010

for 18R martensite or 111 and 201, 115 and 205 for 9R martensite (or bainite) can reflect the extent of the distortion from the ideal martensite structure and the monoclinic distortion is a function of type and degree of atomic ordering [24-261. The splitting parameter rp = sin* &-sin* 8, or the difference in d-spacing (Ad) between pairs of planes are usually used to characterize the change in the monoclinic distortion during martensite aging and it is established that the splitting parameters increase with increasing degree of B2 order or with decreasing degree of D03(L2,) order

(4

x

.Z ii

2

35O’C

25ooc

2OOT

RT

1 24

I

25

!

!

26

21

I 28

29

Two-theta Fig. 5(a). Caption opposite.

I

I

30

31

WEI et

al.:

[24,27]. According to Fig. 4(b), the variation of the splitting parameter rp vs temperature is illustrated in Fig. 6. On the whole, the splitting parameters decrease with increasing temperature, just as demonstrated in the stabilized martensite in CuZnAl alloys [28]. The increase in the splitting above 400°C may be due to the B2 ordering and the formation of bainite. Similar results were obtained for the as-received

3s

w

40

41

1195

REVERSE TRANSFORMATIONS

42

43

martensite when rapidly heated to intermediate temperatures and then rapidly cooled to room temperature. Both atomic disordering and DO3 atomic reordering can result in the decrease in the splitting parameter. However, in the martensite stabilization process during slower heating as shown in Fig. 4, it is atomic disordering rather than DO3 reordering that causes the decrease in the splitting. On the contrary,

44

45

46

41

48

49

Two-theta Figure 5(b). Fig. 5(a, b). The XRD profiles in the range of interest for the as-quenched samples at various temperatures under a heating rate of SO”C/min. (a) Obtained using one sample; (b) obtained using six samples. See text for detail.

1196

WE1 et al.:

REVERSE TRANSFORMATIONS

2

0

100

300

200

Temperature

400

500

(“C)

Fig. 6. The changes in the splitting parameter as a function of temperature for the stabilized martensite in as-quenched sample.

in the case of fast heating and cooling, the decrease in the splitting is, evidently, the result of DO3reordering. TEM observation also confirmed the occurrence of DO, reordering: the average size of DO, and B2 order domains increase with increasing temperature or with increasing duration at the temperatures between 200-4OO”C, indicating an increase in the degree of long range order [l 1,20,29]. Figure 7 shows the ordered domains appeared in the martensite plates after rapid heating to 300°C and then cooling. Such domains can not be found in the as-quenched martensite. In addition, it should be pointed out that at intermediate temperatures, a more stable Cu,MnAl type L2, ordered structure may form as a result of

spinodal decomposition from the DO3into two separated Cu,Al-rich and Cu,MnAl ((Cu, Ni)2TiAl)-rich phases at the temperatures below 300°C according to the phase diagram [19,30]. Accordingly, the sluggish endothermic peak 2 in Fig. 1 may be attributed to the reaction for forming the DO, + L2, single phase state from the two separated phases. The endothermic peak 3 may arise from the short range atomic disordering in the DO, and L21 ordered phase, meaning the introduction of wrong atom pairs or segregation of solute atoms in the long range ordered matrix [31]. However, this does not mean the occurrence of D03(L2,)-B2 reaction. The presence of wrong atom pairs or even clustering of solute atoms

Fig. 7. TEM bright field image showing the highly ordered domains within some martensite plates in the sample which was rapidly heated to 300°C and then cooled to room temperature.

__.__ WC1

*

et al.:

_-x7__“_

KCVCKW

will result in a fluctuation in the chemical composition of the matrix and hence in turn provide nucleation sites for bainite. Also, such an atomic disordering process occurs, as we have found by DSC, prior to the formation of bainite in CuZnAl alloys. This process may be one of the characteristic precursors of bainitic transformation. Figure 4(a) and Fig. 5(a) indicate that the superlattice peak 200 of B2 ordering appears at about 350°C while the 111 peak of DO3 or L2, ordering disappears around SWC, and the 200 peak does not vanish until at 650°C. This fact suggests that the critical temperature for DO, and L2, ordering of the alloy is at about 500°C. To further support this point, in situ X-ray diffraction measurements during cooling from 650°C were performed. Figure 8 shows the changes in the intensity of the 111 and 200 superlattice peaks during the cooling. It can be found that the 111 peak already appears when cooling to 400°C and its intensity gradually increases on further cooling. This result proves that the B2 = D03(L2,) ordering transition occurs between 400 and 500°C and hence peak 5 in Fig. 1 can be ascribed to DOx(L2,) = B2 disordering transition. Apart from the change in the intensity of the superlattice peaks, the X-ray diffraction measurements reveal that there is no other detectable structural changes from 400°C to room temperature. From Figs 4 and 5, it is evident that the equilibrium y2 and tl phases appear at about 500°C. The CIphase is formed by the dissolution of bainite while the yz phase with a CugAl, composition is precipitated from the retained matrix which is enriched with solute content. The y2phase dissolves at temperatures above 550°C and the fraction of u phase also decreases with increasing temperature. This is in good agreement with the phase diagram of the alloys [l 1, 121. Now, the reason that the peak 5 in Fig. 1 shifts to lower temperature with increasing heating rate can be interpreted. On slow heating, the diffusion-controlled bainitic transformation is more complete and hence the retained matrix becomes richer in solute content for the formation of the solute-depleted bainite. On the contrary, the matrix will be less enriched with the solute content in the case of rapid heating because of the incomplete bainitic transformation. It is known that in the alloys the critical temperature for DO, ordering increases with increasing content of Al solute atoms [1 1,30,32]. Therefore, the dependence of peak 5 in Fig. 1 on heating rate can be easily understood. Based on the above results and analyses, the structural evolutions during heating in the as-quenched CuAlNiMnTi alloy can be summarized schematically in Fig. 9, though some of the evolutions may be not so easily distinguished from the DSC curves as shown in Fig. 1. CONCLUSIONS The reverse transformations at elevated temperatures in an as-quenched CuAlNiMnTi shape memory

1197

, TRANSFORMATIONS

I 24

’ 25

I

I

I

I

I

I

I

26

21

28

29

30

31

32

L 3:

Two-theta Fig. 8. The XRD profiles in the ranges of interest at various temperatures during cooling for the sample which was heated up to 650°C.

alloy have been investigated. The following stages of structural evolutions can be distinguished. 1. The reverse martensitic transformation which has a strong dependence upon heating rate. The alloy is susceptible to severe martensite stabilization on slower heating or isochronical annealing. As in most Cu-based alloys, the monoclinic distortion of 18R martensite shows a decrease during the martensite stabilization process and we have evidenced that it is the atomic disordering rather than atomic reordering that take8 place during the process. 2. The occurrence of DO3 atomic reordering in the parent phase matrix. The reordering proceed8 once the martensite reverts to the parent phase state on rapid heating. Both the increases in the intensity of superlattice peaks and in the statistical size of DO, order domains support this point. In addition, on the further DO, reordering, two separated phases,

1198

WE1 et al.:

REVERSE

TRANSFORMATIONS

Heat flow endo>

I

100I

I -

ly\

‘I ‘\ ‘I ‘\ ‘I ---_ ’I ’I ’I &MC’ I 1 I /* ’ I,’ I 200 /

-

;

DO3 atomic reordering and two-phase separation

_--

--__

___-_----

--__

>-

Reverse martensitic transformation

:

I

1

I

;

I

\ I \

:\ I” \ ‘1 I ‘\ \I ‘\ I1, I

!? 2

--__

300

;; 5

Reaction for forming single DO3 + L2, phase from two separated phases

I \I/

$ F

1A

400

The bainitic transformation

cc @ . .

‘I

tic/

\

I -.

-.

,

\

1 ‘\

I Y II

I -

Atomic disordering in the long range ordered matrix

I

\

‘,,i

/ ,’

I’\

‘\

.\ ‘. -.

I%’

500

Fig. 9. Schematic

Precipitation of y2 phase

DSC illustration

/II’\ ___--I II ;v-1 II I/ / \ \ 1’ ‘I’ \ \ \ \ \ \ lt \ / \! showing

--__ ____-=

DO, = B2 disordering

transition

Precipitation of a phase through dissolution of bainite

the overall stages of the structural

Cu, Al-rich and Cu,MnAl-rich, may be formed through spinodal decomposition. 3. The reaction for forming a single DO3+ L2, phase from the two separated phases at a higher temperature. 4. The atomic disordering in the DO, (or DO, + L21) ordered matrix, meaning the introduction of wrong atom pairs or clustering of solute atoms and the decrease in the degree of long range order. This process results in a fluctuation in the chemical composition of matrix and hence provides nucleation sites for the subsequent bainitic transformation. The atomic disordering always occurs prior to the bainitic transformation and it is suggested to be one of the feature precursors of bainitic transformation. 5. The formation of bainite which is poor in solute content. The bainitic transformation is a typical diffusion controlled process. In the case of slower

-

evolutions

during heating.

heating, the bainitic transformation is more complete and the retained matrix is hence more enriched with solute content, and vice versa. 6. The DO3 (or DO, + L21) = B2 disordering transition at the temperatures between 450-5OO”C, depending on the heating rate. The critical temperature decreases with increasing heating rate. This is due to the difference in solute content in matrix which arose from the different extent of the preceding evolutions. 7. The decomposition of retained matrix, dissolution of bainite and precipitation of equilibrium yz and CIphases which are temperature dependent.

Acknowledgements-This work was supported by the National Natural Science Foundation of China. The authors wish to thank MS W. H. Zou and Dr X. L. Ma of Beijing Electron Microscopy Laboratory for their kind assistance in part of the TEM experiment.

WE1 et al.:

REVERSE TRANSFORMATIONS

REFERENCES 1. K. Sugimoto, K. Kamei and M. Nakaniwa, in Engin-

2. 3. 4. 5. 6. I. 8. 9. 10. 11. 12. 13. 14. 15. 16.

eering Aspects of Shape Memory Alloys (edited by T. W. London Duerig et al.), p. 89. Butterworth-Heineman, (1990). S. Eucken, E. Kobus and E. Hombogen, 2. Metallk. 82, 640 (1991). D. W. Roh, J. W. Kim and Y. G. Kim, Mater. Sci. Engng AL%, 17 (1991). M. A. Morris, Scripta metall. mater. 25, 2541 (1991). M. A. Morris and T. Lipe, Acta metall. mater. 42, 1583 (1994). M. A. Morris, Acta metall. mater. 40, 1573 (1992). J. Singh, H. Chen and C. M. Wayman, MetaN. Trans. 17A, 65 (1986). J. [email protected], H. Chen and C. M. Wayman, Scripta metall. 19, 231 (1985). N. Kuwano and C. M. Wayman, Metall. Trans. 15A, 621 (1984). N. F. Kennon, D. P. Dunne and L. Middleton, MetaN. Trans. 13A, 551 (1982). D. P. Dunne and N. F. Kennon, Metals Forum 4, 176 (1981). Y. Itsumi, Y. Miyamoto, T. Takashima, K. Kamei and K. Sugimoto, Mater. Sci. Forum 56-58, 469 (1990). M. A. Morris and S. Gunter, Scripta metall. mater. 26, 1663 (1992). H. Morawiec and M. Gigla, Acta metall. mater. 42,2683 (1994). E. Hornbogen and E. Kobus, Z. Metal/k. 83, 105 (1992). H. Y. Peng, Z. G. Wei and D. Z. Yang, in Shape Memory Materials ‘94 (edited by Y. Chu and H. Tu), p. 423. International Academic Publishers, Beijing (1994).

1199

17. Y. Nakata, T. Tadaki and K. Shimizu, Mater. Trans. J.I.M. 31, 652 (1990). 18. M. Ahmed, S. W. Husain, Z. Iqbal, F. H. Hashmi and A. Q. Khan, Scripta metall. 22, 803 (1988). 19. N. Nakanishi, T. Shigematsu, N. Machida, K. Ueda, H. Tanaka, T. Inaba and 0. Iwatsu, in Proc. of ZCOMAT ‘92 (edited by C. M. Wayman and J. Perkins), p. 1077 (1993). 20. D. Z. Yang and Z. G. Wei, in Shape Memory Materials ‘94 (edited by Y. Chu and H. Tu), p. 319. International Academic Publishers, Beijing (1994). 21. A. Planes, L. Manosa, E. Vives, J. Rodriguez-Carvajal, M. Morin, G. Guenin and J. L. Macqueron, J. Phys.: Condens. Matt. 4, 553 (1992). 22. Y. Hamada, M. H. Wu and C. M. Wayman, in Proc. of ICOMAT ‘92 (edited by C. M. Wayman and J. Perkins), p. 833 (1993). 23. K. Takezawa and S. Sato, Trans. J.I.M. 29,894 (1988). 24. D. W. Roh, E. S. Lee and Y. G. Kim, MetaN. Trans. 23A, 2753 (1992). 25. L. Delaey, T. Suzuki and J. Van Humbeeck, Scripta metall. 18, 899 (1984). 26. Q. Xuan, J. Bohong and T. Y. Hsu, Mater. Sci. Engng 93, 205 (1987). 27. J. Gui, Y. Cui, S. Xu, Q. Wang, Y. Ye, M. Xiang and R. Wang, J. Phys.: Condens. Matt. 6, 4601 (1994). 28. T. Suzuki, R. Kojima, Y. Fujii and A. Nagasawa. Acta metall. 37, 163 (1989). 29. J. Gui, C. Luo, H. Zhang, W. Hu and R. Wang, J. Mater. Sci. 25, 1675 (1990). 30. M. Bouchard and G. Thomas, Acta metall. 23, 1485 (1975). 31. R. Rapacioli and M. Ahlers, Acta metall. 27, 777 (1979). 32. M. Prado, M. Sade and F. Lovey, Scripta metall. mater. 28, 545 (1993).