Effect of mechanical cycling on the pseudoelasticity characteristics of TiNi and TiNiCu alloys

MATERIALS SCIEINE & ENGINEERING ELSEVIER

Materials Science and Engineering A203 (1995) 187 196

A

Effect of mechanical cycling on the pseudoelasticity characteristics of T i - N i and T i - N i - C u alloys B. S t r n a d e P ,

S. O h a s h i b, H . O h t s u k a b, S. M i y a z a k i c, T. I s h i h a r a b

aDepartment of Materials Engineering, Technical University of Ostrava, 70833 Ostrava, Czeeh Republic bNational Research Institute Jot Metals, 1-2-1, Sengen, Tsukuba-Shi, lbaraki 305, Japan ~Institute oJ Materials Science, University of Tsukuba, Tsukuba, Ibaraki 305, Japan

Received 20 March 1995; in revised form 22 May 1995

Abstract

This paper presents the findings of an experimental study of how mechanical cycling of Ti-Ni and Ti-Ni Cu shape memory alloys in the pseudoelastic (PE) state affects their residual elongation after unloading, their critical stress for martensite formation and their hysteresis or amount of energy dissipated during one cycle. Specimens were cycled in two basic modes: hard loading cycles at a constant emax and soft ones at a constant ares. In the hard cycling the authors further investigated how the PE characteristics respond to various strain rates and how the strain rate changes. Each of the examined alloys was cycled in the PE deformation mode at a temperature where each specimen can be deformed at the same constant critical stress for martensite formation in the first cycle of the test. As the number of cycles increases, the residual strain eo grows, while both the stress a .... for martensite transformation and the hysteresis W decrease. The rate at which eo grows depends on o-s, a,ls during cycling and the type of cycling mode. By considering the two factors a s and a .... the rather complicated effect of cyclic deformation on the PE characteristics was explained. Cycling at higher strain rates has been found to increase the residual elongation left after the specimen is unloaded and to cause a more raped decline of the critical stress for martensite formation as cycling continues. After changes in the elongation rate the stability of the cyclic stress-elongation diagram depends on the amount of residual elongation present and on the stability of that diagram during the first cycling at the original elongation rate. Keywords: Mechanical cycling; Pseudoelasticity; Ti-Ni alloys; T i - N i - C u alloys

I. Introduction

The potential of applications utilizing T i - N i and T i - N i Cu shape m e m o r y alloys depends largely on the stability of their basic pseudoelasticity (PE) characteristics during mechanical cycling. The stability of PE can be characterized by the residual elongation eo remaining after the load on the specimen is relieved, the critical stress O'msneeded for martensite to start forming and the hysteresis W, a measure of the energy dissipated in one cycle. The stability of these characteristics is affected by the test temperature, the nickel content and the previous heat treatment of the alloy [1-5], but also, as has recently been discovered [6], largely by the mode in which the specimens are stressed. As for the test temperatures, the stability of cyclic PE 0921-5093/95/$09.50 © 1995 ElsevierScience S.A. All rights reserved SSD1 0921-5093(95)09881 -X

stress-strain diagrams is most easily secured at a temperature that is close to the austenite finish point Af, where the residual deformation after unloading is lowest. Also, if the critical stress for slip is high enough and if the alloy readily undergoes cyclic strain hardening, at this temperature the residual elongation after unloading diminishes further as the number of cycles increases, so that the cyclic stress-strain diagram is more rapidly stabilized. Higher nickel contents and suitable heat treatment induce dislocation and precipitation hardening, increase the critical stress for slip and reduce the residual deformation as cycling progresses [4,5]. How the stability of the cyclic stress-strain diagram responds to the mode of cycling depends on the nickel content and the initial magnitude of the critical stress for slip, but also on the cyclic hardening capacity of the alloy [7].

B. Strnadel et al. / Materials Science and Engineering A203 (1995) 187 196

188

Table 1 Compositions, transformation points and test temperatures of examined alloys Alloy

A B C D E F

Test temperature (°C)

Transformation points (°C)

Composition (at.%) Ti

Ni

Cu

49.1 49.5 50.0 49.0 50.0 48.5

50.9 50.5 50.0 41.0 40.0 41.5

--10.0 10.0 10.0

Mr 115.8 - 77.8 -28.0 7.6 20.9 14.4

So far the effect of cyclic loading on the PE characteristics has been studied either on one alloy at various temperatures [3] or on a set of alloys at a constant temperature [7]. When we want to study the scope of application of shape memory alloys, however, we must often assess their response to cyclic loading at an equal stress or strain condition and also how under these conditions their response to cyclic loading varies with the strain rate. The purpose of the present paper is to clarify these questions.

2. Experimental details The experimental work was performed on three T i Ni alloys with slightly differing but roughly equiatomic compositions and three T i - N i Cu alloys which contained 10 at.% Cu at the expense of their Ni content (see Table 1). The alloys were produced in an induction furnace with an argon atmosphere. Plate tensile specimens [7] were vacuum annealed at 400 °C for 3.6 ks followed by water quenching so as to ensure maximum dislocation hardening. All six alloys are in a single B2 phase according to the equilibrium phase diagrams. However, the Ti-51.0at.%Ni alloy is an exception because it contains metastable Ti3Ni 4 precipitates in the matrix B2 phase if it is heat treated at an intermediate temperature around 400 °C [8]. Such fine Ti3Ni 4 precipitates are effective in raising the critical stress for slip deformation, in addition to which a high density of dislocations is effective [2,4,5]. The martensite finish and start points Mr and Ms and the austenite start and finish points As and At. were in all cases ascertained by differential scanning calorimetry and are listed along with the chemical compositions in Table 1. The mechanical cycling was conducted on a Shimadzu computercontrolled precise universal tester, an Autograph AG-D with a heating chamber. An extensometer was attached directly to the specimen grips so as to measure only the true strain and exclude any non-substantial strains. The elongation rates were 0.01 and 0.1 mm s I, which with an initial distance of 25 mm between the grips represent strain rates of 4 × 10 4 and 4 x 10 3 s - l respectively.

Ms

A~

Ar

- 30.7 - 18.5 37.5 29.8 41.4 37.5

1.9 9.0 48.2 34.5 52.7 42.6

44.6 53.0 77.8 50.0 66.6 60.6

46 61 93 77 96 78

3. Results and discussion The same set of Ti Ni and T i - N i Cu shape memory alloys was investigated in order to clarify their responses to hard and soft loading cycles in an earlier study [7]. To determine how in these alloys the residual elongation, the critical stress ams for inducing martensite transformation and the hysteresis W change with the number of cycles, the cycling tests were done at a constant temperature of 80 °C. However, higher nickel contents reduce the transformation temperatures, increase the critical stress for slip and decrease the transformation-induced elongation, all of which markedly alter the temperature dependence of the relative residual elongation eo/ema, as seen in Fig. 1, where eo is the residual and ema× the maximum elongation and eo generally consists of residual elongation due to non-recoverable slip deformation and recoverable transformation strain. As an example among these six alloys, the reverse transformation start temperature A s and finish temperature Af of alloy A are 1.9 °C and 44.6 °C respectively. Below As, eo/emax is about 0.8 or more, since the transformation strain remains owing to favourably oriented martensite variants. However, eo/emaX decreases with increasing temperature above A s depending on the volume fraction of reverse transformed region and 1 ~.

0,

. "'~

0.6

20

oA

oB

o

II E

"

.... 0.2

\ i

0 -6o

"

i o,..-o i ..e "" .....A

I

I

I

l

I

I

I

l

l

I

-~o

-2o

o

20

~o

6o

so

loo

~2o

1~o

Temperafure [°El

16o

Fig. 1. Temperature dependence of relative residual elongation of Ti Ni and T i - N i Cu alloys.

B. Strnadel et al, / Materials Science and Engineering A203 (1995) 187 196 1200 - 1000

OA oB

eD •E

these alloys, O'ms cannot be constant; it increases with decreasing Ms. The stress strain curve and its cyclic behaviour for shape memory alloys are strongly affected by their critical stress ares. Therefore it is not possible to clarify the effect of alloy composition of the cyclic behaviour by testing these alloys with different M, at the same temperature. In this experiment it was planned to cycle these materials pseudoelastically at such a temperature that the first-cycle ares values would be identical in all the alloys. This equal initial O'ms level could objectively reveal how the stress-strain characteristics varied with the number of loading cycles. All the examined alloys were subjected to 50 cycles in two types of test mode: a hard cycling mode with a constant maximum elongation em~ of l mm and a soft cycling mode with a constant maximum stress am,x of 1.1 a,,ls. The first-cycle ares value was selected as 500 MPa, chiefly to ensure that all the alloys would be tested at a temperature where the PE effect becomes apparent. The maximum stress for the soft cycling mode was chosen as 550 MPa for all the alloys. The test temperatures at which ams equals 500 MPa are listed in Table 1. Fig. 3(a) shows the cyclic stress elongation diagrams for the first and last cycles in both cycling modes, at an elongation rate of 0.01 m m s 1, for the T i - N i alloys. All these diagrams show typical PE behaviour. By comparing the first-cycle diagrams of alloys A (Ti-

,-,.~ "~'~'~V °

o_

~: 8oo

600

...:I:"'" .'" ,.. ..

.<

........= ; ; ~ . ~ : . . , 0

i

-60

-40

I

-20

I

0

i

20

i

i

i

40

60

80

100

I

I

120

140

160

Temperature [ "C] Fig. 2. Temperature dependence o f critical stress for inducing martensite formation in Ti Ni and Ti N i - C u alloys.

reaches its minimum value at Af. Above At, eo/emax mainly consists of residual elongation due to slip deformation; eo/e .... increases with increasing temperature, because the applied stress for martensite transformation increases linearly with temperature following the Clausius-Clapeyron relationship. The critical stress ares for martensite transformation also depends strongly on the Ni content as shown in Fig. 2. This is because of the alloy composition dependence of the martensite transformation start temperature M~. If the cycling test temperature is fixed for all

fh cyde max= I mm

I®

~j~emax=1

mm

®

emax = 1

m~ S

(a) Fig. 3 (a),

189

f

190

B. Strnadel et al. / Materials Science and Engineering A203 (1995) 187-196

emax= I mm

O'max:550 MPa

®

~

O'max=550 MPa

emax= I ram

h cycle

emax= I mm

q~x=550 MPa

(b) Fig. 3. Cyclic stress elongation diagrams for first and 50th cycles in hard and soft cycling modes for (a) T i - N i and T i - N i - C u alloys.

50.9at.%Ni), B (Ti-50.5at.%Ni) and C (Ti-50.0at.%Ni) in both hard and soft cycling modes in Fig. 3(a), it is found that the residual strain is small in the high Ni content alloys A and B, while it is large in the low .Ni content alloy C. This indicates that Ni is effective in increasing the critical stress o-s for slip deformation, since these three alloys were deformed at the same stress level. The stress hysteresis, which is measured by the difference between O'msand the stress aR for reverse transformation upon unloading, is small in alloys A and B, while it is large in alloy C. The large stress hysteresis in alloy C can be attributed to a decrease in ~rk due to the introduction of a higher density of dislocations upon preloading. In the hard cycling mode the PE characteristics after cycling are more stable in alloy A than in alloy B, i.e. eo is smaller and the PE diagram remains less changed in alloy A. This is because as is higher in alloy A than in alloy B. However, alloy C shows a different behaviour, i.e. eo of alloy C is smaller than that of alloy B after

cycling, although as of the former is lower than that of the latter. The PE after cycling in both alloys A and B is characterized by the forward transformation upon loading and the reverse transformation upon unloading, while that in alloy C is not characterized by such transformations but by twinning in the remaining martensite. The reason why the deformation mode in the latter case is twinning is that aR is lOW even in the first cycle and decreases rapidly to below zero, so that the reverse martensite transformation will not occur and the martensite remains during cyclic deformation after several periods of cycling. Therefore the deformation behaviour corresponding to the non-linear stressstrain curve of alloy C for the 10th cycle in Fig. 3(a) is twinning in the remaining martensite. Similar twinning deformation was also observed in a heavily cold-rolled T i - N i alloy [9]. The former deformation behaviour associated with the transformations is accompanied by back and forth movements of the parent-martensite interfaces creating dislocations and hence resulting in

B. Strnadel et al. / Materials Science and Engineering A203 (1995) 187 196

larger eo. However, the latter deformation behaviour associated with the twinning is reversible without creating dislocations, so that e o is smaller after cycling. In the soft cycling mode, however, there are some differences in the cycling effects. By comparing the firstand last-cycle stress elongation diagrams, it is found that after cycling, e,, of alloy A is larger than that of B, which is the opposite result from the hard cycling mode. In the soft cycling mode both alloys A and B show fully transformed PE behaviour for each cycle, since the applied maximum stress a .... is kept constant. However, the PE deformation in alloy A is cycled at higher stresses than in alloy B. Therefore eo is considered to become larger in alloy A than in alloy B, although o-~ of the former is higher than that of the latter. On the other hand, alloy C shows a twinning PE deformation after cycling which is similar to that of alloy C in the hard cycling mode. Since aR is low in alloy C even in the first cycle and will decrease further, possibly becoming negative, with cycling, eo is considered to consist of a considerable amount of transformation strain due to the remaining martensite. However, e,, in alloys A and B is considered to consist mostly of permanent slip deformation, so that the permanent slip deformation in alloy C can be even smaller than that in alloy B. This can be reasonably understood if one considers the difference in PE deformation modes, i.e. the transformation PE in alloy B and the twinning PE in alloy C. Fig. 3(b) presents a similar set of cyclic stress-elongation diagrams for the T i - N i - C u alloys. The Ni contents of alloys F, D and E are 41.5, 41.0 and 40.0at.% respectively. It is observed here too that among the first-cycle deformations in both hard and soft cycling modes the residual strain is less in the higher Ni content alloys F and D than in the lower Ni content alloy E. Basically the same type of discussion which was presented for the Ti Ni alloys in Fig. 3(a) will hold good here to some extent. However, the last-cycle stress-elongation diagrams do not show the typical transformation PE behaviour which was characterized by the transformation under almost constant stress and observed in alloys A and B in Fig. 3(a). It seems that the T i - N i - C u alloys used in the present experiment have low o-~ and show slip deformation more easily than the Ti Ni alloys. Figs. 4 - 6 show plots of the residual elongation eo, the critical stress O-msfor martensite formation and the hysteresis W of the PE diagram respectively in both hard and soft cycling modes. As in previous studies [3-7], it was confirmed that the PE characteristics respond to cycling as follows. As the number of cycles increases, eo grows, while O'ms and W both decrease. Generally speaking, the rate at which eo grows depends on cry, the work-hardening rate, O'ms and the type of cycling mode. In the T i - N i alloys, as and the work-

191

hardening rate are closely related to each other, i.e. the work-hardening rate is high in a high o-~ specimen [10]. Therefore we consider two factors, as and a .... in order to explain the effect of cyclic deformation on eo in each cycling mode. If the effect of a~ is dominant, eo will always be larger in a low Ni content alloy. This is observed in the T i - N i alloys when the number of cycles is less than seven or eight in the hard cycling mode (Fig. 4(a)) or 15 in the soft cycling mode (Fig. 4(b)). In the higher cycle number region, eo becomes larger in a high Ni content alloy than in a low Ni content alloy. In this case the effect of ams during cycling becomes dominant, i.e. am~ is always higher in the higher Ni content T i - N i alloys as shown in Fig. 5, so that slip deformation occurs effectively in the higher Ni content alloys owing to the effect of stress upon cyclic transformation. In the Ti Ni Cu alloys the effect of the number of cycles on eo is rather simple, because c5~ is equally low in all alloys, i.e. eo for the first cycle is considerable in all alloys as shown in Fig. 3(b) and there is no such difference as was observed in the T i - N i alloys. The hysteresis W also shows a specific dependence on cycling as seen in Fig. 6. W depends on the PE strain 1.2 emax=Imm

3A:~6"C = B:61"E ~, C: 93"C

1 " ~ 0.8

• [3:77"C • E:96°C AF: 78"C

......'J , ~,""

~o.6 0.4 0.2 0

ii

0

" ~ .....

t

.... ~ ...........• .......... ~-'--;'""t

10

20

.......... t

.......... t

. . . . . . . . . . t . . . . . . . . . . .4

30

4.0

50

30

/+0

50

CycLe

(o) 2.5

O'max: 550 NPa 2

EI.5 E

oo

I

0.5

0 (b)

10

20

Cyc[e

Fig. 4. Response of residual elongation to (a) hard and (b) soft cycling at an elongation rate of 0.01 m m s ~,

B. Strnadel et al./ Materials Science and Engineering A203 (1995) 187-196

192

1.5 600

e 500 r~

• .

~

max ~

emax:Imm

oA: 4.6"C ,D:77"C

=Imm ~

_

oB: 61"C ,~C: g3"C

oA:46'C o B:61"C aC:93"C

mE: g6"[ * F: 78"E

• D:77"[ mE:96"C * F: 78"C

z,00

b~

0.5

3OO

200 i

0 100

'

10

(o)

'

Cycle

3'o

'

10

'

I

.-.IF

10

i

i

20

i

,

30

i

i

~0

i

50

CycLe

so

(o) 3 o'max= 550 MPa

.

~-

~,

2.5 600

~ax = 550 MPa

2

500 ~ : 1.5

g. 4.00

1

300

0.5 0

200

(b) 100

(b)

.

. 10.

.

.

20

'

10

50

Cycle

0

i;

4

Cycle

Fig. 6. Response o f hysteresis to (a) hard and (b) soft cycling at 0,01 m m s - l

Fig. 5. Response of critical stress for martensite formation to (a) hard and (b) soft cycling at an elongation rate of 0.01 turn s - ].

and the difference between O'ms and o-R. These two factors increase with increasing Ni content of the firstcycle deformation in the T i - N i alloys as shown in Fig. 3(a). Therefore W is larger in a lower Ni content alloy than in a higher Ni content alloy as shown Fig. 6. However, the shape of the stress-elongation diagram of PE deformation changes drastically and becomes smaller for the lower Ni content alloys, so that W of a lower Ni content alloy decreases rapidly as the number of cycles increases• The T i - N i - C u alloys show a simpler behaviour of the change in W as a function of the number of cycles, because the PE characteristics are similar to each other as shown in Fig. 3(b). After cycling in the hard mode with a m a x i m u m elongation of 1 m m at a rate of 0•01 m m s -1, the same specimens were subjected to a further 50 cycles at emax = 1 m m but an elongation rate ~ = 0.1 m m s-~ and at the same temperature as in the previous cycling• To clarify how the alloys respond to elongation rate changes, another set of specimens was hard cycled in a reverse sequence of elongation rates: first 50 cycles at 0.1 m m s i and then another 50 cycles at 0.01 m m s 1. Figs. 7(a) 7(0 show the resultant stress-strain dia-

grams for the two elongation rate combinations. Comparison of the left-hand sides of these diagrams, which show the cycling PE behaviour at k = 0.01 m m s -~ (upper part) and 0.1 m m s ] (lower part), points to the following conclusions on the effect of elongation rate on the cyclic PE behaviour. A higher elongation rate generally entails faster nucleation of new martensite and more rapid propagation of the phase boundaries in the martensite transformation process, but also intensifies hardening and plastic deformation, so that the m a x i m u m effective applied stress is greater. Alloys which have a low critical stress as for slip harden at high elongation rates so strongly that the effect is clearly apparent on the slope of the part of the stresselongation curve where the forward and reverse transformations occur (see Figs. 7(d)-7(f)). The residual elongation eo is larger in the higher elongation rate test than in the lower rate test for all alloys except alloy A, which has a very high as. In alloys C - F , which have low O-s, e o consists of a considerable amount of unrecovered transformation strain. Figs. 8 - 1 0 illustrate how eo, ares and W vary in the course of hard cycling at 0.1 m m s 1. Comparison with the corresponding diagrams for 0.01 m m s - ] in Figs. 4(a), 5(a) and 6(a) shows that in the first and all further

B. Strnadel

et al. i Materials

Science und Engineering

cycles the residual elongation is larger at the higher elongation rate; the e, increment in each further cycle is also larger at the higher elongation rate (see Figs. 4(a) and 8). The notable exception in alloy A, with a dislocationand precipitation-hardened structure, where e, diminishes as the elongation rate increases and eO tends to stabilize as the number of cycles grows (see Fig. 8). The presence of Ti,Ni, precipitates and of more nucleation sites for the phase transformation is thought

@

@

187- 196

IYi

61-C @=O.Olmm/s

b

@

46-C ~=O.lmmls

b

@

b=O.lmmls

ZOOMPa

b

(1995)

at higher elongation rates to promote the development of the transformation in smaller separate zones and to supply an effective back stress for the reverse transformation upon unloading. Consequently, the reverse transformation then starts at a higher stress level and the final result is a lesser residual elongation. Figs. 5(a) and 9 prove that o-‘msdiminishes more rapidly in the course of cycling at the high than at the low elongation rate. The explanation appears to be that the higher

46-c &= O.Olmm/s

A203

&=O.Olmm/s

B=O.:mm/s

b

61-c ~=O.Olmm/s

&=O.lmm/s b

b

kzzif_ e

e

0 I 200MPa

@

93'C ~=O.Olmm/s

T 1200MPa

~=O.lmm/s

I

77'C ~:O.Olmm/s

~=O.lmm/s b

!Il.ic 0.2mm

0

ki?

93'C e=O.lmm/s

@ 6= 0.01 mm/s

b

~=O.lmm/s

z e

Fig. 7 (a)-(d).

e

77-c

b

I

_

t3

~=O.Olmm/s

p.': e

B. Strnadel et al. / Materials Science and Engineering A203 (1995) 187 196

194

(•)

(~

96"C

e:O.O1mm/s

e=O.01 mmls

e=O.1mm/s

200MPa..

e=0.1minis

200MPa

~~~~0

e

~o

b

e=O.O1mmls

7B'C ~:0.I mm/s

b ~ 1 0 f h I///25fh

e

e

e

(~

96"[ ~:O.Immls

O

~~mm

h/5oCfY: f [:yc [e

e

b

78"C

b

cycle cycle

e

e=O.O1mmls I ~~llOth ~ ~ / 2 5 f h

e

cycle cycte

e

Fig. 7. Effect of elongation rate change on stress elongation diagrams for first and 50th cycles, or other cycles if noted, in hard cycling mode in alloys A - F . Diagrams on the left show the results for the first group of cycles, while those on the right show the results for the second group of cycles at a different elongation rate.

1.2

600

emax:lmm

°A:~,6"C

e

oD:77"[

I

oA: 46"[

-Imm

i~ max5OO ['0.

oB:61"[ mE:%'£ ~ C:93"C AF:?8"[

,D:77"[

DB:61"C mE:96"[ ~,C:g3"c '.F:78"[

f@

"-~"0.8

~400

~:~::::::::ZZ~

. . . . . . . . . c . . . . . . . . . . < > . . . . . . . . . . . ( > . . . . . . . . . . < > . . . . . . . . . . ( > . . . . . . . . . . c . . . . . . . . . . . o . . . . . . . . . . . .o

.....-"

~o.6

300 ...

...~ ......

..-

0.~

....e, ....... ~3-~o.

"o

........... o - ....

200

0.2 100 0

i

i

10

i

i

20

i

Cycle

i

30

i

i

~0

i

i

50

Fig. 8. Response of residual elongation to hard cycling at an elongation rate of 0.1 m m s - 1.

applied stress at a higher elongation rate causes slip deformation to occur more easily; this slip deformation creates internal stress which assists the transformation to occur, resulting in decreasing O'mseffectively. The cycling at two different elongation rates, surveyed in Figs. 7(a)-7(f), was aimed at investigating how it affected the stability of the stress-elongation diagrams. This is best ascertained by plots of residual elongation vs. number of cycles such as are presented in Fig. 11 for the 0.01 4-0.1 m m s ~ combination and in Fig. 12 for the 0.1 + 0.01 m m s -~ sequence. Except for the lowest nickel ternary alloy E, which exhibited un-

i

J

10

i

i

20

i

Cycle

i

30

,

i

40

i

i

50

Fig. 9. Response of critical stress for martensite formation to hard cycling at an elongation rate of 0.1 m m s - t .

stable stress elongation curves at both elongation rates, the effects of elongation rate change proved to be considerable. This is not entirely due to the different Ni contents and hence different degrees of hardening after heat treatment. Unfortunately, there is a further factor, one which hampers objective comparisons, and that is the residual elongation remaining after cycling at the first rate is concluded. The amount of this elongation varies and so therefore does the initial elonation at which the second set of cycles at the different rate commences. Even so, there are some valid conclusions to be drawn from these experiments, as follows.

B. Strnadel et al. / Materials Science and Engineering A203 (1995) 187.-196

1.5

f~

e~x=l mrn

o A:~.6"C

• D:77"C

° B:61"C ,~ C: 93"C

= E:96"C * F:78"C

0.5 .. ..... --.i".';'"2..........~ . . . . . . . . .

"~:.:::.....~

..... -

o0

..........

Z

10

........... ,c. ......

,

2O

.~.

,

~

-

:

, .........: ..........: ........... ~0 50

30

Cycte Fig. 10. Response of hysteresis to hard cycling at 0.1 m m s-~.

1.6;

emaxzlrnrn

I.~

°A: t*6"C e0:77"C

": ...o"

.~"

a 8 : 61"C

= E:96"C

~C: 9TC

* F: 78"E

~..-o.~°"~"

0.8

~ / F"

o.2°~°6 0

..................

J

L

i

i

10

i

j

20

i

i

3O

i

i

~.0

.

.

5O

.

.

,

6O

,

70

,

,

8O

,

,

9O

•

100

Cycle Fig.

11. Response

of residual

elongation

to

hard

cycling at

0.01rams ] and then at 0.1mms -1.

1.6

%==lmm

1.k.

o A : 4.6"C ....• . " J O B : 61"C w" ~C 9 3 %

r"r':

• D:77"C =E:96"C " F 78"C

1.2

E

~o.s 0.6 0,4 0.2

0

.~..::~w'r" • p ,

,

lO

,

. . . . 1

?o

,

,

30

~

I

4.o

i

i

~

I

5o 6o Cycle

I

Fig. 12. Response of residual elongation 0.1 m m s - I and then at 0.01 m m s ~.

t

70

J

to

a

so

t

hard

J

90

i

J

loo

L

cycling at

Alloy E, with the lowest nickel content of the ternary alloys and a low critical stress as for slip, failed to undergo any pronounced hardening in the course of its cycling at either of the two rates. Its residual elongation is therefore large and neither of the two combinations

195

of elongation rates stabilized the cyclic stress-elongation diagram of this alloy. Alloy F displayed more plastic deformation at 0.1 m m s 1; at this rate its stress-elongation diagram did not stabilize and it still did not do so even after further cycling at the lower rate (see Figs. 7(f) and 12). However, at 0.01 mm s z this diagram stabilized, the residual elongation was slight and the diagram remained stable even during subsequent cycling at 0.1 mm s ~ (see Figs. 7(f) and 11). Alloy D responded in the opposite way: this is ascribed to the large maximum stress in the second set of cycles at 0.1 mm s-~ (see the upper part of Fig. 7(d)), which causes more residual plastic deformation. The T i - N i alloy A retains so much residual plastic deformation after cycling at 001 mm s ~ that the maximum applied stress in the subsequent cycling at 0.1 mm s ~ is very large. Consequently, the residual plastic deformation, instead of stabilizing, continues to grow with the number of cycles, so that the cyclic stress-elongation diagram is not stabilized either (see the upper part of Fig. 7(a)). When the elongation rate sequence is reversed, the residual deformation at 0.1 mm s l is much lower and so therefore is the maximum effective stress level in the subsequent cycling at 0.01 mm s ~. Consequently, the residual deformation does not grow with the number of cycles as rapidly as in the former case (see Fig. 12). Alloys B and C, which contain less nickel than alloy A, respond much more favourably to the sequence of 0.01 mm s 1 followed by 0.1 m m s ~ than to the reverse sequence, because in the former case the residual plastic deformation grows with the number of cycles more slowly than in the latter case (see Figs. 11 and 12). This is due to the lesser residual deformation at the lower elongation rate and the rapid growth of the maximum stress level during the second round of cycling at 0.1 mm s ]. These effects raise the critical stress o-S for slip and cause the residual deformation to diminish as the cycling progresses. The larger residual plastic deformation during the first group of cycles at the higher elongation rate produces a subsequent quick growth of residual deformation in the second group of cycles at the lower elongation rate. All this can be clearly seen in Figs. 7(b), 7(c), l l and 12. As described in the preceding paragraphs, the effect of elongation rate change appears rather complicated and differs depending on the alloy composition. In order to understand fully the effect of elongation rate change, it is necessary to make a further systematic investigation.

4. Conclusions Three T i - N i and three T i - N i - C u shape memory alloys were cycled in both hard and soft cycling test modes, each alloy at a temperature at which its critical

196

B. Strnadel et al. / Materials Science and Engineering A203 (1995) 187 196

stress for inducing martensite in the first cycle equalled 500 MPa. This revealed general responses of the pseudoelasticity characteristics to mechanical cycling: the residual elongation eo grows, but the critical stress o-ms for martensite formation and the hysteresis W decline as the number of cycles increases. The rate at which eo grows depends on o-s, o-msand the type of cycling mode. Therefore we considered two factors, o-s and o-.... in order to explain the effect of cyclic deformation on eo in each cycling mode. If the effect of o-, is dominant, eo will always be larger in a low Ni content alloy. This is observed in the T i - N i alloys when the number of cycles is less than seven or eight in the hard cycling mode (Fig. 4(a)) or 15 in the soft cycling mode (Fig. 4(b)). In the higher cycle number region, eo becomes larger in a high Ni content alloy than in a low Ni content alloy. In this case the effect of o-ms during cycling becomes dominant, i.e. O'msis always higher in the higher Ni content Ti Ni alloys, so that slip deformation occurs effectively in the higher Ni content alloys owing to the effect of stress upon cyclic transformation. On the other hand, the Ti Ni Cu alloys used in the present experiment have low a s and show slip deformation more easily than the T i - N i alloys. In the former alloys the effect of the number of cycles on e o is rather simple, because o-~ is equally low in all alloys and the PE behaviour during cycling is similar. In the examined alloys, cycling at a higher elongation rate was found to increase not only the amount of residual deformation left after the specimens are unloaded, but also the critical stress for martensite formation. The critical stress diminishes, as cycling proceeds, more rapidly at a high than at a low elongation rate. However, different elongation rates did not appear to have any substantial effect on the response of hysteresis to the progress of cycling.

Changes in the elongation rate affected the cyclic stress-strain diagrams to an extent which depended on the amount of residual deformation and on the stability of those diagrams in the course of cycling at the first elongation rate. Large residual deformation and/or instability of the diagram during cycling at the first elongation rate lead to instability of the cyclic diagrams during subsequent cycling at an altered elongation rate too.

Acknowledgment This work was financially supported by JISTEC, the Japan International Science and Technology Exchange Center.

References [1] S. Miyazaki, K. Otsuka and Y. Suzuki, Scr. Metall., 15 (1981) 287. [2] S. Miyazaki, Y. Ohmi, K. Otsuka and Y. Suzuki, J. Phys. (Paris), Colloq. C4, Suppl. 12, 43 (1982) 255. [3] S. Miyazaki, T. lmai, Y. lgo and K. Otsuka, Metall. Trans. A, 17 (1986) 115. [4] S. Miyazaki and K. Otsuka, ISIJ Int., 29 (1989) 353. [5] S. Miyazaki, in T.W. Duering et al. (eds.), Engineering Aspects of Shape Memory Alloys, Butterworth-Heinemann, Oxford, 1990, p. 394. [6] B. Strnadel, H. Ohtsuka, S. Ohashi, S. Miyazaki, T. Ishihara and S. Kajiwara, Proc. Conf. JIM, Nagoya, 1993, p. 277, [7] B. Stnadel, S. Ohashi, H. Ohtsuka, S. Miyazaki and T. Ishihara, Mater. Sci. Eng. A, in press. [8] M. Nishida, C.M. Wayman and T. Honma, Metall. Trans. A, 17 (1986) 1505. [9] T. Tadaki and C.M. Wayrnan, Scr. Metall. 14 (1980) 911. [10] S. Miyazaki, Y. Kohiyama, K. Otsuka and T.W. Duerig, Mater. Sci. Forum, 56 58 (1990) 765.