The four-step multiple stage transformation in deformed
and annealed Ti
49
Ni
51
shape memory alloy
P.C. Su, S.K. Wu
*Department of Materials Science and Engineering, National Taiwan University, 1, Roosevelt Rd., Sec. 4, Taipei 106, Taiwan Received 7 August 2003; received in revised form 29 October 2003; accepted 30 October 2003
Abstract
A four-step multiple stage transformation is observed in 20% deformed and 500°C annealed Ti49Ni51shape memory alloy. Two
extra B2! B190transformation peaks appear before the previously described B2 ! R and R ! B190peaks while cooling, and these
correspond to one new peak, which appears after the original B190! B2 peak during heating. These two extra peaks are caused by
the combined effect of severe cold-working and long-time annealing on Ti49Ni51alloy, and they come separately from the B2! B190
transformation occurring in regions with low and high dislocation densities, which are originally suppressed by cold-working. Ó 2003 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
Keywords: Cold working; Annealing; Differential scanning calorimetry; Shape memory alloys; Martensitic transformation
1. Introduction
The transformation sequence of TiNi shape memory alloys (SMAs) has been widely investigated in the past decades. In the equiatomic TiNi SMA, the martensitic transformation occurs in a single stage from high
tem-perature parent B2 phase to low temtem-perature B190
monoclinic martensite. After certain thermomechanical treatments, or addition of a third element, the trans-formation sequence of TiNi SMAs can change into a
two-stage B2 to premartensite R-phase to B190 during
cooling [1]. In addition, aged Ni-rich TiNi alloys with scattered Ti3Ni4 precipitates in the matrix can exhibit
either B2$ R $ B190transformation sequence with two
distinct steps during both cooling and heating runs, or a
B2! R ! B190and B190! B2 transformation sequence
with two distinct steps during cooling and only one step during heating [2].
In the past decade, the martensitic transformation was found to be able to appear in more than two distinct steps and is said to be a multi-stage martensitic trans-formation (MST). This MST occurring in TiNi alloys
has been widely observed using differential scanning calorimetry (DSC), although the cause of this behavior is still controversial. Todoroki and Tamura [3] first ex-plained the effect of cold working on the transformation sequence of TiNi SMAs. Lo et al. [4] also found a two-stage transformation in as-quenched Ti40Ni50Cu10with
B2 ! B19 and B19 ! B190 transformations, where B19
is an orthorhombic martensite. Bataillard et al. [5,6] attributed the cause of MST to the stress fields formed around the coherent interfaces between B2 matrix and
Ti3Ni4 precipitates, whereas Morawiec et al. [7–9]
ex-plained MST to be due to the changes of dislocation configuration by low-temperature annealing. In a recent study, Khalil-Allafi et al. [10] found two distinct steps
of B2 ! R and R ! B190, and an additional step with
B2! B190 in the DSC cooling curves of annealed
Ni-rich TiNi alloys. They attributed the cause of MST to
the Ni-concentration inhomogeneity between Ti3Ni4
precipitates and the difference in nucleation barriers
for R-phase and B190 formations. However, they soon
corrected these proposed reasons with the argument that the heterogeneous microstructure between regions with
and without Ti3Ni4 precipitates is responsible for the
MST behaviors [11]. Recently, Chrobak et al. [12] stated that the MST in the early stage of annealed Ti49:3Ni50:7
with 10% cold-working had two transformation peaks
*
Corresponding author. Tel.: 7846; fax: +886-2-2363-4562.
E-mail address:[email protected](S.K. Wu).
www.actamat-journals.com
1359-6454/$30.00Ó 2003 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.actamat.2003.10.044
of B2! R and R ! B19, and that the R! B19 peak consists of two transformation sub-peaks occurring at different temperatures.
According to reports in the literature, a total of three transformation steps have been observed in the DSC cooling curve of TiNi SMAs. In this study, a four-stage MST is found in the deformed and annealed Ni-rich
Ti49Ni51 SMAs. The transformation sequence of this
four-stage MST is discussed to gain a better under-standing of the cause and mechanism of MST behaviors.
2. Experimental
The conventional tungsten arc-melting technique
was employed to prepare the Ni-rich Ti49Ni51 SMA.
Titanium (99.7 wt%) and nickel (99.98%), totaling nearly 140 g, were melted and remelted at least six times in an argon atmosphere with pure Ti as the get-ter. The mass loss during the melting was negligible.
The ingot was hot rolled at 850 °C into a plate with
2 mm thickness and then solution-treated at 800°C for
2 h followed by quenching with water. Thereafter, the solution-treated plate was cold-rolled to obtain 20% thickness reduction and then cut into specimens to conduct DSC, dynamic mechanical analyzer (DMA), and electrical resistivity (ER) tests. The cut specimens
were then annealed at 500°C for various time intervals
between 0.5 and 200 h.
The transformation behavior and characteristic tem-peratures were determined using a DSC cell (TA Q series Q10) with the specimen weight between 40 and 60 mg. In
DSC tests, specimens were heated up to 100°C and kept
isothermal for 3 min to obtain thermal equilibrium, then
cooled down to )120 °C, and also kept isothermal for
3 min and then brought back to 100 °C again. All
heating and cooling rates were kept at 10°C/min.
The changes of mechanical characteristics during the transformation were observed using DMA (TA 2980 model) in the multi-frequency mode. Specimens with the
size of 35 5 1:7 mm were measured in the same
se-quence and temperature range as in the DSC test, except
that the ramp rate was 3 °C/min. The measuring
fre-quency was 1Hz and the amplitude was 2 lm. The volume of DMA specimen chamber was much larger than that of DSC, and the thermocouple was placed farther from the specimen in DMA than in DSC. As a result, for the same thermomechanical treated specimen,
a temperature lag of about 10 °C is observed between
DSC and DMA results.
The martensitic and R phase transformation behav-iors were also measured using a four-probe ER test in constant current mode. The system for measurement was composed of a temperature control device (Cryocon 32B by Cryogenic Control Systems, Inc.), a power supply (GP0250-3R DC power supply), and a data
re-cording system (Keithley 2000 Multimeter for voltage measurement, and the PCI-GPIB interface card of NI
for recording). Specimens with the size of 35 5 1:7
mm were heated up to 100 °C and then cooled down
with a rate of 4 °C/min. Data were recorded from 70 to
)50 °C.
3. Results and discussion 3.1. DSC results
DSC results of 20% deformed Ti49Ni51alloy annealed
at 500°C for different time intervals are shown in Figs. 1 and 2. Fig. 1 indicates that the 0.5-h curve remains a typical transformation sequence of aged Ni-rich TiNi SMAs, with two distinct transformation peaks of
B2$ R and R $ B190. After 3 h of annealing, the two
reverse transformation peaks merge to become a single
peak of B190! B2, and later this single peak in the
heating curve tends to widen towards high temperature after annealing for 24 h, as indicated by the arrow in Fig. 1. When the specimens were annealed further for 48, 72, 100, and 125 h, the transformation behavior changed and evolved to a four-stage transformation in cooling, as shown in Fig. 2(a). The corresponding heating curves of Fig. 2(a) are shown in Fig. 2(b). In Fig. 2, the arrows 1–4, 10, and 20indicate transformation
Fig. 1. DSC cooling–heating curves for 20% cold-rolled Ti49Ni51
peaks 1–4, 10and 20, respectively. In Fig. 2(a) of the 48-h
cooling curve, a new transformation peak (peak 1)
ap-pears before the original B2! R peak (peak 3), and it
also appears as a new peak after the original B190! B2
transformation peak in the corresponding heating curve in Fig. 2(b). After 72 h of aging, another new peak (peak
2) appears between the original B2! R peak (peak 3)
and the peak 1. These results clearly reveal a four-stage
transformation behavior (peak 4 is the original
R! B190 peak shown in Fig. 1). When annealed up to
100 and 125 h, the new peaks in cooling (peaks 1 and 2)
and in heating (peak 20) become more prominent than
the original peaks (peaks 3, 4 and 10). Fig. 3(a) and (b)
plot the peak temperature and total transformation heat in heating DHhvs. annealing time, respectively, from the
data in Figs. 1 and 2. 3.2. DMA results
Fig. 4(a) shows the DMA storage modulus cooling curves of specimens annealed at 48, 72, 100, and 150 h. In the typical DMA curves of TiNi alloys, the storage modulus curve shows a minimum upon transformation,
and the B2! R has a relatively deep minimum
com-pared to that of R! B190 [2]. These characteristics are
also visible in the 48-h curve shown in Fig. 4(a). In the DSC curves of Fig. 2(a), the 48-h annealed specimen already shows a new peak 1, but in Fig. 4(a), no ap-parent transformation minimum is found. This may be
-60 -40 -20 0 20 40 60 80 0 20 40 60 80 100 120 140 Annealing time (hrs) Tem p er at u re ( OC ) peak 1 (B2→B19') peak 2 (B2 B19') peak 3 (B2 R) peak 4 (R B19') peak 1' (B19' B2) peak 2' (B19' B2) 0 5 10 15 20 25 0 20 40 60 80 100 120 140 Annealing time (hrs) ∆ Hh (J/g) → → → → → (a) (b)
Fig. 3. Transformation temperature and heat evolution of 20% cold-rolled Ti49Ni51annealed at 500°C.
due to the fact that the transformation volume is not large enough to appear in the DMA results. In the 72
and 100 h curves, the amount of R! B190
transfor-mation (peak 4) is reduced and thus the minima can no
longer be seen, but the B2! R (peak 3) minima are
clearly affected by the newborn peaks 1 and 2 in such a way that the curves go to the minima less abruptly, as indicated by bold arrows in Fig. 4(a). After 150 h of
annealing, the peak 3 of B2! R transformation is
al-most covered by peak 2 because the relative depth of the minima are smaller compared to the curves obtained at 48–100 h, and thus the two minima in the storage modulus curve should correspond to peaks 1 and 2 in-dividually. The tan delta curves corresponding to Fig. 4(a), which show the damping property changes during transformation, are shown in Fig. 4(b). Typically the tan delta peak corresponding to R-phase transfor-mation is sharper and the transfortransfor-mation is completed within a shorter temperature range, whereas the peak corresponding to martensitic transformation is broader and covers a wider temperature range [2]. In Fig. 4(b), the 48-h curve, no apparent increase in the damping capacity can be found for the appearance of peak 1 for the same reason as stated in the case of Fig. 4(a). At 72
and 100 h, the shoulder before B2! R transformation
peak becomes higher because of the development of peaks 1 and 2, as indicated by the bold arrows in
Fig. 4(b). At 150 h, two distinct peaks corresponding to peaks 1 and 2 are observed.
3.3. ER results
Fig. 5 shows the plots of ER vs. temperature for Ti49Ni51SMAs annealed at 500°C for 48, 72, and 100 h.
In the 48-h curve, the ER does not reflect the existence of peak 1 in DSC curve, and only an abrupt increase in
the ER of B2! R transformation can be observed.
However, in the 72 and 100 h curves, before the B2! R
transformation (peak 3), there can be seen the
B2! B190 martensitic transformation characteristic,
which tends to lower the ER value and thus causes the original uprising curve to decrease and even stop in-creasing, as indicated by bold arrows in Fig. 5. This provides evidence for our viewpoint that peaks 1 and 2
are related to the B2! B190 transformation, as
dis-cussed further in Section 3.4.
3.4. Discussion of the transformation behavior of four-step MST in cooling
3.4.1. Cause of extra peaks
In Figs. 1 and 2, the peaks 3, 4, and 10correspond to
the B2! R, R ! B190, and B190! B2 transformations,
respectively, similar to those already reported [1,2]. Since we have two new peaks (peaks 1 and 2), instead of only one peak as observed in the past studies, we chemically etched off the 50% thickness from both sides of the rolling surface of the 20% deformed specimen, and performed the DSC test to see if these two extra peaks were caused by non-uniform cold-working re-duction between the surface and center areas of the deformed plate. However, the DSC results were the same, thus ruling out the possibility of cold-working non-uniformity to cause the MST in Fig. 2.
Fig. 6 plots the peak height vs. annealing time from the data of Fig. 2. In Fig. 6, the height of peak 3
(B2! R) does not change significantly when MST
ap-pears. Also, its peak temperature remains almost con-stant, as shown in Fig. 3. The peak temperature of peak
4 (R! B190) evolves with time in a similar way as the
R! B190 peak in non-deformed Ti
49Ni51 annealed at
400°C [13]. Also in Fig. 6, the heights of peaks 4 and 10
are clearly lower as extra peaks 1, 2 and 20 appear and
grow prominently. This phenomenon indicates that the MST shown in Fig. 2 is closely related to martensitic transformation, instead of R phase transformation. It is well known that the high density of dislocations induced by cold-working in the specimen will suppress its mar-tensitic transformation. In Fig. 1 (the 1-h curve), for
example, the transformation heat, DHC, of the two
peaks in cooling has a sum of 11.8 J/g, which is smaller than the reverse transformation heat of 16.3 J/g. This
loss of about 5 J/g heat is believed to be covered by the
transformation between the two peaks of B2! R and
R! B190in cooling. That is, the martensite transforms
continuously from R-phase after the appearance of
the B2! R peak. Furthermore, the B190! B2 reverse
transformation heat (16.3 J/g) is much smaller than that of TiNi specimens which have not undergone cold work, which is about 25 J/g [14]. Therefore, there should be a large amount of martensitic transformation which is hindered by the high density of dislocations induced by
cold-rolling. Fig. 3(b) shows the DHhevolution curve in
the reverse transformation of Figs. 1 and 2. In Fig. 3, the
DHhhas an apparent increase as the MST appears after
24 h of annealing; and it is as large as 23.4 J/g at 125 h of annealing, reaching about 25 J/g at 200 h of annealing (not shown in Fig. 2). These phenomena indicate that the originally suppressed martensitic transformation is released after being annealed for a long period of time,
re-appearing and producing new peaks 1, 2, and 20
in-stead of merely increasing the DH of the existing peaks, 3, 4, and 10.
3.4.2. The transformation behavior of the extra peaks The reported MST of Ni-rich TiNi alloy during
cooling consists of a B2 ! R peak, an R ! B190 peak,
and an extra peak that appears after the B2! R peak
[8,10–12]. However, in this study, except for the
B2! R and R ! B190 peaks, two extra peaks (peaks 1
and 2) shown in Fig. 2 are observed before the B2! R
peak but not after the B2! R peak. The reason that
the two extra peaks observed in the present case appear
before the B2! R peak may be due to the long-term
annealing of Ni-rich TiNi specimens and thus the
Ti3Ni4 precipitating effect makes the composition of
the matrix between precipitates reach near equiatomic TiNi. Thus, the B2 matrix has a transformation
40 60 80 100 120 140 Annealing time (hrs) peak 1 peak 2 peak 3 peak 4 peak 1' peak 2' Peak heigh t (arbitrar y u nit)
Fig. 6. Evolution of transformation peak heights for Ti49Ni51annealed
at 500°C for 48, 72, 100, and 125 h. -100 -50 0 50 100 Temperature (˚C) Electrical re s istivit y (arbitrary unit) Peak 3 (RB2) Peak 4 (B19’R) Peak 3 Peak 4 Peak 2 (B19’B2) Peak 1 (B19’B2) Peak 4 Peak 3 Peak 2 Peak 1
Fig. 5. Electrical resistivity vs. temperature cooling curves for Ti49Ni51
temperature (peaks 1 and 2 of Figs. 2 and 3) above room temperature during cooling. We suggest that peaks 1 and 2 in Fig. 2(a) should be two separate
B2! B190 transformations, both corresponding to
peak 20 of Fig. 2(b).
Recently, Chrobak et al. [12] deformed Ti49:3Ni50:7
SMA with 10% reduction and annealed in a much shorter time (5–240 min) and to a lower temperature
(400 °C) than we did. From their TEM observations,
the Ti3Ni4 precipitates would decorate the dislocations
induced by cold-working. The dislocations in some precipitate free areas are bowed after 240 min of an-nealing and hence the deformed sample stores less elastic energy. Chrobak et al. [12] suggested that the sample volume is divided into areas that differ in the
density of precipitates,so two separate R! B190 peaks
occur. Khalil-Allafi et al. [11] also attributed the cause
of MST in non-deformed Ti49:4Ni50:6 SMA to the same
reason, but their three-step transformation peaks are
B2! R and R ! B190, both of which transform in
regions with precipitate; and the other B2! B190 peak
occurs in precipitates free regions. In the present study,
we suggest that our two extra steps of B2! B190
(peaks 1 and 2 of Fig. 2) can also be caused by the difference of the required martensitic transformation energy between different sample areas. The larger cold-working reduction induces relatively higher dislocation density, and these dislocations will recover around the precipitates upon increasing the annealing time. Ac-cording to the argument of [12], we suggest that the volume with higher precipitate density in our long-time annealed sample will accumulate higher dislocation density where more energy is needed to start mar-tensitic transformation. On the other hand, the pre-cipitate free volume in the sample will have low dislocation density and thus needs less energy to transform the martensite. As the annealing time in-creases, the dislocations bend more toward the pre-cipitates, resulting in an increase in the high dislocation volume and causing peak 2 to become more prominent. The volume with low dislocation density also increases upon increasing the annealing time due to dislocation recovery in areas between precipitates, and causes peak 1 to grow higher. From Fig. 6, the peak heights of peaks 1 and 2 increase in the same way, but the relative height of peak 2 is much higher than that of peak 1. This may indicate that the volume with high disloca-tion density is larger than that with low dislocadisloca-tion density.
4. Conclusion
The two extra steps during the forward transforma-tion can be interpreted as follows. Peak 1 corresponds to
the B2! B190 transformation in the low dislocation
density region, and peak 2 also corresponds to the
B2! B190 transformation but now in the high
disloca-tion density region. These peaks are not related to R phase transformation. The one extra step in the reverse transformation corresponds to peaks 1 and 2 and is a
B190! B2 transformation. These two extra
transforma-tion peaks occurring in cooling are related to the reap-pearance of martensitic transformation that is suppressed by dislocations induced by cold-working. The extra
peaks may appear before B2! R transformation peak
due to the higher temperature and longer annealing time applied in this study,which make the composition of matrix very close to equiatomic TiNi SMA. The four-step process of transformation behavior observed in this study, which is even more complex than the results of past studies, is suggested to be the result of combined effect of severe cold-working and long-time annealing, and is not caused by the cold-working non-uniformity between the surface and center areas of the deformed plate.
Acknowledgements
The authors like to acknowledge the financial support of this research from the National Science Council (NSC), Taiwan, Republic of China, under Grant No. NSC91-2216-E002-035.
References
[1] Funakubo H. Shape memory alloys. OPA Amsterdam: Gordon and Breach Science Publishers; 1987. p. 63.
[2] Wu SK, Lin HC, Chou TS. Acta Metall Mater 1990;38:95. [3] Todoroki T, Tamura H. Trans Jpn Inst Met 1987;28:83. [4] Lo YC, Wu SK, Horng HE. Acta Met Mater 1993;41:747. [5] Bataillard L, Gotthardt R. J de Phys III 1995;5:C8–647. [6] Bataillard L, Bidaux J-E, Gotthardt R. Philos Mag 1998;78:327. [7] Morawiec H, Stroz D, Chrobak D. J de Phys IV 1995;5:C2–155. [8] Morawiec H, Stroz D, Goryczka T, Chrobak D. Scr Mater
1996;35:485.
[9] Morawiec H, Ilczukl J, Stroz D, Goryczka T, Chrobak D. J de Phys IV 1997;7:C5–C155.
[10] Khalil-Allafi J, Ren X, Eggeler G. Acta Mater 2002;50:793. [11] Khalil-Allafi J, Dlouhy A, Eggeler G. Acta Mater 2002;50:4255. [12] Chrobak D, Stroz D, Morawiec H. Scr Mater 2003;48:571. [13] Lin HC, Wu SK. Scr Metall Mater 1991;25:1529. [14] Goo E, Sinclair R. Acta Metall Mater 1985;33:1717.
