Physica B 270 (1999) 289}297

Thermal expansion in various phases of Nitinol using TMA

J. Uchil *, K.P. Mohanchandra , K. Ganesh Kumara , K.K. Mahesh ,
T.P. Murali

Department of Materials Science, Mangalore University, Mangalagangotri, Mangalore 574 199, India

Space Mechanisms Group, ISRO Satellite Centre, Bangalore 560 017, India
Received 18 February 1999; received in revised form 22 April 1999

Abstract

Phase transformation in the near-equiatomic, cold-worked Nitinol, exhibiting shape memory e!ect, has been investi-
gated through the study of thermal expansion by employing thermomechanical analyser (TMA). The characteristic
transformation temperatures determined by this method are compared with those obtained through DSC thermograms.
The reliability and sensitivity of the thermal expansion probe have been discussed. The variation of thermal expansion
coe$cient with temperature in di!erent heat treated samples has been studied in the range 30}1203C. Thermal expansion
coe$cient is found to be !ve during MA transformation and #ve in RA transformations. Thermal expansion
coe$cient in martensitic region in the presence and absence of R-phase has been determined.  1999 Elsevier Science
B.V. All rights reserved.

Keywords: Shape memory alloy; Thermal expansion probe; Thermochemical analyser; Phase transformation

1. Introduction clinic unit cell and the intermediate phase consists

of rhombohedral crystallites. During the trans-
Thermal expansion data is one of the important formation of these phases there is a change in the
engineering design parameters, particularly, when size of the sample and hence changes in the coe$-
it involves the solid-state-phase transformations in cient of thermal expansion. The change in the coef-
a metal or in an alloy. Nitinol, an equiatomic shape "cient of thermal expansion can be used for the
memory alloy consisting of Ni and Ti, possesses characterization of di!erent phases in Nitinol.
high- and low-temperature crystalline phases (A The study of crystalline-phase transformation
and M, respectively) in general, and in some cases behaviour involves many techniques, viz., di!eren-
there exists an intermediate (R) phase also. The tial scanning calorimetry, electrical resistivity, in-
high-temperature phase of Nitinol has the BCC ternal friction measurement and thermoelectric
structure, the low-temperature phase has a mono- power. Using these probes, transformation behav-
iour has been studied extensively by many re-
searchers [1}6]. Uchil et al. have also studied the
* Corresponding author. Tel.: #91-824-742249; fax: #91- dependence of heat-treatment temperatures on
824-742367. various phases in Nitinol by electrical resistivity
E-mail address: uchilji@mnglr.ernet.in (J. Uchil) measurements [7]. A comparison with di!erential

0921-4526/99/$ - see front matter  1999 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 1 - 4 5 2 6 ( 9 9 ) 0 0 1 8 6 - 6
290 J. Uchil et al. / Physica B 270 (1999) 289}297

calorimetry and temperature dependence of electri- of thermal expansion. The length of the sample
cal resistance shows that the thermal expansion taken is about 8 mm and the initial length is mea-
measurement yields equivalent results with great sured using the micrometer having the accuracy of
accuracy and consistency. K. Mukherjee et al. 0.01 mm. Cr}Al thermo-couple is used for the
studied the thermal e!ects associated with stress measurement of sample temperature, with an accu-
induced martensitic transformation in Nitinol. The racy of 0.13C. The samples are heated from 303C to
observed development of tensile stress on heating 1253C and then cooled to 303C, the heating and
the sample near A , instead of compressive stress cooling rate kept identical at 23C/min. A small load

was explained on the basis of a possible negative of 0.5 g is applied on the sample through the expan-
coe$cient of thermal expansion around the trans- sion probe to ensure the proper contact of the
formation [8]. Kuroiwa et al. studied the thermal probe and the sample. The change in the length of
expansion near the martensitic transformation in the sample with temperature is measured using
slow cooled and quenched Au}Cd alloy [9]. a linear variable di!erential transformer (LVDT),
In the present study, a thermomechanical ana- the sensitivity of which is 0.1 lm. The results are
lyser (TMA) has been employed to identify di!erent recorded in the computer using the interface TA-50.
crystalline phases of Nitinol through temperature Identically heat-treated samples, of 10 mg in
variation of thermal expansion and determine the weight, are taken for the thermal analysis using a
associated transformation temperatures. In the di!erential scanning calorimeter (DSC-50,
present work, temperature dependence of thermal Shimadzu). Procedure followed in the earlier work
expansion (over a wide temperature region) has [1,7] has been made use of in determining trans-
been used to identify various crystalline phases of formation temperatures.
Nitinol and their transformation behaviour by em-
ploying TMA. The reliability of the results ob-
tained through this method has been con"rmed by 3. Results and discussion
di!erential scanning calorimetry (DSC).
Thermal expansion values of martensitic and The observed temperature dependence of ther-
austenitic phases given by the manufacturers are in mal strain of Nitinol wire heat-treated at di!erent
close agreement with the values obtained in the cur- temperatures reveals certain salient features. Vari-
rent investigation. E!ort has been made to study ation of thermal strain *¸/¸ , where &*¸' is the

systematically the transformation behaviour of change in length and &¸ ', the initial length at room

Nitinol heat treated at various temperatures. The temperature, as a function of temperature in the
validity of this method was veri"ed by comparing range 30}1203C, for the sample heat treated at
with one of the most usual technique } di!erential 4003C, 4803C and 5003C is shown in Figs. 1(a), 2(a)
calorimetry. DSC study has been carried out for the and 3(a), respectively. From the corresponding
similar heat treated samples to con"rm the trans- DSC study of the similar samples (Fig. 4(a)}(d)), it is
formation temperatures. found that the clear transformation leading to SME
takes place only after the sample is heat treated
above 3003C. In the samples, heat treated in the
2. Experimental temperature range 3003C}4803C, the intermediate
R-phase is observable. Only MA transformation
Near-equiatomic Nitinol wires of 2.4 mm dia are is observable in the samples heat treated above
used for the study of thermal expansion. The mater- 4803C. It is a well-accepted fact that heat treatment
ial is obtained from Special Metals Corporation, at higher temperature anneals out dislocations and
USA. The samples are heat-treated by annealing defects present in the cold-worked sample [10}12].
them for 20 min at di!erent temperatures in the At lower heat-treatment temperatures, due to the
range of 300}6003C and then quenched in water at presence of higher dislocation density, the inter-
room temperature. Thermo mechanical analyser mediate R-phase is found to exist [13,14]. Based on
(TMA-50, Shimadzu) is used for the measurement the above "ndings transformation behaviour has
 J. Uchil et al. / Physica B 270 (1999) 289}297 291

Fig. 1. (a) Thermal strain versus temperature and (b) thermal expansion coe$cient versus temperature of the sample heat treated at
4003C.
292 J. Uchil et al. / Physica B 270 (1999) 289}297

Fig. 2. (a) Thermal strain versus temperature and (b) thermal expansion coe$cient versus temperature of the sample heat treated at
4803C.
 J. Uchil et al. / Physica B 270 (1999) 289}297 293

Fig. 3. (a) Thermal strain versus temperature and (b) thermal expansion coe$cient versus temperature of the sample heat treated at
5003C.
294 J. Uchil et al. / Physica B 270 (1999) 289}297

Fig. 4. DSC thermograms for the samples heat treated at (a) 3003C, (b) 4003C, (c) 4803C and (d) 5003C.
 J. Uchil et al. / Physica B 270 (1999) 289}297 295

been studied using thermal expansion as a probe. tenitic and martensitic phases are, respectively,
Figs. 1(b), 2(b) and 3(b) represent the thermal ex- 11;10\ and 7;10\ K\. These values are in
pansion coe$cient (TEC) as a function of temper- close agreement with earlier reported values for
ature for samples heat treated at 4003C, 4803C and Nitinol [15].
5003C, respectively. In the presence of the intermediate R-phase for
In Fig. 3(a) corresponding to a sample heat the samples heat-treated below 4803C, DSC plots
treated at a high temperature of 5003C, the thermal show extra peaks on heating and cooling depend-
strain *¸/¸ is found to increase linearly with ing on the heat treatment condition. For the sample

temperature up to 76.63C and the curve is found to heat treated at 4003C, the DSC plot, in Fig. 4(b),
deviate from linearity and shows an abrupt de- shows peak splitting during heating indicating the
crease in the thermal strain until 92.33C. Above this existence of a mixture of phases in the sample to
temperature the trend reverses and a linear increase start with. The "rst peak represents the RPA
is found to take place until the end of the temper- transformation and the second one the MPA
ature range. In the cooling part of the cycle, starting transformation. But on cooling only one exotherm
from 1253C, the decrease is found to be perfectly peak is observed. This corresponds to an APR
linear until 59.53C and then the nonlinearity begins. transformation. Here APM transformation is not
With further decrease in temperature, the increase observed as it is below the room temperature. The
in thermal strain takes place up to 48.13C and transformation behaviour of a similar heat treated
linear decrease in thermal strain is found to take sample under the thermal expansion probe is
place up to 303C. The abrupt deviation in the shown in Fig. 1(a). To start with, while heating, the
linearity indicates the onset of phase transforma- thermal strain *¸/¸ increases linearly with tem-

tions. In the heating part, the average temperature perature up to 563C. After this temperature, the
at which the decrease in thermal strain takes place variation takes the shape of a cap up to 80.33C.
has been identi"ed as A and that at which the There after, a linear increase in thermal strain with

linear increase is regained as A . So also in the temperature takes place up to 1253C. As usual, the

cooling part of the cycle, the average temperature departure from linearity indicates the onset of
at which the increase in thermal strain begins is phase transformations. In the initial part of the cap
termed M and that at which it is terminated M . there is an abrupt increase in thermal strain and
 
The method followed in DSC thermograms [1] to then a linear part followed by an abrupt decrease in
"x transformation temperature is used in the pres- thermal strain. The temperature at which the in-
ent case. These transformation temperatures are crease begins is identi"ed as A and the point at

found to be in agreement with those obtained from which the linearity in the cap begins as A. So also,

the DSC thermogram for the sample heat treated at the temperature at which the decrease begins is
5003C. noted as A and the point at which the linear

In Fig. 3(b), a TEC is given as a function of increase begins as A . These transformation tem-

temperature as obtained from the data in Fig. 3(a). peratures are in close agreement with A, A and
 
When transformation occurs, a drastic change in A , A determined from the DSC thermogram of
 
TEC is found to take place. Initially in the heating Fig. 4(b). It is now clear that A corresponds to the

part, the #ve TEC is found to gradually decrease start temperature of RPA and A is the "nish

up to A followed by a sudden decrease leading to temperature of the RPA transformation. In the

!ve TEC. The trend is found to get reversed as the cooling part of the curve, the linear decrease in the
temperature approaches A . After A the #ve TEC thermal strain is found to be present up to R (533C)
  
remains fairly constant. While decreasing the tem- where as the sudden decrease in thermal strain
perature, the thermal expansion remains #ve and begins and ends at R (48.63C). R indicates the
 
fairly uniform, a drastic decrease takes place at beginning of the R-phase transformation and R the

M leading to !ve TEC. As M is approached the temperature at which this transformation is com-
 
trend gets reversed and TEC approaches a constant plete. After R thermal strain is found to be linear, it

#ve value after M . The TEC obtained at aus- is still found to be !ve indicating the absence of

296 J. Uchil et al. / Physica B 270 (1999) 289}297

a martensitic phase till room temperature. Thermal transformation takes place, there is a large increase
strain at the beginning of the heating cycle started in &b' and &c' combined with a small decrease in &a'.
from a higher value, presumably, due to the pres- Therefore, the overall change in any direction is an
ence of mixed states of &M' and &R', as the process of increase in the cubic cell edge parameters and con-
quenching took place below M . In the cooling part sequently a net increase in the dimension of the

the lowest temperature reaches well above M . sample. Because of the expansion during a decrease

Fig. 1(b) depicts the variation of TEC with tem- in temperature the !ve TEC results. On the other
perature in a complete thermal cycle. To begin hand, during MPA transformation there is an
with, TEC is #ve and remains fairly constant till overall decrease in the dimension with an increase
A and a surge in TEC is observed during the in temperature which results again in a !ve TEC.

RPA transformation. Furthermore, increasing the However, in APR transformation, the BCC trans-
temperature leads to a MPA transformation forms to a rhombohedral crystal structure resulting
coupled with a dip in the TEC. Beyond A the TEC in small changes in inter axial angles while su!ering

is found to be positive and fairly constant up to no change in unit cell edges [18]. Due to the distor-
1203C. On the other hand, in the cooling part of the tion in the BCC phase during the formation process
cycle TEC remains nearly constant up to R and of the rhombohedral phase, a resultant decrease in

later there is a surge in TEC. After the completion the size occurs in a given direction and hence a con-
of the APR transformation the TEC is found to be traction of the sample with a decrease in temper-
constant up to 303C. ature is observed leading to a #ve TEC. Similarly,
Heat treatment at 4803C is found to be a critical during RPA transformation, the net e!ect is an
point as far as the existence of R-phase in equi- increase in cell parameter with an increase in tem-
atomic Nitinol wire of 2.4 mm diameter is con- perature and therefore a #ve TEC occurs.
cerned. Beyond this temperature the sample is The characteristic transformation temperatures
found to be free from R-phase. The curves corre- obtained in the present study through the thermal
sponding to the sample heat treated at 4803C are expansion probe are compared with those obtained
given in Figs. 2(a) and (b). In Fig. 2(a), there is only from the standard method of DSC thermograms.
MPA transformation in the heating part of the The values are presented in Table 1 and it is found
cycle, whereas APR transformation is barely pres- that the transformation temperatures "xed through
ent followed by RPM transformation in the cool- the thermal expansion probe are in good agree-
ing part of the cycle. The variation of TEC given in ment. Repeated trials have shown that the thermal
Fig. 2(b) shows only a dip in TEC in the heating expansion probe using TMA is equally sensitive
part of the cycle without a prior surge indicating and reliable in identifying the phase transforma-
the presence of only MPA transformation. How- tions in Nitinol. In fact it is di$cult to "x A and

ever, in the cooling part of the cycle as R-phase A using DSC thermograms whereas they are dis-

transformation is approached there is a surge fol- tinctly identi"ed in the case of the thermal expan-
lowed by a steep dip indicating the onset of RPM sion probe. Even in the case of electrical resistivity
transformation. When this transformation is com- probe for determining transformation temperatures
plete the TEC becomes #ve and remains fairly the authors have found that it was not easy to
constant up to 303C. distinguish A from A when there is coexistance of
 
The observed !ve TEC associated with APM M- and R-phase at room temperature [7]. A TEC
transformation can be explained in the following of 11;10\ K\ obtained for the A-phase and
way. During martensitic transformation (APM) 7;10\ K\ for the M-phase are found to be
the BCC of parent austenitic phase transforms into independent of the heat treatment temperature.
a monoclinic crystal structure within a small tem- The pure intermediate state is found to be present
perature interval of 203C. The lattice parameters just above the room temperature in the sample heat
given in the literature for the BCC phase is treated at 4003C (Fig. 1(b)) and the observed TEC is
a "3.0 15 As [16] and for the martensitic phase, 3;10\ K\. But in the heating part of the

a"2.889 As , b"4.120 As , c"4.622 As [17]. When cycle when the R-phase is mixed with the M-phase,
 J. Uchil et al. / Physica B 270 (1999) 289}297 297

Table 1
Transition temperatures of the samples heat-treated at di!erent temperatures obtained by DSC and TMA

Heat treat A A M M R R
     
temperature (3C)
DSC TMA DSC TMA DSC TMA DSC TMA DSC TMA DSC TMA

400 } 67.4 78.5 80.3 } } } } 56.2 53 53.4 48.6

460 69 70 81.2 83.6 44.1 NC 35.1 36.6 55.4 55.5 51.6 NC
480 73 73.3 88.5 87 NC NC 43 41.5 55.1 54.9 NC NC
500 75.8 76.6 95 92.3 59.4 59.5 49.8 48.1 } } } }
600 81.7 80.7 99.5 98.6 66.7 66 52.5 51.6 } } } }

Note: NCP Not clear.

initially, the thermal expansion value is quite large References

and is around 15;10\ K\.
[1] J. Uchil, K.P. Mohanchandra, K.K. Mahesh, K. Ganesh
Kumara, Physica B 253 (1998) 83.
4. Conclusions [2] G.V. Lashkarev, S.M. Solonin, A.V. Brodovoi, I.F. Mar-
tynova, M.V. Radchenko, A.L. Mirets, N.V. Goncharuk,
Sov. Phys. State 34 (2) (1992) 352.
[3] Y. Liu, J. van Humbeek, J. Phys. IV. Collq. C 5 (1997) 519.
(1) TEC undergoes large variations whenever [4] G. Airoldi, D.A. Lodi, M. Pozzi, J. Phys. IV. Collq.
a phase transformation occurs in Nitinol. C 5 (1997) 507.
(2) MA transformation is associated with [5] G. Airoldi, A. Corsi, G. Riva, Mater. Sci. Eng. A 241 (1998)
233.
!ve TEC where as RA transformation is
[6] C.M. Wayman, I. Cornelies, Scripta Metall. 6 (1972) 115.
associated with #ve TEC. [7] J. Uchil, K.P. Mohanchandra, K. Ganesh Kumara,
(3) Thermal expansion probe using TMA is bet- K.K. Mahesh, Mater. Sci. Eng. A 251 (1998) 43.
ter suited to "x transformation temperatures [8] K. Mukherjee, S. Sircar, Narendra B. Dahotre, Mater. Sci.
whenever there is a mixed state. Eng. 74 (1985) 75.
[9] Y. Kuroiwa, T. Finlayson, F. Smith, Solid State Commun.
(4) TEC of martensitic and austenitic phases
106 (8) (1998) 501.
are independent of heat treatment [10] T. Hoshiya, F. Takad, Y. Ichihashi, Mater. Sci. Eng. A 130
temperatures. (1990) 186.
[11] M. Igharo, J.V. Wood, Mat. Sci. Eng. 98 (1988) 433.
[12] Z. Nishiyama, Martensitic Transformations, Academic
Acknowledgements Press, London, p.238}289.
[13] H. Matsumoto, J. Mater. Sci. Lett. 11 (1992) 367.
[14] T. Todoraki, H. Tamura, Trans. Jpn. Inst. Met. 28 (2)
The authors are grateful to Indian Space Re- (1987) 83.
search Organization, Department of Space, Gov- [15] William B. Cross, Anthony H. Kariotis, Fredrick J. Stim-
ernment of India for supporting this work through ler, NASA CR-1433 (1969) 3.
a research grant under their Respond Program. [16] T.V. Philip, P.A. Beck, Trans. Am. Inst. Min. Engrs. 209
(1957) 1269.
They are also thankful to Sri. M.N. Sathyanarayan, [17] K. Otsuka, T. Sawamura, K. Shimizu, Physica Stat. Sol.
Director, Space Mechanisms Group, ISRO Satel- 5 (1971) 457.
lite Center, Bangalore, for useful discussions and [18] T. Tamiya, D. Shindo, Y. Murakami, Y. Bando, K. Ot-
helpful suggestions. suka, Mater. Trans. JIM 39 (7) (1998) 714.