Materials Science and Engineering A 494 (2008) 429–435

Contents lists available at ScienceDirect

Materials Science and Engineering A

journal homepage: www.elsevier.com/locate/msea

Mechanical characterization of NiTi SMA wires using a dynamic

mechanical analyzer
Debdutta Roy a , V. Buravalla b , P.D. Mangalgiri b , S. Allegavi a , U. Ramamurty a,∗
a
Department of Materials Engineering, Indian Institute of Science, Bangalore 560012, India
b
India Science Lab, General Motors R&D, ITPB, Bangalore 560066, India

a r t i c l e i n f o a b s t r a c t

Article history: An investigation into the thermo-mechanical response of trained Ni–Ti shape memory alloy (SMA) wires
Received 6 March 2008 with two-way shape memory effect (TWSME) induced into them was conducted in order to examine the
Received in revised form 16 April 2008 influence of the cold work on quasi-static and fatigue behavior of SMAs. Use of dynamic mechanical ana-
Accepted 21 April 2008
lyzer (DMA) for such tests on thin wires was examined and successfully utilized. Quasi-static stress–strain
responses of the wires at different temperatures were obtained to determine the critical stresses for for-
Keywords:
ward as well as reverse transformations. Mechanical stress-controlled fatigue tests were conducted on
Shape memory alloys
the wires in the austenitic state (130 ◦ C). They showed a two-stage deterioration of fatigue life—a rapid
Fatigue life
Austenite
decay when the maximum stress of the fatigue cycle is higher than the critical transformation stress, and
Martensite a more gradual deterioration at stress levels considerably lower than the critical stress. The fatigue life of
Dynamic mechanical analyzer the wires was found to increase with the load ratio, R, whereas the frequency of cyclic loading had only
a marginal effect. Differential scanning calorimetry (DSC) of the fatigued austenite specimens indicated
the presence of residual stress-induced martensite in the austenitic state. The high concentration of the
austenite–martensite interfaces act as potential sites for stress-concentration and are the main source of
the drastic decrease of the fatigue life in the two-phase region.
© 2008 Elsevier B.V. All rights reserved.

1. Introduction transformation (upon unloading), resulting in a closed stress–strain

loop with relatively large hysterisis. Critical stresses character-
Shape memory alloys (SMAs) undergo a diffusionless marten- ize the forward/backward transformation and loading/unloading
sitic transformation, either upon cooling from the austenitic phase plateaus.
or upon the application of stress. As a result, SMAs exhibit two Considerable amount of research was conducted and reported
distinct mechanical responses depending on the temperature of on the mechanical response of NiTi plates, bars and wires to uni-
deformation. The shape memory effect (SME), which occurs when axial tensile loading, mechanical cycling (mostly pull–pull cycling),
the SMA is deformed in the martensitic condition, is a result thermal cycling, etc. and the underlying mechanisms. Systematic
of the detwinning (i.e., an energetically favored martensite vari- studies of fatigue in pseudo-elastic NiTi were made by Eggeler et
ant growing at the expense of the others) of the martensite. The al. [1,2], Miyazaki et al. [3], and Melton and Mercier [4]. Thermo-
super-elasticity or pseudo-elastic effect arises due to the stress- mechanical response of pseudo-elastic NiTi wires was studied by
induced martensite (SIM) formation when the SMA is deformed in Tsoi et al. [5]. All these studies focused on relating the observed
the austenitic state. Both these processes can be reversed, either phenomena to microstructural processes which are all based on the
by increasing the temperature or releasing the stress, respec- interaction of crystal dislocations with the martensitic transforma-
tively. tion. In spite of such concerted research efforts, understanding of
NiTi SMAs are often used in pseudo-elastic applications where the thermo-mechanical response of NiTi wires under cyclic loading
fatigue limits the service lives of components, especially when is still far from complete particularly for trained NiTi wires, which
SMAs are used to dissipate energy in systems which are sub- are finding increased application as micro-actuators.
jected to cyclic loading. Here, “pseudo-elasticity” refers to a To stabilize the shape memory behavior of an SMA, specific
stress-induced transformation (upon loading) and the reverse thermo-mechanical treatment is imparted to it. This process
induces microstructural modifications such as introduction of
defects/dislocations and texture into the material. These, in
∗ Corresponding author. Tel.: +91 80 2293 3241; fax: +91 80 2360 0472. turn, tend to favor certain orientation of the SIM. The defects
E-mail address: [email protected] (U. Ramamurty). induced during thermo-mechanical treatment also induce some

0921-5093/$ – see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.msea.2008.04.052
430 D. Roy et al. / Materials Science and Engineering A 494 (2008) 429–435

permanently deformed SIM that will not participate in further

transformation either during stress or thermally induced transfor-
mation. The macroscopic manifestation of these is in terms of a set
‘cold-shape’ or a two-way memory (in contrast to the ‘hot-shape’
or one-way memory). From an actuator perspective, this two-way
shape memory effect (TWSME) facilitates operating the device
between two set shapes and hence enhances repeatability. How-
ever, these defects due to cold working can adversely influence the
fatigue behavior of the wires.
In this paper, we examine this particular aspect on an actu-
ator wire that has experienced significant amount of cold work,
and thus exhibits considerable two-way effect. Hence, it does not
exhibit a classical pseudo-elastic behavior in the form of com-
plete closure of the stress–strain loops on loading and unloading.
The present work was initiated with the objectives of character-
izing the thermo-mechanical response of thin trained SMA wires
and to generate currently lacking test data to understand fatigue
and the degradation characteristics in trained NiTi SMA wires.
Further, in contrast to the general approach of using a standard
servo-hydraulic machine or a rotating bending apparatus for con-
ducting the fatigue tests, use of Dynamic Mechanical Analyzer
(DMA) for characterization of SMA thin wires for both quasi-static
and fatigue testing is explored and successfully utilized. The choice
of the DMA was, in particular, dictated by the extremely low loads
required for testing thin wires (0.380 mm diameter in this case)
and the necessity to carry out tests at super-ambient tempera-
tures.

2. Materials and experiments

Commercially available trained NiTi wires (with TWSME

induced into them) of 0.380 mm (15 mil) diameter were investi-
gated. The wires were used with three different pre-conditioning
treatments, viz.,
Fig. 1. (a and b) DSC scans during cooling and heating for heat-treated and as-
(1) As-received (AR) wires that are trained and have TWSME and received wires.
significant cold work in them.
(2) Wires heat-treated (HT) at 450 ◦ C for 30 min in vacuum-sealed
quartz tubes. This condition is used purely for comparison. stress and strain was set to the minimum possible value to ensure
(3) Wires that are thermally cycled (TC) by dipping the wire in hot negligible interference from the dynamic mode. Three tests were
oil (∼150 ◦ C) and held till thermal equilibrium is attained and performed at each temperature.
withdrawn to cool to room temperature. This is repeated a few Fatigue tests were conducted on the TC wires of 25 mm gage
times (5–10 times) so as to ensure removal of any handling length using the DMA at 130 ◦ C (austenitic state). No specific sur-
induced or other extraneous load history effects. face conditioning was performed. Tension–tension fatigue cycling
was conducted with a stress ratio, R (defined as the ratio of the
The stress-free transformation temperatures were obtained by minimum to maximum stresses of the fatigue cycle) of 0.1, 0.25,
differential scanning calorimetry (DSC) carried out on an Univer- and 0.5. The frequency was maintained at 10 Hz for most exper-
sal V3.0GTA Instruments DSC, with a heating and cooling rate of iments except for a set of experiments conducted at 1 Hz for
10 ◦ C/min. Comparison of DSC scans of the AR and HT wires is made studying the frequency effect on the S–N behavior of the wires. Each
in Fig. 1 and their transformation temperatures (As , Af , Ms , and specimen was cycled in stress-controlled mode at a fixed stress
Mf ) are tabulated in Table 1. DSC was also carried out on fatigued amplitude,  =  max −  min , where  max and  min are the maxi-
samples as described later. mum and the minimum stresses of the sine-wave fatigue cycle and
Quasi-static tensile tests both at room temperature as well as the stress–strain loops were monitored till fracture. Three to four
higher temperatures were conducted on 25 mm gage length spec- specimens were tested at each stress level.
imens using a Gabo Eplexor DMA (500 N capacity). The tests were One of the main challenges faced in fatigue testing a thin
conducted in load control mode and the dynamic component of the wire is gripping them. There are two major issues. (a) Allevia-
tion/minimization of the stress concentration associated with the
gripped ends. If this is not addressed, specimens tend to fail prema-
turely, leading to a lower estimation of fatigue life. (b) Tendency for
Table 1
Transformation temperatures (all in ◦ C), obtained using the DSC, of the as-received
slippage of the wires due to repeated loading and unloading, which
and heat-treated wires could lead to overestimation of the strain accumulation. After try-
ing out several different designs, grips shown schematically in Fig. 2
As Af Ms Mf
were found to give the best performance. They consist of two tool
As-received 74 78 48 36 steel plates with EDM cut grooves, the size of the groove being such
Heat-treated 84.6 92.6 55.7 47.7
that the wire fits into the groove completely.
 D. Roy et al. / Materials Science and Engineering A 494 (2008) 429–435 431

Fig. 3. Room temperature quasi-static tensile stress–strain response of the as-

received, heat-treated, and thermally cycled wires.

Fig. 2. A schematic of the grips used for testing the wires. All the dimensions are in
mm.

3. Results
Fig. 4. Quasi-static tensile stress–strain response of the thermally cycled wires at
3.1. DSC various temperatures.

Comparison of DSC scans of the AR and HT wires is made in 200 MPa. The two, however, have different plateau strains (owing
Fig. 1. Both the A → M and M → A transformation start at a higher to martensite variant reorientation), larger in the heat-treated wire
temperature for the HT wire as compared to the AR wire. However, (∼4.5%) in comparison to the thermally cycled wire (∼2.5%).
the HT wire shows a distinct R-phase transformation while cooling, Quasi-static stress–strain responses during loading and unload-
much in contrast to the AR wire wherein the R-phase transforma- ing of the TC wires at different temperatures above Af are shown in
tion is suppressed almost entirely. (Only a small hump in the curve Fig. 4. The curves depict ideal pseudo-elastic behavior with a dis-
corresponding to the R phase formation was seen.) tinct stress plateau during loading owing to SIM transformation.
Data obtained from these curves are summarized in Table 2. It is
3.2. Quasi-static tensile tests seen that the transformation stress increases with increasing tem-
perature whereas the transformation strain remains approximately
Quasi-static tensile stress–strain response of the trained (AR) constant at ∼7.5%.
wire at room temperature is shown in Fig. 3. It does not show a The critical stress for inducing forward transformation,  SIM ,
distinct plateau at a constant stress owing to martensite variant ori- as well as that for inverse transformation,  b , both increase with
entation (lack of detwinning). However, the stress–strain response increase in temperature as shown in Fig. 5. The stress–temperature
at room temperature of both the HT and TC wires (Fig. 3) show a relationship is approximately linear with d/dT of ∼7 MPa/◦ C, as
distinct plateau behavior with the detwinning stress,  DT , of around per the Clausius–Clapeyron relation that describes the thermo-

Table 2
Summary of the data obtained from the qausi-static tensile tests performed at various temperatures

Temperature (◦ C) Starting stress (MPa) End stress (MPa) Starting strain (%) Plateau strain range (%) UTS (MPa)

100 427 524 1.07 7.8 1268

110 509 581 1.87 7.76 1206
120 583 644 2.45 7.04 1196
130 683 740 2.3 7.48 1192
150 816 853 3.22 7.36 1088
432 D. Roy et al. / Materials Science and Engineering A 494 (2008) 429–435

Fig. 5. Variation of the stress-induced martensite start stress on loading and the
reverse transformation stress during unloading as a function of the testing temper-
ature. Fig. 7. Variation of elastic modulus of the as-received wire with temperature.

elastic martensitic transformation. Similar values of d/dT were 3.3. Fatigue

also obtained in the works of Stachowiak and McCormick [6] who
found it to be between 6.4 and 6.6 MPa/◦ C and Miyazaki et al. [7] Fig. 8 shows the maximum stress,  max vs. number of cycles to
who found it to be around 8.28 MPa/◦ C. The quasi-static results failure, Nf , for the wires in their austenitic state corresponding to
of this investigation were found to be in close accordance with R = 0.1, 0.25, and 0.5 at 10 Hz and 130 ◦ C. The fatigue life for the wires
the results of Tsoi et al. [5] who studied the thermo-mechanical increases with R, with the fatigue life enhancement being really sig-
characteristics of a series of NiTi and NiTiCu wires. nificant for R = 0.5. It is apparent from Fig. 8 that the S–N curve of the
While for the forward transformation, the strain in the plateau austenitic wire can be divided into two regimes—a steep decay in
regime remains constant at about 7.5% with increasing temper- the high stress regime ( max > 400 MPa) followed by a slower rate of
ature, the plateau corresponding to the inverse transformation degradation in the low stress regime ( max < 400 MPa). Miyazaki et
slowly disappears. Another interesting observation is the V-shaped al. [7] had also observed similar two stage degradation in εa (strain
evolution of the secant modulus with stress (Fig. 6). Instead amplitude) vs. Nf relationship in the rotary bending fatigue tests
of monotonically decreasing with strain, the modulus initially performed on Ti–50Ni and Ti–40Ni–10Cu (at.%) specimens. Simi-
decreases with the increase in strain but increases in the later lar kind of two-stage deterioration is also observed during thermal
stage. Similar V-shape in Young’s modulus vs. strain relationship cycling of trained NiTi wires with TWSME induced into them [9].
has been observed by Huang et al. [8] for NiTi polycrystalline wires The 106 fatigue limit of the austenitic wires is in close accordance
as well as CuAlNi single crystal. The variation of the secant modu- with the bending rotation fatigue data of NiTi wires obtained by
lus with temperature is shown in Fig. 7. It is seen that the elastic Eggeler et al. [1] for pull–pull cycling in strain controlled mode.
modulus goes through a minimum both during heating and cool- The dependence of fatigue life on frequency of loading is shown in
ing but the temperature at which this minimum occurs is different Fig. 9. As seen, reducing the frequency by an order of magnitude
during heating and cooling in accordance with the transformation decreases the fatigue life only marginally.
hysterisis observed in SMAs.
The DMA offers a specific advantage in that it allows monitoring 3.4. Post-mortem characterization
of the evolution of the modulus with stressing of the wires. It also
allows temperature sweep tests that can be used for measuring the DSC was performed on fatigue-fractured samples in order to
transformation temperatures of SMAs under stressed conditions, examine the changes in transformation behavior, if any, as a result
which is not possible with DSC. of the plastic strain accumulation during fatigue. The DSC scans

Fig. 8. S–N curves generated on the TC wires with different load ratios, R. Here,
for comparison purposes, fatigue life is plotted against the maximum stress of the
Fig. 6. Variation of elastic modulus of the as-received wire with stress at 130 ◦ C. fatigue cycle,  max .
 D. Roy et al. / Materials Science and Engineering A 494 (2008) 429–435 433

amount of remnant martensite due to fatigue. However, fatigue

does not bring about any discernible change in the extent of the
R-phase transformation which suggests that R-phase does not play
a significant role in the micro-mechanisms leading to permanent
damage accumulation during fatigue.

4. Discussion

4.1. Presence of residual martensite

The phenomenon of two-way memory is the result of disloca-

tion arrangements that guide the formation of martensite variants
to a preferred orientation, thereby resulting in an overall macro-
scopic strain during the thermally induced phase transformations.
Two-way shape memory effect has been shown to develop during
Fig. 9. S–N curves generated on TC wires obtained with two different frequencies.
training which involves thermo-mechanical cycling. Plastic defor-
mation occurs during the cycling [10] either due to the residual
obtained during cooling and heating cycles are shown in Fig. 10a stresses that accrue as a result of dislocations or due to residual per-
and b, respectively along with the scans of the as-received trained manent martensite accumulated during training. The dislocations
wires. While the endothermic peaks (during heating cycle) remain associated with the training phenomenon generate an internal
unaffected with fatigue, the exothermic peaks (during cooling stress state which inhibits the martensite from transforming into
cycle) appear to broaden with fatigue. This is to be expected austenite. This martensitic phase remains pinned in the micro-
since the stress-induced and plastically deformed martensite in the structure of the trained wires until the dislocations are removed
alloy will not participate in thermally induced transformation and through an annealing heat treatment. When the latent heat of
thus the enthalpy of transformation is significantly affected due transformation (obtained from the cooling curves, Fig. 1) of the
to fatigue. The broadening of the peaks can be attributed to the AR and HT wires are compared, the former exhibits smaller latent
early nucleation of martensite owing to the presence of significant heat. This may be attributed to less amount of material undergo-
ing the transformation in the trained wires [11], and would thus be
a possible evidence for the presence of residual martensite in the
microstructure of the trained wire.
Further indirect evidence of the presence of residual marten-
site is the difference in the extent of the plateau regime in the
HT and TC wires (Fig. 2). At room temperature, when the wire is
in the martensitic state, the plateau in the stress–strain curve is
due to detwinning and reorientation, i.e., one favorably oriented
martensite variant growing at the expense of the other. A possi-
ble explanation for this difference is that not all the martensite in
the TC wires participate in the variant reorientation process due to
remnant stress-induced martensite. A detailed TEM investigation
is required to confirm this.

4.2. Quasi-static tensile stress–strain response

The stress–strain response of the wires is a function of temper-

ature as is evident from Fig. 3. The individual stress–strain curves
evince the ideal pseudo-elastic behavior with initial elastic defor-
mation of austenite followed by stress-induced transformation of
austenite to martensite associated with a plateau over a certain
stress regime (Table 2) and then finally into the elastic and plastic
deformation of martensite. Here, the critical stress is not the stress
at which significant plastic flow by slip occurs, rather the stress
necessary to initiate stress-induced martensitic transformation of
the existing austenite under the action of the applied load. However,
Miyazaki et al. [7] showed that the elastic region prior to the onset of
stress-induced martensitic transformation generally exhibits two
deformation modes, i.e., pure elastic deformation along with micro-
scopic local stress-induced transformation giving rise to pockets of
martensite within the material before macroscopic stress-induced
transformation begins.
In general, the parent phase (austenite) is stable at temperatures
greater than Af (in this case, 93 ◦ C). Thus, with increase in tem-
perature above Af , the thermodynamic resistance to stress-induced
Fig. 10. DSC scans during (a) heating and (b) cooling of the as-received and fatigued
martensitic transformation increases and hence higher stresses are
wires. required for the transformation. So the critical stress for inducing
434 D. Roy et al. / Materials Science and Engineering A 494 (2008) 429–435

forward as well as reverse transformation increases with increase macroscopic critical stress of the wire at 130 ◦ C) is clear evidence
in temperature. of the formation of stress-induced martensite during these tests
The disappearance of the plateau corresponding to the reverse (Fig. 10a). Also, the pre-existing residual martensite present in the
(martensite to austenite) transformation with increase in temper- microstructure of the trained wire grows with continued cycling.
ature can be attributed to the introduction of dislocations into the The high concentration of the austenite–martensite interfaces act
matrix as the material is stressed to higher levels with increase in as potential sites for stress-concentration and are the main source
temperature. of the drastic decrease of the fatigue life in the two-phase region.
The rapid degradation rate when the wire is cycled at stresses
4.3. Variation of secant modulus higher than the critical stress of the material is in agreement with
the observations of Tabanli et al. [16] who found that the fatigue
The Young’s modulus of SMAs is phase dependant, with the life of NiTi was shorter in the super-elastic plateau regime, which
modulus of the austenite and martensite differing as much as a few they suggest is a consequence of the coexistence of austenite and
times in some SMAs [12]. The V-shaped variation of secant modulus martensite. The slower degradation of the wires cycled at maxi-
(instantaneous slope of the stress–strain curve) with strain (Fig. 6) mum stresses considerably lower than the critical stress (so that
indicates that the SMA initially becomes more complaint before microscopic stress-induced transformation during cycling can be
regaining the stiffness. This can be explained as following. The ruled out) can be explained on the basis that the population of
stress-induced phase transformation from austenite to martensite austenite–martensite interfaces is much lower. In this case, only the
in polycrystalline SMAs is not instantaneous, but occurs gradu- stationary interfaces between austenite and pre-existing residual
ally, i.e., initial phase transformation from austenite to martensite martensite in the trained wire exist as compared to the wires cycled
followed by reorientation of the martensite variants so formed. at high maximum stresses where stress-induced transformation
The stiffening in the later stage is the result of growth of partic- does take place. Thus the magnitude and extent of permanent dam-
ular martensite variants favorable to the applied stress state at age accumulation will be lower and hence higher fatigue life is seen
the expense of the others. The formation and dissolution of inter- for these conditions of cycling.
faces of various variants and their relative motion influences the The improvement in fatigue life with increase in R can be
stress–strain behavior during transformation. The DMA was found explained on the basis that for the same maximum stress, the wire,
to be particularly suited to examine modulus variations as it pro- which has been fatigued at a lower R, has to go through a wider
vides information to obtain the modulus at every load increment stress range in an entire fatigue cycle and hence the extent of micro-
during the quasi-static tensile test. scopic deformation accumulated in it in a cycle is much larger than
the case of a wire fatigued at a higher R.
4.4. Fatigue life The marginal improvement of fatigue life with increase in fre-
quency is possibly because of the higher critical stress (due to
As mentioned earlier, the TWSME generated by training is basi- increased temperature as a result of exothermic SIM transforma-
cally a consequence of a dislocation structure generated during the tion) and hardening rate at high strain rates, which decreases the
training cycle, which provides local stress and favors the forma- transformation and associated microstructural effects that influ-
tion of a preferred orientation to a net shape variation [10]. The ence fatigue degradation.
oriented residual stresses associated with dislocation arrangement
during training are relaxed by the nucleation and growth of cer- 5. Summary
tain preferential martensite variants. The presence of dislocations
does not allow complete martensite–austenite transformation dur- The thermo-mechanical response of thin trained NiTi wires was
ing unloading or heating. Hence, with cycling, there is a progressive examined employing a DMA. The quasi-static tensile stress–strain
increment of residual martensite, leading to permanent damage response of the trained wires with pre-existing residual marten-
accumulation and a deterioration of pseudo-elasticity [13–15]. The site is different from annealed wires in a sense that in the trained
trained wires used in this study therefore had some residual per- wires, owing to the presence of stabilized martensite a priori, there
manent deformation associated with them a priori in the form of is no closure of the super-elastic loop (at temperatures above Af )
dislocations as well as certain fraction of stabilized martensite vari- in contrast to annealed wires where complete closure of the super-
ants. elastic loop is observed. As a result, the extent of the plateau regime
Mechanical cycling of the trained wires at high stresses (owing to martensite re-orientation in case of the trained wires in
( max > 400 MPa) would lead to the generation of dislocations their martensitic state and stress-induced transformation in their
within the wires and dislocation mediated plastic deformation austenitic counterparts) is less in the trained wires compared to
would change the internal stress-fields associated with the trained the annealed ones. An interesting feature, which came up along-
wires. The stress-fields associated with the newly formulated dis- side the stress–strain response of the wires, with the aid of the
locations would possibly aid the nucleation of variants different dynamic mechanical analyzer, was the V-shaped variation of the
from those formed during training cycles and hence are unfavorable secant modulus with strain.
for TWSME. The annihilation and rearrangement of the disloca- Fatigue loading of the wires in their austenitic state revealed a
tion structure impedes the formation of the variants of martensite two-stage deterioration of fatigue life with a rather rapid decay at
responsible for the two-way effect, and with continued cycling, for- maximum stresses higher than the critical stress of the material
mation and stabilization of other martensite variants occurs leading which slows down to a more gradual deterioration at maxi-
to permanent damage accumulation and final failure of the wires. mum stress levels considerably lower than the critical stress for
Although 400 MPa is lower than the critical stress of the material at inducing martensitic transformation. A possible explanation is the
130 ◦ C (around 600 MPa), as mentioned earlier, pockets of marten- population of austenite–martensite interfaces which act as stress
site start forming in the material as a consequence of microscopic concentrators and dislocation mediated plastic deformation which
stress-induced transformation prior to macroscopic transforma- interferes with the two-way memory effect and leads to the func-
tion. The broadening of the peaks and the decrease in latent heat of tional fatigue of the wires. Other interesting feature which came
transformation corresponding to the cooling curves for the wires up with the fatigue test is the improvement of fatigue life with
cycled at a maximum stress of 450 and 550 MPa (lower than the increase in the stress ratio. The frequency of loading had only a
 D. Roy et al. / Materials Science and Engineering A 494 (2008) 429–435 435

marginal effect—10 Hz cycling showed a slightly longer life than References

the 1 Hz loading.
The present attempt to use DMA shows that it is a good tool [1] G. Eggeler, E. Hornbogen, A. Yawny, A. Heckmann, M. Wagner, Mater. Sci. Eng.
A 378 (2004) 24.
to characterize thermo-mechanical behavior of thin SMA wires. [2] M. Wagner, T. Sawaguchi, G. Kaustrater, D. Hoffken, G. Eggeler, Mater. Sci. Eng.
It offers several advantages over conventional fatigue testing that A 378 (2004) 105.
typically involves the use of servo-hydraulic testing machines. [3] S. Miyazaki, T. Imai, I. Igo, K. Otsuka, Metall. Trans. A 17 (1986) 115.
[4] K. Melton, O. Mercier, Acta Metall. 27 (1979) 137.
Instrumented indentation is used extensively to characterize small [5] K.A. Tsoi, J. Schrooten, R. Stalmans, Mater. Sci. Eng. A 368 (2004) 286.
volume structures as well as smart materials [17–20]. However, [6] G.B. Stachowiak, P.G. McCormick, Acta Metall. 36 (1988) 291.
it suffers from the difficulty of interpretation as the stress state [7] S. Miyazaki, K. Mizukoshi, T. Ueki, T. Sakuma, Y. Liu, Mater. Sci. Eng. A 273 (1999)
658.
underneath the indenter is quite complex. Interpretation of the [8] W.M. Huang, J.A. Wu, B.Y. Lim, I.E. Vahhi, J. Alloys Compd. 390 (2005) 175.
results obtained using DMA where samples are tested either in sim- [9] A.G. Mayer, H. Scherngell, A.C. Kneissl, Met. Mater. Trans. A 44 (2004)
ple tension or compression, on the other hand, is straight forward. 432.
[10] P. Filip, K. Mazanec, Scr. Mater. 35 (1996) 349.
Additional advantages include precision in loads (especially at very
[11] D.A. Miller, D.C. Lagoudas, Mater. Sci. Eng. A 308 (2001) 161.
low loads required for thin wires) and temperatures and controlled [12] W.M. Huang, Scr. Mater. 50 (2004) 353.
dynamic loading capability. In addition, DMA allows continuous [13] R. Vaidyanathan, D.C. Dunand, U. Ramamurty, Mater. Sci. Eng. A 289 (2000)
monitoring of secant modulus with cycling. Future directions for 208.
[14] N.Nayan, D.Roy, V.Buravalla, U.Ramamurty Mater. Sc. Eng. A, submitted(2008).
research could be examination of fatigue of martensitic wires and [15] M.F.X. Wagner, N. Nayan, U. Ramamurty, unpublished work.
the effect of remnant cold work on SMAs. [16] R.M. Tabonli, N.K. Simha, B.T. Berg, Mater. Sci. Eng. A 273–275 (1999)
644.
[17] S. Sridhar, A.E. Giannakopoulos, S. Suresh, U. Ramamurty, J. Appl. Phys. 85 (1999)
Acknowledgements 380.
[18] U. Ramamurty, S. Sridhar, A.E. Giannakopoulos, S. Suresh, Acta Mater. 47 (1999)
UR acknowledges the GM R&D, India Science Lab for sponsoring 2417.
[19] J. Basu, N. Nagendra, Y. Li, U. Ramamurty, Philos. Mag. 83 (2003) 1747.
this research at IISc. We are grateful to Mr. S. Sasidhara Pandit and [20] U. Ramamurty, M.C. Kumaran, Acta Mater. 52 (2004) 181.
Mr. N. Suresh for their assistance with the experiments reported in
this paper.