Part I. Thermomechanical characteristics of shape memory alloys

Materials Science and Engineering A368 (2004) 286–298 Part I. Thermomechanical characteristics of shape memory alloys Kelly A. Tsoi a,∗ , Jan Schrooten a , Rudy Stalmans b a Departement MTM, KULeuven, Kasteelpark Arenberg 44, Leuven B-3001, Belgium b FLEXMET,Rillaarsebaan 233, Aarschot B-3200, Belgium Received 11 August 2003; received in revised form 4 November 2003 Abstract Shape memory alloys (SMAs) are a group of alloys that exhibit a phenomenon known as the shape memory effect, (SME). This effect gives the alloys the ability to “recover” their original shape by heating above a certain transition temperature. There is also a large recovery strain, of up to 8%, associated with the transition. Because of this unique property, a large research effort is currently being undertaken, directed towards the use of SMAs in the actuation of smart structures for shape control, vibration control and for damage mitigation. SMAs also have a very high damping capacity due to a superelastic effect. This property of SMAs is extremely useful in vibration damping as well as reducing impact damage in structures. As such there has been much interest in using SMA-composites in structures. With the possibility of using SMA-composites in real structures such as in aviation, high speed transport industry and the automotive industry, there is increasing demands on knowing how the composites will react under everyday conditions. This paper details an investigation into the thermomechanical behaviour of SMA wires, looking at the recovery stresses produced and the stress and strain behaviour with respect to temperature, as well as changes in resistance of the wires with pre-strain. © 2003 Elsevier B.V. All rights reserved. Keywords: Shape memory alloys; Thermomechanical behaviour 1. Introduction The thermomechanical behaviour of shape memory alloys (SMAs) is not a well understood topic. Van Humbeeck and Stalmans [1] have shown that there is a lack of understanding on how the generation of stresses in SMAs is related to the transformational behaviour. This is due in part to the lack of experimental data available on the subject. Several papers on the subject have made some investigation into experimental parameters [2–5], more references are given in [1]. However, these tend to be based on the first heating cycle and not on what happens during repeated cycling. As was detailed in Van Humbeeck and Stalmans, [6], many of the previous experimental studies into thermomechanical behaviour are based on results taken from “hard tensile machines” which do not take into account thermal expansion effects of the cross heads when the wires are heated and cooled. It has ∗ Corresponding author. Present address: DSTO, 506 Lorimer St., Fishermens Bend, Vic. 3207, Australia. E-mail address: [email protected] (K.A. Tsoi). 0921-5093/$ – see front matter © 2003 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2003.11.006 also been shown that when a SMA wire is heated repeatedly, there is a change in the thermomechanical paths experienced by the wire [7]. This data is of great importance when considering the use of SMAs in engineering applications. Many uses of SMAs have been developed, ranging from biomedical, [8–10], to space structures [11]. There has also been growing interest in the possibility of embedding SMA wire elements into a composite matrix [12] in order to alter the vibration frequency of structures [13–18], as well as for shape control of structural elements [19]. Of particular importance for these applications is an understanding of the generation of recovery stresses with respect to temperature, as well as the stress and strain characteristics of the SMAs. Other important parameters to consider are how the wires are to be activated. They can be activated using temperature, via resistively heating the wires. For applications where the wires need to be continuously actuated on and off, the knowledge of whether the wires will fatigue and how they behave under cyclic conditions should be addressed. The aim of this paper is to, primarily, get an overall picture of the basic thermomechanical behaviour of SMA K.A. Tsoi et al. / Materials Science and Engineering A368 (2004) 286–298 wires. Secondly it is to determine what sort of wires are best suited to be embedded into a constraining matrix. Extensive thermomechanical tests on a series of different bare wires were performed and the results are shown in this paper. The stress versus strain behaviour and the effect of temperature on both stress and strain were also investigated along with changes in the resistance of the wires with pre-strain. Interest in the development of recovery stresses within constrained wires is of particular importance if they are to be embedded in a composite material due to the constraining nature of the matrix material. 2. Thermomechanical testing machine The equipment used during this investigation was a fully computerised specially designed tensile testing machine. The machine is capable of accurately measuring a combination of the influence of temperature, stress and strain on the SMA wires and SMA-composites. The specimens are mounted in the grips which are connected to linear variable differential transformers (LVDTs) via quartz rods, to reduce the thermal expansion during heating to a precision of 2 ␮m. As such, very accurate measurements of the stress and strain of the specimen can be measured. The specimen is fully immersed in a silicon oil bath which can be heated and cooled at a rate of 0.5 ◦ C/s with a maximum temperature of 200 ◦ C and a minimum of 14 ◦ C. Further information about the testing machine can be found in Stalmans et al. [20]. 3. Materials used A series of NiTi and NiTiCu SMA wires were investigated. Table 1 gives a description of the wires used. The transformation temperatures and heats of the as-received wires were measured using a differential scanning calorimeter (DSC) and the results are shown in Tables 2 and 3, respectively. The wires were cut to approximately 100 mm long and the ends had a copper slip crimped in place, in order to clamp the wires in the machine grips. 287 4. Thermomechanical behaviour of SMA wires 4.1. Experimental techniques Several load cases were used in order to investigate the different characteristics of the wires. 4.1.1. Stress versus strain Stress–strain experiments were conducted in order to obtain basic information about the wires used in this investigation. The first experiment completed was a tensile test of the SMA wires where they were held at a constant temperature, for a range of temperatures, to determine the stress–strain properties of the wires. A gauge length of 100 mm was used. The thermomechanical experiment consisted of (1) after loading the specimen, applying a small electric current of 0.3 A through the specimen and straining the specimen up to a load of 50 g in order to determine the reference strain (0%), (2) heating the wire from room temperature to the test temperature, T , (3) loading the specimen with a displacement rate of 0.01 mm/s to a strain value of 4%, (4) unloading the specimen back to 0% strain and (5) loading the specimen again until failure. This was completed for temperatures of 25 ◦ C, Af +20 ◦ C and 140 ◦ C. 4.1.2. Strain versus temperature A second series of experiments was performed to investigate the temperature cycling behaviour of the wires at a constant stress. A SMA wire specimen of gauge length 100 mm was placed in the testing machine and the thermomechanical loading consisted of (1) initial heating of the specimen to 130 ◦ C with as small a load as possible on the wire, (2) loading the wire to a constant stress, σ, at 130 ◦ C, (3) cooling the wire to 35 ◦ C with a stress, σ, still applied to the wire, (4) heating to 130 ◦ C with a stress, σ, Table 1 Table showing the wires used in this investigation and a description of each (at room temperature) Wire type Descriptiona NiTiCu-m1 NiTiCu-m2 NiTiCu, Memry, 150 ␮m diameter, martensite, alloy K wire, 35% cold worked and straight annealed, NiTi12 wt.% Cu NiTiCu, Thomas Bolton (Furukawa), 150 ␮m diameter, martensite, NT-H8, straight annealed, 47.17 wt.% Ni, 10.58 wt.% Cu, balance is Ti NiTi, Thomas Bolton (Furukawa), 150 ␮m diameter, R-phase, NT-M2, straight annealed, 55.1 wt.% Ni, balance is Ti NiTiCr, Thomas Bolton (Furukawa), 150 ␮m diameter, austenite, straight annealed, 55.65 wt.% Ni, 0.20 wt.% Cr, balance is Ti NiTi, SMA-INC, 150 ␮m diameter, austenite, alloy H, straight annealed, oxide surface, 55.65 wt.% Ni, 44.09 wt.% Ti NiTi, SMA-INC, 150 ␮m diameter, martensite, alloy H, straight annealed, oxide surface, 54.36 wt.% Ni, balance Ti NiTi-rp NiTi-a1 NiTi-a2 NiTi-m a The description of the wires includes (if available) the supplier, diameter, SMA-state, heat treatment, surface finish and composition, respectively. 288 K.A. Tsoi et al. / Materials Science and Engineering A368 (2004) 286–298 Table 2 Table of transformation temperatures (◦ C) of the as-received wires as measured by a differential scanning calorimeter. Ms , Mf , Mp , are the martensitic start, finish and peak, temperatures, As , Af , Ap are the austenitic start, finish and peak temperatures and Rs , Rf , Rp are the R-phase start, finish and peak temperatures, respectively Wire type NiTiCu-m1 NiTiCu-m2 NiTi-rp NiTi-a1 NiTi-a2 NiTi-m Ms 46.8 39.1 −8.3 – – 39.8 Mp Mf As Ap Af Rs Rp 43.1 32.7 −27 – – 38.2 38.3 27.1 −51 – – 29.7 55.6 42.7 32 – – 74.4 60.7 48.3 50.9 – – 77.9 64.2 54.1 56.3 22.7 – 81.7 – – 53.7 – 19.0 61.2 – – 49 – – 60.1 Rf – – 41 – −2.2 58 where “–” is noted, the transformation temperature was unable to be measured by the DSC, i.e. no signal of the transformation temperature is obtainable for reasons yet to be determined. Table 3 Table of transformation heats (J/g) of the as-received wires as measured by a differential scanning calorimeter. Hm , Ha , Hr , are the martensitic, austenitic and R-phase transformation heats, respectively Wire type Hm Ha Hr NiTiCu-m1 NiTiCu-m2 NiTi-rp NiTi-a1 NiTi-a2 NiTi-m 15 13 – – – 14.0 14.4 12.5 – – – 23.1 na na 5.5 na 2.8 7.2 where “–” is noted, the transformation temperature was unable to be measured by the DSC, i.e. no signal of the transformation temperature is obtainable for reasons yet to be known, where “na” is stated, it is not applicable, since this wire does not show an R-phase transformation. 4.1.3. Resistance versus strain The third set of experiments involved looking at the change in electrical resistance with strain. The wires were tested at a constant temperature of 25 ◦ C, Af + 20 ◦ C and 140 ◦ C and the resistance was measured during the loading cycle. The loading consisted of (1) heating the wire to one of the specified temperatures, (2) loading the wire with a displacement rate of 0.5 mm/s to a strain of 4%, (3) unloading the specimen to 0% strain and (4) loading the specimen again until failure. (5) cooling to 35 ◦ C with a stress, σ, and (6) unloading at 35 ◦ C. 4.1.4. Stress versus temperature Two types of tests were completed to determine the stress versus temperature behaviour of the wires: (i) using direct resistive heating and (ii) heating of the wires in an oil bath. Fig. 2 shows the resistive cycling of a NiTiCu-m1 wire. Steps 3–5 were repeated 23 times. A constant stress, σ, of, respectively, 57, 141, 283, 420 or 566 MPa was used. Fig. 1 shows the complete cycle for a NiTiCu-m1 wire. The numbers shown correspond to the steps 1–6 described above. The wire was loaded with a constant stress of 283 MPa. (1) Initially (point A) the specimen was heated in an oil bath, to a temperature of 110 ◦ C, in order to establish the 0% strain level. (2) The wire was then cooled to room temperature and a pre-strain of 1, 3 or 6% was applied to the wire (point B in Fig. 2). Fig. 1. Strain vs. temperature behaviour of NiTiCu-m1 showing the complete cycle. A constant stress of 283 MPa was applied to this wire. Fig. 2. Thermal cycle of a NiTiCu-m1 wire pre-strained to 3% and heated to 140 ◦ C and then heated resistively with a current of 700 mA. K.A. Tsoi et al. / Materials Science and Engineering A368 (2004) 286–298 (3) The wire was again heated, in an oil bath, in the constrained condition to 140 ◦ C (trajectory C) and then allowed to cool to room temperature (trajectory D). (4) The wire was then resistively heated with a current of 700 mA which was switched on for 10 s and off for 10 s (point E). This cycle was continued for 1500 cycles in order to determine the stability of the recovery stress generation of the wires under thermal cycling. For the NiTiCu-m1 wire, cyclic heating at currents of 600, 700, 800 and 900 mA were also completed in order to determine the effect of resistive heating on the stability of the wires for varying currents. A comparison of the different wire types and the different percent pre-strains was investigated. SMA wires were also placed in the thermomechanical testing machine and the wires were heated in the oil bath to 110 ◦ C in order to establish the 0% strain level. The wire was subsequently cooled to room temperature and a pre-strain (of 1, 3 or 6%) was applied. The wire was then heated to 140 ◦ C and cooled to room temperature, in order to simulate a composite curing cycle, and was followed by cycling between 110 ◦ C and room temperature using the oil bath, to determine the types of recovery stresses obtainable. Information that was obtained from these tests include a comparison of the stress versus temperature behaviour between different wires as well as between different pre-strains for the NiTiCu-m1 wire. 5. Results 5.1. Stress versus strain Fig. 3 shows the stress versus strain behaviour of the different SMA wires at different temperatures. From these charts large variations in the stress–strain behaviour of similar wires, can be seen, highlighting the variations which can occur in wires of similar composition but varying manufacturer. Table 4 shows the Young’s modulus, in the elastic region, for the different wires at different temperatures (25 ◦ C, Af + 20 ◦ C and 140 ◦ C). The stress–strain charts also show the maximum plateau strain values for each of the wires. This maximum plateau strain is dependent on the temperature. It shows that with increasing temperature, the maximum plateau strain value also increases. Table 4 Comparison of Young’s modulus values for SMA wires at different temperatures Wire type 25 ◦ C (GPa) Af + 20 and 140 ◦ C (GPa) NiTiCu-m1 NiTiCu-m2 NiTi-rp NiTi-a1 NiTi-a2 NiTi-m 9.4 4.1 8.3 42.1 47.1 15.1 62.7 54.2 63.9 64.4 72.1 54.1 289 The changes in the maximum plateau strain with increasing temperature are due to the different phases of the material. At low temperature the SMA is in its “soft” martensitic phase. Thus, it is easy to deform. At 25 ◦ C recoverable deformations ranging from 4% (Fig. 3(b); NiTiCu-m2) to about 10% (Fig. 3(c), (e), and (f); Niti-rp, NiTi-a2 and NiTi-m, respectively) are observed. As the temperature increases above Af , the structure becomes austenitic and stiffer, therefore more difficult to deform, requiring a greater stress to obtain the same strain values as was found for the low temperature case. The shape of the unloading curve seen in Fig. 3(a–c) and (f) for the 25 ◦ C curves at around 4.5% is an indication of the reorientation of martensitic plates during loading, [21]. These unload to a non-zero strain, however these strains are recoverable on heating. For Fig. 3(d) and (e) the unloading curves show different behaviour. This occurs since these wires are in an austenitic condition at 25 ◦ C and shows evidence of stress induced martensite (SIM) behaviour, [22]. Similarly the curves for the Af + 20 ◦ C curves (except for Fig. 3(f)) are also indications of the SIM transformation. The increase in stress at around 4% strain in the unloading–reloading cycle is evidence of deformation resulting from stress–strain cycling. This is similar to that found by Miyazaki et al. [23]. This is due to slip deformation which, in turn, increases the stress required to reach a certain strain value. At 140 ◦ C all wires are in their austenitic state, however they do not exhibit SIM behaviour during unloading– reloading since the temperature of the wire is above Md , the temperature below which perfect superelasticity can occur, thus, the strains are not completely recoverable. Fig. 3(f) shows a similar unloading curve since Af +20 ◦ C for this NiTi wire is close to the Md temperature of the wire. 5.2. Strain versus temperature Fig. 1 shows the complete cycle the wires were subjected to. Fig. 4 shows the strain versus temperature behaviour for the different wire types, showing only the second cycle for clarity. From these charts the changes in hysteresis for each wire at a different constant stress can be observed, and with increasing stress, the strain also increases. From these, apart from a few exceptions, there are nearly no changes in the hysteresis with differing stress value. In Fig. 4(c), the size of the 420 MPa curve can be attributed to the fact that when a stress greater than 355 MPa is applied to an R-phase alloy, the alloy undergoes a B2–B19 transformation and does not exhibit an R-phase transformation anymore. The R-phase transformation is dependent on the applied stress as determined by Stachowiak and McCormick, [24]. Fig. 4(d) shows a decrease in the size of the hysteresis for 283 MPa stress which can be attributed to overdeformation of the wire. From Fig. 3(f), the 25 ◦ C curve shows signs of plastic deformation at around 150 MPa, thus, when the wire in the strain–temperature experiment is cooled, any stress 290 K.A. Tsoi et al. / Materials Science and Engineering A368 (2004) 286–298 Fig. 3. Stress vs. strain behaviour of (a) NiTiCu-m1, (b) NiTiCu-m2, (c) NiTi-rp, (d) NiTi-a1, (e) NiTi-a2 and (f) NiTi-m at different temperatures as shown. Af stands for the austenitic finish temperature. NB: no results were obtained at 140 ◦ C for (c). above 150 MPa on that wire will cause overdeformation. This is also the reason that tests greater than 283 MPa are not included in Fig. 4(d). A small hysteresis in SMAs, compared to a large hysteresis, is preferable for control purposes. Small temperature differences, during heating and cooling, enable faster transformations of the wire. The charts show that at Af for each alloy there is a large drop in the strain. This is indicative of a transformation. When the magnitude of the constant applied stress is K.A. Tsoi et al. / Materials Science and Engineering A368 (2004) 286–298 291 Fig. 4. Strain vs. temperature behaviour of (a) NiTiCu-m1, (b) NiTiCu-m2, (c) NiTi-rp and (d) NiTi-m wires under constant loads of 57, 141, 283, 420 and 566 MPa, as shown. Only the second thermal cycle is shown for clarity. Fig. 5. Strain vs. temperature behaviour of different wires at a constant load of (a) 141 MPa and (b) 283 MPa. 292 K.A. Tsoi et al. / Materials Science and Engineering A368 (2004) 286–298 increased, the temperature at which the alloy begins to transform also increases. This is in agreement with the literature, [22,25], where the increase of constraining stress produces a shift in the transformation temperatures to higher temperatures and can be described by the Clausius–Clapeyron equation. A comparison of the hysteresis between the different wires is shown in Fig. 5. From Fig. 5(a) the wire with the greatest hysteresis is the NiTi-m and the smallest is the R-phase NiTi-rp. Fig. 5(b) shows that at 283 MPa, NiTi-rp has the smallest hysteresis with NiTi-m and NiTiCu-m1 also having a small hysteresis. The smaller the hysteresis Fig. 6. Resistance vs. strain behaviour of (a) NiTiCu-m1, (b) NiTiCu-m2, (c) NiTi-rp, (d) NiTi-a1, (e) NiTi-a2 and (f) NiTi-m wires at temperatures of 25 ◦ C, Af + 20 ◦ C and 140 ◦ C as shown. K.A. Tsoi et al. / Materials Science and Engineering A368 (2004) 286–298 the better for embedment into composites. A small temperature hysteresis provides better control over the recovery stresses and the temperature range in which they can be activated. Fig. 1 shows the effect of thermal cycling on the strain–temperature behaviour of the NiTiCu-m1 wire with a constant load of 283 MPa. As the number of cycles increases, the strain at 130 ◦ C also increases starting from 0.3% and reaching a saturation level of 2.3%. The hysteresis also decreases with increasing cycles from an initial 13 ◦ C and stabilises at around 40 cycles at 7 ◦ C. As the wire is continually cycled, defects are produced during the transformations and these defects will reach a saturation level at which point the transformation becomes stable [6]. Vokoun and Stalmans, [7], and Friend, [26], have also found similar behaviour in other SMA materials. 5.3. Resistance versus strain 293 5.4. Stress versus temperature Fig. 2 shows a full thermal cycle of a NiTiCu-m1 wire. Fig. 7 shows the first (a) and second (b) heating cycles for NiTiCu-m1. The wires are pre-strained to 1, 3 and 6% in order to observe the difference in the recovery stress generation. The rate of recovery stress build up changes with pre-strain value. The higher the pre-strain value, the slower the recovery stress build up is. Figs. 8 and 9 show the first and second temperature cycles, respectively, of the stress versus temperature behaviour for four other wires and it is shown that similar results to the NiTiCu-m1 wire are obtained. This provides a valuable insight into the transformation behaviour of these SMAs. The first temperature cycle is considered because SMA wires embedded into composites will undergo an initial curing cycle, and this first cycle gives an understanding of the transformation behaviour that the wires will experience during curing. The second (and The resistance of a SMA wire has been related to the transformation of the wire, [22], and has been used as an indication of the damage state of the wire [27]. Fig. 6 shows the resistance versus strain behaviour for different SMA wires at different temperatures as explained in Section 4.1.3. The charts show that for some of the wires there is a large temperature dependence of the resistance. As the pre-strain increases, the martensitic transformation proceeds and the resistance increases. This is basically related to the higher resistance of the martensitic phase. Generally, there will be a change in the resistance of a wire with change in length, and the slope of the resistance versus strain curve is related to the type of martensitic transformation that is occurring. For example, the electric resistance of preferentially oriented martensite will be different to that of self-accommodating martensite [5]. Hence, by embedding SMA wires into a composite/structure it may also be possible to determine the transformation level of the wires by measuring their resistance. From the charts it is possible to determine the pre-strain for any resistance value. There is also a consistent rate of increase in resistance with strain. This is indicated by the initial straining/unstraining cycle followed by straining to fracture. Table 5 shows the resistance–strain rate for each of the wires at the different temperatures. Table 5 Table of resistance–strain rate, dR/dǫ ( ) Wire type 25 ◦ C Af + 20 ◦ C 140 ◦ C NiTiCu-m1 NiTiCu-m2 NiTi-rp NiTi-a1 NiTi-a2 NiTi-m 5.82 5.09 0.8 2.31 3.04 1.40 3.33 5.39 2.44 3.17 3.12 2.11 2.40 3.98 – 2.88 2.78 1.69 Fig. 7. Stress vs. temperature for NiTiCu-m1 wires pre-strained to 1, 3, and 6% as shown: (a) first heating and cooling cycle (the starting point of the heating cycle and direction of heating for each curve is indicated by the arrow) and (b) second heating and cooling cycle, up to a temperature of 110 ◦ C. 294 K.A. Tsoi et al. / Materials Science and Engineering A368 (2004) 286–298 Fig. 8. Stress vs. temperature for (a) NiTiCu-m2, (b) NiTi-rp, (c) NiTi-a1, and (d) NiTi-a2 wires pre-strained to 1, 3 and 6% as shown for the first heating and cooling cycle. following) cycles are considered in order to understand the cyclic behaviour of recovery stress generation. In general, it is this cyclic behaviour which is of particular interest for applications of SMA wires and SMA-composites. During the initial heating cycle for high pre-strains (6%), the recovery stresses start to build up instantly when the wires are heated, whereas for the 1 and 3% pre-strains the wires only start recovery when the As temperature is reached. It has been suggested, [28], that this effect is due to the different types of transformations that occur within the SMA. It has been shown, [29–31], that an alloy of NiTi12 wt.% Cu undergoes a two step martensitic transformation from cubic (B2) to orthorhombic (B19), which occurs over a small temperature range near room temperature, and orthorhombic (B19) to monoclinic (B19′ ). However, the second transformation, B19-B19′ , is not observable using differential scanning calorimetry because it occurs over a wide temperature range. When the wires are pre-strained to low values, the SMA undergoes the cubic to orthorhombic transformation when heated above its As temperature. How- ever, when the SMA wire is pre-strained to higher values, e.g. 6%, there will be a stress induced martensitic transformation (B19′ –B19), and the alloy will transform at temperatures below As , due to the reverse B19′ –B19 transformation. Fig. 7 also shows that during cooling the recovery stresses decrease to zero for 1 and 3% pre-strains and to around 100 MPa for the 6% pre-strain. The recovery stresses do not return to the same stress that the wires experienced during the initial heating period. This is mainly related once again to the different transformations that the alloy experiences. For the 6% pre-strain, once the B19′ –B19 transformation has occurred, there will be less stress induced martensite available for transformation, therefore the recovery stresses decrease, but not to zero. This transformation is also the cause for differences in the rate of recovery stress with respect to temperature for differing pre-strain values. From Figs. 7 and 8 it can also be observed that there is an increase in the maximum recovery stress between 1% and the higher pre-strains, however, the difference between 3 and 6% is almost non-existent. For low pre-strains the SMA K.A. Tsoi et al. / Materials Science and Engineering A368 (2004) 286–298 295 Fig. 9. Stress vs. temperature for (a) NiTiCu-m2, (b) NiTi-rp, (c) NiTi-a1, and (d) NiTi-a2 wires pre-strained to 1, 3 and 6% as shown for the second heating and cooling cycle. transformation is fully completed at a temperature below 110 ◦ C whereas for the higher pre-strains the SMAs continue to transform above 110 ◦ C. Figs. 7(b) and 9 show the second temperature cycle, heated to 110 ◦ C for the SMA wires. From these curves it can be seen that the maximum recovery stresses generated are equivalent to those generated during the first heat cycle at a temperature of 110 ◦ C. The stress rate, dσ/dT , is the same as for the first cooling cycle. Fig. 10 shows only the second heating and cooling cycle in the oil bath for the different wires which were pre-strained to 3%. From this the maximum recovery stresses of each wire vary, with the NiTi wires (a1 and a2) having the greatest recovery stress and NiTiCu-m1 the lowest at a temperature of 110 ◦ C. The stress rate (dσ/dT ) was determined by calculating the tangent of the curve from the stress versus temperature chart. Fig. 11 shows the variation of the stress rate with pre-strain for the different wires measured between 60 and 110 ◦ C (above Af ) of each wire. It can be seen that dσ/dT decreases with increasing pre-strain. The R-phase Fig. 10. Heating and cooling cycle for different wires pre-strained to 3% and heated to 110 ◦ C. The second cycle only is shown. 296 K.A. Tsoi et al. / Materials Science and Engineering A368 (2004) 286–298 Fig. 11. Stress rate vs. temperature for different wires measured from stress gradient between 60 and 110 ◦ C (above Af ). alloy shows a large decrease between 1 and 6%, and this is reflected in the larger recovery stresses of the 1% alloy (62 MPa greater) compared to that of the 3 and 6% wires, at 110 ◦ C, shown in Fig. 9. For the other wire types the difference between the stress rates of 1 and 3–6% are not large enough for the recovery stress of the 1% wire to be greater than that for 3 and 6% at a given temperature. Figs. 7–9 show that this is generally the case; the recovery stress for 1% is less than for the 3 and 6% cases at 110 ◦ C. The corresponding resistive heating cyclic behaviour over time of the recovery stresses was measured and a representative example of the results is shown in Fig. 12. It was expected that the higher the pre-strain is the more stable the recovery stresses should become. The reason for this is similar to why the strain versus temperature behaviour stabilises over many cycles (Fig. 1). The higher the pre-strain is on the wire, the more defects will be induced, thus, producing stable recovery over time when the maximum number of defects is reached. During temperature cycling this will occur earlier compared with wires pre-strained at lower values. However, the results shown in Fig. 12 indicate that the most stable pre-strain is 3%. This tends to suggest that the optimal pre-strain and the number of defects induced is material dependent and not simply a matter of pre-strain. The NiTiCu-m1 wire was resistively cycled using different levels of current in order to determine the effect of varying the current on the stability of the wires, as shown at point E in Fig. 2. For the 600 and 700 mA currents (not shown) the stability of the wires is good, however as the current is increased to 800 mA the recovery stresses start to decrease and for 900 mA there is a decrease in the recovery stresses of around 14% over 60,000 s (or 4600 cycles). Results for the recovery stress over time using resistive cycling for different wires pre-strained to 3% and activated with a current of 700 mA (not shown) were also obtained. Fig. 12. Stress vs. time long term cyclic behaviour using resistive heating and a current of 700 mA for NiTiCu-m1 wires pre-strained to (a) 1%, (b) 3% and (c) 6%. The cycling data is shown by the grey area and the trends of the maximum and minimum stresses are indicated by the dashed line. It was found that most of the wires are stable over extensive cycling, however there does tend to be a small decrease in the recovery stress at the beginning of the cycling, followed by stable behaviour. The NiTi-m wire showed an initial decrease in the recovery stress of around 50 MPa, but K.A. Tsoi et al. / Materials Science and Engineering A368 (2004) 286–298 this was followed by stable behaviour after about 31,000 s. This initial decrease in the recovery stress is due to the formation of dislocations and defects which stabilises over time [6]. 6. Summary of results The thermomechanical testing of the SMA wires showed the different aspects of the recovery and strain characteristics of the transformations with temperature. The main conclusions reached were: • The stress–strain behaviour of the wires is a function of temperature. • The strain–temperature behaviour of the wires showed that with differing constant load the hysteresis curve of the wires changes. The smallest hysteresis was found for the NiTi-rp wires as well as for the NiTiCu-m1 wires. The smaller the hysteresis, the better for embedding the wires into composites, due to the easier heating capabilities. • The resistance–strain behaviour of the wires showed that with increasing pre-strain the resistance also increases linearly. Thus, it is possible to measure any changes to the pre-strain (for example during damage of the wires) by measuring the resistance and comparing it to the known resistance of a wire with no pre-strain. • The recovery stress–temperature behaviour showed that there was a slight difference in the maximum recovery stresses with differing wire. It was observed that the stress rate (dσ/dT ) decreased with increasing pre-strain. • During the first temperature cycle it was observed that for high pre-strains (6%) recovery stresses start building up instantly when heated and do not return to zero on cooling, whereas for the lower pre-strains, generation of recovery stresses start only after As is reached, and on cooling, stresses returned to zero. This was found to be due to the type of transformations occurring for different initial pre-strain conditions. • During the second and following temperature cycles, recovery stresses were found to be repeatable, and transformations exhibited similar behaviour to the first cooling cycle. • In general, the maximum recovery stresses were slightly higher for the 3 and 6% compared with the 1% specimens, because for the latter the reverse transformation is completed at 110 ◦ C, whereas the higher pre-strains will continue to transform above 110 ◦ C. The recovery stress rate for the wires decreased with increasing pre-strain, however the differences between the rates for 1 and 3–6% were not large enough for the pre-strains to have an influence on the recovery stresses. The exception was the R-phase alloy, which had a large decrease in stress rate between 1 and 6% and the recovery stresses for the 1% were noticeably higher than for 3 and 6% pre-strain (at 110 ◦ C), as expected. 297 • Resistive cycling of the wires showed that with increasing current there was a decrease in the stability of the recovery stress over time. The most stable wires for recovery stress generation over time were found to be the NiTiCu, the R-phase NiTi wire and the superelastic NiTi wire. The NiTi showed an initial degradation in the recovery stress but was then stable over time. • As a result the most appropriate wire for embedment into composite specimens for recovery stress generation was determined to be the NiTiCu due to a small temperature hysteresis and long term cyclic stability. The second part of this paper will discuss the thermomechanical characteristics of SMA-composites where SMA wires were embedded into a kevlar epoxy matrix. Acknowledgements This work has been completed under the framework of the ADAPT project which was funded by the European Commission, in the Industrial and Materials Technologies research and technological programme. Thanks also goes to Dr. Gidnahalli Dayanunda for carrying out the experiments on the SMA wires. K.A. Tsoi is an International Scholar from the School of Aerospace, Mechanical and Mechatronic Engineering, The University of Sydney, Australia and acknowledges Professor Y.-W. Mai, Dr. S.C. Galea and Professor M. Wevers for their generous support during this research and the financial support of the Defence Science and Technology Organisation of Australia and Zonta International through the Amelia Earhart Fellowship Award. References [1] J. Van Humbeeck, R. Stalmans, in: K. Otsuka, C.M. Wayman (Eds.), Shape Memory Materials, Cambridge University Press, 1998, pp. 149–183. [2] J.L. Proft, T.W. Duerig, in: T.W. Duerig, K.N. Melton, D. Stöckel, C.M. Wayman (Eds.), Engineering Aspects of Shape Memory Alloys, Butterworth-Heinemann Ltd., London, 1990, pp. 115–129. [3] T. Borden, in: T.W. Deurig, K.N. Melton, D. Stöckel, C.M. Wayman (Eds.), Engineering Aspects of Shape Memory Alloys, Butterworth-Heinemann Ltd., London, 1990, pp. 158–169. [4] E. Cydzik, in: T.W. Deurig, K.N. Melton, D. Stöckel, C.M. Wayman (Eds.), Engineering Aspects of Shape Memory Alloys, Butterworth-Heinemann Ltd., London, 1990, pp. 149–157. [5] P. Šittner, D. Vokoun, G.N. Dayananda, R. Stalmans, Mater. Sci. Eng. A 286 (3) (2000) 298. [6] J. Van Humbeeck, R. Stalmans, in: M. Schwartz (Ed.), Shape Memory Alloy Types and Functionalities, Encyclopedia of Smart Materials, vol. 2, Wiley, New York, 2002, pp. 951–964. [7] D. Vokoun, R. Stalmans, in: V.V. Varadan (Ed.), Proceedings of SPIE, Smart Structures and Materials: Mathematics and Control in Smart Structures, vol. 3667, 1999, pp. 825–835. [8] J. Haasters, G. Salis-Solio, G. Bensmann, in: T.W. Deurig, K.N. Melton, D. Stöckel, C.M. Wayman (Eds.), Engineering Aspects of Shape Memory Alloys, Butterworth-Heinemann Ltd., London, 1990, pp. 426–444. 298 K.A. Tsoi et al. / Materials Science and Engineering A368 (2004) 286–298 [9] C. Trépanier, M. Tabrizian, L.H. Yahia, L. Bilodeau, D.L. Piron, Mater. Res. Symp. Proc. 459 (1997) 363–368. [10] R.C.L. Sachdeva, S. Miyazaki, in: T.W. Deurig, K.N. Melton, D. Stöckel, C.M. Wayman (Eds.), Engineering Aspects of Shape Memory Alloys, Butterworth-Heinemann Ltd., London, 1990, pp. 452–469. [11] P.P. Jenkins, G.A. Landis, in: NASA Johnson Space Centre, 29th Aerospace Mechanisms Symposium, 1 May 1995, pp. 167– 171. [12] K.A. Tsoi, Ph.D. Thesis, The University of Sydney, September 2002. [13] C.M. Friend, C.R.D. Mattey, in: G.R. Tomlinson, W.A. Bullough (Eds.), Proceedings of 4th ESSM and 2nd MIMR Conference, Harrogate, Institute of Physics, 6–8 July 1998, pp. 107–114. [14] J.E. Bidaux, J.-A.E. Manson, R. Gotthardt, Mater. Res. Soc. Symp. Proc. 459 (1997) 107–117. [15] R. Gotthardt, J.-E. Bidaux, in: K. Inue (Ed.), International Conference on Displacive Phase Transfer and their Application in Material Engineering, TMS Publication, 1988, pp. 157–166. [16] J.-E. Bidaux, J.-A.E. Manson, R. Gotthardt, Third ICIM/ECSSM ’96, Lyon, 1996, pp. 517–522. [17] D.A. Hebda, M.E. Whitlock, J.B. Ditman, S.R. White, J. Intell. Mater. Sys. Struct. 6 (1995) 220. [18] J. Simpson, C. Boller, in: A-M.R. McGowan (Ed.), Proceedings of SPIE; Smart Structures and Materials 2002; Industrial and Commercial Applications of Smart Structures Technologies, vol. 4698, 2002, pp. 416–426. [19] C.H. Beauchamp, R.H. Nadolink, S.C. Dickinson, L.M. Dean, in: 1st European Conference on Smart Structures and Materials, Glasgow, 1992, pp. 189–192. [20] R. Stalmans, J. Van Humbeeck, L. Delaey, Acta Metall. Mater. 40 (3) (1992) 501. [21] P. Filip, K. Mazanec, Scripta Metall. Mater. 20 (1994) 67. [22] C.M. Wayman, T.W. Duerig, in: T.W. Duerig, K.N. Melton, D. Stöckel, C.M. Wayman (Eds.), Engineering Aspects of Shape Memory Alloys, Butterworth-Heineman Ltd., London, 1990, pp. 3–20. [23] S. Miyazaki, T. Imai, Y. Igo, K. Otsuka, Metall. Trans. A 17A (1986) 115. [24] G.B. Stachowiak, P.G. McCormick, Acta Metall. 36 (2) (1988) 291. [25] K.N. Melton, in: T.W. Duerig, K.N. Melton, D. Stöckel, C.M. Wayman (Eds.), Engineering Aspects of Shape Memory Alloys, Butterworth-Heinemann Ltd., London, 1990, pp. 21–35. [26] C.M. Friend, in: 1st European Conference on Smart Structures and Materials, EOS/SPIE, Glasgow, 1992, pp. 181–184. [27] S. Leeman, Masters Thesis, Department of Metallurgy and Materials Engineering, Katholieke Universiteit Leuven, June 1998. [28] Y.J. Zheng, J. Schrooten, K. Tsoi, R. Stalmans, Mater. Sci. Eng. A 335 (2002) 157. [29] W. Tang, R. Sandström, Z.G. Wei, S. Miyazaki, Metall. and Mater. Trans. A 31A (2000) 2423. [30] Y. Zhang, J.L. Lin, Z.Q. Ji, Z. Metallkd. 86 (2) (1995) 91. [31] T.H. Nam, T. Saburi, Y. Nakata, K. Shimizu, Mater. Trans. JIM 31 (12) (1990) 1050.