Thermal stability and some thermodynamics analysis of heat treated quaternary CuAlNiTa shape memory alloy

This study presents the heat treatment effect on a quaternary Cu79Al13Ni4Ta4 (wt%) shape memory alloy. The induction technique was used for melting the alloy and then four pieces of the alloy were heat-treated at (673 K, 873 K, 1073 K, and 1273 K) for one hour. Some physical parameters were characterized using DSC, XRD, and SEM-EDX. The as-cast and heat-treated samples were studied in terms of phase transformation temperatures, enthalpy change, entropy change, Gibbs free energy, and elastic energy. The transformation temperature was increased by applying heat treatment. Martensitic phase transformation at heat treatment temperature of 1273 K was not observed. Besides, after rising heat treatment temperature, some new phases such as ϒ 1 ′ and β 1 ′ were specified in XRD patterns and SEM images. Generally, for heat-treated samples, the transformation temperature remains almost constant after the 3rd cycle. However, the thermal stability of the as-cast alloy was not affected through thermal cycling.


Introduction
Recently shape memory alloys are widely used in modern technological applications such as automotive, aerospace, robotics, and biomedical applications [1,2], because of their well-known properties (superelasticity and shape memory effect), which gives them the ability to return to their original form after deformation [3,4]. In all applications, the effect of temperature on material characteristics is one of the most important factors, especially for the production of primary materials. Although Low-Temperature Shape Memory Alloys (LTSMAs) have transformation temperatures below 100°C, High-Temperature Shape Memory Alloys (HTSMAs) can operate at the temperature range of 100°C [5,6].
After NiTi family, Cu-based SMAs are the most alloy that has attracted the most attention in the technology and industry, because it can be handled easily, and its operation does not need more cost price compared with the other types of SMAs [7]. CuAlNi is the best type of Cu-based HTSMAs because it has high transformation temperature and thermal stability. CuAlNi SMAs can be used for temperatures around 200°C, while NiTi and CuZnAl-based alloys can be controlled for temperatures about 100°C [8,9]. On the other hand, Cu-based SMAs have poor thermal stability, thermal conductivity, and some mechanical properties compared to the other types of SMAs, however, these properties can be improved by adding third and fourth chemical elements [7,[10][11][12]. Saud et al observed that the porosity density and grain size was decreased by adding 2 wt% of Ta to Cu-Al-Ni, while its transformation temperature and the corrosion resistance was increased [11]. Dagdelen and colleagues reported that the grain size and precipitations were decreased through doping CuAlCr alloy with Ni element [13]. In addition, it is reported that microhardness and precipitate particles are increased by substituting Cu with Cr element in CuAl-based SMA [14].
In this study, 4 wt% of tantalum was added to ternary Cu 83−x Al 13 Ni 4 SMA, and the effect of heat treatment has been investigated on transformation temperature, microstructure, and some other thermodynamical properties. The stability of the alloy was studied by applying 10 complete thermal cycles.

Experimental procedure
CuAlNiTa SMA was produced accurately using high pure powder of primary metal elements, which includes 79% wt Cu-13% wt Al-4% wt Ni-4% wt Ta. The powders were pelletized after mixing the powders and pressed with a mechanical hydraulic compressor (SPACAC). The pelleted specimens were melted using the induction melting furnace. The production process was completed after quenching the sample into ice-water. To study the effect of heat treatment on CuAlNiTa SMA, the ingot was cut into some small pieces and kept them at 673 K (sample A), 873 K (sample B), 1073 K (sample C) and 1273 K (sample D), for one hour. Then, the effect of heat treatment on phase transformation temperatures (PTTs) and some thermodynamics parameters, such as entropy change, enthalpy change, Gibbs free energy, and elastic energy was investigated using Perkin Elmer Sapphire Differential Scanning Calorimetry (DSC) under the argon atmosphere at 10°C min −1 heating-cooling rate. In room temperature, the XRD was performed to analyze the crystal structure and different phases of CuAlNiTa SMA for as-cast and heat-treated samples. In addition, the microstructure of treated alloys was investigated using scanning electron microscope (SEM) and energy dispersive scanning x-ray (EDX), model (EVO 40XVP), also to obtain a clear microimage the specimens were firstly polished mechanically and then they etched with 20 ml HCl-96 ml methanol-5 gr Fe 3 Cl-H 2 O solution.

Results and discussions
3.1. Phase transformation temperature and thermodynamic properties The DSC measurements were performed for thermal analysis of all the samples. Figure 1(a) shows the typical DSC curve of as-cast CuAlNiTa alloy. Since the austenite and martensite transformation temperatures are above 400 K, so the alloy can be classified as high-temperature shape memory alloys (HTSMA). Table 1 summarizes the PTTs including austenite start (A s ), austenite peak (A p ), austenite finish (A f ), martensite start (M s ), martensite peak (M p ), martensite finish (M f ) temperatures, and the enthalpy change of phase transformation in both heating/cooling processes. The obtained results showed that all parameters affected by the heat treatment process. To show the thermal cycling behavior, the as-cast sample was subjected to 10 cycles for both exothermic and endothermic processes with a 10 K min −1 heating/cooling rate ( figure 1(b)). The width of the peaks stays constant, while both austenite and martensite peaks start to shift to the higher temperature by increasing the number of cycles. Also, the vibrations in the heating peak were disappeared after completing the first cycle. In addition, sample D has lost its shape memory characteristics, thus, DSC showed peak through neither heating nor cooling process (figure 1(c)). There can be several reasons why martensitic transformation does not occur at 1273 K heat treatment temperature. First of all, this temperature (1273 K) is close to the melting zone and there is a phase transition in this zone, e.g. it can be seen this phase transition in figure 1(c). This phase transition causes the CuAlNiTa alloy to lose its shape memory property. In addition, the precipitation phase in the martensite phase effect the transformation, and thus the composition around the phase is altered. Also, there are interfaces phases in the martensite variants [15,16]. Figure 2(a) illustrates that the PTTs have been generally increased by applying heat treatment on the alloys (except for sample D). The alloy that was aged at 673 K, has the largest temperature hysteresis, however, its value was decreased by increasing the heat treatment temperature ( figure 2(b)). In this study, the chosen heat treatment temperatures are before the eutectoid phase decomposition, which is taking place at 673 K. The eutectoid phase degradation occurs at 838 K [17], thus, it can be concluded that a significant decrease in the value of hysteresis has occurred after the eutectoid transformation took place. In addition, while the value of A p (figure 3(a)) and M p (figure 3(b)) increase gradually with increasing the thermal cycling, in all heat-treated samples these values have diminished. It is well observed that after applying 3 to 4 thermal cycling all cases have been stabilized. Also, the value of M p is converged by repeating the thermal cycling process. For the first cycle, sample C has recorded the highest value of M p , while after three completed cycles its value diminished about 100 K. In general, TTs values of as-cast sample increased while TTs values decreased in the heat-treated alloys, in addition after four complete cycles, the TTs almost stabilized. According to the obtained results, it can be said that the thermal cycling with DSC could regulate the TTs values of all cases.
The heat exchange through the heating process (or enthalpy change ΔH M→A ) was obtained using DSC software program, which based on the integration from austenite start to austenite finish temperature. The area under the DSC peaks represents the enthalpy change [10]: where / dq dt is the derivative of the instantaneous heat absorbed by the sample; T and t are the absolute temperature and time, respectively. In addition, another extensive property is entropy change (DS) that shows the disorderness of microstructure in the alloy. For austenite transformation, DS can be found using the following formula [18]: as where H A and S A are enthalpy and entropy in austenite; H M and S M are enthalpy and entropy in the martensite phase. The pushing force for transforming austenite to martensite can be given as [22]: In addition, elastic energy can be obtained from the differences between Gibss free energy at the beginning of the martensite phase transformation and when the transformation is completed [23]: All calculated parameters are listed in table 2. All of the calculated parameters have been obtained using DSC measurements and the aforementioned equations. In figure 4 it can be seen that all parameters have the same pattern since they are directly function of enthalpy change. Generally, the heat-treated sample at 673 K has recorded the highest value of enthalpy/entropy change with a comparably big D  G A M and G e value. Figure 5 reveals the XRD pattern of all specimens. The peaks were indexed by literature [3,[24][25][26]. The main peaks are b ¢ 1 (thin plate martensite) and g ¢ 1 (thick plate martensite). Also, there are some austenite phases including, g 1 and α. The DSC results support this finding. The matrix of all alloys showed martensite phases with some trapped austenite precipitation phases. Furthermore, the pattern of the alloy has been affected by heat treatment, i.e. some peaks have been strengthened. Moreover, to obtain the grain size of the alloys, the Scherrer equation can be used, which is mostly limited to metallic and ceramic microstructures that have grains in the range of nano-scale. Paul Scherrer proposed his equation, which based on the wavelength of the incident x-ray (λ=1.5406 Å), Braggs angle (θ), shape factor (K=0.9), and wideness at half maximum (B) of the XRD peak. Thus, the equation is as follows [27,28]:

Crystal and microstructural analysis
B cos 8 Figure 6 shows the effect of thermal treatment on the grain size of the alloy. It is obviously can be seen that heat treatment enhanced the size of grains. It is proved that grain size influences the mechanical behavior of all materials [29,30]. In this study, three different tests have been carried out for Vickers microhardness measurements. The standard deviation shows that there are different results obtained from various microstructures of the alloys ( figure 7). Generally, it is found that the value of microhardness has been decreased Table 2. Some calculated thermodynamic parameters for both treated and non-treated samples.  by increasing crystalline size. Also, microhardness depends on the heat treatment temperature. Sugimoto et al showed that microhardness in Cu-Al-Ni-Ti alloy starts to increase for heat treatment at 773 K, however, it falls down for higher temperatures [31]. The same results have obtained for Cu-Al-Ni-Ta alloy. On the other hand, Sampath [32] tried to enhance shape memory characteristics and ductility of CuAlNi-based SMAs through grain  refinement by adding fourth elements. He found that the microhardness was increased by decreasing the grain size of the alloys [32].
The microstructure of the as-cast and heat-treated samples is shown in figure 8. There are two different martensite phases can be seen in the SEM images, where the first one is course-g¢ 1 and the second one is fine b ¢ . 1 Also, there are some precipitation phases such as sphere-like Ta, Al, and Ta(Cu, Al) 2 phases. Since the SEM images, has been taken in room temperature, they show g ¢ 1 and b ¢ 1 phases in the matrix of all specimens, so the results support the DSC measurements which illustrate that the samples have martensite phase at room temperature. It is proved that g ¢ 1 (2 H) has higher Al content compared to b ¢ 1 (18 R) phase [33]. In addition, the amount of these types of martensite phases can influence transformation characteristics such as PTTs [34,35]. Although in the microstructure of CuAlNiTa alloy, which has been heat treated at 1273 K, shows martensite phases, there is no sign of phase transformation in its DSC measurement. In figure 8 grain boundaries cannot be seen on the given scale, while Sari [33] found that as-cast CuAlNi and CuAlNiMn alloys have shown microscopic grain about 1400 and 350 μm, respectively. The production technique is an important parameter that affects the grain size and hence all mechanical properties.

Conclusions
In summary, the main outcomes of this study are as follows: • Martensitic phase transformation was observed in all heat treatment temperatures, while the sample that heat treated at 1273 K gave no result. The PTTs were increased by applying heat treatment.
• It is found that thermal cycling diminished M p temperature in all heat-treated samples, while its value gradually increased in the as-cast alloy. The transformation temperature remains almost constant after the 3rd cycle, i.e., the thermal stability of the alloys increases.
• The XRD has shown various patterns after the heat treatment process was carried out. As the crystallization of the alloy increased, the number of diffraction peaks are also increased.
• The calculated grain size using Scherrer formula showed that the grain sizes are in the range of nanoscale, and the grain boundary cannot visible in the SEM images with Mag=2.50 kx and 500 kx. However, the crystal size is generally increased by increasing the heat treatment temperature.
• The grain size grew up with applying the heat-treatment process. It is obtained that the quantitative value of hardness has been decreased by increasing the crystal size. • From SEM microimages and EDX analysis, it is found that there are two martensite phases in the alloys, including thin-b ¢ 1 (the matrix) and thick-g ¢ 1 phase. In connection with the production method, Tantalumrich regions have been found.