Improvement of High-Temperature Shape-Memory Effect by Multi-Component Alloying for TiPd Alloys

Abstract

The influence of multi-component alloying on the phase transformation and shape-memory effect was investigated to develop new high-temperature shape memory alloys (HT-SMAs). Four alloys—35Ti–20Pd–15Ni–15Pt–15Zr, 40Ti–20Pd–15Ni–15Pt–10Zr (high-entropy alloys, HEAs), 45Ti–20Pd–5Ni–25Pt–5Zr, and 45Ti–20Pd–10Ni–20Pt–5Zr (medium-entropy alloys, MEAs, at%)—were prepared. At room temperature, the B2 structure was stable in the HEAs, and no martensitic transformation (MT) was observed. However, in the MEAs, an MT from the B2 structure to a B19 structure was clearly observed. The MT temperature of the MEAs was comparable to or higher than those of binary and ternary TiPd alloys. The strengths of both the martensite and austenite phases in 45Ti–20Pd–5Ni–25Pt–5Zr were higher than those in 45Ti–20Pd–10Ni–20Pt–5Zr and ternary TiPd alloys. We attempted to explain the high strength using the δ parameter, which indicates the lattice distortion for various atomic sizes, but a clear correlation was not observed, as there were no significant differences in the δ parameter among the tested alloys. The shape recovery was investigated via a thermal cyclic test under an applied stress in the range of 15–200 MPa. Although a small plastic strain was introduced during the thermal cyclic test, a shape recovery over 80% was obtained for both MEAs. Training, that is, the thermal cyclic test under the same applied stress, was conducted to investigate the change of the irrecoverable strain and the work output. For 45Ti–20Pd–5Ni–25Pt–5Zr, the irrecoverable strain was deleted after 50 cycles, and perfect recovery was obtained. The largest work output (3.5 J/cm³) was obtained under 200 MPa. In 45Ti–20Pd–10Ni–20Pt–5Zr, perfect recovery was obtained from the first cycle. However, the recoverable strain was small, and the largest work output was 1.5 J/cm³ under 200 MPa. The shape recovery of 45Ti–20Pd–5Ni–25Pt–5Zr is promising for new HT-SMAs compared with the ternary Ti–Pd–Zr alloys and other HEA-SMAs.

Fig. 7 ST curves of (a–c) 45Ti–20Pd–5Ni–25Pt–5Zr ((a) for 1 cycle and 80 cycles at 200 MPa, (b) for 1 cycle and 103 cycles at 300 MPa, (c) for 1 cycle and 71 cycles at 400 MPa) and (d–f) 45Ti–20Pd–10Ni–20Pt–5Zr ((d) for 1 cycle and 91 cycles at 200 MPa, (e) for 1 cycle and 93 cycles at 300 MPa, (f) for 1 cycle and 39 cycles at 550 MPa).

1. Introduction

Shape-memory alloys (SMAs) undergo a martensitic transformation (MT), in which their crystal structure changes from the austenite phase to the martensite phase without composition change via diffusion during cooling. SMAs can recover the introduced strain in the martensite phase via the reverse transformation. The MT temperature (MTT) controls the operation temperature of SMAs. Presently, TiNi alloys with an MT from a B2 structure to a B19′ structure are available SMAs. For example, they are used in a wide range of fields as hot–cold water mixing faucets, shower valves, and oil flow control devices. However, because the MTT of the TiNi alloy is <100°C, the operation temperature is limited. The development of high-temperature SMAs (HT-SMAs) will allow the use of SMAs in aerospace and automotive applications.¹⁾ Therefore, HT-SMAs with the MTT above 100°C have been developed. In an early stage, increasing the MTT of TiNi via the addition of alloying elements such as Hf, Zr, Pd, Pt, and Au was attempted.^1–10) The MT from the B2 structure to the B19 structure occurs in TiPd and TiPt, and their MTTs are >500°C. TiNi, TiPd, and TiPt are fully solid solutions, and addition of Pd or Pt to TiNi increases the MTT above 500°C.^11,12) However, perfect shape recovery is difficult to obtain. Recently, the addition of Hf and Zr has attracted attention, because nanosized precipitates of (Ti, Hf)₃Ni₄ or (Ti, Zr)₃Ni₄, which are referred to as the H phase, enhanced the strength of TiNi alloys, and as a result, perfect shape recovery was obtained under a high applied stress.^13–19) However, the MTTs of TiNi–Hf and TiNi–Zr with the H phase are <300°C.

We focused on TiPt,^20–24) TiAu,^25,26) and TiPd^27–33) alloys with the MT from the B2 structure to the B19 structure for HT-SMA because their MTTs are approximately 1000, 600, and 570°C, respectively. Because the shape-recovery ratio of the binary Ti–50Pd (at%) determined via a simple compression test was 39.5%,²⁷⁾ improvement of the strength of alloys has been attempted via the addition of alloying elements such as Cr, Nb, Zr, Hf, Mo, W, Ta, Ir, Ru, and Co.^28,31) It was found that Zr is the most promising alloying element, and the shape recovery was improved to 84.3% via the addition of 5 at% Zr.²⁷⁾ Then, the composition dependence of Zr was investigated, and the martensite finish temperatures (M_f) of Ti–50Pd–5Zr, –7Zr, and 10Zr were 445, 400, and 302°C, respectively (reduced from 480°C for Ti–50Pd).³²⁾ Perfect shape recovery was obtained for Ti–50Pd–7Zr and 10Zr alloys via cyclic thermal training, although it was not obtained for Ti–50Pd–5Zr via cyclic thermal training.³²⁾ The simultaneous addition of Zr and V to TiPd was also attempted, because V has the largest strengthening effect on TiPd among alloying elements.^29–31) Perfect recovery was obtained for Ti–50Pd–2.5Zr–2.5V and Ti–50Pd–1Zr–4V via thermal cyclic training, but the M_f decreased to 368 and 424°C, respectively.³²⁾ To maintain a high MTT and strength, the addition of Pd to TiPt was also attempted. The M_f of the Ti–15Pd–35Pt–5Zr was 590°C, which is very high, but perfect shape recovery was not obtained via thermal cyclic training.³³⁾

Recently, high-entropy alloys (HEAs; multi-component equiatomic or near-equiatomic alloys) have attracted attention as new-concept alloys because their high entropy effect, severe lattice distortion, sluggish diffusion, and cocktail effect lead to unexpected properties, for example, compatibility of high strength and high ductility.^34,35) The severe lattice distortion and sluggish diffusion are expected to contribute to strong solid-solution hardening of HT-SMAs. The first attempt of an HEA to an SMA was performed for TiZrHfNiCu. The MT from the B2 structure to the B19′ structure with A_f of 337°C and M_f of 127°C (higher than that of TiNi) was observed.³⁶⁾ The hardness of TiZrHfNiCu was twice that of TiNi, and perfect shape recovery was obtained. However, the recovery strain was 10 times smaller than that of TiNi. Another example was TiZrHfNiCoCu.³⁷⁾ The MT from the B2 structure to the B19′ structure with A_f of 72°C and M_f of −80°C occurred. Large strain recovery of 4.9% was obtained under 650 MPa in TiZrHfNiCoCu, while a similar recovery strain of 4.5% in TiNi was obtained under 200 MPa. The resistance to high stress in TiZrHfNiCoCu was attributed to the significant solid-solution hardening effect. Application of an HEA to a HT-SMA was attempted for NiPdTiHf and NiPdTiHfZr.³⁸⁾ The MT from the B2 structure to the B19 structure, as well as TiPd, occurred. The highest MT, A_f of approximately 780°C, and M_f of approximately 600°C were obtained for NiPdTiHfZr. This indicates that HEAs are promising materials for HT-SMAs.

In this study, the phase transformation and shape recovery of TiPd-based multi-component alloys, i.e., TiPdNiPtZr, were investigated. The potential of TiPdNiPtZr for HT-SMAs was discussed.

2. Experimental Procedure

15-g ingots of TiPdNiPtZr alloys were melted via the arc-melting method. The nominal compositions of the alloys are presented in Table 1. The multi-component alloys are classified according to the mixing entropy using the following equations.³⁹⁾   

\begin{equation} \text{HEA:}\ \Delta S_{\textit{mix}} \geq 1.5R \end{equation} (1)

  

\begin{equation} \text{Medium-entropy alloy (MEA):}\ 1.0R \leq \Delta S_{\textit{mix}} \leq 1.5R \end{equation} (2)

  

\begin{align} &\text{Low-entropy alloy (LEA, conventional }\\ &\quad \text{solid-solution alloy):}\ \Delta S_{\textit{mix}} \leq 1.0R \end{align} (3)

Here, ΔS_mix represents the mixing entropy, and R is a gas constant (8.314 J/K mol).

Table 1 Nominal composition of alloys (at%) and mixing entropy ΔS_mix (J/molK).

ΔS_mix is defined by the following equation.   

\begin{equation} \Delta S_{\textit{mix}} = - R \sum\nolimits_{i = 1}^{n}x_{i}\mathit{ln}x_{i} \end{equation} (4)

Here, x_i is the mole fraction of the component i, and n is the number of constituent elements. The calculated mixing entropies are presented in Table 1. For reference, the mixing entropies of Ti–50Pd–5Zr, Ti–50Pd–7Zr, Ti–50Pd–10Zr, and Ti–50Pt–5Zr are also shown.

The ingots were encapsulated in a quartz tube with Ar gas, heated at 1000°C for 3 h, and then quenched in water. The microstructures of the heat-treated samples were observed using field-emission scanning electron microscopy (JSM 7200 F) with an energy dispersive X-ray spectromentry (EDS) and an electron backscatter diffraction (EBSD) pattern analyzer at 20 kV.

A disc-shaped sample having a diameter of 4 mm and a thickness of 1 mm was cut from the heat-treated ingots using a wire electric discharge machine to measure the MT, the austenite start and finish temperatures (A_s and A_f, respectively), and the martensite start and finish temperatures (M_s and M_f, respectively) via differential scanning calorimetry (DSC) (Material Analysis and Characterization, DSC3200s). The weight of each DSC sample was approximately 100 mg. The measurements were performed with a scanning rate of 10°C/min from room temperature to 700°C. A plate 10 mm long and 1.0 mm thick was cut from the as-cast ingot for X-ray diffraction (XRD) measurements (Rigaku Co., Ltd., RINT TTR-III), which were performed to identify the crystal structure using Cu Kα radiation at 50 kV and 300 mA.

To investigate the strength, a compression test was performed on heat-treated samples with dimensions of 2.0 × 3.0 × 4.0 mm³, with an initial strain rate of 3.0 × 10⁻⁴/s at 30°C below M_f and at 30°C above A_f (Shimazu Corp., AG-X). A single compression test for obtaining the plastic deformation was conducted for the martensite phase. For the austenite phase, a loading–unloading compression test was performed to remove the effect of the stress-induced martensite transformation. The sample was deformed to 0.25% initially and was unloaded, and the applied strain increased for each loading–unloading cycle until plastic deformation occurred. The 0.2% proof stress of the austenitic phase was obtained from the stress–strain curve by plotting the plastic strain for the maximum stress for each loading–unloading cycle.

Thermal cyclic tests were performed in the temperature range of M_f − 30°C to A_f + 30°C under compressive stresses of 15, 50, 100, 150, and 200 MPa to investigate the shape-memory properties. In the thermal cycling test, the sample was first heated to A_f + 30°C, then cooled to A_f − 30°C, and again heated to A_f + 30°C. The sample length was directly measured via image tracking using a charge-coupled device camera. The shape recovery and the work output were calculated according to the strain–temperature (ST) curve using the following equations.   

\begin{equation} \text{Shape - recovery ratio},\ \mathrm{r}\ (\%) = \varepsilon_{r}/\varepsilon_{t} \end{equation} (5)

  

\begin{equation} \text{Work output, w (J/cm$^{3}$)} = \varepsilon_{r} \times \sigma\ \text{(MPa)} \end{equation} (6)

Here, ε_r represents the recoverable strain, ε_t represents the transformation strain, and σ represents the applied stress.

The thermal cyclic test was repeated under the same stress, and the changes in the irrecoverable strain and the work output were investigated. The repeated thermal cyclic test was called “training.”

3. Results and Discussion

3.1 Microstructures

Back-scattered electron images of the samples heat-treated at 1000°C for 3 h followed by water quenching are shown in Fig. 1. The dendrite structure remained in all the samples, even after solution treatment at 1000°C. Because the homogenized microstructure was obtained in Ti–Pd–Zr ternary alloys via the same heat treatment at 1000°C for 3 h, the dendrite structure indicates slow diffusion in the HEAs and MEAs. The composition of the bright phase along dendrite boundaries investigated using EDS was 8∼12 at% for Pd, 2∼7 at% for Ni, 26∼30 at% for Pt, and 15∼23 at% for Zr for all tested alloys as shown in Table 2. Although the composition deviation was found among the tested alloys, the ratio between (Ti, Zr) and (Pd, Pt, Ni) was approximately 57:43. Since the compositions of Zr and Pt were higher in the bright phase along dendrite boundaries than those of the matrix, it is considered that the bright phase is segregation of Zr and Pt. The precipitates with dark contrast were also observed as shown by arrows in Figs. 1(b), (c), and (d). The ratio between (Ti, Zr) and (Pd, Pt, Ni) was approximately 2:1 as shown in Table 2. Then, it is considered these precipitates with dark contrast are Ti₂Pd type precipitates.

Fig. 1

Back-scattered electron of (a) 35Ti–20Pd–15Ni–15Pt–15Zr, (b) 40Ti–20Pd–15Ni–15Pt–10Zr, (c) 45Ti–20Pd–5Ni–25Pt–5Zr, and (d) 45Ti–20Pd–10Ni–20Pt–5Zr.

Table 2 Phase composition (at%) measured by EDS.

In the HEAs, i.e., 35Ti–20Pd–15Ni–15Pt–15Zr and 40Ti–20Pd–15Ni–15Pt–10Zr, the twin structure formed by the MT was not observed, as shown in Figs. 1(a) and (b). While, the twin structure was clearly observed in the images with high magnification in the MEAs, i.e., 45Ti–20Pd–5Ni–25Pt–5Zr and 45Ti–20Pd–10Ni–20Pt–5Zr, as shown in Figs. 1(c) and (d). The crystal structure of the HEAs was investigated via EBSD, and the B2 structure was observed in the whole area, indicating no phase transformation.

3.2 Martensitic transformation

The DSC curves are shown in Fig. 2. Both endothermic and exothermic peaks were observed for the MEAs, indicating that a phase transformation occurred, as shown in Figs. 2(c) and (d). However, the endothermic and exothermic peaks were not clearly observed for 35Ti–20Pd–15Ni–15Pt–15Zr as shown in Fig. 2(a). On the other hand, the small broad peaks were observed in 40Ti–20Pd–15Ni–15Pt–10Zr as shown in Fig. 2(b). However, the EBSD analysis indicated the crystal structure of 40Ti–20Pd–15Ni–15Pt–10Zr was the B2 structure at room temperature. These results indicate that phase transformation from the B2 to B19 structures was not occurred above room temperature in the HEAs. The microstructure observations supported the DSC measurement; that is, the twin structure formed during the MT was observed in the MEAs (Figs. 1(c) and (d)). However, the twin structure was not observed in the HEAs (Figs. 1(a) and (b)). The stable B2 structure in HEAs suggests that the MTT became lower than room temperature. The reason for this is unclear. Since no clear evidence of martensitic phase transformation was found in 40Ti–20Pd–15Ni–15Pt–10Zr, further investigation using high temperature X-ray diffractiometry is necessary to understand the broad peaks shown in Fig. 2(b). In a previous study, the MTT of the equiatomic alloy Ni₂₅Pd₂₅Ti₂₅Hf₂₅ was higher (for example, A_f was 714°C) than that of the near-equiatomic alloy Ni₃₅Pd₁₅Ti₂₀Hf₃₀ (A_f was 686°C).³⁶⁾ In our case, the alloy composition of the HEAs and MEAs was not equiatomic. Even so, the difference of phase transformation appeared. To understand the phase-transformation behavior, the MTT change for various combinations of alloying elements with different compositions must be investigated.

Fig. 2

DSC curves of (a) 35Ti–20Pd–15Ni–15Pt–15Zr, (b) 40Ti–20Pd–15Ni–15Pt–10Zr, (c) 45Ti–20Pd–5Ni–25Pt–5Zr, and (d) 45Ti–20Pd–10Ni–20Pt–5Zr.

The MTTs determined via DSC are presented in Table 3 together with the MTTs of the binary alloy Ti–50Pd,^27,31) the ternary alloy Ti–50Pd–5Zr,^27,31) Ti–50Pd–7Zr,³²⁾ Ti–50Pd–10Zr,³²⁾ and Ti–50Pt–5Zr^20,23) for reference. The MTT decreased with an increase in the Zr content in Ti–50Pd alloys. For example, the M_f and A_f decreased from 480 and 550°C for Ti–50Pd to 302 and 416°C for Ti–50Pd–10Zr, respectively. The decrease in M_f and A_f for 1 at% Zr was 17.8 and 13.4°C, respectively. In the MEA with 5 at% Zr, Pd was replaced with Ni and Pt. Compared with the MTT of Ti–50Pd–5Zr, the A_s, A_f, and M_s of 45Ti–20Pd–5Ni–25Pt–5Zr were higher than those of Ti–50Pd–5Zr, although the M_f was slightly lower than that of Ti–50Pd–5Zr. However, in 45Ti–20Pd–10Ni–20Pt–5Zr, where the amount of Ni increased from 5 to 10 at% and the amount of Pt decreased from 25 to 20 at%, the MTT was drastically reduced compared with 45Ti–20Pd–5Ni–25Pt–5Zr, and they were almost the same as those of Ti–50Pd–10Zr. In comparison with MTT in MEAs and Ti–50Pt–5Zr because MEAs included Pt to increase MTT, MTTs of MEAs were drastically lowered by addition of Pd and Ni. From a different point of view, MTTs of 45Ti–20Pd–5Ni–25Pt–5Zr was held close to those of Ti–50Pd by addition of 25 at% Pt.

Table 3 MTT (°C) and temperature hysteresis (A_f − M_s) of Ti–50Pd and Ti–50Pt alloys and MEAs.

The temperature hysteresis (A_f − M_s) of the binary and ternary Ti–Pd–Zr was approximately 50°C up to 7 at% Zr. It increased to 94°C for Ti–50Pd–10Zr. The temperature hysteresis of the MEA was approximately 100°C for 45Ti–20Pd–5Ni–25Pt–5Zr and 45Ti–20Pd–10Ni–20Pt–5Zr and was closed to that of Ti–50Pt–5Zr. The increase in the temperature hysteresis indicates that a large temperature change is necessary for shape recovery, and it is disadvantageous to use SMAs.

XRD patterns measured at room temperature and 750°C for the as-cast MEAs are shown in Fig. 3. The B19 and the B2 structures were observed at room temperature and 750°C, respectively, in both alloys, although the peaks of TiO₂ were also observed owing to oxidation during the measurement at 750°C. This indicates that the MT occurred during cooling after melting and solidification, and the crystal structure of TiPd was not changed by multi-component alloying.

Fig. 3

XRD patterns at room temperature and 750°C for (a) 45Ti–20Pd–5Ni–25Pt–5Zr and (b) 45Ti–20Pd–10Ni–20Pt–5Zr.

The temperature hysteresis of HE-SMAs was reported in previous studies. For example, the A_f and M_s of Ti_16.667Hf_16.667Zr_16.667Ni₂₅Cu₂₅ were 611 and 499 K, respectively and the temperature hysteresis was 112 K.³⁷⁾ For an alloy obtained by adding Co to TiHfZrNiCu, i.e., Ti_16.667Hf_16.667Zr_16.667Ni₂₅Co₁₀Cu₁₅, the A_f and M_s were 71.1 and 36°C, respectively, and the temperature hysteresis was 35.1°C, which was smaller than that of TiHfZrNiCu.³⁷⁾ The A_f and M_s of Ni₃₅Pd₁₅Ti₂₀Hf₃₀ were 686 and 525°C, respectively, and the temperature hysteresis was 161°C, which is large.³⁸⁾ In the case of the equiatomic composition for NiPdTiHf, the DSC peaks were sharp compared with those of Ni₃₅Pd₁₅Ti₂₀Hf₃₀, and the A_f and M_s of Ni₂₅Pd₂₅Ti₂₅Hf₂₅ were 714 and approximately 620°C, respectively.³⁸⁾ The temperature hysteresis was 94°C, which was smaller than that of Ni₃₅Pd₁₅Ti₂₀Hf₃₀. In most cases, the temperature hysteresis of multi-component alloys above 90°C was larger than that of binary Ti–Pd alloys (40°C). The large temperature hysteresis was attributed to the inhomogeneous microstructure of the as-cast samples.³⁸⁾ Another explanation was that the severe lattice distortion restricted the collective growth of martensite plates, causing large temperature hysteresis.³⁷⁾ In our case, the dendrite structure remained after heat treatment, and the microstructure inhomogeneity and severe lattice distortion may have been the reasons for the large temperature hysteresis. In both TiHfZrNiCu and NiPdTiHf, the mixing entropy of TiHfZrNiCu with Co (1.75R) and the equiatomic NiPdTiHf (1.4R) were slightly higher than those in TiHfZrNiCu (1.5R) and Ni₃₅Pd₁₅Ti₂₀Hf₃₀ (1.3R), respectively. The temperature hysteresis of the alloys with a higher mixing entropy was smaller. However, in our case as shown in Tables 1 and 3, although the mixing entropy of HEAs was higher than those of MEAs, the temperature hysteresis was not observed due to lower MTT below room temperature or no MT in HEAs. Compared with the temperature hysteresis of MEAs and binary and ternary alloys, the temperature hysteresis of MEAs was larger than for binary and ternary alloys up to 7 at% Zr although the mixing entropy of MEAs was higher than those of binary and ternary alloys. To understand the effect of the mixing entropy on the phase transformation, more data are necessary.

3.3 Mechanical property

The 0.2% proof stress of the MEAs and ternary Ti–Pd³²⁾ and Ti–Pt²³⁾ alloys is presented in Table 4. The 0.2% proof stress is plotted with respect to the test temperature in Fig. 4. Although the strength decreased with an increase in the testing temperature, the strength of the martensite and austenite phases of 45Ti–20Pd–5Ni–25Pt–5Zr were higher than those of the ternary TiPd alloys. However, the strength of 45Ti–20Pd–10Ni–20Pt–5Zr was approximately equal to that of the ternary TiPd alloys. The 0.2% flow stress of the martensite phase of Ti–50Pt–5Zr was high for the testing temperature, but the 0.2% flow stress of the austenite phase was smaller than those of other alloys. The behavior of mechanical properties of MEAs was closer to that of Ti–Pd alloys rather than that of Ti–Pt alloys.

Table 4 0.2% proof stress (σ_0.2, MPa) of ternary Ti–50Pd alloys, and Ti–50Pt and the MEAs. Testing temperatures were at M_f − 30°C and A_f + 30°C for Ti–50Pd alloys and MEAs, and at M_f − 50°C and A_f + 50°C for Ti–50Pt–5Zr.

Fig. 4

0.2% proof stress of ternary Ti–50Pd and Ti–50Pt alloys and the MEAs with respect to the testing temperature.

The solid-solution hardening is often governed by the misfit of the atomic size between solvent and solute atoms. However, in multi-component alloys, a solvent element does not exist. Instead of the misfit of the atomic size between solvent and solute atoms, a new parameter (δ, representing the misfit for the average atomic size of the constituent elements) is defined as follows.⁴⁰⁾   

\begin{equation} \delta = 100 \times \sqrt{\sum\nolimits_{\text{i} = 1}^{\text{n}}\mathrm{x}_{\text{i}}\left(1-\frac{\mathrm{r}_{\text{i}}}{\bar{r}} \right)^{2}} \end{equation} (7)

  

\begin{equation} \bar{r} = \sum\nolimits_{\text{i} = 1}^{\text{n}}\mathrm{x}_{\text{i}}\mathrm{r}_{\text{i}} \end{equation} (8)

Here, x_i represents the concentration of element i, r_i represents the atomic radius of element i, and r represents the average atomic radius of the constituent alloying elements. The atomic radii of the constituent elements are presented in Table 5. The average radius and the δ parameters calculated using the atomic radii in Table 5 are presented in Table 6. The average radius was almost the same for all the tested alloys. The δ parameters of the MEAs, Ti–50Pd–5Zr and Ti–50Pt–5Zr were compared. The δ parameters of both MEAs were larger than those of Ti–50Pd–5Zr and Ti–50Pt–5Zr. This indicates that severe lattice distortion occurred in the MEAs. While the δ parameters of the Ti–Pd–Zr ternary alloys increased with the Zr content, owing to the large atomic size of Zr compared with other elements. The δ parameters of Ti–50Pd–10Zr and the MEAs were almost the same. The large δ parameters of the MEAs and Ti–50Pd–10Zr indicate a high degree of solid-solution hardening in both austenite and martensite. However, a large solid-solution hardening effect was observed only for 45Ti–20Pd–5Ni–25Pd–5Zr. The strength of the other alloys exhibited almost the same dependence on the test temperature.

Table 5 Atomic radii of constituent elements.⁴¹⁾

Table 6 Average atomic radius, $\bar{r}$, and the δ parameter.

3.4 Shape-memory effect

To investigate the shape-memory effect, thermal cyclic tests were conducted at 15, 50, 100, 150, and 200 MPa for the MEAs. The ST curves for each applied stress are shown in Fig. 5. The transformation strain increased with increase of the applied stress for both alloys. An almost closed loop, that is, perfect recovery, was obtained up to 100 MPa, but the plastic strain appeared at ≥150 MPa for 45Ti–20Pd–5Ni–25Pt–5Zr. In the case of 45Ti–20Pd–10Ni–20Pt–5Zr, perfect recovery was obtained up to 200 MPa. This is because the MTT of 45Ti–20Pd–10Ni–20Pt–5Zr was lower than that of 45Ti–20Pd–5Ni–25Pt–5Zr, and consequently, the strength of the martensite and austenite phases around the MTT was higher for 45Ti–20Pd–10Ni–20Pt–5Zr than for 45Ti–20Pd–5Ni–25Pt–5Zr. The high strength around the MTT prohibited plastic deformation during the thermal cyclic test. Another difference between 45Ti–20Pd–5Ni–25Pt–5Zr and 45Ti–20Pd–10Ni–20Pt–5Zr was the sharpness of the ST curves. The ST curves were sharp for 45Ti–20Pd–10Ni–20Pt–5Zr, indicating that the temperature hysteresis during the thermal cyclic test was smaller for 45Ti–20Pd–10Ni–20Pt–5Zr than for 45Ti–20Pd–5Ni–25Pt–5Zr.

Fig. 5

ST curves of (a) 45Ti–20Pd–5Ni–25Pt–5Zr and (b) 45Ti–20Pd–10Ni–20Pt–5Zr. From bottom to top, the applied stress is 15, 50, 100, 150, and 200 MPa, where ε_t represents the transformation strain (%) and ε_r represents the recoverable strain (%).

The shape recovery ratio and the work output were calculated using eqs. (5) and (6), as described in the experimental procedure, using the recoverable strain (ε_r), the transformation strain (ε_t), and the applied stress obtained from Fig. 5. The shape recovery ratio and the work output are shown in Fig. 6. It appeared that there was no plastic strain up to 150 and 200 MPa in 45Ti–20Pd–5Ni–25Pt–5Zr and 45Ti–20Pd–10Ni–20Pt–5Zr, respectively. However, the enlarged ST curve indicated small plastic strain for all the tested stresses. For both MEAs, the recovery ratio was not 100%. As shown in Fig. 5(a), the plastic strain became large with an increase in the applied stress for 45Ti–20Pd–5Ni–25Pt–5Zr; then, the recovery ratio decreased significantly above 100 MPa, as shown in Fig. 6(a). The recovery ratio of 45Ti–20Pd–10Ni–20Pt–5Zr was close to 100%, as shown in Fig. 6(b). The work output increased with the applied stress. At 200 MPa, the work outputs of 45Ti–20Pd–5Ni–25Pt–5Zr and 45Ti–20Pd–10Ni–20Pt–5Zr were 3.5 and 4.5 J/cm³, respectively. Because the A_f values of 45Ti–20Pd–5Ni–25Pt–5Zr and 45Ti–20Pd–10Ni–20Pt–5Zr were 600 and 400°C, respectively, during the thermal cyclic test, a large work output (3.5 J/cm³) was obtained in 45Ti–20Pd–5Ni–25Pt–5Zr at 600°C, which is higher than the corresponding temperatures for ternary Ti–50Pd–7Zr (7 J/cm³ at 430°C) and Ti–50Pd–10Zr (5.8 J/cm³ at 400°C). In Ti–50Pt–5Zr, ST curves became trumpet-like shape indicating progress of deformation during thermal cyclic test.³³⁾ It can be said that shape memory effect of MEAs is closed to Ti–Pd alloys.

Fig. 6

Recovery ratio and work output: (a) 45Ti–20Pd–5Ni–25Pt–5Zr and (b) 45Ti–20Pd–10Ni–20Pt–5Zr.

3.5 Training effect

To understand the shape-recovery change during the thermal cyclic test, repeated thermal cyclic tests, i.e., training, was performed. The training conditions are presented in Table 7. The training was performed for 45Ti–20Pd–5Ni–25Pt–5Zr at 200, 300, and 400 MPa for 80, 103, and 71 cycles, respectively. For 45Ti–20Pd–10Ni–20Pt–5Zr, the training was performed at 200, 300, and 550 MPa for 91, 93, and 39 cycles, respectively. Before the training, a stress 50 MPa larger than the testing stress was applied for the first cycle, and then the applied stress decreased to the testing stress for training. The ST curves of the first and last cycles are presented in Fig. 7. For 45Ti–20Pd–5Ni–25Pt–5Zr, plastic strain was observed for the first cycle at the applied stresses of 200, 300, and 400 MPa. However, the plastic strain decreased during the thermal cycles and finally reached 0, and perfect recovery was obtained, as indicated by the last cycle in Figs. 7(a)–(c). Simultaneously, the transformation strain decreased from the first cycle to the last cycle during the training. The transformation strain also decreased with an increase in the applied stress.

Table 7 Training conditions.

Fig. 7

ST curves of (a–c) 45Ti–20Pd–5Ni–25Pt–5Zr ((a) for 1 cycle and 80 cycles at 200 MPa, (b) for 1 cycle and 103 cycles at 300 MPa, (c) for 1 cycle and 71 cycles at 400 MPa) and (d–f) 45Ti–20Pd–10Ni–20Pt–5Zr ((d) for 1 cycle and 91 cycles at 200 MPa, (e) for 1 cycle and 93 cycles at 300 MPa, (f) for 1 cycle and 39 cycles at 550 MPa).

Similar behavior was observed for 45Ti–20Pd–10Ni–25Pt–5Zr, as shown in Figs. 7(d)–(f). The plastic strain was unclear, and almost perfect recovery was obtained for the first cycle up to 550 MPa. The transformation strain decreased during the training and with an increase in the applied stress, similar to the results for 45Ti–20Pd–5Ni–25Pt–5Zr. However, the decrease in the transformation strain was more drastic for 45Ti–20Pd–10Ni–25Pt–5Zr. For example, the transformation strain at 300 MPa (Figs. 7(b) and (e)) was approximately 0.5% for 45Ti–20Pd–10Ni–25Pt–5Zr, while it was approximately 1% for 45Ti–20Pd–5Ni–25Pt–5Zr.

The changes in the plastic strain, recoverable strain, and work output during the thermal cyclic test are shown in Fig. 8 with respect to the cycle number. The plastic strain drastically decreased during the cyclic test and reached 0 after 50 cycles for 45Ti–20Pd–5Ni–25Pt–5Zr, as shown in Fig. 8(a). Perfect recovery was obtained after 50 cycles. The recoverable strain was the largest under 200 MPa and decreased with an increase in the applied stress (Fig. 8(b)). The recoverable strain decreased during the thermal cyclic test, but it was saturated after 50 cycles. Consequently, although the work output decreased during the thermal cyclic test, it was also saturated (Fig. 8(c)). The final work output was approximately 3.5, 3.0, and 2.0 J/cm³ under 200, 300, and 400 MPa, respectively. The largest work output was obtained under 200 MPa. For 45Ti–20Pd–10Ni–25Pt–5Zr, the plastic strain was not observed for the entire thermal cyclic test. The recoverable strain was almost constant under 300 MPa and slightly decreased under 200 MPa during the thermal cyclic test, as shown in Fig. 8(d). The recoverable strain was larger under 200 MPa than under 300 MPa. The work output reflected the recoverable strain; that is, a larger work output was obtained under 200 MPa than under 300 MPa, and the work output was almost constant (Fig. 8(e)). The work output was not estimated for 550 MPa, because the phase transformation was unclear under this condition.

Fig. 8

(a) Plastic strain, (b, d) recoverable strain, and (c, e) work output of (a–c) 45Ti–20Pd–5Ni–25Pt–5Zr and (d–e) 45Ti–20Pd–10Ni–20Pt–5Zr.

Although the behavior observed during the thermal cyclic test was not clearly understood, when the applied stress was increased, it is considered that dislocations were introduced by plastic deformation under 400 MPa, and these dislocations prevented the phase transformation. Consequently, the recoverable strain decreased under a high applied stress. Similar behavior was observed for Ti–35Pt–15Pd–5Zr.³³⁾ Dislocations were introduced in the initial cycles, and then the irrecoverable strain drastically decreased under 200 MPa compared with 50 and 100 MPa. However, the total number of dislocations introduced was higher under 200 MPa, and the decrease in the recoverable strain was large. Thus, a significant reduction in the work output was observed. In the present study, similar behavior may have occurred under 400 MPa.

Perfect recovery was obtained for the MEAs in the thermal cyclic test. Compared with the ternary Ti–Pd–Zr alloys or quaternary Ti–Pd–Zr–V alloys,³²⁾ the MEAs achieved perfect recovery under a larger applied stress (between 200–400 MPa). The largest applied stresses for obtaining perfect recovery were 150 and 200 MPa for Ti–50Pd–1Zr–4V and Ti–50Pd–10Zr, respectively, but they were <65 MPa for other alloys. This is attributed to the effect of the lattice distortion or the cocktail effect with an increase in the mixing entropy, although it is not clear for the solid-solution hardening effect, as shown in Fig. 4. Compared with other HE-SMAs, the transformation strain of Ni₃₅Pd₁₅Ti₃₀Hf₂₀ was approximately 0.8% under 75 MPa. Although the recoverable strain was unclear owing to the limitations of the test machine, if it is considered that the recoverable strain was equivalent to the transformation strain, the work output was approximately 0.6 J/cm³, which is very small.³⁸⁾ Our alloys were MEAs, not HEAs, but the results indicate that the multi-component alloys are promising as HT-SMAs.

4. Conclusions

The phase transformation and shape-memory effect of multi-component alloys were investigated.

1. (1)    Four alloys were prepared: 35Ti–20Pd–15Ni–15Pt–15Zr, 40Ti–20Pd–15Ni–15Pt–10Zr, 45Ti–20Pd–5Ni–25Pt–5Zr, and 45Ti–20Pd–10Ni–20Pt–5Zr (at%). According to the mixing entropy, 35Ti–20Pd–15Ni–15Pt–15Zr and 40Ti–20Pd–15Ni–15Pt–10Zr were identified as HEAs, and 45Ti–20Pd–5Ni–25Pt–5Zr and 45Ti–20Pd–10Ni–20Pt–5Zr were identified as MEAs.
2. (2)    In the HEAs, the martensitic phase transformation was not observed, and the B2 structure was stable at room temperature. However, in the MEAs, the martensitic phase transformation from the B2 structure to the B19 structure was clearly observed.
3. (3)    The A_f and M_f of 45Ti–20Pd–5Ni–25Pt–5Zr were 598 and 432°C, respectively, which were comparable to those of binary Ti–50Pd (480 and 550°C, respectively). The A_f and M_f of 45Ti–20Pd–10Ni–20Pt–5Zr were slightly lower (442 and 256°C, respectively) and comparable to those of Ti–50Pd–10Zr (416 and 302°C, respectively). The temperature hysteresis (A_f − M_s) of the MEAs was two times larger than that of the binary and ternary TiPd alloys. The large temperature hysteresis was attributed to the restriction of the growth of martensite plates during the MT due to the large lattice distortion.
4. (4)    The strengths of both the martensite and austenite phases in 45Ti–20Pd–5Ni–25Pt–5Zr were higher than those for the other three alloys. The δ parameter, which indicates the lattice distortion with various atomic sizes, was used to consider the solid-solution hardening effect. There was no large difference in the δ parameter among the tested alloys, although the δ parameter increased with the Zr content (large atomic size) and for multi-component alloying. Thus, the significant solid-solution hardening effect of 45Ti–20Pd–5Ni–25Pt–5Zr was not explained well using the δ parameter.
5. (5)    The shape recovery of the MEAs was examined using a thermal cyclic test. Although a small plastic strain was introduced during the thermal cyclic test, shape recovery of >80% was obtained for both MEAs.
6. (6)    Training was conducted to investigate the changes in the irrecoverable strain and the work output. For 45Ti–20Pd–5Ni–25Pt–5Zr, the irrecoverable strain was deleted after 50 cycles, and perfect recovery was obtained. The largest work output (3.5 J/cm³) was obtained under 200 MPa. For 45Ti–20Pd–10Ni–20Pt–5Zr, perfect recovery was obtained from the first cycle. However, the recoverable strain was small, and the largest work output was 1.5 J/cm³ under 200 MPa.

Acknowledgments

The study was partly supported by “Precious Metals Research Grant of TANAKA Memorial Foundation”, for which the authors express thanks.

REFERENCES

• 1)   J.  Ma,  I.  Karaman and  R.D.  Noebe: Int. Mater. Rev. 55 (2010) 257–315.
• 2)  G. Bigelow: NASA/TM-2008-214702 (E-15903), NASA Glenn Research Center, (2008) p. 6, pp. 26–34.
• 3)   G.S.  Bigelow,  S.A.  Padula, II,  A.  Garg,  D.  Gaydosh and  R.D.  Noebe: Metall. Mater. Trans. A 41 (2010) 3065–3079.
• 4)   K.C.  Atli,  I.  Karaman,  R.D.  Noebe,  A.  Garg,  Y.I.  Chumlyakov and  I.V.  Kireeva: Metall. Mater. Trans. A 41 (2010) 2485–2497.
• 5)   K.C.  Atli,  I.  Karaman,  R.D.  Noebe and  H.J.  Maier: Scr. Mater. 64 (2011) 315–318.
• 6)   K.C.  Atli,  I.  Karaman and  R.D.  Noebe: Scr. Mater. 65 (2011) 903–906.
• 7)   K.C.  Atli,  I.  Karaman,  R.D.  Noebe,  A.  Garg,  Y.I.  Chumlyakov and  I.V.  Kireeva: Acta Mater. 59 (2011) 4747–4760.
• 8)   Y.  Wang,  Y.  Zheng,  W.  Cai and  L.  Zhao: Scr. Mater. 40 (1999) 1327–1331.
• 9)   X.L.  Meng,  Y.F.  Zheng,  Z.  Wang and  L.C.  Zhao: Mater. Lett. 45 (2000) 128–132.
• 10)   X.L.  Meng,  W.  Cai,  F.  Chen and  L.C.  Zhao: Scr. Mater. 54 (2006) 1599–1604.
• 11)   N.M.  Matveeva,  Yu.K.  Kovneristyi,  A.S.  Savinov,  V.P.  Sivokha and  V.N.  Khachin: J. Phys. Colloques 43 (1982) C4-249–C4-253.
• 12)   O.  Rios,  R.D.  Noebe,  T.  Biles,  A.  Grag,  A.  Palczer,  D.  Scheiman,  H.J.  Seifert and  M.  Kaufman: Proc. SPIE 5761 (2005) 376–387.
• 13)   G.S.  Bigelow,  A.  Garg,  S.A.  Padula, II,  D.J.  Gaydosh and  R.D.  Noebe: Scr. Mater. 64 (2011) 725–728.
• 14)   D.R.  Coughlin,  P.J.  Phillips,  G.S.  Bigelow,  A.  Garg,  R.D.  Noebe and  M.J.  Mills: Scr. Mater. 67 (2012) 112–115.
• 15)   O.  Benafan,  R.D.  Noebe,  S.A.  Padula, II and  R.  Vaidyanathan: Metall. Mater. Trans. A 43 (2012) 4539–4552.
• 16)   H.E.  Karaca,  S.M.  Saghaian,  G.  Ged,  H.  Tobe,  B.  Basaran,  H.J.  Maier,  R.D.  Noebe and  Y.I.  Chumlyakov: Acta Mater. 61 (2013) 7422–7431.
• 17)   O.  Benafan,  A.  Grag,  R.D.  Noebe,  G.S.  Bigelow,  S.A.  Padula, II,  D.J.  Gaydosh,  N.  Schell,  J.H.  Mabe and  R.  Vaidyanathan: Intermetallics 50 (2014) 94–107.
• 18)   A.  Evirgen,  I.  Karaman,  R.D.  Noebe,  R.  Santamarta and  J.  Pons: Scr. Mater. 69 (2013) 354–357.
• 19)   A.  Evirgen,  I.  Karaman,  R.  Santamarta,  J.  Pons and  R.D.  Noebe: Scr. Mater. 81 (2014) 12–15.
• 20)   Y.  Yamabe-Mitarai,  T.  Hara,  S.  Miura and  H.  Hosoda: Mater. Trans. 47 (2006) 650–657.
• 21)   Y.  Yamabe-Mitarai,  T.  Hara,  S.  Miura and  H.  Hosoda: Intermetallics 18 (2010) 2275–2280.
• 22)   Y.  Yamabe-Mitarai,  T.  Hara,  S.  Miura and  H.  Hosoda: Metall. Mater. Trans. A 43 (2012) 2901–2911.
• 23)   A.  Wadood,  M.  Takahashi,  S.  Takahashi,  H.  Hosoda and  Y.  Yamabe-Mitarai: Mater. Sci. Eng. A 564 (2013) 34–41.
• 24)   A.  Wadood and  Y.  Yamabe-Mitarai: Mater. Sci. Eng. A 601 (2014) 106–110.
• 25)   A.  Wadood,  H.  Hosoda and  Y.  Yamabe-Mitarai: J. Alloy. Compd. 595 (2014) 200–205.
• 26)   A.  Wadood and  Y.  Yamabe-Mitarai: J. Alloy. Compd. 646 (2015) 1172–1177.
• 27)   M.  Kawakita,  M.  Takahashi,  S.  Takahashi and  Y.  Yamabe-Mitarai: Mater. Lett. 89 (2012) 336–338.
• 28)   R.  Arockiakumar,  M.  Takahashi,  S.  Takahashi and  Y.  Yamabe-Mitarai: Mater. Sci. Eng. A 585 (2013) 86–93.
• 29)   Y.  Yamabe-Mitarai,  A.  Wadood,  R.  Arockiakumar,  T.  Hara,  M.  Takahashi,  S.  Takahashi and  H.  Hosoda: Mater. Sci. Forum 783–786 (2014) 2541–2545.
• 30)   Y.  Yamabe-Mitarai,  R.  Arockiakumar,  A.  Wadood,  K.S.  Suresh,  T.  Kitashima,  T.  Hara,  M.  Shimojo,  W.  Tasaki,  M.  Takahashi,  S.  Takahashi and  H.  Hosoda: Mater. Today: Proc. 2(Suppl. 3) (2015) S517–S522.
• 31)   Y.  Yamabe-Mitarai: Mater. Sci. Forum 879 (2017) 107–112.
• 32)   H.  Sato,  H.Y.  Kim,  M.  Shimojo and  Y.  Yamabe-Mitarai: Mater. Trans. 58 (2017) 1479–1486.
• 33)   Y.  Yamabe-Mitarai,  W.  Takebe and  M.  Shimojo: Shap. Mem. Superelasticity 3 (2017) 381–391.
• 34)   J.W.  Yeh,  S.K.  Chen,  S.J.  Lin,  J.Y.  Gan,  T.S.  Chin,  T.T.  Shun,  C.H.  Tsau and  S.Y.  Chang: Adv. Eng. Mater. 6 (2004) 299–303.
• 35)   B.  Cantor,  I.T.H.  Chang,  P.  Knight and  A.J.B.  Vincent: Mater. Sci. Eng. A 375–377 (2004) 213–218.
• 36)   G.S.  Firstov,  T.A.  Kosorukova,  Y.N.  Koval and  V.V.  Odnosum: Mater. Today: Proc. 2(Suppl. 3) (2015) S499–S503.
• 37)   C.-H.  Chen and  Y.-J.  Chen: Scr. Mater. 162 (2019) 185–189.
• 38)   D.  Canadinc,  W.  Trehern,  J.  Ma,  I.  Karaman,  F.  Sun and  Z.  Chaudhry: Scr. Mater. 158 (2019) 83–87.
• 39)   J.W.  Yeh: JOM 65 (2013) 1759–1771.
• 40)  B.S. Murty, J.W. Yeh and S. Ranganathan: High-Entropy Alloys, 1st ed., (Elsevier, 2012).
• 41)   A.  Takeuchi and  A.  Inoue: Mater. Trans. 46 (2005) 2817–2829.