TiPd- and TiPt-Based High-Temperature Shape Memory Alloys: A Review on Recent Advances

: In this paper high-temperature shape memory alloys based on TiPd and TiPt are reviewed. The e ﬀ ect of the alloying elements in ternary TiPd and TiPt alloys on phase transformation and strain recovery is also discussed. Generally, the addition of alloying elements decreases the martensitic transformation temperature and improves the strength of the martensite and austenite phases. Additionally, it also decreases irrecoverable strain, but without perfect recovery due to plastic deformation. With the aim to improve the strength of high-temperature shape memory alloys, multi-component alloys, including medium- and high-entropy alloys, have been investigated and proposed as new structural materials. Notably, it was discovered that the martensitic transformation temperature could be controlled through a combination of the constituent elements and alloys with high austenite ﬁnish temperatures above 500 ◦ C. The irrecoverable strain decreased in the multi-component alloys compared with the ternary alloys. The repeated thermal cyclic test was e ﬀ ective toward obtaining perfect strain recoveries in multi-component alloys, which could be good candidates for high-temperature shape memory alloys. 6. The irrecoverable strain of Ti 45 Pd 50 Zr 5 was saturated after 10 cycles, and it remained at approximately 0.1%. The irrecoverable strain of Ti 45 Pd 50 Hf 5 was approximately double that of Ti 45 Pd 50 Zr 5 , and it was unsaturated. This indicates that the strain recovery of ternary alloys is unstable for repeated cycles. The effect of Zr content on the training effect was investigated for Ti 45 Pd 50 Zr 7 and Ti 45 Pd 50 Zr 10 [36]. By increasing Zr addition, the irrecoverable strain finally disappeared during training and perfect strain recovery was achieved [36]. This outcome can be attributed to two reasons: (1) the solid-solution hardening effect is larger at high Zr contents, and (2) MTT reduction advantageously affects the suppression of plastic deformation. the alloys with imperfect recovery and the work output of the alloys with perfect recovery.


Introduction
Shape recovery in shape memory alloys (SMAs) occurs during a reverse martensitic transformation from martensite to austenite phases. Thereafter, the SMA operating temperature is related to the martensitic transformation temperature (MTT). High-temperature shape memory alloys (HT-SMAs) are defined as SMAs that can recover their shapes at temperatures above 100 • C. Several applications of HT-SMAs have been proposed. For example, Ni 30 Pt 20 Ti 50 , whose MTTs include austenite start temperature, A s : 262 • C; austenite finish temperature, A f : 275 • C; martensite start temperature, M s : 265 • C; and martensite finish temperature, M f : 240 • C, was applied for active clearance control actuation in the high-pressure turbine section of a turbofan engine [1]. This indicates that the design can offer a small and lightweight package without requiring motion amplifiers that cause efficiency losses and introduce an additional failure mode [1]. Another example is the helical actuators for surge-control applications in helicopter engine compressors [2]. In this application, Ni 19.5 Ti 50.5 Pd 25 Pt 5 , whose MTTs comprise A s : 243 • C, A f : 259 • C, M s : 247 • C, and M f : 228 • C, was applied because the alloy exhibited good work capabilities, a 2.5% recoverable strain, and a work output of 9.45 J/cm 3 at 400 MPa [2]. Several SMA applications, such as the active jet engine chevron, springs and wires for a general class of high-temperature actuators, oxygen mask deployment latch, SMA-activated thermal switch for lunar surface applications, variable geometry chevrons, and gas turbine variable area nozzles, have also been proposed [3]. Here, HT-SMAs, Ni 19.5 Ti 50.5 Pd 25 Pt 5 or Ni 50.3 Ti 29.7 Hf 20 are used only in springs and

Martensitic Transformation Temperature
The MTTs of the TiPd and TiPt alloys measured through differential scanning calorimetry (DSC) (DSC3200, NETZSCH Japan, Yokohama, Japan) are shown in Table 1. The temperature hysteresis, A f -M s , is also summarized in Table 1. Data without reference are from my group's unpublished work. The addition of a third element, such as Hf, Zr, V, Nb, Ta, Cr, Mo, W, Ru, Ir, and Co, was performed for TiPd, while elements, including Ir, Ru, Co, Zr, and Hf, were added for TiPt. In most cases, the MTT decreased with the addition of an alloying element. In Figure 1a, A f changes with increasing alloying element concentration are shown for some of the ternary alloys. Notably, A f decreased rather linearly with increasing amounts of alloying elements. The same trend appeared for A s , M s , and M f . The average decrease in M s per 1 at% of an alloying element up to 5 at% addition is shown in Table 1. Ta and W did not melt well into TiPd, and their particles remained in the alloy, thereby making the matrix composition of Ti 45 Pd 50 W 5 and Ti 45 Pd 50 Ta 5 approximately close to that of TiPd. Consequently, the decrease in the MTT by W and Ta was small. For other alloying elements, the decrease in MTT was between 15-37 • C. In Figure 1b, the MTT change with respect to Pd concentration is observed for Ti (95 − x) Pd x Zr 5 ; the MTT increased with an increase in the Pd concentration.   [33], TiPd-Zr [33,36], TiPd-Hf [34], TiPd-Ru [37], TiPd-Ir [37], and TiPd-Co [37], and (b) martensite transformation temperature for concentration of Pd of Ti50Pd45Zr5, Ti47Pd48Zr5, Ti45Pd50Zr5 [33], and Ti40Pd55Zr5 [35].
The MTT of TiPt was very high; for example, As and Af were 1000 °C and 1057 °C, respectively, as shown in Table 1. Although the addition of Ru, Co, Zr, and Hf decreased the MTT of TiPt, the addition of Ir increased it.
The MTTs of multi-component alloys are summarized in Table 2. Data without reference are from my group's unpublished work. The multi-component alloys were designed based on TiPd as follows: Pd, a group 10 element in the periodic table, was replaced by Co and Ir, which are group 9 elements; by Ni and Pt which are group 10 elements; and by Au, a group 11 element. Furthermore, Ti, a group 4 element, was replaced by Zr and Hf, which are also group 4 elements. Evident DSC peaks were not observed in the three alloys, Ti40Zr10Pd25Pt25, Ti45Zr5Pd37Ni13, and Ti45Zr5Pd25Pt20Ni5, and their MTTs were estimated from the strain-temperature (S-T) curves during the thermal cyclic test under constant stress.   [33], TiPd-Zr [33,36], TiPd-Hf [34], TiPd-Ru [37], TiPd-Ir [37], and TiPd-Co [37], and (b) martensite transformation temperature for concentration of Pd of Ti 50 Pd 45 Zr 5 , Ti 47 Pd 48 Zr 5 , Ti 45 Pd 50 Zr 5 [33], and Ti 40 Pd 55 Zr 5 [35].
The MTT of TiPt was very high; for example, A s and A f were 1000 • C and 1057 • C, respectively, as shown in Table 1. Although the addition of Ru, Co, Zr, and Hf decreased the MTT of TiPt, the addition of Ir increased it.
The MTTs of multi-component alloys are summarized in Table 2. Data without reference are from my group's unpublished work. The multi-component alloys were designed based on TiPd as follows: Pd, a group 10 element in the periodic table, was replaced by Co and Ir, which are group 9 elements; by Ni and Pt which are group 10 elements; and by Au, a group 11 element. Furthermore, Ti, a group 4 element, was replaced by Zr and Hf, which are also group 4 elements. Evident DSC peaks were not observed in the three alloys, Ti 40   Multi-component alloys are classified according to the mixing entropy, ∆S mix , using the following equations [51].
Here, ∆S mix is defined by the following equation: where x i is the mole fraction of component i, n is the number of constituent elements, and R is the gas constant (8.314 J/Kmol). HEA: Medium-entropy alloy (MEA): 6 of 21 Low-entropy alloy (LEA, conventional solid-solution alloy): The calculated mixing entropies are presented in Table 2. Based on the mixing entropies, the alloy classification is also shown in Table 2. Most quaternary alloys are classified as MEAs, while some are classified as LEAs, i.e., conventional solid-solution alloys. Notably, HEAs have been identified in some multi-component alloys with five or six constituent elements.
The MTTs of the five quaternary alloys, TiZrPdIr, TiZrPdPt, TiZrPdNi, TiZrPdCo, and TiZrVPd, are shown in Table 2. The combination of TiZrPdIr, TiZrPdNi, and TiZrPdCo decreased the MTTs compared with those of Ti 45 Zr 5 Pd 50 . This is because the phase transformations of TiIr, TiNi, and TiCo are different from those of TiPd. For example, TiIr undergoes a two-step phase transformation from the B2 structure at high temperature to a tetragonal structure (distorted B2 structure) at middle temperature, and finally to an orthorhombic structure with space group 65 at low temperature [56]. The MTT depends on Ir concentration. Notably, TiNi undergoes martensitic transformation from the B2 to the B19 structure [4]. The B2 structure is very stable in TiCo from room temperature to the melting temperature [25]; the addition of these elements is considered to stabilize the B2 structure. However, TiZrPdPt increased the MTT relative to Ti 45 Zr 5 Pd 50 . This is because the same martensitic transformation from B2 to B19 structures, as well as TiPd, occur in TiPt, whose MTT is very high, approximately 1000 • C, as shown in Table 1. Furthermore, Pt and Pd exhibit perfect solubility in both austenite and martensite phases. Consequently, MTT gradually increased with Pt content. In the case of TiZrVPd, the total amounts of Zr and V were maintained at 5 at%, and the values of A s and A f were approximately the same as those of Ti 45 Zr 5 Pd 50 , while the M s and M f values of TiZrVPd were lower than those of Ti 45 Zr 5 Pd 50 .
Although the MTTs of multi-component alloys depend on the combination of elements, there are some trends whereby those of TiZrPdPtNi and TiZrPdPtAu are relatively high, whereas that of TiZrPdNiCo is relatively low.
The potential of HEAs as SMAs was presented for the first time by Fistov et al., wherein Ti 16.7 Zr 16.7 Hf 16.7 Ni 25 Cu 25 was investigated and its A s and A f were 184 and 338 • C, respectively [53]. The addition of Co to Ti 16.7 Zr 16.7 Hf 16.7 Ni 25 Cu 25 was investigated, but its MTT drastically decreased and A f was -23 • C [54]. The MTTs of these alloys are summarized in Table 2. The potential of HEAs as HT-SMAs was first shown by Canadinc [55]. The MTT of their alloys was obtained from the DSC curves in reference [55] and is summarized in Table 2. The A s and A f of Ti 16.7 Zr 16.7 Hf 16.7 Pd 25 Ni 25 were 740 and 780 • C, respectively. In my group's experiments, HEAs and some MEAs did not show martensitic transformation up to 700 • C during DSC measurement. However, martensitic transformation may be possible for these alloys at temperatures higher than 700 • C. The MTT measurement of my group's HEAs is presently undergoing ultra-high temperature DSC. We found that some alloys have a high MTT, approximately close to 1000 • C in multi-component alloys.
Notably, the temperature hysteresis of the multi-component alloys becomes larger relative to that of ternary alloys. The number of alloys with temperature hysteresis exceeding 100 • C and their ratios are summarized in Table 3. The ratio of alloys with temperature hysteresis exceeding 100 • C in ternary alloys is 12% and it increases to 28% in LEAs and to 40% in MEAs and HEAs. It indicates that the ratio of alloys with temperature hysteresis exceeding 100 • C increased with increase of amount of constituent elements in alloys. In multi-component alloys, in particularly in MEAs and HEAs, the amount of constituent elements is equivalent or near-equivalent; therefore, it is expected that severe lattice distortion, which is considered to obstruct martensitic transformation, will occur. In many MEA cases, a drastic decrease in M s was observed compared with A f . When SMAs are used as actuators, a smaller temperature hysteresis is necessary to quickly respond to the surrounding environment. Table 3. Ratio of alloys with temperature hysteresis over 100 • C (%).

Number of Alloys with Temperature Hysteresis
over 100 • C

Strain Recovery Determined by Compression Test
In the early stage of my group's research [33][34][35][37][38][39][40][41][42][43][44], the recovery strain was estimated by measuring the sample length before and after the compression test using a strain rate of 3 × 10 −3 /s, and after heat treatment above A f . A sample was deformed by approximately 5% at the test temperature. The deformed sample length (L') was measured after cooling to room temperature. Thereafter, the sample was re-heated over A f , cooled to room temperature, and the recovered sample length (L") was measured. When the initial sample length is L 0 , the applied strain ε a is defined by the following equation: Recoverable strain ε r is defined according to the following equation: Strain recovery ratio, i.e., shape memory effect (SME) was evaluated using the following equation: The applied and recoverable strains, as well as the strain recovery ratios of the ternary TiPd and TiPt alloys, are summarized in Table 4. Data without reference are from my group's unpublished work. For the quaternary alloys, this method was used only for TiZrPdIr, and the results are summarized in Table 5. Although it is difficult to compare the effect of alloying elements due to different applied strains, some alloys have a high strain recovery ratio of above 80%. The quaternary alloys also exhibited a high strain recovery ratio of above 80%, as shown in Table 5. In Figure 2, the strain recovery ratios of Ti 45 Pd 50 × 5 are plotted for the periodic table family; it can be observed that the addition of Zr and Hf is effective toward improving strain recovery. Since the atomic size (metallic) of Zr and Hf are 160 and 159 pm, respectively, and larger than those of Ti (147 pm) and Pd (137 pm) [57], the solid-solution hardening effects of Zr and Hf are considered to exceed those of the other alloying elements. Thereafter, the 0.2% flow stress of the austenite and martensite phases were investigated by compression testing using samples with dimensions of 2.5 × 2.5 × 5 mm 3 . The 0.2% flow stresses of ternary TiPd and TiPt alloys are summarized in Table 6. Data without reference are from my group's unpublished work. When the martensite phase is deformed, a double yielding behavior often occurs. The first yielding behavior represents the rearrangement of the martensite variant and is referred to as the detwinning stress. The second yielding behavior represents the yield stress of martensite. In Table 6, the detwinning stresses of the martensite phase are also shown. In Figure 3, the 0.2% flow stress of the austenite and martensite phases of Ti 45 Pd 50 X 5 (X in group 4-6 elements) and Ti 50 Pd 46 Y 4 (Y in group 8-10 elements) are plotted for the periodic table family. Most alloying elements improved the strength of the austenite and martensite phases compared with those of Ti 50 Pd 50 . The strengthening behavior of the alloying element on the austenite and martensite phases was similar. Comparing  Figures 2 and 3, it is evident that high-strength alloys, such as Ti 45 Pd 50 V 5 , have a lower strain recovery than Ti 45 Pd 50 Zr 5 and Ti 45 Pd 50 Hf 5 with a smaller strengthening effect than Ti 45 Pd 50 V 5 . It is difficult to establish the correlation between strength and strain recovery ratio.    Figure 2. Strain recovery ratio of TiPd [33], Ti45Pd50X5 (X = Zr [33], Hf [34], V [34], Nb [34], Cr [34], and Mo [34]) and Ti50Pd46Y4 (Y = Ru, Co, and Ir) [37] for the periodic table family.     [33], Hf [34], V [34], Nb [34], Cr [34], and Mo [34]) and (b) Ti 50 Pd 46 Y 4 (Y = Ru, Co, and Ir) [37] for the periodic table family.

Strain Recovery Determined by Thermal Cyclic Test under Constant Stress
Recently, strain recovery was investigated through a thermal cyclic compression test under a constant load (Shimadzu AG-X, Shimadzu, Kyoto, Japan). The specimens were heated to 30 • C above  Figure 4b. This is because the compressive plastic deformation is larger than the expansion of the sample during heating above A f . Metals 2020, 10, x FOR PEER REVIEW 12 of 21 Figure 3. The 0.2% flow stress of the austenite and martensite phases of TiPd [33]: (a) Ti45Pd50X5 (X = Zr [33], Hf [34], V [34], Nb [34], Cr [34], and Mo [34]) and (b) Ti50Pd46Y4 (Y = Ru, Co, and Ir) [37] for the periodic table family.

Strain Recovery Determined by Thermal Cyclic Test under Constant Stress
Recently, strain recovery was investigated through a thermal cyclic compression test under a constant load (Shimadzu AG-X, Shimadzu, Kyoto, Japan). The specimens were heated to 30 °C above Af and then cooled to a temperature lower than Mf under a constant load of 15-200 MPa. Five thermal cycles under loads of 15, 50, 100, 150, and 200 MPa were applied for the same samples in the order given with heating and cooling rates of 50 °C/min. The work output (work per volume) was obtained from the product of the recovery strain and the applied stress. The strain temperature (S-T) curves of Ti45Pd50X5 (X = Zr, Hf, V, and Nb) are plotted in Figure 4. The transformation strain increased with an increase in the applied stress. The transformation strains of Ti45Pd50Zr5 and Ti45Pd50Hf5 exceed 2% above 50 MPa, as shown in Figure 4a,b. Nonetheless, the transformation strains of Ti45Pd50V5 are less than 1% at an applied stress of 200 MPa. Phase transformation was not evident in TiPdNb, even at 200 MPa. The perfect strain recovery was obtained under a small applied stress of less than 100 MPa in Ti45Pd50V5, but the irrecoverable strain appeared at 150 MPa and increased at 200 MPa. In Ti45Pd50Zr5 and Ti45Pd50Hf5, the small irrecoverable strain was observed, even at 15 MPa, which increased with the applied stress. At 150 and 200 MPa in Ti45Pd50Hf5, trumpet-shaped S-T curves were obtained, as shown by arrows in Figure 4b. This is because the compressive plastic deformation is larger than the expansion of the sample during heating above Af. The recoverable and irrecoverable strains, as well as the work output of Ti45Pd50X5 (X = Zr, Hf, and V) are plotted as a function of the applied stress, as shown in Figure 5; for reference, those of The recoverable and irrecoverable strains, as well as the work output of Ti 45 Pd 50 X 5 (X = Zr, Hf, and V) are plotted as a function of the applied stress, as shown in Figure 5; for reference, those of Ti 50 Pd 50 at 50 MPa are also plotted [58]. The recoverable strain increased with the applied stress; however, by increasing the irrecoverable strain, the recoverable strain started to decrease after the peak recoverable strain. Although the recoverable strain of Ti 50 Pd 50 is similar to that of Ti 45 Pd 50 X 5 (X = Zr and Hf), the irrecoverable strain is very large compared to that of the ternary alloys. The irrecoverable strain is expected to increase drastically with an increase in the applied stress. It also indicates that the addition of an alloying element is effective toward decreasing the irrecoverable strain. The work output increased with applied stress. Although an irrecoverable strain appeared, a large work output of 7.6 J/cm 3 was obtained for Ti 45 Pd 50 Zr 5 .
Metals 2020, 10, x FOR PEER REVIEW 13 of 21 Ti50Pd50 at 50 MPa are also plotted [58]. The recoverable strain increased with the applied stress; however, by increasing the irrecoverable strain, the recoverable strain started to decrease after the peak recoverable strain. Although the recoverable strain of Ti50Pd50 is similar to that of Ti45Pd50X5 (X = Zr and Hf), the irrecoverable strain is very large compared to that of the ternary alloys. The irrecoverable strain is expected to increase drastically with an increase in the applied stress. It also indicates that the addition of an alloying element is effective toward decreasing the irrecoverable strain. The work output increased with applied stress. Although an irrecoverable strain appeared, a large work output of 7.6 J/cm 3 was obtained for Ti45Pd50Zr5. Figure 5. Changes in (a) recoverable strain and irrecoverable strain, and (b) work output for the applied stress of Ti50Pd50 [58], Ti45Pd50Zr5 [45], Ti45Pd50Hf5, and Ti45Pd50V5 derived from the straintemperature curves shown in Figure 4.

Effects of Training
It is known that repeated thermal cyclic tests, which is referred to as "training," reduces the irrecoverable strain, and a perfect recovery is finally obtained [8,59]. For example, the irrecoverable strain of TiNi-based alloys became approximately 0 after 40 cycles under a load of 80 MPa. The training was effective for the TiNi alloys strengthened by the addition of Pd or Sc. Therefore, training was performed for Ti45Pd50X5 (X = Zr and Hf) under a 50-MPa load. The irrecoverable strain vs. the number of thermal cycles is plotted in Figure 6. The irrecoverable strain of Ti45Pd50Zr5 was saturated after 10 cycles, and it remained at approximately 0.1%. The irrecoverable strain of Ti45Pd50Hf5 was approximately double that of Ti45Pd50Zr5, and it was unsaturated. This indicates that the strain recovery of ternary alloys is unstable for repeated cycles. The effect of Zr content on the training effect was investigated for Ti45Pd50Zr7 and Ti45Pd50Zr10 [36]. By increasing Zr addition, the irrecoverable strain finally disappeared during training and perfect strain recovery was achieved [36]. This outcome can be attributed to two reasons: (1) the solid-solution hardening effect is larger at high Zr contents, and (2) MTT reduction advantageously affects the suppression of plastic deformation.

Effects of Training
It is known that repeated thermal cyclic tests, which is referred to as "training," reduces the irrecoverable strain, and a perfect recovery is finally obtained [8,59]. For example, the irrecoverable strain of TiNi-based alloys became approximately 0 after 40 cycles under a load of 80 MPa. The training was effective for the TiNi alloys strengthened by the addition of Pd or Sc. Therefore, training was performed for Ti 45 Pd 50 X 5 (X = Zr and Hf) under a 50-MPa load. The irrecoverable strain vs. the number of thermal cycles is plotted in Figure 6. The irrecoverable strain of Ti 45 Pd 50 Zr 5 was saturated after 10 cycles, and it remained at approximately 0.1%. The irrecoverable strain of Ti 45 Pd 50 Hf 5 was approximately double that of Ti 45 Pd 50 Zr 5 , and it was unsaturated. This indicates that the strain recovery of ternary alloys is unstable for repeated cycles. The effect of Zr content on the training effect was investigated for Ti 45 Pd 50 Zr 7 and Ti 45 Pd 50 Zr 10 [36]. By increasing Zr addition, the irrecoverable strain finally disappeared during training and perfect strain recovery was achieved [36]. This outcome can be attributed to two reasons: (1) the solid-solution hardening effect is larger at high Zr contents, and (2) MTT reduction advantageously affects the suppression of plastic deformation.

Strain Recovery of Multi-Component Alloys
A thermal cyclic test was performed on the multi-component alloys to investigate strain recovery. Some results have already been published [45][46][47][48][49][50]. The recoverable and irrecoverable strains, as well as the work output of some of the multi-component alloys are shown in Figure 7; the TiZrPdNi and TiZrPdPt alloys are shown in Figure 7a,c,e, and in Figure 7b,d,f, respectively, while Ti45Pd50Zr5 is

Strain Recovery of Multi-Component Alloys
A thermal cyclic test was performed on the multi-component alloys to investigate strain recovery. Some results have already been published [45][46][47][48][49][50]. The recoverable and irrecoverable strains, as well as the work output of some of the multi-component alloys are shown in Figure 7; the TiZrPdNi and TiZrPdPt alloys are shown in Figure 7a,c,e, and in Figure 7b,d,f, respectively, while Ti 45 Pd 50 Zr 5 is shown as a standard sample in all the diagrams. Furthermore, Ti 45 Figure 7a,c,e, the concentrations of Ti and Zr were maintained at 45 and 5 at%, respectively, in TiZrPdNi, and only the concentrations of Pd and Ni were changed. Among the tested alloys in Figure 7a,c,e, TiZrPdNi alloys exhibited a relatively high recoverable strain, although the recoverable strain of Ti 45 Zr 5 Pd 40 Ni 10 [48] was similar to that of ternary Ti 45 Pd 50 Zr 5 . Moreover, Ni addition seems to increase the recoverable strain of Ti 45 Pd 50 Zr 5 . The recoverable strain of Ti 45 Zr 5 Pd 40 Co 10 [48] was very small, thereby suggesting that Co addition drastically decreased the recovery strain. However, the recoverable strain of the multi-component alloys of TiZrPdNiCo was larger than that of Ti 45 Zr 5 Pd 40 Ni 10 [48]. This indicated that the recoverable strain was increased by the effect of the multi-component alloy, i.e., the high-entropy effect. In Figure 7c, the irrecoverable strains of most of the tested alloys were smaller than 0.2%. Only the ternary Ti 45 Pd 50 Zr 5 and quaternary alloy of Ti 45 Zr 5 Pd 45 Ni 5 , which were defined as LEA, represented a large irrecoverable strain exceeding 0.2%. A decrease in the irrecoverable strain of MEAs is also considered to be consequent of the high-entropy effect. As a result of the large recoverable strain, the work output of TiZrPdNi-type multi-component alloys also becomes large, as shown in Figure 7e. The work outputs of some alloys exceeded 10 J/cm 3 . The repeated thermal cycling test, i.e., training was also applied for some alloys. For example, training under 300 MPa was performed for Ti 45 Zr 5 Pd 40 Ni 10 for 100 cycles [48]. Although the transformation strain decreased, the irrecoverable strain decreased, and a perfect recovery was finally achieved during the thermal cyclic test. The final recovery strain was approximately 3.6% under 300 MPa. Thereafter, the work output was 10.8 J/cm 3 at the A f of 258 • C. For Ti 45 Zr 5 Pd 40 Co 10 , training was performed under 700 MPa. After nine cycles, the irrecoverable strain disappeared, and a perfect recovery was achieved. The final recovery strain was approximately 1%, and the work output was 7 J/cm 3 at the A f of 361 • C.
In Figure 7b,d,f, TiZrPdPt alloys are compared. Again, the concentrations of Ti and Zr were maintained at 45 and 5 at%, respectively, in most of the alloys. Compared with the recoverable strain in TiZrPdNi alloys, as shown in Figure 7a and in TiZrPdPt alloys, as presented in Figure 7b, those of TiZrPdPt alloys were smaller than 3% and less than those of TiZrPdNi alloys. This indicates that the addition of Pt decreases the recoverable strain. The same trend can be observed in quaternary alloys by comparing Ti 45 Zr 5 Pd 25 Pt 25 and Ti 45 Zr 5 Pd 45 Pt 5 . A small recoverable strain was achieved in the alloy with high Pt content. The effect of Zr on the recoverable strain in quaternary alloys could be understood by comparing Ti 40 Zr 10 Pd 25 Pt 25 and Ti 45 Zr 5 Pd 25 Pt 25 , and it was found that Zr addition increased the recoverable strain. In the multi-component alloys, it was found that high Pt addition decreased recoverable strain, as shown in Figure 7b. In TiZrPdPt alloys, it was found that the irrecoverable strain decreased in the multi-component alloys including the MEAs and HEAs, except for the LEA, Ti 45 Zr 5 Pd 45 Pt 5 , as shown in Figure 7d. The irrecoverable strain of Ti 45 Zr 5 Pd 25 Pt 25 is also small, which may be due to high MTT. The work outputs of the TiZrPdPt alloys are shown in Figure 7f; they are all between 2-6 J/cm 3  In Figure 7b,d,f, TiZrPdPt alloys are compared. Again, the concentrations of Ti and Zr were maintained at 45 and 5 at%, respectively, in most of the alloys. Compared with the recoverable strain in TiZrPdNi alloys, as shown in Figure 7a and in TiZrPdPt alloys, as presented in Figure 7b, those of TiZrPdPt alloys were smaller than 3% and less than those of TiZrPdNi alloys. This indicates that the addition of Pt decreases the recoverable strain. The same trend can be observed in quaternary alloys by comparing Ti45Zr5Pd25Pt25 and Ti45Zr5Pd45Pt5. A small recoverable strain was achieved in the alloy with high Pt content. The effect of Zr on the recoverable strain in quaternary alloys could be understood by comparing Ti40Zr10Pd25Pt25 and Ti45Zr5Pd25Pt25, and it was found that Zr addition increased the recoverable strain. In the multi-component alloys, it was found that high Pt addition decreased recoverable strain, as shown in Figure 7b. In TiZrPdPt alloys, it was found that the irrecoverable strain decreased in the multi-component alloys including the MEAs and HEAs, except for the LEA, Ti45Zr5Pd45Pt5, as shown in Figure 7d. The irrecoverable strain of Ti45Zr5Pd25Pt25 is also small, which may be due to high MTT. The work outputs of the TiZrPdPt alloys are shown in Figure  7f; they are all between 2-6 J/cm 3 and smaller than that of Ti45Pd50Zr5.
The strength of the austenite and martensite phases of some of the multi-component alloys are investigated, and the results are summarized in Table 7. The strength of the austenite phases in the tested alloys was between 200 and 300 MPa, and they were similar to those of the ternary alloys. This is because the MTTs of the multi-component alloys are relatively high. Therefore, it is difficult to correlate the strength and strain recovery directly. Thereafter, the temperature dependence of the strength of the martensite and austenite phases was investigated for ternary and multi-component alloys [49]. The strength of the multi-component alloys was higher in both martensite and austenite phases when compared at the same temperature. The solid-solution strengthening effect of the multi-component alloys was more evident when the strength of the austenite phase was compared at the same test temperature of 700 • C. The strength of the multi-component alloys was higher than that of the ternary alloys or LEAs. Table 7. Strength of the martensite and austenite phases of the multi-element alloys.

The Difference from Martensite Transformation
Temperature, • C for operating temperatures under 100 °C [61]. The work output of Ni50.3Ti29.7Hf20 is equivalent to that of TiNi alloys at higher Af (220 °C) than TiNi. Although the work outputs of some alloys in my group's research exceed 12 J/cm 3 at approximately 200 °C [48], most of them underwent decreasing work outputs with increasing Af. Nevertheless, it is notable that perfect recovery is achieved at temperatures between 200-600 °C because it is difficult to obtain above 200 °C. The work output obtained at 600 °C was approximately 3 J/cm 3 . .

Conclusions
This paper reviewed TiPd-and TiPt-based HT-SMAs. The effects of alloying elements on phase transformation, strain recovery, and strength are shown in ternary TiPd and TiPt alloys. In most cases, MTT decreased through the addition of alloying elements, as well as an increase in the alloying element content. The alloying element improved the strength of the martensite and austenite phases by the solid-solution strengthening effect. However, it is difficult to correlate with solid-solution strengthening and strain recovery. A thermal cyclic test was performed to investigate the strain

Conclusions
This paper reviewed TiPd-and TiPt-based HT-SMAs. The effects of alloying elements on phase transformation, strain recovery, and strength are shown in ternary TiPd and TiPt alloys. In most cases, MTT decreased through the addition of alloying elements, as well as an increase in the alloying element content. The alloying element improved the strength of the martensite and austenite phases by the solid-solution strengthening effect. However, it is difficult to correlate with solid-solution strengthening and strain recovery. A thermal cyclic test was performed to investigate the strain recovery. The irrecoverable strain decreased with the addition of an alloying element, but it was difficult to achieve perfect recovery, even after repeated thermal cyclic tests (training). Multi-component alloys were also investigated. Notably, MTT can be controlled by a combination of the constituent elements. For example, Zr, Ni, and Co decreased MTT, whereas Pt increased MTT. The effect of alloying elements on the recoverable strain differed from that of the MTT. For example, Ni increased the recoverable strain, while Pt and Co decreased it. The extra solid-solution hardening effect relative to ternary alloys was found in the multi-component alloys. Consequently, the irrecoverable strains were smaller in the multi-component alloys than those in the ternary alloys. In many cases, training was effective in achieving perfect recovery in the multi-component alloys; however, a decrease in the transformation strain also occurred. Multi-component alloys can be good candidates for HT-SMAs; the limitations that need to be overcome entail the suppression of the transformation strain reduction and temperature hysteresis increment.

Funding:
The study was partly supported by "Precious Metals Research Grant of TANAKA Memorial Foundation", for which the author expresses her thanks.