Pt-Fe

One of the best documented martensitic transformations in Pt-based systems is that of PtFe₃. Pt-Fe alloys containing about 25 at.% Pt undergo an order-disorder transformation (2, 3). The disordered alloy has a f.c.c. structure and undergoes a martensitic transformation to a b.c.c. structure at around room temperature (4). Alloys containing more than 25.5 at.% Pt are also reported to transform martensitically from a disordered f.c.c to a b.c.t. (body centred tetragonal) or alternatively to a f.c.t. structure, depending on composition.

Wayman (5) reported that no shape memory effect was observed in the disordered or partially ordered parent phases of PtFe3, but that after ordering, the martensite transformation becomes thermoelastic, proceeding from f.c.c. to b.c.t. and back again, as many times as desired (6—8). Accompanying the increase in ordering is a drop in the M_s (the martensite start transformation temperature). It has been proposed by Muto et al . (4, 9) that the f.c.c. to b.c.t. and the f.c.c. to f.c.t. transformations are independent and competitive. The transformation temperature is very sensitive to changes in the composition and degree of order of the alloy. Oshima et al . (6) observed that the recoverable strain by the shape memory effect is small in the f.c.c. to f.c.t. transformation (<3%). Larger strains would be more desirable in SMAs.

The 24 at.% Pt-Fe alloy is very hard and brittle, but ductility increases with increasing Pt content. The PtFe3 system has some very interesting properties (such as a potential for magnetic-induced shape changes) and is worth exploring further. However, its Ms temperatures are too low to meet the requirements of either a medical in vivo application or of a combustion engine. Therefore, it will not be considered here.

Pt-Al, Pt-Ga

A martensite-type transformation has been reported in both the Pt-Al and Pt-Ga systems. At around 75 at.% Pt, the intermetallic compounds Pt₃Al and Pt₃Gaundergo a phase change from ϒ’, a high temperature cubic (L 1₂) structure, to ϒ₂’, a low temperature tetragonal (DO_c’) structure, see Figure 1. Some workers (10, 11) have reported the existence of an intermediate phase, ϒ₁’, (DO_c) although its existence is uncertain. The consecutive transformations of ϒ’ → ϒ₁’ → ϒ₂’ are reported to be martensitic in nature (10). The Pt-Al martensite phase transformation occurs in the region 73 to 78 at.% Pt, and M_s is reported ≤ 1300°C (12), and between room temperature and 350°C by others (10). For Pt-Al, at higher Pt contents, a Pt-rich solid solution exists which forms a eutectoid mixture with the tetragonal Pt₃Al phase. At least two, rather different, versions of the Pt-Al phase diagram have been published (10, 12).

Pt-Ti, Ni-Ti, Pt-Ti-Ni

A martensite transformation has been reported in the Pt-Ti system at around 50 at.% Pt. It is a displacive transformation from a cubic (B2) high temperature structure (β-TiPt) to an orthorhombic (B19) structure (α-TiPt) at low temperatures, see Figure 2, and has an M_s in the region of 1000°C. The cubic to orthorhombic transformation in this alloy is associated with a smaller temperature hysteresis than occurs, for example, for the cubic (B2) to monoclinic (B19’) transformation in the well-known NiTi SMA (known as Nitinol), which has an M_s in the vicinity of room temperature (13). The M_s in PtTi is a maximum at 50 at.% Pt and decreases as the Pt content varies on either side. Although in NiTi the transformation would normally go from B2 to B19’, it may, depending on alloy composition, do so via an intermediate rhombohedral (R) phase, or, if Pt is alloyed with the NiTi, via an orthorhombic (B19) phase. Depending on the composition, the reaction might not proceed any further, so the final product could be orthorhombic. Lindquist (14) has investigated the substitution of up to 30 at.% Pt for Ni in NiTi, and found that the alloys exhibited the one-way shape memory effect. An initial decrease in M_s was noted, but thereafter an increase in M_s with increasing Pt content was reported. As a smaller temperature hysteresis is associated with the B2 to B19 transformation than the B2 to B19’ transformation, this would affect the range of applications of a shape memory alloy.

Pt-Cr

Twinned microstructures have been observed in the 50 at.% Pt-Cr system by Waterstrat (15). The presence of twins could be due to a number of reasons: for instance, a displacive transformation, ordering, etc. Transformation from a disordered cubic structure to an ordered tetragonal structure occurs at ∼ 1100°C, although dilatome- try did not provide evidence for a displacive transformation (15).

Pt-V

The 50 at.% Pt-V alloy has an orthorhombic structure (B19) at room temperature (16), ostensibly transforming to a high temperature f.c.c. phase (based on the AuCu structure, that is, an ordered L 100000 f.c.c. structure) at ∼ 1500°C (17). However, Waterstrat (16) who examined this system some 14 years earlier, did not record this. In fact, in most displacive transformations a low-temperature B19 phase is generally coupled with a high temperature B2 b.c.c. parent phase, which suggests that the nature of the high temperature phase in this system justifies another investigation. Twinning was observed, but it was concluded that, while this could have arisen from a martensite transformation, it was most likely to have been formed by an order-disorder transformation.

Screening Pt-X Binary Systems for Alloys

Little further information, other than that mentioned above, has been published on Pt systems, and as there may be many other systems with the potential to be developed into SMAs, the authors screened the literature on binary Pt alloy systems, using the following criteria:

In order to decide whether a transformation is displacive, the relationship between the high-temperature parent phase and the product phase needs to be examined. This was often difficult as the parent phase was frequently only stable at exceedingly high temperatures, and more detailed work, including high temperature X-ray diffraction (XRD) and transmission electron microscopy (TEM), would need to be conducted before the true nature of a phase change could be determined. However, preliminary conclusions were made based on morphology and on the crystallographic relationships. At this stage, the aim of the investigation was merely to find some systems that might exhibit displacive transformations and that could have commercial application in the medical and high temperature fields.

Sample Manufacture and Characterisation

Alloy samples (1–3 g buttons) were prepared by arc-melting the individual elements into a button under an argon environment, on a water-cooled copper hearth. Buttons were melted three times to ensure mixing. In most instances, samples were first examined in the arc-melted state. They were subsequently solution treated to be wholly within the parent phase field, for periods of 3 days to a week, at 1200 or 1350°C for the Pt-Ti- and Pt-Al-based alloys, respectively, followed by furnace cooling. The alloys were then subjected to various further heat treatments.

Alloy hardness was evaluated on a Vickers hardness indenter with either 5 or 10 kg loads. At least six (and generally ten) indentations per sample per condition were performed, and an average is reported.

XRD data were gathered from the alloys, and the JCPDS database (the International Centre for Diffraction Data’s database of spectra of all known materials) (18) and Crystallographica (a software package) (19) were utilised in order to model lattices and predict the corresponding X-ray spectra. The modelling gave additional information regarding peak shapes as well as enabling the angular range to be extended beyond that of the JCPDS database thus allowing better comparison with the experimental data.

Samples were polished to a 0.25 uuuuum finish, and some were further polished using colloidal silica. Samples were etched for varying times with a warm solution of 50% aqua regia (3: 1 hydrochloric acid : nitric acid), 50% water. Examinations were performed by SEM. Elemental analysis was undertaken using energy dispersive X-ray spectroscopy (EDS) and pure element standards.

Differential thermal analysis (DTA), with scans taken in an argon atmosphere between 25 to 1500°C, was conducted at heating and cooling rates of ∼ 5°C min^-1. Each sample was repeatedly cycled from room temperature up to 1200 or 1500°C if no melting was anticipated. Some samples also underwent differential scanning calorimetry (DSC) between 20 to 1395°C at a rate of 10°C min^-1.

Ternary phase diagrams ought to be based on quenched samples but due to experimental constraints this was not possible, so diagrams of the phases present after furnace cooling are presented. This has some merit as some of the final systems proposed here, either as possible SMAs or worthy of further study, would probably have to employ furnace cooling if commercialisation were to be considered.

Pt-Al

Samples containing 68 to 85 at.% Pt were studied. Arc-melted samples often had dendritic microstructures that were not completely removed by heat treatment. Samples of composition 68 to 73.9 at.% Pt contained the cubic Pt₃Al phase, while the arc-melted 68.8 at.% Pt sample also contained the Pt₂Al phase. There was no evidence of twinning in samples in this range (68 to 73.9 at.% Pt).

Samples containing 71 to 73.9 at.% Pt were all single phase (cubic Pt₃Al), while those in the 74.6 to 76.6 at.% Pt range contained a phase identified by XRD as the tetragonal Pt₃Al phase. This phase was twinned, as shown in Figure 3.

Samples in the 76.6 to 87.9 at.% Pt range comprised a eutectoid mixture of the tetragonal Pt₃Al phase and the Pt-rich solid solution phase. The eutectoid point was at about 80.6 at.% Pt. Twins were generally observed in the tetragonal Pt₃Al phase.

The high temperature phase fields were generally consistent with the phase diagram of Massalski (17). Attempts were made to measure the phase transformation temperatures and DSC proved to be more successful than DTA as the enthalpy changes were very small, resulting in very small peaks. Mishima et al . (10) proposed that the phase field changed from (cubic Pt₃Al + (Pt)) to (tetragonal Pt₃Al + (Pt)) at ∼ 340°C, whereas Massalski proposed that this change was at 1280°C (17). In fact, samples in the region 74.6 to 85.9 at.% Pt were found in this work to have transformations in the range 132 to 137°C, see Figure 4.

The transformation temperature was very sensitive to composition and, as the Pt content increased from 74.6 to 75.7 at.% Pt, the transformation temperature increased from 132 to 308°C. The 85.9 at.% Pt sample was a eutectoid mixture of Pt₃Al + (Pt) with a transformation temperature ∼ 337°C. These transformation temperatures were more consistent with the phase diagram of Mishima et al . (10) than that of Massalski (17).

Hardness measurements were conducted to gain some idea of the mechanical properties. The measurements were found to depend on stoichiometry, varying between 350 to 650 HV₁₀. Minimum hardness occurred at around 75 at.% Pt. However, the still relatively high hardness of the 75 at.% Pt alloy (350—400 HV₁₀) suggested that the alloy might be difficult to work. Since all the other alloys were harder, it was felt that all the Pt-Al alloy range in this study might be difficult to work.

Nevertheless, limited amounts of cold rolling proved feasible, and the 75 at.% Pt alloy could be reduced by 18% (it hardened to 457 HV₁₀) before cracking occurred. With appropriate thermomechanical treatment, our experience suggests that this alloy could be worked further. It was deduced that hot rolling could also be a possible processing route. While the Pt-Al transformation was martensitic in nature, the phase transformation was only observed in a narrow composition range and M_s was exceedingly sensitive to the chemical composition of these button samples.

These features make Pt-Al an unlikely candidate for commercial application as a binary alloy. Ternary additions, using Ru or Ni, were then considered as a means to gain more control over the transformation by altering phase fields and stabilities.

Pt-Al-Ru

Ru was selected as a ternary alloying option as up to 40 at.% Ru dissolves in the Pt solid solution in the Pt-Ru system. A range of Pt-Al-Ru samples was investigated, containing between 5 and 20 at.% Ru, and a nominal Al content of 25 at.%. The intention was to substitute Ru for Pt in Pt₃Al. The effect of Ru on the phase stability is best illustrated in a proposed isothermal section shown in Figure 5.

For low additions of Ru and high Pt, a mixture of Pt₃Al and a Pt-rich solid solution (f.c.c.) containing Ru was observed. The Pt₃Al matrix was generally twinned, see Figure 6, suggesting it was the tetragonal Pt₃Al phase. In samples where the (Pt) phase was present in a Pt₃Al matrix, heat treatment at 1350°C resulted in the (Pt) solid solution phase decomposing into smaller, very fine, spheres. This heat treatment also caused the Ru to diffuse out of the matrix and into the Pt-rich solid solution. For larger additions of Ru, a mixture of cubic Pt₃Al and a h.c.p. (hexagonal close packed) Ru-rich solid solution containing Pt was observed, see Figure 7. After heat treatment, there was minimal Ru dissolved in the Pt₃Al matrix and minimal Al in the Ru-rich solid solution.

The hardness of all the samples was in the range 425 to 618 HV₁₀ showing that adding Ru increased the hardness considerably. Also, the addition of at least 4 at.% Ru suppressed the cubic-to-tetragonal transformation of Pt₃Al. For additions of less than 4 at.% Ru, a transformation temperature of about 350°C was indicated for the cubic-to-tetragonal change. The transformation temperature (at which the transformation occurred) was difficult to measure accurately because the enthalpy changes were very small.

The results suggest that the addition of Ru did not much extend the composition range of the parent Pt₃Al phase. The appearance of cubic Pt₃Al at room temperature in the samples with the higher Ru additions reflects that the tielines join h.c.p. (Ru) and cubic Pt₃Al, rather than the tetragonal form. This means that there is at least one invariant reaction involving the terminal solid solutions, (Ru) and (Pt), and the two forms of Pt₃Al.

Pt-Al-Ni

Nickel has been considered as a partial substitute for Pt (at least for non-medical applications) due to its similarity in metallurgical properties and its far lower cost. The region of present interest, in the Pt-Al-Ni ternary diagram (21), is dominated by the PtAl, tetragonal Pt₃Al and (Ni, Pt) phases. However, there is an inconsistency between this ternary phase diagram and existing binary phase diagrams (10, 17) in that the ternary section does not show a cubic Pt₃Al phase.

Heat treated samples in the Pt-rich end of the isothermal section, as well as samples containing only the Pt₃Al phase were targeted, and the results are summarised in the proposed isothermal section at 1350°C for Pt-Al-Ni, see Figure 8. The samples in the Pt-rich end of the isothermal section, consisted of a mixture (eutectic or eutectoid) of Pt₃Al and (Pt). As the Pt content increased, the amount of the Pt₃Al phase diminished. The results show that there is a mixture of tetragonal Pt₃Al and (Pt) in the Pt-rich end of the Pt-Al-Ni section and a cubic Pt₃Al and (Pt) mixture at lower Pt contents. The tetragonal Pt₃Al phase underwent a displacive transformation. Nickel partitioned preferentially into the Pt-rich solid solution, but addition of Ni also extended the cubic Pt₃Al phase field into the ternary.

Many phases formed in the Pt-Al-Ni system (21). In addition to the Pt₃Al and Pt phases, a NiPt₂Al (Heusler phase) and a twinned phase (not Pt₃Al) were observed (20). The transformation temperatures were not detectable using either DTA or DSC. Hardness values ranged from 257 to 624 HV₁₀ depending on the phases present.

Pt-Ti

Samples of Pt-Ti containing 30 to 61 at.% Pt were examined. A two-phase region of Ti₃Pt and Ti₄Pt₃ (nominal stoichiometry) was observed between 30 and 43 at.% Pt, while between 43 and 46 at.% Pt, the Ti₄Pt₃ and TiPt phases coexisted. The TiPt phase has a wider stability range, unlike the other phases, and is stable between 46 and 55 at.% Pt. It occurs in a low temperature form (designated α-TiPt) with B19 orthorhombic structure, and in a high temperature form above ∼ 1000°C, (β-TiPt) with the B2 b.c.c. structure.

The β-TiPt structure cannot be preserved down to room temperature by quenching since it undergoes a martensitic transformation to α-TiPt on cooling. The β-TiPt ↔ α-TiPt transformations are complex and, just like those in NiTi, take place via one or more intermediate structures (22). In the region 55 to 63 at.% Pt, both the TiPt and the Ti₃Pt₅ phases were observed. Further discussion on phase relations in the Pt-Ti system, including DTA studies which suggested complex phase relations, can be found elsewhere (20, 22).

The morphology of the α-TiPt phase was dependent on the cooling rate. Figure 9 shows a typical lath-like structure observed in samples that were cooled rapidly. Figure 10 shows a finer twinned structure obtained after furnace cooling. The α/β transformation temperature for the TiPt phase varied from 950 to 1050°C, and was a maximum at the stoichiometric composition and decreased on either side of it. The minimum hardness occurred at the stoichiometric composition. The slower cooled samples (furnace cooled) were softer than faster cooled samples (water quenched or arc-melted). A furnace-cooled 50 at.% Pt-Ti sample was the softest, with a hardness value of ∼ 250 HV₁₀, and could be cold rolled to ∼ 50% reduction before cracking occurred. The sample was also hot rolled at ∼ 1100°C to a reduction of ∼ 90%, although oxidation was observed.

The stoichiometric TiPt composition appears to offer a good prospect for a workable high-temperature SMA, with a transformation temperature of ∼ 1000°C. However, a means to reduce this temperature, and a better level of oxidation resistance, would be very helpful. For these reasons, the addition of ternary elements to Pt-Ti was then investigated.

Pt-Ti-Ru

The phase fields proposed for the heat-treated samples are given in Figure 11. The Pt-Ru system consists of terminal Pt-rich and Ru-rich solid solutions, and a two phase mixture in the region 62 to 80 at.% Ru. In the Ru-Ti system from 45 to 53 at.% Ru, a B2 TiRu phase is observed. The terminal Ti-rich solid solution is b.c.c. at elevated temperatures, undergoing an allotropic transformation to h.c.p. on cooling.

Alloy samples investigated included ones in which Pt was fixed at 50 at.%, with Ru varying from 5 to 20 at.% Ru, as well as samples that were anticipated to be in phase fields surrounding TiRu or TiPt. Samples in the arc-melted state were generally cored and had to be heat treated at 1200°C for 3 days to obtain a reasonably homogeneous microstructure (23).

The addition of 5 to 10 at.% Ru led to a decrease in the TiPt transformation temperature (M_s decreased from ∼ 1000°C to ∼ 700°C, while the hysteresis increased). The enthalpy change also decreased. These properties will affect the potential areas of application and determine whether small additions of Ru are desirable. For coupling applications, a large hysteresis would be desirable, whereas for actuator applications, a small hysteresis may be preferable.

The addition of Ru extended the TiPt phase field into the ternary diagram at least up to 19 at.% Ru. The results are summarised in a proposed partial isothermal section in Figure 11. The TiRu (B2) phase was noted to extend into the ternary up to about 15 at.% Pt. Thus, the two phases do not form a continuous field, despite sharing a common structure.

Pt-Ti-Ni

The addition of Ni to TiPt was also investigated for the same reasons as in the Pt-Al-Ni system. Nitinol (NiTi) is a well-known commercial SMA and some work has already been done on the substitution of Pt for Ni (24, 25). These studies concentrated on low additions of Pt (up to 30 at.%) and suggested that Pt-Ti-Ni alloys would show usable shape memory properties, but there were problems with formability (25).

The present work investigated a range of alloys and two rather different heat treatments. These were a homogenising heat treatment at 1200°C for 3 days followed by furnace cooling, and a solution anneal heat treatment at 1200°C for two hours followed by a water quench. The results are summarised in a proposed partial isothermal section in Figure 12. Nickel stabilised the α-TiPt extending it up to 18.8 at.% Ni into the ternary field. Further details are available elsewhere (20, 26).

Samples containing 5 to 20 at.% Ni (substituting for Pt in TiPt) were cored after arc-melting and had to be heat treated to homogenise the microstructure. If these alloys were to have commercial application, this would be a required processing step. Samples containing about 50 at.% Ti and up to about 5 at.% Ni consisted mainly of a single phase α-TiPt. Samples containing higher Ni contents contained an additional lath-like phase; XRD suggested this was possibly ∼ Ti₃Pt₅. There was also a third phase, rich in a light element or elements (indicated by low EDS totals), which was probably a Ti oxide.

The effect of Ni on the transformation temperature is given in Figure 13. The addition of up to 50 at.% Ni led to a decrease in transformation temperature from nearly 1000°C to room temperature. The apparent lack of influence from Ni at low additions, and the variations in hardness observed in this range are more a consequence of slight variations in Pt and Ti content among the samples. At low Ni additions, the Pt and Ti content have a greater influence on transformation temperature (which is a maximum at about 50 at.% Pt), than the Ni content.

Another important issue to consider is which transformation is being monitored: B2 → B19 (TiPt) or the B2 → B19’ (TiNi). In this study, the B2 → B19 transformation was monitored. The samples containing Ni appeared to have lower enthalpy changes. There was no observed significant trend or change in hysteresis with Ni additions. The addition of 20.3 at.% Ni led to a M_s of 572°C and A_s (austenite start temperature) of 620°C which compared favourably to prior measurements of 626 and 619°C, respectively, for a 20 at.% Ni addition (25).

Hardness values varied from 226 to 626 HV₁₀ for alloys in the Pt-Ti-Ni system, depending on the phases present and on the heat treatment. For samples containing the α-TiPt phase, the addition of Ni generally increased the hardness. As in the binary study, the cooling rate significantly affected the hardness with the furnace cooled samples being the softest. The hardness was also affected by the Ti content; for example, a sample of 10 at.% Ni could have a hardness from 398 to 626 HV₁₀ as the Ti content varied around the stoichiometric point. The lowest hardness values were obtained at around 50 at.% Ti (the balance is Ni + Pt).

Comparisons of SMA Materials

The case for a commercial Pt-based SMA can only be advanced based on some specialised and unique technological property or properties of sufficient utility to overcome the initial cost of Pt. In the absence of such an incentive, it would make more sense to use the existing base-metal SMAs. However, it should be noted that most SMA applications are comparatively high-value in nature and they use rather small SMA components. Thus, the actual price differential between a Pt-based and a conventional component is not nearly as large as the differential in price of the raw material. Therefore, Pt-based SMAs may be acceptable for some applications.

As far as the existing base-metal SMAs are concerned, it is interesting to review briefly their shortcomings. Any Pt-based SMA that could solve one of these problems would have prospects for commercialisation.

Conclusions

The most probable candidates from which to develop commercial Pt-based SMA systems are Fe₃Pt, and TiPt. In each case, commercial alloys would also contain ternary or higher additions. The Fe₃Pt alloys have working temperatures below room temperature, while both and TiPt have characteristics that suggest their use for in vivo and elevated temperature SMA applications.

The Pt-Al system (at around 75 at.% Pt) exhibits a displacive transformation from an ordered f.c.c. to ordered f.c.t. structure in the range 132 to 337°C. It appeared to be rather sensitive to compositional changes, and this might complicate its commercial application.

The Pt-Ti-based alloys (at around 50 at.% Pt) exhibit reproducible displacive transformations from b.c.c. (B2) to orthorhombic (B19) that have the potential for high temperature shape memory applications. Ternary additions of Ru and Ni, and possibly other elements, can be used to reduce the working temperatures to ambient, but this may be at the expense of a loss of ductility. These proposed systems can also be modified by control over their stoichiometry and thermomechanical processing.