Experimental Investigations of Ni–Ti–Ru System: Liquidus Surface Projection and 1150 °C Isothermal Section

Ruthenium addition inhibits the formation of the topologically close-packed phases in Ni-based superalloys and improves the solid solution strength of Ni–Ti shape memory alloys. Therefore, the Ni–Ti–Ru phase stability is a very valuable indicator of the effects of Ru in Ni-based superalloys and Ni–Ti shape memory alloys. In this study, the isothermal section at 1150 °C and liquidus surface projection of the Ni–Ti–Ru ternary system were determined experimentally using the equilibrated alloy method and diffusion couple method, respectively. Alloys were prepared through the arc-melting of Ni, Ti, and Ru (all 99.99% purity), and then vacuum encapsulation in quartz tubes, followed by annealing at 1150 °C for 36 to 1080 h depending on the alloy composition. Diffusion couples were fabricated by joining one single-phase block (τ1) with one two-phase block (Ni3Ti + γ(Ni)), and the couples were annealed under vacuum at 1150 °C for 168 h. Reaction temperatures of as-cast alloys were determined by differential scanning calorimetry performed with heating and cooling rates of 10 °C/min. Scanning electron microscopy and X-ray diffraction were used to analyze the microstructure. Seven three-phase regions were found at the 1150 °C isothermal section. Seven primary solidification regions and five ternary invariant reactions were deduced in the liquidus surface projection. A new ternary compound τ1 was discovered in both the isothermal section at 1150 °C and liquidus surface projection. The results aid in thermodynamic modeling of the system and provide guidance for designing Ni-based superalloys and Ni–Ti shape memory alloys.


Introduction
Ni-based superalloys have been widely used as important components in aerospace engines and various gas turbines because of their good oxidation resistance and excellent mechanical properties at elevated temperatures. The exceptional high-temperature properties of Ni-based superalloys can be attributed to microstructural strengthening. The addition of Ti at low alloying concentrations will cause it to preferentially partition to the γ phase and serve to increase the antiphase boundary energy and strengthen the precipitate's resistance to deformation [1][2][3]. A systematic investigation was carried out by Mishima et al. [4,5] on the temperature dependence of strength in the Ni 3 Al phase with the addition of Ti. The results indicated that as the concentration of Ti increased, residual levels of γ forming solutes may remain within the γ phase and subsequently serve as potent solid solution strengtheners. Then, Christofidou et al. [6] studied the microstructures and mechanical properties of multi-component Ni-based alloys and found that the addition of Ti can promote the formation of the γ and γ phases in Ni-based alloys. Further, the effect of the Ru addition on the microstructures and various mechanical properties of Ni-based alloys has been studied by many researchers [7][8][9][10]. Song et al. [7] found that a small amount of Ru is beneficial in impeding the formation of the topologically close-packed
The Ni-Ru binary system is quite simple: Hallström [32] used the sub-regular solution model to re-evaluate this system, and the calculation result is shown in Figure 1b.
The Ru-Ti binary system was determined by Raub and Roeschel [33], Eremenko et al. [34], and Boriskina and Kornilov [35]. By using the experimental data presented by Eremenko et al. [24], Kaufman and Bernstein [36] and Mazhuga et al. [37] assessed this system in which the RuTi phase was treated as a kind of stoichiometric compound. Then, the Ru-Ti binary system was re-assessed by Gao et al. [38] based on the experimental data [33][34][35]. The solution phases were modeled with the Redlich-Kister equation and Kaptay equation. Then, the compound RuTi was described by a two-sublattice model (Ru,Ti)0.5(Ru,Ti)0.5. Figure 1c shows the evaluated Ru-Ti phase diagram.  [29]; (b) Ni−Ru system presented by Hallström [32]; (c) Ru−Ti system presented by Gao et al. [38].

Ni−Ti−Ru Ternary System
The Ni−Ti−Ru ternary system has been investigated in a few studies [15,[39][40][41]. NiTi and RuTi with a CsCl-type structure are both isostructural phases, and their lattice parameters are also very similar. However, Boriskina [39] found a two-phase region between NiTi and RuTi, and they determined that the following peritectic reaction occurred at 1570 °C: Liq. (liquid phase) + RuTi ↔ NiTi. Eremenko et al. [40] doubted these results because the alloys were annealed at 1200 °C for only 25 h, making it difficult for the alloys to reach phase equilibrium; the authors suggested that there was an isomorphous-type pseudo bi-  [29]; (b) Ni-Ru system presented by Hallström [32]; (c) Ru-Ti system presented by Gao et al. [38].

Ni-Ti-Ru Ternary System
The Ni-Ti-Ru ternary system has been investigated in a few studies [15,[39][40][41]. NiTi and RuTi with a CsCl-type structure are both isostructural phases, and their lattice parameters are also very similar. However, Boriskina [39] found a two-phase region between NiTi and RuTi, and they determined that the following peritectic reaction occurred at 1570 • C: Liq. (liquid phase) + RuTi ↔ NiTi. Eremenko et al. [40] doubted these results because the alloys were annealed at 1200 • C for only 25 h, making it difficult for the alloys to reach phase equilibrium; the authors suggested that there was an isomorphous-type pseudo binary system with a possible phase separation at lower temperatures. To verify the hypothesis, Semenova et al. [41] re-studied the NiTi-RuTi system by using the equilibrium alloy method [41] and differential thermal analysis (DTA). The results indicated that the NiTi-RuTi system was a simple isomorphous system, and a reasonably wide gap existed between the liquidus and solidus lines. In this case, the alloys had dendritic structures easily produced during solidification after smelting, and because the removal of this dendritic structure is difficult within only 25 h of annealing, the alloys in the earlier study [39] possibly appeared to have a two-phase equilibrium. Since the investigation of Semenova et al. [41] was not carried out below 1200 • C, it is not yet certain whether phase separation will occur in the NiTi-RuTi system at low temperatures.

Experimental Methods
The phase relationships in the Ni-Ti-Ru ternary system at 1150 • C were experimentally determined by using the equilibrated alloy method [41] and diffusion couples method [50]. Samples were prepared from 0.5 g each of 99.99% pure Ni particles, Ti particles, and Ru blocks. Each sample was accurately weighed with an electronic balance and determined to be 0.5 g in total. The alloys were prepared by arc melting with flowing argon on a water-cooled copper crucible and remelted at least three times to promote homogeneity. As Ti in the alloy is prone to oxidation which results in alloy composition deviation, pure Ti was smelted once before smelting the alloy to eliminate the possible residual oxygen in the melting furnace.
As-cast alloys were prepared to study the liquidus surface projection. The reaction temperatures of the as-cast alloys were determined by 404F3 DSC with Al 2 O 3 crucibles and with heating and cooling rates of 10 • C/min.
The alloy specimens for the isothermal section were vacuum encapsulated in quartz tubes. The B1 and B2 alloys containing the liquid phase were annealed at 1150 • C for  36 h, while the other alloys were annealed for 1080 h. Finally, the sample capsules were quenched in water at 0 • C to retain the equilibria microstructures at 1150 • C.
The diffusion couples were fabricated across the Ni-Ti-Ru ternary system at 1150 • C. The single-phase alloy (τ1) and two-phase alloy (Ni 3 Ti + γ(Ni)) were sectioned into blocks to fabricate the diffusion couples. Each block was approximately 4 mm, 4 mm, and 3 mm in length, width, and thickness, respectively. The diffusion couples were vacuum encapsulated in quartz tubes and annealed at 1150 • C for 168 h.
All the samples were carefully ground and polished using standard metallographic preparation methods, and the 0.25 mm Al 2 O 3 paste was used in the final polishing step. A chemical etching solution (10% HF + 20% HNO 3 + 70% H 2 O) was used to reveal the microstructure of samples with no obvious atomic number contrast. The microstructure of each sample was examined by SEM (EVO MA10) with EDS to study the morphology and chemical composition of various phases in the samples. The acceleration voltage of 20 kV was applied. The chemical composition of each phase was measured at least thrice. After evaluation and the removal of unreliable data, the average values of the remaining reliable data were accepted and used for further analysis. To further confirm the phases of the samples, XRD analysis was performed using a Rigaku Ultimate IV operating at 100 mA and 40 kV with Cu Kα radiation. The diffraction spectra were collected from 20 • to 100 • with a scan step of 0.02 • .

Isothermal Section at 1150 • C
The constituent phases and compositions of annealing samples are listed in Table 1

Isothermal Section at 1150 °C
The constituent phases and compositions of annealing samples are listed in Table 1, and some typical SEM micrographs and XRD patterns are shown in  As shown in the SEM micrographs of the alloys B1 and B2 (Figure 2a,c), the same typical perieutectic or peritectic microstructures are observed in the alloys, and the formation of these microstructures is due to the phase transformation of the liquid phase during the quenching process. The liquid phase in the alloy B1 is transformed into the NiTi2 phase that exists at low temperatures. Further, in agreement with the XRD patterns in Figure 2b, the three-phase equilibrium of Liq.1 + β(Ti) + (Ni,Ru)Ti can be identified, and the maximum Ru solubility of Liq.1 is 1.85 at.%. In the case of B2, a small amount of dark gray NiTi2 is also precipitated in the final stage of solidification. The two-phase region of Liq.1 + (Ni,Ru)Ti is determined with the SEM micrographs and XRD patterns in Figure 2c and 2d, respectively.  The SEM micrograph of the alloy B6 in Figure 4a shows a small δ(Ru) miscibility gap (marked as δ(Ru) and δ(Ni,Ru) in this isothermal section). Raub and Menzel [51] studied the Ni solubility of δ(Ru) in the Ni−Ru binary system at different temperatures corresponding to the crystal structure parameters of δ(Nix,Ru1−x). The results [51] showed that with the increase in the temperature, the Ni solubility of δ(Ru) gradually increased and the lattice parameter "a" decreased from 2.691 to 2.612. Thus, the change in the lattice parameter will lead to a certain shift of X-ray characteristic peaks, which cannot be ignored in this study. The XRD patterns of the alloy B6 are shown in Figure 4b: the characteristic peaks of δ(Ni,Ru) are similar to those of δ(Ru) owing to the same HCP-A3 crystal structure, but the deviation is significant. In this study, the δ(Ru) phase is inferred to have undergone spinodal decomposition; in other words, δ(Ru) decomposed into two phases with the same crystal structure but different components. However, this hypothesis needs to be verified by performing more in-depth research. Due to the low content of the δ(Ru) phase in B6, the XRD patterns do not show the peaks of the δ(Ru) phase. In addition, it is uncertain whether the miscibility gap between δ(Ru) and δ(Ni,Ru) exists in the area with a higher Ti content.   The SEM micrograph of the alloy B6 in Figure 4a shows a small δ(Ru) miscibility gap (marked as δ(Ru) and δ(Ni,Ru) in this isothermal section). Raub and Menzel [51] studied the Ni solubility of δ(Ru) in the Ni−Ru binary system at different temperatures corresponding to the crystal structure parameters of δ(Nix,Ru1−x). The results [51] showed that with the increase in the temperature, the Ni solubility of δ(Ru) gradually increased and the lattice parameter "a" decreased from 2.691 to 2.612. Thus, the change in the lattice parameter will lead to a certain shift of X-ray characteristic peaks, which cannot be ignored in this study. The XRD patterns of the alloy B6 are shown in Figure 4b: the characteristic peaks of δ(Ni,Ru) are similar to those of δ(Ru) owing to the same HCP-A3 crystal structure, but the deviation is significant. In this study, the δ(Ru) phase is inferred to have undergone spinodal decomposition; in other words, δ(Ru) decomposed into two phases with the same crystal structure but different components. However, this hypothesis needs to be verified by performing more in-depth research. Due to the low content of the δ(Ru) phase in B6, the XRD patterns do not show the peaks of the δ(Ru) phase. In addition, it is uncertain whether the miscibility gap between δ(Ru) and δ(Ni,Ru) exists in the area with a higher Ti content.  There are two typical three-phase regions of (Ni,Ru)Ti + τ1 + δ(Ru) and γ(Ni) + τ1 + δ(Ni,Ru) observed in the alloys B7 and B8 (Figure 5a,c), which are in good agreement with XRD patterns (Figure 5b and 5d, respectively). According to the SEM micrograph and XRD patterns shown in Figure 6a and 6b, respectively, the Ni3Ti, (Ni,Ru)Ti and τ1 phases can be identified clearly in B9. The homogeneity ranges of the τ1 phase are determined to be from 11.06 to 28.19 at.% Ru and 19.69 to 25.14 at.% Ti.     The SEM micrograph and composition profiles of the τ1 and Ni3Ti + γ(Ni) diffusion couples are given in Figure 7. The total range of the interdiffusion zone is approximately 40-50 μm, and the Ni3Ti + γ (Ni) two-phase region gradually transforms into a thick layer Ni3Ti-(Ru-poor). Further, a continuous layer of Ni3Ti-(Ru-rich) of approximately 15-25 μm is formed between the Ni3Ti-(Ru-poor) and τ1. This shows that the flux of Ni atoms causes diffusion into the τ1 phase, which results in the formation of Ni3Ti-(Ru-poor) and Ni3Ti-(Ru-rich). In Figure 7a, the grain boundary between Ni3Ti-(Ru-rich) and τ1 is not perfectly planar. Some Kirkendall voids appear at the boundary between Ni3Ti-(Ru-rich) and Ni3Ti-(Ru-poor) due to interdiffusion. Thus, the diffusion path between τ1 and Ni3Ti + γ(Ni) at 1150 °C is as follows: τ1 single-phase region, Ni3Ti single-phase region, and then the Ni3Ti + γ(Ni) two-phase region. From this result and the two-phase regions τ1 + γ(Ni) and Ni3Ti + γ(Ni) found in B14 and B15 alloys, respectively, a three-phase region τ1 + Ni3Ti + γ(Ni) can be inferred.
In addition, the B10 and B13 alloys support the two-phase regions of (Ni,Ru)Ti + τ1 and δ(Ni,Ru) + τ1, respectively. The B11 and B12 alloys are all located in the two-phase region of δ(Ni,Ru) + γ(Ni).

Liquidus Surface Projection
In this work, 31 as-cast alloys were prepared to determine the liquidus surface projection of the Ni−Ti−Ru ternary system, and the alloy compositions and primary solidification phases are listed in Table 2.
The alloys C3 and C4 are located in the primary solidification region δ(Ru) (Figure  8a,b). Notably, the δ(Ru) phase shows a miscibility gap due to the spinodal decomposition that is discussed in Section 4.1. According to the SEM micrograph and XRD patterns of the alloy C3, the gray phase surrounding δ(Ru) is identified as the τ1 phase that is generated by the peritectic reaction: liq. + δ(Ru) ↔ τ1. As shown in Figure 8c, in the case of C4, in addition to the phases of C3, a grayish-white phase exists, and it was determined to be the (Ni,Ru)Ti phase on the basis of the XRD patterns (Figure 8d). As shown in the SEM micrographs of the alloys B1 and B2 (Figure 2a,c), the same typical perieutectic or peritectic microstructures are observed in the alloys, and the formation of these microstructures is due to the phase transformation of the liquid phase during the quenching process. The liquid phase in the alloy B1 is transformed into the NiTi 2 phase that exists at low temperatures. Further, in agreement with the XRD patterns in Figure 2b, the three-phase equilibrium of Liq.1 + β(Ti) + (Ni,Ru)Ti can be identified, and the maximum Ru solubility of Liq.1 is 1.85 at.%. In the case of B2, a small amount of dark gray NiTi 2 is also precipitated in the final stage of solidification. The two-phase region of Liq.1 + (Ni,Ru)Ti is determined with the SEM micrographs and XRD patterns in Figure 2c and 2d, respectively.
In Figure 3, the SEM micrographs and XRD patterns of alloys B3, B4, and B5 confirm the presence of the two-phase region of (Ni,Ru)Ti + Ni 3 Ti. The two-phase regions of Liq.2 + NiTi and Liq.2 + Ni 3 Ti exist at 1150 • C according to the Ni-Ti binary system presented by Keyzer et al. [29]. Therefore, a narrow three-phase equilibrium region of (Ni,Ru)Ti + Ni 3 Ti + Liq.2 must be present in the Ni-Ti-Ru ternary system.
The SEM micrograph of the alloy B6 in Figure 4a shows a small δ(Ru) miscibility gap (marked as δ(Ru) and δ(Ni,Ru) in this isothermal section). Raub and Menzel [51] studied the Ni solubility of δ(Ru) in the Ni-Ru binary system at different temperatures corresponding to the crystal structure parameters of δ(Ni x ,Ru 1−x ). The results [51] showed that with the increase in the temperature, the Ni solubility of δ(Ru) gradually increased and the lattice parameter "a" decreased from 2.691 to 2.612. Thus, the change in the lattice parameter will lead to a certain shift of X-ray characteristic peaks, which cannot be ignored in this study. The XRD patterns of the alloy B6 are shown in Figure 4b: the characteristic peaks of δ(Ni,Ru) are similar to those of δ(Ru) owing to the same HCP-A3 crystal structure, but the deviation is significant. In this study, the δ(Ru) phase is inferred to have undergone spinodal decomposition; in other words, δ(Ru) decomposed into two phases with the same crystal structure but different components. However, this hypothesis needs to be verified by performing more in-depth research. Due to the low content of the δ(Ru) phase in B6, the XRD patterns do not show the peaks of the δ(Ru) phase. In addition, it is uncertain whether the miscibility gap between δ(Ru) and δ(Ni,Ru) exists in the area with a higher Ti content.
There are two typical three-phase regions of (Ni,Ru)Ti + τ1 + δ(Ru) and γ(Ni) + τ1 + δ(Ni,Ru) observed in the alloys B7 and B8 (Figure 5a,c), which are in good agreement with XRD patterns (Figure 5b and 5d, respectively). According to the SEM micrograph and XRD patterns shown in Figure 6a and 6b, respectively, the Ni 3 Ti, (Ni,Ru)Ti and τ1 phases can be identified clearly in B9. The homogeneity ranges of the τ1 phase are determined to be from 11.06 to 28.19 at.% Ru and 19.69 to 25.14 at.% Ti.
The SEM micrograph and composition profiles of the τ1 and Ni 3 Ti + γ(Ni) diffusion couples are given in Figure 7. The total range of the interdiffusion zone is approximately 40-50 µm, and the Ni 3 Ti + γ (Ni) two-phase region gradually transforms into a thick layer Ni 3 Ti-(Ru-poor). Further, a continuous layer of Ni 3 Ti-(Ru-rich) of approximately 15-25 µm is formed between the Ni 3 Ti-(Ru-poor) and τ1. This shows that the flux of Ni atoms causes diffusion into the τ1 phase, which results in the formation of Ni 3 Ti-(Ru-poor) and Ni 3 Ti-(Ru-rich).
In Figure 7a, the grain boundary between Ni 3 Ti-(Ru-rich) and τ1 is not perfectly planar. Some Kirkendall voids appear at the boundary between Ni 3 Ti-(Ru-rich) and Ni 3 Ti-(Rupoor) due to interdiffusion. Thus, the diffusion path between τ1 and Ni 3 Ti + γ(Ni) at 1150 • C is as follows: τ1 single-phase region, Ni 3 Ti single-phase region, and then the Ni 3 Ti + γ(Ni) two-phase region. From this result and the two-phase regions τ1 + γ(Ni) and Ni 3 Ti + γ(Ni) found in B14 and B15 alloys, respectively, a three-phase region τ1 + Ni 3 Ti + γ(Ni) can be inferred.
In addition, the B10 and B13 alloys support the two-phase regions of (Ni,Ru)Ti + τ1 and δ(Ni,Ru) + τ1, respectively. The B11 and B12 alloys are all located in the two-phase region of δ(Ni,Ru) + γ(Ni).

Liquidus Surface Projection
In this work, 31 as-cast alloys were prepared to determine the liquidus surface projection of the Ni-Ti-Ru ternary system, and the alloy compositions and primary solidification phases are listed in Table 2.
The alloys C3 and C4 are located in the primary solidification region δ(Ru) (Figure 8a,b). Notably, the δ(Ru) phase shows a miscibility gap due to the spinodal decomposition that is discussed in Section 4.1. According to the SEM micrograph and XRD patterns of the alloy C3, the gray phase surrounding δ(Ru) is identified as the τ1 phase that is generated by the peritectic reaction: liq. + δ(Ru) ↔ τ1. As shown in Figure 8c, in the case of C4, in addition to the phases of C3, a grayish-white phase exists, and it was determined to be the (Ni,Ru)Ti phase on the basis of the XRD patterns (Figure 8d).
The alloy C6 is located in the primary solidification region δ(Ru) (Figure 9a,b), and this alloy does not undergo the expected perieutectic reaction: liq. + δ(Ru) ↔ τ1 + γ(Ni); instead, the remaining liquid phase solidifies into γ(Ni) in the peritectic reaction: liq. + δ(Ru) ↔ γ(Ni). According to the SEM micrograph and XRD patterns of the alloy C15 (Figure 9c,d), a large amount of the dendritic primary (Ni,Ru)Ti phase can be observed. The typical eutectic microstructure of this alloy is γ(Ni) + τ1, with the (Ni,Ru)Ti phase enclosed by τ1, which is formed by the peritectic reaction. In addition, some amount of lathy sub-primary solidification phase τ1 also exists. The alloy C6 is located in the primary solidification region δ(Ru) (Figure 9a,b), and this alloy does not undergo the expected perieutectic reaction: liq. + δ(Ru) ↔ τ1 + γ(Ni); instead, the remaining liquid phase solidifies into γ(Ni) in the peritectic reaction: liq. + δ(Ru) ↔ γ(Ni). According to the SEM micrograph and XRD patterns of the alloy C15 (Figure 9c,d), a large amount of the dendritic primary (Ni,Ru)Ti phase can be observed. The typical eutectic microstructure of this alloy is γ(Ni) + τ1, with the (Ni,Ru)Ti phase enclosed by τ1, which is formed by the peritectic reaction. In addition, some amount of lathy subprimary solidification phase τ1 also exists.
Among the series of alloys with the primary solidification phase τ1, the alloy C14 (Figure 10a,b) was considered for the subsequent analyses. Its SEM microstructure is similar to that of the alloy C15, except that the (Ni,Ru)Ti phase is missing. Therefore, the solidification path of the alloy C14 is simple: liq. ↔ τ1, liq. ↔ γ(Ni) + τ1. In addition, Figure  10c shows the primary phase γ(Ni) of the alloy C26 during solidification.  Among the series of alloys with the primary solidification phase τ1, the alloy C14 (Figure 10a,b) was considered for the subsequent analyses. Its SEM microstructure is similar to that of the alloy C15, except that the (Ni,Ru)Ti phase is missing. Therefore, the solidification path of the alloy C14 is simple: liq. ↔ τ1, liq. ↔ γ(Ni) + τ1. In addition, Figure 10c shows the primary phase γ(Ni) of the alloy C26 during solidification.  Based on the XRD patterns and SEM micrograph of the alloy C10 (Figure 11), a few primary solidification phases of (Ni,Ru)Ti and eutectic microstructure Ni3Ti + (Ni,Ru)Ti are identified. In the final stage of solidification, some amount of eutectic microstructure Ni3Ti + NiTi formed at the grain boundary, due to the binary eutectic reaction: liq. ↔ Ni3Ti + NiTi. The alloy C30 is located in the Ni3Ti primary solidification region, while the alloy C31 is located in the τ1 region. From the SEM micrograph and XRD patterns of alloy C30 shown in Figure 12a and 12b, respectively, the light gray phase is determined to be the Based on the XRD patterns and SEM micrograph of the alloy C10 (Figure 11), a few primary solidification phases of (Ni,Ru)Ti and eutectic microstructure Ni 3 Ti + (Ni,Ru)Ti are identified. In the final stage of solidification, some amount of eutectic microstructure Ni 3 Ti + NiTi formed at the grain boundary, due to the binary eutectic reaction: liq. ↔ Ni 3 Ti + NiTi. The alloy C30 is located in the Ni 3 Ti primary solidification region, while the alloy C31 is located in the τ1 region. From the SEM micrograph and XRD patterns of alloy C30 shown in Figure 12a and 12b, respectively, the light gray phase is determined to be the primary solidification phase Ni 3 Ti. The solidification path of alloy C30 is inferred to be as follows: liq. ↔ Ni 3 Ti, liq. ↔ Ni 3 Ti + γ(Ni). The SEM micrograph and XRD patterns of alloy C31 (Figure 12c and 12d, respectively) confirm the existence of the τ1, (Ni,Ru)Ti, and Ni 3 Ti phases, and the solidification path of the alloy C31 is inferred to be as follows: liq. ↔ τ1, liq. + τ1 ↔ (Ni,Ru)Ti + Ni 3 Ti. A comparison of the difference in the paths between alloys C30 and C31 indicate that Ni 3 Ti and τ1 phases can be clearly distinguished and that the existence of a maximum point e max on the univariate line (liq. ↔ Ni 3 Ti + τ1) can be inferred. primary solidification phase Ni3Ti. The solidification path of alloy C30 is inferred to be as follows: liq. ↔ Ni3Ti, liq. ↔ Ni3Ti + γ(Ni). The SEM micrograph and XRD patterns of alloy C31 (Figure 12c and 12d, respectively) confirm the existence of the τ1, (Ni,Ru)Ti, and Ni3Ti phases, and the solidification path of the alloy C31 is inferred to be as follows: liq. ↔ τ1, liq. + τ1 ↔ (Ni,Ru)Ti + Ni3Ti. A comparison of the difference in the paths between alloys C30 and C31 indicate that Ni3Ti and τ1 phases can be clearly distinguished and that the existence of a maximum point emax on the univariate line (liq. ↔ Ni3Ti + τ1) can be inferred. Figure 11. SEM micrograph (a) and XRD patterns (b) of as-cast alloy C10. Figure 11. SEM micrograph (a) and XRD patterns (b) of as-cast alloy C10. Figure 11. SEM micrograph (a) and XRD patterns (b) of as-cast alloy C10. The SEM micrograph and XRD patterns of the alloy C12 are shown in Figure 13a and 13b, respectively. The primary solidification phase of (Ni,Ru)Ti is observed in C12. Obviously, dendritic (Ni,Ru)Ti first precipitates from the liquid phase, and then the peritectic reaction liq. + (Ni,Ru)Ti ↔ β(Ti) occurs. Finally, the remaining liquid phase undergoes binary eutectic reaction liq. ↔ β(Ti) + NiTi2. The SEM micrograph and XRD patterns of the alloy C13 (Figure 13c and 13d, respectively) show that its primary solidification phase is β(Ti). The eutectic microstructure of the alloy C13 is β(Ti) + NiTi2, which is formed by the same eutectic reaction liq. ↔ β(Ti) + NiTi2 (like alloy C12).
According to the SEM micrograph and XRD patterns of the alloy C11, shown in Figure 14a and 14b, respectively, there is a typical eutectic microstructure β(Ti) + NiTi2. The The SEM micrograph and XRD patterns of the alloy C12 are shown in Figure 13a and 13b, respectively. The primary solidification phase of (Ni,Ru)Ti is observed in C12. Obviously, dendritic (Ni,Ru)Ti first precipitates from the liquid phase, and then the peritectic reaction liq. + (Ni,Ru)Ti ↔ β(Ti) occurs. Finally, the remaining liquid phase undergoes binary eutectic reaction liq. ↔ β(Ti) + NiTi 2 . The SEM micrograph and XRD patterns of the alloy C13 (Figure 13c and 13d, respectively) show that its primary solidification phase is β(Ti). The eutectic microstructure of the alloy C13 is β(Ti) + NiTi 2 , which is formed by the same eutectic reaction liq. ↔ β(Ti) + NiTi 2 (like alloy C12).   According to the SEM micrograph and XRD patterns of the alloy C11, shown in Figure 14a and 14b, respectively, there is a typical eutectic microstructure β(Ti) + NiTi 2 .
The SEM micrograph and XRD patterns of the alloy C23 are shown in Figure 14c and 14d, respectively: a perieutectic reaction liq. + (Ni,Ru)Ti ↔ β(Ti) + NiTi 2 and eutectic reaction liq. ↔ β(Ti) + NiTi 2 occur in the alloy.  According to the above experimental results of the primary phases and solidification structures of the as-cast alloys, the solidification paths of the as-cast alloys not presented in the discussion so far are as follows: According to the above experimental results of the primary phases and solidification structures of the as-cast alloys, the solidification paths of the as-cast alloys not presented in the discussion so far are as follows: Alloy C3: liq. ↔ δ(Ru), liq. + δ(Ru) ↔ τ1, liq. + δ(Ru) ↔ τ1 + γ(Ni), liq. ↔ τ1 + γ(Ni); Alloy C4: liq. ↔ δ(Ru), liq. ↔ δ(Ru) + (Ni,Ru)Ti, liq. + (Ni,Ru)Ti + δ(Ru) ↔ τ1, liq. + δ(Ru) ↔ τ1, liq. + δ(Ru) ↔ τ1 + γ(Ni), liq. ↔ τ1 + γ(Ni);  Figure 15. Two peaks starting at 1404 • C on heating and 1398 • C on cooling in Figure 15a correspond to the reaction U1 (liq. + δ(Ru) ↔ τ1 + γ(Ni)). Further, two peaks starting at 1339 • C on heating and 1336 • C on cooling correspond to the eutectic reaction (liq. ↔ τ1 + γ(Ni)). Figure 15b shows the peak starting at 984 • C on heating of the alloy C23 which corresponds to the reaction U4 (liq. + (Ni,Ru)Ti ↔ β(Ti) + NiTi 2 ). In addition, due to the volatilization of Ni at high temperatures, the composition of the C23 alloy changes and the eutectic reaction (liq. ↔ β(Ti) + NiTi 2 ) occurs in the remaining liquid phase, corresponding to the peak starting at 932 • C on cooling. The peaks starting at 1384 • C and 1314 • C on heating and at 1383 • C and 1304 • C on cooling of the alloy C31 (Figure 15c) correspond to the three-phase eutectic reactions (liq. ↔ Ni 3 Ti + τ1) and (liq. ↔ (Ni,Ru)Ti + Ni 3 Ti), respectively. Furthermore, the reaction temperature of P1 and e max are inferred as being about 1500 • C and 1390 • C, respectively. Notably, only the peak starting at 1337 • C on heating corresponding to the reaction U3 (liq. + τ1 ↔ (Ni,Ru)Ti + Ni 3 Ti) is found in the curves ( Figure 15c). This may also be caused by the volatilization of the element. Further, it can be inferred that the reaction temperature of U2 should be about 1330 • C. reaction U4 (liq. + (Ni,Ru)Ti ↔ β(Ti) + NiTi2). In addition, due to the volatilization of Ni at high temperatures, the composition of the C23 alloy changes and the eutectic reaction (liq. ↔ β(Ti) + NiTi2) occurs in the remaining liquid phase, corresponding to the peak starting at 932 °C on cooling. The peaks starting at 1384 °C and 1314 °C on heating and at 1383 °C and 1304 °C on cooling of the alloy C31 (Figure 15c) correspond to the three-phase eutectic reactions (liq. ↔ Ni3Ti + τ1) and (liq. ↔ (Ni,Ru)Ti + Ni3Ti), respectively. Furthermore, the reaction temperature of P1 and emax are inferred as being about 1500 °C and 1390 °C, respectively. Notably, only the peak starting at 1337 °C on heating corresponding to the reaction U3 (liq. + τ1 ↔ (Ni,Ru)Ti + Ni3Ti) is found in the curves (Figure 15c). This may also be caused by the volatilization of the element. Further, it can be inferred that the reaction temperature of U2 should be about 1330 °C. According to the above experimental results and discussion, the isothermal section of the Ni−Ti−Ru ternary system at 1150 °C is presented in Figure 16. The invariant reaction According to the above experimental results and discussion, the isothermal section of the Ni-Ti-Ru ternary system at 1150 • C is presented in Figure 16. The invariant reaction temperatures are shown in Table 3. The predicted liquidus surface projection and complete reaction scheme are shown in Figures 17 and 18 temperatures are shown in Table 3. The predicted liquidus surface projection and complete reaction scheme are shown in Figures 17 and 18, respectively.

Conclusions
The isothermal section at 1150 °C and the liquidus surface projection of the Ni−Ti−Ru ternary system have been established by using SEM/EDS, XRD, and DSC analyses.

Conclusions
The isothermal section at 1150 • C and the liquidus surface projection of the Ni-Ti-Ru ternary system have been established by using SEM/EDS, XRD, and DSC analyses.
The addition of Ti seems to result in a small δ(Ru) miscibility gap due to spinodal decomposition. A new ternary compound τ1 is identified based on the present study on both the isothermal section and liquidus surface projection.