Erosion Resistance of BaZrO₃-Y₂O₃ Two-Phase Crucibles against Highly Active Ti₂Ni Alloys

3.1. Phase and Microstructure Analysis of the Crucibles Figure 2 shows the XRD patterns of BZ~Y100 crucibles after sintering. As can be seen in Figure 2a, the BZ crucible consisted of cubic BaZrO₃ (ICDD PDF# 00-006-0399) and tetragonal ZrO₂ (ICDD PDF# 00-080-2155) phases. Our previous study [23] revealed that the volatilization of BaO during the preparation of BaZrO₃, resulted in the residual of the second phase ZrO₂. When the adding amount of Y₂O₃ addition ranged from 20~50 wt.% (Y20~Y50), the ZrO₂ phase disappeared. In addition, these crucibles consisted of BaZrO₃ and cubic Y₂O₃ (ICDD PDF# 00-043-1036), respectively. From Figure 2b, it can be seen that the characteristic peaks of BaZrO₃ and Y₂O₃ were shifted to lower and higher angles in the Y20~Y50 crucibles, respectively. It is because Zr⁴⁺ (0.72 Å) ions in BaZrO₃ crystal lattice were partially replaced by Y³⁺ (0.90 Å) ions, and Y³⁺ ions in Y₂O₃ crystal lattice were partially substituted by Zr⁴⁺ ions, which could be further verified using the following formulae [25]: where d is the interplanar spacing; λ is the X-ray wavelength; a is the lattice constant of the cubic structure; h, k and l are the Miller indices corresponding to the diffraction peak. Combining Equations (1) and (2) shows that the substitution of Y³⁺ ions for Zr⁴⁺ ions led to an increase in the lattice constant a, and an accompanying increase in the interplanar spacing d, which finally resulted in a decrease in θ, i.e., a leftward shift of diffraction peak. The space group of cubic BaZrO₃ is Pm-3m, and Ba²⁺ ions with coordination number 12 and Zr⁴⁺ ions with coordination number 6 correspond to ionic radii of 1.61 Å and 0.72 Å, respectively [26]. The space group of cubic Y₂O₃ is Ia-3, and the ionic radius corresponding to Y³⁺ ions with coordination number 6 is 0.90 Å [26]. The relative ionic radius difference between Zr⁴⁺ and Y³⁺ ions is about 13%, which is less than 15%, indicating that they can be substituted for each other [27]. Figure 3 shows the inner microstructure of BZ~Y100 crucibles after grinding and polishing. From Figure 3a, it can be seen that the BZ crucible consisted of light gray (spot 1) and gray (spot 2) substances with the appearance of some pores, respectively. EDS results in Table 2 indicated that spots 1 and 2 were BaZrO₃ and ZrO₂ phases, respectively. After adding 20 wt.% Y₂O₃ into BaZrO₃ (Y20 crucible, Figure 3b), the amount of gray substance (spot 4), which was dispersed throughout the crucible matrix, exhibited a significant increase trend and became Y₂O₃ with ZrO₂ in solid solution. EDS results indicated spots 3 and 4 were BaZr_1−xY_xO_3−δ and Y₂O₃(ZrO₂), respectively. According to the phase diagram of ZrO₂-Y₂O₃ system [28] shown in Figure 4a, at 1750 °C, the solubility limit of ZrO₂ in cubic Y₂O₃ is 12.71 at. % (atomic percent) and the solubility limit of Y₂O₃ in tetragonal ZrO₂ is 4.25 at. %. At the same time, the cubic solid solution formed by cubic ZrO₂ and cubic Y₂O₃ ranges from 11.11 at. % to 67.62 at. %. The values of Y/(Zr + Y) (spots 4, 6, 8, and 10) in Table 2 are basically consistent with calculated results in Figure 4a. According to the isothermal section of the BaO-ZrO₂-Y₂O₃ system [28] at 1750 °C (Figure 4b), the solubility limit of Y₂O₃ in BaZrO₃ is 24.43 at. %. The values of Ba/(Zr + Y) and Y/(Zr + Y) (spots 3, 5, and 7) in Table 2 are fully consistent with calculated results in Figure 4b. For the EDS results of spot 9, the values of Ba/(Zr + Y) and Y/(Zr + Y) differ significantly from other results (spots 3, 5, and 7), which is considered to be due to measurement errors. It was caused by the solid solution of Y₂O₃ into BaZrO₃ and ZrO₂ into Y₂O₃, which was consistent with the shifts of observed characteristic peaks in Figure 2b. In addition, through comparison, it can be found that the number of pores in Figure 3b is significantly less than that in Figure 3a. Apparently, the Y₂O₃ addition had two obvious effects. One was to consume the ZrO₂ phase in the crucible, leading to the formation of a more stable Y₂O₃(ZrO₂) phase according to the solid solution effect [24]. The other was to increase the densification of the crucible due to the effect of solid solution sintering. As the amount of Y₂O₃ addition increased from 25~50 wt.% (Y25~Y50 crucibles), the different grains were evenly distributed, and the Y₂O₃(ZrO₂) phase obviously increased with the growth of Y₂O₃ additions, as shown in Figure 3c–e. In addition, the number of pores in crucibles was further decreased, indicating that an increased amount of Y₂O₃ addition could continue to improve the densification of BaZrO₃-Y₂O₃ two-phase crucibles. From Figure 3f, it can be seen that only Y₂O₃ phase (spot 11) could be observed, and a large number of pores were dispersed in the Y100 crucible. The solid solution between Y₂O₃, ZrO₂ and BaZrO₃ during sintering could be described as follows: To further investigate the effect of Y₂O₃ additions on the sintering behavior of BaZrO₃-Y₂O₃ two-phase crucibles, ImageJ software was used to analyze the percentage of pore areas, as shown in Figure 5. Three pictures of each crucible sample were measured and the average percentage of pore areas was calculated. From Figure 5a, it can be seen that the percentage of pore areas in BZ crucible reached 18.2 ± 2.1%. It is because the BaZrO₃ refractory with a high melting point (nearly 2650 °C) was difficult to densify at 1750 °C [23]. Additionally, the presence of ZrO₂ secondary phase partially inhibited the growth of BaZrO₃ grains, leading to the residual of pores. After adding Y₂O₃, from Figure 5b–e, it can be seen that the percentage of pore area in Y20~Y50 crucibles exhibited a significant decrease trend from 3.5 ± 0.6% to 1.2 ± 0.3%, respectively. The lower porosity could have a good effect to resist the erosion of alloy melts. In the BaO-Y₂O₃ phase diagram [28], all eutectic reaction temperatures including the liquid phase are greater than 1750 °C. When sintered at 1750 °C, Y³⁺ replaces Zr⁴⁺ in BaZrO₃, resulting in oxygen vacancies [29]. These factors have led to the decrease in the percentage of pore area. For the Y100 crucible, its percentage of pore area also reached 8.2 ± 1.1%, as shown in Figure 5f. Apparently, the melting point of Y₂O₃ (2410 °C) [23] was lower than that of BaZrO₃ (2650 °C), indicating that the sintering performance of Y₂O₃ was better than that of BaZrO₃ under the same conditions. 3.2. Interaction between the Crucibles and Ti₂Ni Alloy Melts The phases of the inner surfaces of BZ~Y100 crucibles after the melting of Ti₂Ni alloys were characterized by XRD, as shown in Figure 6. For the BZ crucible, not only no new phase was observed, but also the disappearance of ZrO₂ characteristic peaks. The glassy phase is possible, but not easily detected by XRD, which is mainly sensitive to crystalline phases. Chen et al. [30] proposed that the interaction of refractory crucible with Ti melt was caused by physical dissolution of the refractory, it could be confirmed that the Ti₂Ni alloy melt dissolved the ZrO₂ refractory on the surface of the BaZrO₃ crucible. The loose and non-uniform structure of the crucible surface is due to the partial dissolution of the BaZrO₃ refractory. For the Y100 crucible, no new phase was observed compared to the original crucible in Figure 2a, indicating that the interaction of Ti₂Ni alloy melt with Y₂O₃ crucible did not generate any new crystalline product but a glassy phase may exist. For the Y20~Y50 crucibles after melting, there was no appearance of new interaction product on their surfaces in comparison with the original crucibles in Figure 2a. Figure 7 shows the cross-sections of the bottom wall of BZ~Y100 crucibles after melting Ti₂Ni alloys. As shown in Figure 7a, the internal structure of the BZ crucible was damaged due to the infiltration of alloy melts. The generated thickness of the erosion layer was about 658 μm. For the Y20 crucible, the thickness of the erosion layer had a significant decrease to 142 μm, as shown in Figure 7b. As the increase of Y₂O₃ content in the BaZrO₃ crucible, the thickness of the erosion layer was further decreased. For Y25~Y50 crucibles (Figure 7c–e), it ranged from 65 to 7 μm. However, for the Y100 crucible (Figure 7f), the thickness of the erosion layer was only 6.5 μm. The magnified pictures of the erosion layers of BZ, Y20, Y50, and Y100 crucibles are shown in Figure 8. From Figure 8(a1), it can be seen that the erosion layer of BZ crucible exhibited a clear layered structure, which showed a loose and porous phenomenon. The corresponding mapping analysis indicated that only Ba and Zr elements exist in the erosion layer, and there is no residue of alloy melts, as shown in Figure 8(a2). In addition, the EDS result of spot A (Table 3) showed that BaO phase was observed. In our previous studies [23], we had revealed that the decomposition of BaZrO₃ refractory in titanium alloy melts could cause the evaporation of BaO and the Zr and O elements into alloy melts. The same phenomenon was also observed in the melted Y20 crucible, as shown in Figure 8(b1,b2). In the Y50 crucible (Figure 8(c1,c2)), the enrichment of BaO disappeared as the decreasing of BaZrO₃ content. In addition, the Zr elements caused by the dissolution of BaZrO₃ would be partially dissolved into Y₂O₃, which will reduce the Zr pollution to the alloy. As can be seen from Figure 8(d1,d2), the Y100 crucible also did not react with alloy melt and was able to resist the mechanical scouring of alloy melt, indicating that Y₂O₃ refractory has good thermodynamic stability. However, the Y100 crucible has an obvious disadvantage, the Y generated by the dissolution of Y₂O₃ reacts with the O in the alloy melt to generate a large number of oxide inclusions [31]. Actually, the purity degree of titanium alloys was an important evaluation criterion for the practical application. Figure 9 shows the number of melted Ti₂Ni alloys using BZ~Y100 crucibles as well as the O, Zr, and Y element concentrations in each alloy. Figure 9a shows the change trend of O concentration. It can be seen that the O concentration exhibited a decreasing trend as the increase of melting number regardless of the type of crucibles. However, it was more obvious that the BaZrO₃ crucibles with Y₂O₃ addition exhibited a rapid decrease trend for the O concentration. It consisted of the analysis for the erosion layer in Figure 7 and Figure 8. Especially, the alloy melted in the Y50 crucible, its O concentration was only 0.11 wt.% after third melting, which is basically the same as that of alloy melted in the Y100 crucible. The change trend of Zr concentration is shown in Figure 9b, it can be seen that the Zr concentration also exhibited the same change trend as the O concentration. After third melting, the Zr concentration in alloy melted with the Y50 crucible was only 0.05 wt.%, which is much lower than 0.64 wt.% in the BZ crucible. This result was in consistent with the analysis of Figure 8, where the Zr element generated by the dissolution of crucibles was partially solidified into Y₂O₃. From Figure 9c, the Y concentration in alloy exhibited a clear increasing trend with the increase of Y₂O₃ addition. The Y concentration in alloy melted in the Y100 crucible increases from 0.10 to 0.24 wt.% with the increase in melting times, which is due to the large amount of Y₂O₃ dissolution [11,12]. After third melting, the Y concentration in alloy melted in Y20~Y50 crucibles ranged from 0.08 to 0.13 wt.%, and the Y50 crucible with the highest Y concentration was only half of the Y100 crucible. Considering the contamination degree from the three elements, the Y50 crucible clearly showed the best melting performance. Figure 10 shows the microstructure of alloys for the third melting in BZ, Y20, Y50, and Y100 crucibles. From Figure 10(a1,a2), the alloy cooled in the BZ crucible contains some gray particles. Combined with EDS results in Table 4 (spots 1–2), the alloy matrix was Ti₂Ni and the gray particles were residual TiNi. As can be seen in Figure 10(b1), some gray and white particles appeared in the alloy cooled in the Y20 crucible. The gray particles appeared as long strips with a tendency to break off, and most of white particles were attached around gray particles. Combined with the Figure 10(b2) and EDS results (spots 3–5) in Table 4, the white particles were Y₂O₃, the alloy matrix was Ti₂Ni, which was consistent with the target composition, and the gray particles were residual TiNi. From the Ni-Ti phase diagram (Figure 11) [32], it can be seen that the Ti₂Ni alloy firstly entered the two-phase region of liquid and TiNi from the liquid-phase region during solidification process, and then underwent the peritectic reaction Liquid + TiNi → Ti₂Ni at 1258 K, and the generated Ti₂Ni was retained until room temperature. It is hypothesized that the peritectic reaction did not occur completely resulting in a small amount of TiNi remaining in solidified Ti₂Ni matrix. From Figure 10(c1,c2) and EDS results (spots 6–8) in Table 4, the same residual TiNi and inclusions Y₂O₃ were present in Ti₂Ni matrix cooled in the Y50 crucible. As the Y₂O₃ content was increased, the residual long strips of TiNi became fewer and finer, while fracture occurred, indicating that Y₂O₃ promoted this peritectic reaction. The amount and morphology of TiNi remaining in the cooled Ti₂Ni matrix in the Y100 crucible (Figure 10(d1,d2)) was basically the same as that of the Y50 crucible. The presence of Y₂O₃ inclusions was caused by the dissolution of Y₂O₃ refractory in melting process, which was facilitated by the Y100 crucible [33]. 3.3. Erosion Resistance Mechanism From the above analysis, it can be seen that BaZrO₃-Y₂O₃ two-phase crucibles had a certain advantage in comparison with the pure BZ and Y100 crucibles. In particular, the Y50 crucible exhibited the best erosion resistance. In the following, the Y50 crucible was taken as an example to explore the mechanism by which it effectively prevents the erosion of alloy melts. With the addition of Y₂O₃, the average pore diameter of the crucible gradually decreases, and the Y50 crucible reaches the minimum value (1.8 ± 0.6 μm), as shown in Figure 5e. On the whole, combined with Figure 5 and Figure 7, the erosion layer thickness decreased with decreasing average pore diameter, which is consistent with the Washburn model for liquid penetration into a capillary [34]: where h is the erosion layer thickness, r is the pore radius, θ is the contact angle, σ is the surface tension, t is the interaction time, and η is the alloy melt viscosity. Although the porosity also decreases with the addition of Y₂O₃, the Y50 crucible also reaches the minimum value (1.2 ± 0.3%), the erosion layer thickness of the Y100 crucible with a porosity of 8.2 ± 1.1% is basically the same as that of the Y50 crucible. This showed that the average pore size has more influence on the erosion layer than porosity. The erosion resistance of crucible was also affected by its thermodynamic stability and thermal shock resistance [9,35]. Figure 12 shows the Ellingham diagram of TiO₂, TiO, ZrO₂, BaZrO₃ and Y₂O₃. At the experimental temperature (1600 °C), the thermodynamic stability of ZrO₂, BaZrO₃ and Y₂O₃ was significantly higher than that of TiO₂ and TiO. Therefore, no chemical reaction occurred between BZ~Y100 crucibles and alloy melts, which was consistent with the experimental results. In BaZrO₃-Y₂O₃ two-phase crucibles, the presence of solid solution effects led to a further decrease in the Gibbs free energy of BaZr_1−xY_xO_3−δ and Y₂O₃(ZrO₂) [23]. Hence, compared to BZ and Y100 crucibles, the Y50 crucible had a higher thermodynamic stability. It should be noted that physical dissolution of crucible refractory was unavoidable due to mechanical scouring of the inner wall of crucible by alloy melts [36]. Generally, crucibles with low porosity and high thermal shock resistance would have a long service life. In addition, high thermal conductivity and a low linear expansion coefficient were necessary to obtain high thermal shock resistance [37]. High thermal conductivity crucibles had smaller thermal gradients, while low linear expansion coefficient crucibles expanded less and deformed more uniformly, thus reducing thermal stresses and preventing the damage of the crucible. The thermal shock resistance of crucibles could be predicted by calculating the thermal stress resistance parameters R, the following equations [38,39]: where σ_f, E, α, and ν represent the strength, Young’s modulus, coefficient of linear expansion and Poisson’s ratio, respectively. R represents the maximum allowable temperature difference. These parameters could be used to evaluate the cracking sensitivity of crucibles after erosion of molten alloys. Table 5 lists the calculated R values of BaZrO₃ and Y₂O₃. All data in Table 5 are derived from Ref. [9]. The σ_f value is cold modulus of rupture value. The α value ranges from room temperature to 1200 °C. It can be seen that BaZrO₃ and Y₂O₃ possessed similar intrinsic physical properties such as strength, Young’s modulus, coefficient of linear expansion and Poisson’s ratio. Therefore, for the Y50 crucible, using the mixing rule, BaZrO₃ and Y₂O₃ were more easily combined with each other. The similar maximum allowable temperature difference (R) ensured that no cracks occurred between BaZrO₃ and Y₂O₃ in the melting process of the Y50 crucible, thus showing a good thermal shock resistance. In conclusion, the excellent thermodynamic stability and thermal shock resistance of the Y50 crucible made it a promising application in VIM of highly reactive titanium alloys.