Fabrication, Properties of Dense CA₆ Refractory and Its Reaction Behavior with Titanium Aluminum Alloy

Titanium aluminum (TiAl) alloys are characterized by their low density, high specific strength and stiffness, excellent corrosion resistance, and remarkable heat and radiation resistance [1,2,3,4,5,6,7,8]. These properties make them highly promising for aerospace applications, particularly in the manufacturing of engine blades, aircraft, and missile components that must endure high-temperature and high-pressure environments. As a result, the melting and processing of TiAl alloys have become a key research focus in the field of superalloys [9,10].

Currently, there are three primary methods for melting titanium alloy: vacuum arc melting (VAR), cold hearth furnace melting (CHM), and vacuum induction melting (VIM). Among these, vacuum induction melting technology offers significant advantages in terms of energy efficiency and production costs. However, a major challenge in the production and application of vacuum induction melting for titanium alloys is the lack of crucible materials with sufficient stability [11]. Consequently, identifying a suitable crucible material is a critical prerequisite for achieving the industrial-scale production of titanium alloys using vacuum induction melting.

Due to the high chemical reactivity of TiAl alloys at elevated temperatures, they tend to react with nearly all known refractory materials. As a result, refractory materials used for melting TiAl alloys must possess adequate mechanical strength, excellent corrosion resistance, and strong thermal shock resistance.

Scholars worldwide have conducted extensive research on crucible materials for the induction melting of TiAl alloys. Commonly studied oxide crucible materials include Al₂O₃, ZrO₂, CaO, Y₂O₃, etc. [12,13,14,15,16]. Zhang H et al. [17] investigated the interfacial reaction between an Al₂O₃ crucible and a TiAl alloy melt under varying directional solidification temperatures and holding times. The formation of a distinct reaction layer (TiO) was observed at the interface, the thickness of which increased with both temperature and holding time. Zhu J et al. [18] studied the interfacial behaviour between Y₂O₃-stabilised ZrO₂ crucibles and titanium alloys, demonstrating that Y₂O₃-stabilised ZrO₂ possesses superior stability compared to CaO-stabilised ZrO₂, although its erosion resistance remains lower than that of pure Y₂O₃ crucibles. Consequently, ZrO₂-based materials are not considered suitable for industrial-scale mass production. Fu B et al. [19] examined the interface reaction between CaO crucibles and Ti-1100 alloy. Their results indicated that CaO crucibles increase the oxygen content in titanium-aluminium alloys during melting. Moreover, CaO is prone to hydration and difficult to sinter, further limiting its industrial applicability. Tetsui T et al. [20] employed yttrium as a crucible material for the induction melting of TiAl alloys, observing the detachment of crucible material into the final alloy and poor thermal shock resistance. Research findings indicate that Al₂O₃ and ZrO₂ crucibles undergo severe interfacial reactions with TiAl alloy melts, resulting in alloy contamination. While CaO and Y₂O₃ exhibit better thermodynamic stability, CaO is prone to hydration and difficult to sinter, limiting its industrial applicability [17,18,19]. Similarly, Y₂O₃ suffers from poor thermal shock resistance and can dissolve into the alloy melt, forming inclusions that significantly degrade the alloy’s performance [20].

CA₆, the calcium aluminate phase with the highest alumina content in the Al₂O₃-CaO system, exhibits excellent hydration resistance, corrosion resistance, low thermal conductivity, and superior mechanical properties. These characteristics have garnered significant attention in recent years [21,22,23,24]. However, its application in the field of TiAl alloy smelting has yet to be explored. Qi et al. [25] proposed the preparation of a CaO-Al₂O₃-TiO₂ crucible and found that the crucibles with lower TiO₂ content demonstrated good resistance to penetration and erosion by titanium-aluminum alloys. In contrast, pure CaO-Al₂O₃ crucibles showed minimal erosion but severe penetration issues. This indicates that while CA₆ exhibits excellent chemical stability, its low density leads to significant penetration problems.

Therefore, a dense CA₆ crucible refractory was prepared using a two-step sintering method for smelting Ti6Al4V alloy. The reaction behavior of CA₆ crucible refractory with different phases and Ti6Al4V alloy was investigated to develop a material with exceptional chemical stability and thermal shock resistance for titanium-aluminum alloy smelting.

The main raw materials in this study are: industrial Al₂O₃ powder (median particle size: 4–5 μm, purity ≥ 99%, Jiangsu Jingxin New Materials Co., Ltd., Yangzhou, China), nano-CaCO₃ (median particle size: 60–80 nm, purity ≥ 96%, Zhejiang Changshan Jinxiong Co., Ltd., Quzhou, China), and TiO₂ (median particle size: ~10 μm, purity ≥ 99%, Shanghai McLean Biochemical Technology Co., Ltd., Shanghai, China).

CA₆ refractory was synthesized using industrial Al₂O₃ and CaCO₃ as raw materials and TiO₂ as a sintering aid. The raw materials were weighed according to the experimental formula in Table 1, and placed into a ball mill tank. The mixture was dry-ground in a planetary ball mill for 40 min at a rotation speed of 258 rpm, with a powder-to-ball mass ratio of 1:6. The ball-milled powder was then naturally stacked in an alumina crucible and pre-sintered in a medium-temperature furnace at 1300 °C for 3 h under an air atmosphere. After pre-sintering, 1% TiO₂ sintering aid was added to the powder, followed by an additional 3 h of grinding in the planetary ball mill. Subsequently, 1 wt% polyvinyl alcohol (PVA) solution (10 wt% concentration) was added as a binder and mixed thoroughly. The mixture was then pressed into various shapes using a hydraulic press at 150 MPa, including cylindrical billets (Φ 20 mm × 20 mm), strip billets (50 mm × 10 mm × 10 mm and 100 mm × 15 mm × 15 mm), and crucible samples (Φ 50 mm × 50 mm with an inner diameter of Φ (12~15) mm × 15 mm). The green bodies were dried in a vacuum oven at 80 °C and 110 °C for 24 h each. Finally, the samples were heat-treated in a high-temperature sintering furnace at 1750 °C for 3 h under an air atmosphere.

In order to understand the phase transformation pathway of CA₆ during sintering, Figure 1 presents the TG-DSC curves obtained during thermal treatment. The data reveal that after pre-sintering at 1300 °C, the TG curve shows virtually no weight loss, with the low-temperature endothermic peak entirely eliminated. The predominant crystalline phases are calcium aluminate (CaAl₄O₇, CA₂) and α-Al₂O₃, indicating that the system has undergone extensive removal of volatile components and has formed a thermodynamically stable intermediate-phase framework. The powder subjected to 1300 °C pre-sintering is thus in a fully reacted state.

The phase composition of the samples was analyzed using an X-ray diffraction (X′ pert pro, PANalytical, Almelo, The Netherlands). The microstructure and element distribution of the samples were analyzed by a field emission scanning electron microscope field emission equipped with backscattered electron imaging(SEM, FEI Nova 400 Nano-SEM, FEI, Portland, OR, USA) and an X-ray energy-dispersive spectrometer (INCA Penta FET-X3, Oxford Instruments, Oxford, UK).

The dimensions of the samples before and after sintering were measured using a vernier caliper to calculate the linear shrinkage rate. Bulk density and apparent porosity were determined using the Archimedes method. Thermal diffusivity and specific heat capacity were measured from room temperature to various elevated temperatures using an laser thermal conductivity meter (LFA 457 MicroFlash, NETZSCH-Gerätebau GmbH, Selb, Germany). Thermal conductivity was calculated based on the sample’s volumetric density. The coefficient of thermal expansion from room temperature to 1400 °C was measured using a high-temperature thermal dilatometer (DIL502, NETZSCH-Gerätebau GmbH, Selb, Germany) at a heating rate of 10 °C/min.

The bending strength of specimens (50 mm × 10 mm × 10 mm) was evaluated using the three-point bending method. A constant load of 0.5 MPa/s was applied vertically until fracture, and the maximum load F_max(N) was recorded. This quasi-static loading regime ensured that material deformation and failure took place under near-static conditions, thereby accurately capturing the flexural strength behavior of the material. For thermal shock resistance testing, samples were heated in an electric furnace at 1100 °C for 20 min, then rapidly quenched in flowing water at 5–35 °C for 3 min. The residual bending strength was measured to assess thermal shock resistance.

The alloy melting experiment was conducted using a VIF-2 vacuum induction melting furnace. Ti6Al4V alloy was used for the experiment. The self-made crucible containing the alloy was placed inside a graphite crucible, which was then positioned within a quartz crucible. After sealing the furnace, the chamber was evacuated to 1 × 10⁻² Pa using a mechanical pump and a molecular pump. Argon gas was introduced to achieve a pressure of 0.06 MPa, and the system was purged three times with low-vacuum gas. The heating power was then activated to initiate the melting process. The temperature of the alloy melt was monitored using a tungsten-rhenium thermocouple, and the alloy was heated to 1700 °C within 90 min. After melting, the alloy was held in an argon atmosphere for 3 min. The power supply was subsequently turned off, and the melt was allowed to cool in the furnace before removal for further analysis.

In order to clarify the reaction behavior of the CA₆ crucible and Ti6Al4V alloy, three groups of crucible materials with varying phase compositions were prepared. Figure 2 shows the XRD patterns of the three groups of samples after heat treatment at 1750 °C. The results indicate that the phases present in the samples are Al₂O₃, Al₂O₃-CA₆, and CA₆, corresponding to the three groups of refractories. It is evident that the CA₆ phase formed through a high-temperature solid-state reaction between CaCO₃ and Al₂O₃. When the mass ratio of CaCO₃ to Al₂O₃ was less than 1:6, the material contained both the CA₆ phase and residual Al₂O₃ phase. In contrast, when the mass ratio of CaCO₃ to Al₂O₃ was 1:6, the material exhibited a single CA₆ phase. Although the single-phase CA₆ sample was prepared with CaCO₃ and Al₂O₃ in stoichiometric proportions, a trace amount of Al₂O₃ was still detected in the XRD pattern. This discrepancy may be attributed to uneven local dispersion of the mixture during dry mixing, leading to incomplete reaction.

The SEM images and EDS spectra of the three groups of samples with different phases are presented in Figure 3 and Table 2. In the Al₂O₃ sample (Figure 3d), the grains are tightly bonded with minimal porosity, and only a small amount of pores are distributed between the grains. With the addition of CaCO₃ and the corresponding decrease in Al₂O₃ content, the microstructure shows an increase in flake-like grains arranged in an interlocking network. When the mass ratio of CaCO₃ to Al₂O₃ is less than 1:6, the sample exhibits higher porosity with larger size, along with relatively complete grains. The CA₆ phase (points 4 and 7) forms a distinct interlocking lamellar structure, while the Al₂O₃ phase (points 3, 5, and 6) is present at grain boundaries, where the grains are more densely bonded. When the mass ratio of CaCO₃ to Al₂O₃ is equal to 1:6, a significant amount of CA₆ crystals is observed in the sample (Figure 3f). Compared to the Al₂O₃-CA₆ sample, the lamellar structure of the crystal is less pronounced, and the grains exhibit a “bridging” morphology with an interwoven arrangement. The crystals tend to develop into an equiaxed shape, and the pore distribution becomes more uniform.

Figure 4 shows the linear shrinkage, apparent porosity, and bulk density of the three sample groups with different phase compositions after heat treatment at 1750 °C. As shown, with the addition of CaCO₃ and the corresponding decrease in Al₂O₃ content, the linear shrinkage of the sample gradually increases. The apparent porosity increases first and then decreases, while the bulk density remains relatively consistent, ranging from, ranging from 3.45 to 3.62 g·cm⁻³. In comparison, samples consisting of either single-phase Al₂O₃ or single-phase CA₆ exhibit higher density than those containing both phases. Combined with the microstructural analysis in Figure 3, it can be concluded that the coexistence of Al₂O₃ and CA₆ phase in the Al₂O₃-CA₆ sample is the main reason for the increased apparent porosity. In the CA₆ sample, when the mass ratio of CaCO₃ to Al₂O₃ is 1:6, the material consists almost entirely of the CA₆ phase, resulting in a noticeable reduction in apparent porosity.

Apparent porosity is the primary structural factor influencing surface roughness, and the two exhibit a positive correlation. The surface roughness of ceramics essentially represents “microscale irregularities”, whilst open pores (apparent pores) constitute the primary source of surface depressions and defects. The “melt penetration pathway” function of apparent pores is amplified by surface roughness: On highly rough surfaces, depressions connect with open pores, forming “labyrinthine” permeation pathways that facilitate liquid titanium-aluminium alloy infiltration into pores via these depressions. Conversely, on low-roughness surfaces, pores are predominantly isolated closed cells. Even where open pores exist, the absence of depressions to guide the melt significantly increases the difficulty of permeation. CA₆ leverages the synergistic effect of “low apparent porosity + low roughness”, combined with its inherent high chemical stability and high density, to create a compound advantage. This approach blocks the core pathways for alloy contamination at both structural and chemical levels, making it more suitable for application in the field of titanium-aluminium alloy melting.

Figure 5 shows the thermal conductivity and thermal diffusivity of the three sample groups measured at 100 °C, 300 °C, and 500 °C. As the temperature increases, the thermal conductivity and thermal diffusivity of all samples decrease gradually. The single-phase Al₂O₃ sample exhibits the most significant reduction in these properties. In contrast, the Al₂O₃-CA₆ and CA₆ samples show a more gradual decrease with increasing temperature, with nearly identical values between them. Among the three sample types, thermal conductivity decreases in the order: Al₂O₃ > CA₆ > Al₂O₃-CA₆. The results indicate that the thermal conductivity of the single-phase CA₆ material prepared in this study is intermediate between that of single-phase Al₂O₃ and the Al₂O₃-CA₆ composite. A lower thermal conductivity in the crucible material can reduce the cooling rate of the molten alloy and help maintain its superheat. Thermal conductivity is influenced by both the chemical composition and the microstructure of the material. On one hand, the larger relative atomic mass of cations in CA₆ compared to Al₂O₃ contributes to its lower thermal conductivity. On the other hand, phonon scattering at phase boundaries further reduces the thermal conductivity of the Al₂O₃-CA₆ composite below that of single-phase CA₆, though the value remains closer to that of the continuous CA₆ phase.

The heat capacity of a material reflects the change in energy of molecular thermal motion with temperature. When absorbing the same amount of heat, a refractory material with higher heat capacity experiences a smaller temperature increase compared to one with lower heat capacity, thereby improving thermal shock resistance. Table 3 presents the specific heat capacities of the three sample groups after heat treatment at 1750 °C, as measured by a laser thermal conductivity meter at various temperatures. It can be observed that above 300 °C, the heat capacity of single-phase CA₆ exceeds that of single-phase Al₂O₃.

As shown in Figure 6, within the temperature range of 200~1400 °C, the linear expansion rates of the three sample groups with different phase compositions increase with rising temperature, though the differences among them are relatively small. The average coefficient of thermal expansion exhibits a non-monotonic upward trend over the measured range. Between 200 °C and 1200 °C, both the single-phase Al₂O₃ and Al₂O₃-CA₆ samples show a fluctuating increase in thermal expansion. In the lower temperature range (200~600 °C), the average thermal expansion coefficient of the Al₂O₃-CA₆ samples lies between those of the other two groups. However, from 600 °C to 1400 °C, the average thermal expansion coefficient of the Al₂O₃-CA₆ sample becomes lower than that of both single-phase materials. This behavior can be attributed to the compatible thermal expansion of Al₂O₃ and CA₆ at lower temperatures (200~600 °C). As temperature increases, the difference in thermal expansion coefficients between Al₂O₃ and CA₆ becomes more pronounced, inducing microcracks due to thermal stress in the composite material. The presence of these microcracks mitigates the overall thermal stress, thereby slightly reducing the apparent thermal expansion coefficient of the Al₂O₃-CA₆ sample. Additionally, the average thermal expansion coefficient of the single-phase CA₆ sample shows slight decreases in the temperature intervals of 600~800 °C and 1200~1400 °C.

Figure 7 presents the flexural strength at room temperature after heat treatment and the residual flexural strength after three thermal shock cycles for the three sample groups with different phase compositions. The single-phase CA₆ exhibits the highest cold modulus of rupture, while the single-phase Al₂O₃ sample demonstrates the highest residual modulus of rupture after thermal shock. In contrast, the Al₂O₃-CA₆ sample shows the lowest values for both room-temperature flexural strength and residual flexural strength after thermal cycling, indicating that the coexistence of Al₂O₃ and CA₆ phases reduces mechanical strength under both ambient and thermal shock conditions. The residual flexural strength is significantly influenced by the thermal conductivity of the material, which is proportional to its thermal stress fracture resistance (R). Specifically, as thermal conductivity increases (in the order: Al₂O₃-CA₆ < CA₆ < Al₂O₃), the rate of heat transfer within the material increases, leading to a reduced temperature gradient, lower thermal stress, and consequently an increased thermal stress fracture factor. This results in a corresponding improvement in residual flexural strength, which follows the same trend: Al₂O₃-CA₆ < CA₆ < Al₂O₃.

The influence of porosity on the strength of ceramic materials can be expressed by the formula σ_f = σ₀e^−nP, where σ_f denotes the fracture strength of the material, σ₀ represents the strength when porosity P equals zero, and n is a constant. This expression indicates that flexural strength decreases exponentially with increasing porosity. Accordingly, the CA₆ material, which exhibits the lowest porosity among the samples studied, demonstrates the highest fracture strength. In terms of thermal conductivity, Al₂O₃ exhibits the highest value, followed by CA₆, with Al₂O₃-CA₆ showing the lowest. Thermal conductivity critically influences the temperature uniformity within a material under thermal transients. Higher thermal conductivity facilitates faster heat transfer, resulting in a smaller temperature gradient (ΔT) and lower thermal stress (Δσ = αEΔT/(1 − ν)), where α is the coefficient of thermal expansion, E is the elastic modulus, and ν is Poisson’s ratio. This mechanism provides the fundamental basis for thermal shock resistance. Consequently, CA₆ exhibits moderate thermal shock resistance.

Figure 8 shows the microstructure of the reaction interface between the three groups of crucible samples and the Ti6Al4V alloy after melting. The upper region corresponds to the alloy, while the lower region represents the crucible refractory. As shown in Figure 8a, after melting in an Al₂O₃ crucible, noticeable cracks are present in the refractory material beneath the interface. These cracks pose a risk of spalling into the alloy. A magnified view of the interface (Figure 8d) reveals an erosion layer approximately 106 μm thick at the contact zone between the crucible and the alloy. This layer exhibits a relatively loose structure, accompanied by crack formation and potential spalling. However, according to the scanning results in Figure 9a, beneath this eroded region, the Ti6Al4V alloy shows minimal penetration into the dense Al₂O₃ refractory.

Figure 8b illustrates the interface after melting using the CA₆-Al₂O₃ crucible. A clear boundary is visible between the crucible and the alloy. While the crucible surface appears smooth with almost no erosion, the refractory material is less dense, allowing significant alloy penetration exceeding 411 μm, as shown in Figure 8e. In contrast, Figure 8c demonstrates that melting with the CA₆ crucible results in a distinct interface with superior surface smoothness compared to that of the dual-phase sample. The CA₆ refractory remains relatively dense, experiencing negligible erosion and only minor penetration—approximately 100 μm—as depicted in Figure 8f.

A comparison of the interfacial microstructures in Figure 8 indicates that the thickness of the reaction layer follows the order: Al₂O₃-CA₆ > Al₂O₃ > CA₆. Among these, the CA₆ sample exhibits the highest density, best surface finish, and the most favorable interface condition with the alloy. A dense ceramic crucible with a smooth surface effectively mitigates infiltration by the titanium-aluminum melt and reduces the erosion or dissolution of fine particles into the alloy, thereby helping maintain the purity of the ingot.

Figure 9 presents magnified views of the interface and adjacent alloy region for the three types of crucible refractories after interaction with the molten Ti6Al4V alloy. Comparative analysis of the interfacial microstructures (Figure 9a–c) indicates that the alloy melted in the single-phase CA₆ crucible exhibits the least influence from the refractory material and the lowest degree of contamination. In contrast, the single-phase Al₂O₃ crucible shows signs of erosion at the interface (Table 4, point 1), with a risk of refractory particles spalling into the alloy. In the Al₂O₃-CA₆ crucible, Al₂O₃ phases are likely incorporated into the alloy melt under intense electromagnetic stirring.

The single-phase CA₆ crucible refractory demonstrates high density and excellent chemical stability, effectively limiting dissolution and minimizing contamination of the alloy melt. EDS analysis (point 2) suggests that the presence of Al₂O₃ in this region may result from inhomogeneous mixing during preparation. These results indicate that CA₆ refractories, characterized by high density, surface smoothness, and chemical inertness, exhibit significant potential for application in titanium-aluminum alloy smelting.

Observation of the alloy region above the interface (as shown in Figure 9d–f) further confirms that the most severe contamination occurs with the Al₂O₃ crucible (point 5), followed by the Al₂O₃-CA₆ crucible (point 6). The alloy melted using the single-phase CA₆ crucible shows the lowest level of contamination. Figure 10 illustrates the schematic representation of the interface reaction between titanium aluminium alloy and refractory materials, in conjunction with the interface reaction analysis diagram depicted in Figure 9 above. The Al₂O₃ crucible exhibited intense interfacial reduction reactions accompanied by severe particle spalling, with deep penetration occurring between the crucible and the alloy. Erosion effects were diminished in the Al₂O₃-CA₆ crucible, where only minor penetration and erosion were observed. However, under high-temperature conditions, CA₆ underwent slight decomposition, releasing Al₂O₃ phase into the alloy. This variation can be attributed to the strong electromagnetic stirring in the vacuum induction melting furnace, which promotes melt convection, continuously erodes the crucible wall, and enhances dissolution of refractory elements into the alloy. These findings are consistent with the erosion penetration results shown in Figure 8, confirming that the dense CA₆ crucible refractory provides the best performance, with minimal alloy contamination and only slight penetration. CA₆ possesses a hexagonal crystal structure, which provides higher density and better structural integrity than the corundum structure of α-Al₂O₃. With respect to solubility and diffusion, the compact framework of CA₆ significantly hinders atomic mobility. In contrast, Al₂O₃ has a relatively open lattice that allows easier diffusion of Ti and Al atoms into its structure. This incorporation causes lattice distortion and bond rupture, which accelerates reduction reactions at the interface. In comparison, the low diffusion coefficient of CA₆ effectively suppresses atomic exchange across the interface, thereby preventing sustained reaction progress. In addition, the low wettability of CA₆ reduces the effective contact area between the crucible material and the titanium alloy melt, and any reaction products that form can act as a passivation layer, further inhibiting reaction advancement. As a result, CA₆ showed the lowest reactivity in the present experiments.

A dense CA₆ ceramic refractory, with a bulk density of 3.62 g·cm⁻³, was prepared by a two-step sintering method using industrial Al₂O₃ and CaCO₃ as raw materials. Compared to Al₂O₃ and CA₆-Al₂O₃ materials, the fabricated CA₆ refractory showed the highest flexural strength and medium thermal shock resistance. Moreover, the pure CA₆ crucible showed best resistance to Ti6Al4V alloy corrosion. The refractory surface remained smooth and structurally dense, showing almost no erosion and only minimal penetration, with a penetration depth of approximately 100 μm. Furthermore, the alloy melted in this crucible exhibited the lowest level of contamination. These results demonstrate that dense CA₆ material holds significant potential for application in the smelting of titanium-aluminum alloys.

The authors are thankful for the financially support from the National Natural Science Foundation of China (Grant Nos. 52472032 and 52172023), the Key Program of Natural Science Foundation of Hubei Province (Grant No. 2024AFA083) and the Research Project of Hubei Provincial Department of Science and Technology (Grant No. 2024CSA075).

Conceptualization, L.F. and H.G.; Methodology, L.F.; Software, S.H.; Validation, S.H. and N.L.; Formal Analysis, S.H.; Investigation, S.H.; Resources, J.Z.; Data Curation, S.H.; Writing—Original Draft Preparation, S.H.; Writing—Review & Editing, L.F., C.P., N.L., H.G., J.Z. and A.H.; Visualization, S.H.; Supervision, L.F.; Project Administration, L.F.; Funding Acquisition, L.F.

Not applicable.

Not applicable.

The data are available from the corresponding author on reasonable request.

This research was funded by the National Natural Science Foundation of China (Grant Nos. 52472032 and 52172023), the Key Program of Natural Science Foundation of Hubei Province (Grant No. 2024AFA083) and the Research Project of Hubei Provincial Department of Science and Technology (Grant No. 2024CSA075).

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.