Variation in Carbon Content During the Melting of γ-TiAl in a Graphite Crucible

Abstract

Liquid γ-TiAl alloy was prepared by vacuum induction melting within graphite crucibles, then cast using a centrifugal technique. In this process, the degree of superheat (ΔT)—defined as the temperature above the melting point—was carefully controlled, with experiments conducted at ΔT of 200 K (i.e., 200 Kelvin above the melting temperature). It was observed that carbon content in the alloy increased nonlinearly as the melt was held longer in the graphite crucible; for example, carbon concentration rose from an initial value of approximately 0.21 at% to 0.98 at% after 100 s of holding and to 2.11 at% at 650 s of holding. When the melt was held for over 100 s at ΔT = 200 K, titanium carbide (TiC) and titanium aluminum carbide (Ti₂AlC) particles formed along the crucible wall. This resulted in changes to the phase fractions and a corresponding increase in aluminum concentration in the melt. Two types of Ti₂AlC phases were observed: one consisted of coarse Ti₂AlC particles, which were crystallized through peritectic reaction from the TiC carbide and liquid phase. The other consisted of fine Ti₂AlC particles, which were decomposed from the α₂ (Ti₃Al) phase within the interlamellar regions. After 20 s of holding at ΔT = 200 K, carbon rapidly dissolved into a solid solution. Prolonged holding led to significant grain refinement: the microstructure evolved from columnar to equiaxed grains, primarily due to TiC crystallization. This transition is significant because finer, equiaxed grains can enhance mechanical properties such as strength and toughness. The findings provide valuable insight into the interaction between graphite crucibles and γ-TiAl melts, demonstrating how controlled superheat and holding time influence carbon uptake, carbide formation, and microstructural evolution—factors critical for optimizing the performance and manufacturability of γ-TiAl components.

1. Introduction

γ-TiAl alloys have been recognized as promising alternatives to conventional nickel-based superalloys. These alloys offer high-temperature strength and lower density, making them advantageous for engineering applications where weight reduction is critical. In the aerospace sector, γ-TiAl alloys are increasingly considered for components such as turbine blades and turbocharger wheels, where their use can contribute to reduced engine weight and enhanced efficiency. This, in turn, may support improved fuel economy, lower emissions, and operational cost savings, aligning with the industry’s growing emphasis on sustainability and environmental responsibility [1]. However, the adoption of γ-TiAl components remains limited, primarily due to their high production costs. Furthermore, manufacturing these alloys poses significant challenges, stemming from their brittleness at ambient temperatures, limited workability, poor fluidity, and particularly high chemical reactivity in the molten state [2,3,4]. In recent years, the increased adoption of additive manufacturing (AM) has been mostly driven by fast-paced research and development in metal AM, featuring powder metals from Ti alloys to Ni-based superalloys for high-temperature applications. Despite abundant R&D, there are still many challenges in AM of Ti alloys, especially in the inherently rough initial (as-built) surface finishing barriers and the loss of Al elements to be overcome prior to even wider adoption [5].

Investment casting offers high-dimensional accuracy and excellent surface finish at a relatively low cost, particularly for manufacturing components with complex geometries, such as turbochargers [6]. Due to the high chemical reactivity of Ti, its melting and casting must be conducted with great care. To minimize melt contamination, the induction skull melting (ISM) process using a cold crucible has been employed [7]. However, the cold crucibles often lead to casting defects, such as misruns and cold shuts, primarily due to the low ΔT involved.

Vacuum induction melting (VIM) allows for higher ΔT, but it tends to introduce elevated impurity levels into the melts as a result of reactions between the crucible material and the γ-TiAl melts [8]. Consequently, extensive research has been conducted to select suitable crucibles for melting of Ti alloys and to investigate the effect of casting conditions on impurity levels, aiming to produce high-purity castings.

γ-TiAl castings produced using calcium oxide or yttrium oxide crucibles have demonstrated acceptable impurity levels [9,10,11,12,13]. However, the calcium oxide crucibles suffer from low shock resistance and are prone to hydration [9], while the yttrium oxide crucibles are costly and vulnerable to thermal shock [14]. To address these limitations, research has focused on developing low-cost crucibles through the application of yttria coatings. Yttria-coated crucibles have shown promising results in terms of cost-effectiveness and casting purity [15]. Nevertheless, these crucibles require careful handling to prevent cracking, which may result from poor thermal shock resistance or inadequate adhesion of the coating to the crucible substrate. Despite the advantages of graphite crucibles, including ease of handling, low cost, and excellent thermal and erosion resistance, they have not been widely adopted in the melting of Ti alloys. One major limitation is the lack of comprehensive data on the relationship between the variation in melting conditions and the amount of carbon contamination when graphite crucibles are used.

There are two distinct viewpoints on the impact of carbon contamination in molten γ-TiAl alloys. On the one hand, carbon can enhance γ-TiAl alloys through solid solution strengthening and precipitation hardening effects [16]. Recent studies have also explored in situ TiAl matrix composites reinforced with carbides formed during vacuum induction melting (VIM), demonstrating that carbides can improve fracture toughness and slow creep deformation at high temperatures, provided the carbon content remains below 0.5 at% [17,18,19,20,21]. On the other hand, some researchers argue that graphite crucibles are unsuitable for γ-TiAl melting due to the risk of severe carbon contamination, which can reach up to 7 at% [22,23]. Therefore, when melting TiAl in graphite crucibles, it is crucial to clearly analyze how varying carbon content affects the alloy’s microstructure.

This study seeks to assess the interactions between graphite crucibles and γ-TiAl alloy melts under conditions that guarantee thorough mold filling, particularly for turbocharger wheels having intricate geometries and thin-walled structures. Fluidity tests are conducted at various ΔT to determine the optimal pouring temperature. Additionally, the correlation between changes in carbon content and the microstructural evolution of γ-TiAl alloys at different durations of holding (Δt) of melts is investigated. The feasibility of graphite crucibles for melting of γ-TiAl alloys is also examined.

2. Experimental Procedure

2.1. Production of the Investment Mold

The wax pattern and sprue were assembled to have a direct gating system because it could manufacture the intact shape of turbochargers. The turbocharger wax pattern had a diameter of 63.5 mm and a blade tip thickness of 1 mm. The primary slurry comprising the colloidal silica binder (30% SiO₂, 15 nm) and alumina refractories was mixed until the efflux time measured via Zahn cup #4 (ASTM D-4212) was 40 s. The primary coating was performed three times. Backup coating was performed seven times using a mix of chamotte, fused silica refractories, and colloidal silica under constant temperature and humidity conditions. The molds were dewaxed in an autoclave at a pressure and temperature of 0.5 MPa and 423 K, respectively. The firing process was conducted at 1223 K for 2 h in the air atmosphere.

2.2. Induction Melting and Centrifugal Casting

The VAR (vacuum arc remelted) ingot of Ti-48Al-2Cr-2Nb (at%) (Ti4822, purity of 99%, Gfe Gesellschaft für Elektrometallurgie mbH, Nürnberg, Germany) was selected for the evaluation of reactivity with the graphite crucible. Ti4822 alloy (250 g) was melted and cast in vacuum induction melting and vacuum centrifugal casting furnaces. Graphite (99% purity, Dongbang Carbon Co., Ltd., Hwaseong-si, Republic of Korea) was used as a crucible material. The chamber was evacuated at a pressure of 0.001 mbar and refilled with argon gas to 700 mbar to prevent the evaporation of aluminum. The casting was performed at ΔT of 130 K, 170 K, 200 K, and 230 K. The holding-time (Δt) experiment was conducted only at ΔT = 200 K, where Δt was varied from 20 to 650 s to evaluate the interaction between the TiAl melts and the graphite crucible. The temperature of the γ-TiAl melts was measured using a two-color pyrometer (RAYMR1SBSF, Raytek Corp., Santa Cruz, CA, USA).

As the melts heated to the target temperature, the chamber rotated to 300 rpm within 2 s. The rotation was maintained for approximately 1 min, although mold filling was completed within 2 s; the remaining time was used to sustain the applied pressure on the melt. The final pouring temperature was determined from the condition that satisfied mold filling. A detailed schematic of the casting machine is presented in [24].

2.3. Characterization of Castings and Graphite Crucible

Carbon contents of castings were analyzed using a carbon/sulfur analyzer (CS-200, LECO Corp., St. Joseph, MI, USA). The morphologies of castings and interface microstructures between the crucible and melts were observed using an optical microscope (DM 2700M, Leica Microsystems GmbH, Wetzlar, Germany) and field emission scanning electron microscopy (FE-SEM, JSM-7600F, JEOL Ltd., Tokyo, Japan). In addition, a stereomicroscope (EMZ-TR, Meiji Techno Co., Ltd., Saitama, Japan) was used to evaluate the grain size. The phase identification and chemical composition of the castings were analyzed via X-ray diffraction (XRD, M18XHF-SRA, Mac Science Co., Ltd., Yokohama, Japan), energy dispersive X-ray spectrometry (EDS, JSM-7600F, JEOL Ltd., Tokyo, Japan), and spark emission spectrometry (OES, QSN 750-2, OBLF GmbH, Witten, Germany).

3. Results and Discussion

3.1. Optimum Degree of Superheating (ΔT) for Mold Filling of Turbocharger Castings

The fluidities of γ-TiAl at ΔT of 130 K, 170 K, 200 K, and 230 K were evaluated to determine the suitable pouring temperature for the turbocharger casting. Figure 1 shows the exterior of the turbocharger castings on the front, back and blade tip. Misrun at the disk and blade tips and cold shut were observed on the blade when the ΔT was 130 K. At 170 K, misrun and cold shut were sustained in the disk and blade tips, though misrun on the blade tip was reduced. At ΔT above 200 K, the unfilled cavities and cold shut were not observed. Therefore, ΔT above 200 K was required to fill the entire turbocharger cavity.

The final pouring temperature was set to achieve a ΔT of 200 K to ensure proper mold filling, since higher superheating can accelerate melt contamination [25].

3.2. Variation in Carbon Content and Chemical Composition with Respect to Holding Time (Δt)

The reactivity of TiAl melts with the graphite crucible was evaluated with respect to holding the melts at ΔT of 200 K. Figure 2 shows the variation in carbon contamination in the castings. The carbon content increased linearly up to a holding time of 100 s, after which the rate declined with prolonged holding, eventually reducing to one-fifth of the initial rate. Because the rate of carbon increase became much smaller beyond 100 s, the holding experiments in the longer-time region were performed at wider time intervals to confirm the gradual saturation behavior of carbon absorption.

Figure 3 shows OES results indicating the change in the chemical composition of the castings with respect to holding time. The contents of Ti, Al, Cr, and Nb remained relatively constant up to a holding time of 100 s. After that, however, Al concentration gradually increased and finally reached 52.5 at% for a holding time of 650 s, while the Cr and Nb remained constant. For each holding-time condition, three cross-sectional samples were analyzed, and the average values are shown. The measurement error of the OES analysis was within 1.5%, which is smaller than the size of the symbols in Figure 3.

3.3. Compositional Change in Castings and Variation in Carbon Contamination Rate

When the liquid γ-TiAl was held for 100 s at ΔT of 200 K, the carbon concentration measured 0.98 at%. To elucidate the subsequent increase in Al concentration and the reduction in carbon contamination rate beyond Δt of 100 s, the interface between the graphite crucible and the residual melt was examined.

Figure 4 shows the cross-section of the crucible and the interface morphology, along with XRD analysis of the residual melts. Blue square boxes in the Figure 4a–c indicate the observed areas. As Δt increased, the amounts of residual melts gradually rose, leading to significant crucible erosion. At Δt of 650 s, extensive erosion was evident, accompanied by a large accumulation of residual melts and an increased carbon content of 2.11 at%.

The XRD analysis of the residual TiAl performed at Δt of 650 s is shown in Figure 4g. The phases of the residual melt, being cast into the mold, did not correspond to TiAl matrices (α₂ and γ), but large amounts of Ti₂AlC and small amounts of TiC phases were observed.

Figure 5 and

Table 1. EDS analysis results of the phases indicated by arrows A through F in Figure 5 (at%).

Region	A	B	C	D	E	F
C K	50.7	18.7	43.4	24.2	51.9	17.7
Al K	0.4	32.7	0.4	25.8	2.3	14.9
Ti K	42.4	42.6	55.4	50.0	45.8	56.6
Cr K	1.6	1.7
Nb L	4.9	4.3	0.8			10.8
Totals	100	100	100	100	100	100

present the results of FE-SEM and EDS analyses for the interface between the residual TiAl and graphite crucible, respectively. Under all conditions, the melts penetrated the pores of the graphite crucible. In Figure 5, at the boundary between the graphite crucible and residual melts, a thin layer of A, C, and E was observed. This layer was identified as a binary composition of titanium and carbon. The round-edged phases (regions of B, D, and F) grew behind the binary phase and spread toward the melts. The phase proportion of titanium, aluminum, and carbon was confirmed to be 2:1:1. The binary phase finally got thicker to 10 μm, and the ternary compound was observed as a lump without a matrix beyond a holding time of 650 s in the residual melts. The thin layers marked by arrows A and C in Figure 5 correspond to the TiC phase, whereas the rounded feature located behind the interface is identified as the Ti₂AlC phase, as confirmed by XRD and EDS analyses. TiC is the primary solidification phase in TiAl alloys, and Ti₂AlC can be formed from TiC at 1962 K [26,27]. Ti₂AlC originated from the TiC at a molten state, following the formation sequence L + TiC → L + Ti₂AlC [26]. Furthermore, the remaining TiC at the crucible wall could not transform into Ti₂AlC because of the rapid cooling of the crucible after casting.

As shown in Figure 4 and Figure 5 and Table 1, the TiC and Ti₂AlC phases, which are richer in titanium, adhered to the crucible walls and remained in the crucible after casting. Consequently, aluminum content of the melts increased, leading to higher aluminum concentration in the castings.

Because graphite crucibles are used with indirect heating, one of the main drawbacks of a graphite crucible in induction melting is the small degree of force applied by the EM field to melt, which does not allow the metal to be stirred during melting [28]. As the holding time exceeded 100 s, more carbides were incorporated into the melt, which led to increased viscosity [29]. This higher viscosity made electromagnetic stirring less effective; consequently, the stirring force might no longer be sufficient to detach the carbides from the crucible wall. The carbides could adhere to the crucible wall during the melting process. As a result, carbides began to accumulate more readily at the crucible wall, overcoming the stirring action and building up along the wall as the melting process continued.

During melting, a layer of solid TiC and Ti₂AlC progressively formed along the crucible wall, acting as a partial diffusion barrier. Although this layer did not completely prevent carbon ingress, it slowed the rate of carbon transfer from the crucible into the melt.

3.4. Microstructural Evolution of TiAl Castings as a Function of Holding Time, Δt

FE-SEM and EDS analyses of the TiAl castings confirmed the influence of carbon content on microstructural change. The corresponding results are presented in Figure 6 and

Table 2. EDS analysis results of the carbides indicated by arrows 1 through 5 in Figure 6 (at%).

Region	1	2	3	4	5
C K	22.6	30.7	29.9	26.2	31.7
Al K	27.8	25.9	24.2	24.0	27.1
Ti K	49.1	41.4	44.0	49.8	39.2
Cr K	0.5	0.1	1.9		0.3
Nb L		0.1			1.7
Totals	100	100	100	100	100

. Fine carbides formed sparsely within the interlamellar region when the carbon content exceeded 0.36 at% during a holding time (Δt) of 40 s. Klimová et al. observed the decomposition of fine secondary carbide from the α₂ phase at the interface of α₂/γ lamellar in the as-cast state [30]. The formation of secondary carbide occurred owing to the difference in the carbon solubility of γ (0.025 at% C) and α₂ (0.96 at% C) phases in the solid state [30,31]. According to the EDS analysis, the coarse carbide comprised titanium, aluminum, and carbon. As shown in Figure 6d–h, coarse intergranular carbides appeared together with fine interlamellar carbides when the carbon content exceeded 0.66 at% (holding time of 75 s). The fine interlamellar carbides exhibited sizes below approximately 1 μm, whereas the coarse intergranular carbides were about 1 μm in thickness and exceeded 10 μm in length. These quantitative differences indicate that carbon enrichment promotes carbide coarsening and a morphological transition from interlamellar to intergranular types.

Figure 7 shows the X-ray spectra for TiAl castings as a function of carbon content. Due to analytical limitations, Ti₂AlC was detected only when the carbon concentration exceeded 1.56 at% at a holding time of 365 s. Between 0.30 and 0.98 at% C, the Ti₃Al peak intensity increased slightly, indicating that low carbon levels do not destabilize Ti₃Al. However, when the carbon content exceeded 1.56 at%, the Ti₃Al peak decreased as Ti₂AlC formation progressed. Previous studies have established that the feasible carbide phases in TiAl alloys include TiC, Ti₂AlC, and Ti₃Al, with Ti₂AlC being the most thermodynamically stable phase [23]. TiC serves as the primary solidification phase, whereas Ti₂AlC and Ti₃Al can be formed from TiC at transformation temperatures of 1962 K and 1903 K, respectively. However, metastable Ti₃Al decomposes during solidification and does not persist in the final microstructure. The transformation of TiC into Ti₂AlC is promoted by sufficiently low cooling rates during solidification, which provide adequate time for TiC to react with the melt and form Ti₂AlC before the matrix fully solidifies. XRD analysis revealed characteristic diffraction peaks corresponding to Ti₂AlC, while EDS spectroscopy confirmed the presence of titanium, aluminum, and carbon in stoichiometric ratios consistent with Ti₂AlC. These results support the conclusion that both fine secondary and coarse primary carbides in the castings are Ti₂AlC phases, with coarse primary Ti₂AlC forming via the transformation of crystallized TiC in the molten state prior to complete solidification.

Figure 8 presents the grain sizes and morphological variations corresponding to different holding times. McCullough et al. [32] reported that Ti-46~49Al (at%) alloys were solidified by the peritectic reaction. The solidification and solid-state transformation can be expressed as follows [33]:

L → (β) +L → (β + α) + L → (β + α) + γ → (α) + γ → (α₂) + γ → (α₂ + γ) + γ

Columnar microstructures were observed when the carbon content was below 0.63 at% (holding time of 65 s) due to the peritectic solidification reaction that led to Al-rich segregations. However, at a carbon content of 0.66 at% (holding time of 75 s), the microstructure of the castings transitioned to an equiaxed structure. The grain size was quantitatively measured from the optical micrographs in Figure 8: the columnar grains Figure 8a–e exceeded approximately 2000 μm in length, while the equiaxed grains Figure 8f–l gradually refined with increasing carbon content, showing average grain sizes of 1115, 479, 343, 363, 207, 151, and 118 μm, respectively. These quantitative results clearly demonstrate that higher carbon content promotes grain refinement in TiAl castings. Therefore, the holding time (Δt) should be limited to within 65 s at a superheating (ΔT) of 200 K, during which the intrinsic columnar microstructure of the Ti4822 alloy is maintained.

During the transition from a columnar to an equiaxed structure, a sufficient number of nuclei is required in undercooled TiAl melts. Primary TiC and coarse Ti₂AlC particles, which are crystallized through peritectic reaction from the liquid phase and TiC carbide present in the molten state, supply an adequate number of nuclei that promote equiaxed microstructure formation and serve as heterogeneous nucleation sites for β phase development. Consequently, primary TiC or coarse Ti₂AlC plays a critical role in grain refinement and the formation of equiaxed structures by providing effective heterogeneous nucleation sites.

The FE-SEM image was analyzed using an image analysis software (Image Pro-plus 6.0, Media Cybernetics, Rockville, MD, USA) to examine the change in the phase volume fraction. Figure 9 shows the changes in the phase volume fraction with holding time. The volume fraction of Ti₂AlC increased steadily. However, the α₂ (Ti₃Al) phase increased to 20% at 0.98 at% carbon (holding time of 100 s) and then decreased. As carbon is a strong α-phase stabilizer, the eutectic temperature increases, and therefore, the α region is extended. Consequently, the volume fraction of the α₂ phase increases with the carbon content. After holding for 100 s, although the carbide content increased continuously, the volume fraction of the α₂ phase declined sharply. The phase area was shifted to a single γ area because the large amount of carbide in residual melts affected the increase in the aluminum content of the castings.

4. Conclusions

This study aimed to evaluate carbon contamination during the melting of γ-TiAl alloy in a graphite crucible by varying superheating and holding times and to analyze how changes in melt holding time affect the alloy’s microstructure and composition.

• At superheating of 130 K and 170 K, casting defects were observed at the disk and blade tip. However, the misrun and cold shut were resolved at temperatures above ΔT of 200 K. Considering contamination by impurities, it was found that superheating of 200 K is appropriate for casting a sound γ-TiAl turbocharger component.
• As the TiC and Ti₂AlC phases, which are richer in titanium, adhered to the crucible walls during casting, less titanium remained in the melt. Consequently, the proportion of aluminum in the melts increased, leading to higher aluminum concentration in the castings.
• The outer layers of solid TiC and Ti₂AlC, which enclosed the molten volume during the melting process, limited the increase in carbon contamination by creating a stable barrier that prevented carbon diffusion from the crucible into the melt. This reduced the rate of carbon contamination.
• As holding time increased, fine secondary carbides formed first due to differences in carbon solubility between the γ and α₂ phases—specifically, the γ phase could dissolve more carbon than the α₂ phase. As a result, carbon began to precipitate as fine secondary carbides, followed by the crystallization of primary carbides as the process continued. Both the fine secondary and coarse primary carbides were identified as Ti₂AlC phases, ensuring precise phase identification. This sequence led to a transformation of the columnar structure into an equiaxed one through TiC crystallization. Additionally, increased carbon contamination promoted more heterogeneous carbide nuclei in the molten state, resulting in refined grain size.
• The volume fraction of the α₂ phase increased as carbon acted as an α-phase stabilizer, reaching a maximum at a holding time of 100 s. As the holding time increased beyond 100 s, the volume fraction of the α₂ phase decreased sharply. Simultaneously, the γ phase region expanded due to the removal of Ti₂AlC and TiC from the melt, as these carbides adhered to the crucible wall.
• The carbon content was within 0.5 at% in the castings for a melt holding time of up to 50 s at a degree of superheating of 200 K. When the holding time was less than 20 s, carbon was present only as a solid solution in the castings.