Table 1 shows the designated chemical composition of FGH4096 superalloy, the raw materials of which were prepared in purities higher than 99.9%. We investigated three types of crucible materials (MgO, Al2O3 and MgO–spinel) in this study. The dimensions of the crucibles were 80 mm in outer diameter, 60 mm in inner diameter and 130 mm in inner height. The crucibles were composed of chemical reagents with a purity higher than 99%. The properties of the crucibles, as provided by the manufacturer, are shown in Table 2. The chemical compositions of the three types of crucibles were analyzed using an X-ray fluorescence spectrometer (XRF). All crucibles were produced using cold isostatic pressing and ordinary pressure sintering techniques to improve their resistances to thermal shock. The sizes of the ceramic particles (Al2O3, MgO) ranged between 0–1 mm.
Cr Co Mo W Ti Al Nb C Ni 15.0–16.5 12.5–13.5 3.8–4.2 3.8–4.2 3.5–3.9 2.0–2.4 0.6–1.0 0.02–0.05 Bal.
Table 1. Designated chemical composition of FGH4096 superalloy
wt% Crucible MgO / wt% Al2O3 / wt% SiO2 / wt% TiO2 / wt% Porosity / % Bulk density / (kg·m−3) MgO 99.32 0.24 0.21 0.20 17.5 3.0×103 Al2O3 0.09 99.64 0.10 0.11 17.0 2.9×103 MgO–spinel 85.68 13.39 0.44 0.41 17.2 3.7×103
Table 2. Properties of crucibles in current work
A 2-kg-scale VIM furnace was utilized in the experiments. Fig. 1 shows a schematic of the experimental apparatus. A total of 1 kg of raw materials was used in each heat. As late additions, reactive elements (Al and Ti) were added to the melt through the hopper to prevent the formation of oxides and minimize the in-melt time for low-melting elements and thereby prevent burning off. First, the VIM chamber was evacuated to a vacuum degree of 1×10−2 Pa and then backfilled with pure argon to 0.05 MPa before melting. Then, the furnace was heated to 1773 K at a rate of 25 K/min and maintained at 1773 K for 10 min to ensure homogenization prior to start time. The experimental times of each group of crucibles were 5, 10, 30, and 60 min, respectively. When an experiment was finished, the molten alloy was poured into a steel mold of diameter 40 mm. Table 3 lists the experimental conditions.
Figure 1. Schematic of experimental apparatus. 1—High frequency power supply; 2—Hopper; 3—Crucible; 4—Induction coil; 5—Rotating axis; 6—Molten alloy; 7—Steel mold; 8—Viewing window; 9—Pressure gage; 10—Diffusion pump; 11—Mechanical pump; 12—Ar gas; 13—Gas outlet.
Heat No. Crucible material Melting time / min Composition after experiments / wt% Al Ti Mg O A1 Al2O3 5 2.138 3.815 <0.0001 0.0039 A2 Al2O3 10 2.137 3.815 <0.0001 0.0037 A3 Al2O3 30 2.136 3.814 <0.0001 0.0030 A4 Al2O3 60 2.137 3.815 <0.0001 0.0041 MS1 MgO–spinel 5 2.136 3.814 0.0002 0.0030 MS2 MgO–spinel 10 2.135 3.815 0.0004 0.0028 MS3 MgO–spinel 30 2.134 3.814 0.0007 0.0025 MS4 MgO–spinel 60 2.134 3.814 0.0010 0.0030 M1 MgO 5 2.134 3.815 0.0007 0.0023 M2 MgO 10 2.132 3.814 0.0016 0.0015 M3 MgO 30 2.131 3.814 0.0017 0.0019 M4 MgO 60 2.130 3.814 0.0019 0.0016
Table 3. Experimental conditions and composition of FGH4096 superalloy for each heat
After VIM, we analyzed the O content by the LECO (TC-300 combustion analyzer) method, which uses NIST-certified standards with O values accurate to 0.0001wt%. The Al, Ti, and Mg contents in all the samples were determined using inductively coupled plasma-atomic emission spectroscopy (ICP-AES). The compositions of the inclusions of the metallographic samples were characterized using a scanning electron microscope (SEM) equipped with an energy dispersive spectrometer (EDS) with the acceleration voltage of 20 kV. The crucible samples were cut from the same location (10 mm below the liquid-alloy surface). The inner walls of the crucible samples were analyzed using X-ray diffraction (XRD). We also analyzed the interface of the crucible by SEM and EDS using map scanning.
2.1. Raw materials
2.2. Experimental procedure
Based on their morphologies and chemical compositions, the inclusions can be classified into four types, namely, (1) Al2O3 oxides, (2) Al–Ti oxides, (3) Al–Mg oxides, and (4) Al–Mg–Ti oxides. Fig. 2 shows SEM images and the EDS results for these inclusion types. Table 4 shows the specifications of each type of inclusion.
Figure 2. Morphology and EDS result of each type of inclusion: (a) type 1; (b) type 2; (c) type 3; (d) type 4.
Type Specification of inclusions Heats (Al2O3) Heats (MgO–spinel) Heats (MgO) A1 A2 A3 A4 MS1 MS2 MS3 MS4 M1 M2 M3 M4 1 Al2O3 ×× ×× ×× ×× ×× × × × 2 Al–Ti oxide × × × × × × × × 3 Al–Mg oxide × ×× ×× ×× ×× ×× ×× ×× 4 Al–Mg–Ti oxide × × × × × × × × Note: ××—Main type of inclusions; ×—Small amount of inclusions.
Table 4. Types of oxide inclusion in different heats
The inclusions in Heats A1–A4 were mainly Al2O3 with some minor Al–Ti oxide, and were mostly irregular in shape. Some were deoxidation products from the addition of Al and Ti. Others were crucible particles that had become detached because of the physical erosion of the melt under the electromagnetic stirring force. The compositions of the inclusions did not change as a function of the melting time, which suggests that almost no chemical reaction occurred between the alloy melt and the Al2O3 crucible. When the MgO–spinel crucible was applied, Al2O3 comprised the main inclusions at 5 min (Heat MS1), whereas nearly spherical Al–Mg oxides were the main inclusions after 10 min. When the MgO crucible was used, Al–Mg oxides were the main inclusions in Heats M1–M4. At 5 min, Al2O3, Al–Ti oxides, Al–Mg oxides, and Al–Mg–Ti oxides were all observed. At 10, 30, and 60 min, only Al–Mg oxides and Al–Mg–Ti oxides were observed, which indicates that the Al2O3 and Al–Ti oxides had been completely converted to MgO-containing oxides.
Fig. 3 shows the variation in the number densities of oxides with time, in which we can see that the number densities of the oxides in Heats M1–M4 are lower than those in Heats A1–A4 and Heats MS1–MS4 at the same melting time. More inclusions are observed in Heats A1–A4. In addition, the amount of inclusions first decreases with time and then increases at 60 min in the cases of the Al2O3 and MgO–spinel crucibles, whereas in the case of the MgO crucible, the inclusions continue to decrease at 60 min. The fluid flow of the molten superalloy under electromagnetic force favors the floating of the inclusions to the top [24–26].
Table 3 and Fig. 4 show the compositions of the alloys and inclusions, respectively. In Heats A1–A4, the Al, Ti, and Mg contents change only slightly with time. In Heats MS1–MS4 and Heats M1–M4, the Al contents decrease and the Mg contents increase with time, with the Ti contents changing very little. However, the Mg contents in the alloys in Heats M1–M4 are higher than those in Heats MS1–MS4 at the same melting time (see Fig. 4(a)). In the case of the MgO crucible, the Mg content increases sharply up to 10 min of reaction time and then increases slowly after 10 min of melting. The ratio of increase in the Mg content in Heats MS1–MS4 is nearly the same as that in Heats M1–M4 when the melting time is longer than 30 min, which suggests that they have nearly the same reaction rate. As shown in Fig. 4(b), in Heats A1–A4 and MS1–MS4, the O contents first decrease when the melting time is less than 30 min, and then increase at 60 min, similar to the variation in the number densities of the oxides in the superalloy. However, the O contents in Heats M1–M4 decrease with melting time in general. In addition, the O contents in Heats M1–M4 are lower than those in Heats A1–A4 and MS1–MS4 at all four melting times. Furthermore, Fig. 5 shows the Mg/O ratios in the superalloys and the MgO contents in the inclusions, in which we see that the Mg/O ratios in Heats M1–M4 are high, as are the MgO contents. It is interesting that the MgO contents in the inclusions linearly increase with the ratio of Mg/O in Heats MS1–MS4.
Fig. 6(a) shows an elemental map of a cross section of the MgO crucible prior to VIM. After 5 min of reaction time, an Al band appears at the interface. Fig. 6(b) shows an elemental map of the MgO crucible after 60 min of reaction time, in which the Al band is clear and continuous. Studies [18,20–21] have indicated that the alloy–crucible reaction can generate solid or gaseous products, which may adhere to the crucible wall and hinder further reaction. Furthermore, our XRD results (see Fig. 7(a)) show that an MgAl2O4 phase occurs at the crucible walls, which suggests that the Al band is actually a MgAl2O4 layer. In addition, the thickness of the Al band of the MgO crucibles increases as a function of melting time, as shown in Fig. 8. The Al band is discontinuous in Heat M1, but more continuous in Heat M2. Continuous and thick Al bands are evident in Heats M3 and M4. All the above information regarding the MgO crucible indicates that MgO reacts with the FGH4096 superalloy melt. It is interesting that the inner wall of the MgO crucible becomes smooth after VIM, which is considered to be preferable. However, unlike the MgO crucibles described above, no new elemental enrichment is observed at the interface of the Al2O3 crucible after VIM, as shown in Fig. 9. The XRD results of the inner walls of the Al2O3 crucibles (Fig. 7(b)) further demonstrate that no new phase forms. Similar to the Al2O3 crucibles, no new elemental enrichment is observed at the inner walls of the MgO–spinel crucible after VIM, as shown in Fig. 10. Fig. 7(c) shows the XRD results for the inner walls of the MgO–spinel crucible. However, this does not mean that MgO–spinel does not react with the melt, because the reaction products Al2O3 and spinel have the same composition as they do in the crucible. It is difficult to distinguish between them.
Figure 7. XRD spectra of inner walls of the three groups of crucibles: (a) MgO crucibles; (b) Al2O3 crucibles; (c) MgO–spinel crucibles.
Figure 10. Elemental maps of MgO–spinel crucible interface: (a) before VIM; (b) after 60 min of VIM.
Additionally, for all three groups of crucible samples, it is interesting to note that no Ti elemental enrichment was observed at the interface, although Ti was also active and its content (3.8wt%) was higher than Al (2.1wt%).
During the VIM process, the interaction between the crucible and alloy includes a dissolution reaction, chemical reaction, and physical erosion. In many cases, such as industrial production, the use of top slag cannot be fully avoided for economic reasons. The predominant reaction is that between the refractory and the liquid slag [27–28], which makes the interaction more complex. In the present study, to prevent its influence on the chemical composition of the superalloy and crucibles, we used no top slag.
In the dissolution reaction, Mg and Al might be introduced into the melt by the dissolution of their oxides (MgO and Al2O3, respectively), which can be expressed as shown in Eqs. (1) and (2).
for which the equilibrium constants become:
Thus, if the activity product αMg × αO in the alloy is less than K1, Eq. (1) can proceed in the forward direction and MgO can dissolve in the superalloy. Of course, whether Eq. (1) is feasible or not will depend on the solubility of Mg and oxygen in the superalloy and the kinetics of that dissolution. In the present study, the Mg content is very low, so Eq. (1) works. The Al2O3 crucible follows a similar rule, but the Al content is as high as 2.2wt%, which makes the dissolution of Al2O3 impossible. In other words, the dissolution of Al2O3 is very weak.
The chemical reaction between the MgO crucible and the superalloy has two stages, as shown in Fig. 11. In Stage 1, the MgO crucible reacts with the Al of the molten alloy, which can be expressed as shown in Eq. (5) . This reaction generates dissolved Mg in the alloy and Al2O3, which adheres to the inner wall of the crucible to form an unstable Al2O3 layer. The formation of dissolved Mg plays a significant role in the evolution of the inclusions. Studies [29–31] have indicated that Al2O3 inclusions are not stable and transform into MgAl2O4 when dissolved Mg is present in the alloy melt. This phenomenon also occurs in Al–Ti oxide inclusions because the Ti2O3 of the inclusion is less stable than Al2O3. In Stage 2, Al2O3 reacts with the MgO of the crucible to form a stable MgAl2O4 spinel layer, as shown in Eq. (6). On the other hand, dissolved Mg can reduce the Al2O3 and Ti2O3 in the inclusions, which can be expressed as shown in Eqs. (7) [20,28] and (8), respectively. This modifies the composition of the inclusions such that the unstable Al2O3 and Al–Ti oxide inclusions are transformed into stable Al–Mg oxides.
where subscript “incl” is the abbreviation of inclusion. The mechanism of the chemical reaction between the MgO–spinel crucible and the superalloy is similar to that between the MgO crucible and the superalloy. However, the Mg contents in Heats MS1–MS4 are lower than those in Heats M1–M4 for the same melting times. This is due to the presence of spinel, which occupies a large proportion of the inner wall of the crucible and therefore hinders the chemical reaction between the Al of the alloy and the MgO of the crucible. However, we observed no chemical reaction between the Al2O3 crucible and the superalloy, which suggests that Ti does not react with the Al2O3 crucible because no reaction product forms at the interface.
In addition to the dissolution and chemical reactions, physical erosion occurs between the crucible and the melt . In the VIM process, the electromagnetic stirring force applied to the surface causes relative movement between the melt and the crucible, which mechanically erodes the crucible. With continuous stirring, crucible particles become detached and disperse into the melt and a fresh crucible surface is then exposed to further erosion. Some of these detached crucible particles become dissolved in the melt, which contaminates the metal with oxygen. Others remain dispersed in the melt and float to the top surface due to the density difference between the melt and the oxide particles. In the present study, extensive detachment of Al2O3 crucible particles occurred in the melt because the inner surface had already become rough at 5 min, as shown in Fig. 12. This is why the oxygen content and number of inclusions increased after 30 min. However, the inner surface of the MgO crucible remained smooth even at 60 min. This is because the newly formed MgAl2O4 spinel layer can withstand erosion by high-temperature superalloy melt even if its thickness is just 36 μm at 60 min.
In summary, the dominant interaction between the Al2O3 crucible and the superalloy is physical erosion, whereas those of the MgO and MgO–spinel crucibles and the superalloy are the dissolution and chemical reactions. The MgO crucible is preferential when VIM is used with the FGH4096 superalloy because the MgAl2O4 layer that forms prevents contact between the MgO crucible and the superalloy and inhibits further erosion of the crucible.