1 Introduction

The demand for platinum group metals (PGMs, which include Pd and Rh) has been increasing owing to growth in the manufacturing of industrial products that require them, such as catalysts.1 There is a very high demand for PGM recycling because of the uneven distribution and scarcity of mineral resources. In the hydrometallurgical process used for recycling, a residue containing TiO2 is generated along with PGMs. In the present work, an additional pyrometallurgical process of residue treatment was considered in which PGMs are concentrated in the metal phase with Cu as the collector, while TiO2 separates into the slag phase containing SiO2 and CaO as fluxing agents.

Many studies have been conducted on titanium oxides in slag because of their relevance to the steel industry. Devries et al. and Muan measured the liquidus of the SiO2-CaO-TiO2 system.2,3 Mizoguchi and Ueshima reported on the phase diagram of the Ti2O3-CaO-Al2O3 system, Kang and Lee determined the phase diagram of the Fe-Al-Ti-O system, and Shi et al. studied the phase relations of the CaO-SiO2-TiO2-Al2O3-MgO system at 1250–1400 °C. Numerous researchers have investigated the oxidation state of Ti in slag.9,10,11,12,13 They reported that the percentage of Ti4+ decreases while the percentage of Ti3+ increases under the high-temperature reducing conditions typically encountered in steelmaking. Morizane et al. and Datta et al. reported on the formation of TiC in molten slag.10,14

In the current work, a recycling process for PGMs is considered that uses graphite as the refractory of the furnace, which could possibly lead to the generation of solid phases such as TiC or TiOx. Consequently, there is a concern that the presence of solid phases in the slag during melting will increase the viscosity of the slag, thereby preventing slag/metal separation and causing the loss of valuable metal to the slag. However, no study has reported on the equilibrium of carbon or TiC with Ti-containing slag. The goal of this study was to determine the slag composition region that remains liquid at a high C activity during this process. For this purpose, we attempted to experimentally determine the phase diagram of the SiO2-CaO-TiOx system containing TiC.

2 Experimental Procedure

A total of approximately 0.5 g of SiO2, CaO calcined from CaCO3, and TiO2 (rutile-type) powder of special reagent grade was mixed according to the target composition and placed in a graphite crucible with an inner diameter of 8 mm. The sample was then inserted into a reaction tube with a CO (99.95 vol% purity) gas flow rate of 150 mL/min, held at 1773 K (± 3 K) in a furnace using SiC heating elements, then water quenched. By comparing results of samples held at temperature for 60 and 100 h in a preliminary experiment, it was determined that 60 h was sufficient to reach equilibrium. Therefore, all further experiments in this study were held for 60 h.

A schematic of the experimental setup is shown in Fig. 1. After quenching, the samples were embedded in ethyl acetate resin, polished, and sputter-coated with platinum. The compositions of the equilibrium phases were determined by scanning electron microscopy with energy-dispersive X-ray microanalysis (SEM-EDX; JOEL, JSM-6510A). The analytical setting for SEM-EDX was an accelerating voltage of 15 kV, and the obtained characteristic X-rays were corrected using the ZAF method. For the TiC formation experiment, the molar ratio of Ti to C in TiC was determined by electron probe microanalysis (EPMA; JOEL, JXA-8230) because the analytical accuracy of C is inferior in EDX. The analytical setting for EPMA was an accelerating voltage of 15 kV, and the obtained characteristic X-rays were corrected using the ZAF method. To ensure statistical reliability, at least five points were measured in each phase using SEM-EDX and EPMA. TiO2, CaSiO3, and C, which were coated with platinum under the same conditions as the experimental samples, were used as standard samples for EPMA. Additionally, the equilibrium phases were identified by X-ray diffraction (XRD; RIGAKU, SmartLab) with Cu \(\mathrm{K\alpha }\) radiation.

Fig. 1
figure 1

Schematic of the experimental setup

Full size image

3 Results and Discussion

Table 1 shows the initial composition, compositions of the equilibrium phases and compositional standard deviations of each sample. For the samples in which TiC was formed, Table 2 shows the results of quantitative analysis of the TiC phase, the compositional standard deviations, and the molar ratio of Ti to C. Because it is difficult to determine the concentration of O in EPMA from standard samples, the residue of the quantitative values for Ti, C, Si, and Ca was regarded as the concentration of O.

Table 1 Experimentally determined equilibrium phase compositions of the SiO2-CaO-Ti3O5 system containing TiC equilibrated with graphite and CO gas at 1773 K (Carbon-saturated, \({p}_{\mathrm{CO}}=1\))
Full size table
Table 2 Compositions of TiC in equilibrium with the liquid phase at 1773 K
Full size table

The back-scattered electron images obtained by SEM-EDX, and the diffraction patterns obtained by XRD are shown in Fig. 2–4. Figure 2(a) shows the back-scattered electron image of the sample in equilibrium with the liquid phase and SiO2, and (b) shows the XRD results of this sample. Figure 3 and 4 present analogous results for equilibrium conditions involving (liquid + SiO2 + TiC) and (liquid + TiC), respectively. From the results of SEM-EDX and EPMA, it was found that the TiC had oxygen in solid solution. Studies on titanium oxycarbide have also been published.15,16,17,18,19,20 Because the present experiments were conducted under conditions like those studies, it is likely that titanium oxide became titanium oxycarbide when the titanium activity was large. In this study, the ratio of C in the Ti and C rich phase ranged from 0.7 to 0.95, and the diffraction pattern was consistent with that of TiC by XRD, so it was TiC. Si, Ca. O detected in the TiC phase was either solid-solution in TiC or present in the surrounding slag because TiC phase is fine.

Fig. 2
figure 2

Equilibrium with the liquid phase and SiO2; (a) back-scattered electron image, and (b) diffraction pattern

Full size image
Fig. 3
figure 3

Equilibrium with the liquid phase, SiO2, and TiC; (a) back-scattered electron image, and (b) diffraction pattern

Full size image
Fig. 4
figure 4

Equilibrium with the liquid phase and TiC; (a) back-scattered electron image, and (b) diffraction pattern

Full size image

In the case of samples in which the liquid phase and CaSiO3 were in equilibrium, clear back-scattered electron images that could differentiate the two phases could not be obtained, because the average mass numbers of the liquid phase and CaSiO3 phase were very close. Therefore, for these samples, the liquidus composition was assumed to be midway between the initial composition of the sample in which CaSiO3 was confirmed by XRD and the composition of the sample in which it became a homogeneous liquid.

In this experiment, we considered the degrees of freedom from the Gibbs phase rule. This rule is expressed by the following equation:

$$f = c - p + 2$$
(1)

where \(f\) is the degrees of freedom, \(c\) is the number of components, and \(p\) is the number of phases. In this study,\(c=5\) because the components are Si, Ca, Ti, C, and O. Therefore, there are seven maximum possible degrees of freedom. When the sample contained two solid phases different from the liquid phase, in addition to the gas phase and the graphite crucible, \(p=5\) and \(f=2\). When a gas phase at a pressure of 1 atm and a temperature of 1773 K are specified, \(f=0\) and the composition of the liquid phase is fixed. Thus, in this experimental system, the equilibrium compositions of the three phases do not change in the three-phase region.

In this study, we considered the \({p}_{{\mathrm{O}}_{2}}\) and the oxidation state of Ti. From thermodynamic calculations, when the \({a}_{\mathrm{c}}=1\) and the CO partial pressure \(\left({p}_{\mathrm{CO}}\right)=1\) at 1773 K, \({p}_{{\mathrm{O}}_{2}}={10}^{-15.7}\). The reaction is expressed as follows:

$$2{\text{C}}\left( {\text{s}} \right) + {\text{O}}_{2} \left( {\text{g}} \right) = 2{\text{CO}}\left( {\text{g}} \right){ }:{ }\Delta G_{{1773{\text{ K}}}}^{^\circ } = - 533729{ }\left( {\text{J}} \right)^{[21]}$$
(2)

In this study, the oxidation state of Ti in the liquid slag phase was not determined. According to the thermodynamics of the Ti-O binary system, Ti3O5 is the most stable form of titanium oxide at \({p}_{{\mathrm{O}}_{2}}={10}^{-15.7}\) and a temperature of 1773 K.7 Additionally, under similar conditions, it has been reported that the value of Ti3+/Ti4+ in the slag becomes smaller as the slag basicity or \({p}_{{\mathrm{O}}_{2}}\) increases.9,10,11,12,13 Considering these factors, the composition of the liquid phase was calculated from the concentrations determined for each of the metallic elements by SEM-EDX analysis, assuming that Ti in the liquid phase exists entirely as Ti3O5. The resulting phase diagram of the SiO2-CaO-TiOx system containing TiC is shown in Fig. 5. This figure shows that as the basicity \((\mathrm{mass\% CaO}/\mathrm{mass\% Si}{\mathrm{O}}_{2})\) decreases, TiC becomes less stable and more Ti3O5 dissolves in the slag. These phenomena show that the decrease in the slag basicity leads to the decrease of TiOx activity.

Fig. 5
figure 5

Phase diagram of the SiO2-CaO-Ti3O5 system containing TiC equilibrated with graphite and CO gas at 1773 K (Carbon-saturated, \({p}_{\mathrm{CO}}=1\))

Full size image

Because Si, Ca, and O are soluble in solid TiC as shown in Table 2, TiC is not a pure substance, although \({a}_{\mathrm{TiC}}=1\) is assumed. In this case, the TiC-equilibrated liquidus line can be regarded as an isoactivity line of TiOx in the slag. The activity of TiOx in this experiment was calculated using the FTdemoBASE of FactSage database.21 The Gibbs free energies of the reactions for the formation of typical titanium oxides and TiC are as follows:

$${\text{Ti}}\left( {\text{s}} \right) + \frac{1}{2}{\text{O}}_{2} \left( {\text{g}} \right) = {\text{TiO}}\left( {\text{s}} \right){ }:\Delta G_{{1773{\text{ K}}}}^{^\circ } = - 378019{ }\left( {\text{J}} \right)$$
(3)
$${\text{Ti}}\left( {\text{s}} \right) + \frac{3}{4}{\text{O}}_{2} \left( {\text{g}} \right) = {\text{TiO}}_{1.5} \left( {\text{s}} \right){ }:{ }\Delta G_{{1773{\text{ K}}}}^{^\circ } = - 522038{ }\left( {\text{J}} \right)$$
(4)
$${\text{Ti}}\left( {\text{s}} \right) + \frac{5}{6}{\text{O}}_{2} \left( {\text{g}} \right) = {\text{TiO}}_{1.667} \left( {\text{s}} \right){ }:\Delta G_{{1773{\text{ K}}}}^{^\circ } = - 565422{ }\left( {\text{J}} \right)$$
(5)
$${\text{Ti}}\left( {\text{s}} \right) + {\text{O}}_{2} \left( {\text{g}} \right) = {\text{TiO}}_{2} \left( {\text{s}} \right){ }:{ }\Delta G_{{1773{\text{ K}}}}^{^\circ } = - 626576{ }\left( {\text{J}} \right)$$
(6)
$${\text{Ti}}\left( {\text{s}} \right) + {\text{C}}\left( {\text{s}} \right) = {\text{TiC}}\left( {\text{s}} \right){ }:{ }\Delta G_{{1773{\text{ K}}}}^{^\circ } = - 161680{ }\left( {\text{J}} \right)$$
(7)

Using the reactions (3) to (7), the reactions of these titanium oxides with carbon are as follows:

$${\text{TiO}}\left( {\text{s}} \right) + 2{\text{C}}\left( {\text{s}} \right) = {\text{TiC}}\left( {\text{s}} \right) + {\text{CO}}\left( {\text{g}} \right):\Delta G_{{1773{\text{ K}}}}^{ \circ } = - 50526\left( {\text{J}} \right)$$
(8)
$${\text{TiO}}_{{1.5}} \left( {\text{s}} \right) + \frac{5}{2}{\text{C}}\left( {\text{s}} \right) = {\text{TiC}}\left( {\text{s}} \right) + \frac{3}{2}{\text{CO}}\left( {\text{g}} \right):\Delta G_{{1773{\text{ K}}}}^{ \circ } = - 39939\left( {\text{J}} \right)$$
(9)
$${\text{TiO}}_{{1.667}} \left( {\text{s}} \right) + \frac{8}{3}{\text{C}}\left( {\text{s}} \right) = {\text{TiC}}\left( {\text{s}} \right) + \frac{5}{3}{\text{CO}}\left( {\text{g}} \right):\Delta G_{{1773{\text{ K}}}}^{ \circ } = - 41032\left( {\text{J}} \right)$$
(10)
$${\text{TiO}}_{2} \left( {\text{s}} \right) + 3{\text{C}}\left( {\text{s}} \right) = {\text{TiC}}\left( {\text{s}} \right) + 2{\text{CO}}\left( {\text{g}} \right){ }:{ }\Delta G_{{1773{\text{ K}}}}^{^\circ } { } = { } - 68834{ }\left( {\text{J}} \right)$$
(11)

From these reactions, the activity of TiOx on the liquidus line equilibrated with solid TiC at \({a}_{\mathrm{C}}=1, {a}_{\mathrm{TiC}}=1,\) and \({p}_{\mathrm{CO}}=1\) at 1773 K were derived as \({a}_{\mathrm{TiO}\left(\mathrm{s}\right)}=0.0325, { a}_{{\mathrm{TiO}}_{1.5}\left(\mathrm{s}\right)}=0.0666, {a}_{{\mathrm{TiO}}_{1.667}\left(\mathrm{s}\right)}=0.0618,\mathrm{ and }{a}_{{\mathrm{TiO}}_{2}(\mathrm{s})}=0.00938\), respectively. The calculation results suggest that solid TiC is formed under the conditions of higher concentration of titanium oxide and high carbon activities, presumably leading to potential problems with slag/metal separation and efficiency of the proposed recycling process. It is thought that TiC formation can be suppressed by avoiding the use of graphite crucibles and reducing the carbon activity, for example, by using a SiC crucible.

4 Conclusion

The phase diagram of the C and CO-saturated SiO2-CaO-TiOx system containing TiC was prepared by the quenching method at 1773 K. The solubility of TiOx in the slag increased as the slag basicity decreased, and in the case of high TiOx activity, TiOx was reduced by C to form TiC. To form a slag with higher titanium solubility, the use of graphite crucibles should be avoided, and carbon activity should be reduced, for example, by using a SiC crucible.