Experimental Study on the Phase Relations of the SiO2-MgO-TiO2 System in Air at 1500°C

We investigated the phase relations of the SiO2-MgO-TiO2 system in air at 1500°C using the high-temperature isothermal equilibration/quenching technique, followed by x-ray diffraction measurements and direct phase analysis using scanning electron microscopy coupled with x-ray energy dispersive spectrometry. One single liquid phase domain, five two-phase domains (liquid-TiO2, liquid-cristobalite, liquid-MgO·SiO2, liquid-2MgO·SiO2, and liquid-MgO·2TiO2), and five three-phase regions (liquid-TiO2-MgO·2TiO2, liquid-MgO·SiO2-cristobalite, liquid-TiO2-cristobalite, liquid-MgO·SiO2-2MgO·SiO2 and liquid-2MgO·SiO2-MgO·2TiO2) were observed. We constructed a 1500°C isothermal phase diagram based on the experimentally measured liquid compositions. We compared simulations using MTDATA and FactSage thermodynamic software and their databases with the experimental results obtained in this study. These results can be used to provide guidelines for updating the MTDATA and FactSage titania-bearing thermodynamic databases by reassessing the thermodynamic properties of the phase with new experimental data.

The phase relations of the CaO-TiO 2 , 45,46 MgO-TiO 2 , 47-50 SiO 2 -TiO 2 , [51][52][53][54][55] 70,71 Regarding the SiO 2 -MgO-TiO 2 system, Massazza and Sirchia 72 carried out the first study on determining the liquidus of the system in 1957 at 1500-1700°C. The detailed experimental procedures were not reported in their paper. Moreover, a loss due to volatilization of SiO 2 has been found for mixtures rich in silica and melted at high temperatures, which might cause inaccuracies for the determined phases. MacGregor 73 determined the liquidus and solidus relationships in the Mg 2 SiO 4 -MgSiO 3 -TiO 2 -MgTi 2 O 5 domain of the MgO-SiO 2 -TiO 2 system at 1500-1900°C. Hermann et al. 74 measured the solubility of TiO 2 in olivine in the MgO-SiO 2 -TiO 2 system under conditions of atmospheric pressure at 1200-1500°C, but without illustrating the liquid domain. They reported that the highest solubility of TiO 2 was obtained when the olivine was equilibrated with spinel (2MgOAETiO 2 ). 74 Yan et al. 75 determined the phase equilibria of the SiO 2 -MgO-TiO x quasi-ternary system under reducing atmospheres at 1600°C. The isothermal SiO 2 -MgO-TiO x phase diagram at P O2 of 4.85 9 10 À11 atm was constructed and the effect of P O2 on the phase relations was demonstrated by comparison with the results by Massazza and Sirchia. 72 Moreover, the isotherms of the anosovite primary phase field were also determined at P O2 of 1.94 9 10 À9 and 2.75 9 10 À13 atm, suggesting that the solubility of TiO x in the liquid phase decreased with decreasing P O2 .
In summary, the SiO 2 -MgO-TiO 2 system was either not systematically investigated or there are significant discrepancies between the already existing studies. Therefore, the purpose of the present study was to investigate the phase relations in the SiO 2 -MgO-TiO 2 system in air at 1500°C to resolve the current existing contradictions and to provide experimentally reliable, fundamental thermodynamic data for the thermodynamic assessment of the CaO-SiO 2 -MgO-Al 2 O 3 -TiO 2 system. The results are expected to provide guidance for the selective crystallization of TiO 2 and MgOAE2TiO 2 phases from the slags by modifying the composition of industrial titania-bearing blast furnace slags.

EXPERIMENTAL
The starting materials used for synthesizing the slags are listed in Table I. Each powder was weighed in a certain ratio based on the designed composition and mixed thoroughly in an agate mortar before being pressed into a pellet using a hydraulic press with a force of 5 metric tons.
The experiments were followed by high-temperature isothermal equilibration, quenching, and direct phase analysis by applying the XRD and SEM-EDS techniques. A vertical alumina tube furnace (Nabertherm, RHTV 40-250/18, Germany) was used for the high-temperature equilibration experiments, as shown in Fig. 1. Synthetic slag (0.15-0.3 g) was pre-heated at 1600°C for 30 min and then equilibrated at 1500°C for each experiment. A pure platinum foil was used for making crucibles to support the sample pellet. A platinum wire inserted into the furnace from the top end of the guiding tube was used to hold the crucibles. The sample temperature was monitored by an alumina-shielded S-type Pt/90%Pt-10%Rh thermocouple placed next to the sample. The samples were annealed at 1500°C for equilibration, and then quenched in an ice-water mixture to retain the phase assemblies from a high temperature. 76,77 The detailed experimental procedures used were described in our previous studies. [32][33][34]44,64 The quenched samples were dried at room temperature, and part of the sample was mounted in epoxy resin, then ground and polished using a metallographic polishing cloth with diamond sprays. The polished sample surfaces were carboncoated using a LEICA EM SCD50 sputtering device (Leica Mikrosysteme, Austria). A Tescan MIRA 3 scanning electron microscope (SEM, Tescan, Brno, Czech Republic) equipped with an UltraDry silicon drift energy dispersive X-ray spectrometer (EDS, Thermo Fisher Scientific, Waltham, MA, USA) and NSS microanalysis software were used to measure the equilibrium phase compositions. An accelerating voltage of 15 kV, beam current of 20 nA, and a working distance of 20 mm were adopted for the SEM-EDS analysis. The standards and analyzed Xray lines in the EDS analysis were as follows: quartz 48.57 30.14 21.29 Theoretically, a high silica concentration with a rigid silicon-oxygen network and high viscosity results in much slower mass transfer than relatively low silica concentrations. 79,80 Therefore, we used a sample with an initial high silica concentration for the time series. We decided that equilibration had been achieved based on the stabilization of the TiO 2 , SiO 2 , and MgO concentrations in the liquid phase as well as the SiO 2 and TiO 2 concentrations in solid cristobalite, as shown in Fig. 2a. It can be observed that equilibration in the present system was achieved in 8 h. However, with the purpose of ensuring sufficient growth of crystals, all samples were annealed at the experimental temperature for 12 h.
As is known, the element titanium exists as Ti 4+ , Ti 3+ , and Ti 2+ , depending on the prevailing oxygen partial pressure and temperature. Therefore, it was critical to identify the valence state of Ti in the samples of the present study. Based on the prediction of the Ti-O stability phase diagram, 34,44,64 we found that the predominant form for titanium oxide was TiO 2 under the present experimental conditions of air (log 10 P O2 of À 0.6778) at 1500°C. Pure TiO 2 was calcined in air at 1500°C for 24 h to verify the theoretical prediction. The XRD pattern for the calcined pure TiO 2 is shown in Fig. 2b, indicating that pure TiO 2 cannot be reduced to the lower valence states under the present experimental conditions. In our previous studies, 81,82 the oxidation state of Ti was also examined by XPS and it was proved that titanium existed in the samples as Ti 4+ . Thus, it can be concluded that Ti 4+ is the predominant stable valence state for titanium in all samples of the present study. Subsequently, the titanium oxide is presented as TiO 2 in the following sections.

Phase Relations at 1500°C
The equilibrium phase compositions, typical microstructures, and their corresponding XRD patterns are shown in Table II, Figs. 3, and 4, respectively. We observed one homogeneous single liquid domain, five liquid-solid two-phase coexistence domains (i.e., liquid-TiO 2 , liquid-cristobalite, liquid-MgOAESiO 2 , liquid-2MgOAESiO 2 , and liquid-MgOAE2TiO 2 ), and five liquid-solid-solid three-phase coexisting regions (i.e., liquid-TiO 2 -cristobalite, liquid-MgOAESiO 2 -2MgOAESiO 2 , liquid-2MgOAESiO 2 -MgOAE2TiO 2 , and liquid-MgOAE2TiO 2 -TiO 2 ). It should be noted that in the XRD pattern for sample C5 for liquid-TiO 2 twophase equilibrium, there are several unindexed peaks with relatively low intensities, which may have been caused by the introduction of contaminants during the preparation of powder samples for XRD analysis. To demonstrate the reliability of EDS for determining the phase compositions, the measured TiO 2 results were compared with its theoretical value. The deviation between the theoretical    and experimentally measured TiO 2 composition was found to be lower than 0.3%, indicating the present results measured by EDS are reliable enough for constructing the phase diagram of the SiO 2 -MgO-TiO 2 system. As can be observed in Table II, a certain fraction of TiO 2 was contained in the cristobalite when it was equilibrated with the liquid phase, showing a strong relationship with the TiO 2 concentration of the conjugate liquid phase. Similar observations were reported by Kirschen et al. 83 We plotted the experimentally measured TiO 2 concentrations in cristobalite with the obtained standard deviations as a function of TiO 2 concentration in the liquid oxide phase, see Fig. 5. For some points, the standard deviations were too small to be reproduced. The predictions by MTDATA 84,85 using the Mtox 8.2 database 85 were also plotted in the graph for comparison. As shown in Fig. 5, the lowest experimentally measured TiO 2 concentration in the cristobalite phase did not differ significantly from the prediction by MTDATA. However, the maximum TiO 2 content in the cristobalite phase was approximately 2.5 wt.% lower than the value predicted by MTDATA. Furthermore, the minimum and maximum TiO 2 solubilities in the liquid phase calculated by MTDATA were around 4 wt.% and 5 wt.% lower than the present experimental results, respectively.
The liquid-solid-solid three-phase equilibria observed in the present work can be further analyzed by the Gibbs phase rule, 86 described by Eq. 1: where f refers to the degrees of freedom, and C and P refer to the number of independent components and equilibrium phases, respectively. The number 2 refers to the temperature and total pressure variables. As the present study was carried out at a fixed temperature and atmospheric pressure with P = 1 atm in air, Eq. 1 can be further simplified as Eq. 2: According to the Gibbs phase rule, there could be a maximum of 3 phases coexisting in the MgO-SiO 2 -TiO 2 ternary system (f = 0, C = 3, then P = 3) under the present experimental conditions. Therefore, the liquid composition of the liquid-solid-solid threephase equilibrium was constrained to an invariant point in the MgO-SiO 2 -TiO 2 ternary system. The liquid-TiO 2 -MgOAE2TiO 2 three-phase equilibrium was confirmed in samples C7 and C11; the liquidcristobalite-MgOAESiO 2 equilibrium was determined in samples C52, C54, and C55. Thus, the invariant point for the liquid oxide-TiO 2 -MgOAE2TiO 2 coexisting equilibrium was determined by using the average of samples C7 and C11, giving a composition of 50.

Construction of the 1500°C Isothermal Phase Diagram
Based on the present experimental data listed in Table II, a 1500°C isothermal section of the MgO-SiO 2 -TiO 2 system was constructed, as shown in Fig. 6a. The primary phase fields of cristobalite and MgOAESiO 2 were experimentally determined, and the liquid-cristobalite-MgOAESiO 2 three-phase coexisting equilibrium was observed in this study, thus the phase domain for liquid-cristobalite-MgOAESiO 2 coexisting equilibrium can be constructed by connecting the solid composition points of cristobalite and MgOAESiO 2 , and subsequently forming another triangular region #12 for SiO 2 -cristobalite-MgOAESiO 2 three-phase coexisting equilibrium. The area marked with a light gray color was not investigated due to the limited number of experimental points. Data from the literature 72  We found that the experimentally determined isotherm for the cristobalite primary phase field, shown in Fig. 6b, agreed in general with the results by Massazza and Sirchia. 72 However, the present results in the TiO 2 primary phase field shifted slightly to the region with higher TiO 2 content when compared with the observation by Massazza and Sirchia. 72 The 2MgOAESiO 2 primary phase field obtained in the present study was much wider than the results in the literature, 72 expanding toward the region of lower TiO 2 content. The liquid-2MgOAESiO 2 -MgOAE2TiO 2 three-phase invariant point determined in the present study shows a composition with relatively low TiO 2 and high SiO 2 concentrations. The results for the MgOAESiO 2 primary phase field in the present study are narrower than those found by Massazza and Sirchia, 72 whose results covered significantly wider ranges of SiO 2 and TiO 2 content than the present study. Figure 6c shows a comparison of the present results with the calculations made by FactSage. The liquid oxide domain determined in this study was much smaller than that calculated by FactSage. The primary phase field of 2MgOAESiO 2 according to FactSage was wider than the present results, expanding toward both the higher and lower TiO 2 content regions. The present isotherm for the MgOAESiO 2 primary phase field shifted slightly to the area with higher TiO 2 content. The isotherms for the TiO 2 and MgOAE2TiO 2 primary phase fields measured in this study exhibited lower TiO 2 content but higher SiO 2 concentration than the predictions by FactSage. Figure 6d shows a comparison of the present results with the simulations by MTDATA. The liquid domain simulated by MTDATA is larger than that of the present results. The isotherm for the MgOAESiO 2 primary phase field obtained in this study displayed a higher TiO 2 content compared with the results by MTDATA. The present results in the cristobalite primary phase field generally agreed well with the predictions by MTDATA in the lower TiO 2 content region, but the present experimentally measured liquid-cristobalite-TiO 2 three-phase invariant point showed a higher TiO 2 content. The isotherm constructed in the MgOAE2-TiO 2 primary phase field in the present study shifted to a higher SiO 2 content area with a similar trend to the behavior simulated by MTDATA. The present experimental results of the primary phase field of 2MgOAESiO 2 showed a lower MgO content. However, the primary phase fields of MgOAETiO 2 and 2MgOAETiO 2 predicted by MTDATA were not observed in the present study.

SUMMARY AND CONCLUSION
The phase equilibria of the SiO 2 -MgO-TiO 2 system were investigated at 1500°C in air using the high-temperature isothermal equilibration/quenching/SEM-EDS/XRD analysis technique. We determined the isotherms for the cristobalite, MgOAESiO 2 , 2MgOAESiO 2 , MgOAE2TiO 2 , and TiO 2 primary phase field.
Based on the experimental results, we generated a 1500°C isothermal section for the SiO 2 -MgO-TiO 2 system. We compared the present results with the available data from the literature and with simulations by FactSage and MTDATA. The present results in the cristobalite primary phase field agreed well with the observations in the literature, but showed strong discrepancies in the primary phase fields of MgOAESiO 2 and TiO 2 . The present results provide useful thermodynamic information for the selective crystallization of TiO 2 and MgOAE2-TiO 2 phases from the titania-bearing slags by adjusting their compositions. The simulated data by FactSage and MTDATA deviated significantly from the present results. The current experimentally measured data provide guidelines for updating the titania-bearing thermodynamic databases of FactSage and MTDATA thermodynamic software and fundamental information on thermodynamic modeling of more complicated high-order titania systems.  Number 2020TQ0059, 2020M680967] and the Natural Science Foundation of Liaoning Province. Mr. Junmo Jeon is greatly appreciated for his assistance in XRD analyses. The authors declare that they have no conflict of interest.

FUNDING
Open access funding provided by Aalto University.

OPEN ACCESS
This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit h ttp://creativecommons.org/licenses/by/4.0/.