Phase equilibria in TiO2-rich part of the MgO–CaO–TiO2 system at 1500–1600 °C

The equilibrium phase relations of the MgO – CaO – TiO 2 system were investigated at 1500 – 1600 ◦ C in air using the high-temperature isothermal equilibration of samples in platinum crucibles followed by the rapid quenching technique. The phase compositions were analyzed using an electron probe X-ray microanalyzer (EPMA). The isothermal sections at 1500 ◦ C and 1600 ◦ C were constructed based on the experimentally determined liquid phase compositions. The present results deviate significantly from previous observations as well as the predictions by MTDATA and FactSage using their oxide databases, in particular in the perovskite primary phase field


Introduction
Titanium alloys are widely used in the fields of aviation, aerospace, medicine, and construction [1], for example, due to their properties of low density, high strength, good corrosion resistance, and non-toxicity [2].The production of titanium from its mineral ores involves the carbochlorination of rutile and ilmenite to obtain TiCl 4 , followed by metallothermic reduction with magnesium [3,4].However, large amounts of energy consumption and greenhouse gas emissions are associated with this production process.Thus, a cleaner and more efficient way of titanium production is required.
Abundant data are available for the phase relations of the sub-binary systems of CaO-MgO [33][34][35][36][37][38][39][40][41], CaO-TiO 2 [42,43], and MgO-TiO 2 [44][45][46][47].However, it seems that there are limited fundamental data available on the CaO-MgO-TiO 2 system.Coughanour et al. [48] have determined the binary joints in the CaO-MgO-TiO 2 system at 1450-1750 • C using the quenching technique and phase characterization by X-ray and petrographic, although the liquid domain information was not reported in their study.Rouf et al. [49] have investigated the CaO⋅TiO 2 -MgO⋅TiO 2 -TiO 2 portion of the CaO-MgO-TiO 2 system using a hot-stage microscope, micro-differential thermal analysis, and X-ray diffraction techniques.The liquidus temperatures were determined using 5 wt% Rh-Pt/20 wt% Rh-Pt thermocouples.However, they reported that the hot-stage microscope method was not reliable for samples with a TiO 2 content of less than 70 wt%.Furthermore, the method of DTA and visual observations in a hot-stage microscope during cooling the sample with a high cooling rate (10-90 • C/min) brought the difficulties of reaching real equilibrium.They reported the repeatability/accuracy of their temperature data from ±5 to ±10 • C but without examining the effect of cooling rate.Typically, dynamic methods result in lower liquidus temperatures due to undercooling, but that effect was not studied.Shultz [50] have determined the solidus relations of the CaO-MgO-TiO 2 system over the temperature range 1407-1670 • C in air using the quenching method.
In summary, the phase relations of the MgO-CaO-TiO 2 system have not been investigated systematically.Therefore, this study aimed to investigate the phase equilibria of the MgO-CaO-TiO 2 system at 1500-1600 • C in air using the high-temperature equilibration/quenching/SEM-EDS/EPMA technique.The acquired experimental phase equilibrium data are useful for optimizing the thermodynamic properties in higher-order systems.

Experimental
The slag mixtures were prepared using high purity powders of CaO (99.9 wt%, Sigma-Aldrich), MgO (99.95 wt%, Alfa Aesar), and TiO 2 (99.8 wt%, Sigma-Aldrich).For each sample, the powders were weighed in an appropriate ratio, mixed in an agate mortar, and then pressed into a pellet using a hydraulic press.Approximately 0.2 g of molten sample was equilibrated at high temperature for each experiment.
The experiments involved high-temperature isothermal equilibration, quenching in an ice-water mixture, and phase characterization by SEM-EDS/EPMA on polished sections.The high-temperature equilibration of samples was conducted in a vertical tube furnace (Nabertherm, RHTV 40-250/18, Germany) equipped with silicon carbide heating elements.A calibrated S-type Pt/Pt-10% Rh thermocouple was placed next to the sample to measure the sample temperature, as shown in Fig. 1.The samples were pre-melted at a temperature 50 • C higher than the equilibration temperature for 30 min and then lowered to 1500 • C or 1600 • C for equilibration.It had already been proved in our previous studies that at experimental temperatures the system can reach equilibrium within 24 h [18,28,51].Therefore, all samples were annealed at high temperature for at least 24 h.After equilibration, the sample was quenched in an ice-water mixture, dried in air, mounted in epoxy, ground, and polished by wet metallographic methods, and carbon-coated using a carbon vacuum evaporator (LEICA EM SCD050, Austria).The experimental techniques have been described in detail in our previous papers [52][53][54].
The microstructures and phase compositions of samples were preexamined using a scanning electron microscope (SEM, Tescan MIRA 3, Brno, Czech Republic) coupled with an UltraDry Silicon Drift Energy Dispersive X-ray Spectrometer (EDS, Thermo Fisher Scientific, Waltham, MA, USA).The elemental compositions of all phases were accurately determined using a Cameca SX100 Electron Microprobe (Cameca SAS, Genevilliers, France) using the Wavelength Dispersive Spectrometers (WDS).The operation parameters for EPMA analyses were accelerating voltage of 20 kV and beam current of 60 nA.A defocused beam diameter set to 50 μm for the liquid phase and 10-20 μm for the solid phase was used.The standards employed for EPMA were naturally occurring minerals rutile for O Kα and Ti Kα, diopside for Mg Kα, and bustamite for Ca Kα.The PAP-ZAF matrix correction procedure [55] was used for the raw data processing.At least 8 points were selected from the well quenched area of each phase in the EPMA and EDS analyses.The standard deviation (±1σ) of the EPMA measurements is shown as uncertainty of each point.

Equilibrium phase relations at 1500 • C
The microstructures and phase compositions of samples obtained at 1500 • C are shown in Fig. 2 and Table 1, respectively.The valence state of titanium in samples equilibrated in air has been investigated in our previous studies [51,53,54] by XPS, XRD, and the Ti-O stability phase diagram.The main titanium oxide species in the slags in air was found to be TiO 2 .Therefore, the concentration of titanium oxide in the present study was calculated from the primary data as TiO 2 .Several liquid-solid phase equilibria (i.e., liquid-rutile, liquid-pseudobrookite, and liquid-perovskite) and three-phase equilibria, including liquid-rutile-pseudobrookite and liquid-ilmenite-perovskite, were observed at 1500 • C.
The relationship between the number of phases and the number of components of an equilibrium system can be obtained from the Gibbs phase rule [56]: where F refers to the variance of freedom, C = the number of independent components, and P = the number of phases.The number 2 represents the temperature and total pressure of the investigated system.In the present study with fixed temperature and total pressure P total = 1 atm, Eq. ( 1) can be simplified as Eq. ( 2): The number of independent components C in the ternary system investigated in the present study is 3, including CaO, MgO, TiO 2 .The number of coexisting phases P is 3 for the liquid-rutile-pseudobrookite three-phase equilibrium observed in the present study.Therefore, the degree of freedom F could be finally confirmed as F = 3-3 = 0 under the present experimental conditions.As the liquid-rutile-pseudobrookite coexisting equilibrium was observed in samples MCT-3, MCT-5, and MCT-31, the compositions of the liquid and solid phases for liquid-rutilepseudobrookite were determined as the average of samples MCT-3, MCT-5, and MCT-31, which gave compositions of 6.26 wt% MgO-81.72 wt% TiO 2 -12.02 wt% CaO, 0.05 wt% MgO-99.94wt% TiO 2 -0.02

Equilibrium phase relations at 1600 • C
Fig. 3 presents typical BSE (back scattered electron) images of samples equilibrated at 1600 • C. The compositions of the different phases with standard deviations (±1σ) are listed in Table 2. Like the results obtained at 1500 • C, the primary phase domains of rutile, pseudobrookite, perovskite were also observed at 1600 • C.However, the liquidtitania spinel two-phase coexisting equilibrium as well as the liquidpseudobrookite-ilmenite and the liquid-ilmenite-titania spinel threephase coexisting equilibrium, not observed at 1500 • C, were found at 1600 • C.

Discussion
Based on the present experimentally determined equilibrium liquid compositions, the 1500 • C and 1600 • C isotherms were constructed, as shown in Figs. 4 and 5, respectively.The present results were compared with the data from the literature [49], and with simulations by FactSage using the FactPS and FToxid databases [57,58] and by MTDATA using its MTOX database [59,60].

Construction of the 1500 • C isotherm
As shown in Fig. 4(a), the primary phase fields of rutile and perovskite were experimentally determined at 1500 • C in air.However, the domains for liquid-pseudobrookite and liquid-ilmenite primary phase fields were not completely constructed due to the limited number of the experimental sample compositions.
Fig. 4(b) displays the comparison of the present results with the data from the literature [49].It can be seen that the dynamic observations by Rouf et al. [49] suggest a much wider liquid domain than the present results.The isotherms for the primary phase fields of perovskite determined in this study shifted to the region with a lower CaO content but with higher TiO 2 concentrations when compared with the literature data.As for the present isothermal results for the pseudobrookite primary phase field, they exhibited lower MgO concentrations than the findings by Rouf et al. [49].
Fig. 4(c) shows a comparison of the present results with the calculated isothermal section by FactSage.The liquid domain simulated by FactSage is larger than that of the present experimental results.The experimentally measured isotherm for the rutile primary phase field agrees well with the simulations by FactSage within the MgO range of 0-5 wt%.However, the liquid point for the liquid-rutile-pseudobrookite three-phase coexisting equilibrium determined in this study displays a composition with lower TiO 2 but higher CaO concentration when compared with the predictions by FactSage.The present isotherm for the  pseudobrookite primary phase field shows a higher CaO concentration than the predictions by FactSage.A comparison of the present results with the predictions by MTDATA shown in Fig. 4(d) indicates that the present experimentally determined liquid domain is smaller than the predictions by MTDATA.The present experimentally determined isotherms for the primary phase fields of rutile and pseudobrookite fitted well with the calculated results.The present results for the perovskite primary phase field shifted to the region with higher TiO 2 concentration.The presence of the liquidperovskite-ilmenite three-phase equilibrium determined at 1500 • C in this study disproves the formation of the tie-lines for (liquid-titania spinel) and (liquid-monoxide) and the corresponding equilibria predicted by FactSage and MTDATA.

Construction of the 1600 • C isotherm
Fig. 5(a) shows the present experimentally determined 1600 • C isothermal section of the MgO-CaO-TiO 2 phase diagram in air.The primary phase fields of rutile and perovskite were also determined at 1600 • C, expanding toward the region with higher TiO 2 and CaO concentrations, respectively, when compared with the liquidus line obtained at 1500 • C. The liquid point for the liquid-rutile-pseudobrookite equilibrium shifted to the region with higher TiO 2 and MgO concentrations with the increase in temperature from 1500 • C to 1600 • C. The phase domain of ilmenite, which was not observed at 1500 • C, was constructed successfully at 1600 • C. Due to the limited number of experimental points, the titania spinel phase field was not fully and reliably determined.
A comparison of the present results with the data from the literature, shown in Fig. 5(b), indicates that the present 1600 • C isotherm in the rutile primary phase field displayed a lower TiO 2 concentration than the previous observations [49].However, the results for the perovskite primary phase field measured in the present study exhibited a relatively lower CaO but higher TiO 2 concentrations compared with the observations by Rouf et al. [49].The present results for the pseudobrookite primary phase field had a slightly higher CaO concentration than those found by Rouf et al. [49].The ilmenite phase field determined in this study is much narrower than the reported values in the literature, shifting toward the area with lower MgO but higher TiO 2 concentrations.
Fig. 5(c) shows that the liquid phase domain obtained in this study is narrower and smaller than that simulated by FactSage, shrinking toward the higher TiO 2 concentration corner.The present experimentally determined isotherm for the rutile primary phase exhibited lower TiO 2 concentrations than the predictions by FactSage.The present isotherm for the primary phase field of pseudobrookite had a higher CaO content than the FactSage predictions.The results in the perovskite primary phase field obtained in this study exhibited higher TiO 2 concentrations.The titania spinel phase domain simulated by FactSage and MTDATA is wider than the present experimental results, also displaying a much  higher MgO concentration.Fig. 5(d) indicates that the liquid domain simulated by MTDATA deviated significantly from the present observations.The present experimentally determined isotherms in the rutile primary phase field agreed well with the predictions made by MTDATA.The simulated phase domains of pseudobrookite, corundum (i.e., ilmenite in the present study and by FactSage), and titania spinel are wider than that in the present observations and shifted to a higher TiO 2 but lower MgO concentration area.The predicted isotherm for the perovskite primary phase field shifted toward the area with higher CaO but lower TiO 2 concentration when compared with the present results.However, the predictions by FactSage and MTDATA fit well with the primary phase fields of rutile, pseudobrookite, and ilmenite.

Conclusions
The equilibrium phase relations of the MgO-CaO-TiO 2 system were experimentally investigated in air in the temperature range between 1500 • C and 1600 • C. The experimental methodology involved hightemperature isothermal equilibration of the samples followed by rapid quenching of the condensed equilibrium phases, and direct phase characterization by SEM-EDS and EPMA from polished sections.Observations were made of the liquid-solid primary phase equilibria of rutile, pseudobrookite, ilmenite, titania spinel, and perovskite as well as the three-phase equilibria of liquid-pseudobrookite-rutile, liquid-pseudobrookite-ilmenite, liquid-ilmenite-titania spinel, and liquidperovskite-ilmenite.
The 1500 • C and 1600 • C isothermal phase diagrams were constructed based on the experimentally determined liquid compositions, providing information for the selective crystallization of TiO 2 -enriched rutile, pseudobrookite, ilmenite, titania spinel, and perovskite phases from titania-bearing blast furnace slags by adjusting the temperature or slag composition.The obtained results were compared with data from the literature and predictions by FactSage and MTDATA and the current data showed large discrepancies with them in the perovskite, pseudobrookite, and ilmenite primary phase fields.The present results will be used for the further optimization of phase equilibria in the higher-order CaO-SiO 2 -Al 2 O 3 -MgO-TiO 2 systems.

Declaration of competing interest
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.

Fig. 1 .
Fig. 1.A schematic diagram of the sample suspension technique.

Fig. 4 .
Fig. 4. Isothermal section of the CaO-MgO-TiO 2 system in air at 1500 • (a) the present results; (b) comparisons with the data from literature [49]; (c) predictions by FactSage; and (d) predictions by MTDATA.Corundum of MTDATA represents the same solution phase as ilmenite in the present study and in the FactSage predictions.The dash-dotted line in Fig. 4(b)-(d) represents the present experimental results obtained at 1500 • C.

Table 1
Equilibrium phase compositions for the MgO-CaO-TiO 2 system in air at 1500 • C measured by EPMA.

Table 2
Equilibrium phase compositions for the MgO-CaO-TiO 2 system in air at 1600 • C obtained using EPMA.