Phase equilibrium study of the CaO–SiO 2 –CrO x system at 1600°C in reducing atmospheres

Phase equilibria in the CaO–SiO 2 –CrO x system were experimentally investigated at 1600 ◦ C and p O 2 of 10 − 10 to 10 − 11 atm using the high-temperature isothermal equilibration/drop quenching/electron probe X-ray microanalysis technique. The constrained isothermal sections of the CaO–SiO 2 –CrO x system were constructed at 1600 ◦ C and p O 2 of 10 − 10 to 10 − 11 atm based on the experimental results. The primary phase fields including cristobalite (SiO 2 ), larnite (Ca 2 SiO 4 ), (Ca,Cr)Cr 2 O 4 , and corundum (Cr 2 O 3 ) were determined. Simulations by thermodynamic software MTDATA, FactSage, and Thermo-Calc were compared with the present experimental results.

the corrosion resistance of Cr 2 O 3 -containing refractory materials in ferrochromium smelting, it is critical to determine the phase relations of the CaO-SiO 2 -Cr 2 O 3 system under typical ferrochromium smelting conditions.
][10][11][12] However, only limited thermodynamic data are available for the equilibrium phase relations in the metal-free CaO-SiO 2 -Cr 2 O 3 system in the controlled reducing atmospheres.Glasser and Osborn 13 investigated the equilibrium phase relations in the CaO-SiO 2 -Cr 2 O 3 system within the temperature range of 1120-1820 • C in air.The isotherms were mainly determined in the high-CaO part of the system.De Villiers and Muan 14 determined the phase relations in the CaO-SiO 2 -CrO x system in equilibrium with metallic chromium at temperatures of 1180-1800 • C and pO 2 of 10 −10 to 10 −13 atm.On the liquidus surface in 3436 wileyonlinelibrary.com/journal/jaceJ Am Ceram Soc.2024;107:3436-3450.
equilibrium with metallic chromium, two ternary phases including Ca 3 Cr 2 Si 3 O 12 and CaCrSi 4 O 10 with a divalent state (Cr 2+ ) were observed.Additionally, a binary calcium chromite phase with an approximate composition of (Ca 0.4 Cr 0.6 )Cr 2 O 4 was determined within the system. 14retorius and Muan 12 determined the phase relations of the CaO-SiO 2 -CrO x system in equilibrium with Pt-Cr alloys at 1500 • C and pO 2 of 10 −9.56 to 10 −12.5 atm.It was reported that a substantial rise in the solubility of chromium oxide in the liquid phase occurred when the concentration of divalent chromium ions increased, coinciding with a decrease in both oxygen pressure and the basicity of the melt.Nell and De Villiers 15 studied the phase relations in the CaO-SiO 2 -CrO x system in air and in equilibrium with metallic chromium in controlled CO 2 -H 2 gas atmospheres using the temperature-oxygen partial pressure topologic analysis method.They found two phases containing Cr 2+ , that is,  2+  4  10 and ( 0.4  2+ 0.6 ) 3+ 2  4 were stable at low oxygen partial pressures.The liquidus temperature of the system was found to increase with increasing oxygen partial pressure.Li et al. 16 investigated the effects of slag basicity and oxygen partial pressure on the solubility of CrO x in the CaO-SiO 2 -CrO x system in equilibrium with Cr 2 O 3 at 1600 • C.They found that the solubility of Cr 2 O 3 in slag decreased with increasing the slag basicity and the oxygen partial pressure. 16Murata et al. 17 equilibrated the CaO-SiO 2 -CrO x slag and Cr-Pt alloys at 1500-1600 • C and pO 2 of 10 −10 atm.The 1500-1600 • C isothermal sections of the CaO-SiO 2 -CrO x phase diagrams at Cr-Pt alloys saturation were constructed.They reported that CrO x solubility in the slag was higher with slag basicity, similar to the observations by Li et al. 16 As mentioned above, the equilibrium phase relations of the CaO-SiO 2 -CrO x system in reducing atmospheres have not been systematically investigated, especially when it is free of metallic chromium.Hence, the phase equilibria of the CaO-SiO 2 -CrO x system without chromium alloy were experimentally studied in this study at 1600 • C and pO 2 of 10 −10 to 10 −11 atm, related to ferrochromium smelting. 18,19The starting oxide mixtures were equilibrated in molybdenum crucibles under controlled CO/CO 2 -N 2 atmospheres using the high-temperature isothermal equilibration/quenching technique.The equilibrium phase compositions were measured directly from the polished section surfaces using the electron probe X-ray microanalysis (EPMA).

EXPERIMENTAL
Analytical grade oxide powders of CaO (Sigma-Aldrich, 99.9 wt%), SiO 2 (Alfa Aesar, 99.995 wt%), and Cr 2 O 3 (Alfa Aesar, 99.97 wt%) were used to prepare the initial oxide mixtures.Each oxide powder was measured based on the designed compositions.The measured oxide powders were mixed in an agate mortar and then pressed into cylindrical pellets with a diameter of 6 mm.The sample pellets with the mass of approximately 0.2 g were placed into Mo crucibles made of molybdenum foil.The crucibles with sample pellets were hooked by a basket made of molybdenum wire.The oxygen partial pressures of 10 −10 to 10 −11 atm at 1600 • C were controlled using the gas mixture of CO (99.97 vol%, AGA-Linde) and N 2 -CO 2 (95 vol% N 2 -5 vol% CO 2 , Woikoski Oy).The gas flow rates for different target oxygen partial pressures at 1600 • C were reported in detail in our previous work. 20he experimental setup consisted of a vertical furnace (Nabertherm, RHTV 120-150/1) equipped with a gas-impermeable alumina work tube and molybdenum silicide heating elements.A calibrated S-type Pt/Pt-10% Rh thermocouple (Johnson-Matthey Noble Metals) was used for sample temperature measurement.DFC26 digital mass flow controllers (Aalborg) were employed for gas flowrates regulation.Figure 1 shows a schematic diagram of the experimental setup used in this study.The experiments followed the order of high-temperature isothermal equilibration of samples at 1600 • C and pO 2 of 10 −10 to 10 −11 atm for a certain time, drop-quenching in an icewater mixture, and direct phase composition analysis by EPMA from the polished cross sections of samples.Preliminary tests indicated that 4 h was long enough for samples to reach equilibrium. 20,21However, to ensure larger crystal size of solid phases for easy composition measurement, 8 h was selected as the equilibration time for all experiments.After the high-temperature treatment, the samples were quenched in an ice-water mixture and then mounted in epoxy, polished by wet metallographic method, and carbon-coated using a carbon evaporator (JEOL IB-29510VET).The experimental procedures are described in detail in literature. 20,21he phase assemblages of samples were examined using a Tescan Mira 3 scanning electron microscope (SEM, Tescan) equipped with UltraDry silicon drift energy dispersive X-ray spectrometer (EDS, Thermo Fisher Scientific).An accelerating voltage of 15 kV, a beam current of 20 nA, and a working distance of 20 mm were employed in SEM-EDS.A CAMECA (SX100) electron microprobe analyzer (EMPA) was used for determining the equilibrium phase compositions with WDS (wavelength-dispersive) technique.The EPMA analyses were carried out with an accelerating voltage of 20 kV and a beam current of 40 nA.A beam diameter of 5-20 μm was used for the solid and liquid phases.The calibration was performed using external standard materials of quartz (for O-Kα and Si-Kα), diopside (for Ca-Kα), and ferrous chromite (for Cr-Kα).At least six points were randomly selected from the well-quenched areas of each phase.The raw EPMA data processing was carried out using the PAP online matrix correction program.22 Polished sections of selected samples without carbon coating were analyzed by X-ray diffraction (XRD, PANalytical X'Pert Powder XRD [alpha-1]) using Cu Ka radiation (40 kV, 40 mA) with a scanning range of 10 • -90 • (2θ).

Microstructures and equilibrium phase compositions
The typical back-scattered images of samples and equilibrium phase compositions are displayed in Figures 2 and 3, and Tables 1 and 2, respectively.EPMA results alone do not provide insights into the valence state of elements.Consequently, the concentration of chromium oxide was reevaluated from the EPMA data as "CrO" to accurately convey the outcomes derived within reducing atmospheres, where chromium predominantly exists in the valence state of Cr 2+ rather than Cr 3+ in silicate melt.The two-phase equilibria of liquid-cristobalite (SiO 2 ), liquid-larnite (Ca 2 SiO 4 ), and liquid-corundum (Cr 2 O 3 ) were observed at 1600 • C and pO 2 of 10 −10 to 10 −11 atm.However, the liquid-  3E was quite low, which increased difficulties to identify the phases by contrast.However, the (Ca,Cr)Cr 2 O 4 phase had an elongated prismatic shape, and the corundum phase displayed a euhedral shape.Therefore, the corundum and (Ca,Cr)Cr 2 O 4 phases can be differentiated based on the crystal shape.To eliminate the effect of the valence state of chromium on the equilibrium compositions, all the results were calculated into mono cationic mole fractions of CrO x , SiO 2 , and CaO. 21he overall standard deviation of the EPMA results was smaller than 0.5 mol%.It can be clearly seen from the data presented in Tables 1 and 2 that a certain amount of CrO x dissolved into the larnite phase.To determine the relationship between the CrO x dissolution in larnite and the corresponding equilibrium liquid composition, the concentration of CrO x in larnite is plotted against CrO x in the liquid slag, as shown in Figure 5.The endpoints for the results obtained at pO 2 of 10 −10 and 10 −11 atm represent the liquid-larnite-corundum and liquid-larnite-(Ca,Cr)Cr 2 O 4 three-phase equilibrium, respectively, as marked in colorful circles in Figure 5.The dissolution of CrO x in larnite was found to follow a linear relationship with the CrO x concentration in liquid.Within the range of zero to approximately 8 mol% CrO x in liquid, the effect of oxygen partial pressure on the dissolution of CrO x in larnite was not evident at a given equilibrium CrO x concentration in the liquid oxide.However, the CrO x in liquid equilibrated with larnite and (Ca,Cr)Cr 2 O 4 at pO 2 of 10 −11 atm was around 4 mol% higher than that at pO 2 of 10 −10 atm.The corresponding CrO x in larnite equilibrated at pO 2 of 10 −10 atm was approximately 1.6 mol% lower than that at pO 2 of 10 −11 atm. in Figure 6.The primary phase fields of single liquid, cristobalite, corundum, and larnite were constructed at 1600 • C under both pO 2 of 10 −10 and 10 −11 atm.The phase domains for liquid-cristobalite-corundum and liquid-larnite-corundum three-phase equilibria were also determined at pO 2 of 10 −10 atm, as shown in Figure 6A.However, the liquid-cristobalite-corundum three-phase equilibrium did not exist at pO 2 of 10 −11 atm, see Figure 6B.The primary phase field of (Ca,Cr)Cr 2 O 4 as well as the phase domains for liquid-larnite-(Ca,Cr)Cr 2 O 4 and liquidcorundum-(Ca,Cr)Cr 2 O 4 were constructed at 1600 • C and pO 2 of 10 −11 atm.The cristobalite and corundum primary phase fields expanded toward the area with higher CrO x but lower SiO 2 concentrations when oxygen partial pressure decreased from 10 −10 to 10 −11 atm.The isotherm for the larnite primary phase field determined at pO 2 of 10 −11 atm was located almost on the same level of CaO concentration in liquid as at pO 2 of 10 −10 atm, whereas the CrO x concentration in liquid equilibrated with larnite and (Ca,Cr)Cr 2 O 4 was higher than that in equilibrium with larnite and corundum, as can be seen in Figure 5.
The present experimental results were compared with the scarce data from literature, 17 in which the slag was in equilibrium with a metallic Cr-Pt alloys.The isotherm in the primary field of cristobalite determined by Murata et al. 17 displayed a higher SiO 2 concentration than the present experimental results.Their results in the corundum primary phase domain were close to the present data, although their liquid compositions for liquid-cristobalitecorundum and liquid-larnite-corundum three-phase equilibria shifted to a higher SiO 2 concentration.

4.2
Comparison of the present experimental results and simulations

Comparison with simulations by FactSage
Figure 7 shows a comparison of the present experimental results and our predictions by the thermodynamic software FactSage (version 8.2) using the FactPS and FToxid databases. 23,24][27] The liquid domain determined in the present study at 1600 • C and pO 2 of 10 −10 atm was narrower than the simulations by FactSage, as shown in Figure 7A.The isotherm for the larnite primary phase domain obtained in this study exhibited slightly higher SiO 2 concentrations than the modelling by FactSage.The present liquid compositions for liquid-cristobalite-corundum and liquid-larnitecorundum three-phase equilibrium displayed lower and higher SiO 2 concentrations, respectively, than the predictions by FactSage.The dissolution of CrO x in larnite determined in this study was not described and available in the FToxid database of FactSage. 23,24igure 7B shows that the present experimental results at 1600 • C and pO 2 of 10 −11 atm were close to the simulations by FactSage in the cristobalite primary phase field.The isotherm in the larnite primary phase domain determined in the present study displayed higher SiO 2 concentrations than the predictions by FactSage.In the FactSage predictions, the primary phase field of CaCr 2 O 4 was calculated.However, the solid phase with the chemical composition of (Ca,Cr)Cr 2 O 4 was observed to be in equilibrium with liquid oxide in the present experimental work.As for the isotherm in the corundum primary phase domain, the predictions by FactSage had slightly  higher CrO x concentrations than the present experimental results when the CrO x concentration in liquid was lower than 26 mol%.Whereas the predictions by Fact-Sage in the corundum primary phase field exhibited lower CrO x concentrations than the present experimental results when the CrO x concentration in liquid was higher than 33 mol%.

4.2.2
Comparison with simulations by MTDATA  The computations performed by MTDATA relied on an evaluation of the Al 2 O 3 -CaO-Cr-O-MgO-SiO 2 system using data from the Mtox database. 30The liquid domain constructed in this study was narrower than predicted by MTDATA.Our experimental SiO 2 concentration in liquid oxide in equilibrium with solid cristobalite was close to the simulation by MTDTA.However, the present CrO x concentration in liquid for the liquid-cristobalitecorundum three-phase equilibrium was approximately 13 mol% lower than the MTDATA simulation.The predicted isotherm in the corundum primary phase field exhibited a slightly higher CrO x concentration than the present experimental results.The larnite primary phase domain determined in this study shifted toward the area with  CrO x in larnite were not included in the MTDATA Mtox database. 28,29igure 8B indicates that the isotherm in the cristobalite primary phase field determined in this study fitted very well with simulations by MTDATA.Similar to the predictions by FactSage, the primary phase domain of CaCr 2 O 4

Comparison with simulations by Thermo-Calc
The comparison of the present experimental results with predictions by the thermodynamic software Thermo-Calc (version 2023a) 31 using the TCOX12 database 32 is shown in Figure 9. Figure 9A indicates that the present experimental results obtained at 1600 • C and pO 2 of 10 −10 atm were close to the simulations by Thermo-Calc in the primary phase fields of larnite and corundum.However, the results for the cristobalite primary phase domain determined in this study displayed a lower SiO 2 content than the predictions by Thermo-Calc.The present results for the liquid compositions for the liquidcristobalite-corundum three-phase equilibria had a higher CaO but lower SiO 2 concentrations than the Thermo-Calc simulations.
Figure 9B shows that the present experimental results determined at 1600 • C and pO 2 of 10 −11 atm were close to the Thermo-Calc modelling in the primary phase domains of larnite and corundum.The present concentration of SiO 2 in the liquid oxide phase for the liquid-corundum two-phase equilibrium in the CrO x -SiO 2 system was approximately 4 mol% higher than the modelling results by Thermo-Calc.The liquid-CaCr 2 O 4 phase equilibria predicted by Thermo-Calc deviated a lot from the present experimental results, like the modellings by MTDATA and FactSage.
It was reported in literature 14 that two solid solutions, that is, CaCr 2 O 4 and (Ca,Cr)Cr 2 O 4 can be formed in the Cr 2 O 3 -rich side of the CaO-Cr 2 O 3 binary system.The observation of the primary phase field of (Ca,Cr)Cr 2 O 4 as well as the three-phase equilibria of liquid-(Ca,Cr)Cr 2 O 4larnite and liquid-(Ca,Cr)Cr 2 O 4 -corundum is different from the predictions by FactSage, MTDATA, and Thermo-Calc, in which the CaO-Cr 2 O 3 binary solid solution displayed a chemical composition of CaCr 2 O 4 .They displayed higher CaO concentrations in solid than the present experimental observations.The present experimental results can be used to update the MTDATA Mtox, FactSage FToxid, and Thermo-Calc TCOX12 databases.

CONCLUSIONS
The phase relations of the CaO-SiO 2 -CrO x system under pO 2 of 10 −10 to 10 −11 atm at 1600 • C were experimentally determined using the isothermal equilibration and quenching method.The phase compositions were analyzed from the polished cross-sections using EPMA.The two-phase equilibria of liquid-cristobalite, liquid-corundum, liquid-(Ca,Cr)Cr 2 O 4 , and liquid-larnite, as well as the three-phase equilibria of liquid-cristobalite-corundum, liquid-corundum-larnite, liquid-corundum-(Ca,Cr)Cr 2 O 4 , and liquid-larnite-(Ca,Cr)Cr 2 O 4 were confirmed.
The CaO-SiO 2 -CrO x phase diagrams were constructed at 1600 • C and pO 2 of 10 −10 to 10 −11 atm.The experimentally determined phase diagrams were compared with simulations by MTDATA, FactSage, and Thermo-Calc.The predictions by FactSage and Thermo-Calc were closer to the present experimental results than simulations by MTDTA at 1600 • C and pO 2 of 10 −10 atm.As for the results determined at 1600 • C and pO 2 of 10 −11 atm, the present experimental results in the cristobalite primary phase domain fitted well with the simulations by MTDATA.The present experimental results determined in the primary phase field of (Ca,Cr)Cr 2 O 4 displayed significant discrepancies with predictions by MTDATA and FactSage.The present experimental results are useful for updating the CrO x -bearing thermodynamic databases.They provide instructions for preventing the corrosion of CrO x -based refractories in ferrochromium smelting by controlling the smelting conditions, such as the oxygen partial pressure and slag composition of the smelting systems.

A C K N O W L E D G M E N T S
This work was financially supported by the Business Finland-funded TOCANEM project (Grant Number: 2118452).This work utilized the Academy of Finland's RawMatTERS Finland Infrastructure (RAMI) based at Aalto University, GTK Espoo, and VTT Espoo.

C O N F L I C T O F I N T E R E S T S TAT E M E N T
The authors declare no conflicts of interest.

F
I G U R E 1 A schematic diagram of the furnace arrangement; MFC, Mass flow controller.
corundum-cristobalite three-phase equilibrium observed at 1600 • C and pO 2 of 10 −10 atm was not found at pO 2 of 10 −11 atm.When the oxygen partial pressure decreased from 10 −10 to 10 −11 atm, the liquid-(Ca,Cr)Cr 2 O 4 , as well as three-phase equilibria of liquid-(Ca,Cr)Cr 2 O 4 -larnite and liquid-(Ca,Cr)Cr 2 O 4 -corundum, were observed at pO 2 of 10 −11 atm.To verify the crystal structure of the (Ca,Cr)Cr 2 O 4 phase determined in this study, sample CSCr-44 with liquid-(Ca,Cr)Cr 2 O 4 two-phase equilibrium was examined by XRD.The XRD pattern for CSCr-44 in Figure 4 indicates that the (Ca,Cr)Cr 2 O 4 phase in the present study displayed a similar crystal structure for CaCr 2 O 4 .The shift of the analyzed pattern to the right side of the standard peaks for CaCr 2 O 4 was caused by the replacement of Ca 2+ by Cr 2+ , leading to the lattice shrinkage.The contrast difference between corundum and (Ca,Cr)Cr 2 O 4 in Figure

F I G U R E 2
Backscattered electron image of the CaO-SiO 2 -CrO x system equilibrated at 1600 • C and pO 2 of 10 −10 atm.

F I G U R E 3
Backscattered electron image of the CaO-SiO 2 -CrO x system obtained at 1600 • C and pO 2 of 10 −11 atm.
Construction of the 1600 • C phase diagrams at pO 2 of 10 −10 to 10 −11 atm The 1600 • C isothermal sections of the CaO-SiO 2 -CrO x phase diagrams at pO 2 of 10 −10 to 10 −11 atm are shown 15512916, 2024, 5, Downloaded from https://ceramics.onlinelibrary.wiley.com/doi/10.1111/jace.19624 by Duodecim Medical Publications Ltd, Wiley Online Library on [22/04/2024].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License F I G U R E 4 X-ray diffraction (XRD) pattern for CSCr-44 with liquid-(Ca,Cr)Cr 2 O 4 two-phase equilibrium.F I G U R E 5 Solubility of chromium oxide in larnite as a function of CrO x in liquid.

Figure
Figure 8A represents the comparison of the predictions by the thermodynamic software MTDATA (version 7.2) using the Mtox database (version 8.2) 28,29 and the present experimental results at 1600 • C and pO 2 of 10 −10 atm.

F I G U R E 9
Comparison with the simulations by Thermo-Calc.(A) pO 2 = 10 −10 atm; (B) pO 2 = 10 −11 atm.higher SiO 2 but lower CaO concentrations compared with the simulations by MTDATA.The liquid-CaCr 2 O 4 twophase equilibrium as well as the three-phase equilibrium of liquid-larnite-CaCr 2 O 4 and liquid-corundum-CaCr 2 O 4 by MTDATA at 1600 • C and pO 2 of 10 −10 atm were not observed in the present work.The solubility data for 15512916, 2024, 5, Downloaded from https://ceramics.onlinelibrary.wiley.com/doi/10.1111/jace.19624 by Duodecim Medical Publications Ltd, Wiley Online Library on [22/04/2024].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License TA B L E 1 Equilibrium phase compositions for the CaO-SiO 2 -CrO x system at 1600 • C and pO 2 of 10 −10 atm measured by electron probe x-ray microanalysis (EPMA).Equilibrium phase compositions for the CaO-SiO 2 -CrO x system at 1600 • C and pO 2 of 10 −11 atm determined using electron probe X-ray microanalysis (EPMA).Downloaded from https://ceramics.onlinelibrary.wiley.com/doi/10.1111/jace.19624 by Duodecim Medical Publications Ltd, Wiley Online Library on [22/04/2024].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 15512916, 2024, 5, Downloaded from https://ceramics.onlinelibrary.wiley.com/doi/10.1111/jace.19624 by Duodecim Medical Publications Ltd, Wiley Online Library on [22/04/2024].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 15512916, 2024, 5, Downloaded from https://ceramics.onlinelibrary.wiley.com/doi/10.1111/jace.19624 by Duodecim Medical Publications Ltd, Wiley Online Library on [22/04/2024].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License was simulated by MTDATA, but the liquid-(Ca,Cr)Cr 2 O 4 two-phase equilibrium was confirmed in our work to locate at higher SiO 2 concentrations.The primary phase field of larnite in the present experimental work shifted toward the area with lower CaO concentrations than the modelling by MTDATA.The experimentally determined liquid compositions in equilibrium with corundum displayed a lower CrO x concentration than the predictions by MTDATA when CrO x in liquid was lower than 26 mol%, after which it agreed well with the simulations by MTDATA.