Determination of Solubility of Chromium Oxide in the CaO–SiO₂–Cr₂O₃ Slag System at 1873 K under Moderately Reducing Conditions

Abstract

For a better thermodynamic understanding of the optimization of molten slag in the stainless-steel refining process, the solubility of chromium oxide in the CaO–SiO₂–Cr₂O₃ slag system was investigated. A chemical equilibrium technique was employed in which molten slag and solid Cr₂O₃ pellets were equilibrated under a regulated oxygen partial pressure (P_O2) lower than 10⁻¹¹ atm at 1873 K. The mass ratio of CaO to SiO₂ (X_CaO / X_SiO2), as an index of slag basicity, was varied from 0.5 to 1.5. A CO–CO₂–Ar gas mixture was used, and the oxygen partial pressure was precisely controlled by evaluating possible contamination by oxygen that was inevitably introduced into the gas mixture system. A zirconia oxygen sensor was used to directly measure the oxygen partial pressure for this evaluation. The solubility of chromium oxide increased with decreasing slag basicity and oxygen partial pressure. Accordingly, slag design is a huge prospect toward achieving desirable refining conditions.

1. Introduction

It is well known that stainless steel is a typical iron-based alloy containing Cr, causing the formation of a protective and invisibly thin film of Cr₂O₃ on the steel surface, which induces high corrosion resistance.^1,2,3) The production process of stainless steel can be divided roughly into two steps, namely the melting of scrap and ferroalloys in an electric arc furnace (EAF) and a refining process aimed at adjusting the carbon content and removing other impurities.^4,5,6)

Considering that Cr has a strong affinity for oxygen, it is challenging to decarburize stainless steel to a sufficiently low carbon level during the typical refining process while preventing the loss of Cr into the slag caused by the preferential oxidation of Cr. An effective process to suppress this oxidation and Cr loss requires an elevation of the reaction temperature and a reduction in the partial pressure of carbon monoxide. However, the temperature can only rise to a specific point, owing to the influence of refractory damage. Thus, low carbon levels can be achieved by reducing the partial pressure of carbon monoxide in the refining process to ensure preferential decarburization.⁷⁾ This has been conventionally realized and implemented using an argon oxygen decarburization (AOD) furnace with the dilution of CO gas by argon gas and a vacuum oxygen decarburization (VOD) furnace with a reduction in the partial pressure of CO by evacuation during the production of stainless steel.^{8,9,10,11,12)} However, the perfect suppression of Cr loss is not feasible in principle when applying these elaborate decarburization processes because of the intrinsically negative large Gibbs energy for the formation of chromium oxide at steel production temperatures. Hence, the addition of silicon as a reducing agent is necessary to recover Cr from slag to obtain molten crude stainless steel¹³⁾ from the viewpoint of maximizing the yield of Cr in the alloy and possibly lowering the Cr concentration in slag, considering the potential environmental pollution and subsequent harmful effects on human health.¹⁴⁾ This final reduction process leads to an increase in production costs. From another perspective, the slag composition substantially affects the decarburization process. Therefore, it is essential to understand the thermodynamic properties of chromium oxide in molten slag to minimize Cr oxidation.^15,16,17)

Up to now, the solubility of chromium oxide in CaO–SiO₂–Cr₂O₃ slag has been extensively investigated in air and under a moderately low oxygen partial pressure with the coexistence of Cr₂O₃ pellets through a chemical equilibrium technique. However, a change in the oxygen partial pressure lower than 10⁻¹¹ atm has not been attempted because of the difficulty in controlling the oxygen partial pressure and the coexistence of Cr³⁺ and Cr²⁺ in the slag.^18,19,20,21) Although the control methodology of oxygen partial pressure by a CO–CO₂–Ar gas mixture is widely used for thermodynamic experimental implementations,^22,23) the accuracy of this method has not been well verified because it is possible that the oxygen partial pressure in the experimental furnace may not precisely reach the theoretically calculated value. The gas mixing ratio of CO to CO₂ is determined by setting the CO and CO₂ gas flow rates; however, the oxygen contamination and Ar gas in the gas passage influence the gas mixture composition slightly, causing a change in the settled CO/CO₂ gas ratio.

To solve this problem, we used a zirconia oxygen sensor to determine the actual oxygen partial pressure in the furnace; this sensor is widely used for oxygen partial pressure measurements.^24,25) In addition, we developed a new dilution system for CO₂, the dissolved CO₂ in the CO matrix realized a high CO/CO₂ ratio. This method was used in the chemical equilibrium experiment to measure the chromium oxide solubility.

2. Experimental Process

2.1. Chemical Equilibrium Technique

The chemical equilibrium technique is a thermodynamic methodology commonly used to determine the solubilities of components in equilibrated phases in a formulated system. An equilibrium state is the final state in a reversible chemical reaction, where all the reactants and products will not undergo any further change over a time interval. In chemical thermodynamics, when a system is in equilibrium, all subsystem parts will equally be in equilibrium.²⁶⁾ In the ternary system of the CaO–SiO₂–CrO_x slag, the concentration of chromium oxide, CrO_x in the CaO–SiO₂–CrO_x slag phase will remain constant in equilibrium, simply expressing the possible change in oxidation number without an explicit stoichiometric adjustment. The equilibrium state can be conceptually expressed as Eq. (1).   

$${\text{CrO}}_x(\text{in}\mathrm{\hspace{0.17em} }\text{slag})\iff {\text{Cr}}_2{\text{O}}_3(\text{pellet})$$(1)

2.2. Control of Oxygen Partial Pressure

The oxygen partial pressure needs to be controlled in this equilibrium reaction, and it can be evaluated from the composition of the CO–CO₂–Ar gas mixture based on the following equations:²⁷⁾   

$$\text{CO(g)}+ 1/2{\text{O}}_2(\text{g})={\text{CO}}_2(\text{g})$$(2)

  

$$\mathrm{\Delta }G°=-281+0.0853\mathrm{\hspace{0.17em} }T(\text{kJ}\cdot {\text{mol}}^{-1})$$(3)

  

$$P_{{\text{O}}_2}={\left\{\left(P_{\text{CO}}/P_{{\text{CO}}_2}\right)\times exp\left(-(\mathrm{\Delta }G°)/RT\right)\right\}}^{-2}$$(4)

where ΔG° represents the Gibbs energy (kJ·mol⁻¹) of Eq. (2), R represents the gas constant, and P_CO and P_CO2 represent the partial pressure of each gas species, respectively.

2.3. Principle of Measurement with the Zirconia Oxygen Sensor

Figure 1 shows the conceptual cross section and mechanism of an oxygen sensor fabricated with a zirconia solid electrolyte (CSZ). A stabilized zirconia tube with one closed side was used as the solid electrolyte, and platinum wires were adhered to the inner and outer parts of the tip (closed side) of the tube. Platinum paste was used to bond the platinum wire to the zirconia tube. The external atmosphere of the tube was the reference electrode (air), and the inner of the tube was a gas sample whose oxygen concentration was to be measured (oxygen partial pressure controlled by the CO–CO₂ gas mixture). By measuring the electromotive force necessary to cease the movement of oxygen ions, the oxygen partial pressure can be calculated by determining the difference between the inner of the tube and the external reference atmosphere, according to Eq. (5).   

$$E=\frac{RT}{nF}\times \text{ln}\left(\frac{P_{{\text{O}}_2(\text{air})}}{P_{{\text{O}}_2(\text{furnace})}}\right)$$(5)

where P_O2(air) represents the oxygen partial pressure in air as 0.21, P_O2(furnace) represents the oxygen partial pressure in the furnace, F represents the Faraday constant (96485 C/mol), E represents the electromotive force obtained from the monitor, and n represents the number of electrons transferred in the cell reaction associated with one molecule of oxygen; n = 4 in this study.

Fig. 1.

Conceptual cross section and mechanism of zirconia oxygen sensor.

2.4. Experimental Procedure

The solubility of chromium oxide in the slag was determined by equilibrating the CaO–SiO₂–CrO_x slag with a Cr₂O₃ pellet in a Mo crucible at 1873 K. The Cr₂O₃ pellet was prepared by compressing Cr₂O₃ powders followed by sintering under an Ar gas atmosphere with a holding time of 12 h at 1273 K. Calcium carbonate (> 99.9 mass%), SiO₂ (> 99.9 mass%), and Cr₂O₃ (> 98.5 mass%) in reagent grade were used as starting materials to premelt the slag. Calcium oxide was prepared by the calcination of CaCO₃ powder at 1473 K for 12 h and was confirmed by X-ray diffraction (XRD) analysis.

The starting materials were weighed using a balance and sufficiently mixed using an alumina mortar; the mixed powder was placed in a Mo crucible, which was heated to 1873 K in an electric resistance furnace with MoSi₂ heating elements. A CO–CO₂–Ar gas mixture was introduced to control the oxygen partial pressure to obtain the premelted slag. Thereafter, the main experiments were conducted for chemical equilibrium between the premelted slag and Cr₂O₃ pellet.

Figure 2 shows the experimental setup composed of a gas flow system, an oxygen sensor furnace, and a main furnace. The temperature of the uniform zone in the furnace was measured and controlled using a Pt–Rh thermocouple with an error within ±1 K. The gas flow of CO–CO₂–Ar was controlled using a mass flow controller (MFC). We employed a zirconia oxygen sensor to evaluate the oxygen contamination in the system using a temperature range of 1173–1573 K.

Fig. 2.

Schematic representation of the experimental apparatus.

Tables 1 and 2 summarize the initial slag composition and gas mixture, respectively. The holding time necessary to reach equilibrium was determined by preliminary experiments and was observed to be 24 h. After the equilibrium experiment, the Mo crucible containing the sample was removed from the furnace and immediately quenched in water. After removing the slag sample from the Mo crucible, it was subjected to chemical analyses using ICP Optical Emission Spectrometry.

Table 1. Initial slag composition in equilibrium experiments.

mass contents in %			mass%CaO/mass%SiO2
CaO	SiO2	Cr2O3
30.0	60.0	10.0	0.5
37.1	52.9	10.0	0.7
42.6	47.4	10.0	0.9
47.1	42.9	10.0	1.1
53.7	41.3	5.0	1.3
58.2	38.8	3.0	1.5

Table 2. Gas mixture composition in equilibrium experiments.

Gas flow (mL/min)			PCO / PCO2
CO	CO2	Ar
50	1	150	50/1
50	0.1	150	500/1
100	0.1	150	1000/1
200	0.1	150	2000/1

3. Results and Discussion

3.1. Evaluation of Oxygen Partial Pressure

The oxygen partial pressure was measured using different ratios of P_CO / P_CO2, namely 1/1, 500/1, and 1000/1, at temperatures of 1173, 1273, 1373, and 1573 K to evaluate the performance of the zirconia oxygen sensor. According to Eq. (6), the oxygen partial pressure, P_O2, (hereinafter referred to as P_O2(Measured)) was obtained from the value of the electric potential E. We evaluated the oxygen partial pressure by comparing P_O2(Measured)) with the P_O2(hereinafter referred to as P_{O2(Calculation)}) calculated from the thermodynamic data of ΔG°²⁸⁾ regarding gaseous species equilibria.   

$$P_{{\text{O}}_2(\text{Measured})}=P_{{\text{O}}_2(\text{air})}\times \text{exp}\left(\frac{-4FE}{RT}\right)$$(6)

Figure 3 shows a comparison between P_O2(Measured) and P_{O2(Calculation)}. When P_CO / P_CO2 is 1/1, P_O2(Measured) and P_{O2(Calculation)} are practically in good agreement; however, large discrepancies are observed for large ratios, such as 500/1 or 1000/1. These discrepancies can be attributed to the oxygen contamination in the system. Based on the measurement results, the amount of oxygen contamination was estimated from the difference between P_O2(Measured) and P_{O2(Calculation)}.

Fig. 3.

Comparison of measured and calculated oxygen partial pressures at each ratio of P_CO / P_CO2 and temperature.

According to Eqs. (4) and (5), the actual P_CO / P_CO2 at an arbitrary temperature can be calculated using the P_O2(Measured) from Eq. (7).   

$$\text{ln}\frac{P_{\text{CO}}}{P_{{\text{CO}}_2}}=-\frac{1}{2}\text{ln}{P}_{{\text{O}}_2(\text{Measured})}+\frac{\mathrm{\Delta }G°}{RT}$$(7)

The amount of oxygen contamination can be calculated from the deviation between the P_CO / P_CO2 calculated using Eq. (7) and that initially set by the gas flow meter. For this estimation, all the oxygen introduced by contamination is assumed to have reacted with CO to form CO₂ according to Eq. (2). By designating the amount of CO₂ produced by this reaction (equal to the amount of CO consumed by the reaction) on the flow rate basis as x (mL/min), the value of x can be easily calculated using Eq. (8).   

$$\frac{\text{setting}\mathrm{\hspace{0.17em} }\text{value}\mathrm{\hspace{0.17em} }\text{of}\mathrm{\hspace{0.17em} }\text{CO}-x}{\text{setting}\mathrm{\hspace{0.17em} }\text{value}\mathrm{\hspace{0.17em} }\text{of}\mathrm{\hspace{0.17em} }{\text{CO}}_2+x}=\frac{P_{\text{CO}}}{P_{{\text{CO}}_2}}\mathrm{\hspace{0.17em} }\mathrm{\hspace{0.17em} }(\text{calculated}\mathrm{\hspace{0.17em} }\text{from}\mathrm{\hspace{0.17em} }\text{Eq}.\mathrm{\hspace{0.17em} }\text{(}7\text{)})$$(8)

Considering that the amount of oxygen contamination in each composition of P_CO / P_CO2 was observed to virtually be the same (1/2P_O2) = 0.15 ± 0.01 mL/min, it was confirmed that a certain amount of oxygen was introduced through the contamination in the system regardless of the gas composition. Consequently, the actual P_O2 in the furnace was estimated using Eq. (9).   

$$P_{{\text{O}}_2}={\left\{\left(\frac{\text{setting}\mathrm{\hspace{0.17em} }\text{value}\mathrm{\hspace{0.17em} }\text{of}\mathrm{\hspace{0.17em} }\text{CO}-0.15\times 2}{\text{setting}\mathrm{\hspace{0.17em} }\text{value}\mathrm{\hspace{0.17em} }\text{of}\mathrm{\hspace{0.17em} }{\text{CO}}_2+0.15\times 2}\right)\times exp\left(-\frac{(\mathrm{\Delta }G°)}{RT}\right)\right\}}^{-2}$$(9)

Figure 4 shows the re-evaluated values for the oxygen partial pressure, considering the oxygen contamination according to Eq. (9). The re-evaluated lines showed good agreement with the values measured by the oxygen sensor.

Fig. 4.

Comparison of measured oxygen partial pressures and recalculated values considering oxygen contamination at each ratio of P_CO / P_CO2 and temperature.

Another set of experiments was conducted to verify the propriety of this evaluation method for P_O2 in the main furnace (1873 K). Five grams of pure iron was placed in an alumina crucible and held under a CO–CO₂–Ar atmosphere at 1873 K. After being held sufficiently for equilibration, the crucible was removed from the furnace and quenched with water. The oxygen concentration in the iron was then analyzed using a LECO-ONH836 Oxygen/Nitrogen/Hydrogen Elemental Analyzer.

The concentration of the oxygen dissolved in the molten iron under the intended oxygen partial pressure can be calculated using Eq. (13), based on the oxygen dissolution reaction in Eq. (10), the Gibbs energy expressed in Eq. (11),²⁹⁾ the relationship between the oxygen concentration on mass percent basis, [mass%O], and its activity, a_O.   

$$1/2{\text{O}}_2=\underset{\text{\_}}{\text{O}}\mathrm{\hspace{0.17em} }\text{(mass\%}\mathrm{\hspace{0.17em} }\text{in}\mathrm{\hspace{0.17em} }\text{Fe)}$$(10)

  

$$\mathrm{\Delta }G°=-117\mathrm{\hspace{0.17em} }110-3.39\mathrm{\hspace{0.17em} }T\mathrm{\hspace{0.17em} }[\text{J}/\text{mol}]$$(11)

  

$$a_{\mathrm{O}}={[\text{mass\%O}]}_{\mathrm{\hspace{0.17em} }\text{in}\mathrm{\hspace{0.17em} }\text{Fe}}\mathrm{\hspace{0.17em} }(\mathrm{i}\mathrm{n}\mathrm{\hspace{0.17em} }\mathrm{d}\mathrm{i}\mathrm{l}\mathrm{u}\mathrm{t}\mathrm{e}\mathrm{\hspace{0.17em} }\mathrm{s}\mathrm{o}\mathrm{l}\mathrm{u}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n})$$(12)

  

$${[\text{mass\%O}]}_{\mathrm{\hspace{0.17em} }\text{in}\mathrm{\hspace{0.17em} }\text{Fe}}=\text{exp}\mathrm{\hspace{0.17em} }\text{(}-\mathrm{\Delta }G°/RT\text{)}\times P_{{\text{O}}_2}^{1/2}$$(13)

The oxygen concentration, predicted using Eq. (13), which was obtained from the oxygen partial pressure determined by the oxygen sensor, shows good agreement with the analyzed value for the actual iron sample after this confirmation experiment. Therefore, we conclude that the actual oxygen partial pressure in the main experiment can be evaluated in this manner with high accuracy.

3.2. Oxidation States of Cr in Slag Phase

The Cr in the slag phase has two valence states at low oxygen partial pressure, according to Eq. (14):   

$$\begin{array}{l}{\text{CrO}}_x=y\text{CrO}+(1-\text{y}){\text{CrO}}_{1.5},\\ \mathrm{w}\mathrm{h}\mathrm{e}\mathrm{r}\mathrm{e}\mathrm{\hspace{0.17em} }x=1.5-0.5y.\end{array}$$(14)

An accurate quantification of this valency distribution is important in evaluating the solubility of Cr in the slag. There are significant discrepancies in the distribution of the oxidation state of Cr in the slag, owing to the difference in experimental methods, conditions, and analysis methods. For volumetric titration analysis, it is necessary to dissolve the slag sample first and subsequently perform a titration analysis. Therefore, part of the Cr²⁺ will inevitably be oxidized to Cr³⁺ during the analysis process because of its instability. To solve this problem, Wang et al.³⁰⁾ developed a new chemical leaching method to measure the contents of Cr²⁺ and Cr³⁺ in CaO–SiO₂–MgO–Al₂O₃–CrO_x–CaF₂ slag. This analytical technique uses a FeCl₃–HCl solution for the selective leaching of CrO in the slag sample to obtain trivalent chromium ion solution. The divalent vanadium solution was used to reduce CrCl₃, which was generated during the aforementioned chemical leaching process. In this analysis process, only the CrO in the slag was extracted to the solution or settled in the form of CrCl₃, while CrO_1.5 was not involved and leached during the entire process. Thus, the oxidation of Cr²⁺ to Cr³⁺ may be significantly suppressed during the analytical process.

With the development of science and technology in recent years, the X-ray absorption near edge structure (XANES) method has been used to analyze the valence state. Wang et al. studied the valence states of Cr in the multicomponent system CaO–MgO–(FeO–)Al₂O₃–SiO₂–CrO_x at a temperature region up to 2000 K using the XANES analysis method,³¹⁾ and proposed an equation using a variation in the X_CrO / X_CrO1.5 ratio, the slag basicity B (mass%CaO + mass%MgO/mass%Al₂O₃ + mass%SiO₂), and the oxygen partial pressure in the slag:   

$$\begin{array}{l}\text{log}\left({\displaystyle \frac{X_{\text{CrO}}}{X_{{\text{CrO}}_{1.5}}}}\right)=-{\displaystyle \frac{11\mathrm{\hspace{0.17em} }534}{T}}-0.25\times log(P_{{\text{O}}_2})\\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}-0.203\times \text{log(}B\text{)}+5.74\end{array}$$(15)

According to Eq. (15), the X_CrO / X_CrO1.5 increases with an increase in temperature and a decrease in slag basicity and oxygen partial pressure. In this study, the highest oxygen partial pressure and slag basicity are 1.2 × 10⁻¹⁰ atm and 1.5, respectively; therefore, the X_CrO / X_CrO1.5 ratio is considerably high, according to the aforementioned tendency. Thus, Cr²⁺ was predominant under these conditions.

3.3. Solubility of CrO_x in CaO–SiO₂–Cr₂O₃ Slag

Finally, we obtained the solubility of CrO_x in the CaO–SiO₂–Cr₂O₃ slag by evaluating and verifying the oxygen partial pressure and discussing the valence state of Cr in the slag, as described. Figure 5 shows the solubility lines of CrO_x with a saturation of Cr₂O₃ in the CaO–SiO₂–Cr₂O₃ slag system under various oxygen partial pressure at 1873 K. As shown in Fig. 5, the high slag basicity side, the P_O2 has no evident effect on the solubility of CrO_x, and the interval between the isosolubility curves of CrO_x is minimal. Conversely, the low slag basicity side, the solubility is sensitive to the change in P_O2, and the interval between the solubility curves is significantly enlarged. The measured values obtained in this study are considerably lower than the values reported by Morita et al.¹⁹⁾ Two reasons include the volumetric titration methods for the valence state analysis, which might cause an uncertainty for the X_CrO / X_CrO1.5 ratio and CrO_x solubility, and the difficulty in obtaining accurate oxygen partial pressure data, which would have added credibility by applying the oxygen sensor for monitoring during the experimental process. However, the relationship between slag basicity and the CrO_x solubility exhibited the same tendency as the reported values.

Fig. 5.

Solubility lines of CrO_x with saturation of Cr₂O₃ in the CaO–SiO₂–CrO_x slag system under various oxygen partial pressures at 1873 K.

4. Conclusions

The effect of the slag basicity and oxygen partial pressure on the solubility of CrO_x in CaO–SiO₂–Cr₂O₃ slag was determined by equilibrating the CaO–SiO₂–Cr₂O₃ slag with a Cr₂O₃ pellet in a Mo crucible at 1873 K. The oxygen partial pressure was monitored in a furnace using an oxygen sensor. The findings are summarized as follows:

(1) A new method to evaluate the P_O2 in a CO–CO₂–Ar gas mixture system was proposed. This method showed high accuracy and was verified by a thermodynamic experiment. The evaluation analysis leads to the following equation for the control method of the P_O2:   

$$P_{{\text{O}}_2}={\left\{\left(\frac{\text{setting}\mathrm{\hspace{0.17em} }\text{value}\mathrm{\hspace{0.17em} }\text{of}\mathrm{\hspace{0.17em} }\text{CO}-X}{\text{setting}\mathrm{\hspace{0.17em} }\text{value}\mathrm{\hspace{0.17em} }\text{of}\mathrm{\hspace{0.17em} }{\text{CO}}_2+X}\right)\times exp\left(-\frac{(\mathrm{\Delta }G°)}{RT}\right)\right\}}^{-2}$$

where X is the oxygen contamination, depending on the design of the gas passage and the purity of the gas mixture used.

(2) The solubility of chromium oxide decreases with an increase in slag basicity from 0.5 to 1.5. The P_O2 causes the solubility to decrease with an increase in P_O2 from 6.8 × 10⁻¹³ to 1.2 × 10⁻¹⁰ atm.

(3) The solubility of chromium oxide in the slag should be lower during the P_O2 change in the decarburization process to decrease the Cr oxidation loss. Therefore, the slag with a high slag basicity remains in a preferable condition as long as it is maintained in the molten state without any solid phase formation, such as Cr₂O₃ and CaCr₂O₄.

Acknowledgment

The authors would like to express their sincere gratitude to the research group at the Iron and Steel Institute of Japan for the “Melting and refining process of Cr-containing steel with high cleanliness by controlling the slag composition and inclusions” for their financial support.

References

• 1)   E. E.  Rees,  D. S.  McPhail,  M. P.  Ryan,  J.  Kelly and  M. G.  Dowsett: Appl. Surf. Sci., 203–204 (2003), 660.
• 2)   K. H.  Lo,  C. H.  Shek and  J. K. L.  Lai: Mater. Sci. Eng. R Rep., 65 (2009), 39.
• 3)   M.  Ryan,  D.  Williams,  R.  Chater,  B.  Hutton and  D.  McPhail: Nature, 415 (2002), 770.
• 4)   T.  Emi: ISIJ Int., 55 (2015), 36.
• 5)   M.  Venkatraman,  K.  Pavitra, V. Jana and T. Kachwala: Adv. Mater. Res., 794 (2013), 163.
• 6)   C. P.  Manning and  R. J.  Fruehan: JOM, 53 (2001), 36.
• 7)   B.  Arh and  F.  Tehovnik: Metalurgija, 50 (2011), 179.
• 8)   D. R.  Swinbourne,  T. S.  Kho,  B.  Blanpain,  S.  Arnout and  D. E.  Langberg: Miner. Process. Extr. Metall., 121 (2012), 23.
• 9)   R. J.  Choulet,  F. S.  Death and  R. N.  Dokken: Can. Metall. Q., 10 (1971), 129.
• 10)   K.  Takeda: Trans. Iron Steel Inst. Jpn., 19 (1979), 455.
• 11)   N.  Kikuchi: ISIJ Int., 60 (2020), 2731.
• 12)   S.  Kitamura,  K.  Naito and  G.  Okuyama: Miner. Process. Extr. Metall., 128 (2019), 34.
• 13)   Y.  Xiao,  M. A.  Reuter and  L.  Holappa: Metall. Mater. Trans. B, 33 (2002), 595.
• 14)   J.  Li,  B.  Liu,  Y.  Zeng,  Z.  Wang and  Z.  Gao: Environ. Monit. Assess., 189 (2017), 113.
• 15)   T. M.  Besmann,  N. S.  Kulkarni and  K. E.  Spear: J. Am. Ceram. Soc., 89 (2006), 638.
• 16)   P. L.  Dong,  X. D.  Wang and  S.  Seetharaman: Steel Res. Int., 80 (2009), 202.
• 17)   B.  Yan,  F.  Li,  H.  Wang and  D.  Sichen: Metall. Mater. Trans. B, 47 (2016), 37.
• 18)   K.  Morita,  K.  Tsukiashi,  M.  Kimura and  N.  Sano: Steel Res. Int., 76 (2005), 279.
• 19)   K.  Morita,  M.  Mori,  M.  Guo,  T.  Ikagawa and  N.  Sano: Steel Res., 70 (1999), 319.
• 20)   K.  Morita,  T.  Shibuya and  N.  Sano: Tetsu-to-Hagané, 74 (1988), 632 (in Japanese).
• 21)   K.  Morita,  A.  Inoue,  N.  Takayama and  N.  Sano: Tetsu-to-Hagané, 74 (1988), 999 (in Japanese).
• 22)   X.  Lu,  T.  Miki and  T.  Nagasaka: Int. J. Miner. Metall. Mater., 24 (2017), 25.
• 23)   H.  Kimura,  S.  Endo,  K.  Yajima and  F.  Tsukihashi: ISIJ Int., 44 (2004), 2040.
• 24)   J.  Li and  Y.  Kobayashi: ISIJ Int., 60 (2020), 1135.
• 25)   M.  Iwase,  S.  Miki and  T.  Mori: J. Chem. Thermodyn., 11 (1979), 307.
• 26)   Y. K.  Rao: Stoichiometry and Thermodynamics of Metallurgical Processes, 1st ed., Cambridge University Press, Cambridge, UK, (1985), 321.
• 27)   I.  Barin: Thermochemical Data of Pure Substances, 2nd ed., VCH Verlagsgesellschaft mbH, Weinheim, (1993), 271.
• 28)   E. T.  Turkdogan: Physical Chemistry of High Temperature Technology, Academic Press, New York, (1980), 10.
• 29)   Y.  Kobayashi,  N.  Yoshida and  K.  Nagai: ISIJ Int., 44 (2004), 21.
• 30)   H.  Wang,  Y.  Xiao,  K.  Jiang and  D.  Sichen: Metall. Mater. Trans. B, 39 (2008), 390.
• 31)   L.  Wang and  S.  Seetharaman: Metall. Mater. Trans. B, 41 (2010), 946.