Effect of MgO on Formation and Crystallization Behaviors of Calcium Ferrite during Heating and Cooling Processes

Abstract

MgO is one of essential component in blast furnace slag for improving the fluidity and desulfurization. Generally it is added via sintering process. It has been also proved in practice that the addition of MgO affects the quality of sinter significantly. However, the effect of MgO on formation and crystallization behaviors of calcium ferrite in sintering had been scarcely discussed. In this work, the crystalline products in CaO–Fe₂O₃ system with different content of MgO were identified by X-ray diffraction (XRD), scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDS). In order to follow the reactions during heating and cooling, differential scanning calorimetry (DSC) was also conducted. It was found that addition of MgO restrained the formation of CaFe₂O₄ leading to the decrease in content of original melt during sintering process. The contents of Ca₂Fe₂O₅ and solid solution of magnetite (magnetite s.s) were increased and the crystallization of Ca₂Fe₂O₅ was also promoted with increasing MgO. Compared with forming Ca₂Fe₂O₅, addition of MgO facilitated to form magnetite s.s. Content of initial liquid phase was decreased with increasing MgO, whereas new liquid phase was generated at 1609 K when the content of MgO exceeded 7 mass%, due to the reaction between Ca₂Fe₂O₅ and magnetite.

1. Introduction

Iron ore sinter is widely used as a major material of blast furnace for iron-making. During sintering process, iron ore fines or concentrates are converted into sinter with a given size, reliable physical strength, and desirable reducibility.¹⁾ With the growing consumption of high grade iron ore recently, the usage of low grade iron ore with high Al₂O₃ content is increasing.^2,3) Although the Al₂O₃ is essential for sinter, it also leads some negative impacts on reduction degradation property of the sinter, on fluidity and desulphurization of blast furnace slag.^4,5,6,7,8) At present, a main solution way is to introduce MgO addition into sinter. It has been reported in practice that addition of MgO improves low temperature reduction degradation of the sinter. Nevertheless, the strength of iron ore sinter decreases with increasing MgO content.

The quality of iron ore sinter largely depends on bonding phases. Iron ore fines or concentrates are bonded by a matrix containing complex phases with low melting-temperature (~1473 K). Among these bonding phases, silico-ferrite of calcium and aluminum (SFCA) is considered as a desirable one because of its good physicochemical properties for iron-making in blast furnace.^9,10) Many researchers have extensively studied the SFCA.^{11,12,13,14,15,16)} It has been clarified that binary calcium ferrite (Ca₂Fe₂O₅, CaFe₂O₄ or CaFe₄O₇) are formed in the first step, and then gangue components (Si, Al-containing minerals etc.) are dissolved into it to form SFCA.¹⁷⁾ The reaction mechanism between gangue components (SiO₂ or Al₂O₃) and calcium ferrite has been investigated systematically. For examples, Patrick et al.¹⁸⁾ had studied the stability of SFCA between 1513 K and 1663 K, claiming that the substitution mechanism during sintering process is 2(Fe³⁺, Al³⁺) → (Ca²⁺, Fe²⁺) + Si⁴⁺. Scarlett et al.¹⁹⁾ and Webster et al.¹⁾ had reported that the presence of Al₂O₃ can expand the stability range and lower the beginning temperature of the SFCA formation. One of authors^20,21,22) had investigated the effect of SiO₂ on formation of SFC (silico-ferrite of calcium), which is the precursor phase of SFCA, suggesting that Si⁴⁺ can replace Fe³⁺ located on octahedral layers of the binary calcium ferrite to form SFC. The reaction between CaFe₂O₄ and the SFC depresses the melting temperature.

Compared with SiO₂ or Al₂O₃, however, influence of MgO on formation of calcium ferrite is not studied systematically. More attention has been paid on the effect of MgO resources, such as dolomite, serpentine and olivine on the productivity or other industrial indexes of sinter.^{23,24,25,26,27,28)} Due to the complex nature of raw materials, the interactions among different gangue minerals is unavoidable in experimental process. As a result, it leads to different opinions on reaction behaviors of MgO in sinter. Some researchers take the opinion that MgO reacts with calcium ferrite forming a precursor phase of Mg-rich SFCA. For examples, Shigaki et al.²⁹⁾ suggested that MgO dissolved into melt forming di-calcium ferrite. Besides that, Loo et al.²⁴⁾ claimed that solid solution of magnetite (magnetite s.s) is formed as a result of solid-solid reaction between hematite and MgO in the first step, subsequently a solid di-calcium ferrite was formed by a reaction between CaO and magnetite s.s. On the contrary, Higuchi et al. and Panigrahy et al.^30,31) reported that dissolution of MgO into calcium ferrite took place firstly, and then magnetite s.s was mainly formed from the melt in which MgO and FeO_x content were high enough.

Compared with previous studies, ternary system of CaO–Fe₂O₃–MgO mixed by analytical-grade reagents were selected in this work. It can avoid the influence of other gangue components on formation of calcium ferrite. The products crystallized from molten CaO–Fe₂O₃ with MgO addition were identified by using X-ray diffraction and SEM-EDS. The formation and crystallization behaviors of calcium ferrite was recorded by DSC. In addition, the phase diagrams of CaO–Fe₂O₃–MgO at 1573 K and 1673 K are calculated by Factsage 7. Results in this study are expected to extend basis for revealing the reaction mechanisms of calcium ferrite in sintering process when MgO existed.

2. Experimental

2.1. Sample Preparation

Samples were prepared from analytical-grade reagents CaCO₃, Fe₂O_3, and MgO. The chemical compositions of starting samples were listed in Table 1. Among binary calcium ferrite, CaFe₂O₄ was considered to be an initial product during solid-solid reaction between CaO and Fe₂O₃. As a low melting temperature mineral, it was also the basis of liquid phase in sintering process. Therefore, mole ratio of CaO to Fe₂O₃ here was fixed as 1:1. Each sample was homogeneously blended in an agate mortar with anhydrous ethanol as a mixing medium for more than 30 minutes, and then oven-dried at 373 K for 30 minutes. After that, the dried sample was blended again for 10 minutes to ensure the homogenization.

Table 1. Chemical compositions of starting samples (mass%).

Sample No.	CaO	Fe2O3	MgO
S0	25.9	74.1	0.0
S1	25.7	73.3	1.0
S2	25.2	71.8	3.0
S3	24.6	70.4	5.0
S4	24.1	68.9	7.0

An electric resistance furnace with 8 U-shape MoSi₂ heating elements was employed. The isothermal zone was calibrated by S-type thermocouple with a precision of ±2 K. Sample (20 grams) was placed in a Pt crucible and elevated to 1673 K at a rate of 5 K/min. And then, it was isothermally heated at 1673 K for 3 hours. Finally, it was cooled to room temperature at a rate of 5 K/min in air.

2.2. Crystalline Products Identification (XRD, SEM and EDS)

A part of sample was removed from Pt crucible, ground into fine powder in an agate mortar and sieved all through a sieve with 50 μm mesh for XRD analysis. XRD data is collected in a continuous scanning mode at scanning speed of 10 deg/min from 10° to 90° using Rigaku SmartLab X-ray diffractometer (Rigaku Corporation, Tokyo, Japan). CuKα is used as a radiation source (40 kV×150 mA). Remaining sample was mounted in epoxy resin, ground with abrasive paper (400, 600, 800 and 1200 CW successively), polished with diamond slurry (0.5 μm) and prepared for SEM-EDS (MLA 250, FEI Quanta, USA).

2.3. Thermal Analysis (DSC)

Thermal analysis was performed using a simultaneous thermal analyzer instrument (TA-Q600, TA Cooperation, USA). Sample was placed into a Pt pan with another empty Pt pan as reference. The experiment was conducted in air with a gas flow of 100 ml/min. The heating rate of 20 K/min was chosen from 298 K to 1273 K (for a stage of decomposition of CaCO₃). After that a slower rate of 5 K/min was conducted to 1673 K. Finally, the sample was cooled to 1273 K at a rate of 5 K/min. Even though the temperature change in this study is much slower than that in industrial sintering, it is chosen as to detect the phase formation and crystallization with reasonable temperature resolution.

2.4. Quenched Experiment

In order to investigate the phase transformation, samples were quenched at given temperatures. 10 grams of the sample was put in a Pt crucible and hung in the isothermal zone of furnace. The furnace had a quenching pool with liquid nitrogen at the bottom. After being isothermally heated at the given temperature for 3 hours, the Pt crucible was dropped into the quenching pool and the sample was quenched by liquid nitrogen.

3. Results and Discussion

3.1. Thermal Analysis

Formation and crystallization behaviors of calcium ferrite in CaO–Fe₂O₃–MgO system was followed by DSC. The DSC curves were shown in Fig. 1. As shown in Fig. 1(a), there were 3 endothermic peaks, which were referred as A, B and C, appearing at 1478 K, 1491 K and 1609 K respectively during heating process. According to binary phase diagram of CaO–Fe₂O₃ system, the endothermic peaks A and B are corresponding to the eutectic reaction between CaFe₂O₄ and CaFe₄O₇, and decomposing of CaFe₂O₄ respectively as following,   

$${\text{CaFe}}_2{\text{O}}_4+{\text{CaFe}}_4{\text{O}}_7\to \text{L}$$(1)

  

$${\text{CaFe}}_2{\text{O}}_4\to {\text{Ca}}_2{\text{Fe}}_2{\text{O}}_5+\text{L}$$(2)

Fig. 1.

DSC curves of samples during (a) heating and (b) cooling processes.

Where L is liquid phase.

The reaction between CaFe₂O₄ and CaFe₄O₇ (peaks A) occured in sample of S0 (MgO=0 mass%) while disappeared after adding MgO. When the temperature reached 1491 K, a large amount of liquid phase were formed due to the decomposing of CaFe₂O₄ (peak B). With increasing MgO content from 1 to 7 mass%, the intensity of endothermic peak B decreased gradually. It indicates that the additon of MgO inhibited the formation of CaFe₂O₄ which is a precursor of original melt of iron ore sinter.

In addition, there was another endothermic peak C appearing at 1609 K in S4 (MgO=7 mass%). Because of the lack of ternary phase diagram of MgO–CaO–Fe₂O₃ system under curent experiment condition, Factsage 7.0 was used for calculation. The calculated ternary phase diagram of MgO–CaO–Fe₂O₃ system was shown in Fig. 2. The composition of each sample (S0-S4) were marked as hollow blocks.

Fig. 2.

Ternary phase diagram of MgO–CaO–Fe₂O₃ system at different temperature in air. F: Fe₂O₃; C₂F: Ca₂Fe₂O₅; spinel: magniferous magnetite.

As shown in Fig. 2, when temperature was increased from 1573 K to 1673 K, the equilibrium phases of S4 (MgO=7 mass%) changed from liquid, Ca₂Fe₂O₅ and spinel to liquid, spinel and periclase. The spinel in Fig. 2 is inferred as magniferous magnetite and it will be discussed later. The magniferous magnetite [(Mg_x,Fe_1−x)O·Fe₂O₃] is termed³²⁾ to distinguish with other solid solution of magnetite (magnetite s.s.) such as (Ca_x,Fe_1−x)O·Fe₂O₃ and (Ni_x,Fe_1−x)O·Fe₂O₃, though their crystal structure are the same. Considering disappearance of Ca₂Fe₂O₅ and appearance of periclase from 1573 K to 1673 K, it can be inferred that the occurrence of endotherimic peak C is caused by two posibilities. One posibility is that MgO incorporates into Ca₂Fe₂O₅ forming solid solution, and depresses the decomposing temperature of Ca₂Fe₂O₅; The other one is caused by the reaction between Ca₂Fe₂O₅ and magniferous magnetite.

In order to study the first posibiltiy mentioned above, pure Ca₂Fe₂O₅ was synthetized. CaCO₃ and Fe₂O₃ were mixed with the mole ratio of 2:1 and sintered at 1523 K for 4 hours. After that, the sintered sample is pulverised, mixed and sintered at 1523 K for 4 hours again. The pure Ca₂Fe₂O₅ was confirmed by XRD. The Ca₂Fe₂O₅ and MgO were mixed and compacted into a tablet (denoted as S5). Content of MgO addition in S5 was chosen as 10.7 mass% according to the assumption that CaO in S4 was all used to form Ca₂Fe₂O₅. The tablet was heated up to 1623 K with a heating rate of 5 K/min, and then sintered isothermally for 3 hours.

Figure 3(a) shows that the XRD patterns of S5 were almost the same before and after sintering. In additon, the initial external shape of S5 was still maintained after sintering. It indicates that the sample had not melted at 1623 K. Figure 3(b) shows that MgO was not incorporated into Ca₂Fe₂O₅ forming solid solution. The sintering tempature was higher than that of peak C and isothermal time was much longer, it can be conclued that the endotherimic peak C in Fig. 1(a) was not caused by the decomposing of solid solution of Ca₂Fe₂O₅.

Fig. 3.

(a) The external morphology and XRD patterns of S5 before and after sintering; (b) The BSE image of S5 after sintering with EDS result.

To verify the formation of liquid phase, Ca₂Fe₂O₅ and magnetite were blended with mole ratio of 1:1 for DSC experiment. The heating rate of DSC was the same with that in Fig. 1. In order to avoid oxidation of Fe₃O₄, N₂ flow with 100 ml/min was chosen in the experimental process.

As shown in Fig. 4, two endothermic peaks were observed at 1481 K and 1491 K in DSC curve. It indicates that liquid phase was formed by reaction between Ca₂Fe₂O₅ and magnetite. Based on Fig. 5, the melting temperature of ternary calcium ferrite (CWF) depended on mass ratio of Ca₂Fe₂O₅ to magnetite. Therefore, the formation of liquid phase at 1609 K (peak C) was reasonable on thermodynamics. Combinning with the results from Figs. 2, 3, 4, 5, the endotherimic peak C was proved to be caused by following:   

$${\text{C}}_2\text{F}+\text{S}\to \text{L}$$(3)

Where C₂F is Ca₂Fe₂O₅, S is spinel (magniferous magnetite) and L is liquid phase.

Fig. 4.

The DSC curve of mixture of Ca₂Fe₂O₅ and Fe₃O₄ in nitrogen.

Fig. 5.

Phase diagram of CaO–Fe₂O₃–FeO system in air.

Figure 1(b) shows the crystallizaiton behaviors of CaO–Fe₂O₃ melt with different content of MgO. According to binary phase diagram of CaO–Fe₂O₃ system, in curve of S0 (MgO=0 mass%), the exothermic peaks H, I and N are corresponding to crystallization of Ca₂Fe₂O₅, peritectic reaction from liquid and Ca₂Fe₂O₅ to CaFe₂O₄ and eutectic reaction between CaFe₂O₄ and CaFe₄O₇ respectively as following:   

$$\text{L}\to {\text{Ca}}_2{\text{Fe}}_2{\text{O}}_5$$(4)

  

$$\text{L}+{\text{Ca}}_2{\text{Fe}}_2{\text{O}}_5\to {\text{CaFe}}_2{\text{O}}_4$$(5)

  

$$\text{L}\to {\text{CaFe}}_2{\text{O}}_4+{\text{CaFe}}_4{\text{O}}_7$$(6)

After adding MgO into CaO–Fe₂O₃ system, the exothermic peak N disappeared. Moreover, the exothermic peaks during cooling process were classified into 2 groups. One was corresponding to crystallization of Ca₂Fe₂O₅ composing of peaks D, E, F and G. The other was corresponding to peritectic reaction between liquid and Ca₂Fe₂O₅ composing of peaks J, K, L and M. As shown in Fig. 1(b), the crystallization temperature of Ca₂Fe₂O₅ increased from 1498 K to 1583 K with increasing MgO content. However, the peritectic temperature of Ca₂Fe₂O₅ decreased from 1482 K to 1469 K and the intensity of its exothermic peak also decreased gradually.

The crystallization ratio of Ca₂Fe₂O₅ can be calculated by the DSC curve as following:   

$${\text{X}}_{\text{t}}=\frac{{\text{S}}_{{\text{t}}_0}^{\text{t}}}{\text{S}}$$

Where X_t is the crystallization ratio that corresponding to time t. ${\text{S}}_{{\text{t}}_0}^{\text{t}}$ is the area surrounded by DSC curve and the baseline from onset time t₀ to a given time t. S is the total area of crystallization peak from onset time to final time.

The crystallization ratio of Ca₂Fe₂O₅ in melt with different MgO content as a function of time was shown in Fig. 6. It can be seen that the incubation time corresponding to crystallization of Ca₂Fe₂O₅ in the melt becomes shorter after adding MgO. At the primary stage of crystallization (X_t<50%), with increasing MgO, the crystallization ratio of Ca₂Fe₂O₅ became higher. It indicates that MgO addition promoted the crystallization of Ca₂Fe₂O₅ in the melt. When the MgO content exceeded 3 mass%, the crystallization rate of Ca₂Fe₂O₅ had not changed obviously.

Fig. 6.

The crystallization ratio of Ca₂Fe₂O₅ in melt with different MgO content as a function of time.

3.2. Crystalline Products Identification

Aiming to study the effect of MgO on the crystalline products of calcium ferrite, XRD was conducted. The XRD patterns of samples were shown in Fig. 7.

Fig. 7.

XRD patterns of crystalline products of melt with different MgO content (The phases 2 and 3 are tentative and marked as *).

It was shown that CaFe₂O₄ was the solo crystalline product in sample S0 (MgO=0 mass%). The diffraction peaks of Ca₂Fe₂O₅ and magnetite were observed after adding MgO. The contents of Ca₂Fe₂O₅ and magnetite increased simultaneously with increasing MgO content, while the content of CaFe₂O₄ decreased. However, as mentioned in introduction, there were different opinions on reaction behaviors of MgO in sinter. 1) MgO reacted with FeO_x, forming (Mg, Fe)O·Fe₂O₃ as reaction 7;³⁰⁾ 2) MgO reacted with CF (calcium ferrite), forming CFM (calcium ferrite with MgO) as reaction 8.³³⁾   

$$\mathrm{M}\mathrm{g}\mathrm{O}\left(\mathrm{s}\right)+\mathrm{F}\mathrm{e}{\mathrm{O}}_{\mathrm{x}}\left(\mathrm{s}\right)\to \left(\mathrm{M}\mathrm{g},\mathrm{\hspace{0.17em} }\mathrm{F}\mathrm{e}\right)\mathrm{O}\cdot \mathrm{F}{\mathrm{e}}_2{\mathrm{O}}_3\left(\mathrm{s}\right)$$(7)

  

$$\mathrm{M}\mathrm{g}\mathrm{O}\left(\mathrm{s}\right)+\mathrm{C}\mathrm{F}\left(\mathrm{s}\right)\to \mathrm{C}\mathrm{F}\mathrm{M}\left(\mathrm{s}\right)$$(8)

The XRD pattern of magnetite is the same with that of magnesium ferrite and magniferous magnetite. Since XRD patterns could only demonstrate the crystal structures of phases, it failed to distinguish magnetite and others. Therefore, the products in Fig. 7 are tentative. To identify the tentative products, further analysis was conducted by using SEM-EDS.

3.3. Element Distribution in Magnetite Solid Solution

Figure 8 shows the backscattered electron images for samples S1 (MgO=1 mass%), S2 (MgO=3 mass%), S3 (MgO=5 mass%) and S4 (MgO=7 mass%) respectively. The EDS results were listed in Table 2.

Fig. 8.

Backscattered electron images of samples. a (MgO=1 mass%), b (MgO=3 mass%), c (MgO=5 mass%) and d (MgO=7 mass%).

Table 2. EDS results of samples S1, S2, S3 and S4 (mass%) (Number 1, 2 and 3 are corresponding to points in Fig. 8).

	S1			S2			S3			S4
	1	2	3	1	2	3	1	2	3	1	2	3
Ca	21.41	32.90	2.57	21.34	30.72	2.87	20.72	32.21	2.85	21.48	33.27	3.95
Fe	51.87	40.36	59.68	50.79	41.24	57.89	52.17	40.01	58.33	56.20	45.26	62.69
Mg	–	–	9.32	–	–	9.66	–	–	10.01	–	–	9.88
O	26.71	26.73	28.43	27.87	28.04	29.59	27.11	27.78	28.81	22.32	21.47	23.49

It was found that phases No. 1 and No. 2 denoted in Fig. 8 were CaFe₂O₄ and Ca₂Fe₂O₅ respectively. The stoichiometry of No. 3 was closed to magnetite. Combining the results of XRD and EDS, a crystalline product of magnetite s.s with Mg and Ca was confirmed.

As listed in Table 2, Mg only existed on magnetite s.s. Since the ionic radius of Mg²⁺ is closed to that of Fe²⁺ (ionic radius of Mg²⁺ is 0.66 Å and Fe²⁺ is 0.74 Å) and they are all divalent cation, Mg²⁺ was considered to substitute for Fe²⁺ in FeO·Fe₂O₃ forming the magniferous magnetite ((Mg_1−x,Fe_x)O·Fe₂O₃). Because the content of Ca²⁺ in the melt was much higher than Mg²⁺, a little of Ca²⁺ probably diffused into (Mg_1−x,Fe_x)O·Fe₂O₃. Hayashi et al. investigated the equilibrium phases in FeO_x–CaO–SiO₂–MgO system at 1573 K and obtained the same result. Meanwhile, they also found that liquid phase area became smaller with increasing MgO content which agreed with the results in Fig. 1.³⁴⁾ Even though the reaction time in this study is much longer than that in industrial sintering, ternary calcium ferrite bearing MgO was still not found. It can be inferred that MgO may not directly react with calcium ferrite to form a precursor phase unless there are any other gangue elements.

In this study, initial mole ratio of CaO and Fe₂O₃ in sample was fixed as 1:1. Since Fe₂O₃ was consumed to form (Mg_1−x,Fe_x)O·Fe₂O₃, the relative content of CaO in remaining melt was increased. With increasing MgO, the composition of remaining melt moved toward CaO-rich region (from A to B in Fig. 9). Thereby, that is the reason why the content of liquid phase decreased in heating process, and the crystallization temperature of Ca₂Fe₂O₅ increased (from A’ to B’ in Fig. 9) in cooling process. It is considered to be a main reason of sinter degradation. The former resulted in the decrease in binding phase, and the latter one led to shorten the wetting time of the binding phase with iron ore fines.

Fig. 9.

Phase diagram of CaO–Fe₂O₃ system in air.³⁵⁾

4. Conclusions

To investigate the effect of MgO on formation and crystallization behaviors of calcium ferrite, the crystalline products is identified by XRD and SEM-EDS. DSC was also applied to follow the kinetic behaviors of the melt with MgO. The results were summarized as follows:

(1) Compared with forming Ca₂Fe₂O₅, addition of MgO facilitated to form magnetite s.s with Mg and Ca. With increasing MgO content, among the crystalline products, the contents of Ca₂Fe₂O₅ and magnetite s.s increased while the content of CaFe₂O₄ decreased.

(2) Because of the formation of magnetite s.s, the content of liquid phase decreased in heating process. Simultaneously MgO addition increased the crystallization temperature of Ca₂Fe₂O₅ and the crystallization of Ca₂Fe₂O₅ was also promoted. This was considered to be a main reason of sinter degradation with MgO addition.

(3) New liquid phase was generated at 1609 K when the content of MgO exceeded 7 mass%, due to the reaction between Ca₂Fe₂O₅ and magnetite.

Acknowledgments

The authors would like to thank the National Natural Science Foundation of China for providing financial supports (U1460201, 51374017 and 51774029). Department of Material and Engineering in Kyushu University are also acknowledged for their support in the software of Factsage 7.

References

• 1)   N. A. S.  Webster,  M. I.  Pownceby,  I. C.  Madsen and  J. A.  Kimpton: Metall. Mater. Trans. B, 43B (2012), 1344.
• 2)   T.  Maeda,  K.  Nishioka and  M.  Shimizu: ISIJ Int., 49 (2009), 625.
• 3)   J.  He: ISIJ Int., 51 (2011), 696.
• 4)   A.  De,  S. S.  Gupta and  A.  Chatterjee: SEAISI Q., 21 (1992), 35.
• 5)   A. K.  Das,  A.  Kumar,  A. K.  De and  T.  Mukherjee: 60th Ironmaking Conf., Iron and Steel Society, Warrendale, PA, (2001), 973.
• 6)   C. U.  Kumar,  R. V.  Ramana,  S.  Ali,  A. K.  Das,  A.  Kumar,  A. K.  De and  T.  Mukherjee: Tata Search, (1995), 20.
• 7)   L.  Lu,  R. J.  Holmes and  J. R.  Manuel: ISIJ Int., 47 (2007), 349.
• 8)   F.  Shen,  X.  Jiang,  G. S.  Wu,  G.  Wei,  X. G.  Li and  Y. S.  Shen: ISIJ Int., 46 (2006), 1.
• 9)   N. J.  Bristow and  A. G.  Waters: Trans. Inst. Min. Metall. Sect. C, 100 (1991), 1.
• 10)   I.  Shigaki,  M.  Sawada and  N.  Gennai: Trans. Iron Steel Inst. Jpn., 26 (1986), 503.
• 11)   B.  Phillips and  A.  Muan: J. Am. Ceram. Soc., 42 (1959), 413.
• 12)   S.  Watanabe: Tetsu-to-Hagané, 52 (1966), 251.
• 13)   M.  Sasaki and  T.  Nakazawa: Tetsu-to-Hagané, 54 (1968), 1217.
• 14)   L. X.  Yang and  C. E.  Loo: ISIJ Int., 37 (1997), 449.
• 15)   Z. Q.  Wang,  K.  Ishii,  Y.  Sasaki,  T.  Tsutsumi,  K.  Higuchi and  Y.  Hosotani: Tetsu-to-Hagané, 84 (1998), 689.
• 16)   C. E.  Loo and  W.  Leung: ISIJ Int., 43 (2003), 1393.
• 17)   K.  Inoue and  T.  Ikeda: Tetsu-to-Hagané, 68 (1982), 2190.
• 18)   T. R. C.  Patrick and  M. I.  Pownceby: Metall. Mater. Trans. B, 33B (2002), 79.
• 19)   N. V. Y.  Scarlett,  M. I.  Pownceby,  I. C.  Madsen and  A. N.  Christensen: Metall. Mater. Trans. B, 35B (2004), 929.
• 20)   X.  Ding and  X. M.  Guo: Metall. Mater. Trans. B, 45B (2014), 1221.
• 21)   X.  Ding,  X. M.  Guo,  C. Y.  Ma,  K.  Tang and  Y. D.  Zhao: Metall. Mater. Trans. B, 46B (2015), 1146.
• 22)   X.  Ding and  X. M.  Guo: Metall. Mater. Trans. B, 46B (2015), 1742.
• 23)   L. H.  Hsieh and  J. A.  Whiteman: ISIJ Int., 33 (1993), 462.
• 24)   C. E.  Loo,  L. C.  Mackey and  B. M.  England: Trans. Inst. Min. Metall. Sect. C, 105 (1996), 175.
• 25)   E.  Kasai,  Y.  Sakano,  T.  Kawaguchi and  T.  Nakamura: ISIJ Int., 40 (2000), 857.
• 26)   K.  Higuchi,  M.  Naito,  M.  Nakano and  Y.  Takamoto: ISIJ Int., 44 (2004), 2057.
• 27)   M.  Nakano,  M.  Naito,  K.  Higuchi and  K.  Morimoto: ISIJ Int., 44 (2004), 2079.
• 28)   M.  Matsumura,  M.  Hoshi and  T.  Kawaguchi: ISIJ Int., 45 (2005), 594.
• 29)   I.  Shigaki,  M.  Sawada,  M.  Maekawa and  K.  Narita: Tetsu-to-Hagané, 66 (1980), 1612.
• 30)   K.  Higuchi,  T.  Tanaka and  T.  Sato: ISIJ Int., 47 (2007), 669.
• 31)   S. C.  Panigrahy,  P.  Verstraeten and  J.  Dilewijns: Metall. Mater. Trans. B, 15B (1984), 23.
• 32)   Y. F.  Guo and  X. M.  Guo: ISIJ Int., 57 (2017), 228.
• 33)   G. L.  Zhang,  S. L.  Wu,  S. G.  Chen,  J.  Zhu,  J. X.  Fan and  B.  Su: ISIJ Int., 53 (2013), 1515.
• 34)   M.  Hayashi,  H.  Tanaka,  T.  Watanabe and  M.  Susa: ISIJ Int., 57 (2017), 2124.
• 35)   B.  Philips and  A.  Muan: J. Am. Ceram. Soc., 41 (1958), 445.