Intermittent Microscopic Observation of Structure Change and Mineral Reactions of High Phosphorus Oolitic Hematite in Carbothermic Reduction

Abstract

Change of the typical structure of high phosphorous oolitic hematite during carbothermic reduction process have been studied using intermittent observation. It indicated that there were four change stages, including crack formation and growth, mineral diffusion, mineral crystalization, mineral crystal migration and reaction. Eventually, inherent oolitic structure was completely destroyed. Mineral reactions were also investigated and discussed along the structural change, giving much understanding of thermal behavior of the oolitic hematite during reduction. The results showed a variety of mineral reactions occurred, such as decomposition of CaCO₃, reduction of iron oxide, dehydration and decomposition of chlorite, decomposition of fluorapatite, and transformation of gangues. SiO₂ and Al₂O₃ decomposed from chlorite can not only combine with FeO to form Fe₂SiO₄ and FeAl₂O₄, which hindered the reduction of iron oxide, but also promoted the decomposition of fluorapatite to form Ca₃(PO₄)₂ in the presence of SiO₂. Then, Ca₃(PO₄)₂ was reduced to form CaSiO₃ and P₂, which was the main process of dephosphorization. However, a small amount of fluorapatite diffused into iron oxides, which made it difficult to separate by mechanical crushing and screening.

1. Introduction

Deposits of high phosphorus oolitic hematite have been known for many years as a source of raw material for the steel industry. The resources of high phosphorus oolitic hematite are mainly distributed in the north of Western Europe, Ukraine, Canada, and the United States.¹⁾ There are also enormous high phosphorus oolitic hematite deposits with existing reserve of 3.72 billion tons in China. The iron content is up to about 50% and phosphorus content is also high, generally above 0.4 to 1.0%.²⁾ With the rapid development of the iron and steel industry, the demand for iron ore is constantly rising. Development and utilization of complex refractory iron ore has become an important way to solve the contradiction between supply and demand in the steel industry. High phosphorus oolitic hematite as a typical complex refractory iron ore is particularly important to exploit.

Currently, many researches have been carried out on high phosphorous oolitic hematite, which mainly focused on iron increase and dephosphorization, but rarely study about particular structure of high phosphorus oolitic hematite in detail.^3,4,5,6,7) In reduction process of high phosphorus oolitic hematite, influencing factors in reduction of the high phosphorous oolitic hematite were not only mineral types and element contents, but also the distribution of the oolitic structure. Therefore, the study for oolitic structure is important in the study of high phosphorus oolitic hematite.

At present, the studies on the reduction process of oolitic hematite in micro scale have been carried out, with focusing on the reduction mechanism of iron oxide and element migration.^8,9) Our laboratory has been engaged in research of direct reduction process of carbon-bearing pellets, especially high phosphorus oolitic hematite.^10,11) Taking into account that the different mineral types in high phosphorus oolitic hematite resulting in quite different behavior during carbothermic reduction process, it is necessary to carry out individual study by considering containing mineral phases such as phosphorous and others. In the process of carbothermal reduction of high-phosphorus iron ore, not only mineral species and element contents, but also the oolitic structure has a great influence on the reduction. However, little is known about the effect of oolitic structure on reductive dephosphorization during the reduction process. Understanding oolitic structure and mineral changes in the process of reduction, plays an important significance to research and application of dephosphorization of high phosphorus iron -phosphate. In this paper, the effects of oolitic structure change and mineral reaction on reduction and dephosphorization of high phosphorus iron ores have been studied. Understanding of relationships between the oolitic structure changes and mineral reaction and diffusion mechanisms may serves as a basis for control and optimization of the high phosphorus oolitic hematite reduction process, what’s more, it also provides a new method for the research of direct reduction of refractory iron ore.

2. Experimental

2.1. Raw Materials

The high phosphorus oolitic hematite used in this study was produced in western Hubei, China. The mineralogy of oolitic hematite ores is shown in Fig. 1, in which (a) illustrates an oolitic distribution and (b) presents a typical oolitic structure observed in this study. Usually, oolitic structure is composed of one or more elliptical shape ring.¹²⁾ Diameter of the oolitic structure in the sample was about 200 mm. The core of the structure was composed of quartz, and the oolitic ring was composed of hematite, chamosite and fluorapatite. Gangue minerals such as quartz and calcium carbonate were distributed outside of the oolitic structure.

Fig. 1.

Oolitic structure of high phosphorus oolitic hematite. (Online version in color.)

In order to reduce the impact of the elements migration, high purity graphite (99.9 mass% Carbon) with the size of 200 meshesis used as a reducing agent in this experiment.

2.2. Experimental Procedure

To make the mineral reactions and oolitic change process clear, typical high phosphorus oolitic structure was supplied to carbon reduction experiments. High phosphorus oolitic hematite were observed intermittently by using scanning electron microscope (SEM) and energy dispersive spectrometer (EDS) analysis. Mineral reactions in high phosphorus oolitic hematite and the migration of major elements in mineral were analyzed by observing changes of oolitic structure at different time.

Before the reduction, the raw ore were smoothed and polished to sample with a length of 20 mm, a width of 10 mm and a thickness of 3 mm. A smooth surface (20×10 mm) was chosen as an observation plane. Figure 2 illustrates the schematic diagram for sample buried in high purity graphite. The sample with observation plane face-up was put into an alumina crucible, filled with high purity graphite. High purity graphite was milled to below 75 um. The graphite powders were compacted to ensure the contact between the observation plane and the graphite powders. Reaction experiment was performed in a laboratory furnace, the sample was pushed into the laboratory furnace at 1200°C with the protection of nitrogen to prevent from the sample oxidation by air, the sample was moved out of the furnace after roasting for five minutes, and cooled to the ambient temperature in a nitrogen atmosphere. The sample was cleaned with alcohol and dried at 100°C for 5 h in a drying oven and observed. The procedures of the roasting and observation, were repeated until the oolitic structure disappeared.

Fig. 2.

Schematic diagram for sample buried in the pure graphite. (Online version in color.)

3. Results and Discussion

3.1. Oolitic Structure Change in Carbothermic Reduction

In this experiment, the carbothermic reduction which was a typical solid reaction process occurred at the interface between the sample and reducing agent. This reaction process is complex and continuous, including physical changes and chemical reactions.

Figure 3 shows the oolitic structure change in the entire reduction process. At the beginning of the reaction (5–15 min), oolitic structure remained basically intact, but hairline cracks appeared in the oolitic boundary, chlorite ring boundary and CaCO₃ boundary.

Fig. 3.

Structure change of typical high phosphorus oolitic hematite in carbothermic reduction (from 0 to 120 minutes).

At 20 min of the reaction, small cracks began to grow. At 25–55 min of the reaction part of cracks extended from the outside to the inside oolitic, distributing throughout the reaction surface, at this time, oolitic structure started being destroyed. In addition, CaCO₃ particles flaked from the surface of high phosphorus oolitic hematite, leaving a pit. At the same time, graphite powder entered, which increased reduction potential around the pit and aggravating mineral reduction. Therefore, chlorite ring was disconnected and fluorapatite ring was becoming thinner. However, the reduction process was difficult in the area away from the pit, because of the diffusion of fluorapatite and chlorite, the ring became wider.

At 60–70 min of the reaction, the reaction surface of mineral particle was further activated. As a result, crystal nucleus was formed and grew up, granular FeO and rod-shaped fluorapatite were formed in the reaction interface. At this stage, the structure of the mineral grains was incomplete. Ultimately through the lattice correction, mineral grains tended to thermodynamically stable state, therefore, grain structure was gradually completed, and grain boundaries became clear. Since a large number of grains formed and grew, the reaction surface became loose, cracks were filled with grown grains and finally disappeared. At the time of 60 min, a large number of FeO grains migrated to the oolitic core. Diffusing fluorapatite grains were in close contact with the chlorite grains, which made it easy to react (corresponding with the previous study of pure substance), so that fluorapatite and chlorite ring were thinner.

As the reaction proceeded (75–90 min), migration of ions in the reaction surface was intensified, diffusion of grains was promoted, and oolitic structure was completely destroyed. Fluorapatite grains and chlorite grains were continuously reacted, which caused oolitic ring gradually disappeared. Because FeO grains covering at the surface of the quartz core were gradually disappeared, quartz grains were exposed again.

When the reaction time exceeded 95 min, fayalite (Fe₂SiO₄) and hercynite (FeAl₂O₄) were formed inside oolitic structure. The iron oxide was reduced to metallic iron, and a small amount of unreacted fluorapatite grains interspersed among iron oxide grains.

3.2. Mineral Reactions in Carbothermic Reduction

Oolitic structure change was linked to mineral reactions, the study of mineral reactions would gave a reasonable explain for the oolitic structure change process. In this subsection, the figures of mineral reactions were corresponded to the oolitic structure change (A–I in Fig. 3).

3.2.1. Dehydration and Thermal Decomposition of Minerals

As shown in Fig. 4(A), at the beginning of the reaction, cracks appeared on oolitic boundary. This was because there are many small cracks or defects on raw ore oolitic boundary, in the reaction process of high temperature, stress was easy to concentrate in the vicinity of these cracks and defects, when the stress reached an adequate level, cracks began to grow and extend.

Fig. 4.

Cracks at oolitic boundary (A) and internal part of oolitic structure (B) in sample reduced for 5 min.

Figure 4(B) shows the cracking morphology at CaCO₃ boundary and inside of oolitic structure. The formation of cracks may be related to dehydration and decomposition of minerals. Moreover, it may be also related to the expansion and contraction of minerals. In 1200°C, cracks were formed on CaCO₃ boundary. There were two reasons to explain it. First, due to thermal expansion coefficient of CaCO₃ is negative (−60 (10E−6/°C)), at high temperature, the volume of CaCO₃ shrank, which made it difficult to combine with other minerals.¹³⁾ Second, CaCO₃ was easy to decomposed into CaO and CO₂, subsequently CO₂ over-flowed from CaCO₃ minerals, so the close contact between minerals were destroyed, and cracks extended gradually.

Within the oolitic structure, some small cracks appeared in chlorite ring, which is the result of chlorite dehydration reactions. Two dehydration reactions were occurred as shown in Eqs. (1) and (2).¹⁴⁾ Firstly, three molecules of crystal water could be removed from one molecule of chrorite at 500–700°C. Secondly, the last one molecule of crystal water could be removed. Through these dehydration reactions, chlorite became loose and eventually formed the mineral containing Si, Al and Mg.   

$$\begin{array}{l}\mathrm{M}{\mathrm{g}}_2\mathrm{A}\mathrm{l}{(\mathrm{O}\mathrm{H})}_6\cdot \mathrm{M}{\mathrm{g}}_3(\mathrm{S}{\mathrm{i}}_3\mathrm{A}\mathrm{l}){\mathrm{O}}_{10}{(\mathrm{O}\mathrm{H})}_2(\mathrm{s})\\ =\mathrm{M}{\mathrm{g}}_2\mathrm{A}\mathrm{l}{\mathrm{O}}_3\cdot \mathrm{M}{\mathrm{g}}_3(\mathrm{S}{\mathrm{i}}_3\mathrm{A}\mathrm{l}){\mathrm{O}}_{10}{(\mathrm{O}\mathrm{H})}_2(\mathrm{s})+3{\mathrm{H}}_2\mathrm{O}(\mathrm{g})\end{array}$$(1)

  

$$\begin{array}{l}\mathrm{M}{\mathrm{g}}_2\mathrm{A}\mathrm{l}{\mathrm{O}}_3\cdot \mathrm{M}{\mathrm{g}}_3(\mathrm{S}{\mathrm{i}}_3\mathrm{A}\mathrm{l}){\mathrm{O}}_{10}{(\mathrm{O}\mathrm{H})}_2(\mathrm{s})\\ =\mathrm{M}{\mathrm{g}}_2\mathrm{A}\mathrm{l}{\mathrm{O}}_3\cdot \mathrm{M}{\mathrm{g}}_3(\mathrm{S}{\mathrm{i}}_3\mathrm{A}\mathrm{l}){\mathrm{O}}_{11}(\mathrm{s})+{\mathrm{H}}_2\mathrm{O}(\mathrm{g})\end{array}$$(2)

At this stage, decomposition and reduction reactions of other minerals did not occur, so minerals are closely interrelated. This stage could reserve energy for grains formation and mineral reaction in carbothermic reduction process.

3.2.2. Reduction of Iron Oxide

In fact, the reduction process of iron ores is extremely complex, especially when coal is used as the reductant. Figure 5 shows the morphology of the sample after reduction for 15 min. S. Weissberger and Y. Zimmels^8,9) had clarified the impact of oolitic structure on iron oxide reduction process, and observed the formation and growth of metallic iron phase. As shown in Figs. 5(a) and 5(b), we got a similar observation to their previous research. Moreover, it indicated that the dark areas in the middle of the particles are holes which are formed, due to removal of softer hematite core (Fig. 5(a)). Figure 5(b) shows the morphology of the reduced iron, in which the metallic iron particle have appeared.

Fig. 5.

The reduction of iron oxide ((a) dark areas in the middle of the FeO particles, (b) the morphology of metallic iron grain).

In carbothermic reduction process, the reduction follows a series of stepwise reduction reaction (Fe₂O₃→Fe₃O₄→FeO→Fe).^15,16,17,18) In this study, iron oxide reduction was in direct and indirect reduction. The direct reduction process can be the main way, because of the close contact between the surface of sample and graphite.¹⁹⁾ Then, it produced CO gas, which also created indirect reduction, which made the reduction of iron oxides promoted a lot.

3.2.3. Diffusion and Decomposition of Chlorite

Figure 6 shows the diffusion behavior of chlorite at 20 min. Figure 7 shows the surface distribution of Ca, P, Si, and Al at 5 min and 60 min. Along with the reaction, chlorite diffused from ring and quickly decomposed by following Eq. (3) and disappeared.   

$$\begin{array}{l}\mathrm{M}{\mathrm{g}}_2\mathrm{A}\mathrm{l}{\mathrm{O}}_3\cdot \mathrm{M}{\mathrm{g}}_3(\mathrm{S}{\mathrm{i}}_3\mathrm{A}\mathrm{l}){\mathrm{O}}_{11}(\mathrm{s})=2(2\mathrm{M}\mathrm{g}\mathrm{O}\cdot \mathrm{S}\mathrm{i}{\mathrm{O}}_2)\mathrm{\hspace{0.17em} }\text{(}\mathrm{s}\text{)}\\ +\mathrm{M}\mathrm{g}\mathrm{O}(\mathrm{s})+\mathrm{S}\mathrm{i}{\mathrm{O}}_2(\mathrm{s})+\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3(\mathrm{s})\end{array}$$(3)

Fig. 6.

Diffusion of chlorite ring in the sample reduced for 20 min.

Fig. 7.

Decomposition of chlorite ring in the sample reduced for 60 min. (Online version in color.)

As can be seen from Fig. 7, Ca and P had the same distribution area, which indicated that fluorapatite was not reduced. However, the surface distribution of Si and Al were different at 5 min and 60 min.

At 5 min, Al and Si had the same distribution area, which indicated that chlorite was not reacted; At 60 min, Al and Si were separated, Al was mainly distributed in the areas of Fe elements, while Si was distributed in the areas of Ca and P. This is because Al₂O₃ was easier to combine with iron oxide than SiO₂, and formed FeAl₂O₄ eventually, when the decomposition of chlorite was occurred. On the other hand, SiO₂ could promote decomposition of fluorapatite, and then reacted with CaO to form CaSiO₃, which corresponding with the previous study of pure substance.²⁰⁾ The detailed description of fluorapatite reduction process would be given in 3.2.4.

3.2.4. Reduction and Diffusion of Fluorapatite

In high phosphorus oolitic hematite, the phosphorus was mainly existed in the form of fluorapatite, therefore, it is important to study the evolution process of fluorapatite for dephosphorization of high phosphorus oolitic hematite.

Figure 8 shows the distribution of Ca and P from the oolitic core to boundary at different times. Reduction of fluorapatite can be presumed by P/Ca. Before 25 min, the content of P was also high in the place where content of Ca was high, which indicated that fluorapatite ring didn’t reduced and remained intact at this stage. With the proceed of the reduction, P content decreased a lot, while Ca content was essentially the same, resulting in a decrease of the value of P/C. Therefore, the fluorapatite ring outside of the oolitic structure was reduced and became gradually thin. At 45 min, P disappeared, while there were only Ca existed in the fluorapatite ring, resulting in about zero of the vale of P/Ca. The disappearance of P is caused by the fluorapatite decomposition and dephosphorization, ultimately, P discharged form oolitic structure in the form of P₂, and Ca element existed in the form of CaO. In this oolitic structure, the fluorapatite ring is adjacent to chlorite ring, SiO₂ decomposed from chlorite could promote the decomposition of fluorapatite to form Ca₃(PO₄)₂, which is easily reduced using carbon to form CaO and P₂. CaO and SiO₂ combined to form CaSiO₃, which stable the Si element in flourapatite ring. The reaction process is as follows.²¹⁾   

$$\mathrm{C}{\mathrm{a}}_{10}{(\mathrm{P}{\mathrm{O}}_4)}_6{\mathrm{F}}_2(\mathrm{s})\stackrel{\mathrm{S}\mathrm{i}{\mathrm{O}}_2}{⟶}3\mathrm{C}{\mathrm{a}}_3{(\mathrm{P}{\mathrm{O}}_4)}_2(\mathrm{s})+\mathrm{C}\mathrm{a}{\mathrm{F}}_2(\mathrm{s})$$(4)

  

$${\text{Ca}}_3{{\text{(PO}}_4\text{)}}_2\text{(s)}+\text{5C(s)}=\text{3CaO(s)}+\text{5CO(g)}+{\text{P}}_2\text{(g)}$$(5)

  

$$\text{CaO(s)}+{\text{SiO}}_2\text{(s)}={\text{CaSiO}}_3\text{(s)}$$(6)

Fig. 8.

Dephosphorization of fluorapatite in reduction process.

At 55 min, the Ca and P outside of the fuorapatite ring were decreased at the same time, which indicated that CaSiO₃ gradually disappeared due to the diffusion of CaSiO₃, resulting in complete disappearance of the fluorapatite. After 75 min, the internal fluorapatite ring also gradually reacted in this way.

The scanning results of SEM-EDS on I, II, III and IV in Fig. 9 as shown in Table 1. The composition of each point is inferred from the number of atoms in the four points. Figure 9-IV shows the grain morphology of fluorapatite and FeO in fluorapatite ring at 60 min. Then, the fluorapatitering disappeared through the above reaction. However, the further observation indicated that many fluorapatite grains interspersed in iron oxide grains near fluorapatie ring (Fig. 9(E)), which resulted that metallic iron was difficult to separate with phosphorus.

Fig. 9.

Crystallization and diffusion of fluorapatite in direct reduction process.

Table 1. The atomic percentage content of each point.

	Fe (at%)	Ca (at%)	F (at%)	P (at%)	O (at%)
I	46.48	–	–	–	50.56
II	–	24.77	5.25	13.01	56.97
III	47.01	–	–	–	49.28
IV	–	24.83	6.59	12.99	55.59

Thus, the behavior of flourapatite in carbothermic reduction process has been clarified. A part of fluorapatite closed to chlorite was reduced; another part of fluorapatite grains which was not contact with SiO₂ migrated and diffused into iron oxide. Both of them closed the connection between phosphorus and iron, which made it difficult to dephosphorization in carbothermic reduction.

3.2.5. Transformation of Gangue Grains

Due to the high SiO₂ and Al₂O₃ content decomposition from chlorite in high phosphorus oolitic hematite, they would combine with FeO to generate Fe₂SiO₄ and FeAl₂O₄, which hinder the reduction of FeO.

When the grains which had stored adequate energy were contacted, the gangue transformation reaction would happen in contact surface. Figures 10(G) and 10(I) shows that FeAl₂O₄ was formed, when FeO was contacted with Al₂O₃. Because of the high melting point, the FeAl₂O₄ grains remained intact. The formation of FeAl₂O₄ grains fixed Fe²⁺, which hindered the reduction of Fe²⁺. Equations (7) and (8) show two ways about the formation of FeAl₂O₄ grains.   

$$\mathrm{F}\mathrm{e}\mathrm{O}(\mathrm{s})+\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3(\mathrm{s})=\mathrm{F}\mathrm{e}\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_4(\mathrm{s})$$(7)

  

$$\mathrm{F}{\mathrm{e}}_2\mathrm{S}\mathrm{i}{\mathrm{O}}_4(\mathrm{s})+2\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3(\mathrm{s})=2\mathrm{F}\mathrm{e}\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_4(\mathrm{s})+\mathrm{S}\mathrm{i}{\mathrm{O}}_2(\mathrm{s})$$(8)

Fig. 10.

The shape of fayalite and hercynite formed by gangue phase in direct reduction process.

The scanning results of SEM-EDS on a, b, c, d, e, f, g, h and i in Fig. 10 as shown in Table 2. Figure 10(F) shows that FeO and SiO₂ generated a large amount of Fe₂SiO₄, which also fixed Fe²⁺, then, it combined with SiO₂ to form low-melting eutectic mixture FeSiO₃. The molten eutectic mixture flow along grain boundaries and spread gradually (Fig. 10(H)), which improve the reaction dynamics. As a result, unreacted Al₂O₃ contacted with Fe₂SiO₄ to form SiO₂ and FeAl₂O₄. The SiO₂ continued to react. This material is believed to consist of fayalite (Fe₂SiO₄) and hercynite (FeAl₂O₄) which are likely to participate in the process of mass transfer between mineral grains.   

$$2\mathrm{F}\mathrm{e}\mathrm{O}(\mathrm{s})+\mathrm{S}\mathrm{i}{\mathrm{O}}_2(\mathrm{s})=\mathrm{F}{\mathrm{e}}_2\mathrm{S}\mathrm{i}{\mathrm{O}}_4(\mathrm{s})$$(9)

  

$$\mathrm{F}{\mathrm{e}}_2\mathrm{S}\mathrm{i}{\mathrm{O}}_4(\mathrm{s})+\mathrm{S}\mathrm{i}{\mathrm{O}}_2(\mathrm{s})=\mathrm{F}{\mathrm{e}}_2\mathrm{S}\mathrm{i}{\mathrm{O}}_4\cdot \mathrm{S}\mathrm{i}{\mathrm{O}}_2(\mathrm{s})$$(10)

Table 2. The atomic percentage content of each point.

	Fe (at%)	O (at%)	Si (at%)	Al (at%)
a	19.57	56.39	16.88	–
b	47.87	50.08	–	–
c	13.45	63.45	–	23.06
d	47.95	50.46	–	–
e	19.21	56.27	17.01	–
f	19.53	56.33	16.97	–
g	–	65.37	32.04
h	13.53	63.12	–	23.15
i	95.54	–	–	–

4. Conclusions

(1) Oolitic structure change was linked to mineral reactions. Along with the reduction of iron oxide, decomposition of chlorite and reduction of fluorapatite proceeded, resulting in gradual disappearance of oolitic structure.

(2) The reduction of iron oxide follows a series of stepwise reaction. When FeO was formed, it would combined with SiO₂ and Al₂O₃, which hindered the reduction of FeO, eventually, metallic iron would be got from reduction of FeO, FeAl₂O₄ and Fe₂SiO₄.

(3) Chlorite firstly dehydrated at low temperature, then, decomposed into SiO₂ and Al₂O₃. They can not only combine with FeO to form Fe₂SiO₄ and FeAl₂O₄, which hindered the reduction of iron oxide, but also promoted the decomposition of fluorapatite to form Ca₃(PO₄)₂ and CaF₂ in the presence of SiO₂.

(4) A large amount of fluorapatite was reduced before the appearance of metallic iron, and P element discharged from pellets as the form of P₂. However, a small amount of fluorapatite was not reduced and diffused into iron oxides when it was not contacted with gangues. This results in a difficulty of separation by mechanical crushing and screening due to a close connection of grains.

Acknowledgments

Project was funded by the National Natural Science Foundation of China (51374024), Fundamental Research Funds for the Central Universities (FRF-TP-16-019A1) and China Postdoctoral Science Foundation (2016M600919).

References

• 1)   M.  Franceschelli,  M.  Puxeddu and  M.  Carta: Mineral. Petrol., 69 (2000), 267.
• 2)   G. H.  Ai,  X. B.  Li and  Y.  Zhou: Nonferr. Met. Sci. Eng., 2 (2011), 53.
• 3)   E.  Matinde and  M.  Hino: ISIJ Int., 51 (2011), 220.
• 4)   E.  Matinde and  M.  Hino: ISIJ Int., 51 (2011), 544.
• 5)   J. Q.  Yin and  X. W.  Lv: ISIJ Int., 52 (2012), 1579.
• 6)   Y. S.  Sun,  Y. X.  Han and  P.  Gao: Int. J. Miner. Metall. Mater., 20 (2013), 411.
• 7)   D. W.  Yang,  T. C.  Sun and  H. F.  Yang: J. Univ. Sci. Technol. Beijing, Miner. Metall. Mater., 32 (2010), 968.
• 8)   S.  Weissberger,  Y.  Zimmels and  I. J.  Lin: Metall. Mater. Trans. B, 17B (1986), 433.
• 9)   S.  Nadiv,  S.  Weissberger,  I. J.  Lin and  Y.  Zimmels: J. Mater. Sci., 23 (1988), 1050.
• 10)   X. Y.  Ning,  Q. G.  Xue,  G.  Wang,  X. F.  She and  J. S.  Wang: Ironmaking Steelmaking, 41 (2014), 583.
• 11)   J.  Liu: M. S. thesis, University of Science and Technology Beijing, Beijing, (2012), (in Chinese).
• 12)   S. X.  Song,  E.  Fabian,  C.  Toro and  Y. M.  Zhang: Int. J. Miner. Metall. Mater., 20 (2013), 113.
• 13)   Z. D.  Guan: Physical Properties of Inorganic Materials, Tsinghua University press, Beijing, (1992), 3.
• 14)   S. L.  Zheng: Handbook of Surfactant Materials, Chemical Industry Press, Beijing, (2007), 43.
• 15)   G. S.  Liu,  V.  Strezov,  J. A.  Lucas and  L. J.  Wibberley: Thermochim. Acta, 410 (2004), 133.
• 16)   Q.  Wang,  J.  Tian,  Q.  Li and  J.  Sun: Ironmaking Steelmaking, 24 (2004), 457.
• 17)   R. J.  Fruehan: Metall. Trans. B, 8B (1977), 279.
• 18)   M.  Bahgat and  M. H.  Khedr: Mater. Sci. Eng. B, 138 (2007), 251.
• 19)   S. K.  Durra and  A.  Ghosh: Metall. Mater. Trans. B, 25B (1994), 15.
• 20)   C.  Cheng,  Q. G.  Xue,  G.  Wang,  Y. Y.  Zhang and  J. S.  Wang: Metall. Mater. Trans. B, 47B (2016), 154.
• 21)   J.  Mu,  F.  Leder,  W. C.  Park,  R. A.  Hard,  J.  Megy and  H.  Reiss: Metall. Mater. Trans. B, 17B (1986), 861, 869.