Effect of MgO and CaCO₃ as Additives on the Reduction Roasting and Magnetic Separation of Beach Titanomagnetite Concentrate

Abstract

To understand the effect of MgO as additive on the reduction roasting of beach titanomagnetite concentrate, the phase transformations, metallization degree and Fe–Mg–O phase diagram were studied in this paper. Results indicated that with the increasing of MgO dosage, more Mg²⁺ diffused into the magnetite lattice, replaced some of the Fe²⁺, and formed a large of MgFe₂O₄. So adding MgO not only did not effectively improve iron recovery, but baffled the reduction of iron oxides. Besides, the effect of CaCO₃ on the growth of iron particles was also studied by optical microscope and Qwin image analysis software. Tests revealed that an appropriate CaCO₃ dosage can facilitate the growth of iron particles. Because CaCO₃ promoted the reaction of iron oxides, FeO content in the slag decreased and more nuclei of iron crystal was formed, resulting in the growth of iron particles. However, excessive CaCO₃ increased the slag melting point, which hindered the diffusion and growth of iron particles. At last, the best product indexes were obtained by magnetic separation when CaCO₃ dosage was 4%.

1. Introduction

As sources of high-grade titanium minerals (e.g. rutile and ilmenite) are being exhausted worldwide, more attentions have been drawn to processes of dealing with low-grade minerals as alternative resources.¹⁾ Beach titanomagnetite (TTM), which contains the valuable elements such as iron and titanium, has a wide distribution throughout the world, mainly concentrated in Indonesia, China, New Zealand, etc.^2,3) Besides, it has some advantages of high potential value, easy mining, and low exploitation costs. However, the conventional smelting route, which is used to smelt TTM concentrate with carbon in blast furnace to produce pig iron and titanium slag, has many disadvantages. Firstly, the smelting process always requires lots of coke and a high temperature.^4,5) Secondly, the produced titanium slag only contains 20–25% TiO₂, which is difficult to recover commercially.^6,7)

In recent years, most of the studies focus on developing an alternative route to use beach TTM, including direct reduction–electric furnace smelting, direct reduction–magnetic separation, etc.^8,9,10) One of these processes is the direct reduction–electric furnace melting process, which has been used in South Africa and New Zealand. However, this process involved adjustment of the basicity of the slag to an appropriate level to improve the separation behavior of iron and slag, resulting in the content of TiO₂ in the titanium slag being lowered to only 30–33%.¹¹⁾ Hence, the TiO₂ could still not be recovered effectively. Gao et al.¹²⁾ reported on recovering iron and titanium from beach TTM by coal-based reduction followed by magnetic separation. Although a high grade of direct reduction iron powder (DRIP) was obtained, assay of the non-magnetic products showed only 25–30% TiO₂. This may be because a high dosage of reductant was added, and coal ash diluted the concentration of titanium dioxide, which will bring some difficulties to the subsequent processes of extracting titania.¹⁰⁾

In order to promote the enrichment of iron and titanium component, researchers proposed a reduction roasting process by adding external coal. Geng et al.^10,13) investigated this reduction process and found non-magnetic products may contain 35–45% TiO₂. Hence, this process seems to be a promising method to recover titanium. However, there are still some problems to solve, such as low porosity and easily the appearance of iron-joined crystals. Besides, Some Fe impurities are dissolved in (Fe, Mg)Ti₂O₅ solid solution, which lead to relatively low metallization degree and iron recovery. In reduction roasting process, additives have often been used, such as CaF₂,^14,15) CaCO₃,¹⁶⁾ NaSO₄,^10,17) MgO¹⁸⁾ and NaCO₃.¹⁹⁾ These additives may to some extent improve reduction characteristics or facilitate the growth of iron particles. Wang et al.²⁰⁾ examined the interaction between ironsand and flux materials (MgO, CaO and etc.), and found that Mg²⁺ ions significantly diffused into the lattice of ironsand, which hindered the assimilation of ironsand particles. Tang et al.²¹⁾ reported MgO can improve the basicity and other properties viz. strength and physico-chemical properties, especially for lowering reduction degradation index. Though MgO as additive are widely used in sinters and pellet, information on the effect of MgO on the reduction roasting of beach TTM is still limited. Therefore, to understand the effect of MgO on the reduction roasting, the phase transformations, metallization degree and Fe–Mg–O phase diagram were primarily studied in this study. Besides, the literature has very few descriptions of size of iron particles, and the effect of addictive on the growth of iron particles has not yet been clearly studied. This paper, therefore, attempts to use optical microscope and Qwin image analysis software to measure the size of iron particles with different dosages of CaCO₃. At last, the optimal product indexes and CaCO₃ dosage were obtained by magnetic separation.

2. Materials and Methods

2.1. Materials

The beach TTM used in this study was obtained from the Indonesian coast. After drying, grinding and beneficiation, the mineralogical analysis of the TTM concentrate was investigated by x-ray diffraction (XRD). The result indicated that the main crystalline phases were TTM (Fe_2.75Ti_0.25O₄), with small amounts of ilmenite (FeTiO₃) and quartz (SiO₂). The chemical compositions of the TTM concentrate are presented in Table 1. Bituminite was used as reducing agent with a particle size <1 mm. The proximate analysis results of bituminite were composed of 7.25 mass% moisture, 6.55 mass% ash, 29.54 mass% volatiles and 56.66 mass% fixed carbon. MgO and CaCO₃ used as additives were analytical reagent (AR) grade. Besides, sodium carboxymethylcellulose (CMC) which is good adhesion properties was used as organic binder.

Table 1. Chemical compositions of beach TTM concentrate.

Compositions	TFe	FeO	TiO2	SiO2	Al2O3	MgO	CaO	MnO	S	P
Content wt.%	57.29	29.79	11.42	3.06	2.90	2.73	0.27	0.35	0.09	0.05

2.2. Experimental Procedures

The TTM concentrate, CMC (1 mass%) and additive were completely mixed. The obtained mixtures were pelletized in an experimental disc pelletizer so as to produce green pellets with mass of 4 g and height of about 12 mm. The green pellets were dried at 105°C for 4 h in a drying cabinet.

Reduction experiments were performed in a muffle furnace. The pellets of about 25 g were embedded in reductant (C/Fe molar ratio of 1.4), which was filled into in a graphite crucible with a lid, and then the crucible was placed inside the furnace. The crucible charging method is shown in Fig. 1. The furnace was heated from room temperature to 1200°C at a heating rate of 10°C/min. After an adequate holding time (90 min), the crucible was taken out from the furnace and cooled to ambient temperature. Metallization degree (R_m) of reduced pellets was calculated as follows:   

$${\text{R}}_{\text{m}}=\frac{{\omega }_{MF\text{e}}}{{\omega }_{TF\text{e}}}*100\%$$

where: ω_TFe is mass percent of total iron in the reduced sample, mass%; and ω_MFe is that of metallic iron in the reduced sample, mass%.

Fig. 1.

Crucible charging method.

The size of iron particles in the reduced pellets was measured using optical image analysis. The reduced pellets were mounted in epoxy resin and then polished. Afterwards, viewing areas distributed in the left, right, top, bottom and middle of samples, were observed and photographed using an optical microscope with 20x eyepiece and 10x object lens. The total number of each sample was about 100 photos. Then the captured images were processed with the help of Qwin image analysis software. The size (i.e. length, width, area) and number of iron particles were measured automatically. In the end, the data were exported to an Excel for further analysis. In reduction roasting process, the shapes of iron particles are usually long ribbon, so the mean size of iron particles is described by length of them, and it is defined by the following equation:   

$$\overline{L}=\frac{\sum _{\text{i}=1}^{\text{N}}{\text{L}}_i}{N}$$

where: L is the mean size of iron particles; N is the total number of iron particles; L_i is the length of each iron particles.

The other reduced pellets were crushed and ground. The particle size of the reduced samples was less than 81.52 mass% −200 mesh (74 μm), and they were magnetically separated by a magnetic tube with magnetic field intensities of 1800 Oe. The iron concentrates obtained from the first separation were reground until the grinding fineness 83.56 mass% −325 mesh (45 μm), and they were magnetically separated with magnetic field intensities of 1250 Oe. The final magnetic products were designated as DRIP and non-magnetic product, respectively.

2.3. Analysis and Characterization

The particle size of iron particles was analyzed by Leica-DM4500P Optical microscope and Qwin image analysis software. The determination of metallic iron (MFe) content was performed using the potassium dichromate volumetric method. The chemical compositions of the sample and magnetic products were examined by neutralization titration and ICP-AES (Optima 5300DV, PerkinElmer, USA). The morphological changes were observed by Scanning electron microscopy (SEM) with energy dispersive spectrum (EDS) (Carl Zeiss EVO18) analysis. Phase analyses were identified by Rigaku DMAX-RB X-ray Diffract Meter (Cu target).

3. Results and Discussion

3.1. Effect of MgO as Additive

In the presence of different dosages of MgO, the phase transformations of reduced pellets were identified by XRD. As shown in Fig. 2, the main phases without adding MgO are metallic iron and (Fe, Mg)Ti₂O₅. When MgO dosage increases from 0% to 2%, no new phase is observed. The diffraction peaks of MgFe₂O₄ begin to appear when adding 4% MgO. When MgO dosage continues to increase, the peak intensities of MgFe₂O₄ gradually increase, whereas those of (Fe, Mg)Ti₂O₅ and metallic iron become decrease. Beside, some MgTiO₃ phases are detected. These results inferred that adding 4–8% MgO may inhibit the reduction of iron oxides. To further investigate the effects of MgO on the reduction degree of TTM, the metallization degree were calculated.

Fig. 2.

XRD patterns of reduced pellets with different dosages of MgO.

As shown in Fig. 3, it is evident that adding MgO has a detrimental effect on the metallization degree. With increasing the dosage of MgO, the metallization degree decreases. However, metallization degree and the size of iron particles should be considered and controlled seriously during the reduction process, because they both have significant effects on the magnetic separation. A high rate of metallization is essential if people want to get high-grade product index. Therefore, adding MgO didn’t achieve the expect result, which is that metallization degree improved by adding MgO. In order to further explore the mechanism, the binary phase diagram of MgO–Fe₃O₄ and MgO–FeO, which was drawn using the Factsage 7.2 software, are shown in Fig. 4.

Fig. 3.

Metallization degree of reduced pellets with different dosages of MgO.

Fig. 4.

Binary phase diagram of MgO–Fe₃O₄ and MgO–FeO. (Online version in color.)

It is well known that TTM are represented as a series of solid solutions between magnetite (Fe₃O₄) and ulvospinel (Fe₂TiO₄) as follows:²²⁾   

$${({\text{Fe}}_{\text{3}}{\text{O}}_{\text{4}})}_{\text{1}-\text{x}}{({\text{Fe}}_{\text{2}}{\text{TiO}}_{\text{4}})}_{\text{x}}={(\text{FeO}{\text{.Fe}}_{\text{2}}{\text{O}}_{\text{3}})}_{\text{1}-\text{x}}(\text{2FeO}{\text{.TiO}}_{\text{2}})$$

So the reaction of TTM should be divided into two processes which occurred simultaneously, namely the transitions from magnetite to metallic iron and from ulvospinel to titanium-containing phase. Figure 4(a) indicate that because of the similarity of the ionic radii of Fe²⁺ and Mg²⁺ (ionic radius of Mg²⁺ is 0.72 Å and of Fe²⁺ is 0.77 Å), most of the Mg²⁺ can easily enter the magnetite lattice to form MgFe₂O₄ by displacing Fe²⁺ ions. Meanwhile, as show in Fig. 4(b), it can be found that Mg²⁺ and Fe²⁺ can replace each other to form a completely continuous (Fe, Mg)O solid solution. So in this reduction process, with the increasing of MgO dosage, more Mg²⁺ diffused into the magnetite lattice, replaced some of the Fe²⁺, and formed MgFe₂O₄. The process of their migration may be as follow: Fe₃O₄→(Fe, Mg)O.Fe₂O₃→(Mg, Fe)O.Fe₂O₃→MgFe₂O₄. Similarly, high-temperature reduction from Fe²⁺ in ulvospinel to Fe can produce vacancies in crystal, which was allowed to increase diffusion of Mg²⁺ into these regions. This also explains why Mg component was observed in the titanium-containing phase without adding MgO. With the increasing of MgO dosage, finally, the main titanium-containing phases are (Fe, Mg)Ti₂O₅ and MgTiO₃.

In a word, these results suggest that adding MgO not only did not effectively improve iron recovery, but baffled the reduction of iron oxides. The net result may be that MgFe₂O₄ will fall into the magnetic product after magnetic separation and worsen the iron grade.

3.2. Effect of CaCO₃ as Additive

3.2.1. Thermomechanical Analysis

It is generally agreed that coal-based reduction occurs through the gaseous intermediates CO and CO₂.²³⁾ Given that reductant was placed outside pellets, the carbothermal reaction were mainly gas–solid reaction, i.e., via reactions (1) and (2).   

$$\text{TTM}+\text{CO(g)}=\text{reduced}\mathrm{\hspace{0.17em} } \text{TTM}+{\text{CO}}_{\text{2}}(\text{g})$$(1)

  

$$\text{C(s)}+{\text{CO}}_{\text{2}}(\text{g})=\text{2CO(g)}$$(2)

In general, the reduction of TTM proceeds the following path:^24,25)   

$$\begin{array}{l}{\text{Fe}}_{\text{3}-\text{x}}{\text{Ti}}_{\text{x}}{\text{O}}_{\text{4}}\to \text{FeO}+{\text{Fe}}_{\text{2}}{\text{TiO}}_{\text{4}}\to \text{Fe}+{\text{Fe}}_{\text{2}}{\text{TiO}}_{\text{4}}\\ \to \text{Fe}+{\text{FeTiO}}_{\text{3}}\to \text{Fe}+{\text{FeTi}}_{\text{2}}{\text{O}}_{\text{5}}\to \text{Fe}+{\text{TiO}}_{\text{2}}\end{array}$$

The reduction path includes the formation of intermediate phases, such as ulvospinel(Fe₂TiO₄), ilmenite(FeTiO₃) and ferrous-pseudobrookite(FeTi₂O₅).²⁴⁾ So the main reduction reactions of TTM are reactions (3), (4) and (5) without CaCO₃. Similarly, reactions (6), (7), (8) and (9) may occur in the presence of CaCO₃.   

$${\text{Fe}}_{\text{2}}{\text{TiO}}_{\text{4}}+\text{CO}=\text{Fe}+{\text{FeTiO}}_{\text{3}}+{\text{CO}}_{\text{2}}$$(3)

  

$${\text{2FeTiO}}_{\text{3}}+\text{CO}=\text{Fe}+{\text{FeTi}}_{\text{2}}{\text{O}}_{\text{5}}+{\text{CO}}_{\text{2}}$$(4)

  

$${\text{FeTi}}_{\text{2}}{\text{O}}_{\text{5}}+\text{CO}=\text{Fe}+{\text{2TiO}}_{\text{2}}+{\text{CO}}_{\text{2}}$$(5)

  

$${\text{Fe}}_{\text{2}}{\text{TiO}}_{\text{4}}+\text{CaO}+\text{2CO}={\text{CaTiO}}_{\text{3}}+\text{2Fe}+{\text{2CO}}_{\text{2}}$$(6)

  

$${\text{FeTiO}}_{\text{3}}+\text{CaO}+\text{CO}={\text{CaTiO}}_{\text{3}}+\text{Fe}+{\text{CO}}_{\text{2}}$$(7)

  

$${\text{FeTi}}_{\text{2}}{\text{O}}_{\text{5}}+\text{2CaO}+\text{CO}={\text{2CaTiO}}_{\text{3}}+\text{Fe}+{\text{CO}}_{\text{2}}$$(8)

  

$${\text{CaCO}}_{\text{3}}=\text{CaO}+{\text{CO}}_{\text{2}}$$(9)

According to reaction module of Factsage, the standard Gibbs free energy (ΔG^θ) of reaction (3) to (8) with increasing temperature were calculated and shown in Fig. 5. The ΔG^θ of reaction (3), (4) and (5) are almost positive value with entire range of reduction temperature, meaning that these reductions without CaCO₃ are difficult. The ΔG^θ of reaction (6), (7) and (8) constantly decrease as the temperature rises, and is far lower than the ΔG^θ of reaction (3), (4) and (5). The above thermodynamic analysis indicates that TTM is more easily reduced in the presence of CaCO₃, and the formation of CaTiO₃ more easily occurs with increasing temperature.

Fig. 5.

Standard Gibbs free energy (ΔG^θ) of Reactions (3) to (8). (Online version in color.)

3.2.2. Metallization Degree

As shown in Fig. 6, When CaCO₃ dosage increases from 0% to 8%, the metallization degree of reduced pellets rises from 93.86% to 97.70%. After the metallization degree reaches a maximum, it then decreases. As CaCO₃ dosage increases from 8% to 16%, the metallization degree decreases rapidly from 97.70% to 85.37%. The above results show that an appropriate dosage of CaCO₃ improved the reduction of TTM, whereas excessive CaCO₃ inhibited reduction of TTM. To fully explain the experimental data and further explore the mechanism, reduced pellets obtained were studied by XRD and SEM-EDS.

Fig. 6.

Metallization degree of reduced pellets with different dosages of CaCO₃.

3.2.3. X-ray Analysis and Morphology Observations

As shown in Fig. 7, metallic iron, (Fe, Mg)Ti₂O₅ are observed in the reduced products without any additives. With addition of 4% CaCO₃, the diffraction peaks of CaTiO₃ are observed. When CaCO₃ dosage increases from 4% to 8%, the intensity peaks of (Fe, Mg)Ti₂O₅ decrease, whereas intensity peaks of CaTiO₃ increase. This result indicates that adding CaCO₃ easily replace iron oxides from Fe₂TiO₄, FeTiO₃ and FeTi₂O₅ by reactions (6), (7) and (8) to generate CaTiO₃, thereby more FeO was reduced to metallic iron, which may help improve the iron recovery and reduced Fe impurity in Ti-bearing minerals.

Fig. 7.

XRD patterns of reduced pellets with different dosages of CaCO₃.

Figure 8(a) shows SEM image of reduced pellets in the absence of CaCO₃. The microstructure reveals three distinct regions which appear as bright, light gray and gray phases. The bright phase is metal iron. To identify the light gray and gray phase, EDS was performed at regions marked in the SEM image. The results of EDS analysis show that point 1 consists of approximately 7.14% Mg, 62.77% Ti, 24.88% O and 5.20% Fe. Combining with XRD results, it can be concluded that the light gray phases are mainly (Fe, Mg)Ti₂O₅. Point 2 in gray phase mainly consists of Si, Al, Fe, Mg and O. It suggests that a part of iron oxides reacted with gangue minerals which often contain SiO₂, Al₂O₃, MgO. When 8% CaCO₃ is added, Point 3 mainly consists of Ca, Ti and O, it can be seen that CaTiO₃ hardly contain Fe impurities. Figure 8(c) shows Ti region is closely combined with Ca region rather than Fe region. These results further suggest that Ca²⁺ can replace Fe²⁺ and Mg²⁺ impurities in titanium-containing phases, thus Fe²⁺ in the slag decreased and more iron phase was formed. Figure 8(d) shows the SEM images of the reduced pellet with 16% MgO. The results of EDS analysis show that point 4 consists of approximately Ca, Ti, O, Fe and Mg. Besides, a plenty of pores and some fine iron particles could be observed. These phenomena may be speculated that excessive CaCO₃ increased the slag melting point, which hindered the diffusion and growth of iron particles. Because CaO decomposed from CaCO₃ is a substance with a high melting point of 2572°C, CaO dissolves in slag, and naturally the amount of CaO combined with titanium oxidation to generate a large amount of perovskite (CaTiO₃, 1960°C), resulting in bad diffusion and mass transfer. In the end, excessive CaCO₃ inhibited reduction of TTM.

Fig. 8.

SEM images and EDS analysis of the reduced pellets with different dosages of CaCO₃. (a)-0 mass% CaCO3; (b)-8 mass% CaCO3; (c)-elemental maps for Fe, Ti, Mg, Ca of (b); (c)-16 mass% CaCO3; (e)-(h): EDS analysis. (Online version in color.)

Since having a significant influence on recovery of iron and titanium by magnetic separation, particle size of iron was analyzed by microscope. As showed in Fig. 9, the mean size of iron particles shows an increase first and then a decrease with increasing the dosage of CaCO₃. When CaCO₃ dosage increases from 0% to 4%, mean size rises from 29.1 μm to 39.6 μm. After mean size reaches a maximum, it decreases from 29.1 μm to 19.0 μm as CaCO₃ dosage increases from 4% to 16%. So it may be inferred that a suitable CaCO₃ dosage is 4% or 8%. In reduction roasting process, the reduction of TTM could be divided into three stages: chemical reaction stage, nucleation stage, and crystal growth stage.²⁶⁾ During the initial stage of reaction, some iron oxides were reduced to Fe and these Fe grain would act as the nuclei of iron crystal. In the next stage, new formed Fe tended to diffuse to the nuclei of iron crystal and then grew up unceasingly. In the end, plenty of large iron particles were produced. As mentioned above, the addition of CaCO₃ can decrease Gibbs free energy, and the presence of CaCO₃ can easily promote the reaction of iron oxides. So a larger number of FeO were reduced to nuclei of iron crystal, which is beneficial to facilitate the generation and growth of iron particles. However, excessive CaCO₃ increased the slag melting point, which hindered the diffusion and growth of iron particles.

Fig. 9.

Mean size of iron particles in reduced pellets with different dosages of CaCO₃. (Online version in color.)

3.3. Magnetic Separation of Reduced Pellets

To further verify the research results and obtain products, reduced pellets were separated by magnetic separation. Figure 10(a) shows that the iron grade of DRIP increases initially from 87.5% to 92.8% as CaCO₃ dosage increases from 0% to 4%. Afterward, the iron grade decreases from 92.8% to 86.8% when CaCO₃ dosage increases to 16%. The iron recovery of DRIP increases from 82.2% to 90.5% as CaCO₃ dosage increases from 0% to 8%, and then a sharp drop in iron recovery from 90.5% to 81.1% was observed when CaCO₃ dosage increases to 16%. As shown in Fig. 10(b), when CaCO₃ dosage increases from 0% to 4%, the TiO₂ grade increases from 37.8% to 42.3%, but when CaCO₃ dosage further increases, the TiO₂ grade gradually decreased. These results prove that an appropriate CaCO₃ dosage can improve the recovery of iron and titanium. When CaCO₃ dosage was 4%, DRIP with an iron grade of 92.8%, iron recovery of 89.9%, and TiO₂ grade of 42.3% in the non-magnetic product were obtained.

Fig. 10.

Effect of CaCO₃ on the recovery of Fe and Ti. (a) DRIP; (b) non-magnetic product. (Online version in color.)

4. Conclusions

(1) Adding MgO not only did not effectively improve iron recovery, but baffled the reduction of iron oxides. With the increasing of MgO dosage, more Mg²⁺ diffused into the magnetite lattice, replaced some of the Fe²⁺, and formed a large of MgFe₂O₄. In the end, adding MgO baffled the reduction of iron oxides. Besides, Mg²⁺ can also diffuse into titanium-containing phase, finally, the main titanium-containing phases are (Fe, Mg)Ti₂O₅ and MgTiO₃.

(2) An appropriate CaCO₃ dosage can facilitate the growth of iron particles. The presence of CaCO₃ can easily promote the reaction of iron oxide, so more FeO were reduced to Fe and these Fe grain acted as the nuclei of iron crystal, which facilitated the generation and growth of iron particles. However, excessive CaCO₃ increased the slag melting point, which hindered the diffusion and growth of iron particles.

(3) Relatively high grades of DRIP and non-magnetic product were obtained by reduction roasting followed with magnetic separation. When CaCO₃ dosage was 4%, DRIP with an iron grade of 92.8%, iron recovery of 89.9%, and TiO₂ grade of 42.3% in the non-magnetic product were obtained.

Acknowledgement

The authors wish to express their thanks to the Natural Science Foundation of China (No. 51474018 and 51674018) for the finance support for this research.

References

• 1)   B. C.  Jena,  W.  Dresler and  I. G.  Reilly: Miner. Eng., 8 (1995), 159.
• 2)   T. Y.  Hu,  T. C.  Sun,  J.  Kou,  C.  Geng,  X. P.  Wang and  C.  Chen: Int. J. Miner. Process., 165 (2017), 28.
• 3)   L. Z.  Chen,  S. M.  Wen,  G. D.  Xu and  H. Y.  Xie: Miner. Process. Extr. Metall. Rev., 34 (2013), 139.
• 4)   J.  Dang,  G. H.  Zhang and  K. C.  Chou: J. Alloy. Compd., 619 (2015), 443.
• 5)   F.  Pan,  Z.  Du,  M. J.  Zhang and  H. Y.  Sun: ISIJ Int., 57 (2017), 413.
• 6)   D. S.  Chen,  B.  Song,  L. N.  Wang,  T.  Qi,  Y.  Wang and  W. J.  Wang: Miner. Eng., 24 (2011), 864.
• 7)   T.  Hu,  X. W.  Lv,  C. G.  Bai,  Z. G.  Lun and  G. B.  Qiu: Metall. Mater. Trans. B, 44 (2013), 252.
• 8)   J.  Tang,  M. S.  Chu,  C.  Feng,  Y. T.  Tang and  Z. G.  Liu: ISIJ Int., 56 (2016), 210.
• 9)   Y. L.  Huang and  Y. Y.  Zhu: Corros. Sci., 47 (2005), 1545.
• 10)   C.  Geng,  T. C.  Sun,  H. F.  Yang,  Y. W.  Ma,  E. X.  Gao and  C. Y.  Xu: ISIJ Int., 55 (2015), 2543.
• 11)   C.  Lv,  K.  Yang,  S. M.  Wen,  S. J.  Bai and  Q. C.  Feng: JOM, 69 (2017), 1801.
• 12)   E. X.  Gao,  T. C.  Sun,  C. Y.  Xu,  Z. G.  Liu,  Z. Z.  Liu and  C. X.  Yu: Met. Mine, 449 (2013), 46.
• 13)   C.  Geng,  T. C.  Sun,  Y. W.  Ma,  C. Y.  Xu and  H. F.  Yang: J. Iron Steel Res. Int., 24 (2017), 156.
• 14)   Q.  Wang,  Y. Y.  Ge,  J. L.  Kuang,  P.  Jiang,  W. X.  Liu and  W. B.  Cao: J. Alloy. Compd., 696 (2017), 220.
• 15)   T. Y.  Hu,  T. C.  Sun,  J.  Kou,  C.  Geng and  Y. Q.  Zhao: Int. J. Miner. Metall. Mater., 24 (2017), 1201.
• 16)   S. M.  Jung: Metall. Mater. Trans. B, 46 (2015), 1162.
• 17)   H. Y.  Zhao,  Q.  Song,  S. C.  Liu,  Y. H.  Li,  X. H.  Wang and  X. Q.  Shu: Energy Convers. Manag., 161 (2018), 13.
• 18)   Z. G.  Liu,  M. S.  Chu,  H. T.  Wang,  W.  Zhao and  X. X.  Xue: Int. J. Miner. Metall. Mater., 23 (2016), 25.
• 19)   W.  Yu,  T. C.  Sun,  J.  Kou,  Y. X.  Wei,  C. Y.  Xu and  Z. Z.  Liu: ISIJ Int., 53 (2013), 427.
• 20)   Z.  Wang,  D.  Pinson,  S.  Chew,  H.  Rogers,  B. J.  Monaghan and  G. Q.  Zhang: ISIJ Int., 56 (2016), 1315.
• 21)   J.  Tang,  M. S.  Chu and  X. X.  Xue: Int. J. Miner. Metall. Mater., 22 (2015), 371.
• 22)   S. M.  Jung: ISIJ Int., 56 (2016), 1315.
• 23)   Y. Q.  Zhao,  T. C.  Sun,  H. Y.  Zhao,  X. H.  Li and  X. P.  Wang: ISIJ Int., 58 (2018), 1768.
• 24)   E.  Park and  O.  Oleg: ISIJ Int., 44 (2004), 999.
• 25)   F.  Pan,  Z.  Du,  M. J.  Zhang and  H. Y.  Sun: ISIJ Int., 57 (2017), 413.
• 26)   Y. S.  Sun,  P.  Gao,  Y. X.  Han and  D. Z.  Ren: Ind. Eng. Chem. Res., 52 (2013), 2323.