Solid State Reduction of Titanomagnetite Concentrate by Graphite

Abstract

Titanomagnetite concentrate was reduced by graphite isothermally under argon gas in thermogravimetry system at 1373 to 1623 K. The influences of reductive conditions on the reduction and metallization degree including reduction temperature, reduction time and C/O molar ratio were studied. And the characteristics of reduced samples were analyzed by XRD, BES and EDS. Results shown that in the temperatures range of 1423 to 1623 K, initially the reduction proceeded rapidly and after 30 min only a slow increase in reduction was observed. The low reduction degree was owing to the high impurities oxides content such as magnesium oxide in titanomagnetite concentrate and the reduction and metallization degree increased with the C/O molar ratio until up to 1.0 at 1623 K. Above 1373 K, the reduction path is suggested as follow:

  

1. Introduction

The demand for titanium dioxides which are widely used in paint, paper and plastics industries is increasing rapidly.¹⁾ At present, ilmenite (40–65% TiO₂) is the main sources for production of metallic titanium and titanium containing compounds.²⁾ Since the sources of high-grade titania-ferrous ores are decreasing in the world, the utilization of low-grade minerals such as titanomagnetite (TTM, 10–16% TiO₂) has attracted more attention.^3,4) China has the largest titaniaferrous ores reserve about 2×10⁸ t (in TiO₂).⁵⁾ Panzhihua titanomagnetite accounts for more than 90% titanium reserves in China and more than 35% around the world.⁶⁾ At present the TTM concentrate is mainly smelted in the blast furnace (BF) to make iron. In BF process, most of titanium components in the ore are concentrated into molten slag (22–25% TiO₂).⁷⁾ Due to the scattered distribution of titanium components in various fine grained mineral phases (<10µm) with complex interfacial combination, it is difficult to recover the titanium components and metallic iron through traditional separation processes.⁸⁾

In the past decades, a number of researches have been conducted to improve the utilization of titania-ferrous, including smelting,⁹⁾ direct acid leaching,¹⁰⁾ selective chlorination¹¹⁾ and reduction.¹²⁾ Among these, the direct reduced iron (DRI) process is supposed to be a more practical and effective way. Recent years, the Rotary Hearth Furnace (RHF) directed reduction process, a new DRI process which takes advantages of low requirement of raw material performance, high temperature (∼1623 K) and rapid reduction (∼30 min), attracts much attention.¹³⁾ And a relative route, RHF directed reduction - electric arc furnace (EAF) melting separation method, has been proposed for refining both iron and Ti slag from the TTM concentrate.³⁾ In this process, TTM concentrate is first reduced by pulverized coal in RHF and then smelted in EAF. Although the RHF-EAF method is feasible in production, the reductive kinetics of reaction at high temperature between TTM concentrate and carbon are still unclear with little research. The aim of this work is to investigate the isothermal reduction behaviors of TTM concentrate with carbon reductant. The influences of reductive conditions on the reduction and metallization degree of TTM concentrate such as reduction temperature, reduction time and C/O molar ratio were studied. Also the characteristics of reduced samples, reductive kinetics and reduction path were analyzed.

2. Experimental

2.1. Materials

The TTM concentrate used in this study was obtained from Panzhihua, in Sichuan province of China. After drying and grinding, the particle size of concentrate was 5.03–64.79 µm. Chemical composition of concentrate is listed in Table 1. Phase characteristic of sample was investigated by XRD. The result in Fig. 1 indicates that the main crystalline phases of concentrate are TTM (Fe_3–xTi_xO₄, with x=0.27±0.02, or 3(Fe₃O₄)·Fe₂TiO₄) and ilmenite (FeTiO₃ or FeO·TiO₂). The graphite powder (99.9% fixed carbon) in the size range of 19.66-91.09 µm was used as reductant. The particle size distributions of TTM concentrate and graphite powder are shown in Figs. 2 and 3 using LMS-30 laser particle size analyzer.

Table 1. Chemical composition of titanomagnetite concentrate (wt.%).

TFe	FeO	Fe2O3	TiO2	SiO2	MgO	Al2O3	V2O5	CaO	MnO
54.54	32.16	42.18	10.77	3.81	3.72	3.54	0.67	0.39	0.40

Fig. 1.

XRD pattern of titanomagnetite concentrate.

Fig. 2.

The particle size distributions of TTM concentrate.

Fig. 3.

The particle size distributions of graphite powder.

2.2. Experimental Procedures

The TTM concentrate and graphite were thoroughly mixed by stirring over 30 min. The addition amount of graphite is according to the molar ratio of carbon to oxygen corresponding to Fe₂O₃ and FeO in the TTM concentrate. Then the mixtures were pressed in a closed die of 9 mm in diameter under 15 MPa to produce cylindrical pellets with mass of about 2 g and height of 8 mm. The reduction experiments were conducted in a vertical tubal electric furnace with Pt-Rh thermocouples and Si–Mo heating elements. Argon gas with a flow rate 5 L/min was purged into the tube for providing inert ambience. When the furnace temperature reached the reduction temperature and became stable, the pellets put in a pre-heated alumina crucible were lowered into the hot zone of the furnace starting the reduction reaction. During the reduction process, the weight changes of reacting samples were monitored by electronic balance (Mettler-Toledo AL104) and recorded in real-time by computer. When reduction experiments were finished, samples were withdrawn from the furnace and quenched in water for analysis. The mineralogical morphology of sample was examined by XRD (Nikkaku D/max-RB, using Cu Kα), BES (JSM-6480LV) and EDS (Noran System six).

3. Results and Discussion

In analyzing the results, an almost direct correlation between total mass loss and oxygen loss can be assumed since the gas produced during the reduction showed it to consist of CO more than 99%.¹⁴⁾ Therefore the experimental results were presented in terms of the ratio of mass loss against reduction time. The reduction degree (w) is defined as   

$$w=\frac{w_0-w_m}{w_t}\times 100\%$$(1)

where w₀ is the starting mass of sample after removal of moisture, w_m is the mass of sample after reduction time t, and w_t is the maximum possible mass loss contributed by total carbon and oxygen from iron oxides in pellet sample with the molar ratio of carbon to oxygen 1.0 (C/O=1.0).

The metallization degree (η) of each sample was calculated by following formula:   

$$\eta =(MFe/TFe)\times 100\%$$(2)

Where MFe is the weight of metallic iron after reduction, TFe is the weight of total iron after reduction. The total iron content was analyzed by ICP-AES and metallic iron was analyzed by chemical method. The porous samples after reduction were analyzed and the reductive kinetics were discussed in detail as follows.

3.1. Effect of Temperature

Plots of the reduction and final metallization degree of TTM concentrate reduction by graphite with C/O=1.0 at different temperatures as a function of time are shown in Figs. 4 and 6 (Curve A). From figures, it can be seen that temperature obviously influences the reduction and metallization degree. With temperature increasing, the reduction and metallization degree increases. At 1373 K, the reduction degree gradually increases over reduction time. But at higher temperatures from 1423 K to 1623 K, initially the reduction proceeds rapidly and after 30 min only a slow increase in reduction is observed. It indicates that the significant fast reduction occurs at temperatures above 1423 K, which is consistent with the experimental results from Wang¹⁵⁾ and Chen.¹⁶⁾ Compared with the studies of Wang¹⁵⁾ and Chen,¹⁶⁾ the reduction degree of TTM concentrate is lower, especially at the low temperature at 1373 to 1473 K, owing to the high content of impurities oxides including magnesium, manganese and aluminum oxides. This phenomenon can be explained by the barrier effect.¹⁷⁾ It is known from thermodynamic data that during the reduction of TTM concentrate, it is neither possible to reduce magnesium in the TTM to magnesium oxide or magnesium metal nor could magnesium pass into solid solution in titanium oxide.¹⁸⁾ During the reduction of TTM concentrate, the magnesium concentration ahead of the reduction interface increased in TTM concentrate by direct reduction of Fe²⁺. This lowered the thermodynamic activity of the Fe²⁺, making its reduction progressively more difficult. Eventually, the magnesium concentration became so high, and the iron activity was so low that the reduction of Fe²⁺ to metallic iron almost stopped. Similarly, manganese and aluminum oxides have the same effect on the reduction kinetics as magnesium oxide since the impurities oxides content of the briquette increases the barrier effect.¹⁹,²⁰⁾ However, magnesium oxide has somewhat larger effect on the reduction kinetics than manganese oxide. This might be due to the fact that magnesium oxide forms a more stable solid solution with titanium and iron oxide than other impurities oxides.²¹⁾ And compared with the study of Li,²²⁾ the reduction degree of TTM concentrate is higher due to the lower content of impurities such as magnesium and manganese oxides.

Fig. 4.

Effect of temperature on the reduction degree of titanomagnetite concentrate with graphite.

3.2. Effect of Reductant Content

Effect of graphite content on the reduction of TTM concentrate was studied at 1623 K. C/O varied from 0.6 to 2.0. The reduction curves are presented in Fig. 5. As shown in Figs. 5 and 6 (Curve B), increase in C/O from 0.6 to 1.0 causes a sharp increase both on the reduction degree after 10 min and the final metallization degree. This can be considered as reductant content turned from insufficient to saturation. And at initial stage of reduction, the reduction degree also increases with the increasing of C/O from 0.6 to 1.0, although not as sharply, which can be attributed to the faster Boudouard reaction to provide reductant CO. This is caused by the shortening of the diffusion paths for CO that created by reaction between concentrate particles and carbon particles, when an ore particle is well surrounded by carbon particles. But there is no significant effect on the reduction and metallization degree observed when further increasing C/O from 1.0 to 2.0. This is different from Wang’s work¹⁵⁾ that the increase of C/O from 1.0 to 3.0 can also enhance the reduction degree sharply at 1373 K, but it is similar as Chen’s work¹⁶⁾ at 1473 K from 1.0 to 1.5, which can be conclude that the effects of reductant content with excess C/O on the reduction and final metallization degree become weak with temperature increasing.

Fig. 5.

Effect of graphite content on the reduction degree at 1623 K.

Fig. 6.

Effect of temperature and C/O molar ratio on the final metallization degree of titanomagnetite concentrate reduction by graphite.

3.3. Phase Transformation during Reduction of Titanomagnetite Concentrate

In the experimental condition, the reduction temperature affected the reduction and metallization degree as well as the composition of sample significantly. Figure 7 presents the final reduction products at different temperature after 120 min reaction. The sample reduced at 1373 K contains iron and TTM phase. And then ilmenite and ferrous-pseudobrookite appeares in samples reduced at 1423 and 1473 K, respectively. At 1523 K, the intensity of ilmenite peaks become weaker and ferrous-pseudobrookite peaks get stronger, which means that the content of ilmenite falls and that of ferrous-pseudobrookite rises. The intensity for graphite and TTM peaks decrease further and those for iron increase with temperature increasing. At 1623 K, C and TTM phase have disappeared, meaning that reductant gets exhausted and all TTM should be reduced. However, after 120 min reaction, not only metallic iron but also traces of ilmenite and ferrous-pseudobrookite remain in the sample. This is because of the barrier effect of impurities and morphology changes by reduction which will be discussed later. The results of phase analysis and metallization degree of reduced samples at different temperatures are given in Table 2.

Fig. 7.

XRD patterns of samples reduced by graphite at different temperatures after 120 min reaction.

Table 2. Phases and metallization degree of samples reduced by graphite at different temperatures after 120 min reaction (based on XRD analysis, Fig. 7).

Temperature/K	Phases observed	Metallization degree/%
1373	TTM, C, Fe	48.13
1423	TTM, C, Fe, FeTiO3	64.61
1473	TTM, C, Fe, FeTiO3, FeTi2O5	78.57
1523	TTM, C, Fe, FeTiO3, FeTi2O5	92.23
1573	TTM, C, Fe, FeTiO3, FeTi2O5	93.75
1623	Fe, FeTiO3, FeTi2O5	97.50

Samples in the progress of reduction at 1623 K were analysed by XRD as shown in Fig. 8. At the beginning of reduction 2.5 min, there is no new phase observed in sample nearly same as raw TTM concentrate. In the early stage of reduction for 5 min and 10 min, when the metallization degrees are 26.48% and 50.83% respectively, wüstite peaks as a transition phase are detected. And although there are some ilmenite contained in the raw ore of TTM concentrate, ilmenite could also be considered as a produced phase in the progress of reduction with higher peaks. After 15 min, ferrous- pseudobrookite is observed and wüstite disappears in the sample with the metallization degree 80.95%. Similar with the results of reduction by gas,²³⁾ wüstite in TTM is reduced very rapidly, which explains the weakness of wüstite peaks in the XRD patterns. Starting from 20 min, the peak of reductant becomes weaker than that of ilmenite and ferrous-pseudobrookite. Furthermore, no new phases appeared in later stage of reduction. In the XRD pattern of sample by reduced for 30 min, nearly 100% of reduction of iron oxides achieved, but the peaks of Ti oxides are still not detected. This may be due to the low Ti oxides content, and further investigation such as TEM and MLA observation can be used for clarifying the formation of TiO₂.²⁴⁾ The results of phase analysis and metallization degree of reduced samples in the progress are given in Table 3.

Fig. 8.

XRD patterns of samples reduced by graphite at 1623 K in the progress of reduction.

Table 3. Phases and metallization degree of samples reduced by graphite at 1623 K in the progress of reduction (based on XRD analysis, Fig. 8).

Reduction time/min	Phases observed	Metallization degree/%
2.5	TTM, C, FeTiO3	2.38
5	TTM, C, Fe, FeO, FeTiO3	26.48
10	TTM, C, Fe, FeO, FeTiO3	50.83
15	TTM, C, Fe, FeTiO3, FeTi2O5	80.95
20	TTM, C, Fe, FeTiO3, FeTi2O5	92.72
25	TTM, C, Fe, FeTiO3, FeTi2O5	94.52
30	TTM, C, Fe, FeTiO3, FeTi2O5	96.07

In equilibrium state, the reduction of TTM in the Fe–Ti–O system proceeds the following path:²⁵,²⁶⁾   

$$\begin{array}{l}F{e}_{3-x}T{i}_x{O}_4\to FeO+F{e}_{2}Ti{O}_4\to Fe+F{e}_{2}Ti{O}_4\\ \to Fe+FeTi{O}_3\to Fe+FeT{i}_2{O}_5\to Fe+Ti{O}_2\end{array}$$(3)

The reduction path includes the formation of intermediate Ti-containing phases such as ülvospinel (Fe₂TiO₄), ilmenite (FeTiO₃) and ferrous-pseudobrookite (FeTi₂O₅). In equilibrium state, the transformation in structure of Ti-containing phase during the reduction of TTM concentrate can be presented by the following sequence:

TTM (spinel cubic) →ülvospinel (spinel cubic)→ilmenite (rhombohedral)→ferrous-pseudobrookite (orthorhombic)→rutile

The first step includes the transformation of TTM to wüstite and ülvospinel with the rearrangements of Fe²⁺ in tetrahedral and octahedral sites in the lattice, and then it is the transformation from Fe²⁺ in tetrahedral and octahedral sites to the cubic metallic iron as shown in Eq. (4) (The structure of TTM was presented by the Akimoto model^24,27)).

  

$$\begin{array}{l}{\left[F{e}^{3+},F{e}^{2+}\right]}_{tetra}\cdot {\left[F{e}^{3+},F{e}^{2+},T{i}^{4+}\right]}_{octa}{O}^4\to {\left[F{e}^{2+}\right]}_{tetra}\cdot \\ {\left[F{e}^{2+},T{i}^{4+}\right]}_{octa}{O}^4\to {\left[Fe\right]}_{\text{cubic}}+{\left[F{e}^{2+}\right]}_{tetra}\cdot {\left[F{e}^{2+},T{i}^{4+}\right]}_{octa}{O}^4\\ \to {\left[Fe\right]}_{\text{cubic}}+{\left[T{i}^{4+}\right]}_{octa}{O}^4\\ \text{(tetra-tetrahedral site, octa-octahedral site, cubic- cubical site)}\end{array}$$(4)

In experiments, XRD results (Fig. 8 and Table 3) show that the easy step in the reduction is TTM to wüstite starting with the reduction of Fe³⁺ to Fe²⁺ and wüstite to Fe accompanied by the removal of oxygen. Titanium in TTM stabilizes the spinel structure, so the difficult step in the reduction is the release of Fe²⁺ from titania-ferrous phase (that is, from ilmenite and ferrous-pseudobrookite to metallic iron). Base on the above analysis, the reduction path of TTM under the experimental condition is presented as follow:   

$$\begin{array}{l}F{e}_{3-x}T{i}_x{O}_4\to (x+y-1)FeO+F{e}_{4-2x-y}T{i}_x{O}_{4-y}\to \\ (x+y-1)Fe+F{e}_{4-2x-y}T{i}_x{O}_{4-y}\to (3-x)Fe+xTi{O}_2\\ (0<\text{y}<4-2\text{x},\hspace{0.17em}\hspace{0.17em}\text{x}=0.27\pm 0.02)\end{array}$$(5)

3.4. Morphology and Energy Disperse Spectroscopy Analysis

Morphology of particles in the progress of TTM concentrate reduction by graphite at 1623 K were examined by BES (Fig. 9). It can be seen from Fig. 9(a) that the raw ore of TTM concentrate is non-homogeneous particles with an impurities oxides phase having a lamella structure with a vertical or horizontal angle. According to EDS analysis (Table 4), the black impurities oxides region with an atom ratio [Al]/[Mg]=2.62 is mainly consist of magnesia-alumina spinel (MgAl₂O₄, with an atom ratio [Al]/[Mg]=2) with high Al content, and the white background is TTM. In general, four different morphological regions are identified in Fig. 9(b). White region is an iron phase. Dark gray region is slag phase with high Si content, but also there is some unreduced Fe in it. Gray region is the rearrangement TTM. And light gray region is rich in iron with mainly consist of FeO and some reduced Fe with an atom ratio [O]/[Fe]=0.75 (excluding [O] in nonferrous oxides), which is consistent with the XRD detection (Fig. 8 and Table 3). After 5 min reduction (Fig. 9(b)), the reduced iron first appears at outside surface of particle and the phase at the inner of particle is basic invariable. But the phase boundaries become smooth and the impurities oxides phase disappears which could be the results of phase diffusion at the high temperature and formation of low melting point substances. According to the previous analysis, in TTM concentrate, Fe³⁺ is easier to be reduced to Fe²⁺ and further to Fe accompanied by the removal of oxygen. Therefore, in the following progress of reduction, the Fe²⁺ in the light gray region can act as nucleation starting point. Overall, the reduction of the TTM concentrate particles started in a topochemical way. After 10 min reduction (Fig. 9(c)), the internal unreduced phase becomes more homogeneous and the reduced iron begins to nucleate and grow up at the inner of particle. This phenomenon has continued and is more obvious in Fig. 9(d) with 15 min reduction. In the middle and late stages of reduction with metallization degree 50%–90%, basically there are three morphological regions, white iron phase, gray unreduced region and black slag phase (in Figs. 9(c) and 9(d)). From EDS analysis (Table 4), it can be known that Si and Ca are easier to separate from TTM into slag as major nonferrous components. Meanwhile, Mg, Mn and Al are more stable as barrier impurities, and Fe content in the unreduced region is lower with reduction preceding meaning that the activity of Fe²⁺ is lower. This result also validates the previous analysis about the effect of impurities on reduction degree. In the final stage of reaction, as the metallization degree is higher than 90% (Figs. 9(e) and 9(f)), the reduced iron grown by coalescence and the progress of reduction becomes more difficult with lower thermodynamic activity of Fe²⁺. Even after 120 min reduction (Fig. 9(f)), there are still some unreduced phase remained in the particle surrounding by uniform and dense metallic iron which forms a shell can also hinder the reduction.

Fig. 9.

Morphology changes of titanomagnetite concentrate particles during the reduction by graphite at 1623 K.

Table 4. Compositions of titanomagnetite concentrate during the reduction by graphite at 1623 K (EDS analysis, point on Fig. 9, at.%).

Point No.	O	Fe	Ti	Mg	Si	Ca	Al	Mn
1	51.16	20.05	3.62	6.57	0.12	0	17.27	0.14
2	50.85	33.65	4.74	2.05	0.16	0.05	7.35	0.17
3	0	98.56	0.71	0.06	0.37	0.19	0.01	0
4	43.99	33.45	12.95	3.38	0.34	0.12	4.64	0.38
5	46.34	44.27	4.34	3.13	0.09	0	0.75	0.11
6	46.92	26.78	0.89	2.61	14.31	4.64	2.68	0.51
7	0	97.46	0.63	0.64	0.55	0	0.03	0
8	56.42	8.59	4.62	3.26	15.39	3.03	7.23	0.46
9	46.41	23.02	16.45	5.10	0.03	0.15	7.72	0.64

4. Conclusions

(1) In the temperature range of 1423 to 1623 K, initially the reduction of TTM concentrate by graphite proceeded rapidly in the first 30 min, with the reduction time prolonging, the reaction rate decreased and at last the reduction was arrested. The low reduction degree is owing to the high impurities oxides content such as magnesium oxide in TTM concentrate which baffled reduction of Fe²⁺.

(2) The reduction and metallization degree increased with the molar ratio of carbon to oxygen until up to 1.0. However, the effect was negligible with the further increase of the molar ratio.

(3) During the reduction, TTM concentrate was reduced to iron and iron-titanium oxides or titanium oxides, depending on the reduction temperature and time. The reduction path at temperatures above 1373 K is suggested as follow:

  

$$\begin{array}{l}F{e}_{3-x}T{i}_x{O}_4\to (x+y-1)FeO+F{e}_{4-2x-y}T{i}_x{O}_{4-y}\to \\ (x+y-1)Fe+F{e}_{4-2x-y}T{i}_x{O}_{4-y}\to (3-x)Fe+xTi{O}_2\\ (0<\text{y}<4-2\text{x},\hspace{0.17em}\hspace{0.17em}\text{x}=0.27\pm 0.02)\end{array}$$

Acknowledgements

The authors gratefully acknowledge the support of National Natural Science Foundation of China (No. 51090381) and China Postdoctoral Science Foundation (No. 2012m510320).

References

• 1)    Y. M.  Wang,  Z. F.  Yuan,  Z. C.  Guo,  Q. Q.  Tan,  Z. Y.  Li and  W. Z.  Jiang: T. Nonferr. Metal. Soc., 18 (2008), 962.
• 2)    W. S.  Zhang,  Z. W.  Zhu and  C. Y.  Cheng: Hydrometallurgy, 108 (2011), 177.
• 3)    H. Y.  Sun,  J. S.  Wang,  X. J.  Dong and  Q. G.  Xue: Metal. Int., 17 (2012), 49.
• 4)    E.  Park and  O.  Ostrovski: ISIJ Int., 43 (2003), 1316.
• 5)   U.S. Geological Survey: USGS Mineral Commodity Summaries 2011, Unit State Geologic, Washington, (2011), 172.
• 6)    S. L.  Yang and  J. F.  Sheng: Technology of Pig Iron and Titanium Slag Smelting, Metallurgy Industry Press, Beijing, (2006), 19.
• 7)    L.  Zhang,  L. N.  Zhang and  M. Y.  Wang: Miner. Eng., 20 (2007), 684.
• 8)    L. S.  Li and  Z. T.  Sui: Acta Phys. Chim. Sin., 17 (2001), 845.
• 9)    P. C.  Pistorius and  C.  Coetzee: Metall. Mater. Trans. B, 34 (2003), 581.
• 10)    N.  El-Hazek,  T. A.  Lasheen,  R.  El-Sheikh and  S. A.  Zaki:Hydrometallurgy, 87 (2007), 45.
• 11)    A.  Andrew,  Y.  Li,  G. Q.  Zhang and  O.  Oleg: Int. J. Miner. Process., 100 (2011), 166.
• 12)    C. S.  Kucukkaragoz and  R. H.  Eric: Miner. Eng., 19 (2006), 334.
• 13)    S.  Kikuchi,  S.  Ito and  I.  Kobayashi: Kobelco Technol. Rev., 29 (2010), 77.
• 14)    B. B.  Agrawal,  K.  Prasad and  H. S.  Ray: ISIJ Int., 30 (1990), 997.
• 15)    Y. M.  Wang and  Z. F.  Yuan: Int. J. Miner. Process., 81 (2006), 133.
• 16)    D. S.  Chen,  B.  Song,  L. A.  Wang,  T.  Qi,  Y.  Wang and  W. J.  Wang: Miner. Eng., 24 (2011), 864.
• 17)    D. G.  Jones: Trans. Inst. Min. Metall., Sect. C, 83 (1974), 1.
• 18)    R.  Merk and  C. A.  Pickles: Can. Metall. Q., 27 (1988), 179.
• 19)    K. G.  Suresh,  V.  Rajakumar and  P.  Grieveson: Metall. Mater. Trans. B, 18 (1987), 713.
• 20)    K. G.  Suresh,  V.  Rajakumar and  P.  Grieveson: Metall. Mater. Trans. B, 18 (1989), 735.
• 21)    J. F.  Elliott,  M.  Gleiser and  V.  Ramakrishna: Thermochemistry for Steelmaking, Vol. 2, Addison-Wesley, Massachusetts, (1963), 307.
• 22)    W. B.  Li,  Z. F.  Yuan,  C.  Xu,  Y. F.  Pan and  X. Q.  Wang: J. Iron Steel Res. Int., 12 (2005), 1.
• 23)    E.  Park and  O.  Ostrovski: ISIJ Int., 43 (2003), 1316.
• 24)    E.  Park and  O.  Ostrovski: ISIJ Int., 44 (2004), 990.
• 25)    S.  Itoh,  O.  Inoue and  T.  Azakami: Mater. Trans. JIM, 39 (1998), 391.
• 26)    O.  Inoue,  S.  Itoh and  T.  Azakami: J. Jpn. Inst. Met., 60 (1996), 834.
• 27)    S.  Akimoto: J. Geomag. Geoelect., 6 (1954), 1.