Role of MgO in Sinter from Perspective of MgO Distribution between Liquid and Magnetite Phases in FeO x CaO SiO 2 MgO System

Transcription

1 ISIJ International, Vol. 57 (2017), ISIJ International, No. 12 Vol. 57 (2017), No. 12, pp Role of MgO in Sinter from Perspective of MgO Distribution between Liquid and Magnetite Phases in FeO x CaO SiO 2 MgO System Miyuki HAYASHI, 1) * Hiroki TANAKA, 2) Takashi WATANABE 1) and Masahiro SUSA 1) 1) Department of Materials Science and Engineering, Tokyo Institute of Technology, Ookayama, Meguro-ku, Tokyo, Japan. 2) Department of Metallurgy and Ceramics Science, Tokyo Institute of Technology, Now at Muroran works, Nippon Steel and Sumitomo Metal Co. 12 Nakamachi, Muroran City, Hokkaido, Japan. (Received on May 28, 2017; accepted on July 19, 2017) The chemical compositions of liquid in equilibrium with FeO x in the FeO x CaO SiO 2 MgO system at K and oxygen partial pressures of 10 7 atm and 10 6 atm have been determined along with the MgO concentrations in the liquid and FeO x phases to understand the role of MgO in improvement of reducibility of the sinter produced from deteriorated iron ores. The liquid area in the FeO x rich side was decreased with additions of MgO over the measurement range of C/S ratios between 0.32 and 2.28, and was separated into silicate based and calcium ferrite based liquid phases, where 2CaO SiO 2 phase becomes stable thermodynamically in the C/S ratios between 1.66 and The ratio of MgO concentration in the liquid phase to that in the FeO x phase decreased with increasing C/S ratio from 0.36 to 2.28, which trend has been explained from the viewpoint of basicity. On the basis of these findings, the role of MgO in improvement of reducibility of the sinter has been discussed to conclude that additions of MgO make the cohesive zone thinner and improves the gas permeability in a blast furnace. KEY WORDS: iron ore sinter; FeO x SiO 2 CaO MgO system; MgO distribution; basicity. 1. Introduction Steelmaking industry is facing an issue of deterioration in quality of iron ores. Iron ores contains more phosphorus, combined water and alumina concentrations than before, which brings about various problems for the blast furnace operation. 1 4) In particular, the increase of Al 2 O 3 content in iron ore is very serious because Al 2 O 3 lowers the reducibility of iron ore sinter and decreases the gas permeability in a blast furnace. 5,6) To overcome these problems, many attempts have been done, such as introducing micro pores into the sinter, 7) decreasing the volumes of gangue minerals in the sinter 7) and optimizing the compositions 4,8 10) of gangue minerals in the sinter. Among them, additions of MgO into the sinter have been one of the most commonly applied projects because the sinter containing MgO shows superior performance at the lower part of a blast furnace ) For example, Matsumura et al. 11) have tried to add MgO into the sinter in the form of dolomite. They have found that the reducibility of the sinter is improved, and explained that the improvement is due to less formation of silicate slag phase with poor reducibility because both MgO and SiO 2 dissolve into calcium ferrite which is the major part of the sinter. In contrast, Higuchi * Corresponding author: hayashi@mtl.titech.ac.jp DOI: et al. 12) and Sunahara et al. 10) have insisted that MgO dissolves into the slag phase, affecting physicochemical properties of slag, and thereby decreases the dripping temperature and increases the hearth drainage rate of slag. The decrease in the dripping temperature, in turn, makes the cohesive zone thinner, resulting in the increase of gas permeability in a blast furnace. Thus, there is no consensus on the role of MgO although its effectiveness is shared as common understanding. The role of MgO should be investigated from the perspective of kinetics as well as thermodynamics. The present work takes a thermodynamic approach; in that case, on the basis of the previous reports, 10 13) it is strongly required to have information about MgO distributions in mineral phases (hematite, magnetite, calcium ferrite and silicate slag phases) composing the sinter. Thus, the sinter is basically the ternary system FeO x, CaO and SiO 2, and essential is the phase diagram of the system with information on MgO, as shown in Fig. 1: 14) The isotherm of FeO x CaO SiO 2 slag containing MgO at K and oxygen partial pressures (P O2 ) of atm which is similar to that of the sintering process 15) has been calculated by Kongoli and his coworkers, 16,17) and has also been experimentally obtained by Kimura et al. 18) Their isotherms are in good agreement with each other. However, there is no consensus between them in the respect of the continuity of the silicate-based and calcium ferrite-based liquid phases at P O2 = 10 8 atm: 2017 ISIJ 2124

2 Table 1. Initial compositions of samples. Fig. 1. Isotherms for the FeO x-cao-sio 2-5mass% MgO system at P O2 = 10 8 atm and at K experimentally obtained by Kimura et al. 14,18) (solid line) and calculated by Kongoli and Yazawa 17) (dashed line). Kongoli et al. 17) have reported that there are two liquid regions, whereas Kimura et al. 18) has reported the presence of only one liquid phase. In these reports, moreover, the MgO solubility in magnetite phase has not been focused on in spite of the fact that the sufficient amount of MgO is solved in magnetite which is expected to be equilibrated with the liquid phase around P O2 = 10 8 atm. Consequently, the present work aims to determine the chemical compositions of liquid in equilibrium with FeO x at K and oxygen partial pressures of 10 7 atm and 10 6 atm along with the MgO concentrations in the liquid and FeO x phases to understand the role of MgO in improvement of reducibility of the sinter produced from deteriorated iron ores. No. Initial compositions (mass%) Fe 2O 3 CaO SiO 2 MgO C/S Type of crucible P O2 (atm) A A A A A B B A A * A * B * A * A * B * B * C * C * C A * CaCO 3 powders were used instead of CaO. 2. Experimental 2.1. Sample Preparation In the present work, 19 samples were used, as given in Table 1. These samples were prepared from regent grade Fe 2 O 3, SiO 2, MgO and CaO powders, the last being made by thermal decomposition of CaCO 3 ; in some cases, CaCO 3 powders were used instead of CaO. These reagents were weighed to desired compositions and mixed in an alumina mortar Equilibrium Experiment The samples were placed in the respective crucibles, as given in Table 1, and held at K in a flow of Ar O 2 gases (50 ml/min) for 24 h to attain equilibration, where temperature (T) was controlled within ± 2 K in the uniform temperature zone, and partial pressures of oxygen (P O2 ) were controlled to be atm or atm, depending on the sample, which values were recorded in the outlet gas using an oxygen sensor. The equilibration time was determined from preliminary experiments, as mentioned in 3.1. After the equilibration, the crucible containing the sample was quenched into an ice-water mixture. In the present work, three types of crucible (A, B and C) were used depending on the sample, as shown in Table 1 and Fig. 2: Crucible A was folded platinum foil 12 mm 35 mm Fig. 2. Schematic illustration of crucibles mm sized, in which about 0.4 g of a sample mixture was contained. This type of crucible was used for samples having C/S 1, where C/S represents the ratio of (mass%cao)/(mass%sio 2 ). Crucible B was originally developed by Hayes and coauthors 19) and, in the present work, was made of platinum wire having 0.5 mm diameter, on which about 0.22 g of a sample mixture having 10 mm diameter was placed after the sample being uniaxially cold-pressed. This type of crucible was ISIJ

3 used for samples having C/S 1.5, which samples would contain dendritic 2CaO SiO 2 (hereinafter denoted as C 2 S) after quenching: 20) however, dendritic C 2 S is not a phase equilibrated with FeO x at K and is produced during the cooling process. Using crucible B, the molten sample was maintained in the form of film between the gaps of the wire, which resulted in higher quenching rates to minimize C 2 S crystallisation. Crucible C was made of platinum wires and foil, on which about 0.3 g of a sample mixture was placed. This type of crucible was used for samples having C/S 1.7 because samples with higher C/S might be difficult to be maintained by crucible B due to lower viscosities Sample Characterisation X-ray diffractometry (XRD) was applied to identify the phases of samples after equilibrium experiments. The microstructures and chemical compositions of all the phases were measured by an electron probe microanalyzer (EPMA) with an accelerating voltage of 15 kv and a probe current of 20 na, where the ZAF correction was made. For quantitative analyses by EPMA, two glassy samples were prepared as standards: 33.3(mass%)Fe 2 O CaO- 30.9SiO 2-5.0MgO (C/S = 1) and 42.0(mass%)Fe 2 O CaO-22.9SiO 2-3.0MgO (C/S = 1.4), the former was used for the samples with C/S 1 and the latter for the samples with C/S > 1. The composition of the standard sample was calculated using the mixture fractions of the Fe 2 O 3, CaO, SiO 2 and Al 2 O 3 powders. All the iron ions are assumed to be Fe 3+ by considering the Mössbauer spectroscopic analyses. Additionally, magnesio-ferrite sample (88(mass%)Fe 3 O 4-12MgO) was also prepared as standard in reference to the phase diagram of Fe 3 O 4 MgO system 21) because preliminary experiments confirmed that the equilibrated solid phase was magnetite containing a considerable amount of MgO, as mentioned in 3.1. This standard sample was prepared as follows: Weighed mixture of Fe 3 O 4 and MgO powders was uniaxially pressed into the form of tablet having 10 mm diameter. This sample was heated at K in air for more than 16 h and then pulverized, pressed and heated again in the same manner. This sequence was repeated three times to obtain the standard sample. The sample obtained finally consisted of a single phase of magnesio-ferrite, which was confirmed by XRD analysis. concentration in the liquid phase and the MgO concentration in the magnetite phase as functions of holding time, respectively, where the FeO x concentration has been calculated as that for FeO. All the data plotted in both figures are the Fig. 3. Back-scattered electron (BSE) image of the cross section of sample 19 held for 24 h at P O2 = atm in the experimental condition together with a line profile of Mg element. Fig. 4. FeO x concentration in the liquid phase as a function of holding time. 3. Results 3.1. Confirmation of Equilibration by Preliminary Experiments Figure 3 shows a back-scattered electron (BSE) image of the cross section of sample 19 shown in Table 1 held for 24 h at P O2 = atm in the experimental condition together with a line profile of Mg element. It can be seen that the sample is composed of FeO x and slag phases: the former is magnetite according to XRD analysis and the latter used to be liquid. In addition, the Mg distribution is uniform across the magnetite phase, which suggests that the system attained equilibrium in 24 h. To firmly establish the equilibration time, EPMA analyses were made on the liquid and magnetite phases of sample 19 which were held for different times from 24 h to 96 h. Figures 4 and 5 show the FeO x Fig. 5. MgO concentration in magnetite as a function of holding time ISIJ 2126

4 respective averages of values recorded by analyses on ten different positions and the error bar represents the respective scattering range. It can be seen that both FeO x concentration in the liquid phase and the MgO concentration in the magnetite phase become constant in 24 h. Thus, it is reasonable to take 24 h as the equilibration time in Results of All the Other Samples Table 2 shows the chemical compositions of liquid and magnetite phases obtained by EPMA analyses for samples 1 18, where the FeO x concentration in the liquid phase has been calculated as that for FeO. For the magnetite phase, the CaO concentration is less than 3 mass% and the SiO 2 concentration is negligibly small. In addition, BSE images have confirmed that the liquid phase was in equilibrium with the magnetite phase at K in the samples except for samples 8, 14, 16 and 17. For sample 8 the liquid phase coexists with SiO 2 and magnetite phases, as shown in Fig. 6(a), whereas for samples 14, 16 and 17 the liquid phase coexists with C 2 S and magnetite phases, as shown in Figs. 6(b) 6(d). The C 2 S phase observed in the present work is granular; in contrast, it is known that C 2 S produced during the cooling process is dendritic 20) or fine needle-shaped 19) microcrystalline. Thus, it is supposed that the C 2 S phase in samples 14, 16 and 17 is an equilibrium phase. Figure 7 shows the relation between FeO x and MgO concentrations in the liquid phase in equilibrium with magnetite for samples 1 5 with C/S = 1.0 in comparison with reported values for the same sample without MgO 20) equilibrated at K and P O2 = atm. The FeO x concentration in the liquid phase decreases with an increase in the MgO concentration, which means that the liquid phase area equilibrated with the magnetite phase at this temperature becomes smaller with increasing MgO concentration. This is in accordance with the thermodynamic modelling calculation by Kongoli and Yazawa. 16) Fig. 6. BSE images for samples 8 (a) where liquid phase coexists with SiO 2 and magnetite phases, and samples 14 (b), 16(c) and 17(d) where the liquid phase coexists with C 2S and magnetite phases. Table 2. Compositions of liquidus and magnetite phases (mass%), where the FeO x contents in both phases have been calculated by regarding FeO x as FeO, and equilibrated solid phases. No. Compositions of liquid phase Compositions of Fe 3O 4 phase C/S FeO x CaO SiO 2 MgO FeO x CaO SiO 2 MgO Solid phase Fe 3O Fe 3O Fe 3O Fe 3O Fe 3O Fe 3O Fe 3O Fe 3O 4, SiO Fe 3O Fe 3O Fe 3O Fe 3O Fe 3O Fe 3O 4, C 2S Fe 3O Fe 3O 4, C 2S Fe 3O 4, C 2S Fe 3O ISIJ

5 Fig. 7. Relation between FeO x and MgO concentrations in the liquid phase in equilibrium with magnetite for samples 1 5 with C/S = 1.0 in comparison with reported values for the same sample without MgO 20) equilibrated at k and P O2 = atm. Figure 8 shows the isotherm at K, i.e., the liquid compositions in equilibrium with magnetite for the FeO x CaO SiO 2 system with 5 mass% MgO at P O2 = atm (samples 6 18) (thick solid line), in comparison with reported values for the same system without MgO at P O2 = atm 20) (dashed line), where the liquid compositions are plotted assuming that the liquids are composed only of CaO, SiO 2 and FeO. Thin solid lines (straight lines) are those connecting the liquid compositions with the equilibrated solid phases, i.e., magnetite, SiO 2 or C 2 S for the system with 5 mass% MgO. The addition of MgO decreases the liquid phase area equilibrated with magnetite in the C/S range investigated. Different from the system without MgO, the liquid phase area in the present system is separated into two liquid phases in the upper and lower sides of the Alkemade line between FeO x and C 2 S. In addition, in the presence of MgO, the C 2 S phase becomes thermodynamically stable in the C/S range between 1.66 and Figure 9 shows the MgO concentration in the liquid and magnetite phases as functions of the C/S ratio for samples 6, 7, 9 13, 15 and 18. The MgO concentration in magnetite increases with increasing C/S ratio and about 14 mass% at C/S = This high value of MgO concentration in magnetite is consistent with the conclusion by Narita et al., 22) who have reported that the MgO distribution in the magnetite phase is the largest in all the phases relevant to iron ore sinters. Fig. 8. Isotherm at K, i.e., the liquid compositions in equilibrium with magnetite for the FeO x CaO SiO 2 system with 5 mass% MgO at P O2 = atm in comparison with reported values for the same system without MgO at P O2 = atm. 20) Thin solid lines (straight lines) are those connecting the liquid compositions with the equilibrated solid phases, i.e., magnetite, SiO 2 or C 2S for the system with 5 mass% MgO. Fig. 9. MgO contents in the liquid and magnetite phases as a function of the C/S values. 4. Discussion 4.1. Validity of Composition Analysis by EPMA The mass percentage of the liquid phase equilibrated with magnetite can be estimated from mass balance using the analyzed compositions of liquid and magnetite phases in addition to the nominal compositions of the sample. Figure 10 shows the relation between the mass percentages of liquid phase estimated using FeO x and CaO concentrations for samples 6, 7, 9 13, 15 and 18. It can be seen that the plotted data fall down on the straight line passing through the origin and having the slope of unity. This indicates that the composition analysis by EPMA was valid without effects from fluorescent X-rays. Figure 11 shows the average MgO concentration in the sample as a function of the C/S ratio for Fig. 10. Relation between the mass fractions of liquid phase calculated by mass balance of FeO (horizontal axis) and CaO (vertical axis) ISIJ 2128

6 samples 6, 7, 9 13, 15 and 18 having 5 mass% MgO, where the average MgO concentration has been estimated using the MgO concentrations in the liquid and magnetite phases obtained by EPMA analyses in addition to the mass percentages of the liquid and magnetite phases estimated using the FeO x concentration as above. The horizontal dotted line in Fig. 11 represents the nominal MgO concentration of the samples. The deviation of the estimated concentration from the nominal is within ± 20% except for the three samples of C/S = 0.32 and C/S > 2, the deviation for which is around ± 40%. This would be because the C/S ratios of these three samples are very different from those of the standard samples for EPMA analysis, i.e., 1 and 1.4. The agreement in most samples indicates that the composition analysis by EPMA was reasonable MgO Distribution between Liquid and Magnetite Phases Figure 12 shows the distribution ratio of MgO between the liquid and magnetite phases as a function of C/S ratio for samples 6, 7 and In these samples, one liquid phase is in equilibrium with magnetite at K, as shown in Fig. 8. On the other hand, Fig. 13 conceptually illustrates the tie lines connecting the compositions of the liquid and magnetite phases in equilibrium at K for samples 6, 7, 9 13, 15 and 18 in the composition tetrahedron of the FeO x CaO SiO 2 MgO system. Both Figs. 12 and 13 indicate that the distribution of MgO in the liquid phase decreases with increasing C/S ratio from 0.36 to This trend can be explained from the viewpoint of basicity as follows: The basic structure of a liquid slag is the silica network, consisting of SiO 4 4 tetrahedra interconnecting by bridging oxygens, and the cations such as Na + and Ca 2+ breaking the Si O Si covalent bonds to form non-bridging oxygens O. 23) For the FeO x CaO SiO 2 system, SiO 2 is an acidic oxide acting as a network former of silica, CaO a basic oxide acting as a network breaker of silica, and FeO x an amphoteric oxide acting as either a network former or a network breaker. In the present samples, the FeO x concentrations in the liquid phase equilibrated with magnetite are within % while the C/S ratio varies over the range between 0.36 and Therefore, the FeO x concentrations in the liquid phase may be assumed to be identical irrespective of the C/S ratio. The C/S ratio principally determines the degree of depolymerisation of the liquid phase, where the degree of depolymerisation corresponds to the ratio of number of non-bridging oxygen atoms to number of tetrahedrally-coordinated atoms. Although MgO is also a basic oxide and breaks silica network, its basicity is smaller than that of CaO. 24) Thus, with increasing C/S ratio, MgO may be less stable in the liquid phase, resulting in smaller distribution ratios of MgO. Fig. 12. Ratios of the MgO content in liquid phase to that in magnetite phase as a function of the C/S ratio for the samples No. 6, 7 and 9 13 having the initial compositions of 50(mass%)Fe 2O 3 45xCaO 45(1-x)SiO 2 5MgO (x/(1-x) = ), where one solid phase of magnetite is equilibrated with one liquid phase. Fig. 11. Average MgO concentrations in the samples calculated using the respective MgO contents in liquid and magnetite phases analyzed by EPMA as well as the mass fractions of liquid and magnetite phases calculated by the mass balance of FeO x shown in Fig. 10 as a function of the C/S ratio. Fig. 13. Conceptual illustration of the compositions of liquid and magnetite phases equilibrated with each other with the tie lines connecting those compositions on a compositional tetrahedron of the FeO x CaO SiO 2 MgO system on samples No. 6, 7, 9 13, 15 and ISIJ

7 4.3. Composition Design of Sinter Based upon Role of MgO Sinters which have better gas permeability in a blast furnace are designed on the basis of the distribution ratio of MgO between the liquid and magnetite phases. In the cohesive zone of a blast furnace, the olivine-like melt is primarily formed around K, 25) and then, the melt volume increases owing to the two eutectic reactions of the olivine-wüstite-c 2 S pseudo-ternary system in the range K and the C 2 S- wüstite-gehlenite(2cao Al 2 O 3 SiO 2 ) pseudo-ternary system above K. 5,25) In these reactions, wüstite and silicate produce the slag melt and increase the melt volume, which decreases the gas permeability. There are two proposals to avoid such a situation: one is to improve the reducibility of sinters to metallic iron so as to decrease the amount of residual wüstite in reduced sinters, and the other is to dissolve MgO into magnetite in sinters to form magnetio-ferrite. In a blast furnace, magnetio-ferrite is reduced to magnetio-wüstite, which has higher solidus and liquidus temperatures than wüstite without solid solution. Thus, magnetio-wüstite may increase the eutectic temperatures of the reactions related to the slag melt formation, i.e., the softening-melting temperatures, which makes the cohesive zone thinner and improves the gas permeability. Consequently, the present work suggests that the distribution of MgO in magnetite can be increased by increasing the C/S ratio. 5. Conclusions In order to understand the role of MgO in the sinter produced from deteriorated iron ores, the chemical compositions of liquid in equilibrium with FeO x in the FeO x CaO SiO 2 MgO system at K and oxygen partial pressures of 10 7 atm and 10 6 atm have been determined along with the MgO concentrations in the liquid and FeO x phases. The results obtained are summarized as follows. (1) The liquid area in the FeO x -rich side is decreased with additions of MgO over the measurement range of the C/S ratios between 0.32 and 2.28, and is separated into two liquid areas such as the silicate-based and calcium ferrite-based liquid phases, where the C 2 S phase becomes stable thermodynamically in the range of the C/S ratios between 1.66 and (2) The ratio of MgO concentration in the liquid phase to that in the FeO x phase decreases with increasing C/S ratio from 0.36 to This trend has been explained from the viewpoint of basicity: MgO is less basic than CaO and thereby MgO may be less stable in the liquid phase with increasing C/S ratio, resulting in smaller distribution ratios of MgO. (3) On the basis of above findings, the role of MgO in improvement of reducibility of the sinter produced from deteriorated iron ores has been discussed as follows. The MgO distribution in magnetite should be increased with increasing the C/S ratio to form magnetio-ferrite. Magnetio-ferrite is reduced to magnetio-wüstite having higher solidus and liquidus temperatures than FeO. Magnetio-wüstite may increase the eutectic temperatures of the reactions relevant to slag melt formation, which makes the cohesive zone thinner and improves the gas permeability. REFERENCES 1) Y. Nosaka and Y. Hosotani: Tetsu-to-Hagané, 75 (1989), ) T. Kawaguchi, S. Kasama and T. Inazumi: Tetsu-to-Hagané, 78 (1992), ) H. Ueno, K. Yamaguchi, T. Orimoto, Y. Okuno, S. Matsunaga, H. Oda, S. Amano and M. Nose: Proc. 1st Int. Cong. on Science and Technology of Ironmaking, ISIJ, Tokyo, (1994), ) M. Hino, T. Nagasaka, A. Katsumata, K. Higuchi, K. Yamaguchi and N. Kon-no: Metall. Mater. Trans. B, 30B (1999), ) T. Tsutsumi, Z. Wang, Y. Sasaki, Y. Kashiwaya, K. Ishii and N. Konno: Tetsu-to-Hagané, 84 (1998), ) K. Yamaguchi, H. Ueno, T. Kawaguchi, S. Matsunaga, H. Oda and S. Amano: ISIJ Int., 34 (1994), ) K. Yamaguchi: Tetsu-to-Hagané, 87 (2001), ) K. Aketa, M. Takami, I. Shigaki, K. Hoshino, T. Takahashi and K. Kadoguchi: Tetsu-to-Hagané, 73 (1987), ) Y. Niwa, T. Sumigama, A. Maki, H. Ito, H. Inoue and T. Tamura: ISIJ Int., 31 (1991), ) K. Sunahara, K. Nakano, M. Hoshi, T. Inada, S. Komatsu and T. Yamamoto: Tetsu-to-Hagané, 92 (2006), ) M. Matsumura, M. Hoshi and T. Kawaguchi: Tetsu-to-Hagané, 92 (2006), ) K. Higuchi, M. Naito, M. Nakano and Y. Takamoto: ISIJ Int., 44 (2004), ) I. Shigaki, S. Shirouchi, K. Tokutake and N. Hasegawa: Tetsu-to- Hagané, 75 (1989), ) H. Kimura, T. Ogawa, M. Kakiki, A. Matsumoto, T. Hamano and F. Tsukihashi: Tetsu-to-Hagané, 92 (2006), ) H. Yoshikoshi and O. Komatsu: Tetsu-to-Hagané, 63 (1977), S ) F. Kongoli and A. Yazawa: Proc. 6th Int. Conf. on Molten Slags, Fluxes and Salts, KTH, Stockholm-Helsinki, (2000), CD-ROM. 17) F. Kongoli, I. McBow and A. Yazawa: Proc. Yazawa Int. Symp., Vol. 1, TMS, Warrendale, PA, (2003), ) H. Kimura, T. Ogawa, M. Kakiki, A. Matsumoto and F. Tsukihashi: ISIJ Int., 45 (2005), ) P. C. Hayes, J. Chen and E. Jak: Proc. 10th Int. Conf. on Molten Slags, Fluxes and Salts (MOLTEN16), TMS, Pittsburgh, PA, (2016), ) Y. Katahira, T. Watanabe and M. Hayashi: ISIJ Int., 55 (2015), ) M. Kowalski, P. J. Spencer and D. Neuschütz: Slag Atlas, 2nd ed., Verlag Stahleisen, Düsseldorf, (1995), ) K. Narita, M. Maekawa, M. Deguchi and T. Saito: Tetsu-to-Hagané, 66 (1980), ) K. C. Mills: Slag Atlas, 2nd ed., Verlag Stahleisen, Düsseldorf, (1995), 5. 24) K. C. Mills: Slag Atlas, 2nd ed., Verlag Stahleisen, Düsseldorf, (1995), ) M. Hayashi, K. Suzuki, Y. Maeda and T. Watanabe: ISIJ Int., 55 (2015), ISIJ 2130