Development of Iron Recovery Technique from Steelmaking Slag by Reduction at High Temperature

Abstract

In order to develop a new recycling process of steelmaking slag, reduction of (Fe_tO) and (P₂O₅) in steelmaking slag at high temperature have been investigated. In this work, 50 kg-scale experiments which simulated rotary kiln were conducted to investigate the separation behavior between slag and metal. Main results are as follows. (1) Molten iron was tapped out from the experimental furnace under the condition that more than 86% iron was reduced. (2) Weight of reduced iron did not affect the undefined ratio of phosphorus. (3) Common logarithm of phosphorus distribution ratio (logL_P), which was used as an index to explain the effect of slag composition, temperature and oxygen potential, correlates to the undefined phosphorus ratio. (4) In the experiments which simulated rotary kiln treatment, calculated phosphorus distribution ratio (L_P) was smaller than that from experimental results. It can be said that the phosphorus transfer into the metallic phase decreased because the smaller interface between slag phase and metallic phase was obtained due to the slag/metal separation.

1. Introduction

The ironmaking process generates approximately 290 kg/t-steel of blast furnace slag, which is effectively utilized as a raw material for cement, while the steelmaking process generates about 120 kg/t-steel of steelmaking slag, which is used as a backfill material for civil engineering. In order to utilize steelmaking slag as a roadbed material, it is necessary to reduce hygroscopic expansion due to the remaining lime in the slag. On the other hand, some research has examined techniques for effective treatment and utilization of steelmaking slag, for example, in recycling to the ironmaking process,^1,2,3) with the aim of recycling and utilizing its Fe and CaO content, but in recycling to the ironmaking process, the phosphorus in the slag is transferred to the hot metal in the blast furnace under a reducing condition and may have an adverse effect on the quality of steel products. For this reason, the amount of slag recycling to the ironmaking process is limited in the present situation.

Several studies on removal of phosphorus from steelmaking slag have been reported. Shiomi et al.⁴⁾ conducted slag reduction experiments by melting synthetic slags with a slag basicity (CaO/SiO₂ weight percent ratio) of 1.1 to 1.2 in a graphite crucible, and found that a slag dephosphorization rate of 68 to 94% could be achieved in the temperature range of 1773 to 1873 K, and reduction of Fe_tO in the slag preceded that of P₂O₅. They also reported that the reduction rate of P₂O₅ in the slag by carbon is determined by the chemical reaction. Takeuchi et al.⁵⁾ conducted experiments on the reduction of converter slag with carbon in the presence of a Fe–Si alloy, and reported that phosphorus formed P₂ gas which partially dissolved in the molten iron, and at least 60% of the phosphorus could be removed by evaporation and recovered as single phosphorus. Nagata⁶⁾ reported that a vaporized dephosphorization ratio of 70% could be achieved at in the 1896 to 1938 K temperature range by mixing pretreatment slag with a basicity of 2.7 and graphite powder, melting the mixture in a graphite crucible and conducting a reduction experiment, and also mentioned the possibility of further dephosphorization by increasing the reaction interfacial area by slag stirring.

Morita et al.⁷⁾ and Ito et al.⁸⁾ studied dephosphorization from slag by rapid heating. In addition, Kubo et al.⁹⁾ focused on the fact that hot metal dephosphorization slag can be roughly divided into a phosphorus enriched phase containing little iron and an iron-rich phase containing little phosphorus. They carried out experiments in which the two phases were separated magnetically by applying a strong magnetic field gradient, and examined the potential of this technique for recovering phosphorus from steelmaking slag.

Matsui et al.¹⁰⁾ investigated the influence of temperature on the Fe_tO and P₂O₅ reduction behavior of hot metal pretreatment slag and converter slag (basicity = 1.2 to 4.0) under a stirring condition, and reported that the removal ratio of phosphorus from slag exceeded 50% when the activity of Fe_tO in slag was reduced to about 0.01 or less. Nakase et al.¹¹⁾ carried out experiments to reduce steelmaking slag with different Fe_tO concentrations at high temperature under a stirring condition, and reported that the ratio of phosphorus removed into the gas phase increased in slag with a low Fe_tO concentration.

Harada et al.^12,13) carried out experiments in which molten converter slag and coke were added onto a hot metal bath kept in a DC arc furnace and the Fe and P in the slag were recovered to the hot metal, and reported that more than 90% of the Fe and P could be reduced without gas stirring in the hot metal bath and slag foaming during treatment could be suppressed.

As described above, many studies on the reduction behavior of Fe_tO and P₂O₅ in steelmaking slag have been carried out, but few reports have investigated the separation of metallic Fe and the residual slag formed by reduction. In this study, the authors carried out high temperature reduction experiments with steelmaking slag at a 50 kg scale simulating a rotary kiln, and investigated the reduction and separation behavior of iron and phosphorus from steelmaking slag.

2. Experimental Method

These steelmaking slag reduction experiments were carried out using a rotary furnace^14,15) owned by Sumitomo Heavy Industries, Ltd. A schematic diagram of the experimental apparatus is shown in Fig. 1. A reactor with dimensions of φ 1300 × L 500 mm was heated by a LPG burner while rotating at 0.4 rpm. The reactor rotation direction can be changed as shown in Fig. 2. When the reactor is rotated clockwise as viewed from the burner side, the sample in the furnace is indirectly heated by the burner flame, but when rotated counterclockwise, the sample is directly heated by the flame. During the experiment, furnace temperature, CO concentration and CO₂ concentration in the exhaust gas were monitored by a thermocouple and an infrared exhaust gas analyzer, respectively. The LPG flow rate was set to 12 to 15 Nm³/h supplying air and oxygen gas as oxygen sources. The flow rate of the LPG and the flow rates and ratio of air and oxygen gas supplied as an oxygen source were changed so that the furnace temperature reached the target temperature, and the flow rates of the air and pure oxygen were also adjusted so that the CO concentration in the furnace was 10 to 15 vol%. The furnace lid had a raw material charging hole that enabled nitrogen purging, so raw material could be added while preventing air entrainment. A discharge hole was provided in the lower part of the furnace body, making it possible to discharge the sample in the furnace by tapping after the experiment. In the experiment, the molten metal was tapped, and tapping was stopped when the residual slag or charcoal in the furnace began to appear.

Fig. 1.

Experimental apparatus. (Online version in color.)

Fig. 2.

Burner direction during experiment. (Online version in color.)

The chemical compositions of the slag samples used in the experiments are shown in Table 1, and the experimental conditions are shown in Table 2. The dephosphorization slag and the decarburization slag from actual plant operations were mixed so that the slag sample had the prescribed basicity (CaO/SiO₂ wt% ratio). The basicity of slag sample 6 was adjusted by adding silica sand (SiO₂ purity > 95 mass%). The carbon material for reduction was graphite with C purity of 99 mass% or higher, and the blending amount was about 10 times the stoichiometric value necessary for reduction of the (Fe_tO) and (P₂O₅) in the slag. Figure 3 shows the temperature change during the Run-4 experiment, together with the experimental procedure. The furnace was preheated to about 50 K above the experimental temperature, and 5 to 10 kg of carbon material was charged from the raw material charging hole in the furnace lid and burned to form a reducing atmosphere in advance. The time when the CO concentration reached 10% was defined as the experiment start time. The slag sample and reducing carbon were classified so as to have a particle size of 5.0 × 10⁻³ m or less, and were mixed in advance, divided into 10 portions and charged through the raw material charging hole at intervals of 5 min after the start of the experiment. After a predetermined period of time, the burner was extinguished and the sample in the furnace was discharged from the discharge hole in the lower furnace body. The residue in the furnace which was not discharged was recovered by using an iron jig on the next day when the furnace body was cooled. The slag and metal in the samples after the experiment were separated magnetically, and their weights were measured and chemical compositions were analyzed.

Table 1. Chemical compositions of slag for experiments.

Slag sample	Slag composition (mass%)										Slag basicity (%CaO)/(%SiO2)
	CaO	SiO2	Al2O3	MgO	P2O5	MnO	T.Fe	FeO	Fe2O3	M.Fe
1–3	28.72	18.81	5.18	4.35	2.02	4.70	28.50	20.35	10.60	5.27	1.53
4	36.46	16.32	3.43	4.74	2.39	2.77	26.20	12.70	16.14	5.05	2.23
5	26.12	26.57	4.47	3.43	2.67	3.45	26.42	18.71	8.93	5.61	0.98
6	29.76	38.90	2.43	4.64	1.88	1.75	15.11	6.11	13.85	0.67	0.77

Table 2. Experimental conditions.

	Conditions
	Slag sample	Sample weight	Carbon	Temperature		Time	Burner direction
				Target	Actual
Run-1	1–3	50 kg	33 kg	1623 K	1603–1673 K Ave.: 1628 K	90 min	Direct heating
Run-2				1673 K	1643–1753 K Ave.: 1686 K		Indirect heating
Run-3				1623 K	1603–1673 K Ave.: 1632 K
Run-4	4	50 kg	33 kg	1723 K	1648–1773 K Ave.: 1708 K	120 min
Run-5	5	50 kg	33 kg	1673 K	1653–1710 K Ave.: 1683 K
Run-6	6	50 kg	33 kg	1523 K	1444–1561 K Ave.: 1496 K

Fig. 3.

Temperature transition during experiment. (Online version in color.)

3. Experimental Results

3.1. Reduction of (Fe_tO) in Slag

The effects of temperature and slag basicity on the reduction ratio of (Fe_tO) in the slag are shown in Figs. 4 and 5, respectively. As shown in Fig. 4, as the experimental temperature increased, a larger amount of iron oxide was reduced and more metallic Fe was obtained. However, as can be seen in Fig. 5, a comparison of the results of Run-2, 4 and 5, which were carried out at approximately the same experimental temperature, showed that the effect of slag basicity is small.

Fig. 4.

Relationship between temperature and (Fe_tO) reduction ratio.

Fig. 5.

Relationship between slag basicity and (Fe_tO) reduction ratio.

As shown in Figs. 4 and 5, among Run-1 to Run-3, the iron oxide reduction ratio in Run-1 was slightly lower than that in the other runs. The reason for this is considered to be that the LPG did not burn completely and oxidation by combustion air or oxygen gas occurred when the burner flame hit the sample directly in Run-1.

The molten metal could be tapped from the discharge hole in the lower part of the furnace after the experiment, except in the case of Run-6 with the low experimental temperature. As an example, Fig. 6 shows an image of the tapped metal from Run-4, which consisted almost entirely of metal. It is considered that the molten slag and molten metal agglomerated in the furnace due to the rolling of the furnace body, separation proceeded by the difference in specific gravity, and as a result, only the metal, which had a higher specific gravity, was tapped from the discharge hole in the lower furnace body. In other words, the metal can be separated from the slag and collected by reducing the steelmaking slag at a high temperature with a rolling device such as a rotary kiln. Although tapping from the discharge hole was not possible under the low temperature condition in Run-6, the furnace slag recovered after the experiment contained many granular metal particles with a size of several millimeters, as shown in Fig. 7. The reason for this difference is considered to be that both the metal and the slag were not melted under the conditions of Run-6, and aggregation did not progress sufficiently while the metal remained incorporated into the slag.

Fig. 6.

Image of tapped metal (Run-4). (Online version in color.)

Fig. 7.

Image of remained slag with metal droplets (Run-6). (Online version in color.)

3.2. Reduction of (P₂O₅) in Slag

Figure 8 shows the mass balance of phosphorus after the experiment when the weight of phosphorus contained in the slag before the experiment is assumed to be 1. Phosphorus was divided to three phases: a phase that remained in the slag as (P₂O₅), a phase that was captured in the metallic Fe and an undefined component which was different from those two phases. This undefined component is considered to be phosphorous that was vaporized and removed as P₂ gas, as discussed in the previous report.¹¹⁾

Fig. 8.

Mass balance of P in experiments. (Online version in color.)

The effects of temperature and slag basicity on the reduction ratio of (P₂O₅) in slag are shown in Figs. 9 and 10, respectively. Except for the low temperature Run-6, the effect of temperature shown in Fig. 9 is unclear. However, from Fig. 10, it is clear that the reduction ratio of (P₂O₅) increases as the slag basicity decreases.

Fig. 9.

Relationship between temperature and (P₂O₅) reduction ratio.

Fig. 10.

Relationship between slag basicity and (P₂O₅) reduction ratio.

4. Discussion

4.1. Effect of Metallic Weight on Vaporized Dephosphorization

The relationship between the weight of metallic Fe (M.Fe) and the undefined phosphorus ratio after the experiment is shown in Fig. 11. The weight of M.Fe is the value obtained by multiplying the weight of Fe in the slag charged as the sample by the reduction ratio of the iron oxide shown in Figs. 4 and 5, and is the total weight of the tapped metal and the granular iron remaining in the furnace. Here, Run-6 was excluded from Fig. 11 because that run was conducted at lower temperature than the other runs, reduction did not proceed sufficiently, and as a result, the weight of M.Fe after the experiment was as low as about 5 kg. The values in the graph are the mass concentrations of phosphorus in the metal after each experiment.

Fig. 11.

Relationship between weight of M. Fe and undefined ratio of phosphorus.

Comparing Runs-1, -2 and -3, at the same slag basicity, the undefined phosphorus ratio increases with as the weight of M.Fe increases. However, the phosphorus concentration in the metal after the experiment was equivalent at 1.58 to 1.68 mass%, which is contrary to the fact that the unknown phosphorus ratio should decrease as the weight of the metallic Fe increases and the phosphorus contained in the metallic Fe increases, considering the mass balance of phosphorus.

The dissolution reaction¹⁶⁾ of the P₂ gas in the molten iron is represented by the following formula (1), and the standard Gibbs free energy change is represented by formula (2). From formula (2), it is considered that the phosphorus in the metal decreases and the rate of vaporization removal increases as the temperature increases.   

$$\text{1}/\text{2}{\text{P}}_{\text{2}}\left(\text{g}\right)=\underset{\_}{\text{P}}$$(1)

  

$$\mathrm{\Delta }{G}_1^{\circ }=-157\mathrm{\hspace{0.17em} }700+5.4T\left(\text{J}/\text{mol}\right)$$(2)

Because Run-2 was conducted at a higher temperature than Run-1 and Run-3, the undefined phosphorus ratio of Run-2 should be higher than that in the other two runs, but such result was not obtained, as is clear from Fig. 11.

Run-4 was a high basicity, high temperature experiment. The reduction ratio of (P₂O₅) was lower than in the other experiments, and the weight of M.Fe after the experiment was also lower, but the undefined phosphorus ratio was similar. Qualitatively, the undefined phosphorus ratio is expected to decrease under a high basicity condition and increase under a high temperature condition, but it is unclear which condition had a greater effect on Run-4.

Run-5 was carried out under low basicity and high temperature conditions. Since both conditions are qualitatively effective for promoting vaporized dephosphorization, it is reasonable that the undefined phosphorus rate is higher than that in Run-2 with a similar weight of M.Fe.

4.2. Conditions for Vaporized Dephosphorization from Slag

As described in 4.1, the undefined phosphorus ratio cannot be explained only by the weight of M.Fe after the experiment. Therefore, as in the previous report,¹¹⁾ the equilibrium phosphorus distribution ratio L_P between slag and metal was used as an index to show the overall effects of the slag composition, reduction temperature and oxygen potential.

Figure 12 shows the relationship between the undefined phosphorus ratio and the log L_P calculated using Eq. (15) in the previous report.¹¹⁾ There is a good inverse correlation between log L_P and the undefined phosphorus ratio, in which the undefined phosphorus ratio increases as log L_P decreases. Figure 12 also shows the range of the crucible experiment results in the previous report,¹¹⁾ which is in good agreement with the present results. From this fact, it became clear that the undefined phosphorus ratio could also be explained by log L_P in the experiments simulating a rotary kiln.

Fig. 12.

Relationship between logL_P and undefined ratio of phosphorus.

4.3. Effect of Treatment Method on Aggregation of Metals

Figure 13 shows the percentage by weight of the reduced iron collected as metal lumps in the crucible experiment in the previous report¹¹⁾ and in the rotary kiln simulation experiments in the present study. Here, among the experiments in the previous report, the amount of reduced Fe was small in the slag with a low Fe_tO concentration, and the value on the vertical axis of the graph could fluctuate greatly due to a small error in the experiment, so the results for the low Fe_tO concentration slag were excluded. The figure also shows the calculated line of the liquidus ratio of the slag obtained by the thermodynamic calculation software FactSage. The calculated slag compositions were (mass% Al₂O₃) = 8, (mass% MgO) = 6, (mass% MnO) = 6, (mass% FeO) = 1, (mass% Fe₂O₃) = 1, (mass% P₂O₅) = 1, assuming the slag composition after reduction, and (mass% CaO) and (mass% SiO₂) were determined so that the slag basicity was 1.0, 1.5 and 2.0.

Fig. 13.

Aggregation ratio of reduced iron.

As can be understood from Fig. 13, it is difficult to quantitatively examine the difference in the metal aggregation behavior between the crucible experiments and the kiln simulation experiments. Qualitatively, however, as the basicity C/S of the slag decreases, the aggregation ratio of the reduced metal increases. This increase in the aggregation ratio is thought to occur because the low melting point of the slag after reduction and the high liquidus ratio facilitated metal transfer in the slag and thereby promoted aggregation of the metal. Figure 13 shows the calculation result at a low Fe_tO concentration assuming the condition after reduction. However, even if the same calculation is performed up to a Fe_tO concentration of 30 mass% assuming the condition before treatment, the calculated liquidus ratio increases as basicity decreases at the same temperature.

Based on this fact, it is considered that the percentage of slag which was melted or semi-melted at high temperature before and after reduction affected metal aggregation.

4.4. Effect of Treatment Method on Separability of Iron and Phosphorus

The relationship between the calculated log L_P using Eq. (15) in the previous report,¹¹⁾ as in section 4.2, and the actual log L_P calculated from the phosphorus concentration in the metal and slag after the experiments is shown in Fig. 14, which also shows the results of the crucible experiment in the previous report.¹¹⁾

Fig. 14.

Relationship between calculated L_P and L_P from experimental results.

As can be seen from Fig. 14, in the kiln simulation experiments, the actual L_P tends to be larger than the calculated L_P, whereas the crucible experiments showed the opposite tendency. Two possible reasons why the actual L_P is larger than the calculated L_P are conceivable, as follows:

(1) Reduction of (P₂O₅) was insufficient and a slag/metal equilibrium was not reached.

(2) Slag/metal separation proceeded before reduction of (P₂O₅), decreasing the slag/metal interface and thereby suppressing phosphorus transfer to the metal.

As for possible reason (1), as shown in Figs. 9 and 10, more than 40% of (P₂O₅) was reduced except in Run-6, and in particular, 96% of (P₂O₅) was reduced in Run-5. Therefore, it is difficult to think that (P₂O₅) reduction was insufficient compared with the crucible experiment.

Regarding possible reason (2), when transfer of phosphorus from the slag to the metal is suppressed, removal of phosphorus from the slag to the gas phase should be accelerated. This is consistent with the fact that the undefined phosphorous ratio increases as the calculated log L_P decreases, as described in 4.2.

On the other hand, under the conditions of low T.Fe concentration and low basicity in the initial slag in the crucible experiment (◇ in Fig. 14), the actual L_P is about two orders of magnitude smaller than that in the kiln experiments, in spite of the calculated L_P being equivalent to that in the kiln experiments. This large difference is thought to have occurred because the phosphorus concentration in the metal increased to over 10 mass% due to the low T.Fe concentration of the initial slag and the (P₂O₅) reduction ratio of approximately 40 to 80%.

Therefore, it is considered that high temperature reduction of steelmaking slag in equipment where vertical agitation by sample rolling occurs, such as a rotary kiln, reduces the slag/metal interface by advancing slag/metal separation before reduction of (P₂O₅), and this suppresses the transfer of phosphorus to the metal and promotes the removal of phosphorus to the gas phase.

Here, it is necessary to pay attention to the handling of the calculated L_P. As described in detail in the previous report, the calculated L_P used in this study was calculated from the oxygen partial pressure obtained from the phosphate capacity equation proposed by Suito et al.,¹⁷⁾ the standard Gibbs free energy change of the dissolution reaction of P₂ gas into molten iron¹⁶⁾ and the Fe_tO activity calculated by a regular solution model.¹⁸⁾ Since the phosphate capacity according to Suito et al. was measured at a molten steel temperature of 1823 to 1923 K, it is uncertain whether that value can be applied to the experimental temperature range of 1473 to 1708 K in this study.

The phosphorus distribution ratio between slag and metal has been measured^19,20,21,22) at 1473 to 1673 K, which is the temperature range used in hot metal pretreatment. However, since CaO–SiO₂–Fe_tO slag has a wide solid-liquid coexistence region in this temperature range, the data are limited to a composition with a fully molten Fe_tO concentration of more than about 20 mass% or a slag system in which CaF₂ is added as a flux.

Although the phosphorus distribution data cannot be applied directly to the conditions of this study, the following two assumptions were examined, and assumption (1) was adopted in this study because its calculated L_P and actual L_P are more consistent than under assumption (2).

(1) Extrapolation of the measured data at the molten steel temperature and low Fe_tO concentration reported by Suito et al.¹⁷⁾ to the low temperature condition in this study.

(2) Extrapolation of the measured data at the hot metal temperature and high Fe_tO concentration reported by Iwasaki et al.^19,20,21,22) to the low Fe_tO concentration in this study.

5. Conclusion

In order to evaluate the separation behavior of iron and phosphorus from steelmaking slag, experiments with a 50 kg scale simulating a rotary kiln were carried out, and the following results were obtained.

(1) By applying high temperature reduction of steelmaking slag at a slag basicity = 1.0 to 2.3 and temperature of 1630 to 1710 K, it was possible to reduce more than 87% of the (Fe_tO) in the slag and tap the metallic Fe from the discharge hole in the lower part of furnace body.

(2) As the weight of metallic Fe increased after the experiment, the amount of phosphorus in the metal increased and the proportion of phosphorus removed by evaporation decreased.

(3) The equilibrium phosphorus distribution ratio L_P was used as an index to show the overall effects of the slag composition, reduction temperature and oxygen potential. As in the previous report, the ratio of phosphorus removed by evaporation increased with decreasing log L_P.

(4) In comparison with the previous crucible experiments, the calculated L_P was smaller than the actual L_P in the rotary kiln simulation experiments. This difference is attributed to the fact that slag/metal separation proceeded before reduction of (P₂O₅), and as a result, the slag/metal interface decreased and phosphorus transfer to the metal was suppressed.

Acknowledgements

Part of this research was carried out with a grant from the New Energy and Industrial Technology Development Organization (NEDO) in 2009, “Preliminary Research on Phosphorus Separation and Recovery for Steelmaking Slag Recycling Technology.”

References

• 1)   H.  Yahata,  Y.  Kurose,  H.  Okuda,  K.  Oshima and  A.  Aoyagi: CAMP-ISIJ, 8 (1995), 1101 (in Japanese).
• 2)   M.  Kimura,  S.  Nakajima,  T.  Mimura,  I.  Hoshikawa,  R.  Ono and  K.  Semura: Kobe Steel Eng. Rep., 51 (2001), 50 (in Japanese).
• 3)  Development Trends and Subjects for Steelmaking Slag Minimization: Final Report of Steelmaking Slag Minimization Working Group, ISIJ, Tokyo, (1999), (in Japanese).
• 4)   S.  Shiomi,  N.  Sano and  Y.  Matsushita: Tetsu-to-Hagané, 63 (1977), 1520 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.63.9_1520
• 5)   S.  Takeuchi,  N.  Sano and  Y.  Matsushita: Tetsu-to-Hagané, 66 (1980), 2050 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.66.14_2050
• 6)   K.  Nagata: Reduction of Steelmaking Slag Generation and Utilization of Steelmaking Slag, Final Report of Working Group for Basic and Applied Research for Steelmaking Slag, ISIJ, Tokyo, (1997), 49 (in Japanese).
• 7)   K.  Morita,  M.  Guo,  N.  Oka and  N.  Sano: J. Mater. Cycles Waste Manage., 4 (2002), 93.
• 8)   S.  Toishi and  K.  Itoh: CAMP-ISIJ, 17 (2004), PS-1 (in Japanese).
• 9)   H.  Kubo,  K.  Matsubae-Yokoyama and  T.  Nagasaka: Tetsu-to-Hagané, 95 (2009), 300 (in Japanese). https://doi.org/10.2355/tetsutohagane.95.300
• 10)   A.  Matsui,  K.  Nakase,  N.  Kikuchi,  Y.  Kishimoto,  K.  Takahashi and  K.  Ishida: Tetsu-to-Hagané, 97 (2011), 416 (in Japanese). https://doi.org/10.2355/tetsutohagane.97.416
• 11)   K.  Nakase,  A.  Matsui,  N.  Kikuchi and  Y.  Miki: Tetsu-to-Hagané, 102 (2016), 485 (in Japanese). https://doi.org/10.2355/tetsutohagane.TETSU-2016-006
• 12)   T.  Harada,  H.  Hirata,  T.  Arai,  T.  Toh and  T.  Yamada: ISIJ Int., 58 (2018), 1934. https://doi.org/10.2355/isijinternational.ISIJINT-2018-188
• 13)   T.  Harada,  H.  Hirata,  T.  Arai,  T.  Toh and  C.  Shuto: ISIJ Int., 58 (2018), 1943. https://doi.org/10.2355/isijinternational.ISIJINT-2018-189
• 14)   I.  Tetsuyama: Sumitomo Heavy Ind. Tech. Rev., (2005), No. 158, 19 (in Japanese).
• 15)   I.  Tetsuyama and  I.  Kawakami: Sumitomo Heavy Ind. Tech. Rev., (2008), No. 167, 5 (in Japanese).
• 16)  The 19th Committee on Steelmaking, The Japan Society for the Promotion of Science: Steelmaking Data Sourcebook Revised Edition, JSPS, Tokyo, (1984), 30 (in Japanese).
• 17)   H.  Suito and  R.  Inoue: ISIJ Int., 35 (1995), 258. https://doi.org/10.2355/isijinternational.35.258
• 18)   S.  Ban-ya: ISIJ Int., 33 (1993), 2. https://doi.org/10.2355/isijinternational.33.2
• 19)   K.  Iwasaki,  N.  Sano and  Y.  Matsushita: Tetsu-to-Hagané, 67 (1981), 536 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.67.3_536
• 20)   K.  Ito and  N.  Sano: Tetsu-to-Hagané, 69 (1983), 1747 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.69.15_1747
• 21)   M.  Muraki,  H.  Fukushima and  N.  Sano: Tetsu-to-Hagané, 71 (1985), 693 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.71.6_693
• 22)   J.  Im,  K.  Morita and  N.  Sano: ISIJ Int., 36 (1996), 517. https://doi.org/10.2355/isijinternational.36.517