Phosphorus Distribution Behavior of Solid Iron Reduced from Molten Al₂O₃–CaO–Fe_tO–MgO–SiO₂ System at 1623 K

Abstract

During conventional iron making, the temperature exclusively determines the oxygen potential in a blast furnace hearth because of carbon saturation. Consequently, impurities such as phosphorus are reduced and remain in the iron because of the excessively low oxygen partial pressure, which can be controlled using gaseous reductants such as hydrogen and carbon monoxide to produce solid iron because of the low carbon content. We previously investigated the equilibrium distribution of phosphorus between the solid iron and molten slag at 1623 K by varying the oxygen partial pressure and the basicity of the slag, and the phosphorus content in the solid iron was sufficiently low under the experimental conditions. In this study, the phosphorus distribution behavior was investigated when solid iron was obtained by reducing Al₂O₃–CaO–Fe_tO–MgO–SiO₂ molten oxide, and the system was evaluated based on Fe loss.

1. Introduction

Recent integrated steelmaking can be roughly divided into reduction and oxidation processes. An iron ore is first reduced to hot metal in a blast furnace (BF) hearth, and impurities such as C, Si, P, and S are then removed using hot-metal pretreatments, converters, etc. during steelmaking. Iron ore is reduced using cokes, and the hot metal and cokes then coexist in the BF hearth. Therefore, the working temperature determines the oxygen partial pressure in the hearth on the basis of C/CO equilibrium because molten iron is produced under carbon saturation. The hearth temperature is ~1873 K, and so the oxygen partial pressure in the hearth is ~10^–16 atm.¹⁾ Because this is too low, the iron oxide and the impurities in the iron ore are reduced. Consequently, the hot metal contains impurities, which must subsequently be removed. If the oxygen partial pressure could be controlled during reduction, the iron oxide could be reduced without reducing the impurities. Gaseous reductants such as CO and H₂ can be used to reduce only the iron oxide and not the impurities because the oxygen partial pressure can be controlled by changing the gas composition. Thus, using gaseous reductants might enable the production of steel containing fewer impurities without requiring refining. We previously proposed and named such a process “Direct Steelmaking.^1,2)”

Kato et al. proposed a new carbon-recycling-based energy transformation called the “Active Carbon Recycling Energy System” (ACRES)^3,4,5,6,7) as a zero-carbon-dioxide-emission process. In ACRES, energy sources such as iron-steelmaking waste heat, heat from a high-temperature gas-cooled reactor (HTGR), etc., which do not emit CO₂, are used to convert CO₂ into carbon monoxide. In other words, available carbon-dioxide-free primary energy essentially drives ACRES, which can potentially solve the problems of CO₂ emission and primary energy. Combining Direct Steelmaking and ACRES is suitable because ACRES can produce CO by reducing CO₂.

The proposed direct steelmaking method involves first converting iron ore into molten iron oxide by adding CaO to decrease the melting point of the oxide and produce a suitably basic environment. The molten oxide is then reduced using a gaseous reductant. This process is similar to smelting reduction such as the direct iron ore smelting (DIOS) reduction.⁸⁾ The main difference is that in the proposed direct steelmaking, the reduction proceeds under a controlled atmosphere. Furthermore, the obtained reduced iron is solid because it does not contain any carbon, so its melting point is higher than that of the hot metal produced during DIOS or in conventional BF systems. Therefore, the process temperature must be maintained above the melting point of iron so that the iron remains molten, which is unrealistic because heat loss and refractory corrosion both increase with increasing process temperature.

Kashiwaya et al. reported the behavior of impurities when iron ore was reduced using hydrogen or carbon in the range 1173–1373 K as part of direct steelmaking.⁹⁾ The iron ore was first reduced below the melting point of iron, and the reduced iron was then melted at 1873 K in the same crucible without removing the slag. The silicon and manganese contents of the iron reduced using hydrogen were lower than those of the iron reduced using carbon; however, the difference between the phosphorus contents of the iron reduced using hydrogen or carbon was insignificant, indicating that the phosphorus had readily moved to the metal phase during remelting.

Many researchers have reported the influences of slag composition, temperature, and oxygen partial pressure on the equilibrium distribution ratio (L_P) of impurities between molten slag and molten steel.^10,11,12,13) A low temperature, high oxygen partial pressure, and high CaO/SiO₂ ratio are advantageous for steel dephosphorization. Thus, low-temperature processing has many advantages from the perspectives of heat loss, refractory corrosion, and dephosphorization. During direct steelmaking, the reduction temperature is below the melting point of iron, so solid iron is produced. We previously investigated the effects of slag composition and oxygen partial pressure on the equilibrium L_P of phosphorus between solid iron and magnesiowüstite-saturated Al₂O₃–CaO–Fe_tO–MgO–SiO₂ molten slag at 1627 K.²⁾ The L_P increased with increasing oxygen partial pressure and oxide T-Fe content and basicity. Further, the L_P used during BF operation was sufficiently large to produce solid iron without reducing the phosphorus. However, the phosphorus distribution behavior during reduction remains unclear, and it is difficult to analyze the impurity content of the metal dispersed in the oxide phase. Therefore, we have investigated the temporal variation in the phosphorus distribution of the molten oxide and of the solid iron produced by reducing the molten oxide. In the experiment, Al₂O₃–CaO–Fe_tO–MgO–P₂O₅–SiO₂ slag was melted at 1623 K in an iron crucible and subsequently reduced the slag with CO. A method of analyzing the phosphorus content of the metal dispersed in the oxide phase was also developed.

2. Experimental

2.1. Experimental Setup and Procedure

Figure 1 shows a schematic diagram of the experimental apparatus used in this study. The experiments were performed in an electrical resistance furnace equipped with a SiC spiral heating element and a mullite reaction tube (inner diameter (ID) = 42 mm). Reagent-grade Al₂O₃, CaO, 3CaOP₂O₅, Fe, Fe₂O₃, MgO, and SiO₂ were mixed and premelted in an iron crucible under Ar at 1623 K for 1 h. The molten mixture was then quenched in water to prepare 100 g of the oxide system. CaO was then prepared by thermally decomposing reagent-grade CaCO₃. The premelted oxide (30 g) was placed in the furnace in an iron crucible (ϕ26 mm ID, ϕ28 mm OD × 43 mm H) and heated to 1623 K. Ar was flowed for 30 min into the furnace tube from the bottom cap to flush the atmosphere, and then CO (>99.95%) was then flowed at 50 mL/min from the top of the crucible through the mullite tube, whose nozzle was set 3 cm above the slag surface. After 1–10 h had elapsed, the crucible was removed from the furnace and quenched with water.

Fig. 1.

Schematic diagram of experimental apparatus.

The experimental conditions are summarized in Table 1. The oxide composition was basically determined from the composition of the iron ore. CaO was added to set the basicity such that (%CaO)/(%SiO₂) = 1.0. The %FeO of samples 1 and 2 were determined from the Fe content of the iron ore. The %P₂O₅ of sample 1 was set to 1.0 mass% to increase the analytical sensitivity and simulate a high-phosphorus-content iron ore. The %P₂O₅ of sample 2 was set to 0.1 mass%, which is the common content of iron ore. %FeO decreased as the iron ore reduced. The %P₂O₅ contents of samples 3–7 and samples 8 and 9 were determined by reducing samples 1 and 2, respectively, with CO for 1–10 h and subsequently quenching them. Cross-sections of the samples were observed, and the composition of each phase was analyzed using electron probe microanalysis (EPMA). The slag was analyzed after the fine iron particles were removed using Br-methanol. The SiO₂ content of the slag was determined using gravimetric analysis, and the other impurity contents were determined using inductively coupled plasma atomic emission spectroscopy (ICP-AES).

Table 1. Experimental conditions.

No.	Composition (mass%)						Reduction time (h)
	FeO	CaO	SiO2	Al2O3	MgO	P2O5
1	85.9	5.86	5.86	1.29	0.11	1.00	1, 2, 3, 4
2	86.7	5.91	5.91	1.30	0.11	0.10	4
3	3.00	40.3	40.3	8.85	0.74	6.88	10
4	5.00	39.4	39.4	8.67	0.72	6.74	10
5	10.0	37.4	37.4	8.21	0.68	6.38	10
6	20.0	33.2	33.2	7.30	0.61	5.67	10
7	30.0	29.1	29.1	6.39	0.53	4.96	10
8	10.0	39.9	39.9	8.77	0.73	0.66	10
9	30.0	31.1	31.1	6.82	0.57	0.51	10

2.2. Method of Analyzing Phosphorus Content of Metallic Iron Dispersed in Oxide Phases

The metallic iron (m-Fe) produced by reducing molten iron oxide was dispersed in the oxide phases. Although EPMA is useful for determining the compositions of dispersed phases, it is not analytically sensitive enough if the phosphorus content of the m-Fe dispersed in the slag phase is <0.1 mass%. ICP-AES is usually used to chemically analyze the microstructures of crushed samples and determine the average composition of several phases. In this study, the content of the m-Fe dispersed in the slag was low under optimal conditions. It is difficult to completely separate the m-Fe from the oxide phase, and the oxide could significantly affect the chemical analysis of phosphorus content in m-Fe. Therefore, conventional chemical analysis by ICP-AES is unsuitable, so an analysis method was developed as follows:¹⁴⁾

1) The quenched sample was roughly crushed and removed from the crucible.

2) The sample was ground in an agate mortar.

3) A magnet was used to separate the iron particles from the sample.

4) The iron particles were leached using a 1 mass% citric acid solution for 5 min to remove the slag on them. (Citric acid is conventionally used to remove slag.^15,16,17)) The experimental conditions were determined by a pre-experiment. The acid-leached sample was vacuum-filtered from the solution and dried.

5) The dried sample and filter were dipped into a 10 mass% Br-methanol solution for 20 min to dissolve only the m-Fe. Oxide may have affected the analysis if the separated iron was analyzed simply by dissolving it into an acidic solution, because the reduced iron may have contained slag particles.

6) The solution was filtered and the filtrate was heated to evaporate the residual solvent leaving a solid, which was redissolved in diluted aqua regalis. The iron and phosphorus contents of the solution were analyzed using ICP-AES with an ultrasonic nebulizer and standard addition¹⁸⁾ to improve the analytical sensitivity and accuracy, respectively. The phosphorus content was determined from the ratio of iron to phosphorus.

3. Results

3.1. Chemical Analysis

Table 2 shows the slag composition and reduced iron phosphorus content. The basicity (C/S) and distribution ratio of phosphorus (L_P), defined by Eqs. (1) and (2), respectively, are also listed.   

$$\mathrm{C}/\mathrm{S}=\left(\%\mathrm{C}\mathrm{a}\mathrm{O}\right)/\left(\%\mathrm{S}\mathrm{i}{\mathrm{O}}_2\right)$$(1)

  

$$L_{\text{P}}=\frac{\left(\%\text{P}\right)}{\left[\%\text{P}\right]}=0.44\frac{\left(\%{\text{P}}_{\text{2}}{\text{O}}_{\text{5}}\right)}{\left[\%\text{P}\right]}$$(2)

where (%M) and [%M] represent the M contents of the slag and metal, respectively. Samples 3–6 were vertically sliced and several slices were chemically analyzed using ICP-AES, giving the average compositions of the slag and metal phases. The oxide FeO content was calculated by converting the T-Fe not containing any m-Fe.

Table 2. Compositions of slag and metal samples analyzed using ICP-AES after experiment.

No.	Reduction time [h]	Slag composition [%]							P conc. in metal [%]	Lp [–]
		FeO	CaO	SiO2	Al2O3	MgO	P2O5	C/S
1	1	85.4	6.0	5.8	1.33	0.13	1.10	1.03	0.010	49.8
1	2	87.8	4.9	5.0	1.18	0.13	0.88	0.99	0.034	11.2
1	3	86.8	5.7	6.0	1.32	0.16	0.98	0.95	0.013	32.6
1	4	85.7	5.9	6.0	1.34	0.14	0.98	0.98	0.016	26.3
2	4	85.3	7.1	7.2	1.55	0.14	0.16	0.98	0.002	39.4
3-1	10	7.1	40.2	40.7	7.90	0.73	5.10	0.99	0.642	3.5
3-2	10	7.1	39.8	39.8	8.19	0.69	4.53	1.00	0.795	2.5
4-1	10	8.5	40.3	39.1	7.81	0.74	5.36	1.03	0.230	10.2
4-2	10	7.8	39.7	38.7	8.15	0.69	4.90	1.02	2.683	0.8
5-1	10	12.4	38.8	38.4	7.74	0.72	5.79	1.01	0.641	3.9
5-2	10	11.9	37.7	36.7	7.70	0.66	5.47	1.03	0.536	4.5
5-3	10	11.1	36.7	37.9	7.76	0.63	5.11	0.97	0.625	3.6
6-1	10	17.7	35.3	34.4	6.95	0.64	5.37	1.03	0.145	16.1
6-2	10	18.7	33.9	33.0	6.84	0.60	5.13	1.03	0.088	25.4
7	10	27.9	30.2	29.9	6.32	0.52	4.82	1.01	0.022	95.6
8	10	10.5	40.4	39.1	8.23	0.67	0.61	1.03	0.031	8.6
9	10	26.4	33.0	32.5	6.88	0.57	0.52	1.01	0.003	86.3

Figure 2 shows the FeO contents of charged and after the reduction, both of which were almost identical. The significant reduction-induced decrease in FeO was not observed under this experimental condition because the reduction did not proceed very fast. Moreover, although the FeO was reduced, the post-reduction FeO content of the oxide containing <20% FeO increased because the iron crucible had supplied Fe when the oxide was premelted.

Fig. 2.

Comparison of charged FeO content and analyzed FeO content after the reduction experiment.

Figure 3 shows the changes in the slag FeO content during reduction. It did not significantly change with time under the experimental conditions, indicating that it is difficult to change the FeO content of the oxide by drastically reducing the original oxide, although the oxide FeO content might affect the phosphorus content of the m-Fe. Therefore, the lower-FeO-content oxides (i.e., samples 3–9) were used.

Fig. 3.

Changes in FeO content of slag plotted as function of reduction time.

Figure 4 shows the relation between the phosphorus content of the iron and the FeO content of the oxide reduced for 10 h. The phosphorus content increased with decreasing FeO content. The phosphorus content of the iron reduced from the 1.0% P₂O₅ oxide is higher than that of the iron reduced from the 0.1% P₂O₅ oxide.

Fig. 4.

Relation between P content in metal and FeO content in oxide reduced for 10 h.

Figure 5 shows the relation between the phosphorus distribution ratio (L_P) and the FeO content of the oxide for the solid iron and molten oxide reduced for 10 h. L_P increased with increasing FeO content of the oxide. The oxide P₂O₅ content was set to 1.0% to increase the analytical sensitivity, although iron ore typically contains 0.1% P₂O₅; indicating that the experimental conditions were suitable because L_P did not significantly vary with the oxide P₂O₅ content.

Fig. 5.

Relation between phosphorus distribution ratio between solid iron and oxide reduced for 10 h plotted as functions of FeO content.

3.2. Phase Observation

Figure 6 shows compositional (COMP) images of sample 1 at different stages of reduction. The white, dark, and black phases show m-Fe, oxide, and resin, respectively. The continuous outer white area represents the iron crucible, and the central black phase represents a void that had formed during quenching. The reduced m-Fe clearly formed particles, whose number increased with time. The iron particles reduced more than 3 h coalesced into a continuous m-Fe layer at the gas-oxide interface.

Fig. 6.

COMP images of sample 1 reduced for different lengths of time.

Figure 7 shows the increase in the m-Fe phase ratio with time, as analyzed using image analysis software (Image-PRO plus). The ratio increased more slowly with time because the iron oxide was reduced at the molten-oxide/gas interface. A continuous m-Fe layer had formed at the interface, Fig. 6, preventing the mass transfer of oxygen between the oxide and gas phases.

Fig. 7.

Change in the ratio of metallic iron phase with time.

Figure 8 shows COMP images of samples 3–7 reduced for 10 h. Sample 7, containing 30% FeO, showed m-Fe particles dispersed in the oxide phases. Samples 5 and 6 (containing 10 and 20% FeO, respectively) showed m-Fe particles near the oxide/gas interface; however, the samples containing <5% FeO showed >2.5-mm-diameter droplets near the gas/oxide interface.

Fig. 8.

COMP images of samples 3–7 reduced for 10 h.

Figure 9 shows COMP images of sample 1, reduced for 3 h, as analyzed by EPMA. The upper and lower regions of the crucible show m-Fe (white), FeO (gray), and matrix (black) phases. The middle region showed only the FeO and matrix phases. Samples 1 and 2, which contained >80 mass% FeO, show the same trend; however, samples 3–9, which contained <30% FeO, showed only the m-Fe (white) and matrix (gray) phases, as shown in Fig. 10. Thus, the FeO phase disappeared first during reduction.

Fig. 9.

COMP images of sample 1 reduced for 3 h.

Fig. 10.

COMP images of samples 3–9 at the interface between gas and oxide phases reduced for 10 h.

Table 3 shows the post-reduction compositions of the phases, as analyzed by EPMA. The upper, middle, and lower regions of the samples were separately analyzed. The compositions in the table were averaged for at least 3 points. Only samples 1 and 2 showed FeO phases, and “–” means that a particular phase was not observed. The relative magnitude of each value was reliable, although the values <0.1% were inaccurate because of the analytical sensitivity of EPMA. The phosphorus was mainly in the oxide matrix phase, not in the oxide FeO one.

Table 3. Composition of each reduced phase analyzed using EPMA.

No.	Red. time (h)		P in Metal iron	Matrix phase						FeO phase
				FeO	CaO	SiO2	Al2O3	MgO	P2O5	FeO	CaO	SiO2	Al2O3	MgO	P2O5
1	0	Up	0.00	28.0	28.8	30.3	5.22	0.10	5.60	94.5	0.32	0.07	0.39	0.16	0.02
		Mid	0.01	28.5	27.7	31.8	6.29	0.30	3.70	95.7	0.23	0.03	0.39	0.23	0.01
		Low	0.00	29.3	28.5	30.0	5.30	0.15	4.97	95.9	0.17	0.03	0.33	0.28	0.03
	1	Up	0.02	31.8	26.1	27.8	4.94	0.15	4.64	94.6	0.24	0.07	0.47	0.10	0.04
		Mid	0.03	27.8	28.4	29.3	5.60	0.09	3.33	94.7	0.12	0.08	0.40	0.15	0.00
		Low	0.00	28.5	27.3	27.3	5.29	0.19	3.27	93.8	0.11	0.07	0.40	0.15	0.03
	2	Up	0.02	30.2	26.2	27.8	5.15	0.21	4.20	95.1	0.18	0.04	0.44	0.18	0.02
		Mid	–	28.2	27.3	29.6	5.37	0.18	3.51	93.1	0.14	0.10	0.41	0.15	0.00
		Low	0.03	40.1	22.7	25.1	4.63	0.13	2.91	92.9	0.15	0.05	0.45	0.17	0.01
	3	Up	0.03	29.3	27.4	29.2	5.18	0.16	5.24	95.5	0.12	0.09	0.42	0.19	0.03
		Mid	0.01	26.0	28.1	31.0	5.23	0.22	5.37	94.1	0.13	0.04	0.42	0.16	0.00
		Low	0.03	26.4	29.5	31.9	5.35	0.15	4.69	95.1	0.11	0.07	0.45	0.17	0.02
	4	Up	0.02	36.6	24.4	26.3	4.86	0.26	4.81	95.2	0.35	0.10	0.49	0.20	0.06
		Mid	–	25.4	29.0	30.7	5.31	0.17	5.48	92.7	0.15	0.08	0.44	0.09	0.00
		Low	0.07	29.3	27.6	29.9	5.26	0.06	4.10	93.1	0.12	0.09	0.47	0.14	0.01
2	4	Up	0.00	25.2	31.4	30.8	6.20	0.08	0.71	93.6	0.30	0.01	0.43	0.15	0.01
		Mid	0.01	27.5	30.7	30.4	5.92	0.09	0.70	93.6	0.19	0.05	0.38	0.17	0.01
		Low	0.00	33.8	27.1	28.0	5.15	0.06	0.66	94.3	0.27	0.05	0.39	0.11	0.00
3	10	Up	3.04	6.1	38.0	37.7	8.36	0.72	5.60	(mass%)
		Mid	3.40	6.7	37.6	37.4	8.33	0.71	5.47
		Low	2.25	7.2	37.5	37.2	8.38	0.72	4.60
4	10	Up	3.60	6.4	37.1	37.6	8.23	0.71	6.09
		Mid	2.02	7.5	36.7	37.7	8.24	0.71	5.83
		Low	0.27	8.4	36.5	37.1	8.15	0.73	5.18
5	10	Up	0.35	11.3	34.6	36.3	7.92	0.68	6.21
		Mid	0.27	10.7	34.9	35.7	7.93	0.68	6.48
		Low	0.26	11.6	34.7	35.0	7.94	0.69	6.30
6	10	Up	0.12	14.9	32.7	34.7	7.49	0.62	6.33
		Mid	0.03	16.2	33.2	33.2	7.38	0.62	6.15
		Low	0.03	18.3	32.3	32.0	7.14	0.61	6.07
7	10	Up	0.04	16.8	32.2	33.0	7.04	0.63	6.23
		Mid	0.02	22.5	29.9	30.5	6.60	0.62	5.83
		Low	0.00	28.9	28.0	27.8	6.46	0.55	5.00
8	10	Up	0.02	6.5	37.2	40.1	9.45	0.81	0.76
		Mid	0.01	9.0	36.5	38.4	9.01	0.76	0.74
		Low	0.03	9.8	36.4	37.5	8.25	0.70	0.68
9	10	Up	0.01	18.4	33.5	35.4	7.52	0.63	0.65
		Mid	0.01	25.6	30.2	31.7	7.02	0.66	0.55
		Low	0.02	27.7	29.6	30.4	6.81	0.55	0.51

Figure 11 shows the relation between the charged FeO content and the FeO content of the matrix analyzed at different positions by EPMA. Samples 3–9 were used because they only contained m-Fe and matrix phases, i.e., the FeO phase was not observed. The FeO content of the upper phase was lower than those of the middle and lower phases, indicating that the reduction had proceeded from the upper to the lower phase.

Fig. 11.

Relation between charged and matrix FeO contents.

Figure 12 shows the relation between the charged P₂O₅ content and the P₂O₅ content of the matrix, as analyzed by EPMA. When the charged P₂O₅ content was <6%, the matrix P₂O₅ content was higher than the charged P₂O₅ content, indicating that the phosphorus was mainly distributed in the matrix. When the charged P₂O₅ content was >6%, however, the opposite trend was observed. The charged P₂O₅ content was varied by varying the oxide charged FeO content to simulate various degrees of oxide reduction. Figure 13 shows the relation between the iron P content and the charged FeO content. The P content increased with decreasing charged FeO content. The P contents of the upper region were higher than those of the other regions, i.e., the reduction proceeded from the upper to the lower phase. The P was reduced when the FeO content was <10%. Consequently, the P₂O₅ content of the matrix decreased when the charged P₂O₅ content was >6%, as shown in Fig. 12.

Fig. 12.

Relation between charged and matrix P₂O₅ contents.

Fig. 13.

Relation between P content in metal and charged FeO content.

4. Discussion

4.1. Phosphorus Distribution Ratio between Solid Iron and Molten Oxide Compared with Equilibrium Value

Figure 14 shows the relation between the distribution ratio of phosphorus—between the solid iron and molten slag (L_P)—and the FeO content in the slag compared with the equilibrium values of L_P reported by Maruoka et al.²⁾ and Im et al.¹⁷⁾ Note that these values were obtained using ICP-AES, i.e., the phosphorus content of the m-Fe over the whole area was analyzed, and these values are the averages for the upper, middle, and lower regions of the crucible. The equilibrium L_P increased with increasing FeO content and C/S in both this and the equilibrium experiments. The L_P obtained in this study was clearly lower than the equilibrium L_P for C/S ≈ 1.0 because the iron had not been uniformly reduced in the crucible, as shown in Fig. 8 and Table 3.

Fig. 14.

Comparison of L_P with equilibrium as functions of FeO content of slag.

Tagaya et al. reported that the predominant species of phosphorus in oxide phases changed from PO₄^3– to P₂O₇^4– in oxide melts containing ~2 mass% phosphorus,¹⁹⁾ as expressed in the following equations:   

$$\left(\%\mathrm{P}\right)<2\mathrm{\hspace{0.17em} }\mathrm{m}\mathrm{a}\mathrm{s}\mathrm{s}\%:\hspace{1em}\text{[P]+}\frac{\text{5}}{\text{4}}{O}_2+\frac{3}{2}{O}^{2-}=P{O}_4^{3-}$$(3)

  

$$\left(\%\mathrm{P}\right)>2\mathrm{\hspace{0.17em} }\mathrm{m}\mathrm{a}\mathrm{s}\mathrm{s}\%:\hspace{1em}\text{2[P]+}\frac{\text{5}}{\text{2}}{O}_2+2O^{2-}=P_2{O}_7^{4-}$$(4)

In this study, the phosphorus partition should have reacted following Eq. (4) because the P₂O₅ contents of samples 3–7 were 4.5–5.7 mass%, as shown in Table 2. Figure 15 shows the relation between the local L_P^* (= (%P)/[%P]²) and the matrix FeO contents of samples 3–7, which were analyzed by EPMA. The local L_P^* was calculated by dividing the phosphorus content of the matrix phase by the square of the phosphorus content of the m-Fe phase, both of which were analyzed using EPMA. The relation was linear for all the positions in the crucible, indicating that the phosphorus distribution had locally reached equilibrium and was controlled by the oxide phase FeO content. Oxide phase mixing is important because the phosphorus content of the iron in the lower region was lower than that of the iron in the upper one, as shown in Figs. 12 and 13.

Fig. 15.

Relation between local L_P* and FeO content in matrix phase, analyzed by EPMA.

Higher-phosphorus-content m-Fe particles were at the oxide/gas interface because the iron was reduced there. Therefore, the phosphorus concentration was distributed and low- and high-phosphorus-content m-Fe coexisted in the crucible. Figure 16 shows the simulation results for how high-phosphorus-content m-Fe affected the overall L_P. The conditions are listed in Table 4. Most of the m-Fe was assumed to be low-phosphorus-content m-Fe (high L_P metal) whose local L_P was 1000; 0–30 mass% of the high-phosphorus-content m-Fe (low L_P metal), whose L_P was in the range 1–500, coexisted with the low-phosphorus-content m-Fe, indicating that the overall L_P decreased with increasing existence ratio for the low-L_P metal. This trend was enhanced by further decreasing the L_P of the low-L_P metal, indicating that L_P drastically decreases overall when metals are excessively locally reduced and that homogeneous reduction is important.

Fig. 16.

Influence of high phosphorus content metallic particles on overall L_P.

Table 4. Simulation conditions for overall L_P.

Local LP for low P content in metal	1000	[–]
Local LP for high P content in metal	1–500	[–]
Existing ratio of metal with high P content	0–30	[mass%]
Metal/Oxidel ratio	10.0	[–]
(%P2O5) in oxide	5.0	[mass%]

4.2. Iron Droplet Sedimentation

m-Fe formed at the gas/oxide interface during oxide reduction. Figures 6 and 8 show that a continuous m-Fe layer had formed at the gas/oxide interface, >1-mm m-Fe droplets had formed at the gas/oxide interface, and small m-Fe droplets had dispersed in the oxide phase, depending on the experimental conditions. Some researchers used CO to reduce FeO and reported a similar trend.^20,21) The m-Fe phase should settle at the bottom of the oxide phase because m-Fe is denser than the molten oxide. The terminal settling velocity of a m-Fe droplet in the static slag phase was estimated using Stokes’ law:   

$$v_t=\frac{d_P^2\left({\rho }_{iron}-{\rho }_{oxide}\right)}{18\eta }$$(5)

where d_P, g, ρ_iron and ρ_oxide, and η represent the diameter of a m-Fe droplet (m), gravitational acceleration (m/s²), densities of the iron and oxide phases (kg/m³), and slag viscosity (N·s/m²), respectively. In this calculation, ρ_iron and ρ_oxide were 7000 and 2700 kg/m³, respectively, and η was changed depending on the oxide FeO content. The viscosity, estimated by extrapolating literature values,²²⁾ decreased with increasing oxide FeO content. Figure 17 shows how the slag FeO content and the m-Fe droplet diameter affected the m-Fe droplet terminal settling velocity. The terminal settling velocity decreased with decreasing droplet diameter and slag FeO content. In particular, the terminal velocity drastically decreased with decreasing FeO content because the slag viscosity increased when the FeO content was <25 mass%, indicating that free fall cannot be expected when droplets are smaller than 100 μm in diameter or when the oxide FeO content is <25%. Droplets larger than 1 mm in diameter can easily settle regardless of the slag FeO content. However, the gas/oxide interface showed droplets larger than 1 mm in diameter, which had not settled. Poggi et al. reported that a metal-droplet raft had formed on the upper phase surface owing to surface tension, γ_oxide.²³⁾ According to force balancing, the maximum diameter of such a droplet raft can be expressed as follows:   

$$d_{raft}=2{\left(\frac{2{\gamma }_{oxide}}{4/3{\rho }_{metal}g}\right)}^{0.5}$$(6)

The FeO–SiO₂–CaO system surface tension is in the range 0.4–0.5 N/m.²⁴⁾ Thus, according to Eq. (6), the maximum diameter of the droplet raft is 5.6–6.4 mm. The calculated terminal settling velocity and maximum droplet raft diameter are consistent with the experimental results and indicate an agitation requirement.

Fig. 17.

Terminal settling velocity plotted as functions of FeO content of slag for metal droplets of various diameters calculated using Stokes’ law.

4.3. System Evaluation

Reduction distributes the phosphorus in iron ore into the m-Fe and oxide phases. The content and amount of both phases were determined based on the following mass balance:   

$$\begin{array}{l}{W}_{ore}\times {\left[\%P_2{O}_5\right]}_{Ore}\times \frac{2m_P}{m_{P_2{O}_5}}=\\ W_{Metal}\times {\left(\%P\right)}_{metal}+W_{oxide}\times {\left[\%P_2{O}_5\right]}_{Oxide}\times \frac{2m_P}{m_{P_2{O}_5}}\end{array}$$(7)

where W_i and m_i represent the weight and molecular weight of phase i, respectively. [%P₂O₅] and (%P) are defined by L_P, as shown in Eq. (2), which can be calculated based on a function of the oxide phase %FeO, as shown in Fig. 14. The reduction of each component except for the iron in the ore is presumably negligible for this calculation; hence, the ore, metal, and oxide weights are related as follows:   

$$W_{ore}=W_{metal}\times \frac{m_{F{e}_{t}O}}{m_{Fe}}+W_{slag}$$(8)

The m-Fe and oxide phase phosphorus contents can be calculated based on relations (7) and (8) when the iron ore is reduced. The composition of the iron ore used in this study, which is the average of several ores reported by Alejandro Cores et al., is listed in Table 5.²⁵⁾ The iron ore basicity was set to 1.0 by adding CaO, and the FeO content subsequently decreased.

Table 5. Average composition of iron ore reported by Alejandro Cores et al.²⁵⁾

[mass%]
T-Fe	CaO	MgO	Al2O3	SiO2	MnO	P2O5	CL.
62.3	0.45	0.1	1.2	5.46	0.4	0.09	3.78

Figure 18 compares the relation between the m-Fe and oxide phase phosphorous contents obtained in this study with those obtained in equilibrium experiments.²⁾ (%P)_metal decreased and [%P₂O₅]_oxide increased with increasing oxide FeO content because the oxygen potential increased with increasing FeO content. (%P)_metal obtained in the equilibrium experiment is lower than that obtained in the reduction one owing to the higher L_P, as shown in Fig. 14. Oxides containing 16.0 and 11.3 mass% FeO and containing 6.1 and 8.6 mass% FeO must be used under reduction and equilibrium experimental conditions to produce m-Fe containing 0.01 and 0.02 mass% phosphorus, respectively.

Fig. 18.

Relation between phosphorous contents of metallic iron and oxide phases obtained in the reduction (Rd.) and equilibrium (Eq.) experiments.²⁾

Higher-FeO-content oxide can be used to produce lower-phosphorous-content m-Fe. However, the iron loss should be increased by increasing the oxide FeO content. Figure 19 shows simulated Fe loss plotted as functions of m-Fe phosphorus content for reduction and equilibrium experimental conditions. The simulation results showed that 29 and 20 kg/t-steel and 15 and 11 kg/t-steel of iron were lost under the reduction and equilibrium experimental conditions to produce m-Fe containing 0.01 and 0.02 mass% phosphorus, respectively. The simulated losses of Fe as slag in the conventional BF and basic oxygen furnace (BOF) were 1.2 and 19.1 kg/t-steel,²⁶⁾ respectively, indicating that although the proposed direct steelmaking was advantageous for Fe loss under the equilibrium experimental conditions, it was less so under the reduction ones. Hence, it is important for the metallic Fe produced using the proposed direct steelmaking to reach equilibrium.

Fig. 19.

Simulated Fe loss plotted as functions of phosphorus content of metallic iron under reduction and equilibrium experimental conditions.

The results of this study clearly demonstrate that phosphorus-free m-Fe can be produced under suitable conditions; however, high-phosphorus-content m-Fe is produced when the oxide is reduced under extreme conditions at the gas/oxide interface because iron droplets cannot be expected to settle to the bottom of the slag over long distances. Therefore, oxides should not be reduced in deep tank reactors. From this perspective, shallow reactors such as those used to sinter iron ore furnace and rotary hearth furnace (RHF) are suitable for producing phosphorus-free m-Fe. Conventional RHF can be used to produce direct-reduced iron (DRI) under an uncontrolled P_O2 atmosphere. DRI is produced by mixing iron ore and coke powders. Consequently, the phosphorus in iron ore may migrate into the DRI. Phosphorus migration can be suppressed if the RHF is operated under the controlled P_O2 atmosphere used in this study.

5. Conclusions

The phosphorus distribution behavior was investigated when molten iron-ore-based oxide was reduced under CO. The following conclusions were reached:

(1) Solid m-Fe particles dispersed in the oxide phase. Droplets larger than 1 mm in diameter floated on top of the oxide phase. Reducing the oxide produced a continuous m-Fe layer.

(2) The phosphorus content of the m-Fe in the upper region of the crucible was higher than that of the m-Fe in middle and lower regions because the iron was reduced from the gas/oxide interface.

(3) The phosphorus distribution between the oxide and the reduced solid m-Fe was lower than that at equilibrium owing to the high-phosphorus-content iron in the upper region of the crucible.

(4) The local phosphorus distribution between the solid m-Fe and the oxide was controlled by the oxide FeO content.

(5) The proposed direct steelmaking was advantageous, and it is important for the metallic Fe produced using this method to reach equilibrium.

Acknowledgements

We gratefully appreciate the support of grants from KAKENHI (No. 22760574) and ISIJ.

References

• 1)   N.  Maruoka,  S.  Ono,  H.  Shibata and  S.  Kitamura: Bull. Adv. Mater. Process. Build., IMRAM, Tohoku Univ., 66 (2010), 15.
• 2)   N.  Maruoka,  S.  Ono,  H.  Shibata and  S.  Kitamura: ISIJ Int., 53 (2013), 1709.
• 3)   Y.  Kato: ISIJ Int., 52 (2012), 1433.
• 4)   A. L.  Dipu,  J.  Ryu and  Y.  Kato: ISIJ Int., 52 (2012), 1427.
• 5)   N.  Dogan,  G. A.  Brooks and  M. A.  Rhamdhani: ISIJ Int., 51 (2011), 1086.
• 6)   Y.  Kato,  T.  Obara,  I.  Yamanaka,  S.  Mori,  A. L.  Dipu,  J.  Ryu,  Y.  Ujisawa and  M.  Suzuki: Prog. Nucl. Energ., 53 (2011), 1017.
• 7)   Y.  Kato: ISIJ Int., 50 (2010), 181.
• 8)   T.  Kitagawa: Tetsu-to-Hagané, 88 (2002), 430.
• 9)   Y.  Kashiwaya and  M.  Hasegawa: ISIJ Int., 52 (2012), 1513.
• 10)   H.  Suito and  R.  Inoue: Trans. Iron Steel Inst. Jpn., 24 (1984), 47.
• 11)   H.  Suito,  R.  Inoue and  M.  Takada: Trans. Iron Steel Inst. Jpn., 21 (1981), 250.
• 12)   H.  Ishii and  R. J.  Fruehan: Iron Steelmaker, 24 (1997), 47.
• 13)   G.  Li,  T.  Hamano and  F.  Tsukihashi: ISIJ Int., 45 (2005), 12.
• 14)   N.  Maruoka,  S.  Narumi and  S.-y.  Kitamura: CAMP-ISIJ, 26 (2013), 827.
• 15)   K.  Iwasaki,  N.  Sano and  Y.  Matsushita: Tetsu-to-Hagané, 67 (1981), 536.
• 16)   K.  Ito and  N.  Sano: Tetsu-to-Hagané, 69 (1983), 1747.
• 17)   J.  Im,  K.  Morita and  N.  Sano: ISIJ Int., 36 (1996), 517.
• 18)   R.  Wennrich,  A.  Mroczek,  K.  Dittrich and  G.  Werner: Fresenius’ J. Anal. Chem., 352 (1995), 461.
• 19)   A.  Tagaya,  H.  Chiba,  F.  Tsukihashi and  N.  Sano: Metall. Trans. B-Process Metall., 22 (1991), 499.
• 20)   T.  Nagasaka,  Y.  Iguchi and  S.  Banya: Tetsu-to-Hagané, 71 (1985), 204.
• 21)   H. A.  Fine,  D.  Meyer,  D.  Janke and  H. J.  Engell: Ironmaking Steelmaking, 12 (1985), 157.
• 22)  Slag Atlas, 2nd ed., Verlag Stahleisen GmbH, Düsseldorf, (1995).
• 23)   D.  Poggi,  R.  Minto and  W. G.  Davenport: J. Met., 21 (1969), 40.
• 24)   Y.  Kawai,  K.  Mori,  H.  Shiraishi and  N.  Yamada: Tetsu-to-Hagané, 62 (1976), 53.
• 25)   A.  Cores,  A.  Babich,  M.  Muniz,  S.  Ferreira and  J.  Mochon: ISIJ Int., 50 (2010), 1089.
• 26)  Statics related to iron and steel slag, Nippon Slag Association, Tokyo, (2013).