ISIJ International, Vol. 53 (2013), No. 4, pp.

570575

Effect of CO Gas Concentration on Reduction Rate of Major Mineral Phase in Sintered Iron Ore
Daisuke NOGUCHI,1)* Ko-ichiro OHNO,2) Takayuki MAEDA,2) Kouki NISHIOKA3) and Masakata SHIMIZU2)
1) Graduate Student, Kyushu University, Motooka, Nishi-ku, Fukuoka, 819-0395 Japan. 2) Department of Materials Science & Engineering, Kyushu University, Motooka, Nishi-ku, Fukuoka, 819-0395 Japan. 3) Formerly Department of Materials Science & Engineering, Kyushu University. Now at Nippon Steel & Sumitomo Metal Corporation, 16-1, Sunayama, Kamisu, Ibaraki, 314-0255 Japan. (Received on October 31, 2012; accepted on January 4, 2013; originally published in Tetsu-to-Hagan, Vol. 97, 2011, No. 11, pp. 548553)

As a fundamental study for clarifying the reduction phenomena of iron ore sinter in blast furnace, iron oxide (H) and quaternary calcium ferrite (Cf) were prepared and these kinetic behaviors at the final stage of reduction with COCO2 gas mixture were studied. Reduction rate increased with increasing reduction temperature. Moreover, it increased with increasing partial pressure of CO gas. Difference of reduction rate caused by gas composition is much larger than reduction temperature. From comparisons of weight loss curves, reduction rate of H samples was faster than that of Cf samples under the same or similar conditions. Reduction reaction of H and Cf samples proceeded topochemically at higher temperature (1 100C), and didnt proceed topochemically at lower temperature (1 000C). Besides, the reduction reaction of samples with CO rich gas proceeded more topochemically. Structure of iron layer in H samples was affected by temperature and gas composition. On the other hand, structure of iron layer in Cf samples was almost the same in all experimental conditions. Reduction data were analyzed based on one interface unreacted core model, and chemical reaction rate content kc and effective diffusion coefficient in product layer De were determined. The values of kc show Arrhenius-type temperature dependency, and were approximately same tendency except for Cf samples with near equilibriums gas compositions. The values of De of H samples show the temperature and gas composition dependencies, and that of Cf samples were approximately constant in all experimental conditions. KEY WORDS: iron oxide; quaternary calcium ferrite; sinter; reduction rate; kinetic analysis; unreacted core model.

1. Introduction It is necessary to know reduction behavior of self-fluxing iron ore sinter that is main burden of iron, in order to clarify the reaction behavior in a blast furnace. Many experiments for iron ore sinter that produced in sinter plant or pod were carried out to clarify the reduction behavior. However, iron ore sinter has various mineral phases which consist of iron oxide, calcium ferrite, slag and so on. Therefore structure of iron ore sinter is very complex. Analysis of simulated iron ore sinter that has simplified structure is required for the quantitative analysis of reduction rate of iron ore sinter. In particular, clarifying the reducibility of iron oxide and calcium ferrite is most important for understanding the reducibility of iron ore sinter. Many reduction experiments were carried out with pure CO or H2 gas. But actual atmosphere in blast furnace is COCO2 gas mixture, therefore experiments with COCO2 gas mixture are required. Rate analysis
* Corresponding author: E-mail: noguchi@zenith.zaiko.kyushu-u.ac.jp DOI: http://dx.doi.org/10.2355/isijinternational.53.570

of reduction with COCO2 gas mixture near the FeOFe equilibrium gas composition is important especially. Calcium ferrite in iron ore sinter is multicomponent calcium ferrite including SiO2, Al2O3 and so on, and has different reduction mechanism with Fe2O3CaO binary calcium ferrite. One of the authors13) synthesized CaOFe2O3 SiO2Al2O3 quaternary calcium ferrite, then studied its reduction sequence and equilibrium constants in COCO2 gas mixture. In consequence, quaternary calcium ferrite is reduced to iron by way of magnetite and wustite containing CaO, SiO2 and Al2O3 without producing intermediate products CaOFeOFe2O3, CaO3FeOFe2O3 and 2CaOFe2O3, which are produced in the reduction of binary calcium ferrite. The equilibrium CO concentrations for the reduction of quaternary calcium ferrite were higher than those for the reduction of pure iron oxide. Thus, as a fundamental study for clarifying the reduction phenomena of iron ore sinter in blast furnace, experiments on iron oxide and quaternary calcium ferrite were carried out with COCO2 gas mixture, and relationship between the reduction rate at the final stage of reduction of the samples
570

2013 ISIJ

ISIJ International, Vol. 53 (2013), No. 4

and the compositions of COCO2 gas mixture was studied. We investigated at the final stage of reduction of the samples because about 70% of total reduced oxygen in iron oxide and quaternary calcium ferrite are removed in this stage and this reaction is main reduction at chemical reserve zone or more under. 2. Experiments 2.1. Samples Two kinds of samples that were made from iron oxide (H samples) and calcium ferrite (Cf samples) were used in the experiments. Table 1 shows chemical compositions of each samples. H samples were prepared from a reagent grate powder (45 m) of Fe2O3. About 3.0 g of the powder was weighed out and it was made spherical shape about 1 cm by hand roll method. Then the sample was heated up to 1 200C at the rate of 0.33 K/s (20C/min). After being kept for 1 h at 1 200C, the samples were cooled in furnace. This spherical pellet was used as H samples to the reduction experiments. The porosity of H samples were 2732%. Quaternary calcium ferrite was prepared from reagent grade powders of Fe2O3, CaCO3, SiO2 and Al2O3. The powders were mixed to be the composition as shown in Table 1. The mixed powder was fired at 1 000C for 1 h in the air and was followed by crushing and mixing. The operation of firing, crushing and mixing was repeated three times. The powder was put into a magnesia crucible (3 cm 10 cm) and it was then heated up to 1 300C at the rate of 0.17 K/s (10C/min) using a silicon carbide resistance furnace. After being kept for 0.5 h at 1 300C, the synthesized sample was cooled down to 1 100C at the rate of 0.33 K/s (20C/min) and was finally quenched in water. Synthesized sample was crushed into powder of 4575 m in diameter. About 2.3 g of the powder was weighed out and pressed into a briquette of about 1 cm 1 cm. The briquette was used for the reduction experiment without sintering. The porosity of Cf samples were 4449%. Porosity of H and Cf samples were calculated from the apparent and true density respectively. 2.2. Experimental Procedure Reduction experiments carried out at 900, 1 000 and 1 100C using a thermal balance and it was heated up to each experiment temperature in N2 gas stream. Then, a sample was hung on the thermal balance under N2 atmosphere. At first, the sample was reduced to wustite with 50%CO 50%CO2 gas mixture. Next, the sample was reduced to iron with prescribed COCO2 gas mixture. All gas flow rates were 3.33 105 Nm3/s (2 NL/min). Experimental gas composition determined as follows. Equations (1) and (2) show the reaction at final stage reducTable 1. Chemical composition of iron oxide (H) and quaternary calcium ferrite (Cf) samples (mass%). Fe2O3 H Cf 100 65 CaO 0 23.3 SiO2 0 7.8 Al2O3 0 3.9

tion of iron oxide and its equilibrium constant KH.4) FeO (s) + CO (g) = Fe (s) + CO2 (g) ............ (1) KH = exp (2.706 + 2 289 / T) ................. (2) Where T is absolute temperature. Equilibrium gas composition derived from Eq. (2) because total gas pressure was 1 atm. Likewise, Eqs. (3) and (4) show the reaction at final stage reduction of the quaternary calcium ferrite and its equilibrium constant KCf.1) Therefore, equilibrium gas composition derived from Eq. (4). FeO (s) + CO (g) = Fe (s) + CO2 (g)........... (3) KCf = exp (2.785 + 2 042 / T)................. (4) Figure 1 shows equilibrium gas compositions as close and open squares. Experimental gas compositions were equilibrium gas composition + 2% CO gas (as reverse triangles), 100% CO gas (as circles) and intermediate composition of the two (as regular triangles). Table 2 shows equilibrium and experimental gas compositions of H and Cf sample respectively. 3. Results 3.1. Reduction Curves Figures 24 show fractional reduction curves of H and Cf samples with 100% CO, intermediate CO% and equilibrium +2% CO respectively. Reduction rates of both samples increased with increasing reduction temperature with same

Fig. 1.

Equilibrium gas composition-temperature diagram for reduction of quaternary calcium ferrite with COCO2 gas mixture.

Table 2. H 900C 1 000C 1 100C Cf 900C 1 000C 1 100C

Experimental gas composition (vol%).

(Eq.)

70.0 73.3 75.9

81.4 82.9 84.0

68.0 71.3 73.9

(Eq.)

100 100 100

74.0 76.5 78.5

76.0 78.5 80.5

88.0 89.3 90.3

100 100 100

571

2013 ISIJ

ISIJ International, Vol. 53 (2013), No. 4

Fig. 2.

Reduction curves of FeO to Fe with 100% CO gas.

Fig. 5.

Reduction curves of FeO to Fe with COCO2 gas mixture at 1 100C.

Fig. 3.

Reduction curves of FeO to Fe with COCO2 gas mixture of intermediate CO%.

Fig. 6.

Reduction curves of FeO to Fe with COCO2 gas mixture at 1 000C.

Fig. 4.

Reduction curves of FeO to Fe with COCO2 gas mixture of equilibrium +2%CO.

Fig. 7.

Reduction curves of FeO to Fe with COCO2 gas mixture at 900C.

gas composition. Figures 57 show fractional reduction curves of H and Cf samples at 9001 100C respectively. Reduction rates of both samples decreased with decreasing partial pressure of CO at same temperature, and they were especially small near by equilibrium CO%. Furthermore, influence of gas composition on the reduction rate was larger than that of temperature as shown Figs. 24. Reduction rates of H and Cf samples cannot be compared based on the fractional reduction curves, because H and Cf samples were different from initial weight and total reducible oxygen. For this reason, weight loss curves were prepared and reduction rates of H and Cf samples were compared
2013 ISIJ 572

based on these curves. Figure 8 shows the weight loss curves at 1 100C. Reduction rates of H and Cf samples were almost the same near by equilibrium CO%. On the other hand, reduction rate of H samples was faster than that of Cf samples in higher CO%. Same tendency was also observed below 1 000C. The reason for the difference of the reduction rate between H and Cf samples was considered to be not only reactivity of raw materials but also gas diffusivity depend on the sample structure, driving force of reaction depend on gas composition and so on.

ISIJ International, Vol. 53 (2013), No. 4

Fig. 8.

Weight loss curves of FeO to Fe with COCO2 gas mixture at 1 100C.

Fig. 10.

Cross-sectional view of calcium ferrite samples partially reduced with COCO2 gas mixture.

Fig. 11. Fig. 9. Cross-sectional view of iron oxide samples partially reduced with COCO2 gas mixture.

Microstructure of iron oxide samples partially reduced with 100% CO gas.

Macro and Microscopic Observations of Partially Reduced Samples Macroscopic observations were carried out on the partially reduced samples which were prepared by interrupting the reduction at about 70% reduction. Figures 9 and 10 show the cross sections of H and Cf samples. Percentages that shown in these figures represent the fractional reduction of each samples. Cf samples that reduced with pure CO are not shown here because one of authors5) reported that Cf reductions proceeded topochemically with pure CO at 900C or higher. These figures show that the reduction of H and Cf samples proceeded topochemically at higher temperature and CO concentration. By contrast, reduction reaction of H and Cf samples at lower temperature and CO concentration proceeded not topochemically. Figures 1114 show the microstructure of partially reduced H and Cf samples at 900, 1 000 and 1 100C with prescribed COCO2 gas mixture respectively. In these fig573

3.2.

Fig. 12.

Microstructure of iron oxide samples partially reduced with COCO2 gas mixture of equilibrium +2%CO.

2013 ISIJ

ISIJ International, Vol. 53 (2013), No. 4

Fig. 15.

Temperature dependency of chemical reaction rate constants kc.

Fig. 13.

Microstructure of calcium ferrite samples partially reduced with COCO2 gas mixture of intermediate CO%.

Fig. 16.

Temperature dependency of effective diffusivities De.

4. Kinetic Analysis Reduction data were analyzed by one interface unreacted core model because reduction reactions proceeded topochemically at 1 100C. Applying unreacted core model were not suitable because reactions did not proceed topochemically at 1 000C and below. In this study, however, reduction data at all temperature were analyzed by unreacted core model for comparison. Cf sample was analyzed as a spherical approximation based on volume. Chemical reaction rate constants kc and effective diffusivities in product layers De were obtained by mixed-control plot.6) Figures 15 and 16 show the temperature dependency of kc and De. The values of kc of H samples with each CO concentrations show the Arrhenius-type temperature dependency. The values of kc of Cf samples with each CO concentrations also show the Arrhenius-type temperature dependency, but that with near the equilibrium gas composition shows the quite different trend from the others. It is considered that the reduction of Cf sample with the near equilibrium gas composition was not proceeded topochemically compared to H sample. On the other hand, the values of De of H samples show the Arrhenius-type temperature dependency and the values of De of Cf samples did not show the temperature dependency. Figures 17 and 18 show the gas composition dependency of kc and De. The values of kc did not show the gas composition dependency except for the Cf sample reduced with the near equilibrium gas composition, and these values were
574

Fig. 14.

Microstructure of calcium ferrite samples partially reduced with COCO2 gas mixture of equilibrium +2%CO.

ures, an, bn and cn show the microstructure at near the surface, reaction interface and center of samples respectively. In the case of H samples, sintering of reduced iron was observed in all samples. And the morphology of reduced iron and wustite were different. Sintering of reduced iron more proceeded at higher temperature, and grain shapes and porosity appreciably changed especially at 1 100C. On the other hand, when CO gas concentration was low, the sintering of produced iron was more proceeded because the reduction time became longer. Moreover, when the reduction was carried out with CO rich gas and at 1 000C and below, wustite grains surrounded by reduced iron were observed. In the case of Cf samples, sintering of reduced iron was observed in all samples as well as H samples. The morphology of grain and porosity of iron and wustite layer were almost the same. Beside, wustite grain surrounded by reduced iron that observed in H samples were not observed in all Cf samples.
2013 ISIJ

ISIJ International, Vol. 53 (2013), No. 4

and Cf samples, the difference of reduction rate between H and Cf samples as shown in Fig. 8 was influenced by diffusivity rather than reactivity of samples. 5. Conclusions Iron oxide (H) and quaternary calcium ferrite (Cf) samples were prepared and these kinetic behaviors at the final stage of reduction with COCO2 gas mixture were studied. Obtained results are summarized as follows. Reduction rate increased with increasing reduction temperature. Moreover, it increased with increasing partial pressure of CO gas. Difference of reduction rate caused by gas composition is much larger than reduction temperature. From the comparisons of weight loss curves, reduction rate of H samples was faster than that of Cf samples in same conditions. Reduction reaction of H and Cf samples proceeded topochemically at higher temperature (1 100C), and didnt proceed topochemically at lower temperature (1 000C). Besides, the reduction reaction of samples used CO rich gas proceeded more topochemically. Structure of iron layer in H samples was changed by both temperature and gas composition. On the other hand, structure of iron layer in Cf samples was almost the same in all experimental conditions. Reduction data were analyzed by one interface unreacted core model, and chemical reaction rate content kc and effective diffusion coefficient in product layer De were determined. The values of kc show the Arrhenius-type temperature dependency, and were approximately same tendency except for Cf samples in near equilibriums gas compositions. The values of De of H samples show the temperature and gas composition dependencies, and that of Cf samples were approximately constant in all experimental conditions.
REFERENCES
1) T. Maeda and Y. Ono: Tetsu-to-Hagan, 75 (1989), 416. 2) T. Maeda, S. Masumoto and Y. Ono: Technol. Rep. Kyushu Univ., 62 (1989), 697. 3) T. Maeda and Y. Ono: Tetsu-to-Hagan, 80 (1994), 451. 4) W. S. Chung, T. Murayama and Y. Ono: J. Jpn. Inst. Met., 51 (1987), 659. 5) T. Maeda and Y. Ono: Tetsu-to-Hagan, 77 (1991), 1569. 6) J. Yagi and Y. Ono: Trans. Iron Steel Inst. Jpn., 8 (1968), 377.

Fig. 17.

Gas composition dependency of chemical reaction rate constants kc.

Fig. 18.

Gas composition dependency of effective diffusivities De.

almost constant at same temperature. On the other hand, the values of De of H samples show the gas composition dependency, but the values of De of Cf samples did not show the gas composition dependency. According to the microscopic observations, the morphology of product layers in H samples changed by temperature and CO gas concentration, in contrast, it in Cf samples did not change. They suggests that the values of De of H samples have the temperature and gas composition dependencies, and the values of De of Cf samples dont have those. According to the magnitude correlation of kc and De of H

575

2013 ISIJ