2020 Volume 60 Issue 7 Pages 1479-1486
Reduction experiments were conducted with sintered Fe2O3–CaO–SiO2–Al2O3 tablets at (mass%CaO)/(mass%SiO2) (C/S) of 1.5, 2.0, and 2.5 at 1000, 1050, 1100, 1150, and 1200°C. From the reduction behaviors, we investigated the relationship between the reduction rate and C/S at the softening and melting zone temperatures of blast furnaces. The reduction rates at C/S = 2.0 and 2.5 increased with temperature in the range of 1000 to 1200°C. The reduction rate at C/S = 1.5 increased with temperature in the range of 1000 to 1150°C; however, at 1200°C, it decreased to the same value obtained at 1000°C. The microstructures of these samples, after sintering at 1270°C, pre-reduction at 900°C, and reduction at 1200°C, were analyzed through SEM-EDS. Fe2O3 particles, SFCA, slag, and pores among Fe2O3 particles existed in the samples after sintering. The matrix components in the pre-reduced sample were suggested to be calcium silicate slag containing FeOx and Al2O3 at C/S = 1.5, and to be ‘FeOx’ originated from SFCA at C/S = 2.0 and 2.5. The porosity of the open pores at C/S = 1.5 decreased to 16%. It was found that the reduction rate at 1200°C decreased due to this lower porosity. By contrast, the porosity of the open pores after reduction at C/S = 2.0 and 2.5 was much higher than that after pre-reduction. The reduction rates of these samples at 1200°C were found to not decrease as a result of maintaining a higher porosity.
To reduce CO2 emissions, the low-carbon operation of blast furnaces is necessary. The coke layers become thinner in this operation in contrast to the formed softening and melting layers, through which gases cannot permeate when the blast furnace is operated with low carbon content. Because the coke layers act as a gas flow path in the cohesive zone, resistance to the gas flow increases. To achieve efficient and stable operations under such conditions, it is important to clarify and control the phenomena in the cohesive zone. Estimations of the gas permeability,1,2,3) melting-softening behavior,4,5) and reduction behavior6,7,8,9) in the cohesive zone have been the subjects of previous research.10) According to Mori et al.,4) the initial melt formation temperature of the pellets and sinters increases with the degree of reduction. The thickness of the cohesive zone may be decreased by accelerating the reduction rate in the softening and melting temperature regions. It was reported that the reduction rate of liquid FeOx (wustite) is about six times11,13) and twenty times12,13) higher than that of solid state FeOx when it is reduced with CO and H2, respectively. However, it was also reported that the reduction rate of FeOx compacts and agglomerates, including molten slag, decreases at softening and melting temperature because the pores are obstructed by the molten slag.6,7,8,9) Therefore, it is important to accelerate the reduction rate at softening and melting temperatures, as well as to control the melting behavior and pore structure of the sinters. To investigate the reduction behavior of iron oxide at the softening and melting temperatures, reduction experiments were first conducted with sintered Fe2O3–CaO–SiO2–Al2O3 tablets at CaO to SiO2 mass percentage ratios (C/S) of 1.5, 2.0, and 2.5 at 1000, 1050, 1100, 1150, and 1200°C. From the reduction behaviors observed at these temperatures, the relationship between the reduction rate and C/S was investigated. In addition, the influence of the pore structure on the reduction rate was examined.
The compositions of the samples are shown in Table 1. The compositions of Fe2O3 and Al2O3 were fixed at 81.3 and 1.9 mass%, respectively. C/S is basicity, and it is defined as (mass%CaO)/(mass%SiO2). CaO was prepared by the calcination of CaCO3 (purity: 99.5%) at 1050°C in the air. The powders of Fe2O3(purity: 95%, size: under 1 μm), Al2O3 (purity: 99.9%, size: under 75 μm), SiO2 (purity: 99.9%, size: under 100 μm), and CaO (size: 1 μm) were then mixed. Next, this mixed powder was pressed at 5 MPa for 3 min in 3-g quantities using a stainless-steel die with an inner diameter of 15 mm. These pressed samples were sintered at 1270°C for 5 min to obtain Fe2O3–CaO–SiO2–Al2O3 tablets. The tablets were about 13.5 mm in diameter and 6.5 mm in height. Three types of samples with C/S = 1.5, 2.0, and 2.5 were formed.
| C/S (CaO/SiO2) | Composition (mass%) | |||
|---|---|---|---|---|
| Fe2O3 | CaO | SiO2 | Al2O3 | |
| 1.5 | 81.3 | 10.1 | 6.7 | 1.9 |
| 2.0 | 81.3 | 11.2 | 5.6 | 1.9 |
| 2.5 | 81.3 | 12.0 | 4.8 | 1.9 |
Reduction experiments were conducted using an electric resistance furnace with a thermobalance, which is shown in Fig. 1. Figure 2 shows an overview of the procedure used in the reduction experiments. After heating at 10°C/min in an N2 atmosphere, the sample was pre-reduced at 900°C in a 50 vol%CO–50 vol%CO2 atmosphere. The flow rate was 2.3 L/min (n.t.p.). The equilibrated composition of the pure iron oxide after pre-reduction was estimated to be Fe0.925O = FeO1.08 from the phase diagram.14) Pre-reduction experiments were previously conducted, and it was confirmed that the masses of the samples did not undergo further changes after 120 min. Accordingly, the time of the pre-reduction process was determined to be 120 min. Furthermore, the composition of iron oxide in the sample after pre-reduction was estimated to be FeO1.08 from the measured mass change. After pre-reduction, the sample was heated at 5°C/min in an N2 atmosphere. The sample was then reduced to Fe. The following reduction process was carried out at a temperature of 1000, 1050, 1100, 1150, or 1200°C. The composition of the reducing gas is shown in Fig. 3 and Table 2. Gas composition was changed depending on temperature to simulate the atmosphere in the softening and melting zone of blast furnaces. The solid line in Fig. 3 indicates the partial pressure of the gases in equilibrium with FeOx and Fe. It should be noted that the difference between the plots and the solid line correspond to the driving force of reduction. After pre-reduction at 900°C and following reduction at 1200°C for 60 min, the microstructures of the samples were analyzed by scanning electron microscopy and energy dispersive X-ray spectroscopy (SEM-EDS).
Schematic illustration of the experimental apparatus with a thermobalance. (Online version in color.)
Overview of the reduction experiment procedure. (Online version in color.)
PCO + PH2 of the reduction experiments at 1000, 1050, 1100, 1150, and 1200°C.
| Temperature (°C) | Gas composition (vol%) | |||
|---|---|---|---|---|
| CO | CO2 | H2 | N2 | |
| 1000 | 25.9 | 14.1 | 8.0 | 52.0 |
| 1050 | 26.5 | 13.5 | 8.0 | 52.0 |
| 1100 | 27.0 | 13.0 | 8.0 | 52.0 |
| 1150 | 29.9 | 10.1 | 8.0 | 52.0 |
| 1200 | 33.0 | 7.0 | 8.0 | 52.0 |
The reduction curves for the samples at C/S = 1.5 are shown in Fig. 4. The reduction rate is shown to increase in the temperature range from 1000 to 1150°C. The reaction and diffusion rates of gases generally increase with temperature. In addition, the driving force of reduction increases in the temperature range from 1100 to 1200°C as shown in Fig. 3. However, the reduction rate at 1200°C decreased to the same value as that observed at 1000°C. Because the driving force of reduction is small and hardly increases in the temperature range from 1000 to 1100°C, reduction rate dose not significantly increase. Therefore, reduction rate from 1000 to 1050°C seems to be too large, and this is because of experimental error. The reduction curves for the samples with C/S = 2.0 and 2.5 are shown in Figs. 5 and 6, respectively. The reduction rates of these samples increased in the temperature range from 1000 to 1200°C and did not decrease at 1200°C. The reduction rate did not significantly increase from 1000 to 1100°C because the driving force of the reduction is small and hardly increases in the temperature range. These results suggest that the reduction rate is strongly influenced by the microstructures of the samples at 1200°C, which form in the region between the softening and melting temperatures.
Reduction curves of the samples with C/S = 1.5 at 1000, 1050, 1100, 1150, and 1200°C for the FeO1.08 to Fe stage.
Reduction curves of the samples with C/S = 2.0 at 1000, 1050, 1100, 1150, and 1200°C for the FeO1.08 to Fe stage.
Reduction curves of the samples with C/S = 2.5 at 1000, 1050, 1100, 1150, and 1200°C for the FeO1.08 to Fe stage.
The microstructures of the samples following sintering at 1270°C, pre-reduction at 900°C, and reduction at 1200°C for 60 min were analyzed by SEM-EDS. Figure 7 shows the cross-sectional SEM images of the microstructures of all samples, which are the back-scattered electron (BSE) images. Figure 8 shows the SEM image (BSE) of the same region of Fig. 7 at high magnification. Fe2O3 particles, silico-ferrite of calcium and aluminum (SFCA), slag, and pores among Fe2O3 particles existed in the samples after sintering. Some parts of the microstructures of these samples were influenced by the C/S value. The Fe2O3 particles of the C/S = 1.5 sample was covered with the matrix. In the samples with C/S = 2.0 and 2.5, porous matrix existed among the Fe2O3 particles. Small pores and cracks occurred in all particles after pre-reduction. The term ‘FeOx’ indicates FeOx (wustite) containing gangue elements. Reduced Fe and FeOx compounds were observed near the surface and near the center of the samples after reduction at 1200°C, respectively. Furthermore, the formation of a large amount of slag melt was confirmed among the FeOx particles in the sample with C/S = 1.5. The pores near the surface (Figs. 7(j) and 8(j)) were relatively larger than those in the center (Figs. 7(g) and 8(g)) of the sample due to sintering of the reduced Fe particles. On the other hand, the microstructures of the samples with C/S = 2.0 and 2.5 appeared sponge-like. Figure 9(a) shows the phase diagram of the Fe2O3–CaO–SiO2 system.15) The average compositions of the matrix in the samples after sintering were analyzed by EDS, and the average and initial compositions were plotted as circles and diamonds, respectively, in Fig. 9(a). The colored areas in this figure indicate the liquid regions at 1270°C. The average composition of the matrix in the sample with C/S = 1.5 was found to be that of a calcium silicate melt, which equilibrates with Fe2O3 at 1270°C. Therefore, the matrix was suggested to be a calcium silicate slag that contains Fe2O3 and Al2O3. Moreover, the average composition of the matrix in the samples with C/S = 2.0 and 2.5 was that of a calcium ferrite melt, which equilibrates with Fe2O3 at 1270°C. It has been reported that silico-ferrite of calcium and aluminum (SFCA), a kind of calcium ferrite, is formed under similar conditions.16,17,18) To determine if the compositions present in these samples were similar to that of SFCA, the composition of SFCA reported by Nicol19) and the compositions of our samples at C/S = 2.0 and 2.5 were compared in Fig. 10. The compositions of the matrices in both of the C/S = 2.0 and 2.5 samples are 2(CaO·SiO2)-3(CaO·3(Fe,Al)2O3), and this corresponds to the composition of SFCA reported by Inoue and Ikeda,16) which is n(CaO·SiO2)-m(CaO·3(Fe, Al)2O3). The area highlighted with dashed lines in Fig. 9(a) is enlarged and shown in Fig. 9(b). In Fig. 9(b), connecting line of Fe2O3 and 2CaO·SiO2 represents basicity at C/S = 1.87, and it is confirmed that the direction of the cotectic line is reversed across the line at C/S = 1.87. This suggests that calcium silicate and calcium ferrite melts are formed below and above C/S = 1.87, respectively. The compositions of the matrices at C/S = 1.5, 2.0, and 2.5 differ as a result of the melt that is present, which is produced during the sintering of samples.
Cross-sectional SEM images of the samples: (a)–(c) after sintering at 1270°C; (d)–(f) pre-reduction at 900°C; reduction at 1200°C (g)–(i) in the center and (j)–(l) on the surface of the samples.
Cross-sectional SEM images of the samples at high magnification: (a)–(c) after sintering at 1270°C; (d)–(f) pre-reduction at 900°C; reduction at 1200°C (g)–(i) in the center and (j)–(l) on the surface of the samples.
(a) Phase diagram of the Fe2O3–CaO–SiO2 system,15) on which the average compositions of the matrices are plotted. (b) Enlarged portion of the figure in (a) outlined with dotted lines. (Online version in color.)
Plot showing the variation of the chemical composition of SFCA19) with the average composition of the matrices in the samples with C/S = 2.0 and 2.5.
Figure 11 shows the phase diagram of the FeOx–CaO–SiO2 system,20) in which the average compositions of the matrix among the FeOx particles in the pre-reduced samples are plotted. The compositions of the matrices in the pre-reduced samples were similar to those of the sintered samples. Therefore, the matrices in the pre-reduced samples were suggested to be calcium silicate slag containing FeOx and Al2O3 at C/S = 1.5, and to be ‘FeOx’ originated from SFCA at C/S = 2.0 and 2.5. On the other hand, the colored area in Fig. 11 indicates that the liquid region occurs at 1200°C in this system. The composition of the slag in the pre-reduced sample with C/S = 1.5 is similar to the composition of an olivine-like slag melt at 1200°C. This suggests that an olivine-like slag melt is produced from the reaction between the slag and FeOx. Figure 12 shows the cross-sectional SEM images (BSE) of the sponge-like structures at a high magnification around the center of the samples with C/S = 2.0 and 2.5. It was found that the slag was distributed inside of FeOx particles in the sponge-like structure. The composition of ‘FeOx’ after pre-reduction corresponds to the coexistence of FeOx and 2CaO·SiO2 phases at 1200°C. This suggests that the sponge-like structure was formed due to the separation of the gangue elements and FeOx, which elements were contained in the ‘FeOx’ particles. In these samples, slag was observed inside of the FeOx particles, but the penetration of the slag melt to the surface of the samples could not be confirmed.
Average composition of the matrices in the microstructures of the pre-reduced samples plotted on the FeOx–CaO–SiO2 ternary phase diagram. (Online version in color.)
SEM images of the sponge-like structure among the FeOx particles at high magnification for samples with (a) C/S = 2.0 and (b) C/S = 2.5.
The porosities of the samples measured by mercury porosimetry and the SEM analysis of the cross-sectional SEM images are provided in Tables 3 and 4, respectively. In Table 4, the porosities of the centers and surfaces of the samples after reduction at 1200°C are estimated because the main phases differ (Fe is the main phase near the surface, while FeOx is the main phase in the center). Using the mercury porosimetry and SEM analyses, the porosities of only the open pores and all pores (open and closed pores) were respectively found. The total porosities of each of the three sintered samples are almost the same, as they range from 32–36% (Table 4). At C/S = 1.5, the porosity of the open pores was 15% (Table 3), while the total porosity was 35% (Table 4). This result suggests that the slag melt penetrated into some pores between the Fe2O3 particles during sintering at 1270°C. Therefore, the porosity of the closed pores was estimated to be 20%. In addition, the porosities of the open pores were measured to be 23% and 32% at C/S = 2.0 and 2.5, respectively. These results indicate that the same penetration of the melt that occurred at C/S = 1.5 also occurred at these C/S values because SFCA particles existed among the Fe2O3 particles instead of the slag. On the other hand, the porosities of the open pores were found to be almost the same as the total porosities after pre-reduction. In these samples, the closed pores may be connected with the open pores via cracks and pores that are produced during pre-reduction. In the sample with C/S = 1.5, after undergoing reduction at 1200°C, the total porosity in the center of the sample was considerably decreased by 11%. This result corresponds to the slag melt penetrating into some pores during reduction. The total porosity near the sample surface was estimated to be 38%, and this is almost same as that of the pre-reduced sample, although it is much lower than that of the other samples with C/S = 2.0 and 2.5. Some pores and reduced Fe particles might become larger with temperature, and these pores might become obstructed by the slag melt. As a result, the porosity of the open pores at C/S = 1.5 decreased to 16%, as shown in Table 3. It was found that the reduction rate at 1200°C decreased due to this lower porosity. In contrast, in the samples with C/S = 2.0 and 2.5, the porosities of the open pores after reduction were much higher than after pre-reduction. The amount of slag melt that was produced may be small, and therefore many of the pores were not obstructed. As a result, the reduction rates of these samples at 1200°C did not decrease when a higher porosity was maintained.
| C/S | Porosity (%) | ||
|---|---|---|---|
| 1.5 | 2.0 | 2.5 | |
| Sintered sample | 15 | 23 | 32 |
| Pre-reduced sample | 41 | 40 | 44 |
| Sample reduced at 1200°C | 16 | 51 | 61 |
| C/S | Porosity (%) | ||
|---|---|---|---|
| 1.5 | 2.0 | 2.5 | |
| Sintered sample | 35 | 32 | 36 |
| Pre-reduced sample | 40 | 41 | 39 |
| Sample reduced at 1200°C (Center) | 29 | 37 | 50 |
| Sample reduced at 1200°C (Surface) | 38 | 56 | 61 |
Considering these results, it was found that the optimum condition of basicity in the samples was C/S = 2.0–2.5 during reduction at the softening and melting temperatures.
(1) The reduction rates at C/S = 2.0 and 2.5 increased when the temperature ranged from 1000 to 1200°C. The reduction rate at C/S = 1.5 increased when the temperature ranged from 1000 to 1150°C; however, at 1200°C, the rate was the same as that at 1000°C. These results suggested that the reduction rate is strongly influenced by the microstructures in the samples at 1200°C, which form in the region between the softening and melting temperatures.
(2) The matrices in the pre-reduced sample were suggested to be composed of calcium silicate slag containing FeOx and Al2O3 at C/S = 1.5. An olivine-like slag melt was produced by a reaction between the calcium silicate slag and FeOx when this sample was reduced at 1200°C. The porosity of the open pores at C/S = 1.5 decreased to 16% at this temperature. The reduction rate at 1200°C decreased as a result of this lower porosity. Some of the open pores at C/S = 1.5 might become obstructed by the slag melt.
(3) The matrices in the pre-reduced sample were suggested to be ‘FeOx’ originated from SFCA at C/S = 2.0 and 2.5. It was found that the slag was distributed inside of the FeOx particles in the sponge-like structure. In these samples, although the slag was observed inside of the FeOx particles, the amount of slag melt that was produced may be small, and many pores were therefore not obstructed. As a result, the reduction rates of these samples at 1200°C did not decrease when a higher porosity in the samples was maintained.
