2025 Volume 65 Issue 6 Pages 833-838
The quality of iron ore is expected to change, and blast furnace raw materials will diversify in the future as ores are pretreated to improve ore quality. Furthermore, an increase in the proportion of hydrogen-based reducing agents is necessary to satisfy the demand for carbon neutrality. Therefore, the capacity of blast furnace operations to adapt to these external factors must be enhanced. For stable blast furnace operation even when the external factors change, controlling the cohesive zone is crucial. In this study, the ore-packed beds composed of different ores were reduced by heating up to 1200°C in the atmosphere of a hydrogen-enriched blast furnace. Subsequently, the temperature was increased to 1450°C in an inert atmosphere to investigate the softening and melting behaviors of the samples in contact with different ores. During reduction up to 1200°C, metallic iron was bound between ore particles, while no interparticle migration of gangue components occurred. The unreduced oxide core of the ore melted on heating up to 1450°C, although deformation of the ore did not progress considerably owing to the metallic iron structure. The oxide in the reduced metal shell partially melted, and interdiffusion of the gangue components occurred more than 2 mm from the particle interface.
With changes in the quality of raw materials for blast furnaces, such as the depletion of low-phosphorus iron ore1) and progression of inferior quality,2) technologies that allow the use of a diverse range of raw materials are crucial. However, a growing trend has been noted toward carbon neutrality, for which processes such as the enhancement of hydrogen reduction in blast furnaces, melting of direct reduced iron (DRI) in electric arc furnaces, and extensive use of hot briquetted iron (HBI) in blast furnaces are being considered. The production of DRI and HBI requires iron ore of high grade; hence, a beneficiation process is employed to improve the quality of the grade. Beneficiation involves crushing and separation, which reduces the particle size of the raw ore.
In the present blast furnace operation, carbon-based reducing materials such as coke and pulverized coal are used to reduce sinters, pellets, and lump ores. However, possibilities remain of increasing the fraction of hydrogen-based reducing materials and use pulverized ores as pellet feed. With this change in operation, the reduction, softening, and melting behaviors of sinters, pellets, and lump ores in blast furnaces have been widely discussed. When several ores with varying compositions and reduced properties coexist in a blast furnace, the low-melting liquid phase and unmelted ores may react with each other because of the different melting behaviors of each ore. Such interactions among ores potentially result in a behavior different from that of the softening and melting of single ores. Wu et al.3) studied the softening at contact points of different ores at 1200°C using a sinter, two types of pellets, and four types of lump ore. They reported that in samples mixed with lump ore and sinter which have low and high basicity, respectively, 2CaO·SiO2—a high-melting-point slag derived from sinter—and 2FeO·SiO2 —a low-melting-point slag derived from lump ore—reacted during homogenization and generated low-melting-point CaO·FeO·SiO2. This resulted in higher shrinkage than when using a single ore because of the interaction between lump ores and sinter. Hotta et al.4) conducted load softening tests under the conditions of simulating blast furnace processes to analyze the softening and melting behaviors of sinters and pellets. Slag assimilation occurred as the lump ore was reduced and the temperature was increased, and its basic path followed the direction of expansion of the liquid-phase region with heating. Finally, assimilation proceeded toward the composition of the gangue before reduction. Pan et al.5) prepared samples wherein acidic pellets were in contact with the simulated sinter slag. Their findings revealed that the melting of the acidic pellet slag caused an interaction, and the sinter slag exhibited melting behavior at temperatures lower than the melting temperature of the single sinter. They reported that the amount of liquid phase generated increased, which increased the gas flow resistance of the blast furnace. Higuchi et al.6) analyzed the cohesive behaviors of mixed samples comprising a sinter, basic pellets, and high-MgO olivine pellets in a test blast furnace. Kaushik et al.7) reported the melting stages during the mixing of acidic and olivine pellets as (1) sintering of solid-phase metal, (2) initial formation of melt, (3) assimilation of percolated slag, and (4) assimilation of an unreduced core. They also reported that pellet reduction, metal shell structure, slag basicity, and viscosity played competitive roles in different ore interactions and slag percolation. As described above, interactions between ores occur when different types of ores are mixed and reduced in blast furnaces. However, when reduction is carried out under hydrogen-enriched conditions, the reduction progresses and the shapes of the iron and oxide change (Figs. S1 and S2 (Supporting Information)).
Therefore, the present study provides insights into understanding the reaction mechanism between sinters and pellets and lump ores to control the shape of the cohesive zone in hydrogen-enriched blast furnace operations. We investigated the reduction behavior and state of the ores at the contact interface of mixed samples of sinter/acidic pellet and sinter/basic pellet with different chemical compositions reduced under conditions simulating a hydrogen-enriched blast furnace. Furthermore, the softening and melting behaviors of the two types of ore-contact samples were examined in the temperature range of the cohesive zone.
Three types of iron ore—sinter, basic pellet, and acidic pellet—with particle size of 10–15 mm were used; their chemical compositions are listed in Table 1. The packed bed consisted of these ores, and reduction was performed simultaneously by flowing the reducing gas. Furthermore, alumina balls were placed above and below the ore bed to a height of 20 mm each, and a wire mesh was placed between the ore bed and the alumina ball bed to prevent the alumina balls from being embedded in the ore bed. A load of 98 kPa was applied to the entire packed bed in the vertical downward direction. The reduction test conditions (reducing gas and temperature rise conditions) for the samples were optimized based on the results of calculations using a blast furnace mathematical model from the previous study.8) After the entire packed bed was heated from room temperature to 300°C in an N2 atmosphere, the temperature was increased at a constant rate of 3.3°C/min from 300 to 700°C and at 7.5°C/min from 700 to 1200°C in the atmosphere shown in Fig. 1. After reaching 1200°C, the furnace power was immediately turned off, and the packed bed was cooled in the furnace. The packed bed was then cut vertically, embedded in resin, and polished before the microstructural observations. The microstructure was observed using scanning electron microscopy (SEM), and the composition of each phase was analyzed using an electron probe microanalyzer (EPMA).
| T.Fe | FeO | CaO | SiO2 | Al2O3 | MgO | CaO/SiO2 | |
|---|---|---|---|---|---|---|---|
| Sinter | 57.2 | 6.16 | 9.6 | 5.5 | 1.7 | 0.96 | 1.76 |
| Basic pellet | 66.9 | 0.27 | 1.6 | 1.7 | 0.34 | 0.50 | 0.96 |
| Acidic pellet | 65.3 | 0.36 | 0.26 | 5.5 | 0.32 | 0.50 | 0.05 |
The sample reduced by heating up to 1200°C, as described in Section 2.1, was reheated to 1450°C in an inert atmosphere, where the oxide phase melted partially. To observe the interaction of each phase with the deformation and melting near the interface among the different phases, in situ observations during heating and EPMA observations of the cross-sectional microstructure of the sample after heating were conducted. A horizontal electric furnace equipped with a mullite reaction tube with an inner diameter of 37 mm and controllable atmosphere inside the furnace was used for heating. A B-type platinum thermocouple was placed in the soaking zone of the furnace tube and connected to a PID-controllable output device. This allowed the temperature in the vicinity of the sample to remain constant. From the partially reduced sinter/acidic pellet and sinter/basic pellet with heating up to 1200°C, the sample was cut into 10 × 10 × 10 mm3 pieces to encompass the contact surface between the ores and used as a sample for reheating observation. The cut sample was placed on a 25 × 20 × 2 mm3 ZrO2 substrate, which was then placed on an alumina porous brick. This was inserted from the low-temperature zone in the furnace tube into the hot zone of the furnace maintained at 1200°C for approximately 5 s. The furnace atmosphere was controlled using a high-purity N2 flow (200 ml/min). After holding for 2–3 min until the sample temperature stabilized, the temperature was increased from 1200 to 1450°C at 5°C/min. One end of the furnace tube was sealed with a cap with a window for internal observation; a camera (Nikon 1 V3, SIGMA APO MACRO 180 mm F2.8 EX DG OS HSM lens) was set near the window to capture pictures of the sample every 1°C for direct observation. Immediately after the sample temperature reached 1450°C, the cap was opened, and the sample was removed from the furnace and quenched in water. The samples were cut, embedded in resin, and polished; the polished surfaces were used as observation samples. The microstructures of the iron and oxide phases of the cross-section were observed, and the composition of each phase was analyzed using the EPMA.
Figure 2 shows the appearance of the samples during heating. The sample was placed with the basic pellet at the top and the sinter at the bottom; the circular object on the right side of the sample is the tip of the thermocouple. The shape of the sample does not change significantly up to 1300°C. The basic pellet in the upper part of the sample melt and drip oxide above 1350°C. In contrast, the shape of the sinter in the lower part of the sample does not change until 1450°C. Although the slag phase of the reduced sinter, which contains gangue components, also melts, no dripping is observed in the samples used in the heating experiments. In this experimental method, deformation can be observed when a significant portion of the melt forms on the surface. At the experimental temperatures, the reduced iron does not melt, while a low-melting-point slag phase containing a large amount of iron oxide forms in the unreduced regions. Dripping from the sinter is not observed because the large unreduced cores are not exposed on the surface. On the other hand, regions where iron shells form are exposed, but they retain their shape. The shape of the entire sample is retained without significant deformation up to 1450°C due to the solid iron phase. Figure 3 shows the sample after reduction and heating to 1450°C in an inert atmosphere. The upper part is the basic pellet, and the lower part is the sinter. In sample (a) before heating, the light-gray areas on the surface are the metallic iron phase, while the black central part of the basic pellet in the upper right corner is oxide, indicating that reduction is in progress. In sample (b) after heating, the pellet and the sinter are placed in one piece. The oxide melts and drips due to heating, and the area of oxide that existed in the upper right of the sample becomes void. The dripping of the oxide phase of the basic pellet is observed in the temperature range of 1350–1400°C (Fig. 2). Since slag melt was observed on the zirconia substrate after the experiment, the void fraction of the entire reduced ore is expected to increase with heating. However, no significant shape deformation of the iron phase is observed when heating up to 1450°C.
Figure 4 shows the EPMA elemental mapping results around the contact area between the basic pellet (left side) and the sinter (right side) of the sample before heating. The Fe concentration is low in the Ca-containing oxide phase; however, considering the metallic iron site, no clear boundary is observed at the contact region between the basic pellet and the sinter, and the ores appear to coalesce. However, the Ca concentration is high in the sinter area, and the boundary between the basic pellet and the sinter is distinct, indicating that the oxide remains in the elemental distribution of the original ore at the contact interface. Both ores were heated and reduced when in contact; however, the chemical diffusion of gangue components other than metallic Fe is minimal.
To compare the changes in the Ca and Si concentration distribution near the ore interface with heating to 1450°C, line analysis was performed on the region near the ore interface of the samples before and after heating via EPMA. Figure 5(a) shows a cross-sectional image of the basic pellet/sinter sample before heating and the location of the line analysis. The analysis conditions were set as follows: acceleration voltage, 15 kV; irradiation current, 10 nA; beam diameter, 0 μm; analysis interval, 10 μm. Figure 5(b) shows the concentration changes of Ca, Si, O, and Fe obtained by line analysis in units of counts of the detected X-rays; the dotted line indicates the position of the interface between the basic pellet and the sinter. Based on Figs. S1 and S2, the interface between the different ores was defined as the position at which the microshape of the metallic iron phase changed. In both the pellet and the sinter, the oxide phase is dispersed in the porous metallic iron phase. Therefore, the elemental concentrations are discrete because of the presence of oxides, metals, and voids. The oxide phases exhibit high concentrations of Ca, Si, and O, which are almost constant along the analytical lines for each ore. No interaction between the sinter and the basic pellet is evident at 1200°C as the concentrations change sharply through the interface. Figure 6 shows the SEM images of the basic pellet/sinter samples after the heating to 1450°C and the results of line analysis of Ca, Si, O and Fe under the same analysis conditions as in Fig. 5. In the SEM images, the white phase is metallic iron, the light gray phase is wüstite phase, and the dark gray phase is CaO–SiO2–FeO system oxide phase. The wüstite phase is dendrite-like in the CaO–SiO2–FeO oxide, suggesting that it precipitated during the cooling process of the sample. Therefore, the two oxide phases formed a single liquid phase at 1450°C. The size of the precipitated wüstite phase is decreasing from the bulk of the basic pellet to the interface and then to the bulk of the sinter, suggesting a gradual change in the liquid phase composition. From the line analysis results, oxide phases with high concentrations of Ca, Si, and O are almost constant of both ores. In the case of Ca—for which a difference in concentration was observed between the basic pellet and the sinter at 1200°C—almost no difference in concentration is observed between the two ores at 1450°C. Furthermore, the results of the line analysis of Si show that Si migrates about 1 mm from the basic pellet side to the sinter side in the sample after reduction in Fig. 5(b), while Si is detected high on the sinter side in the sample after heating up in Fig. 6(c), indicating that Si migrates at least 8 mm from the interface of the two ores to the sinter side. During the temperature increase to 1450°C, the oxides of both ores melt, Ca and Si diffuse near the interface, and the oxides near the interface are assimilated.
Similar heating experiments were conducted from 1200 to 1450°C with the acidic pellet at the top and sinter at the bottom; however, no clear change in sample shape was observed during heating. Figure 7 shows the photographs of the acidic pellet/sinter samples before and after heating. Before heating (Fig. 7(a)), the acidic pellet was on present the upper part—which was mainly metallic iron on the cut surface of the sample—while the oxide phase was present in the central part of the pellet; the sinter was present in the lower part. As shown in Fig. 7(b), the shape of the pellet after heating is almost the same as that before heating. The surface is smooth and shiny, indicating that part of the oxide phase has melted. As in the heating experiment with the basic pellet and the sinter contact, slag melt was observed on the zirconia substrate after the experiment. Therefore, the void fraction of the entire reduced ore is expected to increase as the temperature was raised. In contrast, the sinter appears to shrink slightly.
The elemental mapping results of the cross sections of the sinter (left side) and the acidic pellet (right side), which were reduced by heating up to 1200°C in a reducing atmosphere, are shown in Fig. 8. The high-iron-concentration area in the sinter is low in oxygen, indicating the formation of metallic iron, whereas the acidic pellet is not partially reduced. As shown in Fig. 8, the iron phase is assimilated at the interfaces between the acidic pellet and the sinter as well as at that between the basic pellet and the sinter, while the oxide phase migrates slightly. The line analysis results for the Ca, Si, O, and Fe concentrations of the samples before and after heating are shown in Fig. 9. The line analysis conditions were set as follows: acceleration voltage, 15 kV; irradiation current, 10 nA; beam diameter, 0 μm; analysis interval, 6 μm. As shown in Fig. 9, the Si and O concentrations in the sample before heating are almost constant along the analysis line on the sinter side, whereas the Si concentration is reduced because of the partially unreduced wüstite phase on the acidic pellet side. However, the Si concentration near the interface is comparable to that of the sinter. In contrast, the Ca concentration is constant along the analysis line of the sinter but decreases across the interface with the acidic pellet. The Ca from the sinter diffuses by approximately 0.3 mm, beyond which it is almost undetectable. Figure 10 shows the SEM images of the acid pellet/sinter samples after heating to 1450°C and the results of line analysis for Ca, Si, O and Fe under the same analysis conditions as in Fig. 9. As same as Fig. 6(b), the white phase is metallic iron, the light gray phase is wüstite, and the dark gray phase is oxide phase of CaO–SiO2–FeO system in the SEM image. The morphology of the wüstite phase suggests that two oxide phases were in a single liquid phase at 1450°C as in Fig. 6(b). The precipitation behavior of wüstite phase varies between the bulk of the acidic pellet and the region near the interface between the acidic pellet and the sinter, as a result of changes in the liquid phase composition caused by the migration of elements from the sinter. The analysis of the acidic pellet/sinter sample after heating to 1450°C in Fig. 10(c) shows that the elemental concentrations on both the acidic pellet and the sinter side demonstrate different values depending on the position. Considering the oxide phase, where Ca, Si, and O show high concentrations, Si and O are almost constant in the analysis lines for both ores. Ca is almost constant on the analysis line of the sinter; however, near the interface with the acidic pellet, the Ca concentration decreases monotonically. The area of Ca that migrated from the sinter is approximately 2 mm from the acidic pellet side. The oxides melt during the temperature increase to 1450°C and assimilate each other near the interface.
Although the migration of elements during this heating process might be affected by contact area and other factors, this experiment suggests that the onset and extent of are important. This is because the dripping of the melt accelerates the mixing of the oxide phases of both ores. Figure 11 shows the relationship between the reduction degree of each ore and the liquid-phase ratio of the oxide phase as calculated by FACTSage 8.09) using the FToxide database. At a reduction degree of approximately 90%—near to the reduction ratio of this study sample—the liquid phase begins to form at approximately 1150°C for both the acidic and basic pellets, and reaches 20% at 1400°C. In the case of the sinter, the liquid phase is almost absent below 1250°C and sharply exceeds 20% as the temperature is above 1250°C. At 1200–1450°C temperature range in this experiment, the formation of the liquid phase is limited; however, above 1250°C, the oxide is present in the solid–liquid coexistence in the iron-phase structure of the pellets and the sinter. At the contact points of the pellets and the sinter, the liquid phases are in contact, allowing for the interdiffusion of the components and homogenization of the Ca concentration at the contact interface (Figs. 6 and 10). However, owing to the oxide phase in the iron-phase structure and the high solid-phase fraction of the oxide phase, the interdiffusion between the liquid phases is limited, resulting in a maximum diffusion of Ca of approximately 4 mm under the heating conditions employed in the experiment. Also, this experiment revealed that elemental migration between the acidic pellet and the sinter is a shorter distance than that between the basic pellet and the sinter when the various ores are reduced in a blast furnace under hydrogen-enriched condition. This change in the migration distance depends on the composition of the pellets when they are prepared, and the elemental migration between the acidic pellet/sinter is considered to be shorter. However, the slag composition varies complexly due to differences in the local reduction degree among various pellets and sinters, as well as liquid mixing after melt generation. Therefore, additional research is needed to identify the specific physical properties that influence migration distance. The influence of slag interdiffusion on the deformation of the ore as a whole was not observed because the iron phase was the main structure. Since some studies10,11) have shown that carburization into pure iron increases when using a reducing gas containing hydrogen (CO–H2 mixture), the carbon concentration in iron produced by reduction might be higher under hydrogen-enriched conditions than in conventional blast furnace operations. Nevertheless, because no change in iron shape was observed in this experiment, it is considered that the shape of reduced iron ores is maintained even under hydrogen-enriched conditions and does not lead to significant pressure loss in the cohesive zone. In this study, the shape changes and elemental migration of reduced iron ore due to reactions between different ores were investigated. Additionally, in actual operation, there is a contact surface between coke and ore. In an experiment12) where pure iron was placed on a coke bed and heated, the amount of carburization was small and no significant change in the shape of the iron was observed. However, the interaction of coke with a single or multiple ores might affect the carburization and shape change. This is an area for further study.
The reduction behavior of sinter/acidic pellet and sinter/basic pellet mixtures under conditions simulating a hydrogen-enriched blast furnace and the softening and melting behaviors at the ore contact interface upon reheating were investigated. The following findings were obtained:
(1) When ore beds were reduced by heating up to 1200°C under hydrogen-enriched blast furnace conditions, ore particles are bound together by metallic iron generated from the ore. In contrast, the interparticle diffusion of gangue components such as CaO does not progress.
(2) When the sample containing pellets and sinter reduced by heating up to 1200°C under hydrogen-enriched blast furnace conditions was heated to 1450°C, no ore deformation is observed due to the presence of metallic iron. The oxide partially melts, and interdiffusion of the gangue components occurs between the particles of more than 2 mm.
The authors have no conflicts to disclose.
Experiments were conducted on the reduction of sinters, basic pellets, and acidic pellets under conventional and hydrogen-enriched blast furnace conditions. SEM observations of ore microstructures revealed that the iron produced under hydrogen-enriched conditions was finer than that produced under conventional conditions, and that the type of iron ore affected the microstructure of the iron produced. This materials is available on the journal website at https://doi.org/10.2355/isijinternational.ISIJINT-2024-363.
This study was carried out as part of a research group of the ISIJ. The authors thank the lead author, Prof. Ohno of Kyushu University, and all those involved.
