Empirical Study on Reduction Behavior and Metallurgical Properties of Vanadia–Titania Magnetite in Blast Furnace

: The loss of permeability affects the reduction of the ferrous burden in the cohesive zone of a blast furnace (BF). Vanadia–titania magnetite (VTM) burden of various chemical compositions have different metallurgical properties. The reduction and softening-melting-dripping properties of different kinds of VTM were investigated. The results showed that the core of sinter or pellet is indirectly reduced to wustite and (Fe,Ti)O x , and the periphery contains interlinked metallic iron and CaSiO 3 in the cohesive zone. Wustite and (Fe,Ti)O x are directly reduced in the melting-dripping zone. The aggregate (Fe, V, Cr) present in the non-dripping causes a loss of valuable components. With the increase in TiO 2 content, the substrate phase of molten slag changes from melilite to titanaugite, and the mass of dripping decreases gradually. In addition, the permeability index S increased and the melting zone widened, which indicates that the increase in TiO 2 content negatively affected the melting-dripping performance. The mass of the dripping is directly proportional to the pellet ratio. Considering the adverse effect of TiO 2 on softening-melting-dripping properties, it is recommended that high TiO 2 VTM is smelted while mixed with ordinary ores or with an increased pellet ratio in the burden structure.


Introduction
Vanadia-titania magnetite (VTM) contains various kinds of valuable elements (Fe, V, Ti, and Cr) and has a high comprehensive utilization value [1,2]. In recent years, the price of vanadium products has remained high, and the smelting of VTM brings huge profits for steel companies. A blast furnace (BF) is the most common method for smelting VTM worldwide [3]. The shape and position of the cohesive zone of the BF, as well as its efficiency, are determined by the melt-down of ores [3,4]. In general, when the thickness of the cohesive zone decreases, the efficiency of the blast furnace increases. The softening and melting temperatures of the burden determine the thickness of the cohesive zone, so the smallest possible difference between them is desired. A reduction degree of a ferrous burden is typically above 50% in the cohesive zone of the BF [4], but the reduction degree of VTM burden is typically lower than this [5]. This leads to the aggravation of the reduction task in the melting-dripping zone. The smelting of VTM is difficult because of the role of titanium in the blast furnace. Thus, the reduction behavior and metallurgical properties of various VTM burden differ from each other markedly.
Titanium is one of the most important influencing factors in the smelting process, which has been studied by domestic and foreign researchers [6][7][8][9][10][11][12]. It not only affects the quality and yield of sinter, but also affects the smelting process in the BF. Cheng [13] studied the effect of TiO 2 on the crushing strength and smelting mechanism of high-chromium VTM pellets. It was found that the crushing strength is closely related to the mineral phase and microstructure, and that the crushing strength decreases when increasing the TiO 2 content. Tang and Zhou [5,14] studied the effect of TiO 2 on the sintering behavior of VTM, which confirmed that the optimum quicklime dosage, mixture moisture, wetting time, and granulation time are 5 wt%, 7.5 wt%, 10 min, and 5-8 min, respectively. Yue [15][16][17] studied the rheological evolution of Ti-bearing slag with different volume fractions of TiN. It was found that increasing the TiN volume increased the viscosity and the shear stress substantially. Ledzki [18] investigated the potential for the formation of high melting titanium compounds in the metal-slag system by TiO 2 reduction from slag phase by solid carbon and carbon dissolved in iron. Furthermore, the abovementioned studies have focused on the agglomeration or metallurgical properties of the specific VTM under laboratory conditions.
In this work, the reduction behavior of VTM was explained by analyzing the phase composition and microstructure of slag, non-dripping, and dripping at various stages. The empirical studies on the reduction behavior and metallurgical properties of varieties of VTM provide a theoretical basis and technical support for BF smelting VTM.

Raw Materials
V-Ti sinter and V-Ti pellet were used in this research as ferrous materials, which were obtained from multiple iron and steel companies in China. The chemical compositions of the ferrous materials and mixed burden (pellet ratio: HJ-45%, CJ-29%, PJ-12.6%) are listed in Table 1. The V-bearing phase is Fe 2 VO 4 , and the Ti-bearing phase is Fe 2 TiO 4 , FeTiO 3 , and Fe 2 TiO 5 in sinter and pellet. The chemical compositions of coke are listed in Table 2. The sizes of the V-Ti sinter, V-Ti pellet, and coke were all 10-12.5 mm. All samples were dried at 110 • C.

Experiment Methods
Softening-melting-dripping experiments were carried out on the samples to measure their high-temperature properties. The schematic experimental apparatus is shown in Figure 1a. The device comprises a heating system, a gas control system, a temperature control system, and a data recording system. MoSi 2 heating elements were used as the core components in the heating system to heat and melt the burden. An R-type thermocouple was placed at the bottom of the compression bar to control the experimental temperature. In order to simulate the blast furnace burden conditions, 500 g of mixed burden was put in the crucible and layers of coke with a depth of 35 mm and 17 mm were placed below and above the samples, respectively. The displacement sensor measured the displacement changes and contraction of the burden when the experiment started. Experimental conditions are shown in Table 3. The heating rates were 10 °C/min below 900 °C, 2 °C/min from 900 °C to 1100 °C, and 5 °C from 1100 °C to the end. To better understand the reduction behavior and softening-melting characteristics of the mixed burden, an interrupted test was also carried out at 1300 °C. The gas was stopped, and N2 was introduced immediately to prevent the reaction from continuing after the test. When the temperature drops to room temperature, the soft slag, molten slag, non-dripping, and dripping products were collected. The chemical composition and microstructure of collected samples were analyzed by chemical analysis and scanning electron microscopy/energy dispersive X-ray spectroscopy (SEM-EDS).

Reduction Behavior of Cohesive Zone
The experiment is terminated when the Burden is heated to 1300 °C, at which point the soft slag in the crucible is taken out. Figure 2a shows the macroscopic morphology of the soft slag. Under the action of external pressure, the upper coke moves downward, and a small amount of molten slag with good fluidity fills the upper coke void, so that the soft slag has been melted into a whole. The outermost edge of the soft slag is a ring of metal shell. When the soft slag is broken, it can be seen from a cross section that the interior is compact, and that the original shape of sinter and pellet are faintly visible with a metal layer between them. Therefore, the outer edge of sinter and pellet is the first to be reduced. In order to simulate the blast furnace burden conditions, 500 g of mixed burden was put in the crucible and layers of coke with a depth of 35 mm and 17 mm were placed below and above the samples, respectively. The displacement sensor measured the displacement changes and contraction of the burden when the experiment started. Experimental conditions are shown in Table 3. The heating rates were 10 • C/min below 900 • C, 2 • C/min from 900 • C to 1100 • C, and 5 • C from 1100 • C to the end. To better understand the reduction behavior and softening-melting characteristics of the mixed burden, an interrupted test was also carried out at 1300 • C. The gas was stopped, and N 2 was introduced immediately to prevent the reaction from continuing after the test. When the temperature drops to room temperature, the soft slag, molten slag, non-dripping, and dripping products were collected. The chemical composition and microstructure of collected samples were analyzed by chemical analysis and scanning electron microscopy/energy dispersive X-ray spectroscopy (SEM-EDS). Table 3. Temperature profile and reduction atmosphere of experiment.

Reduction Behavior of Cohesive Zone
The experiment is terminated when the Burden is heated to 1300 • C, at which point the soft slag in the crucible is taken out. Figure 2a shows the macroscopic morphology of the soft slag. Under the action of external pressure, the upper coke moves downward, and a small amount of molten slag with good fluidity fills the upper coke void, so that the soft slag has been melted into a whole. The outermost edge of the soft slag is a ring of metal shell. When the soft slag is broken, it can be seen from a cross section that the interior is compact, and that the original shape of sinter and pellet are faintly visible with a metal layer between them. Therefore, the outer edge of sinter and pellet is the first to be reduced.

Reduction Behavior of Melting-Dripping Zone
When the Burden reaches the drop temperature, it begins to drop. After 30 min, the experiment ends and the samples in the graphite crucible and bowl are taken out. Figure  3 shows the macroscopic morphology for the molten slag, non-dripping, and dripping. Under the action of external pressure, the molten slag overflows from the small holes and sits on top of the tableting after cooling. The surface of the molten slag is relatively smooth and dense, as shown in Figure 3a. The mobility of the non-dripping is so poor that it cannot drip and stay at the bottom of the crucible. Figure 3b shows that the non-dripping is a dense metal block with coke embedded in it. The dripping drips from the crucible into the graphite bowl. Due to the action of gravity, the dripping is scattered by sputtering, and the surface is in the form of small particles, as shown in Figure 3c.

Reduction Behavior of Melting-Dripping Zone
When the Burden reaches the drop temperature, it begins to drop. After 30 min, the experiment ends and the samples in the graphite crucible and bowl are taken out. Figure 3 shows the macroscopic morphology for the molten slag, non-dripping, and dripping. Under the action of external pressure, the molten slag overflows from the small holes and sits on top of the tableting after cooling. The surface of the molten slag is relatively smooth and dense, as shown in Figure 3a. The mobility of the non-dripping is so poor that it cannot drip and stay at the bottom of the crucible. Figure 3b shows that the non-dripping is a dense metal block with coke embedded in it. The dripping drips from the crucible into the graphite bowl. Due to the action of gravity, the dripping is scattered by sputtering, and the surface is in the form of small particles, as shown in Figure 3c.

Reduction Behavior of Melting-Dripping Zone
When the Burden reaches the drop temperature, it begins to drop. After 30 min, the experiment ends and the samples in the graphite crucible and bowl are taken out. Figure  3 shows the macroscopic morphology for the molten slag, non-dripping, and dripping. Under the action of external pressure, the molten slag overflows from the small holes and sits on top of the tableting after cooling. The surface of the molten slag is relatively smooth and dense, as shown in Figure 3a. The mobility of the non-dripping is so poor that it cannot drip and stay at the bottom of the crucible. Figure 3b shows that the non-dripping is a dense metal block with coke embedded in it. The dripping drips from the crucible into the graphite bowl. Due to the action of gravity, the dripping is scattered by sputtering, and the surface is in the form of small particles, as shown in Figure 3c. The SEM-EDS images of the HJ, CJ, and PJ molten slag are shown in Figure 4. The HJ molten slag has a simple structure and its base phase is melilite, which mainly consists of Mg, Al, Si, Ca and O. The small metallic iron and CaTiO3 is embedded in the melilite. In Figure 4b, the CJ molten slag mainly consist of melilite, titanaugite, and CaTiO3. The CaTiO3 is embedded in the melilite in the form of granules and leaves. In the PJ molten slag, it can be seen that CaTiO3 particles are connected to each other and that their particle The SEM-EDS images of the HJ, CJ, and PJ molten slag are shown in Figure 4. The HJ molten slag has a simple structure and its base phase is melilite, which mainly consists of Mg, Al, Si, Ca and O. The small metallic iron and CaTiO 3 is embedded in the melilite. In Figure 4b, the CJ molten slag mainly consist of melilite, titanaugite, and CaTiO 3 . The CaTiO 3 is embedded in the melilite in the form of granules and leaves. In the PJ molten slag, it can be seen that CaTiO 3 particles are connected to each other and that their particle sizes are large. The basal phase titanaugite consist of O, Mg, Al, Si, Ca, and Ti. The metallic iron is embedded in the titanaugite. The TiC phase surrounds the metallic iron particles. The SEM-EDS images of the HJ, CJ, and PJ molten slag are shown in Figure 4. The HJ molten slag has a simple structure and its base phase is melilite, which mainly consists of Mg, Al, Si, Ca and O. The small metallic iron and CaTiO3 is embedded in the melilite. In Figure 4b, the CJ molten slag mainly consist of melilite, titanaugite, and CaTiO3. The CaTiO3 is embedded in the melilite in the form of granules and leaves. In the PJ molten slag, it can be seen that CaTiO3 particles are connected to each other and that their particle sizes are large. The basal phase titanaugite consist of O, Mg, Al, Si, Ca, and Ti. The metallic iron is embedded in the titanaugite. The TiC phase surrounds the metallic iron particles. The SEM image of the non-dripping is shown in Figure 5. It can be seen that the blocky (Fe, V, Cr) and banded C are embedded in the metallic iron. The presence of C causes the metallic iron to become less fluid, so that it cannot drip from the crucible. Fe, The SEM image of the non-dripping is shown in Figure 5. It can be seen that the blocky (Fe, V, Cr) and banded C are embedded in the metallic iron. The presence of C causes the metallic iron to become less fluid, so that it cannot drip from the crucible. Fe, V, and Cr aggregates are present in the non-dripping, resulting in a decrease in the recovery rate of valuable components.
The Richardson oxide free energy diagram of related elements is drawn using FactSage 7.0 software (ThermFact LTD, Montreal, QC, Canada), as shown in Figure 6. It can be seen that V 2 O 3 and Cr 2 O 3 cannot be directly (with C) reduced at temperatures below 1250 • C. Thus, they are present in wustite and (Fe,Ti)O x . With the increase in temperature, the direct reduction of oxides is further carried out in the melting-dripping zone. The reaction equations are shown in Equations (4), (6) and (7). Vanadium and Chromium are contained in dripping and non-dripping. However, the reduction temperature of titanium oxide is so high that it cannot be reduced, and it reacts to form CaTiO 3 and titanaugite in the molten slag, as shown in Equations (8)- (9).

Softening-Melting-Dripping Performance
The softening-melting-dripping performance of the burden has a dramatic effect on the reduction process, and the permeability and distribution of gas there is also affected. In this work, softening start temperature (T 4 ), softening end temperature (T 40 ), melting start temperature (T s ), dripping temperature (T d ), and pressure drop (∆P m ) are measured to evaluate the softening-melting-dripping properties. T 4 and T 40 are the temperatures at which the shrinkage degree of the burden reaches 4% and 40%, respectively. T s is the temperature when the gas pressure drop begins to increase rapidly. When the temperature reaches T d , the molten iron drips from the crucible. The softening (∆T 1 = T 40 − T 4 ) and melting (∆T 2 = T d − T s ) zones need to be noted. Permeability index (S) is also an important characteristic value. S can be calculated by Equation (10).
where P m is pressure difference at any time, Pa, and ∆P s is the starting melting pressure difference, Pa. The softening-melting-dripping characteristics of the burden are shown in Table 4. T 4 of the burden mainly depends on the melting point of the low melting point slag phase in the reduction process. T 4 and T 40 are affected by so many factors, such as CaO, SiO 2 , TiO 2 , and burden structure, that no obvious pattern has been found. The high T 4 and narrow ∆T 1 are favorable for blast furnace smelting. It can be found that the PJ burden has the highest softening start temperature (T 4 ) and the narrowest softening zone (∆T 1 ) of the three kinds of burden. Therefore, the softening property of the PJ burden is the best. Table 4. Performance data of softening-melting-dripping of burden. In Table 4, it can be found that Ts is closely related to the content of TiO 2 in the burden as seen in Figure 7. With the increase of TiO 2 content, Ts decreases gradually. When the content of TiO 2 in the burden begins to increase, although more high-melting-point CaTiO 3 and titangarnet will be generated, the low-melting-point titanaugite also increases. Titanaugite increases permeability resistance at lower temperatures, leading to a decrease in Ts to a certain extent. PJ burden has the highest T d and the widest ∆T 2 . This is because TiO 2 increases the viscosity of slag and affects the drop of molten iron. When the appropriate fluidity is reached by increasing the temperature, the molten iron drips smoothly. A lower Ts and a wider ∆T 2 are disadvantageous in blast furnace smelting.  The results of the permeability index (S) are shown in Table 4. The smaller the S, the better the droplet performance. It can be seen that the S of the PJ burden is much larger than the others, and that its permeability is the worst. Figure 8 shows the dripping mass of the HJ, CJ, and PJ burden. It can be found that the dripping mass is inversely proportional to the content of TiO 2 in the burden. When the content of TiO 2 is high, the metallic iron is wrapped by the surrounding Ti(C,N), resulting in the loss of dripping. In addition, reducibility is also an important factor affecting the drop of molten iron. The poor reduction of the burden increases the reduction process and difficulty. The pellet ratio of PJ burden with good reducibility is the lowest, so its dripping mass is the lowest.

Conclusions
In this study, the soft slag, molten slag, non-dripping and dripping were analyzed by SEM-EDS. The reduction behavior and softening-melting-dripping performance of the burden was investigated in a cohesive and a melting-dripping zone. The conclusions are as follows: (1) Iron oxides are reduced in the cohesive zone. Wustite and coated (Fe,Ti)Ox are formed in the core of pellet and sinter, while iron oxides and CaFeSiO4 at the periphery is reduced to formed metallic iron and CaSiO3.
(2) With the different content of TiO2 in the burden, the substrate phase of the molten slag is different. The aggregate (Fe, V, Cr) exists in the non-dripping, and metallic iron wrapped by TiC exists in the molten slag, which causes the loss of valuable components.
(3) The mass of the dripping is directly proportional to the pellet ratio and is inversely proportional to the TiO2 content.

Conclusions
In this study, the soft slag, molten slag, non-dripping and dripping were analyzed by SEM-EDS. The reduction behavior and softening-melting-dripping performance of the burden was investigated in a cohesive and a melting-dripping zone. The conclusions are as follows: (1) Iron oxides are reduced in the cohesive zone. Wustite and coated (Fe,Ti)O x are formed in the core of pellet and sinter, while iron oxides and CaFeSiO 4 at the periphery is reduced to formed metallic iron and CaSiO 3 .
(2) With the different content of TiO 2 in the burden, the substrate phase of the molten slag is different. The aggregate (Fe, V, Cr) exists in the non-dripping, and metallic iron wrapped by TiC exists in the molten slag, which causes the loss of valuable components.
(3) The mass of the dripping is directly proportional to the pellet ratio and is inversely proportional to the TiO 2 content.
(4) PJ burden with high TiO 2 content has a high T d and a wide ∆T 2 , which is not conducive to blast furnace smelting.