Desulfurization Ability of Blast Furnace Slag Containing High Al 2 O 3 at 1773 K

: With the consumption of high-quality iron ore resources, the grade of iron ore raw used in the iron making process has gradually decreased. The high aluminum raw materials has led to an increase in the Al 2 O 3 content in the blast furnace slag, which has affected the normal operation of the blast furnace. The activity of the components involved in the desulfurization reaction is an important factor affecting the desulfurization process. In this paper, the effects of B ( w (CaO)/ w (SiO 2 )) and w (MgO)/ w (Al 2 O 3 ) on the desulfurization ability of a CaO-SiO 2 -MgO-Al 2 O 3 quaternary blast furnace slag system were studied by using a double-layer graphite crucible to simulate the process of molten iron dropping through the slag at 1773 K. The desulfurization reaction mechanism of high alumina blast furnace slag is explained from the aspects of slag structure and component activity, which provides a theoretical basis for the reasonable selection of a slag making system under the condition of a high alumina content. The effects of different B and w (MgO)/ w (Al 2 O 3 ) on the slag structure 2343 analyzed by Fourier transform infrared spectroscopy (FT-IR). The results show that when w (MgO)/ w (Al 2 O 3 ) = 0.50 and w (Al 2 O 3 ) = 20%, the desulfurization ability and MgO activity of slag increased with the increase of B ; when w (Al 2 O 3 ) = 20% and B = 1.30, the desulfurization ability and MgO activity of the slag increased with the increase of w (MgO)/ w (Al 2 O 3 ). The FT-IR analysis showed that with the increase of B or w (MgO)/ w (Al 2 O 3 ), the dissociated free oxygen ions (O 2- ) in the slag increased, and the dissociated free oxygen ions (O 2- ) interacted with the bridging oxygen (O 0 ) of silicate, which made the complex Si-O structure in the slag gradually depolymerize and increase the desulfurization ability. bond, and the complex silicate structure of the slag disintegrates into a simple silicate structure. The desulfurization ability of the slag is gradually enhanced. (O 0 ) decreases and the activity of the non-bridge oxygen (O − ) increases. The interaction between free oxygen ions (O 2 − ) and silicate bridge oxygen (O 0 ) breaks the complex Si-O-Si bond, decomposes the complex silicate structure of the slag into a simple silicate structure, and improves the desulfurization ability of the slag.


Introduction
In recent years, with the extensive use of high alumina iron ore, the content of Al 2 O 3 in slag has increased greatly, which inevitably affects the metallurgical properties of blast furnace slag [1][2][3][4][5][6][7][8][9][10]. Among these metallurgical properties, the desulfurization ability of blast furnace slag plays an important role in the quality of hot metal. Many scholars have done a lot of research on the metallurgical properties of blast furnace slag. Shankar A's study on the desulfurization ability of a CaO-SiO 2 -MgO-Al 2 O 3 blast furnace slag system shows that the desulfurization ability of slag increases with the increase of slag B(w(CaO)/w(SiO 2 )) or w(MgO) content [11]. The desulfurization ability of the CaO-SiO 2 -MgO-Al 2 O 3 quaternary slag system at 1773~1823 K was measured by Ma X D et al. [12,13]. In their study, the desulfurization ability of slag increased with the increase of B(w(CaO)/w(SiO 2 )) or w(MgO) content. The desulfurization ability of CaO-SiO 2 -MgO-Al 2 O 3 quaternary high MgO blast furnace slag system was studied by Condo et al. [14]. Many researchers such as Xin H, Wang Y, and Nzotta M M [15][16][17] have also studied the desulfurization ability of slag. Wang L J [18] and others analyzed the structure of chromium containing slag using Raman spectroscopy. The results show that the spectral results are consistent with the results of the sulfide capacity measurements. Lee S [19] studied the viscosity behavior of FeO containing slag melt and a slag structure. Choi J S [20] studied the cationic effect of charge compensation on the sulfide capacity of aluminosilicate slag. Choi J S [21] studied the structure-property relationship amphoteric oxide systems via phase stability and ionic structural analysis. However, for the high alumina CaO-SiO 2 -MgO-Al 2 O 3 slag system, there are few reports on the desulfurization reaction mechanism of high alumina blast furnace slag from the perspective of the slag structure. Therefore, it is necessary to explain the desulfurization reaction mechanism from the perspective of the slag structure, so as to provide a scientific basis and theoretical guidance for the application of high alumina iron ore in the blast furnace.
The main factors affecting slag desulfurization are the furnace temperature and slag chemical composition. The hearth temperature should not be changed excessively. Therefore, improving the slag composition is the main means to improve the desulfurization ability of slag. In this paper, the effects of B and w(MgO)/w(Al 2 O 3 ) on the desulfurization ability of a CaO-SiO 2 -MgO-Al 2 O 3 quaternary blast furnace slag system were studied using a double-layer graphite crucible to simulate the process of molten iron dropping through slag. According to the experimental results, the effects of B and w(MgO)/w(Al 2 O 3 ) on the desulfurization ability of blast furnace slag were discussed. FT-IR was used to analyze the effect of different B and w(MgO)/w(Al 2 O 3 ) on the slag structure. Further study on the microstructure of blast furnace slag can better understand the relationship between the slag chemical composition and metallurgical properties, and can provide a theoretical basis for the reasonable selection of a slag making system under the condition of a high Al 2 O 3 content.

Experimental Principle
The desulfurization reaction in the blast furnace starts after the initial slag is formed. It is mainly carried out in the hearth or the drip zone, and is finally completed in the hearth. The desulfurization reaction is a process of ion migration at the interface between the slag and iron. For the desulfurization reaction, it can be considered that atomic sulfur, which was originally neutral in molten iron, was obtained at the interface between slag and iron, and the electrons in slag became S 2− . The oxygen anion (O 2− ) in the slag loses electrons at the interface, becomes neutral atomic oxygen, and enters into molten iron. The schematic diagram of the desulfurization reaction of slag is shown in Figure 1. The desulfurization of the blast furnace slag mainly makes sulfur enter the slag in the form of sulfide through the reaction of basic oxide in slag and sulfur in molten iron. The theory of molecular ion coexistence holds that only free (Ca 2+ + O 2− ) and (Mg 2+ + O 2− ) ions have a desulfurization ability [22,23]. S 2− exists in slag in the form of (Ca 2+ + S 2− ) and (Mg 2+ + S 2− ) ion pairs, and participates in the desulfurization reaction [22,23].
According to the ion theory, the reaction equation of slag desulfurization can be expressed as follows [22,23]: The sulfur distribution ratio (L S ) is a common evaluation index of the desulfurization ability, and its definition is shown in Equation (4): where (%S) and [%S] are the sulfur content of the slag and hot metal, respectively.
The sulfur distribution ratio (LS) is a common evaluation index of the desulfurization ability, and its definition is shown in Equation (4): where (%S) and [%S] are the sulfur content of the slag and hot metal, respectively.

Experimental Procedure
The desulfurization experiment was carried out in the high temperature tubular furnace (RTW-10, Northeastern University, China), and the temperature of the tubular furnace was controlled by a PID controller. The special double-layer graphite sleeve crucible was used as the container in the experiment. The outer diameter of the upper and lower graphite crucibles was 60 mm, the inner diameter of the upper and lower graphite crucibles was 44 mm, the height of the lower graphite crucible was 40 mm, and the height of the upper graphite crucible was 80 mm. The bottom of the upper crucible was plugged with a 120 mm high graphite plug. The experimental device is shown in Figure 2.

Experimental Procedure
The desulfurization experiment was carried out in the high temperature tubular furnace (RTW-10, Northeastern University, China), and the temperature of the tubular furnace was controlled by a PID controller. The special double-layer graphite sleeve crucible was used as the container in the experiment. The outer diameter of the upper and lower graphite crucibles was 60 mm, the inner diameter of the upper and lower graphite crucibles was 44 mm, the height of the lower graphite crucible was 40 mm, and the height of the upper graphite crucible was 80 mm. The bottom of the upper crucible was plugged with a 120 mm high graphite plug. The experimental device is shown in Figure 2. Figure 3 shows the quaternary phase diagram of CaO-SiO 2 -MgO-Al 2 O 3 [24]. It can be seen from Figure 3 that the suitable slag composition for the blast furnace iron smelting should be in the primary crystal region of plagioclase (2CaO·MgO·2SiO 2 ) [25]. In this region, the melting temperature is low and the isotherm distribution is sparse. Therefore, in this paper, an isothermal single variable desulfurization experiment was carried out for the contents of B and w(MgO)/w(Al 2 O 3 ). The chemical reagents (CaO, CAS no. 1305-78-8, AR, 98.0% purity; SiO 2 , CAS no. 7631-86-9, AR, 98.0% purity; MgO, CAS no. 1309-48-4, AR, 98.0% purity; Al 2 O 3 , CAS no. 1344-28-1, AR, 98.0% purity) employed as the collectors were procured from Sinopharm Reagent Group (Ningbo, China). The slag composition of the desulfurization experiment is shown in Table 1. The eight points in Figure 3 represent the eight groups of blast furnace slag systems in Table 1. In the preparation stage of the experiment, when the furnace temperature of the muffle furnace was raised to 1523 K, the graphite crucible containing the chemical reagents CaO, SiO 2 , MgO, and A1 2 O 3 were put into the muffle furnace for pre-melting treatment to prepare the pre-melted slag. The heating process of the muffle furnace is shown in Figure 4. nace (RTW-10, Northeastern University, China), and the temperature of the tubular furnace was controlled by a PID controller. The special double-layer graphite sleeve crucible was used as the container in the experiment. The outer diameter of the upper and lower graphite crucibles was 60 mm, the inner diameter of the upper and lower graphite crucibles was 44 mm, the height of the lower graphite crucible was 40 mm, and the height of the upper graphite crucible was 80 mm. The bottom of the upper crucible was plugged with a 120 mm high graphite plug. The experimental device is shown in Figure 2.  Figure 3 shows the quaternary phase diagram of CaO-SiO2-MgO-Al2O3 [24]. It can be seen from Figure 3 that the suitable slag composition for the blast furnace iron smelting should be in the primary crystal region of plagioclase (2CaO·MgO·2SiO2) [25]. In this region, the melting temperature is low and the isotherm distribution is sparse. Therefore, in this paper, an isothermal single variable desulfurization experiment was carried out for the contents of B and w(MgO)/w(Al2O3). The chemical reagents (CaO, CAS no. 1305-78-8, AR, 98.0% purity; SiO2, CAS no. 7631-86-9, AR, 98.0% purity; MgO, CAS no. 1309-48-4, AR, 98.0% purity; Al2O3, CAS no. 1344-28-1, AR, 98.0% purity) employed as the collectors were procured from Sinopharm Reagent Group (Ningbo, China). The slag composition of the desulfurization experiment is shown in Table 1. The eight points in Figure 3 represent the eight groups of blast furnace slag systems in Table 1. In the preparation stage of the experiment, when the furnace temperature of the muffle furnace was raised to 1523 K, the graphite crucible containing the chemical reagents CaO, SiO2, MgO, and A12O3 were put into the muffle furnace for pre-melting treatment to prepare the pre-melted slag. The heating process of the muffle furnace is shown in Figure 4.     In the experiment, the iron in a graphite crucible came from a steel plant. The composition of iron before the experiment is shown in Table 2. Before the start of the experiment, iron and pre-melted slag were taken as samples with a mass ratio of 3:1. The slag sample was placed in the lower graphite crucible, while the iron sample was placed in the upper graphite crucible. The special double-layer graphite crucible was put into the high temperature tube furnace one by one. The switch was tightened with a graphite stopper. After the slag sample and iron sample were completely dissolved, when the temperature in the high temperature tubular furnace rose to 1773 K, the graphite stopper switch was turned on to make the molten iron flow in from the gap between the graphite stopper and the crucible, pass through the slag layer, and then flow to the bottom of the lower crucible for deposition, so as to simulate the process of molten iron dropping through the slag. The content of the slag composition of the equilibrium experiment was as follows: w(CaO) = 39%, w(SiO2) = 30%, w(MgO) = 11%, and w(Al2O3) = 20%. The equilibrium times were set at 30, 60, 90, and 120 min, respectively. The relationship between the sulfur content in molten iron and the desulfurization time is shown in Figure 5. It can be seen from Figure  5 that after 80 min, the sulfur content in iron basically did not change. From 90-120 min, the sulfur content in iron decreased by 0.8%, which can be considered as the reaction basically reaching equilibrium. After 90 min of reaction, the crucible was taken out quickly and immersed in oil for rapid cooling to preserve the original state of the slag iron at the end of the reaction. During the experiment, high purity argon was introduced for protection, and then the samples were taken out to separate the slag and iron, and the sulfur content was analyzed to calculate the sulfur distribution ratio. The content of sulfur in the iron alloy was subjected to a chemical analysis using a PE Avio 500 Inductively Coupled Plasma-Atomic Emission Spectrometry (ICP-AES).  In the experiment, the iron in a graphite crucible came from a steel plant. The composition of iron before the experiment is shown in Table 2. Before the start of the experiment, iron and pre-melted slag were taken as samples with a mass ratio of 3:1. The slag sample was placed in the lower graphite crucible, while the iron sample was placed in the upper graphite crucible. The special double-layer graphite crucible was put into the high temperature tube furnace one by one. The switch was tightened with a graphite stopper. After the slag sample and iron sample were completely dissolved, when the temperature in the high temperature tubular furnace rose to 1773 K, the graphite stopper switch was turned on to make the molten iron flow in from the gap between the graphite stopper and the crucible, pass through the slag layer, and then flow to the bottom of the lower crucible for deposition, so as to simulate the process of molten iron dropping through the slag. The content of the slag composition of the equilibrium experiment was as follows: w(CaO) = 39%, w(SiO 2 ) = 30%, w(MgO) = 11%, and w(Al 2 O 3 ) = 20%. The equilibrium times were set at 30, 60, 90, and 120 min, respectively. The relationship between the sulfur content in molten iron and the desulfurization time is shown in Figure 5. It can be seen from Figure 5 that after 80 min, the sulfur content in iron basically did not change. From 90-120 min, the sulfur content in iron decreased by 0.8%, which can be considered as the reaction basically reaching equilibrium. After 90 min of reaction, the crucible was taken out quickly and immersed in oil for rapid cooling to preserve the original state of the slag iron at the end of the reaction. During the experiment, high purity argon was introduced for protection, and then the samples were taken out to separate the slag and iron, and the sulfur content was analyzed to calculate the sulfur distribution ratio. The content of sulfur in the iron alloy was subjected to a chemical analysis using a PE Avio 500 Inductively Coupled Plasma-Atomic Emission Spectrometry (ICP-AES).  In this paper, slag/metal equilibrium experiments were carried out for the deter nation of the activity of MgO in CaO-SiO2-MgO-Al2O3 melts for blast furnace slag at 1773 K, using Sn as the metal solvent and CaF2-MgO slag as reference slag. At equilibri the principal reactions between the molten Sn and reference slag or measured slag can expressed as reactions (5) and (6), respectively: (MgO)mea. + C(graphite) = [Mg]Sn + CO(g) (CO) and CO θ are the pressures of CO and standard atmosphere, respectiv and C is the activity of the carbon relative to the pure solid. When reaction (5) and r tion (6) are at the same conditions, equation K5 = K6 can be written as follows:  Figure 6. Fr the CaF2-MgO phase diagram, when the mass fraction of MgO is greater than 48%,

Experimental Procedure
In this paper, slag/metal equilibrium experiments were carried out for the determination of the activity of MgO in CaO-SiO 2 -MgO-Al 2 O 3 melts for blast furnace slag at T = 1773 K, using Sn as the metal solvent and CaF 2 -MgO slag as reference slag. At equilibrium, the principal reactions between the molten Sn and reference slag or measured slag can be expressed as reactions (5) and (6), respectively: where p (CO) and p θ CO are the pressures of CO and standard atmosphere, respectively, and a C is the activity of the carbon relative to the pure solid. When reaction (5) and reaction (6) are at the same conditions, equation K 5 = K 6 can be written as follows: where a ( The concentration of Mg in Sn is small enough to make the Henry's law suitable, so f [Mg] = 1. The schematic diagram of the experimental principle is shown in Figure 6. From the CaF 2 -MgO phase diagram, when the mass fraction of MgO is greater than 48%, it is the saturated region of MgO. Therefore, the composition of point S(w(CaF 2 ) = 45% and w(MgO) = 55%) was chosen as the reference slag in this study, so a ref(MgO) = 1. Thus, Equation (7) can be simply derived from Equation (8):

Experimental Principle
In order to measure the activity of MgO in the silicate melt, solvent metal (Sn ≥ 99.9%) and the slag (measure slag or reference slag) were equilibrated in a MoSi2 resistance furnace. The chemical reagents (CaO, CAS no. 1305-78-8, 99.99% purity; SiO2, CAS no. 7631-86-9, 99.99% purity; MgO, CAS no. 1309-48-4, 99.99% purity; Al2O3, CAS no. 1344-28-1, 99.99% purity; CaF2, CAS no. 7789-75-5, 99.99% purity) employed as the collectors were procured from Sinopharm Reagent Group (Ningbo, China). The slag composition of the activity experiment is shown in Table 1. A six-hole graphite crucibles was used in the experiment. About 5 g of solvent metal (Sn ≥ 99.9%) and 10 g of the slag (measure slag or reference slag) were filled in a graphite crucible. When the temperature of a tubular furnace sample rose to 1500 °C, CO was introduced. After the balance time was 24 h, the slag sample was oil cooled. Thereafter, the metal and the slag were carefully separated for chemical analyses using Inductively Coupled Plasma-Atomic Emission Spectrometry (ICP-AES).

Results and Discussion
In this paper, a double-layer graphite crucible was used to simulate the process of melting iron dropping through the slag. Isothermal single variable desulfurization experiments were carried out for B and w(MgO)/w(Al2O3). The desulfurization results of each slag sample are listed in Table 3. Previous researchers have done a lot of research on silicate melts [26][27][28], and divided the oxygen ions in the silicate melts into three types: bridging oxygen, non-bridging oxygen, and free oxygen. The silicate melt is a tetrahedral structure, and the different tetrahedrons are connected by bridging oxygen (O 0 ) and non-bridging oxygen (O − ) to form a three-dimensional continuous space network structure. Basic oxides can break the three-dimensional network structure by releasing free oxygen ions

Experimental Principle
In order to measure the activity of MgO in the silicate melt, solvent metal (Sn ≥ 99.9%) and the slag (measure slag or reference slag) were equilibrated in a MoSi 2 resistance furnace. The chemical reagents (CaO, CAS no. 1305-78-8, 99.99% purity; SiO 2 , CAS no. 7631-86-9, 99.99% purity; MgO, CAS no. 1309-48-4, 99.99% purity; Al 2 O 3 , CAS no. 1344-28-1, 99.99% purity; CaF 2 , CAS no. 7789-75-5, 99.99% purity) employed as the collectors were procured from Sinopharm Reagent Group (Ningbo, China). The slag composition of the activity experiment is shown in Table 1. A six-hole graphite crucibles was used in the experiment. About 5 g of solvent metal (Sn ≥ 99.9%) and 10 g of the slag (measure slag or reference slag) were filled in a graphite crucible. When the temperature of a tubular furnace sample rose to 1500 • C, CO was introduced. After the balance time was 24 h, the slag sample was oil cooled. Thereafter, the metal and the slag were carefully separated for chemical analyses using Inductively Coupled Plasma-Atomic Emission Spectrometry (ICP-AES).

Results and Discussion
In this paper, a double-layer graphite crucible was used to simulate the process of melting iron dropping through the slag. Isothermal single variable desulfurization experiments were carried out for B and w(MgO)/w(Al 2 O 3 ). The desulfurization results of each slag sample are listed in Table 3. Previous researchers have done a lot of research on silicate melts [26][27][28], and divided the oxygen ions in the silicate melts into three types: bridging oxygen, non-bridging oxygen, and free oxygen. The silicate melt is a Crystals 2021, 11, 910 8 of 13 tetrahedral structure, and the different tetrahedrons are connected by bridging oxygen (O 0 ) and non-bridging oxygen (O − ) to form a three-dimensional continuous space network structure. Basic oxides can break the three-dimensional network structure by releasing free oxygen ions (O 2− ) and bridging oxygen to form non-bridging oxygen. The degree of polymerization of slags can be expressed by Equation (9).        Figure 10 shows the oil cooled FT-IR of different w(MgO)/w(Al2O3) slag s ples at 1773K when w(Al2O3) = 20% and B = 1.30. From the results of the FT-IR in Fi 10, it can be found that with the increase of w(MgO)/w(Al2O3), the grooves of the [S tetrahedron structure at the wavenumber of 800-1200 cm −1 gradually become shallo and flattened, which indicates that the complex silicate structure polymerized by SiO

Conclusions
In this paper, a double-layer crucible was used to simulate the process of iron droplets passing through the slag. The desulfurization effect of B(w(CaO)/w(SiO2)) and w(MgO)/w(Al2O3) on high alumina blast furnace slag was studied. The conclusions are as follows: (1)

Conclusions
In this paper, a double-layer crucible was used to simulate the process of iron droplets passing through the slag. The desulfurization effect of B(w(CaO)/w(SiO 2 )) and w(MgO)/ w(Al 2 O 3 ) on high alumina blast furnace slag was studied. The conclusions are as follows: