Smelting reduction of MgO in molten slag by liquid ferrosilicon

Qifeng Tang^a,*, Jiacheng Gao^b, Xiaohua Chen^b, Xiaowei Wei^a 

^aXihua University, School of Materials Science and Engineering, Chendu 610039, China

^bChongqing University, College of Materials Science and Engineering, Chongqing 400045, China

*Corresponding author: tqfkmust@126.com

1. INTRODUCTIONTOP

Magnesium is well known for its good capability and widely used in the fields of aviation, aerospace, automotive, electronics, metallurgy and chemical engineering, etc (Friedrich and Schumann, 2001; Mordike and Ebert, 2001). Two processes currently are used in magnesium production-electrolytic reduction of magnesium chloride and thermal reduction of dolomite. In order to promote the further development of magnesium production, it is necessary to carry on fundamental studies as to provide theoretical base.

Several research investigations on kinetic of thermal reduction have been carried out. Morsi et al. (2002) indicated that the silicothermic reduction process under inter atmosphere was controlled by diffusion of reacted species and the apparent activation energy of the reduction process was about 306 kJ mol⁻¹. Barua and Wynnyckyj (1981) carried out relevant research in flowing argon and hydrogen atmosphere, respectively. Some researcher also studied the kinetics of thermal reduction by other reactants, such as Fe-Si-Al alloy, aluminum and carbon (Lan et al., 1999; Li et al., 2002; Lan et al., 2003; Fu et al., 2014). Most information on reduction kinetic of magnesium production focused on solid-solid reaction, while the kinetic of thermal reduction which takes place in liquid phases was seldom studied.

In this present work, the kinetics of smelting reduction of MgO in molten slag by liquid ferrosilicon was investigated at high temperature. Effects of reaction temperature, addition of CaF₂, primary slag composition and MgO/Si molar ratio were discussed.

2. MATERIALS AND METHODSTOP

Calcined dolomite (CaO·MgO) powder supplied by the Chongqing Haibo magnesium smelting Co., Ltd and reagent grade chemical powders (Al₂O₃, SiO₂ and CaF₂) were used to prepare the experimental primary slag. Ferrosilicon (75% Si) powder was used as reductant. All of powders passed through the 200-mesh sieve.

The mixture was prepared by weighing and mixing, and the pellets were compacted into 10 mm-diameter and 30 mm-height. About 150 g pellets charged in graphite crucible. The smelting reduction was carried out in the high-temperature vacuum furnace which was internally heated by using a regulated graphite resistance and had a maximum temperature of 1873 K and a limit vacuum 0.01 Pa. The reduction process was controlled under 4~40 Pa. From previous study (Tang et al., 2013), in order to forming molten slag at 1500~1600 °C, the composition of primary slag was about 55% calcined dolomite (32.0% CaO and 23.0% MgO) and 45% additive (Al₂O₃ and SiO₂). So the primary slag composition and the relevant experimental conditions were list in Table 1. After reduction, the residue slag in graphite crucible was cooled to room temperature and pulverized to powder which passed through the 200-mesh sieve. Chemical compositions of the residue were measured by X-ray fluorescence (XRF-1800). The reduction extent of MgO in slag was defined by the Eq. (1):

η= [1-W_(MgO)t /W_(MgO)0] ×100%      (1)

where W _(MgO)t is the weight of MgO in the slag at any time, t and W _(MgO)0 is the weight of MgO in the slag at t=0, respectively.

3. RESULTSTOP

3.1. Effect of primary slag compositionTOP

Reduction experiments for primary slag with different compositions were carried out at 1550 °C for 1X Si (n_(Si)/n_(2MgO)) =1.0, and the results were shown in Fig. 1. It can be seen from Fig. 1 that the more the mass of SiO₂ in primary slag, the lower the reduction extent of MgO.

In smelting reduction of MgO by ferrosilicon, the reaction was described as follow:

2(MgO) + (Si) → 2Mg_(g) + (SiO₂)      (2)

From Eq. (2), it could be found that SiO₂ produced in reduction. When the mass of SiO₂ in slag was little, SiO₂ diffused away from reaction interface easily and the Eq. (2) favored a move to the right-hand side. Increasing the mass of SiO₂ in primary slag, both of the viscosity of slag and SiO₂ concentration near reaction interface improved. Hence, it was difficult for SiO₂ produced in reduction reaction to remove away from reaction interface, which hindered the reduction reaction processing, so the reduction extent of MgO decreased.

3.2. Effect of calcium fluorideTOP

In the process of magnesium production by solid-solid reaction, CaF₂ was a common additive as catalyst and could greatly improve the reduction extent of MgO (Tian et al., 2012). In present work, the influence of CaF₂ on the reduction extent of MgO was investigated. The experimental results were shown in Fig. 2, indicated that the reduction extent of MgO just increased a little with adding CaF₂. Compared with solid-solid reaction, CaF₂ had not significant effect on the reduction extent of MgO in smelting reduction. In smelting reduction, the reaction rate was very fast due to reduction taking place in liquid state, so the catalysis influence of CaF₂ did not play an important role.

3.3. Effect of silicon stoichiometryTOP

The experiments of smelting reduction with different initial molar ratio of Si/2MgO were carried out and the results of this series of experiments were shown in Fig. 3. When the initial molar ratio of Si/2MgO was less than 1.0, the reduction extent of MgO improved significantly with increasing silicon stoichiometry. Under this condition, a high Si content in the charge provided more active Si and surface area for reaction, which could promote the reaction moving to right-hand side. When the initial molar ratio of Si/2MgO increased from 1.0 to 1.2, the reaction rate had only a small amount of increase.

3.4. Effect of reaction temperatureTOP

The melting point of experimental primary was above 1500 °C, so experiments were conducted at 1550 °C and 1600 °C due to the limitations of the maximum operating temperature of the furnace. The results were shown in Fig. 4 and indicated that the higher the temperature, the faster the reaction. The final reduction ratio was 97% at 1600 °C for 2 hours. With increasing reaction temperature, the viscosity of slag was decreased, mass transfer became easier. Thus, the reaction rate accelerated the reduction extent of MgO.

3.5. Reaction kineticTOP

3.5.1. Reaction orderTOP

Before the start of smelting reduction, the slag and metal layer were formed after the melting of mixture charge. Therefore, the reaction rate of smelting reduction could be expressed by Eq. (3)

where K was apparent rate constant, t was reaction time, C_MgO was MgO consent (weight percent) in slag at time t, and n was reaction order. Integration of Eq. (3) gave:

when n=1 and 2, the Eq. (4) became:

−(ln C_t − ln C₀) =Kt  n=1      (5)

According to the Eq. (5) and Eq. (6), the reaction order could be determined by the relationship between C_t and t in smelting reduction. By dealing with some group data, it was found that the linear relationship between 1/C_t and t was well maintained from Fig. 5. The value of apparent rate constant K and the correlation coefficient r of fitted line at different condition were list Table 2. The correlation coefficient r of fitting line was in the range −0.99< r <−0.97 with an average value of −0.981. Therefore, it could be considered that the reduction of MgO in molten slag was the second order reaction and the reaction rate could be expressed by Eq. (7)

3.5.2. Apparent activation energyTOP

The relationship between apparent rate constant K and reaction temperature T could be expressed by Arrhenius equation as follow:

where A was frequency factor, E was apparent activation energy, R was gas constant. Logarithming of Eq. (8) gave:

Due to the limitation of the maximum operating temperature of the furnace, the experimental temperature was carried at 1550 °C and 1600 °C with the same composition mixture charge (slag 1 and slag 9). The relationship between lnK and T⁻¹ was shown in Fig. 6.

By calculation, the relationship between apparent rate constant K and reaction temperature T was depicted as follow:

From Eq. (9), it was well known that the apparent activation energy of smelting reduction was 586 kJ mol⁻¹.

3.5.3. Reaction mechanismTOP

At high temperature, the reduction of MgO in molten by ferrosilicon was liquid-liquid state reaction. The preceding of settling reaction could be summarized as follows:

The smelting reduction was carried at high temperature, the chemical reaction took place on the metal/slag interface fast. Hence, it was generally considered that chemical reaction was not the controlling step. In this work, the reducing agent was 75% Si-alloy which had about 67% Si and 23% FeSi₂. The activity of Si a_Si was near 1 and the viscosity of Si-alloy was low, which made the mass transfer of Si in liquid ferrosilicon easy, so the overall process was not controlled by step (2). The smelting reduction took place under vacuum, Mg vapor diffused through the slag very rapidly. Therefore the process was not controlled by step (5) either. From the above discussion, the step (1) and/or step (4) might be the rate limiting step. The diffusion activation energy of MgO and SiO₂ in slag had the same order of magnitude, so the total process might be controlled by mass transfer in slag.

Assumed that the overall process was controlled by mass transfer in slag, the rate of mass transfer in slag might be expressed as:

where D was the diffusion coefficient of MgO in slag, δ was the effective boundary layer thickness of slag, S was the area of metal/slag interface, V was the volume of slag, and C_MgO and were MgO content (weight percent) in slag and at the metal/slag interface, respectively. During the overall process, C_MgO was much larger than . Thus, the Eq. (11) could be expressed as:

During the smelting reduction, Mg vapor produced in reaction stirred the slag and changed the effective boundary layer thickness δ. The relationship of δ and the rate of mass transfer in slag could be depicted as followed (Wang and Chou, 1963):

where k was constant. Association Eq. (12) and Eq. (13), the rate of mass transfer in slag could be depicted as:

The Eq. (7) and Eq. (14) were same, which indicated that the total process was controlled by mass transfer in slag.

4. CONCLUSIONSTOP

Magnesium oxide was smelting reducted by liquid ferrosilicon in slag at vacuum. The following results were obtained:

ACKNOWLEDGMENTSTOP

The authors would like to acknowledge financial support by Key Scientific Research Fund of Xihua University (Grant N° Z1320109) and the General Project of Sichuan Province Department of Education Fund for Nature (Grant N° 14201436).

REFERENCESTOP