Effect of Reductant Type on the Metallothermic Magnesium Production Process

Introduction

Mg is a silver grey metal which has a dense hexagonal crystal structure and it has two valance electrons. Melting point and boiling point of magnesium are 650±2 °C (923 K=650 °C) and 1,107±10 °C respectively [1].

The consumption of magnesium in many fields, such as aircrafts, rockets and automobile industry, is expected to increase rapidly for the next decade, because magnesium has the lowest density as 1.738 g/cm³ in all structural metals and its strength/density ratio is very high. It was reported by the USGS in 2015 that the World’s total magnesium production was approximately 878,000 t in 2013 and 907,000 t in 2014, thus the annual supply change is about 3 %. On the other hand, the World’s demand for magnesium increases nearly 10 % per year [2, 3, 4, 5].

Dolomite ore is the most important source to produce metallic magnesium. It is a compound of magnesium carbonate and calcium carbonate with a chemical formula as CaCO₃.MgCO₃. It theoretically consists of 45.3 % MgCO₃ and 54.7 % CaCO₃ by weight. Additionally dolomite is used in various industries such as metallurgy, glass, chemical and paper [6].

The major part of magnesium production is conducted via the Pidgeon Process which is a metallothermic (silicothermic) method [7]. Metallothermic reduction reactions are normally highly exothermic. Thus, the propagation of reactions and the yield of reaction products continue in a self-sustaining mode without requiring any additional heat [8]. On the other hand, the Pidgeon Process is highly endothermic (~ΔH₂₉₈ is about 209 kJ mole of Mg), and heat supply is a critical consideration in industrial retorts (reactors). Magnesium is in gaseous phase at the temperature which is reduced, and it is collected in the form of crown at the cooling part of the retort [9]. The Pidgeon Process is conducted through the reduction of magnesium from calcined dolomite via silicothermic reduction under vacuum atmosphere. Powdered calcined dolomite, ferrosilicon (as reductant) and a slight amount of fluorspar (CaF₂ as catalyst) are mixed prior to the reduction process as raw materials [7, 10]. Morsi et al. investigated to produce magnesium metal from dolomite ore under inert atmosphere in 2002. The highest recovery rate was achieved as 92 % when charge is containing 2.5 mass % CaF₂, CaO/MgO molar ratio was 1.6. Furthermore, Si/MgO ratio was 1.45. The experiment was conducted at 1,300 °C for 5 h [20]. Minic et al. carried out the results of the characterization and thermal analysis of the slag from the magnesium plant named Bela Stena in Serbia. Dicalcium silicate based structure was determined in the solidified slag. Magnesium was mostly in the form of periclase, merwinite and melilite minerals [21]. In 2015, a one-step method, involving dolomite decomposition and magnesium reduction, was examined by Zhang et al. The results showed that the one-step developed technology is effective for both reducing process duration and saving energy [22].

Although Turkey has 16 billion tones (detected and probable) of World’s dolomite reserves [11], Turkey had not produced magnesium metal until 2015. The new Turkish magnesium plant using the silicothermic process has been started with 15,000 t/year production capacity [12]. Yucel et al. have investigated the parameters affecting the silicothermic reduction of calcined dolomite by using Turkish ores since 2002 [4].

In silicothermic Mg production processes, silicon or silicon based materials are used as the reductant of MgO based raw materials such as calcined magnesite and dolomite. Reduction of Mg is generally carried out via Pidgeon process. Mg is in the gas phase at reduction temperatures. The reaction in question is shown by the eq. (1) [6].

This method is inefficient when the Mg source is MgO due to the reaction product which is in the form of MgO.SiO₂. The formation of that compound stops the reduction [13]. To avoid MgO.SiO₂ formation, SiO₂ activity must be decreased with some additives. In this case CaO is the most suitable additive which is provided by using calcined dolomite as reactant. The reduction of Mg from MgO becomes easier and the reduction efficiency of Mg increases in accordance with the formation of CaO.SiO₂ structure in reaction products. Because of that reason, in industrial silicothermic process calcined dolomite is used as a raw material instead of MgO [14].

If ferrosilicon (FeSi) contains higher than 65 mass % Si, activity of Si in ferrosilicon is close to the activity of pure silicon (0.97) at 1,200 °C. Because of the lower price of ferrosilicon, the use of FeSi is more preferable than silicon in commercial applications [15, 16]. The net reaction between ferrosilicon and dolomite is given in eq. (2).

Al can be also used as a reductant in the Pidgeon Process, because it has thermodynamically several advantages. In the aluminothermic reduction process, reduction occurs at lower temperatures than the condition which is FeSi used as a reductant. This situation is explained in detail in the Thermodynamic Background of the present article. Aluminothermic reduction is not preferable for industrial application due to economic reasons. On the other hand some solutions, such as the use of FeAl, can be suggested to make the use of aluminum feasible [14]. The reaction for the use of Al to reduce Mg from calcined dolomite as shown in eq. (3).

Although the use of FeSi is more economical than Si, in the common silicothermic process, it is still needed to reduce the cost of reductant. This work aims to evaluate the possibility of reducing the cost of magnesium production by using CaC₂ with ferrosilicon as much as possible. According to Suchy and Seliger [17], MgO reduction with CaC₂ is possible, but the residue of this process is highly entrained with Mg vapor. Mg vapor has the adverse effect on the reduction when it conjugate with residue, so Mg and lime must be separated from each other in order to obtain high recovery ratios. In addition, there is another disadvantage, the residue caused agglomeration of charge and the reaction mixture substrates are adapted to this agglomeration during reduction. To avoid the agglomeration silica and alumina preferably are added to mixture. With the addition of SiO₂ or Al₂O₃ at the reaction temperature, lime converts into calcium silicate or calcium aluminate. Thus, residue of mixture does not include Mg [17]. In the experiments, increasing proportion of CaC₂ was used with ferrosilicon. Process duration and temperature were carried out as variables in order to obtain high Mg recovery ratios. The reaction for the use of CaC₂ to reduce Mg from dolomite gives metallic magnesium of 3 moles as shown in eq. (4).

Theoretical background

Thermodynamical investigations were conducted by using FactSage 6.4 software. Figure 1 was plotted to understand the change of the reduction conditions of Mg from MgO by Si, Al and CaC₂ under 1 bar and 1 mbar pressures respectively. According to the thermodynamical investigations, the formation temperatures of the reactions are very high under 1 bar atmosphere pressure. In the condition which Al is used as a reductant, reaction begins at 1,475 °C where its Gibbs free energy is 0. The reduction temperatures with others are higher than Al, these are 2,143 °C, 1,789 °C, 1,847 °C for Si, FeSi and CaC₂ respectively. On the other hand, in accordance with the nature of Pidgeon Process, reactions occur at lower temperatures around 1 mbar process pressure than that of 1 bar [4]. According to FactSage 6.4 software, the reaction beginning temperatures are 850 °C for Al, 1,155 °C for FeSi, 1,200 °C for CaC₂ and 1,325 °C for Si under 1 mbar pressure. The effect of vacuum and different reductants on the beginning reduction temperatures of MgO and calcined dolomite is detailed shown in Figure 1 and Table 1. This temperatures are slightly low for the reduction of Mg from calcined dolomite under the reaction pressure of 1 mbar (842 °C for Al, 1,098 °C for FeSi, 1,191 °C for CaC₂ and 1,318 °C for Si).

Figure 1:

The change of gibbs free energies for the reactions between MgO and Si-CaC₂-Al under 1 mbar and 1 bar reaction pressures.

The change of reaction products with increasing temperature is given in Figure 2 for the reduction of Mg from calcined dolomite by using FeSi, CaC₂ and Al as reductant. The diagrams were plotted for the 100 % stoichiometry of the reductants under 1 mbar. Predominance phases in the reaction products are Ca₂SiO₄, Mg and a slight amount of iron (between 800–1,400 °C) for the use of FeSi as reductant Figure 2(a). CaO, Mg and CO with a slight amount of Ca₂SiO₄ are determined as the products for CaC₂ as a reductant Figure 2(b). Ca₃Al₂O₆ and Mg phases exist in the reaction products of in the case of use Al as a reductant Figure 2(c). Moreover, a very low concentration of Ca₂SiO₄ phase occurred up to 1,300 °C and, the phase in question converted to Ca₃SiO₅ at the temperatures above 1,300 °C. According to Figure 2(b), CO_(g) exists with Mg_(g) in the gaseous products. The amount of collected Mg in the form of crown decreases in the presence of CO in the retort. It is predicted that the occurrence of CO in the retort increases the process pressure. Thus, Mg recovery ratio drops via reduction in vapor phase and condensation.

Figure 2:

The change of reaction products with increasing process temperature for different reductant types: (a) FeSi, (b) CaC₂ and (c) Al (Diagrams was plotted for the 100 % of the stoichiometrically required reductant addition and at 1 mbar atmosphere).

In the study which was conducted by Ramachandran and Reddy in 2015 [18], the predominance diagram for Mg-O-C system was calculated and plotted at a constant pO₂ of 10–20 as shown in Figure 3. The diagram shows MgO is stable at lower temperatures and lower CO pressure values. But at the temperatures above 1,600 °C, Mg is the reaction product. MgC₂ is the stable phase under the conditions which the reaction temperature is above 850 °C and the reaction pressure is above nearly 1 bar. The experimental studies in this study, an average process pressure of 1 mbar was used and the process temperature was up to 1,250 °C. Thus, it is clear to see that the formation of MgC₂ phase is not possible under the condition of CaC₂ used as a reductant.

Figure 3:

Predominance diagram for Mg-O-C System [18].

Experimental

Calcined dolomite was provided from domestic sources of Turkey. It was milled by using a laboratory scale Siebtechnic vibratory cup mill. Particle size of milled dolomite was measured by using Malvern Instruments Mastersizer 2000 particle size analyzer and it is calculated in a large distribution varying from 1.855 μm to 74.715 μm. Particle size distribution of milled calcined dolomite is given in Figure 4.

Figure 4:

Particle size distribution of calcined dolomite ore after milling.

Calcined dolomite was analyzed by using chemical analysis and AAS (Perkin Elmer Analyst 800) techniques. Chemical analysis of calcined dolomite is given in Table 2. FeSi (75.0 % Si, 24.0 % Fe, 0.9 % Al, 0.1 % Ca by mass), Si (99.2 % by mass) was used as reductants and ETI Elektrometalurji grade CaC₂ (min. 97.0 % CaC₂ by mass) was also employed as a reductant aiming to decrease the amount of FeSi.

The XRD pattern of calcined dolomite ore was obtained by using PANalytical PW3040/60 XRD. XRD pattern is given in Figure 5. The dolomite ore comprises mainly CaO and MgO according to the XRD pattern.

Figure 5:

XRD pattern of calcined dolomite ore.

Experiments were carried out in cylindrical retorts which are made of stainless steel. The yield magnesium vapor condenses and forms the crown in the water cooled region of the furnace. To obtain vacuum atmosphere in the retort, a two stages integrated rotary vane pump ILMVAC -PK8D used which can hold a final pressure of 2×10⁻⁴ mbar. ILMVAC PIA 100 piezoelectric sensor was used in order to evaluate the extent of vacuum. The retort was externally heated by using a SiC resistance furnace having a maximum temperature of 1,350 °C. The details of the experimental set up can be found elsewhere [4]. Reduction experiments were conducted at two different temperatures, 1,200 °C and 1,250 °C, under vacuum atmosphere. At the end of the reduction, the retort was left in the furnace at the same vacuum values and it was cooled to room temperature. Then, the cover was opened and the condensed magnesium metal on the cooling section was taken. The residue left in the boat were weighted and analyzed. The degree of Mg metal recovery was calculated both from residue and collected metallic magnesium crown by using two different formulas given below.

Where W₀ is the weight of dolomite, Mg₀ % is the weight percentage of magnesium in dolomite, W₁ is the weight of residue, and Mg₁% is the weight percentage of magnesium in residue, W₂ is the weight of crown magnesium, Mg₂% is the weight percentage of magnesium in crown.

In the present study, experimental sets were developed to understand the effects of reductants type on the Pidgeon Process of calcined dolomite. In the first set, the change of Mg recovery was investigated with the increase in charge (reactant) weights in the case FeSi was used as reductant. Two different retorts were used in this experimental set. The experiment with 50 g charge weight was carried out in 1 liter (l) retort, others (2,000 g, 3,000 g and 5,000 g charges) were executed in 10 liter retort at 1,250 °C under 1 mbar. In all experiments process durations were 6 h. In the second experimental set, effects of Si, FeSi and Al reductants were examined. The experiments were conducted in 1 liter retort. Effect of process duration on Mg recovery was investigated for 60, 120, 180, 240 and 300 min. In the third experimental set, 100 % stoichiometric 50 g mixtures were prepared. Stoichiometric amount of the reducing agent was calculated through sum of the reducible oxides (MgO + FeO + SiO₂) contents in the calcined dolomite. The mixture stoichiometric ratios were changed from 100 % FeSi – 0 % CaC₂ to 50 % FeSi – 50 % CaC₂ with 10 % intervals. The change of Mg recovery with increasing CaC₂ addition ratio in FeSi was carried out at 1,200 °C and 1,250 °C under 1 mbar vacuum atmosphere for 6 h. In the last experimental set, the experiments were conducted with increasing CaC₂ addition and in different volumes of retorts as 1 l and 10 l.

Results and discussion

In the first group experimental set, effect of charge amount was investigated on magnesium recovery. Calcined dolomite–FeSi100 % stoichiometric powder mixtures were prepared and experiments were conducted in different charge (reactant) amounts varying from 50 g to 5,000 g, 2,000 g and 3,000 g charge amounts respectively presented the highest recovery rates which were calculated from residue. In all experiments 2.5 % CaF₂ by mass added to charge mixtures to improve Mg recovery. The optimum results were given in the research article where effect of CaF₂ was investigated, which was reported by Demiray et al. [19]. The highest Mg recovery was detected from the charge amount of 50 g as 98 %. In this experiment, crown Mg efficiency cannot be calculated, because crown Mg amount was not enough to carry out analyses. On the other hand, the highest Mg recovery calculated from crown was determined as 90 % for the experiment conducted with 3,000 g charge amount. Figure 6 shows the change of Mg recovery with the increase in charge weight. When the difference is evaluated between the Mg recovery rates calculated from residue and crown for the same conditions (such as reactant weight of 50 g) show that Mg was highly reduced, but the conditions were not enough to collect in the form of crown in a high recovery ratio. It was understood that increasing charge weight firstly affected Mg recovery negatively it reduced from 98 % to 90 % which calculated from residue. Mg recovery increased from 90 % to 92 %. According to the results, reduction duration must be extended with increasing amount of charge.

Figure 6:

The change of Mg recovery calculated both from residue and crown with the increase in charge weight. 50 g Experiment was carried out in 1 l retort, others in 10 l retort (1,250 °C, 1 mbar and 6 h).

In the second experimental set, the effects of Si, FeSi and Al addition and process duration were investigated. In these experiments, the effects of stoichiometric Si, FeSi and Al addition on Mg recovery were examined, and the results were compared to each other. All experiments were conducted under vacuum atmosphere at 1,200 °C and in 1 l retort. Mg amounts in the residues and Mg recovery ratios were given in Figure 7. From the results, the highest recovery was detected at silicon experiments 96 %. It is clear to see that increasing process duration and the use of Al as a reductant instead of FeSi have a positive effect on the recovery of Mg. The highest Mg recovery ratio was determined as 88 % at the experiment conducted with the use of Al and for the process duration of 300 min.

Figure 7:

Mg amounts in residues and Mg recovery ratios for Si, FeSi and Al reductants with increasing process duration and 2.5 % CaF₂ addition at 1,200 °C (50 g, 1 l, 1 mbar). Reductants are Si (□, ■), Al (Δ, ▲) and FeSi (○, ●).

In the third experimental set, the effect of CaC₂ addition and the change of reaction temperature were investigated. In these experiments, CaC₂ was added to charge mixtures in specific proportions. Stoichiometric FeSi/CaC₂ ratio changed from 100 % FeSi – 0 % CaC₂ to 50 % FeSi – 50 % CaC₂. All experiments were conducted under vacuum atmosphere at two different reaction temperatures as 1,200 °C and 1,250 °C. Figure 8 presents effect of CaC₂ amount and reaction temperature on Mg recovery ratio. The experiments show that increasing of CaC₂ addition in the charge decreases the Mg recovery to 82.0 % for CaC₂ addition of 50 % at 1,250 °C. The decrease in terms of Mg recovery with the use of CaC₂ reductant was previously explained. [17].

Figure 8:

The change of Mg recovery calculated from residue with increasing CaC₂ addition ratios at different temperatures (1,200 °C–1,250 °C, 1 mbar and 6 h).

In the last stage of the experiments, effects of CaC₂ addition on different retort volumes were investigated. Magnesium recovery ratios were calculated both from residue and crown. 1 l retort, 10 l retort residue and crown recovery ratios were compared in Figure 9. According to Figure 9, the highest Mg recovery was determined for 1 l retort 100 % FeSi added experiment as 98.2 %. On the other hand, the experiments conducted in the 10 l retort, the highest recovery was calculated at 10 % CaC₂ added experiment as 95.2 % from residue and 94.7 % from crown. Chemical analysis results of all experiments are given in Table 3. The highest MgO amount in residue was obtained in the experiment conducted with 50 % FeSi – 50 % CaC₂ addition ratio as 6.81 %. According to chemical analysis of residues, CaC₂ addition affected amount of MgO in residue, minimum amount of MgO detected in stochiometric 90 % FeSi–10 % CaC₂ experiment with 1.73 %.

Figure 9:

The change of Mg recovery calculated with increasing CaC₂ addition ratios for different experimental volumes conducted in 1 l and 10 l retorts (1,250 °C, 1 mbar and 6 h).

Conclusion

The reduction of Mg from calcined dolomite ores was investigated by using different reductants (FeSi, Si, Al and CaC₂) for process pressure of 1 mbar at 1,200 °C and 1,250 °C temperatures for varying process durations up to 360 min. According to the results of experimental studies, obtained crown Mg was nearly pure and included trace amounts of Fe, Ca, Al and Si. The highest Mg recovery, calculated from residue, was obtained as 98.17 % at the experiment with 50 g charge weight and the use of 100 % stoichiometric FeSi in 1 l retort at 1,250 °C. The use of Al as a reductant was also examined and, 88.0 % Mg recovery ratio was measured at the experiment conducted for 300 min process duration for 100 % Al stoichiometry at 1,200 °C. Although the use of Al is not a cheap choice, it was determined that Al is a more effective reductant than FeSi and it can be evaluated if a low-cost Al source, like FeAl, can be used. During the experiments, increasing amount of charge and increasing CaC₂ quantity reduced Mg recovery. For example, the use of 40 % CaC₂ with 60 % FeSi at the experiment conducted in 1 l retort, Mg recovery was measured as 89.2 %. Thus, when the costs of FeSi and CaC₂ (nearly half of FeSi) are compared, it is clear to understand the use of CaC₂ makes the process economically feasible.