Preparation of vanadium by the magnesiothermic self - propagating reduction and process control

: A new process for preparing vanadium by direct reduction of V 2 O 5 from the Mg – V 2 O 5 self - propagating system is proposed in this article. The reaction behavior and path of V 2 O 5 in the magnesiothermic reduction pro cess were investigated using the XRD, SEM - EDS, laser particle size analyzer, and speci ﬁ c surface area analyzer. The experimental results show that the reaction of the V 2 O 5 – Mg system is a solid - solid reaction, and the initial reaction temperature is 570°C. Although the formation of MgV 2 O 4 spinel cannot be predicted via the calculation of thermodynamics, the presence of MgV 2 O 4 spinel is of great signi ﬁ cance to the V 2 O 5 reduction process. Taking into account the characteristics of the gradual reduction of V 2 O 5 by Mg and the appearance of the MgV 2 O 4 spinel phase, the limiting link of the reaction may be the transi tion from MgV 2 O 4 to V. A reduction path of V 2 O 5 beyond the thermodynamic prediction is proposed: V 2 O 5 → V 3 O 5 → MgV 2 O 4 → V. The reaction temperature and the phase transformation process can be e ﬀ ectively controlled by adjusting the ratio of reactants and additives, and element V can be obtained by a one - step rapid self - propagating reaction and breaking through the reaction restriction link. In this experiment, the vanadium powder with a porous structure, a speci ﬁ c surface area of 3.44 m 2 g − 1 , and the oxygen content of 4.86 wt% were obtained.


Introduction
Vanadium is commonly used as an additive in the iron and steel industry, which can refine the microscopic grains of steel and improve the strength, toughness, and corrosion resistance of steel [1][2][3]. Vanadium is also used as a stable element of the β phase in titanium alloys, which significantly improves the ductility and plasticity of titanium alloys. In addition, vanadium is an important component of superconducting materials, battery materials, phosphors, catalysts, photosensitive materials, and hydrogen storage alloys [4][5][6][7]. Therefore, vanadium has become a strategic material that countries around the world are competing to reserve [8][9][10].
Metal V and Ti-V alloys were obtained by electrolyzing V 2 O 3 in the CaO-CaCl 2 molten salt system based on the principle of the "OS" method [11][12][13][14][15]. Wu et al. studied the electrochemical behavior during the electrolysis of V 2 O 3 and achieved satisfactory results in controlling the current efficiency and energy consumption index [16]. Gussone et al. used the LiCl-KCl molten salt system to electrolyze VCl 3 and TiCl 2 to prepare Ti-V alloy [17]. Weng et al. electrolyzed NaVO 3 from the CaCl 2 -NaCl molten salt to obtain V [18,19]. Cai et al. improved the FFC process, studied the electrochemical behavior of V 2 O 5 in the CaCl 2 -NaCl electrolyte system, and found that V 2 O 5 was first reduced to V 3 O 5 and finally reduced to V [20][21][22][23]. However, the defect of low current efficiency still remained in this method. High-purity metallic vanadium is currently mainly prepared by reducing vanadium chloride with Mg. This idea is derived from the Kroll method to produce sponge titanium, but the inherent problems of the Kroll method such as high energy consumption and high pollution still need to be overcome [24,25]. Inazu et al. [26] proposed a new idea on the production of V, that is, V 2 O 5 and MgO were sintered to obtain the MgV 2 O 4 precursor, and then the MgV 2 O 4 precursor was reduced by the Mg steam in a microwave field to obtain V. In the decomposition method, VN is first obtained by reducing V 2 O 3 /V 2 O 5 with carbon/magnesium in nitrogen, and then VN is decomposed at high temperature to obtain the metal V [27][28][29]. In the silicothermic method, V 2 O 5 is reduced with Si to obtain the primary product, and then high-purity metal V is produced by the deep deoxidation of molten salt electrolysis [30]. Xu used V 2 O 3 as raw material and CaCl 2 as a pore-forming agent, and reduced V 2 O 3 with Ca steam under vacuum to obtain metal V [31]. Lee et al. used hydrogen to reduce V 2 O 5 to obtain V 2 O 3 and then reduced V 2 O 3 at 1,073 K for 48 h with magnesium to obtain metal V [32].
Although metallic vanadium was successfully prepared in the above processes, the intermediate such as VCl x , MgV 2 O 4 , V 2 O 3 , or VN were also produced in the preparation process, and the research on the preparation of metal V using a direct metal thermal reduction is seldom reported. V 2 O 5 is the most stable and common in vanadium-containing compounds, and the preparation process of V 2 O 5 is very mature [33,34]. Therefore, the preparation of the vanadium metal by direct reduction of V 2 O 5 has great application prospects. At present, the preparation of metals and their alloys by direct reduction methods of metal high-valent oxides has gradually matured. For example, Ti, TiAlV, etc. have been successfully prepared by using TiO 2 as raw material and using magnesium thermal self-propagation [35,36].

Experimental
In this study, the raw materials were chosen with chemical reagents of vanadium pentoxide (V 2 O 5 , >99%, Macklin, China), magnesium powder (Mg, >99%, Sinopharm, China), sodium chloride (NaCl, >99%, Sinopharm, China), and hydrochloric acid (HCl, 36%, Sinopharm, China). The ratio of raw materials is listed in Table 1. The raw materials were weighed precisely and mixed thoroughly according to the ingredient ratio shown in Table 1, and then pressed into a cylindrical parison sample. The sample was placed in a self-propagating reactor and performed a self-propagating reaction in a vacuum to obtain a self-propagating reaction product. The reaction product was taken out and pulverized, and then soaked in a dilute hydrochloric acid solution for 3 h. After soaking, the solution was filtered. The filtered product was washed to neutrality, and finally, dried in a vacuum to obtain the reduced product.
The product phase was analyzed by X-ray diffraction (XRD, copper target, Bruke, D8, Germany). The micro-morphology of the product was characterized by field emission scanning electron microscopy (SEM-EDS, Hitachi, su8010, Japan). The specific surface area of the product was obtained using a specific surface area analyzer (ASAP2020 m, U.S.A.). The oxygen content of the product was detected with an oxygen nitrogen hydrogen analyzer (LECO onh836). The particle size distribution of the product was characterized with a laser particle size analyzer (Mastersizer, 2000, UK).

Results and discussion
3.1 Analysis of the reaction kinetics of the V 2 O 5 -Mg system The TG-DTA curve of the V 2 O 5 -Mg system is presented in Figure 1(a). A sharp exothermic peak appeared near 570°C on the DTA curve, indicating that the reduction reaction was initiated at low temperatures, and the sharp exothermic peak indicated that the reaction was rapid and violent. Simultaneously, the obvious weight loss occurred near 570°C in the TG curve, which may be caused by splashing during the violent reaction and the volatilization loss of the metal magnesium in a high-temperature environment. The melting point of Mg is 651°C and that of vanadium pentoxide is 690°C, but the actual reaction temperature was around 570°C, which suggested that the self-propagating reaction was a solid-solid reaction. The DTA curve shown in Figure 1(a) was analyzed using the Freeman-Carroll differential method [37]. It can be seen that the reaction order n = 0.14 and the apparent activation energy E = 1816.227 kJ mol −1 .  ; there was still an unreduced VO phase in the products. This change rule of the equilibrium phase with the temperature showed that high temperatures were not conducive to completing the reduction reaction thoroughly, which was consistent with the strong exothermic thermodynamic properties of the self-propagating reaction system; in other words, increasing the amount of the reducing agent cannot promote the complete reduction reaction. It can be theoretically speculated that the composite oxide phase may be formed in the self-propagating rapid reaction process, and then the composite oxide phase was reduced.
Therefore, the phase evolution path during the reduction process was and MgO, indicating that the actual self-propagating reaction was in the thermodynamical nonequilibrium and free escape system state, and there was a certain lag in the evolution of phases in the actual reaction process. When the ratio of Mg:V 2 O 5 was 8:1 (7#), the diffraction peak intensity of MgV 2 O 4 decreased, and the diffraction peak of V appeared. After adding NaCl, the diffraction peak of MgV 2 O 4 basically disappeared (6#). When the Mg:V 2 O 5 ratio was 10:1 (8#), the diffraction peak of MgV 2 O 4 also disappeared. The XRD pattern of the pickled product is shown in Figure 3(b). It can be seen that after acid leaching, the by-product phases such as NaCl and MgO in the self-propagating reaction products could be effectively removed, but the by-product phases such as Mg 2 V 2 O 7 and MgV 2 O 4 could not be removed; when the Mg:V 2 O 5 ratio was 8:1 (6#) and 10:1 (8#), only the elemental V phase existed, indicating that pure metal V could be prepared by adjusting the material ratio of the self-propagating reaction experiment. The XRD analysis results showed that the evolution path of the product phase in the self-propagating reaction of the Mg-V 2 O 5 process was V 2 O 5 → V 3 O 5 → MgV 2 O 4 → V, which was somewhat different from the thermodynamic equilibrium prediction.  Table 2 shows the chemical composition analysis results of the products after pickling. As can be seen, when Mg:V 2 O 5 ratios were 8:1 and 10:1 (6# and 8#), the mass fraction of oxygen in the metal vanadium was about 5%. When the Mg:V 2 O 5 molar ratios were 2, 3, and 4 (2#, 3#, and 4#), although the acid leaching product phase was only MgV 2 O 4 , the contents of Mg and V in products were significantly different. The three main diffraction peaks corresponding to the XRD pattern in the acid leaching product were analyzed, as shown in Table 3. It can be seen from Table 3 that with the increase of the Mg content in the raw material ratio, the diffraction peak of the products tended to shift to a low angle, the intensity of diffraction peaks near 35 and 18°decreased, and the intensity of diffraction peaks near 43°increased. This may be related to the atomic ratio of Mg and V.

Product morphology analysis
From the macro morphology photo of the self-propagating reaction product in Figure 4, it can be seen that when Mg:V 2 O 5 was less than 5:1 (1#, 2#, 3#, 4#, and 5#), the volume of the self-propagating reaction product shrank significantly. When Mg:V 2 O 5 was greater than 8:1 (6#, 7#, and 8#), the volume of the self-propagating reaction product expanded significantly. Figures 5 and 6 present the SEM images of products before and after pickling, respectively. Table 4 shows the EDS results of the products in Figures 5 and 6. It can be seen that the morphologies of 2# and 4# products before pickling ( Figure 5) and after pickling ( Figure 6) were basically the same, and they were both dense blocks, and the EDS results ( Table 4) indicated that a large amount of Mg element still remained in the 2# and 4# products after pickling, which was consistent with the XRD pattern analyses in Figure 3. The structure of the 5# product was loose after pickling, which was obviously different from the 2# and 4# products ( Figure 6). The surface of the 8# product was dense before pickling, while the surface of the 6# product was attached with small particles (NaCl) before pickling ( Figure 5). After pickling, both the 8# and 6# products were elementary V, proved by EDS analysis ( Table 4), and exhibited porous morphology, which was obviously different from other products ( Figure 6). These holes were divided into two types, one was the closing hole formed by interconnecting adjacent metal vanadium particles and the other was the tunnel-type hole formed by a hole and a skeleton together. In addition, it can be seen from the comparison of the morphology of products after pickling ( Figure 6) that there were more sintered necks in the 8# product after pickling, indicating that the reaction temperature of the 8# product was higher than that of the 6# product; as the Mg content in the reaction system increased, the structure of the product gradually changed from dense blocky to porous after pickling. When the molar ratio of Mg and V 2 O 5 was low (such as 2# and 4#), Mg was wrapped by liquid V 2 O 5 , solid MgO was formed by combining Mg with O 2− , and     vanadium oxide cannot be removed by pickling. Therefore, the 2# and 4# products after pickling were dense and blocky, and the 5# and 7# products showed a "transition from dense blocky to porous" morphology.

Product particle size and specific surface area analysis
Because 1#-5# and 7# products were relatively hard and difficult to break, only the 6# and 8# products were analyzed for the particle size. Figure 7 shows the particle size distribution curves of 6# and 8# products after pickling. Compared with the particle size distribution curve of 8# product after pickling, the particle size distribution curve of 6# product after pickling was obviously shifted to the left, indicating that the particle size of 6# product after pickling was smaller than that of 8# product. Table 5 shows the particle size characteristics of the 6# and 8# products after pickling. The characteristic values of particle size D10, D50, D90, and D [3,4] of the 6# product after pickling were 7.63, 31.69, 73.23, and 36.99 µm, respectively, which were all smaller than the those of 8# product. Figure 8 are the adsorption-desorption curves of 6# and 8# products after pickling. According to the IUPAC classification standard, the adsorption-desorption curves conformed to the characteristics of the type II adsorption isotherm. An obvious inflection point occurred in the lowpressure section of the adsorption curve, suggesting that the single-layer adsorption was completed at this time.   Then, multilayer adsorption occurred as the partial pressure increased. Obvious hysteresis appeared in the highpressure section of the desorption curve because of the occurrence of capillary condensation between the particles and in the macropores. The analysis results of the specific surface area, pore volume, and average pore diameter of the 6# and 8# products after pickling are listed in Table 6. It can be seen that the specific surface areas of the 6# and 8# products were 2.01 and 3.44 m 2 g −1 , respectively. The 8# product had a larger specific surface area and pore volume, which was related to the Mg content in the 8# system and temperature. The higher reaction temperature caused the vaporization of metal Mg, and one part of the gasification Mg escaped from the reaction system, and the other part entered the gap of the raw material to participate in the reduction reaction of V 2 O 5 to form metal V. The reduced metal V was sintered together under the action of high temperatures. Therefore, as more Mg was vaporized, the content of gasification Mg that participated in the reduction reaction of V 2 O 5 increased, which led to an increase in the specific surface area and pore volume of the product after pickling. In the 6# product, NaCl did not participate in the reduction reaction of V 2 O 5 but acted as "pore former," resulting in larger pore size of 6# product compared with that of the 8# product.

Reaction mechanism analysis
Figure 9(a) shows the temperature-time curves during the reaction of 6# and 8# samples, and the illustration on the lower right of Figure 9(a) is a schematic diagram of the temperature measurement. It can be seen that the maximum combustion temperature and combustion rate of the 8# reaction system were higher than those of the 6# reaction system. The highest reaction temperature of both 6# and 8# reaction systems was much higher than the boiling point of Mg and slightly higher than the boiling point of NaCl, but lower than the melting point of metal V. Therefore, in the XRD analyses of the products before pickling, the phase diffraction peak of NaCl was present but no phase diffraction peak of Mg. The self-propagating reaction of the V 2 O 5 -Mg system relied on the flow of liquid/gas Mg and the transfer of heat. During the reaction, NaCl did not participate in the reaction but only played a role in reducing the adiabatic temperature.
Since there was no NaCl in the 8# system and the heat release of the system was larger, the production rate of liquid/gas Mg of 8# system was faster than that of the 6# system, resulting in the higher combustion rate of the 8# reaction system than that of the 6# reaction system. The adiabatic temperature of systems with different Mg and   NaCl contents is shown in Figure 9(b). It indicated that NaCl or excessive Mg could reduce the adiabatic temperature of the system as a diluent, and the dilution capacity of NaCl was larger than that of Mg under the same quantity condition. It was assumed that all the V 2 O 5 in the raw materials were converted into the target product V in the process of calculating the adiabatic temperature, that is, no secondary reactions occurred, but only the experimental results of 8# and 6# systems were close to the ideal process in the present research. The adiabatic temperatures of the 8# and 6# systems were 1714.7°C and 1490.6°C, respectively. The highest temperature of 8# and 6# samples during the reaction (Figure 9a) was lower than the adiabatic temperature. This is because the system dissipated heat as it reacted, and there was dissolved oxygen with a mass fraction of 5% in element V obtained by the reduction reaction, which means that the reaction did not proceed completely. The changing trend of the calculated system reaction temperature in Figure 9(b) was basically consistent with the measured temperature in Figure 9(a), suggesting that the addition of NaCl could reduce the adiabatic temperature and the loss of Mg, so as to save the reducing agent. There was no active addition of MgO in this study, and MgV 2 O 4 appeared when the compound of V 2 O 5 and MgO were heated in the air, so the MgV 2 O 4 in this experiment was likely to come from the reduction of V 2 O 5 [26]. The chemical reactions that may occur in this process are shown in formulas (1)-(3). (1) (2)

Conclusion
In this article, V 2 O 5 , Mg, and NaCl were used as raw materials, and ultrafine metal V powder was successfully prepared by the magnesiothermic self-propagating reaction method. The conclusions are as follows: 1) The TG-DTA analysis showed that the reaction temperature of the V 2 O 5 -Mg system was around 570°C, so the self-propagating reaction of the V 2 O 5 -Mg system was a solid-solid reaction. The reaction order n and the apparent activation energy E calculated by the Freeman-Carroll differential method were 0.14 and 1816.227 kJ mol −1 , respectively. 2) The change path of phase during the reaction was V 2 O 5 → V 3 O 5 → MgV 2 O 4 → V. The formation of the complex MgV 2 O 4 was related to the content of Mg involved in the reaction. MgV 2 O 4 was formed in the case of insufficient Mg content, and MgV 2 O 4 cannot be removed by pickling. With the increase of the Mg content involved in the reaction, the morphology of the product after pickling gradually changed from blocky to porous. 3) The temperature-time curve showed that it was feasible to control the reaction temperature of the system by adding NaCl, and the addition of NaCl could achieve the purpose of reducing the reaction temperature and saving the reducing agent. The finally obtained metal V powder containing oxygen with a mass fraction of 4.86 had a porous network structure with a specific surface area of 3.44 m 2 g −1 and an average pore diameter of 45.718 nm.  Author contributions: All authors have accepted responsibility for the entire content of this manuscript and approved its submission.