Lithium Enrichment by the Carbothermal Reduction of Spodumene Ore and the Preparation of Manganese-Silicon Alloy

: To increase the low utilization rate of spodumene ore during lithium extraction, spodumene ore was subjected to carbothermic reduction to enrich lithium and prepare a manganese-silicon alloy. The experimental results showed that during thermal reduction, lithium was volatilized and collected in the condensation zone. The Li 2 O content in the lithium condensate was 41.72%, which was 10.85 times higher than that of the raw material. The effects of varying reduction temperatures and times on the lithium volatilization rate and direct yield of Mn 5 Si 3 alloy were investigated. The best process conditions were 1873 K for 6 h. Under these conditions, the lithium volatilization rate was 97.65%, and the direct yield of Mn 5 Si 3 was 86.47%.

Owing to the problems associated with the acid-sulfur method, in this paper, lithium was enriched by adding coke and manganese dioxide (MnO 2 ) to spodumene by thermal reduction. SiO 2 and Al 2 O 3 were recovered in spodumene ore. The reported method Owing to the problems associated with the acid-sulfur method, in this paper, lithium was enriched by adding coke and manganese dioxide (MnO2) to spodumene by thermal reduction. SiO2 and Al2O3 were recovered in spodumene ore. The reported method represents a new technology for lithium extraction from spodumene for the comprehensive recovery and utilization of its valuable components.

Materials
The spodumene ore used in the experiment was obtained from Africa. X-ray diffraction was used to analyze the crystal phases of spodumene; the results are shown in Figure  1. The ore was composed of spodumene (LiAlSi2O6) and quartz (SiO2). Inductively coupled plasma emission spectroscopy (ICP-OES) was used for quantitative element analysis of the ore; the results are shown in Table 1. The coke was produced by Ningxia Huiheng (China), and its composition is shown in Table 2.

Methods
First, the spodumene ore and coke were crushed and screened through 200 mesh. The spodumene ore, coke, and manganese dioxide were mixed evenly (w LiAlSi 2 O 6 :w MnO 2 :w C = 5:5:4), pressed into blocks under a pressure of 18 MPa, and placed into the graphite crucible of a vacuum furnace. The vacuum pump was turned on, and after reaching the vacuum state, argon gas was passed until the pressure reached 3 × 10 4 -3.5 × 10 4 Pa (the atmospheric pressure in the experimental area was about 8 × 10 4 Pa). Heating was started according to the preset experimental conditions at a rate of approximately 10 K/min. At the end of the reaction (the pressure in the furnace was about 7 × 10 4 -7.5 × 10 4 Pa), the residue in the graphite crucible and the condensate in the stainless-steel crucible were removed for characterization. The reaction process flow was shown in

Methods
First, the spodumene ore and coke were crushed and screened through 200 mesh. The spodumene ore, coke, and manganese dioxide were mixed evenly (w LiAlSi 2 O 6 : w MnO 2 : w C = 5 : 5 : 4), pressed into blocks under a pressure of 18 MPa, and placed into the graphite crucible of a vacuum furnace. The vacuum pump was turned on, and after reaching the vacuum state, argon gas was passed until the pressure reached 3 × 10 4 -3.5 × 10 4 Pa (the atmospheric pressure in the experimental area was about 8 × 10 4 Pa). Heating was started according to the preset experimental conditions at a rate of approximately 10 K/min. At the end of the reaction (the pressure in the furnace was about 7 × 10 4 -7.5 × 10 4 Pa), the residue in the graphite crucible and the condensate in the stainless-steel crucible were removed for characterization. The reaction process flow was shown in Figure 2. Minerals 2022, 12, x FOR PEER REVIEW 3 of 12 Figure 2. Schematic diagram of the reaction process. The raw materials were first mixed by a blender; then, the mixture was pressed into a round cake shape, entering the vacuum furnace for the preset experimental conditions.

Lithium Volatilization Rate
In the formula, ∂ Li is the volatilization rate of lithium, m Li is the mass of lithium in the reduced residue, and M Li is the mass of lithium in the raw material.

Lithium Enrichment Rate
In the formula, γ Li is the enrichment ratio of lithium, w Li is the content of Li in the condensate, and Li is the content of Li in the raw material.

Alloy Direct Yield
In the formula, β Mn 5 Si 3 is the direct yield of manganese-silicon alloy, m Mn 5 Si 3 is the mass of manganese-silicon alloy in the reduced residue, and M Mn 5 Si 3 is the mass of manganese-silicon alloy in the raw material.

Testing Equipment
The phases in spodumene, condensate, and reduction products were detected by Xray diffractometry (XRD, X'Pert Pro MPD, Nalytical, Heracles, Almelo, The Netherlands). Inductively coupled plasma emission spectroscopy (ICP-OES, Optima 8300, PerkinElmer, Waltham, MA, USA) was used to quantitatively analyze the main chemical components in the raw materials. The morphology of the reduced products was observed and characterized by scanning electron microscopy (TM-3030 Plus, Hitachi, Tokyo, Japan) and energy-dispersive spectrometry (INCA, Oxford, UK). The lithium content of the condensate and reduction products was measured by an atomic absorption spectrophotometer (AAS; iCE 3500, Thermo, Waltham, MA, USA). Schematic diagram of the reaction process. The raw materials were first mixed by a blender; then, the mixture was pressed into a round cake shape, entering the vacuum furnace for the preset experimental conditions.

Lithium Volatilization Rate
In the formula, ∂ Li is the volatilization rate of lithium, m Li is the mass of lithium in the reduced residue, and M Li is the mass of lithium in the raw material.

Lithium Enrichment Rate
In the formula, γ Li is the enrichment ratio of lithium, w Li is the content of Li in the condensate, and W Li is the content of Li in the raw material.

Alloy Direct Yield
In the formula, β Mn 5 Si 3 is the direct yield of manganese-silicon alloy, m Mn 5 Si 3 is the mass of manganese-silicon alloy in the reduced residue, and M Mn 5 Si 3 is the mass of manganese-silicon alloy in the raw material.

Testing Equipment
The phases in spodumene, condensate, and reduction products were detected by X-ray diffractometry (XRD, X'Pert Pro MPD, Nalytical, Heracles, Almelo, The Netherlands). Inductively coupled plasma emission spectroscopy (ICP-OES, Optima 8300, PerkinElmer, Waltham, MA, USA) was used to quantitatively analyze the main chemical components in the raw materials. The morphology of the reduced products was observed and characterized by scanning electron microscopy (TM-3030 Plus, Hitachi, Tokyo, Japan) and energy-dispersive spectrometry (INCA, Oxford, UK). The lithium content of the condensate and reduction products was measured by an atomic absorption spectrophotometer (AAS; iCE 3500, Thermo, Waltham, MA, USA).

Thermodynamic Calculations
HSC6.0 thermodynamic software was used to calculate the possible reactions; the results are shown in Figure 3. SiO 2 in the ore participated during the reaction and preferentially formed LiAlSi 2 O 6 , indicating that Mn 5 Si 3 was preferentially formed over lithium vapor and alumina (Al 2 O 3 ). When MnO 2 was added to the reaction, the reaction temperature dropped significantly to 651 K and 667 K lower than that of SiO 2 + C and LiAlSi 2 O 6 + C, respectively, indicating that adding MnO 2 reduced the reaction temperature and facilitated the reaction. Therefore, this process was thermodynamically feasible.
Minerals 2022, 12, x FOR PEER REVIEW

Thermodynamic Calculations
HSC6.0 thermodynamic software was used to calculate the possible reactio results are shown in Figure 3. SiO2 in the ore participated during the reaction and entially formed LiAlSi2O6, indicating that Mn5Si3 was preferentially formed over l vapor and alumina (Al2O3). When MnO2 was added to the reaction, the reaction tem ture dropped significantly to 651 K and 667 K lower than that of SiO2 + C and LiAl C, respectively, indicating that adding MnO2 reduced the reaction temperature and itated the reaction. Therefore, this process was thermodynamically feasible.

Condensate Analysis
Due to the limited feed amount and the low lithium content in spodumene ore, difficult to collect the lithium condensate in a single step, so the lithium-rich cond was collected through multiple experiments. The phase analysis of the condensa carried out by XRD; the results are shown in Figure 4. Li mainly existed in the cond in the form of Li2SiO3. By referring to the literature [27], we found that Li2O an formed eutectic compounds in the range of 1028-1201 °C. Therefore, it is believed th Li2SiO3 generated in the condensation zone was formed by Li in the condensation and O2 in air to generate Li2O, which then reacted with SiO2 to form eutectic compo The reaction equation is shown in Reaction 4. The Li content in the condensate was 1 as detected by atomic absorption spectrophotometry, and the Li2O content was 4 which was 10.85 times more enriched compared with spodumene. 4Li (g) + 2O2 (g) + SiO2 = 2Li2SiO3,

Condensate Analysis
Due to the limited feed amount and the low lithium content in spodumene ore, it was difficult to collect the lithium condensate in a single step, so the lithium-rich condensate was collected through multiple experiments. The phase analysis of the condensate was carried out by XRD; the results are shown in Figure 4. Li mainly existed in the condensate in the form of Li 2 SiO 3 . By referring to the literature [27], we found that Li 2 O and SiO 2 formed eutectic compounds in the range of 1028-1201 • C. Therefore, it is believed that the Li 2 SiO 3 generated in the condensation zone was formed by Li in the condensation zone and O 2 in air to generate Li 2 O, which then reacted with SiO 2 to form eutectic compounds. The reaction equation is shown in Reaction 4. The Li content in the condensate was 19.47%, as detected by atomic absorption spectrophotometry, and the Li 2 O content was 41.72%, which was 10.85 times more enriched compared with spodumene.

The Effect of Temperature
Based on the theoretical calculations and experimental exploration, time was set to 6 h to ensure the completion of the reaction. After the rea differences in density, the graphite crucible underwent delamination.  Figure 5 shows that LiAlSi2O6 did not begin to r or 1723 K, indicating that the reaction temperature was too low. The Mn5 peak appeared because SiO2 in the ore preferentially reacted with C and MnO consistent with the thermodynamic calculations showing that SiO2 should produce LiAlSi2O6. However, as shown in Figure 6, SiC and Mn2C5 were p upper-layer material, and SiC and Mn2C5 were intermediates during the Mn5Si3. This indicates that the reaction between SiO2 and MnO2 was incom the temperature reached 1773 K and 1823 K, the diffraction peaks shown i changed to those of Mn5Si3, indicating that the lower-layer material was co loys at these temperatures. The diffraction peaks of Al2O3 and LiAlSi2O6 sho 6 indicate that LiAlSi2O6 began to react but did so incompletely. When the was 1873 K, the diffraction peak of LiAlSi2O6 shown in Figure 6 disappeare tensity of the diffraction peaks of Al2O3 and SiC was enhanced. At this tim completely reacted, but the diffraction peak of Mn5Si3 remained, indicating amount of Mn5Si3 was mixed into the upper-layer material. As the reaction continued to increase, the Mn5Si3 diffraction peak remained. When the tem 1923 K, the intensity of the SiC diffraction peak shown in Figure 6 was enha ing that the SiC content increased. The high content of SiC affected the vi upper-layer material, which inhibited the upward diffusion of lithium vapo a decrease in the volatilization rate of lithium. Therefore, the formation of S controlled during the reaction.

The Effect of Temperature
Based on the theoretical calculations and experimental exploration, the reduction time was set to 6 h to ensure the completion of the reaction. After the reaction, due to differences in density, the graphite crucible underwent delamination. Figures 5 and 6 show the XRD patterns of the lower-and upper-layer materials from 1673 K to 1973 K with a holding time of 6 h. Figure 5 shows that LiAlSi 2 O 6 did not begin to react at 1673 K or 1723 K, indicating that the reaction temperature was too low. The Mn 5 Si 2 diffraction peak appeared because SiO 2 in the ore preferentially reacted with C and MnO 2 , which was consistent with the thermodynamic calculations showing that SiO 2 should preferentially produce LiAlSi 2 O 6 . However, as shown in Figure 6, SiC and Mn 2 C 5 were present in the upper-layer material, and SiC and Mn 2 C 5 were intermediates during the formation of Mn 5 Si 3 . This indicates that the reaction between SiO 2 and MnO 2 was incomplete. When the temperature reached 1773 K and 1823 K, the diffraction peaks shown in Figure 6 all changed to those of Mn 5 Si 3 , indicating that the lower-layer material was composed of alloys at these temperatures. The diffraction peaks of Al 2 O 3 and LiAlSi 2 O 6 shown in Figure 6 indicate that LiAlSi 2 O 6 began to react but did so incompletely. When the temperature was 1873 K, the diffraction peak of LiAlSi 2 O 6 shown in Figure 6 disappeared, and the intensity of the diffraction peaks of Al 2 O 3 and SiC was enhanced. At this time, LiAlSi 2 O 6 completely reacted, but the diffraction peak of Mn 5 Si 3 remained, indicating that a small amount of Mn 5 Si 3 was mixed into the upper-layer material. As the reaction temperature continued to increase, the Mn 5 Si 3 diffraction peak remained. When the temperature was 1923 K, the intensity of the SiC diffraction peak shown in Figure 6 was enhanced, indicating that the SiC content increased. The high content of SiC affected the viscosity of the upper-layer material, which inhibited the upward diffusion of lithium vapor, resulting in a decrease in the volatilization rate of lithium. Therefore, the formation of SiC should be controlled during the reaction.  The direct yield of Mn5Si3 was obtained by weighing the alloy in the lower-layer material, as shown in Figure 7. The direct yield of Mn5Si3 was low at temperatures below 1873 K because the spodumene ore had not completely reacted to form an alloy. When the temperature was 1873 K, the direct yield of Mn5Si3 was 86.47%, which was the maximum yield. Continuing to increase the reaction temperature resulted in a slight decrease in the direct yield of Mn5Si3. The reason for the low direct yield of Mn5Si3 was that the formation of SiC caused the loss of Si, and a small amount of Mn5Si3 entered the upper layer. After grinding and mixing the reduced residues in a graphite crucible at varying reduction temperatures, the Li content was detected by atomic absorption spectrophotometry. The influence of varying reduction temperatures on the volatilization rate of Li during the experimental process was investigated; the results are shown in Figure 8. At temperatures below 1723 K, spodumene ore did not participate in the reaction, so the volatilization rate of lithium was 0%. When the reduction temperature was 1773 K, the volatilization rate of Li was 32.75%, indicating that LiAlSi2O6 had begun to be reduced at this temperature, although the reduction degree was low. Upon increasing the temperature, the volatilization rate of Li gradually increased. When the temperature was 1873 K, the volatilization rate of Li was 97.65%. Upon continuing to increase the temperature, the Li volatilization rate remained identical to that at 1873 K, indicating that spodumene completely reacted  The direct yield of Mn5Si3 was obtained by weighing the alloy in the lower-layer material, as shown in Figure 7. The direct yield of Mn5Si3 was low at temperatures below 1873 K because the spodumene ore had not completely reacted to form an alloy. When the temperature was 1873 K, the direct yield of Mn5Si3 was 86.47%, which was the maximum yield. Continuing to increase the reaction temperature resulted in a slight decrease in the direct yield of Mn5Si3. The reason for the low direct yield of Mn5Si3 was that the formation of SiC caused the loss of Si, and a small amount of Mn5Si3 entered the upper layer. After grinding and mixing the reduced residues in a graphite crucible at varying reduction temperatures, the Li content was detected by atomic absorption spectrophotometry. The influence of varying reduction temperatures on the volatilization rate of Li during the experimental process was investigated; the results are shown in Figure 8. At temperatures below 1723 K, spodumene ore did not participate in the reaction, so the volatilization rate of lithium was 0%. When the reduction temperature was 1773 K, the volatilization rate of Li was 32.75%, indicating that LiAlSi2O6 had begun to be reduced at this temperature, although the reduction degree was low. Upon increasing the temperature, the volatilization rate of Li gradually increased. When the temperature was 1873 K, the volatilization rate of Li was 97.65%. Upon continuing to increase the temperature, the Li volatilization rate remained identical to that at 1873 K, indicating that spodumene completely reacted The direct yield of Mn 5 Si 3 was obtained by weighing the alloy in the lower-layer material, as shown in Figure 7. The direct yield of Mn 5 Si 3 was low at temperatures below 1873 K because the spodumene ore had not completely reacted to form an alloy. When the temperature was 1873 K, the direct yield of Mn 5 Si 3 was 86.47%, which was the maximum yield. Continuing to increase the reaction temperature resulted in a slight decrease in the direct yield of Mn 5 Si 3 . The reason for the low direct yield of Mn 5 Si 3 was that the formation of SiC caused the loss of Si, and a small amount of Mn 5 Si 3 entered the upper layer. After grinding and mixing the reduced residues in a graphite crucible at varying reduction temperatures, the Li content was detected by atomic absorption spectrophotometry. The influence of varying reduction temperatures on the volatilization rate of Li during the experimental process was investigated; the results are shown in Figure 8. At temperatures below 1723 K, spodumene ore did not participate in the reaction, so the volatilization rate of lithium was 0%. When the reduction temperature was 1773 K, the volatilization rate of Li was 32.75%, indicating that LiAlSi 2 O 6 had begun to be reduced at this temperature, although the reduction degree was low. Upon increasing the temperature, the volatilization rate of Li gradually increased. When the temperature was 1873 K, the volatilization rate of Li was 97.65%. Upon continuing to increase the temperature, the Li volatilization rate remained identical to that at 1873 K, indicating that spodumene completely reacted at 1873 K. This was consistent with the analysis of the results presented in Figures 5 and 6.

The Effect of Time
In the previous section, the optimal reduction temperature was determined 1873 K by exploring the influence of temperature. In this section, we designed a gr pair experiment with a reduction time of 1-10 h at the optimal reduction temperatu material ratio and detection and treatment method used in the experiment were th as those described in the previous section. Figure 9 shows that before 4 h, in addi the Mn5Si3 diffraction peaks, there were Mn5C2 and SiC diffraction peaks in the layer material. These peaks indicate that the reaction was incomplete. Upon increas reduction time to 6 h, the diffraction peaks of the lower-layer material were due to and Mn5C2. The SiC diffraction peaks disappeared, which showed that 6 h was eno completely separate the upper and lower layers. Figure 10 shows that when the red time was 4 h, the diffraction peak of LiAlSi2O6 disappeared. Thus, LiAlSi2O6 wa pletely reduced, but MnO2 was not completely reduced, indicating that MnO2 req longer reduction time than LiAlSi2O6.

The Effect of Time
In the previous section, the optimal reduction temperature was determined 1873 K by exploring the influence of temperature. In this section, we designed a gr pair experiment with a reduction time of 1-10 h at the optimal reduction temperatur material ratio and detection and treatment method used in the experiment were the as those described in the previous section. Figure 9 shows that before 4 h, in addit the Mn5Si3 diffraction peaks, there were Mn5C2 and SiC diffraction peaks in the l layer material. These peaks indicate that the reaction was incomplete. Upon increasi reduction time to 6 h, the diffraction peaks of the lower-layer material were due to M and Mn5C2. The SiC diffraction peaks disappeared, which showed that 6 h was eno completely separate the upper and lower layers. Figure 10 shows that when the red time was 4 h, the diffraction peak of LiAlSi2O6 disappeared. Thus, LiAlSi2O6 was pletely reduced, but MnO2 was not completely reduced, indicating that MnO2 requ longer reduction time than LiAlSi2O6.

The Effect of Time
In the previous section, the optimal reduction temperature was determined to be 1873 K by exploring the influence of temperature. In this section, we designed a gradient pair experiment with a reduction time of 1-10 h at the optimal reduction temperature. The material ratio and detection and treatment method used in the experiment were the same as those described in the previous section. Figure 9 shows that before 4 h, in addition to the Mn 5 Si 3 diffraction peaks, there were Mn 5 C 2 and SiC diffraction peaks in the lowerlayer material. These peaks indicate that the reaction was incomplete. Upon increasing the reduction time to 6 h, the diffraction peaks of the lower-layer material were due to Mn 5 Si 3 and Mn 5 C 2 . The SiC diffraction peaks disappeared, which showed that 6 h was enough to completely separate the upper and lower layers. Figure 10 shows that when the reduction time was 4 h, the diffraction peak of LiAlSi 2 O 6 disappeared. Thus, LiAlSi 2 O 6 was completely reduced, but MnO 2 was not completely reduced, indicating that MnO 2 required a longer reduction time than LiAlSi 2 O 6 .  The direct yield of Mn5Si3 with varying reduction times is shown in Figure  direct yield of Mn5Si3 increased significantly when the reduction time was increa to 4 h, from 56.33% to 81.55%, indicating that LiAlSi2O6 was involved in the reactio 2 h to 4 h. When the reduction time was 6 h, the direct yield of the alloy was the h When the reduction time was increased to 8 h, the direct yield of Mn5Si3 decreased s but the change was insignificant.
The volatilization rate of lithium with varying reduction times is shown in Fig  There was a large increase in 2-4 h, which further verified the conclusion that Li was considerably reduced with reaction times of 2-4 h. Upon extending the time, latilization rate of lithium increased gradually. When the reduction time was 6 h, latilization rate of lithium reached 97.65%, and the reduction of lithium was comp  The direct yield of Mn5Si3 with varying reduction times is shown in Figure 1 direct yield of Mn5Si3 increased significantly when the reduction time was increa to 4 h, from 56.33% to 81.55%, indicating that LiAlSi2O6 was involved in the reactio 2 h to 4 h. When the reduction time was 6 h, the direct yield of the alloy was the h When the reduction time was increased to 8 h, the direct yield of Mn5Si3 decreased s but the change was insignificant.
The volatilization rate of lithium with varying reduction times is shown in Fig  There was a large increase in 2-4 h, which further verified the conclusion that LiA was considerably reduced with reaction times of 2-4 h. Upon extending the time, latilization rate of lithium increased gradually. When the reduction time was 6 h, latilization rate of lithium reached 97.65%, and the reduction of lithium was comp The direct yield of Mn 5 Si 3 with varying reduction times is shown in Figure 11. The direct yield of Mn 5 Si 3 increased significantly when the reduction time was increased 2 h to 4 h, from 56.33% to 81.55%, indicating that LiAlSi 2 O 6 was involved in the reaction from 2 h to 4 h. When the reduction time was 6 h, the direct yield of the alloy was the highest. When the reduction time was increased to 8 h, the direct yield of Mn 5 Si 3 decreased slightly, but the change was insignificant.

SEM-EDS Analysis
The upper-and lower-layer materials under the abovementioned optimal con were analyzed by SEM-EDS; the results are shown in Figures 13 and 14. Figure 13a that the alloy was bright white with a smooth surface and no cracks or other defec ure 13b,c shows that the coincidence degree of Mn and Si was high. Figure 14a shows the topography of the upper-layer material, which was m gray material, mixed with a small amount of bright white material. Three point selected for point scanning; the results are shown in Figure 14b-d. The compound a 1 was Al2O3, and that at point 2 was Mn5Si3. The inclusions were Al2O3 and SiC. Th pound at point 3 was SiC, and a small amount of Al2O3 was included, which wa sistent with the previous XRD analysis results. Figure 14e-  The volatilization rate of lithium with varying reduction times is shown in Figure 12.
There was a large increase in 2-4 h, which further verified the conclusion that LiAlSi 2 O 6 was considerably reduced with reaction times of 2-4 h. Upon extending the time, the volatilization rate of lithium increased gradually. When the reduction time was 6 h, the volatilization rate of lithium reached 97.65%, and the reduction of lithium was complete.

SEM-EDS Analysis
The upper-and lower-layer materials under the abovementioned optimal con were analyzed by SEM-EDS; the results are shown in Figures 13 and 14. Figure 13a that the alloy was bright white with a smooth surface and no cracks or other defec ure 13b,c shows that the coincidence degree of Mn and Si was high. Figure 14a shows the topography of the upper-layer material, which was m gray material, mixed with a small amount of bright white material. Three point selected for point scanning; the results are shown in Figure 14b-d. The compound a 1 was Al2O3, and that at point 2 was Mn5Si3. The inclusions were Al2O3 and SiC. Th Based on single-factor experimental analysis, it can be concluded that the optimal parameters for studying lithium enrichment by the carbothermal reduction of spodumene to prepare manganese-silicon alloy were 1873 K and 6 h.

SEM-EDS Analysis
The upper-and lower-layer materials under the abovementioned optimal conditions were analyzed by SEM-EDS; the results are shown in Figures 13 and 14. Figure 13a shows that the alloy was bright white with a smooth surface and no cracks or other defects. Figure 13b,c shows that the coincidence degree of Mn and Si was high.

Conclusions
1. Thermodynamic calculations were carried out using HSC6.0 thermodynamic software. We found that SiO2 in the ore preferentially participated in the reaction compared with LiAlSi2O6. Compared with the reaction of SiO2 + C and LiAlSi2O6 + C, the

Conclusions
1. Thermodynamic calculations were carried out using HSC6.0 thermodynamic software. We found that SiO2 in the ore preferentially participated in the reaction compared with LiAlSi2O6. Compared with the reaction of SiO2 + C and LiAlSi2O6 + C, the  Figure 14a shows the topography of the upper-layer material, which was mainly a gray material, mixed with a small amount of bright white material. Three points were selected for point scanning; the results are shown in Figure 14b-d. The compound at point 1 was Al 2 O 3 , and that at point 2 was Mn 5 Si 3 . The inclusions were Al 2 O 3 and SiC. The compound at point 3 was SiC, and a small amount of Al 2 O 3 was included, which was consistent with the previous XRD analysis results. Figure 14e-h shows the element distribution of the upper-layer material. The surface scan of the upper material showed the presence of highly overlapping Al and O elements, indicating that the upper-layer material mainly comprised Al 2 O 3 , with a small amount of Mn 5 Si 3 mixed in. This indicates that even under the optimal conditions, a small amount of Mn 5 Si 3 still entered the upper-layer material because when LiAlSi 2 O 6 formed Mn 5 Si 3 , Al 2 O 3 was also decomposed, which caused a small amount of Mn 5 Si 3 to be carried into the upper layer by Al 2 O 3 . The mechanistic process of the reaction was shown in Figure 15.

Conclusions
1. Thermodynamic calculations were carried out using HSC6.0 thermodynamic software. We found that SiO2 in the ore preferentially participated in the reaction compared with LiAlSi2O6. Compared with the reaction of SiO2 + C and LiAlSi2O6 + C, the decrease was 651 K and 667 K, respectively, indicating that the addition of MnO2 reduced the reaction temperature and increased the feasibility of the reaction. 2. Due to the limited feed amount and the low lithium content in spodumene ore, it was difficult to collect the lithium condensate in a single step, so the lithium-rich

1.
Thermodynamic calculations were carried out using HSC6.0 thermodynamic software. We found that SiO 2 in the ore preferentially participated in the reaction compared with LiAlSi 2 O 6 . Compared with the reaction of SiO 2 + C and LiAlSi 2 O 6 + C, the decrease was 651 K and 667 K, respectively, indicating that the addition of MnO 2 reduced the reaction temperature and increased the feasibility of the reaction.

2.
Due to the limited feed amount and the low lithium content in spodumene ore, it was difficult to collect the lithium condensate in a single step, so the lithium-rich condensate was collected through multiple experiments. The collected condensate contained 19.47% Li, which was 10.85 times more enriched compared with spodumene.

3.
Reduction temperature and time were two important factors affecting the reaction. With increased temperature and time, spodumene ore was gradually reduced; when the reduction temperature was 1873 K and the reduction time was 6 h, the spodumene ore reaction was complete, the volatilization rate of lithium was 97.65%, and the direct yield of Mn 5 Si 3 was 86.47%.

Data Availability Statement:
The data presented in this study are available upon request from the corresponding author. The data are not publicly available because related studies are ongoing.