Research on Reaction Mechanism of Vacuum Carbon Thermal Reduction and Dephosphorization in High Phosphate Iron Ore

Abstract: According to the mineral composition characteristics of high-phosphorus iron ore, the reaction mechanism of fluorapatite was investigated using pure substance and gangue under vacuum carbon thermal reduction (VCTR) conditions. The effects of reduction temperature, basicity, and C/O ratio on the metallization ratio, dephosphorization ratio, and phosphorus content of pellets were studied. The reaction process of fluorapatite in high-phosphorus iron ore was investigated. The results showed that when the metallization ratio of pellets reached maximum (95%), the dephosphorization ratio was only 5.6%, thus indicating adverse result. The reduction processes of high-phosphorus iron ore under vacuum and nitrogen environment were, respectively, compared under the optimal condition. It was found that the metallization ratio of pellets in the vacuum condition was higher than that under the nitrogen condition, while the dephosphorization ratio showed an opposite result. This indicated that in the process of vacuum reduction, fluorapatite not only reacted with carbon to form gaseous phosphide, but also with iron to form compounds containing the Fe–P bond. Therefore, a new mechanism of reduction of fluorapatite was proposed as follows: 2Ca5(PO4)3F + 12Fe + 9SiO2 + 15C = 9CaSiO3 + 6Fe2P + 15CO + CaF2.


Introduction
High-phosphorus iron ore has been paid extensive attention because of the difficulty of effective separating iron and phosphorus.There exists an abundance of high-phosphorus iron ore resources in China, up to 7.45 billion tons distributed primarily in Hubei, Jiangsu, Yunnan, and Anhui provinces [1][2][3].Currently, owing to the high reduction temperature of the ironmaking process, the phosphorus content increases significantly and almost the entire phosphorus is absorbed by the molten iron resulting in difficult separation in a rotary hearth furnace (RHF) [4][5][6].Therefore, high-phosphorus iron ore does not easily meet the requirements for the ironmaking process.Effective achievement of the separation of phosphorus from iron is the foundation of the effective utilization of high-phosphorus iron ore in the ironmaking process [7][8][9][10].
In order to have a better knowledge of the dephosphorization behavior of high-phosphorus iron ore, many separation methods have been proposed.The research methods of separation were classified into the following types: physical method [11][12][13], chemical method [14][15][16][17] magnetization roasting-magnetic separation method [18][19][20][21][22], and coal-based direct reduction method [23][24][25][26][27][28].Among these reported separation methods, the coal-based direct reduction method has better characteristics such as simplicity of the process, a convenient and easy operation, and being In this study, blending of pure substances was carried out to simulate high-phosphorus iron ore.The simulated sample corresponded to high-phosphorus iron ore reserves in western Hubei Province, China.The composition and mineralogy distribution of high-phosphorus iron ore had been detected by X-ray diffraction (XRD, Rigaku, Tokyo, Japan) and electron probe micro-analyzer (EPMA, JEOL, Tokyo, Japan) in our previous paper [26].Experimental materials included analytical pure Fe 2 O 3 , SiO 2 , Al 2 O 3 , and graphite were provided by Sinopharm Group Co., Ltd.(Shanghai, China).The synthetic fluorapatite was analyzed by Raman spectroscopy and XRD.The rationality of the synthetic fluorapatite was demonstrated by comparing with peak of standard material.According to the weight ratios of the material listed in Table 1, the approximate and simulated composition mass ratios were obtained and listed in Table 2.

Procedures
The simulated synthetic fluorapatite, high-purity graphite, CaO, CaF 2 , and Na 2 CO 3 were ground to below 200 mesh.A certain proportion of substances was mixed in accordance with simulate ratio.Then, 5 g mixture was pressed at 16 MPa into cylindrical pellets with a diameter of 15 mm and a thickness of about 8 mm.The pellets were dried at 393 K (120 • C) for 12 h in a drying oven.
The schematic illustration of a vacuum tube furnace was shown in Figure 1.It was prepared and provided by the Shanghai institute of optical precision machinery, Chinese academy of science.During the test, the furnance was first heated to the predetermined temperature and kept for a period of time.Then, the prepared sample was placed in a crucible, which was surrrouded by quartz tube, and sealed by a flange.Furthermore, the quartz tube was vacuumed using a vacuum pump until the vacuum degree reached 100 Pa.Finally, the quartz tube was put into the furnance heated to a predetermined temperature.After a certain reation time, the vacuum was released and the sample was taken out.

Procedures
The simulated synthetic fluorapatite, high-purity graphite, CaO, CaF2, and Na2CO3 were ground to below 200 mesh.A certain proportion of substances was mixed in accordance with simulate ratio.Then, 5 g mixture was pressed at 16 MPa into cylindrical pellets with a diameter of 15 mm and a thickness of about 8 mm.The pellets were dried at 393 K (120 °C) for 12 h in a drying oven.
The schematic illustration of a vacuum tube furnace was shown in Figure 1.It was prepared and provided by the Shanghai institute of optical precision machinery, Chinese academy of science.During the test, the furnance was first heated to the predetermined temperature and kept for a period of time.Then, the prepared sample was placed in a crucible, which was surrrouded by quartz tube, and sealed by a flange.Furthermore, the quartz tube was vacuumed using a vacuum pump until the vacuum degree reached 100 Pa.Finally, the quartz tube was put into the furnance heated to a predetermined temperature.After a certain reation time, the vacuum was released and the sample was taken out.

Dephosphorization ratio η
The phosphorus content in the pellets was obtained by chemical analysis carried out in the National Analysis Center for Iron and Steel (NACIS).The macroscopic analysis of pellets was investigated by XRD and the microscopic behavior of pellets in the process was observed by scanning electron microscopy-energy dispersive spectrometry (SEM-EDS) (JEOL, Tokyo, Japan) and EPMA.The dephosphorization ratio of pellets was calculated by using the following formula:

Dephosphorization Ratio η
The phosphorus content in the pellets was obtained by chemical analysis carried out in the National Analysis Center for Iron and Steel (NACIS).The macroscopic analysis of pellets was investigated by XRD and the microscopic behavior of pellets in the process was observed by scanning electron microscopy-energy dispersive spectrometry (SEM-EDS) (JEOL, Tokyo, Japan) and EPMA.The dephosphorization ratio of pellets was calculated by using the following formula: P t : the phosphorus content in the pellets after reduction; P raw : the phosphorus content in raw pellets.

Metallization Ratio ρ
The metallization ratio is the most direct index to evaluate the reduction degree of iron oxides in the direct reduction process.Its definition is expressed in terms of Equation ( 2): M Fe : the metal iron content in the pellets after reduction; T Fe : the total iron content in the pellets after reduction.

Effect of Reduction Temperature on High-Phosphorus Iron Ore in Vacuum Reduction
The reduction temperature significantly affects the reduction and dephosphorization processes of high-phosphate iron ore.The lower reduction temperature leads to a significant decrease in the reduction rate and the high-phosphorus iron ore could not be entirely dephosphorized.Higher reduction temperature leads to the absorption of gaseous phosphorus by liquid iron and the phosphorus content in the iron bead increases.Therefore, it is very important to determine a suitable temperature for the reduction of high-phosphate iron ore.In our previous study, it was found that the appropriate temperature in the range of 1000 to 1150 • C could effectively inhibit the melting of the iron phase and improve the reduction rate.Therefore, the reduction temperature was fixed at 1000-1150 • C. The ratio of raw material was based on the data listed in Table 2. Figure 2 exhibits that the metallization ratio rapidly increases with the increase of temperature; however, the dephosphorization ratio and phosphorus content of iron bead are weakly dependent on the reduction temperature.It is noteworthy that the dephosphorization ratio decreased at 1100 • C.This result indicated that the reaction rate of carbon-reduced iron oxide rapidly increased at 1100 • C and the carbon was rapidly consumed in a short time, leading to a poor effect involving the reduction of fluorapatite by carbon.Owing to the low dephosphorization rate, a large amount of phosphorus was absorbed by metallic iron during the melting process and the high phosphorus content of the iron bead was obtained.
Figure 3 highlights the comparison of the XRD patterns of the reduced pellets under different temperatures, which is evidence that the peak intensity of metallic iron increases with increase of temperature; and slag phases, such as Ca 4 Si 2 O 7 F 2 , Ca 2 Al 2 SiO 7 , Fe 2 SiO 4 and FeAl 2 O 4 gradually appear.However, a diffraction peak representing element P or the phosphorus-containing phase does not appear.This indicates that temperature exhibited a significant influence on the metallization ratio of pellets; nonetheless, the effect on the dephosphorization ratio was not obvious.
temperature.It is noteworthy that the dephosphorization ratio decreased at 1100 °C.This result indicated that the reaction rate of carbon-reduced iron oxide rapidly increased at 1100 °C and the carbon was rapidly consumed in a short time, leading to a poor effect involving the reduction of fluorapatite by carbon.Owing to the low dephosphorization rate, a large amount of phosphorus was absorbed by metallic iron during the melting process and the high phosphorus content of the iron bead was obtained.Figure 3 highlights the comparison of the XRD patterns of the reduced pellets under different temperatures, which is evidence that the peak intensity of metallic iron increases with increase of temperature; and slag phases, such as Ca4Si2O7F2, Ca2Al2SiO7, Fe2SiO4 and FeAl2O4 gradually appear.However, a diffraction peak representing element P or the phosphorus-containing phase does not appear.This indicates that temperature exhibited a significant influence on the metallization ratio of pellets; nonetheless, the effect on the dephosphorization ratio was not obvious.

Effect of Carbon on High Phosphate Iron Ore in Vacuum Reduction
The effects of the variations in C/O from 0.8 to 1.2 on the VCTR process of high-phosphorus iron ore were investigated.The C/O (the ratio of graphite to Fe2O3) was changed by adjusting the content of high-purity graphite with constant content of Fe2O3.The metallization and dephosphorization ratios of pellets slightly increase with the increase of C/O; however, the phosphorus content of the iron bead is independent of the C/O (Figures 4 and 5).This is attributed to the fact that the content of the reducing agent increased with increasing C/O ratio, contributing to the reduction of iron oxides.Moreover, the generated CO and CO2 gas promoted the volatilization of the phosphorous gas phase, which led to a certain increase in the dephosphorization ratio.There was no significant change in the phosphorus content of the iron bead, which indicated that the iron phase absorbed a large amount of phosphorus during the melting process.Figure 5 shows that the peak intensity of fluoroapatite does not change significantly with the increase of carbon content.This indicated that the increase in carbon content did not promote the reduction of fluoroapatite.On the whole, the metallization ratio of pellets with different C/O ratios was above 90%.With the increase of C/O, the metallization ratio of pellets exhibited relatively less change.

Effect of Carbon on High Phosphate Iron Ore in Vacuum Reduction
The effects of the variations in C/O from 0.8 to 1.2 on the VCTR process of high-phosphorus iron ore were investigated.The C/O (the ratio of graphite to Fe 2 O 3 ) was changed by adjusting the content of high-purity graphite with constant content of Fe 2 O 3 .The metallization and dephosphorization ratios of pellets slightly increase with the increase of C/O; however, the phosphorus content of the iron bead is independent of the C/O (Figures 4 and 5).This is attributed to the fact that the content of the reducing agent increased with increasing C/O ratio, contributing to the reduction of iron oxides.Moreover, the generated CO and CO 2 gas promoted the volatilization of the phosphorous gas phase, which led to a certain increase in the dephosphorization ratio.There was no significant change in the phosphorus content of the iron bead, which indicated that the iron phase absorbed a large amount of Metals 2018, 8, 1003 6 of 17 phosphorus during the melting process.Figure 5 shows that the peak intensity of fluoroapatite does not change significantly with the increase of carbon content.This indicated that the increase in carbon content did not promote the reduction of fluoroapatite.On the whole, the metallization ratio of pellets with different C/O ratios was above 90%.With the increase of C/O, the metallization ratio of pellets exhibited relatively less change.

Effect of Basicity on High Phosphate Iron Ore in Vacuum Reduction
The effects of basicity (the ratio of CaO to SiO2), which changed by regulating CaO content with constant content of SiO2, in the range of 1.1-1.4 on the reduced pellets, were obtained with constant ratios of raw materials.Figure 6 shows that the basicity has little effect on the metallization rate, dephosphorization ratio, and phosphorus content of iron bead.This was mainly attributed to the fact that CaO reacted with SiO2 leading to the inhibition of the reduction effect on iron oxide, and the dephosphorization effect was not obvious.Figure 7 demonstrates that the pellets with different basicity undergo little difference in phase composition after reduction at 1100 °C for 30 min, and the

Effect of Basicity on High Phosphate Iron Ore in Vacuum Reduction
The effects of basicity (the ratio of CaO to SiO2), which changed by regulating CaO content with constant content of SiO2, in the range of 1.1-1.4 on the reduced pellets, were obtained with constant ratios of raw materials.Figure 6 shows that the basicity has little effect on the metallization rate, dephosphorization ratio, and phosphorus content of iron bead.This was mainly attributed to the fact that CaO reacted with SiO2 leading to the inhibition of the reduction effect on iron oxide, and the dephosphorization effect was not obvious.Figure 7 demonstrates that the pellets with different basicity undergo little difference in phase composition after reduction at 1100 °C for 30 min, and the peak intensity of each phase is also similar.This indicated that the basicity had little influence on the

Effect of Basicity on High Phosphate Iron Ore in Vacuum Reduction
The effects of basicity (the ratio of CaO to SiO 2 ), which changed by regulating CaO content with constant content of SiO 2 , in the range of 1.1-1.4 on the reduced pellets, were obtained with constant ratios of raw materials.Figure 6 shows that the basicity has little effect on the metallization rate, dephosphorization ratio, and phosphorus content of iron bead.This was mainly attributed to the fact that CaO reacted with SiO 2 leading to the inhibition of the reduction effect on iron oxide, and the dephosphorization effect was not obvious.Figure 7 demonstrates that the pellets with different basicity undergo little difference in phase composition after reduction at 1100 • C for 30 min, and the peak intensity of each phase is also similar.This indicated that the basicity had little influence on the reduction process of the pellets under vacuum conditions.According to the aforementioned results, the conclusions could be drawn that a higher metallization rate (95%) was obtained at the optimum condition of VCTR (reduction temperature = 1100 °C, basicity = 1.1-1.3,and C/O = 0.8-1.0);however, this failed to achieve the purpose of a high dephosphorization rate.Therefore, the reasons for the low dephosphorization ratio should be discussed by investigating the reduction mechanism of fluoroapatite.

The Reaction Mechanism of High-Phosphate Iron Ore in Vacuum Reduction
Therefore, in order to further investigate the reasons for the low dephosphorization efficiency of fluoroapatite during VCTR, the reduction mechanism of fluoroapatite in the vacuum reduction process of high-phosphorus iron ore was studied.As shown in Table 3, Samples were prepared  According to the aforementioned results, the conclusions could be drawn that a higher metallization rate (95%) was obtained at the optimum condition of VCTR (reduction temperature = 1100 °C, basicity = 1.1-1.3,and C/O = 0.8-1.0);however, this failed to achieve the purpose of a high dephosphorization rate.Therefore, the reasons for the low dephosphorization ratio should be discussed by investigating the reduction mechanism of fluoroapatite.

The Reaction Mechanism of High-Phosphate Iron Ore in Vacuum Reduction
Therefore, in order to further investigate the reasons for the low dephosphorization efficiency of fluoroapatite during VCTR, the reduction mechanism of fluoroapatite in the vacuum reduction process of high-phosphorus iron ore was studied.As shown in Table 3, Samples were prepared According to the aforementioned results, the conclusions could be drawn that a higher metallization rate (95%) was obtained at the optimum condition of VCTR (reduction temperature = 1100 • C, basicity = 1.1-1.3,and C/O = 0.8-1.0);however, this failed to achieve the purpose of a high dephosphorization rate.Therefore, the reasons for the low dephosphorization ratio should be discussed by investigating the reduction mechanism of fluoroapatite.

The Reaction Mechanism of High-Phosphate Iron Ore in Vacuum Reduction
Therefore, in order to further investigate the reasons for the low dephosphorization efficiency of fluoroapatite during VCTR, the reduction mechanism of fluoroapatite in the vacuum reduction process of high-phosphorus iron ore was studied.As shown in Table 3, Samples were prepared according to the optimal conditions (T = 1100 • C, t = 30 min).The samples were, respectively, reduced in vacuum and N 2 atmosphere for a different reduction time (2.5-30 min) under the same conditions.The macroscopic morphology changes of the reduced pellets at different times are, respectively, shown in Figures 8 and 9 under vacuum and nitrogen conditions.It could be found that the reduced pellets in the vacuum condition reacted more uniformly and tended to consolidate; while the reduced pellets under nitrogen condition expressed more cracks and non-uniformity.The results indicated that the reduction reaction of pellets in vacuum conditions was excellent.In general, the control of pellets in the reduction process indicated that reduction temperature was reasonable.Chemical analysis was performed on the reduced pellets after 30 min reaction under the two conditions, and the metallization ratio and dephosphorization ratio were calculated, as presented in Table 4.The macroscopic morphology changes of the reduced pellets at different times are, respectively, shown in Figures 8 and 9 under vacuum and nitrogen conditions.It could be found that the reduced pellets in the vacuum condition reacted more uniformly and tended to consolidate; while the reduced pellets under nitrogen condition expressed more cracks and non-uniformity.The results indicated that the reduction reaction of pellets in vacuum conditions was excellent.In general, the control of pellets in the reduction process indicated that reduction temperature was reasonable.Chemical analysis was performed on the reduced pellets after 30 min reaction under the two conditions, and the metallization ratio and dephosphorization ratio were calculated, as presented in Table 4.       Higher peak intensities of Fe and Fe 3 C obtained in the vacuum condition indicated that the reduction of FeO to Fe was more thorough.However, the peak of FeO could be found in the entire reduction process under the nitrogen condition, which illustrated that the metallization rate in nitrogen condition was lower.Figures 10 and 11 exhibit the appearance of the phosphorus-containing phase (Fe 2 P) in the spectrum; however, the peak intensity does not change significantly due to its low content.No other iron-phosphorus phase is observed in the two Figures, thus it was necessary to conduct further analysis on the phosphorus-containing phase.Furthermore, the spectra of the two cases always exhibited the peak of graphite, which indicated that the carbon content in the reduction reaction was sufficient.
The results summarized in Table 4 indicate a remarkable increase in the metallization ratio of vacuum-reduced pellets to 94.7%, while an undesirable decrease in the dephosphorization ratio from 13.5 to 6.2% compared to that for the nitrogen-reduced pellets.The reasons might be attributed to the following two aspects: (1) with the aggravation of the solution loss reaction of carbon (C + CO 2 = 2CO) in the vacuum condition, a great deal of fixed carbon was consumed to generate CO.In terms of iron oxide, the ability of CO to reduce Fe 2 O 3 was stronger than that of carbon, so the reduction reaction of iron oxide greatly intensified with the increase of CO content.In contrast, the ability of CO to reduce fluorapatite was weaker than that of carbon; therefore, the reduction reaction of fluorapatite was indirectly inhibited due to the large consumption of fixed carbon.(2) The influences of iron oxides and reduced iron on the reduction process of fluorapatite were not considered.The reduction reaction of fluorapatite with fixed carbon and gangue might no longer generate gaseous phosphide, but iron-phosphorus compounds, such as Fe 3 P, Fe 2 P, and FeP.
phase (Fe2P) in the spectrum; however, the peak intensity does not change significantly due to its low content.No other iron-phosphorus phase is observed in the two Figures, thus it was necessary to conduct further analysis on the phosphorus-containing phase.Furthermore, the spectra of the two cases always exhibited the peak of graphite, which indicated that the carbon content in the reduction reaction was sufficient.The results summarized in Table 4 indicate a remarkable increase in the metallization ratio of vacuum-reduced pellets to 94.7%, while an undesirable decrease in the dephosphorization ratio from 13.5 to 6.2% compared to that for the nitrogen-reduced pellets.The reasons might be attributed to the following two aspects: (1) with the aggravation of the solution loss reaction of carbon (C + CO2 = 2CO) in the vacuum condition, a great deal of fixed carbon was consumed to generate CO.In terms of iron oxide, the ability of CO to reduce Fe2O3 was stronger than that of carbon, so the reduction reaction of iron oxide greatly intensified with the increase of CO content.In contrast, the ability of CO to reduce fluorapatite was weaker than that of carbon; therefore, the reduction reaction of fluorapatite was indirectly inhibited due to the large consumption of fixed carbon.(2) The influences of iron oxides and reduced iron on the reduction process of fluorapatite were not considered.The reduction reaction For the second reason, the microscopic behavior of pellets in the process was observed by SEM-EDS. Figure 12 shows no change in the distribution of iron oxide before and after the vacuum reduction reaction.The reduced metallic iron, with small size between 50-80 µm was uniformly distributed in the ore phase.Nonetheless, under the nitrogen condition, the accumulation and growth of metallic iron was found and it eventually formed large clumps with particle sizes between 200 and 300 µm.This was mainly ascribed to the fact that metallic iron grew too late to agglomerate based on the faster reduction rate in vacuum condition.Under vacuum conditions, the fluoroapatite began to be reduced at the 10th minute, and as the reaction progressed, the phosphorus content in the reduced iron also gradually increased.When phosphorus migrated into the iron phase, the high-phosphorus iron phase combined with the low-phosphorus or non-phosphorus iron phase from one side and gradually diffused into the iron phase.The distribution of phosphorus in the iron phase was uniform until the reduction reaction reached 30 min.The fluoroapatite did not start to react until 12.5 min under the nitrogen condition and the migration form of phosphorus entering the iron was the same as under the vacuum condition.This indicated that not all of the reduced phosphorus formed gaseous phosphides (P 2 , P 4 , PO and PO 2 ) during the reduction of fluoroapatite.In contrast, most of the reduced phosphorus entered the metallic iron.The migrated form of phosphorus might diffuse from high-phosphorus iron phase, produced by the reduction process of fluoroapatite, into low-phosphorus iron phase.In order to determine the products in the reaction process, the local micro-areas of the ironphosphorus phase under the vacuum condition and the nitrogen condition were respectively analyzed as presented in Figure 13 and Table 5.Table 5 demonstrates an interesting result that the atomic ratio of iron and phosphorus was about 6-8:1.Therefore, this confirmed that iron phosphorus compounds existed in the form of Fe3P, Fe2P, and FeP.In order to determine the products in the reaction process, the local micro-areas of the iron-phosphorus phase under the vacuum condition and the nitrogen condition were respectively analyzed as presented in Figure 13 and Table 5.Table 5 demonstrates an interesting result that the atomic ratio of iron and phosphorus was about 6-8:1.Therefore, this confirmed that iron phosphorus compounds existed in the form of Fe 3 P, Fe 2 P, and FeP.Based on the aforementioned research results of the reduction process under different conditions, the reaction mechanism of high-phosphorus iron ore was discussed.Under vacuum condition, Fe2O3 was almost completely reduced to FeO after 5 min of reduction, and Fe began to be formed.Then, the stage of rapid formation of Fe occurred in 5-10 min, after which most of the element Fe transformed into the metal Fe.After 10 min, the fluorapatite began to be reduced.At this time, it was easy to ignore the effect of the large amount of reduced metallic iron on the reduction reaction.Some research [26][27][28] considered that iron did not participate in the reaction, and the fluoroapatite reacted only with the fixed carbon and gangue.The reduced phosphorus was volatilized in the form of gas phase such as P2 or PO.However, the test results had shown that the reduced phosphorus was not completely volatilized in the vapor phase, and most of the iron phosphorus compounds were formed.This indicated that a large amount of metal iron formed in 5-10 min, might have participated in the reduction reaction of the fluoroapatite, and the iron phosphorus compounds were directly formed.Therefore, a new reduction mechanism of fluoroapatite was proposed in this study, as shown in Equations ( 3)-( 5) [29][30], and thermodynamic analysis was carried out: 2Ca5(PO4)3F + 18Fe + 9SiO2 + 15C = 9CaSiO3 + 6Fe3P + 15CO(g) + CaF2 (3) 2Ca5(PO4)3F + 12Fe + 9SiO2 + 15C = 9CaSiO3 + 6Fe2P + 15CO(g) + CaF2 (4)  Based on the aforementioned research results of the reduction process under different conditions, the reaction mechanism of high-phosphorus iron ore was discussed.Under vacuum condition, Fe 2 O 3 was almost completely reduced to FeO after 5 min of reduction, and Fe began to be formed.Then, the stage of rapid formation of Fe occurred in 5-10 min, after which most of the element Fe transformed into the metal Fe.After 10 min, the fluorapatite began to be reduced.At this time, it was easy to ignore the effect of the large amount of reduced metallic iron on the reduction reaction.Some research [26][27][28] considered that iron did not participate in the reaction, and the fluoroapatite reacted only with the fixed carbon and gangue.The reduced phosphorus was volatilized in the form of gas phase such as P 2 or PO.However, the test results had shown that the reduced phosphorus was not completely volatilized in the vapor phase, and most of the iron phosphorus compounds were formed.This indicated that a large amount of metal iron formed in 5-10 min, might have participated in the reduction reaction of the fluoroapatite, and the iron phosphorus compounds were directly formed.Therefore, a new reduction mechanism of fluoroapatite was proposed in this study, as shown in Equations ( 3)-( 5) [29,30], and thermodynamic analysis was carried out: 2Ca 5 (PO 4 ) 3 F + 18Fe + 9SiO 2 + 15C = 9CaSiO 3 + 6Fe 3 P + 15CO(g) + CaF 2 (3) 2Ca 5 (PO 4 ) 3 F + 12Fe + 9SiO 2 + 15C = 9CaSiO 3 + 6Fe 2 P + 15CO(g) + CaF 2 (4) 2Ca 5 (PO 4 ) 3 F + 6Fe + 9SiO 2 + 15C = 9CaSiO 3 + 6FeP + 15CO(g) + CaF 2 (5) In these equations, SiO 2 was represented as gangue mineral.According to the type and content of gangue minerals, the products were also different.
Table 6 and Figure 14 show that all reactions could be performed, except for the reaction (5) at 10 5 Pa, in the pressure range of 10 0 -10 5 Pa.This indicated that metallic iron could promote the reduction of fluorapatite, and the effect of metallic iron on the reaction could not be ignored.In order to intuitively show the reduction process of high-phosphorus iron ore, the reduction route is represented in Figure 15 and the schematic illustration of the fluorapatite reduction process is shown in Figure 16.In these equations, SiO2 was represented as gangue mineral.According to the type and content of gangue minerals, the products were also different.
Table 6 and Figure 14 show that all reactions could be performed, except for the reaction (5) at 10 5 Pa, in the pressure range of 10 0 -10 5 Pa.This indicated that metallic iron could promote the reduction of fluorapatite, and the effect of metallic iron on the reaction could not be ignored.In order to intuitively show the reduction process of high-phosphorus iron ore, the reduction route is represented in Figure 15 and the schematic illustration of the fluorapatite reduction process is shown in Figure 16.      Figure 16 demonstrates the completion of the reduction process of iron oxides before the beginning of the reduction of fluorapatite.Fluorapatite, reduced iron, and gangue contacted and reacted with each other with the diffusion of solid phase.Compared to that under nitrogen condition, the particle size of reduced iron in a vacuum was smaller and well distributed.The contact area with fluoroapatite was much larger.Therefore, the vacuum condition with a bad effect on generating gaseous phosphide was more favorable for the formation of FexP.The generated FexP phase attached to the edge of the iron phase and gradually gathered with the aggregation of the iron phase.Moreover, the phosphorus diffused from the high-phosphorus region to the low-phosphorus region during the process of accumulation, and the higher the reduction temperature, the more uniform the distribution of phosphorus.

Conclusions
1.The effects of reduction temperature, basicity, and C/O ratio, on the metallization ratio, dephosphorization ratio, and phosphorus content of pellets were studied under the vacuum carbon thermal reduction (VCTR) condition.It was found that the metallization ratio of pellets reached maximum (95%); however, the dephosphorization ratio of only 5.6% showed an adverse result, when the reduction temperature, C/O ratio, and basicity were 1100 °C, 0.8-1.0, and 1.3, respectively.2. The vacuum condition played an insignificant role in evaluating the dephosphorization ratio, resulting in a higher phosphorus content of iron bead in the reduction process.This was attributed to the fact that during the vacuum reduction process fluorapatite reacted not only with carbon to form gaseous phosphide, but also with iron to form the FexP phase.3. Compared to the nitrogen condition, the metallization ratio of pellets was significantly improved, while the dephosphorization ratio of pellets showed an opposite result.Figure 16 demonstrates the completion of the reduction process of iron oxides before the beginning of the reduction of fluorapatite.Fluorapatite, reduced iron, and gangue contacted and reacted with each other with the diffusion of solid phase.Compared to that under nitrogen condition, the particle size of reduced iron in a vacuum was smaller and well distributed.The contact area with fluoroapatite was much larger.Therefore, the vacuum condition with a bad effect on generating gaseous phosphide was more favorable for the formation of Fe x P. The generated Fe x P phase attached to the edge of the iron phase and gradually gathered with the aggregation of the iron phase.Moreover, the phosphorus diffused from the high-phosphorus region to the low-phosphorus region during the process of accumulation, and the higher the reduction temperature, the more uniform the distribution of phosphorus.

1.
The effects of reduction temperature, basicity, and C/O ratio, on the metallization ratio, dephosphorization ratio, and phosphorus content of pellets were studied under the vacuum carbon thermal reduction (VCTR) condition.It was found that the metallization ratio of pellets reached maximum (95%); however, the dephosphorization ratio of only 5.6% showed an adverse result, when the reduction temperature, C/O ratio, and basicity were 1100 • C, 0.8-1.0, and 1.3, respectively.2.
The vacuum condition played an insignificant role in evaluating the dephosphorization ratio, resulting in a higher phosphorus content of iron bead in the reduction process.This was attributed to the fact that during the vacuum reduction process fluorapatite reacted not only with carbon to form gaseous phosphide, but also with iron to form the Fe x P phase.

3.
Compared to the nitrogen condition, the metallization ratio of pellets was significantly improved, while the dephosphorization ratio of pellets showed an opposite result.4.

5.
The expected result that a vacuum condition promotes dephosphorization of fluorapatite during the reduction stage was not achieved, and phosphorus was not reduced to gas phase, but to iron-phosphorus phase.Based on the new mechanism of fluorapatite reduction, the reduction of fluorapatite which was difficult to separate from the iron phase should be inhibited.Therefore, a new idea was proposed: the reduction temperature of high-phosphorus ore should be less than the temperature of the rapid reduction of fluoroapatite.Decreasing the reduction time and carbon addition would promote the iron increase and phosphorus decrease.Less reducing agent after the reduction of iron oxide should be more beneficial to the inhibition of fluorapatite reduction due to the reduction of iron oxide prior to the reduction of fluoroapatite.These systematic explorations will be carried out in our next study.

Figure 1 .
Figure 1.The schematic illustration of a vacuum tube furnace.

Figure 1 .
Figure 1.The schematic illustration of a vacuum tube furnace.

Figure 2 .
Figure 2. The effects of reduction temperature on the metallization ratio, dephosphorization ratio, and phosphorus content of iron bead.

Figure 2 .
Figure 2. The effects of reduction temperature on the metallization ratio, dephosphorization ratio, and phosphorus content of iron bead.

Figure 4 .
Figure 4.The effects of the C/O ratio on the metallization ratio, dephosphorization ratio, and phosphorus content of iron bead.

Figure 5 .
Figure 5. XRD spectra of reduced pellets at different carbon contents.

Figure 4 .Figure 4 .
Figure 4.The effects of the C/O ratio on the metallization ratio, dephosphorization ratio, and phosphorus content of iron bead.

Figure 5 .
Figure 5. XRD spectra of reduced pellets at different carbon contents.

Figure 5 .
Figure 5. XRD spectra of reduced pellets at different carbon contents.

Figure 6 .
Figure 6.The effects of basicity on the metallization ratio, dephosphorization ratio, and phosphorus content of pellets.

Figure 7 .
Figure 7. XRD spectra of reduced pellets at different basicity.

Figure 6 .Figure 6 .
Figure 6.The effects of basicity on the metallization ratio, dephosphorization ratio, and phosphorus content of pellets.

Figure 7 .
Figure 7. XRD spectra of reduced pellets at different basicity.

Figure 7 .
Figure 7. XRD spectra of reduced pellets at different basicity.

Figure 8 .
Figure 8.The macroscopic morphology changes of the reduced pellets at different times under vacuum condition.

Figure 8 .
Figure 8.The macroscopic morphology changes of the reduced pellets at different times under vacuum condition.

Figure 8 .Figure 9 .
Figure 8.The macroscopic morphology changes of the reduced pellets at different times under vacuum condition.

Figure 9 .
Figure 9.The macroscopic morphology changes of the pellets reduction process at different times under nitrogen condition.

Figures 10 and 11
Figures 10 and 11  show the XRD spectra of the reduced pellets as a function of reduction time under two conditions, respectively.Comparative analysis of the XRD patterns under the two conditions clearly indicates that the reduction reaction of Fe 2 O 3 to FeO was basically completed in the first 5 min.Higher peak intensities of Fe and Fe 3 C obtained in the vacuum condition indicated that the reduction of FeO to Fe was more thorough.However, the peak of FeO could be found in the entire reduction process under the nitrogen condition, which illustrated that the metallization rate in nitrogen condition was lower.Figures10 and 11exhibit the appearance of the phosphorus-containing phase (Fe 2 P) in the spectrum; however, the peak intensity does not change significantly due to its low content.No other iron-phosphorus phase is observed in the two Figures, thus it was necessary to conduct further analysis on the phosphorus-containing phase.Furthermore, the spectra of the two cases always exhibited the peak of graphite, which indicated that the carbon content in the reduction reaction was sufficient.The results summarized in Table4indicate a remarkable increase in the metallization ratio of vacuum-reduced pellets to 94.7%, while an undesirable decrease in the dephosphorization ratio from 13.5 to 6.2% compared to that for the nitrogen-reduced pellets.The reasons might be attributed to the following two aspects: (1) with the aggravation of the solution loss reaction of carbon (C + CO 2 = 2CO) in the vacuum condition, a great deal of fixed carbon was consumed to generate CO.In terms of iron oxide, the ability of CO to reduce Fe 2 O 3 was stronger than that of carbon, so the reduction reaction of iron oxide greatly intensified with the increase of CO content.In contrast, the ability of CO to reduce fluorapatite was weaker than that of carbon; therefore, the reduction reaction of fluorapatite was indirectly inhibited due to the large consumption of fixed carbon.(2) The influences of iron oxides and reduced iron on the reduction process of fluorapatite were not considered.The reduction reaction of fluorapatite with fixed carbon and gangue might no longer generate gaseous phosphide, but iron-phosphorus compounds, such as Fe 3 P, Fe 2 P, and FeP.

Figure 10 .
Figure 10.XRD spectra of reduced pellets at different times under nitrogen condition.

Figure 10 .Figure 11 .
Figure 10.XRD spectra of reduced pellets at different times under nitrogen condition.

Figure 11 .
Figure 11.XRD spectra of reduced pellets at different times under vacuum condition.

Figure 12 .
Figure 12.Microscopic changes during the reduction process of high-phosphorus iron ore.

Figure 12 .
Figure 12.Microscopic changes during the reduction process of high-phosphorus iron ore.

Figure 13 .
Figure 13.Images of the location of the iron-phosphorus phase.

Figure 13 .
Figure 13.Images of the location of the iron-phosphorus phase.

Figure 14 .
Figure 14.Relationships between the theoretical minimum temperature at which the reaction could occur and pressure.

Figure 14 .Figure 14 .
Figure 14.Relationships between the theoretical minimum temperature at which the reaction could occur and pressure.

Figure 15 .
Figure 15.The schematic illustration of the reduction route of high-phosphorus iron ore.Figure 15.The schematic illustration of the reduction route of high-phosphorus iron ore.

Figure 15 .
Figure 15.The schematic illustration of the reduction route of high-phosphorus iron ore.Figure 15.The schematic illustration of the reduction route of high-phosphorus iron ore.

Figure 16 .
Figure 16.Diagram of the fluorapatite reduction process.

Figure 16 .
Figure 16.Diagram of the fluorapatite reduction process.

Table 1 .
Chemical composition of high-phosphorus iron ore.

Table 2 .
The material mass ratio of pure substance simulated high-phosphorus iron ore (mass ratio and mass %).

Table 2 .
The material mass ratio of pure substance simulated high-phosphorus iron ore (mass ratio and mass %).

Table 3 .
The composition of simulated high-phosphorus iron ore.

Table 3 .
The composition of simulated high-phosphorus iron ore.

Table 4 .
The metallization ratio and dephosphorization ratio of pellets.

Table 5 .
Energy dispersive spectrometry (EDS) analysis data of elements in iron-phosphorus phase.

Table 5 .
Energy dispersive spectrometry (EDS) analysis data of elements in iron-phosphorus phase.

Table 6 .
The theoretical minimum temperature (K) for each reaction under different pressure.

Table 6 .
The theoretical minimum temperature (K) for each reaction under different pressure.