Effect of Silica and Carbon-reducing Agents on Ni and Ti iImpurities During Silicon Production


 The Ni and Ti contents of industrial silicon has a significantly affect the of organic silicon. On the basis of large specific production data, the chemical component of silica and the carbon-reducing agent effect of the Ni and Ti contents of silicon were investigated using statistical techniques. Two furnaces were also studied—an 8.5 MVA furnace and a 12.5 MVA furnace. The effects of TiO2 and NiO impurities on the power consumption of both furnaces were also evaluated using the correlation of the TiO2 and NiO impurities in raw materials with specific power consumption. The consumption of raw materials exhibited a high negative correlation with the TiO2 and NiO impurities in industrial silicon, as determined by linear regression—that is, 82% < |r| < 99%. The influence of Ni on industrial silicon production was also stronger than that of Ti. With an increase in Ni, the power consumption of the 8.5 MWA furnace significantly decreased, whereas that of the 12.5 MWA furnace is increased. Adjusting the content of Ni content can reduce the power consumption of industrial silicon production in the large furnace.


Introduction
Organic silicon, the pillar of strategic emerging industries, exhibits superior performance and is widely used in construction [1,2], electronics [3][4][5], transportation [6], energy [7][8][9], and household products [10]. It is typically structured with the Si-C bond (or the Si-O, Si-S, Si-N bond). Most organic silicon compounds are organic, polymers with -Si-O-Si-as the main chain. Organic silicon products chlorosilanes in the upstream products for organic silicon monomer and then used as raw material for processing production of silicone oil, silicone rubber, silicone resin, silane coupling agent four categories of downstream products. Among these products, the organic silicon monomer is mainly crushed and ground into silicon powder via silicon blocks, silicon powder and methyl chloride are directly synthesized into methyl chlorosilane (MCS) under the action of copper-based catalyst [11][12][13][14]. A diagram of the organic silicon industry chain is presented in Fig. 1. However, the reaction during production is more complex, and the quality of silica powder directly affects the synthesis reaction. Gillot [15] examined the reaction of CuCl with silicon containing impurities, such as Al, Fe, Ca, and Ti. They found that the presence of Ti increased the rate of copper formation during CuCl reduction by silicide. However, Tamhankar [16] showed that the presence of large amounts of free copper (not bonded to silicon) hinders the main reaction, leading to the formation of highly chlorinated silanes. Gillot [17] also found that Al promoted the reaction between Cu 3 Si and CuCl, Fe reduced the consumption rate of Cu 3 Si, and the combined action of the two impurities resulted in the formation of more Cu-Si alloys. MCS was synthesized into a siloxane mixture (DMC + D4), which was then polymerized into polysiloxane. Polysiloxane is the main raw material of the downstream products of organic silicon, and its quality directly affects the quality of the downstream products.
Therefore, the analysis of impurities in silicon powder plays a key role in the synthesis of organic silicon monomers and the quality of downstream products.
Silicon is produced by reacting carbonaceous material and silica at high temperatures in an electric arc furnace but with the production of different impurities [18,19]. Owing to the di culty of removing some metal oxides, the calculation of the standard Gibbs free energy of the oxidizing impurity reaction in silicon shows that TiO 2 does not undergo reduction reaction at temperatures in the 0-2000 ºC range but can form a metallic phase (FeSi 2 Ti) with silicon, an inert phase. Studies have shown that Ti can accumulate in the reactor in the form of slag during MCS production. NiO is partly reduced at about 899 ºC; however, during MCS synthesis, Ni can prioritize the formation of methyl hydrodichlorosilane (monomethyl hydrogen).
The iron-philic Ni easily forms a precipitated state with silicon. Ni in large amounts inhibits the production reaction of organic silicone and thus needs to be strictly controlled. Related research shows that impurities in silicon can be classi ed as bene cial and harmful with each part needs to control a certain amount of impurities, depending on the characteristics of organic silicon production process. The Non-ferrous Metal Industry Standard of the People's Republic of China (YS/T 1109-2016) limits the contents of major impurities in silicon powder for organic silicon; for instance, Ti content is not to exceed 0.5%, Ni content is limited to 0.015%, and silicon content has to be in the 99.184-99.724% range. Further, the national standard for industrial silicon (GB/T 2881 − 2014) for the content of trace elements of silicon used in organic silicon and silicon for organic silicon belongs to chemical silicon. The high-precision grade contains Ni ≤ 100×10 − 6 and Ti ≤ 400×10 − 6 , and the general-precision grade contains Ni ≤ 150×10 − 6 and Ti ≤ 500×10 − 6 .
Several studies have been conducted on the in uence of the quality of organic silicon which has been identi ed as a direct factor [20,21]. However, the content and source of trace elements in silicon (such as Ti and Ni) have yet to be clari ed, and the presence of a linear relationship between the two has yet to be determined. These inadequacies hinder the repeatability of the reaction and affect product yield, impeding development of organic silicon. Evaluation and analytical techniques concerning industrial silicon have been developed [22][23][24][25][26][27]. We previously used linear regression to determine the effects of coal, petroleum coke, and silica in different furnaces on the amount of major impurities in industrial silicon [28]. Numerous studies have been performed to measure the effects of major impurities (Fe, Al, and Ca) in silicon production on raw materials and energy; by contrast, no comprehensive research has been reported regarding the effects of other minor impurities (Ti, Ni, K, etc.) on energy consumption and raw materials.
Therefore, from the perspective of raw material and power consumption, the sources of Ti and Ni oxide impurities and their effect on each other are evaluated to provide a theoretical basis for obtaining silicon powders with desirable properties. In the current study, we investigate the relationship between the silicon produced in two submerged arc furnaces and the Ti and Ni oxide impurities in the feedstock to determine how the relationship between the two oxides affects energy and feedstock consumption in silicon production.

Raw materials
Petroleum coke and coal, which were used as reducing agents in the production of industrial silicon (Si > 99%), were sourced from Taiwan and Shanxi. The properties of each raw material are listed in Table 1 and Table 2. The TiO 2 content in coal was 0.025%, and the NiO content in petroleum coke was 0.033%. The electrode material is considered to be 100% carbon xed, but its effect on silicon production is temporarily being ignored.   Figure 2 shows the consumption of different carbon materials and the electric energy required by the 8.5 MVA and 12.5 MVA furnaces. As shown in the gure, the power consumption of the 12.5 MVA furnace uctuates but mostly stays between 1200 and 1400 kwh, whereas that of the 8.5 MVA furnaces is between 1200 and 1370 kwh. The coal consumption of the 12.5 MVA furnace is consistently higher than that of petroleum coke, whereas that of the 8.5 MVA furnace is the opposite.
2.2 Effect of the carbon-reducing agent on TiO 2 /NiO impurity balance in industrial silicon production process Impurities during smelting of industrial silicon mainly originate from raw materials. Figure  impurities in industrial silicon were 66% and 34%, respectively. In general, variations in the ratios of raw materials affect the content of impurities, and the large furnace was bene cial to silicon production. Figure 2 shows that a certain amount of the TiO 2 and NiO impurities attributed to raw materials during the smelting of industrial silicon affects the smelting process of industrial silicon. To explore the sources of impurities in raw materials and ensure that the quality of industrial silicon products could meet the standards of organic silicon production, the sources of different contents of TiO 2 and NiO impurities in different furnace models were analyzed. Figure 4 shows that the sources of TiO 2 and NiO oxides in petroleum coke, coal, and silica during smelting of industrial silicon. In the 8.5 MVA furnace, the content of TiO 2 impurities in the raw materials, ranked in descending order, was as follows: coal > petroleum coke > silica.
The impurities in coal uctuated between 25% and 70%. Petroleum coke varied between 30% and 45%, and silica was about 10%; NiO in petroleum coke differed between 25% and 50%, and coal uctuated between 40% and 70%. In the 12.5 MVA furnace, the content of TiO 2 impurities, ranked in descending order, was as follows: petroleum coke > coal > silica. The TiO 2 impurities in petroleum coke varied between 41% and 75%, and that in coal uctuated ranged between 25% and 35%. NiO in petroleum coke uctuated from 41-67%, and coal varied between 20% and 35%. Therefore, the results provide a reference for adjusting and controlling the contents of impurities in raw materials to address quality problems of industrial silicon products, improve the quality of silicon powder, and effectively reduce the cost.

Data collection
During analysis, continuous production data for the 8.5 MVA furnace and the 12.5 MVA furnace for more than three months were collected, and abnormal data caused by power outages and failures were deleted. All units of data are one ton.

Calculation method
To determine the correlation between raw material consumption and impurity content in industrial silicon, the Pearson correlation coe cient [29] was adopted, expressed in Formula (1). r is the linear correlation coe cient, between − 1 and + 1. When |r| ≥ 0.80 is highly relevant, the following relations apply: 0.5 ≤ |r| < 0.8 for moderate correlation, 0.3 ≤ |r| < 0.5 for low correlation, and |r| < 0.3 considerably weak correlation.

Results And Discussion
3.1 In uence of raw materials on the TiO 2 and NiO total impurity content of silicon shows the absolute value of the slope of the raw material in the small furnace was greater than that of the raw material in the large furnace. This difference suggests that variations in furnace type in uence the production of industrial silicon, but the in uence of impurities in the large furnace is reduced.

In uence of raw materials on the content of TiO 2 impurities in silicon
To understand the correlation between the single impurity in silicon products and the impurities in raw materials, as well as provide theoretical guidance for the effect of the content of TiO 2 and NiO impurities in raw materials on the production process of industrial silicon, the linear tting between TiO 2 and the raw material is presented in Fig. 7. In the 8. moreover, the order of the degree of in uence on silicon impurities was coal > petroleum coke > silica. The TiO 2 impurities were primarily ascribed to coal. The absolute values of slope in the 12.5 MVA furnace were 0.09, 0.13, and 0.04, respectively; the order of the degree of in uence on silicon impurities was petroleum coke > coal > silica. The TiO 2 impurities were mainly attributed to petroleum coke.

In uence of raw materials on the content of NiO impurities in silicon
The linear tting between NiO and the raw material is depicted in Fig. 8

Linear tting of the effect of TiO 2 -NiO interaction in silicon
To understand the interaction between TiO 2 and NiO impurities in industrial silicon, a linear tting diagram of TiO 2 and NiO impurities in the two furnace types of industrial silicon is presented in Fig. 9. In the 8.5 MVA furnace, the two exhibited a low positive correlation, indicating that speci c TiO 2 does not considerably change with an alteration in NiO consumption, but a certain trend of change still exists. With an increase in NiO content, the TiO 2 content also increased; the r value was determined to be 0.28113, and the slope was 0.33. In the 12.5 MVA furnace, the two were positively correlated, with an r of 0.23751. This value is signi cantly lower than that for the 8.5 MVA furnace with a slope of 0.45. The slopes of the two furnaces were compared; that of the 12.5 MVA furnace was much higher than that of the 8.5 MVA furnace, indicating that the NiO content in the larger furnace signi cantly in uenced the production process. Therefore, regulating the NiO content can reduce the effect on industrial silicon production.
3.5 In uence of variations in the combination of reductants on power consumption in industrial silicon production 3.5.1 Effect of TiO 2 and NiO metal oxides in raw materials on power consumption in industrial silicon production To elucidate the in uence of TiO 2 and NiO impurities on the production of industrial silicon, the linear relationship between the two metal oxides and power consumption is depicted in Fig. 10. The gure reveals a moderate positive correlation between metal oxides and power consumption; the power consumption increases with an increase in the content of impurities. In the 8.5 MVA furnace, the r of NiO was 0.58473, the slope was 24546; meanwhile, the r of TiO 2 was 0.53939, and the slope was 16732. For the two linear equations, when the same impurity ∆x 1 was consumed, the effective growth of energy consumption ∆y 1 corresponding to NiO was 874 kwh, and that corresponding to TiO 2 was 584 kwh. This result indicates that the NiO impurity substantially affected power consumption. In the 12.5 MVA furnace, the r of NiO was 0.62033, with a slope of 31989, and the r of TiO 2 was 0.69029, with a slope of 16014. When the two linear equations consumed the same impurity ∆x 2 , the effective growth of energy consumption ∆y 2 corresponding to NiO was 1893 kwh, and that corresponding to TiO 2 was 1010 kwh. A comparison of the two furnace types revealed that the r of TiO 2 in the large furnace was larger but the slope was smaller than that in the small furnace, indicating that the in uence of TiO 2 in the large furnace was smaller than that in the smaller furnace; meanwhile, the r and slope of NiO in the large furnace were signi cantly larger, indicating that the NiO impurity signi cantly affected the power consumption of the large furnace.
3.5.2 In uence of the interaction between two impurity oxides on power consumption in the 8.5 MVA furnace The in uence of the TiO 2 -NiO interaction on power consumption was assessed by contour analysis, as shown in Fig. 11.
The gure on the left is a three-dimensional stereogram, and the gure on the right is a two-dimensional contour map. As shown in the gure illustrating the 8.5 MVA furnace, with an increase in NiO impurity content, power consumption decreases gradually, and particularly when NiO > is 0.14 t, power consumption is signi cantly reduced. In the gure presenting the 12.5 MVA furnace, power consumption gradually increases with an increase in NiO content. Power consumption was at its highest provided that TiO 2 ≥ 0.2943 t and 0.1885 ≤ TiO 2 ≤ 0.3361 t. In summary, with an increase in Ni, the effect on power consumption in the small furnace decreased, whereas that on power consumption in the large furnace increased. Therefore, the production cost can be reduced to a large extent by adjusting the in uence of reduction in Ni content on the production in the large furnace.

Conclusion
The effects of three raw materials in 8.5 MVA and 12.5 MVA ore furnaces on the Ti and Ni contents in industrial silicon were evaluated based on large-scale industrial data. Results indicate that the TiO 2 and NiO impurities and the raw material exhibited a high negative correlation-82% < |r| < 99%; the percentage of impurities decreased with an increase in raw material. The percentage of a single-element impurity exhibited a trend similar to that of the total impurities. The correlation coe cient |r| of the TiO 2 in the 8.5 MVA furnace and coal was the largest-that is, 81.4%; the correlation coe cient |r| of petroleum coke and NiO reached 84.5%. TiO 2 and petroleum coke in the 12.5 MVA furnace showed a correlation coe cient |r| of 74.0%; NiO and coal had a correlation coe cient |r| of 66.7%. The linear t between TiO 2 and NiO in the 12.5 MVA furnace had a slope of 0.45, which was higher than the slope obtained in the 8 MVA furnace (0.33). Thus, compared with the TiO 2 content, the NiO content more strongly in uenced industrial silicon production. Moreover, with an increase in NiO content, the power consumption in the small furnace decreased, whereas that in the large furnace increased. The Ti impurities exerted less in uence on silicon production, compared with the Ni impurities. Thus, the quality of industrial silicon could be improved by adjusting the Ni content, and the production cost could be reduced. Further, the research results bear signi cance for improving the quality of organic silicon and the manufacture of the downstream products of organic silicon.
Declarations Figure 1 Organic silicon industry chain map Linear tting relationship between TiO2/NiO impurity in silicon and raw material Figure 6 TiO2/NiO impurities in silicon products and total impurities in three raw materials of percent  TiO2 impurities in silicon products and total TiO2/NiO impurities in three raw materials of percent Interaction effect between TiO2 and NiO impurities in silicon  Effect of interaction between TiO2 and NiO impurity oxides on power consumption