Machine learning for predicting silica purity based on the chemical composition and granulometry of sand in the Kyrgyz Republic
Keywords:
metallurgical-grade silicon, quartz sand, machine learning, modeling, granulometry, impurities, predictionAbstract
This article presents the results of a study on the influence of chemical composition and granulometric characteristics of quartz raw materials from various deposits in Kyrgyzstan (Tash-Kumyr, Sulyukta, Ozgur) on the purity of metallurgical-grade silicon, using methods of mathematical modeling and elements of machine learning. Elemental analysis revealed significant differences between the deposits: the SiO₂ content in Tash-Kumyr quartz reaches ~94.3 %, while in Sulyukta it does not exceed 85 %, accompanied by a considerably higher level of Al₂O₃ and other impurities. The developed model combines the linear effects of oxide impurities (Al₂O₃, Fe₂O₃, MgO, CaO, K₂O) and a quadratic dependence on particle size, with weight coefficients calibrated using machine learning algorithms. Modeling confirmed that the optimal particle size (0.05 mm) ensures maximum purity, while deviations reduce it by 2–5 %. Predictive curves highlighted the advantage of Tash-Kumyr raw material (up to 98 % purity). The results demonstrate the potential of integrating machine learning to improve prediction accuracy and provide a rational basis for selecting promising deposits.
References
Куропаткин, В. Н. Технологические аспекты производства кремния из кварцевого сырья // Химическая промышленность. – 2008. – № 6. – С. 12–17.
Рогов, А. И., & Иванова, Л. П. Влияние примесей в кварцевом сырье на качество получаемого кремния // Металлургия цветных металлов. – 2012. – № 3. – С. 56–62.
Садыков, К. С. Технология получения технического кремния из кварцевого сырья Кыргызстана // Горный вестник Кыргызстана. – 2015. – № 2. – С. 45–51.
Абдраимов, Ж. Т., Касымбеков, Б. Ж., & Жунушалиева, А. С. Очистка кварцевого сырья для получения кремния высокой чистоты // Вестник Кыргызско-Российского Славянского университета. – 2023. – Т. 23, № 4. – С. 88–94.
Liu, P., Chen, X., & Zhao, F. Influence of particle size and impurities on silicon yield during carbothermal reduction // Ceramics International. – 2019. – Vol. 45, No. 8. – P. 11023–11031. DOI:10.1016/j.ceramint.2019.02.110.
Zhang, Q., Liu, J., Wang, H., & Xu, Y. High-purity quartz: a review of current research and future prospects // Journal of Materials Research and Technology. – 2021. – Vol. 12. – P. 1251–1264. DOI:10.1016/j.jmrt.2020.12.087.
Tuncuk, A., Akcil, A., Yigit, S., & Deveci, H. Iron removal in production of purified quartz by hydrometallurgical methods // Hydrometallurgy. – 2016. – Vol. 161. – P. 1–9. DOI: 10.1016/j.hydromet.2016.01.006.
Kim, J., Park, S., & Choi, Y. Machine learning-based prediction of silicon purity from quartz raw material // Applied Sciences. – 2020. – Vol. 10, No. 22. – Article 8086. DOI: 10.3390/app10228086.
Long, H., Li, W., Sun, J., & others. Advanced processing techniques and impurity management for high-purity quartz in diverse industrial applications // Minerals. – 2024. – Vol. 14, No. 6. – Article 571. DOI: 10.3390/min14060571.
Recent advances in the marketing, impurity characterization and purification of quartz //Physicochemical Problems of Mineral Processing. – 2023. – Vol. 59, No. 1. – P. 56–72. DOI: 10.37190/ppmp/158078.
Омурбекова Г. К., Адылова Э. С., Жапаркулов А. М. Исследование фрактальной размерности частиц кварцевых песков Таш-Кумырского и Озгурского месторождений //Известия национальной академии наук Кыргызской Республики. – 2022. – №. 5. – С. 16-22.
