During the industrial silicon refining process, impurities can affect the selectivity and activity of organic silicon monomer synthesis. This study utilized the simplified molecular interaction volume model (MIVM) alongside actual production data and samples to investigate the effects of different aluminum concentrations on typical impurity phases in industrial silicon. et al. concentrations of 1500-1600 ppmw, the Si7Al8Fe5 phase emerges. As the Al concentration increases to 1700 ppmw, the Si2Al3Fe phase forms. The FeSi2, Si8Al6Fe4Ca, and FeTiSi2 phases are consistently present in industrial silicon. MIVM predictions indicate that the activities of Fe, Ti, and Ca increase with increasing Al concentration, consistent with the actual production trend. However, production data suggest that this correlation is limited under certain conditions. Moreover, MIVM is used to predict the interactions among impurities in industrial silicon. The effects of Al concentration on typical impurity phases in industrial silicon are elucidated using MIVM combined with actual production data. Proper control of Al concentration facilitates the efficient synthesis of organosilicon monomers, increases their yield, and reduces energy consumption. These findings provide a theoretical and technical basis for controlling impurities in organosilicon monomer synthesis.

Metallurgical batching—governing raw material proportioning across sintering, blast furnace ironmaking, converter steelmaking, and non-ferrous smelting—critically determines product quality, energy consumption, and production cost throughout the full process chain. Its inherent complexity, characterized by strong nonlinear physicochemical coupling, measurement delays of up to 1.5 h, and multi-source raw material disturbances, renders conventional linear programming and empirical methods inadequate for dynamic, multi-objective industrial environments. This review systematically examines 98 representative studies (2020–2026) on intelligent algorithms applied to metallurgical batching optimization. A two-dimensional analysis framework of the fusion algorithm function and metallurgical scene is established. All kinds of methods are divided into three categories: prediction-oriented, optimization-oriented and decision-oriented, covering four typical scenes of sintering burdening, blast furnace ironmaking, converter steelmaking and non-ferrous metal smelting. Traditional machine learning models achieve sintering burn-through point prediction with R2 ≈ 0.85 and offer superior interpretability via SHAP analysis. Deep learning architectures deliver blast furnace silicon content prediction with RMSE ≈ 0.04%, while multi-objective evolutionary algorithms provide mature Pareto optimization for batching cost and carbon objectives. Reinforcement learning holds long-term potential for closed-loop adaptive control but remains constrained by Sim-to-Real safety barriers. Converter steelmaking and non-ferrous smelting are identified as underexplored domains. Three priority directions are proposed: domain-adaptive predictive modeling for cross-plant generalization, real-time re-optimization embedding mechanism constraints, and safe reinforcement learning transfer via high-fidelity digital twins.

With the increasing number of retired photovoltaic (PV) modules, their efficient recycling has emerged as a vital research focus. In this study, a green CuCl2-ChCl solvent system was developed to recover silver (Ag) from retired photovoltaic modules. Ultrasonic intensification combined with mechanical stirring resulted in a high Ag leaching rate of 98.81 %. Kinetic analysis revealed that the Ag leaching process in the solvent system was controlled by a chemical reaction mechanism, with an apparent activation energy of 42.19 kJ/mol. An experimental model was established and optimized using response surface methodology to determine the optimal process parameters. The morphology and elemental distribution of samples before and after leaching were analyzed via scanning electron microscopy and energy-dispersive X-ray spectroscopy. AgCl was formed using HCl and NaCl as chlorine sources and then reduced to metallic Ag by ascorbic acid. The reduction process yielded Ag powder with a recovery rate of 98.57 %. This study provides a sustainable and efficient approach for recycling Ag from retired PV modules.

Industrial silicon is a critical raw material for the photovoltaic and electronic information industries. However, current secondary refining processes still face challenges in the effective removal of Ti impurities. During pure-oxygen refining, the Ti content increases and exhibits a strong correlation with Al and Ca removal. This study systematically investigates the effects of Al and Ca on Ti evolution during O2-H2O(g) refining through industrial production analysis data, laboratory experiments, electron probe microanalysis, and first-principles calculations. Ti removal efficiency is negatively correlated with the initial Al and Ca contents. For raw materials with high Al and Ca contents, Ti removal efficiency is below 15% but exceeds 67% for raw materials with low Al and Ca contents. Microstructural analysis reveals that under high- Al and high-Ca conditions, Ti forms stable tau 5-(Fe,Ti)Si2 intermetallic compounds, which hinder its removal. Under low-Al and low-Ca conditions, the independent precipitation of Ti is suppressed, which facilitates its oxidation removal. First-principles density functional theory (DFT) calculations further indicate that Al and Ca inhibit Ti hydroxylation through electronicstructure modulation and competitive OH adsorption, respectively. This study establishes a cross-scale correlation between raw-material composition and atomic-scale interactions and providing theoretical guidance for targeted Ti removal during industrial silicon refining.

Ti-bearing blast furnace slag (TBFS) serves as a valuable secondary Ti resource, and achieving efficient Ti recovery from it is of great significance for enhancing resource utilization and environmental protection. The silicothermic reduction process, using diamond wire saw silicon waste (DWSSW) as a reductant, has emerged as a simple and green route for Ti recovery from TBFS. However, the underlying reaction mechanisms and kinetics of this process have not yet been clearly elucidated, which has severely restricted its industrial application. In this study, the systematic experiments were conducted on the reaction process and kinetic behavior to address these issues. The results indicated that the addition of calcium oxide (CaO) significantly reduces the slag viscosity and increases slag surface tension, thereby effectively enhancing the Ti recovery rate, which reached 90.64 % under optimal conditions. Through systematic analysis of the reaction interface, the reduction mechanism between DWSSW and TBFS was further elucidated. Kinetic studies revealed that the reduction of titanium dioxide (TiO2) by Si follows first-order reaction characteristics, with the diffusion of silicon dioxide (SiO2) in the slag phase being the rate-limiting step. Based on these findings, the established kinetic model shows strong consistency with experimental data (R2 = 0.97), confirming its reliability in describing the reduction process. The research results were expected to not only advance the technological development of solid waste utilization from TBFS and DWSSW, but also provide new insights for the green transformation of industries such as photovoltaics and steelmaking.

The production of high-purity silicon for advanced applications is critically dependent on the removal of boron, a key impurity notoriously difficult to eliminate. While slag refining is a promising industrial route, its optimization is hindered by complex, multi-parameter interactions that challenge traditional experimental methods. This study employs a machine learning (ML) framework to overcome these limitations. A comprehensive database of 4000 entries was constructed by integrating literature data and thermodynamic calculations. Among six ML algorithms evaluated, an optimized XGBoost model achieved superior predictive accuracy for boron removal efficiency (R2 > 0.97, MSE < 0.0003). SHAP analysis identified temperature and the content of the third component (e.g., CaCl₂, CaF₂) as the most influential process parameters. Guided by these insights, an orthogonal experimental design was implemented for optimization. The CaO–SiO₂-CaCl₂ system achieved a peak boron removal efficiency of 88 pct, with the CaO–SiO₂-CaF₂ and CaO–SiO₂-Al₂O3 systems also reaching high efficiencies of ≥ 86 pct and ≥ 75 pct, respectively. This work establishes a data-driven paradigm that integrates machine learning with metallurgical experimentation for intelligent process optimization.

Conventional recycling of spent contact mass (SCM) from the organosilicon industry faces environmental and efficiency challenges. This work proposes a combined process of low-temperature oxidation roasting and selective leaching to recover Si and Cu, which are then used to fabricate Cu-Si phase transition alloys via hightemperature alloying. The process achieved high recovery efficiencies of 85.73% for Si and 90.86% for Cu. The resulting Cu-20wt%Si and Cu-25wt%Si alloys exhibited high melting enthalpies of 187.4 J center dot g-1 and 143.7 J center dot g-1 , respectively, and exhibited excellent cyclic stability with minimal latent heat loss after 100 cycles. Molecular dynamics simulations revealed that higher Si content enhances Cu-Si interatomic interactions, improving lattice stability, and provided atomic-scale insights into the phase transition dynamics. This work enables high-value utilization of industrial waste and demonstrates the potential of these Cu-Si alloys for hightemperature waste heat recovery and peak shaving in new energy systems.

Industrial silicon is the fundamental raw material for the photovoltaic industry, where impurity Ti exerts a remarkable negative effect on the properties of downstream polysilicon and solar cell performance. Nevertheless, Ti removal efficiency is considerably restricted owing to the insufficient kinetic data of Ti during the refining process. In this work, the diffusion and mass transfer behaviors of Ti in molten silicon and three slags (CaO-SiO2, CaO-SiO2-Al2O3, CaO-SiO2-Al2O3-Na2O) were systematically studied. Diffusion coefficients at 1450 degrees C, 1500 degrees C and 1550 degrees C were measured by the melt-capillary method, and diffusion activation energies were calculated via the Arrhenius equation. The results show that Ti diffusion in molten silicon is 7.25 & times; 10- 9-1.12 & times; 10- 8 m2 & sdot;s- 1 with an activation energy of 108.575 kJ & sdot;mol- 1. High CaO content lowers slag viscosity and activation energy, promoting Ti diffusion. Slag refining experiments at 1500 degrees C (slag/silicon ratio 1:1) yield mass transfer coefficients and effective boundary layer thicknesses in silicon and slag phases, achieving 33.7% Ti removal in 150 min. Kinetic analysis reveals that mass transfer in the slag phase is the rate-limiting step. This work is the first to systematically report the diffusion coefficients and mass transfer coefficients of Ti in molten silicon and multi-component slag systems. It fills a critical gap in the kinetic data for this field and provides a theoretical basis for process optimization.

The disposal of end-of-life lithium-ion batteries (LIBs) and silicon solar panels generates a large amount of solid waste. However, the comprehensive recycling of spent lithium-ion batteries and diamond wire saw silicon powder (DWSSP) remains challenging. In this study, Li4SiO4, a CO2 adsorbent, was synthesized through a simplified one-step process using spent LIBs cathode material as the lithium source and DWSSP as the silicon source. The results reveal that the stage heating approach effectively facilitates the silicothermic reduction of the active elements (Ni, Co and Mn) in the LIBs cathode. This enables the successful synthesis of Li4SiO4. Mn doping in Li4SiO4 increases the pore volume and surface area of the synthesized material. The Li4SiO4 sample synthesized under the conditions of 1400 degrees C for 1 h with 3 g of DWSSP reductant and a resultant Mn content of 1.178 wt % exhibited a CO2 adsorption capacity of 0.336 g/g in a pure CO2 atmosphere. This performance is comparable to, or even exceeds, that reported by other research groups. This work presents a potential solution for the simultaneous recovery of valuable metals from spent LIBs and DWSSP.

Energy shortages and environmental concerns have been widely acknowledged as critical challenges facing humanity, consequently driving the development of new energy sources [...]

Accurate prediction of activity interaction coefficients is essential for optimizing the industrial silicon smelting and purification process. Nevertheless, several challenges persist, including the scarcity of experimental data, the complexity of multi-component interactions, prohibitive experimental costs and extended durations, and a substantial divergence between theoretical and experimental values. In this study, six machine learning models including Linear Regression (LR), Support Vector Regression (SVR), Multi-Layer Perceptron Neural Network (MLP), Extreme Gradient Boosting (XGB), Tabnet, and Tab-Transformer (Tabnet-TF) were used to predict both theoretical and experimental activity interaction coefficients. The study found that XGB exhibited the best performance on the theoretical data, achieving determination coefficient (R2) > 0.98, root mean square error (RMSE) < 0.0082, mean absolute error (MAE) < 0.67, and mean absolute percentage error (MAPE) < 0.22. By integrating XGB with a transfer learning approach, pre-training on theoretical data and fine-tuning with sample weighting. The results demonstrate that the model significantly improved prediction accuracy on experimental data, attaining an R2 of 0.9960, compared to 0.6937 with the best standalone model. This work presents an efficient strategy for predicting thermodynamic parameters in silicon refining and validates the effectiveness of transfer learning in materials computation.

Industrial silicon gas blowing has a significant removal effect on impurity P, but the depth of impurity removal is governed by reaction kinetics. In this paper, the structural evolution and dynamic properties of P during industrial silicon gas blowing are investigated by a combination of experimental analysis and ab initio molecular dynamics methods. The results showed that the removal efficiency of impurities was gradually enhanced with increasing temperature. Comprehensive analysis of the properties of the electronic structure at 1723 K and 1823 K shows weak interactions between the P–O bonds and a major structural transition to P2(g). In addition, the kinetic properties of P in silicon melts were obtained, and the diffusivity at 1723 K and 1823 K was calculated to be 4.61 × 10−9 and 1.547 × 10−8m2/s, respectively. This study elucidated the effectiveness of P removal in industrial silicon gas blowing and revealed the removal mechanism of impurity phosphorus in industrial silicon gas blowing.

The study of diffusive mass transfer of impurity phosphorus (P) in silicon melts was a crucial aspect of achieving efficient deep P removal. To elucidate the mass transfer mechanism of P in silicon melts, this paper presents a kinetic analysis of impurity P removal from silicon melts. To minimize the effects of convection on diffusion, a capillary diffusion device was employed to conduct experimental investigations of P diffusion in silicon melts at different temperatures. According to slag–silicon refining experiments, the mass transfer process of P between the silicon melt and the slag phase was characterized and the mass transfer coefficient of P in the silicon melt was calculated. The experimental results indicated that the diffusion coefficients of P in silicon melt at 1450 °C, 1500 °C, and 1550 °C were 2.81 × 10−8, 3.24 × 10−8, and 3.91 × 10−8 m2·s−1, respectively. Through the calculation of the activation energy for P diffusion, the relationship between the diffusion coefficient and temperature was determined as D=1.23×10^-5·e^-87,260/RT . At a refining temperature of 1550 °C, the mass transfer coefficient of P in the silicon melt was found to be 2.11 × 10−5 m·s−1 when calcium silicate slag was used for slag refining. These findings provided essential kinetic data, including diffusion and mass transfer coefficients, to support the advanced removal of impurity P from metallurgical-grade silicon (MG-Si).

Recovering Li from spent battery cathodes is crucial for mitigating scarcity and promoting sustainability. Conventional pyro-hydrometallurgical methods face challenges in Li selectivity due to volatilization losses and inadequate phase control. This study proposes a green approach for sustainable battery recycling. It combines CO2-mediated carbothermal reduction with CO2-saturated leaching to selectively recover Li from spent lithium-ion batteries. By utilizing graphite-mediated carbothermal reduction under precisely controlled CO2 atmospheres at 800 degrees C, Li2O are converted into stable Li2CO3, effectively suppressing Li volatilization. Coupled with CO2-saturated leaching at 30 degrees C, the process achieves 99.8 % Li recovery and 99.9 % Li2CO3 purity. A CO2-controlled "Li2O-Li2CO3 '' dual-phase dynamic equilibrium model was established, effectively restricting the migration path of Li. This study enhances understanding of gas-solid equilibrium in carbothermal processes and advances sustainable, efficient battery recycling.

With the increasing annual output of diamond wire saw silicon powder (DWSSP) from photovoltaic (PV) wafer production, the recovery and purification of silicon from DWSSP are crucial for producing high-purity silicon. This study investigated the migration mechanism of impurity elements during the vacuum refining process of DWSSP through kinetic analysis. The dynamic relationship between impurity removal efficiency and the mass transfer coefficient was examined using a volatile mass transfer model at the gas-liquid interface. At a melting temperature of 1700 K, the removal of impurity elements (including Na, K, and Ca) was controlled by diffusion in the melt boundary layer. In contrast, P and Mn removal were dominated by surface volatilization. Comparative experiments confirmed that the combination of acid leaching pretreatment and vacuum refining eliminated the oxide encapsulation effect, thereby reducing impurity levels. After acid leaching, DWSSP was processed under optimized parameters (a vacuum of 10(-3)-10(-2) Pa, a furnace chamber melting temperature of 1823 K, and a holding time of 120 min). The residual concentrations of P, Na, K, Ca, and Mn were 0.84, 4.33, <2.80, and 0.22 ppmw, respectively, with the Log Reduction Value (removal rates) of 1.058(91.30 %), 2.166(99.32 %), 1.809 (97.80 %), 1.087(91.71 %), and 2.482(99.65 %). These results indicate that acid leaching and fire vacuum melting can effectively eliminate impurity elements from DWSSP, and the recovered silicon is likely to achieve a 6 N purity level. which holds significant practical implications for promoting the sustainable development and low-carbon transformation of the PV industry.

Environmentally hazardous diamond wire saw silicon powder (DWSSP) generated during the production of silicon solar cells possesses a considerable reuse value. However, the presence of a silicon oxide (SiOx) layer and impurities poses a significant barrier to Si recovery and utilization. A novel method that combines acid leaching and spray drying granulation has been proposed to effectively inhibit silicon powder's secondary oxidation. Additionally, vacuum directional solidification is integrated to achieve efficient melting of silicon material and thorough removal of metal impurities. The study investigates the dissolution process of the oxide layer during acid leaching and examines how silicon particle morphology and solidification rate impact both silicon melting rate and metal impurity segregation based on heat transfer and solidification kinetics respectively. The results demonstrate that the dissolution of amorphous silicon dioxide shell and spray granulation increases the melting efficiency of DWSSP by up to 99.89 % and enhances impurity removal from it. Furthermore, an increase in solidification rate improves impurity segregation efficiency significantly, at a solidification rate of 3 mu m/s, total metal impurity content in silicon is reduced from 641 ppmw to 5.75 ppmw. This study showcases the feasibility of utilizing spray granulation combined with vacuum directional solidification technology to achieve low-energy, high-purity recovery of DWSSP.

The production of solar-grade silicon is significantly influenced by the presence of harmful impurities in industrial silicon. However, there is a dearth of comprehensive thermodynamic data on these impurities within the silicon melt, thereby making it challenging to obtain accurate predictions. In this study, five machine learning models including Linear Regression, Support Vector Regression, Multi-Layer Perceptron Neural Network, Random Forest, and Extreme Gradient Boosting, were used to predict the activity of each component of silicon-based binary melts. The model performance was evaluated using determination coefficient (R-2), root mean square error (RMSE), mean absolute error (MAE) and overfitting degree. The Random Forest(RF) model outperforms others, achieving a mean R(2)of over 99.32 % for the training set (RMSE <1.59 %, MAE < 0.97 %) and over 98.5 % for the test set (RMSE <2.39 %, MAE < 1.49 %). The predicted activity value from the RF model was compared with the experimental data, demonstrating a strong correlation and validating the accuracy of the model, reliability, and robustness. The study successfully established the relationship between temperature, compositions and the activity of each component of silicon-based binary melt systems under high-temperature conditions. This study offers valuable theoretical data for the production of industrial silicon.

The solidification process of industrial silicon (Si) is influenced by impurities, which significantly affect the selectivity and activity of silicone monomer synthesis. This study utilized the molecular interaction volume model (MIVM), factory data, and industrial silicon samples to explore the effects of different Fe concentrations on typical impurity phases in industrial silicon during actual production. At a Fe concentration of 3300 ppmw, Si7Al8Fe5 and Si2Al3Fe phases appeared. At a Fe concentration of 3400 ppmw, the Si2Al3Fe phase disappeared. The FeSi2(Al), Si8Al6Fe4Ca, and FeTiSi2 phases were consistently present in industrial silicon. The MIVM model prediction indicated that the activity of Ca was negatively correlated with Fe concentration, while the activities of Ti and aluminum (Al) were positively correlated with iron concentration, which was consistent with the actual production process. This study elucidated the influence of Fe concentration on typical impurity phases in industrial silicon and provided a theoretical and technical basis for impurity regulation in the synthesis of organic silicon monomers.

This study investigated the use of biomass ash as an additive for reducing the boron (B) content in industrial silicon (Si). Coffee shell, a biomass rich in alkali earth metals, was selected for its potential to effectively remove non-metallic impurities from industrial silicon. A series of single-factor experiments was designed to systematically examine the effects of biomass ash content, smelting temperature, and refining time on B removal. The experimental results indicated that under optimized process conditions, the ideal biomass ash addition was 5 wt.