Industrial silicon melt purification method based on directional solidification and composite adsorption

Abstract

The invention relates to the field of industrial silicon melt purification, and discloses an industrial silicon melt purification method based on directional solidification and composite adsorption, which comprises the following steps: S1, multi-stage directional solidification pre-enrichment: carrying out primary directional solidification treatment on an industrial silicon melt, and controlling the solidification rate to be 0.5-5mm / min and the temperature gradient parameter, most of metal impurities in the melt are pre-enriched in a specific area; s2, dynamic targeted adsorption: in or after the primary directional solidification process, an efficient composite adsorbent is dynamically and continuously added into an impurity-rich area of the melt or the whole melt in batches, and the composite adsorbent has selective adsorption capacity for different impurities; and S3, intelligent regulation and control: constructing a full-flow adaptive intelligent dynamic regulation and control system based on digital twinning and AI algorithms. Compared with the prior art, the method has the advantages that the impurities are efficiently and stably removed, the grain structure is accurately modified, and meanwhile, the green and low-carbon production target is achieved.

Description

Technical Field

[0001] This invention relates to the field of industrial silicon melt purification technology, specifically to a method for purifying industrial silicon melt based on directional solidification and composite adsorption. Background Technology

[0002] Industrial silicon is a core raw material for strategic emerging industries such as photovoltaic power generation, semiconductor integrated circuits, and organosilicon new materials. Its purity, crystal structure, and mechanical properties directly determine the performance and reliability of downstream products. Currently, high-purity silicon materials (total impurity content <0.1wt%) required for high-end applications mainly rely on imports, while the domestic industry still focuses on low- to mid-range metallurgical-grade products.

[0003] Existing technologies for deep purification and precise microstructure modification of industrial silicon melts have formed a mature system, enabling efficient impurity removal and intelligent production. However, the traditional flux adsorption method is still used, which has limited effectiveness in removing key impurities such as boron and phosphorus, and is prone to introducing secondary pollution, causing environmental burden. Although directional solidification technology is considered an effective purification method due to its utilization of the different segregation coefficients of impurities in the solid and liquid phases, single directional solidification technology is not effective in removing some impurities with segregation coefficients close to 1, and the traditional solidification process is poorly controlled, resulting in unstable product quality. In addition, most advanced directional solidification and grain modification technologies remain at the laboratory stage and have not yet solved key engineering problems in large-scale industrial production, such as precise linkage control of process parameters, excessive energy consumption, and waste residue and exhaust gas treatment. Summary of the Invention

[0004] (a) Technical problems to be solved

[0005] The technical problem to be solved by the present invention is to provide an industrial silicon melt purification method based on directional solidification and composite adsorption. This method achieves efficient and stable removal of impurities and precise modification of grain structure through the deep synergy of multi-stage directional solidification pre-enrichment of impurities and dynamic targeted adsorption purification, combined with intelligent dynamic control of the whole process based on digital twin and AI algorithm, and finally achieves green and low-carbon production goals.

[0006] (II) Technical Solution

[0007] To solve the above-mentioned technical problems, the technical solution provided by the present invention is: a method for purifying industrial silicon melt based on directional solidification and composite adsorption, comprising the following steps:

[0008] S1. Multi-stage directional solidification pre-enrichment: The industrial silicon melt undergoes primary directional solidification treatment. By controlling the solidification rate to 0.5-5 mm / min and the temperature gradient parameter, most of the metal impurities in the melt are pre-enriched in a specific area.

[0009] S2. Dynamic targeted adsorption: During or after the primary directional solidification process, a highly efficient composite adsorbent is dynamically, in batches or continuously added to the impurity-rich region or the entire melt of the melt. The composite adsorbent has selective adsorption capacity for different impurities. At the same time, stirring conditions are used to ensure that the adsorbent and the melt are in full contact to capture impurities.

[0010] S3. Intelligent Control: A full-process adaptive intelligent dynamic control system based on digital twins and AI algorithms collects sensor data from the production line in real time. It constructs a digital twin of the production process through multi-physics simulation models and historical production data. The AI ​​model analyzes the data to predict key parameters such as the dynamic changes in impurity content in the melt, the position and morphology of the solidification interface, and the evolution of grain size. Based on the prediction results, it automatically optimizes and dynamically adjusts process variables such as heating power curves, cooling rate programs, electromagnetic stirring parameters, and adsorbent dosing rate to achieve closed-loop control.

[0011] As an improvement, the multi-stage directional solidification described in step S1 has at least two stages, and the impurities are enriched in a stepwise and deep manner by controlling the solidification conditions at each stage.

[0012] As an improvement, the high-efficiency composite adsorbent described in step S2 has its components specifically designed according to the target impurity elements, including components that can form stable compounds or eutectics with B, P, Fe, Al, and Ca impurities.

[0013] As an improvement, the stirring conditions described in step S2 are one or more combinations of electromagnetic stirring, gas stirring, or mechanical stirring.

[0014] As an improvement, the digital twin described in step S3 is constructed by integrating a multi-physics field coupled simulation model of thermal field, flow field, and solute field.

[0015] As an improvement, the AI ​​model mentioned in step S3 is a deep learning model or a reinforcement learning model.

[0016] As an improvement, the method also includes a greening process step:

[0017] S4. High-temperature and corrosion-resistant heat exchangers are used to recover waste heat from smelting exhaust gas for preheating raw materials, preheating combustion air, or generating steam, thus realizing the cascade utilization of waste heat.

[0018] S5. The smelting waste slag generated during the purification process is mixed with a carbonaceous reducing agent and subjected to a carbothermic reduction-carbonization reaction under specific atmosphere and high temperature conditions to prepare silicon carbide-based materials or building materials, thereby realizing the resource utilization of waste slag.

[0019] As an improvement, the comprehensive waste heat recovery and utilization rate described in step S4 reaches over 60%.

[0020] As an improvement, the silicon carbide-based material mentioned in step S5 can be silicon carbide abrasive, refractory aggregate, or porous ceramic support.

[0021] (III) Beneficial Effects

[0022] The advantages of this invention compared to the prior art are:

[0023] By combining pre-enrichment through "multi-stage directional coagulation" with purification through "dynamic targeted adsorption," the limitations of single technologies are overcome. Directional coagulation treats most easily segregated impurities, while targeted adsorbents specifically remove elements such as B and P that are difficult to remove through segregation, ensuring that the overall removal rate of major impurities consistently reaches over 90%, and the total impurity content of the final product can be controlled below 0.1 wt%.

[0024] The intelligent control system ensures high stability of the solidification process, can stably control the average grain size below 50μm and ensure uniform size distribution, significantly improves the microstructure and mechanical properties of the product, and greatly improves the stable room temperature tensile strength of silicon ingots. Detailed Implementation

[0025] The invention will now be described in further detail with reference to specific embodiments, but this should not be construed as limiting the scope of the subject matter of the invention to the following embodiments.

[0026] Example 1

[0027] The industrial silicon raw material obtained by smelting (with an initial total impurity content of approximately 1.2 wt%) was used as the treatment target.

[0028] S1. Inject the molten silicon into the directional solidification furnace to initiate primary directional solidification. The solidification rate is controlled at 2 mm / min, and the temperature gradient is 30 K / cm, using an intelligent control system. The solidification process continues, causing metallic impurities such as Fe and Al to accumulate towards the tail of the melt.

[0029] S2. When solidification reaches approximately 70%, the system predicts via digital twin that a tail-end enriched zone has formed. Subsequently, a composite adsorbent designed for B and P (mainly composed of SiO2-CaO-CaF2 system) is continuously added to the tail-end enriched zone of the melt via a spraying system, and an electromagnetic stirring device is simultaneously activated at a stirring intensity of 100 rpm for 20 minutes to ensure that the adsorbent reacts fully with the melt.

[0030] S3. The entire process described above is monitored by an intelligent control system. Based on real-time thermocouple and spectral sensor data, the system uses a pre-trained dynamic convolutional neural network model to predict impurity concentration distribution and grain growth, and automatically fine-tunes the cooling water flow rate in subsequent stages to stabilize the solidification rate within ±3% of the set value.

[0031] S4. The generated 1200℃ high-temperature flue gas enters the waste heat boiler, and the recovered heat energy is used to preheat the next batch of silicon raw materials. According to calculations, the waste heat recovery efficiency of this step reaches 65%.

[0032] S5. After solidification, approximately 20% of the impurity-rich silicon ingot at the tail end was removed. The removed waste residue was mixed with coke at a mass ratio of 5:1 and reacted at 1950℃ for 3 hours under argon protection, successfully synthesizing SiC abrasive.

[0033] Testing revealed that the final high-purity silicon ingot had a total impurity content of 0.08 wt%, an average grain size of 45 μm, a grain size distribution standard deviation of 12%, and a room temperature tensile strength of 125 MPa. The overall removal rate of major impurities reached 92%, and the first-pass yield was 98.5%.

[0034] Example 2

[0035] The steps are basically the same as in Example 1, except that:

[0036] In this case, the intelligent control system adopted a reinforcement learning model. Through continuous interaction and iteration with the digital twin, it autonomously optimized a more energy-efficient combination of process parameters: optimizing the primary solidification rate to 3.5 mm / min.

[0037] The adsorbent was changed to be added in batches. The first batch of 60% of the total amount was added when the solidification was 50%, and the second batch of the remaining 40% was added when the solidification was 80%.

[0038] The final product also met the predetermined targets, and the total energy consumption was reduced by 8% compared to Example 1.

[0039] In summary, the method provided by this invention effectively solves the problems of incomplete impurity removal, unstable product quality, and high energy consumption and pollution in existing technologies. It is an advanced method very suitable for large-scale production of high-purity industrial silicon. Although embodiments of this invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of this invention. The scope of this invention is defined by the appended claims and their equivalents. In short, if those skilled in the art are inspired by this invention and design similar structures and embodiments without departing from the inventive spirit of this invention, they should all fall within the protection scope of this invention.

Claims

1. A method for purifying industrial silicon melt based on directional solidification and composite adsorption, characterized in that, Includes the following steps: S1. Multi-stage directional solidification pre-enrichment: The industrial silicon melt undergoes primary directional solidification treatment. By controlling the solidification rate to 0.5-5 mm / min and the temperature gradient parameter, most of the metal impurities in the melt are pre-enriched in a specific area. S2. Dynamic targeted adsorption: During or after the primary directional solidification process, a highly efficient composite adsorbent is dynamically, in batches or continuously added to the impurity-rich region or the entire melt of the melt. The composite adsorbent has selective adsorption capacity for different impurities. At the same time, stirring conditions are used to ensure that the adsorbent and the melt are in full contact to capture impurities. S3. Intelligent Control: A full-process adaptive intelligent dynamic control system based on digital twins and AI algorithms collects sensor data from the production line in real time. It constructs a digital twin of the production process through multi-physics simulation models and historical production data. The AI ​​model analyzes the data to predict key parameters such as the dynamic changes in impurity content in the melt, the position and morphology of the solidification interface, and the evolution of grain size. Based on the prediction results, it automatically optimizes and dynamically adjusts process variables such as heating power curves, cooling rate programs, electromagnetic stirring parameters, and adsorbent dosing rate to achieve closed-loop control.

2. The method for purifying industrial silicon melt based on directional solidification and composite adsorption according to claim 1, characterized in that, The multi-stage directional solidification described in step S1 has at least two stages, and the impurities are enriched in a stepwise and deep manner by controlling the solidification conditions at each stage.

3. The method for purifying industrial silicon melt based on directional solidification and composite adsorption according to claim 1, characterized in that, The high-efficiency composite adsorbent described in step S2 has components specifically designed according to the target impurity elements, including components that can form stable compounds or eutectics with B, P, Fe, Al, and Ca impurities.

4. The method for purifying industrial silicon melt based on directional solidification and composite adsorption according to claim 1, characterized in that, The stirring conditions described in step S2 are one or more combinations of electromagnetic stirring, gas stirring, or mechanical stirring.

5. The method for purifying industrial silicon melt based on directional solidification and composite adsorption according to claim 1, characterized in that, The digital twin described in step S3 is constructed by integrating a multi-physics field coupled simulation model of thermal field, flow field, and solute field.

6. The method for purifying industrial silicon melt based on directional solidification and composite adsorption according to claim 1, characterized in that, The AI ​​model mentioned in step S3 is a deep learning model or a reinforcement learning model.

7. The method for purifying industrial silicon melt based on directional solidification and composite adsorption according to claim 1, characterized in that, The method also includes a greening process: S4. High-temperature and corrosion-resistant heat exchangers are used to recover waste heat from smelting exhaust gas for preheating raw materials, preheating combustion air, or generating steam, thus realizing the cascade utilization of waste heat. S5. The smelting waste slag generated during the purification process is mixed with a carbonaceous reducing agent and subjected to a carbothermic reduction-carbonization reaction under specific atmosphere and high temperature conditions to prepare silicon carbide-based materials or building materials, thereby realizing the resource utilization of waste slag.

8. The method for purifying industrial silicon melt based on directional solidification and composite adsorption according to claim 7, characterized in that, The comprehensive waste heat recovery and utilization rate described in step S4 reaches over 60%.

9. The method for purifying industrial silicon melt based on directional solidification and composite adsorption according to claim 7, characterized in that, The silicon carbide-based material mentioned in step S5 is silicon carbide abrasive, refractory aggregate, or porous ceramic support.