Data-Driven Optimization and Experimental Validation for the Lab-Scale Mono-Like Silicon Ingot Growth by Directional...

Area of Science:

• Materials Science
• Semiconductor Manufacturing
• Computational Modeling

Background:

• Directional solidification (DS) silicon is promising for high-efficiency solar cells.
• High dislocation clusters in DS silicon limit competitiveness with monocrystalline silicon.
• Optimizing DS silicon growth requires advanced simulation and optimization techniques.

Purpose of the Study:

• To develop and validate an integrated framework for optimizing the directional solidification (DS) silicon growth process.
• To reduce residual stress and dislocation density in mono-like silicon ingots.
• To enhance the quality and efficiency of silicon for solar cell applications.

Main Methods:

• Integration of experimental growth, transient global simulations, artificial neuron network (ANN) training, and genetic algorithms (GA).
• ANN used as an instant predictor of silicon ingot properties based on growth recipes.
• GA employed with the ANN predictor to search for optimal multi-objective growth parameters.

Main Results:

• The integrated framework successfully reduced dislocation density and smoothed the growth rate.
• Optimal recipe implementation led to decreased interface concavity and reduced multi-crystallization in the top ingot region.
• The ANN-GA combination provided instant and quantitative derivation of optimal growth parameters.

Conclusions:

• The developed integrated framework effectively optimizes the DS silicon growth process.
• This approach significantly improves silicon ingot quality by minimizing defects and controlling growth interface.
• The methodology offers a powerful tool for achieving high-performance silicon for solar cell manufacturing.