Machine Learning-Assisted Automated Laser Recovery of Silver Electrode Recovery from Waste Silicon Solar Cells

This study presents a machine learning-assisted automated laser recycling approach for silver electrode recovery from waste silicon solar cells. Traditional methods for silver recovery, such as chemical leaching, are inefficient and environmentally hazardous, while laser-based techniques require precise electrode localization to prevent silicon contamination. A convolutional neural network (CNN) was trained to automate silver electrode detection, overcoming challenges associated with manual alignment and variability in broken end of life solar cells of different shapes and sizes. Previously, we presented a laser-based silver recovery approach from silicon solar cells at the PVSC 52nd conference. The present study advances this work by introducing machine learning-assisted automation for the laser recovery process, laying the foundation for large-scale silver recovery. The trained CNN model successfully identified silver electrodes across diverse conditions, including varying surroundings, lighting environments, and water submerged samples. The detected electrode regions were converted into vector files, enabling precise infrared (IR) nanosecond laser scanning for selective silver removal while minimizing damage to the silicon substrate. Validation demonstrated high detection accuracy, confirming the model' effectiveness in enhancing process efficiency and silver recovery yield. By integrating deep learning with laser processing, this approach eliminates labor-intensive manual alignment, reduces processing time, and improves scalability for industrial-scale recycling. The findings establish AI-driven automation as a cost-effective and sustainable solution for high-purity silver recovery, advancing intelligent material reclamation in photovoltaic waste management and circular economy. In our knowledge, this is the first report in the literature on the application of machine learning for solar cell recovery.