Recent Advances in Silicon Solar Cell Research Using Data Science-Based Learning

The application of machine learning techniques in silicon photovoltaics research and production has been gaining traction. Learning from the existing data has given the potential to research labs and industries of discovering optimized processing parameters, device architectures, and fabrication recipes. It has also been utilized for defect detection and quality inspection. The increasing computational capacities of modern computers and abstraction of machine learning algorithms, along with the increasing community support for open-source software libraries has increased the accessibility of learning-based algorithms that were perceived as complex to be implemented for interdisciplinary research and development just a few years back. In this article, we present a review of the efforts in the literature that have utilized machine learning techniques for commercial silicon solar cell devices in recent times. The emphasis is to categorize and investigate specific learning techniques that are best suited for one particular device or fabrication process parameter optimization. We also provide insight into possible expansions of current research methodologies that can further improve the prediction accuracy while minimizing the computational costs and extract other useful information from a machine learning model, such as prediction uncertainty, scalability, and generalization of a particular model.