Application of Machine Learning for Optimizing Materials Selection and Processing to Enhance Efficiency in Organic Solar Cells

Abstract

The design and optimization of donor and acceptor materials in organic solar cells (OSCs) are crucial for improving their efficiency and commercial viability. This review explores the intersection of machine learning (ML) techniques and material science to address challenges in enhancing low power conversion efficiency, addressing material instability and degradation, balancing light absorption and charge transport properties, and identifying cost-effective, scalable materials for large-scale production compared to traditional silicon cells. To identify the latest advancements in the application of machine learning to organic photovoltaics, a comprehensive search was conducted in the scientific database for papers containing terms related to organic photovoltaics in their titles. From this initial set, papers that also included 'machine learning' in their titles and were published after 2020 were selected for detailed review in this study. The analysis revealed four principal areas of machine learning applications: property prediction, inverse design, machine-learned atomic potentials, and active learning. These emerged as key research areas in the application of machine learning to organic chemistry and related fields. Consequently, these keywords were further searched in scientific databases, and all relevant papers pertaining to organic chemistry were included for review. New machine learning methods have been proposed for areas such as metal-organic frameworks, drug design, and hybrid organic-inorganic perovskites. However, these methods have yet to be applied to the field of organic photovoltaics. Applying these methods to organic photovoltaics could lead to the discovery of new materials that can serve as donors and acceptors, potentially resulting in the development of high-efficiency organic solar cells.