Critical review of machine learning applications in perovskite solar research

Highlights: • Machine learning (ML) works on organolead perovskite solar cells are reviewed. • Both experimental and computational data are used to develop ML models. • Data created in house or extracted from papers and databases are utilized. • Screening perovskites for band gap, structure and stability are common applications. • Machine learning models for cell efficiency and stability were also studied. The astonishing progress achieved in perovskite solar cells in recent years has coincided with the growing interest in machine learning (ML) for material discovery, and the number of papers reporting the use of ML in perovskite solar research has been increased significantly in last two years. ML has been used for various purposes such as discovering new perovskites by screening the large computational or experimental datasets, analyzing the spectroscopic data augmented by data extracted from databases, determining conditions for higher efficiency or stability using experimental data and identifying the basic trends in perovskite solar cell technology by analyzing the published papers and patents. This communication aims to review the research articles as well as the perspectives, comments and opinions, to assess the current directions and summarize the challenges and opportunities for the future works in the field.