Study on the optimization of large-area perovskite fabrication assisted by machine learning and their module performance

Perovskite solar cells have become a frontier research topic due to their low cost and significant advantages in solution processing. However,the scalability of fabricating large-area perovskite films remains a challenge,which hinders the development and commercialization of perovskite solar modules. Developing high-throughput fabrication techniques for perovskite films is an effective approach. In this study,machine learning high-throughput algorithms were used to explored the influence of coating parameters on the preparation of large-area perovskite films prepared by blade coating in ambient air condition. Based on the machine learning results,we selected the optimal coating parameters and successfully fabricated 36 cm² perovskite solar modules. The modules prepared in ambient air condition achieved a photoelectric conversion efficiency of 18.89%. To further enhance the performance of the modules,a layer of phenylethylammonium iodide was decorated on the surface of the perovskite film to reduce interface defects,and the efficiency of the corresponding modules reached up to 19.53%. The module still retaining 95% of the initial efficiency for 480 h under air condition.