Predictions and Strategies Learned from Machine Learning to Develop High‐Performing Perovskite Solar Cells

Abstract Perovskite solar cells (PSCs) have recently received considerable attention due to the high energy conversion efficiency achieved within a few years of their inception. However, a machine learning (ML) approach to guide the development of high‐performing PSCs is still lacking. In this paper ML is used to optimize material composition, develop design strategies, and predict the performance of PSCs. The ML models are developed using 333 data points selected from about 2000 peer reviewed publications. These models guide the design of new perovskite materials and the development of high‐performing solar cells. Based on ML guidance, new perovskite compositions are experimentally synthesized to test the practicability of the model. The ML model also shows its ability to predict underlying physical phenomena as well as the performance of PSCs. The PSC model matches well with the theoretical prediction by the Shockley and Queisser limit, which is almost impossible for a human to find from an ensemble of data points. Moreover, strategies for developing high‐performing PSCs with different bandgaps are also derived from the model. These findings show that ML is very promising not only for predicting the performance, but also for providing a deeper understanding of the physical phenomena associated with the PSCs.