Machine Learning Quantification of Grain Characteristics for Perovskite Solar Cells

Microstructures including grains, grain boundary grooves, and surface fluctuations prevalent in metal halide perovskite (MHP) films and can substantially impact the electronic properties, photovoltaic performance, and stability of solar cells. To surpass the Shockley-Queisser limits of perovskite solar cells (PSCs) and unlock the full potential of MHPs, it is essential to characterize and modify these microstructures. However, the measurement and statistical interpretation are still challenging due to their huge quantities. Here, we developed a machine-learning-based toolkit for extracting and quantifying microstructural characteristics of MHP from atomic force microscope, enabling a reliable statistical analysis. A convolutional neural network with U-Net structure was trained for grain region extraction, and multiple kinds of methodology were produced for quantifying the microstructural characteristics including grain surface area, grain boundary groove angles, groove width, grain surface depression and bulge. Based on this toolkit, we then expanded the study from localized measurement to their statistical distribution over the whole film, and reveal their correlation with the perovskite solar cells performance. This work not only interpret the relationship between microstructure-property-performance of perovskite solar cells, but also reveal potential modification direction for MHP and PSCs.