Exploring Lead-Free Perovskite Solar Cells by Machine Learning

Perovskite solar cells have high energy conversion efficiency and are one of the promising next generation solar cells that are expected to generate renewable resources and solve the global energy problems. The remarkable performance includes a large absorption coefficient in the UV-visible absorption spectra, high carrier conductivity, long electron and hole diffusion lengths, and direct band gap properties. Perovskite solar cells can be easily synthesized at low cost, but the toxicity of typical candidate compound materials due to the presence of lead is a problem. Recently, Nakajima et al. performed DFT calculations on 11025 new candidate materials in a high-throughput study using the supercomputer "K" [1]. Based on the obtained data sets, we used statistical and machine learning methods to explore the suitable candidate compounds for solar cell materials. As a result, we were able to identify the important physical properties for the features we focused on and evaluate the scatter plot to analyze the compounds suitable for solar cell materials.

[1] T. Nakajima, K. Sawada, J. Phys. Chem. Lett., 2017, 8, 4826-4871.