This paper proposes a framework for designing B-site-alloyed ABX3 metal halide perovskites (MHPs) using density functional theory (DFT) and machine learning (ML). A crystal graph convolutional neural network (CGCNN) was trained on 3159 DFT datasets to predict decomposition energy, bandgap, and bandgap type. The trained model explored the chemical space of 41,400 B-site-alloyed MHPs, identifying 10 promising compounds with optimal bandgaps for single-junction and tandem solar cells, namely CsGe0.3125Sn0.6875I3 and CsGe0.0625Pb0.3125Sn0.625Br3.
Publisher
npj Computational Materials
Published On
May 08, 2024
Authors
Jin-Soo Kim, Juhwan Noh, Jino Im
Tags
B-site-alloyed
metal halide perovskites
density functional theory
machine learning
solar cells
bandgap
decomposition energy
Related Publications
Explore these studies to deepen your understanding of the subject.