Machine learning-based ligand engineering of halide perovskite quantum dots for improving photoluminescence quantum yield

Abstract

Halide perovskite quantum dots (h-PQDs) are the next-generation nanomaterials featuring excellent photonic properties and facile processability; however, effective passivation of their surface defects via high-affinity ligands remains a challenge for improving their structural stability and photonic efficiency, known as the photoluminescence quantum yield (PLQY). To accelerate the search process of the optimal ligands, this study employs machine learning to analyze the relationship between the molecular properties of various ligands and PLQY in CsPbBr3 QDs, one of the most intensively studied h-PQDs. Based on the model, several promising candidates have been predicted to yield high PLQY above 80% and are presented for the first time. Finally, the transferability of these predictions to other halide perovskites has been demonstrated in the case study of Cs2PbSnI6 by computing their binding energy using charge-informed molecular dynamics. This approach offers an accelerated framework for ligand engineering for high-performing h-PQDs, with broad implications in various technological applications.