Leveraging Domain Adaptation for Accurate Machine Learning Predictions of New Halide Perovskites

TL;DR

This paper develops a domain-adapted graph neural network approach combined with bond-valence method to efficiently screen a vast number of halide perovskites, identifying promising candidates for photovoltaic applications with improved accuracy over traditional methods.

Contribution

It introduces a novel combination of GNNs and domain adaptation for accurate property prediction in halide perovskites, significantly enhancing screening efficiency.

Findings

Achieved 1.8x improvement in Ef prediction accuracy.

Identified 48 promising compounds, 40 of which are unexplored.

Demonstrated the effectiveness of domain adaptation in materials screening.

Abstract

We combine graph neural networks (GNN) with an inexpensive and reliable structure generation approach based on the bond-valence method (BVM) to train accurate machine learning models for screening 222,960 halide perovskites using statistical estimates of the DFT/PBE formation energy (Ef), and the PBE and HSE band gaps (Eg). The GNNs were fined tuned using domain adaptation (DA) from a source model, which yields a factor of 1.8 times improvement in Ef and 1.2 - 1.35 times improvement in HSE Eg compared to direct training (i.e., without DA). Using these two ML models, 48 compounds were identified out of 222,960 candidates as both stable and that have an HSE Eg that is relevant for photovoltaic applications. For this subset, only 8 have been reported to date, indicating that 40 compounds remain unexplored to the best of our knowledge and therefore offer opportunities for potential experimental examination.