Using automated experiments and machine learning to identify composition-property relationships in halide perovskites
作者
Abigail R. Hering,Mansha Dubey,Elahe Hosseini,Meghna Srivastava,Yu An,Juan‐Pablo Correa‐Baena,Houman Homayoun,Marina S. Leite
出处
期刊: [SPIE] 日期:2025-03-19卷期号:: 8-8
标识
DOI:10.1117/12.3042140
摘要
We use a custom-built, high throughput, in situ photoluminescence (PL) characterization setup to collect sufficient data to train a variety of machine learning (ML) models and to quantify the degradation of several different perovskite compositions. We compare forecasts of PL of several CsyFA(1-y)Pb(BrxI(1-x))3 perovskites in response to temperature cycling with extreme gradient boosting (XGBoost) algorithms. These models enable predictions of PL figures of merit (peak location, area, intensity, and full-width half max (FWHM)), with up to 98% accuracy. Furthermore, we create a generalized model that can predict the PL behavior of ten compositions unseen during model training. The relative feature importance and correlations between environmental inputs and optical performance show that composition dominates sample degradation.