Molecular Design and High‐Throughput Virtual Screening of Electron Donor and Non‐fullerene Acceptors for Organic Solar Cells

• DOI: 10.1002/solr.202400370
• Publication Date: 2024-07-01T04:59:38Z
• Authors: Rui Cao; Cai‐Rong Zhang; Ming Li; Xiao‐Meng Liu; Mei‐Ling Zhang; Ji‐Jun Gong; Yu‐Hong Chen; Zi‐Jiang Liu; You‐Zhi Wu; Hong‐Shan Chen
• Keywords: High-Throughput Screening

The complicated trilateral relationships among molecular structures, properties, and photovoltaic performances of electron donor acceptor materials hinder the rapid improvement power conversion efficiency (PCE) organic solar cells (OSCs). Herein, database 310 non‐fullerene pairs is constructed 39 structure descriptors are selected. Four kinds machine learning (ML) algorithms random forest (RF), extra trees regression, gradient boosting regression trees, adaptive applied to predict parameters. coefficient determination, Pearson correlation coefficient, mean absolute error, root square error adopted evaluate ML performance. results show that RF model exhibits best prediction accuracy. Gini important analysis suggests fused ring aromatic heterocycles critical fragments in determining PCE. unit sets by cutting each molecules database. 31 752 D‐π‐A‐π type 5 455 164 A‐π‐D‐π‐A designed recombination units, 173 212 367 328 donor–acceptor generated combining newly molecules. Based on predicted PCE using trained model, 42 exhibit > 16%, which highest 16.24%.