Fragment-LEGO Strategy toward Seven-Ring-Fused Acceptors for Organic Solar Cells

Abstract

As one major type of non-fullerene acceptors, linear seven-ring-fused derivatives (7RF) integrated organic solar cells (OSC) have displayed superior power conversion efficiencies (PCE). Despite significant synthetic progress, the development of new 7RF from myriads of candidates by trial and error is still expensive. Here, with the aid of machine learning and density functional theory computation, nineteen 7RF acceptors are proposed for OSC applications with predicted PCE ≥ 16%. In addition, we show that the ridge regressor model exhibits the best predictive performance on PCE with the Pearson's correlation coefficient of 0.69 after leave-one-out cross validation. By using the Shapley additive explanation method, we reveal the key fragments of 7RF that are most conducive and adverse to achieve high PCE. This work exhaustively explores the chemical space of the 7RF category based on the fragment-LEGO strategy and suggests new 7RF to be synthesized for developing high-performance OSC.