Abstract

Abstract Non-fullerene materials have attracted attention as high-performance molecular acceptors in organic solar cells (OSCs). A proper understanding of the energy level alignment between donors and non-fullerene acceptors is crucial for photoactive materials selection in designing high-performance non-fullerene OSCs. However, the quantitative assessment for the proper selection of donors and non-fullerene acceptors is still rarely studied, which is seen as time-consuming and complicated tasks. In this study, the optimized Random Forest model based on the electronic descriptors (e.g., highest occupied molecular orbitals levels, lowest unoccupied molecular orbitals levels, and band gap) provides the high predictive power, reaching the coefficient of determination (R2) of 0.85 and 0.80 for the training set and testing set, respectively. The use of machine learning approach benefits the development of non-fullerene OSCs in two ways: (1) it helps to extract complex correlation between various descriptors and device performance, and (2) it indicates that the band gap of acceptors is the more critical feature for improving the efficiency of non-fullerene OSCs. The machine-learning model for predicting the efficiency of non-fullerene OSCs (macroscopic performance) from frontier molecular orbital energy levels of the organic materials (microscopic properties) is developed, as an important guide to design the heterojunction blends and accelerate the research for non-fullerene OSCs.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call