Learning from Fullerenes and Predicting for Y6: Machine Learning and High-Throughput Screening of Small Molecule Donors for Organic Solar Cells

In recent years, research on the development of organic solar cells has increased significantly. For the last few years, machine learning (ML) has been gaining the attention of the scientific community working on organic solar cells. Herein, ML is used to screen small molecule donors for organic solar cells. ML models are fed by molecular descriptors. Various ML models are employed. The predictive capability of a support vector machine is found to be higher (Pearson's coefficient = 0.75). The best small donors with fullerene acceptors are selected to pair with Y6. New small molecule donors are also designed taking into account quantum chemistry principles, using building units that are searched through similarity analysis. Their energy levels and power conversion efficiencies (PCEs) are predicted. Efficient small molecule donors with PCE > 13% are selected. This design and discovery pipeline provides an easy and fast way to select potential candidates for experimental work.

Automated summary

In recent years, there has been significant research on the development of organic solar cells, with machine learning being used to screen and predict the performance of small molecule donors. The study found that a support vector machine had higher predictive capability and selected efficient small molecule donors for further experimentation. This approach of using machine learning and quantum chemistry principles provides a fast and reliable way to identify potential candidates for organic solar cell development.