Machine‐Learning Analysis of Small‐Molecule Donors for Fullerene Based Organic Solar Cells

In recent years, development in organic solar cells speeds up and performance continuously increases. From the last few years, machine learning gains fame among scientists who are researching on organic solar cells. Herein, machine learning is used to screen the small‐molecule donors for organic solar cells. Molecular descriptors are used as input to train machine models. A variety of machine‐learning models are tested to find the suitable one. Random forest model shows best predictive capability (Pearson's coefficient = 0.93). New small‐molecule donors are also designed from easily synthesizable building units. Their power conversion efficiencies (PCEs) are predicted. Potential candidates with PCE > 11% are selected. The approach presented herein helps to select the efficient materials in short time with ease. Machine‐learning analysis is performed to identify descriptors that influence the performance of OSCs. The power conversion efficiencies (PCEs) are predicted using a trained machine‐learning model and top five candidates with PCE > 11% are selected. Machine‐learning requires marginal computational cost to assist the experimentalists to synthesize efficient small‐molecule donors.