Machine Learning‐Driven Design of Unfused Small Molecule Acceptors

The development of organic photovoltaics (OPVs) depends on the properties of small molecule acceptor (SMA). Since OPV devices can be commercialized, unfused‐backbone SMAs were developed because they are easier to synthesize than other types. Machine learning (ML) models are used to predict the efficiency of the OPV devices. Unfused‐backbone SMAs are compiled into a large dataset, and their efficiencies are predicted using pre‐trained ML models. The chemical similarity between these SMAs was analyzed using clustering analysis with the help of chemical fingerprints. Furthermore, the synthetic feasibility of these SMAs was studied. It was observed that majority of these SMAs are easier to synthesize.