Using Computational Tools to Accelerate Discovery of High-Efficiency Solar Cell Materials

High-efficiency organic photovoltaic (OPV) materials made with non-fullerene electron acceptors are exciting because of their potential for realizing solar power that pays for its capital cost in weeks rather than decades. Optimizing these materials is challenging because of the number of (a) possible non-fullerene acceptors, (b) polymers to mix them with, and (c) ways these two compounds can be processed together to make an OPV device. Here we develop computational tools for screening candidate OPV blends that robustly assemble morphologies that convert sunlight into electricity using reproducible high-performance computing (HPC) workflows that enable programmatic specification of OPV structure prediction. We develop open source tools: the Molecular Simulation Design Framework (MoSDeF) for general molecular simulation infrastructure, PlanckTon for launching molecular dynamics simulations of blends, and MorphCT for predicting charge mobilities of the equilibrated morphologies. Specific incorporation of SMARTS and SMILES chemical grammars, new binary file formats, and modular open source quantum chemical calculation engines have contributed to our simplified, more reproducible workflows. This has also helped with containerization of both MorphCT and PlanckTon, which makes them easier to deploy on HPC resources. We briefly describe collections of hundreds of simulation jobs and the understanding of OPV physics enabled by these software developments.