Active Discovery of Donor:Acceptor Combinations For Efficient Organic Solar Cells

Abstract

The structural flexibility of organic semiconductors offers vast a search space, and many potential candidates (donor and acceptor) for organic solar cells (OSCs) are yet to be discovered. Machine learning is extensively used for material discovery but performs poorly on extrapolation tasks with small training data sets. Active learning techniques can guide experimentalists to extrapolate and find the most promising D:A combination in a significantly small number of experiments. This study uses an active learning technique with a predictive random forest model to iteratively find the most optimal D:A combinations in the search space using various acquisition functions. Active learning results with five different acquisition functions (MM, MEI, MLI, MU, and UCB) are compared. Results reveal that acquisition functions that combine exploitation and exploration (MEI, MLI, and UCB) perform far better than purely exploiting (MM) and purely exploring (MU) acquisition functions. Interestingly, the proposed model can overcome the bottleneck of extrapolating small training data sets and find most promising D:A combinations in relatively fewer experiments.

Directed Message Passing Neural Network for Predicting Power Conversion Efficiency in Organic Solar Cells

“…The integration of ML models into the field of OSCs has opened up new avenues for exploring the vast chemical space with greater efficiency and cost-effectiveness. − With the rapid advancement of ML techniques, scientists are now empowered to delve deeper into the properties of chemical systems, taking advantage of the abundance of data, improved algorithms, and exponential increases in computational power . In the field of OSCs, various ML models have been used for predicting key performance metrics such as PCE, − short circuit current density ( J SC ), open-circuit voltage ( V OC ), ,,, fill factor, , non-radiative voltage loss (Δ V NR ), and frontier molecular orbitals (FMO). , High-throughput screening has also become a valuable tool for identifying promising OSC candidates.…”

Comparing the structures and photophysical properties of two charge transfer co-crystals

“…3,[11][12][13][14] High-throughput screening and machine learning methods are also commonly used to identify candidate materials. [15][16][17][18][19] These methods screen combinations of donor and acceptor molecules to predict materials' bulk physical properties from molecular metrics based on geometric and energetic parameters. 20,21 Machine learning and high-throughput screening approaches rely on fundamental studies of materials that employ crystal structure determination, spectroscopic characterization, and electronic structure characterization.…”

Comparing the Structures and Photophysical Properties of Two Charge Transfer Co-crystals

“…3,[11][12][13][14] Modern approaches to exploring chemical space have also emerged, such as high-throughput screening and machine learning. [15][16][17][18][19] These methods screen combinations of donor and acceptor molecules to predict materials' bulk physical properties from molecular metrics such as geometric and energetic parameters. 20 Machine learning and high-throughput screening approaches are continually improving via inclusion of new structure-function relationships, driving a need for fundamental studies that identify such relationships through crystallographic structure determination, spectroscopic characterization, and electronic structure characterization.…”