Rapid Prescreening of Organic Compounds for Redox Flow Batteries: A Graph Convolutional Network for Predicting Reaction Enthalpies from SMILES

Identifying interesting redox-active couples from the vastness of organic chemical space requires rapid screening techniques. A good initial indicator for worthy further investigation is heat reaction ΔH°. Traditional methods calculating this quantity, both experimental and computational, are prohibitively costly at large scale. Instead, we apply a graph convolutional network to estimate heats arbitrary redox orders magnitude faster than conventional computational methods. Our takes only SMILES strings as input, rather full three-dimensional geometries. trained on dataset atomization enthalpies approximately 45,000 hydrogenation reactions, applied separate test set 235 compounds benchmarked against reaction, produces promisingly accurate results, anticipate that methodology can be extended other RFB-relevant reactions. However, lower predictivity in regions not covered by training reinforces pivotal importance particular chemistries presented model during training.