Band gap prediction for large organic crystal structures with machine learning

Large datasets of ab initio calculations have enabled many pioneering studies of machine learning applied to quantum-chemical systems. For example, machine learning models already achieve chemical accuracy on the popular QM9 dataset with small organic molecules. Here, we present a new, more challenging dataset of 12,500 large organic crystal structures and their corresponding DFT band gap, freely available at https://omdb.diracmaterials.org/dataset. The complexity of the organic crystals in this dataset, which have on average 85 atoms per unit cell, makes it a challenging platform for machine learning applications. We run two recent machine learning models, kernel ridge regression with the Smooth Overlap of Atomic Positions (SOAP) kernel and the deep learning model SchNet, on this new dataset and find that an ensemble of these two models reaches mean absolute error (MAE) of 0.361 eV, which corresponds to a percentage error of 12% on the average band gap of 3.03 eV. The models also provide chemical insights into the data. For example, by visualizing the SOAP kernel similarity between the crystals, different clusters of materials can be identified, such as organic metals or semiconductors. Finally, the trained models are employed to predict the band gap for 260,092 materials contained within the Crystallography Open Database (COD) and made available online so the predictions can be obtained for any arbitrary crystal structure uploaded by a user.