DFT and machine learning for predicting hydrogen adsorption energies on rocksalt complex oxides

The prediction of hydrogen adsorption energies on complex oxides by integrating DFT calculations and machine learning is considered. In particular, 14 descriptors for electronic and geometric properties evaluation are adapted within a 336 hydrogen adsorption energy dataset created. Supervised learning techniques were explored to establish an accurate predictive model. With the deep neural network results, a MAE of about 0.06 eV is achieved. This research highlights the synergistic potential of DFT and machine learning for accelerating the exploration of materials for catalysis.