Prediction of various adsorption energies of materials using artificial neural networks

Abstract

Developing catalysts is one of the important research topics in materials science. The various adsorption energies of catalyst materials are closely related with with the catalyst performance. In particular, adsorption energies of *O, *OH, and *OOH are closely related with the catalysts used in fuel cells. We propose an artificial neural networks model (ANN) to predict E(*O), E(*OH), the value of E(*OH) subtracted from E(*O), and E(*OOH) of materials for developing high-performance catalysts. The proposed ANN models show better performance than other machine learning models. Also, the ANN model is much faster than DFT calculation and hence is useful for screening massive materials and selecting candidates for catalysts.