[17p-P04-1~10] 23 Joint Session N "Informatics" (Poster)

English Presentation

[17p-P04-2] Neural Network Potentials for Accurate Activation Energy Calculation of Proton Hopping

〇Thang Duy Dang¹, Naoki Matsumura¹, Yuta Yoshimoto¹, Yasufumi Sakai¹, Atsushi Ishikawa², Junichiro Otomo² (1.Fujitsu Limited, 2.Tokyo Institute of Technology)

Keywords:

Proton Conductor,Neural Network Potentials,Molecular Dynamics Simulation

Perovskites have recently attracted considerable attention as solid proton conductors for a variety of electrochemical applications. One of the key factor that is the calculation of Activation Energy (E_a) of proton hopping for understanding proton displacement. However, there are the vast compositional and configuration spaces by using the first principles calculations (DFT) for capturing the activation energy. Neural Networks Potentials (NNP) in Molecular Dynamics (MD) is emerging as an important role for substituting DFT, especially in measuring the activation energy of proton hopping in reliable results. In this work, we demonstrate the way to leverage NNP for calculating efficiently E_a of proton hopping that reach to the same accuracy as DFT calculation.