Regensburg 2019 – wissenschaftliches Programm
O 17.6: Poster
Montag, 1. April 2019, 17:45–20:00, Poster F
Simulating the scattering of a hydrogen atom from graphene using a high-dimensional neural network potential. — •Sebastian Wille1,2, Marvin Kammler2, Martín L. Paleico3, Jörg Behler3, Alec M. Wodtke1,2, and Alexander Kandratsenka2 — 1Institute for Physical Chemistry, Georg-August University Göttingen, Germany — 2Department of Dynamics at Surfaces, Max Planck Institute for Biophysical Chemistry, Göttingen, Germany — 3Theoretical Chemistry, Georg-August University Göttingen, Germany
To fully understand atom-surface interactions, the availability of an accurate full-dimensional potential energy surface (PES) is crucial. High-dimensional neural network potentials have been shown to provide very accurate PESs for a wide range of systems. We developed a neural network potential for H-atom at a graphene sheet by fitting to density functional theory data calculated on-the-fly in ab initio molecular dynamics simulations. Based on this potential, we studied the scattering under various incidence conditions (like angle, kinetic energy, temperature) and compared the results to experimental data.