Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

O: Fachverband Oberflächenphysik

O 40: Solid-Liquid Interfaces

O 40.1: Poster

Dienstag, 17. März 2015, 18:15–21:00, Poster A

Towards the Simulation of the Water-Zinc Oxide Interface Employing Neural Network Potentials — •Vanessa Quaranta, Tobias Morawietz, and Joerg Behler — Lehrstuhl für Theoretische Chemie, Ruhr-Universität Bochum, D-44780 Bochum, Germany

The reliability of molecular dynamics and Monte Carlo simulations crucially depends on the quality of the underlying potential-energy surface (PES). In recent years, artificial neural networks (NNs) have emerged as a new method to provide accurate PESs for a variety of systems. NNs are flexible functions able to learn the topology of a PES from a set of reference electronic structure calculations. Here, we report first results for a NN potential based on density-functional theory constructed to describe the water-ZnO interface. We find that the NN potential is able to provide a wide range of structural and energetic properties of water, ZnO as well as of interface structures with close to first-principles accuracy.