Dresden 2009 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Downloads | Hilfe

O: Fachverband Oberflächenphysik

O 27: Poster Session I (Methods: Scanning probe techniques; Methods: Atomic and electronic structure; Methods: Molecular simulations and statistical mechanics; Oxides and Insulators: Clean surfaces; Oxides and Insulators: Adsorption; Oxides and Insulators: Epitaxy and growth; Semiconductor substrates: Clean surfaces; Semiconductor substrates: Epitaxy and growth; Semiconductor substrates: Adsorption; Nano- optics of metallic and semiconducting nanostructures; Electronic structure; Methods: Electronic structure theory; Methods: other (experimental); Methods: other (theory); Solutions on surfaces; Epitaxial Graphene; Surface oder interface magnetism; Phase transitions; Time-resolved spectroscopies)

O 27.35: Poster

Dienstag, 24. März 2009, 18:30–21:00, P2

A High-Dimensional Neural Network Potential-Energy Surface for Zinc Oxide — •Nongnuch Artrith, Marcus Maschke, and Jörg Behler — Lehrstuhl für Theoretische Chemie, Ruhr-Universität Bochum, D-44780 Bochum, Germany

Zinc oxide is a technologically important material, for example as support in heterogeneous catalysis. In order to study a variety of interesting problems it is necessary to perform long simulations of large systems, which are beyond the capabilities of established methods like density-functional theory. Recently, artificial Neural Networks (NN) trained to first-principles data have shown to provide accurate potential energy surfaces for condensed systems, which are computationally very efficient to evaluate. However, so far these potentials have been applicable only to elemental systems. We show that by including physically derived terms it is possible to extend the NN methodology to multicomponent systems. The capabilities of the method are illustrated by first applications to zinc oxide.