Dresden 2011 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

O: Fachverband Oberflächenphysik

O 14: Clean surfaces: Metals, semiconductors, oxides and insulators I

O 14.4: Talk

Monday, March 14, 2011, 15:45–16:00, WIL B122

Construction of an Interatomic Potential for Zinc Oxide Surfaces by High-Dimensional Neural Networks — •Nongnuch Artrith, Tobias Morawietz, and Jörg Behler — Lehrstuhl für Theoretische Chemie, Ruhr-Universität Bochum, D-44780 Bochum, Germany

Zinc oxide (ZnO) is a technologically important material with many applications, e.g. in heterogeneous catalysis. For theoretical studies of the structural properties of ZnO surfaces, defects, and crystal structures it is necessary to simulate large systems over long time-scales with sufficient accuracy. Often, the required system size is not accessible by computationally rather demanding density-functional theory (DFT) calculations. Recently, artificial Neural Networks (NN) trained to first principles data have shown to provide accurate potential-energy surfaces (PESs) for condensed systems. We present the construction and analysis of a NN PES for ZnO. The structural and energetic properties of bulk ZnO and ZnO surfaces are investigated using this potential and compared to DFT calculations.