Dresden 2020 – wissenschaftliches Programm

Die DPG-Frühjahrstagung in Dresden musste abgesagt werden! Lesen Sie mehr ...

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

CPP: Fachverband Chemische Physik und Polymerphysik

CPP 113: Nanostructured Surfaces and Thin Films III: Dots, Particles, Clusters (joint session O/CPP)

CPP 113.6: Vortrag

Freitag, 20. März 2020, 11:45–12:00, WIL B321

Global Optimization of Copper Clusters on ZnO Surfaces Utilizing a Neural Network Potential — •Martín Leandro Paleico and Jörg Behler — Universität Göttingen, Theoretische Chemie, Tammannstr. 6, 37077 Göttingen, Germany

The catalyst used in the industrial synthesis of methanol is composed of large copper and zinc oxide nanoparticles. Studying the structure of this system requires a simulation method capable of handling thousands of atoms with ab initio accuracy, but with computational efficiency comparable to classical empirical potentials. To meet these requirements, a Neural Network Potential (NNP) has been trained to reproduce the potential energy surface of the system based on DFT reference calculations.

We have utilized this potential to carry out the tens of thousands of energy and force evaluations required to perform global optimization searches employing genetic algorithms. With this, we are able to optimize pure copper and binary copper-zinc clusters with up to 30 atoms on two different zinc oxide surfaces. This allows us to investigate structural and energetical trends in cluster growth and cluster-substrate interactions, as well as to identify possible active sites and their distribution in the clusters.