Berlin 2018 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
O: Fachverband Oberflächenphysik
O 37: Electronic-Structure Theory: General II
O 37.1: Hauptvortrag
Dienstag, 13. März 2018, 14:00–14:30, MA 141
Unraveling the structure and dynamics at solid-liquid interfaces by machine learning potentials — Matti Hellström1, Vanessa Quaranta2, and •Jörg Behler1 — 1Theoretische Chemie, Universität Göttingen, Germany — 2Theoretische Chemie, Ruhr-Universität Bochum, Germany
Solid-liquid interfaces pose a significant challenge for atomistic simulations. The very different interactions and bonding situations in water and in solid surfaces are best described by electronic structure methods, which can also take into account the dissociation and recombination of water molecules at the interface. On the other hand, long simulations of large systems are required to obtain converged properties of the liquid phase, which is often computationally very demanding. Machine learning potentials offer a solution to this problem by combining a first principles quality description of the potential-energy surface with the efficiency of simple empirical potentials. In this talk, recent results for the interaction of water with metal and oxide surfaces will be presented, which have been obtained in molecular dynamics simulations employing high-dimensional neural network potentials (NNPs), a typical class of machine learning potentials. NNPs are reactive and thus allow to observe proton transfer processes at the interface, which also play an important role in highly concentrated electrolyte solutions.