SurfaceScience21 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
O: Fachverband Oberflächenphysik
O 58: Mini-Symposium: Molecular scale investigations of liquid-vapor interfaces II
O 58.1: Invited Talk
Wednesday, March 3, 2021, 10:30–11:00, R1
Simulating interfacial water with neural network potentials — •Christoph Dellago1, Oliver Wohlfahrt1, and Marcello Sega2 — 1University of Vienna, Faculty of Physics, Vienna, Austria — 2Forschungszentrum Jülich, Helmholtz Institute Erlangen-Nürnberg for Renewable Energy, Nürnberg, Germany
The strong, directed network of hydrogen bonds of water lies at the heart of its rich phase diagram and numerous anomalous properties and is responsible for the peculiar structure of its liquid/vapor interface. Detailed insights into the molecular structure and dynamics of bulk and interfacial water have been gleaned from first principles simulations, which provide an unbiased description of water at the atomic level and yield information on the underlying molecular forces. However, the computational cost of such simulations is still daunting, particularly for interfacial systems that require relatively large systems sizes as they suffer from finite size effects more than bulk systems. Recently, artificial neural networks have been proposed to overcome these limitations, yielding the accuracy of first principles simulations at a fraction of their computational cost. In this talk, I will report on the structural and thermodynamic properties of the water/vapor interface obtained using a neural network potential trained for bulk and interfacial water. In particular, I will discuss the liquid/vapor coexistence curve and its size dependence as well as the structure of the interface in terms of density profiles and molecular orientations.