SKM 2021 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
MM: Fachverband Metall- und Materialphysik
MM 5: Topical Session Interface Dominated Phenomena - Poster
MM 5.8: Poster
Dienstag, 28. September 2021, 10:00–12:45, P
Sampling the parameter space of grain boundaries with a sequential sampling technique - Atomistics meets statistics — •Timo Schmalofski1, Martin Kroll2, Holger Dette2, and Rebecca Janisch1 — 1ICAMS, Ruhr-University Bochum, 44780 Bochum, Germany — 2Department of Mathematics, Ruhr-University Bochum, 44780 Bochum, Germany
The grain boundary energy is a function of five degrees of freedom (DOF). Two DOF describe the inclination of the grain boundary plane and the remaining three DOF the misorientation between both grains (one for the misorientation angle and two for the rotation axis). The sampling of this grain boundary energy space or even subspaces of it has been shown to be very challenging due to the so-called "cusps", steep energy minima at special misorientations and/or inclinations. Several approaches have been tried to sample energy subspaces, but they mostly need a large amount of datapoints for sufficient accuracy, and in addition an a-priori knowledge of the positions of the energy cusps. Therefore, in this work, statistical methods are combined with atomistic simulations and a sequential sampling technique is designed. In this presentation this technique will be introduced and compared to a regular sampling technique, to prove its advantages when sampling a whole subspace with a minimal amount of datapoints and discovering unknown cusps automatically. The presentation will focus on the sampling of only one dimensional subspaces of symmetrical tilt grain boundaries to prove the concept, but the application can be generalized towards multidimensional subspaces.