Taming the Configurational Explosion: Coarse Grained Modeling and Statistical Learning for Surface Structure Search

Hörmann, L. (Speaker), Jeindl, A. (Contributor), Alexander Egger (Contributor), Michael Scherbela (Contributor), Hofmann, O. (Contributor)

Activity: Talk or presentationTalk at conference or symposiumScience to science


Monolayers of organic molecules on inorganic substrates often show rich polymorphism with diverse structures in differently shaped unit cells. Determining the different commensurate structures from first principles is far from trivial due to the large number of possible polymorphs. We introduce the SAMPLE approach[1][2], which employs coarse-grained modeling in combination with Bayesian linear regression to efficiently map the entire the potential energy surface and ab initio thermodynamics to generate phase diagrams.

To this end, we first determine adsorption geometries, which isolated molecules adopt on the substrate. Secondly, we generate all possible unique substrate super cells, whose area lies within predefined boundaries. Thirdly, we combine the local adsorption geometries with each super cell to generate possible structures. For two different subsets of these structures, we calculate the adsorption energies using dispersion-corrected density functional theory (DFT). The first subset serves as training data for a Bayesian linear regression algorithm, which allows to predict the adsorption energies of all possible polymorphs. The second subset allows us to validate the prediction. Finally, we employ ab initio thermodynamics, which assumes that the adsorbed layer is in thermodynamic equilibrium with an ideal molecule gas, to generate phase diagrams.

We demonstrate the power of SAMPLE on the system of naphthalene on Cu(111). For this system we predict the adsorption energies for a large number of structures and determine the phase diagram.

[1] V. Obersteiner, M. Scherbela, L. Hörmann, D. Wegner and O. T. Hofmann, Nano Lett. 17 (7), pp 4453-4460 (2017)
[2] M. Scherbela, L. Hörmann, A. Jeindl, V. Obersteiner and O. T. Hofmann, Phys. Rev. Materials 2, 043803 (2018)
Period13 Sep 2018
Event titleÖPG Jahrestagung
Event typeConference
LocationPetersgasse 16, 8010 Graz, Austria
Degree of RecognitionInternational

Fields of Expertise

  • Advanced Materials Science