Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
TT: Fachverband Tiefe Temperaturen
TT 54: 2D Materials IV: Graphene (joint session O/TT)
TT 54.11: Vortrag
Mittwoch, 20. März 2024, 17:45–18:00, MA 005
Let's Go on Graphs: X-ray Absorption Spectroscopy of Graphene Oxide using Graph Neural Networks — •Samuel J. Hall1, Kanishka Singh1,2, Qinyuan Zhou1,2, and Annika Bande1,3 — 1Helmholtz-Zentrum Berlin, Germany — 2Institute of Chemistry and Biochemistry, Freie Universität Berlin, Germany — 3Leibniz Universität Hannover, Germany
Graphene oxide (GO) materials, while promising for various applications, can be difficult to fully understand and predict its properties due to the highly irregular molecular structure arising from several oxygen functionalizations across the surface. X-ray absorption spectroscopy (XAS) experiments and simulations can help provide valuable insight by characterizing the electronic structure of materials. However, there are problems with complex spectra being hard to interpret and the prohibitive computational simulation cost for large extended systems. We have developed a machine learning model utilizing graph neural networks (GNN) based on a database of 319 GO-derivative molecules, consisting of 7984 individual atomic XAS spectra calculated with time-dependent density functional theory (TDDFT), that can accurately simulate XAS spectra at a significant lower cost. We show how the model can learn through either the combined spectra of the GO-derivative molecules or the individual atomic spectra to make predictions based on either the larger global environment or the local atomic environment and can further be applied to larger extended systems.
Keywords: Graphene Oxide; Graph Neural Networks; X-ray Absorption Spectroscopy