Berlin 2024 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
MM: Fachverband Metall- und Materialphysik
MM 64: Liquid and Amorphous Materials IV
MM 64.5: Vortrag
Donnerstag, 21. März 2024, 17:45–18:00, C 243
Modelling amorphous forms of complex hybrid-inorganic frameworks — •Thomas C. Nicholas, Daniel F. Thomas du Toit, Andrew L. Goodwin, and Volker L. Deringer — Department of Chemistry, Inorganic Chemistry Laboratory, University of Oxford, Oxford, UK
With the continued development of efficient fitting and evaluation frameworks for machine learning potentials (MLPs), modelling complex, multi-component materials is now possible. However, for systems built up of metal nodes connected by organic linkers, such as metal-organic frameworks (MOFs), the time-scales and length-scales required to model and label representative amorphous training configurations using traditional database generation strategies (for example, ab initio molecular dynamics and iterative training) remains a challenge.
We demonstrate a two-stage approach to tackle this, focusing on modelling the amorphous form of a MOF built up from Zn nodes and imidazolate linkers (Zn[C3N2H3]2). Firstly, by exploiting the structural analogy between silica networks and this MOF, we construct a topologically and geometrically diverse database of configurations through a back-mapping scheme whereby we decorate AB2 networks with Zn nodes and imidazolate linkers. Secondly, we introduce an iterative training protocol whereby training configurations are generated using a Monte Carlo simulation refinement that seeks to minimise the difference between the computed and experimental data.
In this way, we demonstrate how our MLP better describes possible amorphous configurations.
Keywords: ML potential; amorphous; metal–organic framework