SKM 2023 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
MM: Fachverband Metall- und Materialphysik
MM 31: Data Driven Materials Science: Big Data and Work Flows – Machine Learning
MM 31.9: Vortrag
Mittwoch, 29. März 2023, 18:00–18:15, SCH A 251
FAIR Modelling Recipes for High-Throughput Screening of Metal Hydrides — •Kai Sellschopp1, Philipp Zschumme2, Michael Selzer2, Claudio Pistidda1, and Paul Jerabek1 — 1Institute for Hydrogen Technology, Helmholtz-Zentrum Hereon, Geesthacht, Germany — 2IAM - Microstructure Modelling and Simulation, Karlsruhe Institute of Technology, Karlsruhe, Germany
A simple modelling recipe for calculating the hydrogenation enthalpy of metal hydrides with ab-initio methods is presented. It consists of the everyday tasks of a computational materials scientist: relaxing a structure, optimising its volume, and calculating vibrational energies. The corresponding workflow is implemented in the framework of KadiStudio[1], where the scientific process is broken down into simple input-processing-output (IPO) tasks. The approach allows to track the inputs and outputs, to easily re-use the modular tasks in other workflows, and to share the workflow as a simple bash or python script. Therefore, not only the generated research data, but also the workflow itself fully comply with the FAIR data principles. As a first application, the recipe is employed to test how the different ingredients of an ab-initio calculation (e.g. xc-functional) affect the accuracy of predicting hydrogenation enthalpies. This helps to make better choices for studying this class of materials in the future and to judge the uncertainty in existing data. Furthermore, the standardized workflow enables a reliable high-throughput screening of new candidate materials for high-density hydrogen storage at near ambient conditions.
[1] L. Griem, et al., Data Science Journal 21, 16 (2022)