Regensburg 2025 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
MM: Fachverband Metall- und Materialphysik
MM 23: Phase Transformations
MM 23.2: Talk
Wednesday, March 19, 2025, 17:30–17:45, H23
From electronic structure to thermodynamic phase diagrams with automated workflows — •Sarath Menon1, Marvin Poul1, Tilmann Hickel2, Ralf Drautz3, and Jörg Neugebauer1 — 1Max Planck Institute for Sustainable Materials — 2Bundesanstalt für Materialforschung und -prüfung — 3Ruhr University Bochum
Phase diagrams are useful for understanding coexistence lines, phase stability, and phase transitions under varying thermodynamic conditions. Calculating phase diagrams involves determining the Helmholtz and Gibbs free energies of different phases and their dependence on thermodynamic state variables - a task that is both intricate and computationally demanding.
In this work, we introduce automated workflows for the calculation of Helmholtz and Gibbs free energies, incorporating configurational entropy, and provide accompanying computational tools. A key component of our approach is the alchemical transformation method, where atomic species are systematically altered along a thermodynamic path to evaluate free energy changes with composition.
We demonstrate the effectiveness of this methodology using an Atomic Cluster Expansion (ACE) machine-learning interatomic potential, parametrized using the ASSYST method, to generate unbiased ab initio structure datasets, and compute the phase diagram of the Au-Cu system. Our workflows are independent of the interatomic potential and the material system, making them readily transferable and paving the way for making the computation of thermodynamic phase diagrams a routine task in the field of atomistic simulations.
Keywords: phase transformation; machine learning; interatomic potentials; free energy