Just another WordPress site - Ruhr-Universität Bochum

Department Atomistic Modelling and Simulation
Research Group

Atomistic Simulation of Thermodynamic Properties

The research group specializes in the development and application of atomistic simulation methods to calculate thermodynamic properties of materials and construct phase diagrams.


Chairman
Dr. Sarath Menon

Research Group Leader

Room: 02-719
Tel.:
E-Mail: sarath.menon@rub.de




Research

Calculation of thermodynamic properties and phase diagrams from atomistic simulations is essential to materials discovery and design, because it enables prediction of formation energies, phase stability and free energy landscapes directly from atomic interactions.

Workflow for calculation of phase diagrams with machine learning interatomic potentials.
Workflow for calculation of phase diagrams with machine learning interatomic potentials.
ICAMS, RUB

Our group develops computational methods for the accurate evaluation of thermodynamic quantities, with particular emphasis on free energy calculations for reliable phase diagram construction. We focus on machine learning interatomic potentials, in particular the atomic cluster expansion and its extensions (ACE, GRACE) to deliver ab initio accuracy at greatly reduced computational cost.

Another key activity of the group is the development and maintenance of scientific software and computational workflows that integrate every step of the research process in a reproducible and reusable manner.

Competences

  • Molecular dynamics calculations for phase diagrams
  • Simulations of phase transitions
  • Machine learning interatomic potentials and applications
  • Reproducible computational workflows
  • Scientific software development
Members
Recent Publications

Research Examples

From electrons to phase diagrams with machine learning potentials using pyiron based automated workflows

We have developed automated workflows covering every stage of phase diagram calculations, from generation of ab initio data and parametrisation of machine learning interatomic potentials, through their validation, to the final computation of binary phase diagrams. These workflows have been applied to the atomic cluster expansion (ACE), high-dimensional neural network potentials (HDNNPs) and classical embedded-atom method (EAM) potentials.

Teaser B1
Automated free-energy calculation from atomistic simulations

We propose efficient algorithms for calculating the Helmholtz and Gibbs free energies as functions of temperature and pressure and provide the necessary computational tools. We apply these methods to compute unary phase diagrams.