Research

We develop simulation methods that are based on density functional theory and enable the calculation of precise and realistic material properties at finite temperatures. Properties – particularly important for material design and, therefore, of particular interest to us – are phase stabilities, diffusion constants, and defect properties. We develop interatomic potentials using "machine learning" to make the methods efficient. We also carry out large-scale molecular dynamics simulations with interatomic potentials, e.g., to understand the movement and interaction of defects (such as dislocations).

• Development of a method to calculate melting properties efficiently using ab initio
• Investigation of dynamically unstable systems with ab initio
• Investigation of diffusion properties in high entropy alloys (DFG-funded project)
• Development of ab initio-based methods for the prediction of phase diagrams (EU-funded project, ERC Consolidator Grant Materials 4.0)
• Calculation of the properties of superdislocations in Ni3Al with machine-learning potentials
• Calculation of Li diffusion in solid electrolytes with machine-learning potentials (within the SimTech excellence cluster)