DigiMatChem
Digital Materials Chemistry is a newly established division in the Materials Chemistry Department at the Bundesanstalt für Materialforschung und prüfung (BAM), Berlin led by Prof. Dr. Janine George. It combines traditional materials science methods with modern digital tools for more efficient, sustainable, and targeted material development. Some of us are affiliated with the University of Jena. Prof. Dr. Janine George is Professor of Materials Informatics there.
news
| Feb 13, 2026 | New preprint on evaluating quantum‑chemical bonding descriptors impact in machine learning material properties We systematically demonstrate that quantum‑chemical bonding descriptors improve machine‑learning models and yield more accurate, interpretable predictions of bonding‑related material properties, including thermal conductivity. Find the preprint here: https://arxiv.org/abs/2602.12109 |
|---|---|
| Dec 15, 2025 | New preprint on assessing novelty and quality of crystal structures generated by generative models. We developed a distributional metric for evaluating material generative models. It is based on the theory of optimal transport. The metric captures both novelty and quality of the generated structures. Find the preprint here: https://arxiv.org/abs/2512.09514 |
| Oct 31, 2025 | New preprint predicting and understanding thermal conductivities of sulfur-based argyrodites We studied Ag₈TS₆ (T = Si, Ge, Sn) argyrodites using two modeling approaches—a Grüneisen-based phonon lifetime model and machine-learned interatomic potentials—to predict lattice thermal cond[…] Find the preprint here: https://doi.org/10.48550/arXiv.2510.23133 |
| Sep 01, 2025 | Namrata Jaykhedkar joins our group as an Adolf Martens fellow!! Namrata started to work as an Adolf Martens fellow in our group. This fellowship is a postdoc fellowship awarded by BAM. She will work on appyling ML interatomic potentials to battery materials! |
| Aug 31, 2025 | New preprint on magnetism heuristics in machine learning We have published a preprint on magnetism heuristics in machine learning of magnetic properties. We show that bond angles are indeed important for magnetic properties. Please take a look at the preprint here: https://doi.org/10.26434/chemrxiv-2025-xj84d |