New developments in Computational Chemistry

Bastian Skjelstad is the new group leader at the MPI für Kohlenforschung

Dr. Bastian Skjelstad is the new group leader at the Max Planck Institut für Kohlenforschung. His research focuses on realistic, computer-aided methods for analyzing chemical reaction mechanisms, and he uses machine learning and automated techniques to better understand structural flexibility, dynamics, and solvent effects.

For several weeks now, the Max Planck Institut für Kohlenforschung has been home to a new group leader: Dr. Bastian Skjelstad from Norway has joined Prof. Frank Neese’s Department of Molecular Theory and Spectroscopy.

“During a stay in Japan as part of my bachelor’s studies, I discovered my passion for computational chemistry,” Skjelstad says. He therefore decided to pursue this field during his master’s studies. “During my studies, which focused on computational studies of reaction mechanisms in homogeneous catalysis, I realized that it’s not so easy to determine whether the calculated mechanism is indeed the mechanism operating in reality,” he says.

This realization led the young chemist back to Japan, where he pursued his Ph.D. at Hokkaido University and subsequently completed a postdoc under the supervision of Prof. Satoshi Maeda, specializing in automated reaction mechanism discovery methods. These automated methods are unbiased and help avoid human errors, such as overlooking possible reaction mechanisms. They can therefore provide a more comprehensive picture of the possible pathways a chemical reaction can take. From 2024 to 2026, he was a Marie Skłodowska-Curie Actions Postdoctoral Fellow at the University of Oxford, where he worked on machine-learned interatomic potentials to simulate chemical reactions in solution, before moving to Mülheim this summer.

Calculations as close to reality as possible

His goal is clear: “I aim to achieve calculations as close to reality as possible,” says Skjelstad. “The Schrödinger equation can be solved very accurately for most molecular systems, and oftentimes, a major source of error is not the quantum chemical calculations themselves, but how we account for structural flexibility and dynamics as well as solvent effects.” For example, how does the choice of solvent influence a particular reaction? Experimental chemists routinely screen reactions in multiple solvents, which may each yield different results, but these effects are currently difficult to explain using conventional quantum chemical calculations. In his research, Skjelstad makes use of artificial intelligence—more specifically, machine learning.

“My approach is not as computationally expensive as quantum chemical calculations, which typically run on high-performance computers and require a lot of energy and time,” says the Norwegian. Compared with conventional theoretical chemistry, he explains, his approach can be thought of as taking a series of photos rather than an entire video of how molecules react. “The machine learning models enable you to deduce what happens between two images. The models can sometimes even make such predictions for molecules they were not trained on. In my work, the computer is the one doing the deducing.”

At the MPI, Skjelstad hopes to have many inspiring conversations with his colleagues, both in the theory department and among the experimental chemists, particularly those working in homogeneous catalysis and transition metal chemistry. “I truly believe we can learn a great deal from one another,” he says.

Bastian Skjelstad has moved to Mülheim an der Ruhr with his wife and son. “We very much like it here and enjoy going to Witthausbusch to visit the animals on the weekends, for example,” he says. And once he has settled in, he hopes to resume his hobby: playing table tennis.