Combining Experimental and Theorical Data for Accurate Predictions and Mechanistic Insights in Heterogeneous Catalysis

TH Department Seminar
Date: Nov 9, 2023
Time: 02:00 PM (Local Time Germany)
Speaker: Dr. Johannes Voss
SLAC National Accelerator Laboratory, CA, USA
Location: https://zoom.us/j/92907992306?pwd=Y3JHY0ZCb0w2M1FVUXVCNFlUTE1YZz09
Room: Meeting ID: 929 0799 2306 | Passcode: 468317
Host: TH Department

First-principles prediction of heterogeneous catalytic performance is challenging due to the complexity of real catalysts and their evolution over time. But even for simple model catalyst surfaces, chemical accuracy (or sufficient accuracy to discriminate the rates of many potentially competing mechanisms) in predictions of reaction energetics on transition metal surfaces is lacking due to difficulties in simultaneously describing metallic states in the catalyst and molecular adsorbate states and the interactions between them.

Here, I present attempts at addressing these issues in atomistic-level catalyst modeling by integrating simulations with experimental benchmark data. I will discuss machine learning-based models which are trained on quantum chemistry data for gas phase and on experimental data for bulk and surface chemistry. These models help at predicting the electronic structure of surface chemical bonds on transition metals and of transition metal oxide chemistry.

Finally, I will show that, even lacking chemically accurate simulations for surface chemistry, the combination of theory and experiment has the potential to discriminate catalytic reaction pathways. For a range of strong to weak interactions of carbon species on metal surfaces, I will present first-principles simulations that allowed us to deduce energy transfer mechanisms from observed ultrafast x-ray adsorbate core-edge spectral evolution in optical pump-XFEL probe experiments.