Homogeneous and heterogeneous catalytic CO2 Reduction Mechanisms wIth Neural networks and Enhanced sampling – HERMINE

This project aims to develop advanced artificial intelligence (AI)-driven molecular simulations to uncover the mechanisms governing carbon dioxide (CO2) reduction, a critical process in the fight against climate change. Converting CO2 into valuable chemical products requires efficient catalysts, yet progress has been hindered by the limited understanding of the molecular interactions and energy barriers involved in catalytic CO2 reduction. While theoretical and computational models can offer insights, all current approaches simplify the role of solvents, neglecting key proton transfer and hydrogen-bonding effects observed experimentally.

The HERMINE project proposes three innovative solutions to address these challenges. First, we will develop a new simulation framework based on neural network potentials (NNPs), previously applied to simpler reactions, and extend it to electron transfer processes. Second, this framework will be used to study CO2 reduction in aqueous microdroplets, a recently observed process, but with key unresolved questions regarding its reaction mechanism and the reducing species. Third, we will explore the paradigm homogeneous electrocatalytic CO2 reduction by iron-porphyrin complexes, providing an explicit description of solvent effects that are essential for catalytic efficiency.

By integrating AI methodologies with molecular dynamics, this project will not only offer a new platform for simulating complex chemical reactions but also deliver crucial molecular-level insights into CO2 reduction. These advancements will pave the way for designing next-generation catalysts with greater efficiency, driving progress in both sustainable chemistry and climate mitigation efforts.