Carbon dioxide hydrogenation on copper and nickel catalysts via a conformal sampling approach

We present a computational workflow, the Conformal Sampling of Catalytic Processes (CSCP) approach, and its application to the case of heterogeneous hydrogenation/reduction of carbon dioxide (CO₂) on copper and nickel catalysts. CO₂ activation is of critical importance for a sustainable global future and one of the major reactions for which sustainable routes must be found urgently. We use the fcc(100) facet of Cu as a worked-out case, and exhaustively derive at the DFT level its reaction mechanisms. We then apply CSCP to first derive a Machine Learning Interatomic Potential (MLIP) for this initial system, and then carry over the knowledge derived on Cu(100) to a pure monometallic facet, Ni(100), to rapidly derive a MLIP for this different system, so as to test the transferability of the approach. The accuracy of the so-derived MLIPs is excellent, predicting both reaction energies and energy barriers for all the mechanistic steps of this complex reaction diagram with consistent and uniform accuracy, with a maximum discrepancy of 0.05 eV for Cu and 0.03 eV for Ni. In the exception cases in which this discrepancy is larger, we show and rationalize that this is due to a change in the reaction mechanism, where the MLIP simulations explore pathways different from the reference DFT ones, without however prejudicing their absolute accuracy. The CSCP-MLIPs are thus shown to be able to assure a transferability en par with the best physics-based models, provide alternative atomistic mechanisms of catalytic processes, and offer themselves as a tool for catalyst rational design on a process societally relevant and exhibiting significant catalytic complexity.