Fe-based catalysts are highly selective for the hydrodeoxygenation of biomass-derived oxygenates but are prone to oxidative deactivation. Promotion with a noble metal has been shown to improve oxidative resistance. The chemical properties of such bimetallic systems depend critically on the surface geometry and spatial configuration of surface atoms in addition to their coverage (i.e., noble metal loading), so these aspects must be taken into account in order to develop reliable models for such complex systems. This requires sampling a vast configurational space, which is rather impractical using density functional theory (DFT) calculations alone. Moreover, "DFT-based" models are limited to length scales that are often too small for experimental relevance. Here, we circumvent this challenge by constructing DFT-parametrized lattice gas cluster expansions (LG CEs), which can describe these types of systems at significantly larger length scales. Here, we apply this strategy to Fe(100) promoted with four technologically relevant precious metals: Pd, Pt, Rh, and Ru. The resultant LG CEs have remarkable predictive accuracy, with predictive errors below 10 meV/site over a coverage range of 0 to 2 monolayers. The ground state configurations for each noble metal were identified, and the analysis of the cluster energies reveals a significant disparity in their dispersion tendency.
© 2024 The Authors. Published by American Chemical Society.