(245f) Coverage-Dependent Site and Adsorbate Configurational Correlations on (111) Surfaces of Late Transition Metals

Correlations between adsorption energies allow for the prediction of a large set of adsorption energies from a few DFT calculations. One popular correlation that significantly reduces the number DFT calculation one needs to perform to evaluate reaction thermodynamics are correlations between adsorption energies of different adsorbates on a single surface -- oftentimes called scaling relationships. Another equally useful but uncommonly investigated correlation are correlations between adsorption energies on geometrically different sites. These site correlations would reduce the amount of calculations one needs to completely assess adsorption behavior on a single surface.

One aspect of either of these correlations that have not been studied are their coverage dependence. We test the robustness of these two types of correlations by performing over 800 density functional theory calculations of the adsorption of H, C, N, O, F, S, Cl, and Br atomic adsorbates on Cu, Rh, Pd, Ag, Ir, Pt, and Au fcc (111) surfaces at a variety of coverages and configurations. We focus our analysis on the coverage dependence of correlations between adsorption energies on different sites and of different adsorbates. We assess geometric requirements for the existence of correlations by looking at the quality of correlations between adsorption on different sites. We then use linear regression techniques to assess the coverage dependence of correlations between different adsorbates by looking at adsorption energies between elemental adsorbates of the same period or the same group at varying coverages.

We find that correlations between both sites and adsorbates are coverage dependent. Though correlations between geometrically similar sites are independent of coverage, coverage plays a key role in reducing the quality of correlations between adsorption energies at very geometrically different sites. For correlations between adsorbates of the same group and same period, we find that changing the coverage universally changes the correlations and in some cases destroy them. We then use a simple electron counting model to relate our results to previously discovered scaling relationships and discuss the ramifications our conclusions have on materials discovery.