Relating the Coverage Dependence of Oxygen Adsorption on Au and Pt Fcc(111) Surfaces Through Adsorbate-Induced Surface Electronic Structure Effects
Spencer D. Miller and John R. Kitchin, “Relating the coverage dependence of oxygen adsorption on Au and Pt fcc(111) surfaces through adsorbate-induced surface electronic structure effects”, Surface Science 603(5), 794-801 (2009). https://doi.org/10.1016/j.susc.2009.01.021
This summary was written by a language model from the paper and has not been checked line by line. Rely on the published paper for details.
Question
Adsorption energies change with how crowded a surface is, yet kinetic models and simple desorption analyses often ignore that dependence or treat it as linear. Oxygen on Pt(111) and on Au(111) look very different experimentally: Pt dissociates O2 readily, while Au barely adsorbs oxygen and forms surface oxides at low coverage. This paper asks whether the coverage dependence of chemisorbed atomic oxygen on these two surfaces can be explained by a common electronic structure mechanism, and whether that shared mechanism makes the energies on one metal predictable from the other (Miller 2009).
Approach
The authors used DFT (VASP, PAW, PW91) on five-layer (111) slabs with the bottom three layers fixed and more than 10 Å of vacuum. Computed lattice constants were 4.18 Å for Au and 4.00 Å for Pt. Cutoffs were 425 eV (Pt) and 475 eV (Au) with k-point density equivalent to 12x12x1 for a 1x1 cell, and adsorption energies were converged to about 0.02 eV/O on Pt and 0.03 eV/O on Au. Oxygen was placed only in fcc hollow sites, which were favored over hcp at full monolayer by 0.24 eV (Pt) and 0.15 eV (Au). Sixteen non-zero-coverage configurations in (1x1), (2x1), (2x2), (3x2), (√3x√3) and (√3x2) cells spanned 0.17 to 1.00 ML. Energies were referenced to gas-phase O2, surface free energies versus oxygen chemical potential were built with atomistic thermodynamics, and atom-projected densities of states were analyzed through the first two moments of the d band (center and width) and the d-band filling.
Key findings
Over 0.17 to 1.00 ML the oxygen adsorption energy weakens by about 1.5 eV/O on Au(111) and by more than 1.0 eV/O on Pt(111). All Pt configurations are exothermic relative to O2 (from about -1.86 eV/O at 0.17 ML to -0.82 eV/O at 1 ML), whereas several high-coverage Au configurations are unstable relative to O2. For the most stable configuration at each coverage the dependence is nearly linear, flattening toward a constant dilute-limit value at low coverage; less stable configurations at a given coverage tend to be locally dense islands. The Au energies agree well with earlier work using the same functional, and the Pt energies are offset by roughly 0.3 eV from a prior ultrasoft-pseudopotential study with similar slopes. The atomistic thermodynamics analysis shows that on Au only a p(2x2) 0.25 ML phase becomes favorable before bulk Au2O3 formation, while on Pt the p(2x2) and p(2x1) phases cover most of the chemical potential range, with p(2x3) structures appearing briefly at intermediate coverage.
On both metals oxygen broadens the surface-atom d band through adsorbate-metal orbital overlap while the d-band filling stays roughly constant. The d-band center and width are linearly correlated, which the authors rationalize with a rectangular band of fixed filling: a wider band must shift down to keep the same number of electrons. The adsorption energy varies nearly linearly with the d-band center on both metals, as the Hammer-Nørskov model predicts, with more scatter and a steeper slope on Au, attributed to stronger Pauli repulsion from its filled d band.
Plotting Au energies against Pt energies for the same configurations gives a near-linear “configurational correlation” with a least-squares slope of 1.26, close to the ratio (1.28) of the coverage slopes of the lowest-energy configurations on Au (1.62) and Pt (1.27). Points sit on average about 0.05 eV from the line (maximum 0.13 eV), and the correlation holds even for non-ground-state configurations. The one notable exception is an inversion of energy ordering between (√3x√3)-O and (3x1)-O, of about 0.1 eV in opposite directions on the two metals. Fitting the correlation to just two configurations (1 ML and (3x2)-O at 0.17 ML) predicted the remaining 14 configurations with mean absolute errors of 0.033 eV for Pt (maximum 0.125 eV) and 0.038 eV for Au (maximum 0.102 eV), less than twice the DFT convergence uncertainty.
Significance
The paper ties coverage-dependent oxygen binding to a concrete electronic structure chain (orbital overlap, d-band broadening, downward d-band shift at fixed filling, weaker bonding) and shows that two chemically dissimilar metals share it. The configurational correlation it introduces suggests that a full coverage dependence on a new metal could be obtained from only a handful of calculations once one reference metal is mapped. The authors limit the claim to geometrically similar sites and d-band dominated bonding, expecting failure for mixed site types, reconstructing surfaces, or metals like Al. Later group papers test exactly this generalization to other adsorbates and other metals.