Modification of the Surface Electronic and Chemical Properties of Pt(111) by Subsurface 3d Transition Metals
J. R. Kitchin, J. K. Nørskov, M. A. Barteau, and J. G. Chen, “Modification of the surface electronic and chemical properties of Pt(111) by subsurface 3d transition metals”, The Journal of Chemical Physics 120(21), 10240-10246 (2004). https://doi.org/10.1063/1.1737365
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
The surface properties of a bimetallic catalyst differ from those of its parent metals for two main reasons: strain, because surface bond lengths change, and the ligand effect, because neighboring atoms of a different element alter the electronic environment of a surface atom. In supported monolayers both act at once, and earlier work had explained trends among late transition metals using strain alone. This paper asks under what conditions the ligand effect alone can dominate, and isolates it by placing a single layer of a 3d metal (Ti through Ni) beneath an unstrained Pt(111) surface layer (Kitchin et al. 2004). The model is motivated by Pt segregation to the surface of many Pt alloys, reports of Pt “skins” on Pt-3d fuel cell catalysts, and the authors’ earlier experiments on subsurface Ni in Pt(111).
Approach
The authors used periodic density functional theory (DACAPO, RPBE functional, ultrasoft pseudopotentials) on four-layer Pt(111) slabs in a 2x2 cell, with all four second-layer atoms replaced by Ti, V, Cr, Mn, Fe, Co or Ni. The bottom two layers were held at the computed Pt-Pt bulk distance of 2.84 Å, so the surface Pt layer carried no lateral strain, while the top two layers relaxed. Calculations were not spin polarized, since earlier Ni-Pt work showed spin did not change the trends. The surface d-band center and mean squared width were obtained as the first and second moments of the projected d density of states, and the d-band filling by integrating up to the Fermi level. Dissociative adsorption energies of H2 and O2 were computed for atoms at 0.25 ML in fcc threefold sites, without zero-point corrections (Kitchin et al. 2004). The authors explicitly set aside alloy lattice-constant changes, reconstructions and adsorbate-induced segregation.
Key findings
Every subsurface 3d metal broadened the surface Pt d band and moved it down in energy relative to pure Pt(111), whose d-band center was computed at -2.44 eV. The effect was smallest for subsurface Ni (center -2.60 eV) and grew toward the early metals, reaching about -3.1 to -3.2 eV for Ti, V and Cr (Kitchin et al. 2004). The surface d-band filling stayed near 9.3 electrons per Pt atom in all cases, so the number of d holes did not change and cannot explain the chemistry. Because filling is conserved, a wider band must sit lower; a rectangular band model predicts that the square root of the mean squared width varies linearly with the d-band center with a negative slope, and the DFT data follow this for a fractional filling of 0.93.
Using a Harrison-type tight-binding matrix element between surface Pt and the subsurface metal, built from characteristic d-orbital lengths and the computed interlayer spacing, the authors showed that the rms surface d-band width is proportional to this Pt-X matrix element. Band broadening therefore reflects strong Pt-3d bonding, which is strongest for early 3d metals and weakest between late transition metals, consistent with strain dominating in the earlier late-metal studies (Kitchin et al. 2004).
The density of states at the Fermi level generally fell as the d-band center dropped, but not monotonically, because it depends on a single state crossing the Fermi level. The authors argue it is a less reliable reactivity indicator than the d-band center. Dissociative adsorption energies of both H2 and O2 decreased nearly linearly as the d-band center became more negative, and for sufficiently low centers dissociative adsorption became endothermic. They also note that since most 3d metals are smaller than Pt, a real alloy lattice would likely compress the surface and push the d band even lower, so the present estimates are an upper bound if an fcc-like structure persists (Kitchin et al. 2004).
Significance
The work shows that strain is not required for a bimetallic surface to behave differently from its parent metals: the ligand effect alone, through heterometallic bonding with subsurface atoms, can shift the d-band center by several tenths of an eV. For PEM fuel cell anodes, lowering the d-band center can improve CO tolerance, but too large a shift makes H2 dissociation unfavorable. For cathodes, weaker O binding may raise the O2 dissociation barrier yet make O adatoms more reactive and could reduce OH coverage, which the authors connect to reported oxygen reduction enhancements on Pt-Fe, Pt-Ni, Pt3Co and Pt3Ni (Kitchin et al. 2004). The matrix-element analysis introduced here was generalized in the companion Letter to treat strain and ligand effects together.