Skip to main content

Relationships between the surface electronic and chemical properties of doped 4d and 5d late transition metal dioxides

The Journal of Chemical Physics (2015)

Zhongnan Xu and John R. Kitchin, “Relationships between the surface electronic and chemical properties of doped 4d and 5d late transition metal dioxides”, The Journal of Chemical Physics 142(10), 104703 (2015). https://doi.org/10.1063/1.4914093

Read the paper

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

Doping one transition metal into the oxide of another is a common way to tune adsorption energies, which govern activity in many oxide-catalyzed reactions. For metal alloys, ligand and strain effects and the d-band model give a framework that connects simple atomic properties to reactivity, but no comparable framework existed for oxides. The paper asks how the chemistry of a particular cation changes when it is placed in another oxide, and whether that change can be split into ligand and strain contributions that are each explained by features of the electronic structure (Xu 2015).

Approach

The authors used VASP with the PBE functional, PAW potentials, a 500 eV cutoff and a 7 x 7 x 1 k-point grid. No Hubbard U or hybrid functional was used, on the argument that U does not change the trends of interest. The systems were all dopant-host pairs among six late 4d and 5d metals (Ru, Rh, Pd, Os, Ir, Pt) as +4 cations in the rutile structure, giving 30 pairs, which fall in the same-valence or flexible-valence dopant categories that lack simple electron-counting rules. Dissociative O adsorption was computed at the 5-fold coordinatively unsaturated (5cus) site of the (110) surface in four-layer slabs (top two layers relaxed), with bridge oxygens present and 0.5 ML 5cus coverage in a 2 x 1 cell.

The “dopant effect” is the change in O adsorption energy on cation M^D when moved from its own oxide into host M^H O2. It is decomposed into a ligand effect (all neighboring cations replaced by M^H at the M^D O2 lattice constant) and a strain effect (pure M^D O2 at the lattice constant of M^H O2). In total 96 adsorption energies were computed. Band centers, rms widths (second moments) and fractional fillings were extracted from atom-projected densities of states for the d, e_g and t_2g states of the site cation and the p states of neighboring oxygens.

Key findings

The sum of the ligand and strain effects closely reproduced the directly computed dopant effect for all 30 pairs. Surprisingly, the ligand effect was up to an order of magnitude larger than the strain effect, even though all cations are formally +4 and are not directly bonded to each other.

Among the descriptors tested (metal d, e_g and t_2g centers, oxygen p center), the t_2g-band center of the adsorption-site cation correlated best across all adsorption energies, with separate trends for 4d and 5d sites; the t_2g states lie near the Fermi level. The ligand effect correlated inversely with the shift of the t_2g-band center, as in the d-band model for metals, and more strongly than with the full d band. Unlike metals, the t_2g filling was not conserved: the center shifted because the number of d electrons on M^D changed. Changes in M^D d filling and in the neighboring oxygen p filling had the same sign, and the oxygen p-filling change was close to the negative of the change in the host cation’s d filling, which the authors take as evidence that charge moves between dopant and host cations with lattice oxygen as the mediator. The ligand effect correlated best with the difference in t_2g-band widths between dopant and host, and, more weakly, with the difference in tabulated effective d-orbital radius raised to the 3/2 power. A dopant with a larger radius than its host loses electrons to it, lowering its t_2g center and weakening O adsorption.

Strain changed the t_2g-band width (broader under compression) and shifted its center linearly, in a roughly element-independent way, without charge transfer, which the authors attribute to strain acting symmetrically on all cations. The adsorption response to strain, however, was element specific: only Pd and Pt bound O more weakly under compression, opposite to the other four and to the general behavior of late transition metals. The slope of the strain effect versus t_2g-center shift varied inversely with the idealized d-band filling, so cations with more than half-filled d bands (Pd, Pt) followed the ligand-effect trend while those at or below half filling showed the opposite sign. The mechanism of the strain effect was left partly unresolved (Xu 2015).

Significance

The work extends the strain/ligand decomposition and band-moment descriptors developed for bimetallic surfaces to doped rutile oxides, and shows that the dominant ligand effect operates through cation-to-cation charge transfer rather than through the constant-filling band-width mechanism of metals. Because the effective orbital radius is a tabulated quantity, the authors propose it as a basis for predictive models of adsorption on doped oxides. The finding that strain and ligand effects follow different electronic structure relationships also argues for treating them separately (Xu 2015).

Related

Concepts: Strain and ligand effects, d-band model, Doped oxides, DFT + U and the Hubbard U. The ligand/strain framework for metals that this paper extends is summarized in kitchin2004a. A companion paper on DFT + U for 3d oxides by the same group is xu2015c.