Tuning oxide activity through modification of the crystal and electronic structure: from strain to potential polymorphs
Zhongnan Xu and John R. Kitchin, “Tuning oxide activity through modification of the crystal and electronic structure: from strain to potential polymorphs”, Physical Chemistry Chemical Physics 17(43), 28943-28949 (2015). https://doi.org/10.1039/c5cp04840k
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
Oxide catalysts are usually tuned by composition, and crystal structure has been explored much less, although methods such as epitaxial thin-film growth can now stabilize metastable polymorphs. The paper asks how the reactivity and stability of hypothetical metastable polymorphs of RuO2, RhO2, IrO2 and PtO2 compare with the stable rutile form. Because every structure considered has octahedrally coordinated cations and three-fold coordinated oxygen, the authors initially hypothesized that the volume difference (strain) relative to rutile alone would explain the differences, and they test that hypothesis (Xu 2015).
Approach
Calculations used VASP with the PBE functional, PAW potentials and a 500 eV plane-wave cutoff. Besides rutile, four polymorphs were modeled for each oxide: anatase, brookite, columbite and pyrite. Strain for each polymorph was defined as its volume relative to the equilibrium volume of rutile for the same metal, and rutile itself was isotropically strained (up to 15% in tension and compression for the IrO2 case study) to give a reference trend. Two chemical properties were computed. The oxygen adsorption energy (relative to half of O2) was evaluated at a five-fold coordinatively unsaturated (5cus) cation site, chosen on facets such that the site has the same number of missing bonds in every structure, so that differences reflect distortions of the MO6 octahedron rather than site coordination; rutile (110) and columbite (101) are examples. Bulk oxygen vacancy formation energies were computed at a fixed 6.25% vacancy concentration. O, OH and OOH adsorption energies, referenced to the standard hydrogen electrode, were fed into the established atomistic thermodynamic model of the oxygen evolution reaction (OER) to estimate theoretical overpotentials. Atom-projected densities of states separated into eg and t2g contributions were used to interpret the results (Xu 2015).
Key findings
For strained rutile, both the O adsorption energy and the vacancy formation energy varied smoothly with strain, but the polymorphs did not fall on these curves. The change in adsorption energy from switching crystal structure was often larger than the change produced by straining rutile. Vacancy formation behaved the opposite way: compressive or tensile strain of rutile made vacancy formation markedly more exothermic, whereas polymorphs, being relaxed at their own equilibrium volume, generally kept more endothermic vacancy energies. The authors attribute this to strained rutile being structurally unstable and relaxing more around a vacancy, and conclude that a globally metastable polymorph could still be reasonably stable under reaction conditions where vacancies form and fill. Strained rutile showed a correlation between adsorption and vacancy energies, but the polymorphs broke it, offering a wider range of adsorption energies without loss of stability.
The O, OH and OOH adsorption energies on polymorphs followed the same scaling relations as on rutile, which the authors link to the similar local geometry of the adsorption site, so the polymorphs land on the same OER activity volcano at new positions. Anatase and columbite IrO2 and brookite RhO2 and PtO2 were predicted to be more active for OER than the corresponding rutile phases. Columbite IrO2, previously proposed by the same group as a thin-film synthesis target, is highlighted as a promising candidate.
In the IrO2 case study, rutile and columbite have similar volumes, yet going from rutile to columbite changed the O adsorption energy by about 0.15 eV while straining rutile had negligible effect. At the adsorption site, strain changed Ir-O bond lengths uniformly (for example from 2.012 Å to 2.090 Å in the surface plane under 15% tension) without altering bond angles, while columbite had unequal bond lengths (about 1.89 to 2.06 Å) and surface O-Ir-O angles near 84 to 98 degrees instead of the rutile 74.6 and 104.9 degrees. Strain mostly reshaped and shifted the eg band and left the t2g band near the Fermi level nearly unchanged, whereas the columbite distortion substantially redistributed the t2g states. The authors conclude that the t2g states govern adsorption strength and that octahedral distortions, not volume change, redistribute them. They tentatively attribute the weaker O binding on columbite IrO2 to new high-energy t2g states at the Fermi level that overlap with antibonding eg states, though no universal electronic descriptor was found across all systems (Xu 2015).
Significance
The study shows that crystal structure is a tuning handle for oxide reactivity that is distinct from, and in some respects more useful than, strain: it moves adsorption energies further while preserving resistance to vacancy formation, suggesting a way to ease the usual activity/stability trade-off. It also reinforces the earlier finding from doped rutile oxides that the t2g states, rather than the eg states that form sigma bonds to oxygen, best track adsorption on these oxides. The OER predictions are framed as a test of principle, and the authors call for more detailed study of columbite IrO2 facets, sites and stability (Xu 2015).