Universality in Oxygen Evolution Electrocatalysis on Oxide Surfaces
Isabela C. Man, Hai-Yan Su, Federico Calle-Vallejo, Heine A. Hansen, José I. Martínez, Nilay G. Inoglu, John Kitchin, Thomas F. Jaramillo, Jens K. Nørskov and Jan Rossmeisl, “Universality in Oxygen Evolution Electrocatalysis on Oxide Surfaces”, ChemCatChem 3(7), 1159-1165 (2011). https://doi.org/10.1002/cctc.201000397
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
Water splitting for hydrogen production loses much of its efficiency to the overpotential at the anode, where the four-electron oxygen evolution reaction (OER) takes place. Earlier computational work by the same groups had treated metals and rutile oxides separately and used the O* binding energy as an activity descriptor. This paper asks whether a single, material-independent picture of OER activity holds across many oxide families, and what that picture implies about the best activity achievable on flat oxide surfaces (Man et al. 2011).
Approach
The authors assembled a large DFT database of HO*, O* and HOO* adsorption energies on rutile, anatase, perovskite, spinel (Co3O4), rock-salt (MnO, NiO) and bixbyite-type manganese oxide surfaces. Rutile (110) surfaces of MO2 with M = Ti, V, Cr, Mn, Nb, Mo, Ru, Rh, Re, Ir, Pt and Sn were modeled on stoichiometric surfaces, while PbO2, SnO2 and NiO2 were treated on reduced surfaces where reaction occurs at bridge sites. Perovskites were LaMO3 and SrMO3 (100) surfaces with M = Ti through Cu. Calculations used Dacapo with ultrasoft pseudopotentials and the RPBE functional, plane-wave cutoffs of 350 to 400 eV, and spin polarization for Cr, Mn, Ni and Co oxides and perovskites where appropriate; DFT+U was deliberately not used because system-specific U values would spoil comparability. The OER was modeled as four sequential proton-electron transfers through HO*, O* and HOO*, with free energies obtained through the computational hydrogen electrode. The theoretical overpotential is the largest of the four step free energies at zero potential, divided by e, minus 1.23 V. Kinetic barriers and explicit water were neglected, so the aim is trends rather than absolute rates (Man et al. 2011).
Key findings
Across all the oxide classes, the HOO* adsorption energy tracks the HO* adsorption energy with a slope of about one and an intercept of 3.2 eV (fit MAE 0.17 eV, with 68% of points within 0.2 eV and 95% within 0.4 eV). The unit slope is attributed to both species binding through a single O-surface bond and preferring the same site type, and the same offset had been noted on metals. An ideal catalyst would need a separation of 2.46 eV between these two intermediates (each of the four steps equal to 1.23 eV), so the excess of roughly 0.74 eV spread over two steps sets a minimum theoretical overpotential of about 0.4 V (with an uncertainty of about 0.2 V from the scatter) even for a material with optimally placed O* (Man et al. 2011).
Because the HO*-HOO* gap is nearly constant, either the HO* to O* step or the O* to HOO* step is potential-determining, and the free energy difference between O* and HO* becomes a single, universal descriptor. Plotting the theoretical overpotential against it gives one volcano shared by all the oxide families. Illustrative free energy diagrams show LaMnO3 binding strongly (limited by HOO* formation), LaCuO3 binding weakly (limited by HO* oxidation) and SrCoO3 near the top, with a descriptor value of 1.48 eV. The predicted perovskite ordering is SrCoO3 > LaNiO3 > SrNiO3 > SrFeO3 > LaCoO3 > LaFeO3 > LaMnO3, which agrees in trend with alkaline experiments by Bockris and Otagawa and by Matsumoto and coworkers at 10 mA/cm2. For the other oxides, Co3O4 and RuO2 lie near the top, followed by rutile PtO2, RhO2 and IrO2, with Ti, Sn, Mo, V, Nb and Re oxides far less active; (001) anatase RuO2 and IrO2 behave similarly to their rutile forms. Discrepancies for Co3O4 (calculated slightly better than RuO2, measured 0.2 to 0.25 V worse) and NiO are attributed to nonstoichiometry, particle size and higher oxidation states in real samples. The authors also note that NbO2, ReO2, VO2, MoO2 and CrO2 are unstable under OER conditions (Man et al. 2011).
Significance
The paper establishes that the HO*-HOO* scaling relation is universal across oxide structures and turns it into a single-descriptor volcano for the OER, which became a widely used framework for oxide electrocatalyst screening. Its central implication is a fundamental limit: tuning intermediate binding alone cannot make planar oxides much better than RuO2. Breaking the limit requires stabilizing HOO* relative to HO*, for example through three-dimensional or rough structures, confinement as in zeolites, or co-adsorbates, effects the authors suggest may operate in water-oxidizing enzymes (Man et al. 2011).