Trends in the oxygen adsorption energy derived from the thermodynamic potential model for the oxygen evolution reaction

Across the last years, substantial efforts were done to understand the limitations of the oxygen evolution reaction (OER) catalysts and to develop robust ones such as to solve the efficiency problem in water splitting. Fundamental understanding represents a pathway towards designing superior catalysts. In this direction, several descriptors (ηTD, ESSI, Gmax(η)) have been derived based on the adsorption energies of the OER intermediate moieties (∆EHO*/∆EO*/∆EHOO*) for faster screening of the materials. A universal scaling between the adsorption energies of HO* and HOO* was established for a wide range of materials and in many cases was shown to govern the minimum theoretical overpotential. On the other hand, the scaling between the adsorption energies of HO* and O* fragments have a large scattering along the trendline. In this work we derive five trends for the O* adsorption energies based on theoretical overpotential intervals, using data collected from the published studies employing DFT calculations for OER. It is shown that the best materials have an adsorption energy for HO* placed within a limited interval, yet sufficiently large for a pool of promising catalysts (-0.5,1.5 eV), and that the adsorption energies of O* scales with that of HO* at a slope of one and an intercept of 1.81 eV. Furthermore, a decrease of the intercept for the scaling between HOO* and HO* from 3.14 (valid for all data analyzed) to 3.03 eV is found. For three of the other four trends, the slope of the adsorption energies of O*/HO*scaling is one with intercepts closer either to HOO* (2.2/2.78) or to HO*(1.43/0.78) adsorption energies. For each set of data, the scaling between HOO* and HO* adsorption energies varies from 3 to 3.37 eV. Separately, the trends for O* adsorption energies were analyzed for the TiO2(110) semiconducting rutile surface when doped with transition metals and modified (i) with HO* or H* co-adsorbed fragments that act as charge donor/acceptor moieties or (ii) when the surface was provided with an excess or a lack of electrons. The analysis was performed using the data collected from the published literature, supplemented by additional calculations. Clear trends based on the amount of charge were obtained when standard GGA functionals were used. An understanding for the origin of the large variation of the oxygen adsorption energies for the systems that have similar adsorption energies for the HO*/HOO* is provided. However, when the Hubbard corrected GGA functional is used, some trends change. Clearly, it is still a challenge describing accurately and in a cost-effective manner the adsorption of OER moieties on the undoped or doped transition metal oxides, especially for O*. A strategy to get further insight into the origin of large variations of oxygen adsorption energies for the materials that have similar adsorption energies of HO* and HOO* is discussed.