Lifting the discrepancy between experimental results and the theoretical predictions for the catalytic activity of RuO2 (110) towards Oxygen Evolution Reaction

Developing new efficient catalyst materials for the oxygen evolution reaction (OER) is essential for widespread proton exchange membrane water electrolyzer use. Both RuO2(110) and IrO2(110) have been shown to be highly active OER catalysts, however DFT predictions have been unable to explain the high activity of RuO2. We propose that this discrepancy is due to RuO2 utilizing a different reaction pathway, as compared to the conventional IrO2 pathway. This hypothesis is supported by comparisons between experimental data, DFT data and the proposed reaction model. Furthermore, our findings indicate that the reaction pathway utilized by RuO2(110) might be pH dependent, following the conventional pathway at high pH.


Introduction
As part of a transition to a future sustainable economy, there is a need for sustainable fuel and energy storage. Hydrogen gas is an ideal candidate for such a fuel and storage compound, as it can be readily produced by the electrolysis of water 1 . Furthermore, the hydrogen is completely sustainable if the source of electricity is renewable 2 . Currently, the major challenge facing widespread electrolyzer use is the sluggish kinetics of the Oxygen Evolution Reaction (OER) at the anode 3 , fundamentally limited by the universal scaling relations 4,5 . Further development of the water electrolyzer thus requires finding efficient and practical catalysts to facilitate the OER. There are three types of water electrolyzers: alkaline water electrolyzers, proton exchange membrane (PEM) water electrolyzers and solid oxide water electrolyzers 6 . Of these three the alkaline water electrolyzer is the most mature and commercialized. Yet PEM technology has many advantages compared to the alkaline electrolyzer. Some examples include a much higher current density, purer gas, a smaller size for the same power and even the ability to operate at high pressure 6,7 . Currently the best candidates for PEM electrolyzer anode material are IrO 2 and RuO 2 , as these are both stable and active 8-13 . Both iridium and ruthenium are however scarce materials and thus expensive 14 . It is therefore unrealistic to expect that these catalysts can be used on an industrial scale that would have an impact on society 1,15 [18][19][20] . This suggests that either the current reaction model is wrong in the case of RuO 2 or the under-evaluation is due to a computational artefact. In this work we propose that the discrepancy is due to RuO 2 utilizing an alternate reaction pathway for oxygen evolution as compared to IrO 2 . We therefore argue that it is not due to a computational artefact.

Results-Discussion
The conventional pathway describing the interaction between water molecules and the surface of an electrocatalyst was suggested in 2004 21 . During this reaction pathway three intermediates are produced via four electron-proton pair exchanges between the anode and the electrolyte, presented in the following reactions: where * indicates an active site of the surface and HO * , O * , HOO * the adsorbed intermediates on that particular site. The above reaction path describes the Oxygen Evolution Reaction taking place in an acidic environment but it can also be used for the thermodynamic description of the procedure happening in alkaline environment 21,22 . A schematic representation of the conventional OER reaction path is depicted in Fig. 1a.   analysis, is that the blue trend line is followed by experimental data produced in the work of Suntivic et.al. 27 . In their experiments, RuO 2 (110) surfaces were synthesized and their electrochemical response in different pH is recorded. Furthermore, they assign the first and second pre-oxidation peaks observed at the cyclic voltammetries, as the HO * and O * intermediates respectively. The experimental HO * energies serve the role of the descriptor for the experimental data at the activity volcano in the above diagram (Fig.2). The red triangles corresponding to IrO 2 , reproduced from another work of Suntivic et. al. 28 , tend as an ensemble to be placed towards weaker HO * binding energies, closer to the strong binding side of the conventional activity volcano. This is an indication that IrO 2 follows the conventional reaction pathway. The RuO 2 experimental data points are however spread. The points corresponding to highly acidic electrolytes are placed right on top of the blue trend-line together with the theoretical prediction for pathway 1b. In contrast, those corresponding to neutral and alkaline electrolytes are placed closer to the conventional activity volcano, suggesting that RuO 2 at those conditions might follow the conventional pathway.
In Fig. 3   It is the relative strong binding of the oxygen intermediate, which makes the calculated activity of RuO 2 smaller than IrO 2 . Whereas the binding of HO * and HOO * on the RuO 2 cus site is similar, the O * binding is much stronger than that on IrO 2 . This could be an artefact of the DFT calculations, however, it is seen to hold across DFT implementations. The binding energies vary between the different methods, but the difference between HO* and O* binding is close to constant.
Previous experimental studies also show that RuO 2 binds oxygen stronger than IrO 2 for the same HO* binding 27 , even if the difference is smaller than that found in the DFT data. In contrast to the DFT data, differences in experimental data is due to varying electrolyte pH. The experiments measure the potentials for the first and second oxidation peaks, those are normally assumed to be

Conclusion
In this work we are studying the discrepancy between DFT and experimental results, regarding the oxygen evolving reactivity of RuO 2 . We propose that the reaction pathway for electrochemical water oxidation on RuO 2 (110) surfaces, at least in acidic conditions, is slightly different from the reaction path on IrO 2 . In particular, the differences are located at the first and third intermediates, where the protons of HO * and HOO * are migrating towards the bridge oxygen surface. The energy inter-dependency of HO * and HOO * is 2.7eV for the RuO 2 pathway, and is much closer to the ideal difference of 2.46eV . As a consequence the DFT activity is much higher than the one produced by the conventional mechanism and thus the structure is placed closer to the apex of the activity volcano. Furthermore the new placement of RuO 2 (110) on the activity volcano, is at the same region of the RuO 2 experimental results for high acidic electrolytes. By using the conventional pathway we have a very weak interaction of HO * with the surface's cus site. On the other hand, using the RuO 2 pathway widens the energy difference between HO * and O * , placing this DFT calculation closer to experimental trend-lines. This theoretical-experimental agreement, indicates that the RuO 2 mechanism, at least for highly acidic environments, is followed.