Journal of Catalysis 396 (2021) 192–201 Contents lists available at ScienceDirect Journal of Catalysis journal homepage: www.elsevier.com/locate/jcat Importance of the oxyl character on the IrO2 surface dependent catalytic activity for the oxygen evolution reaction Danilo González 1, Javier Heras-Domingo 1, Mariona Sodupe, Luis Rodríguez-Santiago, Xavier Solans-Monfort ⇑ Departament de Química, Universitat Autònoma de Barcelona, 08193 Bellaterra, Spain a r t i c l e i n f o a b s t r a c t Article history: The oxygen evolution reaction catalyst optimization is hindered because in the desirable acidic Received 13 November 2020 conditions the sole active catalysts are RuO2 and IrO2. Thus, the understanding of the factors controlling Revised 20 February 2021 the reactivity of these materials is mandatory. In this contribution, DFT (PBE-D2) periodic calculations are Accepted 22 February 2021 performed to analyze the catalytic activities of the main ((1 1 0), (0 1 1), (1 0 0) and (0 0 1)) IrO2 surfaces. Available online 04 March 2021 Results show that the reaction only occurs if the Ir=O species on the surfaces exhibit an oxyl character. The water nucleophilic attack mechanism is the most favorable pathway on the (1 1 0), (1 0 0) and Keywords: (0 0 1) surfaces. In contrast, for the (0 1 1) facet the oxo-coupling is preferred. The required overpotentials Oxygen evolution reaction IrO2 for the four IrO2 surfaces depend on the feasibility to oxidize the Ir-OH to Ir-O species and this is tuned by Metal-oxyl species the coordination of the unsaturated iridium sites: the (1 0 0) and (0 0 1) surfaces appear to be more active DFT than the (1 1 0) and (0 1 1). Ó 2021 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). 1. Introduction Regarding the catalytic performances of different IrO2 materials, it has been found that the crystalline (1 0 0) surface is more active One of the major challenges of the 21st century is the than the crystalline (1 1 0) one.[16] These different catalytic substitution of fossil fuels for sustainable and environmentally activities correlate reasonably well with the number of undercoor- friendly alternative sources of energy [1,2]. One of the most dinated centers per surface area of the crystalline facets [17]. promising solutions is to store energy in the form of chemical Therefore, this suggests that the undercoordinated centers are bonds in a similar manner to what plants do during the photosyn- the main active sites in crystalline materials. Moreover, amorphous thesis [3]. In this way, H2 production by the splitting of water is IrOX tends to be more active, [18–22] but less stable than the seen as one of the most sustainable ways to store energy [4–11]. crystalline analogues, the deactivation mechanisms being a key Water splitting implies two half reaction: i) At the cathode, H+ issue under discussion.[20,23,24] are converted to H2 through the hydrogen evolution reaction In-situ X-ray spectroscopy has been carried out to identify the (HER) and ii) H2O is oxidized to O2 through the oxygen evolution oxidation state of the metal center as function of the applied reaction (OER), at the anode. The OER is considered the bottleneck potential.[18,20,25–30] For IrO2, many authors showed that while of the global process due to the large overpotentials that have to be at potentials below 1.3 V, the metal centers are mainly Ir(IV) or Ir applied. Moreover, acidic conditions are preferred with respect to (III), at higher potentials they are oxidized to Ir(V).[25,31] The Ir(V) basic media for the practical application of water splitting species are proposed to be involved in the catalytic process and [12,13]. However, these acidic media also imply working at harsh they are probably reduced upon O2 release.[26] The data for the reaction conditions which limits the number of potential catalysts most stable (1 1 0) surface is consistent with the IrO2 being to the expensive IrO2 and RuO2 [13–15]. In this context, completely covered with *O (oxo) species coordinated to the understanding the key factors controlling catalyst efficiency is unsaturated centers at reaction conditions (C in Scheme 1).[28] essential to find strategies that could lead to catalyst optimization. In addition, Infrared spectroscopy (FT-IR) allowed the detection of Ir-OOH species (D in Scheme 1) upon water oxidation with an IrO2 nanocluster under pulsed excitation of a light sensitizer.[32] ⇑ Corresponding author at: Departament de Química, Universitat Autònoma de Finally, the use of differential electrochemical mass spectroscopy Barcelona, Carrer dels Tillers s/n, 08193 Bellaterra, Spain. (DEMS) with isotope-labeled electrolytes showed that IrO2 E-mail address: xavier.solans@uab.cat (X. Solans-Monfort). 1 exchanges oxygen with the electrolyte and thus actively partici- D.G. and J. H.-D. have equally contributed. https://doi.org/10.1016/j.jcat.2021.02.026 0021-9517/Ó 2021 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). D. González, J. Heras-Domingo, M. Sodupe et al. Journal of Catalysis 396 (2021) 192–201 Scheme 1. Proposed reaction mechanisms for the Oxygen Evolution Reaction. pates in the catalysis through its lattice oxygen atoms.[20,33,34] (C to E step in Scheme 1) and iii) O2 release from Ir-OOH (E to A Nevertheless, according to the most recent results of Shao-Horn in Scheme 1). Controversy persists, and one can find recent contri- and co-workers on RuO2, this would be the case of none- butions suggesting that either the formation of Ir-OOH or the O2 crystalline materials, since no oxygen exchange is detected on release are the rate determining processes.[53,55,57] rutile-like catalysts.[13,17] In this contribution we report a computational study on the Three different reaction mechanisms have been proposed catalytic activity of IrO2 main crystallographic surfaces (1 1 0), (Scheme 1):[12,13,35,36] i) the water nucleophilic attack (WNA), (0 1 1), (1 0 0) and (0 0 1), with the aim of determining if all low ii) the oxo-coupling mechanism (I2M) and iii) The lattice oxygen miller index facets share the same most favorable reaction mecha- evolution reaction mechanism (LOER). The water nucleophilic nism and rate determining step and understanding the reactivity attack mechanism (Scheme 1a) implies first the formation of an Ir- differences. For that, we include the energy barriers of the water oxo species on the surface (Ir=O) through two proton coupled elec- nucleophilic attack and oxo-coupling chemical steps. Moreover, tron transfer steps (PCET). Then, the Ir=O species suffers the attack the analysis of the electronic structure of the major intermediates of a water molecule leading to the formation of an Ir-OOH hydroper- allows proposing an alternative chemical description of the oxo species that after one additional PCET releases O2 and regener- chemical steps. ates the initial Ir-H2O species. The I2M pathway (Scheme 1b) starts also with the formation of Ir=O species, but, in contrast to the WNA mechanism, O2 is formed by the coupling of two neighbor 2. Computational details Ir=O species. Finally, the LOER mechanism (Scheme 1c) implies the active role of vacant sites on the metal oxide surface[20,33,34], but The main IrO2 surfaces were represented with the slab models their role in highly crystalline materials, as the surfaces considered used in our previous contributions (Fig. 1).[59,60] They were built in this work, appears to be less significant.[13,17] by considering a (2x2) supercell, cut from the optimized bulk Computational chemistry has been widely used to shed light structure. All slabs present a four-layer thickness to ensure a rea- into the most plausible reaction mechanism.[37–57] This is mainly sonable converged surface energy.[59,60] The c value was set to done by computing the thermodynamic cost of each PCET step. 35 Å, the interlayer distance being at least 21 Å. During the opti- Only in a few cases, the energy barriers associated with some of mizations, cell parameters were kept fix, while the atom positions the elementary steps have been computed[38,40,55,58] Remark- were fully relaxed. ably, most of the studies focusing on the reaction mechanism are All DFT periodic boundary calculations were performed with centered on the most stable (1 1 0) surface.[37–40,49,51–55,574 the spin polarized formalism as implemented in the VASP code 1–44] Only a very recent contribution of Rao et al. analyzed the dif- [61,62] and using the GGA PBE functional.[63] Grimme’s D2 empir- ferences in catalytic activity of the (1 1 0), (1 0 0) and (1 0 1) surfaces ical correction was added to account for dispersion forces.[64] D2 of RuO2.[45] Despite the limitations in solvent representation correction leads to bulk cell parameters that are closer to the among others,[39,40] the computed overpotentials are in good experimental data than those obtained with D3.[59,60] Ionic cores agreement with experiments and they support that the WNA were described with projector augmented wave pseudopotentials mechanism is usually the applied mechanism. Indeed, in the crys- [65,66] and the valence electrons were represented through a talline (1 1 0) surface the water nucleophilic attack to the plane wave basis with a kinetic energy cut off of 500 eV. A oxo-species is computed to be kinetically easier than the (4,4,1) Monkhorst  Pack K-point mesh[67] was employed to oxo-coupling of the I2M.[38,40,55] Noteworthily, in some flexible describe the first Brillouin zone. The energy convergence criteria systems such as oxidized iridium surfaces or supported dimeric were fixed to 105 and 104 eV for electronic and geometry relax- Ir species, the I2M mechanism has been suggested to be ations, respectively. This methodology is equivalent to that used in competitive with the WNA one.[46,47,53] Taking WNA as the our previous contributions on RuO2 and IrO2 materials.[59,60,68] applying mechanism, three different rate determining steps have Solvation effects were included by using the implicit model been proposed: i) The Ir=O formation (B to C step in Scheme 1); implemented in VASPsol[69] through single point calculations at ii) the electrochemical step associated with the Ir-OOH formation vacuum optimized geometries. Thermal corrections at 1 atm and 193 D. González, J. Heras-Domingo, M. Sodupe et al. Journal of Catalysis 396 (2021) 192–201 from cooperative hydrogen bonding between water molecules in the models with higher water coverages. However, the computed overpotentials for the four surfaces and the two explored mecha- nism are similar (differences smaller than 0.2 V and 0.1 V for the I2M and the WNA mechanisms, respectively) and the reaction energies of the O-O bond formation either through the oxo- coupling or the water nucleophilic attack mechanisms show little variations (less than 0.2 eV). Therefore, results are not model dependent. 3. Results and discussion We have studied the oxygen evolution reaction catalyzed by the main ((1 1 0), (0 1 1), (1 0 0) and (0 0 1)) crystallographic facets of rutile-like IrO2, with the aim of analyzing their different catalytic activity. In this context, we first determine the potentials required for oxidizing the initial IrO2-H2O material to Ir=O without involving additional water molecules and analyze the electronic structure of the Ir-OH and Ir=O intermediates. In a second step, we compute the catalytic cycle of the two most usually proposed mechanisms (oxo-coupling and water nucleophilic attack) starting from the most stable termination at 1.5 V. We do not consider the Fig. 1. Surfaces considered in this work. Obr refers to the unsaturated surface LOER mechanism since, according to experiments, it appears to be oxygen atoms. Ir5C-ax and Ir5C-eq stands for pentacoordinated iridium centers less relevant in defective free materials such as the models used presenting an axial and equatorial vacant site respectively. Ir4C-eq are tetracoordi- nated iridium sites with two equatorial vacant sites. The density of undercoordi- here. This allows us determining the most favorable mechanism, nated Ir centers per nm2 (blue values) is also included. proposing the rate determining electrochemical step and computing the overpotentials. 3.1. IrO2–H2O oxidation as a function of the applied potential T = 273 K were added considering only the vibrational contribu- Fig. 2 shows the most stable conformation associated with H2O tions of the normal modes associated with the adsorbed species adsorption on the different surfaces. Rutile-like IrO2 structure and the Ir-Oads stretching mode. Entropy contributions for water presents a distorted octahedral environment around the metal and H2 were obtained from the tabulated values for liquid water center with two short Ir-O distances in axial position and four long and H2 gas, respectively. Similarly to previous contributions, ones in the equatorial plane. The most stable (1 1 0) surface and the [37,53–55] the O2 Gibbs energy was computed as the energy differ- (1 0 0) facet present unsaturated pentacoordinated metal sites with ence between the experimental DG° of the global reaction (4.92 eV the vacant site in axial position. The second most stable (0 1 1) for 2H2O ? 2 H2 + O2) and the Gibbs energies of H2O and H2 as surface has also unsaturated pentacoordinated metal sites but with shown in equation 1. No particular treatment was applied to 5OH/ the vacant site in equatorial. Finally, the least stable main OO and 6O/OO as the analysis of the electronic structure suggest a crystallographic orientation, (0 0 1), presents tetracoordinated superoxide (O– 2 ) character with only one single unpaired electron. unsaturated metal centers, the two vacant sites being equatorial. Finally, the DG° for the PCET steps were computed by using the In addition, all surfaces present non-saturated dicoordinated computational standard hydrogen electrode as defined by Ross- oxygen atoms (Obr). We recently showed that the different nature meisl, Nørskov and co-workers.[37] of the vacant at metal center, the basicity of the oxygen bridge atoms and the hydrogen bonding between adsorbed water molecules influence the adsorption strength of a single water Go O2 = 4.92 + 2 Go H2O —2 Go H2 (in eV) ð1Þ molecule and its facility to dissociate forming an Ir-OH moiety Transition states for the oxo-coupling and water nucleophilic and a protonated oxygen bridge.[59,60,72–74] For IrO2, the adsorp- attack chemical steps were located by using either the climbing tion energies are the highest when the vacant site is axial and the image nudged elastic band (CI-NEB)[70,71] strategy or restricted most stable structures are: complete dissociation on the (1 1 0) and geometry optimizations. The highest in energy point of either the (1 0 0) surfaces and 50% or lower dissociation on the other two sur- NEB or the restricted optimization was used for locating the final faces (Fig. 2). At this point it is worth mentioning that, although at transition state by performing a geometry optimization with the least one water molecule dissociates in the majority of cases, the quasi-Newton algorithm implemented in VASP. Convergence was initial species will hereafter be referred as 0H2O/H2O for simplicity. achieved when forces were smaller than 0.02 eV/Å. The nature of The oxidation of 0H2O/H2O to 4O/O is expected to occur through the stationary point was ensured by vibrational analysis that successive PCET steps. The final species has vicinal Ir=O as shown showed that the largest imaginary frequency in all cases corre- in Scheme 2. With the aim of stablishing the most stable sponds to that of the transition state. termination as a function of the applied potential (U) at pH = 0, We also performed calculations on (2x1) supercells models of we represented the relative stabilities of all species in Fig. 3, the (1 1 0), (1 0 0) and (0 0 1) surfaces as well as a (1x1) model of obtained by using equation 2, where NL1/L2 stands for the consid- the (0 1 1) facet to ensure that the density of Ir=O groups has no ered species 1OH/H2O, 2OH/OH, 3O/OH and 4O/O. effect on the main conclusions. In these four models all iridium centers on the surface are fully oxidized. As shown in Table S1 of 0H2O=H2O ! NL1=L2 + ne + nHþ DG0 = G0 N - G0 0 —nU ð2Þ the Supplementary Material, there are some differences associated with the adsorption of water and its oxidation to OH that arise 194 D. González, J. Heras-Domingo, M. Sodupe et al. Journal of Catalysis 396 (2021) 192–201 Fig. 2. Lateral (a) and apical (b) view of the most stable structures for the adsorption of two water molecules on the main IrO2 surfaces. Distances in Å. -1 e- -1 e- -1 e- -1 e- H2O H2O - 1 H+ OH H2O - 1 H+ OH OH - 1 H+ O OH - 1 H+ O O Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir 0H2O/H2O 1OH/H2O 2OH/OH 3O/OH 4O/O Scheme 2. Fig. 3. Surface phase diagram of IrO2 at pH = 0, T = 298 K and P = 1 atm as a function of the applied potential. Grey stands for 0H2O/H2O, green stands for 1OH/H2O, blue stands for 2OH/OH, orange stands for 3O/OH and pink for 4O/O. Three different oxidation behaviors are observed. The (0 0 1) vacant site in axial position, and they get oxidized to 2OH/OH at surface is more easily oxidized to 4O/O than the other facets, the potentials of 1.15–1.16 V and to 4O/O at potentials over 1.46 V. potentials at which 4O/O becomes the most stable termination The (0 1 1) surface, with the vacant site in the equatorial plane, is being 0.93 V. The other three surfaces present pentacoordinated not easily oxidized to 4O/O, which is only predicted to be the most centers and they behave in a more similar manner. However, the favorable species over 1.64 V. As a consequence, at potentials ability of the different surfaces to get oxidized depends also on around 1.5 V, 4O/O is still not formed and 3O/OH is the most stable the vacant site nature. The (1 1 0) and (1 0 0) surfaces have the termination. These values agree well with the computational data 195 D. González, J. Heras-Domingo, M. Sodupe et al. Journal of Catalysis 396 (2021) 192–201 for the (1 1 0) surface reported by Ping, Nielsen and Goddard.[55] 3.2. OER reaction mechanism Moreover, the 2OH/OH to 4O/O transition is computed to occur at 1.46 V, which is only 0.16 V higher than the potential in which irid- Starting from the most stable termination at potentials of 1.5 V ium is oxidized to its higher state according to experiments.[25] (4O/O for the (1 1 0), (1 0 0) and (0 0 1) surfaces and 3OH/O for the Thus, there is a fairly good agreement between our data and the (0 1 1) surface), we have studied the two most commonly proposed existent values, which reinforces the validity of our approach. reaction mechanisms (Scheme 3): the oxo-coupling (I2M) and the Geometric parameters (Table 1 and Supplementary Material) water nucleophilic attack (WNA). The lattice oxygen evolution show that the Ir-OL distance (OL stands for the oxygen of the reaction mechanism (LOER) was not considered, since, according adsorbed species) decreases when going from Ir-H2O to Ir-OH to experiments, it does not take place in a significant manner in and Ir=O. Moreover, the Ir-OL distance tends to be shorter for the defective free materials.[17] For each pathway, we computed the (0 0 1) surface than for the (1 1 0) and (1 0 0) facets, the (0 1 1) sur- thermodynamics of the whole cycle as well as the energy barriers face usually showing the largest values. This is clearly observed associated with the chemical steps. Table 2 reports the energetics for 4O/O where the shortest distances are those of the (0 0 1) surface of the different steps at 0 V potential versus standard hydrogen and the largest ones those of the (0 1 1) facet. Interestingly, the electrode (SHE).[37] Table 3 presents the main geometry parame- shortest distance in the (0 0 1) surface is related with the absence ters and spin densities of the key intermediates and Figs. 4 and 5 of magnetic moment over the oxo ligand (Table 1), thus suggesting show the optimized structures of the transition states. All other the formation of an oxo group with an Ir=O double bond and a data can be found in the Supplementary Material. metal center formally at its +6 oxidation state. In contrast, the lar- The first step of the I2M mechanism is a chemical process in ger distances of the (1 1 0), (1 0 0) and (0 1 1) are associated with which two oxo species on the surface couple forming an O-O bond unpaired electrons both on the metal and the oxo. This indicates (oxo coupling 4O/O to 4O-O). The global process is endergonic for all that such species have an important metal oxyl Ir–O radical char- IrO2 main crystallographic facets, the reaction energies ranging acter where the metal oxidation state is formally +5 (Table 1). This between 0.10 and 2.30 eV. The most unfavorable oxo-coupling interpretation is in agreement with the proposal that Ir(V) species occurs at the (0 0 1), where the Ir=O species has the most pro- are the ones involved in the catalytic cycle[26] and the observation nounced oxo character. The least unfavorable process takes place of oxyl species on iridium surfaces by means of X-Ray spectroscopy at the (0 1 1), where the two Ir=O species are closer (2.887 Å) and and it suggest that the reactivity of the (0 0 1) surface could be dif- the Ir=O species is less stabilized with respect to two adsorbed ferent to that of the other three facets. water molecules. The energy barriers for this step show that the In summary, the oxidation of these facets is largely controlled reaction is kinetically more challenging for the (0 0 1) surface, fol- by the coordination of the unsaturated metal centers and the nat- lowed by the (1 0 0) and (1 1 0) facets and the (0 1 1) surface being ure of their vacant sites. The least coordinated tetracoordinated the one with the lowest barrier. Indeed, the process is only acces- centers of the (0 0 1) surface are the least stabilized ones and, con- sible at the (0 1 1) surface (DGà = 0.33 eV), where oxidation of 3OH/O sequently, those that more easily form the Ir=O group. Within the to 4O/O is needed prior to the oxo-coupling. This process requires other three surfaces, the nature of the vacant site appears to be the potentials above 1.64 V to take place. key factor: Ligands at axial sites present stronger interactions than Analysis of reactants, products and transition states’ structures ligands at equatorial sites. Moreover, the oxo group (or oxyl radi- shows that during the oxo–coupling process the OO distance cal) interacts more strongly than water with the metal center. Con- decreases to 1.387–1.404 Å in 4O-O and this is associated with an sequently, the decrease in strength when going from axial vacant elongation of the Ir-O bond (2.012–2.043 Å). The computed O-O sites ((1 1 0) and (1 0 0) facets) to an equatorial one ((0 1 1)) is more distance in 4O-O suggests that O2 acts as l-g1:g1-O2- 2 peroxo ligand pronounced for Ir=O, the latter becoming destabilized with respect coordinated to two metal centers[75,76] and this is in agreement to 0H2O/H2O and thus, requiring higher potentials to be formed. with the absence of spin moment over the oxygen atoms. Thus, Table 1 Ir-OL distances (in Å) and spin moment for the different surface terminations arising from the oxidation of hydrated IrO2 (1 1 0), (011), (1 0 0) and (0 0 1) surfaces. Species Ir1-OL1 Ir2-OL2 SO1 SO2 SIr1 SIr2 (1 1 0) 0H2O/H2O 2.002 2.002 0.00 0.00 0.00 0.00 1OH/H2O 1.943 1.990 0.00 0.00 0.00 0.00 2OH/OH 1.937 1.938 0.00 0.00 0.00 0.00 3OH/O 1.788 1.925 0.54 0.02 0.48 0.15 4O/O 1.789 1.789 0.67 0.67 0.52 0.52 (011) 0H2O/H2O 2.110 2.077 0.00 0.00 0.00 0.00 1OH/H2O 1.953 2.148 0.00 0.00 0.00 0.00 2OH/OH 1.965 1.940 0.00 0.00 0.00 0.00 3OH/O 1.810 1.933 0.57 0.02 0.55 0.00 4O/O 1.797 1.797 0.59 0.62 0.52 0.57 (100) 0H2O/H2O 2.003 2.003 0.02 0.02 0.08 0.12 1OH/H2O 1.951 1.958 0.01 0.01 0.20 0.15 2OH/OH 1.927 1.927 0.17 0.17 0.41 0.40 3OH/O 1.796 1.924 0.19 0.02 0.24 0.03 4O/O 1.771 1.771 0.57 0.56 0.26 0.26 (001) 0H2O/H2O 2.025 2.099 0.00 0.00 0.02 0.01 1OH/H2O 1.947 2.094 0.00 0.00 0.00 0.00 2OH/OH 1.933 1.900 0.00 0.00 0.00 0.00 3OH/O 1.729 1.898 0.00 0.00 0.00 0.00 4O/O 1.732 1.727 0.00 0.00 0.00 0.00 196 D. González, J. Heras-Domingo, M. Sodupe et al. Journal of Catalysis 396 (2021) 192–201 HO O + 2 H2O -1 e- H2O H2O O O O O OH O OH O - 1 O2 + 1 H2O - 1 H+ Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir 0H2O/H2O 4O-O 4O/O 4OH/OOH 5OH/OO -1 e- -1 e- I2M -1 e- WNA-1 - 1 H+ - 1 H+ - 1 H+ -1 e- O OH H2O OH OH -1 e- O OH -1 e- + 1 H2O O O - 1 H+ O H2O - 1 H+ - 1 H+ - 1 O2 Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir 1OH/H2O 2OH/OH 3O/OH 2O/H2O 6O/OO Scheme 3. Table 2 Reaction Gibbs Energies (in eV) for the chemical and electrochemical steps of the I2M and WNA reaction mechanisms. Values in parenthesis correspond to the Gibbs energy barriers. Reaction conditions are 0 V potential versus SHE, pH = 0, T = 298 K and P = 1 atm. Reaction (110) (011) (100) (001) (001)_1ca I2M 4O/O ? 4O-O 0.67 (0.80) 0.10 (0.33) 0.85 (0.94) 2.30 (2.38) 0.25 (1.15) 4O-O + 2H2O ? 0H2O/H2O + O2 0.89 0.13 0.90 0.72 0.11 0H2O/H2O ? 1OH/H2O + H+ + e- 1.07 0.37 0.90 0.64 0.83 1OH/H2O ? 2OH/OH + H+ + e- 1.15 1.73 1.16 1.25 1.10 2OH/OH ? 3O/OH + H+ + e- 1.56 1.21 1.50 0.52 1.38 3O/OH ? 4O/O + H+ + e- 1.36 1.64 1.41 0.93 1.47 WNA 4O/O + H2O ? 4OH/OOH 0.12 (0.43)b 0.05 (0.42)b 0.08 (0.42)b 1.24 0.25 (0.55)b 4OH/OOH ? 5OH/OO + H+ + e- 1.11 0.85 1.22 0.84 1.20 5OH/OO ? 6O/OO + H+ + e- 1.53 1.52 1.22 0.91 1.24 6O/OO + H2O ? 2O/H2O + O2 0.17 0.02 0.40 0.56 0.21 2O/H2O ? 3O/OH + H+ + e- 1.21 0.98 1.37 0.76 0.97 3O/OH ? 4O/O + H+ + e- 1.36 1.64 1.41 0.93 1.47 a (001)_1c stands for the OER on one single iridium center of the (0 0 1) surface. b The reaction energy is computed as Gsolv(4OH/OOH) – (Gsolv(4O/O) + Gsolv(H2O)), while the Gibbs energy barrier is computed from an intermediate species (4O/O-H2O) where one water molecule is interacting with 4O/O. The Gibbs energy difference between 4O/O + H2O and 4O/O-H2O is always lower than 0.10 eV. Table 3 Ir-OL distances (in Å) and spin moments for the reaction intermediates of the I2M and WNA mechanisms. Species Ir1-OL1 Ir2-OL2 O-O SL1 SL2 SIr1 SIr2 (1 1 0) 4O/O 1.789 1.789 3.182 0.67 0.67 0.53 0.53 4O-O 2.024 2.022 1.395 0.00 0.00 0.00 0.00 4OOH/OH 1.931 1.913 1.417 0.00 0.10 0.00 0.10 5OO/OH 1.931 1,896 1.297 0.22 0.10 0.10 0.10 6OO/O 1.923 1.781 1.289 0.15 0.31 0.12 0.17 (0 1 1) 4O/O 1.797 1.797 2.887 0.62 0.59 0.57 0.52 4O-O 2.043 2.016 1.404 0.00 0.00 0.00 0.00 4OOH/OH 1.961 1.936 1.446 0.00 0.00 0.00 0.00 5OO/OH 1.983 1.939 1.293 0.63 0.00 0.12 0.00 6OO/O 1.941 1.798 1.281 0.08 0.57 0.04 0.54 (1 0 0) 4O/O 1.771 1.771 3.187 0.50 0.50 0.34 0.34 4O-O 2.012 2.015 1.395 0.02 0.02 0.01 0.01 4OOH/OH 1.927 1.914 1.406 0.00 0.00 0.00 0.00 5OO/OH 1.958 1.912 1.286 0.63 0.03 0.17 0.02 6OO/O 1.932 1.760 1.287 0.43 0.05 0.12 0.01 (0 0 1) 4O/O 1.732 1.732 4.304 0.00 0.00 0.00 0.00 4O-O 2.033 2.033 1.387 0.00 0.00 0.00 0.00 4OOH/OH 1.885 1.895 1.464 0.00 0.00 0.00 0.00 5OO/OH 1.853 1.889 1.280 0.00 0.00 0.00 0.00 6OO/O 1.856 1.726 1.273 0.00 0.00 0.00 0.00 the 4O/O to 4O-O process should be viewed as the homolytic Once the peroxide is formed, the subsequent steps are accessi- coupling of two oxyl groups, without reduction of the metal center ble at reaction conditions. The ligand exchange between O2 and at the surface. two water molecules is in all cases exergonic with reaction ener- 197 D. González, J. Heras-Domingo, M. Sodupe et al. Journal of Catalysis 396 (2021) 192–201 are within 0.12 and 1.24 eV (-11.6 and 119.6 kJ mol1). The reac- tion is slightly exergonic at the (1 1 0) (-0.12 eV), essentially iser- gonic at the (0 1 1) and (1 0 0) surfaces (-0.05 and 0.08 eV, respectively) and highly unfavorable at the (0 0 1) surface (1.24 eV). The Gibbs energy barriers from an intermediate struc- ture where the water molecule interacts through hydrogen bond- ing with the two Ir=O species are in all cases low (around 0.42 – 0.43 eV) except for the (0 0 1) surface where the process is thermo- dynamically hampered, and we were not able to localize the tran- sition state. The Ir-O distances at 4OOH/OH are similar to those computed for 2OH/OH and indicative of an Ir-O single bond. Moreover, the O-O dis- tance in the Ir-OOH species ranges from 1.406 to 1.464 Å. These geometry features are associated with the absence of magnetiza- tion in any of the three oxygens of Ir-OOH and Ir-OH species, thus suggesting that Ir-OOH should be viewed as a hydroperoxo (OOH–) species[75]. According to this sequential pathway the H2O dissoci- ation should be viewed as an homolytic splitting of water, which allows understanding the role of the oxyl character on the surface. Fig. 4. Transition states of the oxo-coupling chemical step. Distances in Å. That is, when 4O/O shows an important oxyl character such as on the (1 1 0), (0 1 1) and (1 0 0) surfaces the process is accessible and follows a reasonable correlation (Figure S2 of the Supplementary Materials). In contrast, the Ir=O species in the (0 0 1) surface is mainly an oxo group and the H2O attack is forbidden. From 4OOH/OH, two PCET steps (4OOH/OH to 5OO/OH and 5OO/OH to 6OO/O) take place before the release of O2 occurs. While the O2 release is a favorable chemical process, the two proton transfers are challenging. The H+ + e- transfer from 4OOH/OH involves the H from the hydroperoxo and it is electrochemically accessible on all surfaces at relatively low potentials.[78] The two resulting Ir- O distances in 5OO/OH are similar to those of 2OH/OH and 4OOH/OH, thus suggesting a single bond (See Tables 1 and 3). Moreover, the O-O distance of the originally OOH– species shortens (from about 1.4 to about 1.3 Å) and the magnetization on the OO increases. This is indicative of the formation of O– 2 and thus, the oxidation takes place at the O2-based ligand. The 5OO/OH to 6OO/O step is in all cases more unfavorable than the 4OOH/OH to 5OO/OH one. The geometrical parameters as well as the spin distribution on the ligands of 6OO/O Fig. 5. Lateral (a) and apical (b) view of the transition states associated with the water nucleophilic attack chemical step on the main IrO2 surfaces. Distances in Å. suggest that Ir-O has an important oxyl character on the (1 1 0) and (0 1 1) facets, with a relatively large Ir-O distance. In contrast, Ir=O is best viewed as an oxo group on the (1 0 0) and (0 0 1) surfaces. gies between 0.13 and 0.90 eV. This formally reduces the metal Since the iridium oxo bond (Ir=O) is stronger than the iridium oxyl center and regenerates 0H2O/H2O that as discussed in the first sec- one (Ir-O), the 5OO/OH to 6OO/O step becomes particularly challeng- tion can be easily reoxidized to 4O/O at potentials of 1.5 V with ing for the (1 1 0) and (0 1 1) surfaces, where the required potentials the exception of the (0 1 1) surface. Consequently, the electrochem- are higher than 1.5V. ical steps of the I2M mechanism in the (1 1 0), (1 0 0) and (0 0 1) are In summary, the WNA mechanism is accessible for the (1 1 0), accessible at working potentials. However, in these three surfaces, (0 1 1) and (1 0 0) surfaces at reaction conditions. Despite the this reaction mechanism is hampered by the energy cost of the required overpotentials for the WNA and I2M mechanisms are sim- chemical oxo–coupling step. The I2M mechanism at the (0 1 1) sur- ilar within each particular surface, the Gibbs energy barrier for O-O face appears to be chemically feasible (energy barrier of 0.33 eV), bond formation indicates that the WNA is the preferred mecha- but this surface is not easily oxidized and thus higher nism on the surfaces presenting iridium atoms with an axial vacant overpotentials would be required. site ((1 1 0) and (1 0 0)). In contrast, the I2M pathway is preferred From 4O/O, the WNA mechanism starts with a step where water on the (0 1 1) surface where the vacant site of underoordinated reacts with the surface. This process has been proposed to occur iridium centers is equatorial. Finally, the (0 0 1) surface does not either in a concerted manner or in two steps. In the first case, seem to be reactive if two metal centers are required to achieve the H+ is transferred to the solution and the OH couples with the process and this is associated with the formation of Ir=O oxo Ir=O implying the transfer of one electron to the electrode.[40] In species. the second case, H2O initially dissociates on the surface through We decided to explore if OER could occur on one single Ir center a chemical process that is followed by a proton coupled electron of the (0 0 1) surface, taking into account that Ir is tetracoordinated transfer [58,77] For the (1 1 0) surface, these pathways have been and has two vacant sites on this surface. Results are summarized reported to present low energy barriers. Here, we considered that on Fig. 6 and Table 2. Formation of two Ir-O groups requires over- water initially dissociates as H+/OH– at two oxo sites at the surface, coming higher potentials than those involved for the formation of forming 4OOH/OH. Despite the other possibilities may be competi- one single oxo group. However, the computed values are lower tive or even preferred, the here obtained values can be used as than 1.5 eV. This suggests that bisoxo species on the (0 0 1) are indicative of the feasibility of the water nucleophilic attack. The expected to be formed at reaction conditions. Moreover, structural reaction energies associated with the 4O/O to 4OOH/OH chemical step analysis indicates that the two oxo groups present Ir-O distances 198 D. González, J. Heras-Domingo, M. Sodupe et al. Journal of Catalysis 396 (2021) 192–201 that are significantly larger than that of the single oxo group on the (0 0 1) facet and the magnetization reveals a certain oxyl character (Fig. 6b). As a consequence, the two chemical steps (oxo-coupling and water nucleophilic attack) are now easier than those occurring between two different centers. For the WNA mechanism, formation of 4OOH/OH is slightly endergonic (DG0 = 0.25 eV) and the reaction presents a low Gibbs energy barrier (DGà = 0.55 eV). Therefore, OER on the (0 0 1) is expected to take place involving Ir centers with two oxyl groups and through the WNA mechanism. Fig. 7 and Figure S3 report the Gibbs energy profiles for the dif- ferent surfaces of the whole catalytic process occurring through the water nucleophilic attack (Fig. 7) and the oxo–coupling mech- anisms (Figure S3). Since the energy barriers for the proton coupled electron transfer steps are expected to be small as reported recently,[40] it is reasonable to assume that the largest Gibbs energy difference between two successive intermediates defines the required overpotential (g). In all cases, the computed overpo- tentials for the two reaction mechanisms are similar within each surface and thus, the applying mechanism is determined by the O-O bond formation step that it is likely rate-determining once the onset potential has been reached.[40] The energy barrier for Fig. 6. a) Phase diagram of IrO2 (0 0 1) surface at pH = 0, T = 298 K and P = 1 atm as the O-O bond formation suggests that the OER on the (1 1 0), function of the applied potential; b) Optimized structure for the Ir-bisoxo (1 0 0) and (0 0 1) surfaces proceed through the WNA mechanism intermediate; c) Transition state structure associated with the oxo-coupling chemical process; d) Transition state associated with the WNA step. Values in blue and the I2M pathway is preferred on the (0 1 1) surface. In these correspond to the spin densities over the oxo groups. Distances in Å. cases, the energy span at an applied potential of 1.5 V is small (be- tween 0.33 and 0.55 eV). In the (0 1 1) surface, the spin density over the oxyl radicals is high and the interoxyl distance rather short, 4.92 (110) 4.92 (100) 3.56 = 0.18 3.49 2.52 2.52 2.35 2.12 TS = 0.30 TS 0.43 0.42 1.30 0.00 0.99 0.00 -0.12 0.08 -0.48 -0.20 -0.48 -0.51 -0.65 -0.88 -0.94 -1.01 -1.08 -1.08 O O HO OOH HO OO O OO OH2O O OH O O O O HO OOH HO OO O OO OH2O O OH O O Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir 4.92 4.92 (011) (001)_1c = 0.41 = 0.24 3.28 3.45 2.69 2.32 2.30 2.48 TS TS 1.45 0.42 0.55 0.00 0.80 0.00 0.25 -0.05 -0.68 -0.05 -1.08 -0.31 -0.70 -0.70 -0.52 -1.22 -1.05 -1.08 O O HO OOH HO OO O OO OH2O O OH O O O O HO OOH HO OO O OO OH2O O OH O O Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Ir Fig. 7. Calculated Gibbs energy profile for the whole OER catalytic cycle through the water nucleophilic attack mechanism. Blue lines corresponds to the O-O bond formation, the black lines correspond to the energetics at an applied potential of 0 V and the red values are for an applied potential of 1.5 V. 199 D. González, J. Heras-Domingo, M. Sodupe et al. Journal of Catalysis 396 (2021) 192–201 which appears important for I2M mechanism to occur. The is always accessible and easy, while the oxo-coupling chemical computed overpotentials are: 0.30, 0.41, 0.18 and 0.24 V for the step requires that the Ir-O radicals are also well-oriented. Overall, (1 1 0), (0 1 1), (1 0 0) and (0 0 1) surfaces (Fig. 7). Moreover, the the water nucleophilic attack mechanism is preferred on the (1 1 0), number of Ir vacant sites density varies as: (0 0 1) > (0 1 1) > (1 0 0) and (0 0 1) facets, while the I2M pathway is the main route (1 0 0) > (1 1 0) facet. Therefore, calculations predict that the highest on the (0 1 1) surface. The importance of oxyl groups on molecular activities per surface area would be obtained on the (1 0 0) and or first row transition metal oxide catalysts has been outlined (0 0 1), as they require lower overpotentials for catalyzing the before.[7,81,83,91] Present results suggest that molecular and OER and they present a larger number of active sites per surface heterogeneous catalysts could have larger similarities than initially unit. In contrast, the (0 1 1) and (1 1 0) facets present either a smal- expected and thus strategies for catalyst development on molecu- ler number of active sites per nm2 or require higher overpotentials lar systems may also be suitable to improve heterogeneously for catalyzing the process. These results reproduce the experimen- catalyzed OER tal trend when comparing the (1 0 0) and (1 1 0) surfaces[16] and The required overpotentials for the OER to occur on the four suggest that (0 0 1) surface could be as reactive as the (1 0 0) facet. IrO2 considered surfaces are determined by the feasibility to oxi- Our results suggest that the reactivity on the different surfaces dize the Ir-OH species to Ir-O (either the 3O/OH to 4O/O step or the is highly controlled by two main factors. On one hand, the oxo- 5OO/OH to 6OO/O one). However, if the oxidation is too easy the coupling and the water nucleophilic attack chemical steps are only resulting Ir=O species has no oxyl character and thus, does not fur- feasible when surface Ir-O species has a significant oxyl character. ther react. The iridium coordination and the nature of the vacant Indeed, the computed energy barriers are highly determined by the sites tunes both the oxidation ability and oxyl character. Tetracoor- spin density over the oxyl group and the interatomic distance dinated species are so unsaturated that formation of one Ir=O is between Ir-O species. At this point it is worth mentioning that, easy, but the resulting species do not further react. Pentacoordi- while the role of the oxyl character in determining the catalytic nated centers with equatorial vacant sites are hardly oxidized to activity of molecular complexes[7,79–88] and first row transition Ir=O, due to the weaker Ir-OL interaction. The pentacoordinated metal oxides[89–95] has been discussed and shown to be impor- centers with axial vacant sites show an intermediate behavior tant, its role in IrO2 has been mainly eluded although oxyl species and they appear to be more active. Finally, tetracoordinated irid- on IrO2 has recently been detected by X-Ray spectroscopy. ium sites on the (0 0 1) surface can be oxidized to bisoxo species [58,96,97] According to our data, the IrO2 surfaces and the Ir able to perform OER at low potentials, which appears as promising molecular analogues could have more similarities than one could candidates to be explored experimentally. initially suspect. Funding On the other hand, regardless of the considered surface, the This work was supported by the Ministerio de Ciencia e overpotential required to perform the OER is determined by the Inovación MICINN [CTQ2017-89132-P]; the Generalitat de oxidation of the Ir-OH hydroxyl group to Ir-O species (3OH/O to Catalunya [2017SGR1323] and Red Española de Supercomputación 4O/O step or 5OO/OH to 6OO/O), except if the process is so easy that [QCM-2019–1-0043, QS-2019–2-0031]. the resulting species has no oxyl character. Indeed, the two factors are closely related and a subtle compromise between them is a key Declaration of Competing Interest issue. Analysis of the electronic structure of the four surfaces shows The authors declared that there is no conflict of interest. that the two variables are tuned by the coordination and nature of the vacant sites of the surface iridium centers. Unsaturated Acknowledgements tetracoordinated centers are so poorly stabilized that the forma- tion of one Ir=O group per metal center is easy. However, when this The authors gratefully acknowledge financial support from Ir=O group is formed on these centers, it does not have an oxyl MICINN [CTQ2017-89132-P], the Generalitat de Catalunya character and thus it is not reactive. The oxidation of unsaturated [2017SGR1323] and Red Española de Supercomputación [QCM- pentacoordinated iridium centers is more challenging, the Ir-O 2019-1-0043, QS-2019-2-0031]. XSM is grateful for the Professor group has a significant oxyl character and, consequently, they are Agregat Serra Húnter position. more prone to react. When the vacant site is equatorial the Ir-OL interaction is weaker than for the other cases and, overall, higher Appendix A. Supplementary data overpotentials are required. In contrast, the stronger Ir-OL interac- tion at axial sites, leads to a good compromise between the oxyl Supplementary data to this article can be found online at character and oxidation feasibility. Oxidation of the tetracoordi- https://doi.org/10.1016/j.jcat.2021.02.026. nated sites of the (0 0 1) surface to bisoxo species is as challenging as the oxidation of the (1 0 0) and (1 1 0) surfaces and the resulting References species has enough oxyl character to undergo the WNA. Overall, this turns in very promising candidates that resembles the recently [1] R. Winkler-Goldstein, A. Rastetter, Green 3 (2013) 69–78. reported single-site Iridium species on ITO electrodes.[47] [2] G. Gahleitner, Int. J. Hydrogen Energy 38 (2013) 2039–2061. [3] L. Hammarström, Acc. Chem. Res. 42 (2009) 1859–1860. [4] J.H. Montoya, L.C. Seitz, P. Chakthranont, A. Vojvodic, T.F. Jaramillo, J.K. Nørskov, Nat. Mater. 16 (2017) 70–81. 4. Conclusions [5] C. Spöri, J.T.H. Kwan, A. Bonakdarpour, D.P. Wilkinson, P. Strasser, Angew. Chem. Int. Ed. 56 (2017) 5994–6021. [6] M.G. Walter, E.L. Warren, J.R. McKone, S.W. Boettcher, Q. Mi, E.A. Santori, N.S. The catalytic activity for the oxygen evolution reaction of the Lewis, Chem. Rev. 110 (2010) 6446–6473. (1 1 0), (0 1 1), (1 0 0) and (0 0 1) main surfaces of rutile-like IrO2 [7] X. Sala, S. Maji, R. Bofill, J. García-Antón, L. Escriche, A. Llobet, Acc. Chem. Res. 47 (2014) 504–516. has been studied through spin-polarized PBE-D2 calculations. Both [8] M.D. Kärkäs, O. Verho, E.V. Johnston, B. Åkermark, Chem. Rev. 114 (2014) the oxo-coupling (I2M) and the water nucleophilic attack mecha- (2001) 11863–12001. nisms have been considered and the energy barriers of the associ- [9] J.D. Blakemore, R.H. Crabtree, G.W. Brudvig, Chem. Rev. 115 (2015) 12974– ated chemical processes calculated. Results show that the two 13005. [10] B.M. Hunter, H.B. Gray, A.M. Müller, Chem. Rev. 116 (2016) 14120–14136. mechanisms only take place when the Ir-O species on the surface [11] M.S. Burke, L.J. Enman, A.S. Batchellor, S. Zou, S.W. Boettcher, Chem. Mater. 27 has an oxyl character. In these cases, the water nucleophilic attack (2015) 7549–7558. 200 D. González, J. Heras-Domingo, M. Sodupe et al. Journal of Catalysis 396 (2021) 192–201 [12] E. Fabbri, A. Habereder, K. Waltar, R. Kötz, T.J. Schmidt, Catal, Sci. Technol. 4 [52] Z. Xu, J. Rossmeisl, J.R. Kitchin, J. Phys. Chem. C 119 (2015) 4827–4833. (2014) 3800–3821. [53] M. García-Melchor, L. Vilella, N. López, A. Vojvodic, ChemCatChem 8 (2016) [13] T. Reier, H.N. Nong, D. Teschner, R. Schlögl, P. Strasser, Adv. Energy Mater. 7 1792–1798. (2017) 1601275. [54] R.R. Rao, M.J. Kolb, N.B. Halck, A.F. Pedersen, A. Mehta, H. You, K.A. Stoerzinger, [14] C.C.L. McCrory, S. Jung, I.M. Ferrer, S.M. Chatman, J.C. Peters, T.F. Jaramillo, J. Z. Feng, H.A. Hansen, H. Zhou, L. Giordano, J. Rossmeisl, T. Vegge, I. Am. Chem. Soc. 137 (2015) 4347–4357. Chorkendorff, I.E.L. Stephens, Y. Shao-Horn, Energy Environ. Sci. 10 (2017) [15] J. Yu, Q. He, G. Yang, W. Zhou, Z. Shao, M. Ni, ACS Catal. 9 (2019) 9973–10011. 2626–2637. [16] K.A. Stoerzinger, L. Qiao, M.D. Biegalski, Y. Shao-Horn, J. Phys. Chem. Lett. 5 [55] Y. Ping, R.J. Nielsen, W.A. Goddard, J. Am. Chem. Soc. 139 (2017) 149–155. (2014) 1636–1641. [56] E. Watanabe, H. Ushiyama, K. Yamashita, Y. Morikawa, D. Asakura, M. Okubo, [17] K.A. Stoerzinger, O. Diaz-Morales, M. Kolb, R.R. Rao, R. Frydendal, L. Qiao, X.R. A. Yamada, J. Phys. Chem. C 121 (2017) 18975–18981. Wang, N.B. Halck, J. Rossmeisl, H.A. Hansen, T. Vegge, I.E.L. Stephens, M.T.M. [57] J.A. Gauthier, C.F. Dickens, L.D. Chen, A.D. Doyle, J.K. Nørskov, J. Phys. Chem. C Koper, Y. Shao-Horn, ACS Energy Lett. 2 (2017) 876–881. 121 (2017) 11455–11463. [18] H.-S. Oh, H.N. Nong, T. Reier, A. Bergmann, M. Gliech, J. Ferreira de Araújo, E. [58] H.N. Nong, L.J. Falling, A. Bergmann, M. Klingenhof, H.P. Tran, C. Spöri, R. Mom, Willinger, R. Schlögl, D. Teschner, P. Strasser, J. Am. Chem. Soc. 138 (2016) J. Timoshenko, G. Zichittella, A. Knop-Gericke, S. Piccinin, J. Pérez-Ramírez, B.R. 12552–12563. Cuenya, R. Schlögl, P. Strasser, D. Teschner, T.E. Jones, Nature 587 (2020) 408– [19] D.F. Abbott, D. Lebedev, K. Waltar, M. Povia, M. Nachtegaal, E. Fabbri, C. 413. Copéret, T.J. Schmidt, Chem. Mater. 28 (2016) 6591–6604. [59] J. Heras-Domingo, M. Sodupe, X. Solans-Monfort, J. Phys. Chem. C 123 (2019) [20] O. Kasian, S. Geiger, T. Li, J.-P. Grote, K. Schweinar, S. Zhang, C. Scheu, D. Raabe, 7786–7798. S. Cherevko, B. Gault, K.J.J. Mayrhofer, Energy Environ. Sci. 12 (2019) 3548– [60] D. González, J. Heras-Domingo, S. Pantaleone, A. Rimola, L. Rodríguez-Santiago, 3555. X. Solans-Monfort, M. Sodupe, ACS Omega 4 (2019) 2989–2999. [21] J. Cheng, J. Yang, S. Kitano, G. Juhasz, M. Higashi, M. Sadakiyo, K. Kato, S. [61] G. Kresse, J. Hafner, Phys. Rev. B 47 (1993) 558–561. Yoshioka, T. Sugiyama, M. Yamauchi, N. Nakashima, ACS Catal. 9 (2019) 6974– [62] G. Kresse, J. Furthmüller, Phys. Rev. B 54 (1996) 11169–11186. 6986. [63] J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77 (1996) 3865–3868. [22] E. Willinger, C. Massué, R. Schlögl, M.G. Willinger, J. Am. Chem. Soc. 139 (2017) [64] S. Grimme, J. Comput. Chem. 25 (2004) 1463–1473. 12093–12101. [65] P.E. Blöchl, Phys. Rev. B 50 (1994) 17953–17979. [23] Y. Sun, L. Yuan, Z. Liu, Q. Wang, K. Huang, S. Feng, Mater. Chem. Front. 3 (2019) [66] G. Kresse, D. Joubert, Phys. Rev. B 59 (1999) 1758–1775. 1779–1785. [67] H.J. Monkhorst, J.D. Pack, Phys. Rev. B 13 (1976) 5188–5192. [24] M. Povia, D.F. Abbott, J. Herranz, A. Heinritz, D. Lebedev, B.-J. Kim, E. Fabbri, A. [68] D. González, B. Camino, J. Heras-Domingo, A. Rimola, L. Rodríguez-Santiago, X. Patru, J. Kohlbrecher, R. Schäublin, M. Nachtegaal, C. Copéret, T.J. Schmidt, Solans-Monfort, M. Sodupe, J. Phys. Chem. C 124 (2020) 1227–1237. Energy Environ. Sci. 12 (2019) 3038–3052. [69] K. Mathew, R. Sundararaman, K. Letchworth-Weaver, T.A. Arias, R.G. Hennig, J. [25] A. Minguzzi, O. Lugaresi, E. Achilli, C. Locatelli, A. Vertova, P. Ghigna, S. Chem. Phys. 140 (2014), 084106. Rondinini, Chem. Sci. 5 (2014) 3591. [70] G. Henkelman, B.P. Uberuaga, H. Jónsson, J. Chem. Phys. 113 (2000) 9901– [26] H.G. Sanchez Casalongue, M.L. Ng, S. Kaya, D. Friebel, H. Ogasawara, A. Nilsson, 9904. Angew. Chem. Int. Ed. 53 (2014) 7169–7172. [71] G. Henkelman, H. Jónsson, J. Chem. Phys. 113 (2000) 9978–9985. [27] V. Pfeifer, T.E. Jones, J.J. Velasco Vélez, C. Massué, M.T. Greiner, R. Arrigo, D. [72] R. Mu, D.C. Cantu, X. Lin, V.-A. Glezakou, Z. Wang, I. Lyubinetsky, R. Rousseau, Teschner, F. Girgsdies, M. Scherzer, J. Allan, M. Hashagen, G. Weinberg, S. Z. Dohnálek, J. Phys. Chem. Lett. 5 (2014) 3445–3450. Piccinin, M. Hävecker, A. Knop-Gericke, R. Schlögl, PCCP 18 (2016) 2292–2296. [73] M.-T. Nguyen, R. Mu, D.C. Cantu, I. Lyubinetsky, V.-A. Glezakou, Z. Dohnálek, R. [28] A.F. Pedersen, M. Escudero-Escribano, B. Sebok, A. Bodin, E. Paoli, R. Frydendal, Rousseau, J. Phys. Chem. C 121 (2017) 18505–18515. D. Friebel, I.E.L. Stephens, J. Rossmeisl, I. Chorkendorff, A. Nilsson, J. Phys. [74] R.R. Rao, M.J. Kolb, J. Hwang, A.F. Pedersen, A. Mehta, H. You, K.A. Stoerzinger, Chem. B 122 (2018) 878–887. Z. Feng, H. Zhou, H. Bluhm, L. Giordano, I.E.L. Stephens, Y. Shao-Horn, J. Phys. [29] C.-J. Chang, Y.-C. Chu, H.-Y. Yan, Y.-F. Liao, H.M. Chen, Dalt. Trans. 48 (2019) Chem. C 122 (2018) 17802–17811. 7122–7129. [75] C.J. Cramer, W.B. Tolman, K.H. Theopold, A.L. Rheingold, Proc. Natl. Acad. Sci. [30] V. Pfeifer, T.E. Jones, J.J. Velasco Vélez, C. Massué, R. Arrigo, D. Teschner, F. 100 (2003) 3635–3640. Girgsdies, M. Scherzer, M.T. Greiner, J. Allan, M. Hashagen, G. Weinberg, S. [76] P.L. Holland, Dalt. Trans. 39 (2010) 5415. Piccinin, M. Hävecker, A. Knop-Gericke, R. Schlögl, Surf. Interface Anal. 48 [77] We also explored the possibility that the water attack occurs at 3OH/O in the (2016) 261–273. (011) surface. However, we were unable to find any minimum of 3OOH/H2O, [31] P. Steegstra, M. Busch, I. Panas, E. Ahlberg, J. Phys. Chem. C 117 (2013) 20975– thus suggesting that 3OH/O has to be oxidized before to 4O/O. 20981. [78] We also considered the possibility that process involves H from hydroxyl, but [32] N. Sivasankar, W.W. Weare, H. Frei, J. Am. Chem. Soc. 133 (2011) 12976– the computed energies are similar or higher (see Supplementary Materials). 12979. [79] P.E.M. Siegbahn, R.H. Crabtree, J. Am. Chem. Soc. 121 (1999) 117–127. [33] S. Fierro, T. Nagel, H. Baltruschat, C. Comninellis, Electrochem. Commun. 9 [80] L.-P. Wang, T. Van Voorhis, J. Phys. Chem. Lett. 2 (2011) 2200–2204. (2007) 1969–1974. [81] G. Mattioli, P. Giannozzi, A. Amore Bonapasta, L. Guidoni, J. Am. Chem. Soc. 135 [34] K. Macounova, M. Makarova, P. Krtil, Electrochem. Commun. 11 (2009) 1865– (2013) 15353–15363. 1868. [82] D.W. Crandell, S. Xu, J.M. Smith, M.-H. Baik, Inorg. Chem. 56 (2017) 4435– [35] H. Dau, C. Limberg, T. Reier, M. Risch, S. Roggan, P. Strasser, ChemCatChem 2 4445. (2010) 724–761. [83] P.E.M. Siegbahn, Acc. Chem. Res. 42 (2009) 1871–1880. [36] E. Fabbri, T.J. Schmidt, ACS Catal. 8 (2018) 9765–9774. [84] X. Yang, M.-H. Baik, J. Am. Chem. Soc. 128 (2006) 7476–7485. [37] J. Rossmeisl, Z.-W. Qu, H. Zhu, G.-J. Kroes, J.K. Nørskov, J. Electroanal. Chem. [85] K.R. Yang, A.J. Matula, G. Kwon, J. Hong, S.W. Sheehan, J.M. Thomsen, G.W. 607 (2007) 83–89. Brudvig, R.H. Crabtree, D.M. Tiede, L.X. Chen, V.S. Batista, J. Am. Chem. Soc. 138 [38] Y.-H. Fang, Z.-P. Liu, J. Am. Chem. Soc. 132 (2010) 18214–18222. (2016) 5511–5514. [39] L.G.V. Briquet, M. Sarwar, J. Mugo, G. Jones, F. Calle-Vallejo, ChemCatChem 9 [86] D.W. Shaffer, Y. Xie, J.J. Concepcion, Chem. Soc. Rev. 46 (2017) 6170–6193. (2017) 1261–1268. [87] L. Duan, F. Bozoglian, S. Mandal, B. Stewart, T. Privalov, A. Llobet, L. Sun, Nat. [40] C.F. Dickens, C. Kirk, J.K. Nørskov, J. Phys. Chem. C 123 (2019) 18960–18977. Chem. 4 (2012) 418–423. [41] C.F. Dickens, J.K. Nørskov, J. Phys. Chem. C 121 (2017) 18516–18524. [88] C.N. Brodsky, R.G. Hadt, D. Hayes, B.J. Reinhart, N. Li, L.X. Chen, D.G. Nocera, [42] K. Klyukin, A. Zagalskaya, V. Alexandrov, J. Phys. Chem. C 123 (2019) 22151– Proc. Natl. Acad. Sci. 114 (2017) 3855–3860. 22157. [89] C.Y. Cummings, F. Marken, L.M. Peter, K.G. Upul Wijayantha, A.A. Tahir, J. Am. [43] A. Zagalskaya, V. Alexandrov, J. Phys. Chem. Lett. 11 (2020) 2695–2700. Chem. Soc. 134 (2012) 1228–1234. [44] A. Zagalskaya, V. Alexandrov, ACS Catal. 10 (2020) 3650–3657. [90] B. Klahr, T. Hamann, J. Phys. Chem. C 118 (2014) 10393–10399. [45] R.R. Rao, M.J. Kolb, L. Giordano, A.F. Pedersen, Y. Katayama, J. Hwang, A. Mehta, [91] D.M. Herlihy, M.M. Waegele, X. Chen, C.D. Pemmaraju, D. Prendergast, T. Cuk, H. You, J.R. Lunger, H. Zhou, N.B. Halck, T. Vegge, I. Chorkendorff, I.E.L. Nat. Chem. 8 (2016) 549–555. Stephens, Y. Shao-Horn, Nat. Catal. 3 (2020) 516–525. [92] J.G. McAlpin, T.A. Stich, C.A. Ohlin, Y. Surendranath, D.G. Nocera, W.H. Casey, R. [46] K. Klyukin, A. Zagalskaya, V. Alexandrov, J. Phys. Chem. C 122 (2018) 29350– D. Britt, J. Am. Chem. Soc. 133 (2011) 15444–15452. 29358. [93] D.K. Bediako, C. Costentin, E.C. Jones, D.G. Nocera, J.-M. Savéant, J. Am. Chem. [47] D. Lebedev, R. Ezhov, J. Heras-Domingo, A. Comas-Vives, N. Kaeffer, M. Soc. 135 (2013) 10492–10502. Willinger, X. Solans-Monfort, X. Huang, Y. Pushkar, C. Copéret, ACS Cent. Sci. 6 [94] X. Chen, S.N. Choing, D.J. Aschaffenburg, C.D. Pemmaraju, D. Prendergast, T. (2020) 1189–1198. Cuk, J. Am. Chem. Soc. 139 (2017) 1830–1841. [48] S. Divanis, T. Kutlusoy, I.M. Ingmer Boye, I.C. Man, J. Rossmeisl, Chem. Sci. 11 [95] A.M. Ullman, C.N. Brodsky, N. Li, S.-L. Zheng, D.G. Nocera, J. Am. Chem. Soc. 138 (2020) 2943–2950. (2016) 4229–4236. [49] I.C. Man, H. Su, F. Calle-Vallejo, H.A. Hansen, J.I. Martínez, N.G. Inoglu, J. [96] L.J. Frevel, R. Mom, J.-J. Velasco-Vélez, M. Plodinec, A. Knop-Gericke, R. Schlögl, Kitchin, T.F. Jaramillo, J.K. Nørskov, J. Rossmeisl, ChemCatChem 3 (2011) 1159– T.E. Jones, J. Phys. Chem. C 123 (2019) 9146–9152. 1165. [97] Z. Pavlovic, C. Ranjan, M. van Gastel, R. Schlögl, Chem. Commun. 53 (2017) [50] O. Piqué, F. Illas, F. Calle-Vallejo, PCCP 22 (2020) 6797–6803. 12414–12417. [51] S. Siahrostami, A. Vojvodic, J. Phys. Chem. C 119 (2015) 1032–1037. 201