This is the accepted version of the journal article: González, Danilo; Sodupe Roure, Mariona; Rodríguez Santiago, Luis; [et al.]. «Surface morphology controls water dissociation on hydrated IrO2 nanoparti- cles». Nanoscale, Vol. 13, Issue 34 (September 2021), p. 14480-14489. DOI 10.1039/d1nr03592d This version is available at https://ddd.uab.cat/record/277742 under the terms of the license Please do not adjust margins ARTICLE Surface morphology controls water dissociation on hydrated IrO2 nanoparticles Received 00th January 20xx, Danilo González,a Mariona Sodupe,a Luis Rodríguez-Santiago,a Xavier Solans-Monfort*,a Accepted 00th January 20xx Iridium oxide is a highly efficient catalyst for the oxygen evolution reaction, whose large-scale application requires DOI: 10.1039/x0xx00000x decreasing the metal content. This is achieved using small nanoparticles. The knowledge of the water-IrO2 nanoparticle interface is of high importance to understand the IrO2 behavior as electrocatalyst in aqueous solutions. In this contribution, DFT (PBE-D2) calculations and AIMD simulations on IrO2 nanoparticle models of different sizes ((IrO2)33 and (IrO2)115) are performed. Results show that two key factors determine the H2O adsorption energy and the preferred adsorption structure (molecular or dissociated water): metal coordination and hydrogen bonding with oxygen bridge atoms of the IrO2 surface. Regarding metal coordination, and since the tetragonal distortion existing in IrO2 is retained on the nanoparticle models, the adsorption at iridium axial vacant sites implies stronger Ir-H2O interactions, which favors water dissociation. In contrast, Ir- H2O interaction at equatorial vacant sites is weaker and thus the relative stability of molecular and dissociated forms become similar. Hydrogen bonding increases adsorption energy and favors water dissociation. Thus, tip and corner sites of the nanoparticle, with no oxygen bridge atoms nearby, exhibit the smallest adsorption energies and a preference for the molecular form. Overall, the presence of rather isolated tip and corner sites in the nanoparticle leads to lower adsorption energies and a smaller degree of water dissociation when compared with extended surfaces. . adsorption energies between -211.5 and -152.9 kJ mol-1 and a Introduction degree of water dissociation of 100% on the most stable (110) surface, while this percentage decreases to around 60% in the Iridium oxide (IrO2) is a highly efficient catalyst for the oxygen (011) facet.35 evolution reaction (OER) in acidic media with high activity and Remarkably, despite the importance of IrO2 nanoparticles, to stability1–7 Moreover, it has been recently suggested as the best of our knowledge, the effect of reducing the material potential catalyst for other reactions such as converting light dimensionality from films to nanoparticles in the H2O – IrO2 alkanes in value-added products.8–10 Understanding its chemical interface has not been addressed neither from experiments nor properties is key for the further development of novel iridium calculations. Nanoparticles either amorphous or crystalline are oxide-based substances. This implies nanostructured materials expected to present metal sites that are different from those with a high specific surface areas since the wide application of iridium-based materials suffers from the metal cost.11–17 In this found in extended surfaces, i.e. tip, corner and edges sites. context, one can find a large list of examples where IrO2 has Moreover, the nanoparticle shape can also play a role in been used in the form of nanoparticles of different sizes16,18–29 favoring or decreasing the cooperative effects between including very small and narrowly distributed nanoparticles of adsorbed molecules. Indeed, this has recently been analyzed on less than 2 nm, which show outstanding catalytic activities for H2O-nanoparticle interfaces of related materials.37,38 Therefore, the OER reaction.16,20,23,25,28 it is likely that the local structure of the IrO2 nanoparticle – H2O A deep knowledge of the water-IrO2 interface is required to interface is different from that of films. Indeed, contributions further understand the IrO2 behavior as electrocatalyst in on related hydrated TiO2 nanoparticles highlighted their aqueous solutions.30 Thus, several authors have studied the differences with respect to crystalline facets and allowed to water – IrO2 interaction and water – IrO2 interface.31–33 identify the distribution of OH groups over the surface as well Although the majority of the studies have focused on the most as to analyze size effects.37–41 In this contribution we focus on stable (110) surface, some of the most recent experimental and analyzing how the IrO2 – H2O interface varies when going from extended IrO2 slabs to small Wulff-like nanoparticles. Results computational contributions on related MO2 rutile-like show that IrO2 nanoparticle models are somewhat more ionic materials outlined that the degree of water dissociation is than extended surfaces and this favors the direct H2O-metal orientation and material dependent.34–36 Overall, IrO2 shows interaction. However, the convex shape of the nanoparticle prevents the formation of strong hydrogen bonds either a. Departament de Química, Universitat Autònoma de Barcelona, 08193 Bellaterra, between the surface or adjacent adsorbed molecules. Spain Moreover, the number of accessible oxygen bridge that assist Electronic Supplementary Information (ESI) available: [details of any supplementary information available should be included here]. See DOI: 10.1039/x0xx00000x water dissociation is more limited. Overall, these two factors Please do not adjust margins Please do not adjust margins ARTICLE Nanoscale decrease the degree of dissociated water molecules on the surface of the nanoparticle with respect to thin films and leads to lower adsorption energies per water molecule, particularly at the most accessible corner and tip sites. Computational details Models We use a four-layers thick slab to represent the main crystallographic surfaces. A (2x2) supercell was considered for Figure 1. (a) (IrO2)33 and (b) (IrO2)115 nanoparticle models with the different each facet to avoid lateral interactions with neighbor images. undercoordinated iridium centers (colored atoms). These models are equivalent to those used in our group for modeling RuO2, TiO2 and IrO2 surfaces.34,35,42 For the (IrO2)115, respectively. This ensures at least 18 Å between nanoparticles we use two stoichiometric models of 1.2 and 1.8 images. We used PBE density functional46 with Grimme’s D2 nm large constructed with BCN-M, a python-based empirical correction47 to account for dispersion forces. The computational tool able to construct Wulff-like nanoparticle effect of the dispersion correction was analyzed by performing models (Figure 1).43 For that, we considered the surface test calculations on sites G4ax/eq and I4eq/eq of the (IrO2)33 energies of the four most representative families of IrO2 facets nanoparticle with PBE-D3. Results are reported in Table S1 and ({110}, {100}, {011} and {001}) computed in our previous work show that PBE-D2 and PBE-D3 leads to equivalent results. The (0.150, 0.158, 0.185 and 0.206 eV A-2 for (110), (011), (100) and ionic cores were described with PAW pseudopotentials48,49 and (001), respectively).35 The initial models contain 33 and 115 IrO2 the valence electrons were represented with a plane-wave basis units, respectively, and arise from cutting the bulk, thus the set with an energy cutoff of 500 eV. All calculations were geometry around undercoordinated metal centers is far from performed at gamma point. This methodology is equivalent to optimal and a further geometry optimization accounting for the the one we used for describing the water adsorption on the surface relaxation is performed. Optimization in vacuum implies different crystallographic facets of RuO234 and for studying the reorganization around the nanoparticle tip with the formation OER activity on IrO2 surfaces42 and on isolated iridium atoms of Ir-Ir bonds. This reorganization does not occur when a water supported on iridium tin oxide.15 In the three cases, the monolayer is added on the nanoparticle surface (See details in computational data reproduced the experimental the Electronic Supplementary Information (ESI)). Since IrO2 evidences,15,36,50 indicating that the computational protocol is nanoparticle are usually synthetized in aqueous solution, accurate enough for reproducing main trends on these optimization in the presence of water is more realistic. materials. Therefore, we decided to perform constrained optimizations of Thermal effects at 1 atm and different temperatures were taken (IrO2)33 and (IrO2)115 nanoparticle models taking the geometries into account for the adsorption of two water molecules at the of the optimized (IrO2)33•38H2O and (IrO2)115•88H2O G4ax/eq and I4eq/eq sites of the (IrO2)33 nanoparticle. Only the nanoparticle monolayer systems, respectively, that includes vibrational contributions of the normal modes associated with one water molecule per iridium vacant site. In the constrained the adsorbed species (H2O or OH-/H+) were considered. Table S3 optimizations, the nanoparticle is frozen and only water shows the relative stabilities of the different forms at four molecules and the singly coordinated oxygen atoms are allowed different temperatures, the values suggesting a small influence to relax. Analysis of the different unsaturated metal centers of the temperature in favoring either the molecular or after removing the water molecules reveal that there are eleven dissociated forms. and six different sites for the (IrO2)115 and (IrO2)33 nanoparticles We also performed spin-polarized calculations with the aim of respectively (Figure 1). The existing sites can be divided in five analyzing if reducing the material size has an effect on its groups: i) pentacoordinated centers with an axial vacant site magnetic properties. The results with the spin-polarized (A5ax and B5ax); ii) pentacoordinated centers with an equatorial formalism are reported in the Electronic Supplementary vacant site (C5eq, D5eq, E5eq and F5eq); iii) tetracoordinated Information and show that water adsorption is only marginally centers with one axial and one equatorial vacant site (G4ax/eq affected by the magnetism of the nanoparticle and more and H4ax/eq); iv) tetracoordinated centers with two equatorial importantly, the same preference for the molecular or vacant sites (I4eq/eq and J4ax/eq); and v) metal centers presenting dissociated form is found for each site when using spin- or non- only three bonds with lattice oxygens (K3). spin polarized calculations. Thus, the values reported in the main text are all obtained with the non-spin polarized Level of theory formalism. With the aim of exploring the different water – MO2 interface The electronic structure of (IrO2)N and the water adsorption at high coverages, we also performed ab-initio molecular energies to these nanoparticle models are studied with the dynamics at 300 K of the (IrO2)33•38H2O monolayer in the NVT VASP package44,45 which uses periodic DFT calculations in the ensemble. These simulations were performed at the same level three dimensions. Consequently, calculations are performed by of theory as that of geometry optimizations. Each simulation is considering a 30 and a 35 Å edge large cubic box for (IrO2)33 and 2 | J. Name., 2012, 00, 1-3 This journal is © The Royal Society of Chemistry 20xx Please do not adjust margins Please do not adjust margins Nanoscale ARTICLE 10 ps large and the analysis was performed on the last 5 ps. The and (100) surfaces are high and thus, there is a preference for time step size was fixed to 1 fs. During the AIMD simulations the dis form. On the other side, the (011) surface has only the singly coordinated atoms of the nanoparticle model pentacoordinated iridium centers with the vacant site in an and the water molecules were allowed to move. equatorial position. Consequently, the interaction is weaker Electronic structure analysis is performed by computing the than in the two previous surfaces and the mol form is preferred. iridium d- and oxygen p-orbital band centers of the projected Finally, the (001) surface presents tetracoordinated centers density of states (PDOS) as defined previously:51–53 with two equatorial vacant sites. Therefore, the metals on this !"#$ surface are the most undercoordinated centers and the ∫! &'(&)*+ 𝜀!",$ = "%& !"#$ (1) adsorption of the first water molecule is strong, with the dis ∫! '(&)*+ "%& form being lower in energy by 1.4 kJ mol-1. The adsorption of a second water molecule is not such favorable. The metal center Where 𝜀!",$ is the band center, e is the energy with respect to is less electron deficient after the adsorption of the first water the Fermi level and r(e) is the PDOS. In our case, the emin and molecule and thus the mean adsorption energy per water emax values are set to -10 and 5eV around the Fermi level, molecule decreases significantly. Overall, the preferred respectively. structure on the (001) surface present two non-dissociated water molecules in line with the weak adsorption energy for the Results and discussion coordination to an equatorial site. We organized the results and discussion section in four parts. Table 1. H2O adsorption energies on the main crystallographic surfaces (in kJ mol- 1 ), Ir-O and H···O hydrogen bond distances (in Å) and Iridium d orbital (e ) and First, we describe the intrinsic adsorption of one water Ir oxygen p orbital (eO) band centers of the atoms involved in the adsorption and molecule on the main crystallographic facets of IrO2. In a second potential dissociation. step, we analyze the metal – water interaction at the different sites of the (IrO2)33 Wulff like nanoparticle (Figure 1). In the third Site Eads Ir-O O···H eIr eO part, we describe the effect of increasing the nanoparticle size. (110) Dis -211.5 1.970 2.043 -2.2 -2.5 Finally, we discuss the cooperative and thermal effects on the (100) mol -191.2 2.074 1.683 -2.5 -3.0 smallest nanoparticle by modeling a water monolayer coverage, Dis -193.1 1.960 1.688 formed by adding one water molecule per vacant site (38 water (011) mol -153.0 2.108 1.549 -2.9 -3.7 Dis -146.0 2.025 1.372 molecules on (IrO2)33). (001) mol -201.7 2.086 1.591 -3.0 -3.3 Dis -203.1 1.982 1.632 Intrinsic water – IrO2 main crystallographic facets interaction. Site Eads Ir-O Ir-O eIr eO The iridium centers present a distorted octahedral coordination (001)-2H2O mol/mol -183.1 2.099 2.098 -3.0 -3.3 dis/mol -177.4 1.991 2.103 characterized by a tretragonal compression with two short axial dis/dis -175.5 1.983 2.036 distances and four long equatorial ones, thus leading to two types of vacant sites for the iridium centers on the surface. The The differences in adsorption energies can also be rationalized adsorption of water on IrO2 occurs on these undercoordinated by analyzing the projected density of states (DOS) of the surface iridium atoms. The Ir-H2O adsorption implies charge transfer iridium d and oxygen p orbitals and, particularly the band center from H2O to iridium. This increases the acidity of water and (e) of each type of orbitals (Table 1 and Figure S3 of the ESI). thus, it favors water dissociation by transferring a proton to an According to previous contributions, iridium d-band centers oxygen bridge atom of the surface. Indeed, two conformations close to the Fermi Level (less negative) lead to stronger metal- are plausible: the molecular (mol) and dissociated (dis) forms adsorbate interactions,51–53 since in the case of metal oxides this and their relative stability mainly depends on the iridium-H2O is indicative of the presence of antibonding empty states51 that interaction and the basicity of the oxygen bridge. Table 1 accept the electron density from the adsorbate. The d-band reports the H2O adsorption energies on the four surfaces, the center of unsaturated metals with axial vacant sites ((110) and main geometrical parameters, and the iridium d- and oxygen p- (100) surfaces) have less negative values than those of the (011) DOS band centers. The H2O adsorption energies range between and (001) surfaces. Therefore, the Ir-H2O interaction is higher in -211.5 and -146.0 kJ mol-1. The highest adsorption energy is the formers than in the latters. Moreover, the center of the p- computed for the (110) surface, while the weakest interaction oxygen band can also be used to determine the basicity of the takes place between water and the iridium centers of the (011) oxygen bridge: the closer to the Fermi level the p band center facet. is, the closer to the Fermi level the doubly occupied p orbitals The undercoordinated iridium centers on the main of oxygen bridge are and thus, the higher basic character these crystallographic surfaces present three different coordination oxygen atoms have. Accordingly, the Obr basicity trend is (110) environments and the observed trends can be understood > (100) > (001) > (011) and this agrees with the higher according to the metal coordination. The (110) and (100) adsorption energy and the larger preference for the dissociated surface present pentacoordinated unsaturated centers with the vacant site in axial position. Since Ir-Oax distances are shorter than the equatorial ones, the Ir-H2O interaction in these (110) This journal is © The Royal Society of Chemistry 20xx J. Name., 2013, 00, 1-3 | 3 Please do not adjust margins Please do not adjust margins ARTICLE Nanoscale form at the (110) surface when compared to the (100) one, the two surfaces with axial vacant sites. Water – (IrO2)33 nanoparticle interaction. Results for the single H2O-(IrO2)33 nanoparticle interaction are described site per site starting from the most saturated centres. When possible, relation with similar undercoordinated iridium centres of the main crystallographic surfaces is made. Table 2 presents the adsorption energies and the iridium d and oxygen p DOS band centres and Figure 2 shows the most relevant optimized geometries (other structures are in the ESI). A5ax, B5ax and C5eq are pentacoordinated iridium centres. A5ax corresponds to an undercoordinated metal on the (110) surface of the nanoparticle with an axial vacant site. The adsorption of a water molecule in this A5ax site implies water dissociation, with a formal Ir-OH- fragment and a protonated oxygen bridge of the edge. All attempts to find the mol form evolved spontaneously to dissociated water as in the case of the (110) surface. The Ir-O distance (1.990 Å) is slightly shorter than that computed for the (110) surface (2.018 Å). Moreover, the transferred proton and the Ir-OH fragment interact through a strong hydrogen bond with a short O···H distance (1.774 Å). These geometrical features agree with a slightly higher adsorption energy at the nanoparticle than on the (110) surface (-229.8 vs. -211.5 kJ mol-1 for A5ax and (110), respectively). Therefore, site A5ax behaves similarly to the (110) surface with a small increase on the adsorption energy. The B5ax site is located at the edge defined by two {110} surfaces and presents an axial vacant site. The water adsorbed at B5ax Figure 2. Optimized structure for the molecular (a) and dissociated (b) water adsorption dissociates, similarly to what has been described for A5ax, and at the different (IrO2)33 nanoparticle sites. (c) Most stable structure for the adsorption of this leads to Ir-OH- and a protonated oxygen bridge at the (110) two water molecules in tetracoordinated iridium surface centers. See Figure 1 for labelling. Distances are in Å surface. The Ir-O distance (1.984 Å) and the adsorption energy (-225.5 kJ mol-1) are similar to the value found for A5ax. That is, dissociated forms, as on the (011) facet. However, at variance A5ax and B5ax share the same vacant site and flat morphology to the (011) surface, the dissociated form is preferred over the around the site, thus water adsorption is similar. molecular one and this is indicative of a stronger Ir···OW Table 2. H2O adsorption energies on (IrO2)33 sites in kJ mol-1 and Iridium d orbital (eIr) and interaction. In any case, the adsorption energies (-133.6 and - oxygen p orbital (eO) band centers of the atoms involved in the adsorption and potential 148.7 kJ mol-1 for the mol and dis forms, respectively) are lower dissociation. than those computed for the (011) extended surface, which can Site Emol Edis eIr eO be explained by the convex environment around C5eq that A5ax --a -232.6 -2.6 -2.2 prevents the formation of hydrogen bonds with nanoparticle B5ax --a -229.2 -2.8 -2.1 surface oxygen bridge atoms. Overall, while the preference for C5eq -138.2 -153.0 -2.6 -2.4 the dissociated form suggests a stronger Ir-H2O interaction that G4ax/eq -172.3 -189.2 -2.1 -2.7/3.3b increases water acidity, the absence of hydrogen bonding, I4eq/eq -181.0 -187.1 -2.2 -3.2b globally leads to weaker adsorption energies. K3 --a -192.9 G4ax/eq and I4eq/eq sites are tetracoordinated. G4ax/eq is placed at Emol/mol Edis/mol Emol/dis Edis/dis a corner defined by two {110} and one {011} facets, it has one G4ax/eq -160.2 -167.8 -159.0 -165.7 -2.1 -2.7/3.3b axial and one equatorial vacant sites and thus, has no analogy I4eq/eq -163.6 -162.6 -162.2 N.A. -1.8 -3.2b with any of the surface models. I4eq/eq is located at the tip of the K3 --a --a --a -163.7 nanoparticle, has two equatorial vacant sites and resembles the a All attempts to find the mol form evolved to the diss one; b Values undercoordinated centres of the (001) extended surfaces. corresponding to the dangling oxygen, which is well-placed for stablishing H- Adsorption of the first water molecule to these bonding interactions with the adsorbed H2O molecule. tetracoordinated sites is quite strong as found for the (001) C5eq is located at a corner formed by two {011} and one {110} extended surface. Therefore, although the two potential facets. The iridium centre at this site has an equatorial vacant conformations exist (mol and dis), the dissociated form is site similarly to those of the (011) surface. The adsorption of a preferred in both cases. Since G4ax/eq has one axial vacancy, the water molecule at C5eq can take place both in the molecular and 4 | J. Name., 2012, 00, 1-3 This journal is © The Royal Society of Chemistry 20xx Please do not adjust margins Please do not adjust margins Nanoscale ARTICLE adsorption of the first water molecule is stronger and the preference for the dissociation form higher in G4ax/eq than in I4eq/eq. Similar to C5eq, the convex nanoparticle morphology around these sites makes the hydrogen bond less efficient and this leads to final adsorption energies that are smaller than the adsorption of one water molecule on the (001) surface as well as the face and edge sites of the nanoparticle, although the geometrical parameters suggest that the intrinsic Ir-H2O interaction at these sites may be marginally higher than on a surface model. The adsorption of a second water molecule in these two sites is less favourable than the adsorption of the first water molecule and this makes that the adsorption energy per water molecule decreases significantly. The weaker Ir-H2O interaction when two water molecules are present favours the mol forms. In the case Figure 3. Projected Density of States (PDOS) and the associated band centers of the of I4eq/eq, the most favourable structure is mol/mol as already iridium d- and oxygen p-orbitals of the atoms involved in the H2O adsorption at the found on the (001). In contrast, the preferred structure for the different (IrO2)33 nanoparticle sites. adsorption of two water molecules at G4ax/eq has a molecular insights on the preference for the dissociated or molecular water at the equatorial site and a dissociated water at the axial forms as well as on the intrinsic Ir-H2O interaction (Figure 3). site (dis/mol). This agrees with the global trend that indicates a Comparison of the band centres for the extended surfaces clear preference for the dissociated form at axial sites, while the (Table 1) and those for the different sites of the nanoparticle two structures co-exist in the equatorial ones. Remarkably, due (Table 2) reveals that both the band centre of the iridium d to the intrinsic stronger Ir···H2O interaction at the 1.2 nm orbitals and the band centre of the oxygen p orbitals are closer nanoparticle model when compared with analogous extended to the Fermi level and consequently the intrinsic Ir-H2O surfaces, the energy difference between the molecular and interaction is stronger, and the dissociation feasibility dissociated forms of water adsorbed in equatorial sites is small increased. Although this analysis could be influenced by the fact both at G4ax/eq and I4eq/eq and, indeed, it is significantly smaller that we are performing constrained optimizations, IrO2 at the than the energy difference computed at the (011) surface. 1.2 nm nanoparticle model appears to be more ionic than on Similarly to I4eq/eq, K3 has two equatorial vacant sites and thus extended surfaces, thus increasing metal acidity and oxygen one would expect some analogies on the water adsorption at basicity. Particularly, the oxygen atoms that can be involved in the two sites. Nevertheless, K3 is linked to the nanoparticle the water dissociation at site I4eq/eq are the least basic (dangling through only three Ir-O bonds and the sixth position is occupied oxygen), and thus, I4eq/eq is the only site where the most stable by a singly coordinated oxygen, that shows a short Ir-O distance. structure has only molecular water. This different coordination environment and the morphology of Overall, the analysis of the adsorption of one water molecule on the nanoparticle around K3 site influence significantly the water the 10 sites analyzed so far (four surfaces and six nanoparticle adsorption. In particular, the adsorption of one water molecule sites) suggests that the metal coordination and nature of the at K3 takes place through water dissociation and all our trials to vacant site has an important role in determining the preferred obtain a structure with molecular water evolved to the same structure of water upon adsorption (dis form at the axial ones dissociated form. This suggest that the preference for the and competition between mol and dis forms in the equatorial dissociated form is higher than in the other tetracoordinated sites). Size reduction increases the ionicity of IrO2, which favors sites. Moreover, the transferred proton binds to an oxygen both the intrinsic Ir-H2O interaction and water dissociation. The bridge that is more basic (see eO in Table 2) than the dangling final adsorption energy is also controlled by the hydrogen involved in the H+ transfer at I4eq/eq allowing the formation of a bonding between the adsorbed water and the material surface, hydrogen bond which strengthens the interaction between the and this is strongly influenced by the location of the adsorbed species and the surface. The higher intrinsic Ir-H2O undercoordinated metal center. The convex morphology interaction at K3 when compared to I4eq/eq and the additional around iridium centers located at corner and tip sites of the stabilization through hydrogen bonding at the former site only, nanoparticle makes hydrogen bonding less efficient, thus lead to a higher adsorption energy that is close to that obtained leading to weaker interactions in these sites. at A5ax and B5ax sites. Moreover, although the adsorption of the second water molecule is weaker as in the case of the other Influence of the nanoparticle size in H2O adsorption: H2O-(IrO2)115. tetracoordinated sites, the preferred form for the adsorption of two water molecules at K3 is the dis/dis. All these data indicates With the aim of analyzing how the nanoparticle size tunes IrO2 that the presence of the dangling oxygen atom favours water ionicity and the feasibility of hydrogen bonding, we considered adsorption with respect to similar sites. the adsorption of one water molecule per vacant site in the 11 Analysis of the projected density of states of the d orbitals of different positions of the (IrO2)115 nanoparticle model (Figure 1). iridium centres defining each site and the p orbitals of the The eleven sites are divided in the same five group as before oxygen bridge able to form hydrogen bonds gives additional This journal is © The Royal Society of Chemistry 20xx J. Name., 2013, 00, 1-3 | 5 Please do not adjust margins Please do not adjust margins ARTICLE Nanoscale (Figure 1). For each site, we considered those structures that were found as minima on the (IrO2)33 nanoparticle. Table 3 reports the H2O adsorption energies and the iridium d- and oxygen p-DOS band centers and Figure 4 shows the optimized structures of the preferred forms in each site. All other structures can be found in the Electronic Supplementary Information. Table 3. H2O adsorption energies on (IrO2)115 sites in kJ mol-1 and Iridium d orbital (eIr) and oxygen p orbital (eO) band centers of the atoms involved in the adsorption and potential dissociation. Site Emol Edis eIr eO A5ax --a -231.3 -2.2 -1.8 B5ax --a -218.6 -2.2 -1.9 C5eq -111.9 -64.0 -2.0 -3.4 D5eq -134.3 -84.9 -2.4 -3.3 E5eq -140.6 -168.0 -2.0 -2.6 F5eq -108.5 -183.3 -2.3 -2.5 Emol/mol Edis/mol Emol/dis Edis/dis G4ax/eq -151.6 -155.7 --b --b -2.1 -2.5/--e H4ax/eq --c -190.6 --b -212.2 -2.1 -2.5/-1.8 I4eq/eq -113.8 --b --b --b -2.3 --e/--e J4eq/eq -- d -182.6 -- b --d -2.3 -2.7 K3 --b --b --b --b -2.4 -2.4 a Attempts to find the mol form evolved to the dis one; b Not computed structure; c Attempts to find the mol/mol form evolved to the dis/mol one; d Attempts to find this structure evolved to the dis/mol form; e There are not unsaturated oxygen atoms close to this vacant site. The adsorption energies on the different sites of the (IrO2)115 nanoparticle model range from -231.3 to -111.9 kJ mol-1. These Figure 4. Optimized structure for the most favorable water adsorption form at the values are similar to those computed for the smaller model and different (IrO2)115 nanoparticle sites. See Figure 1 for labelling. Distances are in Å the same general trends observed for the (IrO2)33 nanoparticle are also found for (IrO2)115. H2O adsorbed on metal axial vacant dissociation and leads to a high mean adsorption energy (-212.0 sites dissociates leading to an OH-/H+ structure in which a kJ mol-1). In contrast, no accessible oxygen bridge species are nearby oxygen bridge is protonated and this is regardless site close to G4ax/eq, thus water dissociation does not take place and position and coordination (penta- or tetracoordinated). In the adsorption energy per water molecule is significantly lower contrast, the two structures are usually found for the equatorial (-155.7 kJ mol-1). vacant sites and the preferred form is highly related with the The same reasoning can be applied to the sites presenting only feasibility of forming a strong hydrogen bond that favors equatorial vacancies. Four different pentacoordinated sites dissociation. That is, since hydrogen bonding in the dis form with an equatorial vacant site can be distinguished in (IrO2)115: tends to be stronger than in the mol structure, the presence of C5eq, D5eq, E5eq and F5eq. The adsorption energy in these sites a nearby oxygen bridge favors water dissociation. Moreover, ranges between -183.3 and -111.9 kJ mol-1, the highest value the nanoparticle morphology around the site determines the being obtained at F5eq which is located on the (011) surface and formation/absence of hydrogen bonding and thus the final it presents a well-placed oxygen bridge for stablishing a strong adsorption energy. hydrogen bond that reinforces the H2O-nanoparticle interaction In this context, H2O adsorption at A5ax and B5ax occurs through and favor water dissociation. On the other hand, C5eq, D5eq and the dissociated form, presents strong hydrogen bonding and E5eq are located at edges or corners between different surfaces high adsorption energies. Indeed, the final computed values are and the morphology around these sites is more convex. in between those of the equivalent sites in the (IrO2)33 Consequently, no strong hydrogen bonds can be stablished, nanoparticle model and the (110) slab, which presents the same which leads to smaller adsorption energies and tend to favor type of iridium coordination. Similarly, the adsorption at H4ax/eq the mol form. Similarly, the two tetracoordinated sites with and G4ax/eq always implies a dissociated water molecule in the equatorial vacancies (I4eq/eq and J4eq/eq) present different axial position (as in the (IrO2)33 nanoparticle model). However, preferred forms that can be related with the accessibility to the preferred form for the equatorial position is dis in H4ax/eq oxygen bridge sites. I4eq/eq is located at the tip of the and mol in G4ax/eq. H4ax/eq presents a well-suited oxygen bridge nanoparticle and the convex form around this site prevents the able to interact through hydrogen bonding with the water formation of any hydrogen bond. Consequently, the adsorption molecule in equatorial position and this promotes water energy is low (-113.8 kJ mol-1) and the mol/mol form preferred. 6 | J. Name., 2012, 00, 1-3 This journal is © The Royal Society of Chemistry 20xx Please do not adjust margins Please do not adjust margins Nanoscale ARTICLE J4eq/eq is located at the edge between two {011} surfaces and the protonation of 5 of the 6 dangling oxygens. On the other hand, morphology around the site allows stablishing one hydrogen the optimization starting from the maximum number of bond per water molecule. This results in a higher adsorption dissociated water molecules also implies proton transfer, the energy (-182.6 kJ mol-1) and the preference for the mol/dis final structure presenting 27 dis/11 mol water molecules and all form. Indeed, geometry optimizations starting from a mol/mol dangling oxygens protonated. In both cases, an important and a dis/dis structure, spontaneously evolved to the mol/dis network of hydrogen bonding between the adsorbed species is one. stablished except for some species at the equatorial position of In summary results for the larger nanoparticle indicate that the C5eq and G4ax/eq (Figure S8). Moreover, there is an elongation of main conclusions already drawn with the small nanoparticle the Ir-OW average distance (2.09 Å and 2.05 Å for the 19 dis/19 model are also valid for the larger one. Indeed, the vacant site mol and 27 dis/11 mol structures respectively) when compared nature determines in a large extend the preferred adsorption with the isolated adsorption of one water molecule (2.02 Å). form at low coverages: dis for axial vacant sites and a Therefore, cooperative effects between water molecules competition between the two forms in the equatorial one. decrease the H2O-IrO2 nanoparticle interaction. Remarkably, Indeed, when no strong hydrogen bonds are stablished the adsorption energy per water molecule for the two final between the adsorbed water and the oxygen bridge of the conformations (-172.1 and -176.0 kJ mol-1 for 19 dis/19 mol and nanoparticle, the preferred structure in equatorial sites is the 27 dis/11 mol structures respectively) is very similar to the mol form. In contrast, the presence of strong hydrogen bonds average value considering the individual adsorption in each site, favors the dis form that becomes more favorable. Moreover, -178.2 kJ mol-1. That is, the cooperative effects compensate the the adsorption energy is highly determined by the feasibility to loss of local H2O-IrO2 interaction but does not imply a significant stablish hydrogen bonding between the adsorbed species and increase of the adsorption energy. This was already found for the nanoparticle. Therefore, within centers with the same type the (110) surface and it is in contrast to the results for the (011) of vacant site the adsorption energy varies as face > edge > IrO2 surface.35 corners = tips. This is somewhat counter-intuitive, since the Since the difference in adsorption energy per water molecule in most accessible sites present lower adsorption energies, but the two limit cases is small, they are both expected to outlines the importance of hydrogen bonding in the interaction contribute to the water-monolayer structure. Indeed, other between water and IrO2. Finally, the two nanoparticle models relevant conformations may be close in energy and contribute seem to show a larger ionicity than the extended surfaces, thus to the final interface structure. Moreover, thermal effects are leading to stronger intrinsic Ir-H2O interactions. However, the expected to play a role as evidenced by AIMD simulations on IrO2 ionicity appears to slightly decrease when enlarging the the (110) and (011) surfaces of RuO2, IrO2 and TiO2, which show nanoparticle model, leading to a small decrease of the that cooperative and temperature effects tend to disfavor adsorption energy at equivalent sites of the two nanoparticle water dissociation, particularly on IrO2.35 It is for this reason that models. 10ps AIMD simulations starting from the two limit situations were carried out. Analysis of the degree of dissociation along Cooperative and thermal effects: AIMD simulations of the (IrO2)33 time (Figure S9) shows that equilibration last up to 5ps, thus the monolayer. analysis is made on the 5 to 10ps steps. Figure 5 shows the Interaction between different adsorbed water molecules is frequency of different number of water molecules on the expected to influence the IrO2 nanoparticle - H2O interaction as nanoparticle (including those formed on dangling oxygens) and already described on the IrO2 (110) and (011) surfaces.35 It is for how they are divided in axial and equatorial sites and reports this reason that we explored the formation of a water the mean degree of dissociation for the whole nanoparticle and monolayer around the (IrO2)33 nanoparticle model. This implies as function of each type of sites. 38 water molecules that occupies all iridium vacant sites. Since During the equilibration period, the AIMD simulation starting many different structures are plausible, we decided to optimize from the optimized structure with a larger number of molecular two limit situations: a) all adsorbed molecules are in their mol waters evolve to a structure with a higher water dissociation form and b) a structure with the maximum number of degree. In contrast, the opposite process is observed during the dissociated water molecules (28 dis/10 mol), which is limited by equilibration period of the AIMD starting from the optimized the accessible oxygen dangling and oxygen bridge sites. In a structure with many dissociated water molecules. The average second step, we performed a 10 ps AIMD simulation starting number of H2O species during the production period are 16.8 from the two optimized structures. and 11.4 for the simulation starting with more mol or dis water The two (IrO2)33-monolayer optimized structures show the 38 molecules, respectively. Detailed analysis (Figure S10) shows water molecules interacting with iridium thus, the water that the differences can be attributed in a large extend to C5eq coverage increase does not produce any water desorption and I4eq/4eq sites, where the hydrogen bonding network (Figure S8). However, the resulting structure differs from the between adsorbed water molecules is weaker and the energy initial one in the two cases. The geometry optimization starting difference between mol and dis forms small. The degree of from a monolayer composed of adsorbed molecular water water dissociation (63%; average value considering the two evolves spontaneously to a situation where half of the water simulations) is higher for those water molecules adsorbed at molecules dissociate (19 dis/19 mol) and this is linked with the axial sites (around 72 %) than those adsorbed at equatorial sites This journal is © The Royal Society of Chemistry 20xx J. Name., 2013, 00, 1-3 | 7 Please do not adjust margins Please do not adjust margins ARTICLE Nanoscale stablished with a nearby oxygen bridge atom. However, nanoparticle morphology around corner and tip sites prevents the formation of hydrogen bonds and thus, the mol form becomes predominant in the more convex regions of the nanoparticle. The hydrogen bonding also influences in a large extent the final adsorption energy. Consequently, the highest adsorption energies are obtained at face and edge sites with a flat morphology, and the smallest values are related to corners and tips. That is, the usually less stabilized and more accessible sites present weaker adsorption energies due to the absence of hydrogen bonding. Increasing water coverage from an isolated water to a water monolayer reveals that cooperative effects weaken the IrO2- H2O interaction, which is essentially compensated by the hydrogen bonding between adsorbed species. Furthermore, the global water dissociation degree is highly determined by the accessible unsaturated oxygen atoms on the nanoparticle surface that are not sufficient to account for all the potential dissociations. Overall, the water dissociation degree is between 60 and 70%, which is lower than that obtained for crystalline surfaces. This is in line with the observation that nanoparticles Figure 5. Frequency of the different number of water molecules and the percentage of with large amounts of surface Ir-OH species are highly amorph water dissociation on the nanoparticle surface. The total number of water molecules is and this is related to their higher activity in OER.16 divided as axial and equatorial. Bars and values in blue correspond to the AIMD simulation with a larger fraction of molecular waters, while the bars and values in red correspond to the simulation starting from a large degree of dissociation. Authors contribution (58 %) and these values are similar albeit marginally lower than DG performed all simulations of the present contribution. MS, those arising from the study of individual adsorptions (68 %). LRS and XSM conceived the project, supervised the research In summary, the presence of stabilizing cooperative effects and wrote the paper. between adsorbed water molecules and thermal contributions weakens the IrO2-H2O direct interaction. This produces a decrease of the water dissociation degree (around 63%) that is Conflicts of interest tuned by two additional factors: i) cooperative effects and ii) the There are no conflicts to declare number of accessible oxygen bridge and dangling atoms is smaller than the number of adsorbed water molecules. The latter seems to be dominant and, thus overall, the degree of Acknowledgements dissociation is lower than those of films. The authors gratefully acknowledge financial support from MICINN [CTQ2017-89132-P and PID2020-112715GB-I00], the Conclusions Generalitat de Catalunya [2017SGR1323] and Red Española de Supercomputación [QCM-2019-1-0043, QS-2019-2-0031]. XSM Periodic PBE-D2 calculations and AIMD simulations have been is grateful for the Professor Agregat Serra Húnter position. carried out to study the interface between H2O and Wulff-like stoichiometric IrO2 nanoparticles. For that, different coverages ranging from isolated water molecules to the monolayer as well References as two nanoparticle models ((IrO2)33 and (IrO2)115) constructed 1 Z. W. Seh, J. Kibsgaard, C. F. Dickens, I. Chorkendorff, J. K. with BCN-M have been considered.43 Results show that the Nørskov and T. F. Jaramillo, Science, 2017, 355, eaad4998. adsorption energies range from -232.6 and -111.9 kJ mol-1 and 2 C. C. L. McCrory, S. Jung, I. M. Ferrer, S. M. Chatman, J. C. these values are similar (upper limit) or lower than those Peters and T. F. Jaramillo, J. Am. Chem. Soc., 2015, 137, obtained in extended surfaces. The tetragonal distortion 4347–4357. present on IrO2 rutile bulk is retained on the nanoparticle 3 T. Reier, H. N. Nong, D. Teschner, R. Schlögl and P. Strasser, models and this controls the adsorption strength and the Adv. Energy Mater., 2017, 7, 1601275. preferred structure (molecular (mol) or dissociated (dis)) at 4 J. H. Montoya, L. C. Seitz, P. Chakthranont, A. Vojvodic, T. F. each site. H2O adsorption on axial vacant sites is strong, favoring Jaramillo and J. K. Nørskov, Nat. Mater., 2017, 16, 70–81. water dissociation, which is largely preferred. The H2O-Ir 5 O. Piqué, F. Illas and F. Calle-Vallejo, Phys. Chem. Chem. interaction at equatorial vacant sites is weaker and the mol and Phys., 2020, 22, 6797–6803. dis forms become close in energy. In these equatorial sites, the 6 J. Rossmeisl, Z.-W. Qu, H. Zhu, G.-J. Kroes and J. K. Nørskov, dis form is preferred when strong hydrogen bonding is 8 | J. Name., 2012, 00, 1-3 This journal is © The Royal Society of Chemistry 20xx Please do not adjust margins Please do not adjust margins Nanoscale ARTICLE J. Electroanal. Chem., 2007, 607, 83–89. Adv. Funct. Mater., 2020, 30, 1906670. 7 H. N. Nong, L. J. Falling, A. Bergmann, M. Klingenhof, H. P. 27 M. Kan, X. Qian, T. Zhang, D. Yue and Y. Zhao, ACS Sustain. Tran, C. Spöri, R. Mom, J. Timoshenko, G. Zichittella, A. Chem. Eng., 2017, 5, 10940–10946. Knop-Gericke, S. Piccinin, J. Pérez-Ramírez, B. R. Cuenya, R. 28 F. A. Frame, T. K. Townsend, R. L. Chamousis, E. M. Sabio, Schlögl, P. Strasser, D. Teschner and T. E. Jones, Nature, T. Dittrich, N. D. Browning and F. E. Osterloh, J. Am. Chem. 2020, 587, 408–413. Soc., 2011, 133, 7264–7267. 8 Y. Bian, M. Kim, T. Li, A. Asthagiri and J. F. Weaver, J. Am. 29 J. C. Cruz, V. Baglio, S. Siracusano, R. Ornelas, L. Ortiz- Chem. Soc., 2018, 140, 2665–2672. Frade, L. G. Arriaga, V. Antonucci and A. S. Aricò, J. 9 Y. Tsuji and K. Yoshizawa, J. Phys. Chem. C, 2018, 122, Nanoparticle Res., 2011, 13, 1639–1646. 15359–15381. 30 R. Zhang, P. E. Pearce, Y. Duan, N. Dubouis, T. Marchandier 10 Z. Liang, T. Li, M. Kim, A. Asthagiri and J. F. Weaver, and A. Grimaud, Chem. Mater., 2019, 31, 8248–8259. Science, 2017, 356, 299–303. 31 R. Jaimes, J. Vazquez-Arenas, I. González and M. Galván, 11 H.-S. Oh, H. N. Nong, T. Reier, A. Bergmann, M. Gliech, J. Electrochim. Acta, 2017, 229, 345–351. Ferreira de Araújo, E. Willinger, R. Schlögl, D. Teschner and 32 D.-Y. Kuo, J. K. Kawasaki, J. N. Nelson, J. Kloppenburg, G. P. Strasser, J. Am. Chem. Soc., 2016, 138, 12552–12563. Hautier, K. M. Shen, D. G. Schlom and J. Suntivich, J. Am. 12 T. Audichon, T. W. Napporn, C. Canaff, C. Morais, C. Chem. Soc., 2017, 139, 3473–3479. Comminges and K. B. Kokoh, J. Phys. Chem. C, 2016, 120, 33 S. Siahrostami and A. Vojvodic, J. Phys. Chem. C, 2015, 119, 2562–2573. 1032–1037. 13 L. An, C. Wei, M. Lu, H. Liu, Y. Chen, G. G. Scherer, A. C. 34 J. Heras-Domingo, M. Sodupe and X. Solans-Monfort, J. Fisher, P. Xi, Z. J. Xu and C. Yan, Adv. Mater., 2021, 33, Phys. Chem. C, 2019, 123, 7786–7798. 2006328. 35 D. González, J. Heras-Domingo, S. Pantaleone, A. Rimola, L. 14 H. N. Nong, T. Reier, H.-S. Oh, M. Gliech, P. Paciok, T. H. T. Rodríguez-Santiago, X. Solans-Monfort and M. Sodupe, ACS Vu, D. Teschner, M. Heggen, V. Petkov, R. Schlögl, T. Jones Omega, 2019, 4, 2989–2999. and P. Strasser, Nat. Catal., 2018, 1, 841–851. 36 R. R. Rao, M. J. Kolb, J. Hwang, A. F. Pedersen, A. Mehta, H. 15 D. Lebedev, R. Ezhov, J. Heras-Domingo, A. Comas-Vives, N. You, K. A. Stoerzinger, Z. Feng, H. Zhou, H. Bluhm, L. Kaeffer, M. Willinger, X. Solans-Monfort, X. Huang, Y. Giordano, I. E. L. Stephens and Y. Shao-Horn, J. Phys. Chem. Pushkar and C. Copéret, ACS Cent. Sci., 2020, 6, 1189– C, 2018, 122, 17802–17811. 1198. 37 L. Mino, Á. Morales-García, S. T. Bromley and F. Illas, 16 D. F. Abbott, D. Lebedev, K. Waltar, M. Povia, M. Nanoscale, 2021, 13, 6577–6585. Nachtegaal, E. Fabbri, C. Copéret and T. J. Schmidt, Chem. 38 G. Fazio, D. Selli, L. Ferraro, G. Seifert and C. Di Valentin, Mater., 2016, 28, 6591–6604. ACS Appl. Mater. Interfaces, 2018, 10, 29943–29953. 17 A. W. Jensen, G. W. Sievers, K. D. Jensen, J. Quinson, J. A. 39 D. R. Hummer, J. D. Kubicki, P. R. C. Kent and P. J. Heaney, Arminio-Ravelo, V. Brüser, M. Arenz and M. Escudero- J. Phys. Chem. C, 2013, 117, 26084–26090. Escribano, J. Mater. Chem. A, 2020, 8, 1066–1071. 40 A. Cuko, A. Macià Escatllar, M. Calatayud and S. T. Bromley, 18 E. Oakton, D. Lebedev, M. Povia, D. F. Abbott, E. Fabbri, A. Nanoscale, 2018, 10, 21518–21532. Fedorov, M. Nachtegaal, C. Copéret and T. J. Schmidt, ACS 41 F. A. Soria and C. Di Valentin, Nanoscale, 2021, 13, 4151– Catal., 2017, 7, 2346–2352. 4166. 19 P. Lettenmeier, L. Wang, U. Golla-Schindler, P. Gazdzicki, N. 42 D. González, J. Heras-Domingo, M. Sodupe, L. Rodríguez- A. Cañas, M. Handl, R. Hiesgen, S. S. Hosseiny, A. S. Gago Santiago and X. Solans-Monfort, J. Catal., 2021, 396, 192– and K. A. Friedrich, Angew. Chemie Int. Ed., 2016, 55, 742– 201. 746. 43 D. González, B. Camino, J. Heras-Domingo, A. Rimola, L. 20 F. Bizzotto, J. Quinson, A. Zana, J. J. K. Kirkensgaard, A. Rodríguez-Santiago, X. Solans-Monfort and M. Sodupe, J. Dworzak, M. Oezaslan and M. Arenz, Catal. Sci. Technol., Phys. Chem. C, 2020, 124, 1227–1237. 2019, 9, 6345–6356. 44 G. Kresse and J. Furthmüller, Phys. Rev. B, 1996, 54, 21 Y. Lee, J. Suntivich, K. J. May, E. E. Perry and Y. Shao-Horn, 11169–11186. J. Phys. Chem. Lett., 2012, 3, 399–404. 45 G. Kresse and J. Hafner, Phys. Rev. B, 1993, 47, 558–561. 22 Y. Zhao, E. A. Hernandez-Pagan, N. M. Vargas-Barbosa, J. L. 46 J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., Dysart and T. E. Mallouk, J. Phys. Chem. Lett., 2011, 2, 402– 1996, 77, 3865–3868. 406. 47 S. Grimme, J. Comput. Chem., 2004, 25, 1463–1473. 23 D. Lebedev and C. Copéret, ACS Appl. Energy Mater., 2019, 48 G. Kresse and D. Joubert, Phys. Rev. B, 1999, 59, 1758– 2, 196–200. 1775. 24 H. Jiang, S. Ge, Y. Zhang, M. Dong, S. Wu, M. Wu, J. Zhang, 49 P. E. Blöchl, Phys. Rev. B, 1994, 50, 17953–17979. R. Ge and Z. Guo, J. Phys. Chem. C, 2020, 124, 2–8. 50 A. Minguzzi, O. Lugaresi, E. Achilli, C. Locatelli, A. Vertova, 25 L. Fu, X. Zeng, C. Huang, P. Cai, G. Cheng and W. Luo, Inorg. P. Ghigna and S. Rondinini, Chem. Sci., 2014, 5, 3591. Chem. Front., 2018, 5, 1121–1125. 51 C. F. Dickens, J. H. Montoya, A. R. Kulkarni, M. Bajdich and 26 D. Böhm, M. Beetz, M. Schuster, K. Peters, A. G. Hufnagel, J. K. Nørskov, Surf. Sci., 2019, 681, 122–129. M. Döblinger, B. Böller, T. Bein and D. Fattakhova-Rohlfing, 52 B. Hammer and J. K. Nørskov, Surf. Sci., 1995, 343, 211– This journal is © The Royal Society of Chemistry 20xx J. Name., 2013, 00, 1-3 | 9 Please do not adjust margins Please do not adjust margins ARTICLE Nanoscale 220. 53 Z. Xu and J. R. Kitchin, Catal. Commun., 2014, 52, 60–64. 10 | J. Name., 2012, 00, 1-3 This journal is © The Royal Society of Chemistry 20xx Please do not adjust margins