Spectroelectrochemistry of Water Oxidation Kinetics in Molecular versus Heterogeneous Oxide Iridium Electrocatalysts.

Water oxidation is the step limiting the efficiency of electrocatalytic hydrogen production from water. Spectroelectrochemical analyses are employed to make a direct comparison of water oxidation reaction kinetics between a molecular catalyst, the dimeric iridium catalyst [Ir2(pyalc)2(H2O)4-(μ-O)]2+ (IrMolecular, pyalc = 2-(2'pyridinyl)-2-propanolate) immobilized on a mesoporous indium tin oxide (ITO) substrate, with that of an heterogeneous electrocatalyst, an amorphous hydrous iridium (IrOx) film. For both systems, four analogous redox states were detected, with the formation of Ir(4+)-Ir(5+) being the potential-determining step in both cases. However, the two systems exhibit distinct water oxidation reaction kinetics, with potential-independent first-order kinetics for IrMolecular contrasting with potential-dependent kinetics for IrOx. This is attributed to water oxidation on the heterogeneous catalyst requiring co-operative effects between neighboring oxidized Ir centers. The ability of IrMolecular to drive water oxidation without such co-operative effects is explained by the specific coordination environment around its Ir centers. These distinctions between molecular and heterogeneous reaction kinetics are shown to explain the differences observed in their water oxidation electrocatalytic performance under different potential conditions.

Water oxidation catalysis is a key challenge for hydrogen synthesis from water via electrolysis, with extensive interest in both heterogeneous oxides and molecular catalysts.[1−3] While heterogeneous oxides are already employed in commercial electrolyzers, there have been striking recent advances in the performance and stability of molecular water oxidation catalysts based on iridium.[4−7] Molecular catalysts are particularly attractive as model systems for this reaction, with their well-defined and tunable atomic structures aiding mechanistic insight.[7−10] A key potential mechanistic difference is that molecular catalysts typically consist of isolated metal centers, while heterogeneous oxide films are often reported to promote water oxidation by co-operative effects between adjacent sites.[11−13] Recent work has controversially suggested that long-range interactions, which depend on the coverage of oxidized *O species, can be attributed to solvation effects on IrO surfaces.[11] However, direct comparisons between the water oxidation kinetics of molecular and heterogeneous catalysts operating under similar reaction conditions have been very limited to date.[14,15] Herein we use operando transient spectroelectrochemistry to compare two highly active water-oxidation electrocatalysts based on iridium: a dimeric molecular catalyst [Ir2(pyalc)2(H2O)4-(μ-O)]2+ (Ir, Figure A) (pyalc = 2-(2′pyridinyl)-2-propanolate) and the amorphous hydrous iridium oxide (IrOFigure B).[2,3,7,16−20]

Figure 1

(A) Proposed molecular structure of Ir and (B) schematic of a cluster in electrodeposited IrO, proposed by Pavlovic et al.[13,20,21] (C) Tafel plot from the steady-state current per mol of electrochemically active iridium every 50 mV of Ir and IrO (electrodeposited for 1000s) in aqueous HClO4 0.1 M at pH 1.2. The calculation of electrochemically active iridium is described in the Supporting Information.

It has previously been reported that molecular iridium catalysts can be immobilized on oxide surfaces without the need for additional anchoring groups. When immobilized on high-surface area mesoporous indium tin oxide (mesoITO), Ir can reach turnover frequencies around 7.9 s–1 at an overpotential of 520 mV, and its ligands remain intact for at least 11 h at pH 2.6 under 1.55 VRHE.[16]Ir cocatalysts have also been reported to enhance the performance of hematite photoanodes, shifting the oxygen evolution reaction (OER) onset potential by ∼250 mV.[14,15,21] In our previous work on IrO electrodes, spectroelectrochemical techniques were applied to identify three redox transitions, attributed following the literature to the oxidation of Ir3+ to Ir3.x+, Ir3.x+ to Ir4+, and Ir4+ to Ir4.y+, where Ir3.x+ and Ir4.y+ represent mixed valence states including Ir3+ and Ir4+, and oxidized Ir4+ respectively.[22−39] The final redox transition results in the formation of oxygenated species, *O, whose OER kinetics can also be probed using time-resolved spectroelectrochemistry.[22] Herein, we extend this spectroelectrochemical analysis to the dimeric complex Ir immobilized on mesoITO. By comparing the molecular redox states and their kinetics to those of electrodeposited IrO films,[22] we aim to understand the impact on OER activity of the specific coordination environment of iridium centers in Ir as well as the potential impact of co-operative effects between Ir centers present in the heterogeneous IrO film.

The molecular iridium catalyst Ir was immobilized on mesoporous ITO (mesoITO) following a previously reported procedure, forming a packed monolayer of ∼0.04 μmolIr/cm2geometric, while IrO films were electrodeposited in water from an iridium salt.[14−16] Iridium centers in mesoITO-Ir are expected to be at least 3 times further away from the iridium centers in adjacent molecules than between iridium centers in one Ir catalyst and in IrO because of the bulkiness of the ligands.[14,16] The Tafel plot of the resulting Ir films (Figure C) shows an exponential increase in the[26,27] current at a potential of ∼1.32 VRHE, ∼80 mV lower than in IrO. Comparing the catalytic waves for water oxidation in Figure C, the Ir films exhibit higher performance below ∼1.45 VRHE with a Tafel slope of 174 mV/dec. Above 1.45 VRHE, IrO films exhibit a sharper onset and thus higher performance than Ir, with Tafel slopes of 59 and 69 mV/dec, respectively, in accordance with previous studies.[11,16]

The redox chemistry of Ir on mesoITO was analyzed further spectroelectrochemically, following procedures we have previously reported for IrO (Figures S2–S3).[22] Its UV–vis absorption spectrum was measured as a function of applied potential with 5 mV increments in aqueous 0.1 M HClO4 electrolyte at pH 1.2 (Figure A). For the potential range 0.78–1.08 VRHE, an absorption band at 590 nm dominated the spectrum. A new feature appeared at 800 nm between 1.13 and 1.43 VRHE, and above 1.43 VRHE, absorption changes were primarily detected below 500 nm. These features were deconvolved (Figures B and S3–S7) by fitting a model which assumes three additive contributions to the absorption, each linearly proportional to the concentration of redox states, and where the concentration of redox states formed at each potential follows a Gaussian distribution (SI, Equations S2–7), as reported in our previous work.[22] The deconvoluted spectroelectrochemistry results are shown in Figure B, where the normalized concentration distributions and differential absorption corresponding to the three redox transitions detected in Ir are compared to those in IrO. The optical signals related to the three redox transitions in the two catalysts have broadly similar features (Figure B left). However, the absorption bands are slightly blue-shifted in Ir (shown in more detail in Figure S8). The concentration distributions in Ir are shifted anodically by ∼100 mV with respect to IrO for all the three redox transitions (Figure B right).

Figure 2

(A) Absorption changes at different potentials of Ir on mesoporous ITO. (B) Normalized deconvolution results of the spectroelectrochemical data of Ir and IrO: (left) change in the concentration of the redox states formed at increasing potentials and (right) differential absorption coefficients of the corresponding redox transitions (relative to 0.73 V vs RHE). Measurements done in aqueous HClO4 0.1 M at pH 1.2 under constant applied potentials every 0.05 V.

Considering the assignments of the redox transitions in Figure for IrO, the transitions in the 0.8–1.4 VRHE range have been assigned to intervalence charge transfer within the iridium d orbitals derived from the oxidation of Ir(3+) to Ir(4+) and deprotonation of hydroxyl groups coordinated to the Ir center.[23,24,32,35,40−42] Positive of 1.4 VRHE, redox states of iridium higher than 4+ are expected to be formed.[29,33,34,43−45] Crabtree et al. have reported similar redox transitions in an iridium dimer structurally analogous to Ir: Ir(3+)–Ir(3+) absorbing below 450 nm, Ir(4+)–Ir(4+) absorbing at 600–750 nm, and Ir(4+)–Ir(5+) absorbing at 500 nm.[46,47] By taking this iridium dimer as a reference, the three redox transitions we observe in Ir can be assigned to the sequential oxidation of Ir(3+)–Ir(3+) to Ir(3+)–Ir(4+), Ir(4+)–Ir(4+), and Ir(4+)–Ir(5+), respectively, in good agreement with assignments on IrO. The absolute absorption and the calculated concentrations are smaller for mesoITO-Ir compared to electrodeposited IrO (Figures A and S7 for Ir and S1B and S2 for IrO), indicative of larger geometric densities of electrochemically active Ir centers in IrO, and attributed to its permeability to the electrolyte and electrode morphology. By comparing the electrochemical and the deconvoluted spectroelectrochemical data in Figures C and 2B, respectively, it is apparent for both Ir and IrO that the electrocatalytic current overlaps with the third redox transition detected spectroelectrochemically (discussed further below). This indicates that this transition, assigned to Ir(4+)–Ir(5+) formation in Ir, enables O2 evolution, similar to our findings on IrO. It is also apparent that this transition is shifted anodically for the Ir relative to IrO, indicative of the role of ligands in tuning the redox activity of the Ir centers.

To analyze the kinetics of the redox states in Ir, the optical signal decay was measured after turning an applied potential off, following the same procedure reported for IrO in our previous study (Scheme S1 and Supporting Information).[22] This methodology was used to deduce the OER reaction kinetics of the active redox state as a function of its concentration. Absorption changes upon applying an electric potential from the open circuit potential (OCP) were detected only at applied potentials above ∼1.32 VRHE (iR corrected) (Figure A), which corresponds to the potential at which the active redox state Ir(4+)–Ir(5+) is formed. Notably, the decay kinetics of the active state optical signal in Ir are almost invariant throughout the 1.3–1.6 VRHE range, implying a potential-independent first-order mechanism, where the rate of the reaction (J) is proportional to a potential-independent rate constant (k) and the concentration of oxidized species (θ); i.e., J = kθ. This contrasts with the behavior of IrO, where the decay of the optical signal becomes substantially faster with increasing potentials in this range. The corresponding decay time constants are plotted in Figure B, where the optical signal lifetimes at different potentials were extracted from fitting the signal decays with an initial linear regression (Equations S10–S11). It is apparent that, at potentials below 1.45 VRHE, the lifetimes of the active state in IrO are longer than in Ir, indicative of lower reactivity, while above 1.45 VRHE the lifetimes become shorter, indicative of higher reactivity.

Figure 3

(A) Absorption decay after turning the potential off, and (B) optical signal lifetimes derived from fitting an initial linear regression in Figure 3A in Ir and IrO in 0.1 M HClO4 water at pH 1.2.

Apart from the shift in oxidation potential discussed above, IrO and Ir show similar potential dependencies for the third redox transition assigned to the formation of the oxidized species driving OER (Figure B). As such, the smaller Tafel slope observed for IrO (Figure A) can be assigned to a sharper acceleration of OER reaction kinetics for this film with potential. The absence of any potential dependence of reaction kinetics for Ir suggests that this acceleration for IrO is unlikely to be due to a mechanism change or potential-dependent change in reaction activation energy.

For Ir the third redox transition (Figure B) coincides with the potential required to reach 0.1 A/mmol/cm2geometric (Figure C), consistent with its first-order OER kinetics. In contrast, for IrO the potential to reach 0.1 A/mmol/cm2geometric is shifted anodically by ∼100 mV relative to its third redox transition, providing further confirmation that this heterogeneous catalyst needs to reach a critical concentration of redox active state to trigger the OER reaction. Therefore, it appears likely that this acceleration results from co-operative effects between different oxidized Ir centers on IrO,[48] consistent with a recent report by Nong et al.[11] of co-operative interactions between multiple sites on IrO during OER. Such co-operative effects cannot be realized in the isolated Ir centers of mesoITO-Ir due to steric hindrance by the ligands.

At low overpotentials, in the absence of co-operative effects that can enhance OER activity on IrO, the reactivity for both catalysts would be governed by the chemical environment of isolated oxidized Ir centers, including the ligands and the electrode substrate, as illustrated in Scheme . In the molecular catalyst, the absorption maxima above 500 nm are slightly blue-shifted (Figure B right), and the potential for each redox transition in Ir is shifted by ∼100 mV compared to IrO (Figure B left). Considering that the Ir centers have an octahedral configuration in both Ir and IrO, a plausible cause for these shifts is different solvation environment in Ir and IrO and the stabilization of the t2 d-based iridium molecular orbitals in Ir by pyridine π orbitals (Scheme ).[49,50] On the one hand, this would lead to a larger energy gap between the t2 and e molecular orbitals, which corresponds to the transition observed in the visible range.[49,50] On the other hand, this stabilization of the t2 molecular orbitals, where the valence electrons are located, would make the iridium centers in Ir harder to oxidize than in IrO. This is in agreement with the seminal work of Rossmeisl and Nørskov et al.[51] which suggested that IrO2(110) binds oxygen too strongly relative to the optimal catalyst, resulting in the O–O bond formation step to form *OOH being rate-determining. Consequently, the faster reaction kinetics, and higher OER activity of Ir at low overpotentials (<1.45 VRHE), can thus be explained by the stabilization effect of the pyridine ligand on the valence iridium d orbitals, which increases the oxidizing potential of Ir and decreases the oxygen binding strength.

Scheme 1

Proposed Chemical Effects of the Atoms Coordinated to the Iridium Active Sites on the Redox Potential (E) and Maximum Absorption Wavelength of d–d Transitions (λmax.(d–d)), and lifetimes (τ) in water 0.1 M HClO4 at pH 1.2

In conclusion, by employing operando spectroelectrochemistry, we have analyzed the redox transitions and OER reaction kinetics for IrO and for Ir immobilized on meso-ITO between 1 and 1.6 VRHE. The third redox transition, assigned in both systems to Ir(4+) oxidation, was observed to correlate with the increase in the OER current. Ir is observed to exhibit potential independent OER reaction kinetics, indicative of water oxidation by independent molecular catalysts. This contrasts with IrO films, where the reaction kinetics are observed to accelerate strongly with applied potential, attributed to the OER on this catalyst requiring co-operative interactions between neighboring oxidized Ir centers. The ability of Ir to drive OER without requiring such co-operative effects is attributed to the anodic shift of its redox transitions relative to IrO, resulting from the specific chemical environment of its iridium centers. These differences in redox state energetics and OER kinetics explain the differences observed in OER activity and Tafel slopes between these molecular and heterogeneous catalysts. The absence of co-operative effects, as observed herein for our molecular system, could limit the performance of single atom catalysts[52−55] for OER, particularly on Ir-based centers. Our study has therefore allowed a direct comparison of molecular catalysts and heterogeneous oxide film OER kinetics, thus providing insights into how both the local environment of the catalytic site and co-operative effects between oxidized sites can result in significantly different water oxidation kinetics.