Crossing the divide between homogeneous and heterogeneous catalysis in water oxidation

Significance

An atomic layer deposition (ALD) procedure is described for stabilizing surface binding of a water oxidation catalyst to the surfaces of nanostructured films of indium tin oxide. The catalyst is stabilized on the surface of electrodes by ALD of an overlayer of TiO₂. Stabilization of surface binding allows use of basic solutions where a rate enhancement for water oxidation of ∼10⁶ is observed compared with acidic conditions. There are important implications for stabilizing surface-bound molecular assemblies for applications in dye sensitized solar cells, electrocatalysis, and photoelectrocatalysis.

Abstract

Enhancing the surface binding stability of chromophores, catalysts, and chromophore–catalyst assemblies attached to metal oxide surfaces is an important element in furthering the development of dye sensitized solar cells, photoelectrosynthesis cells, and interfacial molecular catalysis. Phosphonate-derivatized catalysts and molecular assemblies provide a basis for sustained water oxidation on these surfaces in acidic solution but are unstable toward hydrolysis and loss from surfaces as the pH is increased. Here, we report enhanced surface binding stability of a phosphonate-derivatized water oxidation catalyst over a wide pH range (1–12) by atomic layer deposition of an overlayer of TiO₂. Increased stability of surface binding, and the reactivity of the bound catalyst, provides a hybrid approach to heterogeneous catalysis combining the advantages of systematic modifications possible by chemical synthesis with heterogeneous reactivity. For the surface-stabilized catalyst, greatly enhanced rates of water oxidation are observed upon addition of buffer bases and with a pathway identified in which O-atom transfer to OH⁻ occurs with a rate constant increase of 10⁶ compared to water oxidation in acid.

Heterogeneous catalysis plays an important role in industrial chemical processing, fuel reforming, and energy-producing reactions. Examples include the Haber–Bosch process, steam reforming, Ziegler–Natta polymerization, and hydrocarbon cracking (1–8). Research in heterogeneous catalysis continues to flourish (9–15) but iterative design and modification are restricted by limitations in materials preparation and experimental access to surface mechanisms. By contrast, synthetic modification of molecular catalysts is possible by readily available routes; a variety of experimental techniques is available for monitoring rates and mechanism in solution for the investigation of homogeneous catalysis (16–23). Transferring this knowledge and the reactivity of homogeneous molecular catalysts to a surface could open the door to heterogeneous applications in fuel cells, dye sensitized photoelectrochemical cells, and multiphase industrial reactions.

Procedures are available for immobilization of organometallic and coordination complexes on the surfaces of solid supports. Common strategies include surface derivatization of metal oxides by carboxylate, phosphonate, and siloxane bindings (24–27), carbon-grafted electrodes (28–30), and electropolymerization (31–33). These approaches provide a useful bridge to the interface and a way to translate mechanistic understanding and ease of synthetic modification of solution catalysts to heterogeneous applications with a promise of higher reactivity under milder conditions.

A significant barrier to this approach arises from the limited stability of surface binding. Surface-bound carboxylates are typically unstable to hydrolysis in water, whereas phosphonates are unstable in neutral or basic solutions (27, 34). For water oxidation catalysis this is particularly detrimental given the accelerated rates that are accessible for catalytic water oxidation as the pH is increased due to the intervention of base-catalyzed pathways with concerted atom–proton transfer accompanying O—O bond formation (35).

We report here the results of a designed strategy for the systematic surface stabilization of molecular catalysts on solid oxide surfaces. In the strategy we use indium tin oxide (ITO) electrodes and first bind a phosphonate-derivatized molecular water oxidation catalyst to the surface of the electrode. The derivatized electrode is then coated with a conformal nanoscale TiO₂ overlayer applied by layer-by-layer atomic layer deposition (ALD). The overlayer of TiO₂ acts to block hydrolysis of the phosphonate groups from the surface, Fig. 1. ALD stabilization has been used previously to demonstrate significantly enhanced photostability of surface-bound chromophores in acidic and neutral solutions (36–38). In this article we apply the ALD stabilization procedure to surface stabilization of a known water oxidation catalyst and show remarkably enhanced surface binding stability even in basic solutions. Retention of electrocatalytic reactivity on the surface is demonstrated and water oxidation catalysis investigated over a wide pH range. Clear evidence is found in these studies that added proton acceptor bases enhance the kinetic pathways in the key, rate-limiting step (O—O bond formation) via an atom–proton transfer (APT) mechanism (22, 35). In addition, a facile pathway has been identified with direct attack by OH⁻ on an activated oxo form of the catalyst with rate enhancements of up to 10⁶ for water oxidation.

Fig. 1.

Schematic representation of the ALD overlayer protection strategy for a catalyst surface-attached to nanoITO protected by TiO₂. (A) Illustrating the electrochemical device architecture showing the surface derivatized electrode and water oxidation. (B) Underivatized electrodes exposed to basic aqueous conditions showing detachment of the catalysts from the electrode surface. (C) ALD protection of surface attachment even basic aqueous conditions. This figure was adapted from ref. 38.

Results

Single-site, phosphonic acid-derivatized (-PO₃H₂) polypyridyl complexes of ruthenium have been shown to act as water oxidation electrocatalysts on planar ITO and FTO (fluorine-doped SnO₂) electrodes (24, 25, 39) and in mesocopic, nanostructured films of ITO (nanoITO) (40). The structure of the surface-bound catalyst [Ru(Mebimpy)(4,4′-((HO)₂OPCH₂)₂bpy)(OH₂)]²⁺ (Mebimpy is 2,6-bis(1-methylbenzimidazol-2yl); 4,4′-((HO)₂OPCH₂)₂bpy is 4,4′-bis-methlylenephosphonato-2,2′-bipyridine), nanoITO-Ru-OH₂²⁺, and the mechanism of interfacial water oxidation are shown in Scheme 1 (25).

The catalytic cycle in Fig. 2 is initiated by accumulating multiple oxidative equivalents at the catalyst through stepwise proton coupled electron transfer (PCET) reactions, , followed by 1e⁻ oxidation of Ru^IV = O²⁺ to Ru^V(O)³⁺ which occurs at E_1/2 ∼ 1.6 V vs. the normal hydrogen electrode (NHE). The Ru^V(O)³⁺ intermediate is reactive toward O—O bond formation with the initial product being a hydroperoxide intermediate, Ru^III-OOH²⁺. This O—O bond forming step is rate determining in acid solution with k = 0.009 s⁻¹ (22). Oxidation of Ru^III-OOH²⁺ to the peroxide Ru^IV(OO)²⁺ is followed by slow O₂ evolution. Further oxidation of the Ru^IV peroxide to Ru^V(OO)³⁺, which is accessible under oxidative conditions where Ru^V = O³⁺ can be generated, increases the lability of O₂, leading to reentry into the catalytic cycle through O₂ loss and regeneration of Ru^III-OH²⁺ (22, 25, 41).

Fig. 2.

Water oxidation mechanism for nanoITO-Ru-OH₂²⁺.

In acidic conditions, the rate-limiting step for this catalytic process is O—O bond formation, . Based on kinetic isotope effects (35) and quantum mechanics/molecular mechanics (QM/MM)-minimal free-energy path method calculations (42), O—O bond formation occurs in concert with solvation of the released proton by a neighboring water molecule or water cluster (35). However, water is a poor proton acceptor base (pK_a = −1.74 for ) and catalysis is enhanced with the added proton acceptor bases and . These proton acceptor bases activate the concerted APT pathway shown in Eq. 1 in which O—O bond formation occurs in concert with proton loss to the added base (35). Attempts to exploit APT pathways for phosphonate-derivatized catalysts on oxide and semiconductor electrodes have been greatly limited by hydrolytic loss of the catalyst from the surface of the electrode as the pH is increased above 5 with added coordinating buffer base anions such as acetate and (27, 34, 43).

Stabilization by ALD.

We reported earlier on the utilization of Al₂O₃ ALD layers to stabilize surface binding of the chromophore [Ru(bpy)₂(4,4′-(PO₃H₂)bpy)]²⁺on TiO₂ surfaces (36). ALD is a self-limiting thin-film deposition technique in which reactive vapor phase precursors are sequentially exposed to a substrate surface. Through controlled surface reactions and sequential precursor delivery, ALD permits exquisite control over layer thickness and conformality even over complex 3D nanoarchitectures (44), and is useful in a variety of electrochemical system applications (45). Here, we use TiCl₄/H₂O precursor chemistry to deposit TiO₂ films with a thickness control of 0.5 Å per ALD cycle.

The TiO₂ overlayer has no effect on E_1/2 values or on the pH dependences for the surface-bound Ru(III/II) couple. As shown by cyclic voltammetry (CV) measurements, E_1/2 values are consistent with previously reported values and the pH dependence of the surface-bound Ru^III/II varied as expected with the surface pK_a values of 1.8 for and 11.5 for (24, 25, 41). Distortions observed in the waveforms for the couple are consistent with prior observations and are caused by kinetic inhibitions arising from PCET effects (46).

Stabilization Toward Redox Cycling.

Repetitive CV scans through the Ru^III/II redox couple were used to evaluate the surface stabilization effect of the ALD overlayers (Fig. S1) (47). Surface coverage values (Γ in mol/cm²) were determined by integration of the current-potential waveform for the Ru^III/II redox couple and by use of Eq. 2 with Q_CV the integrated charge from the current-potential waveform, n (=1) the number of electrons transferred for the redox couple, F the Faraday constant, and A the surface area of the electrode (39).

Fig. 3A shows a plot of Γ versus number of CV scans for at pH 7. With no overlayer protection, >60% of the catalyst is desorbed from the electrode surface after 50 CV scans from 0 to 1.2 V vs. NHE at a scan rate of 20 mVs⁻¹. With 10 ALD cycles of TiO₂ (∼5.0 Å), only 20% of the surface-bound catalyst is lost. Stability is further increased at 20 ALD cycles (10 Å) where only ∼10% of the catalyst is lost after 50 CVs. Above 30 ALD cycles (≥15 Å), the Ru^III/II wave is difficult to detect, which is attributed to the 12–14-Å diameter catalyst being “buried” beneath the ALD overlayer. Thus, all subsequent experiments are carried out at an ALD thickness of ∼10 Å (20 ALD cycles).

Fig. 3.

(A) Variation in fractional surface coverage Γ/Γ_o with number of CV scans for ALD(TiO₂)-protected as a function of ALD TiO₂ cycles. , ALD cycles as labeled on the figure at pH = 7.1, I = 0.1 M, , total I adjusted to 0.25 M with added LiClO₄, T = 23 °C, 20 mV/s. (B) Variation of surface coverage with number of CV scans for ALD(TiO₂)-protected as a function of pH. Γ_o ∼ 1 × 10⁻⁸ mol/cm²; number of ALD cycles and pH as labeled on the figure. pH = 7.1, I = 0.1 M, ; pH = 9.0, I = 0.1 M, B(OH)₃/B(OH)₂O⁻; pH = 11.1, I = 0.1 M . Total I adjusted to 0.25 M with added LiClO₄, T = 23 °C, 20 mV/s.

Fig. 3B shows that ALD stabilization is effective over a wide pH range even in basic solutions. coated with 10 Å of TiO₂ (20 cycles) exhibited similar stability in solutions from pH 7 to pH 11. Under these conditions, the catalyst is nearly completely lost from the electrode surface at pH 11 without ALD stabilization. Surface binding stability and catalyst complex stability are separate issues. Both surface-binding and catalyst complex stability are observed during CV redox cycling through the wave for the Ru^III/Ru^II couple at all pH values. However, at pH ≥11, a more rapid loss of current response for the catalyst is observed with application of the high positive potentials required for water oxidation. The loss of catalytic activity appears to be due to ligand decomposition, consistent with known reactivity in solution (48, 49), and not to surface detachment due to hydrolysis.

Water Oxidation Catalysis.

With a stabilization protocol in place, we investigated water oxidation catalysis by surface-stabilized (TiO₂) (1-ALD). Following Scheme 1, water oxidation in acidic solution is triggered by oxidation of to at E ∼ 1.6 V in 0.1 M HNO₃ (22). The results of a controlled potential electrolysis experiment for 1-ALD at 1.40 V vs. NHE at pH = 7.1, I = 0.25 M, , are shown in Fig. S2. Gas chromatography analysis of the head space in the electrolysis cell, Fig. S3, revealed that O₂ was produced with a 74% Faradaic efficiency over a period of 1,000 s with a turnover number of 260 moles of O₂ per moles of catalyst, which corresponds to a turnover frequency of 0.26 turnovers per catalyst/s. This value is 2 orders of magnitude greater than rates reported at pH 1 at a constant overpotential of 0.58 V (22). The rate enhancement is consistent with the appearance of the concerted APT mechanism in Eq. 1 with as the proton acceptor base.

Contributions from APT depend both on the concentration of buffer base and the pK_a of the conjugate acid. At a fixed pH of 7.1 and fixed ratio (pK_a = 7.2), the catalytic current i_cat increases linearly with (Fig. S4). This observation is consistent with the relationship between i_cat and k_obs in Eq. 3 where k_obs = k_H2O + k_B . In this expression, k_H2O is the catalytic rate constant in the absence of buffer base, k_B is the first-order rate constant for the APT pathway, and V is the volume of the nanoITO electrode. A similar buffer base effect, but with accelerated kinetics compared with pH 7, was observed at pH 9 with B(OH)₂O⁻ as the buffer base [pK_a = 9.2 for B(OH)₃] (Fig. S5).

Fig. 4 shows CVs for 1-ALD illustrating a decrease in the onset potential for water oxidation of 0.25 V from pH = 7.1 with to pH = 11.1 with . The pH-dependent waves in the CVs from 0.3 to 0.6 V are due to the - coupled with E_1/2 = 0.58 V at pH = 7.1, 0.46 V at pH = 9, and 0.33 at pH = 11.1. The shifts in the E_1/2 values with pH follow the expected Nernstian response for this PCET couple and agree with previously reported potentials for the complex in solution (41).

Fig. 4.

CVs for 1-ALD at 23 °C and 20 mV/s in I = 0.1 M of the added buffers at pH values indicated in figure. The dashed line is the background current at pH 11.1 from a nanoITO electrode with 20 ALD cycles of TiO₂ and no catalyst loading.

By pH 11, the concentration of OH⁻ is no longer negligible and a dependence on [OH⁻] appears in the rate of water oxidation. The dependence on [OH⁻] was investigated between pH 11 and 12.5 by varying the buffer ratio at constant at fixed ionic strength. A Tafel analysis, similar to the recently developed foot of the wave analysis by Savéant and coworkers (50), was used to analyze the data. In this procedure, catalytic currents at the potential for the –Ru^V(O)³⁺/–Ru^IV = O²⁺ couple in Scheme 1 (1.6 V) are determined by extrapolation to minimize complications from the electrochemical background.

A plot of i_cat vs. [OH⁻] is shown in Fig. 5. The first-order dependence on [OH⁻] is consistent with direct attack on the reactive intermediate –Ru^V(O)³⁺by OH⁻ to give the hydroperoxide intermediate as shown in Eq. 4. At the highest concentration of OH⁻ investigated, 0.013 M (pH = 12.1), the catalytic current enhancement compared with acidic solution (pH = 1) is ∼10⁶ with an increase in k_obs from ∼10⁻² ⋅s⁻¹ at pH 1 to ∼10⁴ s⁻¹ at pH 12 at an overpotential of 1.02 V.

Fig. 5.

Variation of i_cat normalized for surface coverage with [OH⁻] in 0.25 M LiClO₄ with . The i_cat values at 1.6 V were obtained by Tafel extrapolation; see text and Fig. S6 for details.

Discussion

We demonstrate here surface stabilization and enhanced performance toward water oxidation catalysis by a surface-attached molecular catalyst. There are important findings in these results: 1) the retention of properties, including reactivity, on the surface; 2) a high degree of stability both toward surface binding and water oxidation reactivity at higher pH values; 3) a general basis for synthesizing tailored surfaces combining features of molecular and heterogeneous catalysis.

Our results have clear implications for utilization of the ALD overlayer strategy for stabilization of molecular catalysts and assemblies on oxide electrode and semiconductor surfaces for possible applications in catalysis, electrocatalysis, and photoelectrochemistry.

Methods

All aqueous solutions were prepared with Milli-Q ultrapure water (>18 MΩ ITO glass, R_s = 4–8 Ω/G) was purchased from Delta Technologies. nanoITO powder (40 nm diameter) was obtained from Lihochem. Optically transparent, electrically conductive, high surface area nanoITO films were prepared as described previously (40). The light blue nanoITO films were prepared with an average thickness of 5 μm with a resistance of ∼200 Ω across a 1-cm section of the film. Synthesis of 1 has been previously reported (41). Stable phosphonate surface binding of 1 on nanoITO electrodes occurred following immersion of the films in solutions containing 0.1 mM catalyst in methanol. Electrodes were immersed for >12 h in the catalyst solution, and immediately upon removal were rinsed with Milli-Q water and loaded into the ALD reactor chamber.

ALD was conducted in a home-built, hot-walled, flow tube reactor. The main reaction chamber is a 24-inch-long, 4-inch inner diameter stainless steel tube. Precursors were delivered into the reaction zone through a manifold constructed from 1/4-inch stainless steel tubing. Nitrogen carrier gas (99.999% purity, National Welders) was metered through a mass flow controller at 300 standard cubic centimeters per minute and exhausted through a rotary vane pump. Precursor gases were pulsed into the reactor using three-way pneumatically actuated diaphragm valves controlled electronically by a LabVIEW sequencer. Gate valves were positioned on either side of the reaction zone to conduct hold steps that permit precursor infiltration into nanoITO structure. For TiO₂ deposition, 99% pure TiCl₄ (Strem Chemicals) is used as the metal precursor and reagent-grade water (Ricca Chemicals) is used as the oxidant. Standard ALD coating conditions were 120 °C and 2 Torr of N₂ carrier gas with a sequence of 0.3-s metal precursor dose, 60-s hold, 180-s N₂ purge, 0.3-s H₂O dose, 60-s hold, 180-s N₂ purge. A 120 °C deposition temperature was selected to improve water desorption during purge steps. Silicon monitor wafers included during deposition on nanoITO electrodes confirmed the expected 0.5 Å per cycle deposition rate for each run.

Electrochemical measurements were performed with a model 601D electrochemical workstation from CH Instruments. The three-electrode system consisted of a nanoITO film on a glass slide working electrode (roughly 0.5 cm² area), a Pt wire auxiliary electrode, and a saturated calomel electrode reference. The potential of the reference electrode was adjusted by 0.24 V for the reported potentials versus NHE. GC analyses were performed on a custom-made Varian 450-GC 220-MS spectrometer with capillary column.