Stabilized photoanodes for water oxidation by integration of organic dyes, water oxidation catalysts, and electron-transfer mediators

Significance

Dye-sensitized photoelectrosynthesis cells (DSPECs) are a very promising approach to convert solar energy into chemical fuels as H₂ or reduced CO₂ species from water or CO₂. Water oxidation occurring in the photoanode, involving 4e⁻ process, is a critical reaction for the artificial photosynthesis. Here, we inserted an electron mediator between light harvester and water oxidation catalyst. In this role it acts as a mimic for the tyrosine (Yz) between the chromophore and catalyst in PSII. The resulting assembly structures were stable for extended periods (3 h) toward water oxidation to O₂, and they also extended the chromophores used in the DSPECs for water oxidation to a phosphonated porphyrin dye.

Abstract

Stabilized photoanodes for light-driven water oxidation have been prepared on nanoparticle core/shell electrodes with surface-stabilized donor–acceptor chromophores, a water oxidation catalyst, and an electron-transfer mediator. For the electrode, fluorine-doped tin oxide FTO|SnO₂/TiO₂|-Org1-|1.1 nm Al₂O₃|-RuP²⁺-WOC (water oxidation catalyst) with Org1 (1-cyano-2-(4-(diphenylamino)phenyl)vinyl)phosphonic acid), the mediator RuP²⁺ ([Ru(4,4-(PO₃H₂)₂-2,2-bipyridine)(2,2-bipyridine)₂]²⁺), and the WOC, Ru(bda)(py(CH₂)_(3or10)P(O₃H)₂)₂ (bda is 2,2-bipyridine-6,6-dicarboxylate with x = 3 or 10), solar excitation resulted in photocurrents of ∼500 µA/cm² and quantitative O₂ evolution at pH 4.65. Related results were obtained for other Ru(II) polypyridyl mediators. For the organic dye PP (5-(4-(dihydroxyphosphoryl)phenyl)-10,15,20-Tris(mesityl)porphyrin), solar water oxidation occurred with a driving force near 0 V.

In dye-sensitized photoelectrosynthesis cells (DSPECs), solar energy is converted into solar fuels, hydrogen, and oxygen by water splitting or carbon-based fuels by CO₂ reduction (1–4). Both are driven by the absorption of visible light by dye molecules on the surfaces of high-bandgap semiconductors (5, 6). At a photoanode for water oxidation to O₂, TiO₂ provides an accessible conduction band and redox levels that are accessible following excited-state injection (7). In DSPEC photoanodes, the redox equivalents for water oxidation are concentrated at a water oxidation catalyst (WOC) where it is used for preparing oxygen by a sequence of multielectron/multiproton reactions (8–10).

In these reactions, a limit arises from the availability of suitable chromophores. They require potentials that are suitable for water oxidation with excited states that can inject electrons into the conduction bands of appropriate semiconductor electrodes (11). In dye-sensitized solar cell applications, Ru(II) polypyridyl dyes have dominated as light absorbers, but organic dyes have also been used (12–18). For DSPECs, significant limitations can arise from the instabilities of the oxidized forms of the dyes in aqueous solution (19–21). Recently, we reported water oxidation from surfaces with organic push–pull dyes with an added WOC (19, 22). The combined light-absorber/catalyst electrodes led to water oxidation but with limited lifetimes because of the instabilities of the cationic forms of the chromophores in water, although stabilization of the dyes on the surface of TiO₂ by atomic layer deposition (ALD) of Al₂O₃ has been reported (19, 20).

Here, we extend the approach by adding an external electron-transfer mediator, RuP²⁺, to surfaces stabilized by the addition of Al₂O₃, as shown in Fig. 1. Addition of the complex, with ${\text{E}}_{1/2}=1.28\text{V}$ versus normal hydrogen electrode (NHE) for the Ru(III/II) couple in 0.1 M HClO₄, creates an electron-transfer pathway for transferring oxidative equivalents from the oxidized chromophore to the catalyst (23, 24).

Fig. 1.

(Left) Organic dye-chromophore catalyst assembly. (Right) Assembly with a Ru(II) polypyridyl, electron-transfer mediator.

Structures are shown in Fig. 2. On these electrodes, RuP²⁺ is a visible light absorber that competes with the organic dye, but, in the high-energy visible region of the spectrum, which is dominated by the dye, it is a minor contributor. Preparation and characterization of the electrodes, FTO|SnO₂/TiO₂|-dye-|Al₂O₃|-RuP²⁺-WOC, is described in the experiment.

Fig. 2.

Chromophore and WOC structures.

As a final step in preparing working electrodes, a derivatized version of the WOC, Ru(bda)(L)₂ (bda is 2,2-bipyridine-6,6-dicarboxylate) $\text{L}=4-\text{pyO}{\left({\text{CH}}_2\right)}_{\left(3\text{,}10\right)}{\text{PO}}_3{\text{H}}_2$ was added to give the final assemblies, FTO|SnO₂/TiO₂|-organic dye-|Al₂O₃|-RuP²⁺-WOC. Of interest in comparing the catalysts was the role of the long-chain -(CH₂)_3,10- structure in dictating photocurrent efficiencies, Fig. 3.

Fig. 3.

Electrode configuration illustrating the -(CH₂)₁₀- extended carbon chain in the catalyst.

In the electrode structures, -RuP²⁺- was added as electron-transfer mediator. With RuP²⁺ in the outer layer, with a redox potential sufficient to transfer electrons away from the surface dye by oxidation to -RuP³⁺-, the mediator provides a basis for subsequent oxidation of the external catalyst, -|-Org1^•+ -|1.1 nm Al₂O₃|-RuP²⁺-WOC → -|-Org1-|1.1 nm Al₂O₃|-RuP³⁺-WOC → -|-Org1-|1.1 nm Al₂O₃|-RuP²⁺-WOC⁺.

Results and Discussion

Electrode Characterization.

The organic dyes have visible absorption maxima at ∼400 nm, depending on the environment and substituents. Absorption manifolds can extend to ∼500 nm with substituents that can vary from hydrogen to methoxide, as shown in SI Appendix, Fig. S1 (19). As obtained by cyclic voltammetry measurements, E_1/2 values for Org1^•+/0and Org2^•+/0 in aqueous 0.1 M KPF₆ are ∼1.04 and 1.03 V versus NHE, with some uncertainty in the values from kinetic distortions in the voltammograms. In acetonitrile, E_1/2 values for the Org1^•+/0, Org2^•+/0, and Org3^•+/0 couples are 1.23, 1.21, and 1.03 V versus NHE, respectively (19).

The influence of surface loading of RuP²⁺on added Al₂O₃ is shown in the UV-vis spectra in Fig. 4. Increasing the ALD-deposited Al₂O₃ overlayer from 0.36 nm to 1.1 nm resulted in enhancement of the RuP²⁺ MCLT absorptions at ∼450 nm. Further increases past 1.1 nm to 1.6 nm resulted in no further increase. Nine ALD cycles of the Al₂O₃ overlayer were equivalent to ∼1.1 nm which is the approximate diameter of Org1.

Fig. 4.

UV-visible measurements on FTO|TiO₂|-Org1-|xAl₂O₃|-RuP²⁺, illustrating the influence of the Al₂O₃ layer on the extent of -RuP²⁺ loading with λ_max ∼ 450 nm. The Al₂O₃ layer was increased by ∼0.12 nm per added Al₂O₃ layer. Full surface coverage occurred at nine cycles of added Al₂O₃.

Photoelectrochemical Water Oxidation.

Water oxidation was explored in core/shell nanoparticle SnO₂/TiO₂ films with the core/shell structure assisting electron–hole separation at the SnO₂/TiO₂ interface (25–29). Surface-loaded versions of related assemblies were investigated previously (30). Based on results from earlier studies, surface loading occurs at a 2:1 ratio of RuP²⁺ compared with the catalyst WOC(A) (30). With the structures of the catalysts shown in Fig. 2, and the supporting –(CH₂)_n- methylene chains of 3 or 10, the catalysts are well separated from the initial oxide surface.

From the data in Fig. 5, photocurrent densities at anodes with different Al₂O₃ thicknesses, and varying amounts of -RuP²⁺, gave similar photoresponses. In the assembly FTO|SnO₂/TiO₂|-Org1-|Al₂O₃|-RuP²⁺-WOC(A), there was no evidence for enhanced photocurrents compared with FTO|SnO₂/TiO₂|-Org1-|Al₂O₃|-WOC(A) except for long-term stability. Similar configurations, but with Org2 or Org3 as the chromophores, in FTO|SnO₂/TiO₂|-Org2 or Org3-|1.1 nm Al₂O₃|-RuP²⁺-WOC(A), gave comparable j-t traces (SI Appendix, Figs. S5 and S11) with no evidence in the photocurrent profiles for a contribution by -RuP²⁺-.

$$\left({\text{Q}}_{\text{Collector}}/{\text{Q}}_{\text{Generator}}\right)/0.70\hspace{0.17em}\times \hspace{0.17em}100\text{\%}=\text{FE}\left(\text{\%}\right).$$ [1]

To test the stability and efficiency of the electrodes toward O₂ generation, collector–generator (C-G) experiments were carried out with a dual-electrode design described elsewhere (31, 32). The results of a 1-h illumination period with an ∼1-sun intensity light source (100-mW cm⁻² and 400-nm long-pass filter) are shown in Fig. 6. At the end of a photolysis cycle, the generator current decayed instantaneously with slower decay at generator electrode as diffusion of O₂ to the electrode occurs. Faradaic efficiencies (FE) for O₂ production were calculated from Eq. 1 with Q_collector and Q_generator the total charge passed at the collector and generator electrodes, respectively. The constant, 0.7, is the experimentally derived, collection efficiency for the cell, as described previously (32). In contrast to measurements on organic dyes without RuP²⁺, the photoanodes were relatively stable over 1-h test periods with FE for O₂ production of ∼100%.

Fig. 5.

Current density−time (j-t) traces over 10-s dark−light cycles for water oxidation by the electrodes, FTO|SnO₂/TiO₂|-Org1-|xAl₂O₃|-RuP²⁺-WOC(A) with variations in x from 0 to 12 cycles at an applied bias of 0.4 V versus Ag/AgCl; pH = 4.65, 0.1 M acetate, 0.4 M NaClO₄.

Fig. 6.

O₂ measurements for water oxidation (black) from, FTO|SnO₂/TiO₂|-Org1-|1.1 nm Al₂O₃|-RuP²⁺-WOC(A), on 1-cm² slides illuminated with 100-mW cm⁻² white light with a 400-nm cutoff filter from 30 to 3,630 s at a bias of 0.4 V versus Ag/AgCl. The current–time response in red is for the O₂ collector electrode held 1 mm from the photoanode poised at −0.85 V versus Ag/AgCl; in 0.1 M acetic acid/acetate buffer at pH 4.65 containing 0.4 M NaClO₄.

Introduction of RuP²⁺ not only improved the stability of the photoanodes, it also led to enhanced O₂ production. Addition of RuP²⁺ resulted in enhancements in the FE for O₂ production from 43 to 100%, in SI Appendix, Figs. S2–S6. Similar results were obtained for assemblies with Org2 and Org3 as the dyes based on photocurrent and O₂ measurements, in SI Appendix, Figs. S7–S16.

Variations in Cell Configuration.

Dye reversal.

In one series of experiments, electrodes were prepared with the position of RuP²⁺ and the dyes reversed, FTO|SnO₂/TiO₂|-RuP²⁺-|xAl₂O₃|-Org1. Before attaching the catalyst to the surface, three cycles (∼0.36 nm) of ALD-Al₂O₃ were added to stabilize Org1 on the surface. As shown by the data in SI Appendix, Figs. S17–S20, inverting the sequence of dye and mediator resulted in a slight increase in photocurrent density. However, the change in sequence caused a notable decrease in stability because of the Org1^•+ reactivity of the cation with the solvent in water oxidation cycles. Photocurrents fell from ∼1 mA/cm² to 65 µA/cm² for FTO|SnO₂/TiO₂|-RuP²⁺-|1.1 nm Al₂O₃|-Org1-|0.36 nm Al₂O₃|-WOC(A) over a period of 1 h, as shown in SI Appendix, Fig. S20.

Dye variations.

The role of other ruthenium polypyridyl complexes as mediators was investigated in films with RuP²⁺ replaced by [Ru(4,4′-CH₂-PO₃H₂-bpy)(bpy)₂]²⁺, RuCP²⁺, and [Ru(4,4′-PO₃H₂-bpy)(4,4′-Me-bpy)₂]²⁺, RuPMe²⁺. The structures of the complexes are shown in SI Appendix, Scheme S2 with E_1/2 values for their Ru(III/II) couples, 1.22 and 1.19 V versus NHE in 0.1 M HClO₄ solution (24). From the data, SI Appendix, Figs. S21 and S22, after 1 h of photolysis under 1-sun illumination, a photocurrent density of 270 µA/cm² was obtained for FTO|SnO₂/TiO₂|-Org1-|1.1 nm Al₂O₃|-RuCP²⁺-WOC(A) and 210 µA/cm² for FTO|SnO₂/TiO₂|-Org1-|1.1 nm Al₂O₃|-RuPMe²⁺-WOC(A). As for RuP²⁺ as the mediator, addition of both complexes greatly enhanced the stabilities of the photoanodes.

Incident Photon to Current Efficiency (IPCE) measurements.

The impact of RuP²⁺ on photocurrent response was also investigated by IPCE measurements with and without RuP²⁺. Results for the electrodes FTO|SnO₂/TiO₂|-Org1-|1.1 nm Al₂O₃|-RuP²⁺-WOC(A) are shown in Fig. 7. Based on the data, there was no evidence for participation by RuP²⁺ as a light absorber in the reactions. Related results are shown for the dyes Org2 and Org3 in SI Appendix, Figs. S23 and S24.

Fig. 7.

IPCE measurements on FTO|SnO₂/TiO₂|-Org1-|1.1 nm Al₂O₃|-RuP²⁺-WOC(A), at an applied bias of 0.4 V versus Ag/AgCl at pH = 4.65 in 0.1 M acetate buffer, 0.4 M in NaClO₄. A 400-nm cutoff filter was used to mimic the conditions in Fig. 2.

Light intensity.

To explore the role of light intensity, the photoelectrochemical response of the electrode, FTO|SnO₂/TiO₂|-Org1-|1.1 nm Al₂O₃|-RuP²⁺-WOC, as a function of incident light intensity from a 1-sun solar lamp, was also explored over range of intensities from 0.08 to 0.4 I_o. Based on the data in SI Appendix, Fig. S25, the variation with light intensity from the 1-sun incident light source was linear from 0.08 to 0.4 I_o, consistent with single-photon excitation and activation of the assembly.

Mechanism.

A possible mechanism for the origin of the photocurrents is shown in Eqs. 2–5 with Org1 as the organic dye. In the scheme, light absorption by the organic chromophore was followed by injection from the excited state, -Org1*-. Following injection, the radical cation Org1^•+ oxidizes RuP²⁺ on the surface of the film, which in turn oxidizes the external catalyst WOC. As suggested in Eq. 4, with the addition of external RuP²⁺ on the outside of the film, injection is followed by electron transfer from the dye cation to RuP²⁺ to give the intermediate, -Org1-|1.1 nm Al₂O₃|-RuP³⁺-WOC, with stabilization of the assembly due to the electron mediator effect of the RuP^3+/2+ couple.

$$\begin{array}{l}\text{FTO}\left|{\text{SnO}}_2/{\text{TiO}}_2\right|-\mathbf{Org1}-\left|1{.1\hspace{0.17em}\text{nm Al}}_2{\text{O}}_3\right|-{\text{RuP}}^{2+}-\mathbf{WOC}+\text{h}v\to \\ \text{FTO}\left|{\text{SnO}}_2/{\text{TiO}}_2\right|-\mathbf{Org1}\mathbf{*}-\left|1{.1\hspace{0.17em}\text{nm Al}}_2{\text{O}}_3\right|-{\text{RuP}}^{2+}-\mathbf{WOC}\end{array}$$ [2]

$$\begin{array}{l}\text{FTO}\left|{\text{SnO}}_2/{\text{TiO}}_2\right|-\mathbf{Org1}\mathbf{*}-\left|1{.1\hspace{0.17em}\text{nm Al}}_2{\text{O}}_3\right|-\mathbf{Ru}{\mathbf{P}}^{2+}-\mathbf{WOC}\to \\ \text{FTO}\left({\text{e}}^{-}\right)\left|{\text{SnO}}_2/\text{TiO}2\right|–{\mathbf{Org1}}^{•+}-\left|1{.1\hspace{0.17em}\text{nm Al}}_2{\text{O}}_3\right|-\mathbf{Ru}{\mathbf{P}}^{2+}-\mathbf{WOC}\end{array}$$ [3]

$$\begin{array}{l}\text{FTO}\left({\text{e}}^{–}\right)\left|{\text{SnO}}_2/{\text{TiO}}_2\right|-{\mathbf{Org1}}^{•+}-\left|1{.1\hspace{0.17em}\text{nm Al}}_2{\text{O}}_3\right|-\mathbf{Ru}{\mathbf{P}}^{2+}-\mathbf{WOC}\to \\ \text{FTO}\left({\text{e}}^{–}\right)\left|{\text{SnO}}_2/{\text{TiO}}_2\right|-\mathbf{Org1}-\left|1{.1\hspace{0.17em}\text{nm Al}}_2{\text{O}}_3\right|-\mathbf{Ru}{\mathbf{P}}^{3+}-\mathbf{WOC}\end{array}$$ [4]

$$\begin{array}{l}\text{FTO}\left({\text{e}}^{–}\right)\left|{\text{SnO}}_2/{\text{TiO}}_2\right|-\mathbf{Org}1-\left|1{.1\hspace{0.17em}\text{nm Al}}_2{\text{O}}_3\right|-\mathbf{Ru}{\mathbf{P}}^{3+}-\mathbf{WOC}\to \\ \text{FTO}\left({\text{e}}^{–}\right)\left|{\text{SnO}}_2/{\text{TiO}}_2\right|-\mathbf{Org}1-\left|1{.1\hspace{0.17em}\text{nm Al}}_2{\text{O}}_3\right|-\mathbf{Ru}{\mathbf{P}}^{2+}-\mathbf{WO}{\mathbf{C}}^{+}.\end{array}$$ [5]

In the scheme, RuP²⁺ may act as an electron-transfer mediator between the oxidized dye and the external catalyst. Based on redox potentials for the RuP^3+/2+ couple in 0.1 M aqueous HClO₄, and for the Org1^•+/0 couple in 0.1 M aqueous KPF₆, oxidation of RuP²⁺ to RuP³⁺ by Org1^•+ is nominally disfavored by ∼0.2 V. There was no evidence for a buildup of RuP³⁺ during the experiment and we assume, that in the film structure, the reaction is either favorable or, if unfavorable, with a sufficiently low driving force that it does not inhibit catalyst activation. Based on the light-intensity experiments described above, there was also no evidence for additional steps in the mechanism involving multiphoton excitation of RuP²⁺ in the photocatalytic cycle.

Transient Dynamics.

Transient absorption measurements were undertaken to explore excited-state processes following excitation of Org1 in FTO|SnO₂/TiO₂|-Org1-|1.1 nm Al₂O₃, and FTO|SnO₂/TiO₂|-Org1-|1.1 nm Al₂O_3|-RuP²⁺. In transient absorption experiments on FTO|SnO₂/TiO₂|-Org1-|1.1 nm Al₂O₃, the kinetics of back electron transfer from the surface to Org1^•+ occur with a half time of 5 µs following 425-nm excitation. In a second series of experiments, transient absorption measurements were conducted on FTO|SnO₂/TiO₂|-Org1-|1.1 nm Al₂O₃|-RuP²⁺, under the same conditions, but at an excitation wavelength of 535 nm, above the absorption maximum for the organic dye. Under these conditions, RuP²⁺ is the dominant light absorber in the electrode. Following excitation at 535 nm only excitation of RuP²⁺ to give RuP^2+* was observed and it decayed with 280 ns, SI Appendix, Figs. S26 and S27. There was no evidence in the data for injection or for the appearance of RuP³⁺.

Structural Variations.

Alkyl chain length.

To further explore the role of structure in WOC, we investigated the role of distance from the surface by varying the bridge to the catalyst from –(CH₂)₃- to –(CH₂)₁₀- in FTO|SnO₂/TiO₂|-Org1-|1.1 nm Al₂O₃|-RuP²⁺-WOC(B). In the fully extended structure, the separation distance from the surface was ∼1.4 nm. The results of earlier experiments have shown that longer alkyl chains can lead to stabilization of the catalyst toward hydrolysis from the surface (33). Photocurrent measurements under the conditions in Fig. 5 resulted in photocurrent densities of ∼0.5 mA/cm² which lasted for 3 h without a significant loss in photocurrent. As shown in Fig. 8, O₂ evolution was retained over the entire photolysis period.

Fig. 8.

O₂ measurements for water oxidation from (black), FTO|SnO₂/TiO₂|-Org1-|1.1 nm Al₂O₃|-RuP²⁺-WOC(B), on 1-cm² slides illuminated with a 100-mW cm⁻² white-light source with a 400-nm cutoff filter for 180 min at a bias of 0.4 V versus Ag/AgCl. The current–time response in red was for an O₂ sensor electrode, 1 mm from the photoanode biased at −0.85 V versus Ag/AgCl in 0.1 M acetic acid/acetate buffer at pH 4.65 in 0.4 M NaClO₄.

Other chromophores.

The electrode procedure was extended to the phosphonate derivatized, free-base porphyrin dye, 5-(4-(dihydroxyphosphoryl)phenyl)-10,15,20-Tris(mesityl)porphyrin, (PP), as the added chromophore, Fig. 9. Porphyrins are attractive chromophores due to their excellent light absorption in the visible but have not been effective as DSPEC water oxidation (34–37). E_p,a for the initial PP^•+/0 oxidation wave was 1.05 V versus NHE in pH 4.65 buffer in 0.1 M acetate, 0.4 M NaClO₄, SI Appendix, Fig. S28. At this pH, electron transfer from the cation PP^•+ to RuP²⁺ following injection, is also disfavored by ∼0.2 V.

Fig. 9.

Electrode configuration for the PP dye as the chromophore for electron transfer in FTO|SnO₂/TiO₂|-PP-|1.1 nm Al₂O₃|-RuP²⁺-WOC(A).

The porphyrin was chosen as the chromophore because the potential for the PP^•+/0 couple is within 0–50 mV of the driving force for water oxidation to O₂ at $\text{pH}=4.65$. As for RuP²⁺ as the electron-transfer mediator, water oxidation does occur for the PP chromophore. Based on the experimental data, without RuP²⁺ as the mediator in FTO|SnO₂/TiO₂|-PP-|1.1 nm Al₂O₃|-WOC(A), O₂ production was minimal, SI Appendix, Fig. S29. After adding RuP²⁺ in a second mediator layer gives FTO|SnO₂/TiO₂|-Porphyrin-|1.1 nm Al₂O₃|-RuP²⁺-WOC(A), photocurrents were initiated at levels of 300–400 µA/cm², SI Appendix, Fig. S30. Utilization of the O₂ measurement procedure after photolysis periods of 10 min gave FE of 78%. IPCE measurements on electrodes in SI Appendix, Fig. S31 with the porphyrin chromophore overlapped with its spectrum without evidence for a significant contribution from -RuP²⁺- and a dramatic example in which photoelectrochemical water oxidation is carried out with a nearly 0-V driving force.

Conclusions

The examples cited here are important in illustrating key structural features for transiently stable photoanodes for water oxidation based on three push–pull organic dyes, a derivative of the Ru(bda) catalyst for water oxidation, and an electron-transfer mediator. The key elements in the structure were the use of ALD to stabilize the surface-bound chromophores, an overlayer of Al₂O₃ for surface stabilization, and surface binding of both a WOC and the electron-transfer mediator. Addition of RuP²⁺ as a mediator removes electrons from the organic dyes following injection stabilizing the structure, allowing for electron-transfer activation of the external catalyst. As for tyrosine in PSII, the mediator is essential in the overall structure by enabling rapid transfer of oxidative equivalents from the oxidized chromophore to the catalyst. The results are impressive with stable photocurrents observed over a 3-h photolysis period with ∼100% O₂ production in the assembly FTO|SnO₂/TiO₂|-Org1-|1.1 nm Al₂O₃|-RuP²⁺-WOC(B).

It is also notable that systematic variations could be made in the assembly structure. Exchange of the organic dye by RuP²⁺ resulted in a photoanode for water oxidation, but with limited stability because of the external instability of the external radical cation of the dye. Notably, exchange of the organic chromophore by the PP dye also resulted in water oxidation, in a photoanode with a driving force near 0 V for O₂ evolution.

Materials and Methods

Fluorine-doped tin oxide (FTO) was purchased from Hartford Glass with a sheet resistance of 15 Ω/sq. TiO₂ films and nanoITO films were made following a well-developed recipe. The preparation of SnO₂/TiO₂ core/shell structures is described below.

Preparation of SnO₂ Films.

SnO₂ paste was prepared by using colloidal suspension (15% w/v) of SnO₂ particles from Alfa Aesar. In the procedure, 1 g of acetic acid was added to 37 g of a SnO₂ colloidal solution to prepare the paste. The resulting mixture was sealed and stirred overnight. In the next step, the as-prepared solution was transferred to an autoclave (100 mL volume). The autoclave was sealed and placed in a box oven and heated for 45 min at 240° for 80 h. After the mixture was cooled, the colloids were redispersed in a sonic hall from Branson Ultrasonics for 2 min. To the dispersed mixture was added mixed polyethylene oxide (mol wt 100,000) and polyethylene glycol (mol wt 12,000). The resulting mixture was covered and stirred a further 48 h. A doctor-blading method was utilized to make the SnO₂ film. The slides were annealed at 450° for 1 h to give 4∼5-µm, 20-nm SnO₂ nanoparticle films.

ALD.

ALD was performed by using a Cambridge NanoTech Savannah S200 instrument with TDMAT [tetrakis(dimethylamino)titanium] as Ti precursor. ALD was performed in a commercial reactor (Savannah S200; Cambridge Nanotech). The chamber pressure is 760 torr when exposed to air. Titanium dioxide (TiO₂) was deposited with Tetrakis (dimethylamido) titanium, Ti(NMe₂)₄ (TDMAT, 99.999%; Sigma-Aldrich), and water. The reactor chamber temperature was 150 °C. The TDMAT reservoir was kept at 75 °C (2-h preheating). ALD coating conditions were 150 °C and 20 torr of N₂ carrier gas with a sequence of 0.5-s metal precursor dose, 20-s hold, 60-s N₂ purge, 0.02-s H₂O dose, 20-s hold, 60-s N₂ purge.

After deposition of 75 cycles of TiO₂ on SnO₂ films, the slides were placed into box oven for 30 min with annealing under 450°.

ALD coating of Al₂O₃ utilized the recipe: chamber temperature 130 °C and 20 torr of N₂ carrier gas with a sequence of 0.015-s metal precursor dose, 20-s hold, 60-s N₂ purge, 0.015-s H₂O dose, 20-s hold, 60-s N₂ purge.

Preparation of Electrodes.

The FTO|SnO₂/TiO₂ slides were prepared by immersing the slides into organic dye solutions (2 mM) with soaking for 2 h. For the PP dye, the slides were soaked in 2 mM PP DMSO/methanol 1:1 solution for 12 h) followed by rinsing with a methanol solution. The dye-loaded slides were placed in an ALD chamber and subjected to deposition of the Al₂O₃ overlayer. After depositing the metal-oxide layer, the slides were immersed in 1 mM methanol solutions of RuP²⁺ for an additional 24 h followed by rinsing with pure methanol. The treated slides were immersed in freshly prepared solutions of the catalyst in methanol at 3 mM for 24 h to give the final assemblies, FTO|SnO₂/TiO₂|-dye-|Al₂O₃|-RuP²⁺-WOC. The addition of catalyst at the end of the assembly was performed in the glovebox.

Measurements.

Electrochemical and photoelectrochemical experiments were performed by using either a CH Instruments 660D potentiostat or a CH Instruments 760E bipotentiostat. A Thor Labs HPLS 30–04 light source was used to provide white-light illumination. For all indicated experiments using 100-mW cm⁻² white-light illumination, the electrochemical cell was positioned an appropriate distance from the light source to receive the indicated light intensity as measured with a photodiode (Newport), and a 400-nm cutoff filter (Newport) was used to prevent direct bandgap excitation of the semiconductor layer.

In the water oxidation experiments, a two-compartment cell with a Nafion membrane was used in a three-electrode configuration, with Ag/AgCl as the reference electrode and a Pt mesh counter electrode for H2 evolution. The experiments were carried out under N₂ at pH = 4.65 in a 0.1 M aqueous sodium acetate buffers in 0.4 M NaClO₄ with a 100-mW/cm² white-light source (400-nm cutoff filter) at a bias of 0.4 V versus Ag/AgCl.