Visible-light-absorbing semiconductor/molecular catalyst hybrid photoelectrodes for H₂ or O₂ evolution: recent advances and challenges

Abstract

The research on the conversion of solar energy and its storage as an eco-friendly and momentarily available chemical fuel, such as H₂, by sunlight-driven water splitting is closely related to the sustainable development of the global economy and to the continuous improvement of the modern living standards of human beings. One of the most promising approaches to sunlight-driven water splitting is to construct a dual-illuminated photoelectrochemical (PEC) cell by integrating a photoanode with a photocathode in a tandem configuration. To this end, the important work is to individually develop highly efficient, durable, inexpensive, and readily scalable photoanodes and photocathodes for each half reaction of water splitting, either O₂ or H₂ evolution reaction (OER or HER). A promising approach emerging in recent years towards OER photoanodes and HER photocathodes is the immobilization of molecular catalysts (MC) onto the surface of visible-light-absorbing semiconductor (VLASC) electrodes. Very recently, some encouraging results have been achieved in the construction of MC-modified VLASC photoanodes and photocathodes. This review is focused on the recent advances in hybrid photoelectrodes for OER and HER, which were built by the integration of MCs with VLASC materials. After a brief introduction of three major units, viz. VLASC materials, MCs, and anchor groups, used to date for fabricating hybrid photoelectrodes for OER and HER, the construction strategy and the performance of the VLASC/MC photoanodes and photocathodes are described in two respective chapters. Finally, challenges and developments in future studies of VLASC/MC hybrid photoelectrodes are discussed.

---

Mei Wang

Prof. Mei Wang received her Ph.D. in 1989 from Dalian Institute of Chemical Physics (China). Afterwards, she worked as an Alexander von Humboldt fellow from 1989 to 1991 at Technische Universität München with Prof. Dr Wolfgang A. Herrmann. She has been a full professor at Dalian University of Technology (China) since 2002. Her research interests are focused on the chemical mimics of [FeFe]-H2ases, and electrochemical and photochemical hydrogen production by water splitting using nonprecious metal-based catalysts.

Yong Yang

Yong Yang received his M.S. degree from Dalian University of Technology in 2013; he majored in applied chemistry. He is currently pursuing his Ph.D. under supervision of Professor Mei Wang. His research interests are focused on photoelectrochemical water splitting devices and non-precious metal-based catalysts for the hydrogen evolution reaction.

Junyu Shen

Junyu Shen received his B.S. degree from Changzhou Institute of Technology in 2012; he majored in chemical engineering and technology. He is currently pursuing his Ph.D. degree under the supervision of Prof. Mei Wang at Dalian University of Technology. His research interests are focused on non-precious metal-based homogeneous and heterogeneous electrocatalysts for water oxidation reaction.

Jian Jiang

Jian Jiang received his B.S. degree from Jiangsu University of Technology in 2014, and majored in chemical engineering. He is currently a Ph.D. candidate in applied chemistry under the supervision of Prof. Mei Wang at Dalian University of Technology. His research interests are focused on the non-precious metal-based electrocatalysts for the hydrogen and oxygen evolution reaction.

Licheng Sun

Prof. Licheng Sun received his Ph.D. in 1990 from Dalian University of Technology. He worked as a postdoc at the Max-Planck-Institutfür Strahlenchemie with Dr Helmut Görner (1992–1993), and then at the Freie Universität Berlin (1993–1995) as an Alexander von Humboldt fellow with Prof. Harry Kurreck. He moved to the Royal Institute of Technology (KTH), Stockholm in 1995 and became assistant professor in 1997, associate professor in 1999 (Stockholm University) and full professor in 2004 (KTH).

---

1 Introduction

The conversion of solar energy into an eco-friendly and momentarily available chemical fuel, in the form of H–H bonds, by sunlight-driven water splitting, and subsequent storage are great aspirations of chemists.^1,2 Over several decades, substantial progress and some breakthroughs have been achieved, especially in recent years, in developing efficient, durable, and cheap photocatalytic systems and devices for solar-to-hydrogen (STH) conversion, thanks to the persistent efforts of a great number of research groups around the world. In general, three technological approaches have been developed for effective light-driven overall water splitting: (i) connecting photovoltaic modules with discrete electrolyzers to build a PV/electrolyzer apparatus;^3–5 (ii) integrating a photoanode with a cathode or a photocathode in a tandem configuration to construct a photoelectrochemical (PEC) cell;^6–8 (iii) assembling a broad band semiconductor material (SCM) with water oxidation catalyst (WOC) and/or water reduction catalyst (WRC) to form a one-pot photocatalyst colloid system.^9,10 Several perspectives and tutorial reviews have commented on the advantages and drawbacks of different device designs.^11–14 Among these three approaches, the integrated PEC cell, an optimum design compromise of a PV/electrolyzer apparatus and a one-pot photocatalyst colloid system, is a most promising approach to large-scale H₂ production from the perspective of the STH conversion efficiency, cost, safety, and availability.¹⁵

PEC water splitting technology allows the harvesting of sunlight and the electrolysis of water to occur simultaneously in a single device in an aqueous electrolyte. In general terms, a PEC cell can be constructed in two ways: by the integration of a photoanode and a photocathode to form a dual-illuminated tandem PEC cell, or by the assembly of a photoactive electrode, either a photoanode or a photocathode, with a photo-inactive counter-electrode to build a single-illuminated PEC cell.¹⁶ Detailed categories and fundamental knowledge of PEC cells have been introduced by some tutorial papers.^17–22 The single-illuminated PEC cell, which was first reported in 1972,²³ required extra electrical energy for driving the water splitting reaction and therefore cannot provide high STH efficiency, due to the narrow absorption of sunlight by a single photoelectrode. Spontaneous water splitting PEC cells, i.e., without the need for bias, with a dual-illuminated tandem device design, have been demonstrated in the laboratory.²⁴ Therefore, dual-illuminated PEC cells have become an attractive target for many researchers in current years.^25–27 To build PEC cells with high STH efficiency, the integrated photoanode and photocathode should have well-matched tandem band energy levels, complementary light absorption properties, compatible H₂- and O₂-evolution rates, and similar durability in testing electrolytes. In the advanced stages towards this end, the important work is to individually develop highly efficient, durable, inexpensive, and readily scalable photoanodes and photocathodes for each half reaction of water splitting.

An electrocatalyst-modified photoelectrode for the half reaction of water splitting, either the O₂ or H₂ evolution reaction (OER or HER), functions in four steps as follows: the absorption of solar energy by a light absorber on the electrode, the charge separation in the light absorber, the charge transport from the light absorber to an electrocatalyst, and the occurrence of OER or HER at the surface of the electrode, driven by the photo-generated redox potential stored in the catalyst. Therefore, light absorbers and OER or HER electrocatalysts are two crucial components for fabricating effective photoanodes and photocathodes. According to the types of light absorbers and electrocatalysts, photoelectrodes can be categorized into all-inorganic electrodes (visible-light-absorbing semiconductor (VLASC)/inorganic material catalyst) and SC-molecular hybrid electrodes, including VLASC/molecular catalyst (MC), dye-sensitized SC/MC, and dye-sensitized SC/inorganic material catalyst. The OER photoanodes and HER photocathodes that were constructed by the combination of SCs or dye-sensitized SCs with inorganic material catalysts as well as the corresponding integrated PEC cells have been widely studied. The advances in this field have been summarized in some recent reviews and perspectives.^28–31 The present review focuses on the recent advances in MC-modified hybrid photoelectrodes for OER and HER, built by the combination of MCs with VLASCs. The MC-based three-component hybrid systems containing either sacrificial electron acceptors or electron donors and the photoelectrodes made of organic dye- and organometallic chromophore-sensitized SCs will not be included in this review; the progress on these topics have been reviewed in several recently published papers.^32–40 In this review, we first make a general introduction of three major components, viz. SCMs, MCs, and anchor units, used to date for fabricating VLASC/MC hybrid photoelectrodes for OER and HER, which is followed by a brief description of the strategies for immobilizing MCs onto the surface of SCMs. The recent advances in MC-modified hybrid photoanodes for OER and photocathodes for HER are presented, respectively, in subsequent two chapters. Photoelectrodes that were fabricated by using organometallic complexes as precursors to deposit inorganic WOCs or WRCs onto SC surfaces will not be included in these two chapters.^41,42 In the last chapter, challenges and developments in future studies of MC-modified hybrid photoelectrodes are discussed.

2 Assembling of VLASC/MC hybrid photoelectrodes

For fabricating eventually available hybrid photoanodes and photocathodes used for PEC cells (Fig. 1), molecular electrocatalysts with high activity at low overpotential for OER or HER and SCs with broad absorption in the solar spectrum, long excited state lifetime, and high charge carrier mobility are eagerly desired. Additionally, both of these components are expected to be inexpensive, eco-friendly, and robust under operating conditions. The intrinsic activity and stability of a hybrid photoelectrode depend predominantly on the properties of SCM and MC, which compose the photoelectrode. In addition to SCM and MC, the length, rigidity, orientation, and electronic properties of the linker between MC and SCM, as well as the type of anchor, also conspicuously influence the PEC performance of VLASC/MC hybrid photoelectrodes. To date, no photoelectrodes made for OER or HER satisfy all the requirements for practical application in a PEC cell for sunlight-driven overall water splitting at zero or low bias.

Fig. 1 Schematic of a hybrid photoanode and photocathode constructed with elemental building units, VLASC, OER-MC or HER-MC, anchor, and linker.

2.1 VLASC materials used for hybrid photoelectrodes

SCMs that are used as sole light absorbers (without sensitization) in photoelectrodes must feature a proper band gap and suitable energy positions of valence and conduction band edges. The SCMs with band gaps larger than 3.1 eV cannot absorb visible light, which accounts for more than 95% irradiation of the solar spectrum, while a narrow band gap will lead to small photovoltage in a photoelectrode under illumination. The energy position of the valence band (VB) upper edges of SCMs used for photoanodes must be more positive than the O₂/H₂O thermodynamic potential, while the position of the conduction band (CB) lower edges of SCMs used for photocathodes should be more negative than the H⁺/H₂ thermodynamic potential. Even though some SCMs satisfy these stringent thermodynamic criteria, the poor kinetics for OER or HER at the surface of these materials constrains their photocatalytic efficiencies.⁴³ Therefore, it is necessary to load the co-catalysts on the SC photoelectrodes to facilitate the OER and HER. Many solid material electrocatalysts for OER or HER have been loaded onto the surface of various SC photoelectrodes, and the advances in this topic have been documented in the literature.^44–46 In the other aspect, organic dye- and organometallic chromophore-sensitized photoelectrodes have been widely studied over the past decade.^32–40 Considering that SCMs generally have better photostability and broader spectral ranges in the solar spectrum than molecular dyes, the combination of MCs with inorganic visible-light-absorbing SCMs can be a promising approach to constructing hybrid photoelectrodes with higher activity and better stability, compared to dye-sensitized photoelectrodes. To date, the sorts of SCMs used for fabricating MC-modified SC hybrid photoelectrodes for OER and HER are still quite limited.

Metal oxides, Fe₂O₃, WO₃, and BiVO₄, are the most commonly used visible-light-absorbing SCMs for fabricating hybrid photoanodes.^47–51 The band gaps are about 2.2 eV for α-Fe₂O₃, 2.6 eV for WO₃, and 2.4 eV for BiVO₄ (Fig. 2),^44,52 corresponding to optical absorption edges of about 560, 480, and 520 nm, respectively, for these materials. The VB upper edges of these metal oxides are more positive than the thermodynamic potential for OER by nearly a volt or greater, while the energy positions of their CB lower edges are less negative than the thermodynamic potential for HER. On the basis of the VB and CB energy positions and the band gaps of these metal oxides, they are suitable for making OER photoanodes. It should be noticed that Fe₂O₃ and BiVO₄ offer good stability in basic and neutral electrolytes, while WO₃, unlike most metal oxides, has good stability towards corrosion in acidic electrolytes, which is a preferred condition for HER. When designing a hybrid photoelectrode, one should consider not only the proper alignment of the band edges of SCMs and the Frontier orbitals of MCs, but also the compatible stability of these two components in contact with electrolytes.

Fig. 2 Band gaps and band positions for the SCMs mentioned in this review, in vacuum or in an aqueous electrolyte at pH 0.

In addition to metal oxides, Ta₃N₅ and n-type Si are also modified by MCs to fabricate hybrid photoanodes. Tantalum nitride (Ta₃N₅) has a band gap of 2.06 eV with VB and CB energy positions straddling the O₂/H₂O and H⁺/H₂ thermodynamic potentials.^53,54 The oxidative corrosion at the surface under photolysis conditions is a fatal shortness for photoanodes made of Ta₃N₅ and n-type Si, which are therefore usually covered with a thin TiO₂ protective layer to improve the stability of such photoelectrodes.⁵⁵

The sorts of VLASCs for constructing hybrid photocathodes are even less than those for photoanodes. Only p-type Si, In- and/or Ga-containing SCMs have been used for VLASC/MC hybrid photocathodes (Fig. 2). Silicon with a narrow band gap of 1.12 eV can capture photons in a significant portion of the solar spectrum.⁵⁶ Although the CB lower edge of p-Si is a few hundred millivolts more negative than the H⁺/H₂ thermodynamic potential, the pristine p-Si is almost inactive for HER under illumination, due to the sluggish charge-transfer kinetics at the Si/electrolyte interface. Several recently reported examples have proved that the modification of the surface of a Si photocathode by molecular electrocatalysts, in addition to widely studied solid material electrocatalysts, can effectively improve the HER kinetics at the Si electrode/electrolyte interface. To overcome the oxidative corrosion problem of Si-based photocathodes, the most commonly used strategy is to decorate a TiO₂ thin layer on the surface of Si electrodes by atomic layer deposition (ALD).^57–60 Such a layer not only protects the surface of Si material from corrosion, but also provides available hydroxyl groups for grafting MCs on the electrode surface. SCs of group III, InP, GaP, and GaInP₂, have desirable band gaps, viz. 1.35 eV for InP,^61,62 2.26 eV for GaP,^52,63 and 1.83 eV for InGaP₂,^64,65 with their CB lower edges several hundred millivolts more negative than the standard H⁺/H₂ potential. In light of the broad absorption of solar energy and high charge carrier mobility of the In- and/or Ga-containing SCs, these materials are attractive candidates for photocatalytic H₂ production. The sulphophile affinity of group III metals and the hydroxyl-terminated surface make both coordinately and covalently anchoring MCs onto the surface of an In- and/or Ga-based SC photocathode practically feasible. Although these photocathodes have demonstrated very high STH efficiencies,²⁴ the corrosion instability and high cost are critical problems that inflict limitations on the large-scale application of the In- and Ga-based SCMs in photocatalytic H₂ production devices.

2.2 MCs used for hybrid photoelectrodes

Although all-inorganic VLASC/solid catalyst photoanodes and photocathodes have been widely studied for decades, there was only one report on the VLASC/MC hybrid photoelectrode for OER and HER before 2010.⁶⁶ The research on MCs for the half reactions of water splitting was concentrated on the homogeneous or heterogeneous photocatalytic systems, where either a sacrificial electron acceptor or a donor was required.^67–70 In recent years, many efforts have been devoted to pushing the research forward from the three-component photocatalytic systems to PEC devices, in which sacrificial electron acceptors and donors are no longer needed.

Decoration of inorganic solid catalysts on the surface of SC photoelectrodes can effectively improve the performance of some photoelectrodes, but may also affect the SC-liquid junction energetics or add deleterious series resistance to the SC-electrocatalyst contact.⁷¹ The surface defect and the grain boundary of solid catalysts could form a barrier hindering the charge transfer between the light absorber and the catalyst. In addition, a large loading amount of non-transparent inorganic catalysts will cause a reduction in the light harvesting efficiency of VLASC photoelectrodes, while a small loading amount of catalysts cannot significantly improve the kinetics of photoelectrodes for OER or HER at the SC/electrolyte interfaces. In contrast to inorganic catalysts, MCs possess the following advantages: (i) the structures of MCs can be modified by ligand design with the purpose of tuning the redox properties, catalytic activity, and stability of catalysts; (ii) some MCs display very high catalytic activity for OER or HER; (iii) clear structures and sufficient purity of MCs provide a requisite condition for a repeatable catalytic performance; (iv) MCs can be covalently or coordinately grafted to the surface of SCMs to form an ultrathin monolayer and thus to realize single-site and atom-efficient catalysis; (v) such a molecular layer is generally light penetrable and electrolyte permeable. An avoidable fatal problem of MCs is that they are generally less stable than inorganic catalysts. In some cases, organometallic compounds serve only as precursors for the in situ formation of electroactive nanomaterial catalysts.^72,73 Noticeably, it was found that immobilization of some MCs to an electrode could make them more stable than their free counterparts in solutions.³⁷

The proper combination of SCM and MC is a key issue in the development of highly efficient hybrid photoelectrodes for PEC water splitting cells. First of all, it must be ensured that the desired charge transport between SC and MC is thermodynamically feasible in the designed hybrid photoelectrodes, while the unwanted reverse charge transfer can be suppressed. This means that in order to build an effective hybrid photoanode, the energy level for the catalysis of OER by the selected OER catalyst must be less positive than the VB upper edge of the SCM to be assembled, but more positive than the standard O₂/H₂O potential. Additionally the lowest unoccupied molecular orbital (LUMO) level of the OER catalyst should be more negative than the CB upper edge of the SCM (Fig. 3). Similarly, to build an effective hybrid photocathode, the energy level for catalysis of HER by the selected HER catalyst must be less negative than the CB lower edge of the SCM to be immobilized, but more negative than the standard H⁺/H₂ potential, and the highest occupied molecular orbital (HOMO) level of the HER catalyst should be more positive than the VB lower edge of the SCM. Additionally, the stability of the selected MC in aqueous solutions of different pH values should be compatible with that of the SCM to be grafted. In order to firmly immobilize MCs onto the surface of SCMs, MCs must be functionalized with anchor groups, which can form strong covalent or coordinate bonds with the surface of SCMs. The following section will briefly introduce the assembling approaches to hybrid photoelectrodes.

Fig. 3 Schematic of the energy level alignments for the VLASC/MC hybrid photoanode and photocathode (EL for OER and EL for HER in the diagram represent the energy level for catalysis of OER and HER by the catalyst, respectively).

2.3 Covalent immobilization of MCs to SCs

When MC and SC are selected in terms of the criteria mentioned above, a critical problem for conceiving an efficient and durable hybrid photoelectrode is how to graft the MC onto the surface of SC. Although physisorption of hydrophobic MCs onto the surface of SC-based photoelectrodes can be conveniently conducted by drop-casting, dip-coating, or spin-coating, many examples have demonstrated that the hybrid photoelectrodes built by these methods are not durable because of the easy desorption of the physically adsorbed catalyst molecules from the electrode surface into the aqueous electrolyte.^54,74,75 In comparison, polymer-coating with the help of Nafion can provide more stable films doped with desired catalysts. However, such films often display very low photocurrent densities, due to the limited number of catalyst host sites within the Nafion,^76,77 and for photoanodes also due to the increase of the overpotential of OER catalysts caused by the strong acidity of hydrophilic head groups (sulfonic acid) in Nafion.^78–80 Compared with physisorption methods, covalent bonding of MCs to the surface of photoelectrodes can afford more stable hybrid electrodes, in addition to providing a better control of the density and orientation of the catalyst molecules at the surface of electrodes. Therefore, the chemical immobilization strategy is one of the crucial problems for constructing efficient VLASC/MC hybrid photoelectrodes for OER and HER. In addition to anchor groups, the length, orientation, flexibility, conjugation unit, and extra functional group of the linker, as well as the electrostatic interaction, viz. attraction or repulsion between the surface of SCM and the ionic MC, will also considerably influence the stability and efficiency of the VLASC/MC hybrid photoelectrodes. For an effective and practical approach to immobilizing MC, the employed reaction conditions should be mild and should not cause detriment to the SCM and MC during the grafting process. All bonds in the linker should withstand prolonged exposure to electrolyte solutions under photocatalytic conditions; all structural elements of the linker should have no deleterious effect on the charge transfer and catalytic activity of the modified photoelectrode.

The anchor group at the ligand of a catalyst should be adopted on the basis of the chemical reactivity of the electrode surface to be modified. In general, –COOH,⁸¹ –PO₃H₂,^82–84 as well as –Si(OH)₃ (ref. 85 and 86) and –CONH(OH)^87,88 are suitable anchor groups for the surface modification of metal oxide SC electrodes by the formation of O–metal iono-covalent bonds. Among these immobilization approaches, the hydroxamate linkage with a bidentate chelating group –C(O)NOH– binds tightly to metal oxides and has a greater resistance to hydrolysis, compared to carboxylate and phosphonate bondings; comparative studies demonstrate that the attachment through phosphonate linkages is significantly more stable than through carbonate linkages, especially to metal cations of high oxidation states.^82,84 Very recent studies by Sakai and co-workers showed that pyridine can function as a much stronger anchor group to the TiO₂ surface than the conventional –COOH and –PO₃H₂ in aqueous electrolytes.⁸⁹ Terminal alkenyl and alkynyl groups are useful for the surface functionalization of Si and In/Ga-containing SCMs by UV-light induced formation of C–Si or C–OM (M = Ga, In) bonds.⁹⁰ Moreover, thiol groups can be used for sulfophilic metal-containing electrodes.⁹¹ According to the assembling operations, MC immobilization has been made by the following four strategies: (i) direct bonding of MCs to the surface of SCMs; (ii) functionalizing MCs with appropriate anchor group(s), which can form covalent bonds with the surface of SCMs; (iii) functionalizing the surface of SCMs with desired ligands, which can readily coordinate to the metal centers of catalyst precursors; (iv) separately functionalizing both the MC and the SC surface, and then integrating the two as-prepared building blocks by certain organic reactions. It is clear that the first strategy is a straightforward approach, with neither MC nor SC surface needing to be extraordinarily functionalized prior to integration; however it is a challenging work to design and prepare MCs that possess such self-assembling ability to a targeted semiconductor material. The approach with complete preparation of the anchor group-functionalized catalyst prior to immobilization is more effective than that with ligand modification of the SC surface prior to metal complexation, because the coordination reaction occurs faster in homogeneous solution than at the electrode surface. Moreover, in the former case, the as-prepared anchor group-functionalized catalysts can be isolated and purified to ensure the quality of catalysts, while in the latter case, incomplete metalation of the grafted ligand is often encountered. The last strategy requires that one of the functional groups on MC or at SC surfaces must be highly reactive, such as an azide or a phthalimide ester moiety, to promote the desired integration reactions. Such strict requirement has limited the wide application of this approach. Furthermore, integration of a VLASC/MC photoelectrode by functionalization of both MC and the SC surface often leads to a long linker that is unfavorable for charge transfer between the SC and the grafted MC.

3 VLASC/MC hybrid photoanodes for OER

Water oxidation is generally acknowledged as the major obstacle for total water splitting, due to the large overpotential required for the four-electron and four-proton transfer process. Therefore, the development of photoanodes featuring large photovoltage and high photocurrent for OER is especially important for constructing eventually available PEC cells for spontaneous water splitting. In the early work on MC-modified photoanodes, catalysts were mostly adhered to the dye-sensitized TiO₂ photoanode surfaces by using Nafion film, or by physisorption.⁷⁷ The performance of dye-sensitized TiO₂ photoanodes was somewhat improved by co-anchoring dye and MC through covalent linkages to the TiO₂ surface, or by attaching MC to the molecular photosensitizer, which was grafted to the TiO₂ surface.^92,93

Reports on the new type of hybrid photoanodes with immobilization of MCs onto the surface of VLASC electrodes have appeared in the literature in very recent years. The most commonly used OER MCs to be anchored to the SC surfaces are Ru and Ir complexes,^68,70 for which pyridine, bipyridine, terpyridine, and their derivatives are the most adopted ligands. The hybrid photoanodes with non-noble metal MCs immobilized on the surface of VLASC electrodes are limited to less than ten examples in the literatures to date. The following subsections are divided according to the metal center of MCs in VLASC/MC hybrid photoanodes for OER. The benchmark of photocurrent densities reported so far for VLASC/MC photoanodes (without an additional hole-storage layer) is 2.1 mA cm⁻² at an applied potential of 1.23 V (potentials reported here are all versus the reversible hydrogen electrode (RHE), unless otherwise noted) under simulated 1-sun illumination.⁹⁴ The performances of reported VLASC/MC hybrid photoanodes are summarized in Table 1, and the structures of MCs listed in Table 1 can be found in Fig. 4–14.

Table 1 Performances and testing conditions of VLASC/MC hybrid photoanodes^a

Photoanode	Anchor or graft method	Testing condition	Photocurrent density^b (mA cm^−2)	Onset potential (CS)^c (V)	Stability	Ref.
a All potentials given in Table 1 are versus RHE. b All photocurrent densities given in Table 1 are taken from linear sweep voltammograms at 1.23 V vs. RHE. c CS = cathodic shift of photocurrent onset potential (Eonset) as compared to the corresponding bare SC photoanode. d PBS = phosphate buffer solution. e NA = nanorod array. f PDC = 2,6-pyridinedicarboxylic acid. g Fh = ferrhydrite.
Fe2O3/MC1	COOH	0.1 M KHC8H4O4–HCl, pH 3, 66 mW cm^−2	0.12	0.71 (0.2)	—	79
Fe2O3/MC2	COOH		0.18	0.6 (0.4)	Stable over 0.28 h
Fe2O3/MC3	Nafion film		0.11	0.6 (0.4)	—
WO3/MC4	PO(OH)2	0.5 M KPF6/0.1 M HNO3, 100 mW cm^−2	1.8	0.6 (0)	Desorbed over 0.5 h	95
BiVO4/MC5	COOH	0.1 M PBS,^d pH 7.1, 100 mW cm^−2	∼1.4	0.4 (0.2)	Photocurrent slowly declined	96
Fe2O3 NA/MC6^e	PDC^f	1 M KOH, λ > 400 nm	1.8	0.8 (0)	Stable over 2.78 h	97
Fe2O3 NA/MC10^e	Direct bonding	0.1 M KNO3, pH 1.01, 100 mW cm^−2	0.9	0.6 (0.25)	Deactivated after 1.33 h	98
H2-treated α-Fe2O3−x/MC10	Direct bonding	pH 7.3, 100 mW cm^−2	∼0.03	1.2 (0.2)	—	99
Ta3N5/TiOx/Fh/Ni(OH)x/MC11^g	Dip-coating	1 M NaOH solution, pH 13.6, 100 mW cm^−2	∼11.5	∼0.65	—	54
Ta3N5/TiOx/Fh/Ni(OH)x/MC12	Dip-coating		∼11.6	∼0.62 (0)	—
Ta3N5/TiOx/Fh/Ni(OH)x/MC11, MC12	Dip-coating		12.1	0.6 (0.02)	Small decrease in current over 0.33 h PEC test at 0.9 V
Ta3N5/TiOx/Fh/Ni(OH)x	No MC		10.7	∼0.62	—
Fe2O3/MC12	Nafion film	0.1 M PBS,^d pH 8, 300 W Xe lamp, λ > 420 nm	0.4	∼0.67 (0.4)	Stable over 2 h PEC test at 1.27 V	80
Fe2O3/MC13	Hydrophobic interaction	0.1 M NaOH, pH 13, 100 mW cm^−2	∼0.5	1.0 (0.03)	—	100
Fe2O3/MC14			1.0	0.8 (0.2)	Stable over 2 h PEC test at 1.37 V
BiVO4/MC15	COOH	0.1 M Na2SO4, pH 6.8, 100 mW cm^−2	2.1	0.3 (0.4)	Stable over 2 h PEC test at 1.4 V	94
BiVO4/MC16	COOH	0.1 M NaClO4, pH 5.5, 100 mW cm^−2, λ > 430 nm	—	∼0.64 (0.35)	Photocurrent declined during PEC test	101
WO3/MC17	PO(OEt)2	0.1 M Na2SO4, pH 3, 100 mW cm^−2	1.1	0.5 (0)	Slow desorption of MC	102
WO3/MC18	PO(OH)2	0.1 M Na2SO4, pH 3, 200 mW cm^−2	1.6	0.5 (0)	Slow desorption of MC	103

---

Fig. 4 Molecular structures of ruthenium OER catalysts MC1–MC3.

3.1 Molecular ruthenium catalyst-modified hybrid photoanodes

The pioneer work in developing MC-modified hybrid photoanodes was reported by Zhuang and co-workers in 2013.⁷⁹ To distinguish the contributions of the catalytic effect of the grafted MC and the heterojunction effect of the SC/molecule interface, three Ru catalysts (MC1, MC2, MC3, Fig. 4) were used to modify α-Fe₂O₃ electrodes. Catalysts MC2 and MC3 bearing a 2-phenylpyridine ligand displayed higher electrocatalytic activity than MC1 containing a 2,2′-bipyridine ligand. MC1 and MC2 were covalently linked to the surface of α-Fe₂O₃ electrodes through a carboxylate linkage, while MC3 without anchor group was adhered to the α-Fe₂O₃ surface with Nafion film. The Fe₂O₃/MC2 photoanode exhibited a photocurrent density of 0.18 mA cm⁻² at 1.23 V in pH 3 buffer solution under 66 mW cm⁻² illumination, which was approximately 1.5- and 2-fold higher than that observed for the Fe₂O₃/MC1 and the pristine α-Fe₂O₃, respectively. Besides, the photovoltages were about 200 mV for Fe₂O₃/MC1 and 300 mV for Fe₂O₃/MC2. The discrepancy between the performances of Fe₂O₃/MC1 and Fe₂O₃/MC2 was attributed to the different electrocatalytic properties of the two OER catalysts. In another aspect, the Fe₂O₃/MC3 displayed a lower photocurrent density (0.11 mA cm⁻²) than the Fe₂O₃/MC2. These comparative studies revealed that proper design of the ligands of MCs could make the hybrid photoanodes benefit from both the high catalytic activity of grafted catalysts and the heterojunction effect of the SC/molecule interface.

In the same year as the publication of Fe₂O₃/MC1–MC3 photoanodes, the first MC-modified WO₃ photoanode was reported by Fujita and co-workers.⁹⁵ The Ru catalyst MC4 (Fig. 5) had two configurational isomers, distal (d) and proximal (p), due to the asymmetric nature of the ligand pynap (pynap = 2-(pyrid-2′-yl)-1,8-naphthyridine). Of these isomers, only the d-MC4 displayed electrocatalytic activity toward WOR. To suppress the photoinduced isomerization of the active d-MC4 to the inactive p-MC4, the d-MC4 was anchored to the surface of the WO₃ photoanode through a phosphonate group. Compared to the pristine WO₃ electrode, the WO₃/d-MC4 exhibited ∼60 mV cathodic shift at 1 mA cm⁻² in the J–E curve measured in 0.5 M KPF₆/0.1 M HNO₃ under simulated 1-sun illumination. Unfortunately, the plateau photocurrent density decreased from 2.1 mA cm⁻² for bare WO₃ to 1.8 mA cm⁻² for the WO₃/d-MC4. The decrease in photocurrent was ascribed to partial visible light absorption by the Ru catalyst and charge recombination within the catalyst layer. An almost complete desorption of d-MC4 from the WO₃ surface was observed over the course of 30 min, indicating the easy hydrolysis of surface PO–W bonds under testing conditions.

Fig. 5 Molecular structures of ruthenium OER catalysts MC4 and MC5.

In 2015, de Respinis, Joya and their co-workers reported the first MC-modified BiVO₄ photoanode.⁹⁶ The Ru catalyst MC5 (Fig. 5) was grafted onto the surface of a planar BiVO₄ photoanode through a carboxylate linkage. The photocurrent density of the BiVO₄/MC5 was significantly enhanced at low potentials (0.6–1.7 V) as compared to that of the pristine BiVO₄ in 0.1 M neutral PBS under 100 mW cm⁻² illumination. Specifically, it was 5.5 times higher than that of the bare BiVO₄ electrode at 0.8 V. Compared with the performance of the BiVO₄ photoanodes modified by CoP_i and RuO₂, two of the most effective inorganic catalysts for OER, the BiVO₄/MC5 displayed a significantly higher photocurrent than BiVO₄/RuO₂ and BiVO₄/CoP_i under front illumination. More importantly, the E_onset (at 0.1 mA cm⁻²) of photocurrent for the BiVO₄/MC5 was about 300 mV more negative than that of bare BiVO₄, and it was 100 mV more negative than that of BiVO₄/CoP_i. No desorption and degradation of the catalyst was observed for the BiVO₄/MC5 in neutral solution at 1.23 V under 100 mW cm⁻² illumination over a period of 1.8 h. It is worth noting that when MC5, CoP_i, and RuO₂ were coated onto FTO, FTO/RuO₂ exhibited a superior electrocatalytic activity with an overpotential of about 0.25 V at 1 mA cm⁻², while FTO/CoP_i and FTO/MC5 displayed overpotentials of 0.7 and 0.8 V, respectively. These results indicate that the electrocatalytic activity of a catalyst is not the only important factor that one should consider when selecting an appropriate catalyst to construct an effective SC/catalyst composite photoelectrode. Some other factors, such as the light absorbing properties of catalysts, the SC/catalyst interface property, the surface charge recombination probability, as well as the penetrating ability of ions in the catalyst layer, also significantly influence the performance of composite photoelectrodes. Here the BiVO₄/MC5 photoanode benefits from the low light absorption of MC5 in the UV-vis-NIR range, the pronounced reduction of surface charge recombination, and the enhanced oxygen evolution kinetics.

The instability of MC-modified SC photoelectrodes with carboxylate or phosphate linkages to metal oxide surfaces in basic solution remains an intractable problem. To improve the stability of the linkages between MC and SCM, 2,6-pyridinedicarboxylic acid (PDC) was used as an anchor to tether a Ru catalyst (MC6, Fig. 6) onto the surface of α-Fe₂O₃ nanorod array (NA) film.⁹⁷ Indeed, the Fe₂O₃ NA/MC6 displayed excellent photocurrent stability in 1 M KOH solution under visible light illumination with a bias of 1.4 V over 2.78 h. The electrochemical impedance spectroscopic (EIS) studies indicated that the resistance at the Fe₂O₃/electrolyte interface was significantly reduced after the surface of the α-Fe₂O₃ film was modified by MC6, and therefore the photocurrent density was increased from 2 mA cm⁻² for bare α-Fe₂O₃ NA photoanode to 3 mA cm⁻² for the Fe₂O₃ NA/MC6. Although immobilization of MC6 on the α-Fe₂O₃ NA led to an apparent increase of photocurrent, the E_onset (0.8 V) of photocurrent displayed by the Fe₂O₃ NA/MC6 was just the same as that observed for the pristine α-Fe₂O₃ NA photoanode.

Fig. 6 Molecular structure of ruthenium OER catalyst MC6.

In addition to metal oxide SC photoanodes, the n-type Si electrode was also integrated with MCs. The Ru complex MC7 was grafted onto the mixed methyl/bipyridyl-functionalized Si surface by ligand substitution of Ru(acac)₂(coe)₂ (acac = acetylacetonate, coe = cis-cyclooctene) with an immobilized bipyridyl moiety (Fig. 7).¹⁰⁴ In a similar approach, the Ir (MC8) and Rh (MC9) catalysts were also attached onto the surface of Si electrodes. After assembly of the metal complexes, the Si surface retained its highly favorable photoelectronic properties. Unfortunately, although the immobilized metal complexes retained their molecular natures, all three MC-modified hybrid Si electrodes were very unstable under electrochemical cycling conditions (in CH₃CN with a sweep width of 0 V to −1.5 V vs. Fc^+/0) due to the rapid detachment of immobilized metal complexes from the Si surface upon electrochemical reduction, possibly caused by ligand-centered reduction. The poor stability of these noble metal-modified Si electrodes thwarted further studies of their photocatalytic activities.

Fig. 7 Schematic of MC7-, MC8-, and MC9-modified n-Si photoanodes.

3.2 Molecular iridium catalyst-modified hybrid photoanodes

Molecular iridium catalysts are the most attractive WOCs used for developing VLASC/MC heterojunction WO photoanodes, in terms of their very high activity and relatively low overpotentials.¹⁰⁵ Unlike most OER MCs, some iridium MCs feature unique stability in aqueous solutions at low pH. In 2015, Brudvig and co-workers found that a di-μ-oxo Ir^IV,IV dimer (MC10), formed by activation of a mononuclear Ir^III precursor, [Cp*Ir(pyalc)OH] (Cp* = pentamethylcyclopentadinenyl, pyalc = 2-(2′-pyridyl)-2-propanolate), with NaIO₄ in water, rapidly self-bound to the surface of metal oxide electrodes (Fig. 8).¹⁰⁶ Such direct binding was stable in aqueous solutions over a wide range of pH values (pH 1–12). Here, the N,O-chelating ligand played a key role for the outstanding stability of the framework of this Ir^IV,IV dimer. For Cp*Ir complexes having only monodentate ligands, the degradation of the complexes was unavoidable, leading to the formation of IrO_x particles under water oxidizing conditions.¹⁰⁷ The immobilized MC10 displayed higher electrocatalytic activity than the bulk IrO_x material for OER in acidic solutions (pH 2.6) and retained its molecular nature over long-term electrolysis.

Fig. 8 Oxidation of the Cp*Ir monomer to the dimer MC10, followed by direct binding of MC10 onto the α-Fe₂O₃ surface.

An ultra-thin monolayer of MC10 was covalently grafted to the surface of the α-Fe₂O₃ photoelectrode. Compared to the PEC performance of the bare hematite electrode, the photocurrent density of Fe₂O₃/MC10 was enhanced by about 0.5 mA cm⁻² (from 0.4 to 0.9 mA cm⁻²), at a bias of 1.23 V in 0.1 M KNO₃ at pH 1 under illumination of simulated sunlight, and the E_onset of photocurrent was cathodically shifted by 0.25 V (from 0.85 to 0.6 V).⁹⁸ The Fe₂O₃/MC10 demonstrated a similar PEC performance to that of the Fe₂O₃/IrO_x under identical conditions, but it was less stable than the Fe₂O₃/IrO_x. During the long-time photolysis of water, the rate of O₂ evolution from the Fe₂O₃/MC10 started to decrease after 60 min and leveled off after 80 min under testing conditions. The decrease in O₂ evolution efficiency was most likely caused by the catalyst detachment, instead of degradation, due to the corrosion of hematite in highly acidic solution. Replacement of hematite with a more acid-stable, narrow-band metal oxide, such as WO₃, may improve the stability of the molecular iridium catalyst-based hybrid photoanodes in highly acidic solutions.

The further in-depth studies, made by Brudvig and Wang, on the kinetics at the Fe₂O₃-based photoanode/electrolyte interfaces of bare Fe₂O₃, Fe₂O₃/MC10, and Fe₂O₃/IrO_x, provided clear experimental evidence for better understanding the mechanism of the effects of molecular (MC10) and material (IrO_x) catalysts modified on Fe₂O₃ photoanodes.¹⁰⁸ The kinetic data obtained from intensity-modulated photocurrent spectroscopic (IMPS) and photoelectrochemical impedance spectroscopic (PEIS) studies revealed that the MC10 and IrO_x attached on Fe₂O₃ surfaces played roles in the PEC performance of Fe₂O₃-based photoanodes in distinct ways. The monolayer of MC10 on the Fe₂O₃ surface did not change the surface electronic states of the Fe₂O₃|H₂O interface; therefore, it could not stop the charge transfer pathways inherent to the Fe₂O₃|H₂O interface, but it could open up additional lower-barrier charge-transfer pathways via the molecular iridium catalyst (Fig. 9). Thus, the MC10 on the surface of the Fe₂O₃ photoanode enhanced the PEC performance by speeding up hole transfer from hematite to the catalyst, without reducing the surface recombination rate. In contrast, the dense IrO_x film on the Fe₂O₃ surface completely changed the characteristic surface states of the Fe₂O₃|H₂O interface, thus leading to the replacement of the charge transfer pathways inherent to the Fe₂O₃|H₂O interface with new pathways, so as to suppress both the direct hole transfer through Fe₂O₃ to H₂O and the surface charge recombination. The IrO_x film on the Fe₂O₃ photoanode therefore improved the PEC performance by both speeding up the forward WOR kinetics and by reducing the rate of surface recombination. Furthermore, studies on the H/D kinetic isotope effects (KIE) unveiled that the OER occurring at MC10 and IrO_x took place in distinct mechanisms. For Fe₂O₃/MC10, the rate-determining step (RDS) of WOR was the hole transfer from Fe₂O₃ to MC10, and WOR occurred rapidly at the molecular iridium catalyst. In contrast, for Fe₂O₃/IrO_x, WOR is the RDS, viz. the hole transfer from IrO_x to H₂O, rather than the hole transfer from Fe₂O₃ to IrO_x.

Fig. 9 Schematic of the kinetic models for (a) bare α-Fe₂O₃, (b) α-Fe₂O₃ with MC10, and (c) α-Fe₂O₃ with IrO_x. Here, k_rec is the rate constant of charge recombination and k_trans is the rate constant of hole transfer from the surface to the solution. Reproduced from ref. 108 with permission from the Royal Society of Chemistry.

Immediately following this work, studies made by Ozin and co-workers provided a deeper insight into the effect of the molecular iridium catalyst on the different hematite materials. The catalyst MC10 was immobilized onto an ultrathin H₂-treated hematite film,⁹⁹ which featured a non-stoichiometric α-Fe₂O_3−x structure with oxygen vacancies. The Fe₂O_3−x/MC10 displayed a larger photocurrent response (0.28 mA cm⁻² at 1.75 V) than bare Fe₂O_3−x electrode, with 200–300 mV cathodic shift in photocurrent onset potential in pH 7.3–7.4 electrolytes under 100 mW cm⁻² illumination. It was found that the monolayer of the iridium catalyst on the H₂-treated hematite film substantially reduced the charge transfer resistance, compared to that for bare Fe₂O_3−x, but did not appreciably alter the capacitance of the electrode, and accordingly, did not affect the flat-band potentials of the Fe₂O_3−x photoanode. Moreover, the effect of the iridium catalyst on thermodynamics and Fermi level pinning was also found to be negligible. The improved PEC performance of the Fe₂O_3−x electrode after modification with the iridium catalyst was attributed to improved interfacial charge transfer and oxygen evolution kinetics, as well as to reduced recombination rate (Fig. 10). In contrast, the modification of an air-treated Fe₂O₃ photoanode with the iridium catalyst apparently increased the electrocatalytic activity of the electrode in the dark, but the modified air-treated Fe₂O₃ electrode displayed very limited photocurrent, due to insufficient band bending and fast charge recombination in the air-treated Fe₂O₃ material. These results reveal that both SCM engineering and MC immobilization are imperative for improving the PEC performance of hybrid photoelectrodes.

Fig. 10 Schematic showing the effect of H₂-treatment and iridium catalyst-modification on the performance of the hematite photoanode for WOR.

Very recently, Li and co-workers constructed a highly efficient Ta₃N₅-based integrated photoanode with its surface decorated with iridium (MC11) and cobalt (MC12) catalysts (Fig. 11).⁵⁴ This photoanode consisted of a Ta₃N₅ layer for light harvesting, two hole-storage layers (Ni(OH)_x/ferrhydrite (Fh)) for mediating interfacial charge transfer, and a TiO_x blocking layer for reducing surface recombination (Ta₃N₅/TiO_x/Fh/Ni(OH)_x); the MCs were adhered to the top surface of the integrated photoanode by physisorption. The Ta₃N₅/TiO_x/Fh/Ni(OH)_x, without decoration of MC, displayed a photocurrent of 10.7 mA cm⁻² at 1.23 V in 1 M NaOH solution, at pH 13.6 under 100 mW cm⁻² illumination. The decoration of MC11 or MC12 onto the integrated photoanode brought about an increase of photocurrent to 11.5–11.6 mA cm⁻² at 1.23 V. Furthermore, the co-decoration of MC11 and MC12 onto this Ta₃N₅-based photoanode resulted in a photocurrent of 12.1 mA cm⁻² at 1.23 V, with a photocurrent onset potential of about 0.6 V. The applied bias photo-to-current efficiency (ABPE) of this photoanode was 2.5% at 0.9 V. Both the photocurrent and the ABPE of Ta₃N₅/TiO_x/Fh/Ni(OH)_x/MC12/MC11 are the state-of-the-art records for VLASC/MC hybrid photoanodes. This work is a perfect example of the construction of highly efficient photoelectrodes by the combination of well-engineered inorganic materials with organometallic MCs. It can be expected that the catalytic activity and stability of this integrated composite photoanode could be further improved by covalently grafting highly active molecular catalysts onto the electrode surface, with a well-designed anchor and an optimized linker.

Fig. 11 Schematic of the integrated Ta₃N₅ photoanode modified with molecular iridium and cobalt OER catalysts (MC11, MC12). Reproduced from ref. 54 with permission from the Royal Society of Chemistry.

3.3 Molecular cobalt and copper catalyst-modified hybrid photoanodes

The first noble metal-free VLASC/MC hybrid photoanode was reported by Li, Sun, and their co-workers in 2014.⁸⁰ In light of the high activity and good stability of Co-oxo cubanes for WOR in homogeneous systems,^109,110 the Co-oxo cubane (MC12) that had a strong electron withdrawing group, –CN, at the para position of the pyridine ligand was used for integration with an Fe₂O₃ photoanode by Nafion film-coating. The Fe₂O₃/MC12 exhibited a cathodic shift of 400 mV in photocurrent onset potential (ca. 0.67 V) for WOR, as compared to that of the Nafion-coated Fe₂O₃ and bare Fe₂O₃ in 0.1 M PBS at pH 8 under illumination (300 W Xe lamp, λ > 420 nm). The transient photocurrent density of Fe₂O₃/MC12 was 0.4 mA cm⁻² at 1.23 V, which was about 8 times that displayed by the Nafion-coated Fe₂O₃. The cobalt catalyst on the Fe₂O₃ surface retained its molecular nature and high O₂-evolution activity for photoelectrolysis at 1.27 V over a period of 2 h. This finding shows that in terms of photocatalytic activity, photovoltage and stability, the integration of highly active molecular non-noble metal catalysts with visible-light-absorbing SCMs, instead of dye-sensitized SCMs, is a promising approach to creating efficient and low-cost photoelectrodes for PEC water splitting cells.

Cobalt phthalocyanine and its derivatives are another type of molecular cobalt catalysts widely studied for WOR in homogeneous systems and on FTO electrodes.¹¹¹ Takanabe and co-workers immobilized hydrogenated and fluorinated Co–phthalocyanines, MC13 and MC14 (Fig. 12), to the surface of an Fe₂O₃ photoanode via hydrophobic interaction.¹⁰⁰ It was noticed that MC14, bearing a perfluorinated phthalocyanine ligand, displayed higher electrocatalytic activity than MC13, having an unsubstituted phthalocyanine ligand in 0.1 M NaOH solution at pH 13; the current onset potential of MC14 was 110 mV less positive than that of MC13, indicative of the merit of MCs in tuning the redox potential by ligand design and functionalization. Accordingly, the photocurrent density of 1.0 mA cm⁻² at 1.23 V for Fe₂O₃/MC14 was about twice as high as that for Fe₂O₃/MC13, and about 4 times of that for bare Fe₂O₃ under 100 mW cm⁻² illumination. Notably, the photocurrent density of Fe₂O₃/MC14 was more than twice as high as that of the fully-inorganic photoanode, Fe₂O₃/Co–Pi (0.45 mA cm⁻² at 1.23 V), partly due to there being no detriment to light absorption by the immobilization of a monolayer of the molecular cobalt catalyst on the surface of Fe₂O₃. Besides the significant augmentation of photocurrent, the current onset potential was cathodically shifted from 1.03 V for bare Fe₂O₃, to 0.8 V for Fe₂O₃/MC14. The evidently improved PEC performance of Fe₂O₃/MC14 was attributed to enhanced WOR kinetics, accelerated hole transfer from the surface of Fe₂O₃ to the catalyst, and ameliorated charge separation at the catalyst-SC interface. The long-time photoelectrolysis of water on the Fe₂O₃/MC14 photoanode demonstrated that this hybrid photoanode, assembled via hydrophobic interaction of the MC to the hematite surface, was quite stable with a sustained photocurrent density of 1.1 mA cm⁻² at 1.37 V under simulated 1-sun illumination over a period of 2 h.

Fig. 12 Molecular structures of cobalt and copper OER catalysts MC13–MC16.

As Co–phthalocyanines, Co–porphyrins are also highly active electrocatalysts for WOR.^112,113 Very recently, Wu and co-workers immobilized a cobalt porphyrin catalyst (MC15, Fig. 12) onto the surface of a vertically aligned BiVO₄ nanosheet (NS) photoanode via covalent linkage through a COOH anchor.⁹⁴ It was found that the coating of a thin layer of Al₂O₃ on BiVO₄ surface played an important role in the adsorption of the cobalt catalyst and in the passivation of the BiVO₄ surface. Immobilization of MC15 to the surface of BiVO₄/Al₂O₃ resulted in a conspicuous increase of photocurrent to 2.1 mA cm⁻² at 1.23 V, which was about 2 times as high as that of the bare BiVO₄/Al₂O₃ in 0.1 M Na₂SO₄ at pH 6.8 under 100 mW cm⁻² illumination. Notably, the photocurrent displayed by this earth-abundant metal catalyst-modified BiVO₄ photoanode is apparently higher than that of the previously reported molecular ruthenium catalyst-modified BiVO₄ electrode (1.4 mA cm⁻² at 1.23 V) under similar conditions. Furthermore, the cobalt catalyst on BiVO₄/Al₂O₃ also brought about a significant cathodic shift of approximately 400 mV in photocurrent onset potential, as compared to that of the bare BiVO₄/Al₂O₃. Sufficient experimental evidence revealed that the improved PEC performance of the BiVO₄-based photoanode by modification with MC15 was attributed to fast WOR kinetics and efficient injection of photogenerated holes into the electrode/electrolyte interface, which greatly suppressed the charge recombination at the BiVO₄ surface. The photocurrent of BiVO₄/Al₂O₃/MC15 declined less than 5% over 2 h of photoelectrolysis at an applied potential of 1.4 V.

A copper catalyst (MC16) containing a porphyrin ligand functionalized with four COOH groups was integrated with BiVO₄ to construct a hybrid photoanode for WOR.¹⁰¹ The approach to immobilization of Cu–porphyrin to BiVO₄ is different from that to attach the Co–porphyrin catalyst (MC15). The Co–porphyrin complex was prepared prior to the process of self-assembly to BiVO₄/Al₂O₃/MC15, while in the case of Cu–porphyrin, the functionalized porphyrin ligand was first immobilized to the surface of the BiVO₄ electrode, followed by a surface coordination reaction of the immobilized porphyrin with Cu²⁺ ion to afford the BiVO₄/MC16 photoanode. It seems that the latter immobilization approach is less effective than the former, due to the incomplete metalation of the grafted porphyrin ligand. A very small photocurrent (<5 μA cm⁻² at 1.23 V) was obtained for BiVO₄/MC16 in 0.1 M NaClO₄ solution at pH 5.5 under the illumination of simulated sunlight (AM 1.5, λ > 430 nm). In addition, the photocurrent gradually declined during extended illumination time, possibly due to the photocatalytic decomposition of the Cu–porphyrin complex.

3.4 Molecular iron catalyst-modified hybrid photoanodes

The first molecular iron catalyst-modified narrow-band SC hybrid photoanode was reported by Bartlett and co-workers in 2014.¹⁰² The N₄-ligand of an iron WOC was functionalized with two PO(OEt)₂ anchors (MC17, Fig. 13),¹¹⁴ and was attached to a WO₃ photoanode via phosphonate linkage. The photocurrent density of the iron catalyst-modified WO₃ electrode was increased by more than 80%, as compared to that of bare WO₃ in 0.1 M Na₂SO₄ solution at pH 1 under 100 mW cm⁻² illumination; specifically, from about 0.7 mA cm⁻² at 1.23 V for bare WO₃ to 1.3 mA cm⁻² for WO₃/MC17. The extent of photocurrent enhancement decreased with increase of the solution pH value from pH 1 to pH 7. Although modification of the WO₃ surface with MC17 apparently increased the rate of PEC water oxidation at WO₃, no appreciable cathodic shift in photocurrent onset was observed under all testing conditions. The photocurrent and the high faradaic efficiency were maintained over a period of 12 h under illumination for WO₃/MC17, notwithstanding the unclearness regarding the authentic active catalyst.

Fig. 13 Molecular structures of iron OER catalysts MC17 and MC18.

Following the work mentioned-above, Reisner and co-workers reported another molecular iron catalyst-modified WO₃ photoanode.¹⁰³ The iron complex (MC18, Fig. 13) containing a PO(OH)₂-functionalized tris(2-picolyl)amine (TPA(PO(OH)₂)) ligand was used to modify a WO₃ electrode in consideration of the activity and stability of MC18 for WOR in homogeneous systems at low pH.¹¹⁴ After modification, the transient photocurrent density was increased from about 0.57 mA cm⁻² at 1.23 V for bare WO₃, to 1.6 mA cm⁻² for the iron catalyst-modified WO₃ photoanode, but with no shift in the onset potential of photocurrent, in 0.1 M Na₂SO₄ solution at pH 3 under 200 mW cm⁻² illumination. The faradaic efficiency of WO₃/MC18 (∼40%) for WOR was doubled as compared to that of bare WO₃, although it was still relatively low.

The most important advantage of molecular iron catalyst-modified WO₃ photoanodes is that such photoanodes operate in acidic media, which is a preferred condition for the proton reduction reaction. In contrast, the Fe₂O₃- and BiVO₄-based photoanodes generally operate in basic and pH neutral solutions. The first noble-metal-free PEC cell for overall water splitting was installed by using WO₃/MC18 as a photoanode and a molecular nickel catalyst-modified TiO₂ electrode (TiO₂/MC19) as a cathode (Fig. 14).¹⁰³ Under 100 mW cm⁻² illumination at a bias of 1.1 V, this PEC cell produced an approximately 2 : 1 ratio of H₂ : O₂ in 0.1 M Na₂SO₄ solution at pH 3, with faradaic efficiencies of about 61% for O₂ evolution and 53% for H₂ evolution. After photoelectrolysis for a period of 1 h, the nickel WRC on TiO₂ retained its molecular entity and activity, while the iron WOC on WO₃ slowly desorbed from the surface of the photoanode under testing conditions, but no iron-based deposit was formed on the electrode.

Fig. 14 Schematic of the PEC water splitting cell with the WO₃/MC18 oxygen evolution photoanode wired to the TiO₂/MC19 hydrogen evolution cathode. Reproduced from ref. 103 with permission from the Royal Society of Chemistry.

4 VLASC/MC hybrid photocathodes for HER

Although the first report on the VLASC/MC photocathodes can be dated back to 1984,⁶⁶ the continuous reports have appeared since 2010. In the reported examples of VLASC/MC photocathodes, the SCMs are limited to Si and In/Ga-containing materials, and the immobilized MCs are mostly limited to cobaloximes, DuBois' nickel complexes, and bio-inspired di-iron dithiolate complexes. Until now, only carboxylate and phosphonate were used as anchors to immobilize MCs to the surface of TiO₂-protected SC photocathodes. The strategy used for attaching MCs to the unprotected Si and GaP surfaces was the self-initiated photopolymerization of functionalized-vinyl monomers at SC photoelectrodes. By this process, the surface of the electrode was coated with polymer brush containing pendent ligands that direct and assemble MCs to visible-light-absorbing substrates. To date, many reported VLASC/MC hybrid photocathodes have exhibited low photocurrent density and/or small open circuit potential for HER in aqueous electrolytes, or displayed catalytic activity in organic media containing strong organic acids. The benchmark of photocurrent densities reported so far for VLASC/MC photocathodes in aqueous solution is 11 mA cm⁻² at zero bias with photocurrent onset potential of 0.75 V for GaInP₂/TiO₂/MC27/TiO₂ under 100 mW cm⁻² illumination.¹¹⁵ The performances of reported VLASC/MC hybrid photocathodes have been summarized in Table 2, and the structures of MCs listed in Table 2 can be found in Fig. 15–21.

Table 2 Performances and testing conditions of VLASC/MC hybrid photocathodes

Photocathode	Anchor or graft method	Testing condition	Photocurrent density^a at 0 V (mA cm^−2)	Onset potential (AS)^b (V)	Stability	Ref.
a Photocurrent densities given in Table 2 are taken from linear sweep voltammograms at 0 V vs. RHE, unless otherwise noted in the brackets. b AS = anodic shift of photocurrent onset potential as compared to the corresponding bare SC photoanode. c ClMePS = poly(chloromethyl)polystyrene. d PVP = polyvinylpyridine. e PVI = polyvinylimidazole. f diMeOPh = 3,5-dimethoxyphenyl.
Si/MC20	Polymerization with ClMePS^c	Neat HBF3OH, 870 mW cm^−2	194 (−0.54 V vs. SCE)		Stable over 5 days of PEC testing	66
InP/MC21	Direct S coordination	0.1 M NaBF4, pH 7, 395 nm LED	0.0003 (−0.2 V vs. NHE)		Stable over 1 h PEC testing at −0.4 V vs. Ag/AgCl	91
GaP(100)–PVP/MC22	PVP^d	1 M PBS, pH 7, 100 mW cm^−2	2.8	0.7 (0.15)	Current declined slowly during CPE testing at 0.17 V	116
GaP(100)–PVP/MC23	PVP^d	0.1 M acetate, pH 4.5, solar simulator 100 mW cm^−2	1.1	0.6 (0.03)	Current declined slowly during PEC test at −0.39 V	117
GaP(100)–PVP/MC23	PVP^d	0.1 M PBS, pH 7, solar simulator 100 mW cm^−2	1.3	0.61	Current declined slowly during PEC test at 0 V, TOF 2.1–2.4 s^−1 from J–E plot at 0 V	118
GaP(100)–PVI/MC24	PVI^e		1.2	0.58
GaP(111)–PVI/MC24	PVI^e	0.1 M PBS, pH 7, solar simulator 100 mW cm^−2	0.89	0.64 (0.62)	Stable after initial 5 min decline in current over 1 h PEC test at 0.01 V, TOF 1.9 s^−1 from J–E plot at 0 V	119
GaP(100)/MC25	Vinylene	0.1 M PBS, pH 7, solar simulator 100 mW cm^−2	1.31	0.61 (0.04)	<10% loss of activity over 4 h of PEC test at 0 V, TOF ≥ 3.9 s^−1 from J–E plot at 0 V	120
GaP(100)/MC26			1.29	0.61 (0.04)	Loss of activity gained by MC26 modification over 4 h of PEC test at 0 V
GaInP2/TiO2/MC27/TiO2	COOH	0.1 M NaOH, pH 13, 100 mW cm^−2	11	0.75 (0.15)	After a slow decay in the first 4 h of PEC testing at 0 V, stable over 16 h at 5 mA cm^−2, TOF 3.4 s^−1 over 20 min	115
GaInP2/TiO2 + MC28	Free MC28		∼0.2	∼0.1	—
GaInP2/TiO2/Pt	ALD		10	∼0.85	Stable during 20 min of PEC testing at 0 V, TOF 2.6 s^−1 over 20 min
Si–Me/MC32	Si–C bond	0.2 M LiClO4/MeCN, 15.6 mM TFA, LED 33 mW cm^−2	6.8 (−0.7 V vs. NHE)	−0.06 V vs. NHE	TOF 285 s^−1 from J–E plot with 91 mM TFA at −0.67 V vs. NHE	121
Si–Me/TiO2/MC33	PO(OH)2	0.2 M LiClO4/MeCN, 30 mM TsOH, LED 33 mW cm^−2	4.7 (at −0.8 V vs. NHE)	−0.19 V vs. NHE	—	122
Si–Me/AZO/TiO2/MC33	PO(OH)2		7.77 (−0.75 V vs. NHE)	−0.12 V vs. NHE	—
Si–diMeOPh^f/AZO/TiO2/MC34	PO(OH)2	0.2 M LiClO4/MeCN, 60 mM TsOH, LED 33 mW cm^−2	7.14 (−0.6 V vs. NHE)	0.03 V vs. NHE	—
Si/mesoTiO2/MC36	PO(OH)2	0.1 M acetic acid buffer, pH 4.5, 100 mW cm^−2, λ > 400 nm	0.34	0.4 (0)	Current declined apparently in the first 10 h of PEC testing at 0 V	123
Si/mesoTiO2/MC37			∼0.33	0.4 (0)	Loss of activity gained by MC37 modification after 1 h of PEC testing at 0 V
Si/MC38	Drop-casting	1.0 M HClO4, 28 mW cm^−2, λ > 620 nm	8	0.15 (0.55)		75
Pillar Si/MC38	Drop-casting		9	0.15 (0.1)	Current declined slowly over 1 h PEC testing at 0 V
Pillar Si/MC38	Drop-casting	1.0 M HClO4, 86 mW cm^−2, λ > 620 nm	30	0.15 (0.1)
n^+p-Si/Ti/TiO2/MC39	PO(OH)2	1.0 M HClO4, AM 1.5, 38.8 mW cm^−2, λ > 635 nm	∼19	0.35	Stable over 1 h PEC testing at 0.2 V	124

---

Fig. 15 Molecular structures of di-iron HER catalysts MC20 and MC21.

4.1 Molecular iron catalyst-modified hybrid photocathodes

The first molecular iron catalyst-modified VLASC photocathode for proton reduction was reported by Mueller-Westerhoff as early as 1984.⁶⁶ In this work, the surface of the p-Si photocathode was coated with a co-polymer formed by the reaction of 1-methyl-[1.1]ferrocenophane (MC20, Fig. 15) with poly(chloromethyl)styrene. The modified Si electrode displayed a photocurrent density of 194 mA cm⁻² at −0.54 V vs. SCE in neat HBF₃OH electrolyte under 870 mW cm⁻² illumination. A saturation current density of 231 mA cm⁻² was attained by using this electrode, and the high activity maintained over 5 days, indicating the excellent stability of this photocathode under testing conditions. The good performance of this photocathode for proton reduction may be partly attributed to the unusual properties of the neat HBF₃OH used as the electrolyte in this test. Hydrogen evolution from neat HBF₃OH on the modified Si electrode was observed, but no gas quantification was presented.

After a long period of silence in the research on VLASC/MC photoelectrodes, Nann and Pickett, together with their co-workers, reported a Fe₂S₂ complex-modified p-type InP hybrid photocathode in 2010.⁹¹ This photocathode was assembled by first building-up a cross-linked InP nanoarray layer-by-layer onto a gold electrode with 1,4-benzenedithiol as a linker, and then incorporating a di-iron catalyst (MC21, Fig. 16) via direct binding of the sulphide bridges to indium. Under illumination with 395 nm LED at a bias potential of 0.21 V, a photocurrent density of 0.3 μA cm⁻² was attained for InP/MC21 in 0.1 M NaBF₄ electrolyte at pH 7, which was sustained for at least one hour with a faradaic efficiency of 60% for H₂ formation.

Fig. 16 Schematic of the di-iron catalyst-modified InP nanoarray on a gold electrode.

4.2 Molecular cobalt catalyst-modified hybrid photocathodes

In recent years, a series of GaP/cobaloxime hybrid photocathodes were reported by Moore and co-workers.^{116–119,125} These photocathodes were constructed by first covalently attaching the polymer brush with multiple pendent pyridyl (polyvinylpyridine, PVP) or imidazolyl (polyvinylimidazole, PVI) groups onto the surface of a p-GaP substrate by surface initiated photografting and photopolymerization. Such a polymer coat on the GaP surface provided both a protective layer and coordination sites for the assembly of MCs. The cobalt catalysts (MC22–MC24, Fig. 17) were grafted to the GaP surface either via base-promoted chloride replacement of [Co(dmgH₂)(dmgH)Cl₂] (dmgH = dimethylglyoximate) or via an aqua ligand replacement of Co(dphgBF₂)₂(H₂O)₂ (dphgBF₂ = (difuoroboryl)diphenylglyoximate) by pyridyl or imidazolyl on the attached polymer brush. The GaP–PVP/MC22 electrode displayed a 2.4 mA cm⁻² current density at 0.31 V in 1 M PBS at pH 7 under 100 mW cm⁻² illumination, while the GaP–PVP exhibited only about 0.1 mA cm⁻² current density at the same potential. In addition, the E_onset of GaP–PVP/MC22 (0.76 V) was positively shifted as compared to that of GaP–PVP (0.69 V). An increase of >200% in the fill factor for the GaP–PVP/MC22, compared to GaP–PVP, implied a dramatic expediting of charge transfer across the SC/electrolyte interface, thanks to the incorporation of the cobaloxime catalyst. The linear dependence of the photocurrent density of GaP–PVP/MC22 on the illumination intensity suggested that the photocarrier transport to the interface was still one of the limiting factors for the performance of cobaloxime-modified GaP photocathodes.^116,125

Fig. 17 Schematic of GaP–PVP/MC22, GaP–PVP/MC23, GaP–PVI/MC24, GaP/MC25, and GaInP₂/TiO₂/MC27/TiO₂ hybrid photocathodes.

The ligand of the anchored cobalt catalyst can influence the photoactivity of the modified GaP electrode. Replacement of the dmgH in the immobilized cobalt catalyst by a more electron-withdrawing ligand, dphgBF₂, improved the stability of cobaloxime complexes towards acid hydrolysis and led to a shift of the reduction potentials of the complexes to less negative, just as the ligand influence observed for the corresponding non-attached cobalt complexes.¹²⁶ The GaP–PVP/MC22 and GaP–PVP/MC23 displayed almost identical photoactivity at pH 4.5 under simulated AM 1.5 illumination, while the former electrode exhibited a higher photocatalytic activity than the latter at pH 7; more specifically, 1.2 mA cm⁻² at 0 V for the former, versus 0.56 mA cm⁻² for the latter.¹¹⁷

The functional units at the attached polymer brush used to assemble the cobalt catalyst can also affect the photocatalytic performance of the modified GaP electrode. The GaP–PVI/MC24 was fabricated by using a similar SC–polymer-catalyst approach, by replacing PVP with PVI.^118,119 A photocurrent density of 1.2 mA cm⁻² at 0 V was obtained with the GaP–PVI/MC24 as the working electrode in 0.1 M PBS at pH 7 under solar simulated illumination at 100 mW cm⁻², which was 3 and 4 times that obtained from the GaP–PVI and bare GaP electrode, respectively. Decoration of the GaP surface with the PVI brush led to a positive shift of current onset potential, from 0.02 V to 0.53 V, and the subsequent incorporation of cobalt catalyst into GaP–PVI brought about a further positive shift of E_onset to 0.58 V. Under identical conditions, the GaP–PVP/MC22 displayed a photocurrent density of 1.3 mA cm⁻² at 0 V, with E_onset at 0.61 V. In comparison, the potential required by the GaP–PVP/MC22 to achieve a 1 mA cm⁻² current density was 170 mV more positive than that required by GaP–PVI/MC24. On the basis of the estimated cobaloxime loadings, the per cobalt centre TOF values were 2.1 and 2.4 s⁻¹ for GaP–PVP/MC22 and GaP–PVI/MC24, respectively. Despite the apparently improved stability of these cobaloxime-modified GaP photocathodes, compared to the bare GaP electrode, a slow decrease in photocurrent was still observed for the GaP–polymer/cobaloxime electrodes during PEC testing, especially in the initial 5 min of illumination, possibly due in part to the loss of surface attached catalyst molecules. The GaP–PVP/cobaloxime(BF₂) electrode displayed somewhat better stability than GaP–PVP/cobaloxime in acidic electrolytes.

Two metalloporphyrin-modified GaP hybrid photocathodes, GaP/MC25 and GaP/MC26 (Fig. 17), were reported very recently, in which metalloporphyrins were tethered to the surface of GaP through a pendent vinylphenyl group at the β-position of the porphyrin ligand by a UV-induced grafting approach.¹²⁰ These photocathodes displayed identical E_onset (0.61 V) and similar photocurrents (1.29–1.31 mA cm⁻²) at 0 V in 0.1 M PBS at pH 7, under 100 mW cm⁻² illumination; however, the applied potential required to reach 1 mA cm⁻² by GaP/MC25 was 0.12 V more positive than that required by GaP/MC26. More importantly, the Co^II–porphyrin-modified GaP was much more stable than the Fe^III–porphyrin-modified GaP under test conditions. The former maintained more than 90% of its H₂ evolution activity over 4 h of PEC testing with about 97% faradaic efficiency within the first 30 min of the PEC test, while the latter almost completely lost the photocurrent gained by catalyst modification, due to the degradation of the Fe^III–porphyrin complex, with ∼45% faradaic efficiency in the initial 6 min of the PEC test. A high STH conversion efficiency of 11% was attained for GaP/MC25 in a half PEC cell, and the estimated TOF of H₂ evolution based on the cobalt porphyrin surface concentration was ≥3.9 s⁻¹, which is the highest TOF reported to date for SC/MC integrated photocathodes operating at 0 V under 1-sun illumination.

Most MC-modified SC photocathodes display photocatalytic activity for HER in organic solvents or neutral and acidic aqueous electrolytes. The p-GaInP₂ electrode modified by {(HOOC)py}Co(dmgH)₂(Cl) (MC27, (HOOC)py = isonicotinic acid, Fig. 17) is the only one that displays high photocatalytic H₂-evolution activity in basic aqueous electrolyte (0.1 M NaOH at pH 13), which is a condition favored by the photoanode counterpart for OER.¹¹⁵ In this photocathode, GaInP₂ was covered with a 35 nm TiO₂ protective layer, and the cobaloxime catalyst was grafted to the TiO₂ surface through the covalent coordination of the carboxylate anchor group. Furthermore, an additional ultrathin TiO₂ layer (∼0.4 nm) was decorated onto the surface of GaInP₂/TiO/MC27 to protect the carboxylate linkage stability in basic solutions. The GaInP₂/TiO₂/MC27/TiO₂ displayed a photocurrent density of 11 mA cm⁻² at 0 V, with E_onset at about 0.75 V under 1-sun illumination. Without the grafted cobaloxime catalyst, the photocurrent of the GaInP₂/TiO₂ rose slowly at 0 V. Noticeably, the system with GaInP₂/TiO₂, plus a free cobaloxime catalyst without anchor group, (py)Co(dmgH)₂(Cl) (MC28), performed almost identically to the GaInP₂/TiO₂ electrode alone, indicating the critical role of the covalent linkage between TiO₂ and the cobalt catalyst. More importantly, the GaInP₂/TiO₂/MC27/TiO₂ exhibited an outstanding incident photon-to-current efficiency (IPCE) of up to ∼80% across the visible range, which was about 20% higher than the GaInP₂/TiO₂/Pt electrode. The high IPCE of GaInP₂/TiO₂/MC27/TiO₂ was attributed to the application of a transparent MC layer and a top antireflective TiO₂ layer. A TOF of 3.4 s⁻¹ was calculated for GaInP₂/TiO₂/MC27/TiO₂, and 2.6 s⁻¹ for GaInP₂/TiO₂/Pt over 20 min of PEC testing at 0 V. Even though two layers of TiO₂ were adopted, a decrease in photocurrent from 10 to 5 mA cm⁻² was observed in the first 4 h of PEC testing with the GaInP₂/TiO₂/MC27/TiO₂ electrode at 0 V, while the current was maintained during a further 16 h of photoelectrolysis experiment, giving a TON of 1.39 × 10⁵. Although the GaInP₂/TiO₂/MC27/TiO₂ displayed remarkable stability compared to the bare GaInP₂ electrode, the detachment of MC from the electrode surface and the deactivation of the grafted catalyst at the GaInP₂/TiO₂/MC27/TiO₂ are still problems that need to be solved by developing more stable MCs and more robust anchor modes.

It is important to understand the influences of the molecular structure and the linker on the kinetics of charge separation and recombination in SC/MC hybrid photoelectrodes. Studies on the interfacial electron transfer dynamics of a series of TiO₂/cobaloxime photocathodes (Fig. 18) revealed that the electron transfer rate from SC to the attached cobalt catalyst was exponentially dependent on the distance between the TiO₂ surface and the cobalt center.¹²⁷ Systematic transient absorption spectroscopic studies indicated that in the presence of a hole scavenger, triethanolamine (TEOA, 0.1 M), the rate of electron transfer from the TiO₂ to the catalyst MC29 was 4 and 10 times faster than to MC30 and MC31, respectively. The results show the importance of the length of the linker and the orientation of the tethered catalyst molecules for controlling the kinetics of charge separation and recombination in SC/MC hybrid photoelectrodes.

Fig. 18 Electron transfer processes in TiO₂ functionalized with a molecular catalyst for H₂ production after UV-light excitation. Molecular structures of cobalt HER catalysts, MC29, MC30, and MC31, showing the distance between the anchoring groups and the catalyst metal center (r, Å).

4.3 Molecular nickel catalyst-modified hybrid photocathodes

DuBois-type nickel complexes bearing diphosphine ligands with pendant amine base(s) are the most active and robust molecular HER catalysts reported to date.^128,129 In very recent years, Rose and co-workers constructed two types of molecular nickel catalyst-modified p-Si photocathodes for HER by covalently tethering a PNP (PNP = Ph₂PCH₂N(Ar)CH₂PPh₂) Ni catalyst to a Si substrate with its surface passivated by methylation.¹²¹ One example was built by anchoring a Ni(PNP)₂ catalyst (MC32) to a Si photocathode via a phenyl ring (Fig. 19). The Si/MC32 was demonstrated to catalyze proton reduction to H₂ in LiClO₄/CH₃CN under the illumination of a broadband LED (33 mW cm⁻²), with E_onset at −0.06 V vs. NHE and a TOF of 285 s⁻¹, estimated from the current density at −0.67 V in the presence of 91 mM trifluoroacetic acid (TFA). Moreover, this photocathode displayed good stability during consecutive PEC-CV scans in LiClO₄/CH₃CN electrolyte. Encouragingly, Si/MC32 exhibited higher activity than the Si–Me/Pt cathode at low applied potential (<−0.3 V vs. NHE), and displayed about 0.2 V more positive onset potential than the corresponding free Ni catalyst on a CH₃-passivated Si electrode under identical conditions.

Fig. 19 Schematic of photocathodes Si–Me/MC32, Si–R/AZO/TiO₂/MC33, and Si/MC35.

Several groups have demonstrated that the decoration of a transparent thin layer of metal oxides onto the surface of p-Si can effectively improve the stability of Si electrodes, as well as beneficially modulate the band-edge position of Si electrodes for HER.^55,130–132 The other advantage of this strategy is that after decoration of the Si surface with a metal oxide layer, MCs can be tethered to the Si surface through M–O bonds instead of being limited to the Si–C linkage. The representatives of Si/metal oxide/MC hybrid photocathodes are Si–Me/TiO₂/MC33 (Fig. 19), Si–Me/AZO/TiO₂/MC33 (AZO = aluminum-doped zinc oxide), and Si–diMeOPh/AZO/TiO₂/MC33 (diMeOPh = 3,5-dimethoxyphenyl),¹²² which were constructed by first anchoring a phosphonate-functionalized PNP ligand to the metal oxide layer of a Si electrode, followed by Ni metalation plus capping of the Ni ionic center with another diphosphine ligand (diPhos). The Si–R/AZO/TiO₂/MC33 electrode displayed higher J_max values than the Si–R/AZO/TiO₂ without or with free catalyst, [Ni{(Ph₂PCH₂)₂N(C₆H₄-p-Br)}₂] (MC34), in LiClO₄/CH₃CN electrolyte under illumination. Various factors, such as the Si–R/AZO heterojunction, the Al : Zn ratio of AZO, the terminal organic group atop the Si surface, the central metal of MC, and the outer capping diphosphine ligand, could influence the performance of Si electrodes for photocatalytic HER. The introduction of a conductive AZO layer between the Si material and the TiO₂ layer had beneficial effects on the performance of the hybrid cathodes, viz. facilitating electron transfer to the electrolyte interface, shifting the E_onset of photocurrent to a more positive position, improving the activity and enhancing the saturation current (J_max), as compared to the corresponding photocathodes with a Si/TiO₂ p/n heterojunction interface. In addition, the Al : Zn ratio of the AZO layer had apparent influences on the resistance of the AZO film on a Si electrode. The terminal organic groups atop the Si surface, as a soft passivated pad within the Si/AZO or Si/TiO₂ p/n heterojunction interface, considerably influenced the morphology of the metal oxide film and the onset potential of the photocurrent. Changing the Ni²⁺ ion to Co²⁺ and replacing the capping ligand PNP to DPPE (1,2-bis(diphenylphosphino)ethane) or perfluoro-DPPE all had negative effects on the performance metrics, E_onset and J_max, of the Si–Me/AZO/TiO₂/(PNP)Ni(diPhos) photocathodes. Under optimized conditions, the Si–diMeOPh/AZO/TiO₂/MC33 electrode displayed a saturated photocurrent density of 7.14 mA cm⁻² with E_onset at 0.03 V vs. NHE in 0.2 M LiClO₄/CH₃CN containing 60 mM TsOH under the illumination of a broadband LED (33 mW cm⁻²). The stability of these Si/metal oxides/nickel catalyst hybrid photocathodes in LiClO₄/CH₃CN electrolyte and their photocatalytic performance in aqueous solutions have not been reported.

Besides the above-mentioned PNP nickel catalyst-modified photoelectrodes, a masked ester-functionalized DuBois-type P₂N₂ nickel catalyst (MC35) was attached to the amine-modified surface of p-Si and p-GaP photocathodes.^63,133 The evidence from grazing angle attenuated total reflection Fourier transform infrared spectroscopy (GATR-FTIR) and X-ray photoelectron spectroscopy (XPS) suggested that the nickel catalyst was covalently linked to the Si surface through direct Si–C bonds, while it was connected to the GaP surface through bridging oxygen atoms. Unfortunately, the MC35-modified photocathodes lacked photocatalytic activity for HER. The authors assumed that the photocatalytic inactivity of this photocathode was due in large part to the structural distortion of the nickel catalyst following surface attachment. In fact, the long linker between the Si surface and the Ni catalyst could also be a possible reason, if the performance of Si/MC35 is compared to that of the following example of a photocathode, Si/mesoTiO₂/MC36, modified with analogous nickel catalyst, but with a shorter linker.

In contrast to Si/MC35, a just published phosphonate-functionalized P₂N₂ molecular nickel catalyst (MC36)-modified p-Si/mesoporous TiO₂ (Si/mesoTiO₂) photocathode (Fig. 20) was demonstrated to be active towards HER in aqueous acetic acid solution at pH 4.5 under UV-filtered, simulated solar light illumination (100 mW cm⁻², λ > 400 nm).¹²³ Attachment of MC36 (about 38.3 nmol cm⁻²) to the Si/mesoTiO₂ surface through the phosphonate linkage led to an apparent increase in photocurrent density, from 0.1 mA cm⁻² at 0 V for Si/mesoTiO₂ to 0.34 mA cm⁻² for Si/mesoTiO₂/MC36, while this hybrid photocathode exhibited almost the same E_onset as Si/mesoTiO₂ did, at about 0.4 V. The controlled potential photoelectrolysis (CPP) experiment of Si/mesoTiO₂/MC36 at an applied potential of 0 V showed that the faradaic efficiency was decreased from 87% in the early stage of CPP to 76% after 24 h, and in the meantime, the photocurrent and the H₂ evolution rate were also apparently decreased in the first 10 h of CPP experiments, due to the initial rapid desorption and/or slow degradation of the catalyst. The TON based on MC36 was about 646 over 24 h of CPP. Post-catalysis characterization by ATR-FTIR and XP spectra suggested that MC36 on the Si/mesoTiO₂ surface maintained its molecular integrity during the CPP experiment. In addition, a specially designed phosphonate-functionalized cobaloxime catalyst (MC37)-modified Si/mesoTiO₂ photocathode displayed slightly lower photocurrent and smaller E_onset as compared to Si/mesoTiO₂/MC36 under identical test conditions. However, the cobaloxime catalyst-modified Si/mesoTiO₂ was much less stable than Si/mesoTiO₂/MC36, due to the ligand degradation of cobaloxime catalyst. Noticeably, the mesoporous TiO₂ layer played important roles in the PEC performance of these Si-based photocathodes as follows: (i) stabilizing the Si surface from passivation, (ii) enabling a high loading of MCs, (iii) affording a fast electron transfer from the light harvester to the bound MCs, and (iv) acting as a hole-blocking layer to reduce the probability of charge recombination between MC and Si. In these photocathodes, the light-generated electrons were first trapped by the CB of TiO₂, and then efficiently transferred to the bound MCs to be used for proton reduction.

Fig. 20 Schematic of photocathodes Si/mesoTiO₂/MC36 and Si/mesoTiO₂/MC37.

4.4 Molecular molybdenum catalyst-modified hybrid photocathodes

In 2011, Chorkendorff and co-workers reported the planar and pillared p-Si photocathodes with their surfaces modified by incomplete cubane-like Mo₃S₄ clusters (MC38, Fig. 21) via physisorption.⁷⁵ The photocurrent density of 8 mA cm⁻² at 0 V was attained for the MC38-modified planar Si photocathode, with a significant anodic shift of the current onset potential, from −0.4 V for bare planar Si to 0.15 V for Si/MC38 in 1.0 M HClO₄ solution under illumination of red-light (λ > 620 nm, 28 mW cm⁻²). The anodic shift of 0.55 V in the photocurrent onset potential is the largest value ever reported that is caused by the modification of SC photocathodes with MCs. The PEC activity was improved by using a Si photocathode of pillar arrays to orthogonalize the light capture and charge carrier collection, but the photocurrent density was still lower than 1 mA cm⁻² at 0 V. With immobilization of MC38 onto the Si pillar electrode, the photocurrent density was enhanced to 9 mA cm⁻² at 0 V, which slowly declined over 1 h of photoelectrolysis experiment. When the intensity of illumination was increased to 85.8 mW cm⁻², the pillar Si/MC38 electrode exhibited a high photocurrent density of 30 mA cm⁻² at 0 V.

Fig. 21 Molecular structures of molybdenum catalysts MC38 and MC39.

The stability of Mo₃S₄ cluster-modified Si photocathodes was significantly improved by adopting two measures: (i) coating the Si surface, first with a thin layer of Ti and then with a 100 nm protecting layer of TiO₂; (ii) using a phosphonate-functionalized Mo₃S₄ cluster (MC39, Fig. 21) to immobilize the MC on the outermost TiO₂ layer through covalent linkage.¹²⁴ The n⁺p-Si/Ti/TiO₂/MC39 electrode displayed a photocurrent density of 19 mA cm⁻² at 0 V, with E_onset at 0.35 V in 1.0 M HClO₄ under illumination of red-light (λ > 635 nm, 38.8 mW cm⁻²). This protected and modified Si electrode was stable at corrosive potentials, and its catalytic activity showed no decline after 1 h of photoelectrolysis at 0.2 V. However, the photocathode was not stable for photoelectrolysis at more reductive potentials, possibly due to molybdenum catalyst detachment and/or degradation. Re-deposition of the catalyst allowed complete recovery of the catalytic activity, indicative of the intact Si substrate during PEC testing. In light of the high photon-to-hydrogen conversion efficiency of these molybdenum catalyst-modified Si electrodes under red-light illumination, they are promising photocathodes for coupling with the large-bandgap photoanodes that absorb the blue-light of the solar spectrum to construct tandem PEC water splitting cells.

5 Concluding remarks and outlook

Immobilization of MCs onto VLASC electrodes is an innovative and promising approach to an efficient water splitting PEC cell. Quite a few comparative results presented in Chapters 3 and 4 have clearly demonstrated that immobilization of MCs has many advantages over free MCs in solution. Firstly, an expedited charge transfer from SC to MC can be expected in virtue of the confined short distance between MC and the surface of SC in a MC-modified SC photoelectrode, which results in the enhancement of photocurrents and the improvement of solar-to-hydrogen/solar-to-oxygen efficiency of photoelectrodes. Secondly, immobilization of MCs on the surface of SC photoelectrodes brings about, in most cases, a cathodic shift of the photocurrent onset potential (E_onset) to different extents, compared to that attained from the corresponding bare SC photoelectrode. Thirdly, it has been found that immobilization of some MCs on an electrode could make them more stable than their counterparts that remain freely in solution.³⁷ Furthermore, immobilization of MCs to form a monolayer on the surface of SC can minimize the amount of catalyst used, which is particularly meaningful for noble metal catalysts in future practical application.

Among the VLASC/MC hybrid photoanodes reported to date, only the well-engineered Ta₃N₅-based composite photoanode with two types of MCs immobilized on its surface (Ta₃N₅/TiO_x/Fh/Ni(OH)_x/MC12/MC11) has exhibited a high photocurrent density (12.1 mA cm⁻² at 1.23 V) under simulated 1-sun illumination;⁵⁴ however, its E_onset value of photocurrent and its stability still need great improvement. Notwithstanding, this example indicates that the immobilization of MCs onto a VLASC photoelectrode with well-engineered protecting layer(s) and hole-storage or electron-storage layer(s) is an effective approach to highly efficient photoanodes and photocathodes. Other VLASC/MC hybrid photoanodes based on metal oxide SCs display photocurrent densities lower than 2.2 mA cm⁻² at 1.23 V (Table 1), even in the case of immobilization of a molecular noble metal catalyst on the surface of SCMs. In addition to photocurrent, the onset potential of the photocurrent is another parameter concern for a photoelectrode. Among the available examples of photoanodes listed in Table 1, the MC-modified BiVO₄ photoanodes display the lowest onset potential of photocurrent, down to 0.3 V.⁹⁴ It is found that in most cases, modification of MCs on the surfaces of Fe₂O₃ and BiVO₄ photoanodes brings about a cathodic shift in the E_onset for photocurrent, generally by 0.2 to 0.4 V, compared to the E_onset observed from the corresponding bare SC photoanode, possibly resulting from the heterojunction formed between the MC layer and the SC surface. Several exceptions have been noticed, viz. the Fe₂O₃ NA/MC6 and all three examples of WO₃ photoanodes modified with either molecular Ru or Fe catalysts,^{95,97,102,103} each of which shows an identical E_onset value to that of the pristine SC photoelectrode; the reason for the unshifted E_onset was not discussed in the original papers. In principle, the E_onset value of the photoelectrode is predominantly determined by the band gap and the photophysical properties of the SC used. In addition, the length of the linker between the catalyst metal center and the SC surface, and the surface coverage of the catalyst on a SC photoelectrode may also influence the E_onset value of MC-modified VLASC photoelectrodes. The linker PDC in Fe₂O₃ NA/MC6 is relatively long, which is suspected of being the cause of unchanged E_onset after MC modification, while for the MC-modified WO₃ photoanodes, the unshifted E_onset might be due to the photophysical properties of WO₃ itself. Systematic studies on the influences of anchor groups and the length of linkers between MC and SC on the E_onset value are very important and can provide some guidance for designing the proper linkers and anchors for efficient VLASC/MC photoanodes.

For VLASC/MC integrated photocathodes reported to date, only p-Si and p-In- and/or Ga-containing SCMs were used as light harvesters. The narrow bang gap SCMs that feature suitable conduction band edges for electron transfer to HER MCs are very limited. The development of new SCMs that can be used as proper light harvesters for VLASC/MC photocathodes are eagerly anticipated. In addition, Cu₂O and some newly published p-type ternary SCMs deserve to be tested as light harvesters for VLASC/MC photocathodes. In the aspect of MCs, studies on VLASC/MC photocathodes have been concentrated on DuBois' nickel complexes and cobaloximes as well as their derivatives, although a number of molecular catalysts were reported to be active for HER. The reasonable grounds for the choice of the nickel catalysts are their low overpotential, high activity, and good stability for HER in aqueous acidic solutions, and the choice of cobaloxime catalysts could mostly be due to their straightforward preparation method and low overpotentials, despite the poor stability of cobaloximes under the conditions of electro- or photoelectrochemical HER.¹²³ Major challenges for adopting other earth abundant MCs to construct VLASC/MC photocathodes for HER include the arduous preparation required for catalysts containing the desired anchor groups, and the difficulty in finding a proper narrow band gap SCM that matches well with the MC in the conduction band of SC and the orbital level of MC for catalysis of HER. In other aspects, the photocathodes reported to date catalyzed HER mostly in neutral and acidic aqueous solutions, or in organic solvents containing strong acids. Considering that the OER, a bottleneck reaction for total water splitting, prefers strong basic conditions, the development of VLASC/MC photocathodes that efficiently operate in basic solutions is especially important for future integration of VLASC/MC photocathodes with photoanodes to construct effective tandem PEC cells. The GaInP₂/TiO₂/MC27/TiO₂ with two TiO₂ layers to protect SC and MC, respectively,¹¹⁵ gives the first very promising example of the construction of integrated photocathodes that are highly efficient and fairly stable for HER in strong basic electrolytes.¹¹⁵ This photocathode displayed a j_sc of 11 mA cm⁻² and a V_oc of 0.75 V with a TOF of 3.4 s⁻¹ at 0 V under simulated 1-sun illumination, which is the best performance among the reported SC/MC photocathodes operating in solutions of all pH values. Such a highly efficient photocathode is promising for integration with a photoanode to construct a PEC cell for total water splitting in strong basic solutions at zero or low bias.

The fundamental work on the development of suitable building blocks, i.e., light harvesters, molecular electrocatalysts for OER or HER, and anchor units, is imperative at the present stage. The improvement of the light absorption and charge separation of very limited n- and p-type classic semiconducting materials and the development of new semiconducting materials that feature desired band-gap and band-edge positions by material engineering are the most challenging and important work. Although some new semiconducting materials have been reported in recent years,¹³⁴ their performances in integration with MCs for OER or HER are still unexplored. The recently emerging perovskite materials are very exciting for application in solar cells. This material can in principle be considered as VLASC in PEC cells for water splitting if the problem of stability against water can be solved. Currently, the majority of OER and HER MCs were studied either in three-component photocatalytic systems with the use of sacrificial electron acceptors for OER or electron donors for HER, or in three-electrode PEC systems with MCs remaining freely in solutions. Some examples have demonstrated that the performance of a MC attached to the SC surface could be quite different from the behavior found for a free corresponding catalyst in organic solvents or in aqueous solutions.^115,133 Importantly, the performance of photoelectrodes is considerably influenced by the length and orientation of linkers. For example, the Si photocathode modified with DuBois' P₂N₂ nickel catalyst MC35 lacks photocatalytic activity for HER, while the Si photocathode modified by an analogous MC (MC36) does display activity for PEC H₂ production under visible-light illumination. The discrepancies in the performance of MC35- and MC36-modified Si electrodes with different lengths of linkers are consistent with the findings by transient absorption spectroscopic studies on the kinetics of electron transfer from SC to the tethered cobaloxime catalyst in a series of photocathodes, TiO₂/MC29, TiO₂/MC30, and TiO₂/MC31.^123,127,133 These results suggest that the proper length of linkers is one of the crucial factors that should be considered for designing new VLASC/MC photocathodes. More systematic studies are needed to reveal the inherent relationship of the length and orientation of linkers and the kinetics of interfacial charge transfer between SC and tethered MC.

One of the problems for both MC-modified SC photoanodes and photocathodes is the deactivation of electrodes with extended photoelectrolysis periods, due to the detachment and degradation of grafted MCs. Therefore, it is important to investigate whether the MCs maintain their molecular integrity after immobilization and long-term PEC testing by using all available advanced detection techniques. Many examples have demonstrated that covering the surface of unstable SC photoelectrodes, such as Ta₃N₅, Si, and In/Ga-containing SCs with broad-band metal oxides, mostly titanium oxides, is an effective method to stabilize the SC materials. With respect to MCs, studies on the immobilization of already known molecular electrocatalysts onto SC electrode surfaces are important for understanding the crucial factors that will influence the performance of the surface-attached MC and the interface electron transfer. On the basis of accumulated knowledge from these in-depth studies, novel OER and HER MCs that are more active and more durable than the reported ones can be expected with rational molecular designs. In addition to the intrinsic stabilities of MCs and SCMs, the anchor group and linker between MC and SC have considerable influence on the activity and stability of hybrid photoelectrodes for OER and HER. Hydroxamate, –CONH(OH), has been demonstrated to form a stable attachment on metal oxides, especially on TiO₂,^87,88 but it has not been used for grafting OER and HER MCs onto the VLASC photocathodes. Better stability is expected for VLASC/MC photoelectrodes with hydroxamate linkages, compared to those with commonly used carboxylate and phosphonate linkages. Besides, the development of new anchor groups with multi-bonding points and the innovation of graft methods for firmly immobilizing MCs on SC surfaces are also important work for the construction of highly efficient and robust photoelectrodes.

In brief, photocurrent, photovoltage, solar-to-hydrogen (or solar-to-oxygen) efficiency, stability, cost, and impact on environment are the major concerns for the construction of MC-modified PEC cells for overall water splitting. Although some encouraging results have been achieved in recent years in the construction of MC-modified VLASC photoanodes and photocathodes, the efficiency and durability of the reported systems are far from the requirements for a viable commercial PEC device used for solar light-induced spontaneous overall water splitting. There is still a long way to go to realize the great aspiration of chemists: using sunlight as the only energy supply to efficiently produce H₂ from water at low cost. With the joint effort of scientists engaged in the research on semiconducting nanomaterials and chemists working on the catalysis and organometallic chemistry, we can expect more innovative approaches to stable and spontaneously effective overall water splitting PEC cells that are integrated with MC-modified SC photoanodes and photocathodes.

Acknowledgements

We are grateful to the Natural Science Foundation of China (Grant No. 21373040, 21673028), the Basic Research Program of China (Grant No. 2014CB239402), the Swedish Energy Agency, and the K & A Wallenberg Foundation for financial support of this work.

References

1. N. S. Lewis and D. G. Nocera, Proc. Natl. Acad. Sci. U. S. A., 2006, 103, 15729.
2. D. G. Nocera, Acc. Chem. Res., 2012, 45, 767.
3. S. Licht, J. Phys. Chem. B, 2001, 105, 6281.
4. M. Grätzel, Nature, 2001, 414, 338.
5. J. Luo, J.-H. Im, M. T. Mayer, M. Schreier, M. K. Nazeeruddin, N.-G. Park, S. D. Tilley, H. J. Fan and M. Grätzel, Science, 2014, 345, 1593.
6. C. R. Cox, J. Z. Lee, D. G. Nocera and T. Buonassisi, Proc. Natl. Acad. Sci. U. S. A., 2014, 111, 14057.
7. M. G. Walter, E. L. Warren, J. R. McKone, S. W. Boettcher, Q. Mi, E. A. Santori and N. S. Lewis, Chem. Rev., 2010, 110, 6446.
8. Y. Tachibana, L. Vayssieres and J. R. Durrant, Nat. Photonics, 2012, 6, 511.
9. J. Yang, D. Wang, H. Han and C. Li, Acc. Chem. Res., 2013, 46, 1900.
10. T. Hisatomi, J. Kubota and K. Domen, Chem. Soc. Rev., 2014, 43, 7520.
11. J. R. McKone, N. S. Lewis and H. B. Gray, Chem. Mater., 2014, 26, 407.
12. Z. Huang, C. Xiang, H.-J. Lewerenz and N. S. Lewis, Energy Environ. Sci., 2014, 7, 1207.
13. T. J. Jacobsson, V. Fjällström, M. Edoff and T. Edvinsson, Energy Environ. Sci., 2014, 7, 2056.
14. M. R. Shaner, H. A. Atwater, N. S. Lewis and E. W. McFarland, Energy Environ. Sci., 2016, 9, 2354.
15. M. S. Prévot and K. Sivula, J. Phys. Chem. C, 2013, 117, 17879.
16. B. Seger, I. E. Castelli, P. C. K. Vesborg, K. W. Jacobsen, O. Hansen and I. Chorkendorff, Energy Environ. Sci., 2014, 7, 2397.
17. Z. Chen, T. F. Jaramillo, T. G. Deutsch, A. Kleiman-Shwarsctein, A. J. Forman, N. Gaillard, R. Garland, K. Takanabe, C. Heske, M. Sunkara, E. W. McFarland, K. Domen, E. L. Miller, J. A. Turner and H. N. Dinh, J. Mater. Res., 2010, 25, 3.
18. S. Hu, C. Xiang, S. Haussener, A. D. Berger and N. S. Lewis, Energy Environ. Sci., 2013, 6, 2984.
19. B. A. Pinaud, J. D. Benck, L. C. Seitz, A. J. Forman, Z. Chen, T. G. Deutsch, B. D. James, K. N. Baum, G. N. Baum, S. Ardo, H. Wang, E. Miller and T. F. Jaramillo, Energy Environ. Sci., 2013, 6, 1983.
20. O. K. Varghese and C. A. Grimes, Sol. Energy Mater. Sol. Cells, 2008, 92, 374.
21. H. Dotan, N. Mathews, T. Hisatomi, M. Grätzel and A. Rothschild, J. Phys. Chem. Lett., 2014, 5, 3330.
22. A. C. Nielander, M. R. Shaner, K. M. Papadantonakis, S. A. Francis and N. S. Lewis, Energy Environ. Sci., 2015, 8, 16.
23. A. Fujishima and K. Honda, Nature, 1972, 238, 37.
24. J. W. Ager, M. R. Shaner, K. A. Walczak, I. D. Sharp and S. Ardo, Energy Environ. Sci., 2015, 8, 2811.
25. J. Brillet, J.-H. Yum, M. Cornuz, T. Hisatomi, R. Solarska, J. Augustynski, M. Graetzel and K. Sivula, Nat. Photonics, 2012, 6, 824.
26. L. M. Peter and K. G. U. Wijayantha, ChemPhysChem, 2014, 15, 1983.
27. C. Ding, J. Shi, Z. Wang and C. Li, ACS Catal., 2017, 7, 675.
28. W. J. Youngblood, S.-H. A. Lee, K. Maeda and T. E. Mallouk, Acc. Chem. Res., 2009, 42, 1966.
29. J. Sun, D. K. Zhong and D. R. Gamelin, Energy Environ. Sci., 2010, 3, 1252.
30. K. S. Joya, Y. F. Joya, K. Ocakoglu and R. van de Krol, Angew. Chem., Int. Ed., 2013, 52, 10426.
31. R. Liu, Z. Zheng, J. Spurgeon and X. Yang, Energy Environ. Sci., 2014, 7, 2504.
32. L. Alibabaei, H. Luo, R. L. House, P. G. Hoertz, R. Lopez and T. J. Meyer, J. Mater. Chem. A, 2013, 1, 4133.
33. Z. Han and R. Eisenberg, Acc. Chem. Res., 2014, 47, 2537.
34. D. L. Ashford, M. K. Gish, A. K. Vannucci, M. K. Brennaman, J. L. Templeton, J. M. Papanikolas and T. J. Meyer, Chem. Rev., 2015, 115, 13006.
35. M. Wang, K. Han, S. Zhang and L. Sun, Coord. Chem. Rev., 2015, 287, 1.
36. Z. Yu, F. Li and L. Sun, Energy Environ. Sci., 2015, 8, 760.
37. N. Queyriaux, N. Kaeffer, A. Morozan, M. Chavarot-Kerlidou and V. Artero, J. Photochem. Photobiol., C, 2015, 25, 90.
38. M. Yamamoto and K. Tanaka, ChemPlusChem, 2016, 81, 1028.
39. N. Jiang, M. Sheng and Y. Sun, RSC Green Chem., 2016, 42, 108.
40. X. Zhang, T. Peng and S. Song, J. Mater. Chem. A, 2016, 4, 2365.
41. R. Liu, Y. Lin, L.-Y. Chou, S. W. Sheehan, W. He, F. Zhang, H. J. M. Hou and D. Wang, Angew. Chem., Int. Ed., 2011, 50, 499.
42. Y.-H. Lai, T. C. King, D. S. Wright and E. Reisner, Chem.–Eur. J., 2013, 19, 12943.
43. F. E. Osterloh, Chem. Soc. Rev., 2013, 42, 2294.
44. J. Li and N. Wu, Catal. Sci. Technol., 2015, 5, 1360.
45. X. Yang, R. Liu, Y. He, J. Thorne, Z. Zheng and D. Wang, Nano Res., 2015, 8, 56.
46. N. Guijarro, M. S. Prévot and K. Sivula, Phys. Chem. Chem. Phys., 2015, 17, 15655.
47. M. J. Katz, S. C. Riha, N. C. Jeong, A. B. F. Martinson, O. K. Farha and J. T. Hupp, Coord. Chem. Rev., 2012, 256, 2521.
48. L. M. Peter, J. Solid State Electrochem., 2013, 17, 315.
49. B. Iandolo, B. Wickman, I. Zorić and A. Hellman, J. Mater. Chem. A, 2015, 3, 16896.
50. K. Sivula, J. Phys. Chem. Lett., 2013, 4, 1624.
51. D. Kang, T. W. Kim, S. R. Kubota, A. C. Cardiel, H. Gil Cha and K.-S. Choi, Chem. Rev., 2015, 115, 12839.
52. D. Jing, L. Guo, L. Zhao, X. Zhang, H. Liu, M. Li, S. Shen, G. Liu, X. Hu, X. Zhang, K. Zhang, L. Ma and P. Guo, Int. J. Hydrogen Energy, 2010, 35, 7087.
53. S. Chen and L.-W. Wang, Chem. Mater., 2012, 24, 3659.
54. G. Liu, S. Ye, P. Yan, F. Xiong, P. Fu, Z. Wang, Z. Chen, J. Shi and C. Li, Energy Environ. Sci., 2016, 9, 1327.
55. D. Bae, B. Seger, P. C. K. Vesborg, O. Hansen and I. Chorkendorff, Chem. Soc. Rev., 2017, 46, 1933.
56. K. Sun, S. Shen, Y. Liang, P. E. Burrows, S. S. Mao and D. Wang, Chem. Rev., 2014, 114, 8662.
57. B. Seger, A. B. Laursen, P. C. K. Vesborg, T. Pedersen, O. Hansen, S. Dahl and I. Chorkendorff, Angew. Chem., Int. Ed., 2012, 51, 9128.
58. B. Seger, T. Pedersen, A. B. Laursen, P. C. K. Vesborg, O. Hansen and I. Chorkendorff, J. Am. Chem. Soc., 2013, 135, 1057.
59. Y. Lin, C. Battaglia, M. Boccard, M. Hettick, Z. Yu, C. Ballif, J. W. Ager and A. Javey, Nano Lett., 2013, 13, 5615.
60. C. G. Morales-Guio, K. Thorwarth, B. Niesen, L. Liardet, J. Patscheider, C. Ballif and X. Hu, J. Am. Chem. Soc., 2015, 137, 7035.
61. M. Hettick, M. Zheng, Y. Lin, C. M. Sutter-Fella, J. W. Ager and A. Javey, J. Phys. Chem. Lett., 2015, 6, 2177.
62. M. H. Lee, K. Takei, J. Zhang, R. Kapadia, M. Zheng, Y.-Z. Chen, J. Nah, T. S. Matthews, Y.-L. Chueh, J. W. Ager and A. Javey, Angew. Chem., Int. Ed., 2012, 51, 10760.
63. D. Cedeno, A. Krawicz and G. F. Moore, Interface Focus, 2015, 5, 20140085.
64. A. Bansal and J. A. Turner, J. Phys. Chem. B, 2000, 104, 6591.
65. B. A. MacLeod, K. X. Steirer, J. L. Young, U. Koldemir, A. Sellinger, J. A. Turner, T. G. Deutsch and D. C. Olson, ACS Appl. Mater. Interfaces, 2015, 7, 11346.
66. U. T. Mueller-Westerhoff and A. Nazzal, J. Am. Chem. Soc., 1984, 106, 5381.
67. A. Singh and L. Spiccia, Coord. Chem. Rev., 2013, 257, 2607.
68. M. D. Kärkäs, O. Verho, E. V. Johnston and B. Åkermark, Chem. Rev., 2014, 114, 11863.
69. S. Berardi, S. Drouet, L. Francàs, C. Gimbert-Suriñach, M. Guttentag, C. Richmond, T. Stoll and A. Llobet, Chem. Soc. Rev., 2014, 43, 7501.
70. J. D. Blakemore, R. H. Crabtree and G. W. Brudvig, Chem. Rev., 2015, 115, 12974.
71. M. R. Nellist, F. A. L. Laskowski, F. Lin, T. J. Mills and S. W. Boettcher, Acc. Chem. Res., 2016, 49, 733.
72. V. Artero and M. Fontecave, Chem. Soc. Rev., 2013, 42, 2338.
73. B. Limburg, E. Bouwman and S. Bonnet, Coord. Chem. Rev., 2012, 256, 1451.
74. C. A. Downes and S. C. Marinescu, J. Am. Chem. Soc., 2015, 137, 13740.
75. Y. Hou, B. L. Abrams, P. C. K. Vesborg, M. E. Björketun, K. Herbst, L. Bech, A. M. Setti, C. D. Damsgaard, T. Pedersen, O. Hansen, J. Rossmeisl, S. Dah, J. K. Nørskov and I. Chorkendorff, Nat. Mater., 2011, 10, 434.
76. R. Brimblecombe, G. F. Swiegers, G. C. Dismukes and L. Spiccia, Angew. Chem., 2008, 120, 7445.
77. R. Brimblecombe, G. C. Dismukes, G. F. Swiegers and L. Spiccia, Dalton Trans., 2009, 9374.
78. L. Li, L. Duan, Y. Xu, M. Gorlov, A. Hagfeldt and L. Sun, Chem. Commun., 2010, 46, 7307.
79. X. Chen, X. Ren, Z. Liu, L. Zhuang and J. Lu, Electrochem. Commun., 2013, 27, 148.
80. B. Zhang, F. Li, F. Yu, X. Wang, X. Zhou, H. Li, Y. Jiang and L. Sun, ACS Catal., 2014, 4, 804.
81. A. D. Weisz, A. E. Regazzoni and M. A. Blesa, Solid State Ionics, 2001, 143, 125.
82. C. Queffélec, M. Petit, P. Janvier, D. A. Knight and B. Bujoli, Chem. Rev., 2012, 112, 3777.
83. P. J. Hotchkiss, S. C. Jones, S. A. Paniagua, A. Sharma, B. Kippelen, N. R. Armstrong and S. R. Marder, Acc. Chem. Res., 2012, 45, 337.
84. K. Hanson, M. K. Brennaman, H. Luo, C. R. K. Glasson, J. J. Concepcion, W. Song and T. J. Meyer, ACS Appl. Mater. Interfaces, 2012, 4, 1462.
85. Y. Gao, X. Ding, J. Liu, L. Wang, Z. Lu, L. Li and L. Sun, J. Am. Chem. Soc., 2013, 135, 4219.
86. Z. Xu and M. Cotlet, Angew. Chem., Int. Ed., 2011, 50, 6079.
87. B. J. Brennan, J. Chen, B. Rudshteyn, S. Chaudhuri, B. Q. Mercado, V. S. Batista, R. H. Crabtree and G. W. Brudvig, Chem. Commun., 2016, 52, 2972.
88. W. R. McNamara, R. C. Snoeberger III, G. Li, C. Richter, L. J. Allen, R. L. Milot, C. A. Schmuttenmaer, R. H. Crabtree, G. W. Brudvig and V. S. Batista, Energy Environ. Sci., 2009, 2, 1173.
89. K. Takijiri, K. Morita, T. Nakazono, K. Sakai and H. Ozawa, Chem. Commun., 2017, 53, 3042.
90. J. M. Buriak, Chem. Rev., 2002, 102, 1271.
91. T. Nann, S. K. Ibrahim, P.-M. Woi, S. Xu, J. Ziegler and C. J. Pickett, Angew. Chem., Int. Ed., 2010, 49, 1574.
92. K. Fan, F. Li, L. Wang, Q. Daniel, E. Gabrielsson and L. Sun, Phys. Chem. Chem. Phys., 2014, 16, 25234.
93. F. Li, K. Fan, B. Xu, E. Gabrielsson, Q. Daniel, L. Li and L. Sun, J. Am. Chem. Soc., 2015, 137, 9153.
94. B. Liu, J. Li, H.-L. Wu, W.-Q. Liu, X. Jiang, Z.-J. Li, B. Chen, C.-H. Tung and L.-Z. Wu, ACS Appl. Mater. Interfaces, 2016, 8, 18577.
95. D. K. Zhong, S. Zhao, D. E. Polyansky and E. Fujita, J. Catal., 2013, 307, 140.
96. M. de Respinis, K. S. Joya, H. J. M. de Groot, F. D'Souza, W. Smith, R. van de Krol and B. Dam, J. Phys. Chem. C, 2015, 119, 7275.
97. K. Fan, F. Li, L. Wang, Q. Daniel, H. Chen, E. Gabrielsson, J. Sun and L. Sun, ChemSusChem, 2015, 8, 3242.
98. W. Li, S. W. Sheehan, D. He, Y. He, X. Yao, R. L. Grimm, G. W. Brudvig and D. Wang, Angew. Chem., Int. Ed., 2015, 54, 11428.
99. J. W. Moir, E. V. Sackville, U. Hintermair and G. A. Ozin, J. Phys. Chem. C, 2016, 120, 12999.
100. K. S. Joya, N. Morlanés, E. Maloney, V. Rodionov and K. Takanabe, Chem. Commun., 2015, 51, 13481.
101. S. Nakashima, R. Negishi and H. Tada, Chem. Commun., 2016, 52, 3665.
102. B. M. Klepser and B. M. Bartlett, J. Am. Chem. Soc., 2014, 136, 1694.
103. T. E. Rosser, M. A. Gross, Y.-H. Lai and E. Reisner, Chem. Sci., 2016, 7, 4024.
104. J. R. C. Lattimer, J. D. Blakemore, W. Sattler, S. Gul, R. Chatterjee, V. K. Yachandra, J. Yano, B. S. Brunschwig, N. S. Lewis and H. B. Gray, Dalton Trans., 2014, 43, 15004.
105. J. M. Thomsen, D. L. Huang, R. H. Crabtree and G. W. Brudvig, Dalton Trans., 2015, 44, 12452.
106. S. W. Sheehan, J. M. Thomsen, U. Hintermair, R. H. Crabtree, G. W. Brudvig and C. A. Schmuttenmaer, Nat. Commun., 2015, 6, 6469.
107. L. Badia-Bou, E. Mas-Marza, P. Rodenas, E. M. Barea, F. Fabregat-Santiago, S. Gimenez, E. Peris and J. Bisquert, J. Phys. Chem. C, 2013, 117, 3826.
108. W. Li, D. He, S. W. Sheehan, Y. He, J. E. Thorne, X. Yao, G. W. Brudvig and D. Wang, Energy Environ. Sci., 2016, 9, 1794.
109. N. S. McCool, D. M. Robinson, J. E. Sheats and G. C. Dismukes, J. Am. Chem. Soc., 2011, 133, 11446.
110. G. L. Ganga, F. Puntoriero, S. Campagna, I. Bazzan, S. Berardi, M. Bonchio, A. Sartorel, M. Natali and F. Scandola, Faraday Discuss., 2012, 155, 177.
111. N. Morlanés, K. S. Joya, K. Takanabe and V. Rodionov, Eur. J. Inorg. Chem., 2015, 49.
112. D. Wang and J. T. Groves, Proc. Natl. Acad. Sci. U. S. A., 2013, 110, 15579.
113. A. Han, H. Jia, H. Ma, S. Ye, H. Wu, H. Lei, Y. Han, R. Cao and P. Du, Phys. Chem. Chem. Phys., 2014, 16, 11224.
114. J. L. Fillol, Z. Codolà, I. Garcia-Bosch, L. Gàmez, J. J. Pla and M. Costas, Nat. Chem., 2011, 3, 807.
115. J. Gu, Y. Yan, J. L. Young, K. X. Steirer, N. R. Neale and J. A. Turner, Nat. Mater., 2016, 15, 456.
116. A. Krawicz, J. Yang, E. Anzenberg, J. Yano, I. D. Sharp and G. F. Moore, J. Am. Chem. Soc., 2013, 135, 11861.
117. D. Cedeno, A. Krawicz, P. Doak, M. Yu, J. B. Neaton and G. F. Moore, J. Phys. Chem. Lett., 2014, 5, 3222.
118. A. M. Beiler, D. Khusnutdinova, S. I. Jacob and G. F. Moore, Ind. Eng. Chem. Res., 2016, 55, 5306.
119. A. M. Beiler, D. Khusnutdinova, S. I. Jacob and G. F. Moore, ACS Appl. Mater. Interfaces, 2016, 8, 10038.
120. D. Khusnutdinova, A. M. Beiler, B. L. Wadsworth, S. I. Jacob and G. F. Moore, Chem. Sci., 2017, 8, 253.
121. J. Seo, R. T. Pekarek and M. J. Rose, Chem. Commun., 2015, 51, 13264.
122. H. J. Kim, J. Seo and M. J. Rose, ACS Appl. Mater. Interfaces, 2016, 8, 1061.
123. J. J. Leung, J. Warnan, D. H. Nam, J. Z. Zhang, J. Willkomm and E. Reisner, Chem. Sci., 2017 10.1039/c7sc01277b.
124. B. Seger, K. Herbst, T. Pedersen, B. Abrams, P. C. K. Vesborg, O. Hansen and I. Chorkendorff, J. Electrochem. Soc., 2014, 161, H722.
125. A. Krawicz, D. Cedeno and G. F. Moore, Phys. Chem. Chem. Phys., 2014, 16, 15818.
126. P. Connolly and J. H. Espenson, Inorg. Chem., 1986, 25, 2684.
127. A. Reynal, J. Willkomm, N. M. Muresan, F. Lakadamyali, M. Planells, E. Reisner and J. R. Durrant, Chem. Commun., 2014, 50, 12768.
128. A. L. Goff, V. Artero, B. Jousselme, P. D. Tran, N. Guillet, R. Métayé, A. Fihri, S. Palacin and M. Fontecave, Science, 2009, 326, 1384.
129. M. L. Helm, M. P. Stewart, R. M. Bullock, M. R. DuBois and D. L. DuBois, Science, 2011, 333, 863.
130. B. Seger, T. Pedersen, A. B. Laursen, P. C. K. Vesborg, O. Hansen and I. Chorkendorff, J. Am. Chem. Soc., 2013, 135, 1057.
131. Y. Lin, C. Battaglia, M. Boccard, M. Hettick, Z. Yu, C. Ballif, J. W. Ager and A. Javey, Nano Lett., 2013, 13, 5615.
132. C. G. Morales-Guio, K. Thorwarth, B. Niesen, L. Liardet, J. Patscheider, C. Ballif and X. Hu, J. Am. Chem. Soc., 2015, 137, 7035.
133. G. F. Moore and I. D. Sharp, J. Phys. Chem. Lett., 2013, 4, 568.
134. K. Sivula and R. van de Krol, Nat. Rev. Mater., 2016, 1, 15010.

---