Multifunctional Coatings from Scalable Single Source Precursor Chemistry in Tandem Photoelectrochemical Water Splitting

The straightforward and inexpensive fabrication of stabilized and activated photoelectrodes for application to tandem photoelectrochemical (PEC) water splitting is reported. Semiconductors such as Si, WO3, and BiVO4 can be coated with a composite layer formed upon hydrolytic decomposition of heterobimetallic single source precursors (SSPs) based on Ti and Ni, or Ti and Co in a simple single‐step process under ambient conditions. The resulting 3d‐transition metal oxide composite films are multifunctional, as they protect the semiconductor electrode from corrosion with an amorphous TiO2 coating and act as bifunctional electrocatalysts for H2 and O2 evolution based on catalytic Ni or Co species. Thus, this approach enables the use of the same precursors for both photoelectrodes in tandem PEC water splitting, and SSP chemistry is thereby established as a highly versatile low‐cost approach to protect and activate photoelectrodes. In an optimized system, SSP coating of a Si photocathode and a BiVO4 photoanode resulted in a benchmark noble metal‐free dual‐photoelectrode tandem PEC cell for overall solar water splitting with an applied bias solar‐to‐hydrogen conversion efficiency of 0.59% and a half‐life photostability of 5 h.


Introduction
Solar-driven water splitting is an attractive technological concept for the generation of sustainable H 2 fuel from sunlight and water. [ 1 ] A photoelectrochemical (PEC) cell is a promising device for fulfi lling this purpose because its capital cost may potentially be lower than that of an electrolyzer wired to photovoltaic (PV) modules, and it has an attractive theoretical effi ciency. [ 2 ] The upper limit effi ciency of a tandem PEC cell with a pair of semiconductors having band gaps of 1.0 and 1.6 eV is close to 30%, [ 3 ] whereas only 13% is achievable for a single light absorber PEC cell consisting of a semiconductor with a band gap of 2.2 eV. [ 2c ] FULL PAPER Herein, we report on a highly versatile and scalable SSP approach for preparing a composite fi lm on photoelectrodes. This composite is multifunctional, serving three main purposes: it protects the photoelectrode from corrosion and can act as both a hydrogen evolution catalyst (HEC) and an oxygen evolution catalyst (OEC) in the same neutral-alkaline solution. Thus, individual photoelectrodes coated with the SSPs can also be arbitrarily combined in tandem PEC cells. The SSPs employed in this study are [Ti 2 (OEt) 9 (NiCl)] 2 (TiNi SSP ) and [Ti 4 O(OEt) 15 (CoCl)] (TiCo SSP ) ( Figure 2 ). [ 12 ] TiNi SSP and TiCo SSP were selected fi rst because their hydrolysis forms amorphous TiO 2 , which is inexpensive and the most widely used protective coating on photoelectrodes that suffer from severe instability. [ 7d , 9 ] Furthermore, decomposition of the SSPs will form Ni [ 7d , 10a,b , 13 ] and Co [ 10d , 14 ] species, which are among the best noble metal-free HECs and OECs and can show bifunctionality for water splitting catalysis. [ 10 ] In addition, TiNi SSP and TiCo SSP can be easily synthesized by a single mid-temperature solvothermal step, with respectable yield. [ 12 ] We demonstrate that the SSPs can be deposited onto Si, WO 3 , and BiVO 4 , where they are activated in situ to protect and catalytically activate the semiconductors for both half-reactions in PEC water splitting. Optimized SSP-modifi ed photoelectrodes allowed for the assembly of a benchmark tandem water splitting cell.

Electrochemical Characterization
The simple coating of the SSPs (TiNi SSP and TiCo SSP ) onto conducting substrates through room temperature deposition allows for the formation of bifunctional composite materials active as HEC and OEC in a pH 9.2 potassium borate (B i ) solution ( Figure 3 ). Hydrolytic decomposition of TiNi SSP (2 × 20 µL, 5 × 10 −3 M in dry toluene) on a fl uoride-doped tin oxide (FTO)-coated glass substrate (1 cm 2 ) has been previously shown to result in an amorphous Ti-and Ni-containing precursor fi lm (FTO|TiNi pre ). [ 11a,c ] Under anodic conditions, a NiO x OEC embedded in a TiO 2 matrix was formed in situ (TiNi OEC ), [ 11c ] whereas an in situ cathodic activation process of the TiNi pre fi lm gave a Ni-based HEC that consists of metallic Ni embedded in an amorphous NiO/Ni(OH) 2 and TiO 2 matrix (TiNi HEC , Figure 3 a). [ 11a ] FTO|TiNi OEC and FTO|TiNi HEC show a catalytic onset potential ( E cat ) of ≈1.7 and −0.1 V versus the reversible hydrogen electrode (vs RHE) for O 2 and H 2 evolution, respectively, and Faradaic effi ciencies (FEs) of more than 90% for both processes ( Table 1 ).
Drop-casting of TiCo SSP (2 × 20 µL, 10 × 10 −3 M in dry toluene) on an FTO-coated glass substrate (1 cm 2 ) resulted in an amorphous precursor fi lm on FTO (FTO|TiCo pre ). TiCo pre is a mixture of agglomerated amorphous particles of TiO 2 and CoO/ Co(OH) 2 , which was confi rmed by scanning electron microscopy (SEM), powder X-ray diffraction (p-XRD), and X-ray photoelectron spectroscopy (XPS; Figures S1 and S2, Supporting Information). Under an anodic potential, TiCo pre converts in situ into TiCo OEC , which contains the well-known CoO x OEC in B i solution (Figure 3 a and Figure S3, Supporting Information). [ 11b, 14d ] FTO|TiCo OEC electro-oxidizes water to O 2 with an onset potential of approximately E cat = 1.6 V versus RHE (Figure 3 b) and a FE of 88% at an applied potential ( E appl ) of 2.0 V versus RHE (Table 1 and Figure S3, Supporting Information). TiCo HEC forms upon a cathodic activation of TiCo pre at E appl of −0.6 V versus RHE for Figure 1. Schematic representation of a tandem PEC cell for solar water splitting consisting of a photocathode coated with a protection layer and a hydrogen evolution catalyst (HEC) and a photoanode integrated with a protection layer and an oxygen evolution catalyst (OEC). A membrane or separator is used for the separation of the gaseous products.  a) Photoelectrodes were irradiated with a solar light simulator (100 mW cm −2 , AM 1.5G). b) N/A indicates that the amount of produced gas was below the limit of quantifi cation.
10 min, also demonstrating the bifunctionality for TiCo. The current density increased from −1.5 to −3.0 mA cm −2 with the formation of H 2 bubbles (confi rmed by gas chromatography) during this pretreatment ( Figure S4, Supporting Information). The active species of TiCo HEC is presumably similar to a previously reported Co-based HEC, metallic Co with a small portion of CoO/ Co(OH) 2 , [ 10d ] that was prepared by electrodeposition of a Co(II) salt. FTO|TiCo HEC electroreduces protons with an onset potential of approximately E cat = −0.2 V versus RHE and a FE of 92% was observed at E appl = −0.6 V versus RHE (Figure 3 b and Table 1 ). The current-voltage characteristics and the near-quantitative FE confi rm that both TiNi SSP and TiCo SSP act as precursors of HECs and OECs and are therefore rare examples of bifunctional water splitting electrocatalysts. [ 10a-d ] Although the catalytic onset overpotential of these electrocatalysts is somewhat higher than that of other benchmark electrocatalysts, the photocurrent of a photoelectrode during irradiation is not necessarily limited by the nonideal response of the composite electrocatalysts. Figure 3 b indicates the relevant minority carrier band positions of the semiconductors used in this study, illustrating that a catalyst with a very small overpotential requirement is not necessary for the HEC and OEC to function effi ciently in such a PEC system. With respect to the standard potential ( E 0 ) of the appropriate half-reaction, an overpotential ( η ) of 0.6 V is available for H 2 evolution on p-Si and an η of 1.8 and 1.3 V for O 2 generation on WO 3 and BiVO 4 , respectively. Furthermore, BiVO 4 and WO 3 can only provide theoretical maximum photocurrent densities of 7 and 5 mA cm −2 , respectively, suggesting that the photocurrent will be limited by light absorption and charge separation of the photoanode rather than electrocatalysis by TiNi OEC and TiCo OEC .

SSP-Coated Photoelectrodes and Performance
Bifunctionality in water splitting catalysis has only been demonstrated in water electrolysis to date. [ 10 ] Here, we explore the utility of composite fi lms from TiNi SSP and TiCo SSP to form protective and bifunctional catalyst layers on state-of-the-art semiconductors in PEC water splitting. While the electrocatalytic showing the anodic response of FTO|TiNi OEC and FTO|TiCo OEC for water oxidation as well as the cathodic response of FTO|TiNi HEC and FTO|TiCo HEC for proton reduction. Bare FTO is also shown. All CV scans were performed at room temperature in an aqueous electrolyte solution (0.1 M B i , 0.1 M K 2 SO 4 , pH 9.2) at a scan rate of 50 mV s −1 . FTO|TiNi OEC and FTO|TiCo OEC exhibit an oxidation wave of Ni III /Ni II and Co III /Co II at approximately E p = 1.58 and 1.41 V versus RHE, respectively. [ 13c , 14d ] iR drop is not compensated.
activities of the TiNi and TiCo composite fi lms have already been evaluated by electrochemical methods (Figure 3 b and  Table 1 ), the coating's success on a semiconductor also strongly depends on the formation of a good interface between the composite and the photoelectrode. [ 15 ] A key advantage of our solution-based SSP approach is its simple application to a wide range of substrates by approaches like spin-coating, dropcasting, dip-coating, or inkjet spraying, followed by hydrolytic decomposition under ambient conditions to form a well-interfaced and multifunctional composite layer.
p-Si has been selected as photocathode due to its near-ideal small band gap of 1.1 eV, providing a theoretical photocurrent density of 44 mA cm −2 . [ 3a , 16 ] WO 3 and BiVO 4 are chosen as state-of-the-art photoanodes having band gaps of 2.6 and 2.4 eV, respectively, thus providing respectable photocurrents under solar light irradiation. Additionally, thin fi lms of these n-type semiconductors are easily prepared without requiring arduous procedures (i.e., no high-temperature annealing above 600 °C), and exhibit promising water oxidation activity. [ 7a , 17 ] For use in PEC hydrogen evolution, the surface of p-Si must be coated with a protective layer and a HEC in order to prevent quenching of photoactivity due to the rapid formation of SiO 2 , and to overcome the kinetic barriers for proton reduction. [ 7b,c ] We have previously demonstrated that TiNi SSP acts as an SSP to form a protective TiO 2 layer and a Ni-based HEC on p-Si. [ 11a ] A p-Si|TiNi HEC electrode was prepared by drop-casting a TiNi SSP solution (8 × 30 µL cm −2 , 2.5 × 10 −3 M in dry toluene) onto a planar p-Si electrode (0.5 cm 2 ) followed by cathodic in situ activation at E appl = 0 V versus RHE under solar light irradiation in a pH 9.2 B i solution ( Figure S5, Supporting Information). [ 11a ] The p-Si|TiNi HEC displayed a promising photoresponse, with a photocatalytic onset potential ( E cat ) of 0.3 V versus RHE, close to its valence band edge of 0.5 V versus RHE. A photocurrent density of approximately j = −5.0 mA cm −2 was achieved at 0 V versus RHE under solar light irradiation (100 mW cm −2 , AM 1.5G) with a quantitative FE ( Figure 4 a, Table 1 ). [ 11a ] p-Si|TiCo HEC that was formed similarly by drop-casting TiCo pre (8 × 30 µL cm −2 , 5 × 10 −3 M in dry toluene) onto p-Si and then activated with the same cathodic in situ activation process showed an E cat of only 0.15 V with j = −3.5 mA cm −2 at 0 V versus RHE (Figures S5 and S6, Supporting Information). Thus, TiCo HEC showed inferior performance compared to TiNi HEC on p-Si and the composite electrode consisting of the latter was employed in PEC water splitting (see below). A bare p-Si electrode does not exhibit a meaningful photocurrent at potentials more positive than −0.2 V versus RHE and displays a photocurrent density less than j = −10 µA cm −2 at 0 V versus RHE (Figures S5 and S6, Supporting Information). Additionally, in the absence of TiNi HEC coating, p-Si lost its activity within minutes. In contrast, TiNi HEC stabilizes p-Si for 4 h with a half-life time of 12 h at 0 V versus RHE. [ 11a ] The physical instability between p-Si and TiNi HEC contributes to longer-term instability, as the catalyst layer visibly detached when vigorous H 2 bubbling was observed.
To analyze the performance of p-Si|TiNi HEC in the bottomabsorber position in a tandem cell, we also recorded the photocurrent with simulated solar light fi ltered by a TiNi-coated nanostructured WO 3 electrode (nanoWO 3 |TiNi OEC ) and a TiCocoated nanostructured BiVO 4 electrode (nanoBiVO 4 |TiCo OEC ; see below for more details about the photoanodes). In such a confi guration, p-Si|TiNi HEC displays an onset potential of 0.25 V versus RHE and photocurrent densities of approximately j = −3.4 and −3.0 mA cm −2 at 0 V versus RHE with the light fi ltered by nanoWO 3 |TiNi OEC and nanoBiVO 4 |TiCo OEC , respectively (Figure 4 a and Figure S7, Supporting Information).
WO 3 is an inexpensive semiconductor material that has a suitable valence band edge ( E VB = 3.0 V vs RHE) to provide enough driving force to photo-oxidize water. [ 6a , 11c , 18 ] Nanosheet-structured and monoclinic WO 3 (nanoWO 3 ) was prepared by a previously reported hydrothermal method [ 6a , 11c ] and characterized by SEM and p-XRD ( Figure 5 a,b and Figure S8, Supporting Information).
A key drawback of WO 3 is its chemical instability at pH > 4. [ 19 ] However, spin-coating of TiNi SSP (4 × 60 µL cm −2 , 5 × 10 −3 M in dry toluene) on nanoWO 3 forms nanoWO 3 |TiNi pre with a uniform Tiand Ni-containing fi lm exhibiting a nanostructured morphology (Figure 5 c), which allows it to be employed in neutral-alkaline solution. The sheet thickness was increased from ≈30 to 300 nm after depositing the Ti-and Ni-containing fi lm. TiNi pre is converted in situ into TiNi OEC , which contains a TiO 2 protection layer and an NiO x OEC on WO 3 . [ 11c ] Illuminated nanoWO 3 |TiNi OEC shows an E cat of ≈0.6 V, a saturation photocurrent density of 0.6 mA cm −2 at 1.23 V versus RHE and a FE of 74% at E appl = 1.23 V versus RHE in aqueous pH 9.2 B i solution (Table 1 , Figure 4
In the absence of TiNi coating, nanoWO 3 lost 50% of its initial photocurrent within 1 h in pH 9.2 B i solution. In contrast, nanoWO 3 |TiNi OEC shows a half-life time of 4 h under the same conditions, demonstrating the composite fi lm's role as protection layer. [ 11c ] In addition, the TiNi coating enhances the photocurrents of nanoWO 3 in the low bias region (<1.15 V vs RHE, Figure S9, Supporting Information). Thus, nanoWO 3 |TiNi OEC serves as a suitable photoanode to pair with p-Si|TiNi HEC in pH 9.2 solution in tandem PEC studies. We note that the TiNi OEC coating likely does not fully prevent WO 3 from direct contact with alkaline solution, contributing to the deactivation of nanoWO 3 |TiNi OEC on longer time scales. We also point out that the lifetime of our protected photoelectrodes is less than that of the top-performing TiO 2 passivated electrodes prepared by ALD technique, [ 7d ] which yields denser and more conformal protective coatings compared to the sol-gel processed SSP chemistry employed here. With further studies, including tuning the molecular ligand and applying a self-assembly strategy to immobilize a monolayer of the SSP onto the electrode before decomposition, a more conformal coating may be achieved and provide better surface protection functions.
BiVO 4 has a smaller band gap and thus a higher theoretical photocurrent density than WO 3 . [ 2c , 7a , 20 ] BiVO 4 also has a more negative conduction band potential ( E CB = +0.1 V vs RHE) than WO 3 ( E CB = +0.4 V vs RHE), which should provide a higher photovoltage and operating photocurrent when paired with p-Si in a tandem PEC cell. Key drawbacks of BiVO 4 are poor carrier mobility and slow water oxidation kinetics, though near-complete suppression of surface recombination by electrodeposited amorphous CoO x on BiVO 4 has been recently demonstrated. [ 4d ] Thus, we were particularly interested in the effect of TiCo OEC on BiVO 4 .
Monoclinic scheelite BiVO 4 was synthesized by a combined electrochemical deposition/metal-organic decomposition synthesis as reported [ 7a ] and characterized by SEM and p-XRD nanoporous structure (nanoBiVO 4 ) prepared in this manner has been shown to improve performance, purportedly by reducing the required hole diffusion length and enhancing the catalytic surface area. NanoBiVO 4 |TiCo OEC was prepared by spin-coating a TiCo SSP solution (4 × 20 µL cm −2 , 5 × 10 −3 M in dry toluene) onto a nanoBiVO 4 electrode. No drastic surface morphology change was observed, but the nanoBiVO 4 appeared to be decorated with an agglomerated Ti-and Co-containing fi lm (Figure 5 f). NanoBiVO 4 |TiCo OEC exhibits an E cat = 0.2 V versus RHE and j = 1.8 mA cm −2 at E appl = 1.23 V versus RHE for water oxidation in pH 9.2 B i solution during irradiation (Figure 4 b). Modifi cation of nanoBiVO 4 photoanode surfaces with TiCo OEC yielded a signifi cant cathodic shift in E cat and a substantial enhancement of the photocurrent compared to bare nanoBiVO 4 (Figure 4 b and Figure S11, Supporting Information). Comparing this composite photoelectrode performance to that of bare nanoBiVO 4 in PEC experiments with Na 2 SO 3 as a hole scavenger reveals that interfacing TiCo OEC with nano-BiVO 4 almost completely eliminates losses due to surface electron-hole recombination. A FE of 78% was observed at E appl = 1.23 V versus RHE for nanoBiVO 4 |TiCo OEC (Table 1 and Figure S12, Supporting Information). A TiNi OEC catalytic fi lm also helps suppress the surface charge recombination of BiVO 4 , [ 6d , 7a , 21 ] but is less effective than TiCo OEC (Figure 4 b).
The catalytic performance of the TiCo OEC composite layer on nanoBiVO 4 appears to compare favorably to that of a recently reported dual-layer FeOOH|NiOOH OEC cocatalyst, in which surface electron-hole recombination is not completely suppressed. [ 7a ] NanoBiVO 4 |TiCo OEC also functions well in a neutral P i solution (0.5 M , pH 7) and shows similar performance compared to nanoBiVO 4 |CoO x -P i , which was prepared by photo-assisted deposition of CoO x on a nanoBiVO 4 electrode from a P i buffer solution containing 0.5 × 10 −3 M Co(NO 3 ) 2 ( Figure S13, Supporting Information). [ 4d ] However, compared to this (photo)electrodeposition method, the SSP approach reported here offers a better metal-atom effi ciency to produce the layer and a higher potential for large-scale production of CoO x OECs by low-temperature inkjet spraying or roll-to-roll processing (i.e., for conductive polymer substrates). The dissolution of a thin CoO x layer in the electrolyte solution during PEC measurements has been recently demonstrated [ 22 ] and might be the reason that TiCo OEC shows negligible effect on the stability of nanoBiVO 4 .

Tandem PEC Cells for Overall Solar Water Splitting
Developing a PEC water splitting device that operates in a near pH-neutral environment is desirable in order to extend the range of potential light absorber and catalyst pairs to those that are not stable under strongly acidic and alkaline conditions. [ 14a ] Additionally, operating at moderate pH allows for the use of natural water resources (including sea water) [ 23 ] and avoids the handling of corrosive solutions. The mass transport limitations imposed by the lack of H + and OH − ions in these conditions can be overcome by adding supporting electrolyte and by employing circulating electrolyte systems for the forced convection of ionic species. [ 24 ] With these considerations in mind, our composite photoelectrode arrays were employed in neutralalkaline conditions, minimizing the mass-transport limitations without compromising much photoelectrode performance.
The tandem PEC cells were subsequently assembled by pairing p-Si|TiNi HEC with nanoWO 3 |TiNi OEC (PEC cell I) and with nanoBiVO 4 |TiCo OEC (PEC cell II) for overall solar-driven (100 mW cm −2 , AM 1.5G) water splitting at room temperature ( Figure 6 ). A two-compartment cell separated by a Nafi on 117 membrane was used with photoelectrodes having geometric surface areas of ≈0.5 cm 2 , in order to minimize efficiency losses due to the solution or material resistance. Since the photoanodes both have larger band gaps than p-Si, light was fi rst absorbed by these photoanodes (nanoWO 3 |TiNi OEC or nanoBiVO 4 |TiCo OEC ) and the attenuated light then arrived at the back photocathode (p-Si|TiNi HEC ). Both cells were operated with the photocatalytic surfaces of the two electrodes facing one another, thus optimizing photoanode performance and minimizing ion transport resistances. It is worth noting that we have employed Nafi on 117 in this study in order to prevent the crossover of the product gases, even though Nafi on 117 is a proton conducting membrane that does not function ideally in neutral-alkaline conditions. Thus, the device effi ciency might Adv. Energy Mater. 2015, 5, 1501668 www.MaterialsViews.com www.advenergymat.de be improved if an appropriate alkali anion exchange membrane or glass frit separator is used. Figure 7 shows the photocurrent density of tandem PEC cell I and II at applied biases from 0 to 1.23 V in an aqueous B i solution (0.1 M , pH 9.2) with K 2 SO 4 (0.1 M ) as supporting electrolyte. An external bias of at least 0.35 V is necessary for tandem PEC cell I to split water, which is consistent with the half-cell performance of p-Si|TiNi HEC placed in tandem cell position and that of nanoWO 3 |TiNi OEC : an E cat of ≈0.25 V versus RHE is required for p-Si|TiNi HEC to photoreduce protons, whereas an E cat of 0.6 V versus RHE is needed for nanoWO 3 |TiNi OEC to photooxidize water (Figure 4 a). In PEC cell I, a photocurrent density of ≈400 µA cm −2 is achievable at an applied bias of 0.8 V, close to the expected required bias of 0.75 V predicted by the half-cell performance of the photoelectrodes (Figure S14, Supporting Information); additional required bias to produce this photocurrent density can be accounted for by the increased ohmic loss incurred in moving to a working two-electrode device.
In the case of PEC cell II, a spontaneous, unbiased photocurrent density of 45 ± 18 µA cm −2 was observed (Figure 7 a) as reasonably predicted from the half-cell performances of p-Si|TiNi HEC and nanoBiVO 4 |TiCo OEC (Figure 4 a). A photocurrent density of ≈1.0 mA cm −2 is achievable at an applied bias of 0.6 V in tandem PEC cell II, which is also in good agreement with the expected required bias predicted by the half-cell performance of p-Si|TiNi HEC and nanoBiVO 4 |TiCo OEC , taking into account some losses in polarization due to solution resistance ( Figure S15, Supporting Information).
A fi rst indication of a PEC cell's performance can be determined by calculating the applied bias photon-to-current conversion effi ciency (ABPE, Equation ( 1)  In this equation, |j | is the photocurrent density (mA cm −2 ), V bias is the applied bias (V) to the tandem PEC cell, and P total is the energy fl ux of the illumination (mW cm −2 ). A maximum ABPE of 0.19 ± 0.01% and 0.65 ± 0.04% was achieved at an applied bias of 0.8 V for PEC cell I and 0.6 V for PEC cell II, respectively (Figure 7 b and Table 2 ).
The stability and performance of the cells were subsequently studied at the applied bias where each tandem PEC cell exhibits the highest ABPE. Both tandem PEC cells exhibit respectable lifetime under continuous solar light irradiation, with a half-life time of 2 and 5 h for PEC cell I and II, respectively; tandem PEC cell II also retains 30% of its initial photocurrent after 24 h (Figure 7 a, inset). To confi rm that the photocurrent of the tandem PEC cell was due to water splitting, the amount of H 2 and O 2 produced was quantifi ed in the reactor's gaseous headspace by gas chromatography and a fl uorescence oxygen probe, respectively ( Figure S16, Supporting Information). During 1 h photoelectrolysis at an external bias of 0.8 V with tandem PEC cell I, a charge density of 1.0 ± 0.1 C cm −2 passed through the external circuit with 5.0 ± 0.5 µmol cm −2 of H 2 and 2.1 ± 0.4 µmol cm −2 of O 2 being detected. The corresponding FEs are 99% for H 2 and 83% for O 2 (Table 2 ). In the case of PEC cell II, a charge density of 3.3 ± 0.2 C cm −2 was generated, with a FE of 91% for H 2 (15.2 ± 0.2 µmol cm −2 ) and 82% for O 2 (6.9 ± 0.1 µmol cm −2 ) at an external bias of 0.6 V after 1 h of photoelectrolysis ( Table 2 ). The near-quantitative FE and the H 2 to O 2 ratio of ≈2 to 1 in both tandem cells confi rm that the passed charge arises mainly from water splitting.
The measurement of the FE allows us to calculate the true solar to fuel conversion effi ciency without relying on the assumption of quantitative product formation. Thus, the "standalone" solarto-hydrogen (STH) effi ciency (Equation ( 2) ) [ 20 ] at zero applied bias can be calculated from the short-circuit photocurrent density ( j SC ) Adv. Energy Mater. 2015, 5, 1501668 www.MaterialsViews.com www.advenergymat.de   PEC cell I did not show an unbiased photocurrent, but 45 µA cm −2 were generated by PEC cell II at zero applied bias corresponding to a STH of 0.05%.
Another meaningful energy conversion effi ciency for a working PEC device is the effi ciency of solar energy conversion to hydrogen fuel under applied bias conditions (AB-STH, Equation ( 3) ). [ 1b , 3b , 25 ] In this calculation, the applied bias is accounted for and only the stored chemical energy from solar photons contributes to the effi ciency; hence, performing such an analysis can identify the applied bias at which a working device should be operated for highest effi ciency solar-to-fuel conversion. Accordingly, the maximum AB-STH effi ciency achieved is 0.19% (at 0.8 V) and 0.59% (at 0.6 V) for tandem PEC cell I and II, respectively.

Performance Comparison with State-of-the-Art Tandem PEC Cells
Several state-of-the-art tandem cells have been reported, including the "Turner cell" (a GaInP 2 photocathode biased by an integrated GaAs PV, with a STH effi ciency of 12.4%), [ 4a ] an amorphous hydrogenated Si integrated with W doped BiVO 4 (3.6%) [ 4c ] and a dye-sensitized solar cell-biased WO 3 (3.1%) or hematite (1.17%). [ 4b ] However, these tandem PEC cells consist of one photoelectrode and one noble metal electrode, and thus are not directly comparable here. A STH effi ciency of 8.2% was achieved by side-by-side irradiation of a dual-photoelectrode combination of p-InP and n-GaAs. [ 26 ] However, this system also contains very expensive components, and both photoelectrodes were illuminated independently, which means that it is also not fully comparable to our stacked confi guration tandem system. Accordingly, the following discussion focuses on PEC cells consisting of one photoanode and one photocathode in tandem confi guration. [ ( Table 3 ). Cu 2 O-BiVO 4 and aSi-Fe 2 O 3 provide bias-free photocurrent densities of 0.32 and 0.74 mA cm −2 , respectively, though the former system employed expensive ALD techniques and the latter used a costly platinum electrocatalyst. Here, tandem PEC cell II generates 45 µA cm −2 (STH = 0.05%) without an applied bias and achieves an AB-STH effi ciency of 0.59%, showing a much better stability than Cu 2 O-BiVO 4 (Table 3 ) without the need for nonscalable techniques and materials. To the best of our knowledge, PEC cell II's AB-STH effi ciency is the highest reported solar-to-hydrogen efficiency in a dual-photoelectrode tandem water splitting system that does not employ noble-metal cocatalysts (Table 3 ). 6,7,14

Medium-Scale Tandem PEC Cell
The success of PEC water splitting devices as a viable technology relies ultimately on the scalability of such systems. [ 27 ] We have therefore investigated a medium-scale tandem system combining the better-performing nanoBiVO 4 |TiCo OEC photoanode with a p-Si|TiNi HEC photocathode in more detail. Since the measured photocurrent density of nanoBiVO 4 |TiCo OEC was smaller than p-Si|TiNi HEC on the small-scale, this electrode's size was maximized to the greatest allowable illumination area (4 cm 2 ) in our PEC reactor.
Subsequently, p-Si|TiNi HEC electrodes of different sizes were studied in order to match the overall photocurrent produced by this photocathode to that of the 4 cm 2 photoanode. Ultimately, a 4-to-1 geometric surface area ratio between the BiVO 4 and p-Si electrodes was employed in order to obtain a system with reasonably matched photocurrents over the practically applicable potential range ( Figure 8 a). In half-cell analysis, the p-Si|TiNi HEC with a geometric surface area of 1 cm 2 generated a photocurrent of ≈−2 mA at 0 V versus RHE, whereas nanoBiVO 4 |TiCo OEC with a geometric surface area of 4 cm 2 generated a photocurrent of ≈3.7 mA at 0.6 V versus RHE. The comparatively reduced photocurrent densities in the larger electrodes are explained by increased iR drop in the system due to increased current loads, and by complications in maintaining uniform composite fi lm loadings on the larger surfaces.
In this mid-scale PEC cell II, where the two photoelectrodes generate comparable photocurrents, the limitation of the twoelectrode tandem cell performance is an operating compromise between the photoanode and the photocathode. This mid-scale tandem device achieved a maximum AB-STH of 0.28%, corresponding to a photocurrent of 1.96 mA at an applied bias of 0.6 V (Figure 8 b). An additional 0.2 V of applied bias are required to produce this photocurrent when compared to the expected required bias derived from the photoelectrodes' half-cell performance ( Figure S17, Supporting Information); this additional device-based overpotential (more noticeable than on the smaller-scale) is attributed to the larger ohmic losses resulting from the increased current loads. Nonetheless, a PEC current of almost 2 mA at an applied bias of 0.6 V under standardized solar light irradiation is the highest reported photocurrent (and thus the highest rate of H 2 generation) for a dual photoelectrode PEC water splitting system, to the best of our knowledge. [ 6a,b , 28 ] While these experiments demonstrate the plausibility of moving toward larger-scale PEC water splitting devices with the current materials, they also serve to highlight some of the challenges that must still be overcome in scaling such systems.

Conclusions
In this work, the application of SSP chemistry for preparing multifunctional composite coatings for photoelectrodes has been reported, along with the use of these composite-coated photoelectrodes in PEC water splitting. Ti-/Ni-and Ti-/Cocontaining fi lms can be easily and inexpensively prepared by drop-casting or spin-coating SSPs onto a range of conductive and semiconducting substrates under ambient conditions. The TiNi and TiCo fi lms act as precursors to bifunctional HECs and OECs for water splitting in pH 9.2 electrolyte solution, demonstrating compatibility for applying these catalyst fi lms onto water splitting photoelectrodes under the same conditions. In addition to serving as bifunctional electrocatalysts, TiNi and Adv. Energy Mater. 2015, 5, 1501668 www.MaterialsViews.com www.advenergymat.de TiCo also act as SSPs to form an amorphous TiO 2 layer for protecting the semiconductor electrodes, thereby enhancing their photostability. We have therefore demonstrated for the fi rst time that a multifunctional material can be integrated with photoelectrodes for application in solar water splitting, while using an approach that does not require prohibitively expensive or nonscalable materials, techniques, or experimental conditions.
Optimized photocathode and photoanode pairs were subsequently combined and tested in tandem PEC water splitting. Close-to-quantitative H 2 and O 2 gases were generated in a near two-to-one ratio with a benchmark AB-STH effi ciency of 0.59% in a PEC cell with p-Si|TiNi HEC wired to nanoBiVO 4 |TiCo OEC in a B i solution (pH 9.2) at room temperature. By scaling to a midsized tandem PEC cell with these electrodes we were able to generate a total photocurrent of ≈2 mA at an applied bias of 0.6 V, which is believed to be the highest operating photocurrent for a dual photoelectrode PEC device to date. Thus, SSP chemistry has been established for the one-step fabrication of cost-effective, scalable, and multifunctional composite materials for PEC water splitting.

Experimental Section
Preparation of TiNi SSP and TiCo SSP : TiNi SSP and TiCo SSP were synthesized and characterized as reported previously. [ 12 ] Preparation of Electrodes : FTO|TiNi pre and FTO|TiCo pre electrodes were prepared by drop-casting a TiNi SSP (2 × 20 µL, 5 × 10 −3 M in dry toluene) or TiCo SSP precursor solution (2 × 20 µL, 10 × 10 −3 M in dry toluene) onto an FTO-coated glass substrate (Pilkington; TEC Glass 7; sheet resistance 7 Ω sq −1 , 1 cm 2 exposed surface area). The as-prepared electrodes were then dried in air at room temperature prior to use. FTO|TiNi OEC and FTO|TiCo OEC electrodes were obtained by cycling the electrochemical potential fi ve times between 0.6 and 1.9 V versus RHE with a scan rate of 50 mV s −1 . FTO|TiNi HEC and FTO|TiCo HEC electrodes were formed electrolytically in an aqueous B i solution (0.1 M , pH 9.2) with K 2 SO 4 (0.1 M ) as supporting electrolyte using E appl = −0.6 V versus RHE for 10 min.
Adv. Energy Mater. 2015, 5, 1501668 www.MaterialsViews.com www.advenergymat.de  The p-Si photocathodes were prepared from commercial boron-doped Si wafers (University Wafers; resistivity of 1-10 Ω cm; (100) singleside polished). The electric contact was formed using a Ga:In eutectic solution (99.99%; Sigma-Aldrich) and a copper wire covered with a conductive silver epoxy resin (RS Components Ltd.). The geometric surface area of 0.5 cm 2 of the photocathodes was defi ned using Tefl on tape for small-scale PEC measurements and was 1 cm 2 for mid-scale PEC measurements. Before the deposition of the SSPs, the silicon surface was cleaned with sequential treatments of H 2 O 2 (30 wt% in H 2 O; Fisher Scientifi c), H 2 SO 4 (95%-98%; Sigma-Aldrich), and HF (65%; Merck Millipore) for 1 min at each step. p-Si|TiNi HEC and p-Si|TiCo HEC were prepared by drop-casting TiNi SSP (30 µL cm −2 , 2.5 × 10 −3 M in dry toluene) and TiCo SSP (30 µL cm −2 , 5 × 10 −3 M in dry toluene) eight times onto the Si substrate, followed by cathodic in situ activation at E appl = 0 V versus RHE under solar light irradiation in a B i solution (0.1 M , pH 9.2) with K 2 SO 4 (0.1 M ) as supporting electrolyte for 10 to 20 min.
Electrochemical and PEC Measurements : All electrochemical and PEC measurements were recorded with an Ivium CompactStat potentiostat with an electrochemical cell with two compartments separated by a Nafi on 117 proton exchange membrane. For three-electrode experiments, an Ag/AgCl/KCl sat electrode was employed as the reference electrode and placed in the same compartment as the working electrode. A platinum foil was used as the counter electrode and was placed in the second compartment. The reported data are not corrected for iR drop. Electrode measurements were carried out in a pH 9.2 B i solution with additional K 2 SO 4 (0.1 M ) as the supporting electrolyte unless otherwise noted. All redox potentials were converted to RHE by using E (V vs RHE) = E (V vs Ag/AgCl/KCl sat ) + 0.197 + 0.059 × pH. [ 29 ] Electrochemical and PEC studies were executed at room temperature.
For tandem PEC cell studies, a two-compartment cell separated by a Nafi on 117 membrane was used with the photoanode as the front electrode and the photocathode as the back electrode in the same light path. A solar light simulator (Newport Oriel, Xenon 150 W) was used as the light source in all experiments. The light intensity was calibrated to 100 mW cm −1 (1 sun). An air mass 1.5 global (AM 1.5G) fi lter and an IR water fi lter (to avoid heating of the electrolyte solution) were used.
Physical Characterization : SEM was conducted to study the surface morphology of electrodes (Phillips XL30-SFGE). p-XRD analyses were carried out using an X'Pert PRO X-ray diffractometer (PANalytical B.V.). Surface compositions of the electrode were verifi ed by XPS (AXIS Nova, Kratos Analytical, with the CasaXPS software) using a high power monochromatic Al Kα radiation (1486.6 eV, 400 µm spot size, 36 W). Survey spectra were collected with a pass energy of 200 eV and 30 sweeps, whereas high-resolution spectra were collected at a pass energy of 40 eV with ten sweeps.
Hydrogen and Oxygen Measurements : Oxygen was analyzed in the headspace of the anodic compartment of the PEC cell using an Ocean Optics fl uorescence oxygen probe (FOXY-R). The probe was inserted through a tightly sealed septum and continuous O 2 readings (O 2 partial pressure) at 1 s intervals were made throughout the experiment. For electrocatalytic O 2 production with FTO|TiCo OEC , a potential of 2.0 V versus RHE was applied between 0.5 and 6.5 h of the experiment with the fi rst 0.5 h as control with no applied potential. For PEC O 2 production by nanoBiVO 4 |TiCo OEC in a three-electrode system, the cell was operated at an applied potential of 1.23 V versus RHE in the dark during the fi rst 0.5 h (control experiment), followed by 1 h under standardized light illumination (100 mW cm −2 ) and another 0.5 h in the dark (control experiment). For O 2 quantifi cation in PEC cell I, an applied bias of 0.8 V was applied in the dark during the fi rst 0.5 h (control experiment), followed by 1 h under illumination and another 0.5 h in the dark (control experiment). In the case of tandem PEC cell II, the cell was operated at an applied bias of 0.6 V with the same dark-light-dark intervals. The control experiment is used for determining leakage of O 2 from the atmosphere into the cell and the resulting data were corrected for the derived rate of O 2 leakage. The total amount of O 2 evolved was determined as the sum of O 2 measured in the headspace using the ideal gas law plus dissolved O 2 in the solution calculated by Henry's Law.
The amount of H 2 generated in the headspace of the cathodic compartment was detected and quantifi ed with an Agilent 7890A Series gas chromatography equipped with a 5 Å molecular sieve column (N 2 carrier gas at a fl ow rate of ≈3 mL min −1 ). The gas chromatography oven kept the columns at 45 °C, and a thermal conductivity detector was used. The electrochemical cell was purged with 2% CH 4 in N 2 for at least 20 min prior to PEC experiments; methane served as an internal standard for H 2 quantifi cation by gas chromatography. Using a syringe, the headspace gas was removed from the airtight electrochemical cell for gas chromatography analysis after electrochemical or PEC experiments. The total amount of H 2 evolved was determined as the sum of H 2 measured in the headspace using the ideal gas law plus dissolved H 2 in the solution calculated by Henry's law.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author. Additional data related to this publication is available at the University of Cambridge data repository (https://www.repository. cam.ac.uk/handle/1810/252335).