La5Ti2Cu0.9Ag0.1S5O7 Modified with a Molecular Ni Catalyst for Photoelectrochemical H2 Generation

Abstract The stable and efficient integration of molecular catalysts into p‐type semiconductor materials is a contemporary challenge in photoelectrochemical fuel synthesis. Here, we report the combination of a phosphonated molecular Ni catalyst with a TiO2‐coated La5Ti2Cu0.9Ag0.1S5O7 photocathode for visible light driven H2 production. This hybrid assembly provides a positive onset potential, large photocurrents, and high Faradaic yield for more than three hours. A decisive feature of the hybrid electrode is the TiO2 interlayer, which stabilizes the oxysulfide semiconductor and allows for robust attachment of the phosphonated molecular catalyst. This demonstration of an oxysulfide‐molecular catalyst photocathode provides a novel platform for integrating molecular catalysts into photocathodes and the large photovoltage of the presented system makes it ideal for pairing with photoanodes.

PEC H 2 production from aqueous solution, with an onset potentialo f0 .8 Vv ersus the reversible hydrogene lectrode (RHE) when using Pt as ac o-catalyst. [8] Surfacem odification of LTC and LTCA photocathodes with thin TiO 2 layers improves their activity due to enhanced charge separation, [10,11] and layers of amorphous TiO 2 have been shownt op rotect LTCA under strongly alkaline (pH 13) conditions. [12] Molecular complexes of Earth-abundant metals are alternatives to preciousm etal nanoparticles, where tuning the primary and secondary coordination spheres allows obtaining very high per site activities. [13] In addition, modification of the outer coordinations hell of molecular catalysts allowsf or specific anchoring onto photocathode surfaces. [14,15] Phosphonic acid anchoring groups have ah igh affinity for metal oxide surfaces under acidic and pH neutral aqueous conditions, [16,17] and are thus well suited for immobilizing molecular catalysts to photocathodes stabilized with TiO 2 protection layers. [18] In this study,w er eport an LTCA photocathode that is stabilized by as puttered TiO 2 layer and modified with am olecular H 2 evolution catalyst. As catalyst, we employaNi bis(diphosphine)-based complex (NiP,F igure 1), which shows optimal activity under mildly acidic conditions, [19] and has previously been employed on am esoporous TiO 2 electrode and aT iO 2protected p-Sip hotocathode. [20,21] The resulting LTCA j TiO 2 j NiP photoelectrode represents an ew type of molecular-semiconductor hybrid photocathode with high PEC performance and stability, as well as expanding the operating conditions of LTCA to acidic aqueous solution.
Particulate LTCA was prepared as previously described, [8,9] and was assembled into freestanding photocathodes with an Au backing layer.N anostructured LTCA films were subsequently modified with at hin (approx. 2nm) layer of TiO 2 by reactive RF magnetron sputtering, and annealed under air to form electrodes referredt oh erein as LTCA j TiO 2 (see SupportingI nformation for details). Sputtering wasu sed because this technique has previously been shown to increase the activity of LTCa nd LTCA photocathodes by minimizing charge recombination. [10,11] The TiO 2 layer also stabilizes the LTCA under the otherwise corrosive acidic conditions required for NiP to operate, [19,22] as well as providing ap rovena ttachment site for this catalyst. [20,21] Scanning electron microscopy (SEM, Figure 1) of these electrodes shows the rod-like structure of the LTCA particles with lengths of approximately af ew micrometers.
Immobilization of NiP on LTCA j TiO 2 electrodes was achieved by submersion in am ethanol solution (0.5 mm)o vernight. The presence of NiP was confirmed by the observation of N, P, and Ni in the XPS spectrum of the modified electrode ( Figure S1 in Supporting Information). The NiP surface loading was quantified as 33.7 AE 2.4 nmol cm À2 by desorptionf rom LTCA j TiO 2 in aqueous NaOH solution (0.1 m), followed by UV/Vis spectroscopic analysis (see Supporting Information). [20] This value is in the expected range for am esostructured electrode modified with aphosphonated metal complex. [17,21] The PEC properties of the LTCA j TiO 2 j NiP electrode were first studied in aqueous Na 2 SO 4 solutiona tp H3under simulated solar irradiation (AM 1.5G,1 00 mW cm À2 ). The linear sweep voltammetry (LSV) scans shown in Figure 2a show ac athodic onset photocurrent at approximately 0.65 Vv ersus RHE with a photocurrent of À0.6 mA cm À2 at 0Vversus RHE. Note that the LSV response does not represent an improvement compared to the LTCA j TiO 2 electrodes withoutc atalyst modification on this timescale (approximately 1min). This mayb ed ue to a photoreductive decomposition process occurring for LTCA j TiO 2 ,l eadingt oi nitially higherb ut short-lived catalytic activity (see below).
Nevertheless,c ontrolled potentialp hotoelectrolysis (CPPE, Figure 2b)d uring irradiation at + 0.3 Vv ersus RHE reveals a differentr esponse between NiP-modified and bare LTCA j TiO 2 electrodes. The latter shows an initial photocurrent of À0.4 mA cm À2 ,w hichd ecays quickly to zero within three hours, indicating instability and/or lack of catalysis. In contrast, prolonged CPPE with LTCA j TiO 2 j NiP under the same conditions showed an initial photocurrent of À0.2 mA cm À2 ,w hich was retained at 50 %a fter 3hours irradiation ( Figure 2b). Product quantificationt hrough gas chromatography during aC PPE experiment with LTCA j TiO 2 j NiP reveals that 1.7 mmol H 2 was produced, representingaturnovern umber (TON) of approximately 50 per initial Ni site. The Faradaic yield was measured to be 87 % ( Figure 2c), whichm atches the values obtained for previously reported NiP catalysts anchored on TiO 2 cathodes [20] and photocathodes. [21] Longer term CPPE ( Figure S2 in Sup-  porting Information) demonstrated continued photocurrent for at least 6h,a lthough with as teadilyd eclining activity that is consistentw ith catalystd esorption or degradation.I nt he absence of NiP,H 2 production had essentially ceased after 2h ( Figure S3 a), suggesting that this is accompanied by ad egradation process that is avoided when the electrons are harvested by the NiP catalyst. It has previously been shown that NiP can efficiently accept electrons from the TiO 2 conduction band, [19,20] which slows the otherwise rapid degradation of the LTCA semiconductor material.
The LTCA j TiO 2 j NiP photocathodes were further characterized by measuring the single wavelength incident photon-tocurrent efficiencies (IPCE) across the visibles pectrum with an appliedp otential of + 0.3 Vv ersus RHE. The photocathodes demonstrated photocathodic current at wavelengths as long as 660 nm, matching the diffuser eflectance UV/Vis spectrum ( Figure 3). An IPCE of 2.4 %was recorded at 440 nm.
Analysis of immobilized NiP on the electrode surface after a three-hour CPPE experiment (quantified by UV/Vis spectroscopy followingd esorption in aqueous NaOH, see above)r evealed as urfacel oading of 16 nmol cm À2 .T his represents al oss of approximately half of the initial catalyst from the electrode surface during CPPE and matches the 50 %d rop in PEC activity (Figure 2b). Moreover,aqualitative resemblance betweent he UV/Vis peaks of the desorbed NiP before and after CPPE (Figure S4), as well as retentiono fp eaks correspondingt oNand P in the XPS spectra after 1h CPPE (Figure S1), support the molecular integrity of NiP during the PEC experiments.N is ignals belonging to either NiP or ad ecomposition product could not be clearly resolved in the post-CPPE XPS spectra.
Control experiments were performed without as puttered TiO 2 layer.ANiP loading of 17.6 AE 3.5 nmol cm À2 was determined for LTCA j NiP,w hich is approximately half that observed for LTCA j TiO 2 j NiP,a nd thus in agreement with the particular affinity between the phosphonate-modified molecules and TiO 2 . [15,23] Moreover,w hen these electrodes were subjected to CPPE at + 0.3 Vv ersus RHE under the same conditions as described above (Figure S5), the initially low photocurrento f À0.15 mA cm À2 approached zero within two hours, demonstratingt he requirement for as tabilizing layer.T hus, TiO 2 adds two further benefits to the increasei np hotocurrent observed at pH 10 [10,11] -it stabilizes LTCA and provides an improved anchorings ite for NiP.
LTCA j TiO 2 electrodes modified with Pt instead of NiP were prepared by PEC reduction of H 2 PtCl 6 (see Supporting Information for details). When studied in pH 10 electrolyte solution, the most commonly used conditions for thesem aterials, [8] the electrodes displayed ap hotocurrento fÀ1.0 mA cm À2 at 0V versus RHE in the LSV and relativelys table H 2 productionf or two hours at E appl =+0.5 Vv ersus RHE ( Figure S6). In pH 3 electrolyte solution,t he LSV of the LTCA j TiO 2 j Pt electrodes displayed as imilar onset potentialt oL TCA j TiO 2 j NiP electrodes with photocurrentr eaching À0.8 mA cm À2 at 0V versus RHE (Figure 2a). However,C PPE at + 0.3V versusR HE showed am uch quicker decay than with the LTCA j TiO 2 j NiP electrodes, falling to only À0.01 mA cm À2 after three hours irradiation and aF aradaic yield of 84 % ( Figure 2b and Figure S3 b). This decay is consistent with previous studies, where the contact between this TiO 2 protecting layer and the Pt catalyst hasl imited the durability of the photocathode, [24] and either an additional Mo/ Ti layer [25] or replacing the Pt co-catalyst with at hick RuO x film [26] has been required to achieve long term performance. Our photoelectrode therefore provides ar are example where the chemical attachment between am olecular catalysta nd the electrode is improved comparedt oapreciousm etal catalyst layer,a nd indeed the molecular catalyst appears essential for operation of the LTCA materialu nder mildly acidic conditions. H 2 production by LTCo rL TCA from acidic solution has not previously been demonstrated, and therefore theu se of both the TiO 2 stabilizing layer and molecular catalyste xpands the possible operating conditions of this class of material.
This work represents an advance in the assembly of hybrid semiconductor/molecular catalyst photocathodes due to its positiveo perating potential and long wavelength activity.T he onset potential of + 0.65 Vv ersus RHE is significantly more positivet han for p-Si modified with TiO 2 and the same NiP cocatalyst, [21] and is comparable to the values reported using GaP-based hybrid photocathodes. [27,28] Additionally,o ur LTCA hybrid photocathodes allow H 2 production up to l = 660 nm, which exceeds that of GaP,w here the band gap limits light absorptiont o5 49 nm. [14] The use of the particle transfer fabrication method, compared to flat semiconductor waferss uch as p-Si and GaP,m akes this electrode scaffold potentially scalable over large areas. The excellent contact between the particles and the contact layer (Au in this case) removes the requirement for high crystallinity across the whole panel, [29,30] and the intrinsically mesostructured surface enablesh igh catalystl oading. However,t he loss of activity is somewhat faster than for the previous( photo)cathodes using the same catalyst. [20,21] Previous systems utilized mesoporous TiO 2 with several-mm-thick films, which trap the catalyst through re-adsorptionf ollowing desorption, when compared to the nm-thick TiO 2 layer reported here. Therefore, increasing the porosity of the TiO 2 layer may improvet he system's longevity in future development. Finally,L TCA j TiO 2 j NiP comparesf avourably in terms of photocurrent and Faradaic yield with "all-molecular" dye/H 2 -catalyst assemblies, which are typicallyimmobilisedonp-type materials such as NiO andC uCrO 2.
[ [31][32][33][34][35] In conclusion, photocathodes based on the oxysulfide La 5 Ti 2 Cu 0.9 Ag 0.1 S 5 O 7 were combinedw ith am olecular catalyst, enablinge arly-onset H 2 fuel synthesis with am olecular/inorganic hybrid.T his was accomplished by employing as puttered TiO 2 layer,w hich protects the material from the aqueous electrolyte solutiona nd provides as uitable attachment site for the phosphonic acid-modified molecularc atalyst NiP.T he latter was crucial for the high performance of the photocathode, which still retained5 0% of its initial activity after three hours photoelectrolysis. An analogous photocathode modified with Pt displayed poor stability,g iving ar are example where am olecular catalyst exceeds the activity ands tabilityo fP tf or H 2 production. [36] Our resultsdemonstrate the possibility of replacing expensive Pt with af irst row transition metal catalystf or oxysulfide-type photocathode materials. The use of aT iO 2 overlayer opens up their use in previously unreported acidic conditions that may be required in tandem PEC device architectures. From am olecular catalysis point of view,t his is the most positive operating potentialf or an inorganic light-harvester/molecular catalyst hybrid, and the system retains high Faradaic efficiency even at this potential. This is therefore an essential step towards constructing an efficient, bias-free molecule-catalysed photoelectrochemical cell.

Experimental Section
Experimental details can be found in the Supporting Information.