Clusters as an Effective H2 Evolution Catalyst on Protected Si Photocathodes

This work shows how a molecular Mo 3 S 4 cluster bonded to a photoelectrode surface via a phosphonate ligand can be a highly effective co-catalyst in photocathodic hydrogen evolution systems. Using a TiO 2 protected n + p Si photocathode, H 2 evolution occurs with an onset of + 0.33 V vs. RHE in an acid solution with this precious metal-free system. Using just the red part of the AM1.5 solar spectrum ( λ > 635 nm), a saturation current of 20 mA/cm 2 is achieved from an electrode containing Mo 3 S 4 dropcasted onto a 100 nm TiO 2 /7 nm Ti/n + p Si electrode.

It is well established that society cannot indefinitely rely on fossil fuels and must switch to a renewable energy source. 1 Solar fuels such as hydrocarbons from CO 2 reduction and H 2 from the water splitting reaction are highly attractive since these molecular products can be stored indefinitely 2 and used directly in the transportation infrastructure. Photocatalytic water splitting using a 2-photon device has the potential for much higher efficiency than a 1-photon device. 3,4 In a 2-photon device the optimal bandgaps for the 2 photoabsorbers are approximately 1.7 eV and 1.0 eV. [3][4][5] This allows for the large bandgap material to absorb the blue light leaving the red light to be absorbed by the small bandgap material. Silicon is very promising as the small bandgap material since it has a bandgap of 1.1 eV. A major problem with Si and other small bandgap materials is that they have stability problems in a water splitting environment. Recently it was shown that depositing a thin film of TiO 2 mitigates stability issues. [6][7][8] However, the rectifying properties of this film may restrict the ability to electrodeposit catalysts onto the protected photoabsorber surface. 9,10 Even though the TiO 2 protection layer can be detrimental to certain electrodeposition procedures, it may also open up a new avenue to help improve the stability of molecular catalysts.
Previous work showed molecular clusters such as Mo 3 S 4 incomplete cubane clusters to be quite effective as H 2 evolution catalysts, 11,12 however the materials are also known to be water soluble. This solubility makes it quite difficult to keep these catalysts adsorbed on an electrode surface in a working environment. A previous approach to resolve this issue involved attaching hydrophobic ligands to the clusters. 13,14 Unfortunately this approach decreased catalytic activity and stability in air. One of the great difficulties in depositing these catalysts was the limited freedom in interacting with the Si surface due to its extreme propensity to oxidize to SiO 2 . However with the emergence of TiO 2 as a protection layer, there now is a much larger freedom to design molecular catalysts that can interact with the electrode.
In the present work we investigated attaching a (hydrophilic) molecular linker to help bind Mo 3 S 4 clusters to the TiO 2 surface. Previous work showed that phosphonate groups are quite effective at binding to TiO 2 in photocatalytic H 2 evolving systems. 15 Thus a N-(phosphonomethyl)iminodiacetate (Hpmida)-Mo 3 S 4 cluster was synthesized and used as a catalyst. Rather than dropcasting the catalyst onto the electrodes, we soaked the electrodes in a solution of the Mo 3 S 4 -Hpmida overnight. We then thoroughly washed off the electrode with ultrapure water (18 M -cm) to ensure that only bound Mo 3 S 4 -Hpmida remained on the 100 nm TiO 2 /7 nm Ti /n + p Si electrodes. This was found to give the same performance as dropcasting, thus indicating an adsorbed monolayer is all that is needed to maximize catalytic efficiency. A photoirradiated p-n junction was used rather than simply a degenerately doped Si to show that potential photon absorption from Mo 3 S 4 clusters is not a significant issue for blocking light to the Si nor in the Mo 3 S 4 catalysis of protons to H 2 . Figure 1 shows a cyclic voltammogram (CV) of red light irradiated (λ > 635 nm) 100 nm TiO 2 /7 nm Ti/n + p Si photocathodes with either Mo 3 S 4 -Hpmida or Pt nanoparticles as catalysts. The electrolyte was 1.0 M HClO 4 . This material has an onset potential (i.e. at current density of 1 mA/cm 2 ) of +0.32(4) V vs. Reversible Hydrogen Electrode (RHE), which is essentially the same as the +0.33(0) V vs. RHE achieved earlier with a MoS x catalyst. However, unlike in our previous work using Mo 3 S 4 clusters with a hydrophobic ligand, 13,14 the Mo 3 S 4 -Hpmida clusters were relatively stable in air, thus no special precautions were necessary to use these materials. It was found that leaving the catalysts in atmospheric conditions for 2 years did result in a slight decrease (∼30 mV) in activity. See Supporting Information for details.
Since the Mo 3 S 4 -Hpmida clusters are water soluble, there is the risk that they dissolve into solution during operation. In previous work, there was substantial current loss after less than 10 scans (corresponding to approximately 5 minutes). To test for durability, a  chronoamperometry test was run. Figure 2A shows a CV before and after a 1 hour chronoamperometry test was ran at a potential of +0.20 V vs. RHE. Figure 2B shows the corresponding 1 hour chronoamperometry test. The variations during the 1 hour chronoamperometry test may be related to mass transfer issues near the catalyst. The CV after 1 hour shows negligible change in catalytic activity. Compared to previous work with water soluble, air stable catalysts, 11 the Hpmida functional group greatly stabilized the catalysts. Further long term testing at more reductive potentials show that some cathodic stability issues remain. Since previous results show that TiO 2 /Ti is stable, 6 we attribute these losses to catalyst detachment/deactivation. However, re-deposition of the catalyst allowed for a completely regain in performance (See Supporting Information).
Previous research has shown that Mo 3 S 4 entities can easily undergo degradation. 14,16 To investigate if this could potentially be an issue, the Mo 3 S 4 -Hpmida catalyst/100 nm TiO 2 /7 nm Ti/n + p Si was photoirradiated and cycled to highly oxidative potentials as shown in Figure 3. The scan before and after the oxidation cycle show no difference indicating that anodic corrosion does not occur on the photo electrode. This inability for the 100 nm TiO 2 /7 nm Ti/n + p Si to corrode the Mo 3 S 4 was actually expected due to the TiO 2 layer forming a depletion layer, which blocks electron transfer. 9 Previous research has shown that when using moderately doped TiO 2 , anodic electron transfer cannot occur at potentials much more anodic H + /H 2 redox potential. 9 In Figure 3 the 5 nm Pt/n + p Si shows anodic current due to H 2 oxidation, however Mo 3 S 4 cannot oxidize H 2 , thus no current is seen for the Mo 3 S 4 -Hpmida/100 nm TiO 2 /7 nm Ti/n + p Si sample. 16 It should be noted that different Si wafers were used in Figure 1 and Figure 3. Each Si wafer has a slightly different photovoltage, thus accounting for the small difference in onset potentials between Figure 1 and Figure 3. In Figure 1 and Figure 3 the electrode with Pt and the electrode with Mo 3 S 4 -Hpmida came from the same wafer so their photovoltages should be equivalent.
While Figure 3 shows that catalyst oxidation is highly unlikely due to blocking action of the TiO 2 layer, there are other potential degradation mechanisms which the catalyst may suffer. In a commercial device, catalysts may degrade by poisoning via electrolyte impurity, mechanical instability, oxidation from oxygen crossover, or many other mechanisms. Thus regeneration/redeposition of the catalyst may be an important process that needs to be taken into consideration. These molecular Mo 3 S 4 -Hpimda clusters have the advantage over the typical nanoparticle catalysts in that they are soluble in water, and will adsorb to the TiO 2 surface of the composite photocathode. Thus deposition (and redeposition) is as simple as soaking/wash-coating the photocathode in a solution of catalyst. Nanoparticles must contend with dispersion issues, thus adding an extra step for optimization. While electrodeposition or photoelectrodeposition also typically uses water soluble precursors, an electrical bias needs to be applied, thus adding another step to this procedure as well. While optimization of both nanoparticles and electrodeposited particles is possible, the simplicity of having an aqueous solution of a catalyst adsorb to the semiconductor surface gives molecular catalysts such as cubane clusters a distinct advantage of simplicity.
In summary, this work demonstrates that a Mo 3 S 4 -Hpmida/100 nm TiO 2 /7 nm Ti/n + p Si composite photocathode can produce an onset of +0.32(4) V vs. RHE for H 2 evolution. Importantly, we demonstrate that anchoring the cubane catalyst to the electrode via phosphonate-TiO 2 gives a very significant improvement in the stability of this efficient, molecular H 2 evolution catalyst, although room for improvement remains. Future work will focus on analyzing other potential deactivation mechanisms and designing the photocathodic system to eliminate them.

Experimental
For anchoring on hydroxylated TiO 2 surfaces, the Mo 3 S 4 cluster was derivatized with N-(phosphonomethyl)iminodiacetate (Hpmida) ligands. A suspension of the free acid H 4 pmida in water was deprotonated by 3 molar equivalents of NaHCO 3 leading to dissolution of the ligand. Dropwise addition of this mixture into an aqueous The electrodes were made and tested in a manner similar to our previous work. 9 The exact details are given in the supporting information.