Dear Matthew T Mayer,

I think it might be role of surface plasmon absorption of Au and charge transfer from Au to fe2o3 or Si might produce enough potential for water splitting eventhough 900 nm photon is not having enough energy.

Firstly, Mathew is correct in his saying. Splitting water involves both reduction (to yield H2) and oxidation (to yield O2). The values for potentials of the two processes are far away, which needs high band gap value (about 2 eV) and consequently needs shorter than 620 nm radiation. This is a well-known fact of life. The 900 nm radiation can not give both H2 and O2 as complementary reactions.

However, the 900 nm is in the IR region, and the 16% efficiency is not well defined here. Suppose that two 900 nm photons are combined, they would yield shorter wavelength effect, and the process may become possible (in theory at least). In practice, the 16% is too good to be good, unless it is meant to be 16% out of the eventually combined photons.

In any way this is an interesting issue, food for thought

Matthew,

The details are not well described in this publication. A typical Xenon lamp has a spike of intensity around 900 nm. The illumination power is 60 mW/cm^2 = 600 W/m^2 (Figure 5.). If all this power belongs to 900 nm photons, then a photon flux is 3e21 photons/s.m^2= 0.005 (moles of photons)/s.m^2. They have a current density (Figure 5a) 2 mA/cm^2 = 20 A/m^2 = 0.0002 (moles of electrons)/s.m^2 at 0 V potential. A formation of each O2 molecules requires 4 photons. Then, IPCE(%)= (100%/4)*(0.0002/0.005) = 1%, which is reasonable but much lower than claimed. Probably they miscalculated IPCE.

As you explained I'm really surprised by this IPCE value. And unfortunately I don't find their experimental setup for this measurements in Suppl. Info. Because for a dual absorber PEC cell, you can't measure ICPE with a simple light source. As show in their figure 1, there are two semiconductors that should absorb photons to have complete water splitting O2 and H2). And at 900 nm Fe2O3 doesn't absorb in theory. They can measure a photocurrent (and IPCE) like this, but to my point of view the hole of Si are not used to make oxidation of water, but must interact with the Fe2O3 or Au ?

For the plasmonic effect and injection of hot electron, I'm also surprised that they don't observe a peak in IPCE and UV vis spectra associated to the plasmon resonnace (around 600 nm for Au I guess).

And the combined photon absorption, suggested by Hikmat, this would be great, but infortunately it is highly improbable: If I remember well the absorption cross section of photon in a semiconductor is 10e-7 nm2 and if it requires two photons at the same place you get a really bad statistic (10e-14). I'm not sure of these values (I think it was for TiO2), but except for really special materials, this process exist but with a really bad statistic and it can be observed only with big Laser.

Dear Thomas Cottineau,

I think SPR peak of Au is tunable to 900 nm. you can refer this paper.

1) Gold nanostructures: engineering their plasmonic properties for biomedical applications,Chem. Soc. Rev., 2006, 35, 1084–1094

2) Enhanced semiconductor optical absorption via surface plasmon excitation in metal nanoparticles, APPLIED PHYSICS LETTERS 86, 063106 (2005)

I believe it is a simple corrosion process. Hydrogen is produced by a reaction of silicon and water; this process can be stimulated by illumination (900 nm is enough for Si) and/or formation of Si/Au galvanic couple.

I agree with Anatoly ! It is Si or may be even Au corrosion.

Hi Matt !

This morning a read the Nano Letters article you mention, because I wondered too about the possibility that plasmon-enhancement improves so much the harvesting and photocurrent... Albeit I can accept that nano-Si could absorb more towards visible; that a p-n junction could help photovoltage; and that AuNPs could enhance light absorption and charge injection, I notoce that Fig.5a is showing a parasitic oxidation phenomenon (peak between -0.5 and + 0.5 V) possibly due to copper (or something else) oxidation (Cu is revealed by EDS Fig. 3d) ! Hence photocurrent density results around 1 mA/cm2 as usual.

Thanks for your I-V PHOCS !

When we say water splitting we should mean production of both H2 and O2. If one reaction is taking place that is not water splitting. In such a case with 900 nm radiation, the only way out (assuming no other errors) would be combination of two photons in one photon (with shorter wavelength). In theory this is OK.

In practice, this would not give 16% conversion, as combining two photons is not easy job and has little probability.

Thanks for the responses! I agree with those who have said Si oxidation is likely a significant process here, contributing to current. After all, either Fe2O3 is fully covering the Si surfaces and therefore prohibiting holes generated in Si from reaching the solution when illuminated at λ>600nm (due to deep valence band of Fe2O3; hence zero IPCE would be expected for the water oxidation half-reaction), or the Fe2O3 does not fully cover the Si which would leave it exposed to strongly basic solution known to readily corrode and dissolve Si. In fact, I/V curves of Si in base show this oxidation current in many publications (eg. http://pubs.acs.org/doi/abs/10.1021/jp052595w), which shows up around 0V vs RHE as reported here.

http://pubs.acs.org/doi/abs/10.1021/jp052595w

The dual absorber behavior is studied in the attached paper. Standard monochromatic IPCE cannot be used to characterize such a device due to the two-photon requirement for charges to move through the materials. It would be interesting to see the IPCE measured under a constant bias of white light, but alas absolutely no experimental details are given for IPCE.

I hoped to learn something from this paper...but instead hopefully we can all learn from each other in this discussion!

Article Hematite/Si Nanowire Dual-Absorber System for Photoelectroch...

We had an interesting discussion thanks to Matthew’s question. Now, it would be nice to hear from authors. Water oxidation catalyst is a bottleneck in water splitting to O2 and H2. A photon at 900 nm has an energy about 1.37 V. The slitting of water requires 1.24 V. Thus, it’s thermodynamically feasible, but the driving force is too small. The problem of this paper is inconsistency of their own raw data with the reported numbers (See my comments above). In addition, the oxygen yield is not quantified, it’s given in arbitrary units. The Faraday efficiency is not reported. We can have thousands hypothesis what is going on in this system, but it would be waste of time if the data are not reliable. The authors did not recognized that in fact they report a supercatalyst for water oxidation. All known water oxidation catalysts are by far less active. If everything is correct (I doubt), the authors should submit the paper to Nature or Science.

Unfortunately I have no access to the full text NanoLett, but any way an efficient water photocatalytic splitting at 900 nm seems very doubtful. On the other hand I know that the light can accelerate a reaction of Si with water and 900 nm is just above Si bandgap.