Direct Observation of Two Electron Holes in a Hematite Photoanode

Article pubs.acs.org/JPCC

Direct Observation of Two Electron Holes in a Hematite Photoanode during Photoelectrochemical Water Splitting Artur Braun,*,† Kevin Sivula,‡ Debajeet K. Bora,†,§,⊥ Junfa Zhu,∥ Liang Zhang,∥,⊥ Michael Graẗ zel,‡ Jinghua Guo,⊥ and Edwin C. Constable§ †

Laboratory for High Performance Ceramics, Empa. Swiss Federal Laboratories for Materials Science and Technology, CH-8600 Dübendorf, Switzerland ‡ Laboratory for Photonics and Interfaces, Ecole Polytechnique Federal de Lausanne, CH-1015 Lausanne, Switzerland § Department of Chemistry, University of Basel, CH-4056 Basel, Switzerland ∥ National Synchrotron Radiation Laboratory, University of Science and Technology of China, Hefei, 230029, China ⊥ Advanced Light Source, Lawrence Berkeley National Laboratory, Berkeley, California 94720, United States ABSTRACT: Visible light active photoelectrodes for hydrogen generation by solar photoelectrochemical water splitting have been under scrutiny for many decades. In particular, the role of electron holes and charge transfer remains controversial. We have investigated the oxygen evolution of hematite in alkaline aqueous electrolyte under a bias potential during visible light illumination in a photoelectrochemical cell operando with soft X-ray (O 1s) spectroscopy. Only under these conditions, two new spectral signatures evolve in the valence band, which we identify as an O 2p hole transition into the charge transfer band and an Fe 3d type hole into the upper Hubbard band. Quantitative analysis of their spectral weight and comparison with the photocurrent reveals that both types of holes, contrary to earlier speculations and common perception, contribute to the photocurrent.

■

thus poor PEC performance.8 Because of inefficient charge separation, the photocurrent is limited by the holes that reach the semiconductor-liquid junction (SCLJ).11 Indeed, recent efforts to optimize the nanostructured morphologies of hematite photoanodes have led to significant improvements in the performance.12 In spite of these efforts, the overall solarto-photocurrent efficiency in hematite remains at only about 1/ 3 of its potential, even at high applied bias. In addition, the water oxidation reaction rate is limiting in aqueous electrolytes due to surface recombination of holes with electronsthe majority carriers in hematite.11,13 Surface states can mediate charge transfer, provide recombination centers, and allow a drop of potential at the surface rather than the space-charge region.14 The kinetics of these competing processes has been of interest for several decades.13,15 Gerischer stated that high efficiencies can only be obtained by minimizing effects that favor recombination of electrons and holes in materials and on surfaces.8 Thus, a complete understanding of the surface electronic states of hematite is necessary but has so-far been a subject of much debate. It is believed that due to the band structure of hematite, valence band (VB) holes would arrive at

INTRODUCTION Hydrogen generation by solar water splitting in photoelectrochemical cells (PECs) has long been the Holy Grail of sustainable energy supply.1−7 The physicochemical processes that enable the water-splitting reaction can be rationalized in the framework of Schottky barrier theory applied to semiconductor (SC) surfaces in contact with aqueous electrolytes. The electric field arising in the depletion layer of an n type or p type SC Schottky junction generally provides an efficient separation of electron hole pairs.8 The search for stable, affordable, and high-performance PEC electrodes has so far failed to identify an ideal material, but hematite has emerged as a promising candidate. Hematite (αFe2O3) is an abundant and electrochemically stable photoanode material with a relatively small band gap (2.0 eV), giving it the capability to harvest over 16% of standard solar illumination (AM 1.5G).9 Hematite is classified by X-ray spectroscopy as a charge transfer or intermediate type insulator.10 As a drawback, the conduction band (CB) of hematite is too low to reduce water, but when a DC bias is applied, such as in a tandem cell, solar water splitting is possible.1 However, the incident-photon-to-current efficiency (external quantum efficiency) of hematite is below that of other metal oxides, intrinsically due to imperfections and its electronic structure, which cause an ultrashort excited state lifetime and © 2012 American Chemical Society

Received: May 2, 2012 Revised: June 11, 2012 Published: June 14, 2012 16870

dx.doi.org/10.1021/jp304254k | J. Phys. Chem. C 2012, 116, 16870−16875

The Journal of Physical Chemistry C

Article

chamber from the UHV environment, with base pressure of 5 × 10−9 Torr during measurements. The energy scale for the O1s absorption range was calibrated by measuring the oxygen NEXAFS spectrum of a TiO2 film as reference. Using a specially designed soft X-ray in situ/operando cell,21,22 we were able to carry out the photoelectrochemical experiment under potentiostatic and galvanostatic conditions (Femtostat FAS2, Gamry Instruments, Warminster, PA), while NEXAFS spectra were recorded either in the dark or during illumination by an 1.5 AM solar simulator (HAL-302 Solar Simulator, 350−750 nm, Asahi Spectra, Japan), the light of which was directed with a glass fiber through the UHV chamber on the sample; 0.4 M NaOH electrolyte was continuously exchanged during the experiment by a peristaltic pump. A Pt wire counter electrode and a flexible Ag/AgCl reference electrode (“FlexRef, World Precision Instruments, Sarasota, Florida) were used in combination with the hematite working electrode. Figure 1 displays a schematic of the substrate−film−cell assembly in contact with the NaOH electrolyte. Overlaid are

the hematite surface in Fe3+ energy levels, rather than in O 2p levels.15 Moreover, the involvement of two different electron holes has been postulated for the photo-oxidation of water on hematite,16 which can potentially be interpreted as the surfacetrapped holes that were recently associated with highly oxidized Fe4+ and Fe5+ surface species.13 The existence of different types of holes with disparate reactivity toward water oxidation has broad implications for the ultimate performance of hematite. However, the experimental detection and verification of surface states and their charge transfer is not trivial.17 Shen et al.18 have demonstrated on n-CuInSe2 how impedance spectroscopy can be employed to map surface states in PEC cells. During the preparation of this manuscript we learned that Klahr et al.19 adopted such approach for the mapping of surface states in hematite. Our recent research efforts have been dedicated to increasing the understanding of the surface electronic states of hematite using soft X-ray analysis. Recently, an ex situ X-ray absorption study indicated the presence of highly oxidized Fe species.20 In particular, we found an extra transition in the upper Hubbard band (UHB) after hematite had been exposed to 600 mV anodic bias. At 200 mV, no such transition is found. This suggested that the surface electronic states can be altered by the PEC operating conditions. In this work, with a specifically designed soft X-ray in situ/ operando cell as outlined by Jiang et al.,21,22 we completed the photoelectrochemical experiment with state-of-the-art Si-doped and nanostructured hematite photoanodes2 under PEC conditions (with an Ag+/AgCl reference electrode and a Pt counter electrode) in alkaline electrolyte while recording nearedge X-ray absorption fine structure (NEXAFS) spectra under simulated sunlight and in the dark. To the best of our knowledge, this is the first ever reported in situ photoelectrochemical NEXAFS study. Our experimental results prove the existence of two different electron holes at the SC liquid junction under operando photoelectrochemical conditions.

Figure 1. Sketch of cell assembly and mapping of Gerischer and Schottky treatment of photoelectrochemical SCLJ. Band levels are adopted from ref 24.

■

EXPERIMENTAL SECTION Hematite films were deposited on Si frames of 5 mm × 5 mm area, 0.5 mm thickness, with a 1 mm × 0.5 mm wide Si3N4 window of 100 nm thickness (Silson Ltd., JBJ Buisness Park, Northampton Road, Blisworth, Northampton, England). Prior to deposition of the hematite film, layers of 1 nm Cr and 10 nm Au were evaporated on the Si3N4 window in order to provide an adhesive metallic current collector between e2O3 and Si3N4. The 1% Si-doped hematite was deposited by ambient pressure chemical vapor deposition with subsequent thermal treatment at 500 °C as described in ref 23. Film thicknesses were selected to allow the penetration of the incident X-rays and the escape of the fluorescence photons. In a recent study on repeated atomic layer-deposited hematite films,11 it was found that thicknesses of 20 ± 2.2 nm films had the maximum internal quantum yield as a result of balancing the development of an electrostatic field and a short collection distance. Films with a thickness around 30 nm had the highest overall efficiency in that study. We therefore decided on 30 nm thickness for our film. Our experiments were performed at beamline 7.0.1 of the Advanced Light Source in Berkeley, CA. Oxygen 1s NEXAFS spectra were recorded in the total fluorescence yield mode using a channeltron detector, and the intensities were normalized to a reference signal recorded simultaneously on a gold mesh. The thin Si3N4 membrane separated the cell

the energy band position for the VB, CB, and Fermi energy EF, plus the relevant energy positions for the flat band and the oxygen and hydrogen evolution according to Gerischer's theory.

■

RESULTS Figure 1 displays a schematic of the substrate−film−cell assembly in contact with the NaOH electrolyte. Overlaid are the energy band positions for the VB, CB, and Fermi energy EF, plus the relevant energy positions for the flat band and oxygen and hydrogen evolution according to the theory of Schottky, Marcus, and Gerischer.24 The oxygen K-shell near edge X-ray absorption fine structure (NEXAFS) is sensitive to the local bonding and symmetry properties of the X-ray excited oxygen atom. The O 1s spectra of hematite (Figure 2) have a well-developed doublet in the pre-edge region at 528−530 eV, originating from transitions of hybridized Fe 3d−O 2p states with t2g and eg orbital symmetry.10,25−27 The O1s prepeak structure of hematite is governed by the Fe 3d components in the hybridized unoccupied pd wave functions.26 The spectra are obtained from the Si-doped hematite film while in contact with the electrolyte under a bias 16871

dx.doi.org/10.1021/jp304254k | J. Phys. Chem. C 2012, 116, 16870−16875

The Journal of Physical Chemistry C

Article

originate from the same electron hole transitions, which cause the photocurrent. Similar to a previous approach,30 we deconvolute the spectra into contributions from specific transitions to quantitatively determine the relative spectral weight of the new pre-edge peaks and the t2g−eg doublet, as shown in Figure 3a,b. Closer inspection of the spectrum collected at 500 mV under illumination (Figure 3a) shows that the new spectral intensity resembles two peaks. We thus identify two transitions at around 525.5 and 527 eV. The same holds for the spectra recorded under illumination at 300 and 700 mV. The leading peak at around 525.5 eV is associated with the energetically higher located hole, and the subsequent peak around 527 eV is the electron hole peak with lesser energy and presumably less effective for water oxidation.15 A similar case with two different types of holes has been found in Sr-substituted La2CuO4,31 where the low-energy peak was assigned to an O 2p type hole transition into the charge transfer band (CTB) and the high-energy peak to a more Fe 3d type hole transition into the UHB. Comparison of the peaks from the deconvoluted spectrum with the energy band diagram of hematite,32 as demonstrated in Figure 4, reveals that the two electron hole peaks coincide with the two lowest known states UHB of tCTB 1u↑ and a1g↑ . We assign the O 2p type hole tCTB 1u↑ , which goes into the CTB the relative spectral weight SCTB = tCTB 1u↑ /t2g↓ + eg↓, and the Fe 3d type electron hole aUHB 1g↑ into the UHB the relative spectral CTB weight SUHB = aUHB 1g↑ /t2g↓ + eg↓. For 300 mV bias, t1u↑ has clearly 16 ; see Figure 2. Disagreement higher intensity than aUHB 1g↑ between the optical absorption coefficients and the photoelectrochemical measurements stimulated speculation that the latter corresponds to a ligand-to-metal charge transfer O2− → Fe3+, whereas the former could originate from a metal-to-metal charge transfer according to a photoinduced disproportionation 2Fe3+ → hv Fe2+ + Fe4+, where Fe2+ stands for Fe3+ + e− and Fe4+ for Fe3+ + h+. This implies that the second hole is linked to an iron site (Fe 3d) but with different energy than the hole in the oxygen VB.16 Figure 2 shows that at around 531.5 eV, there is the highest relative intensity in the minimum of the spectrum when the film is illuminated at 500 mV, corroborating the previously suggested highly oxidized iron species measured in an ex situ experiment.20

Figure 2. O 1s NEXAFS spectra recorded at bias from 100 to 900 mV under dark (left) and light (right) conditions.

potential from 100 to 900 mV vs Ag+/AgCl in either dark or in 1.5 AM illuminated condition. There are no striking spectral differences with the film exposed to 1.5 AM light under 100 mV (Figure 2, bottom spectra) bias. However, at 300 mV, a new spectral feature stands out before the pre-edge doublet at around 525 eV but only when illuminated. Without illumination, the spectrum does not have this new feature. We notice this illumination-induced feature clearly for the bias potentials from 300 to 700 mV (Figure 2, right panel) and to a lesser extent at 100 and 900 mV, where a slightly enhanced intensity before the pre-edge can still be made out, although it is difficult to precisely quantify its small relative spectral weight. Transitions before the pre-edge doublet of oxygen NEXAFS spectra in metal oxides have been associated with doped holes incurred by A- and B-site substitution in perovskites.28−30 While a correspondence between the intensities of pre-edge peaks and the hole states has been postulated,28 a quantitative relationship between conductivity, spectral weight, and hole concentration has only been found recently.30 Given the observation of the new transitions in hematite only when under bias and illumination, the assumption is justified that they

Figure 3. O 1s spectra recorded at 500 mV bias under light (left) and dark (right) conditions. The relative spectral weight S is indicated by the horizontal cyan blue lines. 16872

dx.doi.org/10.1021/jp304254k | J. Phys. Chem. C 2012, 116, 16870−16875

The Journal of Physical Chemistry C

Article

intensity at around 525 eV at 100 mV bias to tCTB 1u↑ . We mentioned in the introduction that due to the electronic structure of hematite, VB holes would arrive at the anode surface in Fe3+ levels, rather than in the O 2p band.15 Consequently, rate constants for water oxidation would become too small to compete with efficient surface recombination routes.15 In contrast, holes arriving at the anode surface in the O 2p band have a very facile faradaic route leading to O2.15 Our NEXAFS data show that two different VB holes are indeed formed. This relationship is sketched in the band diagram in Figure 6, where the high energetic tCTB 1u↑ hole close to the Fermi

Figure 4. Energy band diagram of hematite, reproduced from ref 32, plus aligned peaks from NEXAFS spectrum at 500 mV under illumination.

We have plotted in Figure 5 the spectral weights SCTB (solid green line) and SUHB (dotted green line) from all spectra Figure 6. Energy diagram for PEC water splitting at a hematite CTB photoanode with oxygen evolution. The aUHB 1g↑ and t1u↑ hole states are sketched between EF and EV.

level has the extreme faradaic route for oxygen evolution.15 We now need to discuss which of these holes contributes to the photocurrent. We assume that all holes that can be detected by NEXAFS are formed within a region, D, which comes from the depletion layer width, w, plus a small diffusion region, δ ∼ 2 nm; hence, D = w + δ. We recall that w increases with the square root of the bias V minus the flat band potential Vfb, which is typically in the range of Vfb ≈ 500 mV vs RHE for hematite in 1 M NaOH.15 Hematite has water oxidation photocurrent onset at Von ≈ 1000 mV vs RHE. When V > Vfb, the electric field drives holes formed in w plus a small portion in δ that can diffuse to the edge of w to the SCLJ. We consider a simple balance with the holes at the SCLJ: at low potentials, say V = Vfb + 100 mV, holes formed in the short w layer move to the SCLJ but cannot oxidize water because their free energy cannot overcome the oxygen evolution reaction over potential. This holds not only for the low energy Fe 3d type holes but also for the high-energy O 2p holes (Figure 5). Instead, all holes accumulate and create an electric field that opposes the space charge field, flattens the bands, and stops any further holes arriving at the SCLJ. This phenomenon is observed by the anodic and cathodic current spikes, which are observed in photocurrent transient measurements under light-chopping conditions.34 In this low bias condition (near 0 V vs Ag+/AgCl), the photocurrent is negligible, although there should be a finite steady-state concentration [h+] of accumulated holes at the SCLJ, which, however, we do not yet observe in our NEXAFS data (Figure 5). Only at slightly higher bias potentials (100 mV), we see a finite signature that we assign to tCTB 1u↑ holes. With increasing potential V, w steadily increases, and more holes need to accumulate at the SCLJ to oppose the space charge field. Thus, [h+] increases until V = Von. This potential lifts the holes to a level with sufficient energy to oxidize water, at which the

Figure 5. Comparison of photocurrent (blue line) and spectral weight UHB of tCTB 1u↑ (circles) and a1g↑ (triangles) and their sum (squares). The potential is measured vs the reversible hydrogen electrode RHE.

recorded under illumination versus the bias potential V. We notice that SCTB have their maximum around 400 mV, whereas SUHB have the maximum at around 500 mV. The sum (gray UHB line) S = tCTB 1u↑ a1g↑ /t2g↓ + eg↓ has the maximum at around 500 mV.

■

DISCUSSION In Figure 5, we show the photocurrent under 1.5 AM illumination and the dark current and, for comparison, the spectral weights SUHB and SCTB of both types of holes, which is basically the concentration of hole states, and their sum S = UHB tCTB 1u↑ a1g↑ /t2g↓ + eg↓. On the basis of the conditions reported in previous work,33 the potential 0 V vs Ag+/AgCl in 0.4 M NaOH is close to the flat-band potential; thus, we assume no hole states at this potential (S = 0). For the potentials of 300 and 500 mV, S increases homogeneously. Inspection of the spectra in Figure 2 and the spectral weight in Figure 5 show that the CTB hole reaches a higher concentration at a lower potential, which supports the suggestion that the CTB hole recombines less at lower potentials, assuming equal generation rates. Following this trend, we attribute the slight enhanced 16873

dx.doi.org/10.1021/jp304254k | J. Phys. Chem. C 2012, 116, 16870−16875

The Journal of Physical Chemistry C

Article

irrespective their different reactivity. The O 2p hole appears and also disappears at a potential 100−200 mV below that of the Fe 3d hole, suggesting that the O 2p hole is more reactive than the Fe 3d hole at the same potential. Surprisingly, the Fe 3d hole, less energetic than the O 2p hole but with a larger spectral weight, is also active for water activation, and there is little difference in their activity.

reaction proceeds with a rate proportional to the free energy of the holesessentially V. In our case, the water oxidation rate is still the limiting factor, and holes are still accumulating at the SCLJ, although [h+] decreases as V increase anodic of Von. We notice this in the number of holes from the NEXAFS spectra at around V = 500 mV (Figure 5). When V is sufficiently anodic, [h+] and thus the concentration of accumulated holes have dropped to 0, because every hole arriving at the SCLJ has sufficient free energy to oxidize water, and the photocurrent increases as w increases. Considering this model, the highest steady-state concentration of holes in the electrode should exist at a potential near Von, which we indeed observe as a large number of holes in Figure 5. At higher potentials, [h+] should be decreasing to 0, which is also confirmed as a vanishing NEXAFS intensity at the pre-edge in Figure 5. This model is corroborated by a study of the photocurrent transient peak heights in H2O.34 The potential where the maximum number of holes is observed is the potential where photocurrent onsets. Because the overall sum of the number of holes detected by the NEXAFS experiment matches well with the expected hole concentration as a function of the applied bias, we can now turn our examination to the behavior of each individual type of hole. As stated before, the CTB hole reaches a higher concentration at a lower potential, which supports the suggestion that the CTB hole has a lower recombination rate. In addition, the fact that the CTB hole also disappears at a potential 100−200 mV below that of the Fe 3d hole suggests that the O 2p hole is more reactive than the Fe 3d hole at the same applied potential. The Fe 3d holes reaches a larger spectral weight as compared to the O 2p holes, which is reflective of the larger number of solar photons that can excite these lower energy holes. Surprisingly, this less energetic Fe 3d hole is also active for water activation, based on its reduction in number and with increasing applied bias over the photocurrent onset potential. Moreover, there seems to be little difference in the activity of the different holes as their decrease is at the same rate (same slope with respect to the increasing applied potential) as the O 2p holes. This is in stark contrast to the previous speculations that the Fe 3d holes should not be active for water oxidation and implies that significant improvement in the performance of hematite photoanodes remains possible by increasing the number of holes that reach the SCLJ through nanostructuring techniques or by reducing bulk recombination by other methods, for example, improving the minority carrier lifetime.35

■

AUTHOR INFORMATION

Corresponding Author

*Tel: +41 58 765 4850. Fax: +41 58 765 4150. E-mail: artur. [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The research leading to these results received funding from the European Community's Sixth Framework Marie Curie International Reintegration Program grant no. 042095 (HiTempEchemX-ray and Electrochemical Studies on Solid Oxide Fuel Cells and Related Materials), Seventh Framework Program Novel Materials for Energy Applications grant no. 227179 (NanoPEC−Nanostructured Photoelectrodes for Energy Conversion), Swiss NSF grants 206021-121306, IZK0Z2-133944, 200021-132126, the European Research Council (ERC-2010AdG 267816 Li-Lo), and Swiss Federal Office of Energy contracts 152316-101883, 153613-102809, and 1476-102691. The ALS is supported by the Director, Office of Science/BES, of the U.S. DoE, No. DE-AC02-05CH11231. The potentiostat was provided by Dr. Elad Pollak, LBNL EETD. J.Z. receives support from the Natural Science Foundation of China (Grant No. 20873128), National Basic Research Program of China (2010CB923302), and the “Hundred Talents Program” of the Chinese Academy of Sciences.

■

REFERENCES

(1) Grätzel, M. Nature 2001, 414, 338−344. (2) Tilley, S. D.; Cornuz, M.; Sivula, K.; Gratzel, M. Angew. Chem., Int. Ed. 2010, 49, 6405−6408. (3) Chen, X. B.; Liu, L.; Yu, P. Y.; Mao, S. S. Science 2011, 331, 746− 750. (4) Khan, S. U. M.; Al-Shahry, M.; Ingler, W. B. Science 2002, 297, 2243−2245. (5) Maeda, K.; Teramura, K.; Lu, D. L.; Takata, T.; Saito, N.; Inoue, Y.; Domen, K. Nature 2006, 440, 295. (6) Sivula, K.; Zboril, R.; Le Formal, F.; Robert, R.; Weidenkaff, A.; Tucek, J.; Frydrych, J.; Gratzel, M. J. Am. Chem. Soc. 2010, 132, 7436− 7444. (7) Zou, Z. G.; Ye, J. H.; Sayama, K.; Arakawa, H. Nature 2001, 414, 625−627. (8) Gerischer, H. The Influence of Surface Orientation and Crystal Imperfections on Photo-electrochemical Reactions at Semiconductor Electrodes. In Photoeffects at Semiconductor-Electrolyte Interfaces; Nozik, A., Ed.; ACS Symposium Series; American Chemical Society: Washington, DC, 1981; Vol. 146, Chapter 1, pp 1−16. (9) Osterloh, F. E. Chem. Mater. 2008, 20, 35−54. (10) Wu, Z. Y.; Gota, S.; Jollet, F.; Pollak, M.; Gautier-Soyer, M.; Natoli, C. R. Phys. Rev. B 1997, 55, 2570−2577. (11) Klahr, B. M.; Hamann, T. W. J. Phys. Chem. C 2011, 115, 8393− 8399. (12) Sivula, K.; Le Formal, L.; Grätzel, M. ChemSusChem 2011, 4, 432−449. (13) Wijayantha, K. G. U.; Saremi-Yarahmadi, S.; Peter, L. M. Phys. Chem. Chem. Phys. 2011, 13, 5264−5270.

■

SUMMARY In summary, this is the first time that an analysis of the electronic structure of a PEC photoanode was carried out online and operando by performing soft X-ray spectroscopy under control of the electrochemical potential. We have directly identified two different electron hole transitions in hematite under PEC operating conditions, which arise upon illumination at anodic bias from around 100 to 900 mV vs an Ag+/AgCl reference electrode and thus validated a long-speculated electronic aspect of hematite. One of the holes found with NEXAFS spectroscopy coincides with a previously identified surface trapped state detected with impedance spectroscopy, {UHB} suggesting that a{1g↑} is a surface state. Whereas the parabolic curvature of the t{CTB} {1u↑} state is reminiscent of the curvature of the depletion layer thickness w. Moreover, in contrast to established perception, the two different holes seem to contribute to photoelectrochemical water oxidation, 16874

dx.doi.org/10.1021/jp304254k | J. Phys. Chem. C 2012, 116, 16870−16875

The Journal of Physical Chemistry C

Article

(14) Schrebler, R. S.; Ballesteros, L.; Burgos, A.; Munoz, E. C.; Grez, P.; Leinen, D.; Martin, F.; Ramos-Barrado, J. R.; Enrique, A.; Dalchiele, E. A. J. Electrochem. Soc. 2011, 158 (8), D500−D505. (15) Dare-Edwards, M. P.; Goodenough, J. B.; Hamnett, A.; Trevellick, P. R. J. Chem. Soc., Faraday Trans. 1983, 1 (79), 2027− 2040. (16) Kennedy, J. H.; Frese, K. W. J. Electrochem. Soc. 1978, 125, 709− 714. (17) Ginley, D. S.; Butler, M. A. Photoeffects at SemiconductorElectrolyte Interfaces; Nozik, A., Ed.; ACS Symposium Series; American Chemical Society: Washington, DC, 1981; Vol. 146, pp 79−101. (18) Shen, W.-M.; Tomkiewicz, M.; Cahen, D. J. Electrochem. Soc. 1986, 133 (1), 112−116. (19) Klahr, B.; Gimenez, S.; Fabregat-Santiago, F.; Hamann, T.; Bisquert, J. J. Am. Chem. Soc. 2012, 134 (9), 4294−4302. (20) Bora, D. K.; et al. J. Phys. Chem. C 2011, 115, 5619−5625. (21) Jiang, P.; Chen, J.-L.; Borondics, F.; Glans, P.-A.; West, M. W.; Chang, C. L. Electrochem. Commun. 2010, 12, 820−822. (22) Chen, J.-L.; Wang, W. C.; Glans, P.-A.; West, M. W.; Vayssieres, L.; Guo, J.-H.In-situ cell for soft X-ray spectroscopic study of artificial photosynthesis. J. Electron Spectrosc. Relat. Phenom. 2012, to be published. (23) Sivula, K.; Le Formal, F.; Graetzel, M. Chem. Mater. 2009, 21, 2862−2867. (24) Bak, T.; Nowotny, J.; Rekas, M.; Sorrell, C. C. Int. J. Hydrogen Energy 2002, 27, 991−1022. (25) Colliex, C.; Manoubi, T.; Ortiz, C. Phys. Rev. B 1991, 44, 11402−11411. (26) Pollak, M.; Gautier, M.; Thromat, N.; Gota, S.; Mackrodt, W. C.; Saunders, V. R. Nucl. Instrum. Methods Phys. Res., Sect. B 1995, 97, 383−386. (27) Park, T.-J.; Sambasivan, S.; Fischer, D. A.; Yoon, W.-S.; Misewich, J. A.; Wong, S. S. J. Phys. Chem. C 2008, 112, 10359−10369. (28) de Groot, F. M. F.; Grioni, M.; Fuggle, J. C.; Ghijsen, J.; Sawatzky, G. A.; Petersen, H. Phys. Rev. B 1989, 40, 5715−5723. (29) Sarma, D. D.; et al. Phys. Rev. B 1994, 49, 14238−14243. (30) Braun, A.; et al. Appl. Phys. Lett. 2009, 94 (202102), 1−3. (31) Chen, C. T.; et al. Phys. Rev. Lett. 1991, 66, 104−107. (32) Tossell, J. A.; Vaughan, D. J.; Johnson, K. H. Am. Mineral. 1974, 59, 319−334. (33) Cesar, I.; Sivula, K.; Kay, A.; Zboril, R.; Grätzel, M. J. Phys. Chem. C 2009, 113, 772−782. (34) Dotan, H.; Sivula, K.; Graetzel, M.; Rothschild, A.; Warren, S. C. Energy Environ. Sci. 2011, 4, 958−964. (35) Kleiman-Shwarsctein, A.; Huda, M. N.; Walsh, A.; Yan, Y. F.; Stucky, G. D.; Hu, Y. S.; Al-Jassim, M. M.; McFarland, E. W. Chem. Mater. 2010, 22, 510−517.

16875

dx.doi.org/10.1021/jp304254k | J. Phys. Chem. C 2012, 116, 16870−16875