The Origin of Slow Carrier Transport in BiVO4 Thin Film Photoanodes

Jul 31, 2013 - X-ray diffraction confirms that the films have the desired monoclinic Scheelite structure (Figure S1, Supporting Information). Further ...
6 downloads 7 Views 870KB Size
Download PDF
Letter pubs.acs.org/JPCL

The Origin of Slow Carrier Transport in BiVO4 Thin Film Photoanodes: A Time-Resolved Microwave Conductivity Study Fatwa F. Abdi,*,† Tom J. Savenije,‡ Matthias M. May,§ Bernard Dam,† and Roel van de Krol§ †

Materials for Energy Conversion and Storage (MECS), Department of Chemical Engineering, Delft University of Technology, P.O. Box 5045, 2600 GA Delft, The Netherlands ‡ Opto-electronic Materials (OM), Department of Chemical Engineering, Delft University of Technology, P.O. Box 5045, 2600 GA Delft, The Netherlands § Helmholtz-Zentrum Berlin für Materialien und Energie Gmbh, Institute for Solar Fuels, Hahn-Meitner-Platz 1, 14109 Berlin, Germany S Supporting Information *

ABSTRACT: We unravel for the first time the origin of the poor carrier transport properties of BiVO4, a promising metal oxide photoanode for solar water splitting. Time-resolved microwave conductivity (TRMC) measurements reveal an (extrapolated) carrier mobility of ∼4 × 10−2 cm2 V−1 s−1 for undoped BiVO4 under ∼1 sun illumination conditions, which is unusually low for a photoanode material. The poor carrier mobility is compensated by an unexpectedly long carrier lifetime of 40 ns. This translates to a relatively long diffusion length of 70 nm, consistent with the high quantum efficiencies reported for BiVO4 photoanodes. Tungsten (W) doping is found to strongly decrease the carrier mobility by introducing intermediate-depth donor defects as carrier traps. At the same time, the increased carrier density improves the overall photoresponse, which confirms that bulk electronic conductivity is one of the main performance bottlenecks for BiVO4. SECTION: Energy Conversion and Storage; Energy and Charge Transport

P

nation.9 Tungsten (W) and molybdenum (Mo) have been used as donor-type dopants to improve the overall electronic conductivity of the material. Carrier transport, however, remains an issue, as suggested by the modest carrier separation efficiencies.11,12 Very recently, we showed that the carrier separation efficiency can be further improved by creating a distributed n+−n homojunction throughout the entire thickness of BiVO4 by introducing a gradient in the tungsten dopant concentration.13 This has resulted in an AM1.5 photocurrent of 3.6 mA/cm2 at 1.23 vs RHE (reversible hydrogen electrode), which has set a new performance benchmark for metal oxide photoanodes. With a maximum theoretical AM1.5 photocurrent of 7.5 mA/cm2,11 there is still ample room for improvement. However, further progress is hampered by our limited understanding of the carrier dynamics in BiVO4; the underlying cause of the slow carrier transport in BiVO4 is unclear, and quantitative data on, e.g., carrier mobility and lifetime is still lacking. In this paper we investigate the nature of the carrier transport in BiVO 4 using time-resolved microwave conductivity (TRMC). TRMC is a contactless technique in which the

hotoelectrochemical (PEC) water splitting provides an attractive way of converting solar energy to chemical energy by forming hydrogen and oxygen. Hydrogen can be used directly as a fuel, or it can serve as a feedstock material in the formation of hydrocarbons by Fischer−Tropsch processes.1 PEC water splitting takes advantage of semiconductor materials to absorb sunlight and perform the water splitting reactions at ambient temperature and pressure. Among the different classes of materials, metal oxides have attracted a lot of interest due to their general stability and low cost. One of the most promising metal oxide photoelectrodes is bismuth vanadate (BiVO4). Monoclinic BiVO4 has a bandgap of 2.4 eV and absorbs UV and visible light up to 520 nm.2 The material is stable in aqueous solutions with a pH between 3 and 11.3 The valence band lies significantly positive of the water oxidation potential, and oxygen evolution is readily observed. The conduction band is reported to be close to the H2/H+ redox level,4,5 but direct hydrogen evolution has not been observedpresumably due to the lack of sufficient overpotential. Although relatively low effective carrier masses have been predicted for the monoclinic phase due to extensive mixing of the Bi 6s orbitals with both the valence band O 2p and conduction band V 3d orbitals,6−8 experimental studies show that the overall performance of BiVO4 is limited by poor carrier transport properties.9,10 At light intensities as high as 1 sun (AM1.5, 100 mW cm−2), this limitation causes extensive carrier accumulation and recombi© XXXX American Chemical Society

Received: June 27, 2013 Accepted: July 31, 2013

2752

dx.doi.org/10.1021/jz4013257 | J. Phys. Chem. Lett. 2013, 4, 2752−2757

The Journal of Physical Chemistry Letters

Letter

respectively. The resulting microwave conductance is given by the product of the absorbance-normalized quantum yield (ϕ) and the combined mobility of the charge carriers in the system (Σμ). The minimum mobility of the carriers can be obtained from the peak of the measured signal (maximum ϕΣμ), and the lifetime of the carriers can be deduced from the decay of the signal.25 The maximum ϕΣμ of undoped BiVO4 appears to be a function of the number of incident photons per laser pulse. In undoped BiVO4 (Figure 2a), a ϕΣμ value of ∼2.75 × 10−2 cm2 V−1 s−1 is obtained at 1011 photons cm−2 pulse−1, which is the lowest intensity that still gives a good signal-to-noise ratio. The value of ϕΣμ decreases as the laser intensity increases, which is attributed to fast nongeminate electron−hole recombination. Based on the slope of the logarithmic plot (ϕΣμ ∼ I0α−1, where α is the reciprocal of the order of the electron−hole recombination process), we deduce that a combination of first and second order recombination takes place in this undoped BiVO4 thin film (α = 0.7). Assuming for the moment that the same recombination mechanism also occurs under 1 sun illumination conditions (AM1.5, 100 mW cm−2, ∼109 photons cm−2 pulse−1),26 we extrapolate the ϕΣμ in Figure 2a to arrive at a value of ∼4.4 × 10−2 cm2 V−1 s−1. Since the internal quantum efficiency of our BiVO4 is close to 100%,9 ϕ ∼ 1 and the ϕΣμ value can be taken as the carrier mobility. A carrier mobility of 4.4 × 10−2 cm2 V−1 s−1 for BiVO4 is a few orders of magnitude lower than for typical semiconducting photoelectrodes for PEC water splitting.27−32 This observation thus supports the poor carrier transport properties and the strong dependence of the quantum efficiency of BiVO4 on the light intensity that we reported previously.9 We emphasize that one should be careful when interpreting carrier mobility values measured by TRMC. First and foremost, second-order recombination processes tend to become less pronounced at lower light intensities. This would cause the curve in Figure 2a to flatten, and result in lower mobility values under actual 1-sun conditions. The reported value should thus be considered as an upper limit. At the same time, the value should be considered as a lower limit since the actual quantum yield, ϕ, is always less than one. A final consideration is the frequency of the measurement, and whether or not the measured mobility is equal to the bulk (microscopic) mobility or lower due to scattering at grain boundaries. Based on the high frequency and low mobility, we do not believe grain boundaries play a crucial role in this case. However, more extensive studies on BiVO4 single crystals, or with higherfrequency techniques such as THz spectroscopy, are needed to clarify this issue. Although the carrier mobility is low, Figure 1 indicates that the lifetime of the carriers is relatively long. An exponential carrier lifetime of ∼40 ns is estimated for undoped BiVO4, using fitting procedure previously reported in ref 25 to the microwave conductance transient shown in Figure 1 and taking into account the response time of the cavity. Based on the carrier lifetime, τ, and the carrier mobility, μ, we estimate a carrier diffusion length L, given by L = (Dτ)1/2 with D = kTμ/e, of ∼70 nm. This suggests that despite their slow nature, the carriers are still able to reach the back and front interfaces in a thin film. This is consistent with the appreciable carrier separation efficiency (up to 40%) we observe in 100 nm spray pyrolyzed BiVO4 thin film photoanodes.11 Table 1 summarizes the carrier lifetimes and mobilities of other promising photoelectrode materials, such as Fe2O3, WO3,

change in reflected microwave power from a sample surface is measured after exciting the sample with a nanosecond pulsed laser.14−16 Based on this measurement, the sample’s photoconductance―and essentially its carrier mobility―can be determined. The technique is widely applied in analyzing TiO2based systems, such as dye-sensitized solar cells and photocatalytically active nanoparticles,16−18 amorphous and microcrystalline silicon,19,20 organic solar cells,21,22 and carbon nanotubes;23,24 however, few TRMC studies have yet been reported on photoelectrode materials for PEC water splitting. Here, we report on a TRMC study on undoped and 1% Wdoped BiVO4 photoanode thin films. The measurements reveal a carrier mobility value for BiVO4, which is a few orders of magnitude lower than for typical metal oxides (e.g., TiO2, WO3), confirming the poor carrier transport nature in BiVO4. We show that donor-doping with tungsten increases the material’s conductivity, but also introduces trap states. Based on these findings, the PEC behavior of both undoped and Wdoped BiVO4 can be explained, showing that TRMC is a straightforward yet informative tool to analyze metal oxide photoelectrodes for PEC water splitting. BiVO4 thin films (undoped and 1% W-doped) are deposited on quartz substrates by spray pyrolysis. X-ray diffraction confirms that the films have the desired monoclinic Scheelite structure (Figure S1, Supporting Information). Further details on the synthesis and experimental procedures (TRMC setup, PEC measurement, etc) can be found in the Supporting Information. Upon tungsten doping, W6+ ions substitute for V5+, and the resulting net positive charge is compensated by free electrons. The incorporation of tungsten into the BiVO4 lattice can be described using standard Krö g er−Vink notation:11 2BiVO4

x 2WO3 + Bi 2O3 ⎯⎯⎯⎯⎯⎯→ 2Bi Bi + 2W •V + 8OOx +

+ 2e′

1 O2 (g) 2 (1)

Bi2O3 is added as a codopant in this reaction equation to reflect the fact that the 1:1 stoichiometry of Bi and (V+W) sites is preserved. This prevents the formation of undesired defects in the form of metal ion vacancies, and gives an overall stoichiometry of Bi(V1−xWx)O4. Figure 1 shows the microwave conductance transients of undoped and 1% W-doped BiVO4 obtained after a 3.5 ns laser pulse of 450 nm with an intensity of 2.5 × 1013 photons cm−2 pulse−1. The photon energy used is well above the bandgap of BiVO4; hence, the optical excitation leads to the formation of free electrons and holes in the conduction and valence bands,

Figure 1. Time resolved microwave conductance signals recorded for undoped and 1% W-doped BiVO4 thin films using a 450 nm 3.5 ns laser pulse of 2.5 × 1013 photons cm−2. 2753

dx.doi.org/10.1021/jz4013257 | J. Phys. Chem. Lett. 2013, 4, 2752−2757

The Journal of Physical Chemistry Letters

Letter

Figure 2. Maximum observed TRMC signals as a function of incident photons per laser pulse for (a) undoped BiVO4 on excitation at 450 nm, and (b) 1% W-doped BiVO4 on excitation at 300 (blue square), 450 (green circle), and 475 nm (red triangle). AM1.5 illumination is equivalent to an incident number of photons of ∼109 cm−2 pulse−1.

Table 1. Comparison of Carrier Mobility, Lifetime, and Diffusion Length of Several Promising Metal Oxides

a

photoelectrode material

carrier mobility, μ (cm2 V−1 s−1)

carrier lifetime, τ

diff. length, L (nm)

methods/references

Fe2O3 WO3 Cu2O BiVO4

0.5a 10 6 0.044

3 ps 1−9 nsa 40 psa 40 ns

2−4 150−500 25 70a

transient absorption spectroscopy/33−35 space-charge model/36 THz spectroscopy/37 TRMC/this paper

These values are calculated from the other two reported properties.

is higher than the number of absorbed photons. In this regime, the TRMC signal increases with light intensity since the chance of an electron being trapped decreases as a larger fraction of the traps gets filled. When all the traps are filled, no further trapping occurs and the number of free carriers (i.e., ϕ in the ϕΣμ signal) will thus reach its highest value. When the light intensity increases further, the value of ϕΣμ decreases again due to the same higher order recombination observed in undoped BiVO4. Since evidence of carrier trapping is only observed in the TRMC signal of the W-doped samples, it is reasonable to assume that the nature of the traps is related to the tungsten dopant or an associated defect. Indeed, a light intensity of 5 × 1014 photons cm−2 pulse−1 in these 200 nm thick films corresponds to a carrier concentration of 2.5 × 1019 cm−3, which is of the order of the reported dopant concentration of W-doped BiVO4.41,42 Structural, morphological, or surfacechemical explanations for the decrease in ϕΣμ are not very likely, since no significant differences in the X-ray diffractograms, UV−vis absorption spectra, scanning electron micrographs, or X-ray photoemission spectra are observed between the undoped and 1% W-doped BiVO4 (Figures S1−4, Supporting Information). To further confirm the presence of electron traps in 1% Wdoped BiVO4, the ϕΣμ values are also measured at different laser excitation wavelengths of 300 and 475 nm (Figure 2b). Since the absorption depth (defined as the reciprocal of the absorption coefficient, α) increases with longer wavelengths,8 and assuming that the traps are present throughout the entire film, one would also expect the maximum of the ϕΣμ curve to shift to higher photon fluxes with longer wavelengths. This is indeed observed in Figure 2b; the peak value of ϕΣμ is at 1014 and 1015 photons cm−2 pulse−1 for excitation wavelengths of 300 and 475 nm, respectively. Since the trap levels are unambiguously related to the presence of tungsten and spatially distributed throughout the entire film, we hypothesize that the electron traps are ionized (i.e., empty) tungsten donor states. Trapping of an electron in this empty donor state can be described as follows:

and Cu2O. Although these parameters have been measured under widely varying conditions and should perhaps not be directly compared, one general trend is clear: the other oxides show 1−2 orders of magnitude higher carrier mobilities, while their lifetimes are 1−3 orders of magnitude shorter.33−37 From this, it appears that the relatively high quantum efficiencies reported for undoped BiVO49,39 are primarily due to the long lifetimes of the photogenerated charge carriers in the material. The microwave conductance transient changes significantly upon donor-doping BiVO4 with tungsten (W), as shown in Figure 1. At a laser pulse intensity of 2.5 × 1013 photons cm−2 pulse−1, the maximum ϕΣμ for 1% W-doped BiVO4 is only ∼2.2 × 10−4 cm2 V−1 s−1, a factor of 20 lower than what is observed for undoped BiVO4. Moreover, the lifetime of the carriers is reduced to ∼5 ns, and the TRMC signal is completely gone after ∼50 ns. Despite the significant decreases in both lifetime and mobility of the photogenerated charge carriers, W-doped BiVO4 samples consistently show higher photocurrents than undoped samples.11,12,38−40 The only way we can explain these seemingly contradictory results is by the enhanced electronic conductivity of the material due the tungsten dopant. This is indeed consistent with our earlier assertion that poor electron transport limits the performance of BiVO4.9,11,39 To understand the cause of the lower ϕΣμ values in Wdoped BiVO4, the TRMC signal of 1% W-doped BiVO4 was measured at different laser pulse intensities. The resulting plot of ϕΣμ as a function of incident photons per pulse is shown in Figure 2b. Interestingly, the behavior is very different from what is observed for undoped BiVO4 (Figure 2a). At an excitation wavelength of 450 nm, the ϕΣμ initially increases with light intensity until a clear maximum of ∼4 × 10−4 cm2 V−1 s−1 is reached at 5 × 1014 photons cm−2 pulse−1. A similar dependence of ϕΣμ on the light intensity has been observed in compact and mesoporous TiO2 films,15,16,18 where it is attributed to the competition between electron trap filling and higher-order recombination. We believe the same explanation applies in the case of our W-doped BiVO4. Below a flux of 5 × 1014 photons cm−2 pulse−1, the concentration of electron traps 2754

dx.doi.org/10.1021/jz4013257 | J. Phys. Chem. Lett. 2013, 4, 2752−2757

The Journal of Physical Chemistry Letters BiVO4

W •V + e′ ⎯⎯⎯⎯⎯→ W xV

Letter

wavelength, the absorption depth (1/α) ∼ 100 nm and most of the light is absorbed near one of the interfaces (Figure S5, Supporting Information). The results are shown in Figure 3a. We find that front-side illumination results in ∼30% lower ϕΣμ values compared to back-side illumination. This is indeed consistent with a larger concentration of electron traps at the surface of the material, as illustrated in Scheme 1. No differences between front- and back-side illumination are observed in the ϕΣμ values for undoped BiVO4 (Figure 3b). This is indeed expected, since no electron traps are present in the undoped material. Finally, we relate the knowledge obtained in this study to the actual PEC performance of the BiVO4 photoanodes under continuous illumination. Figure 4 shows the AM1.5 photo-

(2)

In order for this trapping process to be effective, the donor states cannot be too shallow and would have to be located at more than ∼2kT below the conduction band edge. This implies that these intermediate-depth donor states are only partially filled, and that the Fermi level is located within the density-ofstates (DOS) of the trap level. At the surface of the material, however, the donor level is expected to be mostly empty (ionized) due to the presence of a depletion layer. This layer usually forms upon hydroxylation of the metal oxide surface when exposed to air. No depletion layer is expected to form at the back contact due to the passivating nature of the SiO2 substrate. The band diagram shown in Scheme 1 illustrates the Scheme 1. Schematic Energy Band Diagram of TungstenDoped BiVO4, Showing the Presence of Defect States That Could Act as Carrier Trapsa

Figure 4. AM1.5 photocurrent−voltage curve of undoped and 1% Wdoped BiVO4 photoanodes measured under three-electrode configuration in a 0.1 M KPi (pH∼7) electrolyte solution at a scan rate of 50 mV s−1. Both photoanodes were modified with a 30 nm thick cobalt phosphate water oxidation catalyst.

current−voltage (J−V) curve for an undoped and a 1% Wdoped BiVO4 photoanode. Both photoanodes are modified with 30 nm thick electrodeposited cobalt phosphate (Co−Pi) catalyst to ensure fast water oxidation kinetics.11 At V > 1.5 V vs RHE, the photocurrent of both photoanodes approaches a plateau. The photocurrent plateau of the 1% W-doped BiVO4 is ∼30% higher than that of the undoped BiVO4. This is consistent with the fact that poor electron transport limits the photocurrent,9 and that W-doping can solve this by increasing the background concentration of free electrons.11 At potentials below 1 V vs RHE, however, the situation is reversed. Moreover, the photocurrent onset potential for the 1% W-doped BiVO4 is ∼200 mV more positive than for the undoped BiVO4. The lower photocurrents at modest applied potentials are tentatively attributed to trapping of the photogenerated electrons in the intermediate-level donor states, followed by recombination. The comparatively steep increase of

a Black circles represent filled traps (WxV), and white circles represent empty tramps (W•V). CB is conduction band, VB is valence band, and EF is Fermi energy level.

partial occupation of the intermediate-depth donor states. The empty states (open circles) represent active electron traps, and it is clear from the diagram that their concentration is highest near the surface. To test this model, we compare the ϕΣμ values for 1% Wdoped BiVO4 under front- and back-side illumination. To ensure a sufficiently sharp photon intensity profile in the film, an excitation wavelength of 300 nm was chosen; at this

Figure 3. Maximum observed TRMC signals obtained for (a) 1% W-doped and (b) undoped BiVO4 films under front- and back-side laser pulse illumination (λ = 300 nm). 2755

dx.doi.org/10.1021/jz4013257 | J. Phys. Chem. Lett. 2013, 4, 2752−2757

The Journal of Physical Chemistry Letters

Letter

Oxidation under Visible Light Irradiation. Electrochem. Solid State Lett. 2008, 11, H160−H163. (5) Chen, S.; Wang, L. W. Thermodynamic Oxidation and Reduction Potentials of Photocatalytic Semiconductors in Aqueous Solution. Chem. Mater. 2012, 24, 3659−3666. (6) Kudo, A.; Omori, K.; Kato, H. A Novel Aqueous Process for Preparation of Crystal Form-Controlled and Highly Crystalline BiVO4 Powder from Layered Vanadates at Room Temperature and Its Photocatalytic and Photophysical Properties. J. Am. Chem. Soc. 1999, 121, 11459−11467. (7) Payne, D.; Robinson, M.; Egdell, R.; Walsh, A.; McNulty, J.; Smith, K.; Piper, L. The Nature of Electron Lone Pairs in BiVO4. Appl. Phys. Lett. 2011, 98, 212110. (8) Zhao, Z. Y.; Li, Z. S.; Zou, Z. G. Electronic Structure and Optical Properties of Monoclinic Clinobisvanite BiVO4. Phys. Chem. Chem. Phys. 2011, 13, 4746−4753. (9) Abdi, F. F.; van de Krol, R. Nature and Light Dependence of Bulk Recombination in Co-Pi-Catalyzed BiVO4 Photoanodes. J. Phys. Chem. C 2012, 116, 9398−9404. (10) Zhang, K.; Shi, X. J.; Kim, J. K.; Park, J. H. Photoelectrochemical Cells with Tungsten Trioxide/Mo-Doped BiVO4 Bilayers. Phys. Chem. Chem. Phys. 2012, 14, 11119−11124. (11) Abdi, F. F.; Firet, N.; van de Krol, R. Efficient BiVO4 Thin Film Photoanodes Modified with Cobalt Phosphate Catalyst and WDoping. ChemCatChem 2013, 5, 490−496. (12) Zhong, D. K.; Choi, S.; Gamelin, D. R. Near-Complete Suppression of Surface Recombination in Solar Photoelectrolysis by “Co-Pi” Catalyst-Modified W:BiVO4. J. Am. Chem. Soc. 2011, 133, 18370−18377. (13) Abdi, F. F.; Han, L.; Smets, A. H. M.; Zeman, M.; Dam, B.; van de Krol, R. Efficient Solar Water Splitting by Enhanced Charge Separation in a Bismuth Vanadate−Silicon Tandem Photoelectrode. Nat. Commun. 2013, 4 (2195), 1−7. (14) Infelta, P. P.; Haas, M. P. D.; Warman, J. M. Study of Transient Conductivity of Pulse Irradiated Dielectric Liquids on A Nanosecond Timescale Using Microwaves. Radiat. Phys. Chem. 1977, 10, 353−365. (15) Savenije, T. J.; de Haas, M. P.; Warman, J. M. The Yield and Mobility of Charge Carriers in Smooth and Nanoporous TiO2 Films. Z. Phys. Chem. (Muenchen, Ger.) 1999, 212, 201−206. (16) Kroeze, J. E.; Savenije, T. J.; Warman, J. M. Electrodeless Determination of the Trap Density, Decay Kinetics, and Charge Separation Efficiency of Dye-Sensitized Nanocrystalline TiO2. J. Am. Chem. Soc. 2004, 126, 7608−7618. (17) Carneiro, J. T.; Savenije, T. J.; Mul, G. Experimental Evidence for Electron Localization on Au upon Photo-Activation of Au/Anatase Catalysts. Phys. Chem. Chem. Phys. 2009, 11, 2708−2714. (18) Savenije, T. J.; Huijser, A.; Vermeulen, M. J.; Katoh, R. Charge Carrier Dynamics in TiO2 Nanoparticles at Various Temperatures. Chem. Phys. Lett. 2008, 461, 93−96. (19) Kunst, M.; Wunsch, F.; von Aichberger, S. Recombination at High Charge Carrier Concentrations in a-Si:H Films. Thin Solid Films 2001, 383, 274−276. (20) Vanderhaghen, R.; Kasouit, S.; Brenot, R.; Chu, V.; Conde, J. P.; Liu, F.; de Martino, A.; Cabarrocas, P. R. I. Electronic Transport in Microcrystalline Silicon Controlled by Trapping and Intra-Grain Mobility. J. Non-Cryst. Solids 2002, 299, 365−369. (21) Baumann, A.; Savenije, T. J.; Murthy, D. H.; Heeney, M.; Dyakonov, V.; Deibel, C. Influence of Phase Segregation on Recombination Dynamics in Organic Bulk-Heterojunction Solar Cells. Adv. Funct. Mater. 2011, 21, 1687−1692. (22) Moehl, T.; Kytin, V. G.; Bisquert, J.; Kunst, M.; Bolink, H. J.; Garcia-Belmonte, G. Relaxation of Photogenerated Carriers in P3HT:PCBM Organic Blends. ChemSusChem 2009, 2, 314−320. (23) Ferguson, A. J.; Blackburn, J. L.; Holt, J. M.; Kopidakis, N.; Tenent, R. C.; Barnes, T. M.; Heben, M. J.; Rumbles, G. Photoinduced Energy and Charge Transfer in P3HT:SWNT Composites. J. Phys. Chem. Lett. 2010, 1, 2406−2411. (24) Umeyama, T.; Tezuka, N.; Seki, S.; Matano, Y.; Nishi, M.; Hirao, K.; Lehtivuori, H.; Tkachenko, N. V.; Lemmetyinen, H.; Nakao,

the photocurrent at more positive potentials might be a result of field-induced emission from these electron traps. One possibility is a Poole-Frenkel-type emission process, although it seems somewhat doubtful that a sufficiently strong electric field could develop in the bulk of the highly doped W:BiVO4. Another possibility is that under sufficiently large band bending, trapped electrons in the (narrow) depletion layer tunnel to nearby empty conduction band states in the bulk of the material. Further elucidation of the exact nature of these processes is beyond the scope of this study, and calls for further investigation. In conclusion, our time-resolved microwave conductivity study quantitatively confirms previous reports of poor carrier transport in BiVO4 photoanodes. Undoped BiVO4 shows an extrapolated carrier mobility of ∼4 × 10−2 cm2 V−1 s−1 under ∼1 sun illumination conditions, which is at least 1−2 orders of magnitude lower than in typical semiconductor photoelectrodes for water splitting. Despite the low mobility, high quantum efficiencies can nevertheless be achieved in BiVO4 due to the long lifetime of the photogenerated charge carriers (∼40 ns), which leads to relatively long carrier diffusion lengths. The long lifetime implies that there are few intrinsic recombination centers, from which we conclude that BiVO4 is a highly “defect-tolerant” material. Doping with tungsten significantly improves the photocurrent of the material, despite the fact that it decreases the carrier lifetime and mobility with ∼1 order of magnitude due to (de)trapping in intermediatelevel W donor states. This confirms that bulk electronic conductivity is the main performance bottleneck for BiVO4. Finally, these results clearly demonstrate the rich and complex nature of the carrier dynamics in BiVO4, and illustrate the need for further studies to complete our understanding of the BiVO4 photoanode.

■

ASSOCIATED CONTENT

S Supporting Information *

Experimental methods and Figures S1−5. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We gratefully acknowledge the European Commission’s Framework Project 7 (NanoPEC, Project 227179) for financial support of this work.

■

REFERENCES

(1) Haije, W.; Geerlings, H. Efficient Production of Solar Fuel Using Existing Large Scale Production Technologies. Environ. Sci. Technol. 2011, 45, 8609−8610. (2) Tokunaga, S.; Kato, H.; Kudo, A. Selective Preparation of Monoclinic and Tetragonal BiVO4 with Scheelite Structure and Their Photocatalytic Properties. Chem. Mater. 2001, 13, 4624−4628. (3) Abdi, F. F.; Firet, N.; Dabirian, A.; van de Krol, R. Spraydeposited Co-Pi Catalyzed BiVO4: a low-cost route towards highly efficient photoanodes. MRS Online Proceedings Library 2012, 1446, DOI: 10.1557/opl.2012.811. (4) Chatchai, P.; Murakami, Y.; Kishioka, S. Y.; Nosaka, A. Y.; Nosaka, Y. FTO/SnO2/BiVO4 Composite Photoelectrode for Water 2756

dx.doi.org/10.1021/jz4013257 | J. Phys. Chem. Lett. 2013, 4, 2752−2757

The Journal of Physical Chemistry Letters

Letter

Y.; Sakaki, S.; Imahori, H. Selective Formation and Efficient Photocurrent Generation of [70]Fullerene−Single-Walled Carbon Nanotube Composites. Adv. Mater. 2010, 22, 1767−1770. (25) Savenije, T. J.; Van Veenendaal, P. A. T. T.; de Haas, M. P.; Warman, J. M.; Schropp, R. E. I. Spatially Resolved Photoconductive Properties of Profiled Polycrystalline Silicon Thin Films. J. Appl. Phys. 2002, 91, 5671−5676. (26) AM1.5 illumination level of ∼1.34 × 109 cm−2 is estimated by integrating the number of incident photons from a sunlight and multiplying with the laser pulse width of 3.5 ns. (27) Berak, J. M.; Sienko, M. J. Effect of Oxygen-Deficiency on Electrical Transport Properties of Tungsten Trioxide Crystals. J. Solid State Chem. 1970, 2, 109−133. (28) Bosman, A. J.; Vandaal, H. J. Small-Polaron Versus Band Conduction in Some Transition-Metal Oxides. Adv. Phys. 1970, 19, 1− &. (29) Brus, L. Electron Electron and Electron-Hole Interactions in Small Semiconductor Crystallites - The Size Dependence of the Lowest Excited Electronic State. J. Chem. Phys. 1984, 80, 4403. (30) Jeong, B. S.; Norton, D. P.; Budai, J. D. Conductivity in Transparent Anatase TiO2 Films Epitaxially Grown by Reactive Sputtering Deposition. Solid-State Electron. 2003, 47, 2275−2278. (31) Matsuzaki, K.; Nomura, K.; Yanagi, H.; Kamiya, T.; Hirano, M.; Hosono, H. Epitaxial Growth of High Mobility Cu2O Thin Films and Application to p-Channel Thin Film Transistor. Appl. Phys. Lett. 2008, 93, 202107. (32) Sze, S. M. Physics of Semiconductor Devices; John Wiley & Sons: New York, 1981. (33) Joly, A. G.; Williams, J. R.; Chambers, S. A.; Xiong, G.; Hess, W. P.; Laman, D. M. Carrier Dynamics in α-Fe2O3 (0001) Thin Films and Single Crystals Probed by Femtosecond Transient Absorption and Reflectivity. J. Appl. Phys. 2006, 99, 053521. (34) Cherepy, N. J.; Liston, D. B.; Lovejoy, J. A.; Deng, H. M.; Zhang, J. Z. Ultrafast Studies of Photoexcited Electron Dynamics in γand α-Fe2O3 Semiconductor Nanoparticles. J. Phys. Chem. B 1998, 102, 770−776. (35) Kennedy, J. H.; Frese, K. W., Jr. Photoxidation of Water at αFe2O3 Electrodes. J. Electrochem. Soc. 1978, 125, 709−714. (36) Butler, M. A. Photoelectrolysis and Physical Properties of the Semiconducting Electrode WO3. J. Appl. Phys. 1977, 48, 1914−1920. (37) Paracchino, A.; Brauer, J. C.; Moser, J. E.; Thimsen, E.; Grätzel, M. Synthesis and Characterization of High-Photoactivity Electrodeposited Cu2O Solar Absorber by Photoelectrochemistry and Ultrafast Spectroscopy. J. Phys. Chem. C 2012, 116, 7341−7350. (38) Berglund, S. P.; Rettie, A. J. E.; Hoang, S.; Mullins, C. B. Incorporation of Mo and W into Nanostructured BiVO4 Films for Efficient Photoelectrochemical Water Oxidation. Phys. Chem. Chem. Phys. 2012, 14, 7065−7075. (39) Liang, Y. Q.; Tsubota, T.; Mooij, L. P. A.; van de Krol, R. Highly Improved Quantum Efficiencies for Thin Film BiVO4 Photoanodes. J. Phys. Chem. C 2011, 115, 17594−17598. (40) Ye, H.; Park, H. S.; Bard, A. J. Screening of Electrocatalysts for Photoelectrochemical Water Oxidation on W-Doped BiVO4 Photocatalysts by Scanning Electrochemical Microscopy. J. Phys. Chem. C 2011, 115, 12464−12470. (41) Ye, H.; Lee, J.; Jang, J. S.; Bard, A. J. Rapid Screening of BiVO4Based Photocatalysts by Scanning Electrochemical Microscopy (SECM) and Studies of Their Photoelectrochemical Properties. J. Phys. Chem. C 2010, 114, 13322−13328. (42) Li, M. T.; Zhao, L. A.; Guo, L. J. Preparation and Photoelectrochemical Study of BiVO4 Thin Films Deposited by Ultrasonic Spray Pyrolysis. Int. J. Hydrogen Energy 2010, 35, 7127− 7133.

2757

dx.doi.org/10.1021/jz4013257 | J. Phys. Chem. Lett. 2013, 4, 2752−2757