21890
J. Phys. Chem. B 2006, 110, 21890-21898
Room-Temperature Preparation of Nanocrystalline TiO2 Films and the Influence of Surface Properties on Dye-Sensitized Solar Energy Conversion Dongshe Zhang, Jonathan A. Downing, Fritz J. Knorr, and Jeanne L. McHale* Department of Chemistry, Washington State UniVersity, Box 644630, Pullman, Washington 99164-4630 ReceiVed: June 29, 2006; In Final Form: August 31, 2006
An extremely easy method is presented for producing surfactant-free films of nanocrystalline TiO2 at room temperature with excellent mechanical stability when deposited on glass or plastic electrodes for dye-sensitized solar energy conversion. Prolonged magnetic stirring of commercial TiO2 nanoparticles (Degussa P25) in either ethanol or water results in highly homogeneous dispersions which are used to prepare TiO2 films with surface properties which depend on the solvent used for dispersing the particles, even after sintering. The optical and mechanical properties of films cast from ethanol and water dispersions are compared, and differences in the extent of surface defects and dye binding are observed. Optical absorption, photoluminescence, and resonance Raman spectra of TiO2 films sensitized with Ru(4,4′-dicarboxylic acid-2,2′-bipyridine)2(NCS)2 (“N3”) reveal that the electronic coupling of the dye and semiconductor depends on the surface structure of the film which varies with film preparation. Current-voltage data for illuminated and dark dye-sensitized solar cells are obtained as a function of film preparation, and results are compared to spectroscopic data in order to interpret the microscopic basis for variations in solar cell performance, especially with regard to sintered versus unsintered TiO2 films. The results suggest that surface traps associated with oxygen vacancies play a critical role in determining the efficiency of dye-sensitized solar energy conversion through their influence on the binding and electronic coupling of the dye to the semiconductor.
1. Introduction The search for cheap and environmentally friendly energy sources has focused much attention on dye-sensitized solar energy cells (DSSCs) based on nanocrystalline TiO2 films sensitized by chromophores which inject electrons into the semiconductor from excited electronic states.1-5 Optimal lightharvesting properties of semiconductor films coated with a monolayer of dye sensitizer are dependent on the huge surfacearea-to-volume ratio afforded by nanoporous films containing TiO2 nanocrystals. As a result, the performance of DSSCs is strongly dependent on the surface chemistry and defect structure of nanocrystalline TiO2 which can influence dye adsorption, interfacial electron transfer, and carrier transport. Defects resulting from undercoordinated Ti4+ ions, oxygen vacancies, chemisorbed surface species, grain boundaries, etc. result in localized intraband gap states which can serve as electron traps in nanocrystalline TiO2.6-8 Amorphous TiO2 nanoparticles, representing a higher level of defect states, possess a characteristic exponential distribution of midband gap states (“band tails”) and exhibit different kinetics of electron injection and recombination than nanocrystalline semiconductors.9,10 Surface trap states exert an enormous influence on electron transport and (undesirable) recombination11-15 of injected electrons with oxidized dye or redox mediator. Despite the potential for trap states to limit the conductivity of TiO2 films in photoelectrochemical applications, evidence for potentially positive consequences of trap states have been presented in the literature.16,17 In addition, the strong dependence of surface trap states on the preparation and handling of TiO2 nanocrystalline films10,18-20 can be the source of discrepancies in spectroscopic and photoelectrochemical measurements reported by various labs,
providing further impetus for studies aimed at understanding how these defects depend on film preparation. Although TiO2 exists in the anatase, rutile, and brookite crystalline modifications, nanoparticulate anatase, readily prepared from precursors such as TiCl4 or titanium tetraisopropoxide, is the preferred form for solar energy and related photocatalysis applications. Alternatively, commercial nanoparticles such as Degussa P25, which is a mixture of approximately 75% anatase and 25% rutile, are frequently used in solar energy and photocatalysis applications. For photoelectrochemical applications, TiO2 films are typically cast from aqueous dispersions of TiO2 onto transparent conductive glass electrodes. To obtain homogeneous films with good mechanical and electrical properties, binders such as poly(ethylene glycol) or Triton X 100 are added to the TiO2 dispersion to prevent cracking of the film as it dries.5,19,21 These surfactants are subsequently removed by sintering at ∼450 °C in air, a step which also influences the crystallinity and defect structure of the film. There is, however, a great deal of interest in cheaper, flexible solar cells using transparent conductive plastic electrodes in place of the usual fluorine-doped tin oxide glass. These conductive plastics preclude the use of high-temperature sintering, and several low-temperature TiO2 film preparations have been reported that avoid sintering without sacrificing the desired electrical and mechanical properties of the film.22-36 DSSCs based on these low-temperature preparations show lower energy conversion efficiencies than DSSCs using sintered films, presumably because of decreased interparticle electrical contacts and/or more numerous defects in the unsintered films. Better understanding of the microscopic basis for the difference in solar energy conversion using sintered and unsintered TiO2 films will
10.1021/jp0640880 CCC: $33.50 © 2006 American Chemical Society Published on Web 10/05/2006
Nanocrystalline TiO2 Films and Solar Energy Conversion be useful in attempts to optimize the performance of flexible photoelectrochemical cells. Another surface-dependent parameter of great importance to DSSC performance is the electronic coupling of electronically excited sensitizer to the semiconductor conduction band. This coupling strongly influences the rate of electron injection and varies greatly for different sensitizers37-39 and nanoparticle preparations.9,40 Optimization of this coupling is a delicate problem as enhanced coupling which speeds the injection rate can also result in enhanced recombination,41 exemplified by sensitizers which form charge-transfer type complexes with the surface of TiO2. In such systems, strong mixing of the dye and semiconductor orbitals leads to red-shifted sensitizer absorption spectra and rapid electron injection, but overall poor energy conversion efficiency.38 The dye-semiconductor coupling can be a strong function of surface states.9,42 We recently reported that resonance Raman intensities of the sensitizer Ru(4,4′-dicarboxylic acid-2,2′-bipyridine)2(NCS)2 (“N3”) on colloidal TiO2 are enhanced relative to the free dye in solution, an effect attributed to intensity borrowing from the strong TiO2 band gap transition resulting from electronic coupling.43 We also observed a larger red-shift of the N3 optical spectrum on colloidal (amorphous) TiO2 than on nanocrystalline TiO2, indicative of a larger electronic coupling in the former. The purpose of the studies reported here is to further explore the influence of nanoparticle surface properties and crystallinity on electronic coupling and efficiency of solar energy conversion. In the present work, we present spectroscopic and photoelectrochemical studies of the surface properties of sintered and unsintered TiO2 films cast from two different dispersing solvents. To avoid the vagaries of various synthetic procedures for producing nanoparticulate TiO2, we employ widely available commercial TiO2 in the form of Degussa P25. We report a very simple approach that uses prolonged stirring of commercially available P25-TiO2 in ethanol or water to form homogeneous surfactant-free dispersions from which films are cast on transparent conductive glass or plastic electrodes. Our approach provides a degree of tunability in the surface properties that persists even after firing the films in air. We report optical absorption and photoluminescence spectra of the bare TiO2 films and resonance Raman and photoluminescence spectra of films sensitized with Ru(4,4′-dicarboxylic acid-2,2′-bipyridine)2(NCS)2 (“N3”). We compare the current-voltage behavior of illuminated and dark solar cells as a function of film preparation. Significant differences in the efficiency of electron injection and solar energy conversion are found to arise from variation in surface structure with film preparation, and the dyesemiconductor electronic coupling is found to depend on the solvent used in the film preparation step, even for sintered films. Although the goal of the study is to understand the influence of surface properties rather than to optimize the performance of the solar cell, the room-temperature film preparation reported here provides a simple and inexpensive approach for achieving good solar energy conversion efficiency in a plastic solar cell. 2 Experimental Section N3 dye was purchased from Solaronix and used as received. P25-TiO2, nominal particle size of 25 nm, was a generous gift from Degussa Corporation. Acetonitrile (ACN), ethanol (EtOH), and other reagents were reagent grade and used as received from Sigma-Aldrich. P25 powder was heated at 450 °C for 30 min, cooled to 100 °C then added to Millipore water, or dry ethanol, in a mole ratio of about 1:6, followed by magnetic stirring to obtain a homogeneous dispersion. The stirring period ranged
J. Phys. Chem. B, Vol. 110, No. 43, 2006 21891 TABLE 1: Performance of DSSCs Based on TiO2 Films from P25/H2O and P25/EtOH on Conductive Glass, with and without Sintering, in the Absence of TBP film P25/H2O, unsintered P25/EtOH, unsintered P25/H2O, sintered P25/EtOH, sintered
P0, Isc, Voc, dye loading, 10-7 mol cm-2 mW cm-2 mA cm-2
FF
η, %
1.26
100.8
1.91
0.44 0.66 0.57
1.28
99.8
3.28
0.35 0.60 0.69
1.40
101.9
14.2
0.41 0.40 2.29
1.05
112.6
14.3
0.40 0.48 2.42
TABLE 2: Performancea of DSSCs Using Unsintered TiO2 Films on Conductive Glass or Plastic and Containing TBP film
Isc, mA cm-2
Voc, V
FF
η, %
P25/H2O/glass 3.16 0.58 0.61 1.1 P25/EtOH/glass 4.11 0.57 0.60 1.6 P25/EtOH/plastic 2.8 ( 0.4 0.59 ( 0.03 0.54 ( 0.02 0.91 ( 0.14 a
The light power in all cases was 99.6 mW cm-2.
from overnight to several days, depending on the stirring rate. No surfactants or binders were added. Dispersions of P25 in EtOH or H2O were applied to fluorine-doped SnO2 conductive glass (FTO) (Hartford Glass, 15 Ω/square) by the doctor-blade technique, using adhesive tape as a frame to form a film with a thickness of about 8 µm after sintering. For optical absorption measurements on bare and sensitized films, very thin (∼1 µm) transparent TiO2 films were cast on quartz substrate. Especially for thin films cast from water as opposed to ethanol, more prolonged stirring (several weeks) was required in order to obtain relatively homogeneous films with minimal light scattering. Films cast from dispersions in ethanol and water are referred to below as P25/EtOH and P25/H2O. Photocurrents and photovoltages were recorded for DSSCs incorporating both sintered and unsintered films. Sintered films were heated to 450 °C for 30 min and then sensitized by dipping into a 0.3 mM solution of N3 in ethanol while still warm. Unsintered films were sensitized by soaking in N3 solution overnight at room temperature. Photocurrents and photovoltages were determined using sandwich-type solar cells containing either P25 films on FTO and a platinum-coated FTO counter electrode, or P25 films on indium tin oxide/poly(ethylene terephthalate) (ITO-PET) and a Pt-coated ITO-PET counter electrode. The electrolyte consisted of 0.05 M I2, 0.5 M LiI in acetonitrile with or without 0.5 M tert-butylpyridine (TBP) added to suppress recombination. Cells were illuminated with a 75-W xenon lamp which passed through UV and IR filters in front of the sample. The incident power at the cell (after the filters) was about 100 mW/cm2 as noted in Tables 1 and 2 below, and the illumination area was 0.196 cm2. Currents and voltages were measured using a Keithley 2400 source meter, and the incident light power was measured using a Melles Griot bolometer. Dye loading was measured at the conclusion of the current-voltage measurements as described in ref 43. Absorption spectra were recorded using a Shimadzu UV-2501 spectrometer. UV absorption spectra were determined for very thin films of TiO2 (P25/EtOH and P25/H2O) on clean quartz slides before and after sintering at 450 °C for 30 min. Thin films used for optical absorption spectra were prepared without the use of a frame and have variable thicknesses on the order of 1 µm. Absorption spectra of unsensitized films were referenced to bare quartz. For visible absorption spectra of sensitized films, a larger thin film sample of TiO2 on glass was
21892 J. Phys. Chem. B, Vol. 110, No. 43, 2006
Figure 1. Percent of film as a function of ultrasonication time in ethanol (left) and water (right), for unsintered P25/EtOH (squares), sintered P25/EtOH (diamonds), unsintered P25/H2O (triangles), and sintered P25/H2O (circles).
cut into two sections, one of which was exposed to sensitizing solution. Absorption spectra of N3 on TiO2 were then recorded versus the unsensitized TiO2 film on glass. Absorption spectra of sensitized and unsensitized films were recorded in air or in contact with solvent as noted in the figure captions. Photoluminescence spectra of unsensitized TiO2 films deposited on quartz were recorded under an argon atmosphere using excitation at 350 nm from a Krypton ion laser (Spectra Physics Beamlock 2060) at a power of 5 mW. Photoluminescence and resonance Raman spectra of N3-sensitized TiO2 films on quartz and in contact with ACN were excited at 531 nm with 5 mW incident power. The excitation light was focused onto the sample with a cylindrical lens, and the emitted light was focused onto the entrance slit of a single monochromator (Acton Spectro-Pro 2300i) after passing through a long pass filter (for UV excitation) or a notch filter (for green light excitation) to remove the elastically scattered light. The emitted light was detected using a thermoelectrically cooled CCD (Roper Scientific, Spec10: 256E). To compare luminescence and Raman intensities of different samples, care was taken to position the sample in a reproducible fashion by placing the film in a stationary quartz cuvette with the quartz side of the film against the cuvette wall and the TiO2 facing the solvent or argon atmosphere. Samples were illuminated through the quartz substrate in a backscattering geometry. Resonance Raman and luminescence spectra of N3 on TiO2 films were corrected for small differences in dye loading but were not corrected for instrument response or differential extinction of scattered/emitted light. The mechanical stability of the bare films on FTO was tested by sonicating in a bath of water or ethanol and measuring the geometric area of the film as a function of time following ref 28. 3. Results 3.1. Properties of Unsensitized TiO2 Films. Figure 1 shows the results of testing the mechanical stability of the sintered and unsintered nanocrystalline TiO2 films on a conductive glass substrate by ultrasonication in a bath of water or ethanol. The maximum sonication time was slighly more than 3 h. When sonicated in ethanol, the sintered and unsintered P25/EtOH films were found to be more stable than sintered and unsintered P25/ H2O. The unsintered P25/EtOH film began to detach from the glass substrate after about 2 h of sonication in ethanol, while the sintered film just began to detach after 3 h. The sintered and unsintered P25/H2O films, in contrast, began to detach after about 1.5 and 0.5 h of sonication in ethanol, respectively. All films detached much more rapidly when sonicated in water than in ethanol except for sintered P25/EtOH which was quite stable
Zhang et al.
Figure 2. Absorbance of thin films of TiO2 on quartz and in contact with air: (a) P25/EtOH before (dashed line) and after (solid line) sintering, (b) P25/H2O before (dashed line) and after (solid line) sintering, and (c) comparison of sintered films of P25/EtOH (dashed line) and P25/H2O (solid line), where the data have been scaled to give the same maximum intensity. The spectra of the sintered and unsintered films in (b) are indistinguishable on the scale of the plot.
in both solvents. On the basis of experience in handling the different films in numerous experiments as reported below, the mechanical stability of films cast from ethanol was observed to be superior to those cast from water. Films prepared from either P25/EtOH and P25/H2O are optically transparent to the eye. Ultrathin films (on the order of 1 µm) were prepared on quartz slides in order to measure the very intense band gap absorption. Figure 2 compares the absorption spectrum in air of the same film prepared from P25/ EtOH or P25/H2O on quartz before and after sintering. In the case of P25/H2O, there is no change in the absorption spectrum on sintering. In contrast, the absorption spectrum of the film from P25/EtOH shows an increase in intensity after sintering in air at 450 °C. There is no change in the intensity of the visible tail on sintering for either film. However, the visible tail is more prominent for P25/H2O than for P25/EtOH, as shown in Figure 2c, where the absorbance data for the sintered films have been scaled to the same maximum. Band tails could result from defect states in the band gap9 or from scattering. We assume scattering makes little contribution to the total extinction, in accord with the transparent nature of the films and the fact that the band tails are insensitive to the refractive index of the contacting medium, as shown below. Sintering is expected to result in a decrease in defects and improved crystallinity,10,44 which would result in an increase in the extinction coefficient for the band gap absorption.45 The data in Figure 2 may indicate that films prepared from P25/EtOH undergo increased crystallinity upon sintering, while the crystallinity of the P25/H2O film is not significantly improved. Based on the band tails, however, there appear to be more intraband gap states for the P25/H2O preparation. Figure 3 shows the absorption spectra of sintered and unsintered TiO2 films in contact with air, EtOH, or ACN. The sintered and unsintered films in this case are different samples; thus, the thickness and peak absorbance differ slightly. The absorption spectrum of P25/H2O is not very sensitive to contacting solvent either before or after sintering. In contrast, contacting solvents have only a slight effect on the optical spectrum of unsintered P25/EtOH, whereas there is a significant increase in absorbance when the sintered P25/EtOH film is in contact with EtOH instead of air or ACN. It is possible that this increased absorbance results from increased crystallinity
Nanocrystalline TiO2 Films and Solar Energy Conversion
Figure 3. Absorbance of thin films of TiO2 on quartz: (a) P25/EtOH, sintered, (b) P25/H2O, sintered, (c) P25/EtOH, unsintered, and (d) P25/ H2O, unsintered, in contact with air (dotted line), EtOH (dashed line), and ACN (full line). For (b) and (d) the spectra in contact with EtOH and air are indistinguishable on the scale of the plot.
Figure 4. Photoluminescence emission spectra of P25/EtOH and P25/ H2O films in argon on quartz, excited at 350 nm. The dashed lines are for unsintered films and the full lines are for sintered films. The intensity for unsintered P25/H2O has been reduced by a factor of 5 for ease of comparison.
in the presence of EtOH, but this conclusion must be considered tentative until further studies such as X-ray diffraction are performed. Figure 4 compares the photoluminescence (PL) spectra of unsensitized, freshly prepared films of P25/EtOH and P25/H2O on a quartz slide in an argon atmosphere, before and after sintering. The observed PL is nearly completely quenched in the presence of air but is easily observed at room temperature in the presence of argon or solvent. The broad visible luminescence of TiO2 is assigned to surface traps,7,8,46,47 and consistent with this we find the intensity of the photoluminescence is decreased by sintering. The luminescence is generally stronger in P25/H2O than P25/EtOH. Figure 5 shows AFM images of TiO2 films from EtOH and water dispersions before and after sintering. The images are qualitatively similar except for the somewhat greater prevalence of larger clumps of nanoparticles in the films from P25/H2O. Although the resolution is limited, the images do not suggest major changes in particle size or morphology with sintering. This is in agreement with other reports showing changes in crystallinity but not particle shape or size after firing TiO2 films in air.44
J. Phys. Chem. B, Vol. 110, No. 43, 2006 21893 3.2. Optical Spectra of N3-Sensitized TiO2 Films. Figure 6 shows the absorption spectrum of N3 on sintered and unsintered P25/EtOH and P25/H2O in contact with air and with ACN. The absorbance data were taken using thin TiO2 films and are referenced to the unsensitized films in contact with air or ACN. Although there are slight perturbations due to light scattering, there is significant and reproducible dependence of the N3 absorption spectrum on film preparation. For films in contact with either air or solvent, the trend is to observe the maximum absorbance at longer wavelength in the P25/EtOH films than in P25/H2O. On going from air to ACN, the P25/ EtOH films undergo a further red-shift while the P25/H2O films show a slight blue-shift. The absorption spectrum of N3 is reported to red-shift with increasing solvent polarity,48 but increasing electronic coupling (as we have observed previously for N3 on colloidal TiO243) also results in a red-shift of the sensitizer. We consider the differences in the absorption spectra of the dry films in Figure 6a to be the result of differences in coupling, while those of Figure 6b reveal the additional influence of solvent polarity. On this basis, it appears that N3 is more strongly coupled to P25/EtOH than to P25/H2O and that for both films there is an increase in coupling on sintering. The reasons for larger solvent-induced red-shifts for N3 on P25/ EtOH as compared to that on P25/H2O are considered below. Figure 7 shows the resonance Raman and photoluminescence spectra of N3 on sintered and unsintered P25/EtOH and P25/ H2O. The Raman and luminescence spectra were recorded for the same ∼8 µm thick films in contact with ACN in a backscattering geometry using an excitation wavelength of 531 nm. The sample is aligned so that the glass substrate, which is adjacent to the wall of the cuvette, faces the detector, while the sensitized film faces the bulk solvent. For these thick films the transmittance of the incident radiation is essentially zero, and since the laser beam passes through the film before reaching the bulk solvent and is focused on the film, we may assume that the solvent Raman lines of Figure 7a originate from ACN within the pores of the film rather than from bulk solvent. The spectra shown here are similar to our previously reported resonance Raman data for N3 on colloidal (amorphous) TiO2 in ACN,43 although the greater intensity of overlapping solvent lines in our earlier data prevents a thorough comparison of the spectra on crystalline and amorphous TiO2. Apart from an increase in overall Raman intensity for N3 on sintered versus unsintered P25/EtOH, there are no major differences in frequency or relative intensity for the various film preparations. A striking difference in the Raman spectra of the different films is the much larger contribution of solvent Raman scattering in films prepared from ethanol dispersions. Considering our scattering geometry and the high extinction of incident and scattered light by the films, the Raman data strongly suggest a much larger amount of acetonitrile solvent inside the pores of films prepared from ethanolic as opposed to aqueous dispersions of TiO2. This would also explain the trends in the absorption spectra of N3 on the various films, where greater polarityinduced red-shifts on going from air to ACN are observed for P25/EtOH than for P25/H2O. The photoluminescence spectrum of N3, observed in solution at about 800 nm,48 is quenched on TiO2 in proportion to the quantum yield of electron injection, which is nearly 100% in the presence of ACN.5,43 Figure 7b shows the photoluminescence spectrum of N3 on the four different TiO2 films in contact with ACN, excited at a wavelength near the maximum in the N3 visible absorption band. The luminescence data were corrected for differences in dye loading rather than absorbance,
21894 J. Phys. Chem. B, Vol. 110, No. 43, 2006
Zhang et al.
Figure 5. AFM images of sintered (top) and unsintered (bottom) TiO2 films: (a) P25/EtOH, sintered, (b) P25/H2O, sintered, (c) P25/EtOH, unsintered, and (d) P25/H2O, unsintered.
Figure 6. Absorption spectra of N3 on P25/H2O (thin lines) and P25/ EtOH (thick lines) in contact with (a) air and (b) ACN. The dashed lines are for unsintered films and full lines for sintered films.
since the highly scattering nature of the films precludes measurement of the absorbance at the excitation wavelength. We expect the correction for dye loading to be reasonable because the N3 absorption maxima for sintered and unsintered versions of the same film are similar when in contact with ACN. For both P25/EtOH and P25/H2O, there is a decrease in the intensity of emission at shorter wavelengths for the sintered films, and a decrease in the overall luminescence intensity for sintered versus unsintered films. The effect of sintering is larger for the P25/EtOH film than for P25/H2O. The N3 emission spectra show that the efficiency of electron injection is slightly higher for sintered than for unsintered films. 3.3. Dye-Sensitized Solar Cells. Table 1 compares the performance of DSSCs using sintered and unsintered P25 films
Figure 7. (a) Resonance Raman and (b) photoluminescence spectra of N3 on (1) sintered P25/EtOH, (2) unsintered P25/EtOH, (3) sintered P25/H2O, and (4) unsintered P25/H2O on glass. The spectra were excited at 531 nm and are for ∼8 µm thick films in contact with ACN. The asterisks in (a) indicate solvent Raman bands.
on transparent conductive glass electrodes in the absence of TBP. Reported there are the incident light power P0, the shortcircuit photocurrent Isc, the open-circuit voltage Voc, the fillfactor FF, and the energy conversion efficiency η. The dye loading reported in Table 1 is based on the projected area of the film. While it is similar for unsintered P25/EtOH and P25/ H2O, sintering causes an increase in dye loading for P25/H2O and a decrease for P25/EtOH. For both types of films, the efficiency η is greater using sintered TiO2 electrodes which have higher values for Isc and η. In the case of P25/EtOH, sintering
Nanocrystalline TiO2 Films and Solar Energy Conversion
Figure 8. Dark current versus applied voltage for DSSCs containing P25/H2O (thin lines) and P25/EtOH (thick lines) films on conductive glass in the absence of TBP. The dashed lines are for unsintered films, and the full lines are for sintered films.
results in an increase in Voc while for P25/H2O it is decreased. Comparing the two DSSCs with sintered films, the maximum photocurrent, maximum photovoltage, and the efficiency are all similar, even though the dye loading is significantly larger for sintered P25/H2O than for sintered P25/EtOH. For DSSCs with unsintered TiO2 films, the P25/EtOH film produces larger photocurrent, lower photovoltage, and slightly larger efficiency than P25/H2O. Further insight into the properties of the various films in DSSCs is found by measuring the dark current as a function of applied voltage as shown in Figure 8. The onset potential for dark current in P25/EtOH is slightly more negative than that for P25/H2O, indicating that the conduction band is slighly higher for P25/EtOH. For the same applied voltage, the dark current is greater for the unsintered film than for its sintered counterpart, suggesting that the main effect of sintering is to decrease the back electron-transfer rate. The slopes of the steep part of the current-voltage curves reveal changes in series resistance of the various films. The overall steeper slope of the P25/EtOH films indicates lower series resistance compared to that of the P25/H2O films. Note that sintering does not greatly decrease the series resistance as might be expected on the basis of improved interparticle contacts, probably because of the importance of other contributions to the internal resistance, i.e. from the FTO substrate, Pt-coated counter electrode, and the electrolyte solution.49 Analysis of the slopes of the steep part of the dark current curves results in series resistivities ranging from about 9 Ω cm2 for unsintered P25/EtOH to 16 Ω cm2 for unsintered P25/H2O, with intermediate values found for the sintered films. These values are similar to those reported by Hagfeldt et al.35 for a DSSC using an unsintered TiO2 film. Table 2 shows the results for DSSCs prepared from unsintered TiO2 films on conductive glass or plastic using TBP in the electrolyte solution. It was not possible to obtain a well-spread film of P25/H2O on conductive plastic, and therefore, only data for P25/EtOH on ITO-PET is shown. In contrast to the current/ voltage data for glass electrodes, that for the plastic electrode was found to vary with the position of illumination due to film inhomogeneity. Consequently, the values reported for Isc, Voc, FF, and η are averages for six different spots. 4. Discussion Homogeneous, transparent films of nanocrystalline TiO2 were cast from surfactant-free dispersions of P25 in either ethanol or water, but those cast from ethanol exhibit greater mechanical stability as judged by ultrasonication tests and experience with
J. Phys. Chem. B, Vol. 110, No. 43, 2006 21895 routine handling of the films. We also attempted to obtain homogeneous dispersions of P25 in acetonitrile but found that the nanoparticles could not be well-dispersed in this solvent. The choice of dispersing solvent also influences the optical spectrum of the film, even after sintering. Larger photoluminescence intensity and more prominent band tails in the absorption spectrum are found for P25/H2O, indicating a larger number of localized intraband gap states. Reference 50 also noted more facile production of defect states and greater nanoparticle agglomeration in nanocrystalline TiO2 after exposure to water, in agreement with the present work. The strong adsorption of water and alcohols on surface defects of TiO2 is well-known.51-54 The band tails which are found in P25/H2O are not affected by sintering, nor is the optical spectrum of this film very sensitive to contacting solvent. The intensity of band gap absorbance of P25/EtOH, on the other hand, is somewhat solvent-dependent for an unsintered film and moreso for a sintered film. It is interesting that the intensity of the optical spectrum of the sintered film is larger in contact with EtOH than with air or water. As others have seen in preparations of TiO2 nanoparticles from alcohol, the particles as prepared contain adsorbed alcohol molecules and alkoxide radicals which are driven off by sintering or annealing.46,54 The data of Figure 3 suggest that sintering of P25/EtOH films may create surface defects which are healed to some extent in the presence of EtOH solvent. The surface of P25/H2O is likely to contain a larger concentration of very stable Ti-OH groups as well as strongly adsorbed water.52,55,56 Differences in the stability of surface groups may explain differences in the adhesion of the unsintered P25/EtOH and P25/H2O to conductive glass or plastic. More labile surface passivation of TiO2 for dispersions from ethanol as opposed to water may facilitate better bonding to the substrate for unsintered P25/EtOH. Variations in dye loading for the various films also imply differences in surface structure. Sintering leads to a decrease in dye loading for the P25/EtOH films, possibly as a result of adsorption of ethanol from the sensitizing solution, blocking defect sites which would otherwise bind the sensitizer. Alternatively, it may be that adsorbed ethanol in the unsintered film results in less strongly coupled physisorbed dye in agreement with the blue-shift in the N3 absorption spectrum and the increased N3 luminescence on unsintered P25/EtOH compared to that of the sintered film. Given the solubility of N3 in ethanol, it is reasonable to assume that ethanol species on the surface of the unsintered P25/EtOH film, absent after sintering, result in a larger amount of adsorbed dye which is, however, less strongly coupled to the surface than the chemisorbed N3 on the sintered film. In contrast, sintering increases dye loading on P25/H2O somewhat, for reasons which are not presently understood. Although there is no doubt a distribution of binding sites for N3 on the surface of TiO2, ester linkages between carboxylic acid groups of N3 and Ti-OH have been revealed.57,58,59 The larger dye loading of N3 on P25/H2O after sintering could result from an increase in surface hydroxyl groups. The UV-excited photoluminescence (PL) spectrum of both types of unsensitized films was observed to decrease in intensity on sintering, especially for films cast from water instead of ethanol. While the PL intensity of the two sintered films was similar, unsintered P25/H2O showed a much more intense emission than P25/EtOH. The assignment of the broad, Stokesshifted PL of crystalline and nanocrystalline TiO2 is not without controversy, and from a survey of the literature it appears that the observation of this emission is frequently made at cryogenic temperature rather than room temperature as in this work. Tang
21896 J. Phys. Chem. B, Vol. 110, No. 43, 2006 et al.60 assigned the low-temperature emission of crystalline anatase, peaking at about 540 nm, to radiative recombination of self-trapped excitons. Others have correlated the PL intensity of different samples of TiO2 nanoparticles to oxygen vacancies and photocatalytic activity.46,61-63 The nearly complete and reversible quenching of this PL by oxygen, which scavenges photogenerated electrons from TiO2 films,64 shows that the luminescence is associated with surface states. The evidence points to the importance of oxygen vacancies in the mechanism for PL. Reversible production of oxygen vacancies results in undercoordinated Ti4+ sites which serve as sub-bandgap trap states forming Ti3+ centers when occupied.65-67 The apparently larger number of trap states for P25/H2O is consistent with the observation of stronger band tails for this film. In addition, the onset potential for dark current generation is larger in magnitude for P25/EtOH than P25/H2O. Since this potential is the difference between the Fermi levels of TiO2 and the redox mediator, the more prevalent sub-bandgap states in P25/H2O correlate with a lower (less negative) Fermi level as revealed by the dark current onset potential. Surface defects including oxygen vacancies are known to increase recombination current in dye-sensitized solar cells.68 The much stronger PL intensity of unsintered P25/H2O compared to that of unsintered P25/ EtOH, and the similar PL intensities of the sintered films, should be compared to the photocurrent-photovoltage data of Table 1. The larger Isc obtained for solar cells with unsintered P25/ EtOH instead of P25/H2O could derive from enhanced recombination in the latter, whereas the maximum photocurrent for solar cells with sintered electrodes is similar for both types of films. The optical absorption, photoluminescence, and resonance Raman spectra of N3 on TiO2 were found to depend on film preparation, with larger differences between sintered and unsintered films observed for films cast from ethanol rather than water. The smaller dependence on sintering for N3 optical spectra on P25/H2O appears to be consistent with the similar optical spectra of the sintered and unsintered films in the absence of sensitizer. Red-shifts in the optical absorption spectrum of sensitizers adsorbed on dry TiO2, relative to the free dye, are taken as evidence of electronic coupling between the dye and semiconductor by mixing of the dye LUMO with conduction band orbitals of TiO2, resulting in greater stabilization of the dye excited state relative to the ground state on binding. On the basis of the data of Figure 6a, N3 is found to be more strongly coupled to TiO2 films cast from ethanol as opposed to water, and there is an increase in coupling on sintering for both types of films. This same electronic coupling contributes directly to the rate of electron injection as it appears in the preexponential factor of the rate constant, but differences in the Fermi level and N3 excited-state energy level for the various films also influence the driving force and thus the rate and yield of electron injection.69 From the dark current measurements and optical spectra, it can be concluded that the driving force for electron injection is lower for films cast from ethanol (lower dye excitedstate energy, higher TiO2 Fermi level) rather than water. It should be kept in mind that the PL spectra of Figure 7b reveal the distribution of N3 molecules which do not inject electrons into TiO2. The decrease in luminescence of N3 on sintered versus unsintered TiO2 is completely consistent with the changes in electronic coupling inferred from the corresponding absorbance spectra. On comparing the N3 PL on P25/EtOH to that on P25/H2O, the increased electronic coupling for the former appears to be offset by the decrease in driving force, resulting in similar injection yields for various films. However, as
Zhang et al. mentioned previously, the similar maximum photocurrents observed for the sintered P25/EtOH and P25/H2O films, despite smaller dye loading for the former, also points to stronger coupling of N3 to the ethanol-stirred film and/or better collection of injected electrons. One has to keep in mind that electron injection to surface states, rather than the conduction band, also quenches the sensitizer luminescence but does not necessarily correlate with increased collection of electrons in the external circuit. Given that the quantum efficiency of electron injection from N3 to TiO2 in the presence of ACN is already nearly unity, the data of Figure 7 reveal that the large decrease in photocurrent for DSSCs employing unsintered films is not merely the result of a large decrease in the quantum yield for electron injection. We have previously shown that the resonance Raman intensity of N3 adsorbed on colloidal TiO2 is increased relative to the free dye in solution, and concluded that the enhancement results from intensity borrowing via dye-semiconductor electronic coupling.43 Shoute and Loppnow70 made a similar observation but attributed the increased intensity on TiO2 to a reduction in solvent reorganization energy. Unlike the RR spectra for N3 on TiO2 films reported here, the RR intensity data for sensitized colloidal particles are easily converted to absolute intensities using solvent Raman lines as internal standards. The problem of standardizing the RR spectra and correcting for differential self-extinction (i.e., absorption and scattering) of sensitized films is much more complicated. Great care was taken to align the various samples and optimize the Raman spectra in a reproducible fashion, and the only correction applied was a small adjustment for differences in dye loading. With these limitations in mind, we observe a general trend to larger RR intensities of N3 on P25/EtOH compared to those on P25/H2O, and an increase in intensity for N3 on sintered versus unsintered P25/ EtOH. These trends also support the conclusions about dyeTiO2 coupling based on the optical spectra of the sensitized films, where the electronic coupling is observed to be larger for sintered films. However, this conclusion must be considered tentative, owing to the likely strong dependence of the RR intensity on excitation wavelength and the shifts in the absorption spectra of the different films. Note that no significant shifts of N3 vibrational frequencies are found for the various films, revealing that the structure of the sensitizer in its ground electronic state is not very dependent on film preparation. Perhaps the most surprising aspect of the Raman data is the much larger contribution from acetonitrile in the films cast from ethanol instead of water. We conclude there is much more ACN solvent in the pores of the P25/EtOH films than P25/H2O. The preferential adsorption of water over acetonitrile on TiO2 is not unexpected, given the well-known persistence of adsorbed water on TiO2 surfaces; thus, it is possible that the large differences in ACN Raman signal are merely the result of residual water in P25/H2O films. Another possibility is that a larger amount of acetonitrile solvent is trapped in the pores of films prepared from EtOH as a result of larger pore size or more acetonitrilefriendly surface structure. A recent report found that exposure of TiO2 nanoparticles to a hydration-dehydration cycle leads to a 2-fold decrease in pore size.50 Determination of the pore size distribution is beyond the scope of the present work but will be the subject of future studies. In addition to the dye-semiconductor electronic coupling, the rate of interfacial electron transfer also depends on the driving force which is the difference between the free energy of the excited state of the dye and the Fermi level of TiO2. The onset potential for dark current versus voltage as shown in Figure 8 is slightly larger in magnitude for P25/EtOH films
Nanocrystalline TiO2 Films and Solar Energy Conversion compared to that for P25/H2O films. This onset potential should be the difference between the Fermi levels of the (sensitized) TiO2 film and the I-/I3- redox couple in ACN. Thus, the Fermi level of the P25/EtOH film is slightly higher than that of P25/ H2O. If we make the reasonable assumption that relative changes in the absorption maximum of N3 on TiO2 shown in Figure 6 are dominated by perturbations to the excited rather than the ground electronic state, we find the excited state of N3 to be lower on P25/EtOH than on P25/H2O. Taken with the data for the onset potential of the dark current, we deduce that the driving force for electron injection is smaller for N3 on P25/EtOH than on P25/H2O. Were it not for differences in electronic coupling, the inferred differences in driving force would translate into a lower quantum yield for electron injection on P25/EtOH. However, similar short-circuit photocurrents are observed for DSSCs using sintered P25/EtOH and sintered P25/H2O, and higher values are found using unsintered P25/EtOH compared to unsintered P25/H2O. This comparison again argues for greater electronic coupling of N3 to P25/EtOH than to P25/H2O, but future work is needed to determine the collection efficiency, which depends on recombination rates and carrier transport, for DSSCs employing the different films. The overall performance of DSSCs is strongly dependent on sintering and slightly dependent on the dispersing solvent used to prepare the film. We first consider DSSCs in the absence of TBP. Solar cells using sintered films give much larger photocurrents than those using unsintered films. As previously mentioned, sintering results in only a slight increase in the quantum efficiency of electron injection. On the other hand, if the differences shown in Table 1 were the result of improvements in carrier transport for sintered films, greater effects of sintering on the dark current-voltage curves might be expected. We conclude that the dominant effect of sintering is to reduce the recombination current which limits both Isc and Voc. In agreement with changes in the photoluminescence spectra of unsensitized films, sintering reduces the extent of surface defects, and these defects appear to be more prevalent in P25/ H2O than in P25/EtOH. Comparing the two unsintered films on conductive glass, we find larger Isc, smaller Voc and slightly larger efficiency for P25/EtOH, as would be expected on the basis of fewer recombination-enhancing surface defects. The difference in the performance of DSSCs using sintered and unsintered films is also larger for the water-stirred film, in agreement with larger changes in the trap state luminescence of P25/H2O on sintering. For unsintered films, the enhanced current in P25/EtOH comes at the expense of the open-circuit voltage, presumably because the enhanced coupling also permits faster recombination of electrons with oxidized dye, as has been pointed out in ref 41. For sintered films in the absence of TBP, on the other hand, very similar values of Isc and Voc are obtained for both P25/EtOH and P25/H2O, even though the dye loading is larger in the latter. Again, this implies stronger electronic coupling of N3 to P25/EtOH. The slightly greater efficiency of the DSSC containing sintered P25/EtOH compared to sintered P25/H2O is the result of a more favorable fill factor, possibly the result of lower series resistance in films prepared from ethanol. Future work will explore the physical basis for the influence of film preparation on carrier transport. We explore the influence of recombination in DSSCs containing unsintered films on conductive glass and plastic by addition of tert-butylpyridine (TBP), as shown in Table 2. TBP is generally added to dye-sensitized solar cells to retard recombination and improve the kinetically controlled opencircuit voltage.13 It was not possible to obtain a good-quality
J. Phys. Chem. B, Vol. 110, No. 43, 2006 21897 film on ITO-PET from the water-stirred dispersion of P25, but the ethanol-stirred dispersion resulted in a suitable film albeit smaller photocurrent and fill factor compared to the same film on conductive glass. The decreased performance of the DSSC using the flexible electrode is the result of larger series resistance compared to the glass electrode. For conductive glass electrodes, the addition of TBP to the electrolyte results in significant improvements in photocurrent and photovoltage. As for DSSCs in the absence of TBP, the maximum photocurrent is larger for unsintered P25/EtOH than unsintered P25/H2O. The relative improvement in Isc on addition of TBP is greater for P25/H2O than for P25/EtOH. Again, this is consistent with a greater number of traps which serve as recombination centers for films cast from water as opposed to ethanol. Unfortunately, TBP can only partially alleviate the deleterious effects of trap states in unsintered films. On the basis of changes in sintering, the luminescence spectra of TiO2 films are found to be much more sensitive to trap states than are the absorption spectra. Despite negligible changes to the absorption spectrum of either film on sintering, dramatic effects of recombination are inferred from changes to the photocurrent and photovoltage, and smaller changes to the dark current as a function of applied potential. We infer from the luminescence data that these surface traps are more prevalent in P25/H2O than P25/EtOH, particularly before sintering, and result in lower photocurrents for DSSCs using the former. The difference in the optical spectra of P25/EtOH and P25/H2O are also consistent with this conclusion. The effects seen here are similar to the changes in band gap absorption of colloidal TiO2 reported by Hartland et al.,40 where addition of small amounts of water to an ethanol suspension resulted in a red-shift of the band gap transition. Similar to the conclusions of ref 40, we find strong specific interaction of water with the TiO2 surface and a greater number of sub-bandgap states for films cast from water, perhaps from oxygen vacancies which contribute to absorption below the band gap. A reproducible increase in the intensity of the band gap transition was observed for P25/EtOH film in contact with EtOH versus air or ACN. We hypothesize that this is the result of specific surface interactions between EtOH solvent and the film which can propagate through the nanocrystalline structure, and future studies are planned to test this. 5. Conclusions We have demonstrated a straightforward approach for the room-temperature preparation of surfactant-free films of TiO2 for dye-sensitized solar energy conversion. Although most published procedures for preparing TiO2 films use aqueous dispersions (see, however, ref 35), we find that the mechanical properties and dye-semiconductor electronic coupling are more favorable for films produced from ethanolic dispersions of TiO2, regardless of whether the films are fired in air. The choice of ethanol versus water for dispersing the nanoparticles makes a clear difference in surface structure, optical and mechanical properties, and performance of DSSCs. The adhesion of films cast from dispersions of P25 in ethanol to both plastic and glass conductive substrates was found to be superior to that of films cast from water, and apparently fewer photocurrent-limiting defect states are observed for the ethanoldispersed films. The results of the present work highlight the strong dependence of the surface properties of nanocrystalline TiO2 on details of the film preparation, and suggest that the observation of TiO2 photoluminescence is a convenient way to monitor the presence of deleterious surface defect states.
21898 J. Phys. Chem. B, Vol. 110, No. 43, 2006 Acknowledgment. Support from the Washington State University College of Science and Department of Chemistry is gratefully acknowledged. Support from the National Science Foundation though Grant CHE 0234726 for the purchase of the atomic force microscope is gratefully acknowledged. We thank Dr. Louis Scudiero for help with obtaining AFM images. References and Notes (1) O’Regan, B. Gra¨tzel, M. Nature (London) 1991, 353, 737. (2) Gra¨tzel, M. Nature (London) 2001, 414, 338. (3) Hagfeldt, A.; Gra¨tzel, M. Acc. Chem. Res. 2000, 33, 269. (4) Hagfeldt, A.;. Gra¨tzel, M. Chem. ReV. 1995, 95, 49. (5) Nazeeruddin, M. K.; Kay, A.; Rodicio, I.; Humphry-Baker, R. Mu¨ller, E.; Liska, P.; Vlachopoulos, N.; Gra¨tzel, M. J. Am. Chem. Soc. 1993, 115, 6382. (6) Diebold, U. Surf. Sci. Rep. 2003, 48, 53. (7) Tang, H.; Berger, H.; Schmid, P. E.; Le´vy, F. Solid State Commun. 1993, 87, 847. (8) Dittrich, Th. Phys. Status Solidi A 2000, 182, 447. (9) Hao, E.; Anderson, N. A.; Asbury, J. B.; Lian, T. J. Phys. Chem. B 2002, 106, 10191. (10) Benko¨, G.; Skarman, B.; Wallenberg, R.; Hagfeldt, A.; Sundstro¨m, V.; Yartsev, A. P. J. Phys. Chem. B 2003, 107, 1370. (11) Nelson, J.; Haque, S. A.; Klug, D. R.; Durrant, J. R. Phys. ReV. B 2001, 63, 205321. (12) F. Cao, G. Oskam, G. J. Meyer, P. C. Searson, J. Phys. Chem. 1996, 100, 17021. (13) Kalyanasundaraman, K.; Gra¨tzel, M. Coord. Chem. ReV. 1998, 77, 347. (14) Kopadakis, N.; Neale, N. R.; Zhu, K.; van de Lagemaat, J.; Frank, A. J. App. Phys. Lett. 2005, 87, 202106. (15) Schlichtho¨rl, G.; Huang, S. Y.; Sprague, J. Frank, A. J. J. Phys. Chem. B 1997, 101, 8141. (16) Gregg, B. A.; Chen, S.-G, Ferrere, S. J. Phys. Chem. B 2003, 107, 3019. (17) Wu¨rfel, U.; Wagner, J.; Hinsch, A. J. Phys. Chem. B 2005, 109, 20444. (18) Huber, R.; Spo¨rlein, S.; Moser, J. E.; Gra¨tzel, M.; Wachtveitl, J. J. Phys. Chem. B 2000, 104, 8995. (19) Barbe´, C. J.; Arendse, F.; Comte, P.; Jirousek, M.; Lenzmann, F.; Shklover, V.; Gra¨tzel, M. J. Am. Ceram. Soc. 1997, 80, 3157. (20) Zaban, A.; Aruna, S. T.; Tirosh, S.; Gregg, B. A.; Matsai, Y. J. Phys. Chem. B 2000, 104, 4130. (21) Smestad, G. P.; Gra¨tzel, M. J. Chem. Educ. 1998, 75, 752. (22) Zhang, D.; Yoshida, T.; Minoura, H. Chem. Lett. 2002, 874. (23) Zhang, D.; Yoshida, T.; Furuta, K.; Minoura, H. J. Photochem. Photobiol., A 2004, 164, 159. (24) Miyasaka, T.; Kijitori, Y.; Murakami, T. N.; Kimura, M.; Uegusa, S. Chem. Lett. 2002, 1250. (25) Kado, T.; Yamaguchi, M.; Yamada, Y.; Hayase, S. Chem. Lett. 2003, 1056. (26) Nogueira, A. F.; Longo, C.; De Paoli, M.-A. Coord. Chem. ReV. 2004, 248, 1455. (27) Longo, C.; Nogueira, A. F.; de Paoli, M.-A.; Cachet, H. J. Phys. Chem. B 2002, 106, 5925. (28) Pichot, F.; Pitts, J. R.; Gregg, B. A. Langmuir 2000, 16, 5626. (29) Pichot, F.; Ferrere, S.; Pitts, J. R.; Gregg, B. A. J. Electrochem. Soc. 1999, 146, 4324. (30) Park, N.-G.; Schlichtho¨rl, G.; van de Lagemaat, J.; Cheong, H. M.; Mascarenhas, A.; Frank, A. J. J. Phys. Chem. B 1999, 103, 3308. (31) Haque, S. A.; Palomares, E.; Upadhyaya, H. M.; Otley, L.; Potter, R. J.; Holmes, A. B.; Durrant, J. R. Chem. Commun. 2003, 24, 3008. (32) Lindstro¨m, H.; Holmberg, A.; Magnusson, E.; Lindquist, S.-E.; Malmqvist, L.; Hagfeldt, A. Nano Lett. 2001, 1, 97. (33) Lindstro¨m, H.; Holmberg, A.; Magnusson, E.; Malmqvist, L.; Hagfeldt, A. J. Photochem. Photobiol., A 2001, 145, 107. (34) Lindstro¨m, H.; Magnusson, E.; Holmberg, A.; So¨dergren, S.; Lindquist, S.-E.; Hagfeldt, A. Sol. Energy Mater. Sol. Cells 2002, 73, 91. (35) Hagfeldt, A.; Boschloo, G.; Lindstro¨m, H.; Figgemeier, E.; Holmberg, A.; Aranyes, V.; Magnusson, E.; Malmqvist, L. Coord. Chem. ReV. 2004, 248, 1501.
Zhang et al. (36) Gutierre´z-Tauste, D.; Zumeta, I.; Vigil, E.; Herna´ndez-Fenollosa, M. A.; Dome`nech, X.; Ayllo´n, J. A. J. Photochem. Photobiol., A 2005, 175, 165. (37) Huber, R.; Moser, J. E.; Gra¨tzel, M.; Wachveitl, Chem. Phys. 2002, 285, 39. (38) Tae, E. L.; Lee, S. H.; Lee, J. K.; Yoo, S. S.; Kang, E. J. Yoon, K. B. J. Phys. Chem. B 2005, 109, 22513. (39) Clifford, J. N.; Palomares, E.; Nazeeruddin, Md. K.; Gra¨tzel, M.; Nelson, J.; Li, X.; Long, N. J.; Durrant, J. R. J. Am. Chem. Soc. 2004, 126, 5225. (40) Martini, I.; Hodak, J. H.; Hartland, G. V. J. Phys. Chem. B 1998, 102, 607. (41) Durrant, J. R.; Haque, S. A.; Palomares, E. Coord. Chem. ReV. 2004, 248, 1247. (42) Creutz, C.; Brunschwig, B. S.; Sutin, N. J. Phys. Chem. B 2005, 109, 10251. (43) Pollard, J. A.; Zhang, D.; Downing, J. A.; Knorr, F. J.; McHale, J. L. J. Phys. Chem. B 2005, 109, 11443. (44) Hanley, T. L.; Luca, V.; Pickering, I.; Howe, R. F. J. Phys. Chem. B 2002, 106, 1153. (45) Kallioinen, J.; Benko¨, G.; Myllyperkio¨, P.; Khartiachtchev, L.; Skarman, B.; Wallenberg, R.; Tuomikoski, M.; Korppi-Tommola, J.; Sundstro¨m, V.; Yartsev, A. P. J. Phys. Chem. B 2004, 108, 6365. (46) Zhu, Y. C.; Ding, C. X. J. Solid State Chem. 1999, 145, 711. (47) Zhang, W. F.; Zhang, M. S.; Yin, Z. Phys. Status Solidi 2000, 179, 319. (48) Nazeeruddin, Md. K.; Zakeeruddin, S. M.; Humphrey-Baker, R.; Jirousek, M.; Liska, P.; Vlachopoulos, N.; Shklover, V.; Fischer, V.; Gra¨tzel, M. Inorg. Chem. 1999, 38, 6298. (49) Han, L.; Koide, N.; Chiba, Y.; Islam, A.; Komiya, N.; Fukui, A.; Yamanaka, R. App. Phys. Lett. 2005, 86, 213501. (50) Elser, M. J.; Berger, T.; Brandhuber, D.; Bernardi, J.; Diwald, O.; Kno¨zinger, E. J. Phys. Chem. B 2006, 110, 7605. (51) Linsebigler, A. L.; Lu, G.; Yates, J. T. Chem. ReV. 1995, 95, 735. (52) Onda, K.; Li, B.; Zhao, J.; Jordan, K. D.; Yang, J.; Petek, H. Science 2005, 308, 1154. (53) Wang, C.-Y.; Groenzig, H.; Shultz, M. J. J. Phys. Chem. B 2004, 108, 265. (54) Brownson, J. R. S.; Tejedor-Tejedor, I.; Anderson, M. A. J. Phys. Chem. B 2006, 110, 12494. (55) Panayotov, D. A.; Yates, J. T. Chem. Phys. Lett. 2005, 410, 11. (56) Uosaki, K.; Yano, T.; Nihonyanagi, S. J. Phys. Chem. B 2004, 108, 19086. (57) Leo´n, C. P.; Kador, L.; Peng, B.; Thelakkat, M. J. Phys. Chem. B 2006, 110, 8723. (58) Meyer, T. J.; Meyer, G. J.; Pfennig, B. W.; Schoonover, J. R.; Timpson, C. J.; Wall, J. F.; Kobusch, C.; Chen, X.; Peek, B. M.; Wall, C. G.; Ou, W.; Erickson, B. W.; Bignozzi, C. A. Inorg. Chem. 1994, 33, 3952. (59) Falares, P. Sol. Energy Mater. Sol. Cells 1998, 53, 163. (60) Tang, H.; Berger, H.; Schmid, P. E.; Le´vy, F. Solid State Commun. 1994, 92, 267. (61) Liqiang, J.; Yichun, Q.; Baiqi, W.; Shudan, L.; Baojiang, J.; Libin, Y.; Wei, F.; Honggang, F.; Jiazhong, S. Sol. Energy Mater. Sol. Cells 2006, 90, 1773. (62) Nakajima, H.; Itoh, K.; Murabayashi, M. Bull. Chem. Soc. Jpn. 2002, 75, 601. (63) Anpo, M.; Tomonari, M.; Fox, M. A. J. Phys. Chem. 1989, 93, 7300. (64) Lindstro¨m, H.; Rensmo, H.; So¨dergren, S.; Solbrand, A.; Lindquist, S.-E. J. Phys. Chem. 1996, 100, 3084. (65) Earle, M. D. Phys. ReV. 1942, 61, 56. (66) Parker, J. C.; Siegal, R. W. Appl. Phys. Lett. 1990, 57, 943. (67) Howe, R. F.; Gra¨tzel, M. J. Phys. Chem. 1985, 89, 4495. (68) Weidmann, J.; Dittrich, Th.; Konstantinova, E.; Lauermann, I.; Uhlendorf, I.; Koch, F. Sol. Energy Mater. Cells 1999, 56, 153. (69) Newton, M. D. Chem. ReV. 1991, 91, 767. (70) Shoute, L. C. T.; Loppnow, G. R. J. Am. Chem. Soc. 2003, 125, 15636.