An Electrochemical Study on the Nature of Trap States in

Jun 9, 2007 - a TiO2(eCB. -. )(H. +. ) (1). Figure 5. (a) Cyclic ..... T.B. gratefully acknowledges the support of the Austrian Science Fund (Project...
0 downloads 0 Views 417KB Size
Download PDF
9936

J. Phys. Chem. C 2007, 111, 9936-9942

An Electrochemical Study on the Nature of Trap States in Nanocrystalline Rutile Thin Films Thomas Berger, Teresa Lana-Villarreal, Damia´ n Monllor-Satoca, and Roberto Go´ mez* Departament de Quı´mica Fı´sica i Institut UniVersitari d’Electroquı´mica, UniVersitat d’Alacant, Apartat 99, E-03080 Alacant, Spain ReceiVed: February 20, 2007; In Final Form: April 12, 2007

For monocrystalline TiO2 electrodes, capacitive currents are observed at potentials that are negative enough to induce the filling of conduction band states. Nanoparticulate electrodes exhibit, apart from these currents, an additional pair of capacitive peaks at more positive potentials, which can be attributed to charge traps in the band gap. We have taken advantage of the well-defined morphology and crystal structure of three different types of rutile electrodes to investigate the nature of these band gap states. In particular, nanostructured films composed of oriented wires, films of randomly distributed nanoparticles, and smooth single crystals have been used. The analysis of the cyclic voltammetry response reveals a strong dependence of the trap state concentration on the morphological structure of the films. On the basis of results concerning the surface modification of the electrodes, we propose a model with a location of these band gap states at grain boundaries. We report, furthermore, on a new procedure to prepare hierarchically organized nanostructures by direct deposition of nanowires onto nanoparticulate films in aqueous solutions at low temperature. From a practical point of view, this procedure allows for a systematic tuning of the inner surface area and the porosity of the original samples.

1. Introduction Nanosized semiconductors have recently attracted considerable interest because of their altered behavior as compared to bulk systems. Physical and chemical properties often depend on the size and the shape of the oxide nanostructures with a large variety of synthetic routes being available nowadays to systematically tune the particle morphology.1 Application of such materials in solar energy conversion, photoelectrocatalysis, and sensor devices requires their immobilization as thin porous films, which are characterized by a high internal surface area. This high surface area is harnessed in catalytic applications for highly efficient surface reactions and in dye-sensitized solar cells for quantitative adsorption of dye molecules. Electron transport properties of nanoparticulate films differ fundamentally from those observed for bulk semiconductors. Electron diffusion coefficients, for example, were found to be orders of magnitude smaller than in the case of single crystals.2,3 This slow transport has been attributed to a high concentration of localized states, which act as electron-trapping sites. The diffusion coefficient is strongly affected by both the number and the energy distribution of trap states.4 Therefore, great efforts have been made to gain fundamental knowledge of the structure and the energetic position of these trap states.5-11 On rutile TiO2(110) surfaces, oxygen vacancies and hydroxyl groups are the most common point defects, and their electron-trapping nature has been confirmed recently by density functional calculations.5 However, the understanding of charge-transport processes in nanostructured films requires the use of more complex concepts than those employed for single-crystal electrodes, as new types of defects such as undercoordinated surface sites and particleparticle interfaces should be taken into account. In this context, * To whom correspondence should be addressed. Tel: +34 965903536. Fax: +34 965903537. E-mail: [email protected].

the spatial location of transport limiting trap states is an important and controversial issue. Traps in the bulk of the particles,6 at their surface7,10,11 and at the particle-particle interface8,9 (grain boundaries) have been discussed in the literature. However, a thorough knowledge of structure and location of transport limiting traps is lacking so far. It has been shown recently that originally isolated TiO2 nanoparticles get significantly altered by simple hydration and subsequent dehydration at 200 °C.12 A common strategy for the synthesis of nanoporous films consists in the wet deposition of particle suspensions onto conducting substrates followed by thermal annealing and sintering at, typically, 450 °C. The thermal treatment is indispensable to obtain electric conductivity throughout the film and between the film and the substrate. A reduced interparticle distance and the generation of new electronic states indicate intergranular connection. Furthermore, lifetime and diffusion coefficients of charge carriers photogenerated in nanoparticulate films increase with progressive thermal annealing. This behavior has been explained by an improved conductivity between the particles because of both interparticle neck growth and decrease of the carrier trap density at the grain boundaries and at the surface.13 Depending on the initial contact geometry and on the crystallographic alignment between particles, different types of particle-particle interconnection have been observed.14 Unconstrained crystallites showing nearly the same crystallographic orientation can rotate during sintering into complete alignment leading to epitaxial intergrain necks. However, except for crystallites with a high aspect ratio, such a contact geometry seems to be rather a stochastic event. Consequently, the sintering of a random particle agglomeration results rather in a network of crystallographically misaligned crystallites. Partial ordering of nanocrystallites in porous TiO2 films has been obtained by electrophoretic deposition at different temperatures.15 High

10.1021/jp071438p CCC: $37.00 © 2007 American Chemical Society Published on Web 06/09/2007

Trap States in Nanocrystalline Rutile Thin Films degrees of ordering have turned out to correlate with high values of effective electron diffusion coefficients, pointing out the importance of the relative nanocrystal orientation for charge carrier transport in nanoporous films. Therefore, a strategy to overcome the low inherent conductivity of random particle networks in nanoparticulate films is to synthesize structures with a higher degree of order.16 In this context, nanorods aligned in parallel to each other and oriented perpendicularly with respect to the substrate are expected to lead, on the one hand, to a dramatic reduction of grain boundaries and to facilitate, on the other hand, diffusion within the nanoporous network. In agreement, electron transport in films consisting of ZnO columns and applied as electrodes in dye-sensitized solar cells has turned out to be orders of magnitude faster than in nanoparticulate films.17,18 One approach to synthesize ordered nanostructures is the “oriented attachment” mechanism based on surfactantassisted self-assembling processes, which was used to grow a network of anatase TiO2 nanowires.8 High electron-transfer rates were observed in thin film electrodes prepared from those TiO2 network structures. This fast transport was attributed to a low number of grain boundaries, which are expected to act as regions of enhanced recombination because of a high concentration of trapping sites. In this paper, we present a combined structural and electrochemical analysis of rutile electrodes characterized by different morphologies. Nanostructured films consisting of either oriented nanowires or a random particle network and smooth single crystals have been investigated. These rutile samples can be considered as model systems with the following characteristics: single crystal: low surface area, well-defined surface structure, isolated crystal; oriented nanowires: high surface area, well-defined surface morphology, low degree of interconnection of the monocrystalline wires; and nanoparticulate films: high surface area, lower degree of surface definition (elongated in direction [001], but rounded corners), high degree of particle interconnection. To gain additional information, the nanoparticulate samples have been modified by adsorption of an organic molecule (catechol) and by deposition of nanowires onto their surface. The electrochemical characterization in combination with the well-defined crystal structure and morphology of the electrodes under investigation allows us to discuss the spatial location of the band gap states. From a practical point of view, the modification of the nanoparticulate films by nanowire deposition can be highlighted as a new procedure to systematically tune the inner surface area of the TiO2 porous layers, with potential application in different areas such as photocatalysis or dyesensitized solar cells. II. Experimental Section n-Type rutile TiO2(110) single crystals, doped with 0.075 wt % Nb2O5, were purchased from Commercial Crystal Laboratories, Inc. The wafers (10 × 10 × 0.5 mm) were washed with acetone and were activated as described in the literature19 by repeated cycles of chemical etching in 20% HF for 10 min and subsequent thermal treatment at 500 °C for 2 h in air. Rutile nanowire films were prepared by direct deposition from aqueous titanium-oxysulfate solutions onto F:SnO2-coated transparent conducting glass plates (U-type Asahi Glass Co) as previously described in the literature.20 First, a copper wire was attached to the conducting substrate with silver epoxy. The contact area was then sealed by an insulating epoxy resin so that an area of 1 × 2.3 cm was left uncovered. The substrate was washed ultrasonically with 0.1 M NaOH and H2O,

J. Phys. Chem. C, Vol. 111, No. 27, 2007 9937 successively. The precursor solution (3 mM Ti(IV)) was prepared by adding TiOSO4 (Aldrich, 99.99%, 15 wt % solution in sulfuric acid) to aqueous solutions containing hydrochloric acid (HCl). The pH of the solution after 1 h of stirring at room temperature was adjusted to 1. The substrates were immersed into the precursor solution and were maintained at 60 °C for varying deposition times. Nanoparticulate electrodes were prepared by spreading aqueous slurries of commercial rutile TiO2 nanoparticles (NanoRutile, Sachtleben Chemie GmbH) over 1.5 cm2 of F:SnO2coated glass substrates. The suspension was prepared by grinding 1 g rutile powder with 3.2 mL H2O, 60 µL acetylacetone (99+%, Aldrich), and 60 µL Triton X (Aldrich). Typically, 10 µL of this suspension was applied per substrate. Afterward, the films were annealed and sintered for 1 h at 450 °C in air. The coated glass substrates were then electrically contacted as described above. The crystal phase of the TiO2 electrodes was determined by Raman spectroscopy (LabRam spectrometer, Jobin-Yvon Horiba), and the film thickness was measured by scanning electron microscopy (SEM, Hitachi S-3000N). Atomic force microscopy (AFM) was performed using a Nanoscope III (Digital Instruments) operated at room temperature in air. Images were obtained in tapping mode using silicon tips at a driving frequency of ∼270 kHz. After removing the oxide films from the substrates, transmission electron microscopic (TEM) measurements were carried out with a JEM-2010 (JEOL) microscope equipped with an INCA Energy TEM100 (Oxford instruments) for energy-dispersive X-ray analysis (EDX). Images were recorded with a MegaView II camera (SIS). Electrochemical measurements were performed at room temperature in a standard three-electrode cell. All potentials were measured against and are referred to a Ag/AgCl/KCl(sat) reference electrode, whereas a Pt wire was used as a counter electrode. Measurements were performed with a computercontrolled Autolab PGSTAT30 potentiostat. In all experiments, a N2-purged 0.1 M HClO4 solution (Merck, 70%, Suprapur) in ultrapure water (Millipore Elix 3) was used as working electrolyte. Cyclic voltammograms (CV) were recorded using a scan rate of 20 mV/s. The current and charge densities are given on the basis of the geometric area. III. Results and Discussion In this section, we will present first the morphological and structural characterization of the TiO2 samples by AFM, TEM, and Raman spectroscopy. Second, rutile electrodes based on these well-defined morphologies will be characterized by cyclic voltammetry, emphasizing the different magnitude of the capacitive peaks located at the onset of the electron accumulation region for different electrodes. Finally, the results obtained for modified nanoparticulate electrodes will be shown, leading to the discussion about the spatial location of the traps corresponding to these peaks. A. General Characterization of Rutile Electrodes. 1. Single-Crystal Electrodes. Smooth surfaces with a clear-cut terrace structure were observed after repeated cycles of chemical etching and thermal annealing at T ) 500 °C by AFM (Figure 1). The average step height was found to be 0.35 nm in accordance with previous reports.19 Admittedly, numerous atomic clusters still remain on the terraces after such a treatment. The surfaces turned out to be essentially stable in aqueous 0.1 M HClO4 for hours. A Raman spectrum of the Nb-doped rutile TiO2(110) is shown in Figure 2a, together with the corresponding spectra for the other samples employed in this

9938 J. Phys. Chem. C, Vol. 111, No. 27, 2007

Figure 1. AFM image (a) and step profile (b) of a Nb-doped rutile (110) surface after repeated etching in 20% HF and thermal annealing at 500 °C. The straight line in the AFM image indicates the location where the step profile has been determined.

Figure 2. Raman spectra for a rutile (110) single crystal (a), a nanostructured film (b), and a nanoparticulate film (c). The peak marked with the asterisk (Figure 4b) arises from a contribution of the conducting glass substrate to the overall spectrum because of the low thickness of the nanowire films. Excitation line: 632.8 nm.

study (see below). The phonon bands were assigned to rutile according to the literature.21 2. Nanostructured Films Consisting of Oriented Nanowires. Direct deposition of TiO2 onto conducting glass substrates from aqueous titanium-oxysulfate solutions20,22 leads to the formation of oriented nanowires with a mean diameter of 2 nm and an estimated length of up to 350 nm (Figure 3a and 3b). It was shown previously that the nanowires grow in the [001] direction.20,22 The transmission electron micrograph (Figure 3a) as well as the AFM image (Figure 3c) shows that bundles of parallel nanowires form secondary units with a diameter of ∼20-70 nm. The films thus consist of a nanoscaled hierarchical structure. Nanowire formation results from anisotropic crystal growth because of significant differences in the surface free energies of crystallographic planes. The reaction rate and consequently the direction of crystal growth can be controlled by adsorption of surfactant molecules. In the case of nanowire deposition from aqueous titanium-oxysulfate solutions, selective adsorption of SO42- ions on specific surfaces perpendicular to the (001) one causes the pronounced aspect ratio morphology.20,22 These adsorbed species prevent, furthermore, the

Berger et al. contact between the faces onto which adsorption has selectively occurred, inhibiting the aggregation of the wires. After deposition, the oxide film contains around 2.5% of S as determined by EDX. The film thickness can be tuned by varying the deposition time as determined by SEM measurements. During the first 4 h of deposition, the film thickness increases linearly with deposition time and levels off thereafter (see below). After 6 h, the film thickness attains a value of 350 nm. For times up to 6 h, essentially smooth films are formed showing only minor irregularities on a sub-micrometric level because of bundles of nanowires (Figure 3c). However, after prolonged deposition, isolated clusters with a diameter of several micrometers, consisting once again of nanowires as the primary building unit, are formed on top of the smooth films causing a high film roughness. Raman spectroscopy (Figure 2b) as well as X-ray diffraction (not shown) reveals that the films are exclusively composed of rutile phase. This is remarkable as it was previously observed23 that the relative phase stability of rutile and anatase are usually reversed at small particle sizes as a consequence of the contribution of the surface free energy to the total free energy. Because surface energies for anatase faces are generally lower as compared to the thermodynamically stable rutile modification,24 the nanostructure stabilizes the anatase phase, which is not the stable one for a bulk material. In fact, for particle sizes smaller than 14 nm, anatasesalthough metastable from a thermodynamic point of viewsis the stable polymorph.23 For this reason, synthesis of pure rutile nanostructures with particle dimensions smaller than 14 nm is a challenging and important task. In addition, these materials allow one to investigate the influence of the nanostructure on the physical and chemical properties of rutile TiO2 by comparison of well-defined nanocrystals with single crystals, which have been extensively investigated in the past.25 Semiconductor electrodes used in photoelectrocatalysis are complex systems because of both their ill-defined surface structure and their high concentration of different defects. These “real” systems represent a major challenge for the modeling and understanding of the underlying photocatalytic processes, which require a fundamental knowledge of the interplay between the structure and the physical and chemical properties of the materials. However, a promising strategy for the elucidation of these processes is to systematically reduce the complexity of real systems by using model systems such as single crystals or morphologically well-defined nanostructured films, such as those made of nanowires. 3. Nanoporous Films Consisting of a Random Particle Network. Nanoparticulate electrodes were prepared using commercially available nanosized rutile powder (NanoRutile, Sachtleben Chemie GmbH). This powder is characterized, on the one hand, by its pure rutile structure and, on the other hand, by a well-defined morphology. Figure 4a shows a TEM image of the commercial powder as received. It consists of nanorods with a mean width of 7 nm and a mean height of around 4050 nm. In a previous study, it was estimated that around 9598% of the surface of these nanocrystals is formed by (110) faces.26 Films with a thickness of around 10 µm were prepared by wet deposition of the powder onto the conducting substrate and subsequent thermal annealing at 450 °C. This treatment leads to film sintering, that is, particle-particle interconnection by the formation of grain necks, and to a significant change of the particle morphology (Figure 4b). After annealing, the particles exhibit an approximately ellipsoidal crystal shape (prolate

Trap States in Nanocrystalline Rutile Thin Films

J. Phys. Chem. C, Vol. 111, No. 27, 2007 9939

Figure 3. TEM images (a, b) and AFM image (c) of rutile nanowires prepared by direct deposition from a titanium-oxysulfate solution.

Figure 4. TEM images of Sachtleben rutile nanoparticles; (a) native powder, (b) after wet deposition of a particle suspension and thermal treatment at 450 °C, and (c) particle size distribution after annealing. The solid lines are a guide to the eye.

spheroidlike) with a mean width of 21 nm and a mean length of 46 nm (Figure 4c) as determined by TEM. Furthermore, hexagonal structures appear at the surface of the particles after thermal treatment. Using the mean values of the particle diameters and assuming perfect prolate spheroids, a surface area of 57 m2‚g-1 can be calculated. This value corresponds quite well to the specific surface area of 62 ( 2 m2‚g-1 determined by nitrogen physisorption at 77 K. The thin film preparation has no influence on the initial rutile crystal structure of the particles as evidenced by Raman spectroscopy (Figure 2c) and XRD (not shown). B. Electrochemical Characterization. Nanowire electrodes of varying film thickness were prepared by direct deposition from TiOSO4 solutions. CVs of these electrodes in aqueous 0.1 M HClO4 are shown in Figure 5a. The rather symmetric shape of the CVs indicates, on the one hand, the capacitive nature of the film charging and, on the other hand, an effective charge transport throughout the whole film as evidenced by the absence of any significant resistance.27 Thus, there is no need for further thermal treatment to obtain well-conducting electrodes. The direct deposition method constitutes, therefore, a straightforward and economic way to prepare TiO2 electrodes. On the other hand, thermal treatment normally causes a significant alteration of the initial particle morphology. The fact that, in the present case, no annealing step is required opens up the way for the investigation of morphologically well-defined electrodes. The capacitive charging of the films at potentials E < -0.2 V (Figure 5a) has been attributed to electron accumulation in the conduction band, which is accompanied by proton adsorption from the electrolyte:28

TiO2 + e- + H+ a TiO2(eCB-)(H+)

(1)

Alternatively, these currents have been assigned to the filling

Figure 5. (a) Cyclic voltammograms for rutile nanowire electrodes of different thickness (deposition time: 1-6 h). (b) Film thickness and total cathodic charge accumulated (down to -0.6 V) as a function of deposition time. Electrolyte: 0.1 M HClO4, scan rate: 20 mV‚s-1, electrode area: 2.3 cm2.

of an exponential trap distribution (probably surface states) below the conduction band:27

TiIVO2 + e- + H+ a TiIIIO2(H+)

(2)

A perfect correlation of the accumulated charge and the film thickness (Figure 5b) indicates that the entire inner surface is accessible for protons, as corresponds to a nanoporous substrate.

9940 J. Phys. Chem. C, Vol. 111, No. 27, 2007

Figure 6. (a) Cyclic voltammograms for a nanowire electrode (solid line) and a nanoparticulate electrode (dashed line). (b) Rutile (110) single-crystal electrode; electrolyte: 0.1 M HClO4; scan rate: 20 mV‚s-1.

Importantly, and as deduced from eq 1 or 2, the charge integrated below -0.2 V (for instance, between -0.2 V and -0.6 V) should be proportional to the interfacial area, that is, the area of the TiO2 nanostructure in contact with the electrolyte, which is the value relevant in electrochemistry. This proportionality is perfectly reflected by the correlation of the accumulated charge and the film thickness in Figure 5b. Apart from the cathodic currents at E < -0.2 V, an additional pair of capacitive peaks located between -0.3 and 0.1 V is observed. These peaks are characteristic of nanostructured electrodes and have tentatively been assigned to the reversible filling of surface states below the conduction band edge.7,11,29 The intensity of these peaks scales with the film thickness while their location shifts slightly toward more negative potentials. A very similar behavior is observed for nanoparticulate electrodes prepared from Nano-Rutile powder (Figure 6a; the CV for a 350 nm thick nanowire electrode is also shown for the sake of comparison). A significant difference between nanoparticulate and nanowire films consists in the concentration of band gap states as tracked in the CVs by the pair of capacitive peaks located between -0.3 and 0.1 V. In the case of nanoparticulate electrodes, this concentration (per cm2 of projected area) is enhanced by a factor of ∼15 as compared to the nanowire film (Figure 6a) even when the charge exchanged in the accumulation region (E < -0.2 V), which is proportional to the real surface area, is similar in both cases. An estimate of the absolute number of electrons trapped at these band gap states yields a value of around 60 per particle.30 The total number of charges accumulated in the nanoparticulate film in the whole negative-going scan amounts to 2200 ( 200 per particle. The width of the capacitive peaks between -0.3 and 0.1 V observed for nanowire and nanoparticulate electrodes lies between 0.05 and 0.10 eV. Recently, a phenomenological model was developed which allows one to analyze the basic features of experimental CVs.27 Taking into account exclusively capacitive currents and assuming monoenergetic states, peak widths of ∼2kBT are predicted. This is in line with the peak width reported above (2kBT ) 0.052 eV at T ) 298 K). The sharp

Berger et al. energy distribution of band gap states in the TiO2 electrodes is rather remarkable. For example, band widths between 0.2 and 0.5 eV were observed for color centers at the surface of morphologically well-defined MgO nanocubes, which were generated by photochemical reduction in a hydrogen atmosphere and which were detected under high vacuum conditions, that is, in the absence of a surrounding liquid phase.31 In the case of an electrode, however, a significant broadening of the bandwidth of surface states is expected because of interactions with the surrounding electrolyte. The absence of such a broadening in the case of nanowire and nanoparticulate films makes a location of the traps at the particle surface, therefore, unlikely. The TiO2/solution interfacial area of the nanoparticulate electrode can be estimated using the surface area determined by nitrogen physisorption (62 ( 2 m2‚g-1). In fact, around 1.5‚10-3 g of rutile nanopowder are deposited in the preparation of the electrode, yielding a total surface area of 0.09 m2. Taking into account that the geometric (projected) area of the electrode is 1.5 cm2, the ratio of the real surface area (Areal) to the geometric surface area (Aproj) is 600 for a 10 µm thick film. For the nanowire electrodes, on the other hand, this ratio can be estimated from the cathodic currents of the CV according to the charge Qac integrated between -0.2 V and -0.6 V as

ANP real

QNP ac = NW ANW Q real ac

(3)

and

( ) ( ) ( ) ( ) Areal NP Aproj = Areal NW Aproj

Qacl NP qNP Aproj ac ) Qac NW qNW ac Aproj

(4)

where qac is the respective charge density. These charge densities -2 and qNW ) 5.80 mC‚cm-2 for amount to qNP ac ) 6.96 mC‚cm ac the nanoparticulate and the nanowire electrode, respectively. The value of (Areal/Aproj)NW can thus be determined to be ∼500. It is remarkable that similar (Areal/Aproj) ratios for both nanoporous samples are obtained in spite of the fact that the nanoparticulate film is around 30 times thicker than the nanowire film (dNP ) 10 µm, dNW ) 350 nm). This supports the notion that the nanowire films have remarkably high inner surface areas.22 The present voltammetric study also includes the case of a well-defined rutile TiO2(110) single crystal (Figure 6b). Significant capacitive currents are observed only at potentials E < -0.4 V. Most importantly, the pair of capacitive peaks attributed to band gap states is absent in this case. Finally, we comment briefly on the density of states (DOS) as deduced from the voltammetric responses in the accumulation region.27,32 In particular, it can be checked whether the DOS follows an exponential distribution by plotting the logarithm of the current density versus the applied potential (see Supporting Information). It results that the smaller the size of the particle is, the farther is the DOS from an exponential distribution. A similar tendency has been reported recently for anatase nanoparticulate electrodes.32 In conclusion, it can be stated that the density of electron trap states as tracked by capacitive currents between -0.3 and 0.1 V depends strongly on the morphology of the investigated electrode. No peaks have been observed for a single-crystal electrode. Films consisting of oriented single-crystalline nano-

Trap States in Nanocrystalline Rutile Thin Films

Figure 7. TEM images of nanowires grown onto the crystallites of a nanoparticulate film. Deposition time: 30 min (a + b) and 90 min (c + d). The arrow in Figure 7b indicates the orientation as deduced for the observed, although not discernible in the figure, (110) fringes (d ) 3.25 Å).

wires, on the other hand, exhibit them, but with a low intensity relative to Areal. However, films built up of sintered particles show the highest concentration of the corresponding band gap states relative to Areal. To determine their spatial location, that is, whether they are located at the surface or not, two different strategies of surface modification have been applied to nanoparticulate films, the ones with the highest trap state concentration, as described in the following section. C. Surface Modification of Nanoparticulate Electrodes. 1. Deposition of Nanowires onto Nanoparticulate Films. The surface of nanoparticulate rutile films was modified by deposition of nanowires from aqueous titanium-oxysulfate solutions. First, a nanoparticulate electrode was prepared as described in the Experimental Section. Its CV was recorded in 0.1 M HClO4 and was used as reference. Onto this electrode, nanowires were grown thereafter by repeated cycles of the following experimental steps: rinsing the electrode with H2O and immersion into a titanium-oxysulfate solution (pH 1) at T ) 60 °C for varying times, and rinsing with H2O and measurement of the CVs in 0.1 M HClO4. Such a treatment leads to the coverage of the original particle surface and the growth of nanowires in the [001] direction (Figure 7a and b; the arrow indicates the [001] direction as deduced from the observed (110) fringes, not discernible in the figure). It is remarkable that this deposition procedure opens up a new way to prepare hierarchically organized nanostructures. The possibility of systematically tuning the porosity and the inner surface of the original film may, furthermore, comprise an appropriate way to optimize the (photo)catalytic activity of nanostructured films. After a deposition time of 90 min, all the pores formed by the original nanoparticles are filled with bundles of nanowires (Figure 7c and d). Such a treatment leads to a huge increase of the surface area as can be seen by the dramatic increase of the accumulated charge at potentials E < -0.2 V because of conduction band filling (Figure 8a). However, the intensity of the band gap states at -0.3 V < E < 0.1 V does not change significantly. This indicates that the trap states are influenced neither by the coverage of the surface with nanowires nor by the adsorption of sulfate ions (from the deposition solution).

J. Phys. Chem. C, Vol. 111, No. 27, 2007 9941

Figure 8. (a) Cyclic voltammograms for a nanoparticulate rutile electrode before and after nanowire deposition. Deposition times: 5, 10, 15, 30, 60, and 90 min. (b) Cyclic voltammograms for a nanoparticulate electrode in the absence (dashed line) and in the presence (solid line) of 0.1 M catechol. Electrolyte: 0.1 M HClO4, scan rate: 20 mV‚s-1, electrode area: 1.5 cm2.

2. Adsorption of Catechol. Figure 8b shows the CVs of a nanoparticulate film in the absence and in the presence of 0.1 M catechol. Enediol ligands (such as catechol) have a large affinity for undercoordinated surface sites on TiO2 nanoparticles, restoring the coordination of Ti atoms to the octahedral geometry (bulk geometry).33,34 Therefore, by quantitatively modifying the surface with enediol ligands, the contributions of trapping surface sites would be selectively removed.11 The adsorption of catechol onto the nanoparticulate rutile electrode leads only to a slight shift of the band gap states toward negative potentials (-20 mV), probably caused by changes in the double-layer capacitance in the course of catechol adsorption.35 However, the intensity of the capacitive peaks does not change at all, indicating that the respective trap states are not located at the particle surface. The sharp energy distribution of the band gap states under investigation and the fact that their concentration does not change in the course of important surface modifications, such as nanowire deposition or catechol adsorption, indicate anew that their location at the surface must be doubted. Taking into account the dependence of the capacitive currents on the degree of particle-particle interconnection, which was investigated in the present study by using different morphologically welldefined electrodes, a location at the solid-solid interface seems to be very likely. This interpretation is consistent with the recent findings that charge-transfer reductive doping of nanoparticulate electrodes causes not only a dramatic increase of their photoelectrocatalytic activity but also a shift of the respective capacitive peaks toward positive potentials as explained by band bending throughout particle agglomerates.9 In agreement, these peaks have also been detected for other nanoparticulate semiconductor electrodes made of anatase,7,9,11 ZnO,36 or WO3.37 Electrodes prepared from powders can be considered as a random network of crystallographically misaligned crystallites. Sintering induces the contacting of the particles by neck formation. However, as a consequence of the crystallographic misalignment of the particles, lattice mismatches at the grain

9942 J. Phys. Chem. C, Vol. 111, No. 27, 2007 boundary are inevitable. This discontinuity could be the reason for a high concentration of defect states at particle-particle interfaces. A location of band gap states at the grain boundaries could, furthermore, explain the invariance of the characteristic capacitive peaks discussed in this study upon important surface modification as well as their sharp energy distribution. IV. Conclusions In this work, three different types of rutile electrodes, that is, single crystals, films of oriented nanowires, and films consisting of a random nanoparticle network, which are characterized by well-defined morphology and crystal structure, were investigated. The following conclusions have been drawn: (1) Direct deposition of rutile nanowires onto conducting substrates as described in the literature20 results in electrodes characterized by effective charge transport throughout the whole film. This puts aside the need for further thermal annealing and allows for the investigation of morphologically well-defined electrodes. (2) For nanosized electrodes, composed of either wires or particles, a characteristic pair of capacitive peaks is observed in the CVs between -0.3 and 0.1 V. The energy distribution of the respective band gap states is very sharp (0.05-0.10 eV). Their concentration depends strongly on the morphology of the film and was found to be 15 times higher in the case of a random nanoparticle network as compared to a nanowire film. (3) No corresponding capacitive peaks are observed on smooth TiO2(110) single crystals. (4) Surface modification of nanoparticulate electrodes by nanowire deposition and catechol adsorption has no influence on either the energy or the concentration of the respective band gap states. Combining all these pieces of information, we propose a location of these electron trap states at grain boundaries. Finally, from a more practical perspective, the direct deposition of nanowires onto nanoparticulate films used in this work opens up a new strategy for the systematic manipulation of preformed nanostructures. Acknowledgment. We thank C. Almansa and V. Lo´pezBelmonte for performing the TEM and SEM measurements, respectively. J. Winkler and B. Proft from Sachtleben Chemie GmbH (Duisburg, Germany) are acknowledged for kindly providing us with Nano-Rutile powder. This work was financially supported by the Spanish Ministry of Education and Science (MEC) (CTQ2006-06286 (FONDOS FEDER)) and by the Generalitat Valenciana (GV05/119). D. M. S. acknowledges the Spanish MEC for the award of a FPI grant. T.B. gratefully acknowledges the support of the Austrian Science Fund (Project J2608-N20). Supporting Information Available: Plots of the logarithm of the current density versus applied potential together with a brief discussion. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Young-wook, J.; Jin-sil, C.; Jinwoo, C. Angew. Chem., Int. Ed. 2007, 45, 3414-3439.

Berger et al. (2) Solbrand, A.; Lindstrom, H.; Rensmo, H.; Hagfeldt, A.; Lindquist, S. E.; Sodergren, S. J. Phys. Chem. B 1997, 101, 2514-2518. (3) Dloczik, L.; Ileperuma, O.; Lauermann, I.; Peter, L. M.; Ponomarev, E. A.; Redmond, G.; Shaw, N. J.; Uhlendorf, I. J. Phys. Chem. B 1997, 101, 10281-10289. (4) Nelson, J. Phys. ReV. B 1999, 59, 15374-15380. (5) Di Valentin, C.; Pacchioni, G.; Selloni, A. Phys. ReV. Lett. 2006, 97, 166803-166804. (6) Bisquert, J.; Vikhrenko, V. S. J. Phys. Chem. B 2004, 108, 23132322. (7) Wang, H.; He, J.; Boschloo, G.; Lindstrom, H.; Hagfeldt, A.; Lindquist, S. E. J. Phys. Chem. B 2001, 105, 2529-2533. (8) Adachi, M.; Murata, Y.; Takao, J.; Jiu, J.; Sakamoto, M.; Wang, F. J. Am. Chem. Soc. 2004, 126, 14943-14949. (9) Berger, T.; Lana-Villarreal, T.; Monllor-Satoca, D.; Go´mez, R. Electrochem. Commun. 2006, 8, 1713-1718. (10) Kopidakis, N.; Neale, N. R.; Zhu, K.; van de Lagemaat, J.; Frank, A. J. Appl. Phys. Lett. 2005, 87, 2021061-2021063. (11) de la Garza, L.; Saponjic, Z. V.; Dimitrijevic, N. M.; Thurnauer, M. C.; Rajh, T. J. Phys. Chem. B 2006, 110, 680-686. (12) Elser, M. J.; Berger, T.; Brandhuber, D.; Bernardi, J.; Diwald, O.; Kno¨zinger, E. J. Phys. Chem. B 2006, 110, 7605-7608. (13) Nakade, S.; Matsuda, M.; Kambe, S.; Saito, Y.; Kitamura, T.; Sakata, T.; Wada, Y.; Mori, H.; Yanagida, S. J. Phys. Chem. B 2002, 106, 10004-10010. (14) Rankin, J. J. Am. Ceram. Soc. 1999, 82, 1560-1564. (15) Tirosh, S.; Dittrich, T.; Ofir, A.; Grinis, L.; Zaban, A. J. Phys. Chem. B 2006, 110, 16165-16168. (16) Gra¨tzel, M. J. Photochem. Photobiol., C 2003, 4, 145-153. (17) Galoppini, E.; Rochford, J.; Chen, H.; Saraf, G.; Lu, Y.; Hagfeldt, A.; Boschloo, G. J. Phys. Chem. B 2006, 110, 16159-16161. (18) Martinson, A. B. F.; McGarrah, J. E.; Parpia, M. O. K.; Hupp, J. T. Phys. Chem. Chem. Phys. 2006, 8, 4655-4659. (19) Nakamura, R.; Ohashi, N.; Imanishi, A.; Osawa, T.; Matsumoto, Y.; Koinuma, H.; Nakato, Y. J. Phys. Chem. B 2005, 109, 1648-1651. (20) Yamabi, S.; Imai, H. Chem. Mater. 2002, 14, 609-614. (21) Porto, S. P. S.; Fleury, P. A.; Damen, T. C. Phys. ReV. 1967, 154, 522-526. (22) Yamabi, S.; Imai, H. Thin Solid Films 2003, 434, 86-93. (23) Gribb, A.; Banfield, J. F. Am. Mineral. 1997, 82, 717-728. (24) Lazzeri, M.; Vittadini, A.; Selloni, A. Phys. ReV. B 2001, 63, 1554091-1554099. (25) Diebold, U. Surf. Sci. Rep. 2003, 48, 53-229. (26) Rotzinger, F. P.; Kesselman-Truttmann, J. M.; Hug, S. J.; Shklover, V.; Gra¨tzel, M. J. Phys. Chem. B 2004, 108, 5004-5017. (27) Fabregat-Santiago, F.; Mora-Sero´, I.; Garcia-Belmonte, G.; Bisquert, J. J. Phys. Chem. B 2003, 107, 758-768. (28) Boschloo, G.; Fitzmaurice, D. J. Phys. Chem. B 1999, 103, 78607868. (29) Kavan, L.; Kratochvilova, K.; Gra¨tzel, M. J. Electroanal. Chem. 1995, 394, 93-102. (30) The crystal shape was approximated by perfect prolate spheroids with a mean width of 21 nm and a mean length of 46 nm. One and a half milligrams of oxide thus corresponds to 3.34‚1013 crystallites. The number of electrons was determined from the CVs by fitting the cathodic branch between +0.8 and -0.3 V by the superimposition of an exponential curve (conduction band filling) and a Gauss-shaped curve (band gap states). (31) Berger, T.; Sterrer, M.; Diwald, O.; Kno¨zinger, E. J. Phys. Chem. B 2004, 108, 7280-7285. (32) Abayev, I.; Zaban, A.; Kytin, V. G.; Danilin, A. A.; GarciaBelmonte, G.; Bisquert, J. J. Solid State Electrochem. 2007, 11, 647-653. (33) Lana-Villarreal, T.; Rodes, A.; Pe´rez, J. M.; Go´mez, R. J. Am. Chem. Soc. 2005, 127, 12601-12611. (34) Rajh, T.; Chen, L. X.; Lukas, K.; Liu, T.; Thurnauer, M. C.; Tiede, D. M. J. Phys. Chem. B 2002, 106, 10543-10552. (35) Lana-Villarreal, T.; Monllor-Satoca, D.; Go´mez, R.; Salvador, P. Electrochem. Commun. 2006, 8, 1784-1790. (36) Willis, R. L.; Olson, C.; O’Regan, B.; Lutz, T.; Nelson, J.; Durrant, J. R. J. Phys. Chem. B 2002, 106, 7605-7613. (37) Wang, H.; Lindgren, T.; He, J.; Hagfeldt, A.; Lindquist, S. E. J. Phys. Chem. B 2000, 104, 5686-5696.