Preparation and Characterization of Pure Rutile TiO₂ Nanoparticles for Photocatalytic Study and Thin Films for Dye-Sensitized Solar Cells

Abstract

Pure rutile-phase TiO₂ (r-TiO₂) was synthesized by a simple one pot experiment under hydrothermal condition using titanium (IV) n-butoxide as a Ti-precursor and HCl as a peptizer. The TiO₂ products were characterized by XRD, TEM, ESCA, and BET surface area measurement. The r-TiO₂ were rodlike in shape with average size of ∼61×32 nm at hydrothermal temperature of 220°C for 10 h. Hydrothermal treatment at longer reaction time increased the tendency of crystal growth and also decreased the BET surface area. The degradation of methylene blue was selected as a test reaction to confer the photocatalytic activity of as-obtained r-TiO₂. The results showed a strong correlation between the structure evolution, particle size, and photocatalytic performance of r-TiO₂. Furthermore, the r-TiO₂-based solar cell was prepared for the photovoltaic characteristics study, and the best efficiency of ~3.16% was obtained.

1. Introduction

Titanium dioxide (TiO₂), is one of the most popular and promising materials in the field of photocatalytic applications due to its strong oxidizing power, high photostability, and redox selectivity [1]. When TiO₂ is irradiated by photons with an energy higher than or equal to its band gap (~3.2 eV), through photon absorption, the electrons can be promoted to the conduction band, generating holes in the valence band. The photogenerated electrons and holes migrate to the TiO₂ surfaces where they can induce reduction and/or oxidation of adsorbed molecules. TiO₂ is also a commonly used semiconductor for photon-electron transfer processes. In the dye-sensitized solar cells (DSSCs) invented by Grätzel et al., nanosized TiO₂ particles were used for preparing working electrodes, and the cell performance was found to be improved significantly when compared to the flat layered photoelectrodes [2]. The TiO₂ crystal exists in two major forms: rutile and anatase [3, 4]. Anatase is thermodynamically metastable and can be transformed irreversibly to rutile phase at high temperatures [3, 5]. Most of the chemistry researchers have paid greater attention to anatase TiO₂ than rutile TiO₂ (r-TiO₂) in both photocatalytic reactions and photoelectrochemical cell because anatase phase of TiO₂ had been considered to be more active than rutile. Several excellent properties of r-TiO₂, such as chemical inertness, superior light scattering characteristics, and lower cost [3, 6], however, make it a potentially important phase in photocatalytic and photovoltaic applications. Wang et al. reported the high photocatalytic activity of r-TiO₂ for decomposition of rhodamine-B in water under artificial solar light irradiation [7]. Bacsa and Kiwi found that the presence of r-TiO₂ showed enhanced catalytic activity compared to pure anatase TiO₂ during the degradation of p-coumaric acid [8]. Rutile phase has also been shown to be more active than anatase in photodecomposition of H₂S [9], and photooxidation of H₂O with Fe³⁺ [10]. Park et al. showed that the photovoltaic characteristics of rutile TiO₂-based DSSCs are comparable to those of anatase TiO₂-based solar cells [11, 12]. However, due to the insufficient TiO₂ film thickness which is less than 5 μm, the electron injection current and the photon to electron conversion efficiency are limited. It has been proposed that porous film electrodes composed of one-dimensional (1D) nanomaterials provide direct electron conducting channels to the electrodes, thus the solar cell efficiency can be enhanced [13]. Previously, we had successfully prepared the 1D rod-shaped r-TiO₂ nanoparticles by sol-hydrothermal procedure [14]. In continuation of the previous work, we are presenting here with more detailed investigation of the influence of hydrothermal conditions, including the acid concentration and hydrothermal duration on the crystal structure, particle size, particle morphology, and photocatalytic activity of r-TiO₂ nanoparticles. The photocatalytic activity of the derived r-TiO₂ photocatalysts was tested by methylene blue degradation reactions under UV illumination. We also prepared nanocrystalline r-TiO₂ films up to 21 μm in thickness for DSSC study. The effect of r-TiO₂ film thickness and morphology on the photoelectrochemical properties was examined. The overall efficiency of r-TiO₂-based DSSC device is rationalized in terms of r-TiO₂ film thickness and amount of dye adsorbed into the electrodes to find a meaningful property-efficiency correlation.

2. Experimental

2.1. Hydrothermal Synthesis of Rutile TiO₂ Nanorods

Figure 1 shows the schematic diagram for preparing r-TiO₂ nanorods. This procedure is based on the previous sol-hydrothermal process for synthesis of TiO₂ nanoparticles [15] with some modification. Titanium (IV) n-butoxide (Ti(O-Bu)₄, ACROS), as a Ti precursor, was added slowly to hydrogen chloride solution (HCl) under magnetic stirring until a clear sol was formed. Then, distilled water was added dropwise into the sol and continuously stirred for three days. The white mixture formed was then transferred to a Teflon-lined autoclave, 40% filled, and heated at 220°C for various durations. After cooling to room temperature, the products presented in the bottom layer were washed with distilled water several times and finally dried at 150°C to obtain crystallized products. The such-obtained TiO₂ samples were characterized by the transmission electron microscope (TEM, Hitachi, H-7100) for microstructural properties and X-ray diffraction (XRD, Japan MAC Science, MXP 18) for crystalline phase. All peaks measured by XRD analysis were assigned by comparing with those of JCPDS data. The crystal size of the TiO₂ during different states of heat treatment was obtained by the XRD line profile analysis. The Brunauer, Emmett, and Teller (BET) surface area was obtained from nitrogen adsorption-desorption data (Micromeritics, ASAP-2010). The chemical composition was verified using the electron spectroscopy for chemical analysis (ESCA, VG Scientific ESCALAB 250).

Figure 1 

Flow chart of the method for preparing r-TiO2 by sol-hydrothermal synthesis.

2.2. Photocatalysis by Rutile TiO₂ Nanorods

The photocatalytic reaction was carried out in a custom-made photoreactor (FanChun Technology Inc., PR-2000) [16]. The system is open to air atmosphere with sixteen UV-lamps in total (wavelength 253.7±0.8 nm, Sankyo Denki Co., LTD.) circulating a quartz reaction cell. The power at the position of the reactor center, measured in the air by a power meter (Molectron, PM 150x), was about (1.2±0.2) mWcm⁻². UV-Vis spectrometry (JASCO V-630) was used to monitor the absorption spectra of methylene blue (MB) as a function of illumination time. Before the photoreaction experiment, the aqueous solution of MB with initial concentration ([𝐶₀]) of 5×10⁻⁵ mol/l (M) was stirred utterly in the presence of r-TiO₂ sample in the dark to ensure the complete equilibrium of adsorption process. During illumination, a 4 mL aliquot was sampled at various time intervals and centrifuged to separate MB solution for analysis. The photocatalytic efficiencies of the r-TiO₂ samples prepared by hydrothermal treatment for various duration (1, 5, 10, 15, 20, and 24 h) were compared.

2.3. Preparation of Rutile TiO₂ Photoanodes and DSSC Performance Measurement

A doctor-blading technique was used to prepare the r-TiO₂ films on an FTO- (F-doped tin oxide-) coated conductive glass (2×3.3×0.3 cm³, Solaronix, sheet resistance 8 Ωcm⁻²) as DSSC photoanodes. Two edges of the FTO substrate were covered with Scotch tapes. The r-TiO₂ sol obtained after hydrothermal treatment for 10 h was directly applied to one of the bare edges and flattened with a home-made doctor blade by shearing across the tape-covered edges. The resulted r-TiO₂ electrodes were dried at 100°C for 15 min followed by subsequent calcination at 450°C for 30 min in order to remove the organic residues from the final products and to complete the crystallization. Thickness of the film was controlled by multiple coating process in which the coated substrates were subjected repeatedly to doctor-blade coating, drying, and calcination steps (Figure 1). The thickness of the r-TiO₂ films was measured by a Mahr Alpha-step profiler (Perthometer S2) and confirmed by the scanning electron microscopy (SEM, Hitachi S-2400) images of cross sections. The surface morphology and crystal phase of r-TiO₂ films were investigated by SEM and XRD, respectively.

For photosensitization studies, the r-TiO₂ electrodes with working area of 0.25 cm² were immersed in ethyl alcohol containing 3×10⁻⁴ M N3 dye (Ru[L2(NCS)2], L = 2,2^'-bipyridine-4,4^'-dicarboxylic acid, Solaronix) for 24 h at room temperature. The Pt counter electrodes with mirror finish were prepared by sputtering deposition (Hitachi E-1045 ion sputter) of a 20 nm layer of Pt on top of FTO substrates. To assemble the DSSC, the electrolyte of 0.5 M LiI (Acros, 99%), 0.05 M I₂ (Showa, 99.8%), 0.5 M 4-tert-butylpyridine-TBP (Aldrich, 99%), and 0.5 M 1,2-dimethyl-3-propylimidazolium iodide-DMPII (IonLic-Tech. >98%) in acetonitrile was applied to the Pt electrode, which was then placed over the dye-coated r-TiO₂ electrode to form a sandwich-type clamped cell for photovoltaic study.

The photocurrent versus voltage (I-V) curves were measured using a computerized digital multimeter (Keithley, 2400) under the AM1.5 irradiation (1 sun), provided by a class A Thermo Oriel Xenon lamp light source (300 W). The incident power density was 100 mW cm⁻² using NREL-calibrated monocrystalline Si-Solar cell (PVM134 reference cell, PV Measurement Inc.) for calibration. The efficiencies were calculated by Forter software.

3. Results and Discussion

3.1. Characterization of Rutile TiO₂ Nanorods

The sol-hydrothermal reaction employed in the present work led to the formation of nanocrystalline TiO₂. The crystal properties depend on the peptization and hydrothermal treatment, such as acid concentration and time period. The previous study showed that using different acids in a hydrothermal reaction resulted in the formation of different TiO₂ phase [17]. The product was pure rutile from an HCl medium, however, if the concentration of HCl was reduced to 1.5 M, besides rutile, anatase phase was generated as a side product. This implied that the formation of anatase by using HCl as a pepitizer was more difficult than that of rutile phase. Based on this result, we selected an HCl concentration of 3 M for the preparation of r-TiO₂ in the study presented here. Figure 2 shows a series of XRD patterns for r-TiO₂ after hydrothermal treatment at 220°C for various duration. For convenience, the r-TiO₂ samples are hereinafter abbreviated as Ht-A-B, where A and B represent the hydrothermal temperature (°C) and heating duration (h), respectively. It can be seen that pure rutile TiO₂ could be successfully obtained with 3 M HCl for the studied hydrothermal period from 1 to 24 hrs. This result infers that the complete formation of rutile phase could be accomplished in the solution at relatively low temperatures under appropriate conditions. In a general sol-gel process for preparing TiO₂, the primary formed structure phase observed at low temperature is anatase which transforms to thermodynamically most stable and more condense rutile phase only upon calcinating at temperature above 500°C [1]. Note that the rutile XRD peaks became sharper as hydrothermal treatment prolonged indicating the formation of larger r-TiO₂. The crystal sizes of r-TiO₂ obtained by analyzing the half maximum (FWHM) of (110) peak at 2𝜃=27.45 degree using the Scherrer equation with wavelength of the radiation of 1.5405 Å are summarized in Table 1.

Table 1 

The XRD crystal domain, TEM particle size, BET surface area of r-TiO₂ samples, and their photocatalytic activities for photodecomposition of methylene blue in aqueous solution.

Figure 2 

A series of XRD peaks for the prepared r-TiO2 powders after hydrothermal treatment for various duration.

The above XRD results indicate formation of r-TiO₂ crystals that can be confirmed by the ESCA measurement. Figure 3 shows the typical ESCA survey spectra of r-TiO₂ samples. The peaks appearing on the left side showed the Ti2p doublet with bonding energies of 459.4 eV for Ti2p_3/2 and 464.9 eV for Ti2p_1/2. The right side spectrum showed the O1s peak with binding energy of 530.4 eV. Those binding energies of Ti2p and O1s from r-TiO₂ samples are all in good agreement with those in standard spectrum of TiO₂. Compared to the individual photoelectron peaks for O1s and Ti2p from nonbonded Ti and O elements, different binding energies due to the chemical shifts were found. The binding energy of the core level electron depends critically on the species to which it is bonded. Charge transfer from Ti to O leaves Ti (and O) with partial positive (and negative) charges, leading to a shift in core level to higher (Ti2p_3/2, 453.7 to 459.4 eV; Ti2p_1/2, 461.2 to 464.9 eV) and lower (O1s, 531 to 530.4 eV) binding energies associated with increased (and decreased) Coulombic attraction between core electron and the nucleus of Ti (and O) [14].

Figure 3 

The ESCA spectra of r-TiO2 sample for Ti 2p (left side) and O 1s (right side).

The crystal growth in different stage of hydrothermal process was traced by TEM. Figure 4 shows the TEM micrographs of r-TiO₂ samples hydrothermally treated at 220°C for (a) 1, (b) 5, (c) 10, (d) 15, (e) 20, and (f) 24 h. As shown in Figure 4, in the initial stage of newly formed TiO₂ sol, the shape of the TiO₂ was observed as elliptical (Figure 4(a)) with average size of 26.3×10.8 nm. Upon increasing the autoclaving time, the r-TiO₂ crystallites build up rod-like morphology progressively. Prolonging the hydrothermal treatment also increases the average particle dimension based on the weighted-average analysis. In other words, increasing autoclaving time promotes the tendency of crystal growth under the present experimental conditions as expected from XRD analysis. In addition, a broad particle size distribution (not shown) was observed in terms of width and length, indicating that the nucleation of r-TiO₂ is much slower than its growth. The range of particle size is from 26.3×10.8 to 76.9×36.5 nm in length × width as listed in Table 1. Nevertheless, the crystals of small size from XRD indicate the incomplete crystallization for hydrothermal treatment up to 24 h.

(a)

(b)

(c)

(d)

(e)

(f)

Figure 4 

The TEM micrographs of r-TiO2 samples hydrothermally treated at 220°C for (a) 1, (b) 5, (c) 10, (d) 15, (e) 20, and (f) 24 hrs.

The size variation of r-TiO₂ can also be seen from BET surface area measurement. The results listed in Table 1 show the dependence of the surface area of r-TiO₂ on the hydrothermal reaction time. It can be seen that the specific surface area shifts towards smaller values for longer heat treatment. The TiO₂ sample with one hour of hydrothermal treatment at 220°C possesses high specific surface area (162 m²/g) which then decreased appreciably with a limited value around 19 m²/g after 24 h of hydrothermal treatment. This result confirms the observation from TEM and indicates an increase in the particle size of r-TiO₂ in increasing the reaction time. Obviously, this is due to the progressive aggregation of small crystallites into larger particles. It is known that the TiO₂ crystals grow from TiO₆ octahedra that are terminated by the surface Ti-OH groups. During the hydrothermal treatment under acidic condition, the surface hydroxyl groups can be protonated to form Ti-OH⁺₂ which then combines readily with another Ti-OH to form Ti-O-Ti oxygen bridge by eliminating a water molecule (dehydration) through which the crystals grow to a larger size [4]. It is noted that the growth of rutile TiO₂ proceeded via oriented coalescence of the first formed TiO₂ nanorods as demonstrated by many side-by-side aggregated r-TiO₂ (Figures 4(c)–4(f)) and the simultaneous increase of width and length of r-TiO₂ from TEM analysis.

3.2. Photocatalytic Activity of Rutile TiO₂

In order to study the photocatalytic activity of the above-prepared r-TiO₂, the photodecomposition of methylene blue (MB) was investigated in aqueous heterogeneous suspensions under the acidic condition. We chose pH 3.85 to study the photodegradation of MB because it decomposed scarcely in the absence of TiO₂ upon irradiation up to 7 h [16]. The maximal absorption of MB solutions is at 614 and 664 nm under our experimental conditions. The photodegradation was studied by monitoring the variation of intensity at 664 nm. Figure 5 plots the relative concentration [𝐶]/[𝐶₀] of MB against irradiation time of r-TiO₂ samples prepared by hydrothermal treatment at 220°C for different time period. With r-TiO₂, the MB showed significant decrease in the absorbance upon irradiation. It is noted that varying the hydrothermal-treated period changes the photocatalytic efficiency of r-TiO₂. The photocatalytic activity of r-TiO₂ increases with an increase in hydrothermal time from 1 to 15 h, but it decreases for r-TiO₂ with further hydrothermal treatment (20 and 24 h). There are two major variables that can vary during the hydrothermal process: crystallinity and surface area. Increasing the hydrothermal time increases the crystalline domain of r-TiO₂ based on XRD analysis but decreases the specific surface area based on BET measurement (Table 1). Increase in crystallinity is a positive change in photocatalytic activity since amorphous titania is known to have very low photocatalytic efficiency [1, 18]. Decrease in surface area, on the other hand, is a negative change in photocatalytic activity due to the reduction of surface hydroxyl groups (−OH). The photocatalysis is basically a surface phenomenon that is being very sensitive to the amount of surface OH groups which may act as the principal reactive oxidant in the photoreactions of TiO₂ [19]. To derive the kinetic information, the decay of absorption due to the photodecomposition of MB was tentatively assumed to follow the first-order kinetics: rate=−𝑑[𝐶]/𝑑𝑡=𝑘_𝑎[𝐶], where 𝑘_𝑎 is the apparent rate constant for MB decomposition and [𝐶] is the concentration of MB. To determine the reaction rate constant, curves of the variation of MB concentration as a function of illumination time were fit into this model. The rate constants for photodecomposition of MB using various r-TiO₂ samples are also listed in Table 1. At longer hydrothermal-treated period (up to 15 h), the MB decomposition rate increases which is associated with the improvement of r-TiO₂ crystallinity. Further hydrothermal treatment for more than 20 h, the photocatalysis efficiency of r-TiO₂ is deteriorated, which is believed to be due to the decrease in surface area.

Figure 5 

The variation of MB intensity at the 𝜆max=664 nm as a function of UV irradiation time in the absence (black solid square) and presence of r-TiO2 samples prepared for various hydrothermal time.

3.3. Application of Rutile TiO₂ to DSSCs

DSSC is a quite complicated system, and there are many factors influencing the cell efficiency [20]. One of the most important parameters is the TiO₂ electrode. The crystal phase, particle shape, diameter, and surface composition of TiO₂ samples used will affect the dye adsorption, electron transport, and electrolyte diffusion in the cell as well as the light-to-electricity conversion efficiency. In this work, we chose r-TiO₂ samples obtained after 10 h hydrothermal treatment to prepare the photoanodes for DSSC study because of higher surface area with good crystallinity. As shown in Table 1, the specific surface area shifts towards smaller values for further heat treatment. The film preparation procedure and condition also play significant effect in the resultant electrode property, in particular, the film morphology and porosity. The r-TiO₂ electrodes were prepared according to the procedures described in Section 2.3. The XRD pattern as shown in Figure 6 exhibited peaks corresponding to rutile phase TiO₂ indicating the presence of stable rutile phase after 30 min 450°C calcination. Several small peaks appear at 2𝜃=38.0, 42.6, and 51.8 are assigned to SnO₂ from FTO substrate. In order to see the heat treatment effect on r-TiO₂ morphology, the TEM image was acquired from calcinated r-TiO₂ paste. As shown in Figure 7, the r-TiO₂ powders kept the rod shape morphology even with 450°C calcination process. Contrarily, the rod-like microstructure is somehow diminished on calcinated r-TiO₂ electrodes prepared from the same pastes as shown in Figure 8 from SEM. Only irregular and shorter-rod particles with larger particle size (100 nm) were observed. Although it is not clear why the rod shape can not be kept on r-TiO₂ film, there could be one possible reason as depicted in Figure 9. The surface energy effect is beneficial to the side- (length-) by-side (length) arrangement between rod-shaped r-TiO₂ particles. As mentioned earlier, the growth of rutile TiO₂ proceeded via oriented coalescence of the first formed TiO₂ nanorods. In 3 dimensions, the r-TiO₂ nanorods are freely moved and chances are the simultaneous growth of r-TiO₂ in both length and width directions. On r-TiO₂ film, however, one of the dimensions for r-TiO₂ nanorods to move is limited and only nearby r-TiO₂ nanorods aggregate to form larger but shorter r-TiO₂ particles.

Figure 6 

The XRD spectra for r-TiO2 electrode and r-TiO2 powder after calcination at 450°C.

Figure 7 

TEM picture of 450°C calcined r-TiO2 powder.

Figure 8 

A top-view SEM image of r-TiO2 electrodes.

(a)

(b)

Figure 9 

Model of the crystal enlargement of r-TiO2 upon calcination on substrate-free (a) and substrate-limited (b) conditions.

Figure 10 showed the typical current-voltage (J-V) characteristics of N3-sensitized r-TiO₂ solar cells measured at 1 sun light intensity for various r-TiO₂ film thickness. Table 2 lists the photoelectric data, including photocurrent density (J_sc), open-circuit voltage (V_oc), fill factor (FF), apparent cell efficiency (η), and the dye adsorption density (Dye_ads) of the DSSCs in Figure 10. It is generally expected that the DSSC performance largely depends on the TiO₂ film thickness because changing the film thickness changes the amount of dye adsorbed on TiO₂ owing to the change of total TiO₂ surface area. For the homogeneously harvested light by the adsorbed dye molecules, one would also expect the approximately linearly dependence of photocurrent density with the film thickness [21]. It can be seen from Table 2 that when the film thickness increased by 2-fold, from 5.5 to 11.3 μm, the J_sc only improved by 26%, from 5.99 to 7.55 mA/cm². The increase of J_sc, however, is consistent with that of Dye_ads which increases ~27%, from 0.22 to 0.28 μmole/cm² and that of cell efficiency by ~26% from 2.51 to 3.16. These results indicate that the photocurrent as well as cell efficiency are limited by the number of adsorbed dye molecules for film thickness up to 11.3 μm. A small decrease (5%) of V_oc with film thickness may be due to the offsetting effect associated with the J_sc and the number of recombination centers on surface area of the film [12]. Further increase of r-TiO₂ film thickness from 11.3 to 21.0 μm, although the amount of dye adsorption increases by 43%, from 0.28 to 0.40 μmole/cm², the J_sc decreases by 20% from 7.55 to 6.03 mA/cm² as well as cell efficiency by ~15% from 3.16 to 2.68. This is due to the increase in the numbers of recombination centers and the longer mean path of the injected electron to travel inside the cell. Moreover, when the TiO₂ films are thicker, the films become more opaque. Due to the various light absorption mechanism, the irradiation intensity decays significantly upon travelling through the thick films which is therefore detrimental to the DSSC performance [21].

Table 2 

Performance characteristics of DSSCs based on the r-TiO₂ electrode with different thickness.

Figure 10 

Photocurrent-voltage characteristics of the dye-sensitized solar cells with different thickness of r-TiO2 films.

4. Conclusions

Pure rutile phase TiO₂ nanorods have successfully been synthesized under the hydrothermal conditions. Hydrothermal-treated duration shows significant effects on the crystal domain, particle dimension, and photocatalytic activity of r-TiO₂. Hydrothermal treatment at longer reaction time increases the tendency of crystal growth based on TEM/XRD and the BET surface area decreased as well. The photocatalytic activity of r-TiO₂ increases with an increase of hydrothermal time from 1 to 15 h due to the increase of crystal domain, but it decreases for r-TiO₂ with further hydrothermal treatment (20 and 24 h) due to the decrease in surface area. Dye-sensitized solar cells with working area of 0.25 cm² were fabricated from various thicknesses of electrode layers made of r-TiO₂ nanorods. The best-performing DSSC evaluated under 1 sun condition gave current density ~7.55 mA/cm², open circuit voltage ~0.70 V, fill factor ~60%, and energy conversion efficiency ~3.16%.

Acknowledgments

The authors gratefully acknowledge the financial support from the National Science Council of Taiwan, Republic of China (Contract no. NSC96-2113-M-027-005-MY2). The authors also acknowledge the kind help from Professor H.-W. Fang for film thickness measurement.