TiO2 nanotube membranes on transparent conducting ... - CiteSeerX

Home

Search

Collections

Journals

About

Contact us

My IOPscience

TiO2 nanotube membranes on transparent conducting glass for high efficiency dye-sensitized solar cells

This article has been downloaded from IOPscience. Please scroll down to see the full text article. 2011 Nanotechnology 22 285201 (http://iopscience.iop.org/0957-4484/22/28/285201) View the table of contents for this issue, or go to the journal homepage for more

Download details: IP Address: 218.104.96.137 The article was downloaded on 27/06/2011 at 09:11

Please note that terms and conditions apply.

IOP PUBLISHING

NANOTECHNOLOGY

Nanotechnology 22 (2011) 285201 (9pp)

doi:10.1088/0957-4484/22/28/285201

TiO2 nanotube membranes on transparent conducting glass for high efficiency dye-sensitized solar cells Mukul Dubey1, Maheshwar Shrestha1, Yihan Zhong1 , David Galipeau1 and Hongshan He1,2 1

Department of Electrical Engineering, South Dakota State University, Brookings, SD 57007, USA 2 Center for Advanced Photovoltaics, South Dakota State University, Brookings, SD 57007, USA E-mail: [email protected]

Received 15 September 2010, in final form 6 May 2011 Published 31 May 2011 Online at stacks.iop.org/Nano/22/285201 Abstract Crack-free TiO2 nanotube (NT) membranes were obtained by short time re-anodization of a sintered TiO2 NT array on Ti foil, followed by dilute HF etching at room temperature. The resulting freestanding TiO2 membranes were opaque with a slight yellow color having one end open and another end closed. The membranes were then fixed on transparent fluorine–tin-oxide glass using a thin layer of screen-printed TiO2 nanoparticles (NPs) as a binding medium. It was found that low temperature treatment of the resulting NT/NP film under appropriate pressure before sintering at 450 ◦ C was critical for successful fixation of the NT membrane on the NP layer. The resulting films with open-ends of NT membranes facing the NP layer (open-ends down, OED, configuration) exhibited better interfacial contact between NTs and NPs than those with closed-ends facing the NP layer (closed-ends down, CED, configuration). The cells with an OED configuration exhibit higher external quantum efficiency, greater charge transfer resistance from FTO/TiO2 to electrolyte, and better dye loading compared to CED configurations. The solar cells with the OED configuration gave 6.1% energy conversion efficiency under AM1.5G condition when the commercial N719 was used as a dye and I− /I− 3 as a redox couple, showing the promise of this method for high efficiency solar cells. (Some figures in this article are in colour only in the electronic version)

strips one electron from the electrolyte, the electrolyte then recovers its missing electrons from the counter electrode by diffusion. The key part of this kind of cell is the TiO2 NP layer on FTO glass. These particles are interlinked to each other, forming a matrix with high porosity [3–7], which provides not only large surface area for the adsorption of dyes for sunlight harvesting, but also a network for electron migration to the FTO layer then to an external circuit. So far, 11.5% conversion efficiency of photons to electricity has been achieved from this configuration [8]; however, the efficiency needs to be enhanced for cost-effective mass production. One barrier to achieving high efficiency is electron recombination due to the defect states at grain boundaries in NP matrix [8–13], which leads to significant electron loss and, therefore, low electron collection

1. Introduction Dye-sensitized solar cells (DSCs) are cost-effective electrochemical devices that convert solar energy to electricity [1]. A typical DSC, as proposed by Grätzel and O’Regan in 1991 [2], has a three-component structure: a transparent fluorine–tinoxide (FTO) glass with dye-coated TiO2 nanoparticle (NP) matrix on it as the photoelectrode, a platinum-coated semitransparent FTO glass as the counter electrode, and I− /I− 3containing solution as the redox couple. When the device is exposed to sunlight, the excited electrons in the dye are injected into the conduction band of nanocrystalline TiO2 , the electrons then migrate to the bottom of the TiO2 layer and are collected by the FTO layer. Meanwhile, the dye in the oxidized state 0957-4484/11/285201+09$33.00

1

© 2011 IOP Publishing Ltd Printed in the UK & the USA

Nanotechnology 22 (2011) 285201

M Dubey et al

(DMPII) was purchased from Iolitec Inc. All other chemicals were purchased from Acros and used as received.

efficiency. To overcome this problem, nanotubes [14–22], nanofibers [23] and nanorods [23–27] are introduced into the photoelectrode. In particular, vertically aligned TiO2 nanotube (NT) arrays have attracted wide interest due to their versatile preparation and large aspect ratio. More importantly, TiO2 NTs have shown ∼100% charge collection efficiency [28]. It is challenging to have TiO2 NT arrays on FTO glass such that a Grätzel type DSC configuration can be applied during the fabrication. In Grätzel type DSCs, the TiO2 NP film is prepared by applying a viscous TiO2 NP paste on the FTO glass using doctor-blade or screen-printing methods, followed by high temperature sintering. This method is not applicable to TiO2 NTs since TiO2 NTs are usually prepared by anodization of a Ti foil (usually 250 μm) [19, 29]. The TiO2 NTs thus prepared reside vertically on the remaining Ti foil (denoted as TiO2 NT/Ti), which is difficult to lift off. It has been reported that TiO2 NT arrays were lifted off the remaining Ti foil by immersing as-anodized Ti foil in methanol/Br2 solution under a nitrogen atmosphere [30], extensive consecutive rinsing of as-anodized Ti foil in water and methanol [31], etching asanodized Ti foil in aqueous HCl solution [32], or immersing re-anodized Ti foil in H2 O2 aqueous solution [33]. These methods are either time-consuming or lack of reproducibility. It is even more difficult to fix freestanding TiO2 NT membranes on the FTO glass. Park et al [32] applied a small amount of titanium isopropoxide to fix a TiO2 NT membrane on FTO glass; however, the binding was not very strong and most of the membrane came off the FTO glass after sintering. Recently Chen et al [33] reported that a TiO2 NT membrane could be attached to FTO glass by insertion of a thin layer of TiO2 NP at room temperature, which produced 5.4% energy conversion efficiency. In this paper, we describe reliable methods to obtain freestanding TiO2 NT membranes and to fix them on FTO glass. It was found that TiO2 NT membrane was lifted off the Ti foil substrate immediately after it was immersed into dilute HF aqueous solution. The method involved high temperature sintering, short time re-anodization, and quick immersing of TiO2 NT/Ti foil into a HF aqueous solution. The membranes thus obtained were crack-free and were successfully fixed on the FTO glass with a thin layer of commercial TiO2 NP between them. It was found that low temperature treatment before the samples were sintered was critical to obtain strong physical contact between TiO2 NPs and the NT membrane. The DSC with these films produced 6.1% conversion efficiency under the AM1.5G condition, showing the promise of this method for high efficiency solar cells.

2.2. Fabrication of TiO2 nanotube membranes Freestanding TiO2 NT membranes were prepared by electrochemical anodization of Ti foil in an ethylene glycerol solution with 0.2 M NH4 F, 0.01 M H3 PO4 and 2.2% (w/v) H2 O using Pd foil as the cathode [19]. Prior to anodization, both Ti and Pd foils were degreased by ultrasonication in ethanol and deionized (DI) water several times. The backside of Ti foil was covered tightly by commercial Parafilm. The lower part of the frontside was masked by a circular parafilm mask. The anodization lasted for 5, 7, and 10 h at 60 V DC. Current was monitored with an Agilent 4410 digital multimeter. After anodization the samples were flushed with ethanol and water, air-dried at room temperature, and then sintered at 450 ◦ C for 30 min. The samples were re-anodized in the original electrolyte solution for 30 min at 60 V DC, washed with ethanol, and immersed in an aqueous solution of HF (0.07 M) at room temperature. The TiO2 NT arrays lifted off the remaining Ti foil within 1 min after shaking the foil gently. The resulting TiO2 NT membrane was immersed in a dilute NaOH aqueous solution for 15 min to neutralize the acid and then flushed with DI water several times. 2.3. Fabrication of solar cells The freestanding TiO2 NT membrane was first transferred to a piece of filter paper for drying up, then it was put on the top of the TiO2 NP layer that was screen-printed on FTO glass in advance using a commercial TiO2 nanoparticle paste. After air-drying for about 30 min, a small piece of Parafilm was applied to the top of the film and a ∼100 g piece of metal was put on the top to provide some pressure. The film was then kept in a regular freezer (temperature ∼ −20 ◦ C) overnight, then taken out, removed the parafilm and put in an oven for sintering (450 ◦ C for 30 min) in an oxygen ambient. The sample was cooled to room temperature and was immersed in a TiCl4 aqueous solution (20 mM) for 1 h. The film was air-dried and sintered again at 450 ◦ C for 30 min. The film was then put into a dye solution in ethanol (∼2 mM) for 12 h at room temperature. The film was taken out, flushed with ethanol, and vacuum-dried for 1 h. The counter electrode was prepared by sputtering a 10 nm thick film of Pt on FTO glass. A mask made from Parafilm was used as a spacer for the two electrodes and sealing material. The electrolyte solution was injected from two pre-cut channels in the mask and the final cell was sealed by hot glue. The active area was 0.16 cm2 for all samples. The scanning electron microscopic (SEM) images were obtained on a Hitachi S-3400 SEM. The dye loading of TiO2 films was determined by measuring the absorbance change of the dye solution before and after the dye loading. To a quartz spectrophotometer cuvette was added 2.5 ml of an ethanol solution of dye with known concentration. The absorbance was measured. Then sintered TiO2 NT film on a small piece of regular glass (∼2 mm × 2 mm), prepared using the procedure described above for cell fabrication, was put into the cuvette. The cuvette

2. Experimental section 2.1. Materials Ti foil (2 cm × 1 cm × 0.25 cm; 99.9% purity) and Pd foil (2 cm × 1 cm × 0.25 cm; 99.9% purity) were purchased from Sigma-Aldrich. The dyes (N719, N3, Z907 and black dye) and a TiO2 nanoparticle paste (Ti-Nanoxide T20/SP, diameter 20 nm) were purchased from Solaronix SA (Switzerland). 2,3-Dimethyl-1-propyl imidazolium iodide 2


M Dubey et al

was sealed and kept at room temperature for 12 h. Then the cuvette was shaken gently and the absorbance was recorded. From the absorbance change the dye-loading density (mol of dye per gram of TiO2 ) was determined from a predetermined working plot of dye in ethanol. The total amount of TiO2 (NT and NPs) was determined by the weight difference before and after fixing the TiO2 NT film on the FTO glass. This was completed before the glass was immersed into the dye solution. 2.4. Characterization of solar cells The solar cells thus prepared were characterized for current– voltage measurements with a standard light source (xenon arc lamp) having an intensity of 100 mW cm−2 under AM 1.5G conditions using an Agilent 4155C source measurement unit. The overall energy conversion efficiency and fill factor were calculated by η(%) = Pmax × 100/(Pin A) and FF = Pmax /(Isc Voc ), respectively, where Pmax is the maximum output power of cells, Pin is the power density of the light source, Isc is the short-circuit current, Voc is the open-circuit voltage, and A is the active area of the cell. The electrolyte is an acetonitrile/valeronitrile (v/v: 1:1) solution having 0.6 M 1propyl-2,3-dimethylimidazolium iodide, 0.05 M I2 , 0.1 M LiI, 0.1 M guanidine thiocyanate and 0.5 M tert-butylpyridine. The external quantum efficiency (EQE) experiments were performed on an Agilent semiconductor parameter analyzer, a xenon arc lamp and a Newport monochromator. Monochromatic light was incident on the sample through focusing lenses. A National Renewable Energy Laboratory (NREL) calibrated photodetector was used as reference. The samples were scanned from 350 to 850 nm and voltage was recorded. The EQE was calculated using the equation EQE = (sample voltage × reference voltage)/reference EQE. The reference EQE was supplied by a NREL calibrated Hamamatsu S 1133 photodetector. Electron impedance spectroscopy (EIS) measurements were performed on an HP 4192A LF impedance analyzer (5 Hz–13 MHz) set up in the two-electrode configuration. Ten millivolts of AC perturbation was applied ranging between 5 and 105 Hz. The measurements were performed in the dark under forward bias conditions from 0 to 1 V. A simplified Randles type model was used to fit the data to extract the series resistance ( RS ), which accounts for transport resistance of TCO, the combined charge transfer resistance for electron recombination at the TCO/electrolyte/TiO2 interface ( Rct ), and the constant phase element representing capacitance at TCO/electrolyte/TiO2 interface (CPE1).

Figure 1. Schematic of the TiO2 NT membrane lift off process. The light blue and light yellow colors indicate the amorphous and anatase phases of TiO2 . Ti foil, as anodized TiO2 NTs, and sintered TiO2 NTs are shown in grey, purple and yellow, respectively.

photovoltaic performance. Figure 1 shows two methods for obtaining freestanding TiO2 NT membranes. The light blue and light yellow colors indicate the amorphous and anatase phase of TiO2 , whereas grey stands for Ti foil. The first method is direct immersion of as-anodized TiO2 NT/Ti in a dilute HF aqueous solution (0.07 M) under slight shaking using tweezers. The TiO2 NTs lifted off the Ti substrate within 10 min. We found that the NTs were dissolved completely if the concentration of HF is too high (0.7 M) or immersed in dilute HF acid solution (0.07 M) for a long (30 min) time. The obtained TiO2 NT membranes exhibited strong physical strength and could be transferred to different substrates for further analysis. Under SEM, it was found that the NTs were vertically aligned with one end closed (denoted as CE) and the other open (denoted as OE). The CE surface was very smooth without any destructive features, as shown in figure 2. However, significant morphological changes were observed on the OE surface. The OE surface became very rough and looked like ribbons. The length of the TiO2 NTs for 5 h anodization reduced from ∼22 to ∼10 μm, indicating a significant dissolution of TiO2 . The lift off process is proposed as a result of reaction between amorphous TiO2 , Ti and HF. When the sample was immersed in HF solution it started to remove the thin layer of TiO2 on the edges of the foil. Then it started to react very rapidly with Ti underneath the TiO2 NTs with the evolution of H2 . Meanwhile HF also reacted with amorphous TiO2 . Since its reaction was slower than with Ti, the NT arrays came off the Ti before they were completely dissolved. It is unclear why the CE surface was inert to HF. It is possible that the HF concentration underneath the TiO2 NT arrays was low due to the fast reaction of HF with Ti. Though the lifting off process was fast, the formation of disordered surface structure and shortening of the overall length of the TiO2 NTs led us to develop the second method. The second method involves the re-anodization of a sintered TiO2 NT/Ti sample. In this method the amorphous TiO2 NTs on Ti foil were first converted to the anatase phase by sintering a sample at 450 ◦ C for 30 min, which was

3. Results and discussions 3.1. Fabrication of freestanding TiO2 NT membranes The TiO2 NT arrays were fabricated by the anodization of Ti foil at 60 V DC as reported previously [19]. The NTs were vertically aligned on Ti foil with their top ends open. The surface was clean without any debris or bundling. The lengths of the NTs were 15, 22 and 38 μm for 3, 5 and 10 h anodized samples, respectively. The following experiments were mainly focused on 22 μm-long NTs due to their optimal 3


M Dubey et al

Figure 2. Freestanding TiO2 NT membrane from method 1. (A) SEM image of overall features of the membrane. (B) SEM image of OEs of TiO2 NTs. (C) SEM image of CEs of TiO2 NTs. (D) Optical image of freestanding TiO2 membrane on a piece of color-coded paper.

Figure 3. Freestanding TiO2 NT membrane by method 2. (A) SEM image of overall features of the membrane. (B) SEM image of OEs of TiO2 NTs. (C) SEM image of CEs of TiO2 NTs. (D) Optical image of freestanding TiO2 membranes on a piece of color-coded paper with a ruler.

found to be inert to 0.07 M HF aqueous solution. Then the sample was re-anodized in the same electrolyte solution under same conditions for 30 min. This generated a thin layer of amorphous TiO2 NTs underneath the original TiO2 NT arrays. After immersing this sample in 0.07 M HF aqueous solution, the TiO2 NT membrane came off the Ti foil within 1 min. Figure 3 shows the optical image of six freestanding TiO2 NT membranes indicating high reproducibility of this method. The membranes were slightly yellow and are semi-transparent. Under SEM the NT membrane exhibited very good uniformity.

Both sides were intact with one end open and another closed. The TiO2 NTs were packed vertically side by side. The tubes were circular with small variations in diameters. There were some small variations in the length of NTs due to the existence of cracks on the commercial Ti foil. The average length of a NT from 5 h anodization under 60 V DC was ∼22 μm. The membranes had good mechanical strength and could be transferred from preserved aqueous solution to FTO glass by a speculum. The membrane could be kept in solution (methanol, ethanol or water) for several days without any obvious physical 4


M Dubey et al

Figure 4. Cross-sectional SEM images of TiO2 NT membrane on FTO glass. The films were cut in half by a diamond cutter. The cross sections in (A) and (B) were rough compared to (C). (A), (B) SEM images of a OE facing TiO2 NP layer. (C) SEM image of a CE surface facing TiO2 NP layer. (D) Optical image of a N719-coated TiO2 NT film with an OE facing TiO2 NP layer. The impregnation time was 12 h.

changes; however, the membrane broke down into small pieces after preservation in solution for more than a week. It was also noticed that the membrane curled up after drying completely in the air. The degree of curling depends on the thickness of the membrane. The thicker membranes became slightly arched whereas the thinner ones formed a complete column shape. The NT membranes for the following experiments were made by this method. 3.2. Fixation of freestanding TiO2 NT membrane on transparent FTO glass The TiO2 NT membranes themselves are difficult to bind to FTO glass directly. To facilitate this process, a layer of commercial TiO2 NP paste was screen-printed on FTO glass first and TiO2 NT membrane was then placed on its top before the NP paste was dried completely. The samples were kept in a freezer for 12 h with a ∼100 g piece of metal on the top to provide some pressure, followed by sintering at 450 ◦ C for 30 min in an oxygen ambient. The resulting samples were opaque and exhibited excellent physical strength and contact between TiO2 NTs and NPs, as shown in figure 4. The surface of the NT membrane on FTO glass was flat and no curling was observed. The film was less transparent than the TiO2 NT membrane alone. After adsorption of N719 in ethanol for 12 h, the film had a deep red color. The thickness of the TiO2 NP layer, determined from cross-sectional SEM images, exhibited small variations between different batches, but was usually between 2.5 and 4.5 μm. It was found that the TiO2 NTs and NPs have an excellent contact. In some cases, the NTs were embedded inside TiO2 NPs. It should be pointed out that there are two different orientations of NT on FTO glass. The closed-end or open-end surface could face down toward the NP layer (OED and CED, respectively). We found

Figure 5. XRD of a TiO2 NT membrane on FTO glass. (a) Bare FTO glass. (b) TiO2 NT membrane on FTO without a TiO2 NP layer. (c) TiO2 NT + TiO2 NP layer on FTO glass. The * indicates the peaks from FTO glass.

the physical contact in the CED configuration to be poor and the NT membrane came off the FTO glass very easily after taping the sample on the table or during the TiCl4 treatment. The resulting film was also studied by XRD analysis. Figure 5 shows the XRD patterns of film before and after sintering at 450 ◦ C for 30 min. The TiO2 NTs were converted to anatase phase completely after sintering. It should be noted that the thickness of the TiO2 NP layer is also critical for fixation of TiO2 NTs on it. When the thickness of the NP layer was less than 2 μm, the NT membrane was very easily curved up and broke into small pieces during sintering. It was found that 3 μm was an optimum thickness for NPs, providing a good platform for NT binding and less electron 5


M Dubey et al

Table 1. Photovoltaic parameters of devices under AM 1.5G condition. Devices

Dye

Orientation of NT

NP thickness (μm)

NT length (μm)

Jsc (mA cm−2 )

Voc (mV)

FF

η (%)

MT-246 MT-233 MT-241 MP-18a MP-5a MP-6a MP-12a MT-283 MT-309 MT-303 MT-300

N719 N719 N719 N719 N719 N719 N719 N719 Z907 N3 Black dye

OED OED OED — — — — CED OED OED OED

2.5 3.0 4.6 2.8 11 25 30 2.7 3.1 2.3 3.3

15 22 38 — — — — 23 21.5 24.0 21.7

12.3 16.3 13.2 6.23 13.6 4.94 3.58 9.68 10.0 10.0 7.68

680 700 670 689 694 643 641 630 621 647 616

0.58 0.53 0.53 0.61 0.57 0.65 0.67 0.61 0.59 0.53 0.59

4.91 6.12 4.71 2.63 5.42 2.14 1.53 3.75 3.76 3.44 2.80

a

Cells with TiO2 NP only.

Figure 6. The J –V curves of a TiO2 NT-based DSC.

Figure 7. EQE of cells with CED and OED configurations.

recombination for achieving high efficiency. Figure 4 shows cross-sectional SEM images of a TiO2 NT membrane on FTO glass for which the sample was prepared by cutting the films in half with a diamond cutter.

efficiencies, respectively. These results showed that for films of the same thickness (NT plus NP), the cell with NT membranes on a NP layer exhibited much better photovoltaic performance. The cells with a CED configuration gave poor photovoltaic performance under the same conditions. As shown in table 1, the efficiency decreased from 6.1% in the OED configuration to 3.7% in the CED configuration. We also tested three other dyes, i.e. N3, Z907 and black dye. The Z907 performed slightly better than N3 due to the presence of a long alkyl chain reducing the aggregation. Surprisingly, the black dye gave very poor efficiency. We found its loading on TiO2 was poor, which might come from its low solubility in ethanol. The EQE of cells with OED and CED configuration are also different. As shown in figure 7, maximum EQE values of cells with OED and CED configurations were found to be 57.11% and 43.61%, respectively, at a wavelength of 535 nm. This shows ∼31% enhancement of EQE in the OED structure compared to the CED structure. Significant improvement in EQE in the OED configuration can be attributed to the enhanced light scattering as the nanodome structure could lead to higher Jsc and hence better efficiency of the cell compared to the CED configuration. It should be noted that the short-circuit current densities evaluated from the EQE spectra were very close to those obtained from I –V curves (14.56 versus 14.75

3.3. Photovoltaic performance To test the photovoltaic performance of the resulting TiO2 NT films, the Grätzel type solar cells with different NT lengths were fabricated and tested under AM1.5G conditions. The J – V curves are shown in figure 6 with Voc , Jsc , FF, and energy conversion efficiency (η) listed in table 1. Cells with both CED and OED configurations were tested. The energy conversion efficiency of cells with the OED configuration increased from 4.9% to 6.1% when the length of the TiO2 NT increased from 15 to 22 μm, which is attributed to the high dye loading in the longer tube cell. The cell efficiency decreased to 4.7% when the 38 μm-long TiO2 NT was used. This mainly came from electron recombination due to increased surface defects, since the dye loading in this cell should be much higher. As a comparison, four TiO2 NP-based DSCs with film thicknesses between 2.8 μm and 30 μm were also fabricated under the same conditions and 5.4% conversion efficiency was obtained in a cell with an 11 μm thick TiO2 NP film. The cells with 24.5 μm and 2.8 μm thick TiO2 NP films gave 2.1% and 2.6% 6


M Dubey et al

Figure 9. Plot of charge transfer resistance ( Rct ) (top) and capacitance (bottom) versus applied bias ranging from 0 to 1 V.

was found to provide a good fit to the experimental data. Typical Nyquist plots at an applied bias −0.1 V for CED and OED configurations are also shown in figure 8. From these plots, the Rs , Rct and CPE1 were extracted for two cells at different biases. As Santiago et al [35] reported, the TiO2 behaves as an insulator under low applied bias and the recombination is mainly attributed to TCO/electrolyte interface [35]. Figure 9 shows the plot of charge transfer resistance ( Rct ) and capacitance versus applied bias ranging from 0 to 1 V. At low bias, the capacitances of OED and CED are almost the same, indicating that the conduction bands at two configurations remain almost the same. The charge transfer resistance ( Rct ) at low potential is greater for the sample with an OED configuration than with a CED configuration, which indicates that the cell with an OED configuration suffers less recombination at the TCO/electrolyte interface compared to one with a CED configuration. This can be explained by the fact that adhesion of NT with the CED configuration to TCO is weak compared with OED configurations. This could lead to weak electron transport from NT to TCO and therefore more chance of recombination losses at the interface. Therefore, increased current and voltage in the cells with an OED configuration compared to those with a CED configuration come partially from the low recombination between the TiO2 /electrolyte interface. As

Figure 8. Equivalent electrical circuit used for fitting based on a Randles type model and a Nyquist plot of theoretical fitting (——) with the experimental data ( ) at an applied bias of −0.1 V for cells with CED and OED configurations.

◦

for OED and 10.09 versus 9.5 mA cm−2 for CED, respectively) indicating the accuracy of the measurements. There are several possible reasons for the different photovoltaic performance of OED and CED configurations. Ideally the CED configuration should give high photovoltaic performance for several reasons, such as potentially easy electron recovery pathways, easy electrolyte filling, etc. However, its performance is poor compared to the OED configuration. To figure out the cause of these differences, electron impedance spectroscopy (EIS) was carried out to measure the resistance. A simplified Randles type model, as shown in figure 8, was used to fit the experimental data as suggested in the work of Liberator et al [34], where RS is the series resistance accounting for transport resistance of TCO, Rct is the combination of charge transfer resistance for electron recombination at the TCO/electrolyte and TiO2 /electrolyte interfaces, CPE1 is the constant phase element representing capacitance at the TCO/electrolyte/TiO2 interface. The model 7


M Dubey et al

Ti foil to transparent FTO glass is a challenge in this field. In this paper we develop reproducible methods to obtain freestanding TiO2 NT membranes by immersing the TiO2 /Ti foil in dilute HF aqueous solution and subsequent fixation on FTO glass with two different configurations. It was found that conversion of TiO2 from the amorphous to the anatase phase is critical for obtaining non-destructive membranes. We also found that low temperature treatment of TiO2 NT/NP film before high temperature sintering is important for the fixation of NTs on NP layer. The DSCs with an OED configuration perform much better than those with a CED configuration. A 6.1% energy conversion efficiency was obtained in a cell with commercial N719 as the dye and I− /I− 3 as the redox couple, in which a 22 μm thick NT membrane plus a 3.0 μm thick NP layer were used. We attributed this to the advantageous light reflection capability from the caps of the CEs and efficient electron migration in the OED configuration. It was found that high dye loading and high charge recombination resistance are responsible for the better photovoltaic performance of DSCs with an OED configuration.

Figure 10. Simplified DSC structures with CED and OED configurations. The brown color represents I− /I− 3 electrolyte. The red cube represents the dye and red gray ball represents the TiO2 NP. The arrow indicates the electron recovery pathways for the redox couple.

the dye-loading density is also critical for the generation of photocurrent, we measured the dye-loading density on the surface of TiO2 NTs. It was found that dye loading for the OED configuration was 7.16 × 10−6 mol g−1 of TiO2 , whereas it was 3.58 × 10−6 mol g−1 of TiO2 for the CED configuration. As increased dye loading leads to a high photon absorption efficiency, this gives a higher Jsc , as we observed. The high dye-loading density also produces a higher electron density at the conduction band, which increases the Fermi level of the conduction band resulting in a higher Voc . We believe this is the major contribution to higher conversion efficiency of cells with OED configuration. It should also be noted that the compact TiO2 layer at the caps of CEs of the TiO2 NT (with thickness ∼2–3 μm) may also serve as an electron migration barrier, a near-UV light absorber, and light reflector. In 2009, Grimes et al [15] reported a DSC with a CED configuration. The NTs were grown directly on FTO glass and were thought to be in a perfect configuration; however, the cell only gave 6.9% energy conversion efficiency, which we believe is largely due to the existence of a thick layer of TiO2 at the cap of the TiO2 NTs, as shown in figure 10. On the other hand, the OED configuration offers several benefits, including easy and strong binding to the NP layer and good contact with the NP layer. The caps at the CEs can act as a concentrator to reflect light back into the tubes for re-absorption. Currently we are investigating the light absorption and reflection properties in these two configurations. While this paper was in preparation, Chien et al [17] reported 5.3 to 9.1% efficiency from DSCs with a structure very similar to our report except that both ends were open. Even though the results were obtained from very long tubes (∼63 μm), which was time-consuming, their results indicated that efficiency was enhanced by removing the caps of the TiO2 NTs.

Acknowledgments This research was supported by the National Science Foundation/EPSCoR, grant nos 0554609 and 0903804. M Dubey would like to thank Braden Bills and Liping Si of the EECS Department, South Dakota State University, for their help with the EIS and dye-loading measurements. Y Zhong would like to thank SD EPSCoR for providing a summer research scholarship for his work. The authors are grateful to the anonymous referees for their valuable comments and suggestions for the improvements to this manuscript.

References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13]

4. Conclusions

[14]

TiO2 NTs have the potential for high efficiency dye-sensitized solar cells, and how to transform the NTs from non-transparent

[15]

8

Robertson N 2006 Angew. Chem. Int. Edn 45 2338 O’Regan B and Grätzel M 1991 Nature 353 737 Shao W, Gu F, Li C and Lu M 2010 Inorg. Chem. 49 5453–9 Kim Y J, Lee M H, Kim H J, Lim G, Choi Y S, Park N G, Kim K and Lee W I 2009 Adv. Mater. 31 3668–73 Yip C T, Mak C S K, Djuriˇsi’c A B, Hsu Y F and Chan W K 2008 Appl. Phys. A 92 589 Zhao Y, Sheng X, Zhai J, Jiang L, Yang C, Sun Z, Li Y and Zhu D 2007 ChemPhysChem 8 856 Grätzel M 2005 Inorg. Chem. 44 6841 Grätzel M 2009 Acc. Chem. Res. 42 1788 Crossland E J W et al 2008 Nano Lett. 9 2807 Villanueva-Cab J, Wang H, Oskam G and Peter L M 2010 J. Phys. Chem. Lett. 1 748 Wang M, Li X, Lin H, Pechy P, Zakeeruddin S M and Grätzel M 2009 Dalton Trans. 10015 Martinson A B F, McGarrah J E, Parpia M O K and Hupp J T 2006 Phys. Chem. Chem. Phys. 8 4655 Wiberg J, Guo L, Pettersson K, Nilsson D, Ljungdahl T, Martensson J and Albinsson B 2007 J. Am. Chem. Soc. 129 155 Boercker J E, Enache-Pommer E and Aydil E S 2008 Nanotechnology 19 095604 Varghese O K, Paulose M and Grimes C A 2009 Nat. Nanotechnol. 4 592


M Dubey et al

[25] Qiao H, Wang Y, Xiao L and Zhang L 2008 Electrochem. Commun. 10 1280 [26] Pavasupree S, Ngamsinlapasathian S, Nakajima M, Suzuki Y and Yoshikawa S 2006 J. Photochem. Photobiol. A 184 163 [27] Bwana N 2009 J. Nanopart. Res. 11 1917 [28] Jennings J R, Ghicov A, Peter L M, Schmuki P and Walker A B 2008 J. Am. Chem. Soc. 130 13364 [29] Crawford G A and Chawla N 2009 Scr. Mater. 60 874 [30] Albu S P, Ghicov A, Macak J M, Hahn R and Schmuki P 2007 Nano Lett. 7 1286 [31] Wang J and Lin Z Q 2008 Chem. Mater. 20 1257 [32] Park J H, Lee T W and Kang M G 2008 Chem. Commun. 2867 [33] Chen Q and Xu D 2009 J. Phys. Chem. C 113 6310 [34] Liberatore M, Burtone L, Brown T M, Reale A, Carlo A D, Decker F, Caramori S and Bignozzi C A 2009 Appl. Phys. Lett. 94 173113 [35] Santiago F F, Bisquert J, G-Belmonte G, Boschloo G and Hagfeldt A 2005 Sol. Energy Mater. Sol. Cells 87 117–31

[16] Kim D, Ghicov A, Albu S P and Schmuki P 2008 J. Am. Chem. Soc. 130 16454 [17] Lin C-J, Yu W-Y and Chien S-H 2010 J. Mater. Chem. 20 1073 [18] Sadek A Z, Zheng H, Latham K, Wlodarski W and Kalantar-zadeh K 2009 Langmuir 25 509 [19] Allam N K and Grimes C A 2008 Sol. Energy Mater. Sol. Cells 92 1468 [20] Grimes C A 2007 J. Mater. Chem. 17 1451 [21] Stergiopoulos T, Ghicov A, Likodimos V, Tsoukleris D S, Kunze J, Schmuki P and Falaras P 2008 Nanotechnology 19 235602 [22] Shankar K, Mor G K, Prakasam H E, Yoriya S, Paulose M, Varghese O K and Grimes C A 2007 Nanotechnology 18 065707 [23] Das J, Freitas F S, Evans I R, Nogueira A F and Khushalani D 2010 J. Mater. Chem. 20 4425 [24] Schierhorn M, Boettcher S W, Ivanovskaya A, Norvell E, Sherman J B, Stucky G D and Moskovits M 2008 J. Phys. Chem. C 112 8516

9