Effects of a surfactant-templated nanoporous TiO2 interlayer on dye ...

JOURNAL OF APPLIED PHYSICS 101, 084312 共2007兲

Effects of a surfactant-templated nanoporous TiO2 interlayer on dye-sensitized solar cells Kwang-Soon Ahna兲 Energy Laboratory, Corporate R&D Center, Samsung SDI Company, 428-5, Gongse-dong, Giheung-gu, Yongin-si, Gyeonggi-do, 449-577, South Korea

Moon-Sung Kang and Ji-Won Lee Energy and Environment Lab., Samsung Advanced Institute of Technology (SAIT), Gyeonggi-do, 446-712, South Korea

Yong Soo Kang Department of Chemical Engineering, Hanyang University, 17 Haengdang-dong, Seongdong-gu, Seoul 133-791, South Korea

共Received 3 October 2006; accepted 1 March 2007; published online 30 April 2007兲 A 320 nm thick surfactant-templated nanoporous 共STN兲 TiO2 layer prepared from block copolymer P123 关poly共ethyleneoxide兲20-poly共propyleneoxide兲70-poly共ethyleneoxide兲20兴 was used as an interlayer between a 21 ␮m thick mesoporous TiO2 layer and a transparent conducting oxide. The thin STN TiO2 interlayer had well-dispersed nanoporous features, which made it possible to adsorb dye molecules in the interlayer. In addition, it provided an enhanced electron lifetime, resulting in a reduced recombination rate and an increased diffusion length. The dye-sensitized solar cell with the thin STN interlayer resulted in a significantly increased overall energy conversion efficiency from 7.11% to 9.22% with an improvement of all parameters 共short-circuit current, open circuit voltage, and fill factor兲. © 2007 American Institute of Physics. 关DOI: 10.1063/1.2721976兴 I. INTRODUCTION

In recent years, dye-sensitized solar cells 共DSSCs兲 based on mesoporous TiO2 layers have been extensively studied as they show promise as low-cost alternative cells for commercial silicon-based solar cells.1–4 The DSSC consists of a dyesensitized mesoporous TiO2 layer, a Pt layer, and an electrolyte containing a redox couple 共I− / I−3 兲. The cell performance has been improved mainly by controlling the morphology and particle size of the mesoporous TiO2 layer,5 suppressing the recombination rate at the interfaces,6 developing new dyes,7 thermal-resistance transparent conducting oxide 共TCO兲,8 and new electrolytes.9 However, despite these considerable research efforts and advances, studies on the interfacial contact properties between the TiO2 layer and TCO are few. Most of the mesoporous TiO2 layers have been prepared by TiO2 colloids, which are needed to disperse the TiO2 nanoparticles in the solvent by means of physical-mechanical dispersion methods.5 However, in this case, incomplete dispersion and particles aggregation are difficult to overcome and, as a result, may result in bad interfacial contact properties with the TCO and a low energy conversion efficiency. One promising solution to this problem is the use of a surfactant-templated nanoporous TiO2 layer 共referred to as STN layer兲, which is made by a TiO2 sol-gel precursor and a structure-directing block copolymer. Coakley et al.10 report that nanoporous TiO2 films with a uniform pore size 共or without any aggregation of the particles兲 can be made using a兲

Corresponding author. Current address: National Center for Photovoltaics, National Renewable Energy Laboratory, Golden, CO 80401. Fax: ⫹1-303384-6491. Electronic mail: [email protected]

0021-8979/2007/101共8兲/084312/5/$23.00

a block copolymer in a solution of a titania sol-gel precursor. In addition, the inorganic/organic sol will be freely and easily coated on the whole surface of the TCO, which may lead to superior interfacial contact properties with TCO. However, it is difficult to achieve a film thickness above 1 ␮m in a single process,5,11 since the films are mainly made by dipcoating or spin-coating substrates with the inorganic/organic sol. Highly efficient DSSCs have been obtained with a TiO2 film thickness of about 15– 20 ␮m, due to limitations in the electron diffusion length of 7 – 25 ␮m.12 Therefore, it is not practically useful to use only a STN layer for highly efficient DSSCs, because of the requirement that the STN layer be repeatedly coated. In this article, a 320 nm thick STN layer was used as an interlayer between the 21 ␮m thick mesoporous TiO2 layer and the TCO, and its cell performance was compared with that of a conventional DSSC without the STN interlayer. In order to investigate the effects of the thin STN interlayer on the DSSCs, laser-induced photocurrent/photovoltage transient measurements were used and compared with the results of the conventional DSSCs without the STN interlayer. II. EXPERIMENT

Fluorine-doped SnO2 TCO films 共⬃9 ⍀ / 䊐兲 were used as the substrate to prepare both the photoanode and counter electrode. The STN layer was prepared according to the following procedure. First, 7.5 g P123 关poly共ethyleneoxide兲20-poly共propyleneoxide兲70-poly 共ethyleneoxide兲20兴共BASF Co.兲 was dissolved in 40 g of acetylacetone 共ACA兲 by stirring for 2 h. Then, 10.65 g of titanium共IV兲 propoxide 共TTP兲 mixed with 10 g of ACA was added with vigorous stirring. Finally, the sol solution was

101, 084312-1

© 2007 American Institute of Physics

Downloaded 30 May 2007 to 166.104.25.192. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

084312-2

Ahn et al.

stirred at room temperature 共RT兲 until a uniform brightyellow TiO2 sol was obtained. The STN layer was dipcoated, using a mask to affect only the active area, which was subsequently aged at RT for 4 h. The mask was then detached and the sample was calcined at 450 °C for 30 min to remove the block copolymer and form the 320 nm thick STN layer. The preparation of the mesoporous TiO2 layers was initiated from TiO2 共ST-21, Ishihara Sanyo, Ltd.兲 particles dispersed by paint-shaking for 2 h in ethanol,13 which showed an incomplete dispersion and an average aggregate size of about 443 nm 关measured by dynamic laser-light scattering 共Malvern Instrument., Zetamaster兲兴. The TiO2 colloid was thoroughly dispersed using a conditioning mixer 共Thinky Co.兲 by adding ethyl cellulose as a binder and ␣-terpineol as a solvent for the TiO2 paste, followed by concentration using an evaporator. A 21 ␮m thick mesoporous TiO2 layer was prepared from the TiO2 paste coated by an RK print coater on the STN layer followed by calcination at 450 °C for 30 min. For comparison, a mesoporous TiO2 layer was deposited on the TCO substrate without the STN interlayer. A platinum layered counter electrode was prepared by a conventional sputtering method. The photoanodes were immersed in an ethanol solution of N719 overnight for dye adsorption, followed by rinsing with ethanol and drying at 50 °C. The edges of the cell were sealed and separated with 60 ␮m thick Surlyn. The electrolyte was then introduced into the cell, which was composed of 0.6 M 3-hexyl-1,2-dimethyl imidazolium iodide, 0.05 M iodine, 0.05 M LiI, and 0.5 M 4-tert butylpyridine in acetonitrile. Electron transport performances of the photoanodes 共i.e., electron diffusion coefficient and lifetime兲 were evaluated via laser-induced photocurrent/photovoltage transient measurements using a diode laser source 共␭ = 635 nm兲 modulated with a function generator.13–15 The transients were induced by the stepwise change of the laser intensity, which was controlled by a function generator. A set of ND filters was used to change the laser intensity. The laser intensity without ND filter was constant 90 mW cm−2 and was decreased up to about 10 mW cm−2 with the ND filter. The DSSC was irradiated by a laser and a small fraction of the laser intensity, less than 10% of the initial intensity, was stepped down. The active area of the photoanode was carefully measured by the image analyzer. The photovoltaic characteristics of the prepared DSSCs were measured under 1 sunlight intensity 共100 mW cm−2, AM1.5兲 that was verified with an NREL-calibrated Si-solar cell 共PV Measurements Inc.兲. A black mask was applied to the area surrounding the TiO2 active area to avoid penetration of the photons to the dye via reflection and refraction of the glass texture. The current-voltage characteristics were also measured under dark conditions to evaluate the recombination rate.

J. Appl. Phys. 101, 084312 共2007兲

FIG. 1. Cross-sectional SEM images of the photoanodes 共a兲 without and 共b兲 with the STN interlayer, respectively. 共Inset: surface morphology of the STN layer only.兲

TiO2 layer and the TCO due to the incompletely dispersed TiO2 particles and the large aggregate size 共443 nm兲 in the TiO2 paste. Therefore, the large aggregation of TiO2 particles may lead to the bad interfacial contact properties with the TCO, resulting in a bad DSSC performance. Figure 1共b兲 shows that a superior adherence could form between the STN interlayer and the TCO, because the STN interlayer was deposited by dip-coating the TCO in the TiO2 precursor/ block copolymer sol. The STN interlayer was also nanoporous and had a smooth, crack-free surface with a uniform pore size of about 10 nm 共or without any aggregates兲, as shown in the inset of Fig. 1共b兲. Based on the x-ray diffraction pattern 共not shown here兲, it was shown that the STN layer had an anatase structure and a crystallite size of 9.7 nm, calculated according to the Scherrer equation. This welldispersed, crack-free nanoporous layer is the favorable morphology for the DSSCs, because it provides a high surface area and a high pore volume, which lends itself to dye adsorption and the movement of electrolyte ions in the film.11,16,17 To evaluate the recombination rate, dark current-voltage measurements were conducted under dark conditions, because the dark current is a measure of the recombination rate.18 Figure 2 shows the dark current-voltage curves of the DSSCs without and with the STN interlayer. The DSSCs with the STN interlayer had a lower recombination rate than those without the STN interlayer. This indicated that the STN interlayer may provide longer electron life, which will be discussed later.

III. RESULTS AND DISCUSSION

Figure 1 shows the cross-sectional scanning electron microscopy 共SEM兲 images of the photoanodes 共a兲 without and 共b兲 with the STN interlayer, respectively. Figure 1共a兲 clearly indicates the imperfect adherence between the mesoporous

FIG. 2. Dark current-voltage curves of the DSSCs without and with the STN interlayer.


084312-3

Ahn et al.

J. Appl. Phys. 101, 084312 共2007兲

FIG. 3. Typical curves of 共a兲,共c兲 photocurrent transients and 共b兲,共d兲 photovoltage transients for the DSSCs without and with the STN TiO2 interlayer, respectively. The initial laser intensity gradually increased using the ND filters from bottom to top.

The amount of dye adsorbed onto the TiO2 surface was measured using the UV-vis absorption measurement, in which the dye was desorbed into a NaOH solution and this resultant solution was used for the measurement. The photoelectrode with the STN interlayer showed a larger amount of dye 共about 4%–5% on average兲. This indicated that the STN interlayer was not a compact layer but a nanoporous layer, which made it possible to adsorb dye molecules in the interlayer. Figures 3共a兲 and 3共c兲 show typical curves of photocurrent transients, and Figs. 3共b兲 and 3共d兲 show photovoltage transients for the DSSCs without and with the STN TiO2 interlayer, respectively. Laser light 共635 nm兲 was illuminated from the TCO toward the TiO2 active area. The thickness of the STN layer was very thin 共320 nm兲 and the transmittance of the TCO/STN layer without the dye adsorption, as a result, changed only slightly compared with the TCO. This indicated that the STN layer without the dye adsorption was transparent 共not shown here兲. Moreover, the absorption coefficient of the N719 dye at the wavelength of 635 nm was very small.14 Therefore, the electron generation in the mesoporous TiO2 layer may be scarcely affected by the thin STN interlayer, making it possible to compare the photocurrent/ photovoltage transient curves of the DSSCs without and with the STN interlayer. Figures 3共a兲 and 3共c兲 show that the time to reach a constant photocurrent became longer as the initial laser intensity decreased. As a result, the electron diffusion

coefficient 关De = w2共2.77␶c兲兴, where w and ␶c are film thickness and time constant, respectively, in the photoanode, decreased with respect to a decrease in the initial laser intensity, in which all the current transients were well fitted with a single-exponential function, exp共−t / ␶c兲.13–15 The electron lifetime 共␶e兲 at the open circuit was measured from the photovoltage response of the DSSCs against the perturbation of laser light intensity. The photovoltage transient curve was expressed by an exponential decay of the ␶e, exp共−t / ␶e兲.13–15 The ␶e value decreased as the initial laser intensity increased, as shown in Figs. 3共b兲 and 3共d兲, because a high electron concentration at an increased laser intensity can result in a higher probability of the electrons encountering acceptors in the electrolyte. Figure 4共a兲 shows the De values versus short-circuit current and Fig. 4共b兲 shows ␶e values versus open-circuit voltage in the photoanodes without and with the STN interlayer, which were evaluated from the data of Fig. 3. The electron lifetimes 共␶e兲 in the photoanode were calculated from the photovoltage transients with a single-exponential decay, exp共−t / ␶e兲. The x-axis shows 共a兲 short-circuit currents 共Jsc兲 and 共b兲 open-circuit voltage 共Voc兲 under various laser intensities, controlled by a set of ND filters. The Jsc and Voc values in the x-axis increased with an increase of the initial laser intensity. Figure 4共a兲 shows that the STN interlayer had little influence on the De value, indicating that the De value was


084312-4

J. Appl. Phys. 101, 084312 共2007兲

Ahn et al.

FIG. 5. Photocurrent-voltage 共I-V兲 curves of the DSSCs without and with the 320 nm thick STN TiO2 interlayer between the 21 ␮m thick mesoporous TiO2 layer and the TCO. Active area of the dye-adsorbed TiO2 layer was 0.18± 0.005 cm2 for the samples under experimentation here, as carefully measured by the image analyzer.

FIG. 4. 共a兲 Electron diffusion coefficients and 共b兲 lifetimes in the photoanodes without and with the STN interlayer.

dominated by the electron diffusion in the 21 ␮m thick mesoporous TiO2 layer. On the other hand, the ␶e value was improved with the STN interlayer. The ␶e value is the parameter related to the electron recombination. As shown in Figs. 1 and 2, the STN interlayer had a well-dispersed, nanoporous structure with a high surface area, which may enhance the recombination rate. However, the ␶e value increased with the STN interlayer, as shown in Fig. 2共b兲. As mentioned above, the STN interlayer was prepared from an inorganic/organic sol followed by dip-coating the TCO substrate, leading to superior interfacial contact properties with the TCO. This indicated that a reduced recombination rate 共or an increased electron lifetime兲 with the STN interlayer may be provided from the superior interfacial contact properties with the TCO, which corresponds well with the results reported by Frank et al.19 Therefore, the STN interlayer can increase the electron lifetime as well as enhance the movement of electrolyte ions in the interlayer, which are very favorable features for the DSSCs. Figure 5 shows photocurrent-voltage 共I-V兲 curves of the DSSCs without and with the 320 nm thick STN TiO2 interlayer between the 21 ␮m thick mesoporous TiO2 layer and the TCO, which are summarized in Table I. All parameters 关short-circuit current 共Jsc兲, fill factor 共f f兲, and open-circuit voltage 共Voc兲兴 increased with respect to the STN interlayer, resulting in a significant increase from 7.11% to 9.22% for the overall energy conversion efficiency 共␩兲. This indicated

that the improved cell performance with the STN interlayer resulted from the superior interfacial contact properties with the TCO as well as the nanoporous structure of the STN interlayer itself. It is worth noting that the Jsc value was increased by the use of the STN interlayer. Diffusion length 共L兲 is expressed by L = 冑D ⫻ ␶e.13–15 This indicates that the L value is related to both of the De and ␶e values. In this experiment, the ␶e values were measured at open-circuit voltage. The electron lifetime is a function of voltage and may be different for the short-circuit conditions. As shown in Fig. 4, the De value was almost constant but the electron lifetime was enhanced with the STN interlayer. It leads to an increased L value, although the ␶e values were measured at open-circuit voltage. Moreover, the dye molecules can also be adsorbed in the STN interlayer. Therefore, the shortcircuit current may be increased by the use of a STN interlayer due to an increased L value and the adsorbed dye molecules. IV. CONCLUSIONS

In this article, a 320 nm thick STN layer was introduced as the interlayer between the 21 ␮m thick mesoporous TiO2 layer and the TCO. The STN interlayer made it possible to adsorb the dye molecules and move the electrolyte ions in the interlayer, due to the nanoporous structure with a uniform pore size of about 10 nm. In addition, it provided superior interfacial contact properties with the TCO, leading to a suppressed recombination rate and enhanced electron lifetime. As a result, the STN interlayer could increase all parameters, resulting in a significant increase of the energy conversion TABLE I. Photovoltaic characteristics of the DSSCs without and with the STN TiO2 interlayer between the TCO and mesoporous TiO2 layer Photoanodea Without interlayer With interlayer

Jsc / mA cm−2 15.05 18.11

Voc / V 0.713 0.730

ff 0.662 0.698

␩/% 7.11 9.22

a

Active area: 0.18± 0.005 cm2 for the samples used in this experiment, as calculated by the image analyzer.


084312-5

efficiency from 7.11% to 9.22%. We expect that the optimization of the STN interlayer can also provide more efficient DSSCs; related research is currently underway. ACKNOWLEDGMENT

The authors gratefully acknowledge the financial support of the Ministry of Commerce, Industry and Energy in Korea through the Sol-Gel Innovation Project. B. O’Regan and M. Grätzel, Nature 353, 737 共1991兲. M. K. Nazeruddin, A. Kay, I. Podicio, R. Humphy-Baker, E. Müller, P. Liska, N. Vlachopoulos, and M. Grätzel, J. Am. Chem. Soc. 115, 6382 共1993兲. 3 M.-S. Kang, Y. J. Kim, J. Won, and Y. S. Kang, Chem. Commun. 21, 2686 共2005兲. 4 M.-S. Kang, J. H. Kim, Y. J. Kim, J. Won, N. G. Park, and Y. S. Kang, Chem. Commun. 7, 889 共2005兲. 5 S. Ito, T. Kitamura, Y. Wada, and S. Yanagida, Sol. Energy Mater. Sol. Cells 76, 3 共2003兲. 6 S. S. Kim, J. H. Yum, and Y.-E. Sung, J. Photochem. Photobiol. Chem. 171, 269 共2005兲. 1 2

J. Appl. Phys. 101, 084312 共2007兲

Ahn et al.

S. Ferrere and B. A. Gregg, J. Am. Chem. Soc. 120, 843 共1998兲. T. Kawashima, T. Ezure, K. Okada, H. Matsui, K. Goto, and N. Tanabe, J. Photochem. Photobiol., A 164, 199 共2004兲. 9 H. Kusama and H. Arakawa, J. Photochem. Photobiol., A 164, 103 共2004兲. 10 K. M. Coakley, Y. Liu, M. D. McGehee, K. L. Frindell, and G. D. Stucky, Adv. Funct. Mater. 13, 301 共2003兲. 11 M. Zukalová, A. Zukal, L. Kavan, M. K. Nazeeruddin, P. Liska, and M. Grätzel, Nano Lett. 5, 1789 共2005兲. 12 M. Law, L. E. Greene, J. C. Johnson, R. Saykally, and P. Yang, Nat. Mater. 4, 455 共2005兲. 13 K. S. Ahn, M. S. Kang, J. K. Lee, B. C. Shin, and J. W. Lee, Appl. Phys. Lett. 89, 013103 共2006兲. 14 S. Nakade, T. Kanzaki, Y. Wada, and S. Yanagida, Langmuir 21, 10803 共2005兲. 15 J.–H. Yum, S. Nakade, D. Y. Kim, and S. Yanagida, J. Phys. Chem. B 110, 3215 共2006兲. 16 G. Kron, U. Rau, M. Dürr, T. Miteva, G. Nelles, A. Yasuda, and J. H. Werner, Electrochem. Solid-State Lett. 6, E11 共2003兲. 17 M. Dürr, A. Schmid, M. Obermaier, A. Yasuda, and G. Nelles, J. Phys. Chem. A 109, 3967 共2005兲. 18 B. A. Gregg, F. Pichot, S. Ferrere, and C. L. Fields, J. Phys. Chem. B 105, 1422 共2001兲. 19 K. Zhu, E. A. Schiff, N.-G. Park, J. van de Lagemaat, and A. J. Frank, Appl. Phys. Lett. 80, 685 共2002兲. 7 8
