Role of Electrolytes on Charge Recombination in Dye-Sensitized TiO2

Jan 27, 2005 - For the electron diffusion in nanoporous TiO2 electrodes, the diffusion coefficients are influenced by cations through ambipolar diffus...
1 downloads 0 Views 165KB Size
Download PDF
3480

J. Phys. Chem. B 2005, 109, 3480-3487

Role of Electrolytes on Charge Recombination in Dye-Sensitized TiO2 Solar Cell (1): The Case of Solar Cells Using the I-/I3- Redox Couple Shogo Nakade,† Taisuke Kanzaki,† Wataru Kubo,† Takayuki Kitamura,† Yuji Wada,† and Shozo Yanagida*,‡ Material and Life Science, Graduate School of Engineering, Osaka UniVersity, Suita, Osaka 565-0871 Japan, and Center for AdVanced Science and InnoVation, Osaka UniVersity, Suita, Osaka 565-0871 Japan ReceiVed: September 5, 2004; In Final Form: NoVember 19, 2004

Performance of dye-sensitized solar cells (DSCs) was investigated depending on the compositions of the electrolyte, i.e., the electrolyte with a different cation such as Li+, tetra-n-butylammonium (TBA+), or 1,2dimethyl-3-propylimidazolium (DMPIm+) in various concentrations, with and without 4-tert-butylpyridine (tBP), and with various concentrations of the I-/I3- redox couple. Current-voltage characteristics, electron lifetime, and electron diffusion coefficient were measured to clarify the effects of the constituents in the electrolyte on the charge recombination kinetics in the DSCs. Shorter lifetimes were found for the DSCs employing adsorptive cations of Li+ and DMPIm+ than for a less-adsorptive cation of TBA+. On the other hand, the lifetimes were not influenced by the concentrations of the cations in the solutions. Under light irradiation, open-circuit voltages of DSCs decreased in the order of TBA+> DMPIm+ > Li+, and also decreased with the increase of [Li+]. The decreases of open-circuit voltage (Voc) were attributed to the positive shift of the TiO2 conduction band potential (CBP) by the surface adsorption of DMPIm+ and Li+. These results suggest that the difference of the free energies between that of the electrons in the TiO2 and of I3- has little influence on the electron lifetimes in the DSCs. The shorter lifetime with the adsorptive cations was interpreted with the thickness of the electrical double layer formed by the cations, and the concentration of I3- in the layer, i.e., TBA+ formed thicker double layer resulting in lower concentration of I3- on the surface of the TiO2. The addition of 4-tert-butylpyridine (tBP) in the presence of Li+ or TBA+ showed no significant influence on the lifetime. The increase of Voc by the addition of tBP into the electrolyte containing Li+ and the I-/I3redox couple was mainly attributed to the shift of the CBP back to the negative potential by reducing the amount of adsorbed Li cations.

Introduction Intensive attention has been paid to nanoporous metal oxide electrodes in view of various practical applications, since they can provide their large internal surface area which is useful for the sensitivity of sensors,1 optical density of electrochromic window and bi-stable displays,2 photocatalytic efficiency,3 and the light absorption coefficients for solar cells.4 Since all of these applications involve charge transfer in order to realize their functionalities, understanding and controlling electron transport, especially at interfacial surfaces, are important. Dye-sensitized solar cell (DSC) is one of the attractive applications using nanoporous electrodes.4 During the past decade, intensive studies have been carried out because of their high energy-conversion efficiency and the potential of their low production cost.4-5 The solar cell consists of a dye-adsorbed wide band gap semiconductor, typically TiO2, and electrodes immersed in an electrolyte containing a redox couple. Under light irradiation, excited electrons in the dye are injected into the conduction band of the electrode, and the electrons travel to a transparent conductive electrode. The resulting dye-cations are reduced by the redox couple. In this type of solar cell, charge transfer at the electrode/dye/electrolyte interfaces should be * Corresponding author. E-mail: [email protected] † Graduate School of Engineering. ‡ Center for Advanced Science and Innovation.

controlled so that the photoinjected electrons do not recombine with acceptors but are taken out to an external load. Figure 1 shows a schematic view of the charge transfer in the dye-sensitized solar cell. Injection of electrons occurs on the picosecond time scale.5,6 The injection yield depends on the TiO2 surface condition7 and the species and concentration of the cations contained in the electrolytes.8 The injected electrons can transfer to dye cations or I3- in the electrolyte. The recombination with dye cations (R2 on Figure 1) takes tens of nanoseconds to a millisecond, depending on the density of electrons in TiO2,9 while the reduction of the dye cation by the I- ions occurs (Process 5 on Figure 1) in a few to tens of microseconds, depending on the concentration and species of cations.10 In an electrolyte containing a high concentration of the redox couple and under short-circuit conditions, the recombination of electrons with I3- ions (R1) occurs predominantly. The electron lifetime has been measured by various groups, showing that the lifetime scales between a few milliseconds to more than one second with I -a, where I is the light intensity and a is a constant, typically around 0.5.11-13 Species of cations and additives are typically dissolved in the electrolyte of DSC. The roles of the cations have been found to influence the electron injection yield,8 the open-circuit voltage,9,14,15 the electron diffusion coefficient,16,17 and the rate of dye-cation reduction.10 It has been also reported that the cations do not influence the recombination kinetics of the

10.1021/jp0460036 CCC: $30.25 © 2005 American Chemical Society Published on Web 01/27/2005

Role of Electrolytes on Charge Recombination I

Figure 1. Schematic view of the charge transfer in dye-sensitized solar cell.

electrons in TiO2 with dye cations.9 For the additives, tBP is typically added into the electrolytes and the role was mainly attributed to the negative shift of the TiO2 CBP.12 On the other hand, the importance of tBP on the electron lifetime has not been clarified yet,18,19 and the influence of the cations and additives on the electron transfer from TiO2 to I3- has not been studied systematically. It is important to understand and elucidate the influence of the constituents in the electrolyte on the electron lifetime to design the hole conducting layer and the interface at the TiO2 electrode for highly efficient solar cells. The aim of this paper is to clarify the roles of the constituents in the electrolyte and understand the mechanism of the charge recombination in DSCs. Cations and additives can influence the charge traps, the electron diffusion, and the CBP of TiO2 simultaneously, and the electron lifetime could be related to these factors. Therefore, the influences of the cations and additives on the lifetime are discussed with taking account of these parameters. In this paper, we address the roles of electrolytes in the presence of anionic redox couples. Current Proposed Models for the Electron Lifetime in DSCs Electron lifetime in DSCs has two prominent features: powerlaw dependence of the electron lifetime on the electron density, and long electron lifetime up to a few seconds.11 The lifetime is determined by the rate of reaction between the electrons in the TiO2 and dye cation and/or I3-. To explain the observed features, several models have been proposed as summarized below. Model 1. One model is based on the second-order reaction with the electrons in the TiO2 and I3- ions.11,20 When two electrons and at least one I3- ion are needed to form one Iion, the reaction rate scales with n2, where n is the density of electrons. This model assumes that the all electrons in the TiO2 are available for the reaction. If intraband charge traps exist, the traps should be on the surface of the TiO2 and the energy level of the traps has no influence on the electron-transfer rate from the traps to the acceptors. While the model can explain the power-law dependence of the electron lifetime, this model cannot explain the results in a recent report, which showed the power law dependence in DSCs using a polymer hole conducting material.21 Model 2. Another model describes that the electrons recombine through TiO2 surface trap states, and the transfer rate depends on the energy distribution of the traps.12,22 For this case, the traps act as a recombination center, so that a shorter lifetime can be expected from the larger number of the traps. With the increase of electron density, the traps are filled from deep traps and the difference of free energy between the quasi-Fermi level

J. Phys. Chem. B, Vol. 109, No. 8, 2005 3481 of the TiO2 and the redox potential changes. Thus, the lifetime dependence on the electron density is expected. Model 3. The lifetime dependence on light intensity can be explained without taking into account the surface charge traps as a recombination center. The electron diffusion coefficient in thenanoporouselectrodeshasalsoelectrondensitydependence.23-25 Most scientists seem to agree to that the dependence is attributed to the intraband charge traps. The faster diffusion coefficients at higher electron concentration can be explained by the higher ratio of electron density on the conduction band to that of trapped electrons.26 If the electrons do not recombine through the surface state, higher electron concentration on the conduction band will result in the higher probability for the electrons to encounter the acceptors. Relationship between D and τ. In Models 2 and 3, the electron lifetime was related to the intraband charge traps. For these cases, apparent relationship between D and τ can be expected. With the increase of charge traps, Model 2 expects the decreases on both D and τ, and Model 3 expects the decrease of D with the increase of τ. Some experimental results of D and τ have supported Model 3.27,28 It has been reported that surface adsorption of cations and carboxyl group can influence D,29 and cations can influence the charge trap density.30 Based on the consideration described above, the adsorptions would also influence the electron lifetime. Electron-Transfer Rate between e-TiO2 and Acceptors. The transfer rate can be influenced by the distance between the acceptors and TiO2, and the free energy and reorganization energy difference between them. The long recombination lifetime was considered to be at the Marcus “inverted region”.31 Clifford et al. studied the recombination kinetics with various dyes and found that the rate depended on the distance but little on the free energy difference between e-TiO2 and dye cation.32 With the addition of tBP, it has been assumed that the adsorption of tBP increases the distance between the TiO2 and I3-, resulting in the longer lifetime. Experimental Section Nanoporous TiO2 electrodes were prepared on a transparent conductive substrates (Nippon Sheet Glass, SnO2:F, 8 ohm/sq) from a colloidal suspension of TiO2 particles (P25, Nippon Aerosil) by doctor-blade techniques. The electrodes were sintered at 450 °C for 30 min in air. The thickness of the electrodes was measured by a profile meter. Among the electrodes, the electrodes having c.a. 6 µm were selected for DSC. Typical areas of the electrodes were around 0.1 cm2. For sensitization, the electrodes were heated again to around 110 °C, and at the temperature, the electrodes were immersed into an acetonitrile/2-methyl-2-propanol (1:1) solution containing 0.5 mM Ru-dye (Bu4N)2[Ru(Hdcbpy)2(NCS)2] (known as N719, Solaronix) for at least 12 h at room temperature. Then, the electrodes were rinsed with acetonitrile. Solar cells were prepared with the sensitized electrode placed on a platinum sputtered transparent conductive oxide glass as a counter electrode. The edges were sealed by thermal adhesive film. Electrolyte solutions containing redox couple were introduced from a drilled hole on the Pt sputtered electrode. The solvent was acetonitrile unless otherwise stated. Current-voltage curves were measured under simulated 1 sun conditions (100 mW/ cm2, Yamashita Denso, YSS-80). Electron diffusion coefficients were measured by pulsed laser induced current transient measurements. DSCs were irradiated by a laser pulse (Nd:YAG, 7 ns, 532 nm) under the continuous irradiation of a HeNe laser (λ ) 632.8 nm). Current transients

3482 J. Phys. Chem. B, Vol. 109, No. 8, 2005

Figure 2. Electron lifetime vs short-circuit current in DSCs using acetonitrile solutions containing three different cations: Li+ (circle), TBA+ (rectangle), and DMPIm+ (triangle). The concentrations of the cations were 0.7 M. The concentrations of I- and I3- are 0.65 M and 0.05 M. The thickness of electrodes was c.a. 6 µm.

were monitored through a current amplifier (Stanford Research Systems, SR570) on a digital oscilloscope. The current decay was fitted with exp(-t/τD), and the D values were obtained by D ) w2/(2.35 τD). Electron lifetimes were measured at open-circuit conditions by intensity modulated photovoltage spectroscopy (IMVS) or by photovoltage decay measurements. For the IMVS, the cells were irradiated by a diode laser (Coherent, Lablaser, λ ) 635 nm). A small portion of the intensity was modulated by a sinusoidal function. The spot size of the laser was expanded by a lens so that whole DSCs were irradiated by the laser. The photovoltage response of the cells was measured by a lock-in amplifier (Stanford Research Systems, SR810) and plotted on a complex plane. All measurements showed a semicircle on the fourth quadrant. Thus, the lifetime was obtained with the frequency (f) giving a minimum value of the imaginary part, using a relation of τ )1/(2 π f).12 For the photovoltage decay measurements, the intensity of the diode laser was reduced by a step function on a function generator, and the responding photovoltage decay was measured. The intensity was reduced by about 4% of the initial intensity. The amplitudes of the photovoltage decays were less than a few millivolts. The decay signal was amplified by a differential amplifier (NF Electronic Instruments, 5307) and recorded by the oscilloscope. The electron lifetime (τ) was obtained by a fit with exp(-t/τ). Note that the IMVS measurements and the photovoltage decay measurements are related through Fourier transform. All measurements were repeated on at least two samples. Results Electron Lifetime in DSC with Various Cations. First, the influence of the type of cations on the electron lifetime was investigated. The cations were chosen so that each had different properties: Li as an adsorptive hard cation, DMPIm+ as an adsorptive soft cation, and TBA+ as a less-adsorptive soft cation.15,17 Figure 2 shows the electron lifetimes in the DSCs with the three different cations. The horizontal axis shows the short-circuit current (Jsc) under the same laser intensity used to measure the lifetimes at open circuit. The DSCs using Li+ and DMPIm+ showed comparable lifetimes, while TBA+ showed longer lifetime than others. Table 1 shows I-V characteristics of these cells under 1 sun conditions. The open circuit voltage

Nakade et al.

Figure 3. Electron lifetime vs short-circuit current of DSCs containing different concentration of Li+ cations. The concentration of I2 was fixed at 0.05 M.

TABLE 1: I-V Characteristics of the DSC under 1 sun Conditions Used in Figure 2 Li+

DMPIm+ TBA+

Jsc/mA /cm2

Voc/V

FF /

EFF/%

13.4 8.5 6.8

0.53 0.75 0.83

0.58 0.68 0.71

4.2 4.3 4.0

(Voc) was increased in the order of Li+ < DMHIm+ < TBA+. For Jsc, the cell using Li+ showed the highest values.33 The characteristics of I-V curves were similar with those found in the previous report.14 Influence of the concentrations of the cations on the electron lifetime was also investigated as shown in Figure 3. The Li+ concentration was varied by the addition of LiClO4 in these measurements. No significant difference in the electron lifetime was observed by changing the concentrations of Li+. IV curve measurements showed that the Voc of the DSCs was decreased by c.a. 100 mV with the increase of [Li+] from 0.1 to 1 M. For the DSCs using TBA+, no changes in the lifetime and Voc were observed in the range of 0.1 to 1 M of [TBA+]. The results of I-V curves observed here were consistent with the shift of the CBP to the positive side by adsorptive cations, resulting in the decrease of the Voc. It has been reported that TBA+ hardly penetrates the space between the adsorbed dye molecules and the surface of TiO2. To examine the influence of the location of the cations, we prepared TiO2 films having about one tenth of adsorbed dyes in comparison to the films used in Figure 2 and measured the electron lifetime in DSCs prepared from the films with Li+ or TBA+. For the DSCs, comparable lifetimes were found regardless of the cations (Figure 4). For highly efficient DSC, tBP (tert-butylpyridine) is generally added into the electrolyte solution. The influence of tBP was explained mainly as due to the shift of the CBP by its adsorption.12 This phenomenon may be associated with the kind of coexisting cations. Figure 5a and b show I-V characteristics of DSCs with and without the addition of tBP. The electron lifetimes in the DSCs are shown in Figure 5c. The large increase of the Voc was seen without significant change of the electron lifetime. To attribute the increase of the Voc to the change of the lifetime, the electron lifetime must be 2 orders of magnitude larger,34 but the results were much smaller than that. In comparison between TBA+ and Li+, the influence of tBP in the copresence of TBA+ was similar to that found with Li+,

Role of Electrolytes on Charge Recombination I

J. Phys. Chem. B, Vol. 109, No. 8, 2005 3483 electron-hole generation rate. Under one sun conditions, the G in DSC is on the order of 1017/cm2s, when the charge injection efficiency is close to unity. Then, N is estimated to be 1014/ cm3 for 10 µm thick electrodes. Electron density in the TiO2 has been evaluated to be 1017/cm3 by several methods.38 Based on these numbers, the electron lifetime is estimated to be on the order of 10-3/s, which is in agreement with the order under 10 mM of I- (Figure 7a). Thus, the results show that when the concentration of I- is less than 20 mM, the majority of acceptors are dye cations. For the conditions of Figure 7b, there were enough I- ions to reduce the dye cations so that the decrease of the lifetime should be directly related with the increase of I3-. The change of Voc with the increase of [I3-] agreed with the values calculated using the Nernst equation. Discussion

Figure 4. Electron lifetime in DSCs prepared from theTiO2 films having low amount of adsorbed dye. Comparable lifetimes were found between the DSCs using Li+ (open symbols) and TBA+ (closed symbols). Measurements were repeated for four samples (different markers) for each electrolyte condition. Concentrations were 0.7 M of LiI or TBAI and 0.05 M of I2. The film thickness was about 4 µm.

i.e., an increase of Voc without influencing the lifetime. However, the increase of Voc was less prominent than with Li+. Electron Diffusion Coefficients in the DSCs with Various Electrolytes. Correlation between electron diffusion coefficients (D) and lifetimes is expected if the parameters are influenced by the condition of charge traps. Because the cations could increase the density of charge traps,30 measurements of D would provide the information of the influence of electrolytes on the traps. Figure 5 shows D values with various electrolytes. It was found that the D values were independent of [Li+]. Among the three different cations, TBA+ and HMIm+ showed comparable D values, and smaller D values than Li+. If the cations could play a role to form the traps of the electrons in the TiO2, the decrease of the D values was expected with the increase of cation concentration. However, the independence of the D values of the cation concentration and the larger D values with Li+ than those observed for the other two cations suggest that the cations did not change the condition of charge traps. The addition of tBP in the presence of Li+ lowered the D values and resulted in values comparable to those using TBA+ and HMIm+.35,36 Influences of the Redox Couple Concentration. Electron transfers from TiO2 and I- to dye cations are competitive processes (R2 and 5 in Figure 1). It has been reported that more than 30 mM of I- can reduce most of the dye cations under normal solar cell operation.37 Here, we measured the electron lifetimes with various I- concentrations with a fixed I3concentration (Figure 7a), and also with various I3- and Iconcentrations (Figure 7b). Significant increase was observed when [I-] was increased from 10 to 20 mM. This seems to be related to the change of the dye-cation reduction rate. In ref 37, the lifetime of the dye cations without the redox couple was measured, showing that the lifetime was in the order of 10-6/s under the conditions close to 1 sun. The recombination rate can be expressed by

Rate ) kR2nN

(1)

where kR2 is a rate constant, n is the density of electrons, and N is the density of dye cations; the lifetime of dye cations (τDye) is 1/kR2n, and that of electrons is 1/kR2N. The density of dye cations can be estimated from N ) Gτdye, where G is the

In the DSCs addressed here, acceptors for the electron in TiO2 are dye cations and I3-. As it is shown in Figure 7a, for the solution containing more than 20 mM of I-, most of acceptors are I3-. Thus, only the recombination with I3- is considered here and the rate can be written as

Rate ) kR1[e-]R[I3-]β

(2)

For small ∆[e-] and ∆[I3-], R and β can be approximated to be 1. The [e-] can be controlled by varying the irradiating light intensity. Under the same [e-], the influence of electrolytes on the electron lifetime should be attributed to the influence on either kR1 or [I3-]. Equation 2 does not include the influence of charge traps but describes only the rate of interface charge transfer. To explain the electron density dependent lifetime, the ratio of the density of conduction band electrons to the trapped electrons should be taken into account.26 In this paper we use eq 2 to discuss the difference of the electron lifetime measure by small perturbation under the same bulk electron density in the TiO2. Influence of Cations on kR1. The kR1 can be influenced by the reaction free energy (∆G°), reorganization energy (λ), and physical distance between the reactive species. Because no correlation between Voc and lifetime was observed (Figures 2, 3, and Table 1), the free energy difference between the quasiFermi level of TiO2 and the redox potential is less likely to influence the kR1. A similar conclusion can be found in another paper,32 showing that the recombination kinetics with dye cations and eTiO2 is not sensitive to the redox potential of dyes. In the report, it was explained as due to the fact that ∆G° was close to λ so that, the reaction was situated near the peak of the Marcus energy curve. The difference of reorganization energy for the transfer from the TiO2 to I3- was estimated to be c.a. 0.7 V in acetonitrile,39 indicating that the reaction with I3- is also situated near the peak. Influence of Cations on the Local Concentration of I3-. To explain the difference observed in Figure 2, we consider the condition of electric double layer (Helmholtz and diffuse layers) as a parameter influencing the electron lifetime. We define the local (surface) concentration of I3-, which is within the electron transferable distance from the surface of TiO2.40 When electrons are injected into the TiO2, the TiO2 is negatively charged, and electrical double layer must be formed on the surface of TiO2. The density of ions in the layer is determined by the density of electrons in the TiO2 and the kinds and concentrations of ions. For the cations, which are small enough to penetrate between the adsorbed dye molecular and the TiO2, the double layer is formed in about one nanometer at the ion

3484 J. Phys. Chem. B, Vol. 109, No. 8, 2005

Nakade et al.

Figure 5. IV curves of DSCs with and without tBP. The electrolytes were (a) 0.1 M of LiI with the addition of 0.5 M of tBP, (b) 0.1 M of TBAI with the addition of 1.5 M of tBP. 0.05 M of I2 was added for the all samples. (c) Electron lifetime vs short-circuit current in DSCs used for (a) and (b).

Figure 6. Electron diffusion coefficients vs short-circuit current of DSCs containing different electrolytes. The I2 concentration was 0.05 M for all.

concentration of 0.1 M. For this case, negatively charged TiO2 is screened within the size of the dye. Among the cations examined here, Li+ and DMPIm+ are the case. For bulky cations, which cannot penetrate between the dye and TiO2, distance longer than the size of the dye is needed for the screening, and the dyes are located in the double layer. TBA+ has been reported as the case.15 Figure 8 shows the schematic view describing the concentration profile of the ions. As a consequence of the block of the TBA+, anions feel a repulsive force to penetrate between the dye and TiO2. This reduces the local concentration of I3- and results in the longer electron. The condition in Figure 8 also suggests that removing or reducing

the adsorbed dyes should result in similar electron lifetimes for Li+ and TBA+, because they form a similar electric double layer. As expected, comparable lifetimes were found between Li+ and TBA+ (Figure 4). The thickness of the double layer depends on the concentration of ions. By the addition of LiClO4, a thinner electric double layer is expected. This could lead to shorter electron lifetimes, because the anionic redox couple can reach closer to the TiO2 surface. However, the independence of the lifetime from the concentration, seen in Figure 3, implies that the additional ClO4decreased the local concentration of I3-, balancing the increase of the local concentration by thinning the layer. The results of the diffusion coefficient measurements (Figure 5) can be also explained with electric double layer. For the electron diffusion in nanoporous TiO2 electrodes, the diffusion coefficients are influenced by cations through ambipolar diffusion.16,17,25 In Figure 6, no difference was observed between 0.1 and 1 M of Li+ in the DSCs. This is probably because the concentration of Li+ on the TiO2 surface is determined by the condition of the double layer, which is determined by the density of electrons in the TiO2. Thus, the electron density dependent diffusion coefficients were not influenced by the concentration of Li+ cations in the solutions.41 Reduction kinetics of dye cations with various cations in solution has been studied by Pelet et al.10 They found the fastest kinetics with Li+ and attributed it to the adsorbed cations that should attract I- on the surface. It has been clarified that Jsc of the DSCs using only Li+ cations as the counter charge of Igradually decreased over days of storage.42 The electron lifetimes were also decreased with the storage. The results could be interpreted as due to an assumption that the local [I3-] was increased by the increase of the amount of the adsorbed Li+.

Role of Electrolytes on Charge Recombination I

J. Phys. Chem. B, Vol. 109, No. 8, 2005 3485

Figure 7. (a) Electron lifetime as a function of I- concentration in DSC. The concentration of I3- was fixed at 2 mM. (b) Electron lifetime as function of I2 concentration in DSC. The amount of LiI was fixed at 0.7 M and the amount of I2 was varied. The values of the lifetime are the average of two samples.

Figure 8. Schematic view of ion concentration profile in DSC under light irradiation. Li+ can penetrate into the dye so that they can screen the negatively charged TiO2 within the size of the dye. TBA+ is too large for the penetration, i.e., the concentration of TBA+ is nearly zero in range of the dye diameter. Anion concentration profile is determined by the potential of the TiO2 and the concentration profile of cation.

The influences of the adsorbed cations could partially contribute to their shorter lifetimes observed in Figure 2. However, the cation concentration independent lifetime cannot be explained only by the attraction of the anions by the surface adsorbed cations, but the conditions of electronic double layer can provide consistent explanations for the results observed in this paper. The model using the electrical double layer has explained the shift of the redox potentials of the adsorbed dyes in the solution with various cations.15 Our results suggest that the double layer influences not only the charge injection efficiency but also the electron diffusion coefficients and lifetimes. The model used here suggests that, if cationic redox couples are employed instead of anionic redox couples, the influence of cations on the electron lifetime should appear differently. This model would solve the confusion of the interpretation of the role of Li cation on the performance of DSC.19,43 The following paper addresses the issue with cobalt complex redox couples.44 Role of tBP. For DSCs using I-/I3- redox couple and Li+ as their counter charge, the addition of tBP has been known to increase Voc. The effect of tBP was initially interpreted based on the difference of I-V characteristics and was attributed to the suppression of the back reaction caused by the adsorption of tBP on the TiO2 surface.34 However, the interpretation of I-V characteristics is not straightforward because the surfaceadsorbed species can influence both the position of CBP and recombination kinetics, simultaneously. Later, the same group developed IMVS and measured lifetime with and without tBP,

and concluded that tBP increased the lifetime by 1.7 times.12 On the other hand, several recent reports have interpreted the influence of tBP as to suppress the recombination.18,43 Our results have indicated that the addition of tBP did not affect the lifetime in the presence of Li+ or TBA+. Recently, Hara et al. showed by FTIR measurements that tBP was adsorbed on TiO2 surface, and the presence of Li cation increased the amount of the tBP adsorption.18 We have reported that the surface adsorption of Li increased the D values.16,17 The decrease of the D values in Figure 5 implies that surface Li+ concentration was reduced by the interaction with tBP. An absorption peak of tBP by FT-IR seemed to be shifted only with the presence of Li cations.18 These results suggest that, with Li cations, tBP reduces the surface positive charge density by the interaction, resulting in the shift of the CBP back to negative. It was mentioned that the basicity of tBP could also contribute to the increase of Voc.30 Our results showed that with TBA+, the Voc was independent of the concentration of TBA+ and was increased by the addition of tBP. For this case, this increase in Voc is probably induced by the basicity of the tBP and its adsorption on the surface that affects the CBP. No influence of tBP in the copresence of TBA+ on the electron lifetime suggests also that the adsorption of tBP does not influence kR1. Conclusions Electron lifetimes in DSCs employing the I-/I3- redox couple were studied with various compositions of electrolytes. Shorter lifetime was found with adsorptive cations, such as Li+ and DMPIm+ in comparison with less-adsorptive cations of TBA+. The observation was interpreted with the change of the surface concentration of I3-, which depends on the profile of the electrical double layer formed by these cations. For the case of TBA+, it is too large to penetrate between the dye and TiO2, and the cation cannot screen the negatively charged TiO2 within the size of the dye. This makes I3- difficult to approach to the TiO2 surface, resulting in less probability of electron transfer from the TiO2. The role of tBP in the presence of Li+ was found to suppress the adsorption of Li+ on the TiO2 surface. This retards the positive shift of the TiO2 conduction band potential, resulting in higher Voc of DSCs. Appendix Since the DSCs examined here have electron density dependent electron lifetimes, a requirement to compare the electron lifetimes among different samples is to have the same electron

3486 J. Phys. Chem. B, Vol. 109, No. 8, 2005

Figure A1. Electron lifetimes in DSCs using various electrolytes. TBA 0.7 M of TBAI + 0.05 M of I2; DMPIm 0.7 M of DMPIm and 0.05 M of I2; TBA+tBP 0.7 M of TBAI + 0.05 M of I2 + 1.5 M of tBP. The lifetimes were measured at open circuit conditions, and the bottom axis shows the short circuit current under the same light intensities used for the lifetime measurements.

Nakade et al. pared. DSC is connected to a potentiostat (Hokuto Denko, HAB151) and irradiated by a diode laser. The laser intensity and the potential of the potentiostat were controlled by the PC through the output potential of the D/A converter. Initially, the bias potential was applied so that there was no current from the cell, i.e., equivalent to the open circuit condition. Then, the laser intensity and the bias potential were turned off simultaneously by the PC so that the condition was changed to short circuit in dark. Note that the actual response time difference between the two outputs of the D/A card was 15 µs. The resulting current transient was monitored through the potentiostat and stored on a digital oscilloscope. The electron density at the open circuit conditions were estimated by the integration of the current transient. The measurements were repeated with various laser intensities. For the experiments, we prepared DSCs with three different electrolytes: (a) DMPImI (0.7 M), (b) TBAI (0.7 M), and (c) TBAI (0.7 M) + tBP (1.5 M). For all samples, 0.05 M of I2 was dissolved. Electron lifetimes were measured by the same method used in this paper. Figure A1 shows the plot of the lifetime vs Jsc. The trends seen in Figures 1 and 4 were reproduced. The difference between the lifetime with DMPIm+ and TBA+ at the same Jsc was lager than that in Figure 1. We suppose the difference was associated with the different batch of the dye purchased from Solaronix. Figure A2 shows the lifetime as a function of the electron density measured by the extraction method. The trend was the same but the difference was expanded. The result was consistent with the rough estimation using Jsc and τ. References and Notes

Figure A2. Electron lifetimes in the DSCs used in Figure A1 as a function of electron density in the DSC. The electron densities were estimated by a charge extraction method.

density. In practice, several papers including this plot the lifetimes vs Jsc, which is given under the same intensity used for open circuit voltage transient measurements. Here the Jsc and the electron density at open circuit conditions are roughly related with Gτ, where G is an electron generation rate and τ is electron lifetime. Assumptions used here are that the G is proportional to Jsc and the τ measured by small perturbation method can represent all of the electrons in the DSC. The assumption also limits the experimental condition that the sample thickness should be less than the electron diffusion length. If the assumptions are valid, samples having larger τ values under the condition of the same Jsc must have larger electron density. This means that the difference of the lifetimes observed on the plot with Jsc should be larger when it is replotted vs electron density. Thus, a plot vs Jsc can allow us to compare the electron lifetimes semi-quantitatively but does not provide a fixed scale. To measure the electron density, additional experimentation is needed. To check the argument above, we estimated the electron density by the charge extraction method, proposed by Peter et al.20 The experiment was performed as follows. First, a personal computer (PC) equipped with a digital to analog (D/ A) converter (National Instruments, DAQcard-1200) was pre-

(1) Garcia-Belmonte, G.; Kytin, V.; Dittrich, T.; Bisquert J. J. Appl. Phy 2003, 94, 5261. (2) Cinnsealach, R.; Boshloo, G.; Rao, N.; Fitzmaurice, D. Sol. Energy Mater. Sol. Cells 1999, 57, 107. (3) Tachikawa, T.; Tojo, S.; Fujitsuka, M.; Majima, T. J. Phys. Chem. B 2004, 108, 11054. (4) O’Regan, B.; Gra¨tzel, M. Nature 1991, 353, 737. (5) Tachibana, Y.; Moser, J. E.; Gra¨tzel, M.; Klug, D. R.; Durrant, J. R. J. Phys. Chem. 1996, 100, 20056. (6) Asbury, J. B.; Ellingson, R. J.; Ghosh, H. N.; Ferrere, S.; Nozik, A. J.; Lian, T. J. Phys. Chem. B 1999, 103, 3110. (7) Benko¨, G.; Skarman, B.; Wallenberg, R.; Hagfeldt, A.; Sundstrom, V.; Yartsev, A. P, J. Phys. Chem. B, 2003, 107, 1370. (8) Kelly, C. A.; Farzad, F.; Thompson D. W.; Stipkala, J. M.; Meyer, G. J. Langmuir, 1999, 15, 7074. (9) Haque, S. A.; Tachibana, Y.; Willis, R. L.; Moser, Gra¨tzel, M.; J. E.; Klug, R. D.; Durrant, J. R. J. Phys. Chem. B 2000, 104, 538. (10) Pelet, S.; Moser, J.-E.; Gra¨tzel, M. J. Phys. Chem. B 2000, 104, 1791. (11) Fisher, A. C.; Peter, L. M.; Ponomarev, E. A.; Walker, A. B.; Wijayantha, K. G. U. J. Phys. Chem. B 2000, 104, 949. (12) Schlichtho¨rl, G.; Huang, S. Y.; Sprague, J.; Frank, A. J. J. Phys. Chem. B 1997, 101, 8141. (13) Nelson, J.; Haque, S. A.; Klug, R. D.; Durrant, J. R. Phys. ReV. B 2001, 63, 205321. (14) Liu, Y.; Hagfeld, A.; Xiao, X.-R.; Lindquist, S.-E. Sol. Energy Mater. Sol. Cells 1998, 55, 267. (15) Zaban, A.; Ferrere, S.; Gregg, B. A. J. Phys. Chem. B 1998, 102, 452. (16) Nakade, S.; Kambe, S.; Wada, Y.; Kitamura, T.; Yanagida, S. J. Phys. Chem. B 2001, 105, 9150. (17) Kambe, S.; Nakade, S.; Kitamura, T.; Wada, Y.; Yanagida, S. J. Phys. Chem. B 2002, 106, 2967. (18) K. Hara, Y. Dan-oh, C. Kasada, Y. Ohga, A. Shinpo, S. Suga, K. Sayama, H. Arakawa Langmuir 2004, 20, 4205. (19) Sapp, S. A.; Elliott, M.; Contado, C.; Caramori, S.; Bignozzi, C. A. J. Am. Chem. Soc. 2002, 124, 11215. (20) Peter, L. M.; Duffy, N. W.; Wang, R. L.; Wijayantha, K. G. U. J. Electroanal. Chem. B 2002, 524, 127. (21) Kru¨ger, J.; Plass, R.; Gra¨tzel, M.; Cameron, P. J.; Peter, L. M. J. Phys. Chem. B 2003, 107, 7536. (22) Bisquert, J.; Zaban, A.; Salvador. P. J. Phys. Chem. B 2002, 106, 8774.

Role of Electrolytes on Charge Recombination I (23) Cao, F.; Oskam, G.; Meyer, G. J.; Searson, P. C. J. Phys. Chem. 1996, 100, 17021. (24) 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. (25) Kopidakis, N.; Schiff, E. A.; Park, N.-G.; van de Lagemaat, J.; Frank, A. J. J. Phys. Chem. B 2000, 104, 3930. (26) Bisquert, L.; Vikhrenko, V. S. J. Phys. Chem. B 2004, 108, 2313. (27) Kopidakis, N.; Benkstein, K. D.; van de Lagemaat, J.; Frank, A. J. J. Phys. Chem. B 2003, 107, 7759. (28) Nakade, S.; Saito, Y.; Kubo, W.; Kitamura, T.; Wada, Y.; Yanagida, S. J. Phys. Chem. B 2003, 107, 8607. (29) Nakade, S.; Saito, Y.; Kubo, W.; Kanzaki, T.; Kitamura, T.; Wada, Y.; Yanagida, S. Electrochem. Commun. 2003, 5, 804. (30) Wang, H.; He, J.; Boschloo, G.; Lindstro¨m, H.; Hagfeldt, A.; Lindquist, S.-E. J. Phys. Chem. B 2001, 105, 2529. (31) Hagfeldt, A.; Gra¨tzel, M. Chem. ReV. 1995, 95, 49. (32) Clifford, J. N.; Palomares, E.; Nazeeruddin, M. K.; Gra¨tzel, M.; Nelson, J.; Li, X.; Long, N. J.; Durrant, J. R. J. Am. Chem. Soc. 2004, 126, 5225. (33) The difference of I-V characteristics between the DSCs using TBA+ and DMPIm+ is smaller for the recent measurements. The DSCs used in Figures A1 and A2 showed 8.6 mA/cm2 and 0.78 V for TBA+ and 7.2 mA/cm2 and 0.74 V for DMPIm+. (34) Huang, S.; Schlichtho¨rl, G.; Nozik, A. J.; Gra¨tzel, M.; Frank, A. J. J. Phys. Chem. B 1997, 101, 2576. (35) The larger D values with Li+ seem to be due to larger amount of Li+ adsorption on the TiO2 during the time period between the cell fabrication and measurements. D values more compatible with other cations were observed from the cells measured within 1 h from the cell preparation. For the lifetime in the DSCs using Li+, we confirmed the shorter lifetime with Li+ than with TBA+ from the cells fabricated 1 h before the measurements. (36) Probably, a more precise description is that the addition of tBP prevents Li+ from adsorption.

J. Phys. Chem. B, Vol. 109, No. 8, 2005 3487 (37) Montanari, I.; Nelson, J.; Durrant, J. R. J. Phys. Chem. B 2002, 106, 12203. (38) Nakade, S.; Saito, Y.; Kubo, W.; Kitamura, T.; Wada, Y.; Yanagida, S. J. Phys. Chem. B 2004, 108, 1628. (b) Duffy, N. W.; Peter, L. M.; Rajapakse, R. M. G.; Wijayantha, K. G. U. Electrochem. Commun. 2000, 2, 658. (c) Willis, R. L.; Olson, C.; O’Regan, B.; Luz, T.; Nelson, J.; Durrant, J. R. J. Phys. Chem. B 2002, 106, 7605. (39) The value was estimated with eq 8 in ref 3, using ionic radius of 0.5 nm for I3- (Barbe, C. J. et al. J. Am. Ceram. Soc. 1997, 80, 12). (40) With the anion concentraion, nanion(x), where x is the distance from the TiO2 surface, local (surface) concentration would be described by the integral of nanion(x) between 0 and l, divided by the l and a unit area. Here the l is the electron transferable distance. Since the TiO2 is negatively charged, nanion(0) should be smaller than the volume concentration. If the l is much shorter than the thickness of the double layer, the local concentration should be much lower than the volume concentration. (41) We note that, for the measurements performed here, we applied a laser pulse superimposed on a continuous laser irradiation during the measurements. The dye used here has low absorption coefficients against the wavelength of the CW laser so that the nearly constant electron generation in the electrodes is expected. Therefore, the condition of the double layer can also be expected to be uniform in the electrodes. In previous papers (refs 16 and 17), we measured current transients induced by a UV laser pulse without the assist of bias light. For these cases, electrons were generated at outermost layer of the electrodes and diffused in the electrodes, which have no double layer yet. Thus, the transport of electron should be accompanied with that of counter charges, and thus the ambipolar diffusion can be applied with the concentration of cations in the solution. (42) Kambe, S. Ph.D. Thesis, Osaka University, 2002. (43) Nasbaumer, H.; Zakeeruddin, S. M.; Moser, J. E.; Gratzel, M. Chem Eur. J. 2003, 9, 3756. (44) Nakade, S.; Makimoto, Y.; Kubo, W.; Kitamura, T.; Wada, Y.; Yanagida, S. J. Phys. Chem. B 2005, 109, 3488.