Distance and Driving Force Dependencies of Electron Injection and

Apr 9, 2010 - Distance and Driving Force Dependencies of Electron Injection and Recombination Dynamics in Organic Dye-Sensitized Solar Cells...
0 downloads 0 Views 1MB Size
14358

J. Phys. Chem. B 2010, 114, 14358–14363

Distance and Driving Force Dependencies of Electron Injection and Recombination Dynamics in Organic Dye-Sensitized Solar Cells† Joanna Wiberg,‡ Tannia Marinado,|,§ Daniel P. Hagberg,⊥ Licheng Sun,⊥ Anders Hagfeldt,#,§ and Bo Albinsson*,‡ Department of Chemical and Biological Engineering/Physical Chemistry, Chalmers UniVersity of Technology, 412 96 Go¨teborg, Sweden, Center of Molecular DeVices, Royal Institute of Technology, Chemical Science and Engineering, Physical Chemistry, 100 44 Stockholm, Sweden, and Center of Molecular DeVices, Royal Institute of Technology, Chemical Science and Engineering, Organic Chemistry, 100 44 Stockholm, Sweden ReceiVed: January 12, 2010; ReVised Manuscript ReceiVed: March 22, 2010

A series of dyes based on a triphenylamine donor and a rhodanine acetic acid anchor/acceptor for solar cell application has been studied with regards to electron injection and recombination kinetics using femtosecond transient absorption. The series contains three dyes, with estimated electron transfer distances ranging from 17.2 to 11.0 Å, and which have shown significant differences in energy conversion efficiencies. The injection and recombination kinetics were studied in the NIR region where electrons in the conduction band of the TiO2 are suggested to absorb. For all dyes, the injection rate is larger than (200 fs)-1 which implicates a quantitative injection efficiency. Surprisingly, the subsequent recombination reaction has a rate that increases with increasing linker length. On the other hand, this behavior is consistent with the concomitant decrease in driving force for this series of dyes. Moreover, the lifetimes show exponential distance dependence when corrected for driving force and reorganization energy, which indicates a superexchange interaction between the electrons in TiO2 and the radical cations of the dyes. A dependence on probe wavelength of the attenuation factor was found, giving a β value of 0.38 Å-1 at 940 nm and 0.49 Å-1 at 1040 nm. The difference is suggested to be due to the difference in electronic coupling between fully separated dye cations and injected electrons versus geminate electron-hole pairs. Addition of tert-butylpyridine, which from previous work is known to give a substantial drop in the IPCE values for the studied dyes, was found to decrease the amount of long-lived electrons in the TiO2 without affecting the injection rate. Introduction In the vast field of research on dye-sensitized solar cells, more and more attention is drawn to understanding the electron transfer kinetics in this type of cells.1-4 Similar to the field of donor-bridge-acceptor model systems, work is now being made on homologous building block series of dyes. Small structural changes, which give impact on the energy conversion efficiencies can thereby more easily be assessed.5,6 The effect of changing the electron transfer distance has been investigated for a number of ruthenium complexes7-10 and also for organic dyes,11,12 showing anything from no distance dependence to a strong exponential distance dependence with an attenuation factor β of 1 Å-1. We present here a study of the effect of electron transfer distance on the electron injection and charge recombination kinetics in a series of organic dyes for solar cell sensitization. The rates of electron injection and recombination were studied with femtosecond transient absorption measurements on three dyes with different linker lengths: D5L0A3, D5L1A3, and D5L2A3. Each dye, referred to as L0, L1, and †

Part of the “Michael R. Wasielewski Festschrift”. Chalmers University of Technology. § Royal Institute of Technology, Chemical Science and Engineering, Physical Chemistry. | Current address: Division of Chemistry and Materials, Faculty of Textile Science and Technology, Shinshu University, Ueda 386-8567, Japan. ⊥ Royal Institute of Technology, Chemical Science and Engineering, Organic Chemistry. # Current address: Uppsala University, Dept. of Physical and Analytical Chemistry, Box 259, 751 05 Uppsala, Sweden. ‡

CHART 1: Dye Structures (left) with Their Building Block Representation (right)

L2 according to linker length, comprises a triphenylamine donor and rhodanine acetic acid with the dual purpose of anchor and acceptor and is structurally similar to the highly efficient dye D5.13 The difference in linker length gives distances of 17.2, 14.9, and 11.0 Å between the triphenylamine center and the carboxylate carbon in the anchor group binding to the TiO2 surface for L2, L1, and L0, respectively. The orientation of the dyes on the surface will of course add uncertainties to the distances, but XPS studies have shown that the dye molecules are rather vertically arranged with the triphenyl amine pointing out from the surface.14 The structures of the dyes and a building block schematic of the L0-L2 series are given in Chart 1.

10.1021/jp1002963  2010 American Chemical Society Published on Web 04/09/2010

Electron Injection and Recombination Dynamics

J. Phys. Chem. B, Vol. 114, No. 45, 2010 14359

Figure 1. Energy level diagram showing driving forces for injection and recombination, respectively. The addition of tBP is estimated to raise the energy level of the TiO2 CB by 0.15 V.

TABLE 1: Estimations of Driving Forces, Reorganization Energies, and Activation Energies for Electron Injection and Charge Recombinationa molecule

∆G0inj/eV

L0 L1 L2

-0.61 -0.54 -0.60

L0 L1 L2

-0.46 -0.39 -0.45

∆G0rec/eV

λb/eV

∆Gqinjc/eV

∆Gqrecc/eV

0.03 0.04 0.03

0.17 0.13 0.08

0.5 M tBP 0.96 0.07 0.95 0.08 0.94 0.06

0.24 0.19 0.13

0.5 M LiClO4 -1.77 0.96 -1.65 0.95 -1.48 0.94 0.5 M LiClO4, -1.92 -1.80 -1.63

a The E00 values are obtained for the molecules in solution. b λ is assumed to be the same for both charge injection and recombination and independent of addition of tBP. c ∆Giq ) (∆Gi0 + λ)2/4λ.

Solar cell performance measurements on L0-L2 have shown increasing overall cell efficiency with decreasing molecule length. As the conjugation length is increased from L0 to L2, the absorption spectrum is consequently red-shifted; hence, the spectral overlap with the solar spectrum is increased. However, it has been shown by Marinado et al. that the IPCE values are decreased for L2 and L1 compared to L0.5 Suggestions were made that the limiting factor for charge injection observed for the dyes with longer linker length was the lack of sufficient driving force for electron injection. Additionally, the dyes proved to be sensitive to addition of the conduction band raising additive tert-butylpyridine (tBP) which cause a 50% decrease in the IPCE values. The tBP additive is commonly used to increase the open circuit voltage but has also been shown to suppress recombination between injected electrons in the TiO2 and the redox electrolyte.15-17 To correlate the efficiency to the driving forces (∆G0) for electron injection and charge recombination, the potential energy levels were estimated by adding the oxidation potential of the dyes in solution to their excitation energies (E00)srepresenting the LUMO level while the conduction band (CB) energy of the TiO2 was estimated to -0.5 V vs NHE. A schematic of the energy levels and the ET processes is shown in Figure 1, and the estimated values of ∆G0 and reorganization energies (λ) for both electron injection and charge recombination are given in Table 1. The reorganization energies have been estimated previously from DFT calculations for the molecules in solution.18 The discrepancy between the reorganization energies for molecules in solution and adsorbed to the TiO2 surface is difficult to estimate. Rather, as the study focus on the distance dependence and the reorganization energies for all three dyes are so similar, the uncertainty in the values used for λ will not change the qualitative distance dependence but put limitations on the reliability of the reported β-value, Vide infra.

In the Marcus model for electron tunneling, the electron transfer rates are not directly dependent on driving force but rather on the activation energy, ∆Gq, which is related to the sum of driving force and reorganization energy. For comparison, also ∆Gq values are shown in Table 1. The impeding effect of tBP on the energy conversion efficiency has been correlated to a decrease in the injection rate for ruthenium dyes and has been explained in terms of a decrease in driving force for the injection reaction.19,20 It is therefore of interest to study the effect of tBP on this series of dyes, as they, first, have shown great sensitivity to tBP addition and, second, are all organic dyes which may well show different behavior than ruthenium dyes due to their much lower quantum yield for intersystem crossing. Experimental Methods The syntheses of all the compounds used in this study have been reported previously.18 Acetonitrile (MeCN) of spectrophotometric grade (99.5%) from Sigma-Aldrich was used as purchased. Doctor bladed TiO2 photoelectrodes were sintered at 450 °C for 45 min and thereafter cooled down and stored in the dark. Before use, they were reheated to 120 °C for 30 min and subsequently immersed into an ∼0.1 mM solution of the dye in MeCN and left overnight. After dye sensitization, the films were placed in neat MeCN in two consecutive baths for 10 min each. Electrolytes were prepared with 0.5 M LiClO4 (99% Fluka) in MeCN and additionally with 0.5 M 4-tertbutylpyridine (99% Sigma). Ground State Absorption. Ground state absorption spectra were measured on a Cary 5000 with a scan rate of 600 nm/ min. An unsensitized TiO2 film in acetonitrile was used for baseline correction when appropriate. Transient Absorption. The samples were sensitized TiO2 films kept in a 2 mm cuvette with either 0.5 M LiClO4 in acetonitrile or a mix of 0.5 M LiClO4 and 0.5 M 4-tertbutylpyridine in acetonitrile. The femtosecond setup has been described in detail elsewhere.21,22 For all decay traces, the excitation wavelength was set to 475 nm with a pulse intensity of 0.3 µJ per pulse (the beam diameter was estimated to 1 mm, giving 40 µJ cm-2 per pulse). Control measurements were made with increased pump energy to ensure that the pump energy was low enough to avoid effects on the shape of the transient decays. To enable direct comparison between these rather difficult measurements, all transient decay traces were measured on a single day. The measurements have been repeated several times with consistent trend and qualitative shape, although slight variations in absolute values have been recorded. The transient absorption spectra were measured with slightly higher intensity ∼0.4 µJ/pulse pumping at 475 nm for L2 and 484 nm for L1, while L0 was excited at 475 nm with a pump power of 0.2 µJ per pulse. 1000-2000 averages were made for each decay and spectrum. The nanosecond transient spectra were measured using equipment described in previous work.21 Excitation wavelengths were 458, 484, and 500 nm for L0, L1, and L2, respectively, with a pump energy of ∼1 mJ per pulse (the beam diameter was estimated to 5 mm, giving 5 mJ cm-2 per pulse). The signals were detected using a spectrograph and a CCD camera (iStar, Andor). Around 40 measurements were averaged for each delay time depending on signal amplitude. Results The aim of the study is to characterize the distance dependence of both the electron injection into a thin film of TiO2 nanoparticles and the subsequent charge recombination kinetics

14360

J. Phys. Chem. B, Vol. 114, No. 45, 2010

Wiberg et al.

Figure 2. Absorption spectra of L0-L2 adsorbed on TiO2 in the presence of 0.5 M LiClO4 with (dashed) and without (solid) addition of 0.5 M tBP. The difference in absorption amplitude between the samples is due to differences in dye concentrations on the TiO2 film.

for three triphenylamine based organic dyes, using femtosecond transient absorption spectroscopy. In addition, the effect of the open circuit voltage (Voc) increasing additive tBP on said kinetics was investigated. Ground State Absorption. Ground state absorption spectra of the dyes adsorbed on TiO2 were measured in the presence of LiClO4. The lowest energy absorption peak is red-shifted from L0 to L2, as shown in Figure 2, as expected given the increase in conjugation length. When adding tBP, a slight blue shift of the absorption peaks was observed, which can be due to both the change in solvent environment and a shift in the dye energy levels due to the change in surface charge in which the LUMO level is shifted more than the HOMO because of the push-pull character of the dye. Transient Absorption Measurements. Transient absorption measurements on the fs to ns time scale were performed in the visible and NIR region in order to elucidate the kinetic parameters governing this system. The fs and ns transient absorption spectra of L0-L2 adsorbed on TiO2 in the visible region are shown in Figure 3. For L2, the spectral shape remains rather constant up to about 100 ps. A slight red shift is observed which evolves on longer time scales, as indicated by the spectra measured at a 500 ns delay. For L1 and L0, the shifts are faster, and after 1 ns, the spectral shape is the same as that measured after a few hundred nanoseconds. The spectra obtained on the nanosecond time scale overlap well with those obtained with photoinduced absorption (PIA) at much longer time scales. Thus, these spectral features should be the signal from the oxidized dye overlapping with contributions from electrons in the TiO2 CB. In conclusion, both the excited states of the dyes as well as the radical cations absorb in the visible region, which makes it difficult to separate the signals from the different species contributing to the measured differential absorption. The spectral shifts observed on the picosecond time scale can thus not be interpreted as a simple transition from excited state to radical state, since there is a signal present already after 1 ps in the region of radical absorption for the L1 and L2 dyes; cf. Figure 3 at wavelengths above 600 nm. For L0, the interpretation is even more complex, as stimulated emission is covering the positive contribution from the L0 cation in the 700 nm region. We assign the observed spectral evolution to be induced by relaxation effects within the cation and of the cation-TiO2 complex. Due to the complexity of the overlapping signals in the vis region of the spectrum, our focus was shifted to study the kinetics in the NIR region where CB electrons are suggested to

Figure 3. Transient spectra of L0-L2 on TiO2 with 0.5 M LiClO4 in acetonitrile. The spectra were measured with femtosecond TA for delay times up to 1 ns and with nanosecond TA for longer times.

absorb.23-25 This makes it possible to follow the kinetics of the transferred electrons explicitly. The signal buildup in the NIR region can thus directly be assigned to the injection, while the following decay displays the recombination kinetics. In Figure 4, the transient absorption decay traces measured at 940 and 1040 nm are shown. The transient decays were measured at both 940 and 1040 nm, since we have previously found differences in the amount of ultrafast recombination between these wavelengths which were ascribed to differences in absorption between more reactive surface electrons and less reactive bulk electrons.22 The signal buildup at both wavelengths is pulse-width-limited for L1 and L2, while L0 shows a fast rise time in the order of 100-200 fs. Although the addition of tBP has resulted in a decrease in the injection efficiency for ruthenium dyes,20 no increase in the signal rise time was found for either dye. The recombination reactions are from Figure 4 clearly nonexponential and show the presence of an ultrafast component, with fractional amplitudes that differ greatly between the dyes. Ultrafast injection and recombination has been detected for other organic dyes such as fluorescein27,26 and also for porphyrins.12 The decays also exhibit a wavelength dependence where the ultrafast component is more prominent at 940 nm than at 1040 nm. Three exponentials (four for L0 which needed a rise time component) were used to fit the transient absorption decays for L0-L2 in LiClO4, which are given with their respective amplitudes relative to normalized data in Table 2. Interestingly, the amplitude of the ∆A signal increases at 940 nm and decreases at 1040 nm after addition of tBP for L0 and L2.

Electron Injection and Recombination Dynamics

J. Phys. Chem. B, Vol. 114, No. 45, 2010 14361

Figure 4. Transient absorption decay curves for L0-L2 adsorbed to TiO2 in a solution of 0.5 M LiClO4 (black) and with addition of 0.5 M tBP (gray) in acetonitrile. The probe wavelength was 940 nm (left) and 1040 nm (right), while the excitation wavelength was 475 nm.

TABLE 2: Lifetimes and Amplitudes of Exponential Fits to Normalized Transient Decay Traces at 940 and 1040 nm for the Series L0-L2a in 0.5 M LiClO4 molecule τrise/ps L0 L1 L2

0.11

L0 L1 L2

0.13

Arise

τ1/ps

A1

τ2/ps

A2

τ3/ps

A3

y0

940 nm -1.04 1.9 0.20 13 0.24 134 0.30 0.30 0.55 0.48 3.2 0.25 91 0.13 0.22 0.14 0.86 1.01 0.32 58 0.12 0.14 1040 nm -0.82 2.7 0.32 24 0.72 0.20 22 0.44 0.42 3.2

0.28 167 0.26 0.19 0.21 166 0.31 0.32 0.30 46 0.25 0.16

a

The TA signal was fitted with a sum of exponential decays and deconvoluted using a Gaussian shaped pump probe correlation function.

To aid in the comparison of the decay rates between the studied dyes with and without tBP, an averaged lifetime for each sample and wavelength was employed. The averaged lifetimes were estimated by integration of the area under the curve given by the normalized transient absorption decay trace between -1 and 1550 ps which in essence gives an amplitude averaged lifetime of the electrons in the TiO2. The integrated lifetime will clearly depend on the boundaries of the integration, but also a direct calculation of the amplitude averaged lifetime will be subject to an arbitrarily chosen lifetime to cover the amplitude of the long-lived component, y0, in Table 2. Also, with the rapid decays measured, a recombination halftime will not resolve the differences in the amount of remaining signal that we observe between wavelengths as well as after addition of tBP. The integrated lifetimes are presented in Table 3. Also, after addition of tBP, there is a clear dependence both on linker length and probe wavelength on the recombination kinetics. Moreover, for all dyes at both 940 and 1040 nm, the integrated lifetimes are shorter with the addition of tBP, which indicates a faster recombination process induced by the additive.

TABLE 3: Integrated Lifetimes of the Studied Molecules at 940 and 1040 nm in Either 0.5 M LiClO4 or a Mixture of 0.5 M LiClO4 and 0.5 M tBP in Acetonitrile molecule

L0

L1

L2

τavLiClO4/ps τavtBP/ps

940 nm 501 348

346 264

224 181

τavLiClO4/ps τavtBP/ps

1040 nm 332 228

538 371

258 207

Discussion There are two major concerns that must be taken into account when interpreting the transient signals in the NIR region. First, the assignment of the signal to injected electrons must be done with caution, keeping in mind the potential contribution from dye radical cations and the spectral evolution of these. In our case, there is no reason to expect strong cation absorption in the NIR region that would gravely change the interpretation of the results. The trend of the signal amplitude and their decays fits the interpretation of electron absorption too well to be a coincidence with potential spectral evolution of cationic dye radicals. Second, reorganization of electrons within the nanoparticles has been shown to give spectral evolution with a 500 ps decay component in this wavelength region, which is not related to the recombination process.27 However, no 500 ps component was observed; we assume that it is partly covered by other decay components. Moreover, the focus of this study is to compare the electron transfer processes between the dyes, which is why a constant decay component for all three dyes will not affect the qualitative results. The results show ultrafast injection and a following rapid recombination. The amount of long-lived charges at 940 nm, the y0 values given in Table 2, matches the trend of the IPCE values within this series of dyes. However, the absolute values are not consistent which is expected, since the IPCE values are measured at short circuit

14362

J. Phys. Chem. B, Vol. 114, No. 45, 2010

Wiberg et al.

conditions in the presence of the I-/I3- redox mediatorsall of which are expected to affect the system. To seek the distance dependence of the electron injection is not possible with the time resolution used in this study. However, with injection rates of more than (200 fs)-1, the injection yield is quantitative also after addition of tBP. Yet, differences in the recombination rates are clearly resolved, which is why this section will focus on the analyses of the distance dependence of the charge recombination and the change in that dependence caused by the addition of tBP. The distance dependence of electron transfer processes is often analyzed in terms of the attenuation factor β Via the exponential distance dependence of the electronic coupling between the electron donor and acceptor, V2 ∝ exp(-βr). A longer distance will thus give a decrease in V, which decreases the rate constant for electron transfer, kET, and consequently increases the lifetime. In most cases of ultrafast charge recombination between a sensitizing dye and electrons injected into the TiO2, the kinetics cannot be described by a single exponential decay.3,28,29 Attempts have been made to use a sum of exponentials, stretched exponentials, or more complex functions to fit the measured data. Indeed, the fitted rate constants cannot easily be translated to an electronic coupling. By first inspection of the lifetimes given in Table 3, there seems to be an inverse relationship between the recombination rate and the linker length at 940 nm and no apparent linker dependence at 1040 nm. According to Marcus theory, electron transfer rates are highly dependent on the activation energy. In the studied series of dyes, the activation energy increases with decreasing linker length due to the increase in oxidation potential, as indicated in Table 1. Although the recombination reaction is clearly not an electron transfer process between weekly coupled diabatic states, the effect of differences in activation energy needs to be taken into account. For simplicity, the Marcus equation, eq 1, was used, but as the electronic coupling is too large to be estimated with this approach, we have here considered the pre-exponential factor as an activation energy corrected rate constant, kET ) 1/τ*, as given by eq 1.

kET )



(

)

(

1 π -(∆G0 + λ)2 -(∆G0 + λ)2 ) |V| 2 exp exp 4λk T τ* 4λkBT p λkBT B 2

)

(1)

With this approach, a clear exponential distance dependence is found for the lifetimes obtained while probing at 1040 nm. The attenuation factor β at 1040 nm was estimated to 0.49 Å-1 which is similar to the attenuation factors found in donor-bridgeacceptor systems with conjugated bridges. The estimated value of β will naturally depend on the choice of both lifetime averaging and correction for activation energy, but a stringent use of method will enable comparisons within the system. In fact, β was found to be dependent on probe wavelength, giving a lower attenuation factor of 0.38 Å-1 from the lifetimes measured at 940 nm. This wavelength dependency could potentially be caused by the difference in the location of injected electrons. Surface electrons are believed to absorb light with shorter wavelength than bulk electrons, and since the amount of ultrafast recombination at these wavelengths is rather substantial and the attenuation factor at these wavelength is smaller, our results may implicate that these electrons are still part of a geminate e-/h+ pair partly located at the dye molecule. The lack of a clear 500 ps time constant associated with the reorganization of the electrons in the TiO2 in our measurements

Figure 5. Distance dependence of the driving force corrected averaged lifetimes of L0-L2 when probing at 940 nm (9) and 1040 nm (b).

Figure 6. Distance dependence of the driving force corrected integrated lifetimes of L0-L2 at 1040 nm in the presence (b) and absence (9) of tBP. A similar change in slope is observed at 940 nm.

could potentially be caused by static attraction between the injected electrons and the dye cations. The Effect of tBP. Adding the conduction band energy raising agent tBP has for all three dyes studied a negative impact on solar cell energy conversion efficiencies and radical state lifetimes alike. The CB raising effect of the tBP decreases the driving force for injection, but no effect on the injection rates was observed with a time resolution of 150 fs. The rate of the following recombination is expected to decrease by the CB shift, as the recombination reaction is in the Marcus inverted region (|∆G0| > λ). As was shown in the transient absorption decay traces of the electrons injected in the TiO2, Figure 4, the amplitude of long-lived signal was decreased for all dyes in the presence of tBP. This indicates that the recombination rate is increased by the additive. As is also shown, the signal amplitude is decreased at 1040 nm but increased at 940 nm for L0 and L2 when adding tBP. From a surface vs bulk electrons interpretation, these results could indicate an increased tendency to inject electrons into reactive surface states in the presence of tBP, which subsequently give rise to the increased recombination rate observed. With a decrease of the charge separated state lifetime combined with an increase in driving force due to tBP addition and a recombination reaction in the Marcus inverted region, the activation energy corrected rate is increased, as shown in Figure 6. Additionally, the distance dependence of the recombination reaction was substantially increased, from 0.49 to 0.63 Å-1, upon addition of tBP. The increase in distance dependence with the addition of tBP could, by the McConnell model of the superexchange mechanism for electron tunneling,30 be explained by an increase in energy barrier for the CR process, as shown in eq 2.

Electron Injection and Recombination Dynamics

β)

∆E 2 ln R0 ν

( )

J. Phys. Chem. B, Vol. 114, No. 45, 2010 14363

(2)

Here, R0 is the bridge unit size, ν is the inter-bridge-unit coupling and ∆E is the tunneling barrier. However, as the TiO2 CB is raised with addition of tBP, the barrier should be lower, resulting in a decrease in β. One must bear in mind, though, that several steps have been taken from the initial data in order to allow a comparison between the recombination rates of the different dyes. Since the recombination is also not within the nonadiabatic limit, the use of the McConnell model to describe differences in β might well be inapt. The presence of the tBP could also induce a more vertical alignment of the dye molecules and thus cause differences in the distance between the triphenyl amine and the TiO2 surface, why the observed difference in distance dependence could partly be affected by uncertainties in the tunneling distance. However, given the differences found in the distance dependence, the effect of adding tBP is plausibly not only to raise the energy level of the TiO2 conduction band but also to induce much more intricate alterations in the electronic communication within the dye-semiconductor system. Conclusions We have studied the charge injection and recombination kinetics of a series of triphenylamine based dyes for solar cell sensitization with different conjugation lengths. Due to the substantial drop in IPCE found with increasing conjugation length in this series of dyes, which was correlated to the driving force of charge injection, we anticipated inefficient electron injection due to a decrease in injection kinetics. However, the injection rate was ultrafast and larger than (200 fs)-1 for all three dyes studied. The following recombination reaction was, however, also fast which will decrease the energy conversion efficiencies for these dyes. The recombination kinetics was multiexponential, and the amount of long-lived radicals was highly dependent on conjugation length with the shortest dye giving the largest fraction of slowly decaying component. With the use of an averaged activation energy corrected decay constant, the recombination reaction was found to be exponentially dependent on electron transfer distance in accordance with a superexchange interaction between the electrons in the CB and the oxidized dye. The distance dependence was also found to be wavelength dependent with an attenuation factor β of 0.38 Å-1 at 940 nm and 0.49 Å-1 at 1040 nm. We suggest that the difference in distance dependence is correlated with inherent differences between the types of electrons which are contributing to the signal at each wavelengthsreactive surface electrons absorb light of shorter wavelength than the less reactive bulk electrons. Effects on the electron transfer kinetics induced by the CB potential raising additive tBP were also studied. No effect on the injection kinetics was found but rather a net decrease of long-lived separated charges. With an increase of the signal amplitude at 940 nm, an interpretation was put forth that the location of the injected electron was affected rather than the injection efficiency by the tBP additive. Also, an increase in the distance dependence for recombination was found when adding tBP, an effect not further investigated in the scope of

this article but which indicates other implications of the additive than merely increasing the CB potential. Acknowledgment. This work was supported by the Knut and Alice Wallenberg Foundation and the Swedish Energy Agency. References and Notes (1) Wenger, B.; Gra¨tzel, M.; Moser, J. E. J. Am. Chem. Soc. 2005, 127, 12150–12151. (2) Tachibana, Y.; Moser, J. E.; Gra¨tzel, M.; Klug, D. R.; Durrant, J. R. J. Phys. Chem. 1996, 100, 20056–20062. (3) Ramakrishna, G.; Verma, S.; Jose, D. A.; Kumar, D. K.; Das, A.; Palit, D. K.; Ghosh, H. N. J. Phys. Chem. B 2006, 110, 9012–9021. (4) O’Regan, B. C.; Durrant, J. R. Acc. Chem. Res. 2009, 42, 1799– 1808. (5) Hagberg, D. P.; Yum, J.-H.; Lee, H.; De Angelis, F.; Marinado, T.; Karlsson, K. M.; Humphry-Baker, R.; Sun, L.; Hagfeldt, A.; Gra¨tzel, M.; Nazeeruddin, M. K. J. Am. Chem. Soc. 2008, 130, 6259–6266. (6) Galoppini, E. Coord. Chem. ReV. 2004, 248, 1283–1297. (7) Kilså, K.; Mayo, E. I.; Kuciauskas, D.; Villahermosa, R.; Lewis, N. S.; Winkler, J. R.; Gray, H. B. J. Phys. Chem. A 2003, 107, 3379–3383. (8) 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, 5225–5233. (9) Myahkostupov, M.; Piotrowiak, P.; Wang, D.; Galoppini, E. J. Phys. Chem. C 2007, 111, 2827–2829. (10) Wang, D.; Mendelsohn, R.; Galoppini, E.; Hoertz, P. G.; Carlisle, R. A.; Meyer, G. J. J. Phys. Chem. B 2004, 108, 16642–16653. (11) Beek, W. J. E.; Janssen, R. A. J. J. Mater. Chem. 2004, 14, 2795– 2800. (12) Chang, C.-W.; Luo, L.; Chou, C.-K.; Lo, C.-F.; Lin, C.-Y.; Hung, C.-S.; Lee, Y.-P.; Diau, E. W.-G. J. Phys. Chem. C 2009, 113, 11524– 11531. (13) Hagberg, D. P.; Edvinsson, T.; Marinado, T.; Boschloo, G.; Hagfeldt, A.; Sun, L. C. Chem. Commun. 2006, 2245–2247. (14) Marinado, T.; Hagberg, D. P.; Hedlund, M.; Edvinsson, T.; Johansson, E. M. J.; Boschloo, G.; Rensmo, H.; Brinck, T.; Sun, L. C.; Hagfeldt, A. Phys. Chem. Chem. Phys. 2009, 11, 133–141. (15) Boschloo, G.; Ha¨ggman, L.; Hagfeldt, A. J. Phys. Chem. B 2006, 110, 13144–13150. (16) Shi, C. W.; Dai, S. Y.; Wang, K. J.; Pan, X.; Kong, F. T.; Hu, L. H. Vib. Spectrosc. 2005, 39, 99–105. (17) Hara, K.; Dan-Oh, Y.; Kasada, C.; Ohga, Y.; Shinpo, A.; Suga, S.; Sayama, K.; Arakawa, H. Langmuir 2004, 20, 4205–4210. (18) Marinado, T.; Hagberg, D. P.; Hedlund, M.; Edvinsson, T.; Johansson, E. M. J.; Boschloo, G.; Rensmo, H.; Brinck, T.; Sun, L.; Hagfeldt, A. Phys. Chem. Chem. Phys. 2009, 11, 133–141. (19) Katoh, R.; Kasuya, M.; Kodate, S.; Furube, A.; Fuke, N.; Koide, N. J. Phys. Chem. C 2009, 113, 20738–20744. (20) Koops, S. E.; O’Regan, B. C.; Barnes, P. R. F.; Durrant, J. R. J. Am. Chem. Soc. 2009, 131, 4808–4818. (21) Wiberg, J.; Guo, L.; Pettersson, K.; Nilsson, D.; Ljungdahl, T.; Mårtensson, J.; Albinsson, B. J. Am. Chem. Soc. 2007, 129, 155–163. (22) Wiberg, J.; Marinado, T.; Hagberg, D. P.; Sun, L. C.; Hagfeldt, A.; Albinsson, B. J. Phys. Chem. C 2009, 113, 3881–3886. (23) Yoshihara, T.; Katoh, R.; Furube, A.; Tamaki, Y.; Murai, M.; Hara, K.; Murata, S.; Arakawa, H.; Tachiya, M. J. Phys. Chem. B 2004, 108, 3817–3823. (24) Zhao, H.; Zhang, Q.; Weng, Y. X. J. Phys. Chem. C 2007, 111, 3762–3769. (25) Berger, T.; Sterrer, M.; Diwald, O.; Kno¨zinger, E.; Panayotov, D.; Thompson, T. L.; Yates, J. T. J. Phys. Chem. B 2005, 109, 6061–6068. (26) Hilgendorff, M.; Sundstro¨m, V. Chem. Phys. Lett. 1998, 287, 709– 713. (27) Tamaki, Y.; Hara, K.; Katoh, R.; Tachiya, M.; Furube, A. J. Phys. Chem. C 2009, 113, 11741–11746. (28) Benko¨, G.; Hilgendorff, M.; Yartsev, A. P.; Sundstro¨m, V. J. Phys. Chem. B 2001, 105, 967–974. (29) Ghosh, H. N.; Asbury, J. B.; Lian, T. J. Phys. Chem. B 1998, 102, 6482–6486. (30) McConnell, H. M. J. Chem. Phys. 1961, 35, 508–515.

JP1002963