Relating the Photodynamics of Squaraine-Based Dye-Sensitized ...

22 downloads 6252 Views 566KB Size Report
Oct 1, 2012 - The effective lifetimes for the charge regeneration reaction, τobs, range from 2 to. 25 μs for the ..... a digital oscilloscope (Agilent Infiniium DS08064A, 600 MHz,. 4 GSa/s). .... These two peaks correspond to the signature of the.
Article pubs.acs.org/JPCC

Relating the Photodynamics of Squaraine-Based Dye-Sensitized Solar Cells to the Molecular Structure of the Sensitizers and to the Presence of Additives G. de Miguel,† M. Marchena,† B. Cohen,† S. S. Pandey,‡ S. Hayase,‡ and A. Douhal*,† †

Departamento de Química Física, Facultad de Ciencias Ambientales y Bioquímica and INAMOL, Universidad de Castilla-La Mancha, Avenida Carlos III, S.N., 45071 Toledo, Spain ‡ Kyushu Institute of Technology, 2-4 Hibikino, Wakamatsu, Kitakyushu, 808-0196, Japan S Supporting Information *

ABSTRACT: Dye-sensitized solar cells (DSSCs) fabricated with TiO2 nanoparticle thin films and sensitized with four types of indole-based squaraines, SQs (symmetric or asymmetric and varying the length and nature of the alkyl side chain substituents), have been prepared. We have studied the influence of the presence of different additives in the electrolyte solutions (tert-butyl pyridine and/or Li+ cations) on the electron transfer dynamics by means of femtosecond transient absorption spectroscopy and flash photolysis. We obtained the rate constants for the electron injection, kei = 2, 3, 8, and 14 × 1010 s−1, for complete solar cells with an iodide-based electrolyte. The asymmetric SQ showed the largest kei value, 14 × 1010 s−1, in line with a unidirectional flow of electrons from the lowest unoccupied molecular orbital (LUMO) orbital of the SQ to the sub-bandgap states of the TiO2, which leads to a more efficient electron injection than that in the symmetric SQs. Addition of tert-butyl pyridine to the electrolyte solution (I−/I3− in acetonitrile) causes a 5−10-fold deceleration of the electron injection (for example, τobs = 2−11 ps in SQ 41). When including the Li+ cation together with the tert-butyl pyridine, the injection is still slower than in cells without any additive (τobs = 2 vs 7 ps in SQ 41), which reflects a stronger influence of the tertbutyl pyridine in the electron injection process. The effective lifetimes for the charge regeneration reaction, τobs, range from 2 to 25 μs for the complete cells with an iodide-based electrolyte. The fastest regeneration occurs in the SQs with the CF3− groups anchored to the side chains and, especially in SQ 26, with two CF3− groups. This result suggests that the inductive effect of the CF3− groups in the structure of SQ 26 and SQ 41 leads to a higher positive charge density in the π-conjugated system, which promotes a higher local concentration of iodide near the oxidized dye and therefore faster regeneration kinetics. Moreover, addition of Li+ cations to the electrolyte accelerates the regeneration reaction, which is ascribed to its interaction with the backbone of the SQ, favoring the approach of the I− species. Using the transient absorption results, we calculated the electron injection efficiency, φei, and compared it with the short-circuit current density, Jsc, of the complete cells. Thus, in the complete cells sensitized with SQ 41 and SQ 4, φei are the highest ones and present comparable values, 0.93 and 0.90, respectively. On the contrary, cells sensitized with SQ 26 and SQ 2 present lower values, 0.47 and 0.75, respectively. A similar tendency is observed for the values of Jsc. On the basis of this good correlation (φei vs Jsc), we can suggest that the electron injection reaction is partially responsible for the photon losses and derive the reasons why this occurs.

1. INTRODUCTION

to 700 nm (porphyrin-based sensitizer) with an almost 100% efficiency in the generation of free electrons.15 To increase the overall efficiency of the cells up to 15%, it is necessary to harvest photons from the near-infrared (NIR) regions.16−18 Thus, the advantage of using SQs as a sensitizer is the extension of the absorption spectrum to the 700−900 nm region. SQs can be employed not only as the sensitizer in the single junction cell but also in a dye cocktail hybrid cell together with another dye covering the 400−600 nm region, where SQs do not absorb.19,20 So far, the record-efficiency cell (6.29%) with

Squaraine dyes (SQs) are condensation products of electronrich substrates (indole derivatives in this work) and squaric acid, which acts as an electron acceptor.1,2 They present a symmetric donor−acceptor−donor (D−A−D) arrangement although the two donors can bear different substituents (D− A−D′), thus resulting in asymmetric SQs with photochemical properties different from those of the corresponding symmetric SQs.3 SQs have been employed in different cutting-edge technologies such as nonlinear optics,4 photodynamic therapy,5 or biolabeling applications.6−8 However, SQs are recently attracting a lot of attention in the field of photovoltaics, dyesensitized solar cells (DSSCs), or organic solar cells.9−14 So far, the record-efficiency DSSCs (12%) absorb UV−visible light up © 2012 American Chemical Society

Received: June 21, 2012 Revised: September 21, 2012 Published: October 1, 2012 22157

dx.doi.org/10.1021/jp306125y | J. Phys. Chem. C 2012, 116, 22157−22168

The Journal of Physical Chemistry C

Article

Figure 1. Molecular structure of the four studied SQs.

symmetric or asymmetric due to the different effect on aggregate formation.33 Steric hindrance of the lateral chains can help to suppress the formation of aggregates, which is a key point to achieve high efficiencies in SQ-based cells.34 Moreover, the nature of the side groups can also modify the energy levels of the SQs (HOMO and LUMO orbitals), affecting the electron transfer reactions. Finally, asymmetric SQs are capable of generating larger photocurrents than the symmetric analogues due to a unidirectional flow of electrons to the TiO2 surface.3,35,36 In this study, we determined the electron injection rate constant, kei, for complete cells with TiO2 nanoparticles sensitized with a family of indole-based SQs (Figure 1) using femtosecond transient absorption spectroscopy. To obtain the electron injection efficiency, φei, we also carried out the same experiments using ZrO2 as the semiconductor since the higher position of the conduction band edge compared to that in TiO2 hinders the electron injection reaction. A flash photolysis technique was employed to study the dynamics of the recombination, krec, and regeneration, kreg, processes. The values of kei and kreg were determined for complete cells containing the I−/I3− redox electrolyte and both additives, tertbutyl pyridine and Li+ cations. We discuss the influence of these additives on the dynamics of the charge transfer reactions. We obtained a good correlation between the electron injection efficiency of the complete cells, φei, with the short-circuit current density (Jsc) obtained from the characteristic current− voltage (I−V) curves. We believe that this detailed study sheds light on the limitations of SQs as sensitizers and will help to design new SQ molecules with optimized properties for highefficiency DSSCs.

only SQs as the sensitizer has been obtained using asymmetric mixed pyrrole/indole-based SQs.21 In red-absorbing dyes as SQs, it is crucial to carefully design the molecular energy levels to match its LUMO energy with the conduction band of the semiconductor and its HOMO energy with the redox potential of the electrolyte.22 Hence, the photophysical characterization of the electron transfer reactions is essential to gain insight into the relationship between the overall efficiency of the cell and the dynamics of the photogenerated charges.23,24 Kamat et al. carried out the first studies of DSSCs sensitized with SQs to elucidate the ultrafast (picosecond time scale) charge transfer events associated with photosensitization with those dyes.25 Many charge transfer processes are involved in DSSCs, and each step results in an increased spatial separation of electrons and holes, increasing the lifetime of the charge-separated state, but at the expense of reducing the free energy stored in that state. Electron injection (from the dyes to the TiO2 conduction band) and regeneration processes (from the redox electrolyte to the oxidized dyes) are usually investigated by transient absorption spectroscopies.26,27 One of the most interesting issues in the operation of the DSSCs is the influence of different additives on the performance of the device.28 Their effect is understood at a phenomenological level, and it is attributed to the band shift of the semiconducting materials, interaction with the dye or redox electrolyte, changes in the organization of the dye, and/or different coverage of the TiO2 surface.16 Among the different additives, tert-butyl pyridine and Li+ cations are the most employed species, and their effect on the photodynamics of the cells has been widely studied for a broad range of dye molecules.29−32 However, to the best of our knowledge, the influence of these additives on the electron injection and charge regeneration in cells sensitized with SQs has not yet been reported. Another important parameter that influences the efficiency of the device is the type and length of the lateral chains that are attached to the central backbone and, in particular, if the SQ is

2. EXPERIMENTAL PART The four studied SQ molecules (Figure 1) were synthesized as previously described.33,37,38 We have used ((1-butyl-3,3dimethylindolin-2-ylidene)methyl)-4-((5-carboxy-3,3-dimethyl1-(4,4,4-trifluorobutyl)-3H-indol-1-ium-2-yl)methylene)-3-oxo22158

dx.doi.org/10.1021/jp306125y | J. Phys. Chem. C 2012, 116, 22157−22168

The Journal of Physical Chemistry C

Article

pulse·cm2, and the average intensity is 4.1 × 1017 photons/ cm2·s. The remaining fundamental beam goes through a delay line and is focused on a rotating 3 mm thick BaF2 plate to generate the white light continuum. The produced white light is split into two parts to form probe and reference beams, which are directed to the cell, where the probe and the pump beams are overlapped. The cell is placed in a holder connected to a pair of translation stages (MTS series, Thorlabs). They move the cell in the x−y plane perpendicular to the pump beam, preventing it from photodegradation during the measurements. The polarization of the pump is set to magic angle with respect to the probe. The transmitted light is focused to light guides, directed to a spectrograph, and collected by a pair of photodiode arrays. The obtained signal, as a difference in the optical density of the sample probed by the white light continuum with and without pump excitation, is processed by the ExciPRO software (CDP Systems). The zero time of all the analyzed spectra was corrected for the group velocity dispersion effect, according to the standard numerical scheme.40 The chirp of the white light continuum was obtained by measuring twophoton absorption (TPA) in a very thin (150 μm) BK7 glass plate. The UV−visible nanosecond−second flash photolysis setup consists of an LKS.60 laser flash photolysis spectrometer (Applied Photophysics), Vibrant (HE) 355 II laser (Opotek) as a pump pulse source (5 ns time duration), and a 150 W xenon arc lamp as a probe. The signal from OPO (355 nm pumped by Q-switched Nd:YAG laser, Brilliant, Quantel) at 650 nm was used for the sample excitation. The probing light transmitted through the sample (held by tweezers) was dispersed by a monochromator and detected by a photomultiplier coupled to a digital oscilloscope (Agilent Infiniium DS08064A, 600 MHz, 4 GSa/s). The pump energy pulse (5 mJ/pulse) was attenuated by the pair of half-waveplate and a polarizer. The kinetics were recorded in the spectral range of 500−750 nm at 10 nm intervals, and the time-resolved absorption spectra were constructed from the kinetic traces. The quality of the fits was checked by examining the residual distribution and the χ2 value. The pump pulse energy density and pulse power density were 40−80 μJ/cm2 and 0.4−0.8 mW/cm2, respectively. All the experiments were done at 293 K.

cyclobut-1-enolate, SQ 41; ((5-carboxy-3,3-dimethyl-1-(4,4,4trifluorobutyl)-3H-indol-1-ium-2-yl)methylene)-2-((5-carboxy3,3-dimethyl-1-(4,4,4 trifluorobutyl)indolin-2-ylidene)methyl)3-oxocyclobut-1-enolate, SQ 26; ((5-carboxy-3,3-dimethyl-1octyl-3H-indol-1-ium-2-yl)methylene)-2-((5-carboxy-3,3-dimethyl-1-octylindolin-2-ylidene)methyl)-3-oxocyclobut-1-enolate, SQ 4; ((5-carboxy-1-ethyl-3,3-dimethyl-3H-indol-1-ium-2yl)methylene)-2-((5-carboxy-1-ethyl-3,3-dimethylindolin-2ylidene)methyl)-3-oxocyclobut-1-enolate, SQ 2. To fabricate the TiO2 electrodes, Ti-Nanoxide HT paste (Solaronix SA) was employed. First, a conductive FTO glass (Pilkinton NSG TEC 8A 2 × 3 cm2) was cleaned with acetone, then with water, and finally with a detergent with ion-exchanged water using an ultrasonic bath during 10 min each. Afterward, the FTO glass was washed again with water and finally immersed in isopropanol and sonicated for 30 min. After drying the FTO glass, a layer of the TiO2 paste was coated using the doctor blade technique with the help of two parallel adhesive Scotch tapes. The substrate was then baked at 450 °C in a furnace to fabricate TiO2 layers of about 5 μm thickness. The substrate was dipped in an ethanol solution (Scharlau SA, extra pure) containing the SQ dye in the presence of chenodeoxycholic acid (CDCA) coadsorber. The SQ dye concentration was fixed at 0.25 mM, while the CDCA concentration was 2.5 mM. The optical density at the maximum of the absorption peak, ∼660 nm, was around 2 for all the studied cells. No major photodegradation was detected during the measurements. The platinized counter electrode was obtained by spreading a Pt-based solution (Platisol T, Solaronix) on ITO glass, followed by heating at 450 °C for 10 min. To make the complete solar cell, the counter electrode was then placed directly on top of the dye-adsorbed TiO2 nanoparticle electrode and sealed with thermal adhesive films (25 μm Surlyn, Meltronix, Solaronix) that also act as a separator. The different electrolytes were introduced in the complete cells with an analytical syringe through two holes drilled in the counter electrodes, which were later sealed by a piece of Surlyn and a microscope coverslip. The Electrolyte I was obtained from Solaronix (Iodolyte AN50), and it contains the I−/I3− redox couple ([I2] = 50 mM), an ionic liquid, a lithium salt, and a pyridine derivative in acetonitrile (ACN) solution. Electrolyte II is a homemade solution containing the I−/I3− redox couple ([I2] = 50 mM and [NaI] = 0.5 M) and tert-butyl pyridine (580 mM) in ACN solution. Electrolyte III is also a homemade solution containing only the I−/I3− redox couple ([I2] = 50 mM and [NaI] = 0.5 M) in ACN solution. Iodine (I2) and NaI were purchased from Panreac, and ACN, CDCA, and tert-butyl pyridine were obtained from Sigma-Aldrich and utilized without further purification. UV−visible steady-state absorption spectra were measured with a JASCO V-670 spectrophotometer. Femtosecond (fs) transient UV−visible absorption spectra were measured using a two-channel detection system described previously.39 It consists of a Ti:sapphire oscillator (TISSA 50, CDP Systems) pumped by a 5 W diode laser (Verdi 5, Coherent). The seed pulse (30 fs, 450 mW at 86 MHz) centered at 800 nm wavelength is directed to an amplifier (Legend-USP, Coherent). The amplified fundamental beam (50 fs, 1 W at 1 kHz) is split into two beams directed to a femtosecond-transient absorption spectrometer (CDP Systems). The first one goes through an optical parametric amplifier (CDP Systems), and the idler is then frequency-doubled to give 630−670 nm for exciting the sample. The pump pulse intensity is kept constant at 125 μJ/

3. RESULTS AND DISCUSSION 3.1. Steady-State Absorption Spectroscopy. Figure 2 shows the normalized UV−visible absorption spectra of the complete solar cells containing electrolyte I and sensitized with the different SQs, showing maximum intensity at 647, 660, 666, and 663 nm for SQ 41, SQ 26, SQ 4, and SQ 2, respectively. The shoulders at the blue side (600−620 nm) were previously assigned to high-energy vibrational transitions.41 The addition of CDCA strongly reduces the formation of aggregates in the sample as evidenced by the spectra, narrow with no extra absorption bands or shoulders with regard to the spectra in pure solution.42 Thus, the deaggregating effect of the CDCA is based on the prevention of the π−π interactions originating the aggregates.34 Figure S1A (Supporting Information) exhibits the normalized absorption spectra of complete cells with TiO2 (dotted line) and ZrO2 (dashed line) nanoparticles containing electrolyte I and sensitized with SQ 26. Both spectra exhibit the same shape but with a small 4 nm red-shift of the wavelength of the maximum absorption intensity (λmax) of the cells having nanoparticles of ZrO2. Similar shifts of the λmax were found in the other studied SQs. The similar shape and λmax of the 22159

dx.doi.org/10.1021/jp306125y | J. Phys. Chem. C 2012, 116, 22157−22168

The Journal of Physical Chemistry C

Article

absorption spectroscopy.26,27 Here, we have performed such kinds of experiments on complete cells sensitized with the above SQs, with TiO2 NP and in the absence or presence of electrolyte I−III.43 The excitation wavelength was fixed close to the maximum of absorption intensity, 650 nm, and the optical density at that wavelength was around 2. We used a low excitation energy (125 μJ·pulse−1·cm−2) which produces a mean excitation of only 5% of the dyes. In additional experiments carried out with higher excitation energy (250 or 500 μJ·pulse−1·cm−2), we did not observe changes in the decay dynamics, and therefore we rule out the possibility of other power-dependent deactivation channels, like the singlet−singlet exciton annihilation.34 SQ*(S1) + TiO2 → SQ•+ + TiO2 (e−)

Figure 2. Normalized steady-state UV−visible absorption spectra of complete cells prepared with TiO2 nanoparticles, sensitized with SQ 41, SQ 26, SQ 4, and SQ 2 and containing electrolyte I.

(1)

Figures 3A and 4A show the transient absorption spectra at five fixed delays between 0 and 42 ps of complete cells containing electrolyte I and sensitized with SQ 41 and SQ 26, respectively. The transient spectra at 1 ps exhibit positive peaks around 500−530 nm and a very weak signal at 775 nm. These transient features were previously assigned to the singlet excited state of the monomers, S1.19,41 There are also two negative bands in the near-IR region around 650−660 and 710 nm. The strongest one matches in shape and position to the steady-state visible absorption band and therefore is assigned to the ground state bleach of the SQs. The weak band at longer wavelengths is attributed to the stimulated emission from the singlet excited state of the SQs since the fluorescence emission appears in the same region. Evolution of all previous signals at longer delay times gives rise to an asymmetric band peaking at 580 nm for both SQ 41 and SQ 26, together with a broad signal around 750 nm. These two peaks correspond to the signature of the radical cation of the SQs, SQ•+.25,34,35 The direct determination of the rate constant for the electron injection, kei, by measuring the rise of the transient feature of the SQ radical cation at 580 nm is not possible due to the strong overlap with the transient signal of the singlet excited state, S1, and ground state, S0. Therefore, we have analyzed the transient signals at 500 nm to evaluate the relaxation dynamics of the S1 state in these samples, which is related to the electron injection process. The transient absorption decays were fitted to stretched exponential functions (eq 2), which accounts for

absorption spectra indicates that no aggregates are formed in the samples with ZrO2 nanoparticles. Moreover, a different anchoring mode with the surface of the ZrO2 NP could influence the energy of the HOMO and LUMO orbitals of the dye. However, the similar λmax for the absorption peak suggests that the binding properties of the SQs in both types of nanoparticles (TiO2 and ZrO2) are almost identical. The steady-state absorption spectrum of SQ 26 in ACN solution (solid line) is also shown for comparison. We observed a 12 and 9 nm blue-shift of the λmax compared to that of the complete cells with TiO2 NP and sensitized with SQ 26 and SQ 41, respectively. The latter has been previously ascribed to the different environment around the SQ molecules when adsorbed onto the surface of the semiconductor nanoparticles.34 In the complete cells prepared with ZrO2 nanoparticles, we also recorded the steady-state emission spectra (Figure S1B, Supporting Information). We observed a 23 nm red-shift of the λmax with regard to that in ACN solution. This shift in the emission spectra is also attributed to the different environment around the SQ molecules onto the surface of the ZrO2 nanoparticles compared to that in solution, which in turn affects the LUMO orbital of the SQs.42 3.2. Femtosecond Transient Absorption Studies. The most widely used experimental technique to study the electron injection process (eq 1) is the femtosecond (fs) pump−probe

Figure 3. (A) Femtosecond transient absorption spectra (in terms of change in absorbance, ΔA) of a complete cell prepared with TiO2 nanoparticles, sensitized with SQ 41 and containing electrolyte I at five time delays. The inset shows a zoom of the red part of the spectra. (B) Normalized decays of the transient signals (ΔA) at 500 nm of complete cells prepared with TiO2 nanoparticles, sensitized with SQ 41 and containing electrolytes I−III and without electrolyte. The inset shows the initial 5 ps time window. Solid lines are from the best stretched exponential fits of the experimental data. λexc = 650 nm. 22160

dx.doi.org/10.1021/jp306125y | J. Phys. Chem. C 2012, 116, 22157−22168

The Journal of Physical Chemistry C

Article

Figure 4. (A) Femtosecond transient absorption spectra (in terms of change in absorbance, ΔA) of a complete cell prepared with TiO2 nanoparticles, sensitized with SQ 26 and containing electrolyte I at five time delays. (B) Normalized decays of the transient signals (ΔA) at 500 nm of complete cells prepared with TiO2 nanoparticles, sensitized with SQ 26 and containing electrolytes I−III and without electrolyte. The inset shows the initial 5 ps time window. Solid lines are from the best stretched exponential fits of the experimental data. λexc = 650 nm.

Table 1. Effective Lifetimes, τobs, Derived from a Stretched Exponential Fit of the Decay of the Femtosecond Transient Signals (ΔA) at 500 nm of Complete Cells Prepared with TiO2 Nanoparticles, Sensitized with SQ 41, SQ 26, SQ 4, and SQ 2 with Electrolytes I−III and without Electrolytea

decays that occur over a long time scale (Table S1, Supporting Information). f (x) = A exp[( −t /τ )β ]

(2)

where τ is the characteristic mean time of the decays and β is the heterogeneity (dispersion) of the sample describing the Gaussian distribution of unoccupied states in the conduction band, with β values typically between 0.3 and 0.65.29,44 The origin of the multiexponential dynamics in the electron injection process is explained on the basis of an inhomogeneous distribution of the injection energetics.45 However, it is not correct to compare τ values among decays with different β parameters since the former influences strongly the shape of the decay. Thus, to obtain a weighted average effective lifetime, τobs, and from its inverse the “effective rate constant”, kobs (i.e., kobs = 1/τobs), we used eqs 3 and 4, where u is the variable of integration.46 τobs =

τ ⎛1⎞ Γ⎜ ⎟ β ⎝β⎠

⎛1⎞ Γ⎜ ⎟ = ⎝β⎠

∫0



τobs/ps

a

electrolyte

SQ 41

SQ 26

SQ 4

SQ 2

I II III no electrolyte

7 11 2 2

17 19 2 2

11 6 1 1

40 20 4 2

λexc = 650 nm.

effective lifetimes are not largely affected by the presence of the redox couple. They are virtually the same in SQ 4, SQ 41, or SQ 26, while in SQ 2, we find a slightly longer lifetime with electrolyte III. The former indicates a weak effect of the I−/I3− couple on the TiO2 conduction band potential or the energetics of the SQ dye. Similar results have been previously reported in ruthenium dyes, in which the redox couple did not affect neither the excited state dynamics of ruthenium-based dyes (N719) nor the TiO2 electron density in the dark.29 It can be concluded that the presence of the redox couple does not significantly influence the electron injection dynamics, and therefore, the changes in the dynamics of the complete cells with electrolytes I and II can be safely ascribed to the presence of the two additives, tert-butyl pyridine and Li+ cations. Next, addition of tert-butyl pyridine to the redox couple (electrolyte II) gives rise to notable changes in the singlet lifetimes. In that case, the electron injection is retarded 5−10fold with regard to the electrolyte III (for example, from 2 to 11 ps in cells sensitized with SQ 41). The adsorption of the tertbutyl pyridine onto the surface of the TiO2 nanoparticles entails a negative shift of the conduction band potential, which is ascribed to its Lewis base character by direct coordination of the lone pairs on the N atoms to the surface of the TiO2 NP.29,30 The relationship between the electron injection rate constant, kei, and the driving force for the reaction, ΔG0ei, in this kind of device can be expressed by using eq 547,48

(3)

u1/ β − 1e−udu (4)

Figures 3B and 4B depict the transient absorption decays at 500 nm of the complete cells, with and without electrolyte I− III and sensitized with SQ 41 and SQ 26, respectively. It is worth noting that the contribution of the transient signal from SQ•+ at this wavelength is not significant (10−15%), and therefore the signal can be considered to come mostly from the singlet excited state of the SQs.19,41 However, we have utilized an offset in the fits to account for this long-lived signal. The solid lines represent the best fits using a stretched exponential function. The results show the influence of the electrolyte additives, tert-butyl pyridine and Li+, on the electron injection dynamics (Table 1). First, we consider the potential influence of solely the redox electrolyte (I−/I3− in ACN) on the deactivation dynamics of the SQs (electrolyte III vs no electrolyte). Several processes can be responsible for the change in the SQ dynamics, like for example the solvation effect, which decreases the effective charge on the TiO2 surface by polarization and leads to an upward shift of the conduction band. Another such process is the electron transfer from I− to the trap states of the TiO2 nanoparticles that might modify the ion distribution around them. However, it is observed that the

kei =



⎛ −(ΔG 0 + λ)2 ⎞ ei ⎟d E V 2[1 − f (E , E F)]g (E)exp⎜ 4λkBT ⎝ ⎠ (5)

22161

dx.doi.org/10.1021/jp306125y | J. Phys. Chem. C 2012, 116, 22157−22168

The Journal of Physical Chemistry C

Article

Figure 5. (A) Femtosecond transient absorption spectra (in terms of change in absorbance, ΔA) of a complete cell prepared with ZrO2 nanoparticles, sensitized with SQ 41 and containing electrolyte I at five time delays. (B) Normalized decays of the transient signals (ΔA) at 500 nm of complete cells prepared with ZrO2 nanoparticles (nonfull squares) or with TiO2 nanoparticles (full circles), sensitized with SQ 41 and containing electrolyte I. Solid lines are from the best stretched exponential fits of the experimental data. λexc = 650 nm.

Figure 6. (A:) Femtosecond transient absorption spectra (in terms of change in absorbance, ΔA) of a complete cell prepared with ZrO2 nanoparticles, sensitized with SQ 26, and containing electrolyte I at five time delays. (B) Normalized decays of the transient signals (ΔA) at 500 nm of complete cells prepared with ZrO2 nanoparticles (nonfull squares) or with TiO2 nanoparticles (full circles), sensitized with SQ 26 and containing electrolyte I. Solid lines are from the best stretched exponential fits of the experimental data. λexc = 650 nm.

where kei is the electron injection rate constant; V is the electronic coupling between the LUMO orbital of the dye and the different states in the conduction band of the TiO2 NP; ΔG0ei is the driving force for the electron injection; and λ is the reorganization energy. g(E) is the normalized density of states in the TiO2 NP, and f(E, EF) is the Fermi occupancy factor. Thus, according to eq 5, the decrease of ΔG0ei due to the upward shift of the conduction band potential gives rise to a deceleration of the electron injection reaction since this process normally takes place in the normal region of the Marcus parabola (ΔG0ei < λ).29,30 However, the lower density of energetically accessible acceptor states in the TiO2 NP, g(E), plays an important role because it also slows down the interfacial electron injection as seen in eq 5. It is accepted that the density of states in the sub-bandgap of the TiO2 NP increases exponentially with energy, g(E) ≈ exp(−E/E0), where E0 is a parameter with a typical value of 100 meV.47 Thus, assuming that the injection takes place to the sub-bandgap states then, the electron dynamics must be proportional to the number of unoccupied states, and therefore τobs should scale linearly with g(E).47 The relative energies of the sub-bandgap states in the TiO2 NP with and without the addition of tertbutyl pyridine have been reported to vary between 150 and 300 mV.29,30,49 The latter is expected to result in a 5−20-fold deceleration of the electron injection according to the exponential expression of g(E). We find a good agreement with the experimental ratio of the effective lifetimes between the samples with electrolyte II and III (5−10-fold). The relative

differences among the four SQ dyes could be ascribed to the ease of the tert-butyl pyridine to be adsorbed onto the surface of the nanoparticle according to the molecular structure of each SQ. Finally, in the cells containing both additives (electrolyte I), there is no clear tendency in their lifetimes with regard to those with only tert-butyl pyridine (electrolyte II). With SQ 41 or SQ 26, the lifetimes are slightly shorter, but with SQ 4 and SQ 2, we observed longer ones. However, the lifetimes of the cells containing electrolyte III (with only I−/I3− in ACN) are always shorter than those of the cells containing electrolyte I (for example, 2 and 17 ps in cells sensitized with SQ 26). It is wellknown that the small Li+ cations not only adsorb onto the surface of the TiO2 NP but also penetrate into the TiO2 lattice, and thus, the potential of the conduction band results in a positive shift (Li+ is a strong electropositive cation).28 The consequence of the latter is the acceleration of the electron injection since the ΔG0ei and g(E) are larger, in contrast to the effect of tert-butyl pyridine. Thus, these two effects should compete in the electrolyte I, although the net effect compared to that without additives (electrolyte III) is a clear retardation of the electron injection, and therefore we deduced a stronger influence of the tert-butyl pyridine. A similar behavior has been previously reported in ruthenium-based sensitized TiO2 solar cells, in which addition of Li+ to an electrolyte containing tertbutyl pyridine did not affect the injection dynamics.50 The interpretation for the previous results was the appearance of a cooperative effect between Li+ and tert-butyl pyridine, which 22162

dx.doi.org/10.1021/jp306125y | J. Phys. Chem. C 2012, 116, 22157−22168

The Journal of Physical Chemistry C

Article

Table 2. Effective Lifetimes, τobs, Derived from a Stretched Exponential Fit of the Decay of the Femtosecond Transient Signals at 500 nm of Complete Cells Prepared with TiO2 or ZrO2 Nanoparticles, Sensitized with SQ 41, SQ 26, SQ 4, and SQ 2 and Containing Electrolyte Ia SQ SQ SQ SQ

41 26 4 2

τobs(TiO2)/ps

τobs(ZrO2)/ps

10−10kei/s−1

φei

LUMO energy/eV

ΔG0ei/eV

Jsc (mA/cm2)

7 17 11 40

100 32 118 160

14 3 8 2

0.93 0.47 0.90 0.75

−4.45 −4.15 −4.75 −4.50

−0.45 −0.15 −0.75 −0.50

7.50 6.63 7.26 6.36

Electron injection rate constants, kei, were calculated using eq 6 Electron injection efficiencies, φei, were calculated using eq 7. LUMO energies, electron injection driving forces, ΔG0ei, and short-circuit current density, Jsc, were obtained from references 33, 37, and 38.

a

Figure 7. Normalized decays of the femtosecond transient signals (in terms of change in absorbance, ΔA) at 500 nm of complete cells prepared with (A) TiO2 and (B) ZrO2 nanoparticles, sensitized with SQ 41 (nonfull diamonds), SQ 26 (full triangles), SQ 4 (full circles), and SQ 2 (nonfull squares) and containing electrolyte I. Solid lines are from the best stretched exponential fits of the experimental data. λexc = 650 nm.

produced a higher concentration of tert-butyl pyridine molecules at the surface of the TiO2 NP when Li+ was added. Thus, the formation of intermolecular interactions has been suggested between Li+ and tert-butyl pyridine when both additives are included in the electrolyte. This would decrease the concentration of Li+ nearby the TiO2 NP due to the formation of a complex with the tert-butyl pyridine. As a consequence, the higher concentration of Li+ in the solution is suggested to produce the increase in the concentration of tertbutyl pyridine adsorbed on the TiO2 surface.50 To determine the electron injection rate constant, kei, it is necessary to measure the singlet lifetime of the SQ molecules when absorbed onto nanoparticles in the absence of electron injection. Thus, noninjecting cells were fabricated using sensitized ZrO2 nanocrystalline films. ZrO2 nanoparticles exhibit dye binding properties similar to that of TiO2, but they have a conduction band edge potential about ∼1 V more negative than that of TiO2, preventing electron injection from the excited state of the SQs.51 This measurement has been proven to be very valuable in determining the electron injection efficiency.44,52 Figures 5A and 6A illustrate the femtosecond transient absorption spectra of complete cells prepared with ZrO2 NPs containing electrolyte I and sensitized with SQ 41 and SQ 26, respectively. The initial transient spectra at 1 ps are similar to those obtained for the TiO2 NP films, asymmetric peak at 530 nm and a shoulder at 480 nm for SQ 41 (Figure 5A) and symmetric peak at 505 nm for SQ 26 (Figure 6A). These transient features were assigned to the singlet excited state of the SQs. However, in the cells with the ZrO2 NP as semiconductor, the shape of the transient spectra does not change at longer delay times, and no signal remains after the 2 ns time window. Therefore, the SQ•+ species is not formed in

the cells with the ZrO2 NP, confirming that the electron injection does not occur in the samples with the ZrO2 NP. The stretched exponential function was again employed to fit the transient decays in the complete cells with the ZrO2 NP, and the results of the fits are collected in Table 2 (τobs). It has been suggested that the absence of tert-butyl pyridine or Li+ cations in electrolyte II or III should not affect largely the decay process.29,30 Figures 5B and 6B show a comparison of the transient decays at 530 or 500 nm (maximum of the signal) in the complete cells with TiO2 or ZrO2 sensitized with SQ 41 and SQ 26, respectively. Clearly, the singlet excited state lifetimes (τobs) are always longer for the cells with ZrO2 NPs, for example, from 7 to 100 ps and from 17 to 32 ps in cells with SQ 41 and SQ 26, respectively. However, the lifetimes of the singlet excited state of the studied SQs in ACN solution are much longer than those obtained in the ZrO2 cells.41 This shortening in the lifetimes originates due to the different environment of the SQs at the surface of the ZrO 2 nanoparticles.41 Energy transfer quenching among the SQs has been proposed to account for the observed decrease in the lifetimes.53 Moreover, a minor contribution of electron transfer from the excited state of the SQ to trap states of the ZrO2 nanoparticles can also accelerate the decay of the singlet, despite the absence of the transient signal of the radical cation.54 It is probably for these reasons that the decay in ZrO2 cells follows a stretched exponential behavior, and not a singleexponential one. Figures 7A and 7B exhibit the normalized decays of the transient signals at 500 nm of the cells containing electrolyte I, sensitized with the different SQs and prepared with TiO2 and ZrO2 nanoparticles, respectively. It is worth pointing out that similar singlet dynamics were found for all the studied cells prepared with ZrO2 except for the one sensitized with SQ 26 22163

dx.doi.org/10.1021/jp306125y | J. Phys. Chem. C 2012, 116, 22157−22168

The Journal of Physical Chemistry C

Article

Scheme 1. (A) Schematic Representation of the Effects of the tBuPy (tert-Butyl Pyridine) Additive and of the Molecular Structure of the Studied Squaraines on the Electron Injection (EI) Dynamics in the DSSCa and (B) Schematic Representation of the Effects of Li+ Additive and of the Molecular Structure on the Dye Regeneration (DR) Dynamics in the DSSCb

a

The addition of tBuPy causes a deceleration of the electron injection. The asymmetric squaraine, SQ 41, shows faster electron injection than the symmetric ones, SQ 2, SQ 4, and SQ 26.

The presence of two CF3− groups in the structure of SQ 26 (Figure 1) leads to a faster regeneration process than the observed with SQ 2, SQ 4, and SQ 41. The presence of Li+ additive in the electrolyte solution speeds up the regeneration process. CB indicates the conduction band of the TiO2 NP. The molecular structure of the squaraine dyes (SQ 2, SQ 4, SQ 26, and SQ 41) is given in Figure 1.

b

V, between the LUMO orbital of the symmetric SQs and the conduction band of the TiO2 is also a key factor. We can now calculate the electron injection efficiency (φei) for the complete cells containing electrolyte I with eq 7

(Table 2). The latter suggests that some specific interactions take place between the two CF3− (the distinctive feature of the SQ 26) and the ZrO2 surface, making the deactivation of the singlet faster than in the other SQs. On the basis of the previous transient absorption decays, we calculated the electron injection rate constants, kei, by using eq 6 (Table 2) kei =

1 τTiO2



φei =

1 τZrO2

(6)

kei kei +

1 τZrO2

(7)

We obtained values of 0.93, 0.47, 0.87, and 0.75 for the complete cells sensitized with SQ 41, SQ 26, SQ 4, and SQ 2, respectively (Table 2). In general, the electron injection efficiency runs parallel to the value of kei: the larger the kei the greater the φei. However, this is not true for the cell sensitized with SQ 26, where the short singlet lifetime in the cell with the ZrO2 NP is the key factor for the low electron injection efficiency. Thus, the obtained values of φei can give us insight into the measured short circuit current density, Jsc, of each SQ. Table 2 shows the Jsc for the complete cells of the four studied SQs containing electrolyte I. It can be seen that in the cells sensitized with SQ 41 and SQ 4 Jsc is the highest one and with similar values, 7.50 and 7.26 mA/cm2, respectively, while the cells sensitized with SQ 26 and SQ 2 present lower values (6.63 and 6.36 mA/cm2, respectively). Upon comparison of these values with those of the electron injection efficiency, φei, we observed a similar qualitative tendency, which supports the importance of the electron injection to account for the measured short circuit photocurrent. Summarizing this part (Scheme 1), the asymmetric SQ presents the largest kei value (14 × 1010 s−1), in line with an unidirectional flow of electrons from the LUMO orbital of the SQ to the conduction band of the TiO2. On one hand, the presence of tert-butyl pyridine in the electrolyte II slows down the electron injection 5−10 times with respect to the electrolyte without additives (electrolyte III). On the other hand, in the cells containing electrolyte I, with both Li+ and tert-butyl pyridine, the electron injection is always slower than in the cells without additives. The former reflects a stronger influence of

where τTiO2 and τZrO2 are the effective lifetimes (τobs) for the cells with TiO2 and ZrO2 nanoparticles, respectively. We obtained kei values of 14, 3, 8, and 2 × 1010 s−1 for the complete cells containing electrolyte I and sensitized with SQ 41, SQ 26, SQ 4, and SQ 2, respectively. The driving force of the electron injection reaction, ΔG0ei, is also shown in Table 2.33,37,38 The correlation between kei and ΔG0ei is not perfect, and therefore we assume that other factors may control the electron injection process. In particular, kei of the complete cell sensitized with the asymmetric SQ (SQ 41) is the largest one, despite that ΔG0ei is not the greatest one (−0.45 eV). It is widely accepted that in asymmetric dyes there is an unidirectional flow of electrons to the conduction band of the TiO2, which leads to more efficient electron injection as compared to that in symmetric dyes.3,35,36 The next SQ having a high value of kei is SQ 4, 8 × 1010 s−1, which is easily explained by the large ΔG0ei, −0.75 eV. The opposite behavior is found for SQ 26, with a relatively low value of kei, 3 × 1010 s−1, which reflects the small ΔG0ei, −0.15 eV, as expected in the normal region of the Marcus theory. The slowest electron injection corresponds to the cell sensitized with SQ 2, kei = 2 × 1010 s−1, whose value of ΔG0ei, −0.50 eV, is relatively high. It is worth noting that the molecular structure of SQ 2 has the shortest alkyl chain substituent (ethyl), which might influence the coupling with the TiO2 nanoparticles. According to the previous results, we conclude that a high value of ΔG0ei is not sufficient to ensure a fast electron injection process in SQ sensitizers, and therefore the electronic coupling, 22164

dx.doi.org/10.1021/jp306125y | J. Phys. Chem. C 2012, 116, 22157−22168

The Journal of Physical Chemistry C

Article

the tert-butyl pyridine on the electron injection due to a cooperative effect between both additives. The complete cells sensitized with SQ 26 show the lowest overall efficiency, which is attributed to the poor electron injection efficiency. The fast deactivation of the singlet excited state in the samples prepared with ZrO2 determines the low value for the injection efficiency 3.3. Nanosecond Flash Photolysis Measurements. We utilized the flash photolysis technique to study the charge recombination and regeneration dynamics in each sample, exciting at 650 nm. Figure 8 shows three representative

41 containing electrolytes I−III and without electrolyte. In the absence of the redox electrolyte, the ground state of the SQ recovers with the electrons located in the conduction band and the trap states of the TiO2 nanoparticles (recombination reaction, eq 8). To fit the microsecond−millisecond transient data, we have also utilized stretched exponential functions (eq 2 and Table S2, Supporting Information) and the corresponding eqs 3 and 4 to obtain the effective lifetimes, τobs, which can then be compared among all the studied SQs and with the different additives (Table 3). Again, the stretched exponential function Table 3. Effective Lifetimes, τobs, Derived from a Stretched Exponential Fit of the Decay of the Flash Photolysis Signals at 550 nm of Complete Cells Prepared with TiO2 Nanoparticles, Sensitized with SQ 41, SQ 26, SQ 4, and SQ 2 with Electrolytes I−III and without Electrolytea τobs/μs electrolyte SQ SQ SQ SQ a

Figure 8. Nanosecond transient absorption spectra of a complete cell prepared with TiO2 nanoparticles, sensitized with SQ 41 and containing electrolyte I at a 0.1, 1.5, and 5 μs time delay after the laser pulse. λexc = 650 nm.

41 26 4 2

I

II

III

no electrolyte

kreg·10−4/s−1

10 2 20 25

47 12 34 31

13 5 10 13

115 206 188 97

9.1 49.5 4.5 3.0

Regeneration rate constants, kreg, were calculated using eq 10.

accounts for a distribution of lifetimes due to the heterogeneity of the sample. Effective lifetimes of 115, 206, 188, and 97 μs were obtained for TiO2 nanoparticle thin films without redox electrolyte, sensitized with SQ 41, SQ 26, SQ 4, and SQ 2, respectively. These values for the charge recombination are in the same order of magnitude as those obtained for other organic molecules (∼100−500 μs).16 Effective rate constants for the recombination, krec, are calculated as the inverse of the τobs. However, it is important to remark that a recombination reaction (eq 8) is quite sensitive to the electron concentration in the TiO2 NP. Thus, in the cells under operation, the electron concentration reaches a constant value. In our studied systems, the e− density decays with the time since the sample is not continuosly illuminated (dark conditions). This results in higher values of krec in the solar cells under operation, and our calculated krec can be taken as the lowest limit considering the experimental conditions (pump pulse energy density = 40−80 μJ/cm2).

transient absorption spectra at different delay times of a complete cell sensitized with SQ 41 containing electrolyte I. We observed two positive signals around 530−580 and 720−750 nm with a bleach around 650 nm, matching the steady-state absorption spectrum. The same absorption bands were detected for the other SQs independently of the used electrolyte. These features were previously assigned to the radical cation of the SQs, SQ•+, which did not decay in the 2 ns time window of the femtosecond experiments. We observed no change in the shape of the spectra when the delay time becomes longer, and therefore the SQ•+ deactivates straightforward to the ground state. Figure 9A exhibits the normalized transient absorption decays at 550 nm for the complete cells sensitized with SQ

Figure 9. (A) Normalized decays of the flash photolysis signals (in terms of change in absorbance, ΔA) at 550 nm of complete cells prepared with TiO2, sensitized with SQ 41, and containing electrolytes I−III and without electrolyte. (B) Normalized decays of the transient signals (ΔA) at 550 nm of complete cells prepared with TiO2, sensitized with (1) SQ 26, (2) SQ 41, (3) SQ 4, and (4) SQ 2, and containing electrolyte I. Solid lines are from the best stretched exponential fits of the experimental data. λexc = 650 nm. 22165

dx.doi.org/10.1021/jp306125y | J. Phys. Chem. C 2012, 116, 22157−22168

The Journal of Physical Chemistry C

Article

SQ•+ + e−(TiO2 ) → SQ

(8)

SQ•+ + I− → [SQ•+·I−]

(9a)

[SQ•+· I−] + I− → SQ· I 2•−]

(9b)

I 2•− + I 2•− → I3− + I−

complete solar cells sensitized with the different SQs and containing electrolyte I. There is a clear difference in the regeneration dynamics among the four SQs. The complete cells sensitized with the SQ 26 and SQ 41 exhibit shorter lifetimes than those of the SQ 4 and SQ 2 (Table 3). In one of the possible accepted reaction schemes for the regeneration process, the first step is the diffusion of I− to the vicinity of the radical cation of the SQs to form a complex without transferring of charge (eq 9a). The second step involves the approach of another I− species to interact with the complex and produce the charge transfer, releasing the ground state of the SQ and the I2•− (eq 9b). Finally, the disproportionation of I2•− occurs to generate I− and the oxidized species, I3− (eq 9c). The overall regeneration reaction has been recently reported to be of first order with respect to I− in a complete cell sensitized with N719.55 Assuming this behavior in our cells and given that the rate constant is far from the diffusion limit (kdiff ∼ 109−1010 M−1s−1), the rate-limiting step is suggested to be the second one (eq 9b).55,59 Thus, two parameters seem to determine the rate of the regeneration reaction: the strength of the interaction between the first I− and the specific atom in the dye, and the ease of the approach of the second I− species defined by the net charge of the complex. It has been reported that the I− interacts with the π system in metal-free sensitizers.57 Therefore, the more positive the atomic charge on the π system of oxidized SQs, the easier the SQ•+ interacts with I−. The molecular structures of SQ 26 and SQ 41, whose complete cells show the faster regeneration kinetics, differ from the SQ 4 and SQ 2 ones in the presence of CF3− groups attached to the lateral chains. This group has a strong inductive effect, removing negative charge from the SQ backbone, and therefore the interaction between SQ•+ and I− species will be very strong, facilitating a fast regeneration process. Moreover, the resulting higher positive charge density on the SQ backbone will favor the electrostatic interactions with the second I−, facilitating its approach to the complex and accelerating the regeneration reaction. Finally, it would be interesting to apply an expression similar to eq 7 to calculate the efficiency of the regeneration reaction. However, this is not possible with our data since the calculated krec is not the one governing in the cells under operation (vide supra). Despite that, we can utilize the kreg to draw a qualitative tendency of the regeneration efficiency among the four SQs based on the apparently much slower recombination reaction. Complete cells sensitized with SQ 26 present the largest value of kreg = 49.5 × 104 s−1 which suggests the highest efficiency process. The second largest value of kreg corresponds to the cells sensitized with SQ 41 (9.1 × 104 s−1). Finally, two similar values of kreg were obtained for SQ 4 and SQ 2 (4.5 and 3.0 × 104 s−1, respectively), which predicts no major differences in the regeneration efficiency between these two SQs. In conclusion, we have measured the effective lifetimes for the charge recombination and regeneration reactions for the complete cells containing electrolytes I−III and without electrolyte. Addition of tert-butyl pyridine to the electrolyte slows down the regeneration reaction, while Li+ cations produce the opposite effect (Scheme 1). The increase of the local iodide concentration near the TiO2 surface, together with the strengthening of the SQ•+/I− interaction upon addition of Li+, accounts for its faster regeneration dynamics. On the other hand, the inductive effect of the CF3− groups in the structure of SQ 26 and SQ 41 leads to a higher positive charge density in

(9c) •+

In the presence of the redox electrolyte, SQ is reduced by the iodide species (regeneration reaction, eq 9) since this process is usually ∼100 times faster than the recombination one. Table 3 summarizes the effective lifetimes, τobs, obtained from the fits of the transient signals at 550 nm. The use of the stretched function to fit the dynamics of the recombination process is explained due to the exponential distribution of states near the conduction band edge and trap states of the TiO2 nanoparticles. However, in the regeneration reaction, the existence of a range of activation energies due to different possible binding orientations of the SQ or heterogeneous gradients of iodide concentration across the film has been suggested to justify the use of the stretched function.55 Clearly, the lifetimes of the complete cells containing the electrolytes I− III are all much shorter than those without redox electrolyte, which is a requirement for high efficiency cells. Upon comparison of the effective lifetimes obtained for the complete cells containing the different electrolytes (Table 3), we noted a tendency according to the presence of the different additives (tert-butyl pyridine and/or Li+). Addition of tert-butyl pyridine to the electrolyte gives rise to an elongation of the lifetimes (compare electrolytes III and II, for example, from 5 to 12 μs in SQ 26). Furthermore, incorporation of Li+ cations in the electrolyte I causes the opposite effect, a decrease in the regeneration lifetimes (compare electrolytes II and I, for example, from 12 to 2 μs in SQ 26). Thus, tert-butyl pyridine decelerates the regeneration reaction, and Li+ cations increase the rate of the same reaction. These results for the complete cells sensitized with the studied SQs are in line with previous works with other organic dyes that correlate the effect of additives on the regeneration dynamics from a phenomenological level.16 As mentioned previously, both additives adsorb onto the surface of the TiO2, but they can also bind the iodide or the dye molecule.56,57 It has been suggested that when positive charged molecules are adsorbed onto the TiO2 surface the local iodide concentration near the surface is larger, which explains the faster regeneration process upon addition of Li+.58 On the other hand, intermolecular interactions of Li+ cations with a dye have been shown to strengthen the interaction of the oxidized dye with the I−, which reinforces the positive effect of the Li+ on the acceleration of the regeneration reaction.57 In general, both additives will also influence the diffusion of the I−/I3− redox species and local viscosity of the electrolytes, which will affect the regeneration dynamics. However, a more detailed study is needed to identify how the additives affect those two parameters. Equation 10 gives the expression for the rate constant of the regeneration process, kreg, when electrolyte I is used k reg =

1 − k rec τobs

(10)

where τobs is the weighted average effective lifetime from the samples containing electrolyte I and krec is the inverse of the τobs in samples without electrolyte. Figure 9B shows the comparison of the normalized transient absorption decays at 550 nm for the 22166

dx.doi.org/10.1021/jp306125y | J. Phys. Chem. C 2012, 116, 22157−22168

The Journal of Physical Chemistry C



the π system, which explains the faster regeneration kinetics of the complete cells of these two SQs.

ACKNOWLEDGMENTS This work was supported by the MINECO through project PLE2009-0015. We thank Dr. M. Ziolek for fruitful discussions. G.M. is grateful to MINECO for a “Juan de la Cierva” contract.

4. CONCLUSION Complete solar cells fabricated with TiO2 NP thin films, sensitized with a family of four indole-based squaraines dyes, SQs, and containing different additives in the electrolyte solutions have been studied by using femtosecond transient absorption spectroscopy and flash photolysis. We have shown that the combined action of both the molecular structure (symmetric or asymmetric, while varying the length and the nature of the lateral chains) and the composition of the electrolyte (tert-butyl pyridine and Li+ cation) have a notable influence on the electron injection and regeneration dynamics of the SQ-based solar cells. Thus, in the presence of tert-butyl pyridine, the electron injection is retarded 5−10 times, which is accounted for by the negative shift of the conduction band potential caused by the tert-butyl pyridine. In the cells containing both Li+ and tertbutyl pyridine, despite the opposite effect of both additives, the electron injection is always slower than in the cells without additives. This reflects a stronger influence of the tert-butyl pyridine on the electron injection due to a cooperative effect between Li+ and tert-butyl pyridine, which produces a higher concentration of tert-butyl pyridine molecules at the surface of the TiO2 NP. On the other hand, addition of tert-butyl pyridine to the electrolyte gives rise to an elongation of the regeneration lifetimes, whereas incorporation of Li+ cations results in the opposite effect. The Li+ cations interact with the backbone of the SQs, showing that the local [I−] near the dye is larger due to electrostatic attractions, favoring the approach of the I− species, and therefore the regeneration process is faster. Also, the strong electronegative character of the N atom in the tertbutyl pyridine reduces the [I−] near the surface, causing the opposite effect. Regarding the kei and φei values for the four studied SQs, we conclude that a high value of ΔG0ei is not enough to ensure a fast electron injection process in SQs since the coupling, V, also plays a major role. Thus, asymmetric SQs with a long alkyl chain are expected to provide the strongest couplings with the TiO2 surface. Moreover, the presence of the two CF3− groups in SQ 26 entails deactivation channels that quench the S1, competing with the electron injection. In the regeneration process, a higher positive charge density in the π-conjugated system of the SQs favors the interaction of the I− species and the SQ radical cation, SQ•+ . Thus, electron-donating substituents are expected to accelerate the process, like the CF3− group. This information makes it possible to identify the energetic and molecular features of the SQs that optimize the DSSCs.





REFERENCES

(1) Ajayaghosh, A. Acc. Chem. Res. 2005, 38, 449−459. (2) Beverina, L.; Salice, P. Eur. J. Org. Chem. 2010, 1207−1225. (3) Alex, S.; Santhosh, U.; Das, S. J. Photochem. Photobiol., A 2005, 172, 63−71. (4) Beverina, L.; Crippa, M.; Salice, P.; Ruffo, R.; Ferrante, C.; Fortunati, I.; Signorini, R.; Mari, C. M.; Bozio, R.; Facchetti, A.; Pagani, G. A. Chem. Mater. 2008, 20, 3242−3244. (5) Beverina, L.; Crippa, M.; Landenna, M.; Ruffo, R.; Salice, P.; Silvestri, F.; Versari, S.; Villa, A.; Ciaffoni, L.; Collini, E.; Ferrante, C.; Bradamante, S.; Mari, C. M.; Bozio, R.; Pagani, G. A. J. Am. Chem. Soc. 2008, 130, 1894−1902. (6) McEwen, J. J.; Wallace, K. J. Chem. Commun. 2009, 6339−6351. (7) Jisha, V. S.; Arun, K. T.; Hariharan, M.; Ramaiah, D. J. Am. Chem. Soc. 2006, 128, 6024−6025. (8) Gassensmith, J. J.; Matthys, S.; Wojcik, A.; Kamat, P. V.; Smith, B. D. Chem.Eur. J. 2010, 16, 2916−2921. (9) Choi, H.; Raabe, I.; Kim, D.; Teocoli, F.; Kim, C.; Song, K.; Yum, J.-H.; Ko, J.; Nazeeruddin, M. D.; Grätzel., M. Chem.Eur. J. 2010, 16, 1193−1201. (10) Choi, H.; Kang, S. O.; Ko, J.; Gao, G.; Kang, H. S.; Kang, M.-S.; Nazeeruddin, Md. K.; Grätzel, M Angew. Chem., Int. Ed. 2009, 48, 5938−5941. (11) Shi, Y.; Hill, R. B. M.; Yum, J.-H.; Dualeh, A.; Barlow, S.; Grätzel, M.; Marder, S. R.; Nazeeruddin, M. K. Angew. Chem., Int. Ed. 2011, 50, 6619−6621. (12) Ruankham, P.; Macaraig, L.; Sagawa, T.; Nakazumi, H.; Yoshikawa, S. J. Phys. Chem. C 2011, 115, 23809−23816. (13) Paek, S.; Choi, H.; Kim, C.; Cho, N.; So, S.; Song, K.; Nazeeruddin, M. K.; Ko, J. Chem. Commun. 2011, 47, 2874−2876. (14) Ito, S.; Miura, H.; Uchida, S.; Takata, M.; Sumioka, K.; Liska, P.; Comte, P.; Pechy, P.; Grätzel, M. Chem. Commun. 2008, 5194−5196. (15) Yella, A.; Lee, H.-W.; Tsao, H. N.; Yi, C.; Chandiran, A. K.; Nazeeruddin, M. K.; Diau, E. W.-G.; Yeh, C.-Y.; Zakeeruddin, S. M.; Grätzel, M. Science 2011, 334, 629−634. (16) Hagfeldt, A.; Boschloo, G.; Sun, S.; Kloo, L.; Pettersson, H. Chem. Rev. 2010, 110, 6595−6663. (17) Grätzel, M. Acc. Chem. Res. 2009, 42, 1788−1798. (18) Ó Regan, B. C.; Durrant, J. R. Acc. Chem. Res. 2009, 42, 1799− 1808. (19) Choi, H.; Nicolaescu, R.; Paek, S.; Ko, J.; Kamat, P. V. ACS Nano 2011, 5, 9238−9245. (20) Choi, H.; Kim, S.; Kang, S. O.; Ko, J.; Kang, M.-S.; Clifford, J. N.; Forneli, A.; Palomares, E.; Nazeeruddin, M. K.; Grätzel, M. Angew. Chem., Int. Ed. 2008, 47, 8259−8263. (21) Paek, S.; Choi, H.; Kim, C.; Cho, N.; So, S.; Song, K.; Nazeeruddin, M. K.; Ko, J. Chem. Commun. 2011, 47, 2874−2876. (22) Koops, S. E.; Barnes, P. R. F.; Ó Regan, B. C.; Durrant, J. R. J. Phys. Chem. C 2010, 114, 8054−8061. (23) Listorti, A.; Ó Regan, B.; Durrant, J. R. Chem. Mater. 2011, 23, 3381−3399. (24) Cherepy, N. J.; Smestad, G. P.; Grätzel, M.; Zhang, J. Z. J. Phys. Chem. B 1997, 101, 9342−9351. (25) Kamat, P. V.; Das, S.; Thomas, K. G.; George, M. V. Chem. Phys. Lett. 1991, 178, 75−79. (26) Boschloo, G.; Hagfeldt, A. Inorg. Chim. Acta 2008, 361, 729− 736. (27) Anderson, N. A.; Lian, T. Annu. Rev. Phys. Chem. 2005, 56, 491−519. (28) Watson, D. F.; Meyer, G. J. Coord. Chem. Rev. 2004, 248, 1391− 1406. (29) Koops, S. E.; Ó Regan, B. C.; Barnes, P. R. F.; Durrant, J. R. J. Am. Chem. Soc. 2009, 131, 4808−4818.

ASSOCIATED CONTENT

S Supporting Information *

Tables S1 and S2 and Figure S1. This material is available free of charge via the Internet at http://pubs.acs.org.



Article

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest. 22167

dx.doi.org/10.1021/jp306125y | J. Phys. Chem. C 2012, 116, 22157−22168

The Journal of Physical Chemistry C

Article

(30) Haque, S. A.; Palomares, E.; Cho, B. M.; Green, A. N. M.; Hirata, N.; Klug, D. R.; Durrant, J. R. J. Am. Chem. Soc. 2005, 127, 3456−3462. (31) Katoh, R.; Furube, A.; Kasuya, M.; Fuke, N.; Koide, N.; Han, L. J. Mater. Chem. 2007, 17, 3190−3196. (32) Haque, S. A.; Tachibana, Y.; Willis, R. L.; Moser, J. E.; Grätzel, M.; Klug, D. R.; Durrant, J. R. J. Phys. Chem. B 2000, 104, 538−547. (33) Pandey, S. S.; Inoue, T.; Fujikawa., N.; Yamaguchi, Y.; Hayase, S. Thin Solid Films 2010, 519, 1066−1071. (34) de Miguel, G.; Marchena, M.; Ziółek, M.; Pandey, S. S.; Hayase, S.; Douhal, A. J. Phys. Chem. C 2012, 116, 12137−12148. (35) Wojcik, A.; Nicolaescu, R.; Kamat, P. V.; Chandrasekaran, Y.; Patil., S. J. Phys. Chem. A 2010, 114, 2744−2750. (36) Kuang, D.; Walter, P.; Nüesch, F.; Kim, S.; Ko, J.; Comte, P.; Zakeeruddin, S M.; Nazeeruddin, M. K.; Grätzel, M. Langmuir 2007, 23, 10906−10909. (37) Pandey, S. S.; Inoue, T.; Fujikawa, N.; Yamaguchi, Y.; Hayase, S. J. Photochem. Photobiol. A 2010, 214, 269−275. (38) Inoue, T.; Pandey, S. S.; Fujikawa, N.; Yamaguchi, Y.; Hayase, S. J. Photochem. Photobiol. A 2010, 213, 23−29. (39) Gil, M.; Douhal, A. Chem. Phys. 2008, 350, 179−185. (40) Nakayama, T.; Amijima, Y.; Ibuki, K.; Hamanoue, K. Rev. Sci. Instrum. 1997, 68, 4364−4371. (41) de Miguel, G.; Marchena, M. J.; Zitnan, M.; Pandey, S. S.; Hayase, S.; Douhal, A. Phys. Chem. Chem. Phys. 2012, 14, 1796−1805. (42) Martín, C.; Ziółek, M.; Marchena, M.; Douhal, A. J. Phys. Chem. C 2011, 115, 23183−23191. (43) The samples “without electrolyte” were prepared with the TiO2 nanoparticle thin films sensitized with the different SQs but without adding the solvent. (44) Koops, S. E.; Durrant, J. R. Inorg. Chim. Acta. 2008, 361, 663− 670. (45) Tachibana, Y.; Rubtsov, I. V.; Montanari, I.; Yoshihara, K.; Klug, D. R.; Durrant, J. R. J. Photochem. Photobiol. A: Chem. 2001, 142, 215− 220. (46) Nasr, C.; Hotchandani, S.; Kamat, P. V. J. Phys. Chem. B 1998, 102, 4944−4951. (47) Durrant, J. R.; Haque, S. A.; Palomares, E. Coord. Chem. Rev. 2004, 248, 1247−1257. (48) Schmickler, W. Interfacial Electrochemistry; Oxford University Press: Oxford, 1996. (49) Zhang, S.; Yang, X.; Zhang, K.; Chen, H.; Yanagida, M.; Han, L. Phys. Chem. Chem. Phys. 2011, 13, 19310−19313. (50) Katoh, R.; Kasuya, M.; Kodate, S.; Furube, A.; Fuke, N.; Koide, N. J. Phys. Chem. C 2009, 113, 20738−20744. (51) Palomares, E.; Clifford, J. N.; Haque, S. A.; Lutz, T.; Durrant, J. R. J. Am. Chem. Soc. 2003, 125, 475−482. (52) Reynal, A.; Forneli, A.; Martínez-Ferrero, E.; Sánchez-Díaz, A; Vidal-Ferran, A.; Ó Regan, B. C.; Palomares, E. J. Am. Chem. Soc. 2008, 130, 13558−13567. (53) Lutkouskaya, K.; Calzaferri, G. J. Phys. Chem. B 2006, 110, 5633−5638. (54) Kay, A.; Humprhy-Baker, R.; Grätzel, M. J. Phys. Chem. 1994, 98, 952−959. (55) Anderson, A. Y.; Barnes, P. R. F.; Durrant, J. R.; Ó Regan, B. C. J. Phys. Chem. C 2011, 115, 2439−2447. (56) Furube, A.; Katoh, R.; Hara, K.; Sato, T.; Murata, S.; Arakawa, H.; Tachiya, M. J. Phys. Chem. B 2005, 109, 16406−16414. (57) Pastore, M.; Mosconi, E.; De Angelis, F. J. Phys. Chem. C 2012, 116, 5965−5973. (58) Tefashe, U. M.; Nonomura, K.; Vlachopoulos, N.; Hagfeldt, S.; Wittstock, G. J. Phys. Chem. C 2012, 116, 4316−4323. (59) Pelet, S.; Moser, J. E.; Grätzel, M. J. Phys. Chem. B 2000, 104, 1791−1795.

22168

dx.doi.org/10.1021/jp306125y | J. Phys. Chem. C 2012, 116, 22157−22168