An investigation of the role increasing π-conjugation has on the efficiency of dye-sensitized solar cells fabricated from ferrocene-based dyes

In this article, we report the synthesis of new ferrocene incorporating dyes and their evaluation as sensitizers for dye-sensitized solar cells. We have shown that the sequential addition of alkyne units as p-spacers plays an important role in modulating their solution optical and redox properties. Likewise, the most extensively conjugated dye (Fc-D3) provided the highest power conversion efficiency (h) compared to the other lessconjugated dyes in the series. We have also demonstrated that the choice of electrolyte is important in determining h for this series of dyes.


Introduction
The advantages that dye-sensitized solar cells (DSSCs) 1 have over silicon-based photovoltaic devices, such as their ability to operate under low light conditions and their lower cost, has ensured that devices of this type continue to receive signicant attention. 2 Ruthenium-based dyes dominated the early progress in this eld, however, metal-free organic dyes have become increasingly attractive due to their higher molar extinction coefficients, convenient synthesis, purication and structural modication. 3 More recently, porphyrin-based dyes have revitalised research into metal-incorporating dyes and to date have produced headlining power conversion efficiencies of around 13%. 4 The ability of ferrocene (Fc) to undergo reversible one-electron chemical and electrochemical oxidation at low potential has allowed this unit to become a ubiquitous building block in a range of disparate applications including medicine and materials science. 5 Although Fc has proved to be an effective redox electrolyte for DSSCs, 6 its incorporation as a donor unit in photosensitizers has received surprisingly little attention. For example, Chauhan et al. have reported DSSCs fabricated from Fc-based dyes which indicated that both the dye anchoring group and the electrolyte play an important role in determining the power conversion efficiency (h). 7 More recently, Sirbu et al. reported DSSCs fabricated using a trisubstituted ferrocene-based porphyrin derivative with a cobalt(II/III) electrolyte, however, DSSC performance was rather modest (h ¼ 0.0081%). 8 In this study, we describe the synthesis and characterisation of new Fc-based dyes featuring benzothiadiazole and cyanoacrylic acid residues 9 as acceptor units and alkyne residues as p-linker units. 10 We report the signicance of increasing the number of alkyne linker units has on the optical and redox properties of the dyes and the key parameters of their resulting DSSCs. We have also shown that the choice of electrolyte is important in terms of controlling h.

Synthesis
The three dyes were prepared according to the synthetic pathway shown in Scheme 1. Ferrocene boronic acid 1 was coupled with 4,7-dibromobenzothiadiazole 2, 11 via Suzuki crosscoupling reaction, affording 3 in 46% yield. Compound 3 was further reacted with 4-formylphenylboronic acid, affording compound 4 in 84% yield. Compound 6 was obtained, in a 89% yield, via a Sonogashira cross-coupling reaction of ethynylferrocene 5 and 2. 12 This compound was then reacted with 4-formylphenylboronic acid and 4-ethynylbenzaldehyde, via Suzuki and Sonogashira coupling reactions, to give compound 7 and 8, in 85% and 45% yield, respectively. The three aldehydes 4, 7 and 8 were reacted with cyanoacetic acid to afford the corresponding cyanoacrylates Fc-D1 (77% yield), Fc-D2 (98% yield) and Fc-D3 (44% yield). p-p* transition of the conjugated aromatic moieties, 13 whilst the absorption around 400 nm corresponds to an intramolecular charge-transfer (ICT) between the donor and acceptor groups. It is clear that the increase in the conjugation length results in the red shi of the two absorption maxima with higher molar extinction coefficients. The distinct absorption band around 550 nm for Fc-D1 is attributed to a d-p* charge transfer (CT). However, d-p* CT bands are sensitive to adjacent acceptor strength, and are not as pronounced in the spectra of the alkyne containing dyes Fc-D2 and Fc-D3. Compared with the spectra in DMF solution, the ICT absorption band of these dyes on TiO 2 lms exhibits signicant  spectral broadening and red-shi because of the interaction of the anchoring groups with the surface of TiO 2 and J-aggregation of photosensitizer molecules. Unlike the spectra in DMF solution, the absorption related to the metal-acceptor charge transfer is observable in all dyes on TiO 2 lms. From the cut-off wavelengths of the spectra in DMF solution, the estimated values of the optical gaps (E opt ) are listed in Table 1. The electrochemical behaviour of the three dyes in DMF solution was explored by cyclic voltammetry (CV) and square wave voltammetry (SWV) (Fig. 2). The redox-active ferrocenyl group exhibits a single oxidation wave whereas a one electron reduction wave attributed to the BTD acceptor unit is observed. Addition of alkyne p-linker units on going from Fc-D1 to Fc-D3, results in the reduction potentials being shied to more positive values indicating that the benzothiadiazole unit is easier to reduce. The estimated electron affinities (EAs) and ionization potentials (IPs) are presented in Table 1. 14 As shown in Fig. 3, all the IPs are slightly lower in energy than the I À /I 3 À redox couple, ensuring that the generation of all the dyes is energetically favourable. On the other hand, the positions of the EAs are found to be almost 1 eV above the TiO 2 conduction band.

Theoretical calculations
To gain insight on the electronic structure and optical properties of the ferrocene-based dyes, density functional theory (DFT) and time dependent DFT calculations were performed. Fig. 4 shows ground state optimized geometries and electronic density distributions of the HOMO and LUMO of the dyes, which clearly show the role of the acetylene linker on maintaining planarity and facilitating ICT separation. In particular, on going form FC-D1 to FC-D3 the sequential addition of the acetylene linker moiety makes the dihedral angles between the ferrocenyl and benzothiadiazole moieties and between the benzothiazole and phenyl ring moieties signicantly decrease from 18.87 to 0.92 and from 33.23 to 2.12 , respectively. This indicates that acetylene linkers eliminate the steric hindrance between the hydrogen atoms of these rings. The ferrocenyl groups in all dyes adopt an eclipsed D 5h conformation rather than the staggered conformation. 15,16 The electron densities of the HOMOs for all dyes are mainly localized on the ferrocenyl   moiety and are extended along to the benzothiadiazole moiety to the phenyl ring. The LUMOs for all dyes are localised on the benzothiadiazole and cyanoacrylate residues. In all dyes, signicant overlap of the frontier orbitals occurs suggesting that efficient ICT likely occurs from the donor to the terminal acceptor group which anchors to the TiO 2 surface.
Photovoltaic measurements Fig. 5 shows J-V characteristics of DSSCs sensitized with the three dyes. The photovoltaic parameters are summarized in Table 2. The addition of the two acetylene bridging units in Fc-D3 resulted in better photovoltaic performance with the iodide/ triiodide redox couple than Fc-D1: V oc from 0.407 V to 0.434 V, J sc from 0.730 mA cm À2 to 1.070 mA cm À2 and h from 0.180% to 0.279%. The h value of Fc-D3 compared to Fc-D1 is more than doubled which may be ascribed to the improvement of light harvesting efficiency or the electron injection efficiency of Fc-D3.   negatively and thus the increase of E redox À E cb leads to the increase of the photovoltage. 18,19 As electrochemical impedance spectroscopy (EIS) can offer very valuable insight into interfacial charge-transfer processes, we have used this technique to explore DSSCs sensitized with Fc-D3. The semicircle in the intermediate frequency region in Fig. 6a represents the electron transfer process at the TiO 2 /dye/ electrolyte interface and its radius is related to the charge recombination rate, i.e. a larger radius indicates a slower charge recombination. The semicircle in the low frequency region in Fig. 6a is associated with the Warburg diffusion of the redox couple in the electrolyte. The electron lifetime corresponding to the TiO 2 /dye/electrolyte interface, can be estimated from the peak frequency (f peak ) in Fig. 6b according to s e ¼ 1/2pf peak . The calculated values of Fc-D3 (AN-50) and Fc-D3 (Hybrid) are 0.36 ms and 0.71 ms, respectively. The addition of the organic redox couple suppresses the back charge recombination at the TiO 2 /dye/electrolyte and thus this resulted in the increase in J sc .

Materials and methods
All reactions were undertaken under a nitrogen atmosphere. Solvents were puried using a PureSolv solvent purier system. Reagents were purchased from Sigma-Aldrich, and were used without further purication. NMR spectra were obtained with either a Bruker AVIII 400 MHz or a Bruker AVIII 500 MHz spectrometer and all reported chemical shis are relative to TMS. UV-vis spectra were recorded on a Perkin-Elmer Lamda 25 instrument. Optically determined band gaps (E opt ) were estimated using the absorption edge of the longest wavelength absorption (l) using E opt (eV) ¼ (1240/l (nm)). Cyclic voltammetry measurements were undertaken using a CH Instruments 440A electrochemical analyzer using a platinum working electrode, a platinum wire counter electrode and a silver wire pseudo-reference electrode. Ferrocene was used as an internal standard and all redox couples are reported versus the ferrocene/ferrocenium (Fc/Fc + ) redox couple, adjusted to 0.0 V. The solutions were prepared using dry dimethylformamide (DMF) containing electrochemical grade tetrabutylammonium hexauorophosphate (0.1 M) as the supporting electrolyte. The solutions were purged with nitrogen gas for 3 min prior to recording the electrochemical data.

Theoretical calculations
Density functional theory (DFT) calculations were performed using Gaussian 09 to gain insight electronic properties of the dye molecules. 20 Global minimum states were conrmed by absence from imaginary frequencies under ground-state geometry optimization followed by vibrational frequency calculations. All calculations were conducted with Becke's three-parameter hybrid and Lee-Yang-Parr's gradient corrected correlation (B3LYP) functional and 6-311G(d,p) basis set for H, C, N, O, S and Lanl2DZ functional for Fe, under vacuum.

Fabrication of DSSCs and photovoltaic measurements
The TiO 2 photoanodes were screen-printed onto transparent uorine-doped SnO 2 (FTO)-coated conducting glass (TEC 8, Pilkington, 2.2 mm-thick, sheet resistance ¼ 8 U sq. À1 ). 21 The resulting layer was placed in a muffle furnace and gradually heated to 300 C over a 30 min period, then heated at 300 C for 1 hour, and then heated to 575 C for 1 hour, and then cooled to room temperature. The screen printing and calcination processes were repeated until a thickness of approximately 20 mm was obtained. The active area of the electrodes was 0.25 cm 2 . The prepared photoanodes were immersed in a 0.04 M solution of TiCl 4 at 75 C for 1 hour, rinsed with deionized water, and then sintered at 500 C for 30 min. They were exposed to O 2 plasma for 10 min and then immersed for 24 hours in one of three dye-containing ethanol solutions (0.5 mM). The Pt counter electrodes were prepared on the FTO-coated glass with magnetron sputtering and two holes were drilled in the glass. Both the dye-sensitized TiO 2 electrode and Pt counter electrode were sealed with a 60 mm-thick layer of Surlyn (Solaronix, Switzerland). An iodide based redox electrolyte (Iodolyte AN-50, Solaronix) was injected into the rear side of the counter electrode. A hybrid electrolyte, a mixture of the I À /I 3 À and McMT À / BMT couples was also used. The photovoltaic characteristics of the DSSCs under AM 1.5 illumination (equivalent to one sun, 1 kW m À2 ) were investigated with a solar cell current-voltage (I-V) measurement system (K3000 LAB, McScience, Korea). The photocurrent density (J sc ), open-circuit voltage (V oc ), ll factor (FF), and power-conversion-efficiency (h) were simultaneously measured. EQE and IQE were recorded as a function of excitation wavelength (l) by a spectral incident photon-to-current efficiency (IPCE) measurement system (K3100, McScience, Korea).

Synthetic procedures
Compound 3. Ferroceneboronic acid 1 (500 mg, 2.17 mmol) and 2 (960 mg, 3.26 mmol) were dissolved in dry dioxane (15 mL). A 2 M K 2 CO 3 aqueous solution (10.9 mL, 21.7 mmol) was added, and the mixture was degassed for 30 minutes with N 2 . Then Pd(dppf)Cl 2 (791 mg, 0.11 mmol) was added and the resulting mixture was le to stir at reux. Aer 14 h, the mixture was allowed to cool to room temperature, and was then quenched with water (50 mL) and extracted with DCM (3 Â 50 mL). The combined organic extracts were dried over MgSO 4 , ltered and concentrated under vacuum. The crude compound was puried by column chromatography (SiO 2 , petroleum ether : toluene (1 : 1)), affording the title compound as a purple solid (393 mg, 46%  (15 mL). A 2 M K 2 CO 3 aqueous solution (4.9 mL, 9.8 mmol) was added, and the solution was degassed for 30 minutes with 20 mL methanol. The mixture underwent ultrasonic treatment for 2 hours and then iodine (1.25 g) was added to the solution and the reaction was sonicated for a further 10 minutes until the iodine disappeared completely. The solvent was removed under vacuum and the residue was dissolved in excess CH 2 Cl 2 . The solution was washed with water (3 Â 30 mL), and nally the organic phase was collected and dried over anhydrous Na 2 SO 4 , ltered and the solvent evaporated to yield BMT as a white solid which was dried under vacuum at 40 C for 12 hours.

Conclusions
In conclusion, a series of new Fc-functionalised dyes have been synthesised and characterised. We have shown that the sequential addition of alkyne units on going from Fc-D1 to Fc-D3 has a signicant effect on both the light absorption characteristics and LUMO of the dyes. In addition, DSSCs fabricated using dye Fc-D3 display the highest h values of the series, providing a maximum value of 0.325% when a hybrid electrolyte was used. Currently, we are exploring other n-type metal oxide semiconductors in order to avoid direct electron injection from the HOMO of the Fc dyes to the conduction band (CB) of the metal oxide and back electron transfer from the CB to the HOMO. However, an important problem facing the development of ferrocene-based dyes is the well-documented instability of oxidized state of Fc (ferrocenium) to molecular oxygen, 22 which will need to be addressed if efficient and stable DSSCs are to be fabricated in the future.