Improved performance of dye-sensitized solar cells by tuning the properties of ruthenium complexes containing conjugated bipyridine ligands

All materials were reagent grade and were used as received. All synthetic reactions were carried out under a nitrogen atmosphere unless otherwise noted. Solvents were purified according to the published methods. N,N-di-p-anisylamine and N,N-di-p-hexyloxyphenylamine were synthesized according to a procedure in the literature [22, 23]. TiO₂ nanoparticles were purchased from Solaronix S. A. (Nanoxide-T, Lausanne, Switzerland). Fluorine-doped tin oxide (FTO) electrodes with a sheet resistance of 8 Ω per square were supplied from Samsung Corning Co., Korea.

^{1 H}-NMR spectra were recorded on a Varian INOVA 400 MHz NMR spectrometer in ^CDCl ₃ or (DMSO)-d₆ . ESI-MS was performed on a HP 1100 LC/MS with acetonitrile as solvent and FAB-MS on a JMS-700 HRMS. Absorption spectra were recorded on a OPTIZEN 2120 UV-vis spectrophotometer. Electrochemical measurements were performed under a nitrogen atmosphere and the electrolyte used was 0.1 M tetrabutylammonium tetrafluoroborate (^Bu ₄ ^NBF ₄) in acetonitrile. Cyclic voltammetry (CV) was recorded using a three-electrode system consisting of ^Ag/^{Ag +} (0.01 M ^AgNO ₃, 0.1 M ^Bu ₄ ^NBF ₄, ^CH ₃ ^CN) as the reference electrode, a platinum wire as counter electrode and a ^FTO/^TiO ₂/^Dye photoelectrode as the working electrode. All potentials are calibrated by the ferrocene/ferrocenium (^Fc/^{Fc +}) couple potential (E=+212 ^mV versus ^Ag/^AgNO ₃).

Dye-1, Dye-2 and Dye-3 were prepared by a one-pot synthetic route using ligand-1, ligand-2 and ligand-3. The procedure for preparing these dyes is shown in scheme 1 [24].

Photoanodes of the DSSCs were prepared from a colloidal suspension of nanosized ^TiO ₂ particles (Nanoxide-T, Solaronix, average size 13 nm) by a doctor blade technique. The electrodes were sintered at 500 °^C for 30 min in air to obtain ^TiO ₂ films with a thickness of 8 μ^m measured by a Dektek 6M Stylus Profiler. These electrodes were immersed in solution containing 0.5 mM of Ruthenium dyes for one day while still hot (80 °^C). Chenodeoxycholic acid (1 mM) was added into the dye solution as a coadsorbent to prevent dye aggregation [25]. The solvents for the dye solutions were DMF/t-BuOH (3:7, vv) for Dye-1 and ^CH ₃ ^CN/t-BuOH (1:1, vv) for N3, Dye-2 and Dye-3. Platinum-coated counter electrodes were obtained by spreading a drop of 5×10^{−3  M  H} ₂ ^PtCl ₆ (Fluka) solution in isopropanol on the surface of FTO glass and were treated in an oven at 400 °^C for 15 min. The electrolyte was composed of 0.1 M lithium iodide, 0.1 M iodine, 0.5 M 4-tert-butylpyridine and 0.6 M 1,2-dimethyl-3-propyl imidazolium iodide in acetonitrile. Photovoltaic characterization was carried out by illuminating the cell with a 1000 W Xenon lamp (Spectra-Physics) as a solar simulator. The light source was calibrated to 100 ^{mW  cm −2} by a standard NREL Si solar cell equipped with a KG-5 filter to approximate AM 1.5G one sun light intensity. An active area of about 1 ^{cm 2} was irradiated. Current–voltage curves were obtained by measuring the photocurrent of the cell using a Keithley mode 2400 digital source meter (Keithley, USA) under an applied external potential scan. IPCE values were measured as a function of wavelength from 300 nm to 800 nm using a specially designed IPCE system for DSSCs (PV Measurement, Inc.). A 75 W Xenon lamp was used as the light source for the generation of a monochromatic beam. Calibrations were performed with a silicon photodiode using a NIST-calibrated photodiode G425 as the standard and IPCE values were collected at a low chopping speed of 10 Hz.

The molecular structures of Dye-1, Dye-2 and Dye-3 and the preparation of ligands and complexes are given in scheme 1.

The syntheses of conjugated ligands (ligand-1, ligand-2 and ligand-3) are based on a multi-step synthetic pathway, starting from N,N-di-p-anisylamine and N,N-di-p-(hexyloxy)phenylamine. The next reaction step involves the N-arylation of these amines with 4-bromobenzaldehyde (ligand-1 and ligand-2) and 5-bromo-2-thiophene carboxaldehyde (ligand-3). These ligands were synthesized by two steps: nucleophilic addition of the anion of dimethylbipyridine with aldehyde intermediates, followed by dehydration of the corresponding alcohol in refluxing acetic acid. Ruthenium complexes were synthesized in a one-pot synthesis starting from a dichloro(p-cymene)ruthenium (II) dimer in DMF for 13 h. The crude products were purified on a Sephadex LH-20 column using methanol as eluent. The ^{1 H}-NMR profiles of ruthenium complexes (figure 1) are complicated because of the presence of NCS ligands makes pyridine rings magnetically non-equivalent. Hence, the ^{1 H}-NMR spectra exhibit a total number of resonances equal to the total number of aromatic and heterocyclic protons in the complex. There are two 2,2'-bipyridine ligands in different electronic environments. As a result, they show resonance peaks in the aromatic region corresponding to four pyridine rings. The resonance peaks in the aliphatic region are due to methoxy and hexyloxy protons.

Zoom In Zoom Out Reset image size

The absorption spectral data for Dye-1, Dye-2 and Dye-3 are shown in table 1 and figure 2(a). All the dyes show broad and intense visible bands between 393 and 526 nm which are due to metal-to-ligand charge transfer transitions (MLCT) [3, 15]. The absorption bands at about 300 nm are assigned to the intra ligand (π–π^*) transitions of bipyridine ligands. These results are in agreement with the homoleptic N3 dye. The absorption of Dye-1 exhibits three bands centered at 306, 429 and 526 nm. The band at 306 nm is assigned to the π–π^* transition of the ligand-centered transition due to dcbpy units. The second maximum at 429 nm is attributed to the π–π^* transition of the ligand-1 and one of the MLCT transitions for Dye-1. The absorption band at 526 nm corresponds to the MLCT caused by NCS ligands. Similar absorption bands can be obtained for Dye-2 with three bands at 304, 433 and 526 nm. In the case of Dye-3, only two distinct absorption bands could be observed: the band at 302 nm and a broad band from 480 to 540 nm.

Zoom In Zoom Out Reset image size

The π delocalization in conjugated ligands (ligand-1, ligand-2 and ligand-3) creates an increase in the molar extinction coefficient for these dyes. Therefore, very high molar extinction coefficients of MLCT bands of 20 560 ^{M −1  cm −1} (at 526 nm) for Dye-2 in the visible region were obtained.

Figure 2(b) shows the absorption peak wavelengths of adsorbed dyes on ^TiO ₂ films. The enhanced absorptivity of these dyes compared to N3 was observed.

In order to design and synthesize suitable sensitizers for DSSCs, matching between the band structure of the metal complex and the energy level of the semiconductor electrode, the redox electrolyte or the hole conductor is very important. Indeed, the HOMO (highest occupied molecular orbital) level of the excited state of the dye should be well matched with the lower bound of the conduction band of ^TiO ₂ to ensure fast and efficient electron injection. Its redox potential should be sufficiently positive that it can be regenerated through electron donation from a redox couple, such as I⁻/I₃ ⁻ [21].

Energy levels of the HOMO and LUMO (lowest unoccupied molecular orbital) derived from the oxidation potentials of metal complexes are displayed in table 1. The corresponding HOMO of Dye-3 (−5.26 ^eV) is higher in comparison with those of Dye-1 (−5.39 ^eV) and Dye-2 (−5.40 ^eV) with respect to zero vacuum energy level. It can be explained that Dye-3 carries an easily oxidizable ligand.

The HOMO levels of these dyes are much more negative than the conduction band level of ^TiO ₂. Therefore, their LUMO levels are sufficiently negative to inject electrons into the conduction band of ^TiO ₂. Table 1 also gives the band gap energy values obtained from CV results.

Dye-sensitized solar cells were prepared employing Dye-1, Dye-2 and Dye-3 and their performance was compared with the standard N3 dye. The current–voltage characteristics of these solar cells measured under AM 1.5G conditions are shown in figure 3(a). The detailed device parameters are summarized in table 2. Solar cells based on N3, Dye-1, Dye-2, and Dye-3 exhibit efficiencies of 3.86, 4.21, 4.41 and 2.88%, respectively. Both Dye-1 and Dye-2 show higher V ^_OC , J ^_SC and η than those of N3 dye emphasizing the positive effect of these conjugated ligands. The best performance was obtained by Dye-2 with a high short circuit current of 12.35 ^{mA  cm −2} in combination with an open circuit voltage of 0.644 V and FF of about 0.555 leads to a high power conversion efficiency of 4.41%. The structures of Dye-1 and Dye-2 are similar except that methoxy groups in ligand-1 have been replaced by hexyloxy chains in ligand-2. The increase in η value for Dye-2 compared to Dye-1 revealed the role of the hexyloxy substituents on ancillary ligand. The strongly electron donating alkoxy chains on diphenyl amine can avoid the intramolecular electron-hole recombination of Dye-2 and increase the lifetime of the electron on ^TiO ₂ (reducing the possibility of recombination of the electrons on ^TiO ₂ with the holes on the complex dye in excited state). Hydrophobicity is also important in designing new sensitizers for optimization of DSSCs. The Dye-2 dye contains only two carboxylic acids to facilitate least intermolecular hydrogen bondings. We found that the Dye-2-sensitized cell has a relatively lower fill factor compared to other dyes. This may due to the defects of the dye monolayer on the ^TiO ₂ electrode. The use of appropriate co-adsorbents or solvent in the dye assembly process may be able to fix these defects and enhance performance of Dye-2. This work is in progress and the results will be reported elsewhere.

Zoom In Zoom Out Reset image size

The high η values caused by higher values of I ^_SC are mainly contributed by the high absorption coefficients of these complexes. The performance of the device using Dye-3 is unfortunately lower than that of N3. The thiophene moiety incorporated in ligand-3 results in lower power conversion efficiency although it shows better absorptivity than that of N3. The higher device efficiency of dye with a phenyl spacer than that with a thienyl spacer is attributed to the more efficient charge transfer in the former. This is also related to the degree of overlap between the dye and nanocrystalline semiconductor electronic energy levels. The distribution of HOMO and LUMO over the N3 dye is optimal for electron injection into ^TiO ₂ and subsequent reduction of the dye cation by the redox mediator. The HOMO is predominantly distributed over the ruthenium center and thiocyanate ligands, whereas the LUMO is aligned by the bipyridine ligands [4, 20]. It has been proposed that upon excitation the NCS acts as a hole acceptor that can subsequently contact with the outside redox couple, for instance, I⁻/I₃ ⁻, and fulfill the cycle on reduction. In the case of Dye-3, the HOMO level (−5.26 ^eV) is highest among those dyes that have a relatively poor alignment of dye energy level with the ^TiO ₂ conduction band resulting in lower electron injection efficiency as well as dye regeneration efficiency.

The incident photon-to-current conversion efficiency (IPCE) spectra of Dye-1, Dye-2, Dye-3 and N3-sensitized solar cells are illustrated in figure 3(b). Broad bands cover almost the entire visible spectrum from 400 nm to 700 nm with maxima of 66.93% at 540 nm, 66.56% at 530 nm, 51.3% at 560 nm and 67.25% at 540 nm for Dye-1, Dye-2, Dye-3 and N3-sensitized solar cells, respectively. The IPCE values of N3, Dye-1 and Dye-2 are similar in the short wavelength range up to 550 nm. In long wavelengths Dye-2 shows higher IPCE values than those of these dyes. These enhanced IPCE values imply that good electronic coupling between the dye and ^TiO ₂ electrode and the new sensitizer acts as a more efficient electron donor to the conduction band of ^TiO ₂ films upon excitation. The IPCE values of the cells are slightly lower than the short circuit photocurrent density (J ^_SC ) appearing in the I–V characteristic curve because the light intensities used to measure these data are not the same (the incident light employed in IPCE measurement is not continuous and the intensity is lower).