Molecular gelation of ionic liquid–sulfolane mixtures, a solid electrolyte for high performance dye-sensitized solar cells

Laboratory for Photonics and Interfaces Engineering, Swiss Federal Institute of Tech E-mail: shaik.zakeer@ep.ch; michael.graet Center of Nanotechnology, King Abdulaziz Center of Nanotechnology, Department of P University, Jeddah, 21589, Saudi Arabia Center of Nanotechnology, Department of A University, Jeddah, 21589, Saudi Arabia Macromolecular Chemistry I, Bayreuther In and Bayreuther Zentrum für Grenzäche Bayreuth, D-95447, Bayreuth, Germ uni-bayreuth.de Cite this: J. Mater. Chem. A, 2014, 2, 15972


Introduction
Dye sensitized solar cells (DSCs) have attracted wide spread attention in recent years due to their low cost and high efficiency 1 and are presently being employed in commercial products such as light-weight exible devices for powering portable electronic devices and as electric power producing glass panels in building integrated photovoltaic devices. The redox electrolyte plays a key role in the solar light energy conversion process shuttling electric charges from the sensitizer to the counter-electrode. All embodiments that are presently used on the commercial scale employ the triiodide/iodide (I 3 À /I À ) couple as a charge carrier. Various types of electrolytes have been developed employing low viscosity solvents for top-level performance or non-volatile systems based on ionic liquids for high durability. In the latter case mass transport limitations of the photocurrent are frequently encountered depending on the viscosity of the liquid employed. On the other hand, the loss of volatile solvents through leaks is a practical concern for volatile solvent-based cell embodiments. The reduction of volatility and uidity of the liquid electrolyte by jellication is advantageous as it not only prevents leakage but also avoids undesirable contact with solvent sensitive cell constituents, such as the sealing material. In addition, avoiding organic solvents can reduce the risk of dye desorption from the TiO 2 surface into the electrolyte. Considerable effort has been made to replace liquid with quasi-solid or solid-state electrolytes using low molecular weight compounds, polymers or oxide nanoparticles as jellifying materials. [2][3][4][5][6][7][8] Low molecular weight gelators have attracted particular attention due to their numerous industrial applications such as in cosmetics, food processing, lubrications, etc. 9 Solidication occurs upon dissolving a small quantity of the gel forming agent into a hot liquid electrolyte and subsequent cooling below the gel transition temperature. The three-dimensional networks formed in the quasi-solid gel electrolyte immobilize the liquid component to a variable extent. 9 The reversibility of these quasisolid electrolytes depending on the temperature is the attracting feature compared to other quasi-solid electrolytes prepared with metal oxide nanoparticles or polymer electrolytes. Quasi-solid electrolytes prepared using a low molecular weight gelator are liquid at higher than the gel transition temperature and can ll easily the TiO 2 mesopores whereas this is not possible with other nanoparticle based quasi-solid electrolytes. Here we report on the effect of gelation on the photovoltaic performance of a recently reported sulfolane based ionic liquid electrolyte, which exhibits exceptional stability under long term heat stress and light soaking. 10

Results and discussion
We selected cyclohexanecarboxylic acid- [4-(3-tetradecylureido) phenyl]amide coded as CTP as a molecular gel forming agent in order to solidify the ionic liquid-sulfolane composite electrolyte. The molecular structure of the gelator CTP is designed to dissolve at elevated temperatures in a broad range of solvents with different polarities. The amide moiety and the urea moiety enable self-assembly upon cooling. The balance of these hydrogen-bonding units with the terminal cyclohexane group and the linear aliphatic alkyl chain (C14) enables the molecule to form nanobrillar structures, which cause efficient gelation in a large variety of solvents. [11][12][13] The inuence of the solvent polarity and gelator concentration on the dissolution and selfassembly of gelation is concentration dependent. 13 The chemical structure of CTP is shown in Scheme 1. At 2% weight the electrolyte is a liquid when heated to above 100 C but becomes a solid upon cooling to room temperature. The diffusion coef-cients of liquid and gel electrolytes were calculated by electrochemical impedance measurements by using a symmetric cell. 21 The Nyquist plot of the symmetric cells at 0 V is shown in Fig. 1. The diffusion coefficients of triiodide in the liquid electrolyte and in the gelled electrolyte mixture were determined to be 8 Â 10 À7 and 6 Â 10 À7 cm 2 s À1 , respectively. The diffusion coefficients were also calculated by the limiting currents measured by the current-voltage characteristics of such symmetric cells. The trend for liquid and gel electrolytes is similar although the values are somewhat higher compared to the impedance measurement (10 Â 10 À7 and 8 Â 10 À7 cm 2 s À1 , respectively). The newly formulated gel redox electrolyte was employed in DSCs featuring a double layer mesoporous TiO 2 lm a 4 mm light scattering layer of large 400 nm sized particles being superimposed on a 8 mm thick nanoparticle layer. The amphiphilic ruthenium complex C106 was used as a sensitizer. Table 1 compares the open circuit voltage (V oc ), ll factor (FF), short circuit current density (J sc ) and standard power conversion efficiency (PCE), the photovoltaic for the solid and liquid electrolyte devices. The PCE of the liquid electrolyte is a little higher than that of the gel electrolyte due to the difference in the J sc values. Fig. 2 shows J-V characteristics of the two DSCs measured under standard AM 1.5G illumination at 100 mW cm À2 while photocurrent action spectra are presented as an inset in Fig. 2. The incident photon to conversion efficiency (IPCE) covers a broad spectral range from 440 nm to 760 nm, reaching its maximum of 81% at 580 nm. The short circuit photocurrents calculated from the overlap integral of these curves with the standard global AM 1.5 solar emission spectrum agree within 4% with the measured photocurrent showing that any spectral mismatch between our solar simulator and the standard AM 1.5G solar emission is small. The long-term stability of devices with liquid electrolytes is shown to be stable when they are aged at 60 C light soaking as reported. 10 The stability of gel electrolyte devices was excellent and the relative deviation of the power conversion efficiency was within 2% of the initial performance aer aging for 3 weeks under light soaking at 60 C in a solar simulator (100 mW cm À2 ).
We employed electrochemical impedance spectroscopy (EIS) measurements to examine differences in the photovoltaic characteristics of liquid and gel electrolyte based DSCs.
Using the transmission line model [18][19][20] we extracted from the Nyquist plot the key electric device parameters, i.e. the Scheme 1 Molecular structure of the gelator.  transport resistance for the electrons through the nanocrystalline TiO 2 lm (R trans ), the recombination resistance between TiO 2 conduction band electrons and triiodide ions in the electrolyte (R CT ), and the chemical capacitance of the TiO 2 nanocrystals (C chem ). Results are shown in Fig. 3. The R CT values determined are in good agreement with the behaviour of the dark current (see Fig. 3a and b). Since the electrolyte compositions of liquid and gel electrolytes are similar, except for the added molecular gelator, the characteristics of the photoanode are very similar though it seems that in the case of the gel electrolyte the conduction band is slightly shied upwards (10 to 15 mV). On the other hand the recombination rate is increased and the transport seems also a bit slower rendering charge collection less efficient compared to the liquid electrolyte. Nevertheless all the observed differences are small and the conduction band move will be compensated by the faster recombination, which is responsible for obtaining similar V oc values for the two devices (see Fig. 3d). In the case of the resistances contributing to the overall series resistance of the devices (R series (from cables, contacts and the electrolyte resistance), R CE (from the charge transfer at the counter electrode) and R Dif (from the diffusion of the triiodide in the electrolyte)) the characteristics of the devices are very close, though the ones with a quasi-solid electrolyte show a higher R CE and a higher R Dif (see Fig. 3c). The small difference in the photovoltaic performance of liquid and gel electrolyte devices arises from the J sc value. The lower J sc of the gel electrolyte device may be attributed to its larger Warburg impedance indicating an impediment of triiodide diffusion through the network of the gel. Alternatively the higher conduction band position of the TiO 2 in the gel based DSCs may reduce the yield of electron injection of electrons from the excited dye into the TiO 2 conduction band, which is supported by the IPCE spectra.
Outdoor measurements were carried out in Jeddah Saudi Arabia with a DSC employing the C106 ruthenium complex as a sensitizer, the Z988 electrolyte and 2% CTP gelator. Data are shown in Fig. 4. Measurements were carried out on January 2, 2014 from 8 a.m. to 5 p.m. The active area of the cell was a round spot of 6 mm diameter. A black mask with a round aperture having a diameter of 8.5 mm was employed for all the measurements to avoid ination of photocurrents from light piping by uncovered glass. Details of the outdoor conditions are shown in Table 2 along with the measured key photovoltaic performance parameters. Impressively, the PCE of the cells was close to 9% around noontime when the temperature reached nearly 37 C conrming the excellent performance of the new electrolyte formulation under the hot conditions prevailing in Jeddah even in the beginning of January. The laboratory results taken from 5 cells showed the same trend as measured under outdoor conditions. The V oc varied by less than 5 mV during the day, along with small changes in the ll factor and the maximum power point voltage. This has practical advantages as it obviates the need for using a power tracking system to obtain the maximal output. By contrast the J sc was more strongly affected by the AM number and the hour of the day due to variations in the intensity and spectral distribution of the incident sunlight. Importantly the PCE varied over a much smaller range than the photocurrent remaining between 8 and 9% during most of the day except for the rst and the last measurement where the AM number was over 4. Due to the long path through the atmosphere the solar emission shis here to the red and near IR reducing its spectral overlap with the sensitizer and hence the photocurrent.

Summary and conclusions
We show that the molecular gelator CTP is very effective in solidifying the high-endurance electrolyte Z-988 composed of a mixture of sulfolane with an ionic liquid at levels as low as 2%. Solidication has only a small impact on the key photovoltaic performance parameters. The open circuit voltage (V oc ) and ll factor (FF) remain practically unaffected while the photocurrent drops due to an increase in the Warburg diffusion resistance for triiodide ions in the presence of CTP. The rst outdoor experiments performed in Jeddah Saudi Arabia gave very promising results. Importantly power conversion efficiencies exceed those measured under standard AM 1.5 conditions, the maximum PCE of 8.9% being reached at around noontime when the temperature reached 37 C. The V oc varied by less than 5 mV during the day, along with small changes in the ll factor and the maximum power point voltage. This has practical advantages as it obviates the need for using power tracking systems to obtain the maximal output from the PV panel. As expected at high air mass numbers the photocurrent decreased substantially due to the red shi of the solar emission spectrum. Overall the data are of great practical value and forebode well for the large-scale use of dye-sensitized solar cells in desert climates.  Table 2 List of outdoor conditions and key photovoltaic performance parameters for a cell measured in Jeddah Saudi Arabia with a DSC employing the C106 ruthenium complex as a sensitizer, the Z988 electrolyte and 2% CTP gelator. P in ¼ incident power intensity; J sc ¼ short circuit current density; V MP ¼ photovoltage at the maximum output power point; J MP ¼ current density at the maximum output power point; and MP ¼ the maximum output power point The impedance measurement of the dummy devices for the determination of the diffusion coefficient of I 3 À (D I3 À ) was done at 0 V and the data analysed on the basis of the Randles circuit which includes a series resistance, a charge transfer resistance, the double layer capacitance and the element used to extract the diffusion coefficient, the Warburg diffusion element. Applying this equivalent circuit to the impedance spectra yields the Warburg parameter. Aer the tting procedure the diffusion coefficient D I3 À is calculated by the relationship D I3 À ¼ d 2 /W where d is half of the distance between the two Pt electrodes and W the Warburg coefficient extracted from the analysis of the Nyquist plot (20 mm was used instead of the 25 mm since due to the hot melt process the thickness is reduced). 21 The measurements for the determination of D I3 À by cyclic voltammetry were performed in a potential range of À0.9 V to +0.9 V. The limiting current density (J lim ) is obtained by dividing the limiting current (I lim ) by the surface exposed to the electrolyte. The diffusion coefficient is calculated by using the above-mentioned formula, where l is the distance between the two electrodes, n is the number of exchanged electrons, e is the elementary charge, c is the triiodide concentration and N av is the Avogadro number. 21 The impedance spectra of the DSC devices were recorded at potentials varying from 0 V to V oc at frequencies ranging from 1 MHz to 0.1 Hz, with a sinusoidal potential pertubation of 10 mV. The photoanode (TiO 2 ) was used as the working electrode and the Pt counter electrode (CE) was used as both the auxiliary electrode and the reference electrode. These obtained spectra were tted with the transmission line model. [18][19][20] .