Achievement of over 1.4 V photovoltage in a dye-sensitized solar cell by the application of a silyl-anchor coumarin dye

A dye-sensitized solar cell (DSSC) fabricated by using a novel silyl-anchor coumarin dye with alkyl-chain substitutes, a Br3−/Br− redox electrolyte solution containing water, and a Mg2+-doped anatase-TiO2 electrode with twofold surface modification by MgO and Al2O3 exhibited an open-circuit photovoltage over 1.4 V, demonstrating the possibility of DSSCs as practical photovoltaic devices.


Results and Discussion
In ADEKA-3, the introduction of alkyl-chain substituents was performed by linking hexyl-thiophene rings to the coumarin moiety. A methyl group was also added to the coumarin moiety to prevent the co-planar arrangement of the coumarin moiety and the thiophene ring, which will produce a heightening of the HOMO level of the dye through the extension of the coumarin π system to the thiophene ring. The alkoxysilyl coumarin dye of ADEKA-3 exhibited similar UV-visible absorption spectra to SFD-5 in solutions, and a major absorption band of ADEKA-3 solution assignable to the π -π * transition was observed in visible region between 350 and 500 nm. The maximum molar absorption coefficient (ε max ) at λ max was evaluated to be 48,700 at 415 nm (Fig. S4). The energy levels of HOMO and LUMO were estimated to be 1.18 V and − 1.12 V vs. NHE, respectively, for ADEKA-3 (Table S1). The HOMO level is more positive than the redox potential of ~0.9 V vs. NHE of the Br 3 − /Br − redox 10,13,22 , thus providing thermodynamic driving force for the dye regeneration reaction by the electron transfer from the Br 3 − /Br − redox mediator to the oxidized dye formed through the light-excited-electron injection to the TiO 2 electrode. The reliability of the relative positions of HOMO and LUMO levels were supported by the molecular orbital calculations for SFD-5 and ADEKA-3 (Figs S7 and S8).
The Mg-doped TiO 2 crystalline nanoparticles with an increased Mg/Ti atomic ratio to 0.20 were synthesized by the solvothermal method 13,21 . As the reference to the Mg-doped TiO 2 , anatase-TiO 2 nanoparticles without Mg-doping (undoped-TiO 2 ) were also synthesized by the same method. The single phase of anatase structure was confirmed for the Mg-doped TiO 2 crystalline nanoparticles by X-ray diffraction (XRD) experiments, and the particle size was estimated to be ~25 nm by using the Scherrer equation (Figs S9 and S10). The band gap of the Mg-doped TiO 2 was evaluated to be 3.4 eV by a Tauc plot of the diffuse reflectance spectrum (Figs S11 and S12), which is 0.2 eV larger than that of the undoped-TiO 2 consistently with the negative shift of the E C.B. by the Mg-doping 21,23 . Energy levels of the Mg-doped TiO 2 , ADEKA-3, and the Br 3 − /Br − redox mediator are drawn schematically in Fig. 2, which shows the suitability of ADEKA-3 as the sensitizing dye in the cell system with the Mg-doped TiO 2 and the Br 3 − /Br − redox mediator. The results of J-V measurements performed in this work are listed in Table 1 and shown in Fig. S13. The measurements were performed under AM-1.5G one sun illumination (100 mW cm −2 ). To check the performance of ADEKA-3 as a photosensitizer, J-V measurements were carried out for the cells sensitized by SFD-5 and ADEKA-3 as Entry 1 and 2, respectively, at 25 °C with using the TiO 2 electrode without Mg-doping and a Br 3 − / Br − redox electrolyte solution (Electrolyte A: See Methods for the compositions of electrolytes used in this work.). The cell sensitized by ADEKA-3 exhibited 0.1 V higher V oc and smaller short-circuit photocurrent density (J sc ) than the cell sensitized by SFD-5, and light-to-electric energy conversion efficiencies (η) of these cells were almost the same. Since the dark current was smaller in the ADEKA-3-sensitized cell than the SFD-5-sensitized cell (Fig.  S14), the increase of V oc in the ADEKA-3-sensitized cell is considered to be brought by the hexyl-chain substituents introduced in ADEKA-3, which are working as the suppressor for preventing the back electron transfer from the TiO 2 electrode to the Br 3 − /Br − redox electrolyte by covering the naked surface of the TiO 2 electrode between the adsorbed dye molecules [15][16][17][18] . In ADEKA-3, the HOMO level is higher in energy than that of SFD-5 by 0.21 eV, and thus the energy gap between the HOMO level and the redox potential of the Br 3 − /Br − redox mediator is smaller than that for SFD-5 ( Fig. 2 and Table S1). The incident monochromatic photon-to-current conversion efficiencies (IPCE) were observed to tend to be lower in the ADEKA-3-sensitized cell than the SFD-5-sensitized cell (Fig. S15), and thus the smallness of the energy gap is considered as the reason for the lower J sc value in the ADEKA-3-sensitized cell, which produced the delay of the dye regeneration reaction proceeding through the electron transfer from the redox mediator in the electrolyte solution to the dye in the oxidized state 10,22 . Since ADEKA-3 was ascertained as an effective dye for producing high photovoltage, the Mg-doped TiO 2 electrode was applied to the cell sensitized by the dye as Entry 3. The cell exhibited the V oc of 1.23 V, which is about 20% higher than that of the cell using the TiO 2 electrode without Mg-doping. The increase of the V oc is considered to be due to the higher E C.B. of the Mg-doped TiO 2 than that of the TiO 2 without Mg-doping. For a further increment of the photovoltage in the cell, we examined surface modifications of the Mg-doped TiO 2 electrode by wide bandgap metal oxides of MgO and by Al 2 O 3 following the MgO modification (MgO + Al 2 O 3 ) as Entries 4 and 5, respectively. The surface modification by MgO was confirmed to be effective also in the present cell system in improving the photovoltage, and the improvement is understood as the result of the negative shift of the E C.B. of the Mg-doped TiO 2 by the MgO modification (Fig. S16) 20 . More efficient improvement was observed by the twofold surface modification with MgO and Al 2 O 3 . The Al 2 O 3 modification was confirmed not to affect the E C.B. of the MgO-modified Mg-doped TiO 2 by the UV-visible spectra (Fig. S16), and the modification is thought to form a blocking layer on the surface of the Mg-doped TiO 2 electrode suppressing the back electron transfer from the electrode to the redox mediator in the electrolyte solution 19,20 .
In DSSCs, photovoltage is known to be increased by the addition of compounds having coordination ability to the surface of TiO 2 electrodes, such as 4-tert-butylpyridine (TBP), to electrolyte solutions which shift the E C.B. negatively through the coordination. We examined the addition of 4-methylpyridine (MP) and 4-trimethylsilylpyridine (TMSP) to the Br 3 − /Br − redox electrolyte solution 24 , and prepared Electrolyte B with an experimentally optimized composition for high photovoltage. By using the electrolyte solution as Entry 6, the V oc was increased slightly and reached to 1.39 V. As an additive to the electrolyte solution for the improvement of the photovoltage, water is expected to be effective because of its high coordination ability owing to lone pairs on the oxygen atom and small molecular size [25][26][27] . However, the addition of water to electrolyte solutions causes the elimination of sensitizing dyes from the TiO 2 electrodes generally in the case of conventional carboxy-anchor dyes, and the application of electrolyte solutions containing water has been rather limited 28 . On the other hand, alkoxysilyl dyes chemisorb the TiO 2 electrodes by forming Si-O-Ti bonds through the condensation reaction between the alkoxysilyl groups and the hydroxy groups on the TiO 2 surface, and the dye adsorbed electrodes have quite high durability to solvents, e.g. nitrile, water, and mixture of them 7,29-31 . Thus, we attempted to use a Br 3 − /Br − redox electrolyte solution containing water with the concentration of 0.10 M (Electrolyte C) as Entry 7. By the addition of water to the electrolyte, the V oc was improved actually to 1.45 V. The addition of water also brought about a decrement of the photocurrent to the cell, but the η was still to be ~4% (Tables 1 and S2, and Figs 3 and S13).    And further, the V oc reached 1.50 V by lowering the cell temperature to 5 °C (Entry 8) as the result of a possible rise of the E C.B. and a deceleration of the back electron transfer reaction 32 . To the best of our knowledge, the observed V oc of 1.45 V at an ordinary temperature is the highest ever reported for DSSCs with a single-cell structure (Table S3) [7][8][9][10][11]13,14,17,21,33,34 .

Conclusions
We succeeded in producing the photovoltage over 1.4 V with a reasonably high conversion efficiency close to 4% in the DSSC by using the alkoxysilyl-anchor coumarin dye of ADEKA-3, the Mg-doped TiO 2 electrode with the twofold surface modification by MgO and Al 2 O 3 , and the Br 3 − /Br − redox electrolyte solution containing water. The observed V oc is higher than those of other types of solar cells (Table S4), and is comparable to that of a conventional dry cell 5 . The achievement of such a high photovoltage, which is owing to the surpassing property of a silyl-anchor dye as a sensitizing dye for DSSCs, demonstrates the possibility of DSSCs as practical photovoltaic devices.  (Fig. S17). The Mg-doped TiO 2 or TiO 2 electrode sensitized by SFD-5 or ADEKA-3, the counter electrode, and a polyethylene film spacer of 30 μ m thick were assembled, and one of the Br 3 − /Br − redox electrolyte solutions was injected into the space between the electrodes (Fig. S18).

Methods
Photovoltaic measurements. The photovoltaic performances of the fabricated DSSCs were assessed from the IPCE spectra and the J-V properties of the cells with maintaining the aperture area of the cells to be 1.00 × 1.00 cm 2 by the use of a square black shade mask. The IPCE spectra were obtained by using a monochromatic light source of SM-25 (Bunkoukeiki) and an electrometer of R8240 (Advantest) at 25 °C. The J-V properties were measured by using a solar simulator with Class AAA of OTENTO-SUN III (Bunkoukeiki) and a source meter of R6240A (Advantest) under the simulated sunlight irradiation of AM-1.5G one sun condition (100 mW cm −2 ) at 25 or 5 °C. The details were described in Supplementary Information. Molecular orbital calculation. We optimized the molecular structures and calculated the energy levels of frontier orbitals and others for the alkoxysilyl-anchor coumarin dyes on the Gaussian 09 program package by using a density functional theory (DFT) 36 . The details were described in Supplementary Information.