Effect of π-spacers and anchoring groups on the photovoltaic performances of ullazine-based dyes

Abstract

Three ullazine-based organic sensitizers (QD1, QD2 and QD3) have been designed, synthesized, and characterized for dye-sensitized solar cells (DSSCs). Ullazine possesses some attractive properties, such as a planar π-system to promote intensely the intramolecular charge transfer (ICT) between ullazine donor units and cyanoacrylic acid or carboxylic acid acceptor units. The tuning of π-spacers and anchoring groups led to variation of the photovoltaic performances. The DSSC based on QD1 with a cyanoacrylic acid and ethylene, obtained a PCE of 5.28%, which was higher than that of QD2 with a carboxylic acid and phenylethylene, and QD3 with a carboxylic acid and thiophene ethylene. The absorption spectra of QD2 and QD3 with the increasing of the length of the π-spacers are blue-shifted compared to that of QD1. The absorbed amount of QD1 was the highest of all, contributing to the highest J_sc of 12.28 mA cm⁻². The different V_oc values, in the order of QD1 (0.65 V) > QD2 (0.60 V) > QD3 (0.53 V), are caused by the charge recombination rates at the TiO₂/dye/electrolyte interface and the electron lifetimes.

---

Introduction

Dye-sensitized solar cells (DSSCs) have attracted considerable attention since the pioneering report in 1991.¹ As a promising alternative to the conventional silicon based solar cells, DSSCs are endowed with transparency, flexibility, low production cost, ease of synthesis and comparatively high power conversion efficiency (PCE) and so on.^2–6 The visible light collection ability of the sensitizers has great influence on the PCEs of DSSCs. In general, sensitizers can be metal complexes such as ruthenium and zinc or metal free organic dyes.^7–10 Ru-complexes have dominated the DSSCs market, with a power conversion efficiency higher than 11% to date.^11,12 A dye based on zinc-porphyrin, SM315 reported by Grätzel et al., gave a record-high PCE of 13.0%.¹³ However, ruthenium is rare and expensive. Besides, its molar extinction coefficient is rather low. Meanwhile, metal-free organic dyes have emerged as competitive alternatives in virtue of relatively low cost, ease of structural tailor, high molar extinction coefficients, governing absorption wavelength and environment-friendly properties.^14–17 Therefore, recent years have witnessed the strong emergence of research on metal-free organic dyes. Most of the metal-free organic dyes consist of a donor–π–acceptor (D–π–A) framework with relatively high PCEs. These D–π–A systems exhibited generous and strong absorption in the visible region, good stability and excellent performances. In 2015, the DSSC based on co-photosensitized with a carboxy-anchor organic dye LEG4 reported by Minoru Hanaya et al., exhibited a PCE of over 14% under one sun illumination, which can effectively enhance the electron injection from the light-excited dyes to the TiO₂ electrodes.¹⁸ In addition, the DSSC based on C275 with an N-annulated indenoperylene electron donor, triple bond as the π-spacer, benzothiadiazolylbenzoic electron-acceptor, achieved a PCE of 12.5% under irradiance of 100 mW cm⁻² AM 1.5G sunlight.¹⁹ However, in order to achieve the purpose of commercial and industrial applications, it still has tremendous work to do.

Ullazine was first reported in 1983,^20,21 and was first used to develop organic dyes for DSSCs with PCE as high as 8.4%.²² In addition, ullazine possesses some attractive properties, such as a planar π-system to promote intense ICT, both strong donating and electron-accepting properties, and structural tunability.²² In this study, three new based on ullazine-dyes (QD1–QD3) have been designed, synthesized and characterized. According to the relevant literature, the DSSC based on ullazine with hexyl side chains, a cyanoacrylic acid and ethylene gave a good photovoltaic performance.²² Therefore, we designed and synthesized QD1 with 2-ethylhexyloxy side chains, a cyanoacrylic acid and ethylene which could provide steric hindrance and increase intermolecular distances so that it could reduce dye aggregation availably.²³ In QD2 and QD3, we introduced phenylethylene and thiophene ethylene respectively as π-spacers. Compared with benzene, the thiophene can offer more effective conjugation and lower the energy of the charge transfer transition.²⁴ As for electron acceptors namely anchoring groups, QD1 with a strong electron withdrawing cyanoacrylic acid could pull the electron from the donor, narrowing the band gap through the mixing of donor–acceptor orbitals, which promoted absorption in the longer wavelength region and was considered to be facilitate electron injection into TiO₂, just like several D–π–A molecules.^25–27 In QD2 and QD3, we chose a weak acceptor carboxylic acid. On account of gaining further insight into π-spacers and anchoring groups especially, we have taken up this work aiming at studying the effect of the π-spacers and anchoring groups on the photovoltaic performances. Meanwhile, the photophysical, electrochemical properties and photovoltaic performances of this system have been investigated in details.

Experimental section

Materials

All the chemicals were purchased from Alfa Aesar and Chem Greatwall Chemical Company (Wuhan, China) and used without further purification. Tetrahydrofuran (THF) were dried and distilled over sodium and benzophenone. N,N-Dimethylformamide (DMF) was dried over and distilled from CaH₂ under an atmosphere of dry nitrogen. Phosphorus oxychloride and 1,2-dichloroethane were atmospheric distillation. All chromatographic separations were carried out on silica gel (200–300 mesh). All the other solvents and chemicals used in this work were analytical grade and used without further purification. Methyl-4-((diethoxyphosphoryl)methyl)benzoate (12) and methyl-5-((diethoxyphosphoryl)methyl)thiophene-2-carboxylate (17) were synthesized according to the procedure reported in the literatures.²⁸

Instruments and characterizations

¹H NMR and ¹³C NMR spectra were measured on a Bruker Avance 400 instrument. MALDI-TOF mass spectrometric measurements were performed on Bruker Autoflex III. Elemental analysis were measured on an Elementar Vario EL III. UV-Visible spectra of the dyes in diluted solutions and on TiO₂ thin-films were measured on a Perkin-Elmer Lambda 25 spectrometer. The photoluminescence (PL) spectra were obtained using Perkin-Elmer LS-50 luminescence spectrometer. Electrochemical redox potentials were obtained by cyclic voltammetry (CV) using a three-electrode configuration and an electrochemistry workstation. The working electrode was a glassy carbon electrode; the counter electrode was a Pt electrode, and saturated calomel electrode (SCE) was used as reference electrode. Tetrabutyl ammonium perchlorate (TBAP) 0.1 M was used as supporting electrolyte in nitrogen-purged anhydrous CHCl₃. Ferrocene/ferrocenium redox couple was used for potential calibration.²⁹

Fabrication and characterization of DSSC devices

Fluorine-doped SnO₂ conducting glass (FTO) was cleaned, immersed in aqueous 40 mM TiCl₄ solution at 70 °C for 30 min, then washed with water and ethanol, sintered at 450 °C for 30 min. The TiO₂ suspension was prepared from P25 (Degussa AG, Germany) and 1 wt% magnesium acetate solution according to the reported literature.^30,31 Then the paste was deposited onto the FTO glass by blade coating. Subsequently, a 3 μm thick 200 nm particle sized TiO₂ scattering layer was deposited again by blade coating. The TiO₂-coated FTO glass was sintered at 450 °C for 30 min, then treated with TiCl₄ solution at 70 °C for 30 min. Finally, it was calcined at 450 °C for 30 min again. After cooling to room temperature, it was immersed into 0.5 mmol L⁻¹ dye solution with or without chenodeoxycholic acid (CDCA) (0.5 mM) for 12 h in the dark, in THF : ethanol = 1 : 4, which is the optimal condition. The sensitized electrode was rinsed with ethanol, and then dried. A drop of electrolyte was deposited onto the surface of the electrode and a Pt foil counter electrode was clipped onto the top of the TiO₂ electrode to assemble a DSSC to measure their photovoltaic performances. The electrolyte comprised 0.5 M LiI, 0.05 M I₂, and 0.5 M 4-tertbutylpyridine (TBP) in 3-methoxypropionitrile and the efficient irradiated area of the cell was 0.2 cm². The current density voltage (J–V) curves were measured by a Keithley 2602 Source Meter under 100 mW cm⁻² standard AM 1.5G spectrum using a Sol 3A Oriel solar simulator. The incident light intensity was calibrated using a standard Si solar cell. The power conversion efficiency (η) of the DSSC is calculated from short-circuit photocurrent (J_sc), the open-circuit photovoltage (V_oc), the fill factor (FF) and the intensity of the incident light (P_in) in line with the following equation:

The measurement of monochromatic incident photo-to-current conversion efficiency (IPCE) for the solar cell was also detected with a Zolix DCS300PA data acquisition system.

Synthesis

The synthetic routes of the compounds and three dyes are shown in Scheme 1. The detailed synthetic processes are as follows.

Scheme 1 Synthetic routes and molecular structures of the three dyes QD1, QD2 and QD3.

1-Bromo-4-((2-ethylhexyl)oxy)benzene (2). To a 250 mL two necked round bottom flask was added 1,4-bromophenol (10.00 g, 57.80 mmol), K₂CO₃ (11.96 g, 86.71 mmol), and N,N-dimethylformamide (140 mL), and then warmed to 50 °C under an argon atmosphere. After 30 minutes, 2-ethylhexyl bromide (13.40 g, 34.68 mmol), was added slowly into it, and then warmed to 80 °C. After being stirred for 24 h, the product was extracted with dichloromethane. The organic phase was rinsed with water, dried over anhydrous MgSO₄, filtered, and finally removed solvent by rotary evaporation. The crude product was purified on column chromatography using silica gel using petroleum ether as eluent to give compound 2, as a neutral liquid (14.83 g, 90%). ¹H NMR (400 MHz, CDCl₃, δ/ppm): 7.36–7.34 (d, J = 8.0 Hz, 2H), 6.79–6.77 (d, J = 8.0 Hz, 2H), 3.80–3.79 (d, J = 4.0 Hz, 2H), 1.73–1.67 (m, 1H), 1.49–1.31 (m, 8H), 0.93–0.90 (t, J = 6.0 Hz, 6H).

((4-((2-Ethylhexyl)oxy)phenyl)ethynyl)trimethylsilane (3). To a 100 mL two necked round bottom flask was added compound 2 (10.00 g, 35.10 mmol), bis(triphenylphosphine)palladium(II) chloride (PdCl₂(PPh₃)₂, 0.20 g, 0.28 mmol), cuprous iodide, (CuI, 0.15 g, 0.79 mmol), triphenylphosphine (0.30 mg, 1.15 mmol), N-(1-methylethyl)-2-propanamine (50 mL), and then warmed to 80 °C. Ethynyltrimethylsilane (5.44 mL, 38.57 mmol) was added dropwise. After being stirred for 24 h, the product was extracted with dichloromethane. The organic phase was rinsed with water, dried over anhydrous MgSO₄, filtered, and finally removed solvent by rotary evaporation. The crude product was purified on column chromatography using silica gel using petroleum ether as eluent to give compound 3, as a light yellow liquid (7.42 g, 70%). ¹H NMR (400 MHz, CDCl₃, δ/ppm): 7.40–7.38 (d, J = 8.0 Hz, 2H), 6.82–6.80 (d, J = 8.0 Hz, 2H), 3.84–3.82 (d, J = 8.0 Hz, 2H), 1.73–1.68 (m, 1H), 1.53–1.26 (m, 8H), 0.94–0.90 (t, J = 8.0 Hz, 6H), 0.30–0.21 (s, 9H).

1-((2-Ethylhexyl)oxy)-4-ethynylbenzene (4). To a 250 mL round bottom flask was added compound 3 (7.42 g, 24.53 mmol), sodium hydroxide (1.96 g, 49.05 mmol), tetrahydrofuran 50 mL, methanol 50 mL at room temperature under an argon atmosphere. After being stirred for 12 h, the product was extracted with dichloromethane. The organic phase was rinsed with water, dried over anhydrous MgSO₄, filtered, and finally removed solvent by rotary evaporation. The crude product was purified on column chromatography using silica gel using petroleum ether as eluent to give compound 4, as a light yellow liquid (4.80 g, 85%). ¹H NMR (400 MHz, CDCl₃, δ/ppm): 7.42–7.40 (d, J = 8.0 Hz, 2H), 6.84–6.82 (d, J = 8.0 Hz, 2H), 3.84–3.83 (d, J = 4.0 Hz, 2H), 2.98 (s, 1H), 1.75–1.69 (m, 1H), 1.51–1.31 (m, 8H), 0.94–0.90 (t, J = 8.0 Hz, 6H).

1-(2,6-Dibromophenyl)-1H-pyrrole (6). Amounts of 2,6-dibromoaniline (6.00 g, 23.91 mmol), dichloroethane (24 mL), acetic acid (24 mL), and 2,5-dimethoxytetrahydrofuran (7.58 g, 57.38 mmol) were refluxed at 105 °C under an argon atmosphere. During this time, the solution slowly changed colors from clear to yellowish-brown. After being stirred for 10 h, the product was extracted with dichloromethane. The organic phase was rinsed with water, and then rinsed with saturated aqueous potassium carbonate (K₂CO₃), dried over anhydrous magnesium sulphate (MgSO₄), filtered, and finally removed solvent by rotary evaporation. The crude product was purified on column chromatography using silica gel using petroleum ether as eluent to give compound 6, as a white solid (6.48 g, 21.53 mmol, 90%). ¹H NMR (400 MHz, CDCl₃, δ/ppm): 7.68 (d, J = 8.0 Hz, 2H), 7.15 (t, J = 8.4 Hz, 1H), 6.68 (d, J = 2.0 Hz, 2H), 6.42 (d, J = 2.0 Hz, 2H).

1-(2,6-Bis((4-((2-ethylhexyl)oxy)phenyl)ethynyl)phenyl)-1H-pyrrole (7). To a 100 mL two necked round bottom flask was added compound 6 (3.00 g, 9.97 mmol), bis(acetonitrile)palladium(II) chloride (PdCl₂(CH₃CN)₂, 155.66 mg, 0.60 mmol), cuprous iodide (CuI, 76.18 mg, 0.40 mmol), tri-tert-butylphosphine solution (242.78 mg, 1.20 mmol) and compound 4 (5.73 g, 24.92 mmol) under an argon atmosphere. Then diisopropylamine (10 mL) was added into it. After being stirred for 18 h at room temperature, the product was extracted with dichloromethane. The organic phase was rinsed with 10% H₃PO₄ two times, 5% K₂CO₃ one time, dried over anhydrous MgSO₄, filtered, and finally removed solvent by rotary evaporation. The crude product was purified on column chromatography using silica gel using petroleum ether/dichloromethane = 10 : 1 as eluent to give compound 7, as a light yellow viscous liquid (4.78 g, 7.97 mmol, 80%). ¹H NMR (400 MHz, CDCl₃, δ/ppm): 7.53–7.51 (d, J = 8.0 Hz, 2H), 7.30–7.27 (m, 5H), 7.06 (t, J = 4.0 Hz, 2H), 6.82–6.80 (d, J = 8.0 Hz, 4H), 6.36 (t, J = 4.0 Hz, 2H), 3.83–3.82 (d, J = 4.0 Hz, 4H), 1.72–1.70 (m, 2H), 1.52–1.31 (m, 16H), 0.93–0.90 (t, J = 6.0 Hz, 12H). MALDITOF MS (C₄₂H₄₉NO₂) m/z: calcd for 599.840; found 600.410.

3,9-Bis(4-((2-ethylhexyl)oxy)phenyl)indolizino[6,5,4,3-ija]quinoline (8). To a 100 mL round bottom flask was added compound 7 (3.00 g, 5.00 mmol), toluene (35 mL), indium trichloride (InCl₃, 663.00 mg, 3.00 mmol), and then warmed to 100 °C under an argon atmosphere. After being stirred for 24 h, the product was extracted with dichloromethane. The organic phase was rinsed with water, dried over anhydrous MgSO₄, filtered, and finally removed solvent by rotary evaporation. The crude product was purified on column chromatography using silica gel using petroleum ether/dichloromethane = 8 : 1 as eluent to give compound 8, as a yellow viscous liquid (2.40 g, 4.00 mmol, 80%). ¹H NMR (400 MHz, CDCl₃, δ/ppm): 7.73–7.71 (d, J = 8.0 Hz, 4H), 7.46–7.39 (m, 3H), 7.18 (s, 2H), 7.08–7.04 (m, 6H), 3.94–3.93 (d, J = 4.0 Hz, 4H), 1.80–1.75 (m, 2H), 1.51–1.36 (m, 16H), 0.88–0.86 (t, J = 4.0 Hz, 12H). MALDITOF MS (C₄₂H₄₉NO₂) m/z: calcd for 599.840; found 599.529.

3,9-Bis(4-((2-ethylhexyl)oxy)phenyl)indolizino[6,5,4,3-ija]quinoline-5-carbaldehyde (9). To a 250 mL two necked round bottom flask was added compound 8 (2.20 g, 3.67 mmol), dichloroethane (50 mL) under an argon atmosphere at room temperature. And then it was added anhydrous DMF (0.85 mL, 11.00 mmol). After 30 minutes, POCl₃ (0.82 mL, 8.80 mmol) was added slowly into it. After being stirred for 2.5 h at room temperature, sodium acetate solution (100 mL) was added to hydrolyze the reaction for 8 h. The product was extracted with chloroform. The organic phase was rinsed with water, dried over anhydrous MgSO₄, filtered, and finally removed solvent by rotary evaporation. The crude product was purified on column chromatography using silica gel using petroleum ether/dichloromethane = 3 : 2 as eluent to give compound 9, as an orange solid (1.38 g, 2.20 mmol, 60%). It was worth mentioning that compound 10 was formed as the major side product during the formylation reaction of compound 9. ¹H NMR (400 MHz, CDCl₃, δ/ppm): 10.28 (s, 1H), 9.07 (s, 1H), 7.89–7.87 (d, J = 8.0 Hz, 1H), 7.84–7.82 (d, J = 8.0 Hz, 2H), 7.87–7.86 (d, J = 4.0 Hz, 2H), 7.59–7.57 (d, J = 8.0 Hz, 2H) 7.47 (s, 1H), 7.43 (s, 1H), 7.11–7.09 (d, J = 8.0 Hz, 4H), 3.96–3.95 (d, J = 4.0 Hz, 4H), 1.80–1.77 (m, 2H), 1.53–1.37 (m, 16H), 1.00–0.94 (t, J = 12.0 Hz, 12H). ¹³C NMR (100 MHz, CDCl₃, δ/ppm): 191.41, 160.06, 159.98, 135.92, 135.51, 131.76, 130.91, 130.43, 130.28, 130.04, 129.99, 129.82, 127.58, 127.14, 125.78, 121.82, 118.42, 117.11, 116.66, 114.91, 114.85, 109.43, 108.90, 70.72, 39.48, 30.62, 29.19, 23.96, 23.16, 23.15, 14.19, 11.22. MALDITOF MS (C₄₃H₄₉NO₃) m/z: calcd for 627.850; found 627.362.

(E)-3-(3,9-Bis(4-((2-ethylhexyl)oxy)phenyl)indolizino[6,5,4,3-ija]quinolin-5-yl)-2-cyanoacrylic acid (QD1). To a 100 mL two necked round bottom flask was added compound 9 (230.40 mg, 0.37 mmol), cyanoacetic acid (308.38 mg, 3.67 mmol), and piperidine (749.75 mg, 8.81 mmol), CHCl₃ (20 mL), and then were refluxed at 80 °C under an argon atmosphere. After being stirred for 12 h, the solution was poured into a diluted acetic acid (CH₃COOH, 20 mL) solution and stirred for 8 h. The product was extracted with chloroform. The organic phase was rinsed with water, dried over anhydrous MgSO₄, filtered, and finally removed solvent by rotary evaporation. The crude product was purified on column chromatography using silica gel using chloroform/methanol = 80 : 1 as eluent to give compound QD1, as a dark violet solid (198.71 mg, 286.16 mmol, 78%). ¹H NMR (400 MHz, d₆-DMSO, δ/ppm): 8.63 (s, 1H), 7.91–7.89 (d, J = 8.0 Hz, 1H), 7.68–7.60 (m, 4H), 7.40–7.39 (d, J = 4.0 Hz, 1H), 7.32 (s, 3H), 7.09–6.94 (d, J = 5.6 Hz, 6H), 3.94–3.93 (d, J = 4.0 Hz, 4H), 1.73 (m, 2H), 1.45–1.22 (m, 16H), 0.96–0.91 (t, J = 10.0 Hz, 12H). ¹³C NMR (100 MHz, DMSO, δ/ppm): 159.21, 129.47, 114.41, 70.70, 39.52, 30.57, 29.71, 29.19, 23.90, 23.11, 14.13, 11.17. MALDITOF MS (C₄₆H₅₀N₂O₄) m/z: calcd for 694.900; found 694.511. Melting point: 230–233 °C. Anal. calcd for C₄₆H₅₀N₂O₄: C, 79.51; H, 7.25; N, 4.03. Found: C, 80.26; H, 7.26; N, 3.97.

(E)-Methyl-4-(2-(3,9-bis(4-((2-ethylhexyl)oxy)phenyl)indolizino[6,5,4,3-ija]quinolin-5-yl)vinyl)benzoate (13). Compound 12 (386.66 mg, 1.35 mmol) and 9 (565.50 mg, 0.90 mmol) were dissolved in THF (20 mL) and the solution was stirred under the ice bath for 30 min under an argon atmosphere. Then potassium tertbutoxide (151.56 mg, 1.35 mmol) was dissolved in THF (20 mL) and added dropwise to the solution. The reaction mixture was stirred for 5 h at room temperature, and then heated to 50 °C for 24 h. The product was extracted with chloroform. The organic phase was rinsed with water, dried over anhydrous MgSO₄, filtered, and finally removed solvent by rotary evaporation. The crude product was purified on column chromatography using silica gel using petroleum ether/dichloromethane = 3 : 2 as eluent to give compound 13, as an orange red solid (448.40 mg, 0.59 mmol, 65%). ¹H NMR (400 MHZ, CDCl₃, δ/ppm): 7.96–7.94 (d, 2H), 7.76–7.73 (m, 6H), 7.54–7.52 (d, 2H), 7.41 (s, 1H), 7.36–7.34 (d, 1H), 7.13–6.97 (m, 8H), 3.87–3.86 (m, 7H), 1.73–1.70 (m, 2H), 1.52–1.39 (m, 16H), 0.93–0.87 (t, 12H). ¹³C NMR (100 MHz, CDCl₃, δ/ppm): 166.80, 159.68, 159.58, 142.26, 133.03, 132.68, 132.02, 130.90, 130.54, 129.85, 129.37, 129.27, 128.25, 127.27, 126.83, 126.66, 126.02, 125.62, 124.83, 123.54, 120.84, 118.84, 118.77, 114.77, 114.68, 106.64, 106.58, 70.65, 51.95, 39.46, 30.60, 29.16, 23.95, 23.12, 14.17, 11.21. MALDITOF MS (C₅₂H₅₇NO₄) m/z: calcd for 760.010; found 759.516.

(E)-4-(2-(3,9-Bis(4-((2-ethylhexyl)oxy)phenyl)indolizino[6,5,4,3-ija]quinolin-5-yl)vinyl)benzoic acid (QD2). Compound 13 (410.00 mg, 0.54 mmol) and potassium hydroxide (303.00 mg, 5.40 mmol) were dissolved in THF (30 mL) and H₂O (10 mL) under an argon atmosphere, and then refluxed for 12 h. After cooling to room temperature, acetic acid was added into it, and stirred for 5 h. The product was extracted with chloroform. The organic phase was rinsed with water, dried over anhydrous MgSO₄, filtered, and finally removed solvent by rotary evaporation. The crude product was purified on column chromatography using silica gel using chloroform/methanol = 150 : 1 as eluent to give compound QD2, as an orange yellow solid (301.84 mg, 0.4 mmol, 75%). ¹H NMR (400 MHz, CDCl₃, δ/ppm): 8.13–8.11 (d, 2H), 7.92–7.74 (m, 6H), 7.69–7.67 (d, 2H), 7.54–7.49 (t, 2H), 7.27 (s, 1H), 7.24 (s, 1H), 7.14–7.07 (m, 6H), 3.87–3.86 (m, 4H), 1.84–1.79 (m, 2H), 1.58–1.27 (m, 16H), 0.99–0.95 (t, 12H). ¹³C NMR (100 MHz, DMSO, δ/ppm): 171.66, 159.72, 159.57, 142.86, 133.17, 132.82, 132.19, 130.98, 130.57, 130.50, 129.44, 127.58, 127.36, 127.26, 126.78, 126.04, 125.74, 124.87, 123.71, 118.89, 118.81, 114.83, 114.67, 106.58, 70.72, 39.50, 39.47, 31.96, 30.63, 30.56, 29.81, 29.73, 29.63, 29.55, 29.35, 29.19, 29.16, 27.25, 23.96, 23.90, 23.14, 23.12, 22.72, 14.15, 11.20, 11.18. MALDITOF MS (C₅₁H₅₅ NO₄) m/z: calcd for 745.990; found 745.458. Melting point: 221–223 °C. Anal. calcd for C₅₁H₅₅NO₄: C, 82.11; H, 7.43; N, 1.88. Found: C, 81.89; H, 7.53; N, 1.76.

(E)-Methyl-5-(2-(3,9-bis(4-((2-ethylhexyl)oxy)phenyl)indolizino[6,5,4,3-ija]quinolin-5-yl)vinyl)thiophene-2-carboxylate (18). The synthetic procedure for compound 18 was similar to that for compound 13, except that compound 17 (409.21 mg, 1.40 mmol) was used instead of compound 12. The crude product was purified on column chromatography using silica gel using petroleum ether/dichloromethane = 3 : 2 as eluent to give compound 18, as an orange red solid (510.00 mg, 0.67 mmol, 60%). ¹H NMR (400 MHz, CDCl₃, δ/ppm): 7.76–7.67 (m, 7H), 7.48–7.42 (m, 2H), 7.30 (s, 1H), 7.23–7.21 (d, 1H), 7.14–7.05 (m, 7H), 3.96–3.90 (m, 7H), 1.82–1.77 (m, 2H), 1.54–1.26 (m, 16H), 0.99–0.93 (t, 12H). ¹³C NMR (100 MHz, CDCl₃, δ/ppm): 166.67, 159.81, 159.70, 150.47, 134.20, 133.59, 133.16, 132.32, 130.87, 130.60, 130.48, 129.47, 129.36, 127.58, 127.28, 126.96, 126.04, 125.66, 124.33, 123.81, 121.00, 120.82, 119.00, 114.89, 114.79, 106.94, 106.82, 70.72, 52.12, 39.48, 30.63, 29.17, 23.97, 23.13, 14.17, 11.19. MALDITOF MS (C₅₀H₅₅NO₄S) m/z: calcd for 766.040; found 765.555.

(E)-5-(2-(3,9-Bis(4-((2-ethylhexyl)oxy)phenyl)indolizino[6,5,4,3-ija]quinolin-5-yl)vinyl)thiophene-2-carboxylic acid (QD3). The synthetic procedure for compound QD3 was similar to that for compound QD2, except that compound 18 (418.00 mg, 0.55 mmol) was used instead of compound 13. The crude product was purified on column chromatography using silica gel using chloroform/methanol = 100 : 1 as eluent to give compound QD3, as a dark red solid (308.32 mg, 0.41 mmol, 75%). ¹H NMR (400 MHz, CDCl₃, δ/ppm): 7.78–7.69 (m, 7H), 7.46 (s, 1H), 7.40–7.38 (d, 1H), 7.28–7.24 (d, 1H), 7.19 (s, 1H), 7.13–7.04 (m, 7H), 3.97–3.95 (d, 4H), 1.82–1.76 (m, 2H), 1.60–1.26 (m, 16H), 0.99–0.96 (t, 12H). ¹³C NMR (100 MHz, DMSO, δ/ppm): 159.62, 159.42, 132.54, 132.07, 131.51, 130.79, 130.68, 129.29, 126.94, 126.45, 125.15, 124.07, 120.28, 118.70, 118.10, 114.77, 114.55, 106.27, 70.67, 39.55, 39.43, 30.62, 30.56, 29.75, 29.42, 29.22, 29.13, 23.96, 23.90, 23.16, 23.10, 22.75, 14.18, 11.22. MALDITOF MS (C₄₉H₅₃NO₄S) m/z: calcd for 752.010; found 751.358. Melting point: 172–174 °C. Anal. calcd for C₄₉H₅₃NO₄S: C, 78.26; H, 7.10; N, 1.86; S, 4.26. Found: C, 78.57; H, 7.16; N, 1.77; S, 4.13.

Results and discussion

Synthesis

The detailed synthetic routes are provided in Scheme 1. The compound 3 and 7 were synthesized via Sonogashira reaction.³² The compound 9 was synthesized by Vilsmeier reaction. The compound 13 and 18 were gained by Witting–Honor reaction.³³ The target dye QD1 was synthesized via Knoevenagel condensation with cyanoacetic acid in the presence of piperidine. In all reactions, the synthesis of compound 7 is the most difficult, which requires strict feeding sequence and rigorous vacuum. The target dyes QD2 and QD3 were synthesized by ester in alkaline conditions of hydrolysis reactions. The structures of the dyes QD1, QD2 and QD3 were verified by ¹H NMR, ¹³C NMR and elemental analysis.

Photophysical properties

The photophysical properties of the QD1, QD2 and QD3 dyes were investigated by UV-Vis absorption spectra in diluted (10⁻⁵ M) chloroform solution in Fig. 1(a), and the corresponding data were shown in Table 1. The shorter wavelength absorption peaks at ca. 300 nm in ultraviolet region, aroused from π–π* transitions.³⁴ The longer wavelength absorption bands at ca. 400–600 nm could be assigned to the intermolecular charge transfer (ICT) between ullazine donor units and the cyanoacrylic acid or the carboxylic acid acceptor units. The maximum absorption wavelengths (λ_max) of QD1, QD2 and QD3 appeared at 530, 480 and 490 nm, and the corresponding molar extinction coefficients (ε) were 2.42 × 10⁴, 3.18 × 10⁴, 2.34 × 10⁴ M⁻¹ cm⁻¹ respectively. Compared with QD1, the ICT absorption peaks of QD2 and QD3 displayed blue-shifts of 50 and 40 nm, respectively. Maybe they were caused by the introducing of double bond, benzene, and thiophene as π-spacers, and a weak acceptor carboxylic acid as anchoring groups. In addition, the cyanoacrylic acid has stronger electron withdrawing ability than the carboxylic acid, which could effectively pull electron from the donor unit, narrow the band gap. The electron delocalization of QD1 was better than that of QD2 and QD3, which led to a higher ICT absorption peak. The thiophene in QD3 can provide more effective conjugation and lower the energy of the charge transfer transition compared with the benzene in QD2 which contributed to a higher ICT absorption peak.³⁵ As above, the absorption onset of QD1 is broadened to 680 nm, in comparison with QD2 (570 nm) and QD3 (609 nm).

Fig. 1 Absorption spectra of the three dyes (a) in diluted CHCl₃ solutions (10⁻⁵ M) and (b) on TiO₂ thin-films.

Table 1 Maximum absorption and emission data of the three dyes

Dyes	λabs^a/nm (ε/×10^4 M^−1 cm^−1)	λmax^b/nm (on TiO2)	λem^c/(nm)	Γ^d/(mol cm^−2)
a Maximum absorption in CHCl3 solution (10^−5 M) at 25 °C.b Maximum absorption on TiO2 film.c Maximum emission of the dyes in CHCl3 solution (10^−5 M).d Amount of the dyes adsorbed on TiO2 films.
QD1	530 (2.42)	520	638	5.02 × 10^−8
QD2	480 (3.18)	470	583	3.61 × 10^−8
QD3	490 (2.34)	461	614	3.32 × 10^−8

---

Besides, the absorption behavior of these three dyes on TiO₂ films was depicted in Fig. 1(b). Remarkably, the absorption spectra of the films were broadened in comparison with their absorptions in solution, which was beneficial to light utilization. The maximum absorption peaks of three dyes were observed to slight blue-shift by 10, 10 and 29 nm compared with those of solution spectrum respectively, due to the formation of H-aggregation on TiO₂ films.^36,37

Electrochemical properties

To investigate the molecular energy levels, CV was performed in CHCl₃ with 0.1 M TBAP as a supporting electrolyte. As shown in Fig. 2(a) and Table 2, the oxidation potentials (E_ox) of QD1, QD2, and QD3 corresponding to the highest occupied molecular orbitals (HOMO), were found to be 1.02, 0.82 and 0.79 V, respectively, vs. a normal hydrogen electrode (NHE), which were calibrated by addition of 0.63 V to the potential (vs. SCE) vs. Fc/Fc⁺ by CV. All of them were more positive than the redox couple (I⁻/I₃⁻, 0.42 V, Fig. 2(b)), ensuring that the oxidized dyes formed after the injection of electrons into the conduction band of TiO₂, and then accepted the electrons from I⁻ ions thermodynamically. So there was a sufficient driving force for the dye regeneration to efficiently compete with the recapture of the injected electrons by the dye cation radical.³⁸ The zero–zero excitation energy (E_0–0) values are 2.07, 2.30 and 2.21 eV, determined from the intersection of the normalized absorption and the emission spectra (λ_int) in Fig. 2(c). The reduction potentials (E_red vs. NHE), corresponding to the lowest unoccupied molecular orbitals (LUMO), were −1.05, −1.48 and −1.42 eV for QD1, QD2 and QD3, respectively, calculated by E_ox − E_0–0. Distinctly, the LUMO levels of the three dyes were more negative than the conduction band of TiO₂ (about −0.5 V vs. NHE), indicative of thermodynamic feasibility for electron injection from the excited dyes into the TiO₂ electrode. The E_gap between the conduction band and the E_red (vs. NHE) of QD1, QD2 and QD3 are 0.55, 0.98 and 0.92 eV, respectively. Theoretically, an energy gap of 0.2 eV is necessary for efficient electron injection. Therefore, the three dyes were considered to have proper electronic energy levels as promising sensitizers in DSSCs.

Fig. 2 (a) Cyclic voltammogram, (b) energy level diagram and (c) emission spectra in diluted CHCl₃ solutions of the dyes QD1, QD2 and QD3.

Table 2 Electrochemical data of the dyes QD1, QD2 and QD3^a

Dye	λint/nm	E0–0/eV	Eox/V vs. NHE	Ered/V vs. NHE	Egap/V
a E0–0 values were calculated from intersection of the normalized absorption and the emission spectrum (λint): E0–0 = 1240/λint. The first oxidation potential (vs. NHE), Eox was measured in CHCl3 and calibrated by addition of 0.65 V to the potential vs. Fc/Fc^+. The reduction potential, Ered, was calculated from Eox − E0–0. Egap is the energy gap between the Ered of dye and the conductive band level of TiO2 (−0.5 V vs. NHE).
QD1	600	2.07	1.02	−1.05	0.55
QD2	538	2.30	0.82	−1.48	0.98
QD3	561	2.21	0.79	−1.42	0.92

---

Density functional theory (DFT) calculations

In order to gain more insight into the molecular geometries and electronic distribution of the three dyes, the optimized structures of three dyes were calculated geometrically using the density functional theory (DFT) at the B3LYP/6-31G* level. The optimized structures with torsion angles and electron densities of HOMOs and LUMOs of the dyes are shown in Table 3. In the optimized structures, the torsion angles between the ullazine donor and the acceptor, the acceptor and the π spacer, the donor and π spacer were similar, all of which were very small. It demonstrated that all of them possessed good planarity. In dye system, the planar structure could enhance the aromatic character of the heterocyclic atom, increasing the degree of electronic resonance between donor and acceptor moieties in the dye molecules and facilitating the electron transfer from the donor to the acceptor. However, it increased the stacking of the dye molecules, resulting in more dye aggregation and electron recombination.³⁹

Table 3 Optimized structures, torsion angles and electronic distributions in HOMO and LUMO levels of the dyes QD1, QD2 and QD3

Dye	Optimized structure	HOMO	LUMO
QD1
QD2
QD3

---

As indicated from Table 3, the HOMO values of the three dyes were of π-character, and the LUMO values of the three dyes are of π*-character.²⁷ Besides, the HOMO value of QD1 was delocalized over the entire molecule, the HOMO values of QD2 and QD3 were delocalized over the donor units and π-spacers. The LUMO values of all of the three dyes are delocalized over the entire molecules. Hence, the electronic transitions originates from π–π* excitations and intramolecular charge transfer. The results indicated that the electrons of QD1 might be easily transferred from the donor to the acceptor than that of QD2 and QD3.

Photovoltaic properties of the DSSC devices

The photocurrent density–voltage curves (J–V) of DSSCs and the incident photon to current conversion efficiency (IPCE) spectra based on QD1, QD2 and QD3 are shown in Fig. 3 and 4, and the detailed results are summarized in Table 4. As we all know, CDCA can work as an anti-aggregation compound to improve the photovoltaic performance. Therefore, we studied the performance in the DSSC devices with CDCA as the co-adsorbents. From the Table 4, the photovoltaic performances of the three dyes have all been improved when added the CDCA. As we can see from Fig. 4, the IPCE value of QD1 was in the wavelength range of 300–750 nm, with the maximum IPCE value of 80% at 520 nm; the IPCE of QD2 was in the wavelength range of 300–650 nm, with the maximum IPCE value of 20% at 466 nm; the IPCE of QD3 was in the wavelength range of 300–655 nm, with the maximum IPCE value of 19% at 464 nm. The IPCE value of QD1 over 60% was in the wavelength range from 451 to 606 nm, while the IPCE values of QD2 and QD3 over the entire wavelength range did not exceed 30% and 20%, respectively. In addition, the IPCE value and the J_sc of QD1 were consistent with the absorption spectrum in solution and on TiO₂ film. The absorbed amounts (Γ) of the three dyes on TiO₂ films were shown in Table 1. The Γ values of three dyes (QD1, QD2, QD3) were 5.02 × 10⁻⁸, 3.61 × 10⁻⁸, 3.61 × 10⁻⁸ mol cm⁻², respectively. It suggested that QD1 had the best light-harvesting ability on TiO₂ film contributing to a high J_sc. The cells based on QD1, QD2, QD3 gave a short circuit current density (J_sc) of 12.28, 3.82, 2.98 mA cm⁻². Obviously, the J_sc of QD1-based device was the highest of all, which was in good accordance with the J_sc variation in J–V measurements, absorption spectra and the absorbed amounts. As for V_oc, they were in the order of QD1 > QD2 > QD3, which may cause by the difference of charge recombination rate at TiO₂/dye/electrolyte interface. In a word, QD1, QD2 and QD3 yielded a moderate power conversion efficiency (η) of 5.28%, 1.30% and 1.05% respectively.

Fig. 3 J–V curves for DSSCs with the dyes QD1, QD2 and QD3.

Fig. 4 IPCE action spectra for the DSSC based on the QD1, QD2 and QD3 dye-sensitized TiO₂ films.

Table 4 Photovoltaic performances of DSSCs based on QD1, QD2 and QD3 dyes

Dyes	CDCA (0.5 mM)	Jsc (mA cm^−2)	Voc (V)	FF	η (%)
QD1	—	10.23	0.64	0.69	4.50%
QD1	With	12.28	0.65	0.66	5.28%
QD2	—	2.33	0.60	0.54	0.75%
QD2	With	3.82	0.60	0.56	1.30%
QD3	—	1.61	0.54	0.67	0.58%
QD3	With	2.98	0.53	0.66	1.05%

---

Electrochemical impedance spectroscopy studies

To illustrate the difference in V_oc and investigate the interfacial charge transfer process, the electrochemical impedance spectra (EIS) were performed in the dark. Usually, the Nyquist plot of the dye has two semicircles, the small semicircle (R_ce, in high frequency) is related to the charge transfer at the counter electrode/electrolyte interface, while the big semicircle (R_rec, in intermediate frequency) is attributed to the charge transfer at TiO₂/dye/electrolyte interface.⁴⁰ As we can see from the Nyquist plots shown in Fig. 5, all of them had only one semicircle, which was assigned to the charge transfer at TiO₂/dye/electrolyte interface.⁴¹ The charge transfer at the counter electrode/electrolyte interface was not observed because of the quick redox reaction of I⁻ at the platinum counter electrode.⁴² Maybe the charge-transfer resistance at the counter electrode, was too small compared with that at the TiO₂/dye/electrolyte interface.⁴¹ The R_rec was estimated by the large semicircle width, and a large R_rec meant a small dark current and a low charge recombination rate. The R_rec values for QD1, QD2 and QD3 were estimated to 89 Ω, 64 Ω and 44 Ω, respectively. Apparently, the R_rec of QD1 was the largest, which could effectively suppress the charge recombination between injected electron and electron acceptor in the electrolyte, contributing to a higher V_oc.⁴³ Besides, a big R_rec was beneficial to improve charge generation and transport, contributing to a higher J_sc.⁴⁴ When the donor of the three dyes were the same, the electron withdrawing of the acceptor of QD1 was the strongest, the π-spacer of QD1 was the shortest, so the charge transport of QD1 were the more effective. Therefore, the DSSC based on QD1 had the highest J_sc. These results were in accord with that of V_oc and J_sc shown in Table 4.

Fig. 5 EIS Nyquist plots for DSSCs based on QD1, QD2 and QD3 measured in the dark.

The Bode plots for the DSSCs based on QD1, QD2 and QD3 are shown in Fig. 6 with only one semicircle, which is located at intermediate frequency correspond to the large semicircle. The peaks represented the charge transfer at TiO₂/dye/electrolyte interface, whose reciprocal of the peak frequency was regarded as the electron lifetime. As we can see from Fig. 6, the electron lifetimes of the three dyes were in the order of QD1 > QD2 > QD3, which was in consistent with the V_oc. The longer the electron lifetime, the smaller of the electron recombination and the dark current, which contributed to a higher V_oc.⁴⁵ From the Table 3, we found that the planarity of QD2 and QD3 were better than that of QD1, which might caused more dye aggregation and electron recombination. Evidently, this was consistent with our test results and the Nyquist plots.

Fig. 6 EIS Bode plots for DSSCs based on QD1, QD2 and QD3 measured under −0.65 V bias in the dark.

Conclusions

A series of new organic dyes QD1, QD2 and QD3 containing the ullazine unit has been designed and synthesized. The acceptors as well as the anchoring groups were cyanoacrylic acid in QD1, carboxylic acid in QD2 and QD3. The π-spacers were ethylene in QD1, phenylethylene in QD2 and thiophene ethylene in QD3. When the donor unit of the three dyes were same, the DSSCs based on QD1 showed the highest J_sc (12.28 mA cm⁻²) and V_oc (0.65 V) due to the strongest electron acceptor, the shortest π-spacer, which result in more effective charge transport than that of QD2 and QD3. It was worth mentioning that the absorption spectrum of QD1 was red-shifted than that of QD2 and QD3, the IPCE value of QD1 and the absorbed amounts were the highest of all. The order of the V_oc was QD1 > QD2 > QD3, which was consistent with the EIS Nyquist plots and the EIS Bode plots. This was caused by the different charge recombination rates at TiO₂/dye/electrolyte interface and the electron lifetimes. Among the three dyes, the QD1-based cell showed the highest η of 5.28%.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (51173154, 21474081), and Hunan Provincial Natural Science Foundation of China (13JJ2025).

References

1. B. O'regan and M. Grätzel, Nature, 1991, 353, 737–740.
2. Y. Wu and W. Zhu, Chem. Soc. Rev., 2013, 42, 2039–2058.
3. Y. Xie, Y. Tang, W. Wu, Y. Wang, J. Liu, X. Li, H. Tian and W. Zhu, J. Am. Chem. Soc., 2015, 137, 14055–14058.
4. M. Liang and J. Chen, Chem. Soc. Rev., 2013, 42, 3453–3488.
5. S. Fan, K. Lv, H. Sun, G. Zhou and Z.-S. Wang, J. Power Sources, 2015, 279, 36–47.
6. Q. Chai, W. Li, J. Liu, Z. Geng, H. Tian and W. Zhu, Sci. Rep., 2015, 5, 11330.
7. Y. Tang, Y. Wang, X. Li, H. Agren, W. H. Zhu and Y. Xie, ACS Appl. Mater. Interfaces, 2015, 7, 27976–27985.
8. A. Mishra, M. K. Fischer and P. Bäuerle, Angew. Chem., Int. Ed., 2009, 48, 2474–2499.
9. J. Liu, X. Yang, J. Zhao and L. Sun, RSC Adv., 2013, 3, 15734.
10. J. Shi, Z. Chai, R. Tang, J. Hua, Q. Li and Z. Li, Sci. China: Chem., 2015, 58, 1144–1151.
11. L. Han, A. Islam, H. Chen, C. Malapaka, B. Chiranjeevi, S. Zhang, X. Yang and M. Yanagida, Energy Environ. Sci., 2012, 5, 6057–6060.
12. C.-C. Chou, F.-C. Hu, H.-H. Yeh, H.-P. Wu, Y. Chi, J. N. Clifford, E. Palomares, S.-H. Liu, P.-T. Chou and G.-H. Lee, Angew. Chem., Int. Ed., 2014, 53, 178–183.
13. A. Yella, H.-W. Lee, H. N. Tsao, C. Yi, A. K. Chandiran, M. K. Nazeeruddin, E. W.-G. Diau, C.-Y. Yeh, S. M. Zakeeruddin and M. Grätzel, Science, 2011, 334, 629–634.
14. W. Zhu, Y. Wu, S. Wang, W. Li, X. Li, J. Chen, Z.-S. Wang and H. Tian, Adv. Funct. Mater., 2011, 21, 756–763.
15. K. Li, J. Qu, B. Xu, Y. Zhou, L. Liu, P. Peng and W. Tian, New J. Chem., 2009, 33, 2120–2127.
16. R. Katoh, A. Furube, S. Mori, M. Miyashita, K. Sunahara, N. Koumura and K. Hara, Energy Environ. Sci., 2009, 2, 542–546.
17. S. Hwang, J. H. Lee, C. Park, H. Lee, C. Kim, C. Park, M.-H. Lee, W. Lee, J. Park, K. Kim, N.-G. Park and C. Kim, Chem. Commun., 2007, 4887–4889.
18. K. Kakiage, Y. Aoyama, T. Yano, K. Oya, J. Fujisama and M. Hanaya, Chem. Commun., 2015, 51, 15894–15896.
19. Z. Yao, M. Zhang, H. Wu, L. Yang, R. Li and P. Wang, J. Am. Chem. Soc., 2015, 137, 3799–3802.
20. F. Gerson and A. Metzger, Helv. Chim. Acta, 1983, 66, 2031–2043.
21. H. Balli and M. Zeller, Helv. Chim. Acta, 1983, 66, 2135–2139.
22. J. H. Delcamp, A. Yella, T. W. Holcombe, M. K. Nazeeruddin and M. Grätzel, Angew. Chem., Int. Ed., 2013, 52, 376–380.
23. M. Marszalek, S. Nagane, A. Ichake, R. Humphry-Baker, V. Paul, S. M. Zakeeruddin and M. Grätzel, RSC Adv., 2013, 3, 7921–7927.
24. J. T. Lin, P.-C. Chen, Y.-S. Yen, Y.-C. Hsu, H.-H. Chou and M.-C. P. Yeh, Org. Lett., 2008, 11, 97–100.
25. J. Wiberg, T. Marinado, D. P. Hagberg, L. Sun, A. Hagfeldt and B. Albinsson, J. Phys. Chem. C, 2009, 113, 3881–3886.
26. K. Srinivas, K. Yesudas, K. Bhanuprakash and L. Giribabu, J. Phys. Chem. C, 2009, 113, 20117–20126.
27. P. Ganesan, V. S. Rajadurai, J. Sivanadanam, V. Ponnambalam and R. Rajalingam, J. Photochem. Photobiol., A, 2013, 271, 31–44.
28. Z. Gu, P. Shen, S.-W. Tsang, Y. Tao, B. Zhao, P. Tang, Y. Nie, Y. Fang and S. Tan, Chem. Commun., 2011, 47, 9381–9383.
29. H. Tian, X. Yang, R. Chen, R. Zhang, A. Hagfeldt and L. Sun, J. Phys. Chem. C, 2008, 112, 11023–11033.
30. K. Kalyanasundaram and M. Grätzel, Coord. Chem. Rev., 1998, 177, 347–414.
31. X. Liu, J. Long, G. Wang, Y. Pei, B. Zhao and S. Tan, Dyes Pigm., 2015, 121, 118–127.
32. R. Chinchilla and C. Najera, Chem. Soc. Rev., 2011, 40, 5084–5121.
33. Z. Gu, L. Deng, H. Luo, X. Guo, H. Li, Z. Cao, X. Liu, X. Li, H. Huang, Y. Tan, Y. Pei and S. Tan, J. Polym. Sci., Part A: Polym. Chem., 2012, 50, 3848–3858.
34. H. Chen, H. Huang, X. Huang, J. N. Clifford, A. Forneli, E. Palomares, X. Zheng, L. Zheng, X. Wang and P. Shen, J. Phys. Chem. C, 2010, 114, 3280–3286.
35. I.-Y. Wu, J. T. Lin, J. Luo, C.-S. Li, C. Tsai, Y. S. Wen, C.-C. Hsu, F.-F. Yeh and S. Liou, Organometallics, 1998, 17, 2188–2198.
36. J. Tang, W. Wu, J. Hua, J. Li, X. Li and H. Tian, Energy Environ. Sci., 2009, 2, 982–990.
37. L.-Y. Lin, C.-H. Tsai, K.-T. Wong, T.-W. Huang, L. Hsieh, S.-H. Liu, H.-W. Lin, C.-C. Wu, S.-H. Chou and S.-H. Chen, J. Org. Chem., 2010, 75, 4778–4785.
38. X. Liu, Z. Cao, H. Huang, X. Liu, Y. Tan, H. Chen, Y. Pei and S. Tan, J. Power Sources, 2014, 248, 400–406.
39. W. Lee, S. B. Yuk, J. Choi, H. J. Kim, H. W. Kim, S. H. Kim, B. Kim, M. J. Ko and J. P. Kim, Dyes Pigm., 2014, 102, 13–21.
40. Q. Wang, J.-E. Moser and M. Grätzel, J. Phys. Chem. B, 2005, 109, 14945–14953.
41. J. Qu, X. Gao, G. Li, Q. Jiang and T. Yan, J. Phys. Chem. C, 2009, 113, 3359–3363.
42. T. Li, J. Gao, Y. Cui, C. Zhong, Q. Ye and L. Han, J. Photochem. Photobiol., A, 2015, 303–304, 91–98.
43. D. Wang, W. Ying, X. Zhang, Y. Hu, W. Wu and J. Hua, Dyes Pigm., 2015, 112, 327–334.
44. S. Paek, H. Choi, H. Choi, C.-W. Lee, M.-S. Kang, K. Song, M. K. Nazeeruddin and J. Ko, J. Phys. Chem. C, 2010, 114, 14646–14653.
45. Z.-S. Huang, H.-L. Feng, X.-F. Zang, Z. Iqbal, H. Zeng, D.-B. Kuang, L. Wang, H. Meier and D. Cao, J. Mater. Chem. A, 2014, 2, 15365–15376.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra11918b

---