Relationship between measurement conditions and energy levels in the organic dyes used in dye-sensitized solar cells

Abstract

The energy levels of new metal-free organic dyes for dye-sensitized solar cells have been investigated by the photoemission in air, UV-Vis absorption and cyclic voltammetry methods in the solutions of the dye molecules, in films of the pure dyes and in the dyes adsorbed on nanoporous TiO₂. Significant differences of the energy levels have been found depending on the dye environment. For the best level of tuning in a solar cell, the energy levels are to be determined for the dyes adsorbed on the TiO₂ surface. The absorbed photon conversion to current efficiency (APCE) of the solar cells was evaluated and compared with the incident photon quantum efficiency (IPCE). The results obtained show that the IPCE is dependent on the light quanta energy and reaches a maximum value when the light quanta energy is about 0.3 eV higher than the light absorption threshold.

---

Introduction

Dye-sensitized solar cells (DSSCs) have attracted significant attention as appealing alternatives to conventional semiconductor photovoltaic devices,^1,2 promising to offer a solution to low-cost large-area photovoltaic applications. DSSCs are fabricated from cheap, easily processable materials, deriving their competitive performance from judicious molecular design and control of nanoarchitecture.

Ruthenium complexes as molecular sensitizers have shown impressive solar-to-electric power conversion efficiencies (PCEs) in liquid electrolyte based devices, with the PCE reaching over 11% under standard AM1.5G full sunlight.^3–7 Another type of sensitizer that demonstrated impressive performance in liquid electrolyte based devices are dyes based on a porphyrin zinc complex. Porphyrin-sensitized DSSCs with cobalt(II/III) based redox electrolytes have achieved nearly 13% efficiency.⁸

In recent years, metal-free organic dyes have attracted increasing attention as they do not contain any toxic or costly metals and their properties are easily tuned by facile structural modification. To date the most efficient metal-free organic dyes for DSSCs were based upon a D–A architecture. With this construction it is easy to design new dye structures, extend the absorption spectra, adjust the energy levels and complete the intramolecular charge separation. DSSCs employing organic dyes featuring an electron donor and acceptor moiety connected by a π-conjugation bridge have reached 10% efficiency with liquid electrolytes^9–11 and up to 7% with solid HTM.^12–14 Recently it was demonstrated that a large PCE increase, up to 12.75%, is achieved using sensitizers of the D–A–π–A design and with careful structure optimization.¹⁵

In order to make the right choice of the materials for DSSC development, knowledge of the energy levels of the molecules is necessary. While the HOMO levels could be determined directly by the measurement of the ionization potentials (I_p) or from cyclic voltammetry data, the UV absorption spectra are needed to find the LUMO levels. Energy levels are dependent on the aggregate state and environment of the dye molecules. In a solution, the dye molecules are surrounded by the solvent and in the layers they are in contact with molecules of the same material. When adsorbed on TiO₂, they interact with the TiO₂ surface to which they are chemically bound. The dye molecules can form aggregates and this process may also influence their energetic levels. So far, ionization potentials, using the photoemission in air method, have only been determined in the layers of pure dyes, I_p values of the dyes adsorbed on TiO₂ are not known.

In this work the energy levels of a number of hydrazone dyes are determined using the photoemission in air, calculations from UV-Vis absorption spectrum edge and cyclic voltammetry techniques. The measurements are done in different environments (solution, thin solid film, dye adsorbed on nanoporous TiO₂) and the obtained results are evaluated and compared. The absorbed photon conversion to current efficiency (APCE) of the solar cells was also evaluated and compared with the incident photon quantum efficiency (IPCE) results.

Experimental details

Absorption measurements

UV spectra of the dye solutions were recorded on a Perkin-Elmer Lambda 35 spectrometer. A solution concentration of 10⁻⁴ M of the investigated dye in CHCl₃, and a microcell with an internal width of 1 mm were used. For the investigation of the absorption spectra of the films, the dyes were dissolved in tetrahydrofuran at concentration of 0.5 mg ml⁻¹, coated on a Corning glass substrate and dried at 60 °C temperature for 5 minutes under ambient conditions. The measurements were carried out with an Avantes AvaSpec ULS-2048 spectrophotometer.

Ionization potential measurements

The ionization potential (I_p) was measured by electron photoemission in air. Samples for the measurements were prepared by dissolving dyes in tetrahydrofuran at concentrations of 8 mg ml⁻¹. The solutions were coated on an aluminized polyester substrate with a sub-layer of methylmethacrylate and methacrylic acid. The role of the sublayer was to improve adhesion of the sample material, to retard crystallization and to eliminate the electron photoemission from the Al layer through possible sample layer defects. In addition, the adhesion sub-layer was conductive enough to avoid charge accumulation on it during measurement. The thickness of both the sub-layer and the sample layer was 0.4 μm. The ionization potential was measured by the electron photoemission in air method similar to that described in ref. 16. The samples were illuminated with monochromatic light from the quartz monochromator with a deuterium lamp source. A negative voltage of −300 V was supplied to the sample substrate. A counter-electrode with a 4.5 × 15 mm² slit for illumination was placed at an 8 mm distance from the sample surface. The counter-electrode was connected to the input of the BK2-16 type electrometer, working in a charge accumulation mode. A 10⁻¹⁵ to 10⁻¹² A photocurrent flowed in the circuit under illumination. This photocurrent I is strongly dependent on the incident light photon energy hν. The dependence of the square root of the photocurrent on the incident light quanta I^0.5 = f(hν) energy is well described by a linear relationship near the threshold, and the latter indicating the I_p value.

For the I_p investigation of the dyes on TiO₂, the FTO glasses were cleaned and coated with a Ti-Nanoxide T/SC titania paste from Solaronix using a spin coating technique. The titania paste was coated at 1000 rpm for 30 s and sintered at 460 °C temperature for 30 minutes. The thickness of the TiO₂ layer was approximately 100 nm. The dyes were dissolved in tetrahydrofuran at concentrations of 0.5 mg ml⁻¹. The FTO glasses with the TiO₂ layer were dipped in the dye solutions for 12 hours and dried at 60 °C temperature for 5 minutes.

Calculation of the absorbed photon conversion-to-current efficiency (APCE)

The APCE was calculated using the IPCE (incident photon conversion-to-current efficiency) data determined earlier.^17–19 The quantity of light absorbed in the photoactive layers of the cell L_abs is needed to calculate the APCE. The incident light is partly reflected from the cell, partly absorbed in the FTO and TiO₂ layers. Before reaching the active layer the light undergoes interference caused by the light reflections from the TiO₂ layer surfaces. This causes the meandering of the absorption spectra, especially in the regions of weak absorption; it was smoothed by the Lorentz approximation of the Origin program. After reaching the Ag anode, light is reflected back into the active layers of the cell, this increases the light absorption by the dye. These phenomena were taken into account while calculating the amount of the light absorbed in the photoactive layers L_abs, which is expressed by the formula:

L_abs = 10^−D_r(1 − 10^{−2(D−D_r)}) (1)

where D_r is the absorbance of the SC in the red region of the spectrum where there is no light absorption by the dye and D is the absorbance at a given wavelength. The values of the APCE were found dividing the values of IPCE by L_abs.

The main source of the APCE evaluation error is caused by the meandering of the absorbance, which leads to an inaccuracy of the D_r of ±0.01. In Fig. 3 the APCE values are calculated using a D_r equal to the Lorentz approximation at 750 nm. The ends of the error bars correspond to the APCE values calculated using 0.01 larger or smaller D_r values. The APCE evaluation error increases with decreasing cell absorbance making evaluation practically impossible at wavelengths of more than 700 nm.

Cyclic voltammetry (CV) measurements

The electrochemical studies were carried out by a three-electrode assembly cell from Bio-Logic SAS and a micro-AUTOLAB Type III potentiostat–galvanostat. The measurements were carried out with a glassy carbon electrode in dichloromethane solutions containing 0.1 M tetrabutylammonium hexafluorophosphate as electrolyte, Ag/AgNO₃ as the reference electrode and a Pt wire counter electrode.

Potentials measured vs. Fc⁺/Fc were converted to NHE by the application of +0.69 V (ref. 20) and the NHE vs. the vacuum level to 4.5 V.²¹ The ionization potential (I^CV_p) and electron affinity (E^CV_A) levels were calculated according to the formula I^CV_p = −(E_ox + 5.19) eV and E^CV_A = −(E_red + 5.19) eV, where E_ox and E_red are oxidation and reduction potentials respectively.

Fabrication and characterization of the solid-state dye-sensitized solar-cells

A TiO₂ blocking layer was prepared on a fluorine-doped tin oxide (FTO)-covered glass substrate using spray pyrolysis.²² Next, a TiO₂ paste (Dyesol), diluted with terpineol, was applied by screen printing, resulting in a film thickness of 1.7 μm. All films were then sintered for 45 min at 450 °C, followed by treatment in a 40 mM aqueous solution of TiCl₄ at 60 °C for 30 min, followed by another sintering step. The prepared samples with the TiO₂ layers were pretreated with 5 mM solutions of 2-(p-butoxyphenyl)acetohydroxamic acid sodium salt or 2-(p-butoxyphenyl)acetohydroxamic acid tetrabutylammonium salt in ethanol. The electrodes were then dyed in 0.5 mM dye solution in CH₂Cl₂. Spiro-MeOTAD was applied by spin-coating from a solution in DCM (200 mg ml⁻¹) also containing 20 mM Li(CF₃SO₂)₂N. Fabrication of the device was completed by evaporation of 200 nm of silver as the counter electrode. The active area of the ssDSSC was defined by the size of these contacts (0.13 cm²), and the cells were masked by an aperture of the same area for measurements. The current–voltage characteristics for all cells were measured with a Keithley 2400 under 1000 W m⁻² AM 1.5G conditions (LOT ORIEL 450 W). The incident photon to current conversion efficiencies (IPCE) were obtained with an Acton Research Monochromator using additional white background light illumination.

Results and discussion

The dyes investigated

The investigated hydrazone dyes D1–D9 (Fig. 1) were synthesized by the procedures reported earlier.^17–19

Fig. 1 Structures of the investigated hydrazone dyes D1–D9.

Energy levels of the investigated dyes

The ionization potentials of the hydrazone dyes D1–D9 were measured using cyclic voltammetry and photoemission in air techniques, while the electron affinity was measured by cyclic voltammetry and determined from the band edges of the absorption spectra.

The absorption spectra of the dyes were investigated in solution, in a thin layer of dye on glass and in the solar cells before the deposition of the Ag electrode. The results of the investigations for the dye D7 are presented in Fig. 2, spectra of the remaining materials can be found in the ESI.† To better reveal the dependencies at the long wave end of the spectrum, the square root of the normalized absorbance or molar extinction coefficient ε, in case of the dye solution, are plotted. For comparison the values of the square root of incident photon to current conversion efficiencies (IPCEs), determined in the constructed solid-state dye-sensitized solar cells (ssDSSCs), are also presented. The long wave dependencies, as seen in Fig. 2, can be approximated by two linear regions.

Fig. 2 The normalized adsorption spectra of D7 in solution, film of the pure dye and adsorbed on TiO₂ in SC. The IPCE spectrum of the constructed SC is given for comparison.

The intersection of the straight lines with the axis gives two long wave end values of the spectra, which are given in Table 1. In the case of the dye D7 in solution, the intersection points gives the spectrum edge values 2.03 and 1.95 eV. The larger value may be ascribed to the stronger absorption band caused by the main mass of the dye molecules while the lower value is due to absorption of a few aggregated molecules. In the case of the dye film or solar cell (SC) samples, these aggregation bands in the long wave region are stronger and the differences between the two intersections are larger. The increase of shoulder size in the long wave region of the absorption spectrum may be caused by the interactions between the dye molecules and the aggregate formation, which are weak in the solutions, but much stronger in the films and between dye molecules adsorbed on TiO₂. In the case of some dyes, for example D5 (Fig. S9†), the shoulder in the long wave absorption region of the films could be very strong and the corresponding spectrum edges reach to about 1 eV. This may be caused not only by aggregation, but also by the presence of a crystalline phase in the films. The E_g measurement results, calculated from the edges of the absorption spectra (Table 1), show that the dyes in solution demonstrate the largest E_g values, exceeding those determined from the spectra of the films or dyes adsorbed on TiO₂. However, these are smaller than those found in the CV experiments, indicating that the aggregation is largest in the films, lower in the dyes adsorbed on the TiO₂, and lowest in solution. Looking at the energy levels determined using CV and photoemission in air techniques (Table 2), it is evident that the absolute values differ quite noticeably; the CV results are about 0.2–0.3 eV higher. These differences arise from the different measurement conditions; CV is done in solution, while photoemission is measured in the solid film. The dependence between the energy level and structure is more pronounced in the case of the measurements done in solution. Due to significant dilution, the dye molecules have a reduced tendency to form aggregates in the solution which could influence the position of the energy levels. Additionally, the way the molecules pack in the film is also a significant factor. The CV measurements reveal that substitution of the aromatic moieties with aliphatic ones in the hydrazone fragment reduces both the I_p and electron affinity (E_A), while modifications at the triphenylamine donor site affect only the I_p values. It has to be noted that the length of the aliphatic chains has a negligible effect on the energy levels of the dyes in the solution. In the solid state however, the molecules are much more prone to form aggregates and the structure/energy level relationship becomes a lot less predictable. The energy levels of the dye molecules adsorbed on the TiO₂ follow yet another pattern. Due to the different arrangement and intermolecular interactions of the molecules adsorbed on TiO₂, the I_p increases compared to those measured in the films in the dyes D1, D2, and D9 containing a diphenylhydrazone moiety, and decreases if the phenyl ring is substituted by the aliphatic fragment (D3, D4, D6, and D7). Curiously, if longer aliphatic chains are used, the difference between the I_p measured in the film and on TiO₂ becomes smaller (D7) or non-existent (D8), indicating that longer aliphatic chains retard aggregate formation in the films.

Table 1 E_g results from the CV measurements, long wave edges of the spectra of the dye solutions, films of pure dyes on glass substrates and in SC, the edges of IPCE spectra from the solar cell devices, as well as ssDSSC conversion efficiencies under standard AM 1.5G illumination

Dye	CV, eV	Solution, eV	Film, eV	Cell, eV	IPCE, eV	η, %
D1	2.16	2.06	2.03, 1.2	1.87, 1,78	1.81, 1.7	3.8
D2	2.09	2.01, 1.9	1.88, 1.7	1.89, 1.65	1.77, 1.7	3.7
D3	2.15	2.01, 1.5	1.89, 1.1	1.90, 1.6	1.77, 1.6	3.4
D4	2.12	2.02, 1.8	1.85, 1.65	1.89, 1.6	1.73, 1.6	3.2
D5	2.15	2.07, 1.8	1.7, 1.0	1.98, 1.6	1.97, 1.85	3.0
D6	2.10	2.03, 1.75	1.83, 1.0	1.92, 1.6	1.81, 1.7	3.9
D7	2.13	2.03, 1.95	1.95, 1.7	1.90, 1.6	1.76, 1.6	4.5
D8	2.09	2.01, 1.95	1,89, 1.2	1.93, 1.6	1.84, 1.6	3.8
D9	2.05	2.02, 1.75	1.85, 1.0	1.88, 1.75	1.88, 1.76	3.8

---

Table 2 Ionization potential and electron affinity values of the investigated dyes

Dye	I^CVp, eV	Ip of dye films, eV	Ip on TiO2, eV	E^CVA, eV	EA of dye films, eV	EA on TiO2, eV
D1	5.55	5.22	5.29	3.39	3.2, 4.0	3.4, 3.5
D2	5.47	5.04	5.09	3.38	3.2, 3.3	3.2, 3.4
D3	5.47	5.26	5.13	3.32	3.4, 4.2	3.2, 3.5
D4	5.44	5.13	5.06	3.32	3.3, 3.5	3.2, 3.5
D5	5.52	5.36	5.26	3.37	3.7, 4.4	3.3, 3.7
D6	5.46	5.31	5.17	3.36	3.5, 4.3	3.25, 3.6
D7	5.44	5.22	5.18	3.31	3.3, 3.5	3.3, 3.6
D8	5.47	5.17	5.17	3.38	3.3, 4.0	3.2, 3.6
D9	5.41	5.17	5.08	3.36	3.3, 4.2	3.2, 3.33

---

Overall, this investigation shows that the degrees of aggregation of the dyes in the films, in solutions and adsorbed on TiO₂ are different, so it may be incorrect to use ionization potential results measured in the dye films for evaluation of the energy levels in the SC. It may also be incorrect to use the absorption spectrum edges measured for the dye films or solutions to evaluate the electron affinity E_A levels of the dyes in the solar cells. A better way to do this is to measure both the ionization potentials and absorption spectra for the dyes adsorbed on TiO₂, as was done in this investigation.

IPCE and APCE

The efficiency of the investigated cells is 3 to 4.5%, which is considerably less than that achieved in the SC with liquid electrolytes. While a number of factors are important, one of the causes limiting efficiency may be improper tuning of the energy levels inside the cell. In order to better understand the causes limiting the conversion efficiency we investigated the spectra of the absorbed photon conversion to current efficiency (APCE), or the quantum efficiency of the conversion process.

The long wave end of the IPCE spectrum coincides with the edge of the absorption spectrum of the cell, but differs from the edges in the solution. The edge of the IPCE spectrum of the investigated dyes is usually at lower quanta energy than the absorption edge in solution, indicating that the charge carrier generation in some cases takes place not only by the light absorption by single molecules of the dyes, but also by their aggregates. In some cases, as for the dyes D2, D4, and D7 (Fig. 2, S4 and S7†) there is only a small difference between the edges of the absorption spectra in the film and IPCE, but in a number of other cases, as D5, and D3 (Fig. S5 and S11†), the difference is significant. These observations show that not all the aggregation forms are present between the molecules adsorbed on TiO₂ or take part in the charge carrier photogeneration process.

From the data provided in Fig. 3 it is evident that significant light absorption by the nanoporous TiO₂ takes place at light wavelengths below 400 nm. Hole transporting material spiro-MeOTAD, doped with the LiTFSI, also contributes to the overall absorption especially at wavelengths below 600 nm. Curve 3 shows the amount of light absorbed in the active layers of the cell by the dye and spiro-MeOTAD L_abs, which was calculated by formula (1). By dividing the IPCT values, represented by the points 4, the values 5 of APCE were calculated.

Fig. 3 The spectra of the SC with the dye D6. (1) Absorbance of the glass substrate with a TiO₂ nanoporous layer, (2) absorbance of the 2 μm thick spiro-MeOTAD with the LiTFSI additive, (3) quantity of the absorbed light by the active layers of the cell, (4) IPCE values, (5) APCE values.

Some characteristic features of the APCE spectra can be seen in Fig. 3. The APCE values at the long wave end of the spectrum are low. This means that the dye molecules, excited into the energetic level just above the LUMO, possess only a small probability to create free charge carriers capable to have an input into the SC performance. About 0.3 eV more energy is needed to reach the APCE maximum. This needs to be considered while tuning the energy levels in the SC. The electrons from the excited dye molecules need to be injected into the conduction band of TiO₂, the bottom of which is about 4.1 eV. The holes should jump into the HOMO level of the hole transporting material, which is at 5.0 eV. We see from Table 2 that the E_A values for the dyes on TiO₂ are above 3.2 eV for the non-aggregated molecules and above 3.7 eV for the aggregates. Thus, there should be no problem for electrons to be injected into TiO₂, even from aggregates. Similarly, the I_p values of the investigated dyes on TiO₂ are below 5.06 eV. Thus, there also should be no problem with the hole injection from the dye into the transporting material. However the steep IPCE dependence on the excitation quanta energy at the low energy side suggests that there is some mechanism influencing the injection efficiency. Probably, molecular vibrations stimulate the charge carrier injection into either TiO₂, or the transporting layer, or both.

After reaching the maximum, the APCE values decrease with further increase of the excitation light energy. This results in a bend in the APCE spectrum and corresponds to the maximum in the light absorbance spectrum. It may be caused by a lower efficiency of the charge carrier pairs generated near the FTO layer compared with the pairs generated near to the spiro-MeOTAD layer, due to the difficulty for holes to escape from the deep regions of the nanoporous TiO₂ layer. At wavelengths below 450 nm the APCE values once more decrease, this time because of the non-productive light absorption by the doped spiro-MeOTAD.

Conclusions

The results of these investigations show that both the HOMO and LUMO values of the dyes used in SC are dependent on the aggregate state of the dyes. The energy levels differ quite noticeably depending upon the measurement conditions. For this reason the ionization potential and electron affinity of the dyes adsorbed on nanoporous TiO₂ should be evaluated. From the standpoint of the structure/energy level relationship it is evident that long aliphatic chains, positioned at π-conjugated bridges or the donor end of the molecule, inhibit formation of aggregates to some extent. This makes energy levels more predictable and closer related to the molecular structure. The results of this work also show that the absorbed photon conversion-to-current efficiency depends on the energies of the excitation light quanta. The maximum efficiency is reached when the quanta energies exceed the threshold value by about 0.3 eV, which may be caused by some peculiarity of the charge carrier injection from the excited dye into TiO₂ or the hole transporting material.

References

1. B. O’Regan and M. Grätzel, Nature, 1991, 353, 737–740.
2. M. Grätzel, Nature, 2001, 414, 338–344.
3. M. K. Nazeeruddin, F. de Angelis, S. Fantacci, A. Selloni, G. Viscardi, P. Liska, T. Ito, S. Bessho and M. Grätzel, J. Am. Chem. Soc., 2005, 127, 16835–16840.
4. Y. Chiba, A. Islam, Y. Watanabe, R. Komiya, N. Koide and L. Y. Han, Jpn. J. Appl. Phys., 2006, 45, L638–L640.
5. F. Gao, Y. Wang, D. Shi, J. Zhang, M. K. Wang, X. Y. Jing, R. Humphry-Baker, P. Wang, S. M. Zakeeruddin and M. Grätzel, J. Am. Chem. Soc., 2008, 130, 10720–10728.
6. Y. M. Cao, Y. Bai, Q. J. Yu, Y. M. Cheng, S. Liu, D. Shi, F. Gao and P. Wang, J. Phys. Chem. C, 2009, 113, 6290–6297.
7. C. Y. Chen, M. K. Wang, J.-Y. Li, N. Pootrakulchote, L. Alibabaei, C. H. Ngoc-le, J. D. Decoppet, J. H. Tsai, C. Grätzel and C. G. Wu, et al., ACS Nano, 2009, 3, 3103–3109.
8. A. Yella, H.-W. Lee, H. N. Tsao, C. Yi, A. K. Chandiran, M. Nazeeruddin, E. W.-G. Diau, C.-Y. Yeh, S. M. Zakeeruddin and M. Grätzel, Science, 2011, 334, 629–634.
9. S. Kim, J. K. Lee, S. O. Kang, J. Ko, J.-H. Yum, S. Fantacci, F. de Angelis, D. Di Censo, M. K. Nazeeruddin and M. Grätzel, J. Am. Chem. Soc., 2006, 128, 16701–16707.
10. D. P. Hagberg, T. Edvinsson, T. Marinado, G. Boschloo, A. Hagfeldt and L. C. Sun, Chem. Commun., 2006, 2245–2247.
11. W. D. Zeng, Y. M. Cao, Y. Bai, Y. H. Wang, Y. S. Shi, M. Zhang, F. F. Wang, C. Y. Pan and P. Wang, Chem. Mater., 2010, 22, 1915–1925.
12. N. Cai, S.-J. Moon, L. Cevey-Ha, T. Moehl, R. Humphry-Baker, P. Wang, S. M. Zakeeruddin and M. Grätzel, Nano Lett., 2011, 11, 1452–1456.
13. X. Liu, W. Zhang, S. Uchida, L. Cai, B. Liu and S. Ramakrishna, Adv. Mater., 2010, 22, E150–E155.
14. A. Snaith, H. J. Petrozza, S. Ito, H. Miura and M. Grätzel, Adv. Funct. Mater., 2009, 19, 1810–1818.
15. Y. Wu, W.-H. Zhu, S. M. Zakeeruddin and M. Grätzel, ACS Appl. Mater. Interfaces, 2015, 7, 9307–9318.
16. M. Daskeviciene, V. Getautis, J. V. Grazulevicius, A. Stanisauskaite, J. Antulis, V. Gaidelis, V. Jankauskas and J. Sidaravicius, J. Imaging Sci. Technol., 2002, 5, 467–472.
17. S. Urnikaite, T. Malinauskas, V. Gaidelis, I. Bruder, R. Send, R. Sens and V. Getautis, Chem.–Asian J., 2013, 8, 538–541.
18. S. Urnikaite, T. Malinauskas, V. Gaidelis, I. Bruder, R. Send, R. Sens and V. Getautis, J. Phys. Chem. C, 2014, 118, 7832–7843.
19. S. Urnikaite, M. Daskeviciene, R. Send, H. Wonneberger, A. Sackus, I. Bruder and V. Getautis, Dyes Pigm., 2015, 114, 175–183.
20. N. G. Connelly and W. E. Geiger, Chem. Rev., 1996, 96, 877–910.
21. Handbook of electrochemistry, ed. Zoski and G. Cynthia, Elsevier, 2006.
22. B. Peng, G. Jungmann, C. Jager, D. Haarer, H. W. Schmidt and M. Thelakkat, Coord. Chem. Rev., 2004, 248, 1479–1489.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra14030g

---