HPLC coupled with a spectrophotometer as a reliable setup for the study of absorption properties of imidazolium ionic liquids using bmimBF₄ as an example

Abstract

The absorption spectra of the same ionic liquid published by many authors show significant differences, which are interpreted as being caused by the presence of many difficult to eliminate impurities. This paper presents an innovative use of an HPLC system coupled with a spectrophotometer, which allows separation of impurities from bmimBF₄, and to a high degree, separation of bmimBF₄-related species, determination of their relative content and measurement of their absorption spectra. The absorption spectra of bmimBF₄ aggregates and water complexes show a long-wavelength absorption band (λ_max = 278 nm) but with different molar absorption coefficients, ε. The absorption spectrum measured by a spectrophotometer is a sum of the absorption spectra of all species derived from bmimBF₄ and the impurities that it contains, while the use of the proposed setup provides reliable results on the spectral properties of different compounds showing complex and/or difficult to study properties even if they contain numerous impurities.

---

1. Introduction

Ionic liquids (ILs) have been the subject of intense study because of their interesting physicochemical properties and their wide range of applications, in many of which the purity of the ILs is of extreme importance.^1–5 The presence of impurities most often is a consequence of using not pure enough substrates or solvents in the synthesis. Impurities can also appear in the process of synthesis. Therefore, much attention has been paid to the methods of their syntheses and their purification to achieve a purity high enough for spectral investigation. However even if the product of the synthesis appears pure according to the measurements performed by standard analytical methods (such as NMR or IR), it does not mean that it is pure or even colourless.⁶ On the other hand, NIR studies of bmimOTf suggest that this method can distinguish dried and undried IL.^6a Shen et al. have presented an effortless method to prepare colourless ILs using water as a solvent.^6b Furthermore these authors claim that NMR spectroscopy has not been able to distinguish between a coloured and a decolourised IL. The absorption spectra taken before and after purification of ILs were analysed. The authors claim that when the absorption spectrum after the next purification step does not change it means that the product is sufficiently pure.^6c The quality of purification is also evaluated by electronic spectroscopy. In view of the above, detailed knowledge of the absorption and emission properties of the most often studied imidazolium ILs is essential. We have found seven absorption spectra of 1-butyl-3-methylimidazolium hexafluorophosphate (bmimPF₆) reported in the literature,^7–13 five absorption spectra of 1-butyl-3-methylimidazolium bis(trifluoromethanesulphonyl)imide (bmimNTf₂),^8,14–17 and four absorption spectra of 1-butyl-3-methylimidazolium tetrafluoroborate (bmimBF₄).^6b,9–11 Unfortunately, we could not find any paper in which the authors compare the absorption spectrum obtained in their laboratory with the previously measured absorption spectra of the same IL. The only exception known to us is a paper by Katoh,⁸ who compared three absorption spectra of neat bmimPF₆ and found that the absorption spectrum he obtained showed a much lower absorption in the range λ ≥ 250 nm than that reported in two other papers.^12,13 Comparison of the four absorption spectra of bmimBF₄ reported in literature revealed that all of them showed significant differences, despite very careful purification of the IL, as emphasised by the authors.^6b,9–11 Similarly, considerable differences were found when comparing the absorption spectra of the other imidazolium ILs. No two identical, independently measured absorption spectra of the same IL have been obtained as yet! This implies that the imidazolium ILs contain such impurities that are difficult to remove by many purification methods hitherto applied. According to some research groups, impurities are responsible for the formation of a long-wavelength absorption tail.^18–20 Other authors claim that the long-wavelength absorption band (beyond 280 nm) originates from different species formed by the IL itself.^{9,10,12,13,15} Imidazolium ILs undoubtedly show heterogeneous properties as evidenced by efficient association and nanostructural organization.²¹ BmimBF₄ and other imidazolium ILs either neat or in solvent, in particular in water, form aggregates of different structures and properties, in which the solvent molecules are also involved.^22–25 This has been proved by different experimental methods^22–24 and theoretical calculations.^3,25

Electronic absorption spectroscopy offers very high sensitivity, but its main limitation is the additivity of the absorption measured. Thus, on the basis of the absorption spectra measured by a spectrophotometer no conclusions can be drawn about the species formed by a given IL, i.e., their spectral properties and the impurities that the IL contains. A strategy for optimizing the use of this method is to couple it with an HPLC system. A combination of these two independent analytical methods permits the investigation of untypical absorption properties not only of ILs but also of many other compounds, especially those with complex properties. Using HPLC coupled with a spectrophotometer it is possible to separate the different species, formed by the compound under study, from each other and from the impurities in the compound. The absorption properties of all the species present can be analysed simultaneously. A similar system has been earlier employed by us for the investigation of very weakly emitting compounds with elimination of the effect of emissions from impurities.²⁶

The aim of this study is to determine the absorption properties of bmimBF₄ and to assess the influence of the absorbing impurities present in this IL on them. It was highly important to determine the real origin of the long-wavelength absorption tail, always observed in the absorption spectra of imidazolium ILs, especially in view of the serious controversies regarding its origin. Taking into account the differences in literature results for the same IL, we decided to investigate the same IL (bmimBF₄) of the highest available purity purchased from three different companies (specialising in the synthesis of ILs). The three samples are labelled as X, Y, Z. The purity declared by the producers was for X ≥ 99%, for Y ≥ 98% and for Z 99%. The measurements we wanted to undertake were expected to establish which parts of the spectra are definitely the result of absorption by IL and which come from the presence of impurities. It should be emphasised that the only difference between the three samples X, Y and Z could be the type and amount of impurities. Taking great care to use the same concentration of the samples, the same solvent and the same conditions of measurements, the presence of the same IL species is ensured.

2. Results and discussion

2.1 Traditional absorption measurements using a spectrophotometer

We started with the traditional spectrophotometer measurements of the absorption spectra of the three samples of bmimBF₄ used at a concentration of 2.65 M in a 40% acetonitrile (ACN) and 60% H₂O v/v mixture. The absorption spectra of the bmimBF₄ samples labelled as X, Y, Z shown in Fig. 1 revealed considerable differences in the range λ ∼ 245–400 nm. For sample X the absorbance at λ = 278 nm was over twice as high as for the samples Y and Z, which must be a result of the absorbance of impurities.

Fig. 1 Absorption spectra of three samples of bmimBF₄ (c = 2.65 M in 40% ACN and 60% H₂O v/v mixture) purchased from three different companies and labelled X, Y, Z (optical path l = 0.1 cm).

The absorption spectra of all of the samples showed a broad absorption band in the range λ = 250–300 nm, and the absorption spectrum of X showed another band at λ_max ∼ 335 nm, extending to λ > 400 nm. The absorption spectra of samples Y and Z showed a long-wavelength tail reaching λ ∼ 400 nm. A similar long-wavelength absorption has been observed for neat bmimBF₄.^6b,9–11 All absorption spectra of bmimBF₄, both those of X, Y and Z and those reported in above mentioned papers^6b,9–11 differed in shape, intensity and thus molar absorption coefficients ε(λ). In the long-wavelength range, λ ∼ 250–300 nm, the values of ε for bmimBF₄, neat^10,11 and at a concentration of 2.65 M (X, Y and Z) were very small; ε < 1 M⁻¹ cm⁻¹. It can be expected that, because of such a small ε for the IL studied, the influence of impurities on the absorption spectra measured can be considerable and can lead to erroneous interpretations of the results. It is clear that just 1% of some impurities with an ε = 100 M⁻¹ cm⁻¹would give an absorbance of the same magnitude as that of the IL studied! On the other hand, in the short-wavelength range the influence of impurities on the absorption spectra measured must be much weaker because of the much higher ε of the ILs in that spectral range. For bmimPF₆ at λ_max = 210 nm, ε is equal to 4080 M⁻¹ cm⁻¹,⁷ and for bmimNTf₂ (c = 3 × 10⁻⁴ M in ACN solution) it is ε₂₁₁ = 5000 M⁻¹ cm⁻¹.⁸ It should be pointed out that only in two papers,^7,8 have the authors measured absorption spectra for bmimPF₆ and bmimNTf₂, respectively, in the short-wavelength range of λ = 200–250 nm! Thus, in the ESI (Fig. S1†), we present the absorption spectrum of bmimBF₄ (sample Z) measured by us in this spectral range. The use of a cuvette with a very short optical path of 0.001 cm permitted measurements of the absorption spectrum of bmimBF₄ at the highest possible concentration c = 0.53 M. This absorption spectrum is very similar to those of bmimPF₆ and bmimNTf₂, both in shape and position; also the calculated ε₂₁₀ ∼ 3900 M⁻¹ cm⁻¹ is similar. This result confirms that the bmim cation is responsible for the absorption of bmimBF₄ (imidazolium ILs). On the basis of the absorption spectra of bmimBF₄ measured over a wide range of concentrations 0.5 M–1 × 10⁻⁵ M, it can be concluded that the position of the maximum of the absorption band does not depend on the IL concentration, but ε_max slightly increases with decreasing concentration (ε_max ∼ 4200 M⁻¹ cm⁻¹ for c = 1 × 10⁻⁵ M).

2.2 Absorption measurements using HPLC-abs

The system of HPLC coupled with a photodiode spectrophotometer (HPLC-abs) has not been used as yet in studies such as those presented in this work. A single measurement performed with this system gives (1) absorption chromatograms for all wavelengths and (2) absorption spectra for all the retention times (t_R). Analysis of these experimental results permits the determination of the part of the absorption coming from bmimBF₄ and its impurities, the presence of species formed from IL and determination of their absorption spectra. A comparison of the HPLC-abs results following chromatographic separation with those obtained without such a separation should reveal the reasons for the differences in the absorption spectra of samples X, Y and Z (Fig. 1), and for the differences in all hitherto published absorption spectra of bmimBF₄. In particular we anticipated being able to explain the origin of the long-wavelength absorption tail (λ > 250 nm) in the absorption spectrum, which has been commonly observed for bmimBF₄ and other imidazolium ILs.^{6b,9–15,17,27,28}

The very high reproducibility of the absorption chromatograms and absorption spectra measured in the HPLC-abs system lead naturally to the assumption that the parts of the absorption that are identical for the three samples X, Y and Z come from the species formed from bmimBF₄ itself, while the other parts of the absorption must originate from absorbing impurities. Because of the too high viscosity of neat (c = 5.3 M) bmimBF₄ and the characteristics of measurements performed in HPLC-abs, we carried out measurements on the IL in low viscosity solvents (concentration of 2.65 M). The best eluent was a mixture of ACN and H₂O (40/60 v/v), which was also used as the solvent. These solvents have often been used in studies of the spectral properties of many ILs including bmimBF₄. All measurements for X, Y, Z were made under the same experimental conditions.

Because of the high concentration of bmimBF₄ and the high value of its molar absorption coefficient ε₂₁₁ ∼ 3900 M⁻¹ cm⁻¹ at the maximum of the short-wavelength absorption band (λ_max = 211 nm), the absorbance in the spectral range λ = 200–230 nm was too high to be measured for retention times corresponding to the main chromatographic peak. That is why absorption chromatograms were taken at λ = 232 nm (Fig. 2a) (ε₂₃₂ ∼ 160 M⁻¹ cm⁻¹), for which the absorbance at the chromatographic peak maximum was the highest possible and accurately measured (Fig. 2a inset). Taking into account the high purity of Z (99%) and the relatively high value of ε₂₃₂ ∼ 160 M⁻¹ cm⁻¹, it can be assumed that for sample Z (whose absorbance for λ = 232 nm is the lowest, A₂₃₂ ∼ 1.9), the chromatographic peak shows the distribution of the bmimBF₄ concentration with time. A very similar shape of this peak was recorded for sample Z in the absorption chromatograms in the range λ = 232–238 nm (ESI Fig. S2†), which means that in this range of wavelength, the main peak observed in the absorption chromatograms comes almost exclusively from bmimBF₄. For the other two samples, especially for sample X, a small part of the peak, near t_R^max (for which A₂₃₂ ∼ 2) comes from the absorbing impurities (ESI Fig. S3†). Fig. 2b presents absorption chromatograms at λ = 278 nm (corresponding to the maximum of the long-wavelength band in the absorption spectra, as seen in Fig. 2c for samples Y and Z), for which significant differences in the shape and intensity have been noted in the traditionally measured absorption spectra (see Fig. 1 and ref. 6b and 9−11). In both absorption chromatograms (λ = 232 nm and λ = 278 nm) absorption is observed in a wide range of t_R, from 6 min to over 15 min.

Fig. 2 Absorption chromatograms and absorption spectra of three samples of bmimBF₄ purchased from different producers and labelled X, Y and Z, recorded in an HPLC-abs system. Absorption chromatograms measured for (a) λ = 232 nm (inset: the whole range of absorbance); (b) λ = 278 nm (inset: short t_R range, peaks coming from impurities). Absorption spectra measured for (c) t_R = 9.13 min corresponding to the highest concentration of bmimBF₄ (inset: short-wavelength range); (d) t_R = 11.00 min corresponding to broad peak observed in (b), (inset: normalised spectra).

The chromatograms presented in Fig. 2a and b for the two wavelengths of observation (λ = 232 and 278 nm) give an idea of the number of impurities occurring in a given IL. For the three samples, we obtained absorption chromatograms differing in the number and intensity of the peaks. The impurities present in samples X, Y and Z are most often different as follows from their different t_R (Fig. 2b-inset) and their absorption spectra. A single peak does not have to correspond to only one compound, for instance three different absorption spectra were recorded for the retention times corresponding to the peak of t_R^max = 7.95 min (ESI Fig. S4†). As follows from a comparison of absorption chromatograms, sample X contains the greatest number of absorbing impurities (not only for λ = 232 and 278 nm, but also for other wavelengths), while sample Z contains the smallest number of them.

It has been established by the MS method that bmimBF₄ forms dimers, trimers and larger aggregates.^22a,23a Independent fully ab initio large-scale calculations have shown that bmimBF₄ occurs not only in monomers, but it also forms stable dimers, tetramers and octamers.^25a Taking into account (a) the large concentration of bmimBF₄, (b) the very similar shapes of the absorption chromatograms (λ = 232 nm) for samples X, Y and Z in the range t_R = 7.9–9.4 min (for λ = 232 nm) and (c) the almost identical absorption spectra for samples Y and Z recorded for t_R = 9.13 min, it can be assumed that, at the maximum concentration of bmimBF₄ (Fig. 2c), the main peak in Fig. 2a (inset) originates from non-separated dimers, trimers and larger aggregates. Moreover, very similar absorption spectra were recorded for the range of t_R ∼ 8.7–9.3 min of the same peak for samples Y and Z. The absorption spectra of these species show strong short-wavelength absorption (Fig. 2c), a minimum at a wavelength close to 257 nm and two long-wavelength bands: one with a maximum at λ = 278 nm, while the other with no clear maximum (samples Y and Z). The ratio of absorbance at 232 nm and at 278 nm in the absorption spectrum measured for t_R = 9.13 min A_λ=232/A_λ=278 is ∼370, which directly corresponds to the ε ratio for these wavelengths. For sample X, a significant contribution of impurities is observed in the absorption spectrum measured for the same retention time.

As shown in Fig. 2b, the absorption chromatograms (λ = 278 nm) have a very broad peak in the range t_R = 9.9–15 min of the same shape, position and intensity for all three samples X, Y, Z. Moreover the same shape of this broad peak is observed in the absorption chromatograms in the range λ = 260–295 nm, which means that this broad peak comes almost exclusively from bmimBF₄. The large width of this peak, which is much greater than the widths of all the narrow peaks attributed to the impurities present in the IL, suggests that the wide peak comes from a few different species formed from bmimBF₄, itself, which cannot be separated. The low concentration of bmimBF₄ in the range of t_R = 9.9–15 min means that there are many water molecules per one bmimBF₄ molecule. It has been shown, that bmimBF₄ in an aqueous environment forms complexes with water molecules, besides the aggregates made of different numbers of ion pairs.^{3,21a–d,23a,24a–e,25a,29} Thus, it is probable that the untypically broad chromatographic peak, observed over a very wide range of wavelength, comes from such complexes. As presented in ESI Fig. S5,† the shape of the absorption spectra for t_R from the whole range t_R = 10–12 min is very similar, the intensity of the long-wavelength band is high and increases with t_R. The differences in absorbance intensity (and thus in the concentration) for aggregates (main peak in Fig. 2a inset) and for the complexes with water (broad peak, t_R = 9.9–15 min in Fig. 2b), are very large (e.g. for λ = 232 nm, A_aggregates/A_complexes = 880). Thus, it can be assumed that when bmimBF₄ is used in a high concentration (e.g. c = 2.65 M), dimers and greater aggregates dominate, which is in agreement with the model proposed by Dupont et al.²³

The absorption spectra of the aggregates and water complexes of bmimBF₄ presented in Fig. 3 show a long-wavelength band of similar shape and position (λ_max = 278 nm). The relative intensity of this band after normalisation at λ = 232 nm is much smaller (over a 600 times) for the aggregates than for the complexes (Fig. 3).

Fig. 3 Normalised (at the λ = 232 nm) absorption spectra of aggregates formed by bmimBF₄ (t_R = 9.13 min) and of complexes formed by bmimBF₄ with water molecules (t_R = 11.00 min).

3. Conclusions

BmimBF₄ forms different species, such as aggregates (dimers, trimers and larger aggregates) and complexes with water molecules. Thanks to the use of an HPLC-abs setup, the presence of these species and their absorption spectra could be determined. BmimBF₄ was found to contain many impurities whose removal to such a degree that they would not influence the absorption spectrum measured by a spectrophotometer seems to be impossible, at least when the traditional methods of purification are used. It seems that the reason for the presence of so many impurities are the strong interactions of imidazolium ILs, both the cation and the anion, with many compounds both ionic and hydrogen-bond forming. That is why, as suggested by Nockemann et al., the synthesis of imidazolium ILs requires the use of the purest possible substrates.²⁸

The absorption spectra of the species formed by bmimBF₄ show a long-wavelength absorption band with λ_max = 278 nm, however, the shape, and, first of all, the relative intensity of the bands are different for the aggregates and the complexes. The long-wavelength absorption was also recorded in the absorption spectra of many impurities.

The HPLC-abs results obtained for the first time for the three purest commercially available bmimBF₄ samples proved that the reason for the considerable differences between the absorption spectra of the same compound measured by a spectrophotometer was the presence of absorbing impurities. Even for sample Z containing the smallest amount of absorbing impurities, the contribution of their absorption in the absorption spectrum recorded on a spectrophotometer in the long-wavelength range λ ≥ 250 nm, was significant. It explains the differences between the absorption spectrum measured using spectrophotometer and the absorption spectra of species formed from bmimBF₄, measured using HPLC-abs system. The results have shown that the fact that an IL is colourless is not evidence of their spectral purity. The samples studied, X, Y and Z, were colourless but they contained numerous absorbing impurities.

The use of the HPLC-abs method permits the elimination or at least the minimisation of the effect of impurities on the results obtained, in particular for the compounds whose ε(λ) is low and/or when different species are formed. We believe that only with the help of this method is it possible to confirm the formation of different species and to determine their content, concentration and absorption properties.

4. Experimental section

BmimBF₄ was purchased from three well known companies which specialize in synthesis of ILs of high purity – bmimBF₄ of ≥99% purity from Merck (labelled as X), bmimBF₄ BASF quality of ≥98% purity from Aldrich (labelled as Y) and bmimBF₄ of 99% purity from Ionic Liquids Technologies (labelled as Z). All were used as obtained, without further purification. A concentration of 2.65 M of bmimBF₄ in acetonitrile–water (40/60 v/v) solution was used in this work. HPLC gradient grade acetonitrile (99.9% purity) was purchased from Sigma-Aldrich. Water used was deionised in a deionisation unit HLP5 (Hydrolab, Poland) to ultrapure state suitable for HPLC studies. The water used, similarly to ACN, did not contain impurities absorbing in the range λ > 230 nm.

All chromatographic measurements were performed on the Waters' High Performance Liquid Chromatography (HPLC) system consisting of a pump (Waters; type: 1525), an autosampler (Waters; type: 2707) and spectrophotometer UV-VIS (flow cell with 1 cm path length) with photodiode array (PDA) (Waters; type: 2998), with linear working range for absorbance near A ≤ 2. An Empower 2 chromatographic interface was used for data collection. The Gemini C6-Phenyl 250 × 4.6 mm polar RP column (Phenomenex) packed with 3 μm particles with pore size of 110 Å was used. Isocratic elution was carried out with a mixture of acetonitrile (ACN) and water (40/60 v/v) at the flow rate of 0.4 mL min⁻¹. Injections of 3 μL of concentration of 2.65 M were made. All measurements were performed at ambient temperature. It has been shown that imidazolium ILs can interact with silanol groups on the alkylsilica surface and thus, they found important application as silanol-screening agents in improving the chromatographic analysis.^30–32 Therefore, for each sample, before proper measurements, several injections of the same concentration and injection volume were made to make sure that all silanol groups in the column are shielded so that the equilibrium was not disturbed and the results of a very high reproducibility were obtained. Isocratic elution mode was used and no buffer was added to the eluent so that the system peaks (also known as “ghost peaks”) were avoided.³³ Such conditions ensured that all chromatographic peaks, irrespective of their shape and width, come only from the species formed by bmimBF₄ or from impurities.

Absorption spectra of bmimBF₄ were measured using double beam spectrophotometer type V-550 (Jasco) with a 1 nm spectral bandwidth.

Acknowledgements

The authors wish to thank Prof. Jacek Kubicki for helpful discussions and Dr Gordon L. Hug for help in manuscript preparation. This work was financially supported by the Ministry of Science and Higher Education (MNiSW) Poland, within the project no. N204 265 438.

References

1. H. Weingärtner, Angew. Chem., Int. Ed., 2008, 47, 654.
2. J. F. Wishart and E. W. Castner, J. Phys. Chem. B, 2007, 111, 4639.
3. (a) Ionic Liquids in Synthesis, ed. P. Wasserscheid and T. Welton, VCH, Weinheim, 2008; (b) Topics in Current Chemistry: Ionic Liquids, ed. B. Kirchner, Springer-Verlag, Berlin, 2009, vol. 290.
4. R. D. Rogers and G. A. Voth, Acc. Chem. Res., 2007, 40, 1077.
5. N. V. Plechkova and K. R. Seddon, Chem. Soc. Rev., 2008, 37, 123.
6. (a) N. Siraj, G. Grampp, S. Landgraf and K. Punyain, Z. Phys. Chem., 2013, 227, 105; (b) Y. Shen, Y. Zhang, D. Han, Z. Wang, D. Kuehner and L. Niu, Talanta, 2009, 78, 805; (c) T. Singh and A. Kumar, J. Phys. Chem. B, 2008, 112, 4079.
7. A. Berthod and S. Carda-Broch, J. Liq. Chromatogr. Relat. Technol., 2003, 26, 1493.
8. (a) R. Katoh, Chem. Lett., 2007, 36, 1256; (b) R. Katoh and K. Takahashi, Radiat. Phys. Chem., 2009, 78, 1126.
9. E. Binetti, A. Panniello, L. Triggiani, R. Tommasi, A. Agostiano, M. L. Curri and M. Striccoli, J. Phys. Chem. B, 2012, 116, 3512.
10. A. Paul, P. K. Mandal and A. Samanta, J. Phys. Chem. B, 2005, 109, 9148.
11. M. J. Earle, C. M. Gordon, N. V. Plechkova, K. R. Seddon and T. Welton, Anal. Chem., 2007, 79, 758.
12. A. Paul, P. K. Mandal and A. Samanta, Chem. Phys. Lett., 2005, 402, 375.
13. I. Billard, G. Moutiers, A. Labet, A. El Azzi, C. Gaillard, C. Mariet and K. Lützenkirchen, Inorg. Chem., 2003, 42, 1726.
14. P. Illner, A. Zahl, R. Puchta, N. van Eikema Hommes, P. Wasserscheid and R. van Eldik, J. Organomet. Chem., 2005, 690, 3567.
15. A. Paul and A. Samanta, J. Chem. Sci., 2006, 118, 335.
16. L. Berthon, S. I. Nikitenko, I. Bisel, C. Berthon, M. Faucon, B. Saucerotte, N. Zorz and P. Moisy, Dalton Trans., 2006, 2526.
17. I. Billard, S. Mekki, C. Gaillard, P. Hesemann, G. Moutiers, C. Mariet, A. Labet and J. C. G. Bünzli, Eur. J. Inorg. Chem., 2004, 1190.
18. (a) M. J. Muldoon, A. J. McLean, C. M. Gordon and I. R. Dunkin, Chem. Commun., 2001, 2364; (b) N. L. Lancaster, P. A. Salter, T. Welton and G. B. Young, J. Org. Chem., 2002, 67, 8855.
19. A. K. Burrell, R. E. Del Sesto, S. N. Baker, T. M. McCleskey and G. A. Baker, Green Chem., 2007, 9, 449.
20. P. Bonhôte, A. P. Dias, N. Papageorgiou, K. Kalyanasundaram and M. Grätzel, Inorg. Chem., 1996, 35, 1168.
21. (a) H. Wang, J. Wang, S. Zhang, Y. Pei and K. Zhuo, ChemPhysChem, 2009, 10, 2516; (b) Y. Jeon, J. Sung, C. Seo, H. Lim, H. Cheong, M. Kang, B. Moon, Y. Ouchi and D. Kim, J. Phys. Chem. B, 2008, 112, 4735; (c) Y. Jeon, J. Sung, D. Kim, C. Seo, H. Cheong, Y. Ouchi, R. Ozawa and H. Hamaguchi, J. Phys. Chem. B, 2008, 112, 923; (d) C. T. Martins, B. M. Sato and O. A. El Seoud, J. Phys. Chem. B, 2008, 112, 8330; (e) H. Jin, X. Li and M. Maroncelli, J. Phys. Chem. B, 2007, 111, 13473; (f) A. Triolo, O. Russina, H. Bleif and E. Di Cola, J. Phys. Chem. B, 2007, 111, 4641; (g) Y. Wang and G. A. Voth, J. Am. Chem. Soc., 2005, 127, 12192; (h) J. N. A. Canongia Lopes and A. A. H. Padua, J. Phys. Chem. B, 2006, 110, 3330; (i) S. Chen, S. Zhang, X. Liu, J. Wang, J. Wang, K. Dong and B. Xu, Phys. Chem. Chem. Phys., 2014, 16, 5893; (j) K. Shimizu, C. E. S. Bernardes, A. Triolo and J. N. Canongia Lopes, Phys. Chem. Chem. Phys., 2013, 15, 16256.
22. (a) S. Dorbritz, W. Ruth and U. Kragl, Adv. Synth. Catal., 2005, 347, 1273; (b) G. P. Jackson and D. C. Duckworth, Chem. Commun., 2004, 522; (c) R. Bini, O. Bortolini, C. Chiappe, D. Pieraccini and T. Siciliano, J. Phys. Chem. B, 2007, 111, 598; (d) B. K. Ku and J. F. de la Mora, J. Phys. Chem. B, 2004, 108, 14915; (e) Y.-Z. Zheng, N.-N. Wang, J.-J. Luo, Y. Zhou and Z.-W. Yu, Phys. Chem. Chem. Phys., 2013, 15, 18055; (f) A. Stoppa, J. Hunger, G. Hefter and R. Buchner, J. Phys. Chem. B, 2012, 116, 7509.
23. (a) F. C. Gozzo, L. S. Santos, R. Augusti, C. S. Consorti, J. Dupont and M. N. Eberlin, Chem.–Eur. J., 2004, 10, 6187; (b) B. A. DaSilveira Neto, L. S. Santos, F. M. Nachtigall, M. N. Eberlin and J. Dupont, Angew. Chem., Int. Ed., 2006, 45, 7251.
24. (a) J. Wang, H. Wang, S. Zhang, H. Zhang and Y. Zhao, J. Phys. Chem. B, 2007, 111, 6181; (b) J. Bowers, C. P. Butts, P. J. Martin and M. C. Vergara-Gutierrez, Langmuir, 2004, 20, 2191; (c) J. Wang, H. Wang, S. Zhang, H. Zhang and Y. Zhao, J. Phys. Chem. B, 2007, 111, 6181; (d) I. B. Malham, P. Letellier and M. Turmine, J. Phys. Chem. B, 2006, 110, 14212; (e) H. Katayanagi, K. Nishikawa, H. Shimozaki, K. Miki, P. Westh and Y. Koga, J. Phys. Chem. B, 2004, 108, 19451; (f) Y. Wang and G. A. Voth, J. Phys. Chem. B, 2006, 110, 18601; (g) K. Miki, P. Westh, K. Nishikawa and Y. Koga, J. Phys. Chem. B, 2005, 109, 9014.
25. (a) E. Izgorodina, J. Rigby and D. R. MacFarlane, Chem. Commun., 2012, 48, 1493; (b) E. I. Izgorodina, Phys. Chem. Chem. Phys., 2011, 13, 4189; (c) R. P. Matthews, T. Welton and P. A. Hunt, Phys. Chem. Chem. Phys., 2014, 16, 3238.
26. B. Ciesielska, A. Łukaszewicz, L. Celewicz, A. Maciejewski and J. Kubicki, Appl. Spectrosc., 2007, 61, 102.
27. F. Tang, K. Wu, L. Ding, J. Yuan, Q. Liu, L. Nie and S. Yao, Sep. Purif. Technol., 2008, 60, 245.
28. P. Nockemann, K. Binnemans and K. Driesen, Chem. Phys. Lett., 2005, 415, 131.
29. A. Mele, C. D. Tran and S. H. De Paoli Lacerda, Angew. Chem., Int. Ed., 2003, 42, 4364.
30. (a) L. He, W. Zhang, L. Zhao, X. Liu and S. Jiang, J. Chromatogr. A, 2003, 1007, 39; (b) X. Xiaohua, Z. Liang, L. Xia and J. Shengxiang, Anal. Chim. Acta, 2004, 519, 207.
31. (a) M. P. Marszałł, T. Bączek and R. Kaliszan, Anal. Chim. Acta, 2005, 547, 172; (b) R. Kaliszan, M. P. Marszałł, M. J. Markuszewski, T. Bączek and J. Pernak, J. Chromatogr. A, 2004, 1030, 263.
32. A. Berthod, A. J. Ruiz-Angel and S. Carda-Broch, J. Chromatogr. A, 2008, 1184, 6.
33. (a) F. Gritti and G. Guiochon, J. Chromatogr. A, 2004, 1028, 197; (b) D. V. McCalley, Anal. Chem., 2003, 75, 3404; (c) J. Srbek, P. Coufal, Z. Bosakova and E. Tesarova, J. Sep. Sci., 2005, 28, 1263; (d) S. Williams, J. Chromatogr. A, 2004, 1052, 1.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4ra03868a

---