DOI: 10.1002/cphc.201500433
Communications
Vibrational Dephasing in Ionic Liquids as a Signature of Hydrogen Bonding Susana Chatzipapadopoulos,[b] Tobias Zentel,[b] Ralf Ludwig,*[a, c] Matthias Lìtgens,[b] Stefan Lochbrunner,*[b] and Oliver Kìhn*[b] Understanding both structure and dynamics is crucial for producing tailor-made ionic liquids (ILs). We studied the vibrational and structural dynamics of medium versus weakly hydrogen-bonded C¢H groups of the imidazolium ring in ILs of the type [1-alkyl-3-methylimidazolium][bis(trifluoromethanesulfonyl)imide] ([Cnmim][NTf2]), with n = 1, 2, and 8, by time-resolved coherent anti-Stokes Raman scattering (CARS) and quantum-classical hybrid (QCH) simulations. From the time series of the CARS spectra, dephasing times were extracted by modeling the full nonlinear response. From the QCH calculations, pure dephasing times were obtained by analyzing the distribution of transition frequencies. Experiments and calculations reveal larger dephasing rates for the vibrational stretching modes of C(2)¢H compared with the more weakly hydrogen-bonded C(4,5)¢H. This finding can be understood in terms of different H-bonding motifs and the fast interconversion between them. Differences in population relaxation rates are attributed to Fermi resonance interactions.
Vibrational dynamics in condensed matter provides crucial information on microscopic processes and related macroscopic properties. This is particularly true for ILs, which represent a class of liquid materials with unique properties.[1, 2] ILs are salts that consist purely of ions and are liquid at room temperature. The structure and properties of these Coulomb fluids are mainly determined by the type and strength of the intermolecular interaction between the anions and cations. Understanding of the subtle balance between Coulomb interactions, hydrogen bonding, and dispersion forces is crucial to design IL properties and possible applications.[3, 4] Hydrogen bonding in imidazolium-based ILs is reflected in lengthened intramolecular C¢H bonds, downfield NMR proton chemical shifts, and red[a] Prof. Dr. R. Ludwig Institute of Chemistry, University of Rostock Dr.-Lorenz-Weg 1, 18059 Rostock, (Germany) E-mail: [email protected] [b] S. Chatzipapadopoulos, T. Zentel, M. Lìtgens, Prof. Dr. S. Lochbrunner, Prof. Dr. O. Kìhn Institute of Physics, University of Rostock Universittsplatz 3, 18055 Rostock (Germany) E-mail: [email protected] [email protected] [c] Prof. Dr. R. Ludwig Leibniz-Institute of Catalysis, University of Rostock A.-Einstein-Str. 29a, 18059 Rostock (Germany) Supporting Information for this article is available on the WWW under http://dx.doi.org/10.1002/cphc.201500433.
ChemPhysChem 2015, 16, 2519 – 2523
shifted C¢H stretching frequencies at the ring of the imidazolium cation (see Figure 1).[5, 6] By combining various vibrational spectroscopic techniques and anharmonic frequency calculations, we recently analyzed the mid-infrared spectra of imidazolium-based ILs of type [1-alkyl-3-methylimidazolium][bis(trifluoromethanesulfonyl)imide] ([Cnmim][NTf2]).[7] The focus was
Figure 1. Nomenclature used for the ionic liquids [1-alkyl-3-methylimidazolium][NTf2] as investigated in this work.
on the over- and combination tones of the ring vibrational modes and their interference with the C¢H stretch vibrational frequencies at the imidazolium ring. The Raman band at 3180 cm¢1 and the IR band at 3160 cm¢1 were assigned to the symmetric and antisymmetric combinations of the C(4)¢H and C(5)¢H stretching vibration of the imidazolium ring, respectively. The feature around 3120 cm¢1 comprised two bands and was assigned to the C(2)¢H stretching mode and Fermi resonances of the C¢H stretching vibrations with the overtones of in-plane ring deformations. Our results strongly support the important role of hydrogen bonding in these Coulomb fluids, which has been challenged before.[8–12] Recently, this finding was supported by IR spectroscopy in the vapor phase of jetcooled [C2mim][NTf2].[13, 14] In particular, the redshift of the C(2)¢H stretching frequency compared with the C(4)¢H and C(5)¢H vibrational bands clearly indicates a stronger intermolecular interaction with the NTf2¢ anion, which is enhanced by hydrogen bonding. The redshift of the C(2)¢H stretch is related to the downfield NMR proton chemical shift (d1H) of the corresponding ring proton.[9] Both spectroscopic properties indicate local and directional hydrogen bonding via the C(2)¢H bond. However, the static shift in the vibrational transition energies only reflects the average force exerted by the surrounding ions on the C¢H oscillators. Direct time-resolved observations of
2519
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Communications the structural fluctuations in this frequency range are still missing. The IL environment exerts fluctuating forces on the internal degrees of freedom of the cation, causing fast variations in the structure of the latter. These structural fluctuations affect the amplitude and/or phase of the vibration, resulting in energy dissipation and/or loss of phase correlations. The first process is characterized by the vibrational lifetime T1 and the second by the pure dephasing time T2*, resulting together in the decay of vibrational coherences with a total dephasing time T2 [Eq. (1)]: 1 1 1 ¼ þ T2 T2* 2T1
ð1Þ
To study the dephasing properties of vibrational transitions in liquids, time-resolved nonlinear Raman and IR photon echo techniques have been developed.[15–20] As a special version of time-resolved Raman spectroscopy, time-resolved coherent anti-Stokes Raman scattering (CARS) with ultrashort laser pulses was applied to stretching vibrations in mainly apolar solvents or hydrogen-bonded liquids.[15–19] CARS should also be a reliable method to investigate the influence of the overall IL environment on the vibrational modes of the imidazolium cation. It has indeed already been applied to study vibrational energy transfer in ILs.[21] It is the purpose of the present work to study vibrational dephasing in ILs by CARS spectroscopy in combination with quantum-classical hybrid (QCH) calculations of fluctuating fundamental C¢H transition frequencies. CARS spectroscopy allows the study of relaxation mechanisms even if they contribute little to the overall dynamics. This is particularly important for ILs characterized by small self-diffusion coefficients of the ions and relatively high viscosities. It is expected that vibrational dephasing and lifetimes reflect the different strengths of the hydrogen bonds (HBs) formed via the C(2)¢H and C(4/5)¢H bonds at the imidazolium ring. By comparing ILs based on cations with different alkyl chains, the consistency of the results can be checked. For the first time, the structural dynamics of aprotic ILs are measured on a subpicosecond timescale and supported by QCH calculations to understand vibrational dephasing at the molecular level. Time-resolved CARS spectra were measured for the ILs [Cnmim][NTf2] with n = 1, 2, and 8. The spectra for n = 2 are shown in Figure 2 as function of the delay time of the probe pulse. Spectra for n = 1 and 8 display a similar time evolution (see the Supporting Information). The CARS signal is most intense during cross-correlation, with a maximum around time zero, when all three pulses overlap. There, vibrational bands appear redshifted, due to interference between resonant signals and the nonresonant background. At delay times longer than 1 ps only resonant contributions are present. The corresponding CARS spectra are less intense by two to three orders of magnitude, but are much more structured than during the cross-correlation. The width of the resonances is then given by the bandwidth of the probe pulse,[16–18] leading to a rather high spectral resolution of about 20 cm¢1. To extract resonance frequencies and dephasing times the CARS spectra are modChemPhysChem 2015, 16, 2519 – 2523
www.chemphyschem.org
Figure 2. Time series of CARS spectra of [C2mim][NTf2] on the left. In the middle panel the selected spectra (black dots) at the specified probe delays are compared with modeled spectra (red curves). In the right panel the experimental (black dots) and modeled (red curves) time traces at specific frequencies are shown.
eled as described in the Experimental Section of the Supporting Information. The spectroscopic parameters used in these calculations are adapted to obtain the optimal match with the experiment. The resulting fits are shown in Figure 2 as red curves, and the obtained resonance frequencies and dephasing times are listed for all chain lengths in Table 1. By varying the dephasing constants, we observe significant deviations between fits and measurements for changes larger than 0.05 ps, resulting in a confidence interval of + /¢0.05 ps for the T2 times obtained by CARS. The assignment of the bands is based
2520
Table 1. Frequencies and dephasing times of the C¢H stretch vibrations in the ILs [Cnmim][NTf2]
Mode
Amplitude
Frequency [cm¢1]
Dephasing time [ps], + /¢ 0.05 ps
[C1mim][NTf2] shoulder sym. methyl stretch asym. methyl stretch C(2)¢H + overtone C(2)¢H + overtone as. C(4/5)¢H sym. C(4/5)¢H
0.32 1 0.76 0.34 0.40 0.24 0.58
2944 2973 3035 3119 3138 3164 3184
0.29 0.53 0.14 0.25 0.25 0.26 0.35
[C2mim][NTf2] ethyl stretch ethyl stretch sym. methyl stretch C(2)¢H + overtone C(2)¢H + overtone as. C(4/5)¢H sym. C(4/5)¢H
0.28 0.47 1 1.02 0.67 1.04 0.90
2929 2950 2970 3105 3128 3155 3175
0.42 0.59 0.50 0.27 0.28 0.31 0.46
[C8mim][NTf2] stretch in octyl group stretch in octyl group stretch in octyl group stretch in octyl group sym. methyl stretch C(2)¢H + overtone C(2)¢H + overtone as. C(4/5)¢H sym. C(4/5)¢H
0.99 0.19 0.58 0.67 1 0.95 0.47 0.68 0.67
2850 2875 2903 2942 2970 3100 3126 3156 3174
0.31 0.33 0.33 0.45 0.45 0.22 0.23 0.29 0.50
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Communications on calculated Raman spectra, which are shown in the Supporting Information. For all three ILs, two main spectral regions can be distinguished. One extends from 2800 to 3000 cm¢1 and exhibits bands caused by the C¢H stretch modes of the alkyl groups bound to the nitrogen atoms of the imidazolium ring. It shows a strong variation with IL and a pronounced rise of complexity with increasing length of the alkyl group at ring position 1. The second region spans from 3000 to 3200 cm¢1 and displays the spectral signatures of the C¢H stretching vibrations of the imidazolium ring, namely C(2)¢H and the antisymmetric and the symmetric C(4/5)¢H modes. For all ILs, the CARS spectra exhibit a prominent feature at approximately 2970 cm¢1. This mode stems from the methyl group at ring position 3, which is the same for all studied ILs. Accordingly, the corresponding frequency is similar in all cases. The associated dephasing time T2 exhibits a small decrease with increasing length of the alkyl chain and varies from 0.53 over 0.50 to 0.45 ps for n = 1, 2, and 8 (see Table 1). In general, the spectra and their analysis become more complex with increasing chain length and the number of resonances needed to model the CARS response in the alkyl region increases up to five for [C8mim][NTf2]. The number of Raman active vibrations is even higher, but some of them are not strong enough to cause a significant contribution to the CARS signal or they are included in an effective mode, which represents a set of close lying resonances. Despite these complications some trends can be identified. The decrease in T2 indicates that the structural flexibility becomes somewhat larger with increasing alkyl chain lengths, as was also observed by NMR relaxation time experiments.[22] The other prominent feature in the alkyl region stems from the end-standing CH3 group of the ethyl (n = 2) or octyl group (n = 8). In accordance with the calculated Raman spectra (see the Supporting Information), this mode shifts to the red from 2950 cm¢1, for n = 2, to 2942 cm¢1, for n = 8. For this mode, a decrease of T2 with increasing chain length is observed again (see Table 1). In agreement with literature,[8–10] the C(4/5)¢H stretching modes are found at approximately 3180 and 3160 cm¢1. The first stems from the symmetric Raman active C(4/5)¢H stretching mode, whereas the latter originates from the antisymmetric, IR active C(4/5)¢H stretching mode. In the following, only the Raman active C(4/5)¢H mode is discussed, as it has the much larger cross section. With increasing length of the alkyl chain an increase of the corresponding T2 time from 0.35 ps, for n = 1, over 0.46 ps, for n = 2, to 0.5 ps, for n = 8, is observed. This increase is in contrast to the dephasing behavior as a function of the chain length for the alkyl modes. The increase might be related to the increase in viscosity by a factor of four from n = 1 to 8.[23] The most important feature in the imidazolium region is the C(2)¢H mode, which is redshifted compared to the other ring C¢H modes, owing to hydrogen bonding. It is involved in a Fermi resonance with a ring overtone, thus splitting into two subfeatures.[8–10] The two subbands are found at 3119 and 3137 cm¢1 for [C1mim][NTf2], at 3105 and 3128 cm¢1 for [C2mim][NTf2] and at 3100 and 3126 cm¢1 for [C8mim][NTf2], ChemPhysChem 2015, 16, 2519 – 2523
www.chemphyschem.org
thus indicating that the redshift even increases slightly with larger chain length. This finding points to a strengthening of the HBs with increasing chain length, which might result from the formation of a stronger ion-pair, due to steric effects. The most striking fact for the C(2)¢H modes in all studied ILs are the small T2 values compared to the T2 times of the symmetric Raman active C(4,5)¢H modes. This fast dephasing, either resulting from pure dephasing (T2*) or from both, pure dephasing and population relaxation (T1), needs to be further analyzed by considering possible structural motifs for the hydrogen bonding of the C(2)¢H group and their dynamics as it is discussed below. Molecular dynamics simulations were done for [C2mim][NTf2] (see the Supporting Information for details). For selected target molecules QCH calculations of fundamental C¢H transition frequencies were performed. The distributions of frequencies are shown in Figure 3. In all cases, we obtain a broad dis-
Figure 3. Smoothed distributions of transition frequencies calculated for the three imidazolium C¢H bond vibrations of [C2mim][NTf2] centered at the respective average transition frequencies (w0).
tribution, but a detailed statistical analysis gives different full widths at half maximum (FWHM) for the three vibrational modes. Upon converting the FWHM to timescales, the resulting T2* times are 0.42 ps for C(2)¢H, 0.50 ps for the symmetric C(4/5)¢H, and 0.54 ps for the antisymmetric C(4/5)¢H vibration. Although the shape of the distribution for the two C(4/5)¢H modes is similar, the broader C(2)¢H distribution deserves special attention, as it appears to be bimodal. It is tempting to assign this observation to different HB motifs. To scrutinize this assumption we investigated the distribution of the H atom in the plane defined by the HB donor (C) and the two closest acceptor oxygen atoms (see Figure 4). For those configurations that contribute to the main peak, the H atom is on average positioned such as to form a bifurcated HB (Figure 4 A), that is, the maximum of the distribution is midway between the configuration in which the C¢H bond points towards an oxygen atom. In contrast, the blueshifted shoulder is due to configurations in which the maximum H-atom distribution evidences a coordination towards one acceptor oxygen atom (i.e. the left one in Figure 4 B).
2521
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Communications
Figure 4. Density plot of the position of the two closest HB-acceptor O atoms (red) relative to the C(2) atom positioned at (0,0) in the C¢O¢O plane. Notice that along the trajectory the two closest O atoms are either O1 and O2 (see upper panel) or O2 and O3. The density of the projected H-atom position is given in blue. A) Geometries contributing the main peak in the spectrum shown in Figure 3 [(18 + /¢10)cm¢1]. B) Geometries contributing to the blueshifted peak at [(20 + /¢10) cm¢1].
The question arises, of how these finding are related to the dephasing times extracted from the CARS measurements (see Table 1). In principle both theory and experiment predict dephasing times that are shorter for C(2)¢H compared with those for C(4/5)¢H. For the comparison, one should keep in mind, that the present model is able to extract the pure dephasing time, T2*, only, whereas the experimental data also contain the effect of population relaxation according to Equation (1). Hence, a possible interpretation would be that the differences are related to different population relaxations. Estimating the population relaxation times on the basis of the analysis in Figure 3, and using the experimental data for T2, we obtain T1 = 0.42 and 2.9 ps for the C(2)¢H and the symmetric C(4/5)¢H mode. This indicates rather different behavior of the two HBs, which is, however, not unreasonable given the fact that the HB formed by C(2)¢H is stronger and more affected by the Fermi-resonance with the overtone of the imidazolium ChemPhysChem 2015, 16, 2519 – 2523
www.chemphyschem.org
ring vibration.[7] Indeed, ultrafast vibrational relaxation in the few hundred fs range, due to a Fermi-resonance in a Hbonded system, has been demonstrated by IR pump–probe spectroscopy.[24] In this respect one might argue whether the simple Bloch-type relaxation and dephasing model used in the analysis of experimental and theoretical data is adequate. Indeed, the bimodal distributions in Figures 3 and 4 indicate that the situation is much more complicated, comprising of different time scales (fluctuations and change of HB configuration), which adds to the effect of Fermi-resonance interaction. The present study gives insights into the H bonding and vibrational dynamics of the ionic liquids [Cnmim][NTf2] by combining results from time-resolved CARS measurements with QCH simulations. Previous stationary spectroscopic studies showed that the C(2)¢H bond is more strongly H bonded than the C(4/5)¢H ones.[7–10] Here, we investigated how these HB characteristics are reflected by the fluctuation dynamics. Ultrafast CARS measurements showed that the difference in H bonding results in a faster dephasing of the C(2)¢H mode compared with the C(4/5)¢H mode. This is also observed in the QCH simulations on [C2mim][NTf2], which in addition gave evidence for two frequently interconverting motifs of H bonding for the C(2)¢H moiety. One motif consists of a linear HB, whereas the other one is a bifurcated double HB. Upon closer inspection, the CARS measurements reveal dephasing times, T2, that are shorter than the pure dephasing times, T2*, extracted from the QCH simulations. The difference is particularly pronounced for the C(2)¢H mode; this is attributed to the much shorter vibrational lifetime of the C(2)¢H mode compared with the C(4/5)¢H mode. This could be a consequence of the involvement of the C(2)¢H stretching mode in a Fermi resonance with a ring overtone vibration,[7] which opens efficient relaxation channels. This is in agreement with the observation that the band assigned to the C(2)¢H mode assigned in Ref. [7] consists of two subfeatures that reflect the two states resulting from a Fermi resonance. Changing the alkyl-chain length impacts the measured dephasing times of several alkyl modes and of the symmetric C(4,5)¢H mode, indicating a change in the overall structural dynamics of the whole IL. Additionally, a redshift of the C(2)¢H mode is observed with increasing alkyl chain length; this also points to a structural change, for example, a stronger ion-pair formation within the IL. Both facts indicate that the alkyl-chain length can possibly be used to tune structural dynamics in ILs. Keywords: coherent anti-Stokes Raman scattering · hydrogen bonding · ionic liquids · quantum chemical calculations · vibrational dephasing
2522
[1] Ionic Liquids in Synthesis (Ed.: P. Wasserscheid and T. Welton), VCH-Wiley, Weinheim, 2007, 2nd Edition. [2] T. Welton, Chem. Rev. 1999, 99, 2071 – 2084. [3] F. Endres, S. Z. E. Abedin, Phys. Chem. Chem. Phys. 2006, 8, 2101 – 2116. [4] H. Weingrtner, Angew. Chem. Int. Ed. 2008, 47, 654 – 670; Angew. Chem. 2008, 120, 664 – 682. [5] A. Wulf, K. Fumino, R. Ludwig, Angew. Chem. Int. Ed. 2010, 49, 449 – 453; Angew. Chem. 2010, 122, 459 – 463.
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Communications [6] K. Fumino, T. Peppel, M. Geppert-Rybczyn´ska, D. H. Zaitsau, J. K. Lehmann, S. P. Verevkin, M. Kçckerling, R. Ludwig, Phys. Chem. Chem. Phys. 2011, 13, 14064 – 14075. [7] C. Roth, S. Chatzipapadopoulos, D. Kerl¦, F. Friedriszik, M. Lìtgens, S. Lochbrunner, O. Kìhn, R. Ludwig, New J. Phys. 2012, 14, 105026. [8] T. Kçddermann, C. Wertz, A. Heintz, R. Ludwig, ChemPhysChem 2006, 7, 1944 – 1949. [9] A. Wulf, K. Fumino, D. Michalik, R. Ludwig, ChemPhysChem 2007, 8, 2265 – 2269. [10] A. Wulf, K. Fumino, R. Ludwig, J. Phys. Chem. A 2010, 114, 685 – 686. [11] J.-C. LassÀgues, J. Grondin, D. Cavagnat, P. Johansson, J. Phys. Chem. A 2009, 113, 6419 – 6421. [12] Y. Danten, M. I. Cabaco, M. Besnard, J. Phys. Chem. A 2009, 113, 2873 – 2889. [13] R. Cooper, A. M. Zolot, J. A. Boatz, D. P. Sporleder, J. A. Stearns, J. Phys. Chem. A 2013, 117, 12419 – 12428. [14] S. Cha, M. Ao, W. Sung, B. Moon, B. Ahlstrçm, P. Johansson, Y. Ouchi, D. Kim, Phys. Chem. Chem. Phys. 2014, 16, 9591 – 9601. [15] A. Laubereau, W. Kaiser, Rev. Mod. Phys. 1978, 50, 607 – 665.
ChemPhysChem 2015, 16, 2519 – 2523
www.chemphyschem.org
[16] W. Zinth, A. Laubereau, W. Kaiser, Opt. Commun. 1978, 26, 457 – 462. [17] W. Zinth, M. C. Nuss, W. Kaiser, Chem. Phys. Lett. 1982, 88, 257 – 261. [18] M. Lìtgens, S. Chatzipapadopoulos, S. Lochbrunner, Opt. Express 2012, 20, 6478 – 6487. [19] M. Lìtgens, S. Chatzipapadopoulos, F. Friedriszik, S. Lochbrunner, J. Raman Spectrosc. 2014, 45, 359 – 368. [20] E. T. J. Nibbering, T. Elsaesser, Chem. Rev. 2004, 104, 1887 – 1914. [21] M. Namboodiri, M. M. Kazemi, T. Zeb Khan, A. Materny, J. Kiefer, J. Am. Chem. Soc. 2014, 136, 6136 – 6141. [22] J. R. Lyerla, T. T. Horlkawa, J. Phys. Chem. 1976, 80, 1106 – 1112. [23] S. Zhang, N. Sun, X. He, X. Lu, X. Zhang, J. Phys. Chem. Ref. Data 2006, 35, 1475 – 1517. [24] K. Heyne, E. T. J. Nibbering, T. Elsaesser, M. Petkovic´, O. Kìhn, J. Phys. Chem. A 2004, 108, 6083 – 6086.
Received: June 1, 2015 Published online on July 14, 2015
2523
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Communications
Vibrational Dephasing in Ionic Liquids as a Signature of Hydrogen Bonding Susana Chatzipapadopoulos,[b] Tobias Zentel,[b] Ralf Ludwig,*[a, c] Matthias Lìtgens,[b] Stefan Lochbrunner,*[b] and Oliver Kìhn*[b] Understanding both structure and dynamics is crucial for producing tailor-made ionic liquids (ILs). We studied the vibrational and structural dynamics of medium versus weakly hydrogen-bonded C¢H groups of the imidazolium ring in ILs of the type [1-alkyl-3-methylimidazolium][bis(trifluoromethanesulfonyl)imide] ([Cnmim][NTf2]), with n = 1, 2, and 8, by time-resolved coherent anti-Stokes Raman scattering (CARS) and quantum-classical hybrid (QCH) simulations. From the time series of the CARS spectra, dephasing times were extracted by modeling the full nonlinear response. From the QCH calculations, pure dephasing times were obtained by analyzing the distribution of transition frequencies. Experiments and calculations reveal larger dephasing rates for the vibrational stretching modes of C(2)¢H compared with the more weakly hydrogen-bonded C(4,5)¢H. This finding can be understood in terms of different H-bonding motifs and the fast interconversion between them. Differences in population relaxation rates are attributed to Fermi resonance interactions.
Vibrational dynamics in condensed matter provides crucial information on microscopic processes and related macroscopic properties. This is particularly true for ILs, which represent a class of liquid materials with unique properties.[1, 2] ILs are salts that consist purely of ions and are liquid at room temperature. The structure and properties of these Coulomb fluids are mainly determined by the type and strength of the intermolecular interaction between the anions and cations. Understanding of the subtle balance between Coulomb interactions, hydrogen bonding, and dispersion forces is crucial to design IL properties and possible applications.[3, 4] Hydrogen bonding in imidazolium-based ILs is reflected in lengthened intramolecular C¢H bonds, downfield NMR proton chemical shifts, and red[a] Prof. Dr. R. Ludwig Institute of Chemistry, University of Rostock Dr.-Lorenz-Weg 1, 18059 Rostock, (Germany) E-mail: [email protected] [b] S. Chatzipapadopoulos, T. Zentel, M. Lìtgens, Prof. Dr. S. Lochbrunner, Prof. Dr. O. Kìhn Institute of Physics, University of Rostock Universittsplatz 3, 18055 Rostock (Germany) E-mail: [email protected] [email protected] [c] Prof. Dr. R. Ludwig Leibniz-Institute of Catalysis, University of Rostock A.-Einstein-Str. 29a, 18059 Rostock (Germany) Supporting Information for this article is available on the WWW under http://dx.doi.org/10.1002/cphc.201500433.
ChemPhysChem 2015, 16, 2519 – 2523
shifted C¢H stretching frequencies at the ring of the imidazolium cation (see Figure 1).[5, 6] By combining various vibrational spectroscopic techniques and anharmonic frequency calculations, we recently analyzed the mid-infrared spectra of imidazolium-based ILs of type [1-alkyl-3-methylimidazolium][bis(trifluoromethanesulfonyl)imide] ([Cnmim][NTf2]).[7] The focus was
Figure 1. Nomenclature used for the ionic liquids [1-alkyl-3-methylimidazolium][NTf2] as investigated in this work.
on the over- and combination tones of the ring vibrational modes and their interference with the C¢H stretch vibrational frequencies at the imidazolium ring. The Raman band at 3180 cm¢1 and the IR band at 3160 cm¢1 were assigned to the symmetric and antisymmetric combinations of the C(4)¢H and C(5)¢H stretching vibration of the imidazolium ring, respectively. The feature around 3120 cm¢1 comprised two bands and was assigned to the C(2)¢H stretching mode and Fermi resonances of the C¢H stretching vibrations with the overtones of in-plane ring deformations. Our results strongly support the important role of hydrogen bonding in these Coulomb fluids, which has been challenged before.[8–12] Recently, this finding was supported by IR spectroscopy in the vapor phase of jetcooled [C2mim][NTf2].[13, 14] In particular, the redshift of the C(2)¢H stretching frequency compared with the C(4)¢H and C(5)¢H vibrational bands clearly indicates a stronger intermolecular interaction with the NTf2¢ anion, which is enhanced by hydrogen bonding. The redshift of the C(2)¢H stretch is related to the downfield NMR proton chemical shift (d1H) of the corresponding ring proton.[9] Both spectroscopic properties indicate local and directional hydrogen bonding via the C(2)¢H bond. However, the static shift in the vibrational transition energies only reflects the average force exerted by the surrounding ions on the C¢H oscillators. Direct time-resolved observations of
2519
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Communications the structural fluctuations in this frequency range are still missing. The IL environment exerts fluctuating forces on the internal degrees of freedom of the cation, causing fast variations in the structure of the latter. These structural fluctuations affect the amplitude and/or phase of the vibration, resulting in energy dissipation and/or loss of phase correlations. The first process is characterized by the vibrational lifetime T1 and the second by the pure dephasing time T2*, resulting together in the decay of vibrational coherences with a total dephasing time T2 [Eq. (1)]: 1 1 1 ¼ þ T2 T2* 2T1
ð1Þ
To study the dephasing properties of vibrational transitions in liquids, time-resolved nonlinear Raman and IR photon echo techniques have been developed.[15–20] As a special version of time-resolved Raman spectroscopy, time-resolved coherent anti-Stokes Raman scattering (CARS) with ultrashort laser pulses was applied to stretching vibrations in mainly apolar solvents or hydrogen-bonded liquids.[15–19] CARS should also be a reliable method to investigate the influence of the overall IL environment on the vibrational modes of the imidazolium cation. It has indeed already been applied to study vibrational energy transfer in ILs.[21] It is the purpose of the present work to study vibrational dephasing in ILs by CARS spectroscopy in combination with quantum-classical hybrid (QCH) calculations of fluctuating fundamental C¢H transition frequencies. CARS spectroscopy allows the study of relaxation mechanisms even if they contribute little to the overall dynamics. This is particularly important for ILs characterized by small self-diffusion coefficients of the ions and relatively high viscosities. It is expected that vibrational dephasing and lifetimes reflect the different strengths of the hydrogen bonds (HBs) formed via the C(2)¢H and C(4/5)¢H bonds at the imidazolium ring. By comparing ILs based on cations with different alkyl chains, the consistency of the results can be checked. For the first time, the structural dynamics of aprotic ILs are measured on a subpicosecond timescale and supported by QCH calculations to understand vibrational dephasing at the molecular level. Time-resolved CARS spectra were measured for the ILs [Cnmim][NTf2] with n = 1, 2, and 8. The spectra for n = 2 are shown in Figure 2 as function of the delay time of the probe pulse. Spectra for n = 1 and 8 display a similar time evolution (see the Supporting Information). The CARS signal is most intense during cross-correlation, with a maximum around time zero, when all three pulses overlap. There, vibrational bands appear redshifted, due to interference between resonant signals and the nonresonant background. At delay times longer than 1 ps only resonant contributions are present. The corresponding CARS spectra are less intense by two to three orders of magnitude, but are much more structured than during the cross-correlation. The width of the resonances is then given by the bandwidth of the probe pulse,[16–18] leading to a rather high spectral resolution of about 20 cm¢1. To extract resonance frequencies and dephasing times the CARS spectra are modChemPhysChem 2015, 16, 2519 – 2523
www.chemphyschem.org
Figure 2. Time series of CARS spectra of [C2mim][NTf2] on the left. In the middle panel the selected spectra (black dots) at the specified probe delays are compared with modeled spectra (red curves). In the right panel the experimental (black dots) and modeled (red curves) time traces at specific frequencies are shown.
eled as described in the Experimental Section of the Supporting Information. The spectroscopic parameters used in these calculations are adapted to obtain the optimal match with the experiment. The resulting fits are shown in Figure 2 as red curves, and the obtained resonance frequencies and dephasing times are listed for all chain lengths in Table 1. By varying the dephasing constants, we observe significant deviations between fits and measurements for changes larger than 0.05 ps, resulting in a confidence interval of + /¢0.05 ps for the T2 times obtained by CARS. The assignment of the bands is based
2520
Table 1. Frequencies and dephasing times of the C¢H stretch vibrations in the ILs [Cnmim][NTf2]
Mode
Amplitude
Frequency [cm¢1]
Dephasing time [ps], + /¢ 0.05 ps
[C1mim][NTf2] shoulder sym. methyl stretch asym. methyl stretch C(2)¢H + overtone C(2)¢H + overtone as. C(4/5)¢H sym. C(4/5)¢H
0.32 1 0.76 0.34 0.40 0.24 0.58
2944 2973 3035 3119 3138 3164 3184
0.29 0.53 0.14 0.25 0.25 0.26 0.35
[C2mim][NTf2] ethyl stretch ethyl stretch sym. methyl stretch C(2)¢H + overtone C(2)¢H + overtone as. C(4/5)¢H sym. C(4/5)¢H
0.28 0.47 1 1.02 0.67 1.04 0.90
2929 2950 2970 3105 3128 3155 3175
0.42 0.59 0.50 0.27 0.28 0.31 0.46
[C8mim][NTf2] stretch in octyl group stretch in octyl group stretch in octyl group stretch in octyl group sym. methyl stretch C(2)¢H + overtone C(2)¢H + overtone as. C(4/5)¢H sym. C(4/5)¢H
0.99 0.19 0.58 0.67 1 0.95 0.47 0.68 0.67
2850 2875 2903 2942 2970 3100 3126 3156 3174
0.31 0.33 0.33 0.45 0.45 0.22 0.23 0.29 0.50
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Communications on calculated Raman spectra, which are shown in the Supporting Information. For all three ILs, two main spectral regions can be distinguished. One extends from 2800 to 3000 cm¢1 and exhibits bands caused by the C¢H stretch modes of the alkyl groups bound to the nitrogen atoms of the imidazolium ring. It shows a strong variation with IL and a pronounced rise of complexity with increasing length of the alkyl group at ring position 1. The second region spans from 3000 to 3200 cm¢1 and displays the spectral signatures of the C¢H stretching vibrations of the imidazolium ring, namely C(2)¢H and the antisymmetric and the symmetric C(4/5)¢H modes. For all ILs, the CARS spectra exhibit a prominent feature at approximately 2970 cm¢1. This mode stems from the methyl group at ring position 3, which is the same for all studied ILs. Accordingly, the corresponding frequency is similar in all cases. The associated dephasing time T2 exhibits a small decrease with increasing length of the alkyl chain and varies from 0.53 over 0.50 to 0.45 ps for n = 1, 2, and 8 (see Table 1). In general, the spectra and their analysis become more complex with increasing chain length and the number of resonances needed to model the CARS response in the alkyl region increases up to five for [C8mim][NTf2]. The number of Raman active vibrations is even higher, but some of them are not strong enough to cause a significant contribution to the CARS signal or they are included in an effective mode, which represents a set of close lying resonances. Despite these complications some trends can be identified. The decrease in T2 indicates that the structural flexibility becomes somewhat larger with increasing alkyl chain lengths, as was also observed by NMR relaxation time experiments.[22] The other prominent feature in the alkyl region stems from the end-standing CH3 group of the ethyl (n = 2) or octyl group (n = 8). In accordance with the calculated Raman spectra (see the Supporting Information), this mode shifts to the red from 2950 cm¢1, for n = 2, to 2942 cm¢1, for n = 8. For this mode, a decrease of T2 with increasing chain length is observed again (see Table 1). In agreement with literature,[8–10] the C(4/5)¢H stretching modes are found at approximately 3180 and 3160 cm¢1. The first stems from the symmetric Raman active C(4/5)¢H stretching mode, whereas the latter originates from the antisymmetric, IR active C(4/5)¢H stretching mode. In the following, only the Raman active C(4/5)¢H mode is discussed, as it has the much larger cross section. With increasing length of the alkyl chain an increase of the corresponding T2 time from 0.35 ps, for n = 1, over 0.46 ps, for n = 2, to 0.5 ps, for n = 8, is observed. This increase is in contrast to the dephasing behavior as a function of the chain length for the alkyl modes. The increase might be related to the increase in viscosity by a factor of four from n = 1 to 8.[23] The most important feature in the imidazolium region is the C(2)¢H mode, which is redshifted compared to the other ring C¢H modes, owing to hydrogen bonding. It is involved in a Fermi resonance with a ring overtone, thus splitting into two subfeatures.[8–10] The two subbands are found at 3119 and 3137 cm¢1 for [C1mim][NTf2], at 3105 and 3128 cm¢1 for [C2mim][NTf2] and at 3100 and 3126 cm¢1 for [C8mim][NTf2], ChemPhysChem 2015, 16, 2519 – 2523
www.chemphyschem.org
thus indicating that the redshift even increases slightly with larger chain length. This finding points to a strengthening of the HBs with increasing chain length, which might result from the formation of a stronger ion-pair, due to steric effects. The most striking fact for the C(2)¢H modes in all studied ILs are the small T2 values compared to the T2 times of the symmetric Raman active C(4,5)¢H modes. This fast dephasing, either resulting from pure dephasing (T2*) or from both, pure dephasing and population relaxation (T1), needs to be further analyzed by considering possible structural motifs for the hydrogen bonding of the C(2)¢H group and their dynamics as it is discussed below. Molecular dynamics simulations were done for [C2mim][NTf2] (see the Supporting Information for details). For selected target molecules QCH calculations of fundamental C¢H transition frequencies were performed. The distributions of frequencies are shown in Figure 3. In all cases, we obtain a broad dis-
Figure 3. Smoothed distributions of transition frequencies calculated for the three imidazolium C¢H bond vibrations of [C2mim][NTf2] centered at the respective average transition frequencies (w0).
tribution, but a detailed statistical analysis gives different full widths at half maximum (FWHM) for the three vibrational modes. Upon converting the FWHM to timescales, the resulting T2* times are 0.42 ps for C(2)¢H, 0.50 ps for the symmetric C(4/5)¢H, and 0.54 ps for the antisymmetric C(4/5)¢H vibration. Although the shape of the distribution for the two C(4/5)¢H modes is similar, the broader C(2)¢H distribution deserves special attention, as it appears to be bimodal. It is tempting to assign this observation to different HB motifs. To scrutinize this assumption we investigated the distribution of the H atom in the plane defined by the HB donor (C) and the two closest acceptor oxygen atoms (see Figure 4). For those configurations that contribute to the main peak, the H atom is on average positioned such as to form a bifurcated HB (Figure 4 A), that is, the maximum of the distribution is midway between the configuration in which the C¢H bond points towards an oxygen atom. In contrast, the blueshifted shoulder is due to configurations in which the maximum H-atom distribution evidences a coordination towards one acceptor oxygen atom (i.e. the left one in Figure 4 B).
2521
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Communications
Figure 4. Density plot of the position of the two closest HB-acceptor O atoms (red) relative to the C(2) atom positioned at (0,0) in the C¢O¢O plane. Notice that along the trajectory the two closest O atoms are either O1 and O2 (see upper panel) or O2 and O3. The density of the projected H-atom position is given in blue. A) Geometries contributing the main peak in the spectrum shown in Figure 3 [(18 + /¢10)cm¢1]. B) Geometries contributing to the blueshifted peak at [(20 + /¢10) cm¢1].
The question arises, of how these finding are related to the dephasing times extracted from the CARS measurements (see Table 1). In principle both theory and experiment predict dephasing times that are shorter for C(2)¢H compared with those for C(4/5)¢H. For the comparison, one should keep in mind, that the present model is able to extract the pure dephasing time, T2*, only, whereas the experimental data also contain the effect of population relaxation according to Equation (1). Hence, a possible interpretation would be that the differences are related to different population relaxations. Estimating the population relaxation times on the basis of the analysis in Figure 3, and using the experimental data for T2, we obtain T1 = 0.42 and 2.9 ps for the C(2)¢H and the symmetric C(4/5)¢H mode. This indicates rather different behavior of the two HBs, which is, however, not unreasonable given the fact that the HB formed by C(2)¢H is stronger and more affected by the Fermi-resonance with the overtone of the imidazolium ChemPhysChem 2015, 16, 2519 – 2523
www.chemphyschem.org
ring vibration.[7] Indeed, ultrafast vibrational relaxation in the few hundred fs range, due to a Fermi-resonance in a Hbonded system, has been demonstrated by IR pump–probe spectroscopy.[24] In this respect one might argue whether the simple Bloch-type relaxation and dephasing model used in the analysis of experimental and theoretical data is adequate. Indeed, the bimodal distributions in Figures 3 and 4 indicate that the situation is much more complicated, comprising of different time scales (fluctuations and change of HB configuration), which adds to the effect of Fermi-resonance interaction. The present study gives insights into the H bonding and vibrational dynamics of the ionic liquids [Cnmim][NTf2] by combining results from time-resolved CARS measurements with QCH simulations. Previous stationary spectroscopic studies showed that the C(2)¢H bond is more strongly H bonded than the C(4/5)¢H ones.[7–10] Here, we investigated how these HB characteristics are reflected by the fluctuation dynamics. Ultrafast CARS measurements showed that the difference in H bonding results in a faster dephasing of the C(2)¢H mode compared with the C(4/5)¢H mode. This is also observed in the QCH simulations on [C2mim][NTf2], which in addition gave evidence for two frequently interconverting motifs of H bonding for the C(2)¢H moiety. One motif consists of a linear HB, whereas the other one is a bifurcated double HB. Upon closer inspection, the CARS measurements reveal dephasing times, T2, that are shorter than the pure dephasing times, T2*, extracted from the QCH simulations. The difference is particularly pronounced for the C(2)¢H mode; this is attributed to the much shorter vibrational lifetime of the C(2)¢H mode compared with the C(4/5)¢H mode. This could be a consequence of the involvement of the C(2)¢H stretching mode in a Fermi resonance with a ring overtone vibration,[7] which opens efficient relaxation channels. This is in agreement with the observation that the band assigned to the C(2)¢H mode assigned in Ref. [7] consists of two subfeatures that reflect the two states resulting from a Fermi resonance. Changing the alkyl-chain length impacts the measured dephasing times of several alkyl modes and of the symmetric C(4,5)¢H mode, indicating a change in the overall structural dynamics of the whole IL. Additionally, a redshift of the C(2)¢H mode is observed with increasing alkyl chain length; this also points to a structural change, for example, a stronger ion-pair formation within the IL. Both facts indicate that the alkyl-chain length can possibly be used to tune structural dynamics in ILs. Keywords: coherent anti-Stokes Raman scattering · hydrogen bonding · ionic liquids · quantum chemical calculations · vibrational dephasing
2522
[1] Ionic Liquids in Synthesis (Ed.: P. Wasserscheid and T. Welton), VCH-Wiley, Weinheim, 2007, 2nd Edition. [2] T. Welton, Chem. Rev. 1999, 99, 2071 – 2084. [3] F. Endres, S. Z. E. Abedin, Phys. Chem. Chem. Phys. 2006, 8, 2101 – 2116. [4] H. Weingrtner, Angew. Chem. Int. Ed. 2008, 47, 654 – 670; Angew. Chem. 2008, 120, 664 – 682. [5] A. Wulf, K. Fumino, R. Ludwig, Angew. Chem. Int. Ed. 2010, 49, 449 – 453; Angew. Chem. 2010, 122, 459 – 463.
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Communications [6] K. Fumino, T. Peppel, M. Geppert-Rybczyn´ska, D. H. Zaitsau, J. K. Lehmann, S. P. Verevkin, M. Kçckerling, R. Ludwig, Phys. Chem. Chem. Phys. 2011, 13, 14064 – 14075. [7] C. Roth, S. Chatzipapadopoulos, D. Kerl¦, F. Friedriszik, M. Lìtgens, S. Lochbrunner, O. Kìhn, R. Ludwig, New J. Phys. 2012, 14, 105026. [8] T. Kçddermann, C. Wertz, A. Heintz, R. Ludwig, ChemPhysChem 2006, 7, 1944 – 1949. [9] A. Wulf, K. Fumino, D. Michalik, R. Ludwig, ChemPhysChem 2007, 8, 2265 – 2269. [10] A. Wulf, K. Fumino, R. Ludwig, J. Phys. Chem. A 2010, 114, 685 – 686. [11] J.-C. LassÀgues, J. Grondin, D. Cavagnat, P. Johansson, J. Phys. Chem. A 2009, 113, 6419 – 6421. [12] Y. Danten, M. I. Cabaco, M. Besnard, J. Phys. Chem. A 2009, 113, 2873 – 2889. [13] R. Cooper, A. M. Zolot, J. A. Boatz, D. P. Sporleder, J. A. Stearns, J. Phys. Chem. A 2013, 117, 12419 – 12428. [14] S. Cha, M. Ao, W. Sung, B. Moon, B. Ahlstrçm, P. Johansson, Y. Ouchi, D. Kim, Phys. Chem. Chem. Phys. 2014, 16, 9591 – 9601. [15] A. Laubereau, W. Kaiser, Rev. Mod. Phys. 1978, 50, 607 – 665.
ChemPhysChem 2015, 16, 2519 – 2523
www.chemphyschem.org
[16] W. Zinth, A. Laubereau, W. Kaiser, Opt. Commun. 1978, 26, 457 – 462. [17] W. Zinth, M. C. Nuss, W. Kaiser, Chem. Phys. Lett. 1982, 88, 257 – 261. [18] M. Lìtgens, S. Chatzipapadopoulos, S. Lochbrunner, Opt. Express 2012, 20, 6478 – 6487. [19] M. Lìtgens, S. Chatzipapadopoulos, F. Friedriszik, S. Lochbrunner, J. Raman Spectrosc. 2014, 45, 359 – 368. [20] E. T. J. Nibbering, T. Elsaesser, Chem. Rev. 2004, 104, 1887 – 1914. [21] M. Namboodiri, M. M. Kazemi, T. Zeb Khan, A. Materny, J. Kiefer, J. Am. Chem. Soc. 2014, 136, 6136 – 6141. [22] J. R. Lyerla, T. T. Horlkawa, J. Phys. Chem. 1976, 80, 1106 – 1112. [23] S. Zhang, N. Sun, X. He, X. Lu, X. Zhang, J. Phys. Chem. Ref. Data 2006, 35, 1475 – 1517. [24] K. Heyne, E. T. J. Nibbering, T. Elsaesser, M. Petkovic´, O. Kìhn, J. Phys. Chem. A 2004, 108, 6083 – 6086.
Received: June 1, 2015 Published online on July 14, 2015
2523
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
