Anion pairs in room temperature ionic liquids predicted by molecular dynamics simulation, verified by spectroscopic characterization

Abstract

Molecular-level spectroscopic analyses of an aprotic and a protic room-temperature ionic liquid, BMIM OTf and BMIM HSO₄, respectively, have been carried out with the aim of verifying molecular dynamics simulations that predict anion pair formation in these fluid structures. Fourier-transform infrared spectroscopy, Raman spectroscopy and nuclear magnetic resonance spectroscopy of various nuclei support the theoretically-determined average molecular arrangements.

---

Introduction

Room-temperature ionic liquids (RTILs) are salts with a melting point below room temperature. They are composed of bulky organic cations and for the most part of similarly large inorganic or organic anions. Their popularity has increased over the last 15 years owing to their properties, including but not limited to high viscosity,¹ ion conductivity,² thermal^1,3 and electrochemical stability,⁴ and the fact that they are environmentally benign compared to traditional solvents.⁵ Not only are they increasingly studied for energy conversion and storage applications,^6,7 but they also enable the synthesis of phase-pure complex metal oxide materials,⁸ are used as lubricants,⁹ in catalysis,¹⁰ for biological applications,¹¹ or as solvents in biomass conversion.¹²

Despite the expanding number of technical applications for RTILs, the molecular arrangement and interactions between cations and anions and their correlation to macroscopic properties are not well understood. Part of the reason is that traditional means of determining coordination within a salt (or coordination compound) are not suitable for investigating RTILs. In addition to the inherent complexity of this class of materials, their lack of transition metal incorporation and the resulting colorless nature prohibit the use of UV/visible spectroscopy as a tool to identify coordination geometries. Single-crystal X-ray diffraction analysis cannot be performed due to the lack of appropriate crystallites. Nevertheless, many theoretical and experimental studies have been conducted in recent years to better understand the molecular structure and dynamics of RTILs. In summary, it is known from previous studies that (1) no neutral ion pairs are present within liquid RTILs, the ion pairs are dissociated;¹³ (2) the molecular arrangement of ions is governed by the extent of van-der-Waals interactions and hydrogen bonding within the system;^13–15 and (3) the presence of nanoscale structural heterogeneities has been determined for several RTILs using Small-Wide Angle X-ray Scattering (SWAXS).¹⁶ Phase segregation of the non-polar alkyl side chains on the cation into domains within the charged matrix of the liquid leads to these heterogeneities.¹⁶

Classical molecular dynamics (MD) simulations^17,18 as well as ab initio molecular dynamics (AIMD) techniques^19–21 have been applied to study some RTIL systems. However, the inherent limitations of these computational analyses, such as deriving reliable interatomic potentials for classical MD simulations and the large computational demands of AIMD calculations, affect their applicability to many RTILs. For example, computationally-demanding AIMD simulations are often carried out on RTIL systems containing symmetric Cl⁻ and PF₆⁻ anions, which have uniform charge distributions.^19–21 However, the most promising RTILs for energy storage applications contain more complex and asymmetrical anions such as trifluoromethanesulfonate (CF₃SO₃⁻, OTf⁻) or bis trifluoromethanesulfonimide ((CF₃SO₂)₂N⁻, NTf₂⁻), which exhibit non-uniform charge distributions that could lead to unique local structural environments.¹⁶ Isolated cation–anion pairs, as exist in gas-phase RTIL vapors,^13,22 are computationally less demanding to simulate due to the fewer number of atoms in the calculation.^23–25 It is questionable though, if findings from these studies are transferrable to liquid-phase RTILs, because isolated ion pairs do not accurately depict the structural arrangement of RTILs in their liquid state.^13–16,26

We present here the results of classical MD simulations carried out on two commonly used RTILs with asymmetric anions, 1-butyl-3-methylimidazolium trifluoromethanesulfonate (BMIM OTf) and 1-butyl-3-methylimidazolium hydrogensulfate (BMIM HSO₄). Spectroscopic data obtained from BMIM OTf and BMIM HSO₄ at room temperature correlate well with unique motifs of cation–anion arrangements our theoretical simulations predict to be present (Fig. 1). MD calculations reveal that BMIM HSO₄ and BMIM OTf both favour short-range interactions in which two anions are surrounded by a cation cage that on average is composed of 9 BMIM⁺ cations.

Fig. 1 MD simulation snapshots of (a) BMIM OTf and (b) BMIM HSO₄ illustrating the anion pairing in both RTILs. In both cases, 9 BMIM⁺ cations surround an anion pair. Cations are shown with stick model whereas anions are shown with ball and stick model. Carbon atoms are shown in grey, nitrogen atoms in blue, oxygen atoms in red, sulfur atoms in orange, fluorine atoms in light blue, and hydrogen atoms in white.

Combined theoretical and experimental studies have been carried out on a variety of RTILs before.^27,28 This work distinguishes itself in the selection of RTILs that are investigated as well as by the variety of spectroscopic data employed to validate the theoretical findings. With BMIM HSO₄ being a protic RTIL (the anion contains a proton that can readily dissociate from the anion and form hydrogen bonds to the cation) while BMIM OTf is aprotic in nature, this study is aimed at highlighting differences and similarities between the short-range assemblies observed for representative systems of these two classes of ionic liquids. This will elucidate any effect hydrogen bond formation may have on the structures. Since both RTIL systems used in this study contain the same cation, all spectroscopically observable differences can be attributed to the structure of and interactions involving the respective anion.

With nuclear magnetic resonance (NMR) spectroscopy of different nuclei in addition to vibrational spectroscopy data (Fourier-transform Infrared (FTIR) and Raman) this study encompasses a wider range of experimental data to verify short-range structural arrangements proposed based on computational modelling than other studies comparing protic and aprotic RTILs.²⁹

Experimental

For this study, commercially available 1-butyl-3-methyl-imidazolium (BMIM⁺) trifluoromethanesulfonate (TfO⁻) (≥98% purity; Sigma-Aldrich) and 1-butyl-3-methylimidazolium (BMIM⁺) hydrogensulfate (HSO₄⁻) (≥94.5% purity; Sigma-Aldrich) were used without any further purification (see Fig. S1 in the ESI† for nomenclature). The densities reported in this manuscript were determined experimentally (experimental) or by MD simulation (calculated).

Spectroscopic analysis

Fourier Transform Infrared (FTIR) spectra in the range from 1800–600 cm⁻¹ were recorded with a Nicolet iS10 (Thermo Scientific). The spectra shown were recorded at a resolution of 0.5 cm⁻¹, employing a diamond Smart ITR accessory with a diamond window. To obtain interference-free Far-IR spectra, low density polyethylene (LDPE) films were prepared; the LDPE films were folded into pouches into which the respective RTIL could be filled. The total RTIL sample volume was <100 μl. The spectra over a range from 600 to 55 cm⁻¹ were obtained using a Thermo Nicolet Nexus FT-IR Spectrometer equipped with a PE detector and a solid substrate beam splitter. These spectra were recorded at a resolution of ∼1 cm⁻¹. OMNIC spectra software was used for data acquisition and analysis for all IR spectra.

Raman spectra were excited in a backscattering geometry using 100 mW of 785 nm radiation from an SDL-TC30 Tunable CW Laser Diode System (SDL Inc.). This excitation wavelength was chosen to mitigate the strong fluorescence of the RTILs at shorter wavelengths. Scattered light was focused, after filtering through a 785 nm notch filter, onto the slits of a Spex Model 270M single grating spectrometer (Jobin Yvon, Inc.) and detected using a Spec-10:100BR/XP backilluminated, deep-depletion CCD detector (Princeton Instruments). Slit widths were set at 50 μm, which provided a nominal resolution of 0.66 cm⁻¹. Data acquisition and analysis was performed using Winspec software (Princeton Instruments).

Selected spectroscopic data was deconvoluted using Grams/32 AI software. The curvefit application was used to fit the experimental data with a mixed Gaussian and Lorentzian function. The deconvoluted spectra are included in the ESI.†

¹⁵N and ¹⁷O nuclear magnetic resonance (NMR) measurements were done with as purchased ionic liquids (i.e. without any isotope enrichments) using Bruker 750 spectrometer (B₀ = 17.6 T; ¹⁵N and ¹⁷O Larmor Frequencies are 76.01 and 101.7 MHz, respectively). The ¹⁵N and ¹⁷O chemical shifts were externally referenced to 1 M aqueous urea solution (d_iso = 76 ppm) and natural abundance D₂O solution (d_iso = 0 ppm), respectively. To convert the ¹⁵N chemical shifts to the IUPAC suggested CH₃NO₂ standard, one should subtract 381.0 ppm.

Computational methods

All the MD simulations were carried out with the computer code DL_POLY.³⁰ In these simulations, atoms are presented as point-charge particles that interact via long-range Coulombic forces and short-range interactions. The latter are described by parametrized functions and represent the repulsion between electron-charge clouds, van der Waals attraction forces, and many-body terms (e.g. bond bending and dihedral angles). The short-range potential parameters and ionic charges used in this study are those of Canongia Lopes and co-workers.^31–34 The C–H bonds, which are constrained in the force field of Canongia Lopes and co-workers, were assigned the stretching force constant from the OPLS force field.^35,36 The parameters of Canongia Lopes et al.³⁴ for alkylsulfate (C_nSO₄) were used to model HSO₄, whereby the C_n group was replaced by H. An ionic charge of 0.22e was used for H to obtain a net charge of −1e for HSO₄⁻ and the Lennard-Jones parameters, ε and σ, were set to zero for H. O–H equilibrium distance and force constant were set to 1.0 Å and 4184 kJ mol⁻¹ Å⁻², respectively, and the S–O–H equilibrium angle and force constant were set to 109.5° and 1046 kJ mol⁻¹ rad⁻², respectively.

All the simulations were carried out in the NPT ensemble (constant number of particles, constant pressure, and constant temperature) at 298.15 K and zero-applied pressure. The temperature and pressure were kept constant via the Nóse–Hoover thermostat³⁷ and the Hoover barostat,³⁸ respectively. The electrostatic interactions were calculated by means of the Ewald summation method.³⁹ A 12 Å cutoff was used for the short-range interactions and the real part of the Ewald sum. The Ewald sum parameters were chosen to achieve a relative error on the electrostatic energy of at most 10⁻⁷. The Verlet leapfrog integration algorithm was used to integrate the equations of motion with a time step of 2 fs. The simulation cells consisted of 125 ion pairs. The MD simulations were run for 2 ns after a 200 ps equilibration period. Calculated (experimental) densities were 1.351 (1.31) and 1.245 (1.28) g cm⁻³ for BMIM OTf and BMIM HSO₄, respectively.

Results and discussion

Unlike metal salts, RTILs do not exhibit a strongly ordered short-range structure; however, interactions between the functional groups of the cations and anions within an RTIL can lead to favourable short-range molecular arrangements within the fluent system. MD calculations revealed that BMIM HSO₄ and BMIM OTf, a protic and an aprotic RTIL respectively, favour short-range interactions in which two anions are surrounded by a cation cage (Fig. 1).

Electrostatic interactions drive the formation of a cation cage around each anion pair in the MD simulations. Based on the MD calculations, the imidazolium ring, particularly in proximity to the methyl group on N(3), and the SO₃ moieties are the most highly charged parts of the cations and anions, respectively. The molecules arrange themselves to maximize the interactions between these groups and consequently the anion pairs are formed with the CF₃ or OH moieties of two OTf⁻ or HSO₄⁻ anions, respectively, that are in close proximity to each other. BMIM⁺ cations between the CF₃ or OH groups of two anions would experience decreased interaction of the imidazolium ring with the more negatively charged SO₃ groups of the anions and this configuration would therefore not be as favourable. In addition, in the case of BMIM HSO₄, hydrogen bonds can form between two anions, strengthening the anion pairing.

Fig. 2 shows the anion–anion radial distribution functions (RDF) for the two RTILs, illustrating the anion pairing. For BMIM OTf, the F–F RDF shows a peak centred at approximately 3 Å, whereas the O–F and O–O RDFs only show sizeable peaks at longer distances. This indicates that anion pairing is mainly through the CF₃ moieties of two OTf⁻ molecules in the BMIM OTf RTIL system. For BMIM HSO₄, hydrogen bonding is evident from the peaks centred at approximately 2.8 Å in the O–O_H and O_H–O_H RDFs. Coordination numbers of 5.6 and 6.0 for OTf⁻ and HSO₄⁻, respectively, can be determined by integrating the RDF between the S atom of the anion and the C(1) atom of the cation. As two (for BMIM OTf) or three cations (for BMIM HSO₄) usually interact with both anions of a pair, the cation cage consists of 9 BMIM⁺ molecules. Representative snapshots of both RTIL systems, obtained from the MD simulations, are shown in Fig. 1 (one anion pair) and Fig. S2† (two neighbouring anion pairs) to illustrate the cation cage formation around the anion pairs information on the size of the cation cage can be obtained from RDFs between O atoms of the two anions of a pair, as shown in Fig. S3.† Phase segregation on the nanoscale level has been reported before in RTILs¹⁶ and seems similar to the well-researched molecular-level phase separation observed in Nafion, a polysulfonated polymer with fluorinated side chains.⁴⁰ The observed formation of anion pairs in protic as well as aprotic RTILs in this study could possibly be aided by a similar phenomenon, i.e. by the difference in hydrophilicity for the two parts of the anion (CF₃ vs. SO₃⁻) and the BMIM⁺ cation in the case of BMIM OTf.

Fig. 2 Anion–anion radial distribution functions from MD simulations of (top) BMIM OTf and (bottom) BMIM HSO₄.

The following paragraphs contain a comprehensive interpretation of the experimental results obtained by molecular spectroscopy. Aspects pertaining to the theoretical prediction of anion pair formation, and to ion interactions in general, are highlighted and discussed in this context.

Infrared and Raman spectroscopy

Mid-range FTIR spectra of various RTILs containing OTf⁻ anions (CF₃SO₃⁻) have been extensively characterized.^41–43 The vibrations between 1600 and 1300 cm⁻¹, seen in Fig. 3a, can be assigned to the BMIM⁺ cation and will not be discussed further. Similarly, the broad feature at 849 cm⁻¹ (ref. 44) and a sharp resonance peak of medium intensity at 624 cm⁻¹ represent vibration modes originating from the BMIM⁺ cation. Nearly identical resonance peaks are present in the FTIR spectrum of BMIM HSO₄ (Fig. 4a) at 831 cm⁻¹ and 623 cm⁻¹, respectively, confirming the assignment. Because different anions interact with the BMIM⁺ cation in these two systems, these modes should be present in both compounds but with slight variations due to different anion–cation interactions. The broad vibrations at ∼840 cm⁻¹ are most likely due to C–N deformation modes. The sharp vibration peak at 623/624 cm⁻¹ can be assigned to an out-of-plane C–H deformation vibration.

Fig. 3 ATR-FTIR (a) and Far IR (b) spectra of BMIM OTf. The sample preparation for the two measurements was not the same; although the x-axis is continuous, the break indicates the different measurements. Peak intensities of the two spectra are not comparable.

Fig. 4 ATR-FTIR (a) and Far IR (b) spectra of BMIM HSO₄. The sample preparation for the two measurements was not the same; although the x-axis is continuous, the break indicates the different measurements. Peak intensities of the two spectra are not comparable.

Based on literature reports,^41–43 all other vibrations between 1300 and 600 cm⁻¹ that are observable in Fig. 3a originate from OTf⁻ and can be assigned to specific vibrations involving either the CF₃ or the SO₃ moiety of the anion. The asymmetry of the band at 1031 cm⁻¹ suggests a weaker band might be enveloped towards the lower energy side of the vibration. By deconvoluting the envelop curve it can be established that it consists of a more intense mode at 1031 cm⁻¹ (ν_s(SO₃)) and a second mode centered at 1023 cm⁻¹, most likely a weak absorbance caused by the mostly Raman-active C–C stretching modes of the BMIM⁺ cation (see Fig. S4 in the ESI† for deconvolution details). Vibrational frequencies of the anion in BMIM OTf and their respective assigned modes are listed in Table 1.

Table 1 Anion-associated infrared frequencies of BMIM OTf in the range of 55 to 1300 cm⁻¹ (ν, δ, ρ = stretching, deformation and rotational modes, ‘as’ = asymmetric, ‘s’ = symmetric, respectively; s = strong, m = medium, w = weak, v = very, br = broad, sh = shoulder)

Observed wavenumber/intensity	Assignment
1276 sh	νas(SO3)
1263 vs	νas(SO3)
1226 s	νs(SO3)
1167 vs	νas(CF3)
1031 vs	νs(SO3)
843 w, br	BMIM^+; δ(CN)
756 m	δs(CF3)
639 vs	δs(SO3)
624 m	BMIM^+; δ(CH) out-of-plane
574 s	δas(CF3)
518 vs	δas(SO3)
348 w	ρ(SO3)
314 w, br	Lattice mode
210 m	ρ(CF3)
80 m, br	Anion–cation interaction

---

Additional vibration modes are visible in the Far-IR spectrum of BMIM OTf (600 to 55 cm⁻¹; Fig. 3b). Most of them can be attributed to vibrations originating from the OTf⁻ anion as well (Table 1), except for the band below 150 cm⁻¹. This mode has previously been shown to be indicative of hydrogen bonding between anions and cations in RTILs.^45–47 Using the same criteria as Wulf et al., the broad absorption below 150 cm⁻¹ in our RTILs is most likely an envelope curve of different bending and stretching modes of hydrogen bonds between the hydrogen atoms on the cation's imidazolium ring and the anion (⁺C–H⋯A⁻).⁴⁵ Librational modes resulting from oscillatory motions within the ions also appear as a broad vibration at about 80 cm⁻¹ in numerous RTILs.⁴⁸ However, in the spectrum of BMIM OTf, the absence of any other vibrational feature in this region indicates that the absorption maximum at 80 cm⁻¹, which is visible in the measured Far-IR spectrum in Fig. 3b, also envelops modes due to intermolecular hydrogen bonding between the RTIL's anion and cation. FTIR spectra of 1-ethyl-3-methyl-imidazolium trifluoromethanesulfonate (C₂MIM OTf) showed the same vibrational mode for intermolecular hydrogen bonding at 89.5 cm⁻¹.⁴⁶ Given the structural differences of the C₂MIM⁺ and BMIM⁺ cations that result in slightly different reduced masses and interaction forces for the two RTILs, we consider our finding for the intermolecular hydrogen bond vibration in good agreement with literature data.⁴⁶ As mentioned above, the SO₃⁻ groups of the OTf⁻ anions are oriented to face BMIM⁺ cations and our MD simulations indicate that the oxygen atoms of the anion's SO₃⁻ group interact strongly with the hydrogen atoms on the imidazolium ring (see RDF in the ESI, Fig. S5†).

Mid-range FTIR and Far-IR spectra of BMIM HSO₄ are shown in Fig. 4. Interestingly, even though BMIM HSO₄ is also commercially available this compound has not been studied as intensely as its OTf⁻-based counterpart. The most thorough analysis of FTIR and Raman data for a HSO₄⁻-containing RTIL was carried out by Kiefer and Pye, who studied 1-hexyl-3-methylimidazolium hydrogensulfate (HMIM HSO₄) in depth.²⁷ While the overall appearance of the FTIR data they report is very similar to our findings, the individual vibration modes do not match well between the two data sets. We can only speculate that the slight differences might be due to the fact that the cation is different in the two RTIL systems, which might lead to a different short-range molecular order in the RTIL. Alternatively, the differences can possibly be traced back to different sample synthesis or preparation techniques. The synthesis paper Kiefer and Pye reference states that HMIM HSO₄ is stable after elimination of water, but should preferably be handled under an inert gas atmosphere.⁴⁹ Our BMIM HSO₄ is a commercially available material and was exposed to air during the measurements. Furthermore, our vibrational spectroscopy data was recorded at higher resolution than the spectra presented by Kiefer and Pye.²⁷ A lower resolution can significantly alter the line shapes of the vibration modes and possibly obscure additional features. This is an especially important factor for identifying HSO₄⁻-related vibration modes. The symmetry distortion in HSO₄⁻, compared to SO₄²⁻, leads to further degeneration of the already multiply degenerated SO₄²⁻ vibration modes.^50,51 Therefore, we will identify wavenumber ranges for the different vibration modes instead of assigning the individually observable peaks. A more detailed analysis and subsequent assignment of the different modes would require isotope labelling experiments⁵² or polarized IR analysis,⁵¹ both of which are beyond the scope of this work.

Based on comparison to FTIR analysis of other hydrogensulfate salts, vibration modes in the region between 1300 and 1100 cm⁻¹ are assigned to asymmetric stretching vibrations, ν_s(SO₄); between 1100 and 900 cm⁻¹ symmetric stretching vibrations (ν_as(SO₄)) are visible (Fig. 4a).^51,53 Symmetric vibrations observed in FTIR spectra usually show lower or at most similar intensity than asymmetric vibrations, therefore we suggest that not all of the modes in this region are due to HSO₄, but the very strong band at 1006 cm⁻¹ might be attributed to a C–N stretching vibration, ν_as(CN).⁵¹ The assignment is tentative, because a vibration of similar intensity is not visible in the FTIR spectrum of BMIM OTf (Fig. 3a). The band observed at 732 cm⁻¹ can be assigned to a bisulfate S–OH stretching mode, ν(S–OH), and is therefore direct evidence that bisulfate ions instead of sulfate ions are present in the system.⁵⁴ The vibration modes at 653 and 610 cm⁻¹ are assigned to ν(SO₄) core stretching vibrations based on literature reports.^51,53 As discussed above, the BMIM⁺-related δ(CH) out-of-plane mode is visible at 623 cm⁻¹.

The intense modes visible at 579 and 553 cm⁻¹ in the Far-IR spectrum of BMIM HSO₄ (Fig. 4b) are most likely part of the same multiply degenerated ν(SO₄) core stretching vibration as the modes at 653 and 610 cm⁻¹ (Fig. 4a). Similarly, the vibrations visible around the intense mode at 431 cm⁻¹ (500 to 400 cm⁻¹ range) can be attributed to a different degenerated ν(SO₄) core stretching vibration.^51,53 The remaining vibration modes below 350 cm⁻¹ have been termed ‘lattice modes’ in literature reports about Far-IR investigations of MHSO₄ salts (M = organic or inorganic cation; not an ionic liquid).⁵¹ As discussed for the Far-IR data of BMIM OTf, these vibrations comprise O=S=O wagging modes (above 200 cm⁻¹),⁴⁸ oscillatory motions, tentatively assigned to the mode at 77 cm⁻¹,⁴⁸ and various hydrogen bond related modes.⁴⁵ A study by Fumino et al. in 2013 unambiguously established that stronger anion–cation interactions lead to a shift of the corresponding mode towards higher wavenumbers.⁴⁸ For BMIM OTf, we observed the vibrational mode for anion–cation interactions at 80 cm⁻¹, it therefore can be expected that this mode will be visible at higher wavenumbers in the spectrum of BMIM HSO₄, a compound with stronger hydrogen bonding possibility between anions and cations. The most intense vibration below 150 cm⁻¹ is centered at 103 cm⁻¹. Assigning the anion–cation interaction (⁺C–H⋯A⁻) to the vibration mode at 103 cm⁻¹ is in good agreement with observed values for similar compounds. For C₂MIM EtSO₄ and C₂MIM BuSO₄ the location of an intermolecular frequency has been reported at 106.4 and 105.1 cm⁻¹, respectively.⁴⁶

Raman data for BMIM OTf (Fig. 5a) was recorded over a range from 530 to 1250 cm⁻¹. As expected, only two strong vibration resonances are visible in this range: at 757 cm⁻¹ the combination of a symmetric CF₃ stretching mode and symmetric bending of the CF₃ group is visible;⁵⁵ the vibration at 1032 cm⁻¹ can be assigned to the symmetric SO₃ stretching mode, ν_s(SO₃), of the OTf⁻ anion.^43,56 Although it should be pointed out that the vibration mode assigned to ν_s(SO₃) is not symmetric, it is an envelop curve composed of modes at 1023 and 1032 cm⁻¹ (see Fig. S6 in the ESI† for deconvolution details). The vibration at lower energy can be attributed to C–C stretching modes of the BMIM⁺ cation (ν(CC)).⁵⁷ The location of both bands associated with OTf⁻ is in excellent agreement with previously reported data for other OTf⁻-containing compounds.^43,55,56 Vibration modes at 745, 824, 1007(sh), 1019, 1045, 1057 and 1115 cm⁻¹ are visible in the Raman spectra of BMIM HSO₄ (Fig. 5b). As can be seen from the two panels of Fig. 5, the S/N ratio for the Raman spectrum of BMIM HSO₄ is larger than for the one obtained for BMIM OTf. Therefore, several additional vibration modes associated with BMIM⁺ are visible in this spectrum. The modes visible at 1019 and 1115 cm⁻¹ are assigned to the BMIM⁺-related vibrations.⁵⁷ The band at 745 cm⁻¹ can be assigned to out-of-plane C–N vibrations between the N(1) and N(3) atom and the C(6) and C(10) atoms of the organic side chains, respectively.⁵⁸ The vibration at 824 cm⁻¹ can be attributed to an S–OH stretching mode, ν(S–OH), therefore offering a direct proof of the presence of the bisulfate anion, and the intense vibrations at 1007(sh), 1045 and 1057 cm⁻¹ are S=O stretching modes, (ν(SO)).

Fig. 5 Raman spectra of (a) BMIM OTF and (b) BMIM HSO₄.

In summary, vibrational spectroscopy provides us with a comprehensive picture – including specific information about anion–cation interactions – that does not oppose the hypothesis of anion pair formation proposed based on MD simulations. The additional structure observed for the lattice modes below 350 cm⁻¹ in the Far-IR spectrum of BMIM HSO₄, compared to the featureless envelop curve for BMIM OTf, strongly suggests the presence of hydrogen bonds between at least two HSO₄⁻ anions. However, without carrying out measurements on deuterium labelled compounds and in the absence of reference data this assignment remains speculative. Our experimental data is in good agreement with spectroscopic observations from other RTILs, which leads us to assume that our theoretical means, which match well, might be suitable to model RTIL structures in general.

NMR spectroscopy

For the two RTILs, we are here focusing on the less commonly reported ¹⁵N NMR and ¹⁷O NMR data. As shown in Fig. 6a, two distinct resonance peaks are visible in the ¹⁵N NMR spectra of both compounds. Based on a report by Besnard et al.⁵⁹ the peaks at 182.7 ppm (BMIM OTf, black spectrum) and 182.9 ppm (BMIM HSO₄, red spectrum), respectively, are assigned to the N(1) atom in the imidazolium ring; the resonance at 170.2 ppm (BMIM OTf) and 170.6 ppm (BMIM HSO₄), respectively, originates from the N(3) atom in the imidazolium ring. Both peaks in the spectrum of BMIM HSO₄ are shifted downfield to higher ppm compared to what can be detected in the spectra of BMIM OTf, indicating a lower electron density on both nitrogen atoms in the imidazolium ring of BMIM HSO₄. In other words, the nature of the anion has a more pronounced effect on the chemical environment of the N(3) atom. This is not surprising considering the steric environment around the imidazolium ring and comparing our results to previously reported ¹⁵N NMR data of BMIM CH₃CO₂ strengthens our hypothesis further. Chemical resonances at 183.7 ppm (δ(¹⁵N(1))) and 171.6 ppm (δ¹⁵(N(3))) have been observed for BMIM CH₃CO₂.⁵⁹ Considering all three compounds, the overall chemical shift range is 1.0 ppm for δ(¹⁵N(1)) and 1.4 ppm for δ(¹⁵N(3)). This data point, which is specific to anion–cation interactions, also emerges from our MD simulations (see RDF in the ESI, Fig. S7†).

Fig. 6 ¹⁵N NMR spectra (a) and ¹⁷O NMR spectra (b) of BMIM OTf (black) and BMIM HSO₄ (red).

The distance between the two resonance peaks (Δ(δ(¹⁵N)) = δ(¹⁵N(1)) − δ(¹⁵N(3))) in the two ¹⁵N NMR spectra differs by only 0.2 ppm, Δ(δ(¹⁵N)) is 12.3 ppm for BMIM HSO₄ and 12.5 ppm for BMIM OTf, but the larger distance observed for BMIM OTf reflects the overall higher shielding of this imidazolium ring. Although other researchers reported that the difference in ¹⁵N chemical shifts is affected to a larger degree by different alkyl substituents on the nitrogen atoms,⁶⁰ our research, both the experimental data and the MD simulations, indicates that the anions still play a significant role in creating an asymmetric charge distribution on the imidazolium ring of the BMIM⁺ cation.

¹⁷O NMR directly probes the chemical and electronic environment of the two different anions and the resonance peaks seen in the spectra of BMIM OTf (Fig. 6b, black spectrum) and BMIM HSO₄ (red spectrum) show the expected difference in chemical shift, which is due to different functional groups on the anion. The higher electron density on the oxygen atoms of the HSO₄⁻ anion causes the observed chemical shift towards lower ppm compared to the peak originating from the SO₃⁻ group of the CF₃SO₃⁻ anion. Because of the large quadrupolar moment of ¹⁷O, the NMR spectra naturally exhibit line broadening. However, the observed full width at half maximum is about 570 ± 20 Hz for BMIM OTf and 875 ± 20 Hz ppm for BMIM HSO₄. The additional peak broadening observed for BMIM HSO₄ indicates hindered rotation of the HSO₄⁻ anion.⁶¹ This is in agreement with the anticipated and theoretically confirmed hydrogen bonding between two HSO₄⁻ anions. If the rotational hindrance was due to anion–cation interaction, it should be approximately the same for both RTILs and the peaks in the ¹⁷O NMR spectra in Fig. 6b would display approximately the same line width, just like the difference in chemical shifts in the ¹⁵N NMR spectrum (Fig. 6a) does not change very much for the two different anions. ¹⁵N and ¹⁷O NMR results confirm our MD predictions of hydrogen bonding between HSO₄⁻ anions in BMIM HSO₄ and similar interactions and distances between anions and cations in both RTIL systems.

The influence of different anions on the cations' ¹H NMR signature has been reported before⁶² and studied by ab initio calculation.⁶³ The effect of the OTf⁻ and HSO₄⁻ anions, respectively, on the BMIM⁺ cation is most pronounced for the protons on the imidazolium ring (see ¹H NMR in the ESI, Fig. S8†). This is in good agreement with our MD simulations, which calculated a higher interaction between the respective anions and the protons on the imidazolium ring or directly adjacent to it (see RDF in the ESI, Fig. S5†) as well as with literature reports.^1,64 The differences observed in the ¹³C NMR spectra of the two compounds are negligible (see ¹³C NMR in the ESI, Fig. S9†).

In summary, the NMR data confirm the predicted stronger anion–anion interactions for BMIM HSO₄, as well as the existence and directionality of anion–cation interactions for this RTIL. The thinner observed line width in the ¹⁷O NMR spectrum of BMIM OTf does not rule out the anion pair formation our MD simulation indicates for this system and confirms that interactions between the OTF⁻ anions are weaker than observed for the strongly hydrogen bonded HSO₄⁻ pairs. The evidenced anion–cation interactions are slightly weaker for BMIM OTf than for BMIM HSO₄.

Conclusions

Using molecular spectroscopy as benchmarks, we have shown that MD simulations can predict molecular-level arrangements and orientations in BMIM OTf and BMIM HSO₄ accurately. FTIR, Far-IR, Raman and NMR spectroscopic data confirm the calculated ion arrangements and interactions to the extent possible and certainly do not contradict the modelling results. We therefore conclude that, at the molecular level, both RTILs we investigated, BMIM OTf and BMIM HSO₄, exhibit a high probability of forming anion pairs within the fluent system. On average, these anion pairs are surrounded by 9 cation molecules. The anion to cation ratio may vary depending on the steric requirements of the cations within a given RTIL, but the structural model proposed here, as well as the use of MD calculations to predict the short-range structure of complex RTIL systems, have the potential of significantly impacting the area of RTIL research.

Acknowledgements

This work was supported by the Pacific Northwest National Laboratory (PNNL) under the open call Laboratory Directed Research and Development (LDRD) program. PNNL is a multi-program laboratory operated by Battelle Memorial Institute for the U.S. Department of Energy (DOE) under Contract DE-AC05-76RL01830. The NMR measurements and MD simulations were performed using EMSL (http://www.emsl.pnl.gov), a national scientific user facility sponsored by the DOE's Office of Biological and Environmental Research (BER) and located at PNNL.

Notes and references

1. A. Bagno, C. Butts, C. Chiappe, F. D'Amico, J. C. D. Lord, D. Pieraccini and F. Rastrelli, Org. Biomol. Chem., 2005, 3, 1624.
2. J. Pitawala, J.-K. Kim, P. Jacobsson, V. Koch, F. Croce and A. Matic, Faraday Discuss., 2012, 154, 71.
3. J. L. Anderson, R. Ding, A. Ellern and D. W. Armstrong, J. Am. Chem. Soc., 2005, 127, 593.
4. P. A. Z. Suarez, V. M. Selbach, J. E. L. Dullius, S. Einloft, C. M. S. Piatnicki, D. S. Azambuja, R. F. de Souza and J. Dupont, Electrochim. Acta, 1997, 42, 2533.
5. C. K. Z. Andrade and L. M. Alves, Curr. Org. Chem., 2005, 9, 195.
6. A. Lewandowski and A. Świderska-Mocek, J. Power Sources, 2009, 194, 601.
7. A. Brandt, S. Pohlmann, A. Varzi, A. Balducci and S. Passerini, MRS Bull., 2013, 38, 554.
8. D. C. Green, S. Glatzel, A. M. Collins, A. J. Patil and S. R. Hall, Adv. Mater., 2012, 24, 5767.
9. M. Palacio and B. Bhushan, Tribol. Lett., 2010, 40, 247.
10. T. Welton, Chem. Rev., 1999, 99, 2071.
11. M. Naushad, Z. A. Alothman, A. B. Khan and M. Ali, Int. J. Biol. Macromol., 2012, 51, 555.
12. P. A. Suarez and H. F. Ramalho, Curr. Org. Chem., 2013, 17, 229.
13. J. P. Armstrong, C. Hurst, R. G. Jones, P. Licence, K. R. J. Lovelock, C. J. Satterley and I. J. Villar-Garcia, Phys. Chem. Chem. Phys., 2007, 9, 982.
14. H. Tokuda, S. Tsuzuki, M. A. B. H. Susan, K. Hayamizu and M. Watanabe, J. Phys. Chem. B, 2006, 110, 19593.
15. I. Skarmoutsos, D. Dellis, R. P. Matthews, T. Welton and P. A. Hunt, J. Phys. Chem. B, 2012, 116, 4921.
16. O. Russina, L. Gontrani, B. Fazio, D. Lombardo, A. Triolo and R. Caminiti, Chem. Phys. Lett., 2010, 493, 259.
17. S. M. Urahata and M. C. C. Ribeiro, J. Chem. Phys., 2004, 120, 1855.
18. Z. Liu, T. Chen, A. Bell and B. Smit, J. Phys. Chem. B, 2010, 114, 4572.
19. M. Bühl, A. Chaumont, R. Schurhammer and G. Wipff, J. Phys. Chem. B, 2005, 109, 18591.
20. M. G. Del Pópolo, R. M. Lynden-Bell and J. Kohanoff, J. Phys. Chem. B, 2005, 109, 5895.
21. B. L. Bhargava and S. Balasubramanian, J. Phys. Chem. B, 2007, 111, 4477.
22. K. Dong, L. Zhao, Q. Wang, Y. Song and S. Zhang, Phys. Chem. Chem. Phys., 2013, 15, 6034.
23. P. A. Hunt, B. Kirchner and T. Welton, Chem. – Eur. J., 2006, 12, 6762.
24. P. A. Hunt, J. Phys. Chem. B, 2007, 111, 4844.
25. E. Rezabal and T. Schäfer, J. Phys. Chem. B, 2013, 117, 553.
26. F. Malberg, A. S. Pensado and B. Kirchner, Phys. Chem. Chem. Phys., 2012, 14, 12079.
27. J. Kiefer and C. C. Pye, J. Phys. Chem. A, 2010, 114, 6713.
28. E. Bodo, S. Mangialardo, F. Ramondo, F. Ceccacci and P. Postorino, J. Phys. Chem. B, 2012, 116, 13878.
29. S. Tsuzuki, W. Shinoda, M. S. Miran, H. Kinoshita, T. Yasuda and M. Watanabe, J. Chem. Phys., 2013, 139, 174504.
30. W. Smith, T. R. Forester and I. T. Todorov, The DL_POLY Classic User Manual, Daresbury Laboratory, United Kingdom. http://www.ccp5.ac.uk/DL_POLY_CLASSIC/.
31. N. Canongia Lopes, J. Deschamps and A. A. H. Pádua, J. Phys. Chem. B, 2004, 108, 2038.
32. J. N. Canongia Lopes, J. Deschamps and A. A. H. Pádua, J. Phys. Chem. B, 2004, 108, 11250.
33. J. N. Canongia Lopes and A. A. H. Pádua, J. Phys. Chem. B, 2004, 108, 16893.
34. J. N. Canongia Lopes, A. A. H. Pádua and K. Shimizu, J. Phys. Chem. B, 2008, 112, 5039.
35. W. L. Jorgensen, D. S. Maxwell and J. Tirado-Rives, J. Am. Chem. Soc., 1996, 118, 11225.
36. G. Kaminski and W. L. Jorgensen, J. Phys. Chem., 1996, 100, 18010.
37. W. G. Hoover, Phys. Rev. A, 1985, 31, 1695.
38. S. Melchionna, G. Ciccotti and B. L. Holian, Mol. Phys., 1993, 78, 533.
39. P. P. Ewald, Ann. Phys., 1921, 64, 253.
40. E. J. Roche, M. Pineri and R. Duplessix, J. Polym. Sci., 1982, 20, 107.
41. W. Huang, R. Wheeler and R. Frech, Spectrochim. Acta, Part A, 1994, 50, 985.
42. T. Iwahashi, T. Miyamae, K. Kanai, K. Seki, D. Kim and Y. Ouchi, J. Phys. Chem. B, 2008, 112, 11936.
43. D. H. Johnston and D. F. Shriver, Inorg. Chem., 1993, 32, 1045.
44. K. Noack, P. S. Schulz, N. Paape, J. Kiefer, P. Wasserscheid and A. Leipertz, Phys. Chem. Chem. Phys., 2010, 12, 14153.
45. A. Wulf, K. Fumino and R. Ludwig, Angew. Chem., Int. Ed., 2010, 49, 449.
46. K. Fumino, A. Wulf, S. P. Verevkin, A. Heintz and R. Ludwig, ChemPhysChem, 2010, 11, 1623.
47. K. Fumino, K. Wittler and R. Ludwig, J. Phys. Chem. B, 2012, 116, 9507.
48. K. Fumino, V. Fossog, K. Wittler, R. Hempelmann and R. Ludwig, Angew. Chem., Int. Ed., 2013, 52, 2368.
49. J. Fraga-Dubreuil, K. Bourahla, M. Rahmouni, J. P. Bazureau and J. Hamelin, Catal. Commun., 2002, 3, 185.
50. G. Sivanesan, P. Kolandaivel and S. S. Pandian, Mater. Chem. Phys., 1993, 34, 73.
51. M. Drozd and J. Baran, Spectrochim. Acta, Part A, 2006, 64, 867.
52. S. Rivera-Rubero and S. Baldelli, J. Phys. Chem. B, 2006, 110, 4756.
53. T. Mhiri and P. Colomban, Solid State Ionics, 1991, 44, 235.
54. T. I. Yacovitch, T. Wende, L. Jiang, N. Heine, G. Meijer, D. M. Neumark and K. R. Asmis, J. Phys. Chem. Lett., 2011, 2, 2135.
55. M. E. Tuttolomondo, A. Navarro, E. L. Varetti and A. B. Altabef, J. Raman Spectrosc., 2005, 36, 427.
56. F. Alloin, G. Hirankumar and T. Pagnier, J. Phys. Chem. B, 2009, 113, 16465.
57. M. C. C. Ribeiro, J. Phys. Chem. B, 2012, 116, 7281.
58. K. Malek, A. Puc, G. Schroeder, V. I. Rybachenko and L. M. Proniewicz, Chem. Phys., 2006, 327, 439.
59. M. Besnard, M. I. Cabaço, F. V. Chávez, N. Pinaud, P. J. Sebastião, J. A. P. Coutinho, J. Mascetti and Y. Danten, J. Phys. Chem. A, 2012, 116, 4890.
60. A. Lyčka, R. Doleček, P. Šimůnek and V. Macháček, Magn. Reson. Chem., 2006, 44, 521.
61. J. Tritt-Goc and D. Fiat, Magn. Reson. Chem., 1991, 29, 156.
62. V. Klimavicius, Z. Gdaniec, J. Kausteklis, V. Aleksa, K. Aidas and V. Balevicius, J. Phys. Chem. B, 2013, 117, 10211.
63. S. Chen, R. Vijayaraghavan, D. R. MacFarlane and E. I. Izgorodina, J. Phys. Chem. B, 2013, 117, 3186.
64. A. Bagno, F. D'Amico and G. Saielli, J. Phys. Chem. B, 2006, 110, 23004.

---

Footnote

† Electronic supplementary information (ESI) available: Radial distribution functions obtained from the MD simulations; atom numbering for BMIM⁺, deconvolution of selected vibrational modes, ¹H NMR and ¹³C NMR data. See DOI: 10.1039/c3ra46069j

---