Structure and vibrational spectroscopy of methanesulfonic acid

In this work, we have used a combination of vibrational spectroscopy (infrared, Raman and inelastic neutron scattering) and periodic density functional theory to investigate the structure of methanesulfonic acid (MSA) in the liquid and solid states. The spectra clearly show that the hydrogen bonding is much stronger in the solid than the liquid state. The structure of MSA is not known; however, mineral acids typically adopt a chain structure in condensed phases. A periodic density functional theory (CASTEP) calculation based on the linear chain structure found in the closely related molecule trifluoromethanesulfonic acid gave good agreement between the observed and calculated spectra, particularly with regard to the methyl and sulfonate groups. The model accounts for the large widths of the asymmetric S-O stretch modes; however, the external mode region is not well described. Together, these observations suggest that the basic model of four molecules in the primitive unit cell, linked by hydrogen bonding into chains, is correct, but that MSA crystallizes in a different space group than that of trifluoromethanesulfonic acid.


Introduction
Methanesulfonic acid (MSA, see inset in figure 1 for the structure) is a molecule of interest particularly in atmospheric chemistry due to its formation by oxidation of dimethylsulfide (DMS, mainly produced by marine biota [1]) in the geochemical sulfur cycle. Although DMS accounts for less than 10% of the total nonsea-salt mass of aerosol sulfur, it is the only known source of MSA in marine air and could be a vital tracer for oceanic emissions and atmospheric reaction pathways of organic sulfur compounds [2]. This can then be used to determine the impact of sulfur-containing aerosols on the global climate [3]. In addition, it has found use in linking atmospheric conditions to natural events such as El Niñ o [1] as DMS emissions are significantly modulated by short-term as well as long-term climatic changes in the past, [4], particularly with regards to the ocean climate [5].
& 2018 The Authors. Published by the Royal Society under the terms of the Creative MSA has also been explored due to its environmental benefits (especially when compared to F-or Bcontaining alternatives) as it is readily biodegradable and can be considered to be a natural product, being part of the natural sulfur cycle. Moreover, it has good aqueous solubility, low toxicity and is easy to waste treat. Its high electrolyte conductivity requires less electricity and hence lower fuel consumption for electrochemical processes, in particular, as the commercial electrolyte standard for Sn/Pb solder electroplating (although it also has emerging uses in electroplating pure tin on sheet steel) [6].
As a catalyst, MSA has also been widely used in many pathways, for example, in cyclization reactions such as 3-arylpropanoic and 4-arylbutanoic acids to 1-indanones and 1-tetralones, respectively [7] and ring-opening polymerization, e.g. of trimethylene carbonate [8]. MSA is also a new and efficient catalyst for pathways such as the one-pot synthesis of 2-amino-4H-chromenes, [9] which have medical applications.
The vibrational spectroscopy of MSA has been studied for many years [10][11][12][13][14]. However, all of these relate to surface spectroscopy of aqueous MSA by vibrational sum frequency spectroscopy [10], or are of the gas or liquid states [11 -13]: there are no published solid-state spectra. Here, we report a comprehensive study of the liquid and solid state of MSA, using a combination of infrared, FT-Raman and inelastic neutron scattering (INS) spectroscopies, with the assignments supported by density functional theory calculations.

Materials
MSA was used as received from Sigma-Aldrich (better than 99.0% purity).

Vibrational spectroscopy
The INS spectrum was recorded at less than 20 K using TOSCA [15] at ISIS (http://www.isis.stfc.ac.uk/). The spectrum is available at the INS database: http://wwwisis2.isis.rl.ac.uk/INSdatabase/. Infrared spectra were recorded using a Bruker Vertex70 FTIR spectrometer, over the range of 100-4000 cm 21 at 4 cm 21 resolution with a DLaTGS detector using 64 scans and the Bruker Diamond ATR. The use of the ultra-wide range beamsplitter enabled the entire spectral range to be recorded without the need to change beamsplitters. Variable temperature ATR infrared spectra (150-300 K) were recorded using a Specac Golden Gate accessory. The spectra have been corrected for the wavelength-dependent variation  in path length using the Bruker software and also baseline corrected. FT-Raman spectra were recorded with a Bruker MultiRam spectrometer using 1064 nm excitation, 4 cm 21 resolution, 500 mW laser power and 64 scans. The liquid MSA Raman spectrum was measured in air at room temperature in a quartz cuvette; the solid MSA Raman spectrum was obtained by cooling the filled cuvette in liquid N 2 .

Computational studies
The plane wave pseudopotential-based program CASTEP was used for the calculation of the vibrational transition energies and their intensities [16,17]. The generalized gradient approximation (GGA) Perdew-Burke-Ernzerhof (PBE) functional was used in conjunction with optimized norm-conserving pseudopotentials. Electronic supplementary material, table S1 gives the details of the calculations. All of the calculations were converged to better than j0.0035j eV Å 21 . After geometry optimization, the vibrational spectra were calculated in the harmonic approximation using density-functional perturbation theory [18]. This procedure generates the vibrational eigenvalues and eigenvectors, which allows visualization of the modes within Materials Studio (http://accelrys.com/products/collaborative-science/ biovia-materials-studio/) and is also the information needed to calculate the INS spectrum using the program ACLIMAX [19]. We emphasize that the transition energies have not been scaled. Figure 1 shows a comparison of INS, infrared and Raman spectra obtained from the liquid and solid states. There are obvious differences between the solid and liquid states of MSA. In the Raman spectra, the peaks are sharpened and more readily resolved in the solid state, often changing from a single broader peak in the liquid state spectra to two distinct peaks in the solid state. The observed splitting of peaks can be explained by how the molecules become fixed in the solid, which destroys the symmetry seen in each individual molecule. Hence, the doubly degenerate peaks in the liquid state are no longer present in the solid state. The effects of the hydrogen bonding are best seen in the infrared spectra, figure 2. As the hydrogen bonding strength increases, the O -H stretch peak downshifts and the SZOZH bending motions upshift. This can be seen in figure 2, where the broad O -H stretch peak in the liquid spectrum is downshifted from approximately 3000 cm 21 to 2600 cm 21 in the solid spectra. In addition, the 762 cm 21 (liquid) peak is upshifted to 790 cm 21 (solid) and the 1171 cm 21 shoulder (liquid) is resolved into the 1224 cm 21 peak (solid). Clearly, in the solid state, the hydrogen bond strength present in MSA increases. The presence of hydrogen in MSA has a considerable effect on the spectra which can be seen by contrasting it with that of metal methanesulfonates [20]. Most of the modes are not affected and therefore appear at similar values. However, there is a large noticeable upshift of the SO 3 symmetric and asymmetric stretching modes from approximately 1000 to 1150 cm 21  Previous literature has suggested that MSA exists as a dimer structure with C i point group symmetry as a result of comparisons between calculations of frequency [12]. This conclusion may have been based on the well-known dimer formations present in carboxylic and benzoic acids [21]. However, there are exceptions to this in the form of monocarboxylic acids such as formic acid [22], acetic acid [23,24] and b-tetrolic acid [25], which preferentially form catemer motifs. This is possible as the molecules are small and non-bulky, therefore enabling them to pack within the limits of 5-8 Å that can be accommodated by the hydrogen bonds [26].

Results and discussion
A linear chain structure is commonly found in mineral acids and, in particular, is seen in the closely related molecules fluorosulfuric acid and trifluoromethanesulfonic acid [27]. In the previous literature, nitric acid was also believed to exist as dimers in its solid state [28,29], but more recent findings are that it too follows a catemer motif [30], hence it is not unreasonable to question the dimer MSA structure given the advances in the understanding of these solid-state structures. Moreover, inconsistencies between the predicted spectra for a MSA dimer and the observations have even been noted in previous work [12].
The crystal structure of MSA is not known and the reasons for this are not clear. MSA is a liquid at room temperature, thus the procedure for obtaining crystals suitable for X-ray diffraction is more complex, although there are many examples in the literature of such materials. However, the crystal structure of the closely related trifluoromethanesulfonic acid, which is also a liquid at room temperature and for which the solid-state structure is known [27], was used as the basis for our model. The structure is monoclinic, space group P2 1 /c, with four molecules in the primitive cell. By replacing the F with H in each molecule, we were able to construct an approximate structure for MSA in the linear chain structure. This structure was then geometry and lattice parameter optimized, noting that the space group was maintained. This resulted in a reduction of cell volume from 465.0 to 385.8 Å 3 . Vibrational analysis revealed all real modes present across the entire Brillouin zone, showing that the structure is dynamically stable, figure 3; the final structure is shown as the inset in figure 4. The calculated density of the solid is 1.65 gm cm 23 , the density of the liquid is 1.48 gm cm 23 , the stronger hydrogen bonding in the solid state would probably increase the density above that of the liquid, further towards the calculated value, suggesting that the basic model with four molecules in the primitive cell is correct. peak wavenumbers is due to hydrogen bonding. It is possible that this is a consequence of using CASTEP for the calculations, the reasons of which will be expanded upon later. The most striking feature of the INS spectrum is the massive CH 3 torsion mode, which is well replicated in the calculations. Owing to this extremely high intensity, it is likely that strong overtones would be observed as additional peaks in the experimental spectrum. The fundamental and first overtone are easily seen, and the anharmonicity correction (harmonic frequency ¼ v e , anharmonicity ¼ x e v e ) was calculated using the standard formulae [31] to deduce the locations of higher overtones. The values predicted (with v e ¼ 253 cm 21 , x e v e ¼ 8.5 cm 21 ) were 236, 455 and 657 cm 21 , which are clearly visible as the 236, 455 and 655 cm 21 peaks seen in the experimental spectrum, suggesting that the first, second and third overtones are present. The fourth overtone was predicted to appear at 842 cm 21 which is close to the weak feature lying at 860 cm 21 . However, this is more likely to be a phonon wing from the SZOZH out of plane deformation. The small value of the anharmonicity correction, approximately 3% v e, is consistent with weak intermolecular interactions. In methanesulfonate salts and complexes [20], the methyl group  projects into empty space, as is also the case for trifluoromethanesulfonic acid [27] and this propagates through the calculation to our model structure. That this feature is conserved, suggests that it is also likely to occur in the real (unknown) structure.
The Raman spectrum (figure 5) is consistent between both the observed and experimental data. This is because the features due to hydrogen bonding only appear weakly in the spectrum and the main features result from the CH 3 and SO 3 groups. However, there is a major miscalculation of intensity for the O -H stretch modes: in the calculation, these occur as extremely strong modes at 2622 and 2756 cm 21 , whereas only a weak broad feature centred at 2821 cm 21 is seen in the experimental data. The very weak peaks seen at 2666 and 2815 cm 21 in the experimental spectrum are probably overtones of the symmetric and asymmetric methyl deformations enhanced in strength by Fermi resonance with the intense symmetric methyl stretch mode at 2942 cm 21 .
The infrared spectrum (figure 6) appears to have a few inconsistencies between the calculated and experimental spectra, the most affected area appearing at 500-900 cm 21 . It is important to note that the calculations did not take into account any line width mechanisms (e.g. electrical anharmonicity), which probably explains the difference in peak widths [32], especially for the SO 3 symmetric and asymmetric stretching modes at 1088 and 1326 cm 21 . Additionally, CASTEP does not incorporate hydrogen bonding well in its calculations, as found previously from work on metal methanesulfonates [20]. Therefore with this context in mind, the CH 3 and SO 3 modes match reasonably well. The broad peak due to OH has been calculated to be in the correct position, although this is likely to be fortuitous due to the cancellation of errors in the calculation [33].
We have previously investigated how the mode of coordination influences the vibrational spectra of methanesulfonates [20]. We found that only the asymmetric SZO stretch of the [CH 3 SO 3 ] 2 ion was systematically perturbed. Figure 7 compares the infrared spectra of MSA, Cu(H 2 O) 4 (CH 3 SO 3 ) 2 and Cs(CH 3 SO 3 ). In the copper complex, the methanesulfonate is monodentate [34], and the caesium salt is ionic [35]. A proton can be considered as the extreme of coordination, and it can be seen that, in contrast to the previous examples, both the symmetric stretch and the asymmetric stretch modes are strongly affected: the splitting of the asymmetric stretch mode is more than twice as large as seen previously and the symmetric mode is downshifted by more than 100 cm 21 . Visualization of the mode shows that while it is complex, also involving the C-S stretch and the methyl rock, only the S-OH bond is involved, the two nominally S¼O bonds do not participate. This decoupling is not the result of the increased mass of the OH group relative to an O atom, as this would result in only a 4% or so shift. The large width of the components of the asymmetric mode are qualitatively reproduced by our model. Figure 3 shows the dispersion curves and it can be seen that at the G-point (0,0,0) the factor group splitting is 75 cm 21 for the mode at approximately 1100 cm 21 , and 30 cm 21 for the mode at approximately 1300 cm 21 , the experimental values are 106 and 62 cm 21 .

Conclusion
We report the first spectra of MSA in the solid state. Comparison with the liquid state shows that the hydrogen bonding is stronger in the solid state, as shown by the upshift of the SZOZH deformation modes and the downshift of the O -H stretch mode. The solid-state structure of MSA is unknown; however, a periodic density functional theory (CASTEP) calculation based on the linear chain structure found in the closely related molecule trifluoromethanesulfonic acid gave good agreement between the observed and calculated spectra, particularly with regards to the methyl and sulfonate groups. The model accounts for the large widths of the asymmetric SZO stretch modes; however, the external mode region is not well described. Together, these observations suggest that the basic model of four molecules in the primitive unit cell, linked by hydrogen bonding into chains is correct, but that MSA crystallizes in a different space group than that of trifluoromethanesulfonic acid. The linear chain structure of MSA is probably retained in the liquid state, although with some disorder to account for the weaker hydrogen bonding.