Raman spectroscopy of cryosolutions: the van der Waals complex of dimethyl ether with fluoroform

Raman spectra in the 3200–350 cm–1 range were obtained for solutions in liquid krypton, at 131 K, containing dimethyl ether, fluoroform and mixtures of them.

In the spectra of the mixed solutions bands were detected that must be assigned to the 1∶1 complex between the ether and fluoroform.

The fluoroform C–H stretching in the complex is detected at 18.7 cm–1 above its value in monomer fluoroform, confirming the blue shifting hydrogen bonding in the complex.

Analysis shows that the Raman intensity of this mode is not significantly affected by complexation.

The advantage of the complementarity of the IR and Raman spectra of cryosolutions is demonstrated by the detailed analysis of the methyl deformation modes of the ether.


Molecular cryospectroscopy using infrared (IR) spectroscopy to detect solutes in low temperature rare gas solutions has been proven to be a viable technique in the study of weak molecular complexes.1,2

In contrast, Raman spectra of cryosolutions have been studied far less, and the Raman literature on the formation of weak complexes in cryosolution is virtually non-existent.

The reasons for this are the weakness of the Raman effect and the low solubility of most compounds in typical cryosolvents.

While the second drawback can be overcome in IR by using cells with longer pathlengths, the low concentrations call for the use of high sensitivity detection in Raman.

State-of-the-art detectivity is offered by the use of a CCD camera, and we have used a detector of this type in this study.

A complex which has recently drawn attention is the one formed between fluoroform and dimethyl ether (DME), F3CH·O(CH3)2.

The complexation enthalpy of this complex was determined to be 12.6(3) kJ mol–1 in liquid argon,3 which puts it in the realm of weak complexes.

The complex is held together by a hydrogen bond formed between the fluoroform hydrogen and the DME oxygen atoms.

This interaction has been characterized by both calculation and experiment3 to belong to the class of blue shifting hydrogen bonds.

The blue shift of the fluoroform C–H stretch, by 17.7 cm–1 for F3CH·O(CH3)2 in liquid argon, is accompanied by a decrease of the IR intensity of the fundamental stretching transition of the hydrogen bonded C–H bond.

This decrease, by an order of magnitude, makes the transition hard to detect in the IR spectra.3

In this study we have investigated the possibility of (i) recording Raman spectra of relatively dilute solutions, in a typical cryosolvent, of fluoroform and dimethyl ether, and (ii) observing the formation of a complex between them.

In view of its unusual behaviour in IR, we have studied in some detail the Raman intensity of the fluoroform C–H stretch upon formation of the complex.

Ab initio predictions of spectral intensities for complexes held together by blue shifting hydrogen bonds hitherto have exclusively concentrated on the IR spectra.

In view of our interest in the Raman characteristics of blue shifting hydrogen bonds we filled this gap by carrying out MP2 calculations of the Raman intensities of monomers and complex.


Dimethyl ether and fluoroform, with stated purities of 99+% and 98+%, respectively, were obtained from Sigma Aldrich.

The solvent gas, krypton, with stated purity of 99.998% was provided by L’Air Liquide.

All compounds were used without further purification.

Raman spectra were obtained by using a Dilor XY spectrometer, consisting of a subtractive double f = 80 cm monochromator coupled to a single f = 80 cm spectrograph.

The scattered Raman signal was imaged onto a liquid nitrogen cooled charge-coupled device (P88231 EEV CCD) with a resolution of 770 × 1152.

The 514.5 nm line of a Spectra-Physics argon ion laser was used for Raman excitation.

The power of the incident laser beam was set at 1 W. The polarization of the laser beam was controlled with a polarization rotator and measurements were made with the incident beam either perpendicular (“polarized spectrum” or parallel (“depolarized spectrum”) to the direction of measurement.

The light scattered at 90° was focused on the entrance slit, where a polarization scrambler was inserted to eliminate the intensity distortion by apparatus polarization.

The spectra were measured with resolutions of 1.5 and 0.8 cm–1.

The Raman cell consisted of a 70 cm long Pyrex tube with an outer diameter of 10 mm.

The tube is sealed at the lower end to contain the solution.

At the other end the tube is fused to a 1 L glass flask, which can be accessed via a Young PR10RA stopcock.

The tube is fitted on the outside with a PT 100 thermoresistor for temperature measurements.

The cell is dimensioned to be inserted into a Leybold exchange gas cryostat, using standard vacuum fittings.

In a typical run the evacuated cell is filled at room temperature with the desired gas mixture to a pressure of 2 bar.

Cooling of the cell to a temperature below the normal boiling point of krypton then causes condensation of an amount of liquid that fills the cell to approximately 2 cm above the bottom of the tube, allowing convenient laser excitation of the sample.

Raman measurements in this study were made in the 3200–350 cm–1 range, with the solutions kept in a 1 K temperature interval around 131 K. Full thermodynamic characterization of the complex equilibium requires spectra of the solutions to be recorded in a sufficiently wide temperature interval.

Due to the very limited liquid range of krypton at 1 bar, such a temperature study imposes the use of pressures inside the cell of up to 15 bars.

The present cell has not been designed to safely hold such pressures.

Consequently, no variable temperature studies can be performed with it, so that the Raman spectra could not be used to determine the complexation enthalpy.

The previous cryosolution study of the complex3 in liquid argon focused on the fluoroform C–H stretch.

For comparison, we have repeated that study using liquid krypton as solvent, and using concentrations similar to those for the Raman spectra.

The latter are significantly higher than those used previously,3 so that an IR cell with a pathlength of 1 mm had to be used.

The IR experiments were performed as described before.3

Theoretical information on the geometry of the complex and on its vibrational spectra was obtained by carrying out ab initio calculations, at the MP2 = FULL/6-31+G(d,p) level.

During the geometry optimization and the force field calculation, corrections for basis set superposition error were taken into account explicitly using the CP-corrected gradient techniques developed by Simon et al.4

All calculations were performed using Gaussian.985

The Raman intensities were calculated by numerical differentiation of the dipole derivatives with respect to the electric field, using the FREQ = RAMAN and DENSITY = CURRENT keywords.

Results and discussion

Using ab initio calculations at the MP2/6-311++G(d,p) level, the structures of the monomers and of the 1∶1 complex have been optimized.

The resulting geometries are very similar to those obtained at the MP2/6-31++G(d,p) level3 and require no further comment.

The more important vibrational frequencies and corresponding IR and Raman intensities for these molecules have been collected in Table 1.

The intermolecular van der Waals modes of the complex are not listed, because they are not directly of interest to the present study.

In the discussion of the spectra below, the fundamentals of the monomers are described by their number in the Herzberg system, as in Table 1, with the monomer indicated as as superscript.

The symmetry species of the normal coordinate is indicated, when deemed necessary, in parentheses.

The symmetry species for monomer DME have been derived with the heavy atom skeleton in the (y,z)-plane.

The point groups for DME and fluoroform are C2v and C3v, respectively, but the complex has the reduced symmetry Cs.

To facilitate comparison, however, fundamentals of the complex are described by the symbol, including the symmetry species, of the corresponding monomer transition.

The actual species for the complex can easily be derived using the correlations a1, b1 → a′ and a2, b2 → a″ for the DME moiety and a1 → a′ and e → a′ + a″ for the fluoroform moiety.

The occurence of bands in the spectra of mixed solutions that are not present in either of the single monomer solutions allows the conclusion that a complex is formed between DME and fluoroform.

The Raman frequencies of these observed complex bands are compared with the corresponding monomer frequencies in Table 2, together with the Raman, IR and ab initio complexation shifts.

Figs. 1 and 2 illustrate the new bands observed in the regions of (a1) and (a1), respectively.

In both figures the top panel gives the Raman spectra while the lower panel gives the IR spectra.

In order to allow direct comparisons of complex/monomer relative intensities, the spectra shown were recorded from solutions in which the concentration of the solute(s) was the same in both types of spectroscopy.

The Raman spectra in Fig. 1 show the presence of a complex band in the spectrum of the mixed monomer solution at 3051.8 cm–1, on the high frequency side of the monomer (a1) band.

With the highest DME mode observed at 2990.5 cm–1, the assignment of that complex band to a DME CH3 stretching mode requires a complexation shift of more than 60 cm–1.

In view of the weakness of the complex and of the weak perturbation of the DME methyl groups, such a shift is unrealistically high.

Therefore, we assign the 3051.8 cm–1 band to (a1) in the complex.

Its frequency is 18.7 cm–1 higher than the corresponding monomer band at 3033.1 cm–1, in very good agreement with the value of 17.7 cm–1 determined at 92 K using IR spectroscopy in liquid argon.3

Thus, the 3051.8 cm–1 Raman band clearly confirms the blue shifting hydrogen bond between DME and fluoroform.

The experimental conditions used in the present experiments are such that for the mixed solution the complex band is hardly visible in the IR spectrum, which exemplifies the above mentioned problem caused by the decrease of the IR intensity of the C–H stretch upon complexation.

From the relative intensities in the top panel then follows that the Raman intensity of this mode is affected much less.

A more quantitative appreciation of this will be discussed in a following paragraph.

Other complex bands observed in the CH stretching region are compiled in Table 2.

They are due to DME, but have not been assigned in more detail.

This is because of the problems with the assignments of the DME fundamentals in this region.6

Table 2 indicates that the highest and the lowest of the bands in this region appear at 2990.5 and 2811.7 cm–1.in the monomer, and at 2992.6 and 2811.7 cm–1 in the complex.

In the depolarized Raman spectrum these bands are considerably weaker than in the polarized spectrum, which shows that they must be assigned as νDME1(a1) and νDME2(a1), respectively.

The six complex bands found between νDME1(a1) and νDME2(a1) are also polarized in Raman, and, therefore, must be due to overtones or combination bands that derive their intensity from Fermi resonance with the a1 fundamentals.

These resonances do not permit meaningful comparison between the complexation shifts of the outer two complex bands and the (harmonic) ab initio results.

For the complex band in the IR spectrum, shown in the lower panel of Fig. 2, is much sharper than the corresponding monomer mode: the measured full width at half height equals 1.0 cm–1 for the complex band, against 5.0 cm–1 for the monomer band.

In the Raman spectrum, top panel of Fig. 2, both monomer and complex bands are very narrow.

Because polarized Raman band widths of totally symmetric bands are predominantly determined by vibrational relaxation, this shows that the monomer IR contour is significantly broadened by rotational relaxation.

The ab initio results predict that for the Raman intensity in the monomer differs little from that in the complex, the ratio being 1.051.

This ratio is in line with expectations for a mode of a weak complex that is localized in atoms not directly involved in the hydrogen bond.

If the ab initio ratio is accepted to be correct, it can be used to transform the experimental band area ratio for this mode into the monomer/complex concentration ratio.

The monomer/complex Raman band area ratio for the 698/694.7 cm–1 doublet assigned to was determined from least squares band fitting to be 1.03(5).

Combining this value with the ab initio Raman intensity ratio leads to a monomer/complex concentration ratio for the studied solution equal to 1.08(5).

Also using least squares band fitting, the monomer/complex Raman band area ratio for in the same solution is calculated to be 0.99(7).

With the above concentration ratio this can be transformed into the Raman intensity ratio for this doublet, resulting in a value of 0.92(11), which is in reasonable agreement with the ratio derived from the ab initio Raman intensities in Table 1, equalling 0.70.

Hence, always in the assumption that the ab initio intensity ratio for the doublet is valid, the above result confirms what was anticipated in the previous paragraph: in contrast with the IR intensity, the Raman intensity of the C–H bond stretching of fluoroform engaged in a blue shifting hydrogen bond with DME, does not significantly differ from that of the monomer.

Fig. 1 then makes clear that blue shifting hydrogen bonding may escape detection in IR as a consequence of the intensity behavior, but that this is much less likely to happen when Raman spectroscopy is used.

The data in Table 2 show that complex bands have been observed for a number of other fundamentals.

It can be seen that in all cases the observed directions match the ab initio ones, and that also the quantitative agreement in general is good.

Over the years, dimethyl ether has been used as a model compound in numerous studies, many of which focus on the Lewis base properties of the compound.

It would seem that this popularity has led to a reliable assignment of all the normal modes of DME.

Inspection of the literature, however, shows that this is not the case.

For instance, in recent years, at least four different studies have reported on the IR spectra of matrix isolated monomeric DME.7–10

In two of these studies9,10 a detailed assignment of the bands observed in the CH3 deformation region, between 1500 and 1400 cm–1 was given.

Of the six fundamentals expected in this region, spanning the representation 2a1 + a2 + b1 + 2b2, five have been assigned by Asselin et al.,10 and four by Wrobel et al.,9 which differ from each other because the assignments of the b1 and one of the b2 fundamentals have been reversed.

Their observed frequencies agree well with those of a somewhat older matrix study,11 who identified and assigned five of the fundamentals.

In none of these studies, the IR forbidden νDME9(a2) mode was assigned.

This mode was identified in the spectrum of solid DME isotopically diluted in DME-d6, by Allan et al.,12 at 1462.4 cm–1.

Although the positions of the bands in this environment differ somewhat from those in an argon matrix, detailed comparison suggests that this 1462.4 cm–1 transition must correspond to the 1456 cm–1 band observed in the argon matrices, which was assigned to a b1 mode,10 or to a b2 mode9 in that environment.

Raman spectra of solid DME isotopically diluted in DME-d6 have also been reported.13

Although νDME9(a2) is Raman active, it was not assigned, and the assignments proposed for the remaining five modes in this region disagree with those of the corresponding IR study.12

Hence, it is clear that the assignments of the methyl deformations have not yet converged, and this is another instance where the present study can contribute.

In Fig. 3 the methyl deformation region is shown in Raman, top panel, and in IR, lower panel, for a solution in liquid krypton.

Obviously, the higher and lower bands, at 1474.8 and 1426.3 cm–1, respectively, are present in Raman as well as in IR.

In the Raman spectrum two more bands are observed, with maxima at approximately 1452.5 and at 1445.1 cm–1.

Also in the IR spectrum two more bands are observed, at 1457.5 and 1454.8 cm–1.

The non-coincidence of the central Raman and IR doublets shows they are due to different fundamental modes of DME.

Hence, in the combined spectra a total of six different transitions is observed, equal to the number expected in this region.

The assignment of the highest and lowest bands in this region to ν3(a1) and ν19(b2), respectively, is straightforward and requires no comment.

Apart from the IR forbidden ν9(a2) mode, the ν4(a1) is predicted, Table 1, to have significantly lower IR intensity than the other methyl deformation modes.

We, therefore, assign the central ir doublet, in the predicted frequency order, to the transitions with the higher ab initio ir intensity, i.e. ν18(b2) and ν13(b1).

The ab initio Raman intensity of ν18(b2) suggests that this mode may contribute some intensity to the high frequency shoulder of the central Raman doublet, even when no separate maximum is observed for this mode.

The components of that Raman doublet must be due to ν4(a1) and ν9(a2).

In contradiction with the predicted frequency order in Table 1, we prefer the assignment of ν4(a1) to the intense low frequency component at 1445.1 cm–1, as this leads to a more satisfactory agreement with the predicted Raman intensities of these modes.


In the paragraphs above we have demonstrated the feasiblity of Raman cryospectroscopy, with solute concentrations in the 1 × 10–3 mole fraction range, in the field of van der Waals molecules.

The spectra recorded for mixtures of fluoroform and dimethyl ether confirm the blue shifting nature of the hydrogen bond between these monomers.

Finally, exploiting the sharpness of the vibrational bands in cryosolution, the use of combined IR and Raman spectra in removing uncertainties for some assignments has been illustrated.