AIMD IR
0 5 10 15 A(˜ν)/ m2 mol−1
Experiment IR
0.00 0.02 0.04
Absorbance
AIMD mode spectra
0 500 1000 1500 2000 2500 3000 3500 4000
0 2 4
ν˜/ cm−1
P(˜ν)/Kcm 1 2 3
4 5 6
7 8 9
10 11 12
0 200 400
A/kmmol−1
Figure 4.13: Spectra of methanol in carbon tetrachloride: IR spectrum and normal coordinate analysis from an AIMD simulation of one methanol molecule in 32 carbon tetrachloride molecules. The exper-imental IR spectrum has been provided by the research group of Prof. Dr. Peter Vöhringer. Due to the strong absorption bands of carbon tetrachloride, it is shown only above 2000 cm−1. See figure 4.1 for graphical representations of the normal coordinate vectors.
andν1+ν4306causes a splitting of the band at 786 cm−1in the experiment. The simulation shows a broader peak instead, which consists of both the fundamental transition and the combination band. The first overtone ofν3and the combinationν1+ν3+ν4, which are found as a weak band around 1540 cm−1in the experiment, are also predicted by the AIMD, though with a significant redshift due to the underestimated fundamental wavenumber ofν3.
No. ν˜/ cm−1 hTi/ K Description
1 ≈270 395 HCOH torsion
2 956 109 CO stretching
3 1059 93 CH3bending + COH bending
4 1135 100 CH3bending
5 1322 116 COH bending + CH3bending
6 1447 101 CH3bending
7 1473 111 CH3bending
8 1474 108 CH3bending
9 3002 117 Symmetric CH3stretching 10 3080 131 Antisymmetric CH3stretching 11 3109 145 Antisymmetric CH3stretching
12 3681 179 OH stretching
Table 4.6: Normal coordinates of methanol in carbon tetrachloride: peak maxima ˜νand average mode temperatureshTi.
methanol clusters should be avoided, a thick solution layer with a low methanol concentration was measured. Thus, the region below 2000 cm−1is totally covered by the carbon tetrachloride bands. The intensity ratios of the OH stretching and the symmetric CH3stretching vibrations are, however, in very good agreement. A blueshift of these modes in the simulation is ob-served in the same way as in the gas phase (see section 4.1.1). As already mentioned before, the broader second peak of the CH stretching modes is largely influenced by Fermi resonances with bending overtones300. Since these are insufficiently included in the AIMD, the simulated spectrum does not show the second peak.
Compared to the gas phase simulation, the HCOH torsional mode gives rise to a broader band in the solution, showing that the torsion is significantly influenced by the surrounding carbon tetrachloride molecules. Furthermore, it possesses a much larger IR intensity, and the main reason for that is the significant excess energy of this mode in comparison with the other vibrations (see table 4.6). However, as the simulation was performed with the thermo-stat temperature set to 400 K, the energy of the other vibrations is actually too low. This is a general artifact of the simulation: although massive thermostatting was used to approximately equilibrate the system, the energy flows from the high-frequency modes of methanol to the low-frequency modes of the carbon tetrachloride. A potential solution would be the applica-tion of two thermostats, one for the carbon tetrachloride molecules and one for the methanol molecule. However, since the temperature of the CH stretching and OH stretching modes is similar, no major influence on the intensity ratios in the spectral region important for the comparison with the experiment is expected in the particular case studied here. Nevertheless, simulations of solute molecules have to be checked carefully for this issue in general. A side effect of the lowered methanol temperature is the slight blueshift of all bands with respect to
AIMD IR
0 10 20 30 40
A(˜ν)/ m2mol−1
Experiment IR
0 500 1000 1500 2000 2500 3000 3500 4000
0.0 0.2 0.4 0.6
ν˜/ cm−1
Absorbance
0 500 1000 1500
A/kmmol−1
Figure 4.14: Spectra of pinacol in carbon tetrachloride: IR spectrum from an AIMD simulation of one pinacol molecule in 32 carbon tetrachloride molecules. The experimental IR spectrum has been provided by the research group of Prof. Dr. Peter Vöhringer. Due to the strong absorption bands of carbon tetrachloride, certain regions of the experimental spectrum are cut out.
the gas phase, since a smaller part of the anharmonic potential energy surface is sampled (see section 4.1.2).
4.3.2 Pinacol in Carbon Tetrachloride
The second example to be investigated is pinacol solved in carbon tetrachloride. Pinacol is studied in the research group of Prof. Dr. Peter Vöhringer as a model system for extended hydrogen bond wires307,308. In the synclinal conformation, the two hydroxyl groups of pinacol form an intramolecular hydrogen bond where one hydroxyl group acts as donor while the other one is the acceptor. The solution in a nonpolar solvent such as carbon tetrachloride largely suppresses the influence of the surrounding molecules, allowing to study the hydrogen bond dynamics of the almost isolated system.
The comparison of simulation and experiment (see figure 4.14) shows a very good agree-ment of most bands. The five experiagree-mental lines between 1000 cm−1and 1500 cm−1are repro-duced very well by the simulation, only the intensity of the peak in the middle of the higher triplet is underestimated and the bands are slightly red-shifted. Also the lower-wavenumber region, which is covered by the carbon tetrachloride here, coincides with experimental data obtained in the gas phase and in a carbon disulfide solution309,310. The OH stretching band around 3600 cm−1clearly shows a splitting of 47 cm−1in the experiment. The narrow peak at 3626 cm−1is connected to the dangling hydroxyl group that acts as the hydrogen bond accep-tor, while the slightly broader peak at 3579 cm−1belongs to the donating hydroxyl group307. Due to the noise in the simulated spectrum, it is hard to exactly quantify the splitting there, but the width of the band indicates that it is in the order of 50 cm−1, which is very similar to
N Me
H OAc
N⊕ Et
N Me
C HOAc
N Et
Figure 4.15: Carbene formation by a proton transfer in [C2C1Im][OAc].
the experiment. A clear deficiency of the simulation is, however, the strong underestimation of the CH stretching vibrations around 3000 cm−1. This is in line with earlier observations (see section 4.1.1) though the effect is much more pronounced here. On the one hand, this might be caused by the underlying electronic structure method, but a static calculation within the harmonic approximation using the same method provides larger IR intensities for these modes. On the other hand, it is likely that anharmonicity effects in terms of Fermi resonances play a significant role similar to the methanol system, and these are insufficiently included in the AIMD, leading to the strongly reduced intensities. Moreover, the integral over the CH stretching bands in the power spectrum is equal to around nine modes instead of the expected twelve modes, so this is another reason for the underestimation of the IR intensity, but it cannot explain the complete effect.
It is important to note that the simulation of pinacol in carbon tetrachloride does not show the same temperature shifting effect as the simulation of methanol in carbon tetrachloride.
The total integrals of the power spectra indicate that both the pinacol molecule and the carbon tetrachloride molecules have temperatures of approximately 400 K. A possible reason might be that pinacol, in contrast to methanol, possesses many modes at low wavenumbers close to carbon tetrachloride. This largely facilitates the exchange of energy between the solute and the solvent, allowing to maintain equipartition more easily.