The barriers to internal rotation of the free methanol molecule in the B3LYP/cc-pVTZ, B3LYP/acc-pVTZ, wB97X-D/cc-pVTZ, wB97X-D/acc-pVTZ approximations. Then we calculated the energy of torsional states.
As seen from calculations the results obtained in the wB97X-D/acc-pVTZ approximation are close to those in B3LYP/acc-pVTZ. The same can be said of the wB97X-D/cc-pVTZ and B3LYP/acc-pVTZ approxi-mations. The results of the calculations are more sensitive to basis and cc-pVTZ significantly more preferable than acc-pVTZ. Comparing the calculations in B3LYP/cc-pVTZ and wB97X-D/cc-pVTZ should be noted that in the first case the results are somewhat close to the experimental. Complete information not only about the width but also the shape of the potential curve as a whole, can be obtained expanding the potential energy function in a Fourier series. In accordance with the results of [1], obtained using microwave and rotational spectra data of CH3OH, CD3OH, CH3OD and CD3OD, the potential curve can be represented as follows: $f(t) = V_{0}+V_{3} \cos(3t)+V_{6} \cos(6t)+V_{9} \cos(9t)$, where $V_{0}= V_{3}=186.13$; $V_{6}=-0.035$; $V_{9}=0.0$ cm-1. As noted in [1] "Thus the value of $V_{6} = -0.035\pm0.105$ cm-1 should measure the true deviation from a sinusoidal barrier potential. Our result indicates that the deviation is indeed very small for methanol." Let us now compare the Fourier series of potential functions, calculated in different approximations:
1.B3LYP/cc-pVTZ: $V_{0}= V_{3}=185.698$; $V_{6} =0.471$; $V_{9}=0.0898$.
2. wB97X-D/cc-pVTZ: $V_{0}= V_{3}=183.641$; $V_{6} =3.186$; $V_{9}=1.7081$.
3. B3LYP/acc-pVTZ: $V_{0}= V_{3}=170.774$; $V_{6} =1.569$; $V_{9}=-1.3609$.
4. wB97X-D/acc-pVTZ: $V_{0}= V_{3}=169.727$; $V_{6} =-1.722$; $V_{9}=2.0041$.
Considering as a deviation from experimental potential curver the expression: $$A = \sum_{i}(V_{i}^{exp} -V_{i}^{calc})^{2}$$ one can obtain:
1. B3LYP/cc-pVTZ: $\Delta = 0.6373$
2. wB97X-D/cc-pVTZ: $\Delta = 25.6826$
3. B3LYP/acc-pVTZ: $\Delta = 476.5911$
4. wB97X-D/acc-pVTZ: $\Delta = 544.9792$
Thus, despite the fact that the energy values calculated at any point in the approximations 2-4 lower, than in approximation 1, the relative positions of points on the potential curve is reproduced better by using the approximation 1. In addition, the deviation from sinusoidal is minimal for potential curve, calculated in approximation 1, too. The choice in favor of B3LYP/cc-pVTZ approximation in particular based on the fact that it has established itself well in our calculations of force fields and vibrational spectra of some organic molecules [2,3].