Ab initio computational study of beta-cellobiose conformers using B3LYP/6-311++G**

The molecular structure of 27 conformers of beta-cellobiose were studied in vacuo through gradient geometry optimization using B3LYP density functionals and the 6-311++G** basis set. The conformationally dependent geometry changes and energies were explored as well as the hydrogen-bonding network. The lowest electronic energy structures found were not those suggested from available crystallographic and NMR solution data, where the glycosidic dihedral angles fall in the region (phi, psi) approximately (40 degrees, -20 degrees ). Rather, 'flipped' conformations in which the dihedral angles are in the range (phi, psi) approximately (180 degrees, 0 degrees ) are energetically more stable by approximately 2.5 kcal/mol over the 'experimentally accepted' structure. Further, when the vibrational free energy, deltaG, obtained from the calculated frequencies, is compared throughout the series, structures with (phi, psi) in the experimentally observed range still have higher free energy ( approximately 2.0 kcal/mol) than 'flipped' forms. The range of bridging dihedral angles of the 'normal' conformers, resulting from the variance in the phi dihedral is larger than that found in the 'flipped' forms. Due to this large flat energy surface for the normal conformations, we surmise that the summation of populations of these conformations will favor the 'normal' conformations, although evidence suggests that polar solvent effects may play the dominant role in providing stability for the 'normal' forms. Even though some empirical studies previously found the 'flipped' conformations to be lowest in energy, these studies have been generally discredited because they were in disagreement with experimental results. Most of the DFT/ab initio conformations reported here have not been reported previously in the ab initio literature, in part because the use of less rigorous theoretical methods, i.e. smaller basis sets, have given results in general agreement with experimental data, that is, they energetically favored the 'normal' forms. These are the first DFT/ab initio calculations at this level of theory, apparently because of the length and difficulty of carrying out optimizations at these high levels.