Submitted to: Canadian Journal of Chemistry
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: January 3, 2006
Publication Date: April 1, 2006
Citation: French, A.D., Johnson, G.P. 2006. Quantum Mechanics Studies of Cellobiose Conformations. Canadian Journal of Chemistry. 84:603-612 Interpretive Summary: Many of the physical characteristics of cotton are unexplained but are almost certain to result from their three dimensional molecular structures of its constituent cellulose molecules. If these structures could be understood, it should be possible to either do a better job of taking advantage of the unique traits, or to carry out some modification of the structure so that the fiber could carry out its functions better, or even take on new functions. Such structures are difficult to study by experiment, but can be studied with computerized molecular modeling. The current paper discusses the results of a modeling study carried out with electronic structure theory, or quantum mechanics. Such calculations take extremely long times. The study, on a very short version of the cellulose molecule, cellobiose, was surprisingly predictive of the experimentally observed shapes of cellobiose and similar molecules, including cellulose itself. This work provides a benchmark for faster, empirical methods. It also suggests that water is not likely to affect the shape of the cellulose molecule much. This has been a continuing question. This work is of interest to scientists working with structure-related problems in cotton and other cellulosic fibers, as well as to computational chemists.
Technical Abstract: Three regions of the Phi,Psi space of cellobiose were analyzed with quantum mechanics. A central region, in which most crystal structures are found, was covered by a 9 x 9 grid of 20° increments of Phi and Psi. Besides these 81 constrained minimizations, we studied two central sub-regions and two regions at the edges of our maps of complete Phi,Psi space with unconstrained minimization, for a total of 85 target geometries. HF/6-31G(d) and single-point HF/6-311+G(d) calculations were used to find the lowest energies for each geometry. B3LYP/6-31G+G(d) and single point B3LYP/6-311+G(d) calculations were also used for the unconstrained minimizations. For each target, 181 starting geometries were tried (155 for the unconstrained targets). Numerous different starting geometries resulted in the lowest energies for the various target structures. The starting geometries came from five different sets that were based on molecular mechanics energies. Although all five sets contributed to the adiabatic map, use of any single set resulted in discrepancies of 6 to 11 kcal/mol with the final map. Generally, other starting geometries gave lower energies at the three other types of calculation. However, each type gave the same overall lowest energy structure that was found previously by Strati et al. This global minimum, stabilized by highly cooperative hydrogen bonds, is in a region that is essentially not populated by crystal structures. HF/6-31G(d) energy contours of the mapped central region were compatible with the observed crystal structures. Observed structures that lacked O3…O5' hydrogen bonds were about 1 kcal/mol above the map’s minimum, and observed structures that have a pseudo two-fold screw axis ranged from about 0.4 to 1.0 kcal/mol. The HF/6-311+G(d) map accommodated the observed structures nearly as well.