Lone pairs vs. covalent bonds: conformational effects in bicyclo[3.3.1]nonane derivatives

Abstract

Investigations on the relative energy of two least-strain conformers for bicyclo[3.3.1]nonane 1, bicyclo[3.3.1]nonan-9-one 2, and their heteroanalogues: 3,7-dimethyl-3,7-diazabicyclo[3.3.1]nonane 3, 3,7-dimethyl-3,7-diazabicyclo[3.3.1]nonan-9-one 4 were performed using the calculations from the first principles (ab initio, DFT) as well as by semiempirical (NDDO, DFTB) and empirical (molecular mechanics, MM) techniques. For these quite simple structures, serious discrepancies in results of modeling between methods of different origins were revealed. Nonempirical calculations state that the “double chair” (CC) form is the most favorable for carbobicyclic structures 1 and 2, while 3,7-dimethyl-3,7-diaza compounds 3 and 4 are in general more prone to adopt the “chair-boat” (CB) conformation. The classical rationalization of these quantum chemistry results leads to the hypothesis similar to one that underlies the Gillespie VSEPR concept, namely that the 3,7-repulsion of lone electron pairs is stronger than the corresponding interaction of hydrogen atoms of C–H bonds. The semiempirical NDDO calculations retain the qualitative correspondence of the results to those of the ab initio calculations, while the results of more recent DFTB approaches are closer to MM in their qualitative inconsistency with high-level ab initio methods. In particular, for 4 the relative energy of CC is severely underestimated, erroneously predicting the predominance of this form over CB. The origin of this failure could lie in the relatively coarse parameterization of common force fields when concerning the subtle interplay between different types of interatomic interactions and could be recovered, although only partially, by the proper choice of the charge scheme to use the atomic-centered charges in the explicit account for the non-valency interactions in the Coulombic form.