A MOLECULAR MECHANICAL MODEL THAT REPRODUCES THE RELATIVE ENERGIES FOR CHAIR AND TWIST-BOAT CONFORMATIONS OF 1,3-DIOXANES

We present molecular mechanics calculations on the conformational energies of several `2,2-dimethyl-trans-4,6-disubstituted-1,3-dioxanes. Previous studies by Rychnovsky et al. have suggested that the relative conformational energies of chair and twist-boat forms of these 1,3-dioxanes were poorly represented by the molecular mechanical models MM2* and MM3* (MacroModel implementations of MM2 and MM3) both when compared to experiment and to high-level quantum mechanical calculations. We have studied these molecules with a molecular mechanical force field which features electrostatic-potential-based atomic charges. This model does an excellent job of reproducing the relative conformational energies of the highest level of theory (MP2/6-31G*) applied to the problem. Furthermore, when empirically corrected using the MP2/6-31G* relative conformational energies of the unsubstituted compound 2,2,4-trimethyl-1,3-dioxane, the absolute energy differences calculated with this new model between the chair and twist-boat conformers for five substituted compounds are within an average of 0.30 kcal/mol of the MP2/6-31G* values. The correlation with experiment is also very good. One can, however, modify the initial molecular mechanical model with a single V-1(-O-C-O-C-) torsional potential and do an excellent job in reproducing the absolute conformational energies of the dioxanes as well, with an average error in conformational energies of 0.45 kcal/mol. This same torsional potential was independently developed by comparing ab initio and molecular mechanical energies of the molecule 1,1-dimethoxymethane. Thus, we have succeeded in developing a general molecular mechanical model for 1,3-dioxoalkanes. In addition, we have compared the standard MM2 and MM3 models with MM2* and MM3* (ref. 2) and have found some significant differences in relative conformational energies between MM2 and MM2*. MM2 has an improved correlation with the best ab initio data compared to MM2* but is still significantly worse than that found with lower-level ab initio or AM1 semiempirical quantum mechanics or the new molecular mechanical model presented here. MM3 leads to conformational energies very similar to MM3*. Energy component analysis suggests that the single most important element in reproducing the conformational equilibrium is the electrostatic energy. This fact rationalizes the success of AMBER models, whose fundamental tenet is the accurate representation of quantum mechanically calculated molecular electrostatic effects. (C) 1995 by John Wiley and Sons, Inc.