RESUMO
The potential of the approach combining nuclear magnetic resonance (NMR) spectroscopy, relaxed grid search (RGS), molecular dynamics (MD) simulations, and quantum mechanical (QM) calculations for the determination of diastereomer configurations is demonstrated using four diastereomers of a trisubstituted epoxide. Since the change in configuration of the chiral center is expected to change the distribution of conformer populations (including those of side-chain rotamers), changes in NMR parameters [chemical shifts, J couplings, and nuclear Overhauser effects (NOEs)] are expected. The method therefore relies on (1) identification of possible conformations in each diastereomer using relaxed grid search analysis and MD simulations; (2) geometry optimizations of conformers selected from step (1), followed by calculations of their relative energies (populations) using QM methods; (3) calculations of averaged NMR parameters using QM methods; (4) matching calculated and experimental values of NMR parameters of diastereomers. The diastereomer configurations are considered resolved, if three NMR parameters different in nature, chemical shifts, J couplings, and NOEs, are in agreement. A further advantage of this method is that full structural and dynamics characterization of each of the diastereomers is achieved based on the joint analysis of experimental and computational data.
RESUMO
The potential of an approach combining nuclear magnetic resonance (NMR) spectroscopy, molecular dynamics (MD) simulations, and quantum mechanical (QM) calculations for full structural characterizations in solution is assessed using cyclic organic compounds, namely, benzazocinone derivatives 1-3 with fused five- and eight-membered aliphatic rings, camphoric anhydride 4, and bullvalene 5. Various MD simulations were considered, using force field and semiempirical QM treatments, implicit and explicit solvation, and high-temperature MD calculations for selecting plausible molecular geometries for subsequent QM geometry optimizations using mainly B3LYP, M062X, and MP2 methods. The QM-predicted values of NMR parameters were compared to their experimental values for verification of the final structures derived from the MD/QM analysis. From these comparisons, initial estimates of quality thresholds (calculated as rms deviations) were 0.7-0.9 Hz for (3)J(HH) couplings, 0.07-0.11 Å for interproton distances, 0.05-0.08 ppm for (1)H chemical shifts, and 1.0-2.1 ppm for (13)C chemical shifts. The obtained results suggest that the accuracy of the MD analysis in predicting geometries and relative conformational energies is not critical and that the final geometry refinements of the structures selected from the MD simulations using QM methods are sufficient for correcting for the expected inaccuracy of the MD analysis. A unique example of C(sp(3))-H···N(sp(3)) intramolecular noncovalent interaction is also identified using the NMR/MD/QM and the natural bond orbital analyses. As the NMR/MD/QM approach relies on the final QM geometry optimization, comparisons of geometric characteristics predicted by different QM methods and those from X-ray and neutron diffraction measurements were undertaken using rigid and flexible cyclic systems. The joint analysis shows that intermolecular noncovalent interactions present in the solid state alter molecular geometries significantly compared to the geometries of isolated molecules from QM calculations.
