SIMULATION OF C-13 NUCLEAR-MAGNETIC-RESONANCE SPECTRA OF POLYCYCLIC AROMATIC-COMPOUNDS

Spectra-structure relationships are developed that allow carbon-13 nuclear magnetic resonance chemical shifts to be simulated in a set of diverse polycyclic aromatic compounds. Employing 33 compounds that encompass 24 different aromatic ring backbones, six linear models are generated that allow complete spectra to be simulated to an average error of 0.68 ppm. Successful model generation is found to be dependent upon grouping atoms of similar chemical environments together and the use of structural parameters based on Huckel molecular orbital and molecular mechanics calculations. The computed models are evaluated by use of a separate prediction set of 11 compounds not included in the model generation work. The average spectral prediction error for these compounds is 1.02 ppm, even though seven of the 11 compounds contain ring backbones not found among the 33 compounds used in computing the models.