MULTIVARIATE-ANALYSIS OF C-13 NUCLEAR MAGNETIC-RESONANCE SPECTRA - IDENTIFICATION AND QUANTIFICATION OF AVERAGE STRUCTURES IN PETROLEUM DISTILLATES

The {C10, C11,…, C25} fractions from two North Sea crude oils are investigated. For each fraction CHn(n = 1, 2, 3) 13C nuclear magnetic resonance subspectra are obtained with the use of distortionless enhancement by polarization transfer subspectral editing. The combined spectral profiles for each fraction, consisting of 9651 data points, are reduced to 1024 pooled variables by using a “soft” maximum entropy criterion. A principal component analysis is performed. Two principal components account for 85.9% of the variance in the reduced data. Spectral assignments show that the two principal components are related to variations in the abundances of n-alkanes and isoalkanes. After the variables are divided by their means, the second principal component discriminates between the two oils in all fractions. The loading profile forthis principal component shows that the two oils differ mainly in the content of saturated soprenold structures. The content of the saturated isoprenoid structures is quantified by using the marker-object projection technique, with the spectrum of pure pristane as a marker object. © 1990, American Chemical Society. All rights reserved.