Raman
spectroscopy is largely informative about organic matter
(OM) chemical composition and structure and can therefore be applied
to evaluate the thermal maturity of coals, carbonaceous materials
and kerogens, i.e., their degree of evolution during burial heating.
The evaluation of OM maturity is commonly performed by band-fitting
followed by the measure of suitable spectral parameters (typically
band separations and area ratios). However, this procedure can introduce
some subjectivity both in the number and line-shape of the fitting
bands and in the definition and selection of the most meaningful spectroscopic
parameters. Here, a principal component analysis–partial least
squares regression (PCA–PLS) chemometric approach for the treatment
of spectra is presented, that is intrinsically unaffected by such
arbitrariness. In fact, the total spectrum is analyzed in order to
extract the spectroscopic ranges of maximum variance in a multivariate
approach. In addition, being automated, the treatment is well-suited
to huge sets, as those commonly collected to appropriately and reliably
evaluate the chemical–physical and geological variables in
an exploration well. As a case study the maturity profile (maximum
paleotemperature) of OM from a well drilled in the Lower Congo basin
(offshore Angola) is evaluated, both by spectral parameters and by
PCA–PLS analysis, with results in good agreement on the basis
of the prediction error. A set of coals is adopted as a reference
to predict the reflectance in oil (%R
o) profile from the Raman data. The comparison
of the calculated %R
o values with the experimental ones, presenting, in this case, some
suppressed values, shows that the Raman analysis is not affected by
underestimation of maturity as the vitrinite reflectance may be.