Application of low-field and medium-resolution 1H NMR spectroscopy combined with chemometric methods for automotive gasoline quality control

Abstract

This work describes the application of low-field and medium-resolution 1H NMR (LF-1H NMR) combined with PLS and SVM multivariate regression techniques for fast prediction of five fundamental quality parameters of commercial Brazilian gasoline (specific gravity, distillation temperatures of 50 and 90 percent of recovered and olefinic and aromatic content). All these five parameters are of paramount importance to predict the fuel quality and the application of the developed PLS and SVM models showed prediction errors compared to the reference methodologies applied for these essays. Nevertheless, it is important to highlight that in a different way of the reference methodologies, such as distillation or chromatographic methods, the application of LF-1H NMR combined with PLS or SVM models enables the fast determination of those five parameters directly from a single LF-1H NMR spectrum, which is acquired without any sample pre-treatment nor dilution in deuterated solvents, and this whole process takes no more than 15 s.