Analysis of 1H NMR spectra of diesel and crambe biodiesel mixtures using chemometrics tools to evaluate the authenticity of a Brazilian standard biodiesel blend

Abstract

A methodology was developed to monitor the content of crambe biodiesel in mixtures with conventional diesel using hydrogen nuclear magnetic resonance (1H NMR) spectroscopy combined with the orthogonal projections on the latent structure-discrimination analysis (OPLS-DA). The efficiency of the developed OPLS-DA model was analyzed based on the criteria of true response statistics: false positive and false negative rate, sensitivity, specificity, efficiency and Matthew's correlation coefficient, where the sensitivity (true positive rate) and specificity (true negative rate) were both equal to 1 and the false positive and false negative rates were both equal to 0, which means that all samples to be predicted as belonging to the diesel class of interest, B10 (containing 10% biodiesel and 90% pure diesel), were predicted in class 1, and all samples to be considered as belonging to the diesel class, not of interest, BX (biodiesel content less and greater than in B10), were predicted in class 0. These results showed 100% correct classification of the training and test set samples for B10 and BX, demonstrating a high efficiency of the OPLS-DA model in the monitoring of crambe methyl biodiesel content when mixed with diesel in various proportions. The excellent results in the application of this model suggest that this analytical methodology is feasible, efficient and suitable for use by inspection agencies to control the quality of this fuel.