Autenticação do óleo de andiroba por RMN e ferramentas quimiométricas

Abstract

Andiroba oil is widely known and traded in Brazil and abroad. The most abundant andiroba species is Carapa guianensis, which is considered native from the Amazon region, and can be found mainly in the northern region of the country. Most of the studies of C. guianensis oil are related to its chemical constitution, along with their medicinal properties, especially anti-inflammatory, analgesic, insecticidal, repellent, antibacterial, antiparasitic and anticancer properties. Up to 95% of andiroba oil is constituted by fatty material (palmitic, oleic, stearic and arachidic acid) and less than 5% of unsaponifiable matter, mostly limonoids. Due to its great commercial and medicinal value, this work aimed to employ one-dimensional 1H NMR spectroscopy for the evaluation and control of adulteration of andiroba by soybean oil. This technique allowed to characterize and to identify adulterations of andiroba oil with soybean oil identified with the help of chemometric study employing both exploratory and classificatory methods. The characterization of the fatty chain of the andiroba and soybean oil was determined by 1D and 2D hydrogen NMR experiments (COSY), as well as 2D heteronuclear experiments (HMBC and HSQC). These experiments confirmed the different characteristics of the fatty chain of both vegetable oils. Andiroba oil has a very higher percentage of saturated chains and oleic acid, whereas soybean oil a higher proportion of unsaturated chains. The increasing area of the unsaturated hydrogens peaks in the 1H NMR spectra help to identify the adulteration of andiroba by soybean oil. In addition, the soybean also contains the linolenic acid that is not present in the andiroba oil. The 1H NMR data were analyzed by statistical exploratory methods to evaluate the behavior and trends of the samples by Hierarchical Cluster Analysis (HCA), Fuzzy C-Means and Principal Component Analysis (PCA). From the exploratory results it was possible to perform the interval least squares regression (I-PLS) method, in which a model was proposed to estimate the adulteration in each commercial sample acquired in the market and fairs of the city of Manaus. Firstly, a known adulteration database was built to validate the calibration method, from a combination of standard andiroba oil with increasing amounts of soybean oil. This enabled us to classify the samples as non-adulterated and adulterated, and also to estimate the adulteration detected in commercial samples.