Simultaneous Spectrophotometric and Chemometric Determination of Oleic, Linoleic, and Linolenic Fatty Acids in Vegetable Oils

Simultaneous spectrophotometric methods are described for the determination of oleic, linoleic, and linolenic fatty acids in vegetable oil samples using neural network (NN), principal component regression (PCR), partial least squares (PLS1 and PLS2), and K-matrix (KM) algorithms. The assay used to obtain the absorbance spectrum unique for each fatty acid is selective to the -CH=CH-CH2 that reaches spectral maturity after 15 minutes. Results show that the root mean square error of prediction (RMSEP) compared quite equally well for PCR, PLS1, and PLS2 algorithms for the three components, with these algorithms outperforming NN and KM. In sunflower and vegetable oil unknown samples, PLS2 mostly yielded a better performance than PLS1 and PCR algorithms when validated with the USDA database. This paper shows how the novel assay coupled with chemometric algorithms might provide faster and cheaper methods for simultaneously quantitating oleic, linoleic, and linolenic fatty acids in vegetable oil samples.