Factors influencing the fatty acid determination in fats and oils using Fourier transform near-infrared spectroscopy

Fourier transform near-infrared (FT-NIR) technology is matrix dependent and thus highly dependent on factors that influence the absorption spectra. Ignoring these factors during the development of FT-NIR models will affect the accuracy and reliability of the classification of fats and oils and the determination of their fatty acid (FA) composition. Four factors were studied: the temperature at which samples are scanned, differences in FA chain length and number of double bonds, and the presence of non-triacylglycerol components. The results showed that an increase in the recording temperature decreased the absorption peak intensity, but not the position. FT-NIR spectral differences were linked to variations in molecular vibrations resulting from the number of carbon atoms or double bonds in the FA. The FT-NIR method could clearly differentiate between chain lengths from 10:0 to 18:0 and numbers of double bonds from zero (18:0) to three (18:3). Contaminants in triacylglycerols altered the FT-NIR spectra, resulting in increased errors in the FA content. An increased concentration of P-sitosterol in triolein decreased or increased the observed contents of cis9-18:1 and cis11-18:1, respectively. An FT-NIR model adjusted for the phytosterol content corrected this discrepancy. The revised FT-NIR model was successfully used to provide the accurate FA compositions of commercial sunflower oils.