Quantitative Analysis of Sunflower Lecithin Adulteration with Soy Species by NMR Spectroscopy and PLS Regression

NMR spectroscopy was used to distinguish pure sunflower lecithin from that blended with soy species, and to quantify the degree of such adulteration. Sample preparation included liquid extraction of lecithin blends and measurement of polar and non‐polar fractions using ³¹P and ¹H NMR spectrometry. Several phospholipid species, linolenic acid and stachyose were found to be characteristic for sunflower lecithin authentication. For quantitative analysis, partial least squares regression (PLS) was utilized for modeling NMR data of authentic lecithin samples and in‐house prepared blends (n = 80). The models based on phospholipid, fatty acid and saccharide distributions were validated using independent test sets. PLS based on saccharide composition is able to estimate lecithin falsification regarding its vegetable origin with the sensitivity and root mean square error of validation below 1 % w/w and 3.5 % w/w, respectively. Prediction error was improved by modeling the whole lecithin profile (phospholipids, fatty acids, and saccharides). Repeatability and precision expressed in coefficients of variations was estimated to be below 8 %. The developed approach allowed the evaluation of the composition of lecithin blends of unknown soy and sunflower content on the basis of multiple chemical components.