This study (purchase required – USD$24.95) developed and validated a classification model using portable near-infrared spectroscopy (NIRS) to detect adulterants in butter. Seven adulterants were studied in a range between 2% and 100%: palm fat, margarine, and cottonseed, canola, sunflower, corn, and soybean oils. A total of 412 samples were analyzed, including 12 adulterated samples seized in an operation of the Brazilian Federal Police.
The authors report that their model performed an almost perfect classification, with the discrimination corresponding to key variables associated mainly with C–H (fat) bonds. This interpretation was corroborated by an exploratory principal component analysis (PCA) model. Samples seized by the Brazilian police were effectively detected as non-authentic. The estimate of quantitative parameters, decision limit (CCα) and detection capability (CCβ), for qualitative methods allowed to establish semi-quantitative models.
They conclude that this approach provides a practical, non-destructive, and environmentally friendly solution for food authentication, addressing the urgent need for reliable methods in combating butter adulteration.
Photo by Sorin Gheorghita on Unsplash
