Geographical Classification of Authentic Moroccan Argan Oils and the Rapid Detection of Soya and Sunflower Oil Adulteration with ATR-FTIR Spectroscopy and Chemometrics

For the purpose of implementing low-cost, field-deployable analytical techniques to ensure the authenticity and traceability of argan oil, a comprehensive approach that combined mid-infrared spectroscopy data with discriminant and modeling classification methods—including principal component analysis (PCA), soft independent modeling of class analogy (SIMCA), and data-driven soft independent modeling of class analogy (DD-SIMCA)—was applied to classify and check the authenticity of 78 argan oil samples according to their geographical origins and distinguish them from two sets of 24 argan oil samples that contained 5–100% w/w soya or sunflower oil. Optimal models were selected as combinations of many wavelength ranges and data pre-processing methods, thus leading to maximum efficiency for cross-validation. The discrimination approach provided satisfactory classification results with good efficiency for determining argan oil authenticity and detecting adulteration. In addition, an adulteration quantification study was performed with the help of partial least square (PLS) regression of binary mixture, with this demonstrating good linear regression for actual values against predicted ones. The coefficient of determination (R 2 ) was 0.999, while the root mean square errors of calibration (RMSEC) were low at 0.389% and 0.685% w/w and the root mean square errors of validation (RMSEV) were 0.639 and 0.863% w/w for soya and sunflower adulteration, respectively. Moreover, the PLS models best predicted adulterant content, with the R 2 and root mean square error of prediction (RMSEP) being 0.998 and 1.067% (w/w), respectively, for soya and 0.997 and 1.199% (w/w) for sunflower.