Comparison of spectroscopic techniques using the adulteration of pumpkin seed oil as example

Abstract

The aim of the present study was to compare different spectroscopic techniques using the example of adulteration of pumpkin seed oil with rapeseed oil in combination with a multivariate regression method. A total of 124 pure seed oils and 96 adulterated samples (adulteration levels from 0.5 to 90.0% w/w) were analyzed using mid infrared, Raman, and 1H-nuclear magnetic resonance spectroscopy. To build quantification models, partial least squares regression (PLS-R) was used. The regression performance parameters, latent variables, and the detection limits (in terms of root mean square error of PLS prediction) calculated when applying the different spectroscopic approaches were compared. For the studied example (pumpkin seed oil adulterated with refined rapeseed oil), the lowest detection limit (3.4% w/w) was obtained for 1H-nuclear magnetic resonance spectroscopy. For the mid infrared and Raman spectroscopy, detection limits of 4.8% w/w and 9.2% w/w, respectively, were obtained, which might be used as screening methods.