Rapid quantitative determination of butter adulteration with palm oil using the DSC technique

Butter, a dairy product containing a large amount of milk fat, is often the object of economic fraud involving replacement with cheaper animal fats or vegetable oils (mainly palm oil). Determination of adulterants in butter continues to be an analytical challenge. In order to obtain a rapid method, the aim of this study was to evaluate the applicability of the DSC technique for quantitative assessment of butter adulteration with palm oil (PO). Validation of the developed DSC method required the determination of linearity and repeatability, as well as the accuracy by performing the recovery experiments. The addition of PO to butterfat (BF) results in an altered shape of the melting curve. With an increase of PO content in BF from 0% to 35%, the melting temperatures: T1 (peak of low melting fraction, LMF) and Tend increased by approx. 2 °C, the area of LMF (ΔH1) from 34 to 55% and the peak height (h1) from 145 to 234 mW g−1. For the peak of the medium melting fraction (MMF) temperature T2 decreased by 2 °C, the area (ΔH2) dropped from 31% to 10% and the peak height (h2) decreased from 345 to 200 mW g−1. Linearity was calculated measuring DSC parameters (5 repeats) of nine concentration levels of PO in butterfat within the range from 2 to 35%. Various DSC parameters of the melting curve, such as the peak area of LMF and MMF (ΔH1, ΔH2), peak heights (h1, h2), as well as complex variables: (ΔH1/ΔH2, ΔH1/ΔH3, h1/h3, h1/h2) were used for simple and multiple linear regression (MLR) analyses. The obtained equations were used for the prediction of PO determination in six adulterated BF samples. Among all of the regression equations, the best suitability was found for the MLR model with complex variables, for which high correlation coefficient between the true and measured values was found, amounting to 0.999, as well as low average value of bias (0.84). The accuracy of the DSC method, expressed as bias (b) and recovery (R), was compared with the results recorded by the Official method, for which the bias value was equal to 2.2. When comparing R values, for the Official method it was 89.2%, whereas for the DSC method (MLR model) it was 90.7%. The results clearly indicate that the DSC technique is applicable for the quantitative detection of PO in BF within the range of concentrations from 2 to 35%, based on the parameters of peak area and peak height.