Prediction of the identity of fats and oils by their fatty acid, triacylglycerol and volatile compositions using PLS-DA

The identity of a variety of animal fats and vegetable oils was predicted by three different analytical techniques with help of chemometrics. The sample material of animal origin consisted of milk fat, cow fat, pig fat and poultry fat. The vegetable oils comprised coconut, palm and palm kernel oils. Each product group was composed of at least eight samples of independent batches. For the identity prediction of the fats/oils several (combinations of) datasets were used: absolute and relative measurements of fatty acid compositions, of triacylglycerol compositions, and of combined fatty acid and triacylglycerol compositions. Volatile organic compound compositions were used as well. Fatty acid and triacylglycerol compositions were determined by gas chromatography. Fingerprints of volatile compositions were acquired using Proton transfer reaction-mass spectrometry. The rates of successful prediction were high and varied between 89 and 100%. The 100% rate was obtained for the absolute combined fatty acid/triacylglycerol dataset. Proton transfer reaction-mass spectrometry resulted in 89% correct classifications, has the advantage that it allows very rapid measurements compared to the other techniques, but requires further studie