Classification of Vegetable Oils by Principal Component Analysis of FTIR Spectra

Principal component analysis (PCA) of infrared spectra of different known vegetable oils is used to determine the identity of several unknown vegetable oils. The experiment requires access to an FTIR spectrometer with 1-cm-1 resolution and peak-finder capability, and a software package that can perform principal components analysis. Students acquire IR spectra of a series of known vegetable oils, choose spectral features to be analyzed by PCA, and create scatter plots of principal component scores of each oil. The unknowns are then analyzed, plotted, and identified based on their proximity to the knowns in principal component space. The PCA data analysis extracts the useful information from a highly correlated data set; it is easier to identify the unknown oils by looking at the plot of principal component scores than by looking at the IR spectra. This practical application of PCA in an instrumental laboratory introduces students to chemometrics and allows them to experience first-hand the utility of a multivariate data analysis technique.