AbstractExtra virgin olive oil is expensive to produce and its rising popularity as part of a healthy diet over the last two decades has increased world demand. This has led to olive oil being diluted with cheaper vegetable oils in an attempt to increase output and, therefore, profit. Some oil producing areas are considered to produce a superior quality olive oil and some suppliers import oil from other, less fashionable, areas and export the oil as 'bottled in' the more favoured region. This thesis examines the questions that arise concerning the authenticity of olive oil. It investigates whether spectroscopic data (NMR and IR) analysed by multivariate analysis (Unscrambler 7.5, Camo ASA) can reliably characterise Greek olive oil samples by their area and year or identify adulteration with sunflower oil. Extraction of the volatile compounds would also be considered to discover whether this fraction of olive oil could be useful in the determination of oil origin.
The PLS1 regression models constructed using IR data for a set of samples from the Crete 1995/96 harvest gave a low error of prediction for the measured percentage sunflower adulteration against the predicted values. An RMSEP of 0.795% sunflower oil with an offset of 0.147% using 11 regression components gave good prediction of adulteration of unknown samples.
PCA models of proton NMR data produced classification plots with clear separation between oils originating from Crete and those from other areas of Greece. They also gave unequivocal classification of each year's harvest, making it possible to ascertain the year and area of olive growth.
SFE and SPME were both used for the extraction of the volatile fraction of olive oil with separation and detection by GC-MS. Hierarchical clustering (SPSS) showed rudimentary grouping of the oils by area of origin using both techniques. Coupling the two extraction techniques gave a novel methodology, which with improvements, could become an exciting new procedure.
|Date of Award