MULTIVARIATE DATA VISUALIZATION METHODS BASED ON MULTIELEMENTAL ANALYSIS OF WINES AND COFFEES USING TOTAL-REFLECTION X-RAY-FLUORESCENCE ANALYSIS

The potential of applying TXRF for multi-elemental analysis and chemom etric techniques for the classification of wines and coffees has been investigated. The wine samples were all of the same grape variety and coffee samples were sourced from various manufacturers. Elemental conc entrations were determined using a vanadium standard. Principal compon ents analysis (PCA) of the wine data showed that each wine type could be clearly distinguished, and cluster analysis showed the separation b etween the sample types. Star plots were produced for the wines and th ey showed that the variability in the replicate samples was much less than the differences between plots for each wine type. The coffee samp les were not so clearly separated by PCA and cluster analysis, but mos t of the samples were found to be grouped by manufacturer, rather than country of origin. Again, it was demonstrated that star plots could b e used to show the differences between samples. Clearly the applicatio n of simple chemometric techniques to multi-elemental analysis based o n TXRF data is useful for the rapid screening and classification of sa mples such as wine and coffee and this type of methodology is enhanced through the use of descriptive visualisation techniques such as star plots.