Multielemental analysis of coffee samples using Total Reflection X-ray Fluorescence (TXRF) and Energy-dispersive X-ray Fluorescence (EDXRF) methods

Abstract

This Master’s thesis focuses on the analysis of the multi-elemental com-position of commercial co ee powder samples from Croatian supermar-kets. The co ee samples come from three di erent categories: pure cof-fee, pure instant co ee and co ee with additives (e.g. milk, proteins). The analysis was performed using a Bruker Ranger S2 bench-top energy dispersive X-ray fluorescence (EDXRF) system and a total reflection X-ray fluorescence (TXRF) spectrometer custom-built at the Institute of Atomic and Subatomic Physics, TU Vienna, based on the WOBISTRAX design. Since there is no certified reference material (CRM) for co ee, the results obtained were verified and compared with inductively cou-pled optical plasma spectroscopy (ICP-OES) and flame atomic absorp-tion spectroscopy (FAAS) measurements. The concentrations obtained showed acceptable agreement with the reference measurements. One of the novelties of this study is the detailed discussion of di erent sample preparation procedures. For the EDXRF analysis, pellets with and with-out di erent amounts of wax added for stabilization were compared to lose powder samples. For TXRF, we took the digest previously prepared for ICP measurements and added gallium as an internal standard. For the quantification of the EDXRF and TXRF measurements, we focused on the elements K, Ca, Mn, Fe, Cu, Zn, Sr and Rb. We applied a va-riety of fundamental parameter methods as well as empirical calibration methods. The obtained spectra and concentrations were then used for chemometric analysis, namely Principal Component Analysis (PCA) and Partial Least Square Discriminant Analysis (PLS-DA). The chemometrics were performed using SPSS, a commercial statistical software package. In contrast to many previous works on chemical fingerprinting of food sam-ples, we also obtained chemometric results by inserting the raw spectra directly into the SPSS algorithms. This has the advantage that elemental concentrations do not have to be determined beforehand.