RESUMO
In this study an analytical methodology for food analyses combining X-ray spectroscopy (XRS) with partial least-squares (PLS) data treatment was developed. Fifteen tea samples were purchased at a local market, and XRS spectra were obtained without sample pretreatment. For comparison of the metal concentrations, the samples were also mineralized, and six elements were determined using flame atomic absorption spectrometry (Ca, Fe, Mg, and Mn), flame atomic emission spectrometry (K), and thermospray flame furnace atomic absorption spectrometry (Zn). The spectral information obtained from XRS and the metal concentrations found using the alternative techniques were employed to generate six PLS models. The Ca and Mn models required four latent variables (LV), Fe, five LV, K, two LV, and Mg and Zn, three LV. The limits of quantification for these models were 614, 134, 761, 140, 85, and 1 mg kg(-1) for Ca, Fe, K, Mg, Mn, and Zn, respectively.