Partial least squares modelization of energy dispersive X-ray fluorescence

Abstract

As a proof of concept, a green methodology has been developed for the energy dispersive X-ray fluorescence (ED-XRF) determination of calcium, potassium, iron, magnesium, aluminum, chromium, strontium, phosphorus and nickel in the peel of untreated kaki fruit (Diospyros kaki. L) samples. ED-XRF spectra of fifty-six kakis purchased in the local area of LLombay (Valencia) were obtained directly from samples without any previous treatment and without sample damage just after cleaning the fruit with distilled water. Inductively Couple Plasma Optical Emission Spectrometry (ICP-OES) was used as a reference method to determine the mineral elements after microwave assisted acid digestion. XRF spectra and concentration values obtained by ICP-OES were processed using partial least squares (PLS) data treatment to build the corresponding chemometric models for prediction of mineral profile of samples. PLS-ED-XRF permits a direct and accurate determination of Ca and K in kaki peel. For Al, Fe, Mg, Ni and Sr screening semiquantitative results were obtained. Concentrations obtained directly by the internal calibration of instrument, using GeoChem Trace model, were also compared with data predicted by chemometric models being found that PLS models must replace the calibration of the instrument for thus kind of analysis.