Improving the accuracy of EDXRF results in gold alloy analysis by matrix effect correction

Energy-Dispersive X-Ray Fluorescence (EDXRF) is a faster non-destructive analysis and demands less sample preparation than the fire assay technique. This research sought to minimise the matrix effect (inter-element) by studying the role of matrix-specific materials and thus improving the accuracy of...

Full description

Bibliographic Details
Published in:Spectrochimica Acta - Part B Atomic Spectroscopy
Main Authors: Mazuki A.A.M., Mahat M.M., Abdullah S., Ramli R., Nor F.M.
Format: Article
Language:English
Published: Elsevier B.V. 2023
Online Access:https://www.scopus.com/inward/record.uri?eid=2-s2.0-85147124801&doi=10.1016%2fj.sab.2023.106629&partnerID=40&md5=31874430cffa8356ceab5c6a0adcb7ad
Description
Summary:Energy-Dispersive X-Ray Fluorescence (EDXRF) is a faster non-destructive analysis and demands less sample preparation than the fire assay technique. This research sought to minimise the matrix effect (inter-element) by studying the role of matrix-specific materials and thus improving the accuracy of EDXRF in gold alloy analysis. The combination of empirical calibration with standardless Fundamental Parameter (FP) was expected to fine-tune the analysis of precious metals, such as gold (Au), silver (Ag), and copper (Cu). The acquired R2 of 0.9999 for all metals signalled an alignment between the certified value and EDXRF measurement. The technique was validated with CRMs to determine corrective K-factors to realign systematic measurement errors due to alloy mixture. The relative error (rel%) for all measurements was <0.1 rel%. The relative standard deviation (%rsd) was reduced to <0.11%rsd for pure Au, 0.16%rsd for Au-Ag and Au-Cu mixture, and <0.20%rsd for Au-Ag-Cu mixture with K-factor correction. A similarity to the fire assay might reduce bias. The null hypothesis of equal mean values of fire assay and XRF data was confirmed by the student's t-test. This outcome showed that the XRF provided acceptable precision and verified the null hypothesis on the equivalence between mean results. © 2023 Elsevier B.V.
ISSN:05848547
DOI:10.1016/j.sab.2023.106629