Optimization of portable X-ray fluorescence spectrometry for the assessment of soil total copper concentrations: application at an ancient smelting site

Abstract

Purpose Copper (Cu) is the earliest anthropogenic metal pollutant, but knowledge of Cu soil concentrations at ancient metalworking sites is limited. The objective of this work was to examine the ability of portable X-ray fluorescence to quantify Cu in soils at such sites. Materials and methods Using a Bruker Tracer III-SD pXRF, we examine factory "scan" settings versus simple instrument parameter changes (a reduction in energy settings from 40 to 12 kV) to target analysis for Cu. We apply these to a set of uncontaminated samples (n = 18, < 92 mg Cu kg-1) from Central Thailand and compare results to standard wet chemistry analysis (aqua regia digestion and ICP-OES analysis). We then apply the optimized method to a set of highly contaminated samples (n = 86, < 14,200 mg Cu kg-1) from a known ancient smelting site. Results and discussion We demonstrate that simple changes to factory recommended "scan" settings can double the sensitivity of Cu determination via pXRF ("optimized limit of determination" of 19.3 mg kg-1 versus an initial value of 39.4 mg kg-1) and dramatically improve the accuracy of analysis. Changes to other results for other elements are variable and depend on concentration ranges, soil matrix effects, and pXRF response for the individual element. We demonstrate that pXRF can accurately determine Cu across a wide concentration range and identify grossly contaminated soil samples. Conclusions We conclude that pXRF is a useful tool to rapidly screen and analyse samples at remote sites and can be applied to ancient metalworking sites. Simple optimization of the pXRF settings greatly improves accuracy and is essential in determining comparative background concentrations and "unaffected" areas. Application to other elements requires further element and matrix specific optimization.