USING PXRF TO MAP TRACE METAL DISTRIBUTIONS IN URBAN AGRICULTURAL SOILS

In Detroit, Michigan, USA and its surrounding metropolitan area, trace metal concentrations in soils are a function of both their glacial origin as well as deposition from anthropogenic sources associated with the region’s industrial heritage. Like many post-industrial cities in the glaciated regions of the USA and Canada, land use repurposing in Detroit includes urban agriculture, which is growing in popularity but also raising concerns about the safety of urban soils. Although Portable X-Ray Fluorescence (pXRF) has been used to characterize metal concentrations in heavily contaminated soils surrounding mines, smelters, and other industrial facilities, pXRF application to urban garden soils where concentrations are orders of magnitude lower (often approaching manufacturer reported detection limits) poses new challenges in developing appropriate methods for trace metal measurement and interpretation.

In this study, soil concentrations of thirteen elements (As, Cr, Cu, Mn, Pb, Rb, Sr, Th, Ti, U, V, Zn, and Zr) were measured in three urban gardens using a Thermo Scientific Niton XL3t 950 GOLDD+ XRF analyzer. Detailed sampling with 60 or more composite samples arranged in a nested grid demonstrated meter-scale trace metal concentration variability in each garden. Although most metals were found in trace amounts, several (e.g., Pb, As) were observed at concentrations exceeding relevant soil safety criteria.

The reliability of our pXRF measurements was assessed using certified reference materials for calibration, ICP-MS analysis of digested samples for confirmation, and nested ANOVA for evaluation of geochemical variation. The quality of our results varied for the different trace metals investigated. Based on project specific data quality objectives and QA/QC procedures, measurements of Pb, Zn, Rb, and Sr were found to be most reliable, V, Ti, and Zr were acceptable, and As, Cr, Cu, Mn, Th, and U, were unreliable for small scale concentration mapping. We attribute these differences in pXRF measurement reliability to a combination of matrix interference effects and decreased pXRF sensitivity as concentrations approach instrument detection limits.