Information depth of elements affects accuracy of parallel pXRF in situ measurements of soils

Abstract

Though not litigable in most European countries, portable X-ray fluorescence spectrometers (pXRF) provide cost- and time-effective as well as prompt information about hot spots of inorganic soil contaminants. The quality of aqua regia analysis of contaminants can be approximated by a thorough sample preparation, i.e., homogenization, grinding, and sieving of the examined soil before pXRF measurement is carried out. However, elaborate sample preparation causes a trade-off in terms of the desired straightforwardness of the pXRF method. For a first assessment of the in situ accuracy of pXRF measurements, two equal pXRF devices were used in parallel to determine the contents of As, Cu, Fe, Mn, Pb, Rb, Sr, Zn, and Zr of 9 identical points of a riparian soil profile. Maximum measurement values were not restricted to one pXRF device, but changed from element to element. Pearson correlation coefficients of the parallel measurements varied between 0.07 (Cu) and 0.80 (Zn), reflecting small-scale heterogeneity of the soil constituents as well as element-specific interferences. For each element, overall deviations between measurement parallels were expressed as the root-mean-square error (RMSE) and contrasted against the element-specific information depth in soil, i.e., the depth interval, from which the received spectral signals originate. From this, a gradual relation could be derived: The greater the information depth, the more stable the measured element value turns out. This context should be taken into account, when interpreting contents of elements with small atomic numbers.