Determination of aerosol composition by ED-XRF on Teflon and quartz substrates: Potentialities and limits

Abstract

Energy-dispersive X-ray fluorescence (ED-XRF) is a versatile non-destructive technique to evaluate elemental composition of atmospheric particulate matter (PM) without the need for sample preparation and with high potentiality in source apportionment studies. It is usually applied on Teflon or polycarbonate substrates; however, it would be preferable to use quartz substrates for the possibility to use the same substrate also for carbon detection. In this work an intercomparison among five laboratories on PM10 samples collected on Teflon and quartz filters was done with the specific purpose of understanding the performance of the ED-XRF technique applied to samples collected on quartz substrates. Limits of detection (LODs) on quartz substrates were significantly larger than those on Teflon for the majority of the elements with the exclusion of Cl, Mn, Cu, and Rb, which had comparable LODs for the two substrates. Repeatability on PM10 samples collected on quartz and Teflon substrates was comparable and, on average, better than 10 % for the majority of the elements analysed and better than 5 % for several elements. Comparisons of analysis on Teflon filters for 20 elements obtained by the different laboratories were in the range of ±15 % of the 1:1 line for most of the elements and laboratories. Comparison of measurements on samples collected on quartz and Teflon substrates showed that 17 elements were well correlated (R > 0.7) with average Cquartz/CTeflon ratios in the range 0.6 ± 0.1 (for light elements, due to self-absorption effects) to 1.1 ± 0.1 for the majority of the cases. This suggested that reasonable results could be obtained on quartz substrates for 17 elements, including Na, Mg, and Al, using calibration on Teflon and the ratios Cquartz/CTeflon as correction factors. However, these correction factors were dependent on the instrument and method used for the analysis.