Inter-Laboratory Comparison of Air Particulate Monitoring Data

Abstract

This paper compares three analytical methods that are often used to analyze composition of atmospheric aerosol: Ion Chromatography (IC), Proton Induced X-ray Emission (PIXE), and X-ray Fluorescence (XRF). Three monitoring studies are discussed: (1) a comparison of air particulate data collected by several independent sampler/analytical technique suites run by different laboratories; (2) a study involving two identical samplers and a single suite of analytical techniques; and (3) analysis of identical aerosol samples by two different techniques (XRF vs. PIXE). While the XRF versus PIXE project shows a very good agreement for most elements, the first interlaboratory study demonstrates the "real-life" noise introduced into the final data set by various sampling complications and different collection characteristics of the samplers used. The XRF versus PIXE study also revealed an unexplained deviation in measured sulphur concentrations for very lightly loaded samples. In the five-sampler comparison, two data sets provided by IC were approximately 20% lower than the three data sets obtained by PIXE and XRF. When two identical IMPROVE-compatible samplers were used and samples were subjected to similar procedures and the same analytical techniques, the variability between the two air concentration data sets significantly decreased.

PIXE and XRF analysis of particulate matter samples: an inter-laboratory comparison

“…XRF and PIXE measurements are in good agreement: differences between concentrations obtained by the two methods are reported in Table 1 and are always within 10%, except in two cases (being anyway at maximum 15%): this result is comparable with those reported in literature [13,14]. These discrepancies can be due to different X-ray spectra fittings, and to sample and blanks Table 1 Results of the XRF versus PIXE comparison on 14 aerosol samples, collected on Teflon, polycarbonate and cellulose mixed esters filters: slopes and correlation coefficients, obtained by a fitting procedure, are reported (linear regression plots are shown in Fig.…”

Organic and inorganic decomposition products from the thermal desorption of atmospheric particles

“…Multiple LV-OOA components were observed and combined into a single source type called cLV-OOA, which accounted for 31.3 % of total observed OA and whose mass spectrum contained a dominant contribution from CO + 2 . This aerosol type is thought to be the most oxygenated and aged aerosol component and has been observed in many studies Ng et al, 2010). Here, we observe an increased correlation between TAG decomposition m/z 44 signal and cLV-OOA (r = 0.73) and an even higher correlation with one of the individual LV-OOA components (r = 0.90) (see Table 1), although there is no known explanation in the source or process differentiation between the multiple LV-OOA components.…”

“…Fine-mode particulate matter is derived from a range of primary sources, such as combustion, and secondary sources where gases oxidize in the atmosphere to produce lower-volatility products that create secondary aerosol. Globally, the majority of fine PM is secondary in nature and made up of thousands of individual chemicals Ng et al, 2010;Goldstein and Galbally, 2007), creating challenges in apportioning the original emission sources of this material, formation pathways, and oxidative evolution in the atmosphere.…”

Estimating Precision Using Duplicate Measurements

“…Air monitoring is usually concerned with high concentrations and its practitioners thus express precision in relative, dimensionless terms. [3][4][5][6][7][8][9][10] To obtain a stable estimate of relative precision, the concentration range with uncertainties dominated by additive uncertainties must be eliminated from the estimates; however, exactly which concentrations to exclude is not clear. The additive and multiplicative uncertainties are unknown and can vary by species, site, filter lot, and even analysis date.…”