Experimental characterization of sulfur interference in IMPROVE aluminum and silicon XRF data

Silicon (Si) and aluminum (Al) are important elements for evaluating soil content in particulate matter. However, these elements are in some cases difficult to measure by X-ray fluorescence (XRF), the method used by the two large particulate matter networks in the U.S., the Interagency Monitoring of PROtected Visual Environments (IMPROVE) network and the Chemical Speciation Network. The IMPROVE network has issued data advisories on Al and Si based on analysis of data from the networkʹs ambient samples. The analysis shows that when sulfur concentrations are high, silicon concentrations can be larger than expected and aluminum may not be detected when it is likely to be present (vista.cira.colostate.edu/improve/data/QA_QC/Advisory.htm). In this research, a laboratory study was undertaken to determine the conditions and extent of measurement uncertainty in Al and Si when sulfur concentrations are high. Three possible causes of the uncertainty are spectral interference due to a low-energy tail on the S peak in the region of Al and Si peaks, attenuation of Si and Al fluorescence by S and the low instrument responses of Si and Al. Ambient samples encompassing the Al and Si mass range in the IMPROVE network and with low sulfur were collected. A layer of ammonium sulfate was sampled onto the deposit in the laboratory. Although the real Si concentrations were unchanged by the addition of pure ammonium sulfate, the Si concentrations reported by XRF were substantially higher. For samples with S/Fe ≥ 8, Si is over-reported by 6–300% and the Al concentrations have large uncertainties. No sulfur interference is observed for samples with S/Fe < 8. These measurement errors are expected to occur in 50% of IMPROVE data from December 1, 2001 through December 31, 2010. New XRF instruments with different hardware and software are being utilized by IMPROVE to measure elemental masses for samples beginning January 1, 2011.