Filter artifacts associated with particulate measurements: Recent evidence and effects on statistical relationships

• Published in: Atmospheric environment (1994) 1994, Vol.28 (20), p.3233-3249
• Main Author: Lipfert, Frederick W.
• Format: Article
• Language: English
• Subjects: Analysis methods; Applied sciences; Atmospheric pollution; Exact sciences and technology; Nitrates; Pollution; Q1; Sulfur compounds
• ISSN: 1352-2310; 1873-2844
• DOI: 10.1016/1352-2310(94)00167-J

Suspended particulate data from five sites in New York State and from national networks are analyzed to deduce the magnitudes of artifacts created by sampling media. Such artifacts are inferred by comparing results from co-located samplers that used different protocols, including high-volume total suspended particulate (TSP) samplers with alkaline glass-fiber filters and the nearly neutral Teflon or quartz filters used in size-classified (PM 2.5, PM 10, and PM 15) sampling. Comparisons are presented for sulfate ion, nitrate ion, total mass, and the “remainder” mass, defined here as the total mass less the contributions of SO 4 2− and NO 3 −. The sulfate artifact implied from hi-vol sampling was found to be about 5 μg m −3, which is larger than some previous investigations had indicated and constitutes a substantial fraction of the indicated average concentrations. A reliable estimate of the nitrate artifact could not be made from these data, since the TSP NO 3 − values appeared to be inflated due to collection of HNO 3 and the PM 10 NO 3 − values appeared to reflect loss of material, possibly from volatilization. Thus, TSP total mass values may be overstated by as much as 10–20μg −3 in summer and PM 10 values may be understated by an undetermined amount. The relationship between TSP and PM 10 was seen to be site- and year-specific, and differences among cities in average particle loading were seen to be more pronounced for the larger particles. Estimates of the magnitudes of ecological or health effects based on data obtained from TSP sampling should be adjusted to account for filter artifacts.