Estimating Uncertainties for IMPROVE_A Carbon Fraction Measurements in PM2.5 Across Analyzer Brands

摘要

Carbon fractions measured during organic carbon/elemental carbon (OC/EC) analysis of PM2.5 quartz filter samples are not independent analytes but rather are defined by the specific conditions (primarily temperature and trace oxygen contamination level) under which they are evolved from the sample during analysis. In OC/EC analysis, none of the carbon fractions except for total carbon (TC) is independent of the other fractions. For example, TC is calculated as the sum of the seven carbon peaks, and TC is always equal to the sum of OC and EC. Because of this, all carbon fractions except for TC are not independent analytes in the same sense that an ion or an element can be measured independent of all other ions or elements present in the sample. In previous work (Peterson and Richards, A&WMA Annual Meeting, Portland, OR, 2008), the authors demonstrated an empirical technique for estimating uncertainties for non-independent carbon fractions measured by the EPA’s Chemical Speciation Network/Thermal-Optical Transmittance (CSN/TOT) method using multiple analyzers from the same manufacturer. This paper will apply the technique to estimate uncertainties for IMPROVE_A carbon fractions across two different thermal-optical carbon aerosol analyzers from different manufacturers. The procedure allows for estimating extended uncertainties that are desirable when human health or safety is a concern and for estimating more typical uncertainties that could be used for samples collected where health and safety are not a concern.