Organic tracer-based source analysis of PM2.5 organic and elemental carbon: A case study at Dongguan in the Pearl River Delta, China

Abstract

Organic carbon (OC) and elemental carbon (EC) are major constituents of PM2.5 and their source apportionment remains a challenging task due to the great diversity of their sources and lack of source-specific tracer data. In this work, sources of OC and EC are investigated using positive matrix factorization (PMF) analysis of PM2.5 chemical composition data, including major ions, OC, EC, elements, and organic molecular source markers, for a set of 156 filter samples collected over three years from 2010 to 2012 at Dongguan in the Pearl River Delta, China. The key organic tracers include levoglucosan, mannosan, hopanes, C27–C33n-alkanes, and polycyclic aromatic hydrocarbons (PAHs). Using these species as input for the PMF model, nine factors were resolved. Among them, biomass burning and coal combustion were significant sources contributing 15–17% of OC and 24–30% and 34–35% of EC, respectively. Industrial emissions and ship emissions, identified through their characteristic metal signatures, contributed 16–24% and 7–8% of OC and 8–11% and 16–17% of EC, respectively. Vehicle exhaust was a less significant source, accounting for 3–4% of OC and 5–8% of EC. Secondary OC, taken to be the sum of OC present in secondary sulfate and nitrate formation source factors, made up 27–36% of OC. Plastic burning, identified through 1,3,5-triphenylbenzene as a tracer, was a less important source for OC(≤4%) and EC (5–10%), but a significant source for PAHs at this site.The utility of organic source tracers was demonstrated by comparing PMF runs with different combinations of organic tracers removed from the input species list. Levoglucosan and mannosan were important additions to distinguish biomass burning from coal combustion by reducing collinearity among source profiles. Inclusion of hopanes and 1,3,5-triphenylbenzene was found to be necessary in resolving the less significant sources vehicle exhaust and plastic burning. Inclusion of C27–C33n-alkanes and PAHs can influence the source profiles resolved by PMF and thereby affect the source contributions to OC and EC. Considerably more OC (44% vs. 27% of OC) was apportioned to the secondary factors when only major components were considered in comparison with the PMF analysis with the full suite of organic tracers, mainly at the expense of coal combustion and industrial emissions. EC apportionment to the few major combustion sources was found more sensitive to inclusion of organic tracers than OC apportionment, with PAHs playing a prominent role. This work demonstrates the importance of having distinct organic tracers in identifying and quantifying OC and EC sources.