A Coupled Volatility and Molecular Composition Based Source Apportionment of Atmospheric Organic Aerosol

Abstract

We apply non-negative matrix factorization (NNMF) to molecular composition and volatility measurements of ambient sub-micrometer particles made using a high-resolution time of flight chemical ionization mass spectrometer (HRToF-CIMS) equipped with a custom filter inlet for gases and aerosols (FIGAERO) as part of the Southern Oxidant and Aerosol Study (SOAS). The resulting factors have a representative thermogram, which carries information on the factor volatility and unique weights for individual ions corresponding to molecular components of measured organic aerosol (OA). These properties and the diurnal patterns of factor weights are used to assign a specific source to each factor. With no a priori information used as input, the routine produces a set of factors with spectra that align well with those previously determined from several laboratory chamber experiments. The factorization routine gains relevance and separation of OA composition when using resolved thermograms as input rather than integrated thermogram time series. Of the seven factors produced by NNMF using the thermogram data, three are attributed to monoterpene-derived OA with extremely low, low, and semivolatile volatility. These three factors together represent 68% of the total organic aerosol mass examined, consistent with previous studies using a spectral basis set. Additionally, two factors were sourced to isoprene chemistry, one representing IEPOX-derived SOA, and the other relating to other oxidation products exhibiting relatively low volatility. The two isoprene-related factors account for 22% of OA mass. Notably absent is a category exclusively capturing the behavior of particulate organic nitrates (PON), which may be consistent with the relatively low local concentrations of PON observed and points to limitations of factorization to fully characterize OA. However, NNMF applied to the volatility and molecular level information from the FIGAERO HRToF-CIMS can resolve dominant precursors and chemical properties of ambient OA components with minimal assumptions.