Source apportionment of organic gaseous and particulate compounds using a combined positive matrix factorization approach in summer (2020) in the Paris region (France)

Gaseous and particulate organic compounds are key components of atmospheric chemistry and better understanding their composition, sources and processes is essential to limit their impacts. Source apportionment using positive matrix factorization (PMF) is customarily performed for such studies. Combining organic aerosol data with their gaseous precursors was shown to be a promising approach, however very rarely attempted so far, to refine their origins using PMF. In this study, co-located continuous proton-transfer-reaction mass spectrometer (PTR-MS) and aerosol chemical speciation monitor (ACSM) measurements were performed at the suburban SIRTA station located in the Paris region. A combined dataset using both instruments during summer (June–August 2020) was then used in an exploratory PMF analysis to investigate the sources and processes of organic compounds, particularly the influences of biogenic emissions and important photochemical reactions on the formation of secondary organic aerosol (SOA) in this period of the year. Specific parameters and evaluation procedures were needed to ensure an equivalent representation of both instruments in the PMF model. A weighing factor was applied to the PTR-MS uncertainties which was controlled and optimized based on the analysis of the modelled scaled residuals. Seven main factors were obtained, describing anthropogenic sources (hydrocarbon-like organics, cooking-like organics), primary biogenic volatile organics, nighttime VOC, oxidized organics, aged organics and a specific isoprene oxidation factor which contributed 11 % to VOC and 4 % to OA. Compared to single-instrument PMFs, more factors were obtained, notably including the cooking-like factor which is not usually resolved using only ACSM data. This method also showed a better mathematical performance for the PTR-MS variables (mean absolute scaled residuals lower for the combined PMF (11.2) than for the PTR-MS-only PMF (38.2)) and a better separation of the factors for the ACSM variables in the combined PMF.