A synergic approach to perform source apportionment of organic aerosol using offline and online measurements in Positive Matrix Factorization

Understanding the sources and processes responsible of atmospheric PM composition and concentration is required to implement effective PM control strategies. During the last decade, the use of online non-refractory submicron aerosol mass spectrometer (AMS) and aerosol chemical speciation monitor (ACSM) measurements have successfully allowed real time measurements of organic fraction. However, to profoundly understand the sources and formation processes of organic aerosol, a comprehensive source apportionment analysis is still needed. The combination of different datasets from several measurements to refine the apportionment of OA sources, and notably secondary ones, is probably one of the best way to achieve this goal. To the best of our knowledge, this has never been performed extensively (Crippa et al. 2013; Huang et al. 2014; McGuire et al. 2014; Petit et al. 2014; Sun et al. 2012). In the present study, we propose a novel approach of combining online and offline measurements in statistical source-receptor model such as Positive matrix factorization (PMF). An intensive campaign was performed at the SIRTA atmospheric research observatory, representing the suburban background air quality conditions of the Paris area (about 25 km SW from Paris city center). PM10 samples were collected every 4 hours over a period of intensive PM pollution events (PM > 50 μg m-3 over several days) on March 6- 24 2015, concomitantly with online measurements, carried out using ACSM, 7Aethalometer, TEOMFDMS, NOx and O3 analyzers. Regular PMF was first performed on organic matrix obtained from offline measurements. Figure 1 shows the eight different factors obtained using the specific primary (i.e. levoglucosan (biomass burning), methane sulphonic acid (MSA) (marine), 1-nitropyrene (traffic)) and secondary organic molecular markers (i.e. hydroxyglutaric acid (α-pinene), 3-methyl,5- nitrocatechol (biomass burning), α-methyl glyceric acid (isoprene)). PMF was also performed on ACSM OA matrix and has been deconvolved into four factors including HOA, BBOA1, BBOA2 and OOA (Figure 1). These results show that PMF performed on individual dataset is not truly viable to procure complete information about the different atmospheric processes. Lack of high-resolution time to understand rapid atmospheric processes in case of filter measurements, and lack of specific markers to validate oxygenated factors while using online measurements, have emerged out as a critical limitation of regular PMF. Here, the synergic approach proposes to combine traditional off-line PMF factors, such as primary biomass burning, primary biogenic, secondary biogenic, traffic, with OA matrix from ACSM measurements. The unified matrix was again deconvolved with PMF in order to retrieve factors such as HOA, BBOA, OOA, and to explore additional information about OA sources. Discussion will further focus on the factors obtained by combining online and offline measurements in PMF, and their benefits over the conventional approach