Linking Precursors and Volatility of Ambient Oxygenated Organic Aerosols Using Thermal Desorption Measurement and Machine Learning

We conducted thermal desorption measurements and machine learning analysis to investigate the volatility and precursors of ambient oxygenated organic aerosols (OOA), with a focus on the link between them, in a variety of urban and marine locations. We found that the OOA species measured by an iodide-based Chemical Ionization Mass Spectrometer equipped with a Filter Inlet for Gases and AEROsol (FIGAERO-CIMS) accounted for 16 ± 13% of OA in those urban and marine locations and represented mostly the secondary and moderate-volatility portion of ambient OA. On average, 25.1% in species number and 26.8% in mass of the OOA species detected by the FIGAERO-CIMS in a winter campaign at an urban site in Wuhan, a megacity in central China, might be attributed to thermal decomposition fragments. Our results show that the volatility and precursor of ambient OOA differed systematically according to location, season, and pollution level. The OOA in the ocean atmosphere was characterized by high fractions of extremely low volatility organic compounds (ELVOC) and aliphatic species, while the inland urban OOA was characterized by aromatic species and fell primarily into the low volatility organic compounds (LVOCs) and semivolatile organic compounds (SVOCs) range. The volatilities of OOA in the inland urban atmosphere in summer, pollution days, and daytime were lower than those in winter, clean days, and nighttime. When the PM episode developed from clean to particle growth and then to pollution period, the OOA species shifted from Low-Mw OOA species to Median-Mw OOA species and then to highly nonvolatile species. The study of ambient OOA volatility in this work also provides important data for future closure studies of SOA formation and its precursors.