Insights from an Integrated LC–MS/MS and Machine Learning Analysis into the Role of Long-Chain CHO-Containing Compounds in Ambient Light-Absorbing Aerosols

Fossil fuel combustion, biomass burning, and cooking are major emission sources of urban organic aerosols, which have a significant impact on human health and global radiative forcing. This study collected ambient PM2.5 in Shanghai, China, along with particulate matter from biodiesel/diesel-fueled vehicles, bio-oil pyrolysis, and laboratory-scale experiments conducted using a Miniature Combustion Aerosol Standard (lab-MINICAST) experiments. Targeted MS/MS using HPLC-Q-TOF-MS methodology and machine learning were applied to investigate and decode the light-absorbing chromophores of methanol-extracted organic aerosols. The results showed emissions from biomass burning and vehicles are the main contributors to ambient organic aerosols. CHO compounds accounted for the largest proportion. A fraction of the identified light-absorbing chromophores were long-chain CHO-containing compounds, including fatty acids (C18H34O2, C18H32O2, C16H30O2, C5H8O2, and a subset of CnH2nO2 (12 < n < 25)) and fatty acid methyl esters (FAMEs, the remaining CnH2nO2 (12 < n < 25)) originatimg from vehicle emissions, accounting for 65.256% of the organic aerosol light-absorbing during nonbiomass burning days, but only accounting for 7.12% of the CHO compounds and 3.06% of the organic aerosols. Long Short-Term Memory (LSTM) models revealed that all the light-absorbing molecules of the methanol-extracted samples identified in this study typically exhibit higher weights of atomic around oxygen-containing functional groups (>0.5), contrasting with non-light-absorbing substances, which display more uniformly distributed lower atomic weights. This may provide some explanation for the light-absorbing ability of fatty acids and FAMEs.