A fast and automated strategy has been developed for identifying unknown substances in the atmosphere (concretely, in the particulate matter, PM10) using LC-HRMS (MS3). A total of 15 samples were collected in three different areas (rural, urban and industrial). A sampling flow rate of 30 m3 h-1 was applied for 24 h, sampling a total volume of around 720 m3. A total of 49 compounds were tentatively identified using very restrictive criteria regarding exact mass, retention time, isotopic profile and both MS2 and MS3 spectra. Pesticides, pharmaceutical active compounds, drugs, plasticizers and metabolites were the most identified compounds. To verify whether the developed methodology was suitable, 11 substances were checked with their analytical standards and all of them were confirmed. Different profiles for industrial, rural and urban areas were examined. The Principal Component Analysis (PCA) model allowed us to separate the obtained data of the three assessed area. When the profiles obtained in the three evaluated areas were compared using a Volcano plot (the rural area was taken as reference), 11 compounds were confirmed as being discriminant: three of them (3-hydroxy-2-methylpyridine, 3-methyladenine and nicotine) were more likely to be found in industrial sites; ten compounds (3-hydroxy-2-methylpyridine, 3-methyladenine, azoxystrobin, cocaine, cotinine, ethoprophos, imidacloprid, metalaxyl-M, nicotine and pyrimethanil) were more probable in the case of urban sites; finally, triisopropanolamine was more likely to be detected in rural locations.
Keywords: HRMS; PM10; confidence level; pollutants profiles; unknown analysis.