Highlighting molecular markers of PM sources: Benefits of non-target screening strategy based on HRMS combined with multivariate statistical analyses

International audienceThe negative impact of particulate matter (PM, aerosols) on air quality, and so health, is largely recognized. The implementation of policies to reduce their concentrations in ambient air requires the knowledge of their sources. The study and evaluation of PM sources can be achieved by monitoring marker compounds or chemical signatures further used in statistical source-receptor models. However, some PM sources are still not discriminated due to the lack of specific molecular markers. High resolution mass spectrometry (HRMS) enables to obtain a more complete characterization of environmental matrices. Using HRMS data, non-target screening (NTS) approaches combined with multivariate statistical analyses can be used to compare the chemical fingerprints of samples to highlight and identify substances of interest. In this work, we evaluated the potential of a fingerprinting comparison approach to highlight specific molecular markers of PM sources currently unresolved. Two sources of PM were studied considering their significant contributions to the ambient air concentrations: biomass burning and vehicular exhaust emissions. On the former, the objective was to discriminate between residential wood burning and garden green waste burning emissions and, on the latter, the discrimination of both primary and secondary sources of PM from exhaust emissions of light-duty diesel and gasoline vehicle engines was assessed. Several experiments under simulated real-world conditions were performed to collect PM samples from both sources. Residential wood heating (fireplace and woodstove) and garden green waste burning (hedge trimming, fallen leaves) experiments were carried out in a large combustion chamber to simulate ambient air dilution conditions. Vehicular exhaust samples were collected from chassis dynamometer tests using two modern (EURO 5 standard) gasoline and diesel vehicles. Their emissions were also aged using an oxidation flow reactor to simulate atmospheric photochemical processes (associated to OH radical reactivity). Samples were then characterized using non-targeted analytical methods using HRMS coupled with both liquid and gas chromatography (LC and GC). The analytical procedures were optimized to detect as many species as possible. Chemical fingerprints obtained were compared combining several multivariate statistical analyses (e.g., partial least squares discriminant analysis (PLS-DA) and cluster analyses). Overall, about thirty chemical features were highlighted as characteristic of garden green waste burning and four of residential wood heating. Tentative identification or potential molecular formulas were proposed for the different marker candidates. Similarly, about ten chemical features specific of primary or secondary gasoline and diesel vehicle emissions have been determined. The combination of HRMS NTS chemical characterization with multivariate statistical analyses shows promise for uncovering organic aerosol fingerprinting and highlighting key molecular markers of PM sources. Such species might be used for further PM source apportionment purposes in ambient air