Evaluation of sequential solvent and thermal extraction followed by analytical pyrolysis for chemical characterization of carbonaceous particulate matter

Abstract

A stepwise analytical protocol is presented for recognition of uncharacterized organic chemical entities comprising carbonaceous air particulate matter (PM). In this method, thermal treatment of PM samples starts with thermal extraction (TE) using temperatures below 300°C to evolve volatile high molecular weight (LMW) chemicals followed by pyrolysis (Py) of non-volatile high molecular weight (HMW) compounds to yield signature patterns of their decomposition products. Both volatile species and pyrolyzed products are characterized using gas chromatography (GC) with a mass spectrometric (MS) detection. The efficacy of sequential TE (at 200 and 300°C) and Py (at 400, 500, 600, 700, 800, 900 and 1000°C) was demonstrated on a defined mixture of LMW species and non-volatile polystyrene, a model HMW compound. When a solid matrix, in a form of silica or graphite particles, was introduced to this system, we determined that a solvent extraction (SE) applied prior to a sequential TE/Py-GC/MS sample analysis was an essential step to eliminate the analyte–matrix interactions, which hindered the efficient vaporization of LMW compounds from a solid matrix resulting in a poor TE/Py separation. This SE/TE/Py-GC/MS method was then successfully applied to PM samples representing real-world PM matrices to discover several novel features of PM composition. HMW species were observed in both water-soluble (extracts) and unextractable (solid residues after an exhaustive solvent extraction) fractions of wood smoke and urban PM. The species evolving upon pyrolysis were associated with the corresponding precursor polymers, such as oxidized fragments of lignin, polysaccharides and lipids, in wood smoke PM and non-volatile anthropogenic hydrocarbons in urban PM. The proposed sequential SE/TE/Py-GC/MS protocol is capable of revealing otherwise unobserved chemical components while providing useful information on their interactions within a complex and poorly characterized PM matrix. Signature SE/TE/Py profiles, i.e., PM “fingerprints,” provided by our approach may prove useful in source apportionment studies.