Compositional variance in extracted particulate matter using different filter extraction techniques

Collection and subsequent extraction of particulate matter (PM) from filter substrates is a common requirement for in vivo and in vitro toxicological studies, as well as chemical analyses such as ion chromatography and inductively coupled plasma mass spectrometry. Several filter extraction protocols exist and different laboratories employ different methods, potentially biasing inter-study comparisons. Previous studies have shown significant differences in extraction efficiency between techniques and identified the relevant extraction artifacts. However, a comprehensive inter-comparison of different methods based on the chemical composition of the extracted PM has never been conducted. In the current study, an exhaustive suite of chemical analyses is performed on PM extracted from glass micro-fiber filters using techniques commonly employed in different laboratories: Multi-solvent extraction (MSE) and spin-down extraction (SDE). PM samples were collected simultaneously during field studies conducted in an urban and rural setting using a high-volume PM2.5 sampler. Results show remarkable compositional variance between the PM extracts for all chemical components analyzed, including metals, water soluble ions, polycyclic aromatic hydrocarbons, non-aromatic organics, elemental carbon and organic carbon. Mass closure was greater than 90% for MSE but deviated substantially for SDE. Detailed retrospective gravimetric analysis of archived SDE samples revealed that a process-based loss of PM mass is the root cause of the differences. These losses are shown to be compositionally biased, both externally between different PM mixtures and internally within a given PM mixture. In combination, the results of this study are the first to demonstrate (i) an exhaustive chemical characterization of a single PM extract, (ii) the significance of directly characterizing the extracted PM used in toxicological studies, (iii) the existence of substantial compositional biases between different filter extraction techniques and (iv) the importance of standardizing filter extraction objectives and procedures to avoid introducing study bias into toxicological studies.