Chemical Composition of PM<sub>2.5</sub> in October 2017 Northern California Wildfire Plumes

Abstract

Abstract. Wildfires have become more common and intense in the western US over recent decades due to a combination of historical land management and warming climate. Emissions from large scale fires now frequently affect populated regions such as the San Francisco Bay Area during the fall wildfire season, with documented impacts of the resulting particulate matter on human health. Health impacts of exposure to wildfire emissions depend on the chemical composition of particulate matter, but the molecular composition of the real biomass burning organic aerosol (BBOA) that reaches large population centers remains insufficiently characterized. We took PM2.5 (particles having aerodynamic diameters less than or equal to 2.5 μm) samples at the University of California, Berkeley campus (~60 km downwind of the fires) during the October 2017 Northern California wildfires period, and analyzed molecular composition of OA using a two-dimensional gas-chromatography coupled with high resolution time-of-flight mass spectrometer (GC×GC ToFMS). Sugar-like compounds were the most abundant component of BBOA, followed by mono-carboxylic acids, aromatic compounds, other oxygenated compounds and terpenoids. The vast majority of compounds detected in smoke have unknown health impacts. Regression models were trained to predict the saturation vapor pressure and averaged carbon oxidation state of compounds. The compounds speciated have a wide volatility distribution and most of them are highly oxygenated. In addition, time series of primary BBOA tracers observed in Berkeley were found to be indicative of the types of plants in the ecosystems burned in Napa and Sonoma, and could be used to differentiate the regions from which the smoke must have originated. Commonly used secondary BBOA markers like 4-nitrocatechol were enhanced when plumes aged, but their very fast formation caused them to have similar temporal variation as primary BBOA tracers. Using hierarchical clustering analysis, we classified compounds into 7 factors indicative of their sources and transformation processes, identifying a unique daytime secondary BBOA factor. Chemicals associated with this factor include multifunctional acids and oxygenated aromatic compounds. These compounds have high average carbon oxidation state, and are also semivolatile. We observed no net particle-phase organic carbon formation, which indicates an approximate balance between the mass of evaporated primary and secondary organic carbonaceous compounds to the addition of secondary organic carbonaceous compounds.

Ch3MS-RF: a random forest model for chemical characterization and improved quantification of unidentified atmospheric organics detected by chromatography–mass spectrometry techniques

“…The test set compounds were quantified using two alternative quantification methods, i.e., manual or closest proxy quantification (described in Liang et al, 2021, which utilizes a combination of both), to benchmark random forest model performance. Manual proxy quantification entails manually assigning a compound to a chemically similar external standard based on researcher judgment on what chemical class the unidentified compound would likely belong to, based on some combination of location in GCxGC space and mass spectrum.…”

“…That said, these instrumental configurations are rare, and fragmentation under 14 eV is still sufficiently significant to leave the formulae of many species not identifiable and therefore still uncharacterized (Worton et al, 2017). Recent efforts to embrace a larger fraction of the full complexity of chemical information yielded by highly spe-ciated organic aerosol measurements (on the scale of lowto mid-hundreds of compounds) have categorized unidentified species by likely source groups and chemical families through time series correlations with known tracer species (Zhang et al, 2018) or by manual group assignments by individual researcher judgments based on mass spectral features (Liang et al, 2021). These methods are difficult to standardize and reproduce and become prohibitively inefficient when pushing towards the full chemical complexity of speciated observations produced from typical atmospheric samples, which extend into the low-to mid-thousands of species.…”

“…Where possible, compounds in GC and GCxGC-MS are directly quantified by the calibration curves of authentic standards, but direct quantifications are limited by standard expense and availability, even for species that can be positively identified. Compounds that cannot be directly quantified, both in GCxGC-MS and in GC-MS, are most commonly quantified by assigning quantification factors from compounds resolved closely in chromatographic space, compounds that are identified as sharing chemical structures, or some interpolation of multiple nearby proxies (Hatch et al, 2015;Jen et al, 2019;Liang et al, 2021;Zhang et al, 2018). The errors associated with these assignments/choices are usually estimated from the range of quantification factors of close or chemically similar species and are assumed to be high (up to a factor of 2, depending on the degree of certainty in assigning chemical class, as described in Jen et al, 2019 andLiang et al, 2021).…”

“…Compounds that cannot be directly quantified, both in GCxGC-MS and in GC-MS, are most commonly quantified by assigning quantification factors from compounds resolved closely in chromatographic space, compounds that are identified as sharing chemical structures, or some interpolation of multiple nearby proxies (Hatch et al, 2015;Jen et al, 2019;Liang et al, 2021;Zhang et al, 2018). The errors associated with these assignments/choices are usually estimated from the range of quantification factors of close or chemically similar species and are assumed to be high (up to a factor of 2, depending on the degree of certainty in assigning chemical class, as described in Jen et al, 2019 andLiang et al, 2021). To our knowledge, this work presents the first quantitative error analysis of these techniques based on applying proxy quantification techniques to compounds with known quantification factors.…”

Emissions of organic compounds from western US wildfires and their near-fire transformations

“…The filters were stored at −20 • C prior to analysis. Filters were analyzed using an offline GC × GC coupled to an electron impact/vacuum ultraviolet light ionization source and a high-resolution time-of-flight mass spectrometer (GC × GC EI/VUV HRToFMS), following the same protocols as and Liang et al (2021). Small punches of each filter (with added isotopically labeled internal standards) were thermally desorbed at 320 • C in helium flow using a Gerstel Thermal Desorption System.…”