Identification of significant precursor gases of secondary organic aerosols from residential wood combustion

Organic gases undergoing conversion to form secondary organic aerosol (SOA) during atmospheric aging are largely unidentified, particularly in regions influenced by anthropogenic emissions. SOA dominates the atmospheric organic aerosol burden and this knowledge gap contributes to uncertainties in aerosol effects on climate and human health. Here we characterize primary and aged emissions from residential wood combustion using high resolution mass spectrometry to identify SOA precursors. We determine that SOA precursors traditionally included in models account for only ~3–27% of the observed SOA, whereas for the first time we explain ~84–116% of the SOA by inclusion of non-traditional precursors. Although hundreds of organic gases are emitted during wood combustion, SOA is dominated by the aging products of only 22 compounds. In some cases, oxidation products of phenol, naphthalene and benzene alone comprise up to ~80% of the observed SOA. Identifying the main precursors responsible for SOA formation enables improved model parameterizations and SOA mitigation strategies in regions impacted by residential wood combustion, more productive targets for ambient monitoring programs and future laboratories studies, and links between direct emissions and SOA impacts on climate and health in these regions.

Liquide) 2 into the chamber begins and the contents of the chamber are irradiated with UV light for 4-6.5 h.
Particles are dried prior to analysis by AMS (Nafion membrane, Perma Pure LLC) and AMS data are analyzed in Igor Pro 6.3 (WaveMetrics, Inc.) using the SQUIRREL (version 1.53F) and PIKA (version 1.12F) analysis programs. An AMS collection efficiency of 1 is applied based on previous biomass burning studies [3][4][5] . Equivalent black carbon (eBC) is quantified using a 7 wavelength Aethalometer (2 l min -1 , AE33, Magee Scientific Company) 6 . Particle wall loss rates in the chamber are determined using the decay of eBC at the end of each experiment assuming all particles are lost equally to the walls and that condensable material partitions only to suspended particles 7 . All literature SOA yields, except the yields from Hildebrandt et al. 8 for toluene, are determined using the decay of an inert seed aerosol and with no consideration of partitioning of condensable vapors to the walls, as done in the current study. The average particle half-life in the chamber is 3.4±0.7 h. NMOG wall losses are inferred by monitoring NMOG concentrations prior to initiating photochemistry and by assessing the smog chamber conditions affecting loss rates 9 . Measurements of NMOGs in the chamber prior to aging are stable, indicating that the chamber walls are not a sink for NMOGs, but rather that NMOGs are in equilibrium with the chamber walls, particles and the gas phase. Zhang et al. 9 show that the bias created by the wall loss is inversely proportional to seed aerosol concentration and OH concentration, both of which were relatively high in the current experiments. Seed aerosol concentrations are given in Table 1 and OH concentrations during the experiments were   S4   ~1.4x10 7 molec cm -3 . Under these experimental conditions, NMOG wall losses are not expected   to be large and thus no corrections were applied. Gas-phase species are sampled from the chamber through a Teflon sample line heated to 323 K after exiting the temperature-controlled chamber housing. A filter (Tissuquartz, Pall Corporation) upstream of the inlets prevents particles from entering the instruments monitoring gas-phase species. NMOGs are characterized using a high resolution proton transfer reaction time-of-flight mass spectrometer (PTR-ToF-MS 8000, Ionicon Analytik G.m.b.H.), operating with hydronium ions as the reagent and with a drift tube pressure of 2.2 mbar, voltage of 543 V and temperature of 90°C. The ratio of the electric field (E) and the density of the buffer gas (N) in the drift tube, which dictates the ion drift velocity in the drift tube, is 137 Td. The PTR-ToF-MS transmission function is determined using six NMOGs in a gas standard (methanol, acetaldehyde, propan-2-one, toluene, p-xylene, 1,3,5-trimethylbenzene; Carbagas). PTR-ToF-MS data are analyzed in Igor Pro 6.3 (WaveMetrics, Inc.) using the Tofware analysis platform (version 2.4.5, Tofwerk). Mass spectral data from m/z 33 to m/z 130 are fit, as well as the 18 O isotope of the reagent ion. Peak widths and possible molecular formulas increase with increasing m/z, making accurate peak assignments difficult in the higher m/z range. However, signal above m/z 130 corresponding to compounds previously identified during residential wood combustion are fit [10][11][12][13][14][15][16] . Isotopic contributions are constrained during peak fitting and accounted for when determining parent peak concentrations. The minimum detection limit is taken as three standard deviations above the background, where the standard deviation is determined from the background measurements of each ion in the chamber prior to emission injection. There is a small continuous dilution in the chamber during aging due to the constant nitrous acid injection and NMOG time traces are corrected for this dilution using CO as an inert tracer.

S5
Assigning structures to ions detected using the PTR-ToF-MS is critical for understanding the conversion of NMOGs to SOA during aging. Structural assignments are guided by previously identified compounds emitted during residential wood combustion [10][11][12][13][14][15][16] . The contribution of an individual NMOG to the measured SOA is determined using the temporal evolution of the compound during aging and published SOA yields 8,[17][18][19][20][21][22][23][24] , which are available for 18 of the 59 identified compounds. SOA yields can depend on experimental conditions, including the ratio of NO x to NMOG, presence of seed aerosol and total organic aerosol mass 8,18,19,22,25 . When a range of yields are available, the average of literature values from experiments most similar to the current study in terms of seed aerosol and NO x /NMOG is applied. The average NO x /NMOG for the set of experiments is ~150 ppb ppmC -1 and individual values ranges from ~35-350 ppb ppmC -1 . In addition to the 18 species with SOA yields in the literature, isomers of 2,4-/2,5dimethylfuran, styrene, benzaldehyde and isomers of 4-(2-hydroxyethyl)phenol/2-methoxy-4methylphenol are identified as likely contributors to SOA due to relatively high concentrations and structural similarities to the compounds for which SOA yields are known. The SOA yields for these four compounds are taken as the average of the published SOA yields for the NMOGs with at least 6 carbon atoms per molecule (≥C 6 ). This estimated SOA yield is also applied to the sum of compounds with lower relative concentrations and at least six carbon atoms per molecule (structurally assigned ≥C 6 compounds), as well as signal above m/z 130 which is not fit, but is expected to be due to compounds with at least six carbon atoms per molecule and could contribute to SOA formation (structurally unassigned ≥C 6 compounds).
Uncertainties arise from the inability to resolve isomers using the PTR-ToF-MS. For example, SOA yields are available for both 2-methylprop-2-enal and (2E)-2-butenal and the applied yield is taken as an average of the values from the two isomers. SOA yields are relatively low for both S6 of these compounds and the impact on results is negligible. In cases of possible isomeric contributions where a SOA yield is available for only one isomer, the approach assumes that either 1) the isomers have the same SOA yield or 2) the signal is entirely due to the compound with the known SOA yield. The possibility of isomers is not expected to influence the results considerably, however, as the majority of compounds are not suspected to have significant isomer contributions (Table S2) and the compounds that contribute the most to the SOA have no previously detected isomers in residential wood combustion emissions.
The reaction rate constant of each species with the reagent in the drift tube is needed to convert the raw PTR-ToF-MS signal to concentration. The relatively limited availability of applicable measured reaction rate constants precludes assignment of a reaction rate constant to each ion.
When available, individual reaction rate constants are applied 26 and a default reaction rate constant of 2×10 -9 cm 3 s -1 is applied to all other ions. In cases where ions could correspond to several isomers, the reaction rate constant is taken as the average of available values. Reaction rate constants varying from the default for species of interest for SOA formation are presented in Table S3.

Monoterpene emissions
Monoterpenes have been previously measured in residential wood combustion 10,12 , but are below the detection limit in all experiments, which is likely largely due to the type of wood burned.
Previously reported emission factors (mg kg -1 should be read as mg emitted species per kg combusted fuel) for wood stove burning of beech wood are below the detection limit for 3carene and limonene and very low for α-pinene (0.506 mg kg -1 ) 12 . α-Pinene emissions of this magnitude would contribute less than 0.2% to SOA in the current study. The emission factors S7 for other compounds reported by Evtyugina et al. 12 for wood stove burning of beech wood, except naphthalene which was below the detection limit, are within a factor of ~1-3 of those in the current study, which is reasonable considering the large impact of burn parameters on emissions 27 . Although it is reasonable that monoterpenes are not detected based on previous findings 12 , these species can be emitted in much larger quantities during the burning of other wood types, particularly softwoods 10 which can emit over ten times more terpenes than during comparable burning of hardwoods, and should be considered in models in addition to the 22 individually identified NMOGs.  n/a n/a n/a n/a n/a 0.32 d styrene n/a n/a n/a n/a n/a 0.32 d benzaldehyde n/a n/a n/a n/a n/a 0.32 d 4-(2-hydroxyethyl)phenol/2-methoxy-4methylphenol n/a n/a n/a n/a n/a 0.32 d structurally assigned ≥C6 compounds n/a n/a n/a n/a n/a 0.32 d structurally unassigned ≥C6 compounds n/a n/a n/a n/a n/a 0.