Large contribution of in-cloud production of secondary organic aerosol from biomass burning emissions

Organic aerosol constitutes a large fraction of submicron aerosol mass, and accurately apportioning its sources is critical for constraining climate radiative forcing and informing air quality policy. OA comprises primary organic aerosol (direct emissions) and secondary organic aerosol formed via oxidation and partitioning of VOCs. Multiple lines of evidence suggest SOA can contribute up to ~76% of ambient OA, yet the role of in-cloud aqueous-phase processing in SOA formation is comparatively understudied relative to gas-phase pathways. Water-soluble organic gases can dissolve into cloud or fog water and undergo reactions to yield highly oxygenated, low-volatility aqSOA. Field observations during haze and fog events (e.g., China, Po Valley) point to substantial aqSOA, often coincident with biomass burning emissions; however, the specific vapors and their contributions remain unclear. Biomass burning is the second largest global NMOG source after biogenic emissions, but the water solubility and aqSOA formation potential of complex, real biomass burning emissions have not been comprehensively quantified. Compounding this, laboratory approaches either generate very short-lived clouds (chambers) or unrealistically high liquid water contents (bulk solutions), impeding atmospherically relevant simulation of in-cloud processes. This study addresses these gaps by simulating in-cloud aqSOA production from residential wood burning using a wetted-wall flow reactor under near-atmospheric conditions, characterizing emitted gases, quantifying their solubility and uptake, and determining aqSOA composition and yields.

Prior work has significantly advanced understanding of gas-phase SOA formation, while aqueous-phase SOA has received less attention. Literature largely focuses on individual dissolved compounds such as glycolaldehyde, acetic acid, levoglucosan, phenol, vanillin, and guaiacol, with mechanistic insights into aqueous oxidation and oligomerization. Comprehensive characterizations of biomass burning NMOGs exist (e.g., PTR-MS and GCxGC-TOF studies) with extensive identification and emission factors, and updated compilations of Henry’s law constants facilitate solubility estimates. Field studies in haze and fog (Po Valley, China) have inferred important contributions from aqueous processing to ambient SOA, with one EESI-based molecular-level report during Chinese haze showing aqSOA constituting ~54% of total SOA under high-NOx and RH. However, studies of aqSOA formation from real, complex mixtures of BB-derived WSOGs are scarce, and traditional lab systems either limit cloud lifetimes or overestimate LWC, motivating development of more representative experimental platforms.

The study used a wetted-wall flow reactor (WFR) to simulate in-cloud aqSOA formation and aging from residential wood burning emissions under near-atmospheric humidity and kinetic conditions. Emissions generation: Pine and spruce logs (0.5–1.2 kg) were burned in a modern woodstove (Avant, 2009, Attika). Flue emissions were sampled via heated lines, diluted ~10× with an ejector diluter, and injected into a ~1 m³ heated stainless steel holding tank to stabilize NMOG concentrations for WFR input. Combustion conditions yielded modified combustion efficiency MCE ~0.9–0.95, indicating mainly flaming with some smoldering. WFR configuration: A rotating quartz cylinder (length 125 cm, ID 6 cm) with sandblasted inner surface maintained an ~85 µm water microlayer at ~15 rpm. Relative humidity in the reactor was 95–100% using humidified clean air (9.5–9.75 L min⁻¹). A Xe-excimer laser (172 nm, 7.2 eV) photolyzed H2O and O2 to generate OH, O3, and HO2. Temperatures were 290–295 K. OH concentrations at steady state were ~3×10⁻¹³ to 5×10⁻¹² M (aqueous) and ~3×10⁹ molecules cm⁻³ (gas). Despite higher LWC in the WFR than in clouds, the non-equilibrium system captures atmospheric-like gas–liquid partitioning and processing. Experimental sequence: Two experiment types were conducted. (1) Pure uptake (blank): with laser off (no oxidants), primary emissions were introduced, then 20 mL water was injected to form the microlayer; after 1 h, water was collected to evaluate aerosol formation upon nebulization without oxidation. (2) Oxidation experiments: primary emissions first with laser off, then laser on to generate oxidants; after ~15 min to steady gas composition, 20 mL water was injected and maintained with the laser on for 1, 3, 5, 7, or 9 h; water was collected and stored at 4 °C, then analyzed within 24 h. Instrumentation: Gas-phase composition and oxidation products were characterized using PTR-TOF-MS (Ionicon Series 8000), Vocus PTR-TOF-MS (TOFWERK/Aerodyne) for broader VOC/IVOC coverage, and a dual-phase extractive electrospray ionization TOF-MS (Dual-EESI). Ozone was monitored with a commercial analyzer. Uptake ratio for each species was calculated as (C_no-cloud − C_cloud)/C_no-cloud after specified cloud-processing durations. Aqueous-phase products were assessed via nebulization (Apex Q, 60 °C, N2) followed by drying and analysis with a high-resolution time-of-flight aerosol mass spectrometer (AMS, Aerodyne) and particle-phase EESI-TOF-MS (Particle-EESI) for molecular-level composition. Determination of solubility and partitioning: Henry’s law constants (H) were estimated for detected species using EPI Suite group and bond contribution methods, with a simplified parameterization applied across the detected formulae; modeled uptake ratios were compared to measured net uptake after cloud processing. A kinetic model (QEMRA) described partitioning behavior under WFR conditions, considering diffusion limitations and yielding thresholds for non-, moderately-, and fully soluble compounds as functions of log10H. Yield quantification: Aqueous extracts from various oxidation times were spiked with 5 ppm solutions of isotopically labeled (NH4)2SO4 and NH4NO3 for quantitative AMS analysis. The aqSOA yield was defined as integrated aqSOA mass divided by the integrated mass of all NMOGs diffusing to the wetted wall during the experiment. For comparison, gasSOA yields were estimated using an oxidation flow reactor (OFR) on the same emissions for ~5 days equivalent photochemical age. Data handling and controls: Backgrounds were established with humidified air. SVOCs in the particle phase were filtered prior to WFR, and SVOC gas-phase contributions were low. Blank experiments showed negligible aqSOA formation without OH. Cleaning protocols between experiments included triple rinses with ultrapure water and heating to 200 °C under dry air to prevent carry-over.

- Primary gas composition: Among detected species, acetic acid (C2H4O2) contributed 10.4% of total concentration, acetaldehyde (C2H4O) 7%, methyl acetate (C3H6O2) 6.1%, acetone (C3H6O) 5.8%, and furfural (C5H4O2) 5.5%. The CHO family dominated primary gases (87.4% of total mixing ratio), followed by CH (7.5%), CHN (2.6%), and CHON (2.5%). Combining Vocus and Dual-EESI detected 680 elemental formulae between m/z 40–350; IVOCs accounted for ~15.7% of total vapor concentration. CHO compounds spanned a wide volatility range with ~14% in IVOC/SVOC classes. - Solubility and partitioning: Based on parameterized Henry’s law constants, under WFR conditions, non-soluble and moderately soluble gases constituted ~51.9% and 45.0% of total mixing ratio, respectively; fully soluble comprised the remainder. Mapping to atmospheric LWC indicated that in thick clouds (LWC ~1 g m⁻³) the non-soluble fraction would be higher (~79.6%) and moderately soluble lower (~19.3%) than in WFR, implying WFR overestimates moderately soluble fractions but remains close to ambient for thick clouds. Uptake ratios increased with log10H, approaching constant values above ~5.5 M atm⁻¹, consistent with QEMRA predictions. - aqSOA composition: AMS showed strong oxygen-containing fragments (CHO+, C2H3O+, CO2+), and aqSOA had high O/C = 0.9 ± 0.1, consistent with field observations. Particle-EESI revealed rapid molecular complexity growth: detectable ions increased from 34 (pure uptake) to 307 (1 h oxidation) to 1275 (3 h), stabilizing thereafter; CHO dominated (~80% of signal). Double-bond equivalent was ~4.5–4.7 and stable after ~3 h, indicating compositional stabilization. Detected products included guaiacol-derived dimers and ring-opening products, matching lab and field observations of phenolic aqueous oxidation. Compared to dissolved gases, aqSOA molecules had higher carbon numbers (C≥8 constituted ~85% of aqSOA vs dissolved gases dominated by C≤5 monomers, 53%) and more oxygen atoms per molecule (#O 3–7 vs 1–3), providing evidence for extensive aqueous-phase oxidation and oligomerization. The top 50 aqSOA molecules accounted for ~36.7% of aqSOA signal and were mainly oxidized monomers with 4–6 O atoms. - Yields: Normalized integrated aqSOA mass increased with aqueous processing time, while composition stabilized after ~3 h. aqSOA yield (integrated aqSOA mass / integrated NMOGs to the wall) decreased from 0.43 after 1 h to 0.20 after 9 h, reflecting diminishing contributions from moderately soluble gases over time and sustained formation from fully soluble NMOGs. Gas-phase SOA yield from the same emissions in the OFR reached a maximum of ~0.17, lower than aqSOA yields. - Comparative aging: For similar equivalent processing (~5 days), aqSOA exhibited more highly oxygenated and higher-carbon-number species than gasSOA, consistent with aqueous oligomerization pathways and extensive oxygenation. - Scaling implications: Extrapolation to other wood-burning experiments indicates aqSOA contributions can be 2–7 times higher than primary BBOA, aligning with ambient haze conditions where aqSOA can exceed BBOA by factors of 2–4.

The experiments demonstrate that in-cloud aqueous processing of biomass burning vapors, driven by OH oxidation and oligomerization, is an efficient SOA pathway, yielding 0.20–0.43 under the studied conditions and surpassing gas-phase SOA yields (~0.17). The low gas-phase SOA yield implies that a majority of oxidation products remain in the gas phase; owing to increased oxygenation, these products have enhanced solubility and can partition irreversibly into cloud or fog water to form additional aqSOA upon further aqueous oxidation. This mechanism helps explain field observations of substantial aqSOA during haze and fog events in regions impacted by biomass burning. The high oxygenation (O/C ~0.9) and prevalence of high-carbon-number, low-volatility species in aqSOA underscore the importance of oligomerization in the aqueous phase, distinguishing it from gasSOA composition. The solubility-based partitioning framework (via Henry’s law and QEMRA) provides a predictive link between NMOG properties, cloud LWC, and aqSOA formation, aiding interpretation of ambient measurements and model parameterization. Regionally, enhanced aqSOA formation in clouds and fog has implications for radiative forcing and air quality, especially near emission sources and under high humidity. The results likely extend to wildfire plumes, where diverse NMOGs are abundant; increased oxidation state downwind further amplifies aqueous partitioning and aqSOA production during daytime high-RH conditions, while primary emissions may dominate precursors at night. Overall, the study provides molecular-level evidence and yields needed to better represent in-cloud aqSOA in atmospheric models.

This work provides a comprehensive, atmospherically relevant simulation of in-cloud aqSOA formation from residential wood burning emissions using a wetted-wall flow reactor. It quantifies the solubility distribution of emitted gases, links uptake to Henry’s law behavior, and shows that aqueous OH oxidation and oligomerization produce highly oxygenated, higher-carbon-number SOA with yields (0.20–0.43) exceeding gas-phase SOA yields. The findings indicate that in-cloud chemistry is a major pathway for transforming biomass burning vapors into SOA, with significant implications for climate and air quality. Future research should: (1) extend solubility and yield characterization to wildfire emissions and broader BB fuel types; (2) investigate effects of environmental variables (UV spectrum including UVB, pH, NOx, temperature, LWC) on aqSOA formation and composition; (3) reconcile laboratory conditions with ambient fog/haze/cloud variability; and (4) collect additional field molecular-level aqSOA data to validate laboratory-derived mechanisms and parameterizations.

- WFR water film thickness and LWC are higher than typical atmospheric clouds, leading to overestimation of moderately soluble partitioning relative to thick-cloud conditions; a thinner water film would better represent cloud uptake but precluded yield and composition quantification with current setup. - Nebulization and aerosol drying may induce some oligomer formation, potentially biasing oligomer contributions upward, though such drying also occurs in the atmosphere. - Experimental conditions (pH 4–5, RH ~100%, temperature 290–295 K, OH levels) may not capture the full range of ambient variability, including effects of NOx, UVB, and temperature; additional pathways may operate in the field. - SVOCs in the particle phase were filtered out and not considered; low gas-phase SVOCs were present but may still contribute under some conditions. - Partitioning analysis assumes parameterized Henry’s law constants with minor deviations from ideality; uncertainties in H estimation and diffusion limitations in the WFR can affect inferred soluble fractions. - Cloud chamber analog limitations remain: although non-equilibrium conditions better reflect atmospheric processes than bulk solutions, cloud lifetime, microphysics, and LWC differ from natural clouds and fog.