Insight into wet scavenging effects on sulfur and nitrogen containing organic compounds in urban Beijing

Winter haze pollution driven by submicron aerosol (PM1) remains a major challenge for air quality in the North China Plain, with organic aerosols (OA) typically accounting for 31–59% of PM1 mass during winter haze in Beijing. Although secondary organic aerosol (SOA) comprises a large fraction (about 43–78%) of OA in Asia, its molecular-level composition, sources, and ageing processes—especially for sulfur-containing organic compounds (SOCs) and nitrogen-containing organic compounds (NOCs)—are not well constrained. SOA forms via oxidation of volatile organic compounds (VOCs) and through aqueous-phase processing on wet particles and droplets. Photochemical processing often yields less-oxidized OOA (LO-OOA), while aqueous-phase processing tends to form more-oxidized OOA (MO-OOA), with interplay between these pathways. SOCs and NOCs are ubiquitous and diverse OA constituents and useful tracers for SOA formation, yet their atmospheric formation mechanisms and evolution remain insufficiently understood at the molecular level. Precipitation can remove gases and aerosols via wet scavenging, potentially suppressing formation of SOCs and NOCs. This study asks how wet scavenging modulates the formation, composition, and evolution of OA—particularly SOCs and NOCs—in urban Beijing. The objective is to characterize OA composition and sources and to diagnose formation pathways of SOCs and NOCs before and after rain and snow events during autumn–winter conditions.

Prior work shows OA dominates submicron aerosol in Beijing winter haze and that SOA formation proceeds via multiple pathways, including photochemical and aqueous-phase processes. Photochemistry is linked to LO-OOA formation, while aqueous-phase processes commonly yield MO-OOA, although LO-OOA can evolve into MO-OOA and photochemical aqueous-phase reactions also occur. SOCs identified in Beijing include glycolic acid sulfate (GAS) formed via glyoxal uptake on sulfate aerosols, along with organosulfates from anthropogenic aromatics/PAHs and biogenic monoterpenes/isoprene via OH, NO3, and O3 oxidation. Previous studies have reported substantial organosulfate burdens in urban areas. NOCs have been observed in gas and particle phases; FIGAERO/ToF-CIMS studies report particulate organic nitrogen, including nitrophenols and monoterpene-derived NOCs, and chamber studies indicate nitrophenol formation from aromatic + NOx chemistry. Organonitrates can form from alkane/olefin oxidation by OH/NO3 in the presence of NOx (day) and via NO3 + olefins (night). Despite these findings, molecular-level field characterization of SOCs and NOCs under real atmospheric conditions, and how precipitation modifies their formation, remains limited, motivating the present real-time, high-resolution investigation across wet scavenging episodes.

Field observations were conducted on the rooftop of the Environmental Technology Building at the Research Center for Eco-Environmental Sciences (RCEES; 40.0°N, 116.3°E), Chinese Academy of Sciences, in downtown Beijing, from 2021-10-01 to 2021-11-12. The site is surrounded by dense residential areas and heavy traffic. Non-refractory PM1 (NR-PM1: organics, sulfate, nitrate, ammonium, chloride) was measured with an Aerodyne high-resolution time-of-flight aerosol mass spectrometer (HR-TOF-AMS). Data were analyzed using PIKA (v1.25G) in Igor Pro. A collection efficiency (CE) of 0.5 was applied based on aerosol composition. Relative ionization efficiencies (RIE) used: nitrate 1.1, sulfate 1.2, chloride 1.3, organics 1.4; ammonium RIE = 4.0 (determined with NH4NO3). PM2.5 was obtained from a nearby (∼5 km) state control station (Olympic Center). VOCs (isoprene, benzene, toluene, naphthalene, monoterpene) were measured by PTR-TOF-MS. Gases (SO2, NO2/NOx, CO, O3) were monitored with Thermo Scientific analyzers (models 43i, 42i-TL, 48i, 49i). Formaldehyde (HCHO) was measured by a Picarro G2307; NH3 by QCTILDAS. Meteorological parameters (RH, temperature, wind speed/direction) were measured with a Vaisala weather station. UV radiation (290–400 nm) was measured with a CUV3 radiometer at 1-min intervals. Air mass transport before/after precipitation was assessed with 48-h HYSPLIT back trajectories (version 5) using GDAS, initialized four times daily (00, 06, 12, 18 UTC) at 50 m AGL. OA source apportionment: High-resolution AMS spectra were analyzed by PMF (PET v2.06). Five OA factors were resolved: two secondary OA (OOA and aqueous-related OOA, aq-OOA) and three primary OA (hydrocarbon-like HOA, cooking-related COA, biomass burning BBOA), supported by diagnostic tracers and spectra. Elemental ratios: H/C, O/C, OA/OC were computed via the Improved-Ambient method; N/C and S/C via the Aiken-Ambient method. Aerosol liquid water content (ALWC) was estimated as the sum of inorganic water (Wi) from ISORROPIA II (forward mode, using NH3) and organic water (Wo) calculated via a simplified k-Köhler approach. Estimation of SOCs: Following Song et al., SOCs mass was estimated assuming SOCs yield SO2+ and SO2−4 fragments in AMS. Observed SO2 and SO4 were combined with molar masses and reference ratios (from clean, dry periods: NR-PM2.5 < 2 µg m−3 and RH < 20%) to compute SOCs. Estimation of NOCs: NOCs mass (including amines, amides, amino acids, organonitrates) was estimated from OA/OC and N/C (NOCs = OA/OC × N/C). Particulate organonitrates mass fraction and mass were further derived from NO2+/NO+ ratios using calibration and field-determined parameters. Formation pathway proxies: Photochemical activity was represented by O3, Ox (= O3 + NO2), and UV[O3]; nocturnal NO3 radical by [NO2][O3]. Aqueous-phase SOC formation proxies used products of ALWC, glyoxal-related fragments (CH2O2+, C2H2O2+, C2O2+), and sulfate (SO4 2−). Correlation analyses (Pearson r, r2 with significance tests) were used to diagnose relationships between SOCs/NOCs and precursors/conditions across periods (haze/clean; before/after rain and snow; day/night).

• Two pronounced wet scavenging events (rain ~9 h; snow ~20 h) dramatically reduced NR-PM1 and OA. Average NR-PM1 before/after: rain 26.5 ± 8.2 → 2.5 ± 1.1 µg m−3; snow 54.5 ± 13.4 → 2.3 ± 1.6 µg m−3; wet scavenging efficiency >90% for NR-PM1. In the snow episode, scavenging efficiencies were 96% (NR-PM1), 91% (OA), 99% (SOCs), and 95% (NOCs).
• SOCs and NOCs decreased strongly after precipitation: SOCs averages before/after rain 0.24 ± 0.14 → below detection; before/after snow 0.78 ± 0.25 → 0.01 ± 0.01 µg m−3. NOCs averages before/after rain 0.22 ± 0.06 → 0.03 ± 0.01 µg m−3; before/after snow 0.51 ± 0.10 → 0.03 ± 0.002 µg m−3. SOCs scavenging ≈ 99%; NOCs 85–95%.
• Over the haze period, NR-PM ranged 0.9–96.6 µg m−3 (means: haze 46.9 ± 18.3; clean 6.9 ± 7.4; overall 19.2 ± 21.8 µg m−3). Estimated SOCs in haze were 0.01–1.45 µg m−3 (up to 10% of OA and 14% of total particulate sulfur), while NOCs were 0.07–0.80 µg m−3 (up to 4% of OA and 10% of total particulate nitrogen).
• OA fraction in NR-PM1 was 29% (haze) and 52% (clean), reaching up to 68% after rain, indicating stronger removal of inorganics and their precursors by precipitation.
• OA composition shifted: prior to precipitation, SOA dominated (aq-OOA and OOA; before rain 20% and 28%; before snow 44% and 20%), reflecting higher oxidation and aqueous processing. After precipitation, POA fractions increased (after rain COA 31%, HOA 11%; after snow COA 33%, HOA 21%) while BBOA decreased, indicating dominance of stable local primary sources and reduced ALWC conditions for SOA formation.
• Alternating dominance: SOCs proportion exceeded NOCs during haze, but after wet scavenging NOCs proportion exceeded SOCs. This reflects stronger removal of SOCs and more favorable post-precipitation photochemical conditions (higher UV, O3/NOx) for NOCs formation.
• SOCs formation: SOCs increased with RH; diagnostic fragment ratios indicated SOC presence and aqueous formation. Glycolic acid sulfate (GAS) was the dominant SOC throughout haze, supported by strong correlations with glyoxal-related fragments (e.g., CH2O2: r2 = 0.66; C2H2O2+: r = 0.84; C2O2+: r = 0.76). Aqueous-phase GAS proxies ALWC × [glyoxal fragments] × [SO4 2−] correlated with SOCs before rain (r = 0.37–0.76) and before snow (r = 0.51–0.98) during both day and night; after snow, correlations were strong at night (r = 0.53–0.93), implying nighttime accumulation. Additional photochemical SOCs were indicated before snowfall via correlations with aromatics on Nov 3 (benzene r2 = 0.91; toluene r = 0.93; naphthalene r = 0.93) and with monoterpene on Nov 4 (r = 0.82). Photochemical proxies (Ox, UV[O3]) contributed especially before snow; NO3 radical chemistry contributed little to SOCs in both episodes.
• NOCs speciation and formation: N/C increased when RH > 60%, with NOCs composition ranked amino acids > amides > amines > organonitrates. Amines increased with RH; amino acids, amides, and organonitrates peaked at RH 40–50% and increased again for RH > 60%. High oxidized NOCs (amino acids, amides) dominated during haze, indicating secondary formation consistent with aq-OOA behavior.
• NOCs correlations: Haze daytime NOCs correlated with amide fragment CH2NO+ (r = 0.85). Before rain: daytime NOCs correlated with tertiary amine C2H7N+ (r = 0.65); nighttime with amide C4H8NO+ (r = 0.70) and with VOCs (isoprene, benzene, toluene, naphthalene; r = 0.96–0.99). Before snow: daytime correlations with CH2NO and amino acid C2H5NO2 (r = 0.84, 0.81); strong daytime correlations on Nov 3 with amides (CH2NO+ r2 = 0.95; C2H4NO+ r2 = 0.95; C2H3NO+ r2 = 0.92) and isoprene (r = 0.94), indicating photochemical processing.
• Oxidant control for NOCs: Before rain, NOCs correlated with O3 (r = 0.40), while after rain with NO3 proxy [NO2][O3] (r = 0.62). Before snow, NOCs correlated with O3 during daytime (r = 0.63–0.87) and with NO3 proxy at night (r = 0.63–0.74); after snow, with UV[O3] during daytime (r2 = 0.59–0.68). O3 played a crucial role in NOCs formation via daytime photochemistry and nighttime NO3 chemistry.
• Air mass transport influenced pollutant loadings: During rain, ~50% of air masses originated from NW Mongolia, with additional short-range transport from E/NE. Before snow, 78% were local; after snow, air masses were predominantly from cleaner NW long-range transport, consistent with NR-PM1 evolution.

The study demonstrates that wet scavenging not only removes particulate matter and gaseous precursors but also reshapes OA composition and the dominant chemical formation pathways. Before precipitation, elevated ALWC and stagnant/local conditions favored aqueous-phase processing, yielding more oxidized OA (aq-OOA) and abundant SOCs dominated by glycolic acid sulfate from glyoxal–sulfate aqueous chemistry, with additional photochemical SOCs from aromatics and monoterpenes under high O3/UV conditions prior to snowfall. After precipitation, strong removal of inorganics and SOCs, coupled with reduced ALWC, suppressed aqueous SOC formation. In contrast, post-precipitation photochemical environments with higher UV and available O3/NOx supported formation of highly oxidized NOCs (amides, amino acids), leading to a compositional shift where NOCs became more prominent relative to SOCs. The observed correlations with oxidants (O3, UV[O3]) and NO3 proxies across diurnal periods confirm oxidant-driven NOCs formation via gas-phase production and gas–particle partitioning or further oxidation. These findings address the research question by detailing how precipitation modulates the formation mechanisms and relative abundances of SOCs and NOCs, improving mechanistic understanding of OA evolution in urban winter haze. The results are relevant for air quality management, as they highlight the dependence of SOCs on aqueous conditions and of NOCs on photochemical regimes, implying that meteorology and oxidant control can shift OA composition and properties even when emissions remain similar.

This work provides high-resolution, real-time evidence that wet scavenging events in urban Beijing drive a transition in OA composition from SOC-dominated before precipitation to NOC-dominated afterward. SOCs were largely governed by aqueous-phase formation of glycolic acid sulfate under high ALWC, with contributions from photochemical reactions of aromatics and monoterpenes prior to snowfall. NOCs were dominated by highly oxidized amides and amino acids formed primarily via photochemical pathways influenced by O3 (day) and NO3 (night), with enhanced NOCs formation after precipitation due to stronger photochemical conditions. The study clarifies the meteorology–chemistry coupling that controls SOCs and NOCs formation pathways and OA evolution across precipitation events. Future research should: (1) expand molecular-level speciation of SOCs/NOCs using complementary techniques (e.g., offline HRMS) to validate AMS-based estimates; (2) investigate seasonal variability and multi-year representativeness; (3) quantify kinetics and yields of key aqueous and gas-phase pathways under varying particle acidity and ALWC; (4) assess the role of nocturnal chemistry and multiphase radical sources; and (5) integrate observations with process models to predict OA compositional shifts under changing meteorology and emission controls.

• Single urban site and a limited autumn–early winter period may constrain spatial and temporal representativeness; results may not generalize to other seasons or regions without caution.
• SOCs estimation relies on AMS fragment-based assumptions (e.g., SOCs producing SO+ and SO2+ only) and clean/dry reference ratios, which introduce uncertainty and may bias absolute mass estimates.
• NOCs mass estimation using OA/OC and N/C ratios aggregates diverse N-containing species; individual NOC classes are inferred from characteristic fragments with potential overlap and interferences.
• HR-TOF-AMS is less specific for certain organosulfur and organonitrogen compounds; species-level attribution (e.g., distinguishing HMS, MSA, sulfones) is limited, especially at low concentrations.
• Wet scavenging periods and air mass changes co-occur, making it difficult to fully isolate scavenging from transport effects on observed concentrations.
• Lower HCHO levels during the study compared with prior winters may have reduced HMS formation, affecting comparability with previous studies.