Relative humidity driven nocturnal HONO formation mechanism in autumn haze events of Beijing

China’s stringent air quality controls have reduced emissions and PM2.5, yet annual PM2.5 in the Beijing-Tianjin-Hebei region remains well above WHO guidelines. Elevated primary emissions and an increased atmospheric oxidation capacity exacerbate haze. Hydroxyl radicals (OH) are the main atmospheric oxidant, and HONO photolysis is often the dominant primary OH source in polluted conditions. However, existing models fail to reproduce observed HONO, indicating missing sources or poorly constrained processes. This study addresses the research question: what mechanisms, particularly the role of relative humidity in nocturnal heterogeneous NO2 conversion, explain observed HONO during autumn haze in Beijing, and how can improved parameterization reduce model biases and quantify contributions to OH and secondary pollution?

Prior studies identify multiple HONO sources: direct emissions; homogeneous NO + OH; heterogeneous NO2 reactions on humid surfaces; photo-enhanced surface reactions with photosensitizers; photolysis of nitric acid, particulate nitrate, and nitrogen-containing organics. Despite including many mechanisms, models still underpredict HONO, particularly in complex, polluted environments. Relative humidity strongly modulates heterogeneous NO2-to-HONO conversion, and accounting for RH has improved model performance in recent works. Debate persists regarding whether aerosol or ground surfaces dominate heterogeneous HONO production, with contributions varying by location and pollution level. Reported vehicle HONO emission factors vary widely (0.03–2.1%) depending on fuel and driving conditions, underscoring the need for site-specific constraints. Evaluated kinetic data show wide variability in NO2 uptake coefficients under different NO2, RH, and irradiation conditions.

Field observations were conducted on the rooftop of an eight-story building at the Research Center for Eco-Environmental Sciences (RCEES), Chinese Academy of Sciences in Beijing (40.02°N, 116.35°E) from 5 Oct to 4 Nov 2021. The urban site is near multiple roads and lacks nearby industrial sources. Measurements included HONO (WLPAP), NOx, SO2, O3 (Thermo analyzers), NH3 (QC-TILDAS), NR-PM1 composition (HR-TOF-AMS), particle size distributions (two SMPS, 5–700 nm), VOCs (PTR-TOF-MS calibrated with PAMS standards), photolysis frequencies (UF-CCD spectrometer), and meteorology (T, RH, P, WS, WD). PM2.5 and CO were from a nearby station (~5.5 km), and boundary layer height from reanalysis. A zero-dimensional AtChem2-MCM v3.3.1 box model was used, constrained by observed trace gases, VOCs, meteorology, and photolysis frequencies. The analysis focused on 26 Oct–4 Nov (stagnant conditions; complete HONO data), with a 4-day spin-up (22–25 Oct). Three model cases were evaluated: Base (MCM default HONO chemistry via NO + OH only), EGAN (Base plus seven additional sources: direct emission, heterogeneous NO2 on ground/aerosol with and without photo-enhancement, photolysis of particulate nitrate and adsorbed HNO3, and losses including photolysis, reaction with OH, dry deposition), and EGANwRH (EGAN with an RH enhancement factor applied to nocturnal heterogeneous NO2 reactions on ground and aerosol surfaces). Parameter constraints from observations were derived to reduce uncertainties: (1) Direct vehicle emission factor for HONO estimated by screening 10 fresh nocturnal plumes (criteria: NOx > 63 ppb; ΔNO/ΔNOx > 0.8; duration < 2 h; strong HONO–NOx correlation R2 > 0.7), yielding ΔHONO/ΔNOx = 1.23% ± 1.09%. (2) Ground-surface NO2 uptake coefficient γg estimated from 5 nocturnal cases with minimal aerosol influence (PM2.5 < 15 µg m^-3; stable NO2 and CO; WS < 1 m s^-1; NO < 5 ppb; sustained HONO and HONO/NO2 increase > 2 h). NO2-to-HONO conversion rate CHONO was derived and used with boundary layer height H to compute γg via CHONO = γg × (1/8) × (ν/ H), where ν depends on temperature (νNO2 = sqrt(8RT/πM)); at night H was set to BLH (97% < 100 m) and daytime H = 100 m. The resulting γg averaged 8.25 × 10^-5. To capture missing nocturnal sources, the discrepancy HONOGRH = HONOobs – HONOEGAN was attributed to RH-enhanced heterogeneous NO2 conversion on the ground. The ratio HONOGRH/HONOG defined an RH enhancement factor EFRH relative to the ground-heterogeneous HONO component (HONOG). Analysis showed a piecewise-linear dependence of EFRH on RH with a breakpoint at ~50%. The parameterization used: EFRH = 0.59 × RH + 0.46 for RH < 50%, and EFRH = 8.34 × RH – 3.29 for RH ≥ 50%. Given strong nocturnal NH3–RH correlation (R2 = 0.75), EFRH was interpreted to include combined RH and co-varying factors (e.g., NH3 effects). Sensitivity tests varied aerosol uptake parameters (γa, Sa) to assess aerosol-surface contributions. The mechanism and parameters were validated at a suburban Beijing site (BIPT Qingyuan campus) during three late-summer/early-autumn campaigns (2018–2019), using the same scheme except a site-specific direct emission factor (0.85%).

• The default MCM (Base) severely underpredicted HONO with NMB = -92.8%.
• Adding seven HONO sources (EGAN) improved NMB to -46.1% but still missed nocturnal HONO, especially during high RH/PM2.5.
• Site-constrained parameters: vehicle HONO emission factor = 1.23% ± 1.09%; ground-surface NO2 uptake coefficient γg = 8.25 × 10^-5 (range ~3.06 × 10^-5 to 2.21 × 10^-4).
• A nocturnal RH enhancement factor EFRH, piecewise-linear with RH (break at ~50%), further improved performance (EGANwRH): NMB = -5.1%, IOA = 0.97.
• Source apportionment: Nighttime HONO dominated by heterogeneous NO2 reaction on ground surfaces (85.6%), with vehicle emissions contributing 10.9%; homogeneous NO + OH and aerosol heterogeneous pathways each < 3% at night. Daytime HONO sources: homogeneous NO + OH (41.8%), ground heterogeneous (31.3%), vehicle emissions (21.3%), particulate nitrate photolysis (3.4%), aerosol heterogeneous (1.6%); adsorbed HNO3 photolysis negligible.
• Aerosol-surface contribution was limited during the study; HONOcorr/NO2 increased with RH but decreased when PM2.5 exceeded ~100–180 µg m^-3; sensitivity tests showed little impact of increasing γa or Sa on HONO trends, attributed to relatively low Sa.
• OH budget: overall, HO2 + NO contributed 85.3% of OH; daytime primary OH production dominated by HONO photolysis (73.6%), with OVOCs/HNO3/H2O2 photolysis 22.3% and O3 photolysis 4.1%; ozonolysis negligible.
• The impact of added HONO sources on HONO and OH increased with pollution level. In P1, HONOEGANwRH-Base contribution to HONO rose from 43.5% to 67.2% as PM2.5 increased from 4 to 136 µg m^-3; corresponding OH contribution increased from 5.4% to 25.1%. In P2, OH contribution increased from 4.7% to 23.8% as PM2.5 increased from 5 to 114 µg m^-3. Despite lower OH concentrations during pollution, total OH production rates were higher, implying elevated OH reactivity and enhanced secondary formation.
• Validation at a suburban site showed improved agreement using EGANwRH with IOA > 0.82 and NMB within ±35%, particularly effective when RH > 60%.

The study demonstrates that nocturnal HONO deficits in models stem largely from underrepresented RH effects on heterogeneous NO2 conversion, primarily on ground-related surfaces (buildings, vegetation, pavements). By deriving site-specific parameters (vehicle emission factor, γg) and implementing an RH-driven enhancement factor, the model reproduced observed HONO with small bias and high agreement. The ground surface dominates nocturnal HONO formation under the observed autumn haze conditions, while aerosols contributed little, likely due to relatively low aerosol surface area. The enhanced nocturnal HONO accumulates under stagnant, high-RH conditions and photolyzes after sunrise to produce OH, which accelerates secondary pollutant formation (nitrate, sulfate, SOA), reinforcing haze through a feedback mechanism. The parameterization, validated at another site and season, suggests broader applicability, especially under RH > 60% and in NH3-rich environments where RH and NH3 covary. These findings refine our understanding of HONO formation pathways, improve model predictability, and highlight RH as a critical environmental control, informing pollution mitigation strategies and chemical mechanism development.

This work identifies and parameterizes an RH-driven enhancement of nocturnal heterogeneous NO2 conversion to HONO, constrained by observations in urban Beijing. Incorporating this factor, along with site-specific vehicle emission and ground uptake coefficients, substantially improves HONO simulations (NMB from -92.8% to -5.1%) and clarifies source contributions (ground heterogeneous dominating at night; NO + OH in daytime). The enhanced HONO photolysis is a major daytime OH source, increasing atmospheric oxidation and secondary aerosol formation during haze. The parameterization was validated at a suburban site, indicating broader applicability under humid conditions. Future research should investigate RH effects on daytime HONO chemistry, further disentangle the roles of NH3 and aerosol properties, expand multi-layer/vertical observations to quantify surface-type contributions, and test the scheme across seasons and regions with varying emissions and meteorology.

• The RH enhancement was applied to nocturnal heterogeneous NO2 reactions; daytime RH effects were not explored and may also be important.
• Site- and period-specific conditions (autumn 2021, urban Beijing) and relatively low aerosol surface area may limit generalizability to other seasons or regions with different aerosol properties.
• PM2.5 and CO were from a nearby station (~5.5 km), introducing potential representativeness uncertainty.
• The attribution of missing HONO primarily to ground-surface processes assumes minor aerosol contributions under observed conditions; different aerosol regimes might yield other outcomes.
• The Jun Liu affiliation was not explicitly indicated in the provided text; author affiliations may have minor ambiguities in the source formatting.