Introduction

Chlorine radicals (Cl·) are an important atmospheric oxidant, affecting atmospheric chemical components and deteriorating air quality1,2,3,4. In the polluted troposphere, Cl· can react with volatile organic compounds (VOCs), resulting in the formation of primary ROx radicals (including OH·, HO2·, and RO2·) and O3, thereby enhancing atmospheric oxidation capacity5,6,7.

Nitryl chloride (ClNO2) and molecular chlorine (Cl2) are the major Cl· precursors in the troposphere and have been investigated on a global scale5,8,9,10,11,12,13,14,15,16,17,18,19. ClNO2 is generally generated through the heterogeneous uptake of dinitrogen pentoxide (N2O5) on the chloride-containing aerosols8, with its highest levels occurring after sunset and photolyzing rapidly after sunrise7,20. However, high daytime concentrations of ClNO2 have also been observed in previous research, mainly due to transport from upper atmosphere or reduced photolysis rates6,17,21. Cl2 can be produced from both primary and secondary sources12,18,22, with peak concentrations occurring either during the daytime or nighttime5,23,24. During the daytime, photochemical processes within aerosol particles, such as those associated with O325,26,27, OH·28, Fe(III)29, nitrates18, titanium dioxide (TiO2)30, ClNO231,32, hypochlorous acid (HOCl)33, and chlorine nitrates (ClONO2)34 on chloride-containing aerosols, are considered as important sources of Cl2. At night, heterogeneous reactions of N2O5 also contributed to Cl2 generation in addition to ClNO223,35. Besides ClNO2 and Cl2, HOCl is also a daytime source of Cl·36. It is generally produced by the reaction of ClO and HO2·, which did not fully account for the observed HOCl levels in the eastern tropical Atlantic33. Recent research suggested that the HOCl levels can be explained by a mix of Saharan dust and sea spray aerosols activated by sunlight37. Even nitrate photolysis also contributed to daytime HOCl generation36.

Currently, numerous observational studies have been conducted on the distribution characteristics and sources of individual chlorine species (ClNO2 or Cl2)6,7,18,19,21,23, but the distributions and formation mechanisms of multiple reactive chlorine species (e.g., ClNO2, Cl2, and HOCl) are not well understood. Previous studies have found that during the COVID-19 lockdown, the reduction in anthropogenic activities led to a sharp decline in NO concentrations, resulting in significant increases in ClNO2 and Cl2 levels35. Another study indicated that the rise in vehicle emissions enhanced nitrogen-related chemical processes associated with NO2 concentrations after the Spring Festival, further increasing the levels of ClNO2 and Cl2·38. These findings highlight the critical regulatory role of anthropogenic activity-driven chemical processes in the formation and transformation of reactive chlorine species.

Coastal regions influenced by anthropogenic activities and sea-salt aerosols have witnessed emerging insights into the sources of reactive chlorine species. Our previous autumn observations revealed that nitrate photolysis and aerosol iron photochemical processes serve as major sources of Cl2 and HOCl at coastal sites36,39. Similar with these findings, field measurements coupled with simulations in the coastal atmospheres of Hong Kong further underscore the key contribution of nitrate photolysis to Cl2 production18,40, yet the sources of HOCl remain unresolved40. Given the complex interplay between marine and continental air masses in coastal environments, systematic investigations under varying air mass conditions are urgently needed to elucidate the formation mechanisms and atmospheric impacts of reactive chlorine species.

In this study, we conducted intensive observations of reactive chlorine species (ClNO2, Cl2, and HOCl) during summer in a southeastern coastal city. The results indicated that the concentrations of ClNO2, Cl2, and HOCl were very low under the influence of marine air masses, whereas under the influence of continental air masses, the concentrations of these chlorine species increased sharply, accompanied by secondary pollution. Through the combination of field observations and model simulations, we elucidated the distribution patterns, formation and loss processes, and differential impacts on atmospheric oxidation capacity of ClNO2, Cl2, and HOCl under the influence of marine and continental air masses. This study not only advances the scientific understanding of the formation mechanisms of multiple reactive chlorine species but also provides important insights for formulating regional photochemical pollution control strategies.

Results and discussion

Significant enhancement of chlorine chemistry

Figure S1 presents the time series of ClNO2, Cl2, HOCl, and related parameters observed in a coastal city of southeastern China from July 19th to August 9th. Interestingly, the study site was alternately influenced by marine air masses (Type 1) and continental air masses (Type 2) during the observation period from mid-July to early August (Fig. 1a, b). This provided a unique opportunity to investigate the influence of marine and continental air masses on the distributions of reactive chlorine species.

Fig. 1: Distribution characteristics of reactive chlorine species (ClNO2, Cl2, and HOCl) and related parameters under the influence of marine and continental air masses, respectively.
figure 1

(a, c, e, g, i) Period of marine air mass influence (July 19th–26th, 2023); (b, d, f, h, j) Period of continental air mass influence (August 1st–9th, 2023).

Full size image

Under the influence of marine air masses, particulate chloride (Cl−) exhibited a strong correlation with sodium ions (Na+) (Fig. S2a, r = 0.86) but no significant correlation with potassium ions (K+) (Fig. S2b, r = 0.06). In contrast, during the continental air mass influence period, the correlation between Cl− and K+ became statistically significant (r = 0.49) (Fig. S2d). This difference confirms that the study site was predominantly influenced by marine and continental air masses during the Type 1 and Type 2 periods, respectively. Notably, the diurnal variation of Cl− concentrations during the Type 2 period exhibited lower daytime levels (Fig. S3), suggesting the enhanced photochemical loss, which likely promoted the formation of gaseous reactive chlorine species (ClNO2, Cl2, and HOCl). The average hourly concentrations of ClNO2, Cl2, and HOCl during the Type 1 period were 0.69, 1.7, and 4.8 ppt, respectively, while during the Type 2 period, they increased markedly to 21.5, 14.3, and 49.9 ppt, respectively (Fig. 1 and Table S1). Concurrently, substantial increases were observed in the concentrations of O3, PAN, and PM2.5, with PM2.5 levels rising from a maximum of 20 μg·m−3 during the Type 1 period to 60 μg·m−3 during the Type 2 period, reflecting a pronounced deterioration in air quality during the Type 2 period at the study site.

During the Type 1 period, the concentrations of reactive chlorine species were consistently low and exhibited no significant diurnal variations. However, during the Type 2 period, ClNO2 displayed a distinctive diurnal variation pattern characterized by a nighttime peak and sustained elevated levels in the afternoon. We can exclude instrumental artifacts as an explanation for the summer ClNO2 distribution, as evidenced by the characteristic isotopic signature of ClNO2 (Fig. S4). Prior observations reported nighttime peaks and accumulations in ClNO2 concentrations followed by rapid photolysis after sunrise5,8,13,19. In this study, the limited nocturnal accumulation of ClNO2 weakened its impact on daytime ClNO2 levels, thereby facilitating the identification of other possible processes related to daytime ClNO2 generation. As indicated by prior studies41,42,43, the daytime availability of NO3 influences ClNO2 concentrations by modulating the formation of N2O5 (Fig. S5). Using a box model, we quantitatively assessed the contribution of heterogeneous N2O5 uptake to daytime ClNO2 concentrations (Text S1). In our calculations, we adopted an N2O5 uptake coefficient (γ(N2O5)) of 0.03 and a ClNO2 production efficiency (ϕ(ClNO2)) of 0.6, representing moderate estimates6,32,44,45. The simulated daytime ClNO2 levels accounted for only 20% of the observed ClNO2 concentrations (Fig. S6). Even when extreme values of γ(N2O5) (0.06) and ϕ(ClNO2) (1.0) were applied, the observed daytime ClNO2 peak could not be fully explained by N2O5 chemistry alone (Fig. S6), suggesting the existence of unknown daytime ClNO2 formation pathways. Additionally, Cl2 and HOCl also exhibited daytime peaks. The persistent daytime peaks during the Type 2 period suggest that, despite rapid photolysis, the sources of reactive chlorine species were very strong, continuously replenishing their concentrations. Furthermore, statistical analysis of daytime data (07:00–17:00 Local Time) during the Type 2 period revealed the strong linear correlations among these species: the correlation coefficient (r) between Cl2 and ClNO2 was 0.75, between HOCl and Cl2 was 0.66, and between HOCl and ClNO2 was 0.82 (Fig. S7), respectively, implying their common sources.

To investigate the reasons for the elevated concentrations of reactive chlorine species during the Type 2 period, we compared meteorological parameters and related species between the two periods (Table S1). The results showed that photochemical reaction conditions (temperature, UV radiation, and JNO2 values) and diffusion conditions (wind speed) differed only slightly, suggesting that these meteorological factors were not the primary drivers of the significant differences in the concentrations of reactive chlorine species. However, NOx levels, aerosol iron (Fe), particle-related components, such as nitrates (NO3⁻) and ammonium (NH4+), all exhibited higher concentrations during the Type 2 period. Given the observed daytime concentration peaks of ClNO2, Cl2, and HOCl, coupled with their short photochemical lifetimes in summer, ClNO2, Cl2, and HOCl production may be predominantly driven by local photochemical processes. Therefore, the impacts of marine and continental air masses on the concentrations of ClNO2, Cl2, and HOCl may be indirect, primarily through influencing precursor concentrations rather than directly changing the concentrations of ClNO2, Cl2, and HOCl. Based on the comprehensive analysis, it is inferred that under the similar conditions of photochemical reaction and atmospheric diffusion, the elevated concentrations of both primary and secondary pollutants during the Type 2 period are likely to facilitate the enhancement of ClNO2, Cl2, and HOCl production. Pollution-driven chlorine activation was also observed in coastal areas of Hong Kong18,40. These results underscore the need for further investigation into the sources and mechanisms driving the formation of daytime reactive chlorine species.

Production and loss processes of reactive chlorine species

Due to the considerable enhancement in the concentrations of reactive chlorine species during the Type 2 period, this study focused on exploring the sources of daytime Cl2, HOCl, and ClNO2 during this period.

In the box model, Cl2 concentrations were simulated by incorporating gas-phase reactions (e.g., ClO + ClO and ClO2 + Cl·) and heterogeneous uptake reactions of ClNO2, HOCl, O3, and OH· on chloride-containing aerosol surfaces, alongside loss pathways such as Cl2 photolysis, Cl2 + OH·, and negligible deposition. Despite including these reactions, the simulated Cl2 concentrations accounted for only 30% of the observed concentrations (Fig. 2a), implying substantial uncertainties in the uptake coefficients of known pathways or the presence of significant unknown sources. Among the known Cl2 formation pathways (Fig. 2b), O3 uptake was the dominant contributor (60.0%), followed by ClNO2 (28.0%) and HOCl (11.0%) uptake, while OH· uptake (<2.0%) and gas-phase reactions played minor roles. Previous studies have demonstrated that incorporating the heterogeneous uptake reaction of O3 (γO3 = 1 × 10−3) into the CMAQ model could explain over 80% of Cl2 formation46, but the selection of uptake coefficients requires further validation. Given that photolysis is the primary daytime loss pathway for Cl2 and its photolysis lifetime is short, Cl2 is generally considered to be in a quasi-steady state, where its production rates approximate its loss rates. By subtracting the total loss rates from the total production rates, we estimated the production rates of the unknown Cl2 sources. Recent studies proposed that nitrate photolysis and aerosol iron photochemistry may serve as significant daytime Cl2 sources18,37,47. Accordingly, we analyzed the correlations between the unknown Cl2 sources and the approximate values of the nitrate photolysis and aerosol iron photochemistry indicator (Fig. 2c, d), revealing good correlations (r = 0.41 and 0.45), respectively. This suggests that these processes may contribute to additional daytime Cl2 sources. Notably, although the uptake coefficients for HOCl, ClNO2, OH·, and O3 are supported by literature23, they may not be fully applicable to the specific ambient conditions of our study site. Therefore, we have established a series of different scenarios to evaluate variations in unknown sources of Cl2 (Table S2). As shown in Fig. S8, daytime Cl2 concentrations could only be reasonably simulated when the uptake coefficients for O3, ClNO2, HOCl, and OH· were set to four times the literature-reported values. However, we consider these adjusted uptake coefficients to be unrealistically high. Based on scenario analysis, the estimated production rates of the unidentified Cl2 sources reach up to 0.22 ppb·h−1.

Fig. 2: Formation mechanisms of Cl2.
figure 2

a Comparison between observed and simulated Cl2 concentrations; b Production and loss rates of Cl2; c Relationship between the unknown sources of Cl2 and the indicator of nitrate photolysis (NO3−×JNO2×Sa); d Relationship between the unknown sources of Cl2 and the indicator of aerosol iron photochemistry (Fe×JNO2×Sa). Here, Sa signifies aerosol surface area.

Full size image

HOCl was primarily generated via the reaction of ClO with HO2·, where ClO is formed from the reaction of Cl· with O3, though ClO can also be consumed by NOx. Observational results support this mechanism (Fig. S9), showing a strong correlation between daytime O3 and HOCl concentrations, with high HOCl levels often coinciding with low NOx concentrations. The photolysis of ClNO2 and Cl2 provides a substantial source of Cl·, accelerating HOCl formation via the reaction of ClO and HO2·. Using the box model, we evaluated the contributions of the ClO + HO2· reaction and other gas-phase pathways (e.g., Cl2/ClNO2 + OH·, OH· + OClO) to HOCl concentrations. The results indicate that the simulated HOCl concentrations accounted for only 5–10% of the observed daytime concentrations (Fig. 3a). Similarly, unknown sources of HOCl were also found in marine air affected by ship exhaust in Hong Kong40 and in the marine boundary layer of the eastern tropical Atlantic33. These findings highlight a considerable gap in the current understanding of HOCl sources. In this study, the loss rates of HOCl reached 0.12 ppb·h−1, with photolysis as the primary loss pathway (81.0%), followed by HOCl uptake (17.0%) and the reaction of HOCl with OH· (2.0%) (Fig. 3b). Similar to the analysis of unknown Cl2 sources, we estimated the formation rates of the unknown HOCl sources by subtracting the loss rates from the formation rates. Recent studies suggested that nitrate photolysis and aerosol iron photochemistry not only contribute to Cl2 formation but may also increase HOCl concentrations36,37,47,48. To validate these mechanisms at our study site, we analyzed the correlations between the unknown HOCl sources and the approximate values of the nitrates photolysis and aerosol iron photochemistry indicator (Fig. 3c, d), finding their strong correlations (r = 0.69 and 0.56), respectively. Therefore, in addition to Cl2, nitrate photolysis and aerosol iron photochemistry also resulted in HOCl formation.

Fig. 3: Formation mechanisms of HOCl and ClNO2.
figure 3

a Comparison between observed and simulated HOCl concentrations; b Production and loss rates of HOCl; c Relationship between the unknown sources of HOCl and the indicator of nitrate photolysis (NO3−×JNO2×Sa); d Relationship between the unknown sources of HOCl and the indicator of aerosol iron photochemistry (Fe×JNO2×Sa); e Production and loss rates of ClNO2; f Relationship between the unknown sources of ClNO2 and the indicator of nitrate photolysis (NO3−×JNO2×Sa).

Full size image

As mentioned before, observed daytime concentrations of ClNO2 could been not explained by heterogeneous N2O5 processes when considering large loss rates of ClNO2 during the daytime primarily contributing from ClNO2 photolysis and uptake (Fig. 3e), revealing unknown production pathways of daytime ClNO2. Notably, strong correlations among HOCl and ClNO2, Cl2 during the daytime imply that these species may share similar daytime sources. Recent experimental and modeling studies indicated that nitrate photolysis can produce ClNO2 in addition to Cl2 and HOCl48. Specifically, O(3P) generated by nitrate photolysis oxidizes Cl− to form Cl2 and HOCl, further reacting with nitrite to produce ClNO248. In our study, the production rates of unknown ClNO2 sources exhibited a strong correlation with the approximate value of the nitrate photolysis indicator (r = 0.59) (Fig. 3f), indicating that nitrate photolysis may be an important source for the elevated daytime ClNO2 concentrations at our study site.

Figure S10 illustrates the daytime chemical cycling mechanisms of ClNO2, Cl2, and HOCl at our study site. The photolysis of ClNO2, Cl2, and HOCl releases Cl·, which react with O3 to form ClO. This ClO subsequently interacts with HO2·, NO, and NO2 to produce HOCl, Cl·, and ClONO2, respectively. The heterogeneous uptake of HOCl, ClNO2, O3, and OH· on chloride-containing aerosol surfaces contributes to the formation of Cl2. Furthermore, the heterogeneous uptake of N2O5 results in the production of ClNO2. Additionally, the photolysis of nitrate facilitates the simultaneous generation of ClNO2, Cl2, and HOCl, while aerosol iron photochemistry promotes the formation of Cl2 and HOCl. Overall, these processes significantly enhance reactive chlorine chemistry during the Type 2 period, potentially having profound implications for photochemical oxidation processes.

Enhanced oxidation capacity

Based on the box model, we systematically compared the changes in atmospheric oxidation induced by reactive chlorine chemistry (the photolysis of ClNO2, Cl2, and HOCl) and their differential impacts on O3 and PAN formation during the Type 1 and Type 2 periods. The average maximum Cl· production rates increased significantly from 2.7 × 105 molecules·cm−3·s−1 during the Type 1 period to 4.5 × 106 molecules·cm−3·s−1 during the Type 2 period, representing an approximately 15-fold increase (Fig. S11). As illustrated in Fig. 4a, source analysis revealed that Cl2 photolysis dominated Cl· production during both periods, contributing over 80%, followed by HOCl photolysis (~15%), with ClNO2 photolysis contributing the least (~5%). Higher contribution from Cl2 photolysis have also been found in northern and southern China18,24. Differently, several previous studies reported larger contribution of ClNO2 photolysis to Cl· generation in the early morning5,10,49. However, in this study, the limited nighttime accumulation of ClNO2 resulted in minimal contributions to ClNO2 concentrations in the morning. The contribution of HCl + OH to Cl· generation was not considered in this study based on two key considerations: (1) the absence of direct HCl measurements, and (2) the substantial uncertainties of simulated HCl concentrations.

Fig. 4: Impacts of reactive chlorine chemistry on atmospheric oxidation capacity under the influence of marine and continental air masses, respectively.
figure 4

a Contributions of ClNO2, Cl2, and HOCl photolysis to Cl· production rates in continental air masses; b–f Differences in the contributions of reactive chlorine chemistry to OH· (b) HO2· (c) RO2· (d) PA· concentrations (e) and net O3 production rates (net P(O3)) (f) under the influence of marine (Type 1) and continental (Type 2) air masses.

Full size image

The Cl· released from the photolysis of ClNO2, Cl2, and HOCl promoted the formation of HO2· and RO2· through VOC oxidation. Consistent with previous studies18,38, it is found that Cl· primarily participated in alkane oxidation (>50%), followed by the oxidation of OVOCs and alkenes (Fig. S12). Notably, HOCl photolysis not only releases Cl· but also generates OH·. Additionally, HO2· and RO2· generated by chlorine chemistry can react with NO to form OH·, which further promotes HO2· and RO2· formation through VOC oxidation, thereby intensifying the cycling of OH·, HO2· and RO2·. During the Type 1 period, low concentrations of ClNO2, Cl2, and HOCl resulted in limited enhancement of OH· (1.4%), HO2· (<1.0%), and RO2· (<1.0%) concentrations via Cl·-mediated VOC oxidation (Fig. 4b–d). In contrast, during the Type 2 period, higher concentrations of reactive chlorine compounds led to average increases of 3.0%, 5.4%, and 13.0% in OH·, HO2· and RO2· concentrations, respectively. Importantly, the concentrations of PA· within RO2· increased markedly during the Type 2 period due to enhanced reactive chlorine chemistry, with an average maximum increase of 14.2% (Fig. 4e).

Based on these changes in radical concentrations, we further quantified the impacts of reactive chlorine chemistry on photochemical O3 and PAN formation. By calculating the difference between O3 production and loss rates, we assessed its effect on net O3 production rates (P(O3)): the average increase was 1.4% during the Type 1 period but rose to 10.5% during the Type 2 period (Fig. S13 and Fig. 4f). Simultaneously, as PAN is formed through the reaction of PA· with NO2, the increased PA· concentrations induced by reactive chlorine chemistry greatly enhanced PAN concentrations, with a maximum increase of 14.1% during the Type 2 period.

The sensitivity of O3 and PAN generation

Several studies have revealed the role of reactive chlorine chemistry in modulating O3 production sensitivity5,50,51, yet its contribution to PAN formation sensitivity remains unclear. Therefore, we evaluated the impacts of VOC and NOx reduction on net P(O3) and PAN concentrations after incorporating reactive chlorine chemistry mechanisms into the box model (Fig. 5).

Fig. 5: EKMA curves of net O3 production rates (net P(O3), ppb·h−1) and PAN concentrations (ppb) under different simulation scenarios during the Type 2 period.
figure 5

(a, c, e) Net P(O3) (Unit: ppb·h−1); (b, d, f) PAN concentrations (Unit: ppb). (a, b) Scenarios without reactive chlorine species constraints; (c,d) scenarios constrained by ClNO2, Cl2, and HOCl; (e,f) the individual impacts of chlorine chemistry on O3 and PAN (△Cl). △Net P(O3) equals net P(O3) with chlorine chemistry minus net P(O3) without chlorine chemistry, and △PAN equals PAN concentrations with chlorine chemistry minus PAN concentrations without chlorine chemistry. The white squares in the Fig. 5 represent the average concentrations of VOCs and NOx during the Type 2 period.

Full size image

Given higher O3 and PAN concentrations during the Type 2 period and the significant enhancement effect of reactive chlorine chemistry during this period, we focused on assessing its impacts on EKMA curves for net P(O3) and PAN concentrations during the Type 2 period. The results show that in scenarios without reactive chlorine chemistry, O3 and PAN generation is primarily controlled by VOCs, and a reduction in NOx concentrations exacerbates their formation (Fig. 5a, b). Further analysis reveals that when reactive chlorine chemistry is introduced into the box model (Fig. 5c, d), although O3 and PAN production remains predominantly VOC-sensitive, net P(O3) and PAN concentrations increased, highlighting the strong influence of reactive chlorine chemistry on photochemical oxidation processes. To directly compare the effects of including or excluding reactive chlorine chemistry on O3 and PAN production, we conducted a difference analysis of net P(O3) and PAN concentrations between scenarios with and without reactive chlorine species constraints and subsequently generated EKMA curves based on this analysis. Notably, when reactive chlorine chemistry is considered, the sensitivity of O3 and PAN shifts to co-control by VOCs and NOx (Fig. 5e, f). This result underscores the critical role of VOCs and NOx in the positive impact of reactive chlorine chemistry on photochemical O3 and PAN production.

Previous studies based on seasonal observations of ClNO2 and Cl2 in Beijing have shown that the inclusion of reactive chlorine chemistry alters O3 sensitivity5. Research in southern coastal China also indicated that Cl2 generated from nitrates photolysis limits O3 increases in VOC-sensitive regions and exacerbates O3 decreases in NOx-sensitive regions51. However, these studies did not comprehensively consider reactive chlorine species, thereby, potentially resulting in the underestimation for the effects of reactive chlorine chemistry in atmospheric oxidation. Notably, previous studies indicated that the molybdenum conversion method tends to overestimate NO2 concentrations52,53,54, which may affect the sensitivity of O3 and PAN formation. To evaluate this uncertainty, we reduced the observed concentrations of NO2 by 20%. As shown in Fig. S14, compared with the original results, the decreased concentrations of NO2 indeed altered the sensitivity of O3 and PAN formation.

In summary, this study revealed that the elevated concentrations of precursors promoted the formation of reactive chlorine species by accelerating heterogeneous chemical processes. The pollution-driven chlorine chemistry moderately enhanced atmospheric oxidation capacity, resulting in O3 and PAN formation and shifting their control regime. Importantly, we identified unknown sources of reactive chlorine species linked to nitrate photolysis and aerosol iron photochemistry. However, the lack of well-constrained parameterization schemes for these heterogeneous processes limits the accurate quantification of their contributions. Future research should prioritize elucidating these mechanisms to refine parameterizations and improve the precision of model simulations, thereby advancing our understanding of reactive chlorine chemistry in the atmospheric environment.

In our previous studies on Cl2 and HOCl during the autumn36,39, the impacts of different air mass sources on the distributions of reactive chlorine species were not well revealed due to the constant influence of coastal air masses during this period. However, in this study, by systematically comparing the distribution patterns and photochemical impacts of reactive chlorine species under the influence of marine and continental air masses, this study emphasized the pivotal role of anthropogenic pollutants in driving reactive chlorine chemistry, with important implications for air pollution control policies. Although sea salt aerosols serve as an abundant source of particulate chloride, they are not the primary limiting factor for the generation of reactive chlorine species18,36. Instead, anthropogenic emissions, particularly NOx and iron-containing aerosols, play a critical role in the chlorine chemistry. These findings underscore the necessity of strict controls on anthropogenic pollutants (e.g., VOCs, NOx, and Fe) to mitigate chlorine-mediated adverse environmental impacts. Furthermore, we propose extending control measures to anthropogenic chlorine emissions in air pollution mitigation strategies in the future.

Methods

Field Measurements

Measurements of reactive chlorinated compounds and associated parameters were conducted from July 19th to August 9th, 2023, in Xiamen, a major coastal city in Southeast China. The study site, located over 70 meters above ground level in an urban area, is surrounded by Xinglin Bay, educational and residential buildings, and traffic roads. Throughout the observation period, ClNO2, Cl2, HOCl, N2O5, VOCs, and other trace gases (including NOx, CO, SO2, and O3), as well as PM2.5, aerosol chemical components (e.g., Fe and water-soluble ions), aerosol surface area (Sa), and meteorological parameters, were monitored at the site. Detailed descriptions of observation instruments are provided in Text S2.

An iodide-adduct Chemical Ionization-Atmospheric Pressure Interface-Time of Flight mass spectrometer (I−-ToF-CIMS) was employed to measure ClNO2, Cl2, HOCl, and N2O5. The operational principles of I−-ToF-CIMS and related details have been described in previous studies and Text S136,38,55. Targeted species selectively reacted with reagent ions (I− and I(H2O)−), forming iodide clusters that were detected by the mass spectrometer. Calibrations of ClNO2, Cl2, HOCl, and N2O5 were performed using established methods7,18,19,22,33,56. Briefly, ClNO2 was synthesized through the reaction of Cl2 with a moistened mixture of NaNO2 and NaCl. N2O5 was generated by the reaction of O3 with an excess of NO2. Cl2 was produced using a Cl2 permeation tube (supplied by VICI Metronics, Inc.) with a permeation rate of 98 ng·min−1. HOCl calibration was achieved by converting HOCl to Cl2. By analyzing the reduction in the HOCl signal and the corresponding increase in the Cl2 signal as they passed through a chloride solution, the relative sensitivities of HOCl and Cl2 were determined. A summary of the calibration procedures and uncertainty analysis is provided in Text S3 and Fig. S15.

The box model

An observation-based model (OBM) was employed to investigate the effects of ClNO2, Cl2, and HOCl photolysis on photochemical oxidation processes. The box model incorporated the Master Chemical Mechanism (MCM Version 3.3.1) and integrated established chlorine chemistry mechanisms based on previous studies18,49. Detailed configurations of the box model have been thoroughly described in our earlier publications38,57. The photolysis rates of ClNO2, Cl2, and HOCl (JClNO2, JCl2, and JHOCl) were derived from the Tropospheric Ultraviolet and Visible Radiation (TUV) model under clear-sky conditions, with adjustments made using measured JNO2 values at our study site. In this study, the box model was constrained by observed data, including ClNO2, Cl2, HOCl, VOCs, HCHO, HONO, O3, NO2, NO, CO, SO2, and meteorological parameters, all provided at 1 h time resolution, ensuring consistency between the simulated and observed diurnal variations of these species’ concentrations.

Notably, the dissociation of N2O5 into NO2+ and NO3− initiates competitive reactions between Cl⁻ and other species with NO2+, which directly modulates ClNO2 production41. Following previous studies6,58, we employed the simplified reaction (N2O5 + Cl− → ClNO2) in our model, while addressing potential uncertainties through varying N2O5 uptake coefficients and ClNO2 yields. Quantifying Cl2 and HOCl sources and accurately simulating their observed concentrations remain a great challenge in reactive chlorine chemistry. Existing studies have yet to fully resolve their sources or achieve precise simulations of these species’ concentrations40,58. Our study accordingly distinguishes between the known and unknown sources. For the known sources, we quantified formation and loss rates using established laboratory parameterizations. Sensitivity analysis was performed by testing different uptake coefficients and yields. The unknown source strength was derived by subtracting known formation rates from total production, followed by correlation analysis with potential mechanisms to assess their possibility.

Based on the box model, this study analyzed the generation of OH·, HO2·, RO2·, O3, and PAN under the influence of marine and continental air masses. To quantify the effects of chlorine chemistry on atmospheric oxidation processes, two distinct chemical scenarios were examined: one incorporating the reactive chlorine chemistry mechanism and another excluding it. To further elucidate the impacts of reactive chlorine chemistry on the control strategies for O3 and PAN precursors, we employed the empirical kinetic modeling approach (EKMA) during the period dominated by continental air masses, which are typically associated with enhanced photochemical pollution. Through a comprehensive 20 × 20 concentration matrix, this study quantitatively simulated the response patterns of O3 and PAN formation across a wide range of concentration gradients (both reduction and increase) for VOCs and NOx. Additionally, we conducted a mechanistic analysis to evaluate the relative contributions of current atmospheric processes, including (1) gaseous reactions and heterogeneous uptake of ClNO2, HOCl, OH·, and O3, (2) ClO + HO2· and other gaseous reactions, and (3) heterogeneous N2O5 uptake, to the daytime concentrations of Cl2, HOCl, and ClNO2, respectively. Detailed descriptions of the model configurations, input parameters, heterogeneous reaction settings, and output parameters are provided in Text S1.

HYSPLIT backward trajectory

The 24 h backward trajectories of air masses arriving at 500 meters above ground level were calculated using the Hybrid Single-Particle Lagrangian Integratesd Trajectory (HYSPLIT) model.