Key role of continental inorganic halogens in the evolution of global air quality Authors: Qinyi Li, Xiao Fu, Xianyi Sun, Javier A Barrera, Rafael P Fernandez, Carlos A Cuevas, Chuang Qin, Yee Jun Tham, Shanshan Wang, Douglas Kinnison, Jean-Francois Lamarque, Likun Xue, Tao Wang, Alfonso Saiz-Lopez Published: 2026 ABSTRACT AbstractHalogen (chlorine, bromine, and iodine) species actively interact with other atmospheric constituents such as nitrogen oxides, organic gases, and ozone. Previous field observations confirmed the ubiquitous existence of halogens in various environments and modeling simulations quantified halogen impacts on air pollutants at present time. However, the abundance and impacts of continental inorganic halogens (CIH) on global air quality throughout history remain unexplored. Here, we present a global inventory of CIH emissions from anthropogenic and biomass-burning sources from 1970 to 2015, with average fluxes of 4302, 207, and 98 kt/yr for chlorine, bromine, and iodine, respectively, representing ~15%, 5-10%, and ~3% of the corresponding global short-lived chlorine, bromine, and iodine emissions. Incorporating these emissions into a global chemistry-climate model reveals a substantial increase ( > 50%) in inorganic halogen levels over continents, helping to reproduce the observed halogen levels. Our results also suggest substantial effects of CIH on air pollutant concentrations with significant spatio-temporal variations, e.g., the largest perturbations in nitrogen oxides and secondary aerosols migrating from Europe and North America in the 1980s-1990s, to East Asia in the 2000s, then to South Asia in the 2010s, suggesting a movement of CIH-associated air quality impact hotspots from the developed regions to the developing areas. This study highlights the under-appreciated role of halogen chemistry in air quality and its evolution in the recent four decades and calls for attention to the potential CIH impacts in emerging regions where field observations are not yet available.Subject terms: Atmospheric chemistry, Atmospheric science Introduction Halogen (chlorine-, bromine, and iodine-containing) species profoundly affect atmospheric chemistry, air quality, and the climate system through chemical reactions with various atmospheric constituents1–4. Halogen atoms react with ozone (O3), forming halogen oxides that subsequently react with each other, nitrogen monoxide (NO), or nitrogen dioxide (NO2), ultimately resulting in the loss of NO2 and O3 and the formation of ammonium nitrate aerosol (Eqs. 1–6); halogen oxides also modulate the ratio of OH/HO2 by transforming HO2 into OH (Eqs. 7–8); in the presence of volatile organic compounds (VOCs) including methane (CH4), chlorine and bromine atoms oxidize VOCs, leading to the formation of key atmospheric oxidants, e.g., OH, and air pollutants, e.g., secondary organic aerosol (SOA), sulfate aerosol, and O3 (Eqs. 9–16). Some air pollutants in turn activate less reactive halogens and produce photolabile reactive halogen species, e.g., nitryl chloride (ClNO2; Eqs. 17–20). More detailed descriptions of the halogen-initiated reactions affecting the budgets of atmospheric oxidants and air pollutants can be found in previous reviews1,2. Field measurements, supported by recent developments in instrumentation, have revealed the widespread existence of inorganic halogens in various environments, vertically from Earth’s surface, free troposphere, to the stratosphere, and horizontally from polar regions, remote oceans, to continental areas5–13. Modeling simulations, incorporating laboratory-derived chemical mechanisms and validated by field observations, have revealed the critical and complex impacts of halogens on the atmospheric and climate systems3,4,14–19. Anthropogenic emissions have undergone significant changes since the Industrial Revolution, particularly in the last half-century. Recent literature has quantified the large-scale changes in routine air pollutants (e.g., VOCs, NOx, sulfur dioxide, carbon monoxide, and black carbon) emission fluxes20,21 and concentration distributions22,23. In the 20th century, air pollutant emissions peaked in developed regions (e.g., Western Europe and North America) and later in the 21st century in developing regions (e.g., East and South Asia). As a result, the hotspots of severe air pollution have gradually shifted from developed to developing regions. Reports on global halogen emissions are very limited, and most of these reports focused on oceanic sources or organic halogens, which are far away from human habitats or have limited effects on air pollutants in the boundary layer. Much less is known about global continental inorganic halogens (CIH), which are mostly co-emitted with the air pollutants and their precursors. Keene et al.24 compiled continental chlorine sources for the year 1990; Zhang et al.25 reported continental chlorine emissions from 1960 to 2014; Saiz-Lopez et al.3 and Fu et al.26 included CIH sources representing present-day conditions in their simulations. However, global CIH emissions and their impacts on the evolution of air quality in the recent four decades, during which the global landscape of air pollution has been undergoing sharp changes, have remained unquantified until now. Here, we compile a global inventory of CIH emissions from 1970 to 2015, utilizing activity data from publicly available databases and emission factors from literature. We then conduct simulations for the years 1980, 1990, 2000, and 2015 using a global chemistry-climate model (Community Atmosphere Model with Chemistry, CAM-Chem) with (Full case) and without (noCNT case) the CIH emission inventory (Materials and Methods). Our simulation results show that the inclusion of CIH significantly improves the capability of the CAM-Chem model in reproducing the observed halogen concentrations over the continents. We find that CIH induces noticeable effects in air pollutant levels, particularly nitrogen oxides, secondary pollutants like surface O3, and secondary inorganic and organic aerosols. Such CIH impacts exhibit strong spatio-temporal variations that closely resemble those of air pollutants, suggesting that the co-emitted halogens exacerbate air pollution. Our results highlight that CIH is a key component of global air quality and its evolution, and that including CIH in air quality models is necessary to more accurately simulate the formation of air pollutants and their temporal evolution in different regions around the world. Results Continental sources of inorganic halogens We estimated the global emission of continental inorganic halogen species from 1970 to 2015 using the emission factor method, namely multiplying activity levels by emission factors26–28. The inventory includes inorganic chlorine species (HCl and fine-particle chloride) from coal burning, open waste burning, and biomass burning, as well as bromine (HBr and Br2) and iodine (HI and I2) from coal burning. Particulate bromide and iodide were excluded, as previous measurements29,30 have indicated that bromine and iodine emissions from coal burning occur predominantly in the gas phase. For activity data, we derived the national-level coal and solid biofuel consumption data from International Energy Agency (IEA) (http://www.iea.org), the gridded open burning masses of biomass from the historic global biomass burning emissions for CMIP6 (BB4CMIP) database31, and the gridded open burning masses of waste from an updated global anthropogenic emission inventory based on the Community Emissions Data System (CEDS)21. Following our previous studies26–28, emission factors for gaseous halogens from coal burning were calculated based on halogen content in coal, halogen release rates, installation rates of air pollution control devices, and their removal efficiencies. The year- and region-specific installation rates of air pollution control devices were estimated using the S-shape curves from the GEMS global database (http://inventory.pku.edu.cn/home)32,33, and the other parameters were obtained from our previous studies26–28. Emission factors of HCl for open biomass burning were derived from the updated compilation of biomass burning emission factors by ref. 34. Other emission factors were also from our previous studies26–28. For biomass and waste open burning, gridded emissions were calculated directly. For coal and solid biofuel burning, we have calculated the halogen emission for each subsector (e.g., powerplant, industry, residential, etc.) and each country, which are then spatially interpolated into the 1° × 1° grid. The global 0.1° × 0.1° gridded SO2 or PM2.5 emissions for each subsector from the GEMS database are used as spatial interpolation weights for gaseous or particulate halogen, respectively. These emissions are reports on continental inorganic bromine and iodine for the recent decades, while two previous studies24,25 have reported global continental inorganic chlorine emissions. Our chlorine emission agrees well with the more recent study25 and is lower than the earlier one24, although these two studies did not quantify the global atmospheric impacts of continental inorganic chlorine emissions. A more detailed analysis of the emission comparison is provided in the Methods section. As shown in Fig. 1, the hot spots of halogen emissions are located in East Asia, South Asia, Europe, and North America, with the largest emission intensity exceeding 0.2 t km−2 yr−1 for Cl and 0.01 t km−2 yr−1 for Br and I. Africa and Southeast Asia also emit a considerable amount of Cl, due to large contributions from open biomass burning. Halogen emissions in Europe and North America peaked around 1980–1990 and then decreased afterward due to the wide application of air pollution control devices for coal burning, and halogen emissions from East Asia increase until late 2000s and fluctuate afterwards as a result of the growing economy and increasingly stringent control policy, while those from South Asia continue to increase throughout the period because of the increasing population and growing economy. Globally integrated continental emissions of inorganic chlorine account for ~15% of global total (anthropogenic and natural) short-lived (lifetime <180 days) chlorine emissions in our study period, while continental inorganic bromine and iodine emissions represent between 5% to 10% and ~3% of total short-lived bromine and iodine emissions, respectively (Table 1). Inorganic halogen emission from 1970 to 2015. Global and regional a Cl, c Br, and e I emissions from 1970 to 2015. The definition of selected regions is shown in Supplementary Fig. S1a, including North America, South America, Europe, Africa, South Asia, and East Asia. Spatial distribution of b Cl, d Br, and f I emission averaged from 1970 to 2015. Table 1Global continental inorganic halogen emissions and the corresponding contribution to short-lived halogen global emissions for 1980, 1990, 2000, and 20151980199020002015Continental inorganic Cl (Gg/yr)3687.74018.74385.55409.5Global short-lived Cl (Gg/yr)23,243.124,792.125,354.031,780.0Continental inorganic Cl fraction15.9%16.2%17.3%17.0%Continental inorganic Br (Gg/yr)239.1231.3184.2147.1Global short-lived Br (Gg/yr)2552.82521.22029.63066.0Continental inorganic Br fraction9.4%9.2%9.1%4.8%Continental inorganic I (Gg/yr)100.3107.194.582.4Global short-lived I (Gg/yr)2668.02676.12822.22816.5Continental inorganic I fraction3.8%4.0%3.3%2.9% Global continental inorganic halogen emissions and the corresponding contribution to short-lived halogen global emissions for 1980, 1990, 2000, and 2015 Impact of CIH sources on halogen loadings We implemented historical CIH emissions into a global chemistry-climate model (CAM-Chem, which incorporates state-of-the-art oceanic sources and chemistry of halogens)26, and conducted simulations with (FULL cases) and without (noCNT cases) these CIH emissions for the years 1980, 1990, 2000, and 2015 (Table S1). The simulated average changes in inorganic halogens at the surface layer show large spatial heterogeneity (Fig. S1), closely following the emission distribution (Fig. 1), due to the short lifetime of hours to days for these inorganic halogens. The CIH emission-induced changes in global average surface mixing ratios of total inorganic chlorine (Cly; Methods), bromine (Bry), and iodine (Iy) over the continents are 14.9, 0.7, and 0.2 pptv, respectively (Table S2), or in relative terms, 116.5%, 113.0%, and 56.8%, respectively (Table S3). Such large increases in simulated halogen species help to close the existing gap between the simulated and observed inorganic halogens over the continents (Fig. S2), suggesting that CIH is necessary to reasonably reproduce the observed levels of reactive halogens over the continents. Note that these continental field measurements of inorganic halogens were conducted in the recent two decades, mainly in North America, Europe, and East Asia. Based on the model evaluation, our model simulations, incorporating CIH sources, provide a realistic representation of halogen species and chemistry over the continents in the recent past. It is noteworthy that we simulated a substantial increase in inorganic halogens levels due to CIH sources in economically emerging regions (Table S3), e.g., South America (219.1% for Cly), Africa (223.9% for Cly), and South Asia (185.8% for Cly, 149.5% for Bry, and 77.6% for Iy), where there are no reports on field observations of inorganic halogens. Effects of CIH on air pollutants The significant increase in inorganic halogens across continents caused widespread impacts on air pollutants, particularly the secondary pollutants, with large spatial heterogeneity (Fig. 2 and S3). Averaged among all simulated periods (1980, 1990, 2000, and 2015; Table S1), CIH overall induced decreasing effects on the surface O3 over the global continents by −0.2 ppbv or −0.84% (Table S2, 3), with isolated increases ( ~1%) in East Asia and Western Europe. The large-scale decrease in surface O3 was mainly due to the direct loss of O3 via halogen atom reactions with O3 molecules (Eq. 1) and the indirect loss of O3 via NOx reduction in NOx- and mixed-limited O3 regime regions where reducing NOx leads to O3 loss (Eqs. 3–6 and Fig. S3); the increase in O3 was mainly due to O3 production via VOCs oxidation by halogen atoms (Eqs. 9–14) and NOx reduction in VOCs-limited regions where reducing NOx results in O3 increase (Eqs. 3–6 and Fig. S3). On the contrary, CIH enhanced the formation of secondary aerosols, including a global 1.35% increase in secondary inorganic aerosols (sulfate, nitrate, and ammonium, hereafter SNA) and a 0.47% increase in secondary organic aerosols (SOA), with largest enhancements ( ~10% in SNA and ~20% in SOA) occurring over China and Europe (Fig. 2). The increase in SNA and SOA concentrations was caused by the transformation of NO2 to HNO3 (and nitrate aerosol) by halogen oxides (Eqs. 5–6) as well as the CIH-induced increases in the oxidants, including OH radical (Fig. S4) and Cl and Br atom and their precursors (Figure S1) that enhanced the oxidation of the precursors of secondary aerosols (Eq. 11 and Eq. 16). The CIH-induced changes in the total fine aerosol (PM2.5) are similar to those in SNA and SOA in terms of spatial pattern but with a smaller magnitude (Fig. 2e). Simulated continental inorganic halogen-induced relative changes (%) in key air pollutants averaged in all simulated periods. a NOx, b O3, c sulfate, nitrate, and ammonium (SNA), d secondary organic aerosol (SOA), and e PM2.5 (particulate matter with an aerodynamic diameter ≤ 2.5 μm) in the surface layer averaged from 1980 to 2015. Simulated levels of these species in the noCNT and FULL cases, as well as the continental inorganic halogen-induced absolute changes, are shown in Fig. S3 in the Supporting Information. We also investigated the temporal variation of CIH impacts on air pollutants (Fig. 3 and Figs. S5–9) under the changing climate and emissions of both routine air pollutants and halogens. Globally, CIH-induced effects on air pollutants did not change sign (keeping negative impacts on gaseous pollutants and positive impacts on particulate matters) but decreased in magnitudes from 1980 to present day, e.g., impact on NOx from −3.9% to −1.5%, impact on O3 from -0.8% in 1980 to −0.7% in 2015, impact on SNA from 1.6% to 1.2%, and that on SOA from 0.7% to 0.3%, and that on PM2.5 with smaller variation between 0.10% to 0.15% (Fig. 3a). Regarding the 6 selected regions, the CIH impacts showed greater complexity than in the global results. The CIH impacts in North America and Europe generally decreased in magnitude from 1980 to present-day; those in South America and Africa are maintained in limited scales; those in South Asia showed an increasing trend; while those in East Asia increased in magnitude from 1980 to 2000 but kept at similar levels afterwards (Fig. 3b–g). For instance, the regional average CIH impacts on NOx in Europe went from −10.3% in 1980 to −1.4% in 2015 (Fig. 3d), and those on SOA in South Asia expanded from 0.2 to 4.2% (Fig. 3f). The global CIH impact on air quality showed decreasing trends from 1980 to 2015, which might be attributed to the decreases in CIH impacts in Europe and N. America outweighing the increases in Asia, and the decreases in bromine effects outweighing the increases in chlorine effects. It is noteworthy that the maximum daily 8-hour average O3 (MDA8 O3) is more commonly used in the air quality community than the annual average O3; therefore, we output the hourly average data in the 2015 simulations for completeness. The results showed that the relative changes in MDA8 O3 in 2015 (Fig. S10d) are very similar to those in annual average O3 in 2015 (Fig. S6d). Global and regional average of continental inorganic halogen (CIH)-induced impacts (%) on air pollutants in 1980, 1990, 2000, and 2015. a Global, b North America, c South America, d Europe, e Africa, f South Asia, and g East Asia. The error bars represent the ranges (5th percentile and 95th percentile) of CIH impacts on each pollutant for the grid points in each region in each year. Note the different scale for the Y axis for each panel. The values for the higher limits (95th percentile) of CIH impacts on SOA are shown in each panel. SNA denotes sulfate, nitrate, and ammonium; SOA represents secondary organic aerosol; PM2.5 stands for particulate matter with an aerodynamic diameter ≤ 2.5 μm. The spatial distributions of CIH impacts on individual air pollutants in the years 1980, 1990, 2000, and 2015 are shown in Figs. S5–9. The monthly variations of CIH impacts on each air pollutant in selected cities in each region in each year are shown in Figs. S11–16. We note that CIH induced much more substantial changes in air pollutants at individual locations than in global and regional averaged changes. The error bars in Fig. 3 represent the ranges of CIH impacts for the grid points in each region and each year and instead of showing the minimum and maximum changes, we use 5th and 95th percentiles to avoid extreme values. For instance, the global impacts of CIH on NOx extended well beyond the global average to -8.0%, that on O3 to -2.3%, that on SNA to 1.8%, that on SOA to 36.7%, and that on PM2.5 to 1.1%, all occurring in 1980 (Fig. 3a). Similarly, more extensive impacts on air pollutants can be found for each region (Fig. 3b–g). Zooming into the typical cities in each region in each month (Figs. S11–16), we find that CIH impacts on air quality present clear seasonal variations (typically peaking in winter seasons) with magnitudes much larger than the annual and regional average, particularly those cities with elevated emissions in the past decades, e.g., New York, Paris, New Delhi, and Beijing. We also conduct a sensitivity case with only continental inorganic chlorine emissions (CNT-Cl; Table S1) that allows to separate in intermediate steps the impacts of continental inorganic chlorine (changes between noCNT and CNT-Cl) from those of continental inorganic bromine and iodine (changes between CNT-Cl and FULL). We find that while continental inorganic chlorine predominantly decreases NOx levels and increases the atmospheric abundances of O3, SNA, SOA, and PM2.5 (Fig. S17), continental inorganic bromine and iodine induce more complex changes in these species (Fig. S18) that are similar, in terms of spatial pattern, to the overall CIH impacts (Fig. 2). Integrating over continents, continental inorganic chlorine also causes less noticeable changes (Table S4) than continental bromine and iodine (Table S5). These analyses suggest that to fully understand the CIH impacts on atmospheric compositions, in terms of both distribution and magnitude, the continental bromine and iodine contribution to global CIH emissions must be taken into account. Model uncertainty analysis Our previous works (ref. 3 and the references therein) comprehensively analyzed the model uncertainty and evaluated the performance of the CAM-Chem model in reproducing the ambient levels of halogens and their impact on atmospheric compositions. Although some sources of uncertainties remain in the halogen chemical mechanisms, particularly the multiphase reactions, the CAM-Chem model is consistently capable of providing realistic estimates of the ambient atmospheric constituents critical to the current paper, including reactive halogens (e.g., ClNO2, BrO, IO), tropospheric oxidants (e.g., OH), and air pollutants (e.g., O3). The advancement of the current study lies in the newly compiled historical CIH emissions and their associated impacts on the evolution of global air quality. Therefore, here we primarily quantify the uncertainties associated with CIH emissions. Our estimate of continental inorganic halogen emissions was subject to uncertainty due to incomplete activity data and emission factor estimates. Following our previous global study26, we quantified the uncertainty in our Cl emission estimates by conducting a Monte Carlo analysis with 10,000 simulations, where input parameters were randomly sampled from their respective probability distributions. Uncertainty in activity data was characterized by normal distributions, with coefficients of variation (CV) of 5% for coal consumption in power plants and industry, 20% for coal and biomass consumption in domestic sectors, 50% for biomass and waste by open burning. The technology split was represented by a uniform distribution with a CV of 50%. Cl content followed a log-normal distribution with a CV of 100%. The uncertainties of other factors (e.g., release rate, removal efficiency, fine particle Cl percentage, etc.) were set as our previous study27. Overall, the uncertainties of our global Cl emission estimation were estimated to be within the range from −50 to +100% at a 95% confidence interval (specifically, −37 to 64% for 1980, −38 to 69% for 1990, −41 to 66% for 2000, and −44 to 74% for 2015). Quantifying the uncertainty in Br and I emissions is more challenging due to the limited number of measurements of their emission characteristics from continental sources. In this study, we considered only emissions from coal combustion and used the highest measured Br or I content in coal for each region to account for other potential, unknown sources. Here, we assume the same uncertainty ranges (−50% to +100%) for continental Br and I emissions as for continental Cl emissions. Based on the uncertainty range of CIH emissions, we conducted sensitivity simulations using 50% and 200% of the proposed CIH emissions (0.5CIH case and 2.0CIH case; Table S1) for all simulation years (in total eight extra cases) and explored the resulting uncertainty in the CIH-induced impacts on air quality. Our sensitivity tests suggest that by halving and doubling the CIH emissions, the simulated impacts of CIH on reactive halogen levels (Fig. S19) are approximately 50% and 200% of those using the standard CIH emissions (Fig. S1). These CIH-induced changes in halogen levels subsequently cause perturbation in air pollutant levels (Figs. S20, 21, Table S6 and S7) at similar rate as the changes in halogen concentrations. These extra sensitivity simulation results show that even at the lower-limit emission, CIH still cause a substantial effect on the levels of reactive halogens as well as key air pollutants, highlighting the importance of expanding observational networks, particularly in developing regions. Discussion In the current study, as a report on continental inorganic halogen impacts on global air quality over the past four decades, we aim to provide a total impact and its spatiotemporal variation, along with a corresponding uncertainty analysis. We adopted the latest pollutants emissions and chemical mechanisms to represent the evolution of CIH sources and atmospheric fate, and our validation results suggested that although our model setup was able to reproduce the magnitude of the inorganic halogens observed in continental regions (Fig. S2), the simulated halogen mixing ratios were still lower than the observed levels, possibly due to missing direct emissions, secondary production, or recycling processes in the model. We note that if these missing terms are confirmed and quantified, e.g., via future research developments, the CIH impacts on air quality are expected to be even larger than those presented here. Such a possibility has also been proposed in a recent study on the global impact of chlorine species for the year 201435. Ocean-originated halogens, enhanced by anthropogenic air pollution, have been proposed to reduce the levels of key air pollutants, such as O3, throughout the globe, via the so-called buffering effects of halogens36. However, the present study showed that the CIH increased surface O3 mixing ratios in the polluted regions characterized by elevated NOx and VOCs emissions. This finding suggests that, contrary to the buffering effects of oceanic halogens on pollutants, CIH has an amplifying effect on air pollution. In addition, this work also shows a significant change in secondary aerosols caused by CIH, which have not been accounted for in previous global studies. Another point worth mentioning is that although our modeling setup is based on the latest research findings, noticeable uncertainty exists in the emission inventory and halogen chemical mechanism (Model uncertainty analysis section). Here, we call for future studies to further refine the CIH emissions using local emission factors, to fully consider the impacts of halides on thermodynamic equilibria of ions in aerosols with the respective gases, and to elucidate the multiphase halogen mechanisms through field-derived and laboratory-based findings. The final discussion point is about the spatio-temporal shift of the hot spots of CIH impacts on halogen abundance and air pollution in the past decades (Fig. 4). Following the changes in economic sectors in each region, CIH impacts peak in Europe in the 1980s and 1990s, and change to East Asia in the 2000s, and more recently shift to South Asia. We anticipate that, with the ongoing migration of manufacturing industries from East Asia to South Asia, Southeast Asia, Central America, South America, and Africa, the hotspots of CIH impacts will follow. The potential for larger CIH impacts on air quality in these economically emerging and highly populated regions warrants future efforts to monitor ambient levels of short-lived halogens, estimate local CIH emissions, investigate halogen chemistry, and evaluate CIH-induced impacts on air pollution. The variation of continental inorganic halogen emission impacts on inorganic halogen abundance and air quality. The inserted figure illustrates the continental inorganic halogen impacts on SOA (secondary organic aerosol) in percentage in each region for the years 1980, 1990, 2000, and 2015. Please refer to Fig. S1 for the region ranges. Methods CAM-Chem simulations The Community Atmosphere Model with Chemistry (CAM-Chem) is the atmosphere module in the Community Earth System Model (CESM), and the formation mechanisms of key air pollutants in CAM-Chem are described in previous literature37. Briefly, atmospheric oxidants (OH, O3, and NO3) oxidize VOCs, SO2, and NOx to form oxidized organic gases, sulfuric acid, and HNO3 that subsequently produce O3, sulfate, nitrate, ammonium, and SOA. In the current study, we employed the CAM-Chem model coupled with comprehensive reactive halogen chemistry3 to evaluate the impacts of CIH emissions on global air quality. The reactive halogen chemistry and its connection to key pollutants, i.e., NOx (NO and NO2), O3, SNA, and SOA, considered in CAM-Chem have been summarized in the Introduction section. Notably, halogen chemistry on one hand directly perturbs the budget of air pollutants, and on the other hand induces significant impacts on the levels of other atmospheric oxidants (OH, O3, and NO3), subsequently influencing the formation and loss of air pollutants. In addition to secondary aerosols (SNA and SOA), the total fine particles (PM2.5) also include primary aerosols, i.e., black carbon, organic carbon, dust, and sea-salt aerosols. Atmospheric concentrations of primary aerosols are driven by physical processes (emissions, transport, and depositions) but not chemistry; hence in the current study, we mainly explore the role of halogen chemistry in the formation of secondary pollutants (O3 and secondary aerosols). Halogen impacts on air pollutants are primarily driven by inorganic rather than organic halogens. Therefore, we focus on the changes in total inorganic halogens caused by CIH emissions. Total inorganic chlorine (Cly), bromine (Bry), and iodine (Iy) are defined as below: We conducted CAM-Chem simulation cases with (FULL) and without (noCNT) continental inorganic halogen emissions for the years of 1980, 1990, 2000, and 2015, 8 simulation cases in total (Table S1). Each case was run with an additional 12-month spin-up period. The difference in atmospheric compositions between the noCNT and FULL cases represents the impacts of CIH. We also conducted sensitivity simulations with only continental inorganic chlorine (CNT-Cl) to isolate the impacts of continental inorganic chlorine on atmospheric compositions by comparing the noCNT and CNT-Cl results and to quantify the impacts of continental inorganic bromine and iodine through comparing the CNT-Cl and FULL cases. To analyze the influence of uncertainty in the newly proposed CIH emissions on the simulated results, we further ran sensitivity cases with half (0.5CIH) and twice (2.0CIH) CIH emissions. By comparing noCNT with 0.5CIH as well as noCNT with 2.0CIH, we estimated the possible lower limit and upper limit of the CIH impacts on the atmospheric abundance of key constituents. All CAM-Chem simulations were conducted with a spatial resolution of 0.9° × 1.25° and 56 vertical layers using the specified dynamic setup, in which offline meteorological data are used to drive the dynamic evolution. Global meteorological dataset from the Modern-Era Retrospective Analysis for Research and Applications Version 2 (MERRA2; https://gmao.gsfc.nasa.gov/reanalysis/MERRA-2/docs/) was used to constrain the model dynamics for all simulation years. Global emission data for routine air pollutants, such as CO, VOCs, NOx, SO2, and NH3, in each simulation year from the Community Emissions Data System (CEDS)21 were adopted in this study. All simulation cases have the same long-lived halogen emissions (e.g., CFC), continental organic halogen emissions (e.g., CH2Cl2), oceanic halogen emissions (e.g., CHBr3, HOI), as well as halogen precursor sources (e.g., sea-salt aerosol). We have not included halogen emissions from waste incineration plants. Stricter control measures are required in the waste incineration plants than in industrial sectors, in order to control the emissions of toxic substances (e.g., dioxin), thus leading to small chlorine emissions. In our previous study focusing on China27, the estimated chlorine emissions from the waste incineration plants account for less than 5% of the chlorine emissions from open waste burning. Meanwhile, to the best of our knowledge, long-term global data on burned waste amounts at waste incineration plants are not available. Therefore, we do not include halogen emissions from waste incineration facilities, and we believe that this omission will not alter our conclusion regarding the role of continental inorganic halogens in global air quality. Reactive halogen emissions from chemical industry38,39 and disinfectants production/usage40 have been excluded from our compiled global continental inorganic halogen emission inventory, due to limited activity data on global scale. However, these emissions of reactive halogens could lead to noticeable changes in atmospheric chemistry and air quality at local to city scales, and future work on the roles of halogens in air quality in cities with strong reactive halogen emissions from industrial activity and disinfectant production/use should take these emissions into account. Emission and model validation We compiled continental inorganic chlorine, bromine, and iodine emissions for the last few decades, while there have been two previous reports on historical chlorine emissions. Our chlorine emissions agree well with the more recent report by Zhang et al.25, hereafter Z2022, but are lower than the report by Keene et al.24, hereafter K1999. For the year (1990) that all three studies provided chlorine emissions, we estimated 4019 kt of chlorine, Z2022 reported a similar amount of 4781 kt chlorine, while K1999 reported a much larger emission of 12900 kt chlorine. Lower chlorine emissions from our study and Z2022 are mainly caused by the different calculation methods. For biomass burning, K1999 estimates the inorganic chlorine emissions based on a biomass-chlorine-content approach, where we (and Z2022) use measured HCl or particle Cl emission factors; for coal burning, K1999 estimates the flue-gas-desulfurization application rate based on a previous estimation in several countries, where we (and Z2022) use more recent studies with information from more countries. In principle, our newly proposed emission inventory and Z2022 represent a more realistic estimate of the chlorine emissions than K1999. The CAM-Chem model used in the current study has been intensively validated with surface and aircraft measurements in recent studies3,18,41,42, which show that the CAM-Chem model with halogen chemistry is able to reproduce the key features of global atmospheric compositions, including oxidants (e.g., OH), air pollutants (e.g., O3), greenhouse gases (e.g., CH4), sea-salt aerosols, as well as inorganic and organic halogen species. Here, we further evaluate the CAM-Chem model performance in inorganic halogen species over the polluted environments (Fig. S2). The consideration of CIH emissions (FULL case) induced global increases in simulated reactive halogen levels over the continents compared to the noCNT case. Depending on whether the noCNT results compare favorably with the observations (overestimate) or unfavorably (underestimate), considering CIH emissions could improve (reducing the deviation) or degrade (enlarging the deviation) the model performance. Here, we adopt the concept of the improvement rate (Ir), defined as Eq. 24, in which CnoCNT is the simulated concentration in the noCNT case, Cobs is the observed concentration, Cfull represents the simulated concentration of halogens in the FULL case, and abs is used to calculate the absolute value. When Ir is larger than 0, the deviation of the noCNT simulation from the observation is larger than that of the FULL case, meaning that including CIH emissions improves the model performance; when Ir is lower than 0, the deviation of the noCNT simulation is smaller than that of the FULL case, meaning that including CIH emissions deteriorates the model performance. By including CIH emissions, CAM-Chem model performance improved at 89% (48 of the 54 sites) locations with the greatest improvement rate of 77.4%; large improvements (on average 16.8%) occurred in East Asia, while the improvements in West Europe and North America were mostly below 10%, which are in line with the larger CIH emissions in East Asia than in West Europe and North America in the recent decade (when most field observations were collected). Overall, our results suggest that our model setup is able to reproduce the spatial difference in atmospheric halogen levels induced by varied CIH emissions across the globe, and considering CIH (FULL case) is critical for reproducing the observed inorganic halogens over continental regions, particularly in those regions with large CIH emissions. We also collected ground-level continental observations of O3 and PM2.5 from U.S. (https://www.epa.gov/outdoor-air-quality-data/download-daily-data), Europe (https://eeadmz1-downloads-webapp.azurewebsites.net), and China (https://air.cnemc.cn:18007), as well as Tropospheric Ozone Assessment Report (TOAR; https://toar-data.org/) datasets for the year 2015, to validate the model performance on air pollutants. The direct comparisons of observed and simulated O3 and PM2.5 are shown in Figures S22 and S23, which show that both model cases (FULL and noCNT) are capable of simulating the magnitude and variation of MDA8 O3 and PM2.5 across regions; averaging in each region, FULL case slightly improves the model performance in MDA8 O3 in Europe and U.S. and PM2.5 in China and U.S., but slightly decreases the performance in MDA8 O3 in China and PM2.5 in Europe (Table S8). Similar to the impacts of CIH emissions on halogen simulations, considering CIH emissions in the CAM-Chem model induced diverse changes in O3 and aerosols (as shown in Figs. 2 and 3 and discussed in the main text). Here we also calculate the improvement rate for each month-site data point, suggesting that the consideration of CIH in CAM-Chem model improved the performance in some month-sites, but degrade in others, and overall CIH emissions reduced the model bias for O3 and PM2.5 at most of the observational sites ( ~60%; Table S9), suggesting that CIH emissions should be included in reproducing the spatial distribution and temporal evolution of air quality. Supplementary information Supplementary Information Transparent Peer Review file