The influence of ammonia emission inventories on size-resolved global atmospheric aerosol composition and acidity

Ammonia (NH₃) is an abundant alkaline gas in the atmosphere and a key precursor in the formation of particulate matter. While emissions of other aerosol precursors such as SO₂ and NO_x have decreased significantly, global NH₃ emissions are stable or increasing, and this trend is projected to continue. This study investigates the impact of NH₃ emission changes on size-resolved aerosol composition and acidity using the atmospheric chemistry–climate model EMAC (ECHAM5/MESSy Atmospheric Chemistry). Rather than directly perturbing NH₃ emissions, we employ three distinct emission inventories: two bottom-up inventories and one derived using an updated top-down method. The results reveal that sulfate–nitrate–ammonium aerosols in two fine-mode size ranges (0–1 and 1–2.5 µm) show the greatest sensitivity to NH₃ emission changes. Regional responses vary depending on the local chemical environment of secondary inorganic aerosols. In “NH₃-rich” regions (e.g., East Asia and Europe), the abundance of NH₃ partially offsets the effects of reduced NH₃ emissions when NO_x and SO₂ are available, especially for aerosols in the 1–2.5 µm range. This underscores the importance of coordinated control strategies for NH₃, NO_x, and SO₂ emissions. Further, we find that NH₃ has a buffering effect in densely populated areas, maintaining aerosol acidity at moderate levels and mitigating drastic pH shifts. While pH changes correlate strongly with NH₃ variability, they are also influenced by concurrent changes in SO₂ and NO_x emissions. These results highlight the critical role of NH₃ in shaping aerosol acidity, arguing for size-specific approaches to managing particulate matter.

1 Introduction

As an abundant alkaline gas in the atmosphere, ammonia (NH₃) acts as a precursor in the formation of particulate matter by neutralizing atmospheric acids (e.g., H₂SO₄, HNO₃) to form sulfate–nitrate–ammonium (SNA) aerosols (Li et al., 2018; Chen et al., 2016; Wang et al., 2013), which are the main secondary inorganic components of PM_2.5 (particulate matter with a diameter of 2.5 µm or less). By condensing onto freshly nucleated particles, NH₃ enhances the growth rate of new particles as well as their hydrophilicity (Wang et al., 2020; Li et al., 2018). This can degrade air quality and change the solar radiative balance by interaction with radiation and clouds (Che et al., 2009; Zhao et al., 2011; Yao et al., 2018). In addition, the conjugate base–acid pair NH₃ $/$ NH${}_{\mathrm{4}}^{+}$ acts as the major buffer that inhibits changes in aerosol acidity (Chen et al., 2019; Zheng et al., 2020; Karydis et al., 2021).

Anthropogenic emissions are the main source of atmospheric NH₃, with an average global contribution of 76 %, dominated by agricultural activities, including livestock farming and the fertilization of soils (Schlesinger and Hartley, 1992; Dentener and Crutzen, 1994; Bouwman et al., 1997; Olivier et al., 1998; Van Aardenne et al., 2001; Sutton et al., 2013). Meanwhile, the importance of non-agricultural sources, such as industrial emissions and fossil fuel combustion has been highlighted by studies focusing on severe haze episodes in East Asia (Chang et al., 2019; Liu et al., 2018a; Pan et al., 2016). The construction of high-resolution regional or global datasets has characterized the spatiotemporal patterns of NH₃ emissions, with the most common datasets consisting of bottom-up inventories and top-down modeling inversion methods. Bottom-up emission inventories rely on activity data and emission factors, the latter being sensitive to assumptions about fertilizer types, local soil, and meteorological properties (Bouwman et al., 2002; Søgaard et al., 2002; Xu et al., 2024). Zhang et al. (2018) evaluated discrepancies between bottom-up NH₃ emission inventories to exceed a factor of 2 due to the uncertainties in emission factors, meteorological properties, and agricultural statistics (Beusen et al., 2008). Crippa et al. (2018) pointed out that the uncertainty in estimated NH₃ emissions is largest among all pollutants in the Emissions Database for Global Atmospheric Research (EDGAR v4.3.2), with a range of variation from 186 % to 294 % in 2012. In contrast, the ability of satellites to measure NH₃ abundance combined with numerical simulations allows a better characterization of the spatial distribution and seasonality of NH₃ emissions, but the low signal-to-noise ratio over low-emission areas limits the accuracy of retrieval products (Morán et al., 2016; Xu et al., 2016; Kong et al., 2019). Using a hybrid inverse modeling approach, Chen et al. (2021) optimized the NH₃ emission inventory (NEI) over the United States by combining CMAQ model simulations with constraints from the Infrared Atmospheric Sounding Interferometer (IASI) retrieval product. They found a 26 % low bias of NH₃ emissions in the NEI, and the optimized NH₃ emission inventory improved the model performance of PM_2.5 mass concentration in the Midwest USA, with the normalized mean bias of NH${}_{\mathrm{4}}^{+}$ and NO${}_{\mathrm{3}}^{-}$ decreasing from 27 % to 22 % and from 64 % to 55 %, respectively.

Recently, the emissions of SO₂ and NO_x have decreased due to the implementation of related clean-air policies in East Asia, North America, and Europe (Zheng et al., 2018; Hand et al., 2012; Russell et al., 2012; Aas et al., 2019; Gong et al., 2024), while the emissions of NH₃ have remained stable or have slightly increased in major agricultural regions such as China (Xu et al., 2016), the United States (Yu et al., 2018), and Europe (Fortems-Cheiney et al., 2022). The increase in NH₃ emissions is associated with increasing fertilizer use and local temperatures (Warner et al., 2017; Xu et al., 2016; Skjøth and Geels, 2013; Xu et al., 2019), and such trends are expected to continue on a global scale throughout the century (IPCC, 2014). Driven by such emission trends of SO₂, NO_x, and NH₃, the main composition of SNA has changed in large regions with a shift from an ammonium sulfate to an ammonium nitrate formation regime (Tsimpidi et al., 2024; Lei et al., 2021; Jo et al., 2020; Zhou et al., 2019; Shah et al., 2018; Wang et al., 2013; Hauglustaine et al., 2014). Li et al. (2017) evaluated that during the period 1989 to 2013, the increasing trend of sulfate (SO${}_{\mathrm{4}}^{\mathrm{2}-}$) and ammonium (NH${}_{\mathrm{4}}^{+}$) mass concentration in India and China occurred at a rate > 0.1 $\mathrm{\mu }\mathrm{g}{\mathrm{m}}^{-\mathrm{3}}{\mathrm{yr}}^{-\mathrm{1}}$, while decreasing trends were found in North America and Europe at a rate of about 0.1 $\mathrm{\mu }\mathrm{g}{\mathrm{m}}^{-\mathrm{3}}{\mathrm{yr}}^{-\mathrm{1}}$. The reduction in SO₂ and NO_x emissions decreases the atmospheric availability of H₂SO₄ and HNO₃, thereby reducing the demand for NH₃ to neutralize these acids and form particulate NH${}_{\mathrm{4}}^{+}$. As NH₃ emissions have remained stable or slightly increased, this leads to an accumulation of NH₃ in the gas phase (Liu et al., 2018b). Warner et al. (2017) assessed that the average increase rate of NH₃ mass concentration from 2002 to 2016 in the United States, Europe, and China was 2.6 % yr⁻¹, 1.8 % yr⁻¹, and 2.3 % yr⁻¹, respectively. Some studies further pointed out that the increase in NH₃ concentration may offset the effectiveness of PM_2.5 control achieved via SO₂ and NO_x emission reduction by promoting the formation of nitrate (NO${}_{\mathrm{3}}^{-}$) (Huang et al., 2021; Cai et al., 2017; Zhang and Geng, 2019; Fu et al., 2017).

Several studies concluded that reducing NH₃ emissions would be a cost-effective way to control PM_2.5 concentrations (Gu et al., 2021; Tsimpidi et al., 2007; Erisman and Schaap, 2004). However, the response of SNA to changes in its precursors is not linear (Pozzer et al., 2017; Wang et al., 2011, 2013; Tsimpidi et al., 2007; West et al., 1999) because the gas–particle partitioning of NH₃ $/$ NH${}_{\mathrm{4}}^{+}$ and HNO₃ $/$ NO${}_{\mathrm{3}}^{-}$ is influenced by several parameters, such as temperature, liquid water content, and aerosol acidity (Xu et al., 2020; Nenes et al., 2020; Guo et al., 2018). Nenes et al. (2020) developed a conceptual framework to describe the sensitivity of particulate matter to NO_x and NH₃ emissions, highlighting the critical influence of aerosol acidity and liquid water content on particulate matter formation. Based on sensitivity tests, Guo et al. (2018) evaluated that the response of NO${}_{\mathrm{3}}^{-}$ to NH₃ reduction shows an apparent decrease only when the aerosol pH falls below the value of 3.

On the other hand, reductions in SO₂ and NO_x emissions are expected to reduce aerosol acidity, but recent studies revealed that aerosol acidity does not decrease as expected (Chen et al., 2019; Guo et al., 2017b; Zheng et al., 2022; Karydis et al., 2021). Aerosol acidity affects many processes involving the atmosphere and various aspects of the Earth system (Pye et al., 2020; Tilgner et al., 2021; Karydis et al., 2021) as well as human health (Dockery et al., 1993, 1996; Thurston et al., 1994; Spengler et al., 1996). Weber et al. (2016) found that the acidity of PM_2.5 in the southeastern United States has remained at a relatively constant level with a pH value of 0–2 over the past 15 years, despite a 70 % reduction in SO${}_{\mathrm{4}}^{\mathrm{2}-}$ concentration. A lack of aerosol acidity trend was also reported in China (Zhou et al., 2022). This is mainly caused by the buffering effect of NH₃ $/$ NH${}_{\mathrm{4}}^{+}$ (Chen et al., 2019; Zheng et al., 2022; Zheng et al., 2020). To investigate this, Song et al. (2019) derived an equation from the partitioning of NH₃ and estimated that a unit increase in pH requires a 10-fold increase in NH₃ concentration, which is consistent with the findings of Guo et al. (2017b). The regional variation in aerosol acidity is considerable, and the pH of PM_2.5 in northern China is in the range of 4 to 5, which is higher than in Europe and the United States (Shi et al., 2017, 2019; Liu et al., 2017; Guo et al., 2015, 2016, 2017a; Karydis et al., 2021). This is caused by multiple driving factors, including aerosol mass concentration and composition, NH₃ mass concentration, aerosol water content, and meteorological factors (Ding et al., 2019; Zhang et al., 2021a). However, the main driver of the difference in aerosol acidity remains controversial. Zheng et al. (2020) pointed out that aerosol water content is the most important factor causing the regional variation in aerosol acidity, while Zhang et al. (2021a) emphasized the equal importance of aerosol mass concentration and chemical composition for the aerosol acidity contrasts between China and the United States.

Almost all recent studies that discuss the response of aerosol composition and acidity to changes in NH₃ emission trends focus on the fine mode (e.g., PM_2.5). The size-resolved composition of SNA is not uniform (Karydis et al., 2016; Fang et al., 2017; Karydis et al., 2011, 2010; Guo et al., 2017a), and NH${}_{\mathrm{4}}^{+}$ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ are mainly concentrated in the fine mode (Wang et al., 2012; Seinfeld and Pandis, 2016), while NO${}_{\mathrm{3}}^{-}$ aerosol can be formed on the surface of supermicron particles via heterogeneous chemistry (Allen et al., 2015). Furthermore, Milousis et al. (2024) revealed that the acidity of fine-mode aerosol is more sensitive to NH₃ emission than coarse-mode aerosol. Reducing the NH₃ emissions by half, the simulated pH of fine- and coarse-mode aerosol decreased by up to 3 and 1.5 units, respectively. Aerosol acidity tends to decrease with increasing particle size, with pH varying by up to 6 units (Craig et al., 2018; Fang et al., 2017; Bougiatioti et al., 2016). Size-resolved aerosol acidity is associated with different formation pathways (Tilgner et al., 2021; Cheng et al., 2016). Ding et al. (2019) found that the coarse-mode aerosol shifted from neutral to weakly acidic with the increase in NO${}_{\mathrm{3}}^{-}$ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ during severe hazy days. Cheng et al. (2016) further pointed out that the dominant oxidant in SO${}_{\mathrm{4}}^{\mathrm{2}-}$ production by SO₂ oxidation changes with the ambient aerosol acidity. Therefore, it is necessary to comprehensively investigate the response of size-resolved chemical composition and acidity to changes in NH₃ emissions.

In this study, we investigate the influence of NH₃ emissions on size-resolved aerosol composition and acidity using the EMAC (ECHAM5/MESSy Atmospheric Chemistry) model. The EMAC model, coupled with the ISORROPIA II thermodynamic module, provides a robust framework for simulating global aerosol–chemistry–climate interactions and estimating aerosol pH under varying environmental conditions. Rather than directly perturbing NH₃ emissions, we employ three distinct emission inventories: two bottom-up inventories (CAMS and CEDS_GBD) and one derived using an updated top-down approach (Luo et al., 2022). These scenarios allow us to assess the sensitivity of size-resolved SNA aerosols to realistic variations in NH₃ emissions, in conjunction with co-varying SO₂ and NO_x trends. Model results are evaluated against satellite-derived NH₃ columns and ground-based aerosol composition observations. Our analysis spans four aerosol size ranges (0–1, 1–2.5, 2.5–5, and 5–10 µm) and includes regional assessments over North America, Europe, and East Asia to explore chemical regime shifts and buffering capacity.

2 Modeling description

EMAC (ECHAM5/MESSy) is a global atmospheric chemistry and climate model that includes a number of submodels describing atmospheric processes and interactions among oceans, land, and anthropogenic influences (Jöckel et al., 2016). These submodels are linked to the base model, the fifth-generation European Centre Hamburg general circulation model (ECHAM5; Roeckner et al., 2006), via the Modular Earth Submodel System (MESSy; Jöckel et al., 2005). In this study, the horizontal resolution of the EMAC model is T63L31, which corresponds to a grid resolution of about 1.875° × 1.875° (Jöckel et al., 2010) and 31 vertical layers extending up to 25 km altitude. EMAC is applied for 4 years, from 2009 to 2012 with the first year used as spin-up. The ERA5 meteorological reanalysis data (Hersbach et al., 2020), with a significantly enhanced horizontal resolution of 31 km and hourly output throughout, are used in EMAC to nudge the simulation.

In EMAC, organic aerosol (OA) formation is simulated by the ORACLE module (Tsimpidi et al., 2014, 2018), where logarithmically spaced saturation concentration bins are used to describe the organic aerosol components based on their volatility. The aerosol microphysics and gas–aerosol partitioning are calculated by the Global Modal-aerosol eXtension (GMXe) module described by Pringle et al. (2010), which has the same microphysical core as the M7 submodel (Vignati et al., 2004). The aerosol size distribution is treated by 7 lognormal modes, including 4 hydrophilic and 3 hydrophobic modes, covering nucleation (soluble only), Aitken, accumulation, and coarse modes (both soluble and insoluble). To determine size-resolved aerosol composition and pH, we sum the contributions of each aerosol component, water content, and H⁺ concentration across all GMXe modes corresponding to a given size range. This is achieved by calculating the volume fraction of the lognormal distribution of each mode that falls within the specified size limits. The atmospheric chemistry module MECCA (Module Efficiently Calculating the Chemistry of the Atmosphere), which contains a comprehensive atmospheric reaction, is used to calculate the gas concentrations (Sander et al., 2019). The SEDI module is used to compute aerosol particle sedimentation (Kerkweg et al., 2006). Dry deposition and wet deposition of gas and particle species are calculated by the DRYDEP module (Kerkweg et al., 2006) and the SCAV module (Tost et al., 2006), respectively. The CLOUD submodel (Roeckner et al., 2006) is used to calculate cloud properties and microphysics, utilizing the microphysical scheme of Lohmann and Ferrachat (2010) and a physically based treatment of liquid (Karydis et al., 2017) and ice crystal (Bacer et al., 2018) activation processes.

An advanced parameterization scheme is incorporated into the EMAC model to calculate the dust emission flux online (Astitha et al., 2012). The scheme uses the online meteorological fields from the EMAC model, such as temperature, pressure, relative humidity, soil moisture, and surface friction velocity, to calculate the threshold friction velocity, which is the initial step of dust production. Above the threshold friction velocity, dust emission is possible. Following Karydis et al. (2016), the emissions of individual crustal species in this study are estimated as constant fractions of the dust emission (Klingmüller et al., 2018). These fractional factors depend on the geological information, which includes different dust emission sources. Karydis et al. (2021) pointed out that the crustal ions (Ca²⁺, Mg²⁺, K⁺, and Na⁺), especially Ca²⁺, have significantly contributed to maintaining the particle pH value at the level of 4.5–5 in East Asia during the last decade. The importance of crustal ions in determining aerosol acidity and factors such as liquid water content, aerosol mass concentration, and chemical composition, has been highlighted in other studies (Zheng et al., 2020; Zhang et al., 2021a; Ding et al., 2019).

2.1 NH₃ emission scheme

In this study, three NH₃ emission schemes are implemented in specific model simulation cases to assess the impact of NH₃ emission variability on aerosol composition and acidity (Table 1). These include two bottom-up inventories used in the base and CEDS cases and one top-down inventory applied in the Top-Dep case. In addition, two sensitivity simulations are conducted: the Meta case, which assumes metastable aerosol thermodynamics (described in detail in Sect. 2.2), and the noNH₃ case, in which all NH₃ emissions are turned off. The latter is used to provide insights into the role of NH₃ in shaping the simulated size-resolved aerosol acidity.

Table 1 Overview of simulation cases.

2.2 Bottom-up scheme

Bottom-up schemes are applied in the base and the CEDS case, with different anthropogenic emission inventories. The anthropogenic emission inventory utilized in the base case simulation is CAMS-GLOB-ANT (v4.2, https://eccad.sedoo.fr, last access: 8 September 2025, abbreviated as “CAMS”), which contains 17 sectors with a spatial resolution of 0.1° × 0.1° and monthly temporal resolution. With the basic data of 2010 from the Emission Database for Global Atmospheric Research (EDGAR, v4.3.2), CAMS extends the period to recent years based on the trend factors derived from the Community Emissions Data System (Hoesly et al., 2018). Meanwhile, Hoesly et al. (2018) pointed out that there are limitations to the system, especially in the emission trends for specific sectors, and emphasized the need for more detailed data to be incorporated into regional emission inventories. The emission inventory used in the CEDS sensitivity simulation case for aerosol precursor emissions (NH₃, SO₂, and NO_x) is CEDS_GBD-MAPS (McDuffie et al., 2020), abbreviated as “CEDS_GBD”. The CEDS_GBD is developed using the Community Emissions Data System and is reported as a function of 11 anthropogenic sectors and 4 fuel categories, with a spatial resolution of 0.5° × 0.5° and monthly temporal resolution. Assuming that the specific regional emission inventories are more accurate, McDuffie et al. (2020) updated the activity data and the core scaling procedure, modified the final emission gridding and aggregation procedures, and then utilized several regional emission inventories to improve the previous version of CEDS_GBD via the scaling procedure, which can not only reduce the discrepancy with other global emission inventories but also help to maintain the timeliness and regional accuracy of the global estimates. However, they also pointed out that the sources of uncertainty in the CEDS_GBD are similar to those in the CEDS.

Other NH₃ emissions include aircraft emissions from the CAMS-GLOB-AIR inventory (v1.1, https://eccad.sedoo.fr, last access: 8 September 2025, land and water biological emissions from the Global Emissions Inventory Activity (GEIA) inventory, and biomass burning emissions calculated by the BIOBURN submodel (Kaiser et al., 2012). BIOBURN determines the flux based on biomass burning emission factors and dry matter combustion rates from the Global Fire Assimilation System (GFAS), which calculates biomass burning emissions by assimilating fire radiative power (FRP) observations from MODIS (Andreae, 2019).

2.2.1 Top-down scheme

The NH₃ emission inventory over land is updated by a top-down method with the constraint of IASI satellite observations (https://iasi.aeris-data.fr/, last access: 8 September 2025 developed by Luo et al. (2022). This fast top-down method updates the prior NH₃ emissions ($E_{{\mathrm{NH}}_{\mathrm{3}},\mathrm{mod}}$, molec. m⁻² s⁻¹) with a correction term positively proportional to the biases between observed ($C_{{\mathrm{NH}}_{\mathrm{3},\mathrm{obs}}}$, molec. m⁻²) and simulated ($C_{{\mathrm{NH}}_{\mathrm{3},\mathrm{mod}}}$, molec. m⁻²) monthly averaged NH₃ total column densities and inversely proportional to the NH₃ lifetime (${\mathit{\tau }}_{{\mathrm{NH}}_{\mathrm{3}},\mathrm{mod}}$, s) (Eq. 1). The ${\mathit{\tau }}_{{\mathrm{NH}}_{\mathrm{3}},\mathrm{mod}}$ is calculated as the ratio of the simulated NH₃ column and the sum of the simulated loss rate of the NH_x family (NH_x ≡ NH₃ + NH${}_{\mathrm{4}}^{+}$) through dry and wet deposition (Eq. 2).

$$\begin{array}{}\text{(1)}& {\displaystyle }{E}_{{\mathrm{NH}}_{\mathrm{3}}}=E_{{\mathrm{NH}}_{\mathrm{3}},\mathrm{mod}}+{\displaystyle \frac{C_{{\mathrm{NH}}_{\mathrm{3}},\mathrm{obs}}-C_{{\mathrm{NH}}_{\mathrm{3}},\mathrm{mod}}}{{\mathit{\tau }}_{{\mathrm{NH}}_{\mathrm{3}},\mathrm{mod}}}}\text{(2)}& {\displaystyle }{\mathit{\tau }}_{{\mathrm{NH}}_{\mathrm{3}},\mathrm{mod}}={\displaystyle \frac{C_{{\mathrm{NH}}_{\mathrm{3}},\mathrm{mod}}}{D_{{\mathrm{NH}}_{\mathrm{3}},\mathrm{mod}}+D_{{\mathrm{NH}}_{\mathrm{4}}^{+},\mathrm{mod}}}}\end{array}$$

The fast top-down method relies on the total column concentrations retrieved by IASI. According to Dammers et al. (2019), the dominant source of uncertainty in the IASI observational product stems from the systematic bias, with the negative bias estimated to be between 25 % and 40 %, compared to site observations (Dammers et al., 2017). While this method simplifies the chemical and physical processes governing NH₃, Luo et al. (2022) identified large uncertainties in regions like Central Asia and tropical Africa due to poorly constrained sources by IASI observations in these areas. Nonetheless, they demonstrated that simulations driven by the updated top-down emission inventory show better consistency with satellite observations compared to those driven by the prior emission inventory.

2.2.2 Emission comparison

Overall, the global NH₃ emissions used in this study in all simulation cases range from 73 to 85 Tg yr⁻¹ (Table 2), which is within the reported range of the current literature from 52 to 91 Tg yr⁻¹ (Schlesinger and Hartley, 1992; Dentener and Crutzen, 1994; Bouwman et al., 1997; Olivier et al., 1998; Van Aardenne et al., 2001; Sutton et al., 2013). The distribution of the global NH₃ emission flux derived from the base case and the absolute emission flux difference between the sensitivity simulation cases and the base case are shown in Fig. 1. Significant regional variations in NH₃ emission flux are found in Fig. 1a, with the maximum emission flux exceeding 7 g m⁻² yr⁻¹ in northern India, eastern China, and central Europe, all regions with high population density. Other emission hotspots include the eastern United States, southeastern Latin America, and central and eastern Africa. The base case is able to capture the global NH₃ emission hotspots reported by Van Damme et al. (2018). Meanwhile, the minimum flux below 0.01 g m⁻² yr⁻¹ is located in the Antarctic and Arctic regions and the Sahara, as well as in remote oceans, where there is little impact of human activities. Agricultural activities including livestock and fertilization are the main source of NH₃ emissions in China, India, and the United States (Liu et al., 2022; Khan et al., 2020; Van Damme et al., 2018; Sahoo et al., 2024), while soil emissions, biomass burning, and domestic fires are the main contributors to NH₃ emissions in central and eastern Africa (Hickman et al., 2021; Delon et al., 2012).

Table 2 Comparison of annual NH₃ emissions (Tg yr⁻¹) across global and regional scales between the two sensitivity cases and the base case.

^a Absolute difference between sensitivity cases and the base case.
^b Relative difference between sensitivity cases and the base case.
^c Below the 0.1 precision threshold.

Figure 1 Global distribution of the annual average NH₃ emission flux for (a) the base case and the absolute differences between (b) the base and CEDS and (c) the base and Top-Dep cases during 2010–2012.

The global NH₃ emissions in both the CEDS and the Top-Dep cases exceed those in the base case, with increases of approximately 5 % and 16 %, respectively. The lower NH₃ emission fluxes of the CEDS case are found in North America, Europe, and China except for the northeastern and southeastern coastal areas, while emission fluxes are higher throughout India (Fig. 1b). Slightly higher fluxes are also found in western and eastern Africa, the northern Middle East, and southeastern Asia. According to McDuffie et al. (2020), the NH₃ emission flux from the Multi-resolution Emission Inventory for China (MEIC, http://www.meicmodel.org, last access: 8 September 2025, European Monitoring and Evaluation Programme (EMEP), and US Environmental Protection Agency (EPA) are used to scale the previous emission (the basic data of 2010 from EDGAR) over mainland China, Europe, and the USA. The NH₃ emission flux from India and Africa remains the same as that of the original inventory. Constrained by the IASI satellite observation, the NH₃ emission flux of the Top-Dep case is elevated in most regions of the world, but lower emission fluxes are estimated in regions such as western North America, western and southern Europe, India, China except the northeastern and southeastern coastal areas, and western Australia. A comparison of the absolute and relative difference between the base case and the other emission schemes is summarized in Table 2.

2.3 ISORROPIA II

The thermodynamic equilibrium model ISORROPIA II is used to calculate the multiphase mass transfer of the K⁺–Ca²⁺–Mg²⁺–NH${}_{\mathrm{4}}^{+}$–Na⁺–SO${}_{\mathrm{4}}^{\mathrm{2}-}$–NO${}_{\mathrm{3}}^{-}$–Cl⁻–H₂O aerosol system (Nenes et al., 1998; Fountoukis and Nenes, 2007). The process of gas–aerosol partitioning is calculated in two steps (Pringle et al., 2010). In the first step, the quantity of gas-phase species kinetically able to condense on the aerosol within one time step is calculated (Vignati et al., 2004). ISORROPIA II then re-distributes the mass between the gas and aerosol phase. In this study, ISORROPIA II runs in the forward mode with the input of relative humidity, temperature, and concentration of aerosol- and gas-phase species. ISORROPIA II determines the subsystem set of equilibrium equations and solves the equilibrium state by the chemical potential method. It outputs the equilibrium concentration of species in gas, solid, and liquid phases by assuming that the particle phase is in the thermodynamically stable-state mode where salts precipitate once the aqueous phase becomes saturated (Fountoukis and Nenes, 2007).

Meanwhile, a sensitivity case assuming the particle phase is in the thermodynamically metastable-state mode is performed with the same emission scheme as the base case (Meta case). In the metastable state, the aerosol may be supersaturated with respect to dissolved salts and always consists only of an aqueous phase (Fountoukis and Nenes, 2007). Karydis et al. (2021) pointed out that more acidic particles (up to 2 pH units) are derived from the metastable assumption in regions affected by high concentrations of crustal cations and consistently low relative humidity values.

According to past studies, the treatment of crustal species (e.g., Ca²⁺, K⁺, Mg²⁺) in ISORROPIA II can improve model predictions (Karydis et al., 2010, 2011), as both the phase partitioning of NO${}_{\mathrm{3}}^{-}$ and the thermodynamic interaction between NH${}_{\mathrm{4}}^{+}$ and the remaining ions in the aqueous phase are significantly affected. Karydis et al. (2016) found that when these crustal species are included in the EMAC model, the increase in global NO${}_{\mathrm{3}}^{-}$ tropospheric load can be up to 44 %, while the global NH${}_{\mathrm{4}}^{+}$ tropospheric load decreases by 41 %.

2.3.1 pH calculations

The pH is calculated from the negative decimal logarithm of the hydrogen ion activity:

$$\begin{array}{}\text{(3)}& \text{pH}=-{\log }_{\mathrm{10}}\left(\mathit{\gamma }{x}_{{\mathrm{H}}^{+}}\right),\end{array}$$

where $x_{{\mathrm{H}}^{+}}$ and γ represent the molality of hydrogen ions in the solution and the ionic activity coefficient of hydrogen, respectively. With γ assumed to be unity, the pH value is derived using the hydrogen ion concentration in the aqueous-particle-phase output by ISORROPIA II (in mol m⁻³) and the aerosol water content output by GMXe (in mol mol⁻¹). The effect of water-soluble organic aerosols on aerosol water content is accounted for via the GMXe module, which includes both inorganic and organic contributions. Organic aerosol formation is simulated using the ORACLE module, and the associated water uptake is calculated assuming a κ-hygroscopicity value of 0.14 for all organic components (Tsimpidi et al., 2017). This influences the total aerosol water content used in pH calculations. Both hydrogen ion and aerosol water content are output every 5 h, following Karydis et al. (2021). In addition, the temperature threshold of 269 K is set to ensure that the calculations are performed only when liquid water is present in the aerosol.

Discrepancies among thermodynamic models tend to increase as RH decreases, primarily due to differing assumptions about activity coefficients (Pye et al., 2020). ISORROPIA II assumes constant mean activity coefficients and a unity activity coefficient for H⁺, which results in relatively invariant and often lower pH predictions under dry conditions. Additionally, the choice between stable and metastable phase state assumptions introduces further uncertainty. Under low RH (e.g., < 35 %), the metastable assumption allows for supersaturated solutions, enabling pH calculations even at low aerosol liquid water content, often resulting in low pH values. Conversely, the stable state permits salt crystallization, and in cases where the aerosol becomes fully solid, pH may no longer be defined. To assess the sensitivity of our results to this assumption, we conducted a simulation using the metastable assumption (Meta case; see Table 1).

The EMAC model does not account for chemical interactions between ammonia and organic acids (e.g., ammonium oxalate formation) and treats inorganic and organic aerosol phases separately. As a result, while organics can indirectly influence pH through their contribution to aerosol water content, changes in NH₃ emissions do not affect organic aerosol formation or the associated water content in our simulations. Furthermore, although the influence of organics on hydrogen ion activity coefficients is not explicitly represented, previous studies have shown that water-soluble organic aerosols exert only a minor effect on aerosol pH (Guo et al., 2015; Pye et al., 2018; Vasilakos et al., 2018).

2.3.2 Two factors affecting pH value change

According to Eq. (3), the pH value is determined by the concentrations of H⁺ and H₂O. To evaluate the impact of each factor on the pH value, we independently calculate the changes in pH arising from two pathways: one driven by H⁺ and the other by H₂O.

$$\begin{array}{}\text{(4)}& {\displaystyle }\mathrm{\Delta }{\text{pH}}_{{\mathrm{H}}_{\mathrm{2}}\mathrm{O}}={\log }_{\mathrm{10}}{\displaystyle \frac{{\mathrm{H}}_{\mathrm{2}}\mathrm{O}+\mathrm{\Delta }{\mathrm{H}}_{\mathrm{2}}\mathrm{O}}{{\mathrm{H}}^{+}}}-{\log }_{\mathrm{10}}{\displaystyle \frac{{\mathrm{H}}_{\mathrm{2}}\mathrm{O}}{{\mathrm{H}}^{+}}}\text{(5)}& {\displaystyle }\mathrm{\Delta }{\text{pH}}_{{\mathrm{H}}^{+}}={\log }_{\mathrm{10}}{\displaystyle \frac{{\mathrm{H}}_{\mathrm{2}}\mathrm{O}}{{\mathrm{H}}^{+}+\mathrm{\Delta }{\mathrm{H}}^{+}}}-{\log }_{\mathrm{10}}{\displaystyle \frac{{\mathrm{H}}_{\mathrm{2}}\mathrm{O}}{{\mathrm{H}}^{+}}}\end{array}$$

Here, $\mathrm{\Delta }{\text{pH}}_{{\mathrm{H}}_{\mathrm{2}}\mathrm{O}}$ and $\mathrm{\Delta }{\text{pH}}_{{\mathrm{H}}^{+}}$ represent the changes in pH caused by variations in H₂O and H⁺ concentrations, respectively. The base case concentrations of H₂O and H⁺ are used as references, and the changes in concentration are expressed as ΔH₂O and ΔH⁺, corresponding to the deviations in H₂O and H⁺ from their base case levels.

3 Observations

Multiple observational datasets are used in this study to validate the model simulation in different regions of the world as defined by the IPCC (2022). The information of each dataset is summarized in Table 3, and the site distribution is plotted in Figs. 2 and S1 in the Supplement. These datasets include satellite retrievals from the Infrared Atmospheric Sounding Interferometer (IASI; https://iasi.aeris-data.fr/NH3_IASI_A_L3_data/, last access: 8 September 2025) and the following observation site networks: the Nationwide Nitrogen Deposition Monitoring Network (NNDMN; https://figshare.com/articles/dataset/Data_Descriptor_Xu_et_al_20181211_Scientific_data_docx/7451357/5, last access: 8 September 2025), the European Monitoring and Evaluation Programme (EMEP; https://ebas-data.nilu.no/Default.aspx, last access: 8 September 2025), the Central Pollution Control Board (CPCB; https://cpcb.nic.in/, last access: 8 September 2025), the Acid Deposition Monitoring Network in East Asia (EANET; https://www.eanet.asia/, last access: 8 September 2025), the Ammonia Monitoring Network (AMoN; https://nadp.slh.wisc.edu/networks/ammonia-monitoring-network/, last access: 8 September 2025), the US Environmental Protection Agency (EPA; https://www.epa.gov/data, last access: 8 September 2025), and the Clean Air Status and Trends Network (CASTNET; http://www.epa.gov/castnet, last access: 8 September 2025).

Table 3 Information for each observation dataset used to validate model simulation during 2010–2012.

^a Only the data over land are used.
^b Measurements refer to a chemical or physical property of the total aerosol particle phase in the size fraction with less than 2.5 µm median aerodynamic diameter.

Figure 2 Regions and observation sites used in this study. North America, South America, Europe, the Middle East, South Asia, and East Asia are represented by purple, dark blue, sky blue, orange, light green, and yellow, respectively. In North America AMoN and CASTNET sites are shown as blue and red circles, respectively (EPA sites are provided in Fig. S1). In Europe, EMEP sites are shown as blue circles for NH₃ and red circles for PM_2.5 components. In South Asia, CPCB sites are marked by green circles. In East Asia, NNDMN and EANET sites are indicated by dark-green and dark-blue circles, respectively.

Due to its better precision resulting from favorable thermal contrast conditions (Clarisse et al., 2009), only the morning (around 09:30 local time) overland IASI data are used in this study. The original temporal resolution of the various datasets includes hourly, 3 d, daily, weekly, bi-weekly, and monthly, which we uniformly convert to monthly. The mean bias (MB), mean absolute gross error (MAGE), normalized mean bias (NMB), normalized mean error (NME), and root mean square error (RMSE) are calculated to evaluate the model performance:

$$\begin{array}{}\text{(6)}& \text{MAGE}={\displaystyle \frac{\mathrm{1}}{N}}{\sum }_{i=\mathrm{1}}^N\left|P_i-O_i\right|,\end{array}$$

$$\begin{array}{}\text{(7)}& \text{MB}={\displaystyle \frac{\mathrm{1}}{N}}{\sum }_{i=\mathrm{1}}^N\left(P_i-O_i\right),\end{array}$$

$$\begin{array}{}\text{(8)}& \text{NME}={\displaystyle \frac{{\sum }_{i=\mathrm{1}}^N\left|P_i-O_i\right|}{{\sum }_{i=\mathrm{1}}^N{O}_i}},\end{array}$$

$$\begin{array}{}\text{(9)}& \text{NMB}={\displaystyle \frac{{\sum }_{i=\mathrm{1}}^N\left(P_i-O_i\right)}{{\sum }_{i=\mathrm{1}}^N{O}_i}},\end{array}$$

$$\begin{array}{}\text{(10)}& \text{RMSE}=[{\displaystyle \frac{\mathrm{1}}{N}}{\sum }_{i=\mathrm{1}}^N{\left(P_i-O_i\right)}^{\mathrm{2}}{]}^{\frac{\mathrm{1}}{\mathrm{2}}},\end{array}$$

where P_i and O_i represent the monthly value of model simulation and measurement, respectively. N is the total number of data points used for comparison.

4 Model evaluation

4.1 Aerosol composition

The comparison between the simulation derived from the base case and the observational datasets is summarized in Tables 4–6. Compared with the NNDMN dataset, which primarily includes sites located in urban and rural areas across China, the base case overestimates the NH₃ mass concentration (NMB = 0.19) and underestimates the NH${}_{\mathrm{4}}^{+}$ mass concentration (NMB = −0.41), though it reproduces the NO${}_{\mathrm{3}}^{-}$ mass concentration well (NMB = −0.02). Xie et al. (2022) summarized the NMB between observed NO${}_{\mathrm{3}}^{-}$ and simulated values in China as ranging from −0.97 to 1.90 based on modeling studies in the last decade. The negligible bias of the simulated NO${}_{\mathrm{3}}^{-}$ shows the good performance of the EMAC model in this region. However, the biases in the simulation of NH₃ and NH${}_{\mathrm{4}}^{+}$ indicate that the NH₃ $/$ NH${}_{\mathrm{4}}^{+}$ partitioning treatment is not efficient enough or is oversimplified, as less NH${}_{\mathrm{4}}^{+}$ is produced even with sufficient NH₃. Similarly, in Europe and North America, we obtain a positive bias of NH₃ mass concentration (EMEP dataset: NMB = 2.26; AMoN dataset: NMB = 0.58) and a comparably low NH${}_{\mathrm{4}}^{+}$ mass concentration (EMEP dataset: NMB = 0.05; AMoN dataset: NMB = −0.23). On the other hand, the dry deposition of NH₃ in China is lower than reported from observations (NMB = −0.28; not shown), which contributes to a higher atmospheric NH₃ concentration.

Table 4 Comparison of the base case predictions with monthly average observations from China and Europe during 2010–2012.

Table 5 Comparison of the base case predictions with monthly average observations from India and eastern Asia (PM_2.5 matrix) during 2010–2012.

Table 6 Comparison of the base case predictions with monthly average observations from America (PM_2.5 matrix) during 2010–2012.

In East and Southeast Asia, the mass concentration of NH${}_{\mathrm{4}}^{+}$ is well reproduced (NMB = −0.02), while high and low discrepancies are found in the mass concentrations of NO${}_{\mathrm{3}}^{-}$ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ (NMB = 0.71 and −0.49, respectively). Similar results were also found in Europe, with agreement for NH${}_{\mathrm{4}}^{+}$ (NMB = 0.05) but an overestimation of the NO${}_{\mathrm{3}}^{-}$ mass concentration (NMB = 1.58) and underestimated SO${}_{\mathrm{4}}^{\mathrm{2}-}$ (NMB = −0.41). The positive bias of the simulated NO${}_{\mathrm{3}}^{-}$ is reported by many studies (Xie et al., 2022; Heald et al., 2012; Colette et al., 2011; Bian et al., 2017; Pozzer et al., 2022; Milousis et al., 2025). The negative bias of SO${}_{\mathrm{4}}^{\mathrm{2}-}$ is considered a reason for the positive bias of NO${}_{\mathrm{3}}^{-}$ regarding the thermodynamic equilibrium between NH${}_{\mathrm{4}}^{+}$, SO${}_{\mathrm{4}}^{\mathrm{2}-}$, and NO${}_{\mathrm{3}}^{-}$. The discrepancy in SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and NO${}_{\mathrm{3}}^{-}$ is also due to the missing heterogeneous oxidation reactions for SO₂ and NO_x in the model. Several studies have concluded that adding multiphase chemistry can significantly improve the model performance (Zheng et al., 2015; Zhang et al., 2021b). Cheng et al. (2016) and Zheng et al. (2024) pointed out that the multiphase reactions act as an important SO${}_{\mathrm{4}}^{-}$ source in haze pollution, while Guo et al. (2017b) argued that the multiphase reactions are not likely limited by the required alkaline environment.

In North America, the base case reproduces the mass concentration of SO${}_{\mathrm{4}}^{\mathrm{2}-}$ (NMB = −0.26) but overpredicts the mass concentrations of NH${}_{\mathrm{4}}^{+}$ and NO${}_{\mathrm{3}}^{-}$ (NMB = 0.23 and 0.79, respectively), which is in line with the findings of Walker et al. (2012). The excess NH${}_{\mathrm{4}}^{+}$ promotes the formation of NO${}_{\mathrm{3}}^{-}$, and the uncertain uptake coefficient of N₂O₅ used in models may contribute to more NO${}_{\mathrm{3}}^{-}$ (Walker et al., 2012). The highest NH₃ mass concentration is found in India, especially in the Indo-Gangetic Plain. Our model basically captures the “hotspot”, with a slight negative bias (NMB = −0.26). However, the scarce observation sites and the lack of observed NH${}_{\mathrm{4}}^{+}$, NO${}_{\mathrm{3}}^{-}$, and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ hinder further evaluation of the model performance.

4.2 Aerosol acidity

Due to the lack of direct measurement of aerosol acidity, we collect the PM_2.5 pH values from related studies to compare them with our model simulation in Table 7. These pH values are calculated using the thermodynamic equilibrium model ISORROPIA or E-AIM with input from observational datasets. Compared with Guo et al. (2017a), our simulated pH value from the base case is higher in the western USA (4.3 vs. 2.7), but the value from the Meta case is much closer (2.6). Both the base and the Meta cases predict the same aerosol water content, and the high Ca²⁺ concentration from the Great Basin Desert leading to the precipitation of CaSO₄ is the main reason for the lower aerosol acidity in the base case (Karydis et al., 2021). It is worth noting that the effect of Ca²⁺ on aerosol acidity was not considered by Guo et al. (2017a). The summertime southeastern USA is characterized as relatively clean with aerosol dominated by NH${}_{\mathrm{4}}^{+}$ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ but with negligible mineral cations, and the aerosol is acidic with the observed pH value around ∼ 0.6 (Guo et al., 2015; Pye et al.,2020). However, due to a low bias in simulated SO${}_{\mathrm{4}}^{\mathrm{2}-}$ mass concentration, the model underestimates aerosol acidity, resulting in a higher simulated pH value of approximately 2.0 during the summer period. In Europe, although the good agreement between our simulated pH and the result of Guo et al. (2018) (both are 3.9) supports the model simulations, we note a significant overestimation of alkaline species (sum of NH${}_{\mathrm{4}}^{+}$ and NH₃, NMB = 17 %) and an underestimation of acidic species (sum of NO${}_{\mathrm{3}}^{-}$ and HNO₃, NMB = −57 %; SO${}_{\mathrm{4}}^{\mathrm{2}-}$, NMB = −51 %) in December. These discrepancies emphasize the need for improved representation of aerosol composition in the model. In northern China, Wang et al. (2016) reported pH values that are about 2 units lower than our simulations (4.4 vs. 6.7), and Guo et al. (2017b) found moderate acidic aerosol (pH = 5) during wintertime. We attribute this high bias in simulated pH to the underestimation of cation mass concentrations (e.g., Ca²⁺, Mg²⁺) in PM_2.5, as the mass concentrations of NH${}_{\mathrm{4}}^{+}$, SO${}_{\mathrm{4}}^{\mathrm{2}-}$, and NO${}_{\mathrm{3}}^{-}$ are in the observational ranges reported by Wang et al. (2016).

Table 7 Simulated pH value of PM_2.5 at certain locations compared with observational estimates of particle pH.

Note that the table was extracted in part from Karydis et al. (2021).

5 Secondary inorganic aerosol composition

5.1 Size-resolved composition

The regional mass fractions of size-resolved inorganic aerosol components (NH${}_{\mathrm{4}}^{+}$, SO${}_{\mathrm{4}}^{\mathrm{2}-}$, and NO${}_{\mathrm{3}}^{-}$) are presented as bar charts in Fig. 3, while global distribution maps of their size-resolved mass concentrations are shown in Fig. S2; both the simulation results are derived from the base case. To assess ammonia neutralization of sulfuric and nitric acids, we applied the chemical domain framework defined by Ge et al. (2022) based on SNA molar concentrations in PM₁₀ with a threshold of > 1 µg m⁻³. The four chemical domains, illustrated in Fig. 4, are defined as follows: SO${}_{\mathrm{4}}^{\mathrm{2}-}$ very rich (totNH₃ $/$ totSO₄ < 1, where totNH₃ is the sum of NH₃ and NH${}_{\mathrm{4}}^{+}$ and totSO₄ is the sum of SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and HSO${}_{\mathrm{4}}^{-}$), SO${}_{\mathrm{4}}^{\mathrm{2}-}$ rich (totNH₃ $/$ totSO₄ between 1 and 2), NO${}_{\mathrm{3}}^{-}$ rich (free NH₃ $/$ totNO₃ between 0 and 1, where free NH₃ is totNH₃ minus double totSO₄ and totNO₃ is the sum of NO${}_{\mathrm{3}}^{-}$ and HNO₃), and NH₃ very rich (free NH₃ $/$ totNO₃ > 1). Figure S3 shows the ratios used to define these domains.

Figure 3 (a–c) Global distribution of SO${}_{\mathrm{4}}^{\mathrm{2}-}$, NO${}_{\mathrm{3}}^{-}$, and NH${}_{\mathrm{4}}^{+}$ mass concentration in the size range of 0–10 µm, (d–f) bar plots for mean mass fractions of size-resolved NH${}_{\mathrm{4}}^{+}$, SO${}_{\mathrm{4}}^{\mathrm{2}-}$, and NO${}_{\mathrm{3}}^{-}$ over the globe, land, ocean, and regions (marked in Fig. 2); the size ranges of 5–10, 2.5–5, 1–2.5, and 0–1 µm are marked by green, orange, purple, and blue, respectively (NAm: North America; SAm: South America; Eu: Europe; ME: Middle East; SAs: South Asia; EAs: East Asia).

Figure 4 Distribution of the ammonia neutralization state of sulfuric and nitric acids, based on SNA molar concentrations in PM₁₀ with a threshold of ≥ 1 µg m⁻³, where the SNA mass concentration less than the threshold is measured by blank zone. SO${}_{\mathrm{4}}^{\mathrm{2}-}$ very rich represents ratios of totNH₃ to totSO₄ less than 1 (totNH₃: sum of NH₃ and NH${}_{\mathrm{4}}^{+}$; totSO₄: sum of SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and HSO${}_{\mathrm{4}}^{-}$), SO${}_{\mathrm{4}}^{\mathrm{2}-}$ rich represents ratios of totNH₃ to totSO${}_{\mathrm{4}}^{\mathrm{2}-}$ between 1 and 2, NO${}_{\mathrm{3}}^{-}$ rich represents ratios of free NH₃ to totNO₃ between 0 and 1 (free NH₃: totNH₃ minus double totSO₄; totNO₃: sum of NO${}_{\mathrm{3}}^{-}$ and HNO₃), and NH₃ very rich represents ratios of the free NH₃ to totNO₃ over 1.

Regions with low inorganic aerosol concentrations (< 1 µg m⁻³) are found in the Southern Hemisphere oceans and remote areas such as the North Pole and South America, including the Amazon Basin. These areas represent relatively pristine baselines for evaluating anthropogenic impacts (Andreae et al., 1990; Andreae, 2007). In low northern latitude oceans, the SO${}_{\mathrm{4}}^{\mathrm{2}-}$ very rich and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ rich domains dominate, where NH₃ is insufficient to fully neutralize sulfuric acid. As a result, sulfuric acid remains only partially neutralized, contributing to aerosol acidity. This pattern reflects low NH₃ emissions over oceans (Fig. 1) and the contribution of biogenic dimethyl sulfide (DMS) to marine SO${}_{\mathrm{4}}^{\mathrm{2}-}$ (Fiddes et al., 2018; Jackson et al., 2020). Additional NO_x and SO₂ from shipping contribute to marine SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and NO${}_{\mathrm{3}}^{-}$ (Wang et al., 2023; Burgard and Bria, 2016). Average PM₁₀ SNA concentrations are lower over oceans compared to land (NH${}_{\mathrm{4}}^{+}$: 0.06 µg m⁻³ vs. 0.36 µg m⁻³; NO${}_{\mathrm{3}}^{-}$: 0.37 µg m⁻³ vs. 1.24 µg m⁻³; SO${}_{\mathrm{4}}^{\mathrm{2}-}$: 0.60 µg m⁻³ vs. 0.95 µg m⁻³). Marine SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and NO${}_{\mathrm{3}}^{-}$ are predominantly in the supermicron mode, with mass fractions of 70 % and 58 %, respectively. Coastal areas exhibit a NO${}_{\mathrm{3}}^{-}$ rich domain due to continental outflows (Fig. S3) (Prospero et al., 1985), consistent with prior findings that marine aerosol in the supermicron mode primarily comprises inorganic salts, including sea salt, non-sea-salt sulfate, and nitrate, while organic matter is concentrated in the submicron range (Russell et al., 2023; Cavalli et al., 2004).

On land, except for areas such as northern Russia, central Africa, and the Arabian Peninsula, the aerosol typically falls within the NH₃ very rich domain. In this domain, SO${}_{\mathrm{4}}^{\mathrm{2}-}$ is neutralized by the available NH₃, with sufficient NH₃ remaining to neutralize NO${}_{\mathrm{3}}^{-}$. As a result, NO${}_{\mathrm{3}}^{-}$ becomes the limiting factor in the formation of NH₄NO₃. More than 60 % of SNA mass is in the submicron mode on land, while the supermicron modes (i.e., 1–2.5, 2.5–5, and 5–10 µm) account for a smaller fraction (7 % for NH${}_{\mathrm{4}}^{+}$, 15 % for SO${}_{\mathrm{4}}^{\mathrm{2}-}$, and 18 % for NO${}_{\mathrm{3}}^{-}$). In polluted regions such as East and South Asia, PM₁₀ SNA concentrations are 3 times higher than the global land average. For example, East Asia shows NH${}_{\mathrm{4}}^{+}$ of 1.88 µg m⁻³, NO${}_{\mathrm{3}}^{-}$ of 5.31 µg m⁻³, and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ of 2.29 µg m⁻³, while South Asia records NH${}_{\mathrm{4}}^{+}$ of 1.58 µg m⁻³, NO${}_{\mathrm{3}}^{-}$ of 3.68 µg m⁻³, and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ of 3.18 µg m⁻³. Free NH₃ is abundant in regions like southern North America, Europe, South Asia, and East Asia, with mean free NH₃ $/$ total NO${}_{\mathrm{3}}^{-}$ ratios of 2.11, 3.77, 5.30, and 3.78, respectively. Over 75 % of SNA mass in these regions is concentrated in the 0–1 and 1–2.5 µm size ranges. The NH₃ surplus reflects recent trends in precursor emissions, with stable or increasing NH₃ emissions contrasting with declining SO₂ and NO_x emissions (Zheng et al., 2018; Hand et al., 2012; Russell et al., 2012). In the Middle East, particularly the Arabian Peninsula, aerosol is dominated by desert dust with negligible NH₃ emissions. The SO${}_{\mathrm{4}}^{\mathrm{2}-}$ rich and NO${}_{\mathrm{3}}^{-}$ rich domains predominate, where NH₃ levels are insufficient to neutralize acidic components, limiting NH₄NO₃ formation. Mean concentrations of NH${}_{\mathrm{4}}^{+}$, NO${}_{\mathrm{3}}^{-}$, and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ in this region are 0.72, 2.07, and 3.19 µg m⁻³, respectively. Over 70 % of NO${}_{\mathrm{3}}^{-}$ resides in the supermicron modes due to interactions with sea salt and crustal dust, which shift NO${}_{\mathrm{3}}^{-}$ from submicron to supermicron modes (Chen et al., 2020; Koçak et al., 2007; Karydis et al., 2016). Submicron NH${}_{\mathrm{4}}^{+}$ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ remain dominant, accounting for 96 % and 79 % of their respective fractions, consistent with Osipov et al. (2022), who identified anthropogenic sources as the primary contributors to fine particles in the region.

5.2 NH₃ column concentration

Figure 5a shows the global distribution of NH₃ column concentration. The global area-weighted mean value of NH₃ column concentration is 0.80 mg m⁻², with the highest value up to 30 mg m⁻² in the Indo-Gangetic Plain of India and the lowest value of less than 0.01 mg m⁻² in the remote oceans of the Southern Hemisphere and the South Pole. Compared with previous studies that investigated NH₃ column concentrations based on satellite retrievals (Van Damme et al., 2014, 2015; Zhou et al., 2024), our results can capture the distribution of NH₃ hotspots worldwide, including the Indo-Gangetic Plain, the North China Plain, and West Africa and Amazonia, where biomass burning is dominant (Van Damme et al., 2018). However, Van Damme et al. (2018) pointed out that two-thirds of the NH₃ emission hotspots are underestimated by at least 1 order of magnitude in the NH₃ emission inventory EDGAR (CAMS-GLOB-ANT used in this study is derived from EDGAR; see Sect. 2.1.1). Given such an underestimation in the current NH₃ emission inventory, we further improve the emission by applying a new inventory and updating the current inventory using a top-down method introduced in Sect. 2.1. The simulation results are discussed in Sect. 7.

Figure 5 Global distribution of (a) the average NH₃ column concentration (mg m⁻²) for the base case and (b) the absolute difference between the IASI satellite retrieval and base case, from 2010 to 2012.

Our simulated NH₃ column concentrations show good agreement with the IASI satellite observations (Fig. 5b), with global land mean values at a comparable level of 1.66 mg m⁻². However, regional biases can be significant, with an NMB of −0.32. The model overestimates NH₃ column concentrations over regions such as India, China, western America, and northeastern Africa, while it has a negative bias in high-latitude regions. It is also important to note that biases in the IASI satellite products have been identified in previous studies (Dammers et al., 2017, 2019); more details are discussed in Sect. 2.1.2.

5.3 Global NH₃ budget

The global budget for NH₃ and NH${}_{\mathrm{4}}^{+}$ is summarized in Table 8. Given a global NH₃ emission input of 73.3 Tg yr⁻¹, the global burden and lifetime derived from the base case simulation are 0.4 Tg and 2.0 d, respectively; for NH${}_{\mathrm{4}}^{+}$, the global burden and lifetime are 0.3 Tg and 3.1 d, respectively. Based on the simulations of nine models, Bian et al. (2017) assessed that given the average NH₃ emission input of 76.4 Tg yr⁻¹, the average global burden and lifetime for NH₃ and NH${}_{\mathrm{4}}^{+}$ are 0.2 Tg and 0.7 d and 0.3 Tg and 4.3 d, respectively. Compared to Bian et al. (2017), our study uses the same NH₃ emission input; however, the global burden and lifetime of NH₃ derived from the base case are larger. This potential overestimation may be attributed to the wet-deposition scheme used in EMAC. The scavenging scheme (SCAV) accounts for pH adjustments in NH₃ dissolution. More specifically, the EMAC model implicitly determines the effective Henry's law constant by solving a system of coupled ordinary differential equations, explicitly representing liquid-phase processes in clouds and raindrops, including dissociation, acid–base equilibria, redox reactions, and photolysis (Tost et al., 2006). This approach ensures a comprehensive calculation of total wet deposition for NH${}_{\mathrm{4}}^{+}$ and NH₃. Notably, the NH₃ burdens simulated in the AeroCom model intercomparison by Bian et al. (2017) exhibit significant variability, spanning a factor of 17. This wide range underscores the sensitivity of atmospheric NH${}_{\mathrm{4}}^{+}$ and NH₃ burdens and lifetimes to model domain definitions, deposition pathways, and NH₃ chemical processes across different models (Ge et al., 2022).

Table 8 Atmospheric budget of NH₃, NH${}_{\mathrm{4}}^{+}$, and NH_x (NH₃ + NH${}_{\mathrm{4}}^{+}$).

^a NH₃ lifetime = burden $/$ emission; NH${}_{\mathrm{4}}^{+}$ lifetime = burden $/$ (dry deposition + wet deposition).
^b NH_x lifetime = NH_x burden $/$ NH₃ emission; in the parentheses, NH_x lifetime = NH_x burden $/$ (dry deposition + wet deposition).

Considering the SCAV scavenging scheme, we further calculate the global budget for NH_x (NH₃ + NH${}_{\mathrm{4}}^{+}$) in Table 8. The global burden of NH_x is 0.7 Tg. Wet deposition and dry deposition contribute almost equally to the sink, accounting for 54 % and 46 %, respectively. The lifetime of NH_x is 3.5–4.0 d, depending on the calculation method. Ge et al. (2022) estimated a global budget for NH_x, with an input NH₃ emission of 64.5 Tg yr⁻¹. They calculated the global burden of NH_x to be 0.9 Tg, with a lifetime of 4.9–5.2 d, and found its wet and dry deposition contribute equally to the sink (i.e., 51 % and 49 % of total deposition, respectively).

6 Aerosol acidity

Figure 6 illustrates the global distributions of size-resolved aerosol pH, with regional averages summarized in Table 9. Aerosol pH exhibits marked spatial variations, influenced by land–sea contrasts and regional sources. Over land in the Northern Hemisphere, excluding deserts, aerosols are generally acidic, whereas marine aerosols are alkaline due to sea salt influence. High-latitude marine aerosols are generally more acidic than those over remote ocean regions, primarily due to the long-range transport of anthropogenic pollutants such as H₂SO₄ and HNO₃ from continental sources (Myhre et al., 2013; Karydis et al., 2021). The average pH values for land-based aerosols are 4.2 (5–10 µm), 4.2 (2.5–5 µm), 3.9 (1–2.5 µm), and 4.3 (0–1 µm). In contrast, marine aerosol pH values are 6.2 (5–10 µm), 5.8 (2.5–5 µm), 5.1 (1–2.5 µm), and 5.1 (0–1 µm). The deserts of North Africa, the Middle East, and the Gobi maintain the highest pH values (> 7) across all size ranges, driven by non-volatile cations (e.g., Ca²⁺, Mg²⁺) that neutralize acidic components and enhance water uptake.

Figure 6 Global distribution of surface average aerosol pH values in the following size ranges: (a) 0–1 µm, (b) 1–2.5 µm, (c) 2.5–5 µm, (d) and 5–10 µm. Derived from the base case from 2010 to 2012.

Table 9 Size-resolved pH values across the globe and regions from simulation cases.

^* Average pH value.

Regionally, higher pH values (5–5.7) in the Middle East are attributed to airborne dust, while coastal areas like the western Arabian Gulf have lower aerosol pH (< 3 in 0–1 µm) due to elevated sulfate concentrations. In South Asia, abundant NH₃ emissions keep pH values between 5.4 and 4.9 despite high SO₂ and NO_x emissions. East Asia shows a distinct pH gradient, with desert regions in the northwest reaching pH > 7 across all sizes and southeast coastal areas exhibiting low pH (2–4), linked to significant SO₂ emissions and sulfate formation. Europe (pH 3.7–4.1) and North America (pH 3.2–3.6) demonstrate moderate acidity, with the western USA exhibiting higher values (pH 4–6) due to desert influences. In South America, pH ranges from 3.9 to 4.5, with coastal regions exceeding 6 due to sea salt.

An increasing trend in pH is observed from North America to Europe and East Asia, consistent with prior studies (Zhang et al., 2021a; Ding et al., 2019; Guo et al., 2017b). Aerosol alkalinity is driven by NH${}_{\mathrm{4}}^{+}$ and non-volatile cations, which neutralize SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and NO${}_{\mathrm{3}}^{-}$, while water-soluble ions (WSIs) enhance liquid water uptake. East Asia exhibits the highest pH among regions, facilitated by relatively low SO${}_{\mathrm{4}}^{\mathrm{2}-}$, abundant NH${}_{\mathrm{4}}^{+}$ and WSIs, and significant NO${}_{\mathrm{3}}^{-}$ and non-volatile cations in the coarse modes. These chemical properties outweigh the influence of meteorological effects such as differences in temperature and humidity.

Contrary to previous findings (e.g., Kakavas et al., 2021; Ding et al., 2019), which suggest decreasing aerosol acidity with increasing particle size, pH in the base case (stable state) for 0–1 µm exceeds that of 1–2.5 µm over many regions, excluding oceans and the Middle East. This is examined through three perspectives: first, the assumption of the phase state – as described in Sect. 2.2.1, the pH predictions based on the metastable state tend to be lower than those based on the stable state under low-RH conditions. Although previous studies concluded that the choice between stable and metastable assumptions in ISORROPIA has no significant impact on pH predictions (Song et al., 2018), our sensitivity simulation assuming metastable conditions (Meta case) yields consistently lower pH values across all particle size ranges. The differences in pH relative to the base case are shown in Fig. 7 and Table 9. The most substantial reductions are observed in regions with high mineral cations and/or low RH, such as South Asia and the Middle East, where the pH decreases by more than 2 units. These results are consistent with the findings of Karydis et al. (2021) and Milousis et al. (2024). For instance, Karydis et al. (2021) reported that the stable-state assumption yields global average pH values approximately 0.5 units higher than those from the metastable assumption, comparable to the magnitude of differences calculated in our study (Table 9). While the Meta case explains some discrepancies, pH in the 0–1 µm size range remains higher relative to the 1–2.5 µm range in several regions. Another sensitivity simulation, removing NH₃ emissions (noNH₃ case), significantly reduced pH across all sizes over land, especially in NH₃-rich regions like South Asia and the Middle East, where 0–1 µm pH dropped below zero. This indicates increased pH sensitivity in the fine size ranges (i.e., 0–1 and 1–2.5 µm), particularly in regions where NH₃ availability is abundant. Excluding NH₃ results in a consistent pH increase with particle size across all regions. Finally, as discussed in Sect. 4.1, the comparison with observed size-resolved mass concentrations and the NH₃ $/$ NH${}_{\mathrm{4}}^{+}$ partitioning ratio reveals an underestimation of acidic components (SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and NO${}_{\mathrm{3}}^{-}$) and an overestimation of alkaline components in the 0–1 µm size range, along with an inadequate treatment of NH₃ $/$ NH${}_{\mathrm{4}}^{+}$ partitioning. This chemical imbalance contributes to the anomalously high simulated pH in the fine mode (0–1 µm). For instance, the observed SO${}_{\mathrm{4}}^{\mathrm{2}-}$, NO${}_{\mathrm{3}}^{-}$, and NH${}_{\mathrm{4}}^{+}$ at 0–1 µm accounts for 91 %, 59 %, and 67 % of 0–2.5 µm, respectively, at the Montseny site during 2010 compared to the accounting of 79 %, 54 %, and 77 % in the simulation, highlighting a need for improved representation of aerosol composition in fine modes. In addition, the observed mean NH${}_{\mathrm{4}}^{+}$ partitioning ratios (NH${}_{\mathrm{4}}^{+}$ $/$ (NH₃ + NH${}_{\mathrm{4}}^{+}$), in µmol m⁻³) during 2010–2012 from EMEP, NNDMN, and AMoN are 49 %, 49 %, and 50 %, respectively. Our model shows low biases, with mean ratios of 32 %, 39 %, and 41 %. These results reinforce the necessity of refining the treatment of NH₃ $/$ NH${}_{\mathrm{4}}^{+}$ partitioning to better capture the size-resolved trend of aerosol acidity.

Figure 7 Global distribution maps of pH value absolute difference between Meta case and base case in the size ranges of 0–1, 1–2.5, 2.5–5, and 5–10 µm, averaged from 2010 to 2012.

Part of the discrepancy in size-resolved pH calculations between models may arise from differences in how particle size distributions and gas–particle partitioning are treated. EMAC model uses a lognormal size distribution and applies ISORROPIA separately to each size mode to calculate gas–aerosol partitioning. In contrast, other studies (e.g., Kakavas et al., 2021) employ sectional approaches, where gas–aerosol partitioning is first performed on the bulk aerosol phase, and the resulting condensed mass is then distributed across size bins based on the available surface area. These fundamental differences in modeling assumptions can lead to variations in the predicted distribution of aerosol components across size ranges, which in turn affects the calculated size-resolved pH. Moreover, previous studies investigating size-resolved aerosol acidity also have certain limitations. For instance, a laboratory study observed increasing aerosol acidity with decreasing particle size, but this trend only holds for pH below 2 (Craig et al., 2018). Similarly, a field study by Fang et al. (2017) reported the same trend but lacked concurrent measurements of key gas-phase species, which limits the interpretability of the results.

7 Emission sensitivity analysis

The formation of secondary inorganic aerosols is strongly linked to NH₃ emissions (Wu et al., 2016; Chen et al., 2019; Liang et al., 2024). Xu et al. (2020) and Wang et al. (2015) have highlighted the critical role of gas–particle partitioning of NH₃ $/$ NH${}_{\mathrm{4}}^{+}$ in SNA formation, which is influenced by factors such as temperature, aerosol water content, and aerosol acidity (Nenes et al., 2020). Notably, the NH₃ $/$ NH${}_{\mathrm{4}}^{+}$ partitioning process buffers aerosol acidity, maintaining stability even amid shifts in acidic species like NO${}_{\mathrm{3}}^{-}$ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ (Karydis et al., 2021). These findings suggest that the effects of NH₃ emissions on SNA formation and aerosol acidity remain understudied (Weber et al., 2016; Zheng et al., 2022; Fu et al., 2017; Zou et al., 2024). However, most studies have focused on fine particles (PM_2.5 or PM₁), polluted regions (e.g., northern China), and haze episodes (Ge et al., 2019; Gao et al., 2016). Modeling studies often evaluate the effects of NH₃ emission changes by uniformly altering emission levels within each grid (Pozzer et al., 2017; Fu et al., 2017). While such approaches provide valuable insights, they may lack feasibility as NH₃ abundance correlates with population density, making uniform changes less representative of real-world scenarios. To address this, we compare the effects of two NH₃ emission schemes, CEDS and Top-Dep, on size-resolved SNA mass concentration and aerosol acidity relative to the base case.

The NH₃ mass concentration serves as a proxy for NH₃ emissions, and differences in NH₃ mass concentrations between the scenarios and the base case (Fig. S4) align with corresponding emission differences (Fig. 1). Figures 8 and 9 show the responses of size-resolved NH${}_{\mathrm{4}}^{+}$, NO${}_{\mathrm{3}}^{-}$, and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ mass concentrations to NH₃ emission changes in the CEDS and Top-Dep cases, respectively. Similarly, Fig. 10 illustrates the size-resolved aerosol acidity responses for the two cases. Additional details are provided in the Supplement: Figs. S5–S7 (CEDS case) and Figs. S8–S10 (Top-Dep case) depict the size-resolved responses of NH${}_{\mathrm{4}}^{+}$, NO${}_{\mathrm{3}}^{-}$, and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ mass concentrations, while Figs. S11 and S12 highlight changes in size-resolved aerosol acidity.

Figure 8 Bar plots for regional surface SNA mass concentration (µg m⁻³) absolute difference between the CEDS case and base case in the four size ranges (a–d); change ratio in the four size ranges (e–h). The calculation of the change ratio in the size range of 0–1 µm is based on the mask of 0.1 µg m⁻³; the change ratio in the size range of 1–2.5, 2.5–5, and 5–10 µm is based on the mask of 0.05 µg m⁻³ (Land: global land; NAm: North America; SAm: South America; Eu: Europe; ME: Middle East; SAs: South Asia; EAs: East Asia).

Figure 9 The same as Fig. 8 but for the difference between the Top-Dep case and base case.

Figure 10 Bar plots for the pH value absolute difference between the CEDS/Top-Dep case and base case in the four size ranges (a–d), with the two dashed lines representing the value of 0.1 and −0.1, respectively (Land: global land; NAm: North America; SAm: South America; Eu: Europe; ME: Middle East; SAs: South Asia; EAs: East Asia).

The regional gas–particle partitioning ratios for NH₃ $/$ NH${}_{\mathrm{4}}^{+}$ (ε(NH${}_{\mathrm{4}}^{+}$) = NH${}_{\mathrm{4}}^{+}$ $/$ (NH${}_{\mathrm{4}}^{+}$ + NH₃)) and NO${}_{\mathrm{3}}^{-}$ $/$ HNO₃ (ε(NO${}_{\mathrm{3}}^{-}$) = NO${}_{\mathrm{3}}^{-}$ $/$ (NO${}_{\mathrm{3}}^{-}$ + HNO₃)) are shown in Fig. S13. Regional emission amounts of NO_x and SO₂ from the CAMS and CEDS_GBD inventories are detailed in Table S1.

7.1 Size-resolved SNA response

Regarding the atmospheric budget of NH₃ (Table 8), a modest increase in global NH₃ emissions (CEDS case) slightly raises deposition of NH₃ and shortens the lifetime. Meanwhile, both the burden and the deposition of NH${}_{\mathrm{4}}^{+}$ are increased, with a longer lifetime. In contrast, the larger emission increase in the Top-Dep cases does not alter the NH₃ burden but leads to higher deposition and a shorter lifetime for both species.

Across land regions, a small increase in NH₃ emissions (CEDS case), along with rising NO_x and SO₂ emissions, slightly raises ε(NH${}_{\mathrm{4}}^{+}$) while marginally lowering ε(NO${}_{\mathrm{3}}^{-}$) (Fig. S13). The SNA mass concentration increases consistently across size ranges, with the most notable growth in NH${}_{\mathrm{4}}^{+}$ and NO${}_{\mathrm{3}}^{-}$ in the 1–2.5 µm range (19 % and 11 %, respectively) and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ in the submicron particles (9 %). Under a larger NH₃ emission increase (Top-Dep case), ε(NH${}_{\mathrm{4}}^{+}$) drops significantly and ε(NO${}_{\mathrm{3}}^{-}$) decreases slightly. The SNA response becomes more pronounced, with substantial increases in the 1–2.5 µm range (NH${}_{\mathrm{4}}^{+}$: 104 %; NO${}_{\mathrm{3}}^{-}$: 41 %; SO${}_{\mathrm{4}}^{\mathrm{2}-}$: 23 %), while NO${}_{\mathrm{3}}^{-}$ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ decrease in coarser particles. These shifts are linked to higher NH₃ emissions in relatively clean, high-latitude regions with low NH₃ flux (Fig. 1c), which elevate NH₃ concentrations (Fig. S4) and enhance NH${}_{\mathrm{4}}^{+}$ formation (Fig. S8).

Regional analysis of SNA responses in Europe shows that a 23 % reduction in NH₃ emissions, along with increases in NO_x and SO₂ emissions (CEDS case), raises ε(NH${}_{\mathrm{4}}^{+}$) slightly while causing a minor drop in ε(NO${}_{\mathrm{3}}^{-}$). The SNA mass increases mainly in the 1–2.5 µm range, with a slight decrease in SO${}_{\mathrm{4}}^{\mathrm{2}-}$ in coarser particles. Since Europe has abundant NH₃, reductions are offset by existing availability and rising NO_x and SO₂ levels, leading to additional NH₄NO₃ and (NH₄)₂SO₄ formation. With higher NH₃ emissions (Top-Dep case), the response becomes more complex: SNA decreases in the submicron particles (Fig. 9e) but increases in larger particles (Fig. 9f–g), with the largest growth seen in NH${}_{\mathrm{4}}^{+}$ (∼ 50 %).

In East Asia, reducing NH₃ emissions while increasing NO_x and SO₂ emissions (CEDS case) raises ε(NH${}_{\mathrm{4}}^{+}$) and slightly increases ε(NO${}_{\mathrm{3}}^{-}$). SNA grows across all particle sizes, with the most significant changes in the 1–2.5 µm range (Fig. 8f). A similar NH₃ reduction in the Top-Dep case produces comparable ε(NH${}_{\mathrm{4}}^{+}$) and ε(NO${}_{\mathrm{3}}^{-}$) changes, with SNA mainly decreasing in the submicron particles. The response in East Asia resembles that of Europe, where abundant NH₃ buffers SNA changes. In South Asia, NH₃ and NO_x emissions rise in the CEDS case, while SO₂ emissions decline, slightly lowering ε(NH${}_{\mathrm{4}}^{+}$) and increasing ε(NO${}_{\mathrm{3}}^{-}$). NH${}_{\mathrm{4}}^{+}$ and NO${}_{\mathrm{3}}^{-}$ concentrations grow across all size ranges (Fig. 8), with the largest NH${}_{\mathrm{4}}^{+}$ increase found in 1–2.5 µm particles (66 %) and NO${}_{\mathrm{3}}^{-}$ rising in both submicron and 1–2.5 µm particles (∼ 60 %). SO${}_{\mathrm{4}}^{\mathrm{2}-}$ decreases in submicron particles but increases in coarser ones (25 %). Conversely, reducing NH₃ emissions in the Top-Dep case raises ε(NH${}_{\mathrm{4}}^{+}$) and lowers ε(NO${}_{\mathrm{3}}^{-}$), leading to SNA declines, especially in submicron particles. South Asia, with abundant NH₃, shows NO${}_{\mathrm{3}}^{-}$ as the limiting factor for NH₄NO₃ formation, driving strong ε(NO${}_{\mathrm{3}}^{-}$) and NO${}_{\mathrm{3}}^{-}$ responses in finer particles.

In North America, NH₃ and SO₂ emissions decrease, while NO_x emissions slightly rise (CEDS case). ε(NH${}_{\mathrm{4}}^{+}$) increases marginally, whereas ε(NO${}_{\mathrm{3}}^{-}$) declines slightly. NH${}_{\mathrm{4}}^{+}$ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ show minor reductions in submicron particles, while NO${}_{\mathrm{3}}^{-}$ increases. In the Top-Dep case, a sharp rise in NH₃ emissions, mainly over Canada and Greenland, significantly lowers ε(NH${}_{\mathrm{4}}^{+}$), stabilizes ε(NO${}_{\mathrm{3}}^{-}$), and increases NH${}_{\mathrm{4}}^{+}$ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ in 1–2.5 µm particles. In South America, a small NH₃ emission rise (CEDS case) has little effect on ε(NH${}_{\mathrm{4}}^{+}$) or ε(NO${}_{\mathrm{3}}^{-}$), resulting in minimal aerosol composition changes. However, a substantial NH₃ increase (Top-Dep case) significantly boosts NH₃ concentrations, reduces ε(NH${}_{\mathrm{4}}^{+}$) and ε(NO${}_{\mathrm{3}}^{-}$), and shifts aerosol partitioning to smaller particles, particularly in central regions. Decreased NO${}_{\mathrm{3}}^{-}$ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ in surrounding areas (Figs. S9–S10) suggest NH₃ is neutralizing transported species, explaining the observed ε(NO${}_{\mathrm{3}}^{-}$) reduction. Changes in other particle size ranges are minimal. In the Middle East, NH₃ emissions rise moderately (20 %), along with slight NO_x and SO₂ increases (CEDS case). ε(NH${}_{\mathrm{4}}^{+}$) and ε(NO${}_{\mathrm{3}}^{-}$) remain stable, with NH${}_{\mathrm{4}}^{+}$ and NO${}_{\mathrm{3}}^{-}$ increasing mainly in submicron particles (∼ 40 %), while SO${}_{\mathrm{4}}^{\mathrm{2}-}$ grows across sizes (∼ 15 %). A minor NH₃ emission drop (Top-Dep case) reduces NH₃ concentrations but enhances SNA in 1–2.5 µm particles, highlighting compensatory NH₃ effects in NH₃ very rich domains.

Overall, higher NH₃ emissions enhance SNA formation, particularly in the fine size ranges (0–1 and 1–2.5 µm). Greater NH${}_{\mathrm{4}}^{+}$ formation (e.g., Top-Dep case) depletes NO${}_{\mathrm{3}}^{-}$ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ from coarse size ranges, leading to decreases in the 2.5–5 and 5–10 µm ranges. In low-SNA regions (e.g., South America, Greenland), NH₃ increases have limited SNA impacts but substantially elevate NH₃ concentrations. In NH₃ very rich regions (e.g., East Asia, Europe), NH₃ reductions alone may still increase ε(NH${}_{\mathrm{4}}^{+}$), promoting further SNA formation. These findings align with Zou et al. (2024), emphasizing the greater effectiveness of NO_x reductions in PM_2.5 mitigation compared to NH₃ or SO₂ control in ammonia-rich environments. Coordinated control of precursor emissions is crucial for mitigating air pollution, especially in heavily polluted regions (Wen et al., 2024).

7.2 Aerosol acidity response

7.2.1 Size-resolved pH changes

Among atmospheric buffering agents (e.g., conjugate acid–base pairs like NH₃ $/$ NH${}_{\mathrm{4}}^{+}$ and CO₂ $/$ HCO${}_{\mathrm{3}}^{-}$, as well as organic acids), the NH₃ $/$ NH${}_{\mathrm{4}}^{+}$ acid–base pair exhibits the largest buffering capacity for aerosols, dominating much of the continental urban area (Zheng et al., 2023, 2020). When NH₃ emissions are entirely eliminated, changes in marine aerosol pH values are negligible compared to the base case (noNH₃ case, Table 9 and Fig. S14). However, aerosols over land become significantly more acidic, with pH values in the 0–1 µm size range dropping to −0.2 in South Asia and −0.17 in the Middle East (Fig. 10). The most substantial pH decreases occur in 0–1 µm aerosols, primarily in South Asia, East Asia, Europe, North America, and South America, consistent with the “NH₃-buffered regions” identified by Zheng et al. (2020).

Interestingly, while neither the Middle East nor northern Africa is categorized as “NH₃-buffered regions”, noticeable pH decreases are observed in the 0–1 and 1–2.5 µm size ranges over the Middle East, including Egypt and Libya. In these regions, the SNA is dominated by SO${}_{\mathrm{4}}^{\mathrm{2}-}$ rich and NO${}_{\mathrm{3}}^{-}$ rich chemical domains (Fig. 4), where NH₃ levels are insufficient to neutralize acidic components, resulting in available NO${}_{\mathrm{3}}^{-}$ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$. This results in excess acidic components, particularly in the 0–1 and 1–2.5 µm size ranges (Fig. S3). Without NH₃ emissions, the abundance of NO${}_{\mathrm{3}}^{-}$ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ further increases aerosol acidity.

Compared to the base case, size-resolved pH values in the CEDS case show an average decrease of −0.13 to −0.15 units over land. The most pronounced decreases occur in remote regions of the Northern Hemisphere (Fig. S11), where NH₃ emission flux is relatively low. In contrast, the Top-Dep case exhibits an average pH increase of 0.24 units in the 0–1 µm range over land, driven by the higher NH₃ emission flux. This NH₃-driven alkalization effect is less pronounced in the 1–2.5 µm range due to counteracting acidification effects from newly formed acidic components, as suggested by Zheng et al. (2024). Notably, pH decreases of −0.10 units are found in the 2.5–5 and 5–10 µm size ranges.

In East Asia, the pH changes in the CEDS case are concentrated in the 2.5–5 and 5–10 µm ranges (−0.10 units in both), with minimal changes in the 0–1 and 1–2.5 µm ranges. In the Top-Dep case, however, pH changes are more drastic across all size ranges: 0–1 µm (−0.17 units), 1–2.5 µm (−0.31 units), 2.5–5 µm (−0.35 units), and 5–10 µm (−0.33 units). These discrepancies between cases highlight that a reduction in NH₃ emissions enhances aerosol acidity (as in the Top-Dep case), but this effect can be partially offset by shifts in SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and NO${}_{\mathrm{3}}^{-}$ mass concentrations (as in the CEDS case). For instance, in the CEDS case, decreases in SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and NO${}_{\mathrm{3}}^{-}$ concentrations result in pH rises of 0.15–0.25 units across all size ranges compared to the Top-Dep case.

Our simulation results align with other studies. For example, Song et al. (2019) found that a ∼ 0.3 increase in log ₁₀[NH₃] contributed to a 0.3–0.4 unit rise in PM₁ pH in Beijing during winter between 2014/2015 and 2018/2019. Concurrent changes in aerosol composition (increased NO${}_{\mathrm{3}}^{-}$ and reduced SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and Cl⁻) led to a modest increase in pH of 0.1 units. Similarly, Zhou et al. (2022) observed a minor PM_2.5 pH decrease of −0.24 units in the Yangtze River Delta region from 2011 to 2019, despite significant changes in aerosol composition resulting from the Air Pollution Prevention and Control Action Plan.

7.2.2 Drivers of pH changes

We further quantify the changes in pH values between the CEDS/Top-Dep cases and the base case, isolating the contributions from changes in H⁺ and H₂O concentrations (detailed in Sect. 2.2.2). The results are presented in Figs. S15 and S16.

In the CEDS case, the decreased pH values across all four aerosol size ranges are primarily driven by changes in H⁺ concentrations, particularly over the remote regions of the Northern Hemisphere. The effects of H₂O concentration changes are particularly relevant in the Middle East and South Asia. Notably, a substantial pH increase is observed in South Asia, with increases of 0.35, 0.43, 0.44, and 0.63 units in the 0–1, 1–2.5, 2.5–5, and 5–10 µm size ranges, respectively. These increases are predominantly caused by changes in H⁺ concentrations, which fully counteract the pH-decreasing effects of changes in H₂O concentrations.

In the Top-Dep case, although changes in H⁺ concentrations significantly enhance pH values in the 0–1 µm size range, the pH-decreasing effects of changes in H₂O concentrations are distributed across all four size ranges over the entire land surface. In East Asia, both H⁺ and H₂O concentration changes contribute substantially to the observed pH decreases, highlighting their combined impact.

In summary, NH₃ emissions play a crucial role in maintaining terrestrial aerosols at moderately acidic levels, particularly in the fine-mode size range of 0–1 µm. When NH₃ emissions are completely removed in the model, the largest pH decreases are found in this size range. In general, pH changes demonstrate a consistent trend with variations in NH₃ emissions in the fine-mode ranges (0–1 and 1–2.5 µm) based on our findings. Overall, pH changes in the fine-mode aerosols (0–1 and 1–2.5 µm) show a consistent response to variations in NH₃ emissions. However, increases in SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and NO${}_{\mathrm{3}}^{-}$ associated with elevated NH₃ emissions can partially offset the expected rise in pH (e.g., over land areas in the 1–2.5 µm range in the CEDS case). Furthermore, in regions with high SO₂ and NO_x emissions, such as East Asia under the CEDS inventory, the enhanced formation of SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and NO${}_{\mathrm{3}}^{-}$ can counteract the pH-lowering effect of reduced NH₃ emissions. In the coarser size ranges (2.5–5 and 5–10 µm), however, pH changes are governed by the competing influences of H⁺ and H₂O concentration changes.

8 Conclusion

This study applied three distinct NH₃ emission schemes to the global atmospheric chemistry and climate model (EMAC) to assess the impact of NH₃ emissions on the size-resolved sulfate–nitrate–ammonium (SNA) aerosol composition and aerosol acidity. It also explores the synergistic effects arising from concurrent changes in NH₃, SO₂, and NO_x emissions. The emission schemes include two bottom-up inventories and one inventory updated through a top-down approach. By simulating the aerosol size ranges of 0–1, 1–2.5, 2.5–5, and 5–10 µm, the study provides a comprehensive assessment of how NH₃ emissions influence global aerosol chemistry and acidity across different particle size fractions.

The model accurately captures the distribution of global NH₃ hotspots, but comparisons with observational datasets reveal positive biases in NH₃ concentrations and underestimations of NH${}_{\mathrm{4}}^{+}$ in some regions such as China, North America, and Europe. This discrepancy suggests inefficiencies or oversimplifications in the NH₃ $/$ NH${}_{\mathrm{4}}^{+}$ partitioning treatment, with insufficient NH${}_{\mathrm{4}}^{+}$ generated despite NH₃ availability. In East and Southeast Asia, NH${}_{\mathrm{4}}^{+}$ concentrations are well-simulated, but discrepancies were found for NO${}_{\mathrm{3}}^{-}$ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$, likely due to the absence of heterogeneous oxidation processes in the model. The simulated global NH₃ burden and lifetime are higher than reported in related studies, attributed to the wet-deposition scheme used in EMAC, which accounts for pH adjustments for NH₃ dissolution.

Over oceans, NH₃ is entirely converted into NH${}_{\mathrm{4}}^{+}$, while SO${}_{\mathrm{4}}^{\mathrm{2}-}$ remains partially neutralized due to low NH₃ emissions and high SO${}_{\mathrm{4}}^{\mathrm{2}-}$ levels from biogenic dimethyl sulfide (DMS) emissions and NO_x and SO₂ from shipping. Marine SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and NO${}_{\mathrm{3}}^{-}$ are dominant in the supermicron modes, with mass fractions of 70 % and 58 %, respectively. Over land, total ammonia (the sum of NH₃ and NH${}_{\mathrm{4}}^{+}$) generally exceeds SO${}_{\mathrm{4}}^{\mathrm{2}-}$, enabling more complete neutralization, except in regions such as northern Russia, central Africa, and the Arabian Peninsula. Terrestrial NO${}_{\mathrm{3}}^{-}$ is also largely neutralized by NH₃, resulting in over 60 % of SNA concentrated in the submicron mode. In polluted areas such as East and South Asia, submicron SNA fractions exceed 70 %, with South Asia exhibiting nearly 90 % submicron NH${}_{\mathrm{4}}^{+}$. In contrast, the Middle East, dominated by dust and with minor NH₃ emissions, shows incomplete neutralization of SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and NO${}_{\mathrm{3}}^{-}$. Here, nearly 70 % of NO${}_{\mathrm{3}}^{-}$ resides in the supermicron modes, while NH${}_{\mathrm{4}}^{+}$ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ dominate the submicron mode with respective mass fractions of 96 % and 79 %.

In the Northern Hemisphere, terrestrial aerosols are generally more acidic than marine aerosols, except in desert regions. Remote marine and desert aerosols remain neutral due to the alkaline influence of sea salt and non-volatile cations in dust, which enhance water uptake and neutralize SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and NO${}_{\mathrm{3}}^{-}$. However, at high latitudes, marine aerosols become more acidic due to the long-range transport of anthropogenic pollutants from continental sources. The 0–1 µm size range exhibits higher pH values than the 1–2.5 µm range in many regions, influenced by several factors. These include the assumption of a thermodynamically stable aerosol phase in the ISORROPIA model, the high sensitivity of aerosol acidity to NH₃ in the 0–1 µm range, and observed lower concentrations of acidic components (SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and NO${}_{\mathrm{3}}^{-}$) in this size range compared to larger sizes.

An 18 % increase in NH₃ emissions over land leads to a significant increase in SNA concentrations in the 1–2.5 µm size range (NH${}_{\mathrm{4}}^{+}$: 104 %; NO${}_{\mathrm{3}}^{-}$: 41 %; SO${}_{\mathrm{4}}^{\mathrm{2}-}$: 23 %), while coarse-mode SNA (2.5–10 µm) decreases. However, changes in size-resolved pH remain minimal, with the largest increase of 0.24 pH units occurring in the 0–1 µm range. In regions with low SNA, such as South America and Greenland, increased NH₃ emissions only lead to higher NH₃ concentrations due to the limited availability of HNO₃ and H₂SO₄ for SNA formation. Conversely, in NH₃-rich regions such as East Asia and Europe, reductions in NH₃ emissions trigger compensatory effects, as excess NH₃ interacts with available SO₂ and NO_x to maintain SNA formation despite the reduced NH₃ supply. This highlights the buffering capacity of NH₃ in stabilizing aerosol acidity in densely populated regions. However, it is important to note that this buffering effect may be partially offset by concurrent increases in SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and NO${}_{\mathrm{3}}^{-}$ concentrations.

In summary, this study underscores the critical influence of NH₃ emissions on global aerosol composition and acidity, with pronounced impacts on the size-resolved composition and properties of SNA aerosols. NH₃ emissions significantly modulate aerosol acidity, particularly in submicron ranges, highlighting the sensitivity of fine-mode aerosols to NH₃ levels. By buffering changes in aerosol pH, NH₃ emissions contribute to maintaining a relatively stable aerosol acidity, especially in densely populated and heavily industrialized regions. These findings emphasize the importance of accurately representing NH₃ dynamics in models for predicting atmospheric chemistry and climate interactions.