The shifting of secondary inorganic aerosol formation mechanisms during haze aggravation: the decisive role of aerosol liquid water

Although many considerable efforts have been done to reveal the driving factors on haze aggravation, however, the roles of aerosol liquid water (ALW) in secondary inorganic aerosol (SIA) formation were mainly focused on the condition of aerosol liquid water content (ALWC) < 100 µg m⁻³. Based on the in situ high-resolution field observations, this work studied the decisive roles and the shifting of secondary inorganic aerosol formation mechanisms during haze aggravation, revealing the different roles of ALWC on a broader scale (∼500 µg m⁻³) in nitrate and sulfate formation induced by aqueous chemistry in the ammonia-rich atmosphere. The results showed that chemical domains of perturbation gas limiting the generation of secondary particulate matter presented obvious shifts from a HNO₃-sensitive to a HNO₃- and NH₃-co-sensitive regime with the haze aggravation, indicating the powerful driving effects of ammonia in the ammonia-rich atmosphere. When ALWC < 75 µg m⁻³, the sulfate generation was preferentially triggered by the high ammonia utilization and then accelerated by nitrogen oxide oxidation from clean to moderate pollution stages, characterized by nitrogen oxidation ratio (NOR) < 0.3, sulfur oxidation ratio (SOR) < 0.4, ammonia transition ratio (NTR) < 0.7 and the moral ratio of ${\mathrm{NO}}_{\mathrm{3}}^{-}/{\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}=\mathrm{2}:\mathrm{1}$. When ALWC > 75 µg m⁻³, the aqueous-phase chemistry reaction of SO₂ and NH₃ in ALW became the prerequisite for SIA formation driven by Henry's law in the ammonia-rich atmosphere during heavy and serious stages, characterized by high SOR (0.5–0.9), NOR (0.3–0.5) and NTR (>0.7), as well as the high moral ratio of ${\mathrm{NO}}_{\mathrm{3}}^{-}/{\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}=\mathrm{1}:\mathrm{1}$. A positive feedback of sulfate on nitrate production was also observed in this work due to the shift in ammonia partitioning induced by the ALWC variation during haze aggravation. It implies the target controlling of haze should not simply focus on SO₂ and NO₂, but more attention should be paid to gaseous precursors (e.g., SO₂, NO₂, NH₃) and aerosol chemical constitution during different haze stages.

1 Introduction

Fine particulate matter (PM_2.5) presented a close link with several environmental issues, such as visibility reduction and climate change (Zhang et al., 2015; Shang et al., 2020; Wang et al., 2016, 2020; Nozière et al., 2010). Epidemiological studies have stated the association of PM with public health and even adverse birth outcomes (Gwynn et al., 2000; Lavigne et al., 2016; Zhao et al., 2020). As the most abundant secondary inorganic aerosols (SIAs) in PM_2.5 during Chinese winter haze episodes (Fu and Chen, 2017; Liu et al., 2019), the formation of sulfate and nitrate plays a key role during haze aggravation, as well as having an impact on the oxidants in gas and aqueous phases; the characteristics of pre-existing aerosols, fog and cloud; and meteorological conditions. Recently, aerosol liquid water content (ALWC) was reported as being associated with SIA formation, especially sulfates and nitrates, during the haze periods (Wu et al., 2018; B. Zheng et al., 2015; Wang et al., 2016; Cheng et al., 2016; Carlton and Turpin, 2013; Nguyen et al., 2014; Xue et al., 2014; Tan et al., 2017; Z. Liu et al., 2017). Atmospheric aerosol liquid water (ALW), which is determined by ambient relative humidity (RH), has been proposed as a container since it could provide the reaction medium for the multiphase chemistry during the haze process (Ansari and Pandis, 2000; Shiraiwa et al., 2012; Davies and Wilson, 2015). The roles of ALWC on the generation of particulate sulfate (Wang et al., 2016; Cheng et al., 2016) and global secondary organic aerosols (Hodas et al., 2014; McNeill, 2015; Wong et al., 2015) were reported. Thus, fully understanding ALW and its roles during haze aggravation is fundamentally important for atmospheric physicochemical processes, especially the liquid chemical transformation of SO₂ and NO_x in ALW.

Ammonia is the most important alkaline gas, neutralizing with acidic species to form ammonium salts. Due to little attention having been paid to NH₃ emissions by the Chinese government, atmospheric NH₃ experienced a significant increasing trend (Ge et al., 2019; Tan et al., 2017). Although the increase in atmospheric NH₃ is beneficial to reduce atmospheric acidity (Liu et al., 2019), its chemical behavior in regional haze formation is still being debated. Cheng et al. (2016) indicated that the fast transform of gaseous SO₂ to particle sulfate under polluted conditions is attributed to the neutralization of NH₃, which raises particle pH and thereby facilitated the aqueous oxidation of S(VI) by NO₂. Fang et al. (2017) stated that NH₃ partitioning significantly modified aerosol pH and thereby adjusted the partitioning of SO₂ and NO₂. Although the role of NH₃ has been identified from a theoretical perspective, the lack of NH₃ emission control sets barriers for a more effective reduction of PM_2.5. Therefore, it is urgent to fully understand the reactive gases' behavior and the chemical mechanisms of SIA formation during different pollution stages, which will be helpful in proposing reasonable strategies for each stage.

So far, the SIA formation has been extensively studied during short-term, continuous or persistent haze episodes, and several heterogeneous and homogeneous oxidation pathways for sulfate and nitrate formation have been proposed (Guo et al., 2014, 2017; G. J. Zheng et al., 2015; Huang et al., 2014; Liu et al., 2019, 2021; Yao et al., 2020; Zhou et al., 2018). In the ammonia-rich atmosphere, NH₃ partitioning significantly modified aerosol pH, adjusted the partitioning of SO₂ and NO₂ (Fang et al., 2017), and promoted the aqueous oxidation of S(VI) by NO₂ (Wang et al., 2016; Cheng et al., 2016). Although many considerable efforts have been done to reveal the driving factors of haze aggravation, however, the roles of ALW in SIA formation were mainly focused on the condition of ALWC < 100 µg m⁻³ (Nenes et al., 2020; Wu et al., 2018; Bian et al., 2014; Jin et al., 2020). Therefore, the roles of ALWC on a broader scale and the shifting mechanisms of secondary inorganic aerosol formation during haze aggravation in the ammonia-rich atmosphere need to be understood in depth. Based on a continuous observations with 1 h resolution from December 2019 to January 2020, this work discusses the shift in the dominant mechanisms with ALWC variation during the time window of haze aggravation processes, which will be helpful in proposing more effective PM_2.5 control strategies for each pollution stage.

2 Sampling and experiment methods

2.1 Description of sampling site

Hohhot, the capital city of the Inner Mongolia Autonomous Region, is the central city of the Hohhot–Baotou–Ordos group, as well as an important northern Chinese city with a population of more than 3.126 million and an area of 17 224 km² (Fig. 1). This region has a continental climate with marked seasonality changes, which is characterized by a long-lasting cold, humid winter and short other seasons. Thereby, to survive the cold season, approximately half a year of coal-fired heating events (15 October to the following 15 April) occur, which emit gaseous pollutants, as well as PM, around the clock. The main industries include thermal power plants, the coal-energy-based biochemical industry, the dairy industry, the petrochemical industry, etc., which also emit atmospheric pollutants continuously. Thus, high concentrations of PM pollution dominated the major contamination cases during the winter season (data obtained from the Department of Ecology and Environment of Inner Mongolia Autonomous Region, http://sthjt.nmg.gov.cn/, last access: 16 October 2020) and gradually emerged as the limiting factor to regional ambient air quality and human health.

Figure 1 Map of sampling sites and coal-fired enterprises.

In this study, observations were conducted at the Inner Mongolia Environmental Monitoring Center (40^∘49^′22^′′ N, 111^∘45^′2^′′ E) on top of a 16-story building (∼40 m a.g.l. – above ground level) located in the eastern part of the downtown area near the People's Government of Inner Mongolia Autonomous Region building near the 2nd Ring Road from 1 December 2019 to 31 January 2020. Residential and administrative regions were characterized by the major functional domain near the sampling site, with no direct industrial regions nearby.

2.2 Data acquisition and analysis methods

2.2.1 Data acquisition

An online ion-chromatograph instrument (MARGA ADI 2080, Metrohm Applikon, Switzerland) was employed to simultaneously determine the water-soluble inorganic ions (Na⁺, ${\mathrm{NH}}_{\mathrm{4}}^{+}$, Mg²⁺, Ca²⁺, K⁺, Cl⁻, F⁻, ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$, ${\mathrm{NO}}_{\mathrm{3}}^{-}$) in PM_2.5 and corresponding trace gases (SO₂, HNO₂, HNO₃, HCl, NH₃). This instrument has been widely used in previous work (Rumsey et al., 2014; Nie et al., 2015; Huang et al., 2020), and the details are listed in the Supplement (Sect. S1.1). Correspondingly, gaseous pollutants (e.g., NO_x, CO, PM₁, PM_2.5, PM₁₀) and meteorological datasets (e.g., wind speed, wind direction, RH, temperature), as well as the adopted models, can be found in our previous work (Xie et al., 2021). In addition, peroxyacetyl nitrates (PANs), nitrous oxide (N₂O) and solar spectrophotometry were measured by PAN 100 (Focused Photonics Inc.), N₂O monitor (LSE Monitors) and CE-318T (CIMEL), respectively.

2.2.2 Analysis methods

Generally, the sulfur oxidation ratio (SOR) and the nitrogen oxidation ratio (NOR) were calculated as follows, which were used to indicate the contribution of secondary transformation during the haze events (Song et al., 2007; Zhou et al., 2018).

$$\begin{array}{}\text{(1)}& {\displaystyle }& {\displaystyle }\mathrm{SOR}={\displaystyle \frac{n\left({\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}\right)}{n\left({\mathrm{SO}}_{\mathrm{2}}\right)+n\left({\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}\right)}}\text{(2)}& {\displaystyle }& {\displaystyle }\mathrm{NOR}={\displaystyle \frac{n\left({\mathrm{HNO}}_{\mathrm{3}}\right)+n\left({\mathrm{NO}}_{\mathrm{3}}^{-}\right)}{n\left({\mathrm{NO}}_{\mathrm{2}}\right)+n\left({\mathrm{HNO}}_{\mathrm{3}}\right)+n\left({\mathrm{NO}}_{\mathrm{3}}^{-}\right)}}\end{array}$$

Meanwhile, as an indicator of ammonia conversion efficiency, the ammonia transition ratio (NTR) was calculated as the following equation (all units are µg m⁻³).

$$\begin{array}{}\text{(3)}& \mathrm{NTR}={\displaystyle \frac{{\mathrm{NH}}_{\mathrm{4}}^{+}/\mathrm{18}}{{\mathrm{NH}}_{\mathrm{4}}^{+}/\mathrm{18}+{\mathrm{NH}}_{\mathrm{3}}/\mathrm{22.4}}}\end{array}$$

In addition, as the fractions of ammonia, nitrate and sulfate in deliquesced aerosol, ε (${\mathrm{NO}}_{\mathrm{3}}^{-}$), ε (${\mathrm{NH}}_{\mathrm{4}}^{+}$) and ε (${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$) were expressed as follows.

$$\begin{array}{}\text{(4)}& {\displaystyle }& {\displaystyle }\mathit{\epsilon }\left({\mathrm{NO}}_{\mathrm{3}}^{-}\right)={\displaystyle \frac{n\left({\mathrm{NO}}_{\mathrm{3}}^{-}\right)}{n\left({\mathrm{HNO}}_{\mathrm{3}}\right)+n\left({\mathrm{NO}}_{\mathrm{3}}^{-}\right)}}\text{(5)}& {\displaystyle }& {\displaystyle }\mathit{\epsilon }\left({\mathrm{NH}}_{\mathrm{4}}^{+}\right)={\displaystyle \frac{n\left({\mathrm{NH}}_{\mathrm{4}}^{+}\right)}{n\left({\mathrm{NH}}_{\mathrm{3}}\right)+n\left({\mathrm{NH}}_{\mathrm{4}}^{+}\right)}}\text{(6)}& {\displaystyle }& {\displaystyle }\mathit{\epsilon }\left({\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}\right)={\displaystyle \frac{n\left({\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}\right)}{n\left({\mathrm{SO}}_{\mathrm{2}}\right)+n\left({\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}\right)}}\end{array}$$

2.2.3 Aerosol pH

In this work, a widely used thermodynamic model, ISORROPIA-II (Song et al., 2018; Gao et al., 2020), was employed to establish aerosol acidity. Including the concentrations of water-soluble ions (WSIs) in PM_2.5 and gaseous pollution (e.g., NH₃, HCl), the simultaneously measured temperature and RH data were imported into its Na⁺–K⁺–Ca²⁺–Mg²⁺–${\mathrm{NH}}_{\mathrm{4}}^{+}$–${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$–${\mathrm{NO}}_{\mathrm{3}}^{-}$–Cl⁻–H₂O aerosol system. According to a previous study (Song et al., 2018) and our data profiles, “forward mode” and “metastable state” were selected in the model of ISORROPIA-II to calculate aerosol acidity (${\mathrm{H}}_{\mathrm{air}}^{+}$, H⁺ loading per volume air – µg m⁻³) and ALWC. Then the aerosol pH was calculated by the following equation.

$$\begin{array}{}\text{(7)}& \mathrm{pH}=-{\log }_{\mathrm{10}}{\displaystyle \frac{\mathrm{1000}{\mathrm{H}}_{\mathrm{air}}^{+}}{\mathrm{ALWC}}}\end{array}$$

The concentrations of NH₃, ${\mathrm{NH}}_{\mathrm{4}}^{+}$, ${\mathrm{NO}}_{\mathrm{3}}^{-}$ and ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ modeled by this model significantly correlated with their measured values with correlation coefficients of 0.971–0.999, indicating the accuracy and acceptability of the model in this work (Fig. S1 in the Supplement).

2.2.4 Heterogeneous sulfate production

Due to the necessity of precise ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ generation, heterogeneous sulfate production (P_het) was parameterized and calculated according to the following equation (Jacob, 2000; B. Zheng et al., 2015).

$$\begin{array}{}\text{(8)}& \begin{array}{rl}{P}_{\mathrm{het}}=& {\displaystyle \frac{\mathrm{3600}{\mathrm{sh}}^{-\mathrm{1}}\times \mathrm{96}\mathrm{g}{\mathrm{mol}}^{-\mathrm{1}}\times P}{R\times T}}{\left({\displaystyle \frac{R_{\mathrm{p}}}{D_g}}+{\displaystyle \frac{\mathrm{4}}{\mathit{\upsilon }\mathit{\gamma }}}\right)}^{-\mathrm{1}}\\ & S_p\left[{\mathrm{SO}}_{\mathrm{2}}\left(g\right)\right],\end{array}\end{array}$$

where P_het is presented in micrograms per cubic meter per hour (µg m⁻³ h⁻¹), 3600 sh⁻¹ is the time conversion factor, 96 g mol⁻¹ is the molar mass of ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$, P is atmospheric pressure (in kPa), R is the gas constant with the value of 8.31 Pa m⁻³ mol⁻¹ K⁻¹, T is the temperature (in K), R_p represents the radius of aerosol particles (m), D_g is the SO₂ molecular diffusion coefficient, and υ is the mean molecular speed of SO₂ with the typical tropospheric value of $\mathrm{2}\times {\mathrm{10}}^{-\mathrm{5}}$ m² s⁻¹ and 300 m s⁻¹. γ is the uptake coefficient of SO₂ on aerosols, and S_p is the aerosol surface area per unit volume of air (m² m⁻³) (Jacob, 2000). PM_2.5 mass concentrations (µg m⁻³) and mean radius (m) during the campaign were roughly calculated utilizing the following empirical formula published by Guo et al. (2014).

$$\begin{array}{}\text{(9)}& R_{\mathrm{p}}=\left(\mathrm{0.254}\times C_{\left({\mathrm{PM}}_{\mathrm{2.5}}\right)}+\mathrm{10.259}\right)\times {\mathrm{10}}^{-\mathrm{9}}\end{array}$$

The mean density of particles ρ was calculated as 1.5×10⁶ g m⁻³ using the volume and surface area formulas of a sphere (Guo et al., 2014). S_p was estimated from the following formula.

$$\begin{array}{}\text{(10)}& S_p={\displaystyle \frac{C_{\left({\mathrm{PM}}_{\mathrm{2.5}}\right)}\times {\mathrm{10}}^{-\mathrm{6}}\mathrm{g}\mathrm{\mu }{\mathrm{g}}^{-\mathrm{1}}}{\mathrm{4}/\mathrm{3}\cdot \mathit{\pi }{R}_{\mathrm{p}}^{\mathrm{3}}\cdot \mathit{\rho }}}\cdot \mathrm{4}\mathit{\pi }{R}_{\mathrm{p}}^{\mathrm{2}}\end{array}$$

Relative-humidity-dependent γ values were derived according to B. Zheng et al. (2015) during the campaign in this work and are shown as the following formula.

$$\begin{array}{}\text{(11)}& \begin{array}{rl}& \mathit{\gamma }=\\ & \left\{\begin{array}{ll}\mathrm{2}\times {\mathrm{10}}^{-\mathrm{5}},& \mathrm{\Psi }\le \mathrm{50}\mathit{\%},\\ \mathrm{2}\times {\mathrm{10}}^{-\mathrm{5}}+\frac{\mathrm{5}\times {\mathrm{10}}^{-\mathrm{5}}-\mathrm{2}\times {\mathrm{10}}^{-\mathrm{5}}}{\mathrm{100}\mathit{\%}\text{–}\mathrm{50}\mathit{\%}}\times (\mathrm{\Psi }-\mathrm{50}\mathit{\%}),& \mathrm{50}\mathit{\%}\le \mathrm{\Psi }\le \mathrm{100}\mathit{\%},\end{array}\right.\end{array}\end{array}$$

where ψ refers to RH (with the unit of %).

3 Results and discussion

Based on National Ambient Air Quality Standards of China (HJ633-2012) (https://www.mee.gov.cn/ywgz/fgbz/bz/bzwb/jcffbz/201203/t20120302_224166.shtml, last access: 28 May 2020), an air quality index (AQI) was introduced in this work to classify pollution levels (Wang et al., 2015; Kanchan et al., 2015; Xu et al., 2017) and outline the characteristics of atmospheric pollutants. Briefly, daily concentrations of PM_2.5 ranging from 0–75, 75–115, 115–150, 150–250 and >250 µg m⁻³ were classified as clean (C), lightly polluted (L), moderately polluted (M), heavily polluted (H) and seriously polluted (S) periods, respectively.

Figure 2 Nitrate to sulfate molar ratio as a function of ammonium to sulfate molar ratio.

3.1 The observed evidence for the ammonia-rich atmosphere

The characteristics of atmospheric pollutants and meteorological parameters during the studied period were summarized (Sect. S2.1). In this work, molar ratios of ${\mathrm{NH}}_{\mathrm{4}}^{+}$ vs. anions were used to identify the chemical species of ammonium salts (Zhou et al., 2018; Wang et al., 2021; Z. Liu et al., 2017; Shi et al., 2019). The calculated results (Sect. S2.2) showed the predominant chemical species of ammonium gradually varied from the coexistence of ammonium sulfate ((NH₄)₂SO₄) and ammonium nitrate (NH₄NO₃) to the coexistence of (NH₄)₂SO₄, NH₄NO₃ and ammonium chloride (NH₄Cl) with haze aggravation (Fig. S5). Further, the slope of the fitted equation between excess ${\mathrm{NH}}_{\mathrm{4}}^{+}$ and anions was still lower than the 1:1 line after all the measured anions were neutralized, indicating an ammonia-rich atmosphere (Fig. S5c). Considering the national demand for ultra-low-emissions activities (nearly 2 times lower than the former national standard) for gaseous pollutants, the heavy usage of ammonia-containing compounds in the process of desulfurization and denitrification (Solera García et al., 2017; Tan et al., 2017) at widely distributed thermal power plants (>300 000 kWh) and the close-set coal-fired heating stations (Fig. 1) resulting in ammonia fugitives provided a reasonable explanation for this ammonia-rich atmosphere. Even though retrofitting because of the national demand for ultra-low-emissions activities has been completed, distributed coal-based enterprises could also emit substantial SO₂ and NO₂ and, being subjected to heterogeneous reactions, further generate sulfate and nitrate, which aggravated the haze events (Fig. S7a and b).

To show the reaction between ammonia and nitric acid and the other formation processes of nitrate in different (relative) concentrations of sulfate, the data of previous studies and different pollution levels (C, L, M, H, S) in this work are plotted in Fig. 2. When $\left[{\mathrm{NH}}_{\mathrm{4}}^{+}\right]/\left[{\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}\right]\le \mathrm{1.5}$, the nitrate formation was associated with crustal elements rather than ammonium; when $\left[{\mathrm{NH}}_{\mathrm{4}}^{+}\right]/\left[{\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}\right]>\mathrm{1.5}$, the homogeneous gas-phase reactions between NH₃ and HNO₃ became the major pathway for atmospheric ammonia to form NH₄NO₃ (Pathak et al., 2009; Liu et al., 2019). The results illustrated that the ammonia-rich regimes not only were found in Hohhot but were also observed in Guangzhou (Huang et al., 2011), Chengdu (Huang et al., 2018), Lanzhou, western and eastern USA, India, Ireland, Europe, Qingdao, Italy and Lin'an (Pathak et al., 2009) in recent decades (Fig. 2). It suggested that atmospheric oxidative modifications in the ammonia-rich atmosphere should be a widespread atmospheric issue with significant contributions to SIA generation. It is worth noting that the slopes of our data were becoming steeper, coupled with the ${\mathrm{NO}}_{\mathrm{3}}^{-}/{\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ ratio change from ∼4 to about 1, as the pollution levels increased. The high PM_2.5 nitrate concentration during heavy and serious stages cannot be explained by the homogeneous gas-phase reaction involving ammonia and nitric acid, which may be associated with the heterogeneous reaction of ALW on the surface of the preexisting aerosols.

Figure 3 (a) Correlations between ALWC and ratios of PM response to relative humidity evolutions; (b) correlations between ALWC and NOR and SOR in their response to different pollution stages; the pentagrams are colored as a function of relative humidity.

3.2 Driving mechanisms of SIA formation

3.2.1 Aerosol liquid water

Our results showed that SOR, NOR and SIAs in PM_2.5 presented increasing trends with the increasing ALWC during the five pollution levels. The variation in predominant chemical species of ammonium (Fig. 2) indicated more SIAs will be generated on particles with the simultaneous increase in ALWC and PM_2.5 (Fig. 3b). Theoretically, the inorganic compound conversion was enhanced via aqueous-phase chemistry on moist particles owing to the continuous partitioning of gaseous pollutants (e.g., SO₂, NO₂, N₂O₅) in ALW and then disrupted the equilibrium between the gaseous and condensed phases, resulting in the aggravation of haze events (Xue et al., 2014; Wu et al., 2018; G. J. Zheng et al., 2015; Wang et al., 2016). Considering seasonal heating characteristics, the shift in the equilibrium between gaseous and condensed phases was enhanced with the increasing atmospheric pollutant concentrations due to the coal-fired combustion events in winter. In detail, owing to their hygroscopic nature, the particles must increase their water contents via ALW along with RH (Fig. S8a) to maintain thermodynamic equilibrium and water vapor and simultaneously enhance the oxidation and dissolution of precursors in the micro-solution (the ALW) of the particulates. This process elevated the inorganic mass fraction, as well as particulate mass concentrations, during different pollution stages (Fig. S8b) (Bertram et al., 2009; Wang et al., 2016; B. Zheng et al., 2015; Cheng et al., 2016). Due to the larger affinity of H₂SO₄ for NH₃ (aqueous – aq), sulfate was preferentially and fully neutralized by ammonium in the ammonia-rich atmosphere to generate the non-volatile nature of (NH₄)₂SO₄ (Z. Liu et al., 2017; Zhou et al., 2018; Wang et al., 2021). Thus, SOR presented higher exponential growth, with the elevated ALWC coupling with more sulfate production (Fig. 3b). Concomitantly, the preferentially generated (NH₄)₂SO₄ further enhanced the hygroscopicity of particulate matter, in turn helping more ammonia partition into moist particulate matter and generating ammonium salts, which accelerated haze aggravation (Figs. S6 and S8c). Thus, most important of all, the sharp increase in inorganic compounds associated with the elevated ALWC significantly modified the specific surface area of particulates and further accelerated the hygroscopic aerosol growth, which simultaneously provided a substrate for the ensuing heterogenous reaction and accelerated the evolution of haze events. Previous work reported that particles of different modes made different contributions to ALWC, with the contributions of nuclear, Aitken, accumulation and coarse modes assessed at <1 %, 3 %, 85 % and 12 %, respectively, indicating that the contribution of accumulation mode particles to ALWC dominated among all the aerosol particle modes (Tan et al., 2017). It indicated that secondary aerosol formation mainly happens on these fine particles, as the surface area and volume of the fine particles are much larger than those of the coarse particles. Thus, the observed significant correlations of ALWC with the ratios (PM_1.0 $/$ PM_2.5 and PM_2.5 $/$ PM₁₀) in this work also indicated that the hygroscopic growth of fine particulate matter (D_p≤2.5 µm) is strongly associated with ALWC (Fig. 3a). Both previous work and our monitoring results suggested that the ratios of PM_1.0 $/$ PM_2.5 and PM_2.5 $/$ PM₁₀ presented the potential possibility to index the hygroscopic growth of particulate matter.

Figure 4 Chemical domains of aerosol response to ammonia and nitrate emissions.

3.2.2 Perturbation gases

Due to the strict control of SO₂, atmospheric concentrations of NO₂ and NH₃ gradually became the decisive reactive precursors for regional atmospheric secondary particulate matter generation. Thus, the state-of-the-art framework proposed by Nenes et al. (2020) was carried out to examine the chemical domain classifications and the decisive precursor based on the datasets of previous studies (Nenes et al., 2020) and this work (Fig. 4). Due to the thermodynamically stable property of the preferentially generated (NH₄)₂SO₄, the semi-volatile NH₄NO₃ dominates the partitioning of NH₃^T (sum of NH₃ and ${\mathrm{NH}}_{\mathrm{4}}^{+}$, same as NO₃^T) and NO₃^T. Although aqueous ${\mathrm{NO}}_{\mathrm{3}}^{-}$ concentrations varied with haze processes, the calculated ε (NO₃^T) (detailed calculated method can be found in Sect. S1.2), which was an equilibrium parameter between gaseous HNO₃ and particle-phase ${\mathrm{NO}}_{\mathrm{3}}^{-}$ (Guo et al., 2016; Fang et al., 2017), consistently presented full loadings of nitrate on the existing particulates during the studied period (Fig. S9a and b). This could provide clear evidence for the initial HNO₃-sensitive area and continuous control of HNO₃ during the studied periods. However, with haze aggravation, significant elevated ALWC resulted in more precursors partitioning into micro-droplets to maintain water vapor. This process induced a positive shift in HNO₃ dissolution equilibrium and led to more HNO₃ partitioned on particles driven by Henry's law (e.g., HNO_3(g)↔HNO_3(aq), K_H=2.07 mol (L Pa)⁻¹. Meanwhile, HNO₃ and HONO could also be produced through the reactions of ${\mathrm{NO}}_{\mathrm{2}}+{\mathrm{H}}_{\mathrm{2}}\mathrm{O}\stackrel{\mathrm{Het}}{⟶}{\mathrm{HNO}}_{\mathrm{3}}+\mathrm{HONO}$ (Huang et al., 2018). Accordingly, the OH radicals generated by HONO photolysis also contributed to these oxidation processes (Yue et al., 2020; Zhu et al., 2020). These aqueous oxidation processes were evidenced by the observation of significantly elevated HONO and PANs during the haze aggravation (Fig. S7c and d). Accordingly, the equations of ${\mathrm{NH}}_{\mathrm{4}}^{+}+{\mathrm{NO}}_{\mathrm{3}}^{-}+{\mathrm{H}}^{+}+{\mathrm{OH}}^{-}{\mathrm{NH}}_{\mathrm{4}}{\mathrm{NO}}_{\mathrm{3}}+{\mathrm{H}}_{\mathrm{2}}\mathrm{O}$ and ${\mathrm{NH}}_{\mathrm{4}}^{+}+{\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}+{\mathrm{H}}^{+}+{\mathrm{OH}}^{-}\to ({\mathrm{NH}}_{\mathrm{4}}{)}_{\mathrm{2}}{\mathrm{SO}}_{\mathrm{4}}+{\mathrm{H}}_{\mathrm{2}}\mathrm{O}$ were shifted to generate more NH₄NO₃ and (NH₄)₂SO₄ (Nenes et al., 2020; Xie et al., 2020) due to the driving force of more ammonia partitioned in elevated ALWC (${\mathrm{NH}}_{\mathrm{3}}+{\mathrm{H}}_{\mathrm{2}}\mathrm{O}{\mathrm{NH}}_{\mathrm{3}}\cdot {\mathrm{H}}_{\mathrm{2}}\mathrm{O}$, ${\mathrm{NH}}_{\mathrm{3}}\cdot {\mathrm{H}}_{\mathrm{2}}\mathrm{O}{\mathrm{NH}}_{\mathrm{4}}^{+}+{\mathrm{OH}}^{-}$). Therefore, NH₃ and NO_x became the decisive factors for regional atmospheric oxidability in the ammonia-rich regime (Zhai et al., 2021; Tan et al., 2017; Liu et al., 2019; Li et al., 2019).

Figure 5 Correlations between (a) O_x and SOR, (b) NO₂ and SOR, and (c) NOR and ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$.

Generally, both NH₃ and HNO₃ were the limiting factors governing the generation of aerosols for cities of North China due to high loadings of atmospheric ammonia, while NH₃ governed PM formation in the southeast USA (SAS) (Zhao et al., 2020). Thanks to the raw data of Shenzhen (SZ) (Wang et al., 2022), we also calculated the ALWC and aerosol pH using ISORROPIA-II, and the scatters of SZ suggested an obvious chemical transition from a HNO₃–NH₃-sensitive regime to a NH₃-sensitive regime due to the differently originating air masses. Although both cities are located in the USA, the findings in California (CNX) were quite interesting and distributed in the insensitive region and the combined NH₃–HNO₃-sensitive region due to the moderate NH₃ levels and the complicated atmospheric conditions during the observations (Nenes et al., 2020). In our work, the data points ($\mathrm{541}/\mathrm{744}$) in summer (pH = 3.47±1.29) mostly laid in the HNO₃-sensitive region, while chemical domains of perturbation gas limiting the generation of secondary particulate matter presented obvious shifts from a HNO₃-sensitive to a HNO₃- and NH₃-co-sensitive regime with the haze aggravation in winter. Some data points of this work laid in the combined NH₃–HNO₃ region in winter owing to the more acidic conditions. Under the stable pH of aerosols in winter in Hohhot (pH = 4–5), the more important detail is that a fraction of points will distribute in the combined NH₃–HNO₃ region when ALWC > 75 µg m⁻³, which may be attributed to the aqueous chemical transformation driven by Henry's law mentioned above due to the elevated ALWC. Comparatively, the aerosol pH in summer was significantly lower than that in winter in Hohhot. Compared to TJ (Tianjin) and SZ (Shenzhen), the aerosols pH of Hohhot in winter was also significantly higher (Fig. 4) due to the acidity of atmospheric PM being largely dependent on the alkaline material in surface soils in arid and semi-arid regions and the elevated atmospheric ammonia. In terms of seasonal characteristics, the higher temperature in summer elevates the volatility of NH₄NO₃ and dominates the partitioning of NH₃^T in the atmospheric phase to decrease the pH of aerosols. Therefore, as can be seen from Fig. 4, the data points measured in winter in Hohhot are characterized by higher pH and lower ALWC than those in summer (Hohhot, SAS, CNX, SZ). According to the framework of Nenes et al. (2020), the transition points of Hohhot (whether winter or summer) between NH₃-dominated and HNO₃-dominated sensitivity also occur at a pH of around 2 but at lower levels of ALWC. Theoretically, it should be associated with the higher aridity of Hohhot, located in the arid and semi-arid regions of China. Our results provide the evidence for “the additional insight” proposed by Nenes et al. (2020) that the transition ALWC varies with season change and the aridity of sites in response to seasonal variability and climate change. Although this effort could provide a sound explanation for limiting gaseous pollutants in PM formation, mechanisms on their chemical domains, especially the roles of ALW in different locations under various conditions, need further study in the future.

3.2.3 The shifting of SIA formation mechanisms driven by ALW

It is worth noting that two independent correlations were found between SOR and odd oxygen (O_x, ${\mathrm{O}}_x={\mathrm{NO}}_{\mathrm{2}}+{\mathrm{O}}_{\mathrm{3}}$) during the aggravating processes of haze events, indicating the differential mechanisms of atmospheric oxidability in sulfate generation at different stages (Fig. 5a). Different to inefficient homogeneous sulfate oxidation efficiency (Fig. S10), significant-correlation pairs of NO₂ with SOR (Fig. 5b) and NOR with ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ (Fig. 5c) suggested the haze aggravation was largely related to the regional NO₂ levels due to the regulating effects on atmospheric oxidability. Thus, the aqueous-phase oxidation of S(IV) by NO₂ (aq) was triggered and accelerated by the increasing ALWC and the following equation (Yao et al., 2020; Wang et al., 2016) (Fig. S11a).

$$\begin{array}{}\text{(R1)}& \mathrm{S}\left(\mathrm{IV}\right)+{\mathrm{NO}}_{\mathrm{2}}\left(\mathrm{aq}\right)+{\mathrm{H}}_{\mathrm{2}}\mathrm{O}\to \mathrm{S}\left(\mathrm{VI}\right)+{\mathrm{H}}^{+}+{\mathrm{NO}}_{\mathrm{2}}^{-}\end{array}$$

Meanwhile, sharp logarithmic increases between SOR and ${\mathrm{NH}}_{\mathrm{4}}^{+}$ were also observed from the clean to moderate pollution stages (Fig. S12). Due to the joint effects of the ammonia-rich atmosphere and ammonia's extremely water-soluble property, sufficient hydroxide generated by ammonia dissolution forced the NO₂ partitioned in ALW to maintain pH through neutralization, producing sulfate via Reaction R1. Thus, the following Reaction R2 was derived by considering the processes of ammonia hydrolysis, which is evidenced by Fig. S11b.

$$\begin{array}{}\text{(R2)}& \begin{array}{rl}\mathrm{S}\left(\mathrm{IV}\right)& +{\mathrm{NO}}_{\mathrm{2}}\left(\mathrm{aq}\right)+{\mathrm{NH}}_{\mathrm{3}}\left(\mathrm{aq}\right)+{\mathrm{H}}_{\mathrm{2}}\mathrm{O}\\ & \to \mathrm{S}\left(\mathrm{VI}\right)+{\mathrm{NH}}_{\mathrm{4}}^{+}+{\mathrm{NO}}_{\mathrm{2}}^{-}+{\mathrm{H}}^{+}+{\mathrm{NO}}_{\mathrm{3}}^{-}\end{array}\end{array}$$

Generally, NOR < 0.1 means insignificant nitrogen oxide oxidation; therefore the observed regime shift in nitrate and ammonia chemical behavior in sulfate generation suggested sulfate generation was preferentially triggered by the high ammonia utilization and then accelerated by the co-effects of ammonia utilization and nitrogen oxide oxidation (Fig. 5c).

Figure 6 The characteristics and formation mechanisms of SIAs during haze aggravation.

Accordingly, Reaction R2 was activated due to the increased ALWC forcing more ammonia to partition into moist particulate matter driven by Henry's law in the ammonia-rich atmosphere (NH_3(g)→NH_3(aq)) (Fig. S9c) (Clegg et al., 1998; Wu et al., 2018; Xie et al., 2020). Meanwhile, our calculated aqueous-generated ${\mathrm{NO}}_{\mathrm{3}}^{-}$ nicely matched the theoretical aqueous-nitrate generation curve (the solid blue line in Fig. S9b) proposed by Guo et al. (2017), suggesting the pathway of fast sulfate formation from the oxidation of S(IV) by NO₂ to generate HONO (Wang et al., 2020) (Fig. S11) via Reaction R2. As a result, the thermodynamically stable (NH₄)₂SO₄ would be preferentially formed to maintain its water vapor pressure and thermodynamic equilibrium and then trigger the haze formation. Thus, the mentioned effects resulted in a pronounced increase in NH₃ partitioning with the haze aggravation, suggesting the importance of ammonia partitioning in sulfate generation, namely, the NTR-controlled regime with ALWC < 75 µg m⁻³. In summary, when ALWC < 75 µg m⁻³, the sulfate generation was preferentially triggered by high ammonia utilization and then accelerated by nitrogen oxide oxidation from clean to light pollution stages (Fig. 5c) with NOR < 0.3, SOR < 0.4 and NTR < 0.7. In this period, the chemical composition of SIAs was characterized by the moral ratio of ${\mathrm{NO}}_{\mathrm{3}}^{-}:{\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}=\mathrm{2}:\mathrm{1}$ (Fig. 6).

When ALWC > 75 µg m⁻³, the haze was aggravated from moderate to serious stages along with the increasing ALWC. As a result of an increase in ALW, a large amount of H⁺ was dissociated during the generation of ammonium sulfate (Fig. S13a). From light to moderate pollution stages, the solubility of SO₂ driven by Henry's law was self-limiting due to the acidity effect in low ALWC (with ALWC < 75 µg m⁻³). Therefore, low sulfate concentrations were coupled with low ALWC at the beginning of a haze event (Fig. S13a). However, due to the co-effects of elevated ALWC and the hygroscopic nature of pre-generated ammonium sulfate, H⁺ concentrations were diluted and had nearly constant in situ pH with the increase in ALWC during heavy and serious pollution stages (Fig. S14) (Wang et al., 2016; Clifton et al., 1988; Huie and Neta, 1986; Lee and Schwartz, 1983). Hence, the significantly elevated ALWC provided more chances for the partitioning of SO₂, NO₂ and NH₃ in ALW from moderate to serious pollution stages. Theoretically, Henry's constant of NO₂ ($\mathrm{9.74}\times {\mathrm{10}}^{-\mathrm{8}}$ mol (L Pa)⁻¹) is 3–4 orders of magnitude lower than those of SO₂ ($\mathrm{1.22}\times {\mathrm{10}}^{-\mathrm{5}}$ mol (L Pa)⁻¹) and NH₃ ($\mathrm{6.12}\times {\mathrm{10}}^{-\mathrm{4}}$  mol (L Pa)⁻¹); however, it is worth noting that the aqueous-generated ${\mathrm{NO}}_{\mathrm{3}}^{-}$ from moderate to serious stages rapidly increased 2–5 times higher than clean and light stages (Fig. S9b). Meanwhile, according to our monitoring results, the solar spectrophotometry at 380 nm during moderate to serious stages was significantly lower than that in the clean stage (Fig. S15), suggesting the aqueous oxidation of NO₂ was predominant compared to chain photolysis (Huang et al., 2018). Accordingly, it could be deduced that the aqueous-phase chemistry reaction of SO₂ and NH₃ in ALW, driven by Henry's law, became the dominant mechanism for sulfate formation due to more NO₂ being required to take part in the fast sulfate formation with the increase in ALWC in the ammonia-rich atmosphere by Reaction R2. Thus, with the increasing ALWC, high concentrations of sulfate and nitrate with high SOR (0.5–0.9), NOR (0.3–0.5) and NTR (>0.7) induced the haze events to reach heavy and serious levels (Fig. 5c). Simultaneously, the calculated heterogeneous sulfate production rate (Jacob, 2000; McNeill, 2015) (Fig. S16) presented similar trends to the impacts of ammonia on sulfate production during different pollution stages (Xue et al., 2016; Cheng et al., 2016; Liu et al., 2020). It further stated the environmental significance of the partitioning of SO₂ and NH₃ between gas and aqueous (ALW) phases for SIA formation and haze aggravation. Our results provided the evidence of significant negative correlations between HONO and N₂O (Fig. S17) from moderate to serious stages and positive correlations between HONO and SOR (Fig. S11a), highlighting the recently reported secondary aqueous-phase oxidation pathway of SO₂ by HONO from the moderate pollution period ($\mathrm{2}\mathrm{N}\left(\mathrm{III}\right)+\mathrm{2}\mathrm{S}\left(\mathrm{IV}\right)\to {\mathrm{N}}_{\mathrm{2}}\mathrm{O}\uparrow +\mathrm{2}\mathrm{S}\left(\mathrm{VI}\right)+\mathrm{other}\mathrm{products}$) (Wang et al., 2020). In summary, when ALWC > 75 µg m⁻³, the aqueous-phase chemistry reaction of SO₂ and NH₃ in ALW became the prerequisite for SIA formation driven by Henry's law in the ammonia-rich atmosphere during heavy and serious stages with high SOR (0.5–0.9), NOR (0.3–0.5) and NTR (>0.7). In this period, the chemical composition of SIAs was characterized by the moral ratio of ${\mathrm{NO}}_{\mathrm{3}}^{-}:{\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}=\mathrm{1}:\mathrm{1}$ (Fig. 6).

3.2.4 The positive feedback of sulfate on nitrate production

Previous works suggested that the homogeneous reaction of NO₂ with OH radicals during daylight and heterogeneous hydrolysis of N₂O₅ at night were the main routes of nitrate formation during haze episodes (He et al., 2018; Liu et al., 2019, 2020). Unsurprisingly, higher nitrate production rates (Δ${\mathrm{NO}}_{\mathrm{3}}^{-}$, the difference in hour concentrations and matrixing afterwards) were frequently observed under ammonia-rich conditions due to the ammonia-rich regime being more conducive to nitrate generation. However, the high level of nitrate production rates (Δ${\mathrm{NO}}_{\mathrm{3}}^{-}$) was found in the area characterized by high ammonium and low sulfate levels, suggesting that utilizing ammonium and pre-generated sulfate promotes particle-phase nitrate generation (Fig. 7).

Figure 7 Direct observations of nitrate production facilitated by initially generated sulfate. (All data were matrixed to avoid interference caused by dimensionality.)

Here, we proposed a hypothesis about the hydrogen ion concentration to respond to the above observations. As is known to all, apart from the extremely low levels of crustal elements, ammonia is the only alkaline gas to neutralize the acidic gases in the atmosphere and generate ammonium ions (Xie et al., 2020). Thus, the concentrations of particulate sulfate and nitrate are affected by the partitioning of ${\mathrm{NH}}_{\mathrm{4}}^{+}$ and NH₃. Thereby, higher values of Δ${\mathrm{NO}}_{\mathrm{3}}^{-}$ and Δ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ always occurred in the regions with more ammonium ions and were not confused (Figs. 7 and S18). According to both our results and published laboratory work (Wang et al., 2016), the acidity of the particulate matter could be significantly modified by the bulk aqueous reaction between NO₂ and SO₂, in which this reaction could be further enhanced due to the presence of NH₃. As a result of the increase in RH, the partitioning of atmospheric ammonia was broken to a great extent, which enhanced the neutralization of S(VI) by ammonia at the particle surface to generate ammonium sulfate and dissociate large H⁺ (Fig. S13b, red part). Simultaneously, the ALWC did not rise significantly (Fig. S14b) at the beginning of haze events with relatively low sulfate concentrations. Thus, hydrogen ions generated from sulfate dissociation absorb ammonia more effectively from the ammonia-rich atmosphere at low relative humidity during the early pollution stages, which significantly promotes the net nitrate production. However, due to the co-effects of elevated RH and the hygroscopic nature of pre-generated ammonium sulfate, H⁺ concentrations were diluted and shown as nearly constant in situ pH (Fig. S14a). According to previous works, the reaction between sulfate generated first and bisulfate with ammonia was treated as the determination reaction for particle acidity (Weber et al., 2016; M. Liu et al., 2017). This reaction is self-limiting due to the acidity effect, namely that it increases the acidity of the aqueous phase and in turn reduces the efficiency of Henry's constant for SO₂ solubility and reaction rate, as well as reduces the H⁺ formation rates from moderate periods compared with clean periods (Fig. S13b, blue) (Wang et al., 2016; Clifton et al., 1988; Huie and Neta, 1986; Lee and Schwartz, 1983). Due to the co-effects of RH increase and the hygroscopic nature of sulfate, the ALWC was significantly elevated with the worsening of haze. Although more H⁺ was generated in this process, no significant decrease in pH was found with the haze aggravation due to the dilution effect of ALWC on H⁺. Previous works suggested that, in the case of ALWC increase, nitrate production is controlled by elevated H⁺ associated with the increase in sulfate; namely, ${\mathrm{NO}}_{\mathrm{3}}^{-}$ presented an elevating trend with the increases in H⁺ concentration (Xie et al., 2020). Thus, although H⁺ from the dissociation of sulfuric acid and fully loaded particle nitrate in conjunction with the haze aggravation generate particle HNO₃ (Fig. S19a) could force more ammonia partitioned on the particles to generate ammonium nitrate (Fig. S19b), net nitrate production (Δ${\mathrm{NO}}_{\mathrm{3}}^{-}$) was nearly consistent.

4 Conclusions

The formation of SIAs, especially sulfates and nitrates, was inherently associated with ALWC during the haze aggravation, in which the roles of ALWC should be more significant in the ammonia-rich atmosphere. The novelty of our work was finding the shifting of secondary inorganic aerosol formation mechanisms during haze aggravation and explaining the different roles of ALWC on a broader scale (∼500 µg m⁻³) in the ammonia-rich atmosphere based on the in situ high-resolution online monitoring datasets. The results showed that chemical domains of perturbation gas limiting the generation of secondary particulate matter presented obvious shifts from a HNO₃-sensitive to a HNO₃- and NH₃-co-sensitive regime with the haze aggravation, indicating the powerful driving effects of ammonia in the ammonia-rich atmosphere. When ALWC < 75 µg m⁻³, the sulfate generation was preferentially triggered by the high ammonia utilization and then accelerated by nitrogen oxide oxidation from clean to moderate pollution stages, characterized by NOR < 0.3, SOR < 0.4, NTR < 0.7 and the moral ratio of ${\mathrm{NO}}_{\mathrm{3}}^{-}:{\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}=\mathrm{2}:\mathrm{1}$. When ALWC > 75 µg m⁻³, the aqueous-phase chemistry reaction of SO₂ and NH₃ in ALW became the prerequisite for SIA formation driven by Henry's law in the ammonia-rich atmosphere during heavy and serious stages, characterized by high SOR (0.5–0.9), NOR (0.3–0.5) and NTR (>0.7), as well as the high moral ratio of ${\mathrm{NO}}_{\mathrm{3}}^{-}:{\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}=\mathrm{1}:\mathrm{1}$. A positive feedback of sulfate on nitrate production was also observed in this work. Our work provides a potential explanation for the interactive mechanisms and feedback between nitric aqueous chemistry and sulfate formation in the ammonia-rich atmosphere based on high-resolution field observations. It implies the targeted controlling of haze should not simply focus on SO₂ and NO₂, but more attention should be paid to gaseous precursors (e.g., SO₂, NO₂, NH₃) and aerosol chemical constitution during different haze stages.