1 Introduction

The frequency and intensity of haze events in China have increased significantly in recent years as a result of economic development and the acceleration in urbanization. The large amounts of suspended fine particles with aerodynamic diameters of 2.5 μm (PM_2.5) or less during haze events not only has the potential to harm human health by penetrating deeply into the lungs and contributing to cardiovascular morbidity and mortality (Brook et al., 2010; Cheng et al., 2013), but may also affect the climate by scattering and absorbing solar radiation (Seinfeld et al., 2004). Severe haze events with particulate matter as the dominant pollutant often occur in China, especially over the North China Plain (He et al., 2014; Jiang et al., 2015; Yang et al., 2015). The rapid increase in PM_2.5 over the Beijing–Tianjin–Hebei region is related to stagnant meteorological conditions and regional transport from sources of large anthropogenic emissions (Liu et al., 2013; Sun et al., 2016). PM_2.5 pollution during haze events is always contributed from secondary species, including both inorganic and organic species. Secondary inorganic aerosols (mainly SO₄^2–, NO₃^–, and NH₄⁺) account for a major fraction of PM_2.5 and increase with increasing levels of pollution (Quan et al., 2014; Zhang et al., 2014). Secondary inorganic aerosols have been suggested to be the principal driving force for the rapid increase in PM_2.5 concentrations (Wang et al., 2006; Zheng B. et al., 2015). The formation of SO₄^2–, NO₃^–, and NH₄⁺ (SNA) ions in fine particles is important during the evolution of haze.

Aerosol acidity is the aqueous concentration of H⁺, a crucial parameter in aerosol water. Aerosol acidity may influence the pH-dependent heterogeneous chemical processes that take place on aerosol surfaces, which contribute to the secondary growth of fine particulate matter. SO₄^2– is mainly produced in the aqueous phase via oxidation by H₂O₂, O₃, NO₂, and O₂ and is catalyzed by transition metals such as Feⁱⁱⁱ and Mnⁱⁱ in cloud droplets and aerosol water. Aqueous phase reactions are thought to be the dominant pathway for the formation of SO₄^2– (Alexander et al., 2009; Zhang R. Y. et al., 2015), which is highly sensitive to pH (Chameides, 1984; Cheng et al., 2016). Aerosol acidity also directly affects the gas-particle partitioning of semi-volatile species, including NH₃ and HNO₃ (Keene et al., 2004) and the gas-to-particle conversion of these species (Van Oss et al., 1998; Nemitz et al., 2004). The formation of NO₃^– via the heterogeneous hydrolysis of N₂O₅ on the surface of moist aerosols is also influenced by the acidity of the aerosols (Pathak et al., 2009).

A severe haze event with a long duration and strong intensity blanketed North China, East China, and central China from 16 to 22 December 2016, covering an area > 1.0 × 10 ⁶ km². The peak concentrations of daily PM_2.5 reached 500 μg m^–3 and above, about 14 times the daily average PM_2.5 concentration recommended by the World Health Organization (WHO, 2006). Beijing, located in the northern part of the polluted area in Hebei Province, suffered a periodic cycle of episodes of particulate matter pollution in autumn and winter (Guo et al., 2014) and six haze events occurred in December alone in 2016. Peripheral regions to the south of Beijing in Hebei Province, which are both highly populated and heavily industrialized, experienced more severe air pollution. Studies focused on the ionic composition during this extreme haze event may help to characterize the formation and evolution of secondary inorganic aerosols in PM_2.5 and provide observational data for the government to adopt effective measures to improve air quality in the Beijing–Tianjin–Hebei region.

The concentration of water-soluble ions was measured in PM_2.5 samples collected from an urban site in Beijing and a rural site in Gucheng, Hebei Province from November 2016 to January 2017. Aerosol acidity was also investigated to provide an insight into the mechanism of formation of secondary species in PM_2.5 during winter haze events. To assess the changes in aerosol pH due to the change in SNA under haze conditions, we investigated the sensitivity of pH to changes in SO₄^2–, NO₃^–, and NH₄⁺ concentrations, focusing on haze data from both sites. The PM_2.5 concentrations, meteorologi-cal conditions, and levels of gaseous pollutants, including SO₂, NO₂, and CO, were collected synchronously.

2 Materials and methods 2.1 Sampling

Aerosol samples of PM_2.5 were collected continuously over 24-h periods from 26 November 2016 to 12 January 2017 by using a high volume air sampler (Model TH-1000C II with PM_2.5 impactor, Wuhan Tianhong Instruments, China) at a flow rate of 1.05 m³ min^–1. The samples were collected on Whatman quartz microfiber filters (Whatman, UK). After sampling, the filters were placed in aluminum foil bags and sealed in clean plastic bags, which were then stored in a refrigerator at –18 °C until analysis.

The aerosol samples were collected at two sampling sites. The urban site was located on the roof top of the Chinese Academy of Meteorological Sciences (39°57'N, 116°20'E; about 40-m high) in Haidian District of Beijing between the 2nd and 3rd Ring Roads (CMA site). The rural site was located on a platform at the China Meteorological Administration farm (39°09'N, 115°44'E) in the town of Gucheng, Hebei Province, about 100 km southwest of Beijing (GC site).

2.2 Chemical analysis of ions

The ions were determined chemically with the method reported by Ye et al. (2015). A square sample (2 cm × 2 cm) of each filter was extracted under sonication with 20 mL of Millipore water (≥ 18 MΩ) for 80 min in an ice–water bath. The extracts were filtered through 0.45-μm filters before analysis. Na⁺, K⁺, Mg²⁺ Ca²⁺, NH₄⁺, SO₄^2–, NO₃^–, Cl^–, and NO₂^– were determined by using a Dionex ICS-2100 ion chromatography system (Thermo Fisher Scientific, USA). The ion composition of PM_2.5 at the CMA and GC sites was determined from 10 and 13 valid samples, respectively.

2.3 Calculation of PM2.5 acidity

A thermodynamic equilibrium model (ISORROPIA II; Fountoukis and Nenes, 2007) was used to predict the acidity of the PM_2.5 samples for the K⁺–Ca²⁺–Mg²⁺– NH₄⁺–Na⁺–SO₄^2––NO₃^––Cl^––H₂O aerosol system. The model was applied in the forward mode in a metastable state to give the best prediction of the aerosol pH (Hennigan et al., 2015), which required both the aerosol and gas phase concentrations. Gaseous NH₃ was not measured in our campaign and therefore we used the average NH₃ concentration obtained by Wang et al. (2016) in Beijing during 2015 as the input for gaseous NH₃; this varied with different atmospheric conditions. The sum of the concentrations of the gas phase and inorganic ions were input into the model as total concentrations, toge-ther with the relative humidity and temperature.

2.4 Other data

PM_2.5, PM₁₀, CO, NO₂, and SO₂ concentrations were observed at Wanliu station in urban Beijing by the Beijing Municipal Environmental Monitoring Center, about 3 km from the CMA site, and by the Baoding Environmental Protection Bureau at their Baoding monitoring station, about 30 km from the GC site. Meteorologi-cal data—including relative humidity, wind speed, and visibility—were derived from online data (www.wunderground.com) from weather stations located at 40°04'N, 116°35'E) (ZBAA) and 38°17'N, 114°42'E (ZBSJ). The cloud water content at the two sampling sites was observed by the MODIS (Moderate Resolution Imaging Spectroradiometer) instrument onboard the Terra satellite with a resolution of 0.1° × 0.1° on each day of the sample period (data available at http://neo.sci.gsfc.nasa.gov). The time used in this paper is local time (UTC + 8).

3 Results and discussion 3.1 Ambient concentrations of PM_2.5, gas pollutants, and SNA

Figures 1 and 2 show the PM_2.5 concentrations and the concentrations of gaseous pollutants and SNA in Beijing and Gucheng from December 2016 to January 2017. The properties of particulate matter in Beijing have been shown to have a periodic cycle of 4–7 days (Guo et al., 2014). The hourly PM_2.5 concentrations in Beijing showed periodic fluctuations from 3 to 585 μg m^–3 with a mean of 141 ± 132 μg m^–3; similar variations from 13 to 557 μg m^–3 with a mean of 193 ± 109 μg m^–3 were measured in Gucheng.

Gucheng is located on the transport axis for pollutants from Shijiazhuang and Baoding to Beijing and, as a result, Beijing has less pollution from severe fine particulate matter than Gucheng. Clean air conditions with a PM_2.5 concentration < 50 μg m ^–3 were observed at the start of each cycle at the CMA site in Beijing. PM_2.5 concentrations increased sharply and doubled within a very short time. This rapid increase in particulate matter contributed to the decrease in atmospheric visibility and caused haze pollution (He et al., 2014). The ratio of PM_2.5 to PM₁₀ during the sample period was about 70% at both sites and increased with PM_2.5 concentrations, especially at the urban site in Beijing, suggesting that fine particles accounted for most of the PM₁₀ during the observation period. The statistics also showed that only 38% and 8% of daily PM_2.5 concentrations at the CMA and GC site, respectively, were less than the Chinese National Ambient Air Quality Standard for daily PM_2.5 (NAAQS, Grade II) of 75 μg m^–3, indicating frequent pollution by fine particles during our observation period.

During the massive haze event occurring from 16 to 22 December 2016, hourly PM_2.5 concentrations > 500 μg m ^–3 were recorded at both sites. The maximum daily PM_2.5 concentrations at sites CMA and GC were 383 and 394 μg m^–3, respectively, 5 times higher than the daily average in the standard Chinese regulation. Gaseous pollutants such as NO₂, SO₂, and CO were also greatly increased at this time, with a peak concentration of 234 μg m^–3 for NO₂ in Beijing. Over the entire sample period, the NO₂ concentrations generally exceeded those of SO₂, in agreement with previous studies (Meng et al., 2009). The gaseous pollutants showed similar variations to PM_2.5, in particular the high concentrations of NO₂ and CO (Figs. 1, 2).

To facilitate comparison, the samples from the CMA site in Beijing were divided into two groups based on the PM_2.5 concentrations—namely, clear days (daily PM_2.5 < 75 μg m ^–3) and polluted days (daily PM_2.5 ≥ 75 μg m^–3). No clear day was observed for the samples collected from the GC site and therefore a concentration of 150 μg m^–3 was used to distinguish light–middle polluted days (LMPD) and heavy–severe polluted days (HSPD). Table 1 shows that precursors such as SO₂ and NO₂ were significantly higher at the CMA site on polluted days than on clear days (t-test, p < 0.05), favoring the formation of SO ₄^2– and NO₃^–. The concentrations of SNA on polluted days were about six times the levels on clear days, much higher than the increase in the precursors. The ratio of PM_2.5 to PM₁₀ was also much higher on polluted days than on clear days, showing that the fine particles made up a larger proportion of PM₁₀ in haze episodes. The only difference for the GC site was that LMPDs had higher concentrations of SO₂ and lower concentrations of SO₄^2– (Table 2) compared with HSPDs, implying that more SO₄^2– accumulated under more polluted conditions. The concentration of SNA is therefore suggested to be the principal driving force for the increase in PM_2.5 concentrations (Wang Y. X. et al., 2014; Zheng B. et al., 2015).

3.2 Composition of water-soluble ions and synoptic conditions

Table 3 lists the maximum, average, and minimum concentrations of water-soluble inorganic ions in PM_2.5 during the sampling period for the two sites. SNA are the predominant inorganic species of PM_2.5. The average concentration of SO₄^2– at the CMA site was 8.3 μg m^–3, accounting for 21% of the total mass of inorganic ions. This is lower than the previously reported annual or winter average concentrations in Beijing (Wang et al., 2006; Yang et al., 2011; He et al., 2012). At the GC site, SO₄^2– made up 23% of the total inorganic ions, with an average of 14.0 μg m^–3. The average NO₃^– concentrations at sites CMA and GC were 12.5 and 14.2 μg m^–3, making up 31% and 24% of the total mass of inorganic ions, respectively. This was higher than the concentrations observed during 2001–06. The average concentrations of NH₄⁺ were 14.1 and 24.2 μg m^–3 at sites CMA and GC, accounting for 35% and 40% of the total mass of inorganic ions, much higher than the previously reported data. The sum of the concentrations of Na⁺, K⁺, Mg²⁺, Ca²⁺, and Cl^– contributed about 13% of the total concentration of inorganic ions at these two sites.

SO₄^2– has previously been reported to be the dominant component of SNA in PM_2.5 over Beijing from 2000 to 2006, followed by NO₃^– and NH₄⁺ (Wang et al., 2006; Yang et al., 2011; He et al., 2012). Our results at both sites showed that NH₄⁺ accounted for the highest proportion of inorganic ions, which was comparable with the proportion contributed by NO₃^– followed by SO₄^2–. This result may be explained by the variation in SO₂ and NO_x emissions over the last decade (Sun et al., 2016). SO₂ emissions have decreased as a result of desulfurization, whereas NO_x emissions have increased as a result of industrialization and an increase in the number of vehicles. At the CMA site, SO₄^2– was linearly correlated with SO₂ with a correlation coefficient (r) of 0.77 (p < 0.01). When the concentrations of SO ₄^2– decreased with the decrease in SO₂ emissions, free NH₃ was released and more NH₄⁺ accumulated in the form of NH₄NO₃, in addition to the increase resulting from an increase in NO_x emissions (Wang et al., 2013).

Recent research has shown that emissions from the burning of fossil fuels (including traffic emissions) and NH₃ from fossil fuel power plants contribute most of the air pollution in urban Beijing in the winter months (Pan et al., 2016), which could result in the NH₄⁺-rich aerosols at the CMA site. The high concentrations of NH₄⁺ at the rural GC site may be associated with both agricultu-ral sources and emissions from the burning of fossil fuels in the surrounding highly industrialized areas. The high levels of NH₄⁺ at both sites therefore suggests high emissions of NH₃ over the North China Plain, in turn favoring the formation of more NO₃^–. The findings reported here emphasize the significance of controlling NH₃ emissions in the future, given the contribution of NH₄⁺ to aerosol growth in haze events.

Figure 3 shows the variation in PM_2.5 with meteorological conditions. At the CMA site, PM_2.5 and the ratios of PM_2.5 to PM₁₀ increased with increasing relative humidity, whereas the PM_2.5 concentrations showed a decreasing trend with wind speed. A high relative humidity may enhance the liquid water content of aerosols and thus enhance the heterogeneous uptake of gas precursors on their surface (Wang X. F. et al., 2012), which favors the increase in PM_2.5 concentrations. The stable synoptic conditions (relatively low wind speed) during the polluted period were unfavorable toward the horizontal transport and vertical mixing of pollutants, which played an essential part in the accumulation of pollutants and the formation of haze (Wang Z. F. et al., 2014; Zheng B. et al., 2015; Zheng G. J. et al., 2015).

At the GC site, the patterns of variation in PM_2.5 concentrations were different from those in the urban Beijing. The PM_2.5 concentrations showed no obvious trend below 75% relative humidity and reached their peak at the highest levels of humidity. A minor increase in the concentration of PM_2.5 with increasing wind speed was observed, indicating that atmospheric transport from the surrounding polluted areas had a key role, which corresponded to the location of this site in the NE–SW pollution transport channel. The daily average wind speed reached a maximum of 3.3 m s^–1 on 26 November 2016 at site GC, corresponding to turbid air conditions with a 24-h PM_2.5 concentration of 222.2 μg m^–3. The concentration of Ca²⁺ reached a maximum of 3.0 μg m^–3 on this day, together with high concentrations of SO₄^2–, NO₃^–, and NH₄⁺ of 27.5, 20.9, and 26.0 μg m^–3, respectively. The turbid air conditions at the GC site during the sampling period resulted in less variation in the PM_2.5 concentrations than at the urban site in Beijing.

The mass ratio of NO₃^–/SO₄^2– in aerosols can be used to evaluate the relative importance of mobile versus stationary sources (Arimoto et al., 1996; Gao et al., 2009). Our results showed that the NO₃^–/SO₄^2– ratio changed with the PM_2.5 concentrations (Fig. 4). The NO₃^–/SO₄^2– ratio was > 1 in urban Beijing, with an average of 1.6 on the polluted days, suggesting that mobile sources were important sources of emission during haze events in Beijing. On the clear days, the NO ₃^–/SO₄^2– ratio showed a good positive correlation with PM_2.5 (r = 0.85, p < 0.05) and the minimum ratio was 0.7 when the PM _2.5 concentration was 14 μg m^–3. Sun et al. (2013) found similar results in Beijing, with NO₃^–/SO₄^2– ratios as low as 0.52 on a non-haze day with a PM_1.8 of 35 μg m^–3 (Sun et al., 2013). High wind speeds not only favor the transport of SO₂ and SO₄^2– into Beijing, but can also dilute local NO_x emissions from vehicles, contributing to the relatively low NO₃^–/SO₄^2– ratio on windy days (Sun et al., 2013). The NO₃^–/SO₄^2– ratios (0.7–2.0) were higher than the mean values of 0.58 and 0.67 measured in Beijing during 1999–2000 and 2001–03, respectively, which may demonstrate the response of the particle composition to changes in the precursor emissions (Yao et al., 2002; Wang et al., 2005). For the GC site, the NO₃^–/SO₄^2– ratio was relatively low, with an average of 1.0 under haze conditions, indicating that the GC site was mainly influenced by stationary emission sources.

3.3 Formation of SNA

The concentrations of SO₄^2–, NO₃^–, and NH₄⁺ in PM_2.5 during haze periods were about 4–6 times those on the non-haze days at the CMA site; similar results were obtained for the GC site (Table 3). The high ratio of SNA on polluted days to that on clear days suggested that more secondary inorganic aerosols were formed under haze conditions, as observed previously in similar campaigns (Shen et al., 2009; Zhang Q. et al., 2015). Figures 5a and b show that there was a good correlation between the NH₄⁺ concentration and the sum of SO₄^2– and NO₃^– concentrations (r= 0.99 for the CMA site, r= 0.84 for the GC site, p< 0.01), indicating that NH₃ had an important role in neutralizing the SO₄^2– and NO₃^– formed. Figure 5c shows that high SNA concentrations generally occurred under conditions of low visibility and that they decreased sharply when the visibility was < 10 km.

The concentration of OH radicals is strongly dependent on the amount of ultraviolet radiation (Ehhalt and Rohrer, 2000; Lu et al., 2013) and previous results for Beijing have shown that the concentration of OH radicals is low on haze days as a result of the aerosol dimming effect (Zheng B. et al., 2015; Zheng G. J. et al., 2015). Therefore, the increase in SO₄^2– and NO₃^– concentrations during haze events was unlikely to be driven via gas phase oxidation by OH radicals and the formation of SO₄^2– and NO₃^– was more likely to be via the aqueous phase pathway.

Aqueous phase reactions may enhance the formation of SO₄^2– and NO₃^– via heterogeneous reactions in clouds or rain droplets, although there was no precipitation during the sampling period. In addition, the satellite data for the cloud water content (Fig. 6) did not show a significant difference under different atmospheric conditions, suggesting that this could not account for the increase in SO₄^2–, NO₃^–, the sulfur oxidation ratio [SOR = nSO₄^2–/ (nSO₄^2– + nSO₂), where n is the molar concentration] and the nitrogen oxidation ratio [NOR = nNO₃^–/(nNO₃^– + nNO₂)] from clear days to polluted days. The SOR and the NOR could be used as indicators of secondary transformation (Sun et al., 2006). Figures 7a and b show that the SOR and NOR increased from clear days to polluted days and from LMPD to HSPD. In urban Beijing, the NOR showed a good linear correlation with the relative humidity (r= 0.79, p< 0.01) and the SOR showed a moderate linear correlation with relative humidity (r= 0.67, p< 0.05) (Fig. 7c), consistent with the dependence of heterogeneous reactions on relative humidity. Heterogeneous reactions have been suggested to play an essential part in producing high SO₄^2– and NO₃^– concentrations during haze pollution events (Wang Y. X. et al., 2014).

The heterogeneous reaction is usually parameterized in models as follows (Jacob, 2000; Wang K. et al., 2012):

(1)

where k is the rate constant (s^–1), R is the radius of the aerosol particles (m), D is the gas phase molecular diffusion coefficient (m² s^–1), vis the mean molecular speed (m s^–1), γis the uptake coefficient, and S is the aerosol surface area per unit volume (m² m^–3).

The uptake coefficient γ is assumed to depend on relative humidity in modeling studies (Song and Carmichael, 2001; Wang Y. X. et al., 2014; Zheng B. et al., 2015). For example, the value of γ for SO₂ was set between 10^–3 and 10^–2 by Wang Y. et al. (2014), a γ value of 5 × 10^–3 was used for SO₂ when the relative humidity was < 50%, and 5 × 10 ^–2 was used when relative humidity was > 50% ( Song and Carmichael, 2001). Taking SO₂ as an example, the difference in k between polluted and clear days can be expressed as follows:

(2)

The hygroscopic growth factor (GF) of 1.1 was taken from earlier studies in Beijing (Meier et al., 2009). A γ value of 5 × 10^–3 for SO₂ was used on clear days and a γ value of 5 × 10^–2 was used on polluted days (Song and Carmichael, 2001). Values of D of 2 × 10^–5 m² s^–1 and v of 3 × 10² m s^–1 were used (Jacob, 2000) and an R_CD value of 1 × 10^–6 m. Equation (2) shows that the heterogeneous reaction of gaseous pollutants may be significantly increased on polluted days.

3.4 Aerosol acidity and sensitivity analysis

Aerosol acidity is the measurement of the aqueous concentration of H⁺ and is correlated with the ion distribution and levels of liquid water in particles. Combined with measurements of the in-situ ion composition, the thermodynamic equilibrium model ISORROPIA II (Fountoukis and Nenes, 2007) was used to predict the pH of aerosols. For the CMA site, we obtained mean (range) pH of 5.0 (4.9–5.2) under haze conditions and 6.9 (6.0–7.7) on clear days. The clean aerosol samples seemed to be more alkaline than those formed under haze conditions. All the PM_2.5 samples at the GC site were acidic, with an average pH of 5.3 (range 4.6–6.3), in agreement with haze conditions for all observed days. The pH for these two sites under haze conditions agreed with previously reported results of 5.4–6.2 on the North China Plain in winter (Cheng et al., 2016) and exceeded the particle pH of 0–3 in North America (Guo et al., 2015; Xu et al., 2015; Weber et al., 2016), indicating the neutralization of aerosol particles under haze conditions. China has high emissions of NH₃ as a result of agricultural activities, especially in Henan, Shandong, and Hebei provinces (Streets et al., 2003; Huang et al., 2012). The high neutralizing capacity of aerosols might be attributed to the high NH₃ emissions on the North China Plain, corresponding to high NH₄⁺ concentrations.

We have shown earlier that heterogeneous reactions have an essential role in producing high concentrations of SO₄^2– and NO₃^– under haze conditions. The acidity of the particles affects the chemical processes in the aerosol water where the heterogeneous reactions occur. Based on the predicted pH, we tried to predict the major pathways for the aqueous phase oxidation of SO₂ and NO₂. According to recent research, in the range of pH obtained here, dissolved NO₂ should be the most effective oxidant in the aerosol water and could generate SO₄^2– and NO₂^– under haze conditions in Beijing (Cheng et al., 2016). In addition, when the NO₃^– is correlated with relative humidity and aerosol loadings, the heterogeneous hydrolysis of N₂O₅ on the acidic surface of aerosols has been shown to be a major route to forming NO₃^– in Beijing haze (Pathak et al., 2009). Nevertheless, the predicted pH is just one factor affecting the chemical processes in aerosol water and the proposed reaction pathways are speculative and without supporting evidence.

Based on the SNA fractions discussed here, the contribution of SO₄^2– has decreased in the past decade, whereas the concentrations of NO₃^– and NH₄⁺ have increased. We attempted to assess the sensitivity of the pH of aerosols under haze conditions at both sites to the changes in the concentration of SO₄^2–, NO₃^–, and NH₄⁺, i.e., the acidic samples. SO₄^2–, NO₃^–, and NH₄⁺ concentrations were independently changed from 10% to 200% of the initial measured value and were used as inputs to ISORROPIA II. The predicted aerosol pH values are shown in Fig. 8, where the 100% data represent the predicted pH based on the in-situ measurement of the particle bulk composition. The decrease below the measured SO₄^2– levels corresponds to a higher pH value with an amplitude of less than one unit of pH. Similarly, slight decreases in pH were obtained when the SO₄^2– concentration increased up to twice the initial value. In contrast with the changes in SO₄^2–, a higher NH₄⁺ concentration was correlated with higher pH values, indicating greater neutralization of the aerosols, whereas lower pH values were obtained with a reduction in NH₄⁺ concentrations. A similar trend was found for NO₃^– at the GC site which was less sensitive than NH₄⁺, whereas minor fluctuations were seen at the CMA site. The pH appeared to be more sensitive to SO₄^2– and NH₄⁺ than to NO₃^–. Specifically, a 50% decrease in NH₄⁺ concentration resulted in a 0.28 unit reduction in pH, whereas a 50% increase in NH₄⁺ concentration resulted in a 0.17 unit increase in pH at the CMA site. For the GC samples, a 0.33 unit increase and a 0.25 unit reduction in pH was seen for a 50% increase and a 50% decrease in SO₄^2– concentration, respectively. These results show the relatively weak sensitivity of pH to a wide range of SO₄^2–, NO₃^–, and NH₄⁺ concentrations. This is attributed to the buffering effect of NH₃ partitioning as result of the semi-volatility of NH₄⁺ and is consistent with similar studies of NH₃ and SO₄^2– (Weber et al., 2016; Zhao et al., 2016). The predicted pH varied over a narrow range under haze conditions at both sites.

4. Conclusions

SO₄^2–, NO₃^–, and NH₄⁺ were the dominant ions in PM_2.5 at both an urban Beijing site and a rural site in Gucheng. The average SO₄^2–, NO₃^–, and NH₄⁺ concentrations during the sampling period were 8.3, 12.5, and 14.1 μg m^–3 at the CMA site and 14.0, 14.2, and 24.2 μg m^–3 at the GC site. The NO₃^– concentrations exceeded those of SO₄^2– at both sites, in contrast with the results of previous studies, indicating the variation in SO₂ and NO_x emissions. The finding that NH₄⁺ was the dominant component of SNA emphasized the importance of controlling future NH₃ emissions. The NOR and SOR were linearly correlated with relative humidity (r= 0.79 and r= 0.67) in urban Beijing and were related to the aerosol loadings, suggesting that heterogeneous reactions in the aerosol water played an essential part in the significant increase in SO₄^2– and NO₃^– in haze pollution. A parameterized model was applied to calculate the rate constant (k) of the heterogeneous reaction. For gaseous SO₂, the value of k on polluted days was about 10 times higher than that on clear days.

The aerosol acidity was predicted by using the ISORROPIA II model in the forward mode with a metastable state. Under haze conditions, the pH ranged from 4.9 to 5.2 with a mean of 5.0 at the CMA site and from 4.6 to 6.3 with a mean of 5.3 at the GC site. The aerosol particles in haze were acidic and retained a high neutralizing capacity associated with the high NH₃ emissions on the North China Plain. Based on the pH environment, we predicted that oxidation by dissolved NO₂ and the hydrolysis of N₂O₅ on the acidic aerosol surface should be the major pathways for the heterogeneous formation of SO₄^2– and NO₃^–. Under haze conditions, the sensitivity analysis of aerosol pH suggested a relatively weak sensitivity of pH to a wide range of SO₄^2–, NO₃^–, and NH₄⁺ concentrations associated with the buffering effect of NH₃ partitioning. The pH was more sensitive to SO₄^2– and NH₄⁺ concentrations with opposing trends, than to NO₃^– concentrations. Future work exploring the mechanisms of the heterogeneous reactions in aerosol water of particles is required to elucidate the reasons for the rapid increase in SNA under haze conditions.

Acknowledgments. We thank Haochi Che of the Chinese Academy of Meteorological Sciences and Sanxue Ren of the China Meteorological Administration farm at Gucheng for carrying out the sampling at the Beijing and Gucheng site, respectively.