Atmospheric measurements have demonstrated rapid sulfate production during severe haze events in China (Guo et al., 2014; Wang et al., 2014, 2016; Zhang et al., 2015; Cheng et al., 2016). For example, Wang et al. (2016) showed that, during pollution episodes in Xi'an, the SO42- mass concentration increased markedly from less than 10, from 10 to 20, and to greater than 20 µg m-3, with the corresponding increases in the mean PM2.5 mass concentrations from 43, 139, to 250 µg m-3 from clean, transition, to polluted periods, respectively. Among the PM2.5 species in Xi'an, organic matter (OM), nitrate (NO3-), and SO42- were most abundant, with the mass fractions of 55, 14, and 14 %, respectively, during the polluted period. In addition, the work of Wang et al. (2016) demonstrated that the molar ratio of SO42- to SO2, which reflects sulfur partitioning between the particle and gas phases, exhibited an exponential increase with relative humidity (RH), with the values of less than 0.1 at RH < 20 % to 1.1 at RH > 90 % in Xi'an. Similar evolutions in SO42-mass concentrations and the molar ratio of SO42- to SO2 were shown during the pollution development in Beijing (Sun et al., 2013; Wang et al., 2014, 2016). The rapid sulfate formation measured in China could not be explained by current atmospheric models and suggested missing sulfur oxidation mechanisms (Wang et al., 2014). Typically, high sulfate levels during haze events in China occurred concurrently with elevated RH, NOx, and NH3 (Wang et al., 2014, 2016; Zhang et al., 2015), implicating an aqueous sulfur oxidation pathway. On the basis of complementary field and experimental measurements, Wang et al. (2016) concluded that the aqueous oxidation of SO2 by NO2 is key to efficient sulfate formation, but is only feasible under two atmospheric conditions, i.e., on fine aerosols with high RH and NH3 neutralization or under cloud conditions.

Several recent studies estimated the particle acidity and aqueous sulfate production during severe haze events in China using the thermodynamic model (Cheng et al., 2016; Guo et al., 2017a; M. Liu et al., 2017). For example, Cheng et al. (2016) estimated a pH range of 5.4 to 6.2 using a thermodynamic model (ISORROPIA-II) in Beijing. On the basis of their estimated pH and the previous experimental rates of SO2 oxidation by NO2 and the Henry's Law constants for sulfur dioxide (SO2), bisulfite (HSO3-), and sulfite (SO32-) from the literature (Lee and Schwartz, 1983; Clifton et al., 1988; Seinfeld and Pandis, 2006), the authors derived a sulfate production rate and concluded that reactive nitrogen chemistry in aerosol water explained the sulfate formation during polluted periods in Beijing. In contrast, other recent studies by Guo et al. (2017a) and M. Liu et al. (2017) adopted a similar method to Cheng et al. (2016), but reported significantly different values of pH and the sulfate formation rates by the aqueous SO2 oxidation by NO2 in China. Those two later studies determined a pH range of 3.0–4.9 and suggested that fine particles were moderately acidic and the aqueous SO2 oxidation by NO2 was unimportant during severe wintertime haze periods in China.

In this article, we conducted laboratory measurements of the hygroscopicity for oxalic acid particles and particle growth of ammonium sulfate particles upon exposure to SO2, NO2, and NH3 at high RH conditions, in order to evaluate the dominant factors regulating the aqueous oxidation of SO2 by NO2. In addition, field measurements of chemical compositions of water-soluble fraction for fine PM (including oxalic acid) in Beijing, Hebei Province, and Xi'an were performed during the winter haze episodes, showing significantly enriched water-soluble organic matter (WSOM). The implications for the multi-aerosol chemical compositions on the pH value and sulfate production during winter pollution periods in China are discussed.

Several recent studies using thermodynamic models (Wexler and Clegg, 2002; Fountoukis and Nenes, 2007) estimated the particle acidity and sulfate production during pollution episodes in China (Cheng et al., 2016; Guo et al., 2017a; M. Liu et al., 2017). Those previous studies treated the PM exclusively as a mixture of inorganic salts dominated by ammonium sulfate and neglected the effects due to the presence of organic compounds. Apparently, the conclusions by those modeling studies hinge on the validity of several critical assumptions in their analyses, including the application of the thermodynamic model, the accuracy in determining the aerosol water content (AWC), and the applicability of the earlier experimental measurements for the aqueous oxidation of SO2 by NO2 to atmospheric conditions.

Estimation of the pH values using the thermodynamic models is typically of considerable uncertainty, because of several intricate difficulties. For example, the ISORRPIA-II model includes two modes, i.e., metastable (aerosols are assumed to be in the liquid-phase only and may reach supersaturation) and stable (aerosols are assumed in the liquid- and solid phases that are in equilibrium) (Guo et al., 2017b). As the thermodynamic model is established on the basis of the equilibrium principles, its application to non-equilibrium conditions needs to be rigorously assessed. In addition, the phase (e.g., liquid, amorphous, or crystalline) and mixing state of ambient aerosols are highly complex because of the presence of multi-component organic and inorganic species (Qiu and Zhang, 2013; Zhang et al., 2015), inevitably rendering high uncertainty in the thermodynamic calculations.

Guo et al. (2017a) suggested that the pH predictions using the metastable mode would be more reliable than that using the stable mode, on the basis of model evaluation from measured and predicted NO3- and NH4+ during the winter of 2012 in Xi'an. Figure 4 compares the concentrations of NH3 (g) and aerosol species predicted by ISORROPIA-II with the field measurements under the metastable and stable modes in Xi'an during the winter of 2012. As evident in Fig. 4a and b, NH3 predicted is similar to the measured value with the metastable or stable mode. Furthermore, the predicted concentrations of NO3- and NH4+ using both the metastable and stable modes are nearly identical (Fig. 4c–f). Guo et al. (2017a) only compared the liquid NH4+ and NO3- predicted by the model under the stable mode with the field measured aerosols composed of both liquid and solid compounds, and thus their predicted concentrations were lower than those of the measurements (see Fig. S1 in Guo et al., 2017a). As a result, their statement that pH prediction with the metastable mode would be more reliable than that with the stable mode was unjustified. Noticeably, the pH values estimated by the ISORROPIA-II model under the two modes are significantly different, with the values of 4.57 ± 0.40 under the metastable mode and 6.96  ± 1.33 under the stable mode. Most recently, it was suggested that the large discrepancy in predicting pH is attributable to the model code errors (Song et al., 2018).

In addition, the pH estimation by the thermodynamic model is highly dependent on the ratio of the concentration of hydrogen ions in the liquid-phase to AWC. Guo et al. (2017a) and M. Liu et al. (2017) assumed negligible particle water associated with the organic aerosol mass. Such an assumption is clearly invalid, as aerosols typically contain a large portion of WSOM in China (Fig. 3), including organic nitrogen species (G. Wang et al., 2010; Wang et al., 2013) and acids (G. Wang et al., 2009, 2010; Wang et al., 2006). Also, organic acids engage in particle-phase reactions with the basic species (i.e., NH3 and amines), significantly enhancing the particle hygroscopicity and reducing the acidity (Gomez-Hernandez et al., 2016). In addition, because of their strong basicity and high abundance, amines likely play a key role in reducing the particle acidity in China (L. Wang et al., 2010a, b; Qiu et al., 2011; Qiu and Zhang, 2012; Dong et al., 2013; Zheng et al., 2015; Yao et al., 2016; F. Liu et al., 2017). Consequently, the acidity for organics-dominated aerosols is considerably different from that of ammonium sulfate aerosols, as demonstrated in our experimental results. While effort has been made to account for the effects of organic species on the aerosol properties (Clegg et al., 2013), the available thermodynamic models are still inadequate in representing complex multi-component aerosols. An inconsistency of the ammonium–sulfate ratios using the thermodynamic models was identified in the eastern US, also suggesting a possible role for organic species (Silvern et al., 2017).

Furthermore, the chemical mechanism leading to the aqueous conversion of SO2 to sulfate by NO2 is not well understood. The previous modeling studies adopted the aqueous reaction rate constants previously measured (Lee and Schwartz, 1983; Clifton et al., 1988), while the applicability of the earlier experimental studies to atmospheric conditions is uncertain. For example, Lee and Schwartz (1983) examined the oxidation of S(IV) by NO2 in the liquid phase by flowing gaseous NO2 through a NaHSO3 solution at a constant pH by regulating NaOH and determined the rate constant of 1.4 × 105 M-1 s-1 at pH = 5 and with a lower limit of 2 × 106 M-1 s-1 at pH = 5.8 and 6.4 from measuring the electrical conductivity of the solution. Clifton et al. (1988) measured the rate constant for the reaction of NO2 with S(IV) over the pH range of 5.3–13, by producing NO2 from irradiation of NaNO2 and N2O solutions and mixing with Na2SO3 solutions, and obtained the second-order rate constant of 1.24 × 107 and 2.95 × 107 M-1 s-1 from the decay of NO2 monitored by absorption spectroscopy. The results of the measured rate constants between the two earlier experimental measurements differed by 1–2 orders of magnitude (Lee and Schwartz, 1983; Clifton et al., 1988). Also, both kinetic experiments employed bulk solutions and did not account for the gaseous uptake process (Lee and Schwartz, 1983; Clifton et al., 1988).

Wang et al. (2016) obtained the SO2 uptake coefficient for sulfate production from combined field measurements and laboratory experiments, and their laboratory experiments using aqueous oxalic acid particles reproduced the rapid sulfate production measured under polluted ambient conditions in China. The SO2 uptake coefficient on oxalic acid particles in the laboratory reaction chamber is 8.3 ± 5.7 × 10-5 (Wang et al., 2016) under the humid conditions and similar to that (4.5 ± 1.1 × 10-5) (Wang et al., 2016) observed in Beijing during the haze period of 2015. The results of the SO2 uptake coefficients determined by Wang et al. (2016) are also consistent with the modeling studies in quantification of the sulfate formation using atmospheric models in the country (e.g., Wang et al., 2014), indicating the applicability of their proposed mechanism to haze conditions in China. In contrast, M. Liu et al. (2017) invoked the experimental work by Hung and Hoffmann (2015) as a plausible cause for rapid SO2 oxidation by O2 in the absence of photochemistry, but without noting the high acidity as a necessary condition in that experimental work (i.e., pH ≤ 3). Most recently, Li et al. (2018) suggested an indirect mechanism of SO2 oxidation by NO2 via HONO/NO2- produced in fast-hydrolytic disproportion of NO2 on the surface of NaHSO3 aqueous microjets. In addition, another recent theoretical work by Zhang et al. (2018) indicated that under weakly acidic and neutral conditions (pH ≤ 7) the oxidation of HOSO2- by dissolved NO2 is a self-sustaining process, where the produced cis-HONO, HSO4-, and H2SO4 promote the tautomerization from HSO3- to HOSO2- as the catalysts.

In this paper we have presented experimental measurements of the growth of ammonium sulfate seed particles exposed to vapors of SO2, NO2, and NH3 at variable RH, the HGF of oxalic acid particles, and field measurements of WSOM for PM2.5 during the severe haze events in Beijing, Hebei Province, and Xi'an. Our experimental results reveal that sulfate production on fine particles is dependent on the particle hygroscopicity, phase-state, and acidity, as well as RH. The acidity and sulfate formation for ammonium sulfate seed particles are distinct from those of oxalic acid seed particles. Aqueous ammonium sulfate particles show negligible growth because of low pH, in contrast to aqueous oxalic acid particles with significant dry-size increase and sulfate formation because of high pH. In addition, our atmospheric measurements show significant concentrations of WSOM (including oxalic acid) in fine PM, indicating multi-component haze particles in China. Our results reveal that a particle mixture of inorganic salts adopted by the previous studies using the thermodynamic model does not represent a suitable model system and that the particle acidity and aqueous sulfate formation rate cannot be reliably inferred without accounting for the effects of multi-chemical compositions during severe haze events in China. Our combined experimental and field measurements corroborate the earlier finding that sulfate production via the particle-phase reaction involving SO2 and NO2 with NH3 neutralization occurs efficiently on organics-dominated aerosols (Wang et al., 2016) but are in contradiction to the most recent studies using the thermodynamic model (Guo et al., 2017a; M. Liu et al., 2017).

In conclusion, while the particle acidity or pH cannot be accurately determined from atmospheric field measurements or calculated using the thermodynamic models, our combined experimental and field results provide the compelling evidence that the pH value of ambient organics-dominated particles is sufficiently high to promote SO2 oxidation by NO2 with NH3 neutralization under polluted conditions in China.