Persistent sulfate formation from London Fog to Chinese haze

Significance

Exceedingly high levels of fine particulate matter (PM) occur frequently in China, but the mechanism of severe haze formation remains unclear. From atmospheric measurements in two Chinese megacities and laboratory experiments, we show that the oxidation of SO₂ by NO₂ occurs efficiently in aqueous media under two polluted conditions: first, during the formation of the 1952 London Fog via in-cloud oxidation; and second, on fine PM with NH₃ neutralization during severe haze in China. We suggest that effective haze mitigation is achievable by intervening in the sulfate formation process with NH₃ and NO₂ emission control measures. Hence, our results explain the outstanding sulfur problem during the historic London Fog formation and elucidate the chemical mechanism of severe haze in China.

Abstract

Sulfate aerosols exert profound impacts on human and ecosystem health, weather, and climate, but their formation mechanism remains uncertain. Atmospheric models consistently underpredict sulfate levels under diverse environmental conditions. From atmospheric measurements in two Chinese megacities and complementary laboratory experiments, we show that the aqueous oxidation of SO₂ by NO₂ is key to efficient sulfate formation but is only feasible under two atmospheric conditions: on fine aerosols with high relative humidity and NH₃ neutralization or under cloud conditions. Under polluted environments, this SO₂ oxidation process leads to large sulfate production rates and promotes formation of nitrate and organic matter on aqueous particles, exacerbating severe haze development. Effective haze mitigation is achievable by intervening in the sulfate formation process with enforced NH₃ and NO₂ control measures. In addition to explaining the polluted episodes currently occurring in China and during the 1952 London Fog, this sulfate production mechanism is widespread, and our results suggest a way to tackle this growing problem in China and much of the developing world.

Fine particulate matter (PM), which typically contains a complex mixture of inorganic and organic species, has important implications for several environmental issues (1–3). Presently, the mechanisms leading to PM formation remain uncertain, particularly under highly polluted conditions, hindering efforts in developing effective mitigation policies to reduce their local, regional, and global impacts (1). It is well established, though, that sulfate (SO₄²⁻) is ubiquitous and is a key PM constituent in the atmosphere. Moreover, hygroscopic sulfate aerosols serve as efficient cloud condensation nuclei, affecting cloud formation, precipitation, and climate (4–8). A major fraction of regional acid deposition is attributed to the sulfate content that exerts debilitating effects on acid-sensitive ecosystems (9). Furthermore, high levels of fine PM have been implicated in adverse human health issues (1), as exemplified by high fatality during the 1952 London Fog (1, 10). Sulfur compounds are emitted globally from many natural and anthropogenic sources (1–3, 11), and there have been high SO₂ emissions from combustion of coal and petroleum products in developing countries (such as China) spurred on by fast economic development (12).

Gaseous SO₂ is converted to particulate sulfate through gas-phase oxidation or aqueous reactions, but the detailed chemical mechanisms remain controversial (1–3, 13, 14). The gas-phase oxidation of SO₂ is dominated by its reaction with the OH radical, with a lifetime of ∼1 wk at the typical tropospheric level of OH radicals. The aqueous pathways of SO₂ oxidation may occur much faster, including reactions with dissolved ozone, hydrogen peroxide, organic peroxides, OH, and NO₂ via catalytic or noncatalytic pathways involving mineral oxides (15–20). Most recently, an interfacial SO₂ oxidation mechanism involving O₂ on acidic microdroplets has been suggested (16).

It has been hypothesized that aqueous SO₂ oxidation by NO₂ can be an important pathway for sulfate formation under urban conditions and in the presence of sufficient neutralizing agents such as NH₃ (2). Several earlier experimental studies, in which gaseous NO₂ was exposed to bulk solutions containing sulfite (SO₃²⁻) and hydrogen sulfite (HSO₃⁻) ions prepared from Na₂SO₃, investigated the aqueous sulfur oxidation by NO₂; the measured rate constants differed by 1–2 orders of magnitude (17–19). Typically, this aqueous oxidation has been neglected in atmospheric models because of limited water solubility of NO₂ (1, 13, 20). A model simulation of dissolution of NO₂ in cloud droplets under NO_x-rich environments has shown enhanced regional wintertime sulfate by up to 20%, resulting in better agreement between simulations and observations (13). Also, atmospheric measurements have revealed high sulfate production during severe haze events in China (21–25), which cannot be explained by current atmospheric models and suggests missing sulfur oxidation mechanisms (14). Typically, high sulfate levels during haze events in China occurred with concurrently elevated RH, NO_x, and NH₃ (24, 25), implicating an aqueous sulfur oxidation pathway. However, elucidation of the sulfur oxidation mechanisms from available atmospheric measurements remains challenging, particularly under polluted conditions because of multiple highly coemitted primary gaseous pollutants (1, 21). In this work we investigated the sulfur oxidation mechanism and its role in severe haze formation, by combining field measurements of gaseous pollutants and aerosol particle properties in two Chinese megacities (Xi’an and Beijing) and complementary laboratory experiments (Materials and Methods and SI Appendix).

Results

Sulfate Evolution During Pollution Episodes.

The pollution episodes in Xi’an exhibit a periodic cycle of 4–5 d, which is reflected in the temporal evolutions of the mass concentrations of SO₄²⁻ and PM smaller than 2.5 μm (PM_2.5) (Fig. 1A and SI Appendix, Fig. S1A and Table S1). For each pollution episode, the SO₄²⁻ mass concentration increases markedly from less than 10 μg m⁻³ (clean), 10–20 μg m⁻³ (transition), to greater than 20 μg m⁻³ (polluted), with the corresponding increases in the mean PM_2.5 mass concentrations from 43, 139, to 250 μg m⁻³, respectively. Among the main nonrefractory PM_2.5 species in Xi’an (Fig. 1B), organic matter (OM), nitrate (NO₃⁻), and SO₄²⁻ are most abundant throughout the pollution episode. The SO₄²⁻ mass fraction increases during the transition and polluted (hazy) periods, whereas there is a slight decrease in the OM mass fraction. We quantified the molar ratio of SO₄²⁻ to SO₂, which reflects sulfur partitioning between the particle and gas phases. This ratio ranges from less than 0.1 at relative humidity (RH) <20% to 1.1 at RH >90% in Xi’an, exhibiting an exponential increase with RH (Fig. 1C). During the pollution development, there is increasing RH (Fig. 1D and SI Appendix, Fig. S1B), and the concentrations of SO₂, NO_x (NO + NO₂), and NH₃ are highly elevated (Fig. 1D and SI Appendix, Fig. S2 A–C). Clearly, the larger conversion of SO₂ to SO₄²⁻ during the hazy periods is responsible for the enhanced SO₄²⁻ formation, i.e., with high mass concentrations and mass fractions (Fig. 1 A and B). Field measurements in Beijing also show a similar SO₄²⁻ evolution. There are noticeable increases in SO₄²⁻ and PM_2.5 mass concentrations during the pollution development (Fig. 1E and SI Appendix, Fig. S3A and Table S2). The SO₄²⁻ mass fraction increases from clean to polluted periods, in contrast to a decreasing OM mass fraction (Fig. 1F). During the hazy periods in Beijing, the molar ratio of SO₄²⁻ to SO₂ also exhibits an exponential increase with RH (Fig. 1G), and RH and the concentrations of SO₂, NO_x, and NH₃ are high (Fig. 1H and SI Appendix, Figs. S3B and S4 A–C).

Fig. 1.

Sulfate production during pollution episodes in Xi’an and Beijing. (A–D) Measurements in Xi’an from 17 November to 12 December 2012, and the particle properties correspond to those in PM_2.5. (E–H) Measurements in Beijing from 21 January to February 4, 2015, and the particle properties correspond to those in PM₁ (particles smaller than 1 μm). In A and E, the dates on the x axis correspond to midnight local time. B and F show the mass fractions of the five main nonrefractory constituents from 5 to 12 December 2012 in Xi’an and from 21 January to 4 February 2015 in Beijing, respectively. The lines in C and G represent the exponential fits through the data, i.e., y = 0.07 + 1.0 × 10⁻⁴exp (x/11) with R² = 0.60 in Xi’an and y = 0.05 + 7.0 × 10⁻³exp (x/15) with R² = 0.88 in Beijing. Except for the colors in B and F depicting the aerosol compositions, the blue, orange, and black colors correspond to the SO₄²⁻ mass concentrations of less than 10 μg m⁻³ (clean), 10–20 μg m⁻³ (transition), and greater than 20 μg m⁻³ (polluted), respectively. The top and bottom of the vertical line for each box in D and H correspond to the 95th and 5th percentiles, respectively, and the top, middle, and bottom horizontal lines of the box mark the 75th, 50th, and 25th percentiles of the data range. The white dot in each box represents the mean value.

Our field measurements demonstrate that efficient conversion of SO₂ to SO₄²⁻ occurs at high RH and concurrently with elevated concentrations of SO₂, NO_x, and NH₃, implicating aqueous sulfate production from the participation of these species. Furthermore, the enhanced sulfate formation during the hazy periods is also accompanied by simultaneously increased formation of particulate NO₃⁻ and OM (SI Appendix, Fig. S5). The concentration of ozone is low during the hazy periods in both locations (i.e., a few parts per billion in SI Appendix, Figs. S2D and S4D and Tables S1 and S2), and the visibility is considerably reduced (SI Appendix, Figs. S1C and S3C and Tables S1 and S2), both indicating weak photochemical activity (26, 27). Further examination of the measurements in Beijing reveals markedly continuous growths in the PM_2.5 mass concentration and the average particle size throughout the pollution episodes (SI Appendix, Fig. S6 A and C), which are attributable to efficient formation of SO₄²⁻, NO₃⁻, and secondary organic aerosol (SOA) during the hazy periods. The considerably reduced photochemical activity during the hazy periods is also reflected in the measured decrease in the photolysis rate coefficient of NO₂ (J_NO2) (SI Appendix, Fig. S6B), evident from the anticorrelation between J_NO2 and PM_2.5. Clearly, the efficient PM mass and size growths at high RH and low photochemical activity during the hazy periods are indicative of an increasing importance of aqueous phase oxidation not only for SO₄²⁻ but also for NO₃⁻ and SOA.

A comparison between the two field studies reveals some distinctions. For example, the ratio of SO₄²⁻ to SO₂ at 70% RH is 0.8 in Beijing, much larger than the corresponding value of 0.1 in Xi’an. In addition, the SO₄²⁻ mass fraction and the total inorganic mass fraction in Beijing are larger than those in Xi’an, indicating that fine PM is more hygroscopic in Beijing (21). The measured contents of Fe and Mn of fine PM are small during the hazy periods in Xi’an (SI Appendix, Table S1), consistent with size-resolved composition measurements showing that the mineral elements are usually enriched in coarse particles, because of their dust origins in China (1, 21, 23). With negligibly low concentrations of water-soluble Fe and Mn (SI Appendix, Table S1), the catalytic capability of the mineral elements in fine PM is limited.

Ammonia Neutralization.

To evaluate the PM acidity during the field campaigns, we calculated the equivalent ratio of ammonium (NH₄⁺) to the sum of SO₄²⁻ and NO₃⁻ (SI Appendix, Fig. S7), because these species represent the dominant nonproton cations and anions in fine PM, respectively. During the hazy periods in Xi’an, this ratio remains near unity (Fig. 2A). Hence, SO₄²⁻ and NO₃⁻ in fine PM are completely neutralized, because of the presence of high levels of gaseous ammonia (17–23 parts per billion, ppb) during the hazy periods. Further analysis of the PM_2.5 chemical compositions reveals that the equivalent ratio of the total nonproton cations (NH₄⁺, Na⁺, Ca²⁺, Mg²⁺, and K⁺) to anions (SO₄²⁻, NO₃⁻, and Cl⁻) is also near unity (Fig. 2B and SI Appendix, Fig. S8A), with the mean values of 1.15 ± 0.14 and 1.06 ± 0.06 during the transition and polluted periods, respectively. The close balance between these cations and anions in Xi’an further confirms that fine PM exhibits negligible acidity. Similarly, the equivalent ratio of NH₄⁺ to SO₄²⁻ and NO₃⁻ is slightly larger than unity throughout the pollution episodes in Beijing (Fig. 2A and SI Appendix, Fig. S7B). When the chloride anion, which likely exists as NH₄Cl in ambient PM, is included, the ratio is reduced to 1.09 ± 0.11 during the polluted period (Fig. 2B and SI Appendix, Fig. S8B). Hence, fine PM in both locations is effectively neutralized by ammonia with a calculated pH ∼7 (SI Appendix, Tables S1 and S2), when rapid sulfate production occurs during the polluted period. Interestingly, our results of fully neutralized fine PM in China are in contrast to a recent study showing highly acidic aerosols in the southeast United States, despite declining atmospheric sulfate concentrations over the past 15 years (28).

Fig. 2.

Neutralizing effect of ammonia on fine PM. (A) Equivalent ratio of NH₄⁺ to the sum of SO₄²⁻ and NO₃⁻ in Xi’an and Beijing. (B) Equivalent ratio of the total cations to anions in Xi’an and equivalent ratio of NH₄⁺ to the sum of SO₄²⁻, NO₃⁻, and Cl⁻ in Beijing. The blue, orange, and black colors correspond to the clean, transition, and polluted periods, respectively, as defined in Fig. 1.

An Aqueous Synergetic SO₂ Oxidation Pathway.

To elucidate the mechanism of SO₂ oxidation and interpret the rapid sulfate production in our field measurements, we conducted a series of laboratory experiments by exposing pure water or ammonium (3 wt %) solutions under dark conditions to gaseous SO₂ and NO₂ in a reaction cell. Sulfate formation was quantified (SO₄²⁻ at m/z = 96, SI Appendix, Fig. S9A) by thermal desorption-ion drift-chemical ionization mass spectrometry (TD-ID-CIMS) (29, 30). When the pure water or ammonium solutions were exposed simultaneously to SO₂ and NO₂ using either N₂ or air as the buffer gas, significant SO₄²⁻ production was detected, and the signal was higher in the ammonium solution than in pure water (SI Appendix, Table S3). In contrast, the SO₄²⁻ production was absent for only SO₂ exposure under similar conditions, indicating the oxidizing role of NO₂. Also, there was little difference in the measured SO₄²⁻ production between experiments using N₂ and air, suggesting negligible SO₂ oxidation by O₂ molecules.

We performed additional experiments by exposing seed particles to gaseous SO₂, NO₂, and NH₃ under dark and variable RH conditions in a reaction chamber (SI Appendix, Fig. S10). Size-selected oxalic acid particles, which were used to represent organic aerosols that dominate the early stages of haze development in China (21) (see also Fig. 1 B and F), were simultaneously exposed to SO₂, NO₂, and NH_3, while the variation in the dry particle size was monitored. The evolution in particle size distributions measured after exposure to three different RH conditions is depicted in Fig. 3A: The size distribution remains unchanged at 30% RH (i.e., identical to that of the initially seeded particles), whereas exposures at 60% and 70% RH lead to dramatic shifts to larger size distributions. We also conducted experiments to analyze the chemical composition of exposed particles. SO₄²⁻ production in collected particles after the exposure at high RH is clearly evident (Fig. 3B and SI Appendix, Fig. S9B).

Fig. 3.

Aqueous sulfate formation in the reaction chamber. (A) Evolution in the dry particle size distribution when sized selected oxalic acid particles are exposed to SO₂, NO₂, and NH₃ under three different RH conditions in a 1-m³ reaction chamber. (B) Desorption spectra of particles collected by TD-ID-CIMS after exposure to SO₂, NO₂, and NH₃ at 65% RH. (C) Particle growth factor after exposure to SO₂, NO₂, and NH₃ as a function of RH. The exponential fit is y = 1.05 + 4.0 × 10⁻⁵exp (x/6.8) with R² = 0.96. Each point corresponds to three measurements, and the error bar denotes the SD (1σ). (D) Integrated desorption peak areas of particles collected by TD-ID-CIMS after exposure to SO₂, NO₂, and NH₃ as a function of RH. The exponential fit is y = 1.8 × 10⁶ + 417exp (x/7.9) with R² = 0.97. All experiments were performed under the dark condition and at temperature of 298 K. The exposure time was 60 min, and the initial particle size was 45 nm. The initial gas concentrations were 250 ppb for SO₂ and NO₂ and 1 ppm for NH₃.

We quantified the growth of seed particles after exposure to SO₂, NO₂, and NH₃, on the basis of the measured ratio of the dry particle sizes (D_p/D_o), where D_p and D_o are the values after and before the exposure, respectively. The growth factor increases with RH, with the values of near unity (no growth) at RH< 20% and 2.3 at 70% RH (Fig. 3C). The size growth after exposure is explained by SO₄²⁻ production, as depicted in the corresponding increase of the integrated SO₄²⁻ desorption peak areas (Fig. 3D). Both the particle size growth and the SO₄²⁻ formation after the exposure to SO₂, NO₂, and NH₃ display a similar exponential increase with RH, consistent with increasing hygroscopic growth of oxalic acid particles (31). We performed additional measurements when one or both of NO₂ and NH₃ were excluded from the exposure: In the absence of NO₂, NH₃, or both from the exposure, no observable particle growth or SO₄²⁻ production was measured (SI Appendix, Table S4).

On the basis of our combined field and laboratory measurements, we establish the occurrence of an overall aqueous reaction between SO₂ and NO₂,

$${\text{SO}}_2\left(\text{g}\right)+{2\text{NO}}_2\left(\text{g}\right)+{2\text{H}}_2\text{O}\left(\text{aq}\right)\mathrm{\to }{2\text{H}}^{+}\left(\text{aq}\right)+{\text{SO}}_4^{2-}\left(\text{aq}\right)+2\text{HONO}\left(\text{g}\right).$$ [1]

Because this reaction is second order with respect to NO₂ accommodation, its rate is strongly dependent on the gaseous NO₂ concentration, i.e., only proceeding efficiently under NO₂-rich conditions. This reaction is also dependent on pH, which not only governs the solubility but also the aqueous reaction rate. For example, when the pH value is varied from 6 to 4, the effective Henry’s constant of SO₂ decreases by more than two orders of magnitude (1–3), leading to a decreased oxidation rate of approximately two orders of magnitude (17–19). In addition, the solubility is dependent of temperature, i.e., increasing with decreasing temperature (32–34). Also, the reaction is self-limiting because of the acidity effect, namely that its occurrence increases acidity in the aqueous phase and in turn reduces the solubility and reaction rate. Nitrous acid (HONO) generated from this reaction is likely released into the gas phase because of its limited solubility in aqueous solution (34). The measurements in Xi’an show that the HONO concentration increases during the pollution development, reaching 2–3 ppb during the hazy periods (SI Appendix, Fig. S11A). Also, the measured HONO concentration is correlated with increasing RH but is inversely correlated with gaseous SO₂, likely supporting its formation from the aqueous SO₂ oxidation by NO₂ (SI Appendix, Fig. S11B). The previously proposed HONO formation mechanisms include heterogeneous conversion of NO₂ on ground or aerosol surfaces (1–3).

Our results indicate that aqueous SO₂ oxidation by NO₂ is favored in two atmospheric scenarios (Fig. 4), i.e., under cloud/fog conditions and on fine aerosols with high RH (> 60–70%) and sufficient neutralization (pH ∼ 7). In-cloud SO₂ oxidation proceeds in the presence or absence of a neutralizing agent, when elevated levels of SO₂ and NO₂ coexist (Fig. 4A). Cloud droplets are large (exceeding several tens of micrometers in sizes), and the amount of sulfate formed is sufficiently diluted and does not appreciably alter the particle acidity (1–3). Consequently, water evaporation from cloud droplets under unsaturated conditions leads to concentrated sulfuric acid particles (33), contributing to acid rain and regional acid deposition (9, 13). Also, in-cloud SO₂ oxidation can be further enhanced in the presence of the basic species, as demonstrated in our laboratory work for NH₃ (SI Appendix, Table S3). On the other hand, the oxidation on fine PM is inhibited by the acidity effect, and the presence of basic species (i.e., NH₃) is necessary to maintain the oxidation (Fig. 4B),

$${2\text{NH}}_3\left(\text{g}\right)+{\text{SO}}_2\left(\text{g}\right)+{2\text{NO}}_2\left(\text{g}\right)+{2\text{H}}_2\text{O}\left(\text{aq}\right)\mathrm{\to }{2\text{NH}}_4^{\mathrm{+}}\left(\text{aq}\right)+{\text{SO}}_4^{2-}\left(\text{aq}\right)+2\text{HONO}\left(\text{g}\right).$$ [2]

The acidity effect is relevant to the ionic strength, which is highly dependent on the particle size, i.e., decreasing by two to four orders of magnitude from submicrometer aerosols to cloud droplets (1–3, 35). Note that this acidity effect on aerosol sulfate formation (i.e., decreasing with increasing particle size) is analogous to the Kelvin (curvature) effect, which represents a major barrier in aerosol nucleation (36, 37). In addition, neutralization of fine PM can be facilitated by amines, albeit at a lower atmospheric concentration than NH₃ (38, 39).

Fig. 4.

Schematic of the sulfate formation mechanisms. Variations in temperature, RH, and particle size and acidity for the aqueous reactions between SO₂ and NO₂ leading to SO₄²⁻ formation under in-cloud conditions (A) and on fine PM (B). The red, black, yellow, and green colors in A and B represent SO₄²⁻, H⁺, NH₄⁺, and SOA, respectively. (C) Anticorrelation between the photochemical activity and aqueous chemistry during the severe haze evolution (i.e., from the clean, transition, to polluted periods) in China, displaying the central role of the SO₂ to SO₄²⁻ conversion in facilitating aqueous production of the major secondary constituents.

Hence, the acidity, hygroscopicity, and RH represent the key factors for sulfate formation on fine PM, explaining the differences in the various ambient measurements (14, 20–25). For example, the acidity effect on fine aerosols is effectively overcome by NH₃ neutralization in Xi’an and Beijing (Fig. 2). Also, the noticeably earlier increase of the SO₄²⁻ to SO₂ ratio with RH in Beijing than in Xi’an (Fig. 1 C and G) is attributable to more hygroscopic aerosols, because of a larger inorganic mass fraction in Beijing (Fig. 1 B and F) (21).

We derived the equivalent SO₂ uptake coefficient (γ) for sulfate production from our field and laboratory results in SI Appendix, Tables S5 and S6, respectively. The γ-values derived from the Beijing measurements are (2.1 ± 1.6) × 10⁻⁵ and (4.5 ± 1.1) × 10⁻⁵ during the transition (41% RH) and polluted (56% RH) periods, respectively, compared with (8.3 ± 5.7) × 10⁻⁵ at 65% RH and (3.9 ± 1.2) × 10⁻⁴ at 70% RH derived from the laboratory measurements. Hence, our laboratory experiments reproduce the rapid sulfate production measured under polluted ambient conditions, and these kinetic data are applicable for quantifying sulfate formation in atmospheric models (1, 13, 14).

The Central Role of Sulfate Production in Severe Haze Development.

Our results indicate that the formation of the various secondary organic and inorganic constituents in fine PM is mutually promoting and the severe haze development involves a transition from photochemical to aqueous phase processes (Fig. 4C). During the early stage, efficient photochemical oxidation of volatile organic compounds (VOCs) leads to SOA formation (Fig. 1 B and F), which provides an aqueous media for subsequent SO₄²⁻ production. With high RH and low photochemical activity during the later hazy periods, continuously large PM growth (i.e., the SO₄²⁻, NO₃⁻, and OM mass increases) is maintained by the aqueous chemistry (SI Appendix, Fig. S5). In particular, the SO₄²⁻ production likely represents the most critical step in initializing the aqueous chemistry, because of increasing particle hygroscopicity. Efficient SO₂ to SO₄²⁻ conversion not only contributes to the high SO₄²⁻ production rate, but also enhances formations of NO₃⁻ and SOA on aqueous particles, explaining the sustained high production of the major secondary constituents during the hazy periods in our current field measurements and those of the previous studies in China (1, 21, 40). For example, with reduced photochemistry during the hazy periods, the measured large NO₃⁻ mass concentration is attributable to an enhanced heterogeneous conversion of NO_x to HNO₃, because the hydrolysis reaction of N₂O₅ occurs efficiently on sulfate aerosols (41). Also, hydration and oligomerization reactions of glyoxal and methyglyoxal, which are produced with high yields by aromatic hydrocarbon oxidation from traffic emissions, are enhanced by sulfate formation, because these reactions are highly dependent on particle hygroscopicity (30, 32, 42). Furthermore, gaseous HONO formed from the aqueous SO₂ oxidation with NO₂ provides an additional photochemical OH source that enhances the atmospheric oxidizing capability during the hazy periods (43). It should also be pointed out that severe haze formation in China is characterized by a complex interplay between meteorological, thermodynamic, and chemical processes (1, 21, 44).

Conclusion

Atmospheric sulfur chemistry has remained an open problem (1, 13, 14). The formation of the 1952 London “Killer” Fog is still mysterious in terms of the detailed chemical mechanism for SO₂ conversion to sulfate (1, 45). Our results indicate that the formation of London Fog was similar to in-cloud SO₂ oxidation by NO₂ (Fig. 4A), because both species were present in highly elevated levels as the coproducts of coal burning. The sulfate formation was greatly facilitated by high RH, low temperature, and the presence of large fog droplets (45), yielding elevated sulfuric acid levels that persisted throughout the event. The particle acidity was regulated by temperature, and water evaporation from fog droplets at warmer temperature resulted in concentrated sulfate acid particles (33), explaining the highly acidic nature of the London Fog (45).

Interestingly, we show that the same sulfur problem persists presently to contribute to severe haze formation in China, although the fine PM is mainly nonacidic. Major emission sources in China include industry (for SO₂, VOCs, and NO_x) and traffic (for VOCs and NO_x), because of its fast-growing economy and urbanization (1, 23, 46, 47). Also, there has been a rapid increase in the production and use of nitrogen fertilizers in China, leading to high NH₃ emissions (48). For example, the emissions of SO₂, NO_x, and NH₃ in China are estimated to be about 22 Tg S y⁻¹, 19 Tg N y⁻¹, and 15 Tg N y⁻¹ in 2010, respectively (48). In addition, traffic emissions have been suggested to represent an important urban NH₃ source (49). High emissions of these organic and inorganic PM precursors result in large secondary production of SO₄²⁻, NO₃⁻, NH₄⁺, and SOA in China (Fig. 4C), via the combined atmospheric photochemical and aqueous processes (1, 21, 40, 50–53). Our results indicate that sulfate production is key to the formation of persistent severe haze in China (Fig. 4 B and C). Whereas current efforts have been focused primarily on minimizing SO₂ emissions (1, 14, 21), significant haze reduction may only be achievable by disrupting this sulfate formation process. For example, controlled NH₃ emissions may be important, because the acidity effect represents the key rate-limiting factor in sulfate production on fine PM. Also, because of the second-order nature of NO₂ in the aqueous SO₂ oxidation (i.e., reactions 1 and 2), reduction of the NO_x level is likely effective in lowering sulfate formation. In light of large contributions to urban NO_x, VOC, and NH₃ levels from transportation (1, 21, 49), regulatory actions in minimizing traffic emissions may represent the critical step in mitigating severe haze in China. These measures are clearly supported by our experimental results, showing no particle growth or sulfate formation at high RH when oxalic acid particles were exposed to high levels of SO₂ in the absence of NH₃, NO₂, or both (SI Appendix, Table S4).

In addition to polluted urban areas, efficient sulfate production is also expected in the proximity of power plant and biomass burning plumes and ship tracks (1–3, 9, 13), where SO₂ and NO_x are coemitted. Because of increasingly high SO₂, NO_x, VOC, and basic species (NH₃ and amine) emissions in many developing countries (1, 38, 48, 49), the synergetic sulfate formation pathway identified in our work is likely widespread globally, contributing not only to air quality problems but also to enhanced nitrogen (i.e., NH₄⁺ or NO₃⁻) or acid (in the absence of basic species) deposition, with major implications for the ecosystem vitality, greenhouse gas budgets, and biological diversity (48). Our results highlight the necessity for comprehensive understanding of the atmospheric aerosol chemistry in the development of effective pollution mitigation policies (1), to minimize the impacts of fine PM on visibility, human health, ecosystems, weather, and climate.

Materials and Methods

Field measurements of gaseous and PM pollutants were performed in Xi’an and Beijing. The sampling site in Xi’an (from 17 November to 12 December 2012) was located on the rooftop (around 10 m above the ground) of a three-story building on the campus of the Institute of Earth Environment of Chinese Academy of Science (CAS) in the southwest of the city (54). The sampling site in Beijing (from 21 January to 4 February 2015) was located on the campus of Peking University in northwestern Beijing (21). Gaseous species and PM properties were monitored by a suite of instrumentations and methods (54–64). Laboratory experiments were performed to evaluate SO₂ oxidation by NO₂ on bulk solutions and aerosols under dark conditions (see also SI Appendix). Pure water or 3 wt % NH₃ solution was exposed to SO₂ and NO₂ in N₂ or pure air using a reaction cell. The exposed solution from the reaction cell was analyzed by TD-ID-CIMS for sulfate formation. To evaluate the conversion of SO₂ into SO₄²⁻ on aerosols under conditions relevant to the atmosphere, we conducted experiments by exposing seed particles to SO₂, NO₂, and NH₃ and measuring the size variation and sulfate formation on the exposed particles in a 1-m³ Teflon reaction chamber covered with aluminum foil (SI Appendix, Fig. S10). Size-selected oxalic acid particles (45 nm) were used as model aerosols in the reaction chamber for the aqueous conversion of SO₂ to sulfate, by exposing to SO₂, NO₂, and NH₃ at variable RH. The variations in the dry particle sizes and sulfate formation were measured by a differential mobility analyzer and TD-ID-CIMS, respectively. Additional descriptions of the instrumentation and procedures of the field and laboratory measurements are provided in SI Appendix.