Has climate change affected the formation of sulfate through the oxidation of copper SO2 by NO2 at aerosol levels? If so, in which regions? Is there a way to prevent it? What are the effects of humans and greenhouse gases?

Severe urban air pollution in China is driven by a synergistic conversion of SO2, NOx, and NH3 into fine particulate matter (PM2.5). Field studies indicated NO2 as an important oxidizer to SO2 in polluted atmospheres with low photochemical reactivity, but this rapid reaction cannot be explained by the aqueous reactive nitrogen chemistry in acidic urban aerosols. Here, using an aerosol optical tweezer and Raman spectroscopy, we show that the multiphase SO2 oxidation by NO2 is accelerated for twoorder-of-magnitude by a copper catalyst. This reaction occurs on aerosol surfaces, is independent of pH between 3 and 5, and produces sulfate by a rate of up to 10 µg m-3 air hr-1 when reactive copper reaches a millimolar concentration in aerosol water – typical of severe haze events in North China Plain. Since copper and NO2 are companion emitters in air pollution, they can act synergistically in converting SO2 into sulfate in China’s haze.Air pollution is a persistent problem in developing countries, such as China and other emerging economies experiencing rapid industrialization1–3 . Among the pollutants, the principal culprit is fine particulate matter (PM2.5), the airborne particles that can penetrate into human lungs, leading to premature deathsfrom cardiovascular and respiratory diseases and lung cancer4,5 . To mitigate PM2.5 and its public health impacts, the Chinese government renewed its air-quality policies in 2021, aiming for a 10% reduction of urban PM2.5 concentration by 20256 . Achieving this objective requires a clear understanding of the atmospheric chemistry producing PM2.5 in urban haze. China’s haze differs from London fog or Los Angeles smog in several ways. First, gas pollutants coexist at high concentrations3,7–9 , including SO2, NOx (NO and NO2), and NH3, emitted from industry, traffic, and agriculture10–12. These gases convert synergistically into PM2.5 through atmospheric multiphase reactions3,7–9 . Second, the multiphase reactions occur rapidly, much faster than what aqueous chemistry predicts8,13–15. Such rapid kinetics may result from many factors, including enhanced chemical reactivities at the air-water interface14,16,17, the catalytic effects of transition metal ions (TMI)13,18, or the salt effects in the oversaturated aerosol water15,19 – or all these factors acting simultaneously. Recognizing these characteristics, scientists coined the term haze chemistry to describe how PM2.5 is formed during the severe urban air pollution in China3,7–9,20. A decade-long debate in haze chemistry research concerns whether NO2 can effectively oxidize SO2 into sulfate, thereby contributing to PM2.5 formation. Why is NO2 considered an oxidizer of SO2? First, this redox reaction can occur in the atmospheric environments; NO2 can oxidize SO2 on the surfaces of primary particles (i.e., soot21 and dust9 ) and, more prevalently, in aerosol water7,22–24:2NO2 aq ð Þ þ HSO 3 aq ð Þ þ H2O ! Hþ ð Þ aq þ 2HONOð Þ aq þ SO2 4 aq ð Þ

Reaction Additionally, NO2 is abundant in the urban haze, especially when photochemical oxidizers such as O3, H2O2, and OH are inhibited in the polluted troposphere dimmed by haze8 or at night25. Field campaigns in China8,26,27 showed that sulfate and NO2 concentrations are positively correlated. A Beijing campaign25 found that the HONO produced by Reaction 1 can even further oxidize SO2. An air-quality model8 predicted that, at pH 5.8, the reaction between HSO 3 and NO2 (hereafter, the HSO 3 /NO2 reaction, and so forth) would produce sulfate by a rate of 10 µg m−3 air hr−1 . A laboratory study14 found that, at pH 6, a SO2 3 /NO2 reaction would produce sulfate by 90 µg m−3 air hr−1 .These studies suggested that sulfate PM2.5 in China’s haze was produced mainly through the NO2 reaction pathway. Yet equally compelling evidence indicates that NO2 contributed to SO2 oxidation negligibly. Although Reaction 1 can occur in the aqueous phase, it is unlikely to occur through a direct electron transfer, because the redox potentials between aqueous HSO 3 and NO2 are close28,29. The reaction instead occurs through the formation of [NO2-SO3] 2- adducts, which decompose to SO 3 radicals slowly28. This kinetic constraint rules out a rapid sulfate formation through Reaction 1. Additionally, the average pH of urban aerosols in China30 is approximately 4, which is more acidic than what previous studies have assumed8 . At acidic conditions, SO2 has limited solubility, leaving too few S(IV) ions (HSO 3 and SO2 3 ) to facilitate a rapid sulfate formation13,31. At pH 4, the sulfate formation rate via HSO 3 /NO2 8 and SO2 3 /NO2 14 reactions are respectively 0.04 and 0.15 µg m−3 air hr−1 . Recent air-quality models13,32 showed that the HSO 3 /NO2 reaction contributed approximately 0.1% of sulfate13; the SO2 3 /NO2 reaction, approximately 0.4%32. A source apportion study31 showed that the NO2 reaction pathway contributed at most 1% of sulfate in China haze. A recent globalscale study33 found that the NO2 reaction pathway is unimportant unless aerosol pH is above 5, a condition rarely met worldwide. These disagreements8,25–30,33 indicate a knowledge gap regarding how sulfate is produced in urban air pollution. Why does it matter whether NO2 contributes to sulfate formation? If so, then both SO2 and NO2 would be sulfate precursors, and effective abatement would require closer coordination between the industry and transportation sectors3,7–11. Bridging this knowledge gap requires us to answer the following question: Can the multiphase SO2 oxidation by NO2 occur rapidly at acidic conditions? Here, we show that NO2 can oxidize SO2 into sulfate rapidly at acidic conditions when the reaction is catalyzed by copper (hereafter, Cu). Cu, albeit a transition metal, is a weak catalystfor S(IV) oxidation by O2 34. But Cu is a strong catalyst for NO2 reduction by S(IV) in flue gas de-nitrification35,36. Additionally, Cu is ubiquitous in urban air pollution37. A field campaign38 reported that Cu elements were on the orders of hundreds of ng m−3 air during the air pollution in North China Plain (NCP). In Beijing, Cu mainly originates from traffic emissions, i.e., brake and tire wear39; in the broader NCP region, Cu mainly originates from coal combustions39. On the other hand, NO2 originates from both industrial and traffic emissions11,12, and its concentration can reach 40-to-80 ppb during heavy air pollution in NCP8 . In other words, Cu and NO2 are companion emitters, and they may synergistically convert SO2 into sulfate during urban haze. Furthermore, we show that the kinetics of the ternary Cu/SO2/NO2 reaction depends more sensitively on NO2 concentration, rather than on SO2 concentration. This may explain why, over the past decade, a substantial decrease in SO2 emission has not led to a proportional decrease in sulfate concentration in China. Results Method summary We studied the Cu-catalyzed reaction with Raman micro-spectrometry (hereafter, micro-Raman) and an aerosol optical tweezer (hereafter, AOT). The micro-Raman experiments provided information on the reaction mechanism, including the reaction products, the catalytic effect of Cu(II) ions, and kinetic dependence on droplet size (radius 5–30 µm) and acidity (pH 3–5). The AOT experiments provided kinetic data for the reactions in levitated droplets, under conditions closely mimicking urban air pollutions, such as droplet solute ((NH4)2SO4), acidity (pH 4), relative humidity (RH 60%), and reactant gases mixing ratio (SO2, 5–200 ppb; NO2, 50–500 ppb), and reaction time (hours). We designed these experiments based on literature values of aerosol pH13,30, gas concentrations8 , and RH conditions14. Specifically, the ranges of these parameters encompass their average values during severe pollution events in Beijing (i.e., pH 4, SO2 40 ppb, NO2 66 ppb). Refer to the Methods section for details. Copper-catalyzed SO2 oxidation by NO2 Figure 1A shows the Raman spectra of microdroplets, which served as reactors for the oxidation of SO2 (500 ppb) by NO2 (500 ppb). Droplet pH was buffered at approximately 4 with 400 ppb NH3 40. Ambient RH was approximately 80%. The left panel represents the reaction catalyzed by Cu(II) ions in the microdroplet seeded with a mixture of NH4Cl/HCl/CuCl2 (1:0.005:0.001). Here, the Raman spectrum exhibits a peak around 980 cm−1 , indicating SO2 4 formation (See Figure S1 for the full spectrum). This catalyzed reaction produced approximately 0.4 M sulfate in 240 min. Contrastingly, the right panel represents the uncatalyzed reaction in the microdroplet seeded with NH4Cl/HCl (1:0.005). This uncatalyzed reaction was too slow to be measured with the micro-Raman. In Figure S2, the AOT data shows that the reaction catalyzed by 0.1% Cu-in-solute was faster than the uncatalyzed reaction by two orders of magnitude. In both cases, the reaction did not produce NO 3 , which would exhibit a Raman peak at 1050 cm−3 . In other words, NO2 served only as an oxidizer of SO2 and did not undergo disproportionation at our experimental conditions. Figure S3 shows another control experiment, where NO2 was not applied, and no sulfate was produced within 240 min. Figure 1B–D show the kinetic dependence on droplet size and acidity. These experiments were conducted in NH4Cl/HCl/CuCl2 droplets (1:0.005:0.001), with a radius (hereafter, a) between 5 and 30 µm. Other conditions were 500 ppb SO2, 500 ppb NO2, 40-to-4000 ppb NH3, and 80% RH. Figure 1B shows that the reaction is faster in smaller droplets. Specifically, the SO2 4 formation rate, d SO2 4=dt, (unit: M s−1 ) is inversely proportional to droplet radius, a (See the dotted line in Fig.1C). This relationship indicates that the reaction rate is proportional to the droplet surface-area-to-volume ratio, such as A=V: Hereafter, we will normalize kinetic data as below:Droplet ambient conditions. The aerosolized droplets, led by an N2 flow, were then delivered to the optical trap inside the sample cell of the AOT system. This sampling process was considered successful when one of the droplets was captured by the optical trap. At this stage, the composition of the droplet is highly sensitive to the ambient gas phase, which should be maintained at stable conditions throughout the measurement. Specifically, the relative humidity (RH, 60 ± 1%) in the cell was controlled by mixing dry and humidified N2 gases. The RH was monitored with a hygrometer (CENTER-313, Qunte Technology Co., LTD). The temperature was maintained at room temperature (298 K) Reactant gases flowed through the sample cell with a prescribed mixing ratio. When investigating the Cu-catalyzed reaction, we applied the reaction gases per the following arrangements: SO2 was at 5, 10, 15, 25, 40, 50, 75, 100, 150, or 200 ppb; NO2 was at 50, 100, 250, or 500 ppb; NH3 was at 50, 100, 200, 400, or 800 ppb to buffer the (NH4)2SO4/NH4HSO4 droplets at pH 2.8, 3.1, 3.4, 3.7, or 4.0, respectively. When investigating the uncatalyzed reaction, we applied the gases per the following arrangements: SO2was at 0.1, 0.25, 0.5, 0.8, 1.0, 1.5, 2.0, 3.0, 5.0, 7.5, 10.0, or 20.0 ppm; NO2 was at 10 ppm; NH3 was at 0.8 or 8.0 ppm to buffer the droplets at pH 4.0 or 5.0 conditions, respectively. The detailed experimental conditions for the AOT study can be found in Table S7. Droplet pH was calculated with E-AIM40. Raman spectral data collection and analyzes. The backscattered Raman signal was collected with a time resolution of one second. During the reaction, the SO2 was continuously converted into SO2 4 , causing a continuous increase in droplet radius, a. Such droplet growth, albeit slight in magnitude, can be precisely determined by observing the redshift of the whispering gallery mode (WGM) in the stimulated Raman spectra. At each time step, t, we inverted the WGM wavelength λ to droplet size a, by using the Mie-scattering calculation algorithm provided in ref. 47. Next, the increase in droplet volume dV, during a time interval dt, can be quantified as: dV ¼ 4πa2 t atþdt a EquationS1Þ This increase in volume was contributed by the (NH4)2SO4 produced by the reaction, and the corresponding increase in the mole of (NH4)2SO4 is therefore: dn ¼ ðNH4Þ2SO4 dV ðEquationS2Þ Here, ðNH4Þ2SO4 is the molar concentration at an approximately 60% RH condition, calculated with E-AIM40. In summary, the reaction rate can be calculated from droplet growth rate per the following relationships: dn dt mols2 0da dt × L 1015μm3 ðEquationS3Þ and R mols1 μm1L 1015μm3 ðEquationS4Þ Here, a0 is the initial radius of droplets (unit, µm). The droplet growth rate da=dt (unit, µm s−1 ) was determined by linearly fitting the aðtÞ dataset. It is worth noting that Eqs. S1, S3, and S4 hold true only when the value of at a0 is much smaller than a0 (so that the curvature of the droplet surface can be ignored). In the AOT experiments, the at a0 did not exceed 50 nm, which is approximately 1% of the a0. For each experiment, the values of da=dt, the uncertainties (95% confidence interval values of the linear fitting), and the number of data points used in the fitting can be found in Table S7. Also, note that the treatment of Eqs. S3 and S4 requires that SO2 4 is the sole product remaining in the condensed phase. Such a condition has been confirmed in our Micro-Raman study (See Fig.S7). We also assumed that the productwas always (NH4)2SO4when NH3was in the ambient gases. Aqueous copper speciation model Visual MINTEQ. Following the method introduced in refs. 63,64, we estimated the chemical speciations of Cu(II) in the aqueous phase of Beijing PM2.5 by using Visual MINTEQ model version 3.1. The visual MINTEQ model, which was originally designed for chemical speciation analysis in natural aquatic systems, has also been successfully utilized for estimating metal speciation in aerosol water63,64. In other words, this model accounts for metal-organic complex formation and calculates the fraction of metals existing as organic complexes. The model input parameters included the aqueous concentrations of secondary inorganic matters (SO4 2−, NO3 −, and NH4 +), dicarboxylic acids (oxalate, malonate, succinate, and glutarate), and metal ions (Na+, K+, Mg2+, Ca2+, Al3+, Mn2+, As3+, Cr2+, Cu2+, Ni2+, Pb2+, Sb3+, Se4+, Zn2+, Fe2+, and Fe3+), as well as aqueous pH (fixed at 4) and temperature (fixed at 25 °C). These PM2.5 composition data were acquired from field campaigns conducted in Beijing38,66,67. Details of the composition data can be found in Table S4. Following that recommended in ref. 64, we adopted the specific interaction theory (SIT) for the ionic strength correction of the stability constants of the metal complexes. SIT correction was preferred because it is more appropriate for the high ionic strength condition (>1 M) of urban aerosols64. Sulfate, nitrate, and ammonium. We estimated the aqueous concentrations of inorganic matters with the E-AIM model40 according to the hygroscopicity of a SO4 2−/NO3 −/NH4 + mixture at a molar ratio of 1:1.5:3.5 and at an ambient RH of 80%. The molar ratio of the mixture was determined according to the mass fractions of SO4 2− (19.2%), NO3 − (18.5%), and NH4 + (12.6%) in Beijing PM2.5 at heavily polluted conditions67.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.