Formation and Evolution of Secondary Particulate Matter During Heavy Haze Pollution Episodes in Northeast China in Winter

Based on the simultaneous observation of ne particulate matter (PM 2.5 ) and its chemical components in four at 14 sampling sites in northeast China from 2017 to 2019, the formation and existence of sulfate (SO 42- ) and nitrate (NO 3- ) secondary contaminants under different stages of the pollution episodes, and different meteorological and emission conditions were compared. The results yielded three main ndings. (1) Organic carbon (OC) was the most important component of PM 2.5 , followed by NO 3- ,SO 42- ,and ammonium (NH 4+ ). Nitrate surpassed sulfate as the most important secondary inorganic component over the study period. (2) The signicant increase in atmospheric OC, SO 42- , and NO 3-concentrations was an important reason for haze formation. Meteorological factors such as wind direction, wind speed, temperature (T), relative humidity (RH), and atmospheric oxidability played an important role in secondary pollutant formation. (3) There were two potential SO 42- formation mechanisms. The rst was the gas-phase reaction of the hydroxyl radical(OH·) leading to the oxidation of nitrogen dioxide (NO 2 ) and sulfur dioxide (SO 2 ),and high ozone (O 3 ) concentrations. A high atmospheric oxidability and high winter Ts were very important for SO 42- formation. The second mechanism occurred under neutral or weakly alkaline conditions when large amounts of SO 2 could enter aerosol droplets, and NO 2 was more likely to react in the aqueous phase with SO 2 to increase the output of SO 42- . Nitrate formation was may be mainly due to the homogeneous gas-phase reaction of OH· with NO 2 , SO 2 , and ammonia(NH 3 ). The highest NO 3 concentration was observed under mild winter Ts, high RH, high atmospheric oxidability (O 3 and nitrous acid (HONO)), high NH 3 concentrations, and suitable light conditions. The differences in SO 42- formation between northeast China and other regions were mainly a result of the suppression of the aqueous reaction of SO 42- due to the low T in winter and low-sulfur coal emissions, which resulted in the gas-phase oxidation process with the highest SO 42- production capacity becoming an important process. However, the aqueous reaction process 42− (Sun et al., 2013). The NO 3− formation was more complicated than that of SO 42− . Section 3.3.1 of this article shows that a high atmospheric oxidability, high NH 3 , high NO 2 , and high [NH 4+ ] excess were all important in NO 3 formation.


Introduction
There has been an increase in haze pollution in northeast China in recent years. The causes of haze pollution and the formation of the major chemical pollutants are not well understood. Northeast China is located in the mid-high latitudes of the northern hemisphere and is an area with large anthropogenic emissions (van Donkelaar et al., 2016). It is also one of the regions of the world where the terrestrial climate is warming the most rapidly due to global climate change. The rate of increase in the winter surface temperature (T) is signi cantly higher than the Chinese and global average (Zhao et al., 2007). This change will have an impact on the regional and global climate (Sobhaniet al., 2018). During heavy haze periods in northeast China, aerosols make the most signi cant contribution to organic carbon (OC) and secondary inorganic components (e.g., sulfate (SO 4 2− ) and nitrate (NO 3 − )) in the atmosphere . Both can account for more than 50% of ne particulate matter (PM 2.5 ) and play a pivotal role in haze formation (Zhang et al., 2020). Therefore, clarifying the causes of secondary aerosol formation is the key to understanding the causes of the persistent heavy haze in this region.
The occurrence of haze pollution is closely related to the chemical reactions between polluted particles in the air. Various physical processes and chemical reactions occur between different ions under different meteorological conditions, which strengthens the photochemical effect of haze particles and is the internal promoting factor of the "explosive" nature and "sustainability" of strong haze pollution . Sulfate in the atmosphere is mainly formed by the photochemical oxidation of sulfur dioxide (SO 2 ), which can be formed either by gas-phase homogeneous reactions or aqueous heterogeneous oxidation reactions. Studies have shown that the gas-phase oxidation is mainly the gas-phase oxidation of SO 2 by the hydroxyl radical (OH·), and aqueous oxidation occurs mainly in cloud/fog water droplets or aerosols, under the catalysis of transition metal ions (TMIs), e.g., Fe(III) and Mn(II), the oxidation of S(IV) by hydrogen peroxide (H 2 O 2 ), ozone (O 3 ), organic peroxides and oxygen (O 2 ), of which aqueous oxidation is the main SO 4 2− formation pathway worldwide (Wang et al., 2021). Its generation rate is closely related to the pH of the particulate matter.
Under acidic conditions, the catalytic oxidation of transition metals accounts for more than 90% of the generation, while under near-neutral conditions, the oxidation of H 2 O 2 and nitrogen dioxide (NO 2 ) each account for 50% (Wang, 2021a). A typical feature of the heavy pollution in northern China in winter is a rapid increase in the SO 4 2− concentration, and the sulfur oxidation rate increases exponentially with an increase in the relative humidity (RH) (Elser et al., 2016). This indicates that the heterogeneous chemical reaction under a high RH is an important pathway for SO 4 2− formation during heavy haze pollution episodes . Through a smoke box simulation study, Wang (2016a) found that the liquid(aqueous) phase oxidation of SO 2 by NO 2 on PM 2.5 in China's atmosphere is an important SO 4 2− formation mechanism during haze periods in China. The Ge research group has reported (Wang et al., 2021b) that the SO 4 2− formation mechanism in eastern China in winter is mainly due to the rapid catalytic oxidation of SO 2 by manganese(Mn) ions on the surface of aerosols under low T and high RH conditions. The high ionic strength of the aerosol greatly enhances this reaction. The reaction rate is two to three orders of magnitude higher than that of the traditional aqueous reaction, which can explain 92.5% of the Nitrate in the atmosphere is mainly derived from the photochemical generation of NO 2 and OH· during the day and the heterogeneous hydrolysis of dinitrogen pentoxide (N 2 O 5 ) at night (Seinfeld, 2006 In addition, northeast China is the northernmost part of the country, and the winter is long and severely cold. The speci c emissions and meteorological conditions of the region will have an impact on secondary particulate matter formation. In addition, monitoring stations are mainly located in big cities, with few observations in small cities, suburbs, and rural areas. It is therefore di cult to fully understand the mechanism of regional haze formation from the distribution of pollution sources, atmospheric physics, and chemical processes. Therefore, this study established 14 sampling sites in three different cities, and their suburbs and surrounding rural areas in northeast China to simultaneously observe the atmospheric PM 10 and PM 2.5 mass concentrations, their chemical components, and the precursor gas concentrations. The characteristics and production mechanisms of secondary particulate matter (i.e., SO 4 2− and NO 3 − ) under low-sulfur emission and low-T conditions has led to differences in the haze pollution process compared to other regions of China. The changes in secondary particulate matter concentrations due to meteorological factors and precursor gases, and the differences with other regions of China were investigated. The important role of the increase in the secondary particulate matter concentration during the formation of regional persistent haze pollution was also examined.
Sampling And Analysis

Sampling and analysis
A total of 14 sampling sites were set up in urban, suburban, and rural locations in northeast China. Simultaneous observations of PM 10  A Dandong Smart Medium Flow Sampler was used for PM 2.5 sampling, with a ow rate of 28.3Lmin −1 . Whatman's quartz lm was used to collect particle samples (Whatman, UK). The sampling frequency was once a day for 24 h. The lter membrane was wrapped with tin foil before sampling and roasted in a mu e furnace at 550°C for 5 h to remove the residual organic matter and impurities. Before and after sampling, the quartz membrane was placed in a constant T and RH box (50% RH, 25°C) for 48 h, weighed with a 100,000 balance (accurate to 0.01 mg), and the mass of the collected particulate matter was determined by subtraction. Using the ow rate and duration of sampling, the PM 2.5 mass concentration was calculated. The sample was stored in tin foil in a refrigerator (-20°C) until analysis.
The chemical analysis included water-soluble ions(F − , Cl − , NO 3 − , SO 4 2− , NH 4 + , Na + , K + , Ca 2+ , and Mg 2 ) and OC and elemental carbon(EC). The detection limit of each ion was less than 0.003 mgL −1 . For the analysis, 1/4 of the lter membrane was placed in 20 ml ultrapure water (Millipore, USA, 18.2MΩ) and treated three times in an ultrasonic bath. The extract was measured with ion chromatography(ICS2500 and ICS2000, Dionex, USA). The OC and EC analysis was conducted with a thermal/optical carbon analyzer (Sunset Laboratory Inc., USA). The CLEANOVEN program was run prior to the analysis to remove possible interferences in the instrument. To detect the amount of particulate organic carbon (POC)generated, the sample was irradiated with a 633 nm laser to measure the change in re ected light intensity during heating. The return of the re ection intensity was de ned as the start point of the EC, and POC was isolated from EC1 to accurately determine the separation of OC and EC. Ultimately, the total OC of the particulate samples was de ned as OC=OC1+OC2+OC3+OC4+ POC, and EC was de ned as EC=EC1+EC2+EC3-POC.

Quality control and quality assurance
Quality control was performed in strict accordance with the relevant technical speci cations. Blank samples were analyzed, and each analytical instrument was calibrated with standard materials. The correlation coe cient of the water-soluble ion calibration curve was not less than 0.995, and the standard recovery rate was between 80% and 120%. The OC/EC ratio was manually calibrated at the beginning and end of the daily analysis. The deviation in the corrected peak area was within 3.0%, and was quantitatively calibrated with a 5% CH 4 /He internal standard to ensure that it was within the quality control range. The correlation between manual sampling and online air monitoring results was signi cant (P<0.05), and 14 instruments were compared in parallel before sampling to ensure that the results were accurate and comparable.    Figure 3). Harbin and its nearby areas had the highest OC and EC levels, the SO 4 2− and NO 3 − concentrations were highest in Shenyang, and the K + concentration was higher in rural areas than in urban and suburban areas. Table 2 The concentrations of PM 2.5 and its chemical components in three pollution episodes (µgm -3 ) Comparing the three pollution episodes shown in Table 2 Figure 5). The wind turned to the north on January 1, the wind speed increased, and the PM 2.5 concentration decreased rapidly, with the pollution episode ending on January 9. Whenever the wind direction turned to the south and the wind speed decreased, the PM 2.5 concentration gradually accumulated, but when the wind direction was northerly the wind speed increased and the concentration decreased rapidly (Figures 4 and 5). The trends of SO 2 and NO 2 were similar to those of PM 2.5 , indicating that emissions had a great impact on pollution. The SO 2 and NO 2 levels in EP1 were the highest of the entire three years. The boundary layer height (BLH) followed the opposite pattern to that of the pollution concentration ( Figure 4x). A low BLH results in poor diffusion of pollutants and high pollutant concentrations, which plays an important role in the rise and fall of the pollutant concentration ( Figure 4v). The effects of air T and RH were also signi cant, and pollution in EP1 increased as the T increased ( Figure 4p). However, the T on January 2-8 was higher than on the most polluted day December 31-January 1. It was apparent that T did not play a decisive role in pollution episodes. When the RH was high, the pollution was heaviest (Figure 4w). A high RH was very important for pollution formation.
The OC/EC ratio is used to characterize carbon aerosol emissions and conversion characteristics. Different ratios represent different sources of and the pollutant concentration will then increase rapidly (see section 3.3 for details). Therefore, it was concluded that EP1 was a composite pollution process formed by local emissions from coal combustion, secondary conversion under a low T and high RH, and regional transmission.
EP2 was expected to be dominated by straw burning pollutants due to the large-scale straw burning that occurred during the sampling process. The chemical component with the highest concentration was OC (Figure 4g),with a peak concentration of 180µgm −3 at HST. The concentrations at the surrounding BL, BX, and SJ sites were also very high, indicating that straw burning made a signi cant contribution to OC. The most representative ion for straw burning is K + and its concentration was very high, with only EP4 having a higher concentration (Figure 4m). The K + concentrations at HST, SJ, SP, BX, BL, and ZD were 5-7 times higher than the typical atmospheric concentration in the main straw burning area. The different sampling sites that were affected by straw burning displayed different characteristics. At HST and BX, OC and K + increased signi cantly at the same time, while the K + concentration at SP increased more signi cantly than the OC concentration. There was a signi cant increase in OC at BL, whereas K + did not increase signi cantly. These results may be related to the different local emission sources, with HST receiving a large contribution from straw combustion and motor vehicle emissions, SP being impacted by straw combustion, and BL being impacted by residential coal burning in the suburbs.
EP3 was a cross-regional transmission process. Figure 6a clearly shows the gradual reduction of PM 2.5 and other pollutants as they were transported from southwest to northeast. The jump point of the PM 2.5 concentration changed from SY→BH→SJ→HST (Figure 6b), which indicated that EP3 was clearly dominated by a secondary transmission process. The SO 4 2− /EC and NO 3 − /EC ratios were the highest among the four pollution episodes, and the peak values of the four sampling sites were in the range of 6.7-18.9 (Figure 4p, q). This further con rms that the event was a regional transfer process.
EP4 was the pollution episode with the highest PM 2.5 concentration and the heaviest pollution during the entire observation period. Its main features were a high PM 2.5 and O 3 concentration, and a high mid-winter T. The SO 2 and NO 2 concentrations were lower than those of EP1 and EP3. The period was characterized by low emissions, and therefore the high levels of pollution were mainly a consequence of the strong oxidation conditions and high winter T.
There were three pollution peaks in EP4 (Figure 4), with high concentrations of the main chemical components (OC and SNA). The rst peak was on February 15. The PM 10 concentration was highest at BX and YS, which were the representative rural sampling sites, and it was much higher than the PM 2.5 concentration (Figure 4) . The YS site also had high concentrations of chemical components. These sites were located in rural areas, with few motor vehicles, and the high concentrations were likely caused by high emissions of loose coal and biomass combustion.
The second peak occurred on February 21. The wind direction at each station was southerly. The PM 2.5 concentration at BX, HST, and SJ was high (Figure 4b), the weather conditions were stable, and the wind speed was low to calm(i.e., less than 2m/s) ( Figure 5). The T continued to increase and the RH was conducive (about 50%)to the conversion to secondary pollutants. The high concentration at HST was related to the high SO 2 , NO 2 , and CO emissions on the previous day. On February 20th in BX, the NO 3 − and SO 4 2− concentrations were high, the OC concentration at SJ was high, and the EC concentration at BX and SJ was high. These results A large amount of secondary generation of SNA. The SNA concentration was only higher in EP3 due to regional transmission. However, the SO 2 concentration was the lowest among the four pollution episodes, and the NO 2 concentration was the same as in EP1. Due to the low emission levels, the high pollution load was created by high secondary pollutant concentrations. There were two main reasons for this. First, under a relatively high T (0-5°C) and RH (40-60%), which was not high enough to limit the reaction, the gas emitted by local coal and motor vehicles generated a high SNA concentration through homogeneous and heterogeneous phase reactions (see section 3.3 for details).
Similarly, SJ, HST, DQ, ZD, and SY had high PM 2.5 concentrations, which may be related to their higher local emissions. The OC, EC, and Cl − concentrations at SJ and ZD were high on February 26-28, the OC and EC concentrations at DQ and HST were high on February 28, and the SNA concentration at SY had been continuously increasing since February 27. This indicates that the peaks were mainly the local secondary conversion of primary emissions. At this time, the SO 4 2− concentration was high (Figure 4k, peak value 6.1-20.4µgm −3 ), but still lower than the NO 3 − concentration.
The reason why the NO 3 − concentration was higher than the SO 4 2− concentration was related to the low SO 2 emission concentration, which limited SO 4 2− formation. It may also be related to the high O 3 concentration and the strong gas oxidation was conducive to the gas-particle conversion of The superimposed effect of cross-regional transmission has caused the rapid growth of secondary pollutants. The SO 4 2− /EC and NO 3 − /EC ratios in Figure 4q and r are very high, indicating that secondary formation occurs through the superimposition of regional transport and secondary reactions. The OC/EC ratio was not high (Figure4s, between 1.7 and 8.4, with an average of 4.9), indicating that it was mainly controlled by emissions from coal combustion and motor vehicle exhaust emissions. It can be seen from the graphs of the PM 2.5 concentration and wind eld in Figure 7 that the most polluted area was near Harbin (HST). At this time, the wind direction and wind speed of each station changed from northerly to southerly, and the wind speed decreased rapidly. This shows that the atmosphere was in a static and stable state, and the static and stable meteorological conditions were also conducive to the accumulation of secondary particulate matter. In addition, the wind eld at the high-concentration center of the BL site had an obvious counterclockwise convergence, and the wind speed was higher than that in the surrounding areas, which accelerated the accumulation of pollutants at the site.
It can be seen that EP4 was a compound pollution process. The Cl − and OC concentrations were the highest among the four episodes, indicating that coal combustion made a substantial contribution, and the K + and Mg 2+ concentrations were also the highest among the four episodes. Biomass combustion and industrial emissions also made a large contribution. Except for the high Ca 2+ concentrationatthe BX and ZD sites on February 21st, the levels at the other stations were low. The main source of Ca 2+ is usually sand and dust emissions (Xu et al., 2017b), indicating that the impact of sand and dust was small in EP4. Therefore, the pollution formation during this episode occurred through the homogeneous and heterogeneous reaction of pollutants emitted by local coal combustion, biomass combustion (K + ), and industrial emissions (Mg 2+ ) under a moderate T and strong gas oxidizing conditions (high O 3 ). It was formed by a signi cant increase in the secondary pollutant concentrations, such as SOC and SNA, and then superimposed on a composite pollution process controlled by regional transmission. In EP1 and EP4, the slopes were 1.1 and 1.14, and the intercepts were 0.09 and 0.12, respectively, indicating that some of the NO 3 − +Cl − was not completely neutralized by NH 4 + . In Figure 4n, it can be seen that the Ca 2+ concentration in EP1 was higher than in the other pollution episodes, and the contribution of sand dust was also greater. The remaining NO 3 − and Cl − may exist in the form of calcium nitrate (Ca(NO 3 ) 2 ) and calcium chloride (CaCl 2 ). The lowest slope of EP2 was 0.68, indicating that after all the NO 3 − +Cl − was neutralized, there was still a large amount of NH 4 + . The main feature of EP2 was that straw burning produced a large amount of NH 3 , and therefore the NH 4 + concentration was also high. The slope of EP3 was 1.04, indicating that NH 4 + was present in excess and Cl-emissions from coal combustion were not high. The K + concentration in EP4 was high. The contribution of biomass combustion was also high, which may have enabled potassium to exist in the form of potassium nitrate (KNO 3 ) and potassium chloride (KCl).
Nitrate in the EP1 process was formed under the conditions of a medium T and high RH. At this time, the warming period in winter was occurring ( Figure 8d, 5v), and the RH was also high when NO 3 − peaked (Figure 8c, 5w). This shows that NO 3 − formation occurred when the T was mild in winter and the RH was high. A high T was conducive to the homogeneous gas-phase reaction of NO 2 reacting with OH to produce gaseous HNO 3 (Harris et al., 2013). A T not higher than 4°C is also conducive to NO 3 − formation. The reaction therefore proceeded in the positive direction .
The subsequent reaction between gaseous HNO 3 and NH 3 needs to occur at a higher RH, because although NO 3 − is mainly formed through a homogeneous reaction in the gas-phase, a high RH is conducive to the accumulation of NO 3 − in the particles (Shi et al., 2014). Therefore, a high RH promotes NO 3 − formation.
The EP2 process was clearly impacted by straw burning emissions, with rapid NO 3 − formation in an NH 3 -rich environment. Straw burning in the vicinity of several cities, coupled with southerly winds, was conducive to the spread of pollution from south to north. In particular, the wind speeds at several sampling sites were high on December 15 ( Figure 5, 4-5ms −1 ), which caused pollution to rapidly spread to the area. On the December16 the wind speed decreased rapidly, and the calm winds were conducive to the accumulation of pollutants. At this time, at SP, YS, and other locations, the RH was high (40-60%, Figures 4w and 8c), and the T was low (-15 to -11°C, Figure 4v), and therefore straw burning pollutants accumulate in wet weather conditions to produce haze. There are records of haze in various locations.
In the EP3 pollution process, the NO 3 − concentration was the highest among the four pollution episodes (Figure 4j, Figure 9a). It was formed by the secondary pollutants in the high RH and medium T environment during the transmission process, and the RH has a great in uence on it. The RH was also the highest of the four periods at around 50-80% (Figures 9a, c and Figure 4w), and the T was centered at -10-0°C (Figure 4v).
The NO 3 − formation in the EP4 process occurred under a relatively high T and moderate RH in winter. Only EP3 had a higher NO 3 − concentration, with the NO 3 − concentration in the third peak period being especially high (Figures 4j, 9a). The RH was only between 40-60% (Figures 4w, and 8a, c).
Although the RH was not high, the deliquescent RH of the particulate matter was still higher than the weathered RH, which caused it to continue to deliquesce without being weathered, and a high water content was maintained in the particulate matter. In conjunction with the high winter T and strong atmospheric oxidability, serious haze pollution could still occur. The pollution concentration sometimes exceeded the concentration recorded when the RH was higher and the T was lower (such as in EP1). It can be seen that the in uence of RH on NO 3 − in northeast China was not as obvious as its in uence on SO 4  Similarly, during EP4, a high NH 3 concentration was observed, which reacted with HNO 3 (Figure 9b) and promoted massive NO 3 − aerosol formation.
Due to the use of low-sulfur coal and the widespread use of desulfurized rubber as pollution control measures, SO 2 concentrations have decreased, but

The relationship between the SO 4 2concentration and water vapor pressure
To characterize the relationship between ion concentrations and RH, the relationship between ambient water vapor pressure(e*) and the SO 4 2− concentration was determined for different pollution periods, under different RH, T, and sulfur oxidation rate (SOR) conditions, and with different precursor gas concentrations (mainly SO 2 and NO 2 ). The ambient water vapor pressure was calculated using the formula e*=e×RH/100 (e refers to the saturated water vapor pressure, where e=6.112exp (17.67t/(t+243.5), t refers to the T (°C)). The SOR refers to the molar ratio of sulfur in SO 4 2− to the total sulfur, and is calculated with the formula SOR= SO 4 2− /( SO 4 2− +SO 2 ). Figure 10 shows the changes in the SO 4 2− concentration with e*, with air temperature, different pollution periods, RH, and the precursor gases shown in different colors.
The trend of the tted line in Figure 10a shows that SO 4 2− increased with an increase in e*, but the increase was not very obvious, indicating that SO 4 2− formation was complicated. Taking e*=1.5 and the SO 4 2− concentration=10µgm −3 as boundaries, the gure could be divided into four areas. Figure 11 shows the statistical values of various pollutants, T, and RH in the four areas.
Area . This feature mainly occurred in EP4 and EP3 (Figure 10a, b). This area of the gure had the highest SO 4 2− concentration (14.7µgm −3 ), the highest T and O 3 (3.7°C and 71.6µgm −3 , Figure 10a, Figure 11), coupled with high HONO and NH 3 concentrations (Figure 9b). The degree of oxidation of NO 2 and SO 2 was enhanced, and the SOR (0.3, Figure 10d), SO 4 2− (14.7µgm −3 ), and NO 3 − (22.3µgm −3 ) increased the most. Although the SO 2 concentration was low, the NO 2 concentration was high, and the prevailing conditions still supported SO 4 2− formation. At the same time, the RH was not high (Figure 10d, Figure 11, 63.8%), which affected the aqueous phase reaction of SO 4 2− to a certain extent. In area , the correlation between SOR and O 3 was highest (0.473, Table 3), and the correlation with T (0.275) was higher than the correlation with RH (0.110), In area , the correlation between SOR and O 3 was highest (0.473 , Table 3), and the correlation with T (0.275) was higher than with RH (0.110), indicating that SO 4 2− was mainly transformed by the SO 2 gas-phase instead of aqueous oxidation. Therefore, the SO 4 2− formation in area was mainly the homogeneous gas-phase reaction of the OH· oxidation of NO 2 and SO 2 . The source of OH· was the photolysis of O 3 and HONO, which plays a key role in atmospheric chemistry . The oxidation rate of OH· is usually several times or even hundreds of times higher than in other pathways (Tang et al., 2006 concentrations. The SO 4 2− concentration in area was slightly lower than that of area , and these conditions mainly occurred in EP1. The conditions that favored low O 3 , and high RH) during Beijing haze, and speculated that NO 2 may be an important oxidant for ne aerosols under conditions with high RH and NH 3 neutralization. A high pH value will draw more SO 2 into the aerosol droplets, thereby increasing the rate of SO 4 2− formation in the NO 2 reaction pathway. It is more likely that NO 2 will oxidize SO 2 in the aqueous phase on ne aerosols with a high RH (>60-70%) and su cient neutralization (pH~7) (Wang et al., 2016a). Therefore, it was speculated that the increase in SO 4 2− production under high NO 2 and SO 2 concentrations, high RH, and low O 3 levels in northeast China haze events may have a speci c mechanism. Under neutral or weakly alkaline conditions, the high pH will pull more SO 2 into the aerosol water, and the high RH will also make it easier for NO 2 to react with SO 2 droplets in an aqueous oxidation to increase the SO 4 2− production rate. This is another important reaction pathway for the production of high SO 4 2− concentrations in northeast China in addition to the gasphase transformation under the in uence of the photochemical activity in area . However, due to the unique low-T environment and low SO 2 emissions in northeast China in winter, the aqueous phase reaction of SO 4 2− in area was actually substantially weakened, and the gas-phase SO 4 2− formation in area , with its high atmospheric oxidability, was greater than in area . in Beijing-Tianjin-Hebei and the entire North China region in autumn and winter. They also found and con rmed that the aqueous oxidation of NO 2 on atmospheric ne particles is an important SO 4 2− formation mechanism during haze episodes in China. The gas-phase oxidation process was not important. However, in northeast China, due to the inhibitory effect of the low T in winter on the SO 4 2− aqueous reaction, there is rapid SO 4 2− and NO 3 − production, and the rapid gas-phase oxidation process plays an important role. This is the main difference between northeast China and other regions.
However, the aqueous reaction is still the most common way of producing SO 4 2− in northeast China.
The SO 4 2− concentrations in areas and were similar. The largest advantage of area in terms of SO 4 2− formation was the high RH, but there were low SO 2 and NO 2 concentrations and a low T, which inhibited the heterogeneous oxidation of NO 2 and SO 2 to a certain extent, resulting in a low SO 4 2− concentration. The biggest advantage of area in terms of SO 4 2− formation was the high T in winter, but the low SO 2 and NO 2 concentrations and slightly lower O 3 concentration than in area resulted in the degree of homogeneous oxidation of NO 2 and SO 2 in the gas-phase being less than that of area . In addition, the low RH affected the aqueous phase oxidation process causing a low SOR. Therefore, the SO 4 2− concentration was not high.

The in uence of T and RH on the concentration of important chemical components
(1) Changes with T: The NO 3 − concentration increased with increasing T (Figure 12). Higher Ts are conducive to NO 3 − formation in the gas-phase, but Ts below 4°C will not support volatilization and therefore will affect the gas particle distribution in the NH 3 +HNO 3 ↔NH 4 NO 3 reaction. The formation of SO 4 2− was similar to that OC and followed a V-shaped pattern. Higher Ts in winter were bene cial to SO 4 2− formation because with high Ts the solar radiation is strong, the rate of photochemical oxidation is rapid, and the OH· concentration increases, which in turn strengthens the oxidation of SO 2 . Under a high RH, the water content of aerosols is relatively high, and O 3 is dissolved in aerosol water where it reacts with dissolved SO 2 to promote the liquid-phase oxidation process . Heating emissions increase as the T falls, and therefore large amount of SO 2 are generated. The OC increases with an increase in T, with a higher T meaning that the photochemical reaction was strong, which was conducive to OC formation ).
(2) Changes with RH: The highest OC, NO 3 − , and SO 4 2− concentrations were observed when RH was in the range of 80-90% (Figure 12). Under a high RH, many atmospheric heterogeneous reactions are accelerated, and the reaction products are conducive to water absorption and the deliquescence of particulate matter. They will therefore have a signi cant role in promoting haze formation. The coexistence of pollutant particles and fog forms a positive feedback mechanism, which will continuously promote the conversion of the gaseous pollutants discharged from the primary to secondary aerosol. This will result in their concentration continuing to increase. The SO 4 2− concentration in this study increased with an increase in RH. This shows the importance of the heterogeneous reactions (liquid phase reactions) under a high RH in SO 4 2− formation.
Unlike OC and SO 4 2− , more than 70-60% of the NO 3 − was produced when the RH was 50-30%. The phenomenon in which the NO 3 − concentration produced by the EP4 process (low RH) was higher than that of produced during the EP1 process (high RH) was explained in section 3.  Nitrate formation was mainly due to the homogeneous gas-phase reactions of OH· with NO 2 , SO 2 , and NH 3 . The highest concentration occurred under mild winter Ts, high RH, high atmospheric oxidability (O 3 and HONO), high NH 3 concentrations, and suitable light conditions.
The SO 4 2− concentration tended to increase with an increase in vapor pressure (e*). There were two main SO 4 2− formation pathways. One was the gas-phase reaction of the OH· oxidation of NO 2 and SO 2 . High O 3 concentrations, higher winter Ts, and high atmospheric oxidability and photochemical activity were very important for SO 4 2− formationduring haze events. The second pathway operated under neutral or weakly alkaline conditions, during which SO 2 enters aerosol droplets, and NO 2 is more likely to react in the aqueous phase with SO 2 to increase the output of SO 4 2− .
The difference between the SO 4 2− formationin northeast China and other regions was re ected by the suppression of the aqueous reaction of SO 4 2− due to low Ts in winter and low-sulfur coal emissions in northeast China. This resulted in the rapid generation of gas-phase oxidation processes being important for high levels of SO 4 2− production in the region. However, the aqueous oxidation reaction was still the most common SO 4   The average concentrations of PM2.5and its chemical components at each sampling site (Note: There are three columns for each sampling site, representing 2017, 2018, and 2019 respectively).

Figure 3
The concentrations of PM2.5 and its chemical components (OC, SO42-, and NO3-: µgm-3)  The NO3-and SO42-concentrations and their precursor gases at the SY site Figure 10 The relationships between SO42-and the sulfur oxidation rate (SOR), ambient water vapor pressure(e*), relative humidity (RH), and temperature (T)