Characteristics of Particulate Matter at Different Pollution Levels in Chengdu, Southwest of China

: Air pollution is becoming increasingly serious along with social and economic development in the southwest of China. The distribution characteristics of particle matter (PM) were studied in Chengdu from 2016 to 2017, and the changes of PM bearing water-soluble ions and heavy metals and the distribution of secondary ions were analyzed during the haze episode. The results showed that at different pollution levels, heavy metals were more likely to be enriched in ﬁne particles and may be used as a tracer of primary pollution sources. The water-soluble ions in PM 2.5 were mainly Sulfate-Nitrate-Ammonium (SNA) accounting for 43.02%, 24.23%, 23.50%, respectively. SO 42 − , NO 3 − , NH 4+ in PM 10 accounted for 34.56%, 27.43%, 19.18%, respectively. It was mainly SO 42 − in PM at Clean levels (PM 2.5 = 0~75 µ g/m 3 , PM 10 = 0~150 µ g/m 3 ), and mainly NH 4+ and NO 3 − at Light-Medium levels (PM 2.5 = 75~150 µ g/m 3 , PM 10 = 150~350 µ g/m 3 ). At Heavy levels (PM 2.5 = 150~250 µ g/m 3 , PM 10 = 350~420 µ g/m 3 ), it is mainly SO 42 − in PM 2.5 , and mainly NH 4+ and NO 3 − in PM 10 . The contribution of mobile sources to the formation of haze in the study area was signiﬁcant. SNA had signiﬁcant contributions to the PM during the haze episode, and more attention should be paid to them in order to improve air quality. order of ion content, PM 42 Ca 2+ K + Cl − Na + Mg 2+ F − , while SO 42 − , NO 3 − , NH 4+ accounted for 34.56%, 27.43%, 19.18% of the total ion content, respectively. The results showed that the secondary ions (SO 42 − , NO 3 − , NH 4+ ) were the main ions in Chengdu atmospheric particles.


Introduction
PM 10 refers to particles with an aerodynamic equivalent diameter less than or equal to 10 µm in ambient air, and PM 2.5 refers to particles with an aerodynamic equivalent diameter less than or equal to 2.5 µm in ambient air. The composition of atmospheric particles is complex, including heavy metals, water-soluble ions, carbonaceous components, and so on from multiple sources [1,2]. In addition, with small particle sizes and large surface area, atmospheric particulates have adverse effects on the atmospheric environment and public health. In recent years, there have been many haze pollution incidents occurring in a number of locations across China, which has caused more and more public concerns and attention paid to air pollution [3][4][5].
The Sichuan Basin is in southwestern China. The topography of hills and basins, coupled with the climate conditions of high humidity and low wind speed, leads to atmospheric pollution easily in this area [6,7]. It is the fourth highest haze area following the Beijing-Tianjin-Hebei area, Yangtze River Delta, and Pearl River Delta. Its pollution characteristics are of high particle concentration and low visibility [8][9][10]. The special terrain and the humid climate of Chengdu are not conducive to the diffusion of particulate matter and are prone to secondary pollutant (SNA) conversion and generation [6, 10,11].
In recent years, although the implementation of pollution prevention and control measures has improved the air environment in Chengdu, the region still has a problem with air pollution [12]. The research topics in this region mainly include the analysis of particulate pollution characteristics [13,14], the impact of meteorological conditions on particulate pollution [15], the lag effect of particulate pollution on related diseases [16], and source apportionment [17]. It is reported that the heavy metals in the atmospheric particulate matter (PM) in Chengdu are mainly arsenic (As), lead (Pb), copper (Cu), nickel (Ni), zinc (Zn), iron (Fe), and manganese (Mn) [18,19]. Among them, arsenic (As) mainly comes from industrial smelters. Pb, Cu, Ni, and Zn mainly come from the exhaust of motor vehicles and the wear of tires and brake pads whereas Fe and Mn are mainly from dust generated during vehicle driving [20,21]. Sulfate-Nitrate-Ammonium (SNA) are water-soluble ions that greatly contribute to PM concentration [22,23]. Air pollution in Chengdu has obvious seasonal distribution characteristics, which are closely related to the meteorological factors of the city [24]. It is reported that the mass concentration of SNA is the highest in winter and the lowest in summer [25]. The high temperature in summer and autumn is conducive to the conversion of sulfur dioxide (SO 2 ) to sulfate (SO 4 2− ) while adverse to the stable existence of ammonium nitrate (NH 4 NO 3 ) . Although the low temperature in winter inhibits the conversion of gaseous precursors, it is beneficial to the stable existence of NH 4 NO 3 [26,27].
However, it is known from the previous studies that the concentration of particulate matter increases with the increase of pollution, but the mechanism of particle concentration and composition change is different under different pollution levels [28]. A recent study showed that the rapid increase of PM 2.5 at light pollution level in Beijing was caused by regional transportation, while the rise from heavy to severe was mainly caused by an increase in the proportion of secondary inorganic components [29]. The air pollution in cities in southern China has been easily overlooked. Up to now, there is no detailed report on the various characteristics of atmospheric particulate matter at various pollution levels in Chengdu, southwest China according to our investigation.
The purpose of this study is to find out how heavy metals and water-soluble ions in PM in Chengdu, China during the haze periods are distributed and changed at different pollution levels. Therefore, we investigated PM 2.5 and PM 10 in Chengdu in southwest China. The changes in heavy metal elements and water-soluble ions corresponding to the pollution level and their contribution to particulate matter are discussed. The effects of SO 4 2− , NO 3 − , and NH 4 + on the particulate matter were emphatically explored. The secondary production of sulfate and nitrate will be shown to be important in high pollution level scenarios, and the same with the heavy metal analysis.

Study Site and Sample Collection
Chengdu is located in the western part of the Sichuan basin, surrounded by the western part of the Longquan Mountains and the eastern part of the Qionglai Mountains. The sampling site was located in Shilidian, Chenghua District, Chengdu (104 • 08 E, 30 • 40 N), the capital of Sichuan Province in the western part of the Sichuan Basin. Chengdu is densely populated, about 1000 people/km 2 [30,31], with the annual temperature 15.2~16.6 • C, the annual precipitation 873 mm~1265 mm, the annual sunshine 23-30%, the average annual wind speed 1.3 m/s, and the average annual relative humidity 80% [32]. Shilidian is surrounded by major cities in Sichuan, including Deyang and Mianyang ( Figure 1).  From March 2016 to January 2017, PM2.5 and PM10 were collected at the Chengdu University of Technology in the urban area of Chengdu with no chemical enterprises and tall buildings. The sampling period was 24 h and we collected 72 PM2.5 samples and 72 PM10 samples at the same time, with a total of 144 samples. The sampling instrument was a TH-150C medium flow atmospheric sampler (Wuhan Tianhong, Wuhan, China), with a calibrated flow rate of 100 L/min. Two kinds of filter membranes made of quartz and Teflon, respectively, (Whatman, Buckinghamshire, UK) were chosen. The Teflon filters are used for the heavy metal analysis because Teflon filters have low heavy metal background content, and the quartz filters are used for the water-soluble ion analysis. After the samples were collected, the sampling membranes were placed in clearly marked sample boxes immediately. At the same time, the meteorological data at the Shilidian meteorological monitoring station were recorded, including temperature, air pressure, wind speed, relative humidity, etc. Samples were collected under stable weather conditions, with weak wind at speeds less than 1.5 m/s, thus the contribution from pollutants transported long distances are likely small. The samples in this study mainly represent the local atmospheric conditions in Chengdu.

Mass Concentration Analysis
Before sampling, the Teflon filter membrane (Whatman, Ф90 mm, Buckinghamshire, UK) is equilibrated for at least 24 h at a temperature of 20 ± 5 °C and a relative humidity of 50±5%. Quartz filter membranes (Whatman, Ф90 mm) were wrapped in aluminum foil, baked in a muffle furnace (SX-8-13, Beijing) at 500 °C for 4 h to remove the background organic matter, and then placed in the same environment as the Teflon filters for at least 24 h. After the filter membrane, use a one-hundred thousandth balance (Sartorius, Göttingen, Germany, CPA225D) was used to weigh each filter 3 times to ensure that the difference between any two weighing values did not exceed 0.04 mg. After the filter membranes were weighed, they were all wrapped in aluminum foil, and put in a sealed bag, and stored at −4 °C until analyzed. A pretreated blank filter membrane was used as a background. Before sample collection, the cutting head of the sampler filter membrane grid, sealing gasket, and other places that may be in contact with the filter membrane were wiped two to three times with high-grade pure absolute ethanol to prevent impurities from entering the filter membrane during the sampling process. Refer to "Ambient Air PM10 and PM2.5 Measurement-Gravimetric Method" (HJ618-2011) for details on the method used to calculate the mass concentrations of PM2.5 and PM10.

Heavy Metals Analysis
The concentrations of heavy metals of the samples were then analyzed. Before the experiment, all Teflon vials were thoroughly cleaned with 20% hot nitric acid solution (70 °C) and deionized water to avoid contamination. Subsequently, 1/2 of a Teflon filter was dissolved with 1 mL of nitric acid (HNO3) and 1 mL hydrofluoric acid (HF) in a From March 2016 to January 2017, PM 2.5 and PM 10 were collected at the Chengdu University of Technology in the urban area of Chengdu with no chemical enterprises and tall buildings. The sampling period was 24 h and we collected 72 PM 2.5 samples and 72 PM 10 samples at the same time, with a total of 144 samples. The sampling instrument was a TH-150C medium flow atmospheric sampler (Wuhan Tianhong, Wuhan, China), with a calibrated flow rate of 100 L/min. Two kinds of filter membranes made of quartz and Teflon, respectively, (Whatman, Buckinghamshire, UK) were chosen. The Teflon filters are used for the heavy metal analysis because Teflon filters have low heavy metal background content, and the quartz filters are used for the water-soluble ion analysis. After the samples were collected, the sampling membranes were placed in clearly marked sample boxes immediately. At the same time, the meteorological data at the Shilidian meteorological monitoring station were recorded, including temperature, air pressure, wind speed, relative humidity, etc. Samples were collected under stable weather conditions, with weak wind at speeds less than 1.5 m/s, thus the contribution from pollutants transported long distances are likely small. The samples in this study mainly represent the local atmospheric conditions in Chengdu.

Mass Concentration Analysis
Before sampling, the Teflon filter membrane (Whatman, Φ90 mm, Buckinghamshire, UK) is equilibrated for at least 24 h at a temperature of 20 ± 5 • C and a relative humidity of 50 ± 5%. Quartz filter membranes (Whatman, Φ90 mm) were wrapped in aluminum foil, baked in a muffle furnace (SX-8-13, Beijing) at 500 • C for 4 h to remove the background organic matter, and then placed in the same environment as the Teflon filters for at least 24 h. After the filter membrane, use a one-hundred thousandth balance (Sartorius, Göttingen, Germany, CPA225D) was used to weigh each filter 3 times to ensure that the difference between any two weighing values did not exceed 0.04 mg. After the filter membranes were weighed, they were all wrapped in aluminum foil, and put in a sealed bag, and stored at −4 • C until analyzed. A pretreated blank filter membrane was used as a background. Before sample collection, the cutting head of the sampler filter membrane grid, sealing gasket, and other places that may be in contact with the filter membrane were wiped two to three times with high-grade pure absolute ethanol to prevent impurities from entering the filter membrane during the sampling process. Refer to "Ambient Air PM 10 and PM 2.5 Measurement-Gravimetric Method" (HJ618-2011) for details on the method used to calculate the mass concentrations of PM 2.5 and PM 10 .

Heavy Metals Analysis
The concentrations of heavy metals of the samples were then analyzed. Before the experiment, all Teflon vials were thoroughly cleaned with 20% hot nitric acid solution (70 • C) and deionized water to avoid contamination. Subsequently, 1/2 of a Teflon filter was dissolved with 1 mL of nitric acid (HNO 3 ) and 1 mL hydrofluoric acid (HF) in a closed-cap Teflon vial for 48 h at 180 • C. After that, the mixed solution was steamed to near dry, and then re-dissolved twice with 1 mL HNO 3 (120 • C). After the last re-dissolution, HNO 3 (1 mL), Rh solution (1 mL of 1000 ng/mL), and 5 mL deionized water were added and kept in Teflon vials for 6 h (100 • C). At this point, the sample pre-treatment was completed. The concentrations of the heavy metals were analyzed by inductively coupled plasma-mass spectrometry (ICP-MS, Perkin Elmer Corp., Norwalk, USA). The reference material GSS-4 was used to ensure the analytical accuracy with recovery between 94.3% and 103.6%. In addition, for 10% of the samples analysis was repeated and reagent blanks were also used to check the quality of the analysis. A total of eight metal elements were measured, including arsenic (As), cadmium (Cd), chromium (Cr), copper (Cu), nickel (Ni), lead (Pb), vanadium (V), and zinc (Zn). Their detection limits are: As (0.

SOR and NOR Analysis
The concentrations of sulfate, nitrate, and ammonium are related to the concentration of gaseous precursors: sulfur dioxide (SO 2 ), nitrogen oxides (NO x ), and ammonia (NH 3 ), and their conversion rates to particles generated in the atmosphere. Here SOR (sulfur oxidation rate) and NOR (nitrogen oxidation rate) are used to describe the formation of secondary aerosol species. The measured values of SO 2 and NO 2 come from the Chengdu Shilidian permanent monitoring site. Based on Ma et al. [29], the calculation formulas of SOR and NOR are: where n is the molar concentration of the species. When SOR > 0.1, it indicates that there is a process of SO 2 oxidation to SO 4 2− in the particles. When NOR > 0.001, it is said that there is a process of oxidation of NO 2 to NO 3 − in the particulate matter. The higher value of SOR or NOR, the higher the oxidation rate of the pollutant [33].

PM Mass Concentration
Thirty samples, eight samples, twenty-one samples, and thirteen samples were analyzed in spring, summer, autumn, and winter, respectively. The concentrations of PM 2.5 and PM 10 showed obvious seasonal distribution characteristics. The changes of PM 2.5 concentration with seasons (spring to winter) were: 98.62 µg/m 3 , 66.75 µg/m 3 , 84.02 µg/m 3 , and 159.74 µg/m 3 . PM 10 concentration changes with seasons (spring to winter) were: 169.87 µg/m 3 , 107.22 µg/m 3 , 167.16 µg/m 3 , and 260.30 µg/m 3 . The concentrations of PM 2.5 and PM 10 were the highest in winter and the lowest in summer. An inversion easily forms in winter, which prevents particles from diffusing. While the movement of atmospheric molecules and the atmospheric oxidation capacity is enhanced because of the high temperature in summer, which is conducive to the diffusion of atmospheric particles.
According  Figure 3a showed that there were 32 samples at the clean levels, 55 samples at the Light-Medium levels, and 12 samples at the severe levels. It is worth noting that the pollution level based on PM 2.5 did reach the Severe levels on 3 January 2017.
The concentrations of PM 2.5 that increase with the change in pollution levels were on average 48.80 µg/m 3 , 109.84 µg/m 3 , and 186.21 µg/m 3 for Clean, Light-medium, and Heavy levels, respectively. The concentrations of PM 10 increase with the change in pollution levels were 98.49 µg/m 3 , 226.53 µg/m 3 , and 383.21 µg/m 3 for Clean, Lightmedium, and Heavy levels, respectively (Figure 3b). From the perspective of the increase in particle concentration, the growth rate of PM 10 is faster than the growth rate of PM 2.5 , indicating that coarse particles (PM 2.5-10 ) have a certain contribution to the growth of PM 10 . PM 2.5 /PM 10 from Clean to Heavy pollution decreases first and then increases slightly, indicating PM 2.5 contributed the most to PM 10 at Clean levels and the least to PM 10 at Severe levels.

PM Mass Concentration
Thirty samples, eight samples, twenty-one samples, and thirteen samples were analyzed in spring, summer, autumn, and winter, respectively. The concentrations of PM2.5 and PM10 showed obvious seasonal distribution characteristics. The changes of PM2.   There is a correlation between the concentration of PM and related climatic conditions ( Table 1). The concentration of PM is significantly negatively correlated with wind speed, temperature, and ozone, and significantly positively correlated with relative humidity, atmospheric pressure, CO, NO2, and SO2. The correlation for PM2.5 and PM10 with temperature, CO, and NO2 are similar. PM2.5 has a stronger correlation with relative humidity, atmospheric pressure, ozone, and SO2, while PM10 has a stronger correlation with wind speed. This shows that the influence of meteorological conditions on fine particles is greater.

Heavy Metals Characteristics and the Potential Use
The content of heavy metals in PM at different pollution levels is shown in Figure 4. The content of heavy metals varies greatly at different levels of pollution (average values are shown in Tables S1 and S2). At each pollution level, the heavy metal content in PM10 was significantly higher than that of PM2.5. With the increase of pollution level, the total amount of heavy metals in the particles gradually increased, but the degree of increase gradually decreased. In PM2.5 and PM10, the order of heavy metal content at each pollution level was Zn > Cu> Pb > Cr > As > Ni > V > Cd. It is reported that Zn, Cu, Cr, Pb mainly come from exhaust emissions of motor vehicles or the wear of brake pads and tires [34][35][36], and Pb, As, Ni come from coal and petroleum combustion [13,37]. Cd is related to industrial processes [13,38], and V may come from mining or soil fertilizer use [39]. Lead, zinc, and copper account for a relatively high proportion, which is related to automobile exhaust. Urban traffic jams are becoming more and more serious, leading to frequent braking and start-up of vehicles, which aggravates the emission of heavy metals in the exhaust gas. Beijing is the city with the largest number of cars in China, and car exhaust has been studied in Beijing as a factor [40,41]. Chengdu is the second-largest city in the country for car ownership, so the contribution of car exhaust to Chengdu's atmospheric particulate matter is also significant [42].
The relative percentage content of all heavy metals is almost constant at each pollution level. The content of heavy metals (HM) per particle at different pollution levels is shown in Figure S1 (Supplementary Information). It can be seen that the heavy metals per particle changes with the increase of particle concentration. The heavy metals content per particle in PM2.5 is always higher than that for PM10. There is a correlation between the concentration of PM and related climatic conditions ( Table 1). The concentration of PM is significantly negatively correlated with wind speed, temperature, and ozone, and significantly positively correlated with relative humidity, atmospheric pressure, CO, NO 2 , and SO 2 . The correlation for PM 2.5 and PM 10 with temperature, CO, and NO 2 are similar. PM 2.5 has a stronger correlation with relative humidity, atmospheric pressure, ozone, and SO 2 , while PM 10 has a stronger correlation with wind speed. This shows that the influence of meteorological conditions on fine particles is greater.

Heavy Metals Characteristics and the Potential Use
The content of heavy metals in PM at different pollution levels is shown in Figure 4. The content of heavy metals varies greatly at different levels of pollution (average values are shown in Tables S1 and S2). At each pollution level, the heavy metal content in PM 10 was significantly higher than that of PM 2.5 . With the increase of pollution level, the total amount of heavy metals in the particles gradually increased, but the degree of increase gradually decreased. In PM 2.5 and PM 10 , the order of heavy metal content at each pollution level was Zn > Cu> Pb > Cr > As > Ni > V > Cd. It is reported that Zn, Cu, Cr, Pb mainly come from exhaust emissions of motor vehicles or the wear of brake pads and tires [34][35][36], and Pb, As, Ni come from coal and petroleum combustion [13,37]. Cd is related to industrial processes [13,38], and V may come from mining or soil fertilizer use [39]. Lead, zinc, and copper account for a relatively high proportion, which is related to automobile exhaust. Urban traffic jams are becoming more and more serious, leading to frequent braking and start-up of vehicles, which aggravates the emission of heavy metals in the exhaust gas. Beijing is the city with the largest number of cars in China, and car exhaust has been studied in Beijing as a factor [40,41]. Chengdu is the second-largest city in the country for car ownership, so the contribution of car exhaust to Chengdu's atmospheric particulate matter is also significant [42]. est, indicating that heavy metals enriched in coarser particles may be discharged into the atmosphere at this pollution level. At the severe level, the heavy metal content increased sharply. For example, on 3rd January 2017, it was found that the wind speed was the lowest during the study period (0.5 m/s). The wind speed on the previous day (2 January) was relatively higher (0.9 m/s) and from the northwest. It is speculated that the heavy metal content on January 3rd sharply increased due to the metal sources carried by the wind from the northwest of Chengdu.
However, unlike the heavy metal percentage distribution, the relative percentage of each ion varies significantly at different pollution levels. In PM10, the percentage of Ca 2+ was significantly higher than that of PM2.5, indicating that Ca 2+ is more likely to be enriched in coarse particles. Coarse particles often are dust, and CaCO3 is a major component of dust, which is the same as previous studies in Chengdu [13,27]. From Clean to the subsequent pollution levels, the relative percentages of NO3 − , NH4 + gradually in- The relative percentage content of all heavy metals is almost constant at each pollution level. The content of heavy metals (HM) per particle at different pollution levels is shown in Figure S1 (Supplementary Information). It can be seen that the heavy metals per particle changes with the increase of particle concentration. The heavy metals content per particle in PM 2.5 is always higher than that for PM 10 .
At the Light-Medium level, the ratio of heavy metals in PM 2.5 to heavy metals in PM 10 is the largest, indicating that heavy metals are mainly concentrated in fine particles at this pollution level. At the Heavy levels, the content of heavy metals in PM 10 and PM 2.5 is the smallest, and the contribution of heavy metals in PM 2.5 to that in PM 10 is the smallest, indicating that heavy metals enriched in coarser particles may be discharged into the atmosphere at this pollution level. At the severe level, the heavy metal content increased sharply. For example, on 3rd January 2017, it was found that the wind speed was the lowest during the study period (0.5 m/s). The wind speed on the previous day (2 January) was relatively higher (0.9 m/s) and from the northwest. It is speculated that the heavy metal content on January 3rd sharply increased due to the metal sources carried by the wind from the northwest of Chengdu.

Ions Characteristics at Different Pollution Levels
The ions in PM have significant differences at different pollution levels ( Figure 5). From Clean to the subsequent pollution levels, the ion content in the particles increased gradually (see Tables S3 and S4  different pollution levels. The content of ions in particles is obviously different at different pollution levels. From Clean to Heavy or Severe, the content of ions in PM2.5 and PM10 decreased gradually. The ion content per particle in PM2.5 is always greater than that for PM10 at each pollution level, but the ratio of ion content per particle in PM2.5 to PM10 decreases gradually from Clean to Heavy. The results indicate that ions may mainly enrich fine particles, but the proportion of ions in coarser particles gradually increases as the particle concentration increases.  Figure 6 shows the changes of parameters related to SNA (SO4 2− , NO3 − , NH4 + ) at different pollution levels. From Clean to the subsequent pollution levels, SOR is always greater than NOR, but the degree the two increases with the pollution levels are different. NOR in PM2.5 increased from 0.11 (Clean) to 0.22 (Severe), and SOR increased from 0.43 (Clean) to 0.61 (Severe). NOR in PM10 increased from 0.16 (Clean) to 0.29 (Heavy), and the SOR increased from 0.52 (Clean) to 0.60 (Heavy). Early studies have shown that, when SOR is greater than 0.1, there is a photochemical reaction of SO2 in the atmosphere [47]. This result indicates that SO2 is more susceptible to secondary conversion than NO2. The sulfate and nitrate in this study were largely formed through secondary reactions. However, unlike the heavy metal percentage distribution, the relative percentage of each ion varies significantly at different pollution levels. In PM 10 , the percentage of Ca 2+ was significantly higher than that of PM 2.5 , indicating that Ca 2+ is more likely to be enriched in coarse particles. Coarse particles often are dust, and CaCO 3 is a major component of dust, which is the same as previous studies in Chengdu [13,27]. From Clean to the subsequent pollution levels, the relative percentages of NO 3 − , NH 4 + gradually increased, but the relative percentages of Ca 2+ , K + , Na + , and SO 4 2− gradually decreased, and the relative percentages of Mg 2+ , Cl − , and F − were basically stable. This result demonstrated atmospheric polluting processes in Chengdu were mainly caused by particles with ions such as NO 3 − and NH 4 + during the research period. It is reported that with the control of SO 2 pollution in China, the sulfate content in PM has been significantly reduced [43,44]. At the same time, NO 3 − and SO 4 2− will interact, and NO x will catalyze the conversion of SO 2 to SO 4 2− [45]. The oxidation of a large amount of SO 2 will not only produce SO 4 2− but also promote the formation of NO 3 − on water particles [46]. Therefore, SO 2, as the precursor of sulfate, is oxidized, as NOx is converted to NO 3 − , and the conversion of SO 2 should be slow and reduced. Figure S2 (Supplementary Information) shows the content of ions per particle at the different pollution levels. The content of ions in particles is obviously different at different pollution levels. From Clean to Heavy or Severe, the content of ions in PM 2.5 and PM 10 decreased gradually. The ion content per particle in PM 2.5 is always greater than that for PM 10 at each pollution level, but the ratio of ion content per particle in PM 2.5 to PM 10 decreases gradually from Clean to Heavy. The results indicate that ions may mainly enrich fine particles, but the proportion of ions in coarser particles gradually increases as the particle concentration increases.

Characteristics of Sulfate-Nitrate-Ammonium (SNA)
3.3.2. Characteristics of Sulfate-Nitrate-Ammonium (SNA) Figure 6 shows the changes of parameters related to SNA (SO 4 2− , NO 3 − , NH 4 + ) at different pollution levels. From Clean to the subsequent pollution levels, SOR is always greater than NOR, but the degree the two increases with the pollution levels are different. NOR in PM 2.5 increased from 0.11 (Clean) to 0.22 (Severe), and SOR increased from 0.43 Atmosphere 2021, 12, 990 9 of 13 (Clean) to 0.61 (Severe). NOR in PM 10 increased from 0.16 (Clean) to 0.29 (Heavy), and the SOR increased from 0.52 (Clean) to 0.60 (Heavy). Early studies have shown that, when SOR is greater than 0.1, there is a photochemical reaction of SO 2 in the atmosphere [47]. This result indicates that SO 2 is more susceptible to secondary conversion than NO 2 . The sulfate and nitrate in this study were largely formed through secondary reactions. with the change of pollution level. Sulfate and nitrate are important hygroscopic ions, which can promote the hygroscopic growth of atmospheric particles and have a great impact on visibility and temperature [3,54,55]. The NO3 − /SO4 2− ratio has large differences at different pollution levels, which gradually increase with the increase of pollution levels, and the aerosol ions will be easier to absorb moisture [56]. NO3 − represents mobile source, and SO4 2− represents fixed source. The NO3 − /SO4 2− ratio is often used to indicate whether particulate matter is dominated by mobile source or fixed source. The NO3 − /SO4 2− ratio increased from 0.52 (Clean) to 0.95 (Severe) in PM2.5, and increased from 0.57 (Clean) to 1.20 (Heavy) in PM10. The results show that the contribution of pollution caused by mobile sources to the increase of PM is gradually increasing. In the fine particles, it is a mainly fixed pollution source at different pollution levels. While in the coarse particles, it is a mainly fixed pollution source at Clean and Light-Medium levels, and mainly mobile sources at the Heavy.  Figure 7 shows the correlation between SNA and PM at different pollution levels. The relative contribution of SNA to the increase of PM2.5 and PM10 at each pollution level is different. At Clean levels, the contribution of SO4 2− to the increase of PM2.5 is 11.5%, which is much larger than the contribution of other ions, and the relative contribution of SNA to PM10 is very small. At Light-Medium levels, the contributions of SO4 2− , NH4 + , NO3 − to PM differ (PM2.5: 7.18%, 10.9%, 9.93%; PM10: 10.4%, 14.0%, 14.0%). Sulfates contributed more to PM2.5 at the Clean levels because of the trend to form ammonium sulfate during the formation of SNA in the inhomogeneous phase, which impeded the formation of ammonium nitrate. This result is reflected in more contribution of nitrate to The formation of SNA is closely related to meteorological conditions (relative humidity and temperature) [48][49][50]. When the relative humidity is low, the main reaction is a gasphase reaction, and when the relative humidity is high, the main reaction is a heterogeneous reaction on particles [51,52]. According to Pandis and Seinfeld [53], the liquid-phase oxidation of SO 2 may be an important way to generate SO 4 2− , while NO 3 − is mainly generated by gas-phase oxidation of NO x . So, the effect of humidity on SO 4 2− is more significant. With the increase in pollution levels, the relative humidity increased from 76% to 83%. As the air approached saturation, the particle concentration increased, and the temperature decreased (PM 2.5 : 21.7-11.8 • C; PM 10 : 19.8-17.8 • C). It can be seen that nitrate and sulfate in this study tended to form through heterogeneous reactions with the change of pollution level. Sulfate and nitrate are important hygroscopic ions, which can promote the hygroscopic growth of atmospheric particles and have a great impact on visibility and temperature [3,54,55]. The NO 3 − /SO 4 2− ratio has large differences at different pollution levels, which gradually increase with the increase of pollution levels, and the aerosol ions will be easier to absorb moisture [56]. NO 3 − represents mobile source, and SO 4 2− represents fixed source. The NO 3 − /SO 4 2− ratio is often used to indicate whether particulate matter is dominated by mobile source or fixed source. The NO 3 − /SO 4 2− ratio increased from 0.52 (Clean) to 0.95 (Severe) in PM 2.5 , and increased from 0.57 (Clean) to 1.20 (Heavy) in PM 10 . The results show that the contribution of pollution caused by mobile sources to the increase of PM is gradually increasing. In the fine particles, it is a mainly fixed pollution source at different pollution levels. While in the coarse particles, it is a mainly fixed pollution source at Clean and Light-Medium levels, and mainly mobile sources at the Heavy. Figure 7 shows the correlation between SNA and PM at different pollution levels. The relative contribution of SNA to the increase of PM 2.5 and PM 10 at each pollution level is different. At Clean levels, the contribution of SO 4 2− to the increase of PM 2.5 is 11.5%, which is much larger than the contribution of other ions, and the relative contribution of SNA to PM 10 is very small. At Light-Medium levels, the contributions of SO 4 2− , NH 4 + , NO 3 − to PM differ (PM 2.5 : 7.18%, 10.9%, 9.93%; PM 10 : 10.4%, 14.0%, 14.0%). Sulfates contributed more to PM 2.5 at the Clean levels because of the trend to form ammonium sulfate during the formation of SNA in the inhomogeneous phase, which impeded the formation of ammonium nitrate. This result is reflected in more contribution of nitrate to PM at the Light-Medium levels, compared to the Clean levels. At Heavy levels, the contributions of SO 4 2− , NH 4 + , NO 3 − to PM are significantly different (PM 2.5 : 24.0%, 3.76%, 11.7%; PM 10 : 24.0%, 40.8%, 59.7%). This result shows sulfate is more likely to be enriched in fine particles at each level. Nitrate and ammonium salts are easily concentrated in fine particles at Clean and Light-medium pollution levels, while they are easily concentrated in coarse particles at Heavy pollution levels (NO 3 − : 59.7%; NH 4 + : 40.8%). The secondary conversion of SO 2 is mainly liquid-phase reaction, which is closely related to relative humidity. The relative humidity of the Heavy level is the largest (83%), so the contribution of sulfate to particles is also the largest at this level. The secondary reaction of NO 2 is a mainly gas-phase reaction. The atmospheric temperature is lower than other levels at Heavy levels, which is not conducive to the secondary generation of NO 2 . However, it has been reported that it is conducive to the stable existence of NH 4 NO 3 . Therefore, the contribution of NO 3 − to PM is relatively large at Heavy levels [57].
phere 2021, 12, x FOR PEER REVIEW 10 of 13 PM at the Light-Medium levels, compared to the Clean levels. At Heavy levels, the contributions of SO4 2− , NH4 + , NO3 − to PM are significantly different (PM2.5: 24.0%, 3.76%, 11.7%; PM10: 24.0%, 40.8%, 59.7%). This result shows sulfate is more likely to be enriched in fine particles at each level. Nitrate and ammonium salts are easily concentrated in fine particles at Clean and Light-medium pollution levels, while they are easily concentrated in coarse particles at Heavy pollution levels (NO3 − : 59.7%; NH4 + : 40.8%). The secondary conversion of SO2 is mainly liquid-phase reaction, which is closely related to relative humidity. The relative humidity of the Heavy level is the largest (83%), so the contribution of sulfate to particles is also the largest at this level. The secondary reaction of NO2 is a mainly gas-phase reaction. The atmospheric temperature is lower than other levels at Heavy levels, which is not conducive to the secondary generation of NO2. However, it has been reported that it is conducive to the stable existence of NH4NO3. Therefore, the contribution of NO3 − to PM is relatively large at Heavy levels [57].

Conclusions
The present study analyzed the distribution and changes of heavy metals and water-soluble ions in PM2.5 and PM10 during the haze periods from March 2016 to January 2017 in Chengdu, China at different pollution levels. It revealed the concentration of PM was closely related to meteorological conditions and the effect on fine particles is more significant. Heavy metals were more easily enriched in fine particles at different pollu-

Conclusions
The present study analyzed the distribution and changes of heavy metals and watersoluble ions in PM 2.5 and PM 10 during the haze periods from March 2016 to January 2017 in Chengdu, China at different pollution levels. It revealed the concentration of PM was closely related to meteorological conditions and the effect on fine particles is more significant. Heavy metals were more easily enriched in fine particles at different pollution levels, and the relative percentage content was basically stable. However, the relative percentage of water-soluble ions varied with the pollution level, and the relative percentage of NO 3 − and NH 4 + increased gradually. The water-soluble ions in the particles during the study were mainly SO 4 2− , NO 3 − and NH 4 + and mainly from secondary reactions. Furthermore, the contribution of SNA to the increase of PM was variable at different pollution levels. It was mainly SO 4 2− in PM at Clean levels, and mainly NH 4 + and NO 3 − at Light-Medium levels. At Heavy levels, it is mainly SO 4 2− in PM 2.5 , and mainly NH 4 + and NO 3 − in PM 10 . Mobile sources are contributing more to the occurrence of haze in Chengdu, which should have more attention paid to it. The results of this research not only enrich the air pollution research in Chengdu, China, but also provide a reference for the urban air pollution research with the same background. The deficiency lies in the lack amount of PM 10 samples under Heavy and Severe pollution levels. The next step will be to study the source analysis of PM quantitatively.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/atmos12080990/s1, Figure S1: Changes of heavy metals per particle at different pollution levels, Figure S2. Change in ions per particle at different pollution levels, Table S1: Average mass concentrations of heavy metals in PM 2.5 at different pollution levels (ng/m 3 ), Table S2: Average mass concentrations of heavy metals in PM 10 at different pollution levels (ng/m 3 ), Table S3: Average mass concentration of ions in PM 2.5 at different pollution levels (µg/m 3 ), Table S4: Average mass concentration of ions in PM 10 at different pollution levels (µg/m 3 ).