Evolution of Key Chemical Components in PM2.5 and Potential Formation Mechanisms of Serious Haze Events in Handan, China

Handan has been one of the most polluted cities in China since 2013 and became the top city for PM2.5 in 2017. In this research, we observed coarse particulate matter (PM10), fine particulate matter (PM2.5), submicron particulate matter (PM1), and the chemical composition of PM2.5 from November 16, 2015, till March 14, 2016, in Handan. During the observation period, hourly concentrations of PM10, PM2.5, and PM1 peaked at 1070.1, 864.4, and 519.5 μg m, respectively. Severe pollution occurred on a large fraction of days in the heating season, which was characterized by frequent and longlasting pollution episodes. A large fraction of the transport trajectories during nine typical episodes during that period in Handan were from the northwest. Water-soluble ions (sulfate, nitrate, and ammonium) in PM2.5 accounted for the largest proportion at all pollution levels. The highest proportion of SIA occurred in a heavily polluted episode, during which it was as high as 50.0% (sulfate: 18.8%, nitrate: 18.7%, and ammonium: 12.5%). The sulfate and ammonium in PM2.5 increased gradually while the nitrate decreased as the level of pollution rose from clean to heavily polluted. The fraction of SOA and OM to PM2.5 decreased as the pollution level increased, indicating a weakening of photochemical reactions. The POA in PM2.5 increased with the aggravation of haze, and the heterogeneous chemistry was enhanced by the aggravation of pollution. Liquid reactions were important in the formation of sulfate during pollution and non-pollution stages. Liquid reactions of NO2 are enhanced in the pollution stage during the heating season in Handan.


INTRODUCTION
Environmental problems seriously affect air quality and public health in China (Tie et al., 2009;Zheng et al., 2014;Lin et al., 2015;Wang et al., 2015a).Aerosol pollution characterized by high concentration and low visibility frequently occurs in China (Tao et al., 2014;Wang et al., 2015c), and many areas have often experienced severe haze pollution in recent years (Cheng et al., 2012;Liu et al., 2013;Che et al., 2015).The North China Plain (NCP) is recognized as the hardest hit area (Zhang et al., 2015a), owing to the combination of the high emission of air pollutants that result from heating and frequent stable meteorological conditions in this area (Wang et al., 2015b;Elser et al., 2016).Handan is located on the margin of the NCP region, adjacent to the four provinces of Henan, Hebei, Shandong, and Shanxi (Wei et al., 2010).Handan was one of the ten most seriously polluted cities in China from 2013 till 2017 (http://www.mep.gov.cn/hjzl/zghjzkgb/lnzghjzkgb), with the highest concentration of PM 2.5 (more than 700 µg m -3 on an hourly average) and a persistent haze phenomenon (Wei et al., 2014a, b).In 2017, it had the highest PM 2.5 concentrations within all the monitored cities in China.Consequently, the air quality of this city attracts wide attention from scholars.
Handan is a typical urbanized city with rapid industrial development and high population density.It has the characteristics of high emission, a complicated emission environment, complicated sources, and a special geographical position.Relevant studies have been conducted to understand the particulate matter problem in Handan.Previous studies include the use of MM5-models-3/CMAQ models to analyze the contribution rates of regions and sources to PM 2.5 during a special period (Wang et al., 2014a).It was proved that the two major local sources from SHB (Southern Hebei), industrial and domestic, contributed 28.8% and 25.1%, respectively, of the total PM 2.5 concentrations in Handan.In addition, previous studies used the Positive Matrix Factorization (PMF) model to understand major sources during December 2012 till January 2013 in Handan.The major sources were coal combustion (25.9%); secondary emission (21.8%); industry (16.2%); Ba, Mn, and Zn (12.7%); motor vehicles (7.7%); road dust (10.9%);K + , As, and V (6.3%); and fuel oil combustion (2.5%) (Wei et al., 2014a, b).In recent years, extreme haze episodes have been attracting great scientific interest.The regional transport of pollutants contributed significantly to the concentration of PM 2.5 (Wang et al., 2014a;Wang et al., 2014b), dust (Yang et al., 2013), and SO 2 (Yang et al., 2013).For example, the regional sources contributed from 27.5% (Shijiazhuang) to 28.6% (Handan) of the PM 2.5 concentrations in three cities (Shijiazhuang, Handan, and Xingtai) in January 2013 (Wang et al., 2014c).
During haze pollution periods, relative humidity is usually high, and ozone concentration is usually low.Most increases in secondary inorganic species are likely to be led by heterogeneous reactions rather than photochemical reactions (Wang et al., 2012;Sun et al., 2013a;Huang et al., 2014;Han et al., 2015;Quan et al., 2015;Zhang et al., 2016).Cheng et al. (2016) and Wang et al. (2016a) confirmed that the production of sulfate and nitrate via heterogeneous reactions on aqueous particles leads to severe haze pollution in polluted environments with high relative humidity.Thus, the haze pollution in Beijing during winter was attributed to the combination of meteorological conditions, primary emissions (coal combustion), heterogeneous reactions, and regional transport (Sun et al., 2013b(Sun et al., , 2015)).
PM 2.5 concentration is known to increase with the aggravation of haze pollution.The concentration of PM 2.5 increases as the level of pollution increases by virtue of heterogeneous reactions and regional transport (Zheng et al., 2015;Ma et al., 2017).In this study, PM 10 , PM 2.5 , PM 1 , and the chemical compositions of PM 2.5 were observed online during the heating season in Handan to address the following questions: (1) characteristics of particulate pollution and meteorological conditions in multiple episodes during the heating season; (2) evolution of key chemical components in PM 2.5 ; and (3) the potential formation mechanisms of serious haze pollution.

Experimental Site
Online ambient observation was carried out from November 16, 2015, till March 14, 2016, on the campus of Hebei University of Engineering.The observation site is situated on the rooftop of the Urban Construction Laboratory building in Hebei University of Engineering (36°34′N, 114°29′E), approximately 12 m above the ground (Fig. 1).There are no obvious pollution sources and tall buildings that impede air circulation at this site, which is a typical mixed area of cultural and educational residents.The collected samples have a certain authenticity and represent the air pollution level of Handan City to a certain extent.

Online Monitoring of Pollutants and Meteorological Parameters
Hourly concentrations of particulate matter were monitored continuously for the entire research period using dichotomous monitors (PM-712 and PM-714; Kimoto Electric Co., Ltd., Japan) at a flow rate of 16.7 L min -1 .The PM-712 and PM-714 monitors measured particle mass concentration directly, including the weight of water in the particles.The particulate matter online monitor works under the principle of the βray absorption method, and the monitor is equipped with a US-EPA inlet and a virtual impactor (Kimoto Electric Co., Ltd., 2012;Kaneyasu et al., 2014).We used the hygroscopic growth correction formula to achieve dehumidification: Dehumidified PM x mass conc.= Measured PM x mass conc.× [1/(1 + 0.010 × e 6.000RH/100 )] (1) RH refers to the relative humidity in %, whereas 0.010 and 6.000 are semi-empirical constants derived by comparing the results of PM-714 and PM-712 with the results obtained using the US Federal Reference Method.As the influence of chemical compositions on particulate hygroscopicity is not accurately considered in this equation, it is not applicable to every sampling condition.However, in general, this equation can provide a satisfying estimation of Aerosol Water (AW), as validated by the Ministry of Environment in Japan.AW was estimated by the hygroscopic growth correction formula (Zheng et al., 2016): AW = Measured PM 2.5 mass conc.× [0.010 × e 6.000RH/100 / (1 + 0.010 × e 6.000RH/100 )] (2) PM 10 dehumidification was thus achieved by deducting AW from the measured PM 10 data, which was acquired by summing PM 2.5 and PM 2.5-10 .All PM data hereinafter refer to the dehumidified PM data.
The carbonaceous aerosol content in PM 2.5 was obtained from the APC-710 Monitor.This analyzer collects aerosol samples on the PTFE filter tape and then performs optical analysis at different wavelengths, as introduced in the previous patent of Kimoto Electric, Ltd., Japan (application number: 2015-039568, Japan Patent Office, https://www.jplatpat.inpit.go.jp/web/all/top/BTmTopPage).SO 4 2-and NO 3 in PM 2.5 were obtained from the ACSA-08 Monitor.In winter, NH 4 + of the North China Plain is always in excess (Zheng et al., 2015;Meng et al., 2016;Zheng et al., 2016).This study used the same method as in related past studies for estimating NH 4 + (Zheng et al., 2015(Zheng et al., , 2016)).NH 4 + was predicted to be 1.09 times of that required to fully neutralize SO 4 2-and NO 3 -.The coefficient 1.09 was dependent on the analysis of our PM 2.5 off-line samples in January, April, July, and October 2015 in Handan, and the calculated coefficient 1.09 is the same as that in related research in Beijing (Fig. 2) (Zheng et al., 2015(Zheng et al., , 2016)).PM 2.5 off-line samples were also collected from the observation site.Details on the PM 2.5 off-line data collection, analysis of water-soluble ions, quality assurance (QA), and quality control (QC) are provided in our previous study (Meng et al., 2016).The method of estimating NH 4 + is described in detail in a related study (Zheng et al., 2016).As an online automatic meteorological observation monitor, WXT (Kimoto Electric Co., Ltd., Japan) was used to obtain meteorological parameters, such as atmospheric pressure, temperature, RH, wind speed, and wind direction.SO 2 and NO 2 concentrations were obtained from the online monitoring instruments SA-731 and NA-721.Detailed information of the Kimoto instrument can be obtained from: http://www.kimoto-electric.co.jp.In addition, the online date of visibility was obtained using the SWS-100 visibility meter (BIRAL, England).The specifications of the instruments are listed in Table 1.The measurement procedure and the quality assurance/quality control measures (QA/QC) for online data are available in previous studies (Duan et al., 2006;Li et al., 2017a).

Formulas of the Main Components in PM 2.5
The formulas of the major components in PM 2.5 are listed in Table 2. Sulfate (SO 4 2- ) and nitrate (NO 3 -) were formed by chemical conversions of SO 2 and NO x precursors.The conversion ratio of sulfate (SOR) and nitrate (NOR) were estimated to describe the formation of secondary aerosol species.High SOR and NOR mean more SO 2 and NO 2 that are converted into secondary inorganic aerosols (Han et al., 2015).

Trajectory Calculation and Clustering
We used the Hybrid Single Particle Lagrangian Integrated Trajectory (HYSPLIT) model for 24 h calculations, in which 24 trajectories were calculated each day with a starting height of 500 m.HYSPLIT is a professional model developed by the Air Resources Laboratory of National  (Wang et al., 2009).

Overview of Meteorological Conditions and Aerosol Characteristics during the Heating Season
Particulate matter pollution was serious in Handan during the heating season of 2015-2016(November 14, 2015-March 14, 2016).The average concentrations of PM 10 , PM 2.5 , and PM 1 were 227.9, 132.8, and 105.5 µg m -3 , respectively.PM 1 constituted about 80% of PM 2.5 that constituted about 60% of PM 10 .Hourly concentrations of PM 10 , PM 2.5 , and PM 1 peaked at 1070.1, 864.4,and 519.5 µg m -3 (excluding the period of the Spring Festival), respectively.The average concentrations of PM 10 and PM 2.5 were 1.5 and 1.8 times higher than the Grade II national standards for air quality (PM 10 : 150 µg m -3 , PM 2.5 : 75 µg m -3 , China).Most of the pollution events occurred frequently during the heating season and lasted for a long time (Zheng, et al., 2015;Ma et al., 2017).Wang et al. (2014b) proposed the definition of haze episodes as two successive days with daily PM 2.5 exceeding 75 µg m -3 .The same definition was applied in this study with the addition of visibility below 10 km.Details of the pollution episodes are listed in Table 3(a).In total, nine episodes were identified in the heating season of 2015-2016.Maximum episode-averaged PM 10 , PM 2.5 , and PM 1 concentrations reached 447.9, 252.8, and 198.8 µg m -3 , respectively, in Episode III.Another unique feature of PM 10 , PM 2.5 , and PM 1 during haze periods in the heating season was their dramatic hourly fluctuation.The maximum hourly variations were 842.2, 838.7, and 510.2 µg m -3 , respectively, during the research period.Hourly data of changes in PM concentration over 100 µg m -3 were observed over 60 times during pollution periods.PM 2.5 had a good correlation with OC, EC, and nitrate (R 2 > 0.6 for these nine species), but poor correlation with sulfate, reflecting larger episodic variations.The average concentrations of NO 2 and SO 2 were different in each pollution episode (Table 3(b)).In addition, the mass concentration of SO 2 was higher than of NO 2 in all pollution episodes.SO 2 exceeded most NO 2 by 250% in Episode VII.Episode V to Episode IX was much drier than Episode I to Episode IV.The relatively high SO 2 and low NO 2 concentrations in episodes may indicate that the fixed source emissions contained more coal combustion in Handan.The emission of SO 2 was generally higher than that of NO 2 in the North China Plain (Wang et al., 2013(Wang et al., , 2016b;;Li et al., 2017b; MEIC, http://www.meicmodel.org).

Backward Trajectory Analysis
In order to investigate the transport characteristics of the air mass trajectory in the heating season of Handan, HYSPLIT was used to analyze air mass trajectories of each episode during the heating season by conducting cluster analysis.Considering the passage through the area, moving speed, and direction of air mass trajectories, the trajectories were divided into four categories and calculated from a height of 500 m, at which the wind field can not only reduce the influence of ground friction on the air flow trajectory but also accurately reflect air mass transport characteristics Episode IV and Episode VI were influenced by air mass trajectories from Tianjin that accounted for 11% and 23%, respectively.Episode V and Episode IX were influenced by air mass trajectories from Shanxi Province that accounted for 19% and 21%, respectively.The third category of Episode VII was from Henan Province and accounted for 21%.The distances of the transport trajectories were short and slower, which led to the accumulation of in Handan, the southern region of Hebei Province.

Chemical Profiles at Different Pollution Levels
Haze periods are characterized by higher RH and lower radiation intensity (Crutzen and Birks, 1982;Ramanathan et al., 2001;Cheng et al., 2008;Ramanathan and Carmichael, 2008;Wendisch et al., 2008;Zheng et al., 2015;Li et al., 2017b;Ma et al., 2017).From the clean level to the heavily polluted level, RH increased by 44.1%, from 38.2% to 55.1%.The main components of PM 2.5 are secondary aerosols (inorganic and organic) in China (Yang et al., 2011).The variation in the chemical composition accompanying the change in PM 2.5 pollution level among episodes was explored in this research.According to the Air Quality Index (http://kjs.mep.gov.cn/hjbhbz/bzwb/dqhjbh/jcgfffbz/201203/t20120302_224166.htm?COLLCC=2906016564&), PM 2.5 pollution is classified into four levels: clean (PM 2.5 ≤ 35 µg m -3 ), slightly polluted (35 < PM 2.5 ≤ 115 µg m -3 ), polluted (115 < PM 2.5 ≤ 350 µg m -3 ), and heavily polluted (PM 2.5 > 350 µg m -3 ), where PM 2.5 refers to the hourly concentration.The wind speed was lower in the pollution episodes than in the normal period, whereas relative humidity showed a continuous increase trend with the increase in the pollution level (Figs. 4,5,and Table 3(b)), which indicates that the meteorological conditions became unconducive to horizontal diffusion of pollutants with the increase in the pollution level.The concentrations of PM 10 , PM 2.5 , PM 1 , and chemical components in PM 2.5 were low during the clean period, but they significantly increased as the pollution level increased (Fig. 5).The concentrations of PM 10 , PM 2.5 , and PM 1 increased from 61.1, 24.8, and 20.1 µg m -3 to 705.1, 432.4,and 328.2 µg m -3 .Water-soluble ions (sulfate, nitrate, and ammonium) of PM 2.5 accounted for the largest proportion in all pollution levels.The highest proportion of SIA was 50.0%(sulfate: 18.8%, nitrate: 18.7%, and ammonium: 12.5%) in the heavily polluted level.Although the mass concentrations of PM 10 , PM 2.5 , and PM 1 and the chemical composition of PM 2.5 all showed increases as the pollution level increased, the relative contributions of chemical compositions varied (Fig. 5).For example, when the fraction of organics to PM 2.5 decreased from 37.9% (clean) to 26.1% (heavily polluted) as a whole, the fraction of EC increased from 3.6% to 4.3%, which is not a significant change.Water-soluble ions in PM 2.5 clearly increased as the pollution level increased.The SIA contribution increased in PM 2.5 and RH (Fig. 5), which indicated that the dependence of SIA on relative humidity is more likely attributable to heterogeneous reactions (Zheng et al., 2014;Ma et al., 2017).The fraction of POA increased from 9.0% to 12.4% (Fig. 5), indicating that the increase of POA had contributed to the aggravation of haze in Handan (Shen et al., 2009;Zhang et al., 2015b;Li et al., 2017b).
During a haze period, the higher mass concentration of particles will weaken the radiation, while the decrease of radiation intensity will change the photochemical reaction and oxidant concentration.O 3 was reduced by 91.3%, dropping from 42.4 µg m -3 at the clean level to 3.7 µg m -3 at the severely polluted level; this consequently changed the production and aging of secondary organic aerosols (SOAs) (Hallquist et al., 2009;Jimenez et al., 2009).O 3 is considered to be the crucial oxidant for the formation of SOAs (Jimenez et al., 2009).Since SOA is a product of photochemical reactions, the contribution of SOA to PM 2.5 decreased from 28.9% (clean level) to 13.8% (heavily polluted level).The results show that with the increase of particles, the concentration of photochemical oxidants decreased, photochemical reactions weakened, and the concentration of SOA also decreased (Zheng et al., 2014).
To assess the role of chemical outputs, the EC-scale concentrations were analyzed for every compound.People tend to use the EC-scale concentration to remove the influence of different dilution/mixing conditions on the changes in pollutant concentrations that were observed.Changes in the pollutant concentration are determined not only by chemical reactions but also by boundary layer developments.In terms of chemical production rate and the same emission rate, various mixing conditions will lead to different air pollutant levels.Thus, it is extremely difficult to conclude that a stronger/weaker chemical production relies on pure concentration data when the boundary layer effect is not taken into consideration.Because EC is only produced by primary emission and is relatively inertial to chemical reactions, its changes show in detail the influence of atmospheric physical processes (dilution/mixing effect).The ratio between other pollutants and EC will remove the changes to a large extent because of mixing/dilution and perfectly present the contribution of chemical reactions.If the main reason for the increase was the reduction in the boundary layer thickness, the PM 2.5 /EC ratio would be close to constant during changes from the clean level to the polluted level.It is noteworthy that the PM 2.5 concentration is standardized by EC so as to off-set the effect exerted by the reduced boundary layer (Ma et al., 2017).The PM 2.5 /EC ratio was not constant with an increasing pollution level in this study (Fig. 6).Therefore, the reduction of the boundary layer could not be the dominant reason for the PM 2.5 increase.In addition, Zheng et al. (2015) found that regional transport and chemical conversions could cause the increase of PM 2.5 during haze pollution periods in Beijing; these processes may also aggravate particulate matter pollution in Handan (Wang et al., 2012;Zheng et al., 2014;Quan et al., 2015;Sun et al., 2015;Zhang et al., 2015a;Hua et al., 2016;Wang et al., 2016a).
As shown in Fig. 5, SOA/EC showed a decreasing trend (Fig. 6), which reflects the reduced photochemical production (Ma et al., 2017).However, the sulfate and ammonium in PM 2.5 gradually increased from the clean level to the heavily polluted level.Relative contributions of nitrate to PM 2.5 decreased as a whole as the pollution level increased (Fig. 5).However, the increase of SO 2 4-/EC and the decrease of NO 3 -/EC slowly changed from clean periods to heavily polluted periods (Fig. 6).In addition, the effect of heterogeneous reactions was discussed.SOR and NOR have been used as indicators of secondary transformation (Sun et al., 2006).The distinct increase in SOR, NOR, SO 2 , and NO 2 with pollution events becoming more severe (Fig. 6) reflects secondary formations of sulfate and nitrate during severe haze events.

Enhanced Heterogeneous Chemistry
SOR increased as RH increased and reached the maximum at 80% (excluding the data for VIS > 10 km).NOR firstly increased and then decreased, with the maximum at 60-70%.SOR and NOR depended on RH and corresponded with the concentration of aerosols well.The maximums of SOR and NOR were 0.55 and 0.48, respectively (Fig. 7).SO 2 /EC decreased significantly, and NO 2 /EC slightly changed.SO 4 2-/EC prominently increased, and NO 3 -/EC decreased slowly as RH increased, indicating that SO 2 was dissolved in the liquid water of aerosols and promoted liquid phase reactions to produce more SO 4 2-as the pollution level increased (Zheng et al., 2015).The influence of temperature on SOR and NOR was rather complex during the heating season.High temperatures and solar radiation promoted SO 2 and NO 2 with OH to produce more SO   -by gas-phase oxidation (Zheng et al., 2015), but the effect of temperature was not obvious under the condition of high RH.POA, NH 4 -, and SO 4 2-increased obviously and SOA decreased as the pollution level increased.This implies that photochemical reactions weakened and the heterogeneous chemistry was enhanced in Handan.With the aggravation of haze, NO 3 -in PM 2.5 decreased, which may be caused by the effect of heterogeneous chemistry, and photochemical reactions ceased.
The time series of SOR, NOR, and aerosol water content is shown in Fig. 8 (excluding the effect of precipitation).The moisture content of aerosol was positively correlated with SOR and NOR during both the non-pollution and pollution periods, with correlation coefficients of 0.85 and 0.44, respectively, for the non-pollution and 0.86 and 0.53, respectively, for the pollution period.The correlation between SOR and aerosol water content remained the same throughout the pollution and non-pollution periods in Handan.Liquid-phase oxidation to sulfate formation was very important, and there was no obvious change between the pollution and non-pollution periods.SO 2 was absorbed by aerosol moisture and oxidized by H 2 O 2 or O 3 to SO 4 2- (Seinfeld et al., 1998).The liquid chemical to sulfate pollution played an important role in the pollution process under high humidity conditions (fog or wet haze) in many other cities (Sun et al., 2006;Du et al., 2011;Sun et al., 2013a).The correlation coefficient of NOR in the pollution stage was significantly higher than that in the non-pollution stage, and the correlation coefficient value was less than that of SOR.NO 2 was oxidized to HNO 3 by OH mainly through gas-phase reaction when increased haze would promote NO 2 for liquid-phase oxidation (Zheng et al., 2015).

CONCLUSIONS
The evolution of key chemical components in PM 2.5 and the potential formation mechanisms of serious haze pollution were discussed in this study.We observed online particulate matter, gaseous pollutants, meteorological parameters, and SO 4 2-, NO 3 -, OC, and EC in PM 2.5 from November 16, 2015, till March 14, 2016, in a highly polluted city of the North China Plain.Handan experiences high levels of particulate matter pollution that occur frequently and last for long periods.The concentration of SO 2 was obviously higher than that of NO 2 during the heating season in Handan.We analyzed air-mass trajectories of Handan during the haze period using HYSPLIT and found that they were mainly from the northwest.The relative contribution of nitrate to PM 2.5 decreased with the aggravation of pollution.The proportion of POA in PM 2.5 increased with the level of pollution.Furthermore, the weakening of photochemical reactions caused by the aggravation of haze led to the reduction of SOA in PM 2.5 .Heterogeneous reactions were enhanced under high RH that produced more SO 4 2-and NO 3 -, further aggravating the PM 2.5 pollution.Liquid reactions were important in the formation of SO 4 2-in the pollution and non-pollution stages.Liquid reactions of NO 2 would be enhanced in the pollution stage during the heating season in Handan.

Fig. 1 .Fig. 2 .
Fig. 1.(a) Location of the sampling site in Handan in the North China Plain.The map is color-coded by annual organic carbon emission rates modeled by the Multi-resolution Emission Inventory for China (MEIC, http://www.meicmodel.org;Li et al., 2017b).The grid size is 0.05° × 0.05°.(b) The monitoring site and equipment.

Fig. 3 .
Fig. 3. Backward trajectories of the transport paths to Handan during the nine pollution events in the heating season (November 14, 2015-March 14, 2016).

Fig. 4 .
Fig. 4. Time series of PM 10 , PM 2.5 , PM 1.0 , and major components (OC, EC, SO 4 2-and NO 3 -) of PM 2.5 and meteorological data (wind speed, wind direction, temperature and relative humidity) for the heating season in Handan.

Fig. 5 .
Fig. 5.Chemical compositions of PM 2.5 at different pollution levels in Handan.

Fig. 6 .
Fig. 6.(a) Particulate matter concentration normalized by EC at different pollution levels.Note that in order to eliminate the effect of atmospheric physical processes (mixing/dilution), SO 4 2-, NO 3 -, SIA, SOA, and PM 2.5 concentrations are normalized by EC to better represent the contribution of chemical reactions.(b) Variation of SO 4 2-, NO 3 -, SO 2 , NO 2 , SOR and NOR with the pollution level.

Fig. 7 .
Fig. 7. (a-b) Hourly SOR and NOR plotted against RH and colored with temperature.(c) Relationship between RH and aerosols (d) EC-scaled precursors (SO 2 and NO 2 ) and products (SO 4 2-and NO 3 -) plotted against RH.

Table 1 .
Specifications of the on-line monitoring instrument used in this study.

Table 2 .
Formulas and references of the main components of PM 2.5 .
Oceanic and Atmospheric Administration (NOAA) and Bureau of Meteorology Australia for the calculation and analysis of air pollutant transport and diffusion trajectory.It was used to calculate backward air trajectories arriving at the monitoring site.The related data were downloaded from ftp://arlftp.arlhq.noaa.gov/pub/archives/gdas1.Further details on HYSPLIT can be found in some related papers

Table 3 (
a). Information about pollution episodes in Handan.
category of Episode II was from Hubei Province, passed through Henan Province, and accounted for 17%.The third category of Episode III was from Hebei Province and accounted for the largest proportion at 52%.In addition, these air mass trajectories had the largest impact on Handan.