Pollution Characteristics, Chemical Compositions, and Population Health Risks during the 2018 Winter Haze Episode in Jianghan Plain, Central China

To determine the pollution characteristics, chemical compositions, and population health risks of PM2.5 at different pollution levels, PM2.5 samples were intensively collected during the long-lasting winter haze episode from 13–23 January 2018 in Xiantao in Jianghan Plain (JHP), central China. The higher PM2.5 levels during the severe pollution period were dominated by the WNW-NNE air-masses, whereas the lower PM2.5 concentrations during other pollution periods were mainly affected by the NE, S, and NW air-masses. The NO3−/SO42− and OC/EC ratios indicated a mixed contribution of intensive vehicle exhaust and secondary formation. The enrichment factor and geo-accumulation index for assessing the PM2.5-bound metal(loid)s contamination levels were positively correlated. Ingestion is the dominant exposure pathway of PM2.5-bound metal(loid)s for children and adults, followed by inhalation and dermal contact. As, Cr, and Pb may pose carcinogenic and non-carcinogenic risks, whereas Sb and V may only pose non-carcinogenic risks for children and adults. The population health risks may not depend on the pollution levels but depend on the PM2.5-bound metal(loid)s concentrations. PM2.5-bound metal(loid)s may pose much higher population health risks for adults compared to children. More attentions should be paid to the population health risks of PM2.5-bound metal(loid)s during a long-lasting winter haze episode in JHP.


Introduction
Haze is associated with the explosive growth of airborne fine particulate matter (PM 2.5 ) in ambient air [1]. PM 2.5 has attracted worldwide concerns over the several years because of its adverse effects on atmospheric visibility and human health [2]. For example, PM 2.5 can deposit in lungs through the inhalation exposure pathway and then result in lung cancer and acute respiratory infections [3]. In addition, it was classified as belonging to cancerogenic substance Group I by the International Agency for Research on Cancer [4]. However, the impacts of PM 2.5 on population health rely on its chemical compositions, such as metal(loid)s, which represent all kinds of PM 2.5 -bound metals and metalloids [5], and organic components [6]. It was reported that PM 2.5 -bound metal(loid)s have influences on acute changes in cardiorespiratory physiology [7] and allergic airways disease [8], whereas PM 2.5 -bound elemental carbon is the main contributor to epidemiological disease [9]. Hence, necessary knowledge of PM 2.5 chemical compositions is beneficial for the public and government to evaluate the population health risks.
Jingmen, Wuhan, and Xiantao. Therefore, we conducted the intensive sampling campaign in 13 January 2018. PM 2.5 field sampling campaign was carried out at an urban site from 13-23 January 2018 in Xiantao in JHP, central China. The geographical location of this sampling site is plotted in Figure 1. There is no obvious pollution source which may have effect on the sampling work around this urban sampling site, because the scale of industries near the sampling site, such as Ruiyang Automotive Parts (Xiantao) Co. Ltd. and Xiangyuan Electromechanical Equipment Co. Ltd. (Xiantao, China), are very small or do not have an independent production process. To track the haze bloom-decay process, samples were simultaneously collected on the rooftop of the library of Xiantao Vocational College (~20 m height, 30 • 20 27" N, 113 • 25 16" E) four times a day (at local times 06:00-11:00, 11:30-16:30, 17:00-22:00, and 22:30-05:30 the following day) with two medium-volume samplers (TH-150F, Wuhan Tianhong Instrument Co., Ltd., Wuhan, China) at an air flow rate of 100 L min −1 to increase the number of PM 2.5 samples classified at different pollution levels as much as possible. This prevented the stoppage of sampling due to excessive filter resistance caused by a haze episode. However, heavy rains and snowy weather were observed from the evening of 23 January till 27 January 2018. So, we terminated this intensive sampling process on 24 January 2018 after the fourth sample collection was completed on 22:30 23 January-05: 30 24 January 2018. To subtract possible contamination occurring during or after sampling, blank samples were collected for about 20 min by mounting blank filters onto the samplers without pumping into any air separately before and after the sampling [21]. In sum, a total of 43 pairs of PM 2.5 samples as well as three pairs of blank samples were effectively collected on quartz fiber filters (Whatman, UK) and Teflon filters (Munktell, Sweden), respectively. The sample data on 06:00-11:00 21 January was missing because the power supply line was cut off by accident, which meant that the two samplers could not work normally during this period. The quartz fiber filters were baked at 500 • C for 6 h in a muffle furnace and the Teflon filters were prepared at a constant temperature and relative humidity (25 ± 1 • C, 50 ± 5%) for 48 h before use, respectively. An electronic microbalance with a resolution of 1 µg (Sartorius SECURA 125-1S, Sartorius Lab Instruments GmbH & Co. KG, Göttingen, Germany) was used to determine the PM 2.5 mass on Teflon filters. After sampling, all the quartz fiber filters folded, wrapped with aluminum foil, then were sealed in plastic bags, and finally were stored in a refrigerator at −18 • C as well as Teflon filters for chemical analysis. According to the Technical Regulation on Air Quality Index of China (HJ 633-2012, http://www.mee.gov.cn/gkml/hbb/bgg/201203/t20120302_224146.htm), the sampling periods were divided into four pollution levels: mild, moderate, heavy, and severe pollution when the PM 2.5 concentrations were between 75 and 115, 115 and 150, 150 and 250, and greater than 250 µg m −3 , respectively. The sample numbers for mild pollution, moderate pollution, heavy pollution, and severe pollution levels are 17, 10, 10, and 3, respectively.
Atmosphere 2020, 11, x FOR PEER REVIEW 3 of 19 Province, especially in Xiangyang, Jingmen, Wuhan, and Xiantao. Therefore, we conducted the intensive sampling campaign in 13 January 2018. PM2.5 field sampling campaign was carried out at an urban site from 13-23 January 2018 in Xiantao in JHP, central China. The geographical location of this sampling site is plotted in Figure 1.

Chemical Analysis
The analysis method and procedures of WSIIs, CAs, and TE have been described in our previous studies [22][23][24][25]. To determine the WSIIs, a punch (3.14 cm 2 ) of quartz fiber filter was extracted using 20 mL ultrapure Milli-Q water (18.25 MΩ cm Analyses of filed blank samples were performed using the above same methods. All the reported data of WSIIs, CAs, and TE were corrected by the field blanks.

Meteorological Parameters, Trace Gases, and Air Mass Back Trajectory Analysis
Hourly meteorological parameters in January 2018 (Supplementary Figure S1), including wind direction (WD), wind speed (WS), temperature (Temp.) relative humidity (RH), precipitation (Prec.), and visibility (Vis.) were collected from the nearest meteorological station to the sampling site (3 km, Figure 1), which can represent the ambient meteorology for the Xiantao site. Trace gases (Supplementary Figure S2), including SO 2 and NO 2 , during the sampling time at Xiantao site were collected from the same place of Xiantao Industrial Park Station, which is a provincial controlled air quality monitoring station. To determine the long-range transport of air masses from different potential regions, 3 days (72 h) air mass back trajectories were calculated every 4 h (00:00, 06:00, 12:00, and 18:00 UTC) at 500 m A.G.L and were cross-checked at 1000 m A.G.L using the hybrid single particle Lagrangian integrated trajectory (HYSPLIT) model (https://www.arl.noaa.gov/hysplit/), provided by the US NOAA. All the air mass back trajectories were clustered into four typical types, including NW (20%), S (24%), WNW-NNE (50%), and NE (7%), by HYSPLIT 4.8 and GIS 10.2 software using a hierarchical clustering method.

Geo-Accumulation Index
To compare the concentration of PM 2.5 -bound metal(loid)s in ambient air with the concentration in the earth's crust, the geo-accumulation index (I Geo ) was used to evaluate contamination levels of metal(loid)s in PM 2.5 aerosol samples [26] as follows:  [27]. The constant of 1.5 was to verify the natural fluctuations of a specific metal(loid) in the environment. The I Geo values of PM 2.5 -bound metal(loid)s were classified as uncontaminated, uncontaminated to moderately contaminated, moderately contaminated, moderately to heavily contaminated, heavily contaminated, heavily to extremely contaminated and extremely contaminated, when I Geo values were less than 0, between 0 and 1, 1 and 2, 2 and 3, 3 and 4, 4 and 5, and greater than 5 [28]. Generally, it may indicate the influence of anthropogenic emissions when I Geo values were higher than 1.

Population Exposure Assessment Model
Population Exposure Dose Due to long-term expose to PM 2.5 -bound metal(loid)s in ambient air through ingestion, inhalation, and dermal contact exposure pathways, it may cause potentially adverse population health risks, including non-carcinogenic and carcinogenic risks. A population exposure assessment model was used to evaluate the health risks of PM 2.5 -bound metal(loid)s in ambient air for humans, including children and adults. According to Technical Guideline for Population Exposure Assessment of Environmental Pollutant (HJ 875-2017, http://www.mee.gov.cn/gkml/hbb/bgg/201711/t20171129_427128. htm), population exposure was defined as average daily population exposure dose (ADPED) of each metal(loid) and was then calculated individually for each metal and exposure pathway, including ingestion (ADPED Ing ), inhalation (ADPED Inh ), and dermal contact (ADPED Der ) as follows: where R Ing and R Inh represent the ingestion and inhalation rate, respectively; ABS represents the dermal absorption factor; EF represents the exposure frequency; ED represents the exposure duration; BW represents the body weight; AT represents the averaging time; PEF represents the particle emission factor; SA represents the skin surface area in contact with air; AF represents the adherence factor for fine particulates to skin; CF represents the conversion factor. The Cr concentration was typically equal to the one-seventh of the total Cr concentration because only Cr(VI) was carcinogenic, while Cr(III) was non-carcinogenic [29]. These variables have been described in previous studies [5,[30][31][32] and are listed in Supplementary Table S1.

Population Health Carcinogenic and Non-Carcinogenic Risk
The carcinogenic risks (CRs) due to exposure to As, Cd, Co, Cr(VI), Ni, and Pb were equal to ADPED multiplied by a specific slope factor (SF). The non-carcinogenic risks due to exposure to As, Cd, Cr(III), Co, Cu, Mn, Ni, Zn, Pb, Ag, Al, Ba, Mo, Sb, Sr, U, and V were evaluated using the hazard quotient (HQ), which were calculated by dividing ADPED into a specific reference dose (RfD) [29,30,33]. The total carcinogenic risk (TCR) and total hazard index (THI) indicated the mixed carcinogenic and non-carcinogenic risk due to exposure to an individual metal(loid) and multiple metal(loid)s through three exposure pathways in ambient air, respectively, which were estimated as follows: Generally, acceptable CR and TCR values ranging from 1 × 10 −6 to 1 × 10 −4 suggest that the PM 2.5 -bound metal(loid)s in ambient air does not cause a carcinogenic risk for population health. The carcinogenic risk is categorized as very low, low, moderate, high, and very high for population health when the CR values were less than 1 × 10 −6 , between 1 × 10 −6 and 1 × 10 −4 , 1 × 10 −4 and 1 × 10 −3 , 1 × 10 −3 and 1 × 10 −1 , and greater than 1 × 10 −1 [34]. There was no significant population health non-carcinogenic risk when the HI values were less than 1. Otherwise, there may be non-carcinogenic risks. When compared to other cities around the world (Table 1) (http://sthjt.hubei.gov.cn/fbjd/zwgk/jcsjfb/hjkq/), respectively, indicating a heavy PM2.5 pollution during wintertime in JHP, central China. The average daily WSIIs, OC, EC, and TE concentrations were 83.6 ± 46.7, 16.4 ± 5.75, 5.56 ± 1.78, and 7.97 ± 2.36 μg m −3 , respectively. The PM2.5 mass concentrations during severe pollution period (277 ± 20.2 μg m −3 ) was approximately 2-3 times higher than the levels during mild, moderate, and heavy pollution periods (95 ± 10.5, 131 ± 8.52, and 175 ± 27.3 μg m −3 , respectively). When compared to other cities around the world (Table 1), the PM2.5 concentration in Xiantao was comparable to the level in Tianjin, China (124 μg m −3 ) [11]. The level was much higher compared  The PM 2.5 concentration reached the highest value on 20 January during severe pollution period. The phenomenon could have been caused by the air masses from the WNW-NNE direction (first comes from the WNW direction and then turn around from the NNE direction to approach the Xiantao site), accounting for 50% of all the trajectories (Figure 3), which could be found by the highest average concentrations of PM 2.5 (153 ± 69.5 µg m −3 ) and WSIIs (104 ± 55.6 µg m −3 ) (Supplementary  Table S2). These air-masses originated from Xiangyang, passed through Suizhou, Xiaogan and Wuhan, and then approached the Xiantao site with a much lower transport speed during severe pollution period (Supplementary Figure S1), which was not conducive to the horizontal transport for ambient air pollutants. In contrast, the PM 2.5 concentration reached the lowest value on 21 January during mild pollution period. This phenomenon could have been partially caused by the air masses from the NE direction, accounting for 7% of all the trajectories (Figure 3), which could be found by the lowest average concentration of PM 2.5 (96.7 ± 19.8 µg m −3 ) (Supplementary Table S2). These air-masses initiated in Anhui Province, then passed through Jiangsu province, and finally entered Hubei Province to approach Xiantao site with a higher transport speed during mild pollution period, which was conducive to the horizontal transport for ambient air pollutants. In addition, rainy and snowy weather (0.30 mm prec.) was observed from midnight of 20 January until the afternoon of 21 January, and thus significantly reduced the PM 2.5 concentration through the wet deposition (Supplementary Figure S1).

PM 2.5 Mass Concentrations and Pollution Characteristics
Atmosphere 2020, 11, x FOR PEER REVIEW 8 of 19 and thus significantly reduced the PM2.5 concentration through the wet deposition (Supplementary Figure S1).  Figure 4a presents the WSIIs concentration, Secondary inorganic aerosol (SNA, including SO4 2− , NO3 − , and NH4 + )/WSIIs and WSIIs/PM2.5 ratios at different pollution levels. SNA accounted for 91.4 ± 4.89% of total WSIIs, ranging from 76.9% to 96.3%. The average WSIIs/PM2.5 ratios were 60.8 ± 9.82%, indicating that WSIIs accounted for the most percentage of PM2.5 mass concentration. It should be noted that the WSIIs concentration increased as the pollution levels increased: mild pollution (54.7 ± 11.0 μg m −3 ) < moderate pollution (76.3 ± 14.7 μg m −3 ) < heavy pollution (119 ± 23.5 μg m −3 ) < severe pollution (208 ± 10.1 μg m −3 ). Moreover, the SNA/WSIIs and WSIIs/PM2.5 ratios increased as pollution levels increased, which indicated the intensification of secondary formation during the winter haze episode in Xiantao, consistent with our previous simultaneously field sampling campaign conducted in Xiangyang [18]. . Moreover, the SNA/WSIIs and WSIIs/PM 2.5 ratios increased as pollution levels increased, which indicated the intensification of secondary formation during the winter haze episode in Xiantao, consistent with our previous simultaneously field sampling campaign conducted in Xiangyang [18]. The acidity of PM 2.5 (Supplementary Text S1) could be roughly evaluated through the balance of anions and cations [42]. The average anions/cations ratios during mild, moderate, heavy, and severe pollution periods were 1.04 ± 0.06, 1.00 ± 0.09, 1.05 ± 0.02, and 1.09 ± 0.03, respectively. All these ratios approached 1 and strong positive correlations (r = 0.93, 0.86, 0.99, and 0.82, respectively, p < 0.05) between anions and cations were observed at different pollution levels in Xiantao (Figure 4b), which indicated that almost all the WSIIs were identified and these WSIIs were important alkaline and acidic species in the PM 2.5 aerosol [43]. Moreover, all these ratios were also greater than 1, which may be due to a deficiency of H + in the calculation and/or to NH 4 + being converted into the gaseous phase [22]. The anions/cations ratios is also a good indicator for studying the acidity of PM 2.5 aerosols [22], which indicated that the PM 2.5 aerosols were acidic during the winter haze episode in Xiantao.

WSIIs
4.89% of total WSIIs, ranging from 76.9% to 96.3%. The average WSIIs/PM2.5 ratios were 60.8 ± 9.82%, indicating that WSIIs accounted for the most percentage of PM2.5 mass concentration. It should be noted that the WSIIs concentration increased as the pollution levels increased: mild pollution (54.7 ± 11.0 μg m −3 ) < moderate pollution (76.3 ± 14.7 μg m −3 ) < heavy pollution (119 ± 23.5 μg m −3 ) < severe pollution (208 ± 10.1 μg m −3 ). Moreover, the SNA/WSIIs and WSIIs/PM2.5 ratios increased as pollution levels increased, which indicated the intensification of secondary formation during the winter haze episode in Xiantao, consistent with our previous simultaneously field sampling campaign conducted in Xiangyang [18]. The acidity of PM2.5 (Supplementary Text S1) could be roughly evaluated through the balance of anions and cations [42]. The average anions/cations ratios during mild, moderate, heavy, and severe The NO 3 − /SO 4 2− ratio has been generally used as an indicator to evaluate the relative contribution of local stationary sources (e.g., coal-fired power plant) and local mobile sources (e.g., vehicle exhaust) to sulfur and nitrogen in the ambient air [44,45]. In this study, the average NO 3 − /SO 4 2− ratios during mild, moderate, heavy, and severe pollution periods were 2.35 ± 0.78, 2.39 ± 1.08, 2.31 ± 0.62, and 1.70 ± 0.15, respectively, which indicated that local mobile sources contributed more to PM 2.5 compared to local stationary sources during the winter haze episode. Generally, the sulfur oxidation ratio (SOR) and nitrogen oxidation ratio (NOR) values (Supplementary Text S1) were less than 0.10 for primary pollutants, whereas they were greater than 0.10 when gaseous precursors (SO 2 and NO 2 ) were photochemically oxidized in the ambient air [46,47]. The SOR and NOR values during mild, moderate, heavy, and severe pollution periods were 0.38 ± 0.25 and 0.28 ± 0.09, 0.45 ± 0.27 and 0.30 ± 0.09, 0.65 ± 0.11 and 0.47 ± 0.05, and 0.66 ± 0.04 and 0.47 ± 0.05, respectively (Supplementary Figure S2), which indicated that the significantly secondary transformations of SO 2 to SO 4 2− and NO 2 to NO 3 − were observed during the winter haze episode.

CAs
The average OC and EC concentrations during the winter haze episode were 16.4 ± 5.74 and 5.56 ± 1.78 µg m −3 , respectively, which accounted for 74.3 The OC/EC ratios increased as the pollution levels increased from mild to moderate pollution, whereas the ratios decreased as the pollution levels increased from moderate to severe pollution, thus reached the highest levels during moderate pollution period ( Figure 5, which indicated that more SOA were formed during moderate pollution period compared to the other three pollution levels. Moreover, the average OC/EC ratios at four pollution levels (3.06 ± 0.85, 3.42 ± 0.87, 2.81 ± 0.39, and 2.42 ± 0.46, respectively) were higher than 2, indicating that secondary organic aerosol (SOA) were formed during the winter haze episode. OC was positively correlated with EC (r = 0.71, p < 0.05), which indicated that there may be a certain same emission source during the winter haze episode. 32.6% of OC/EC ratios ranged from 1.5 to 2.5 (r = 0.95, p < 0.05) (Supplementary Figure S3), it may indicate the contribution of intensive vehicle exhaust to CAs, whereas 67.4% of the OC/EC ratios were higher than 2.5 (r = 0.83, p < 0.05), which might suggest a rapid transformation of secondary organic carbon (SOC) precursors such as VOCs [48].

TE
The TE concentrations during mild, moderate, heavy, and severe pollution periods were 6.64 ± 2.18, 9.26 ± 1.89, 9.59 ± 1.70, and 6.52 ± 2.30 μg m −3 , respectively. The proportion of TE to PM2.5 was 6.04 ± 1.79%. The sum of Sb, Al, Si, Zn, Pb, Co, Cr, Mn, Sn, and Cu concentrations accounted for 95.7% of total TE. Figure S4a shows the dynamic variations of TE mass concentrations at the four ambient air quality levels. The Si, Mn, Ni, Cu, Zn, As, Se, Sr, Ag, Cd, Sn, Ba, Pb, and Bi concentrations increased as the pollution levels increased from mild to moderate pollution, whereas these concentrations decreased as the pollution levels from moderate to severe pollution. The Al, Cr, Co, and Sb concentrations increased as the pollution levels increased from mild to heavy pollution, whereas these concentrations decreased as the pollution levels from heavy to severe pollution. The other TE concentrations fluctuated at four pollution levels. The average TE concentration reached the higher value during 18 January to 23 January. The phenomenon could have been partially caused by the air masses from the NE and WNW-NNE directions, accounting for 7% (8.93 ± 1.91 μg m −3 ) and 50% (8.54 ± 2.31 μg m −3 ) of all the trajectories (Figure 3 and Supplementary Table S2), respectively. The former air-masses originated from Anhui Province, passed through Shandong Province, and then turn around to approach the Xiantao site with a much higher transport speed during mild pollution period (Supplementary Figure S1) due to long-range transport, which may carry lots of ambient air pollutants to the region. The latter air-masses originated from originated from Xiangyang, passed through Suizhou, Xiaogan, and Wuhan, and then approached the Xiantao site with a much lower transport speed due to combined effect of long-range transport and local emissions (Supplementary Figure S1). Moreover, the rains and snowy weather occurred in the evening of 23 January may increase the hygroscopic growth. In contrast, the average TE concentration reached the lower value during 13 January to 17 January. This phenomenon could have been partially caused by the air masses from the S and NW directions, accounting for 24% (7.59 ± 2.76 μg m −3 ) and 20% (6.68 ± 2.10 μg m −3 ) of all the trajectories (Figure 3 and Supplementary Table S2). The former air-masses initiated in Shiyan, then passed through Jingmen, and finally approached the Xiantao site with a lower transport speed (Supplementary Figure S1). The latter airmasses originated from originated from Xiantao, passed through Yueyang and Changsha, and then turn around to approach the Xiantao site with a much higher transport speed (Supplementary Figure S1). Both may be due to the combined effect of longrange transport and local emissions.
The Pb/Cd ratio is generally considered as an indicator to identify the sources of TE [49]. In our

TE
The TE concentrations during mild, moderate, heavy, and severe pollution periods were 6.64 ± 2.18, 9.26 ± 1.89, 9.59 ± 1.70, and 6.52 ± 2.30 µg m −3 , respectively. The proportion of TE to PM 2.5 was 6.04 ± 1.79%. The sum of Sb, Al, Si, Zn, Pb, Co, Cr, Mn, Sn, and Cu concentrations accounted for 95.7% of total TE. Figure S4a shows the dynamic variations of TE mass concentrations at the four ambient air quality levels. The Si, Mn, Ni, Cu, Zn, As, Se, Sr, Ag, Cd, Sn, Ba, Pb, and Bi concentrations increased as the pollution levels increased from mild to moderate pollution, whereas these concentrations decreased as the pollution levels from moderate to severe pollution. The Al, Cr, Co, and Sb concentrations increased as the pollution levels increased from mild to heavy pollution, whereas these concentrations decreased as the pollution levels from heavy to severe pollution. The other TE concentrations fluctuated at four pollution levels. The average TE concentration reached the higher value during 18 January to 23 January. The phenomenon could have been partially caused by the air masses from the NE and WNW-NNE directions, accounting for 7% (8.93 ± 1.91 µg m −3 ) and 50% (8.54 ± 2.31 µg m −3 ) of all the trajectories (Figure 3 and Supplementary Table S2), respectively. The former air-masses originated from Anhui Province, passed through Shandong Province, and then turn around to approach the Xiantao site with a much higher transport speed during mild pollution period (Supplementary Figure S1) due to long-range transport, which may carry lots of ambient air pollutants to the region. The latter air-masses originated from originated from Xiangyang, passed through Suizhou, Xiaogan, and Wuhan, and then approached the Xiantao site with a much lower transport speed due to combined effect of long-range transport and local emissions (Supplementary Figure S1). Moreover, the rains and snowy weather occurred in the evening of 23 January may increase the hygroscopic growth. In contrast, the average TE concentration reached the lower value during 13 January to 17 January. This phenomenon could have been partially caused by the air masses from the S and NW directions, accounting for 24% (7.59 ± 2.76 µg m −3 ) and 20% (6.68 ± 2.10 µg m −3 ) of all the trajectories (Figure 3 and Supplementary Table S2). The former air-masses initiated in Shiyan, then passed through Jingmen, and finally approached the Xiantao site with a lower transport speed (Supplementary Figure S1). The latter airmasses originated from originated from Xiantao, passed through Yueyang and Changsha, and then turn around to approach the Xiantao site with a much higher transport speed (Supplementary Figure S1). Both may be due to the combined effect of long-range transport and local emissions.
The Pb/Cd ratio is generally considered as an indicator to identify the sources of TE [49]. In our study, the Pb/Cd ratios during mild, moderate, heavy, and severe pollution periods were 35.9 ± 33.7, 34.7 ± 9.14, 40.5 ± 7.94, and 44.4 ± 2.89, respectively. These ratios were close to the ratios for anthropogenic emissions (46), indicating that Pb and Cd may be emitted from anthropogenic emissions. Moreover, the V/Ni ratio is commonly used to distinguish industrial emissions (0.7-1.9) and shipping emissions (2.1-3.1) [49,50]. The V/Ni ratios were found to be within the range of 0.7-1.9 during mild and severe pollution periods (1.80 and 1.59, respectively), which indicated that V and Ni may come from industrial emissions during these two periods. This might be mainly affected by the airmasses from the NE and WNW-NNE directions due to combined effect of long-range transport and local emissions as mentioned above. In addition, Pb was positively correlated with Cd (r = 0.96, p < 0.05) during the moderate pollution period, which indicated that the Pb and Cd may have same emission sources during this period. Similar results for V and Ni (r = 0.69, p < 0.05) were observed during heavy pollution period.  (Figure 6a), which indicated that PM 2.5 -bound metal(loid)s were affected by natural and anthropogenic activities. The EF values of U, Ba, Sr, Mn, Li, and V were within the range of 2-10, which suggested that these metal(loid)s were mixed affected by crustal sources and anthropogenic emissions. In contrast, the EF values of most metal(loid)s, including Be, Cr, Ni, Cu, Mo, Tl, Zn, Ag, As, Pb, Co, Bi, Cd, Sn, Se, and Sb were significantly higher than 10, indicating the predominate influence from anthropogenic emissions on these metal(loid)s (Supplementary Figure S5). As shown in Figure 6a, the I Geo values of Li, Al, Th, U, Ba, Sr, V, and Mn were lower than 0, indicating that these metal(loid)s were uncontaminated. The I Geo values of Be ranged from 1 to 2, which indicated that there was moderate contamination for Be. The metals of Cr and Ni were moderately to heavily contaminated because the I Geo values of them ranged from 2 to 3. Mo was heavily contaminated because the I Geo values ranged from 3 to 4. Cu and Tl were heavily to extremely contaminated because the I Geo values of them ranged from 4 to 5. What calls for special attention is the extreme contamination for Zn, As, Ag, Pb, Co, Bi, Cd, Sn, Se, and Sb because the I Geo values of these metals were higher than 5, which may have caused population health carcinogenic or non-carcinogenic risk to some extent. It should be noted that EF values were positively correlated with I Geo values at the four pollution levels (r = 0.99, p < 0.01) (Supplementary Figure S4b), which indicated that the dynamic variation trends of EF and I Geo values remained consistent during the winter haze episode.

Chemical Mass Closure
The chemical mass closure method (Supplementary Text S3) was used to better understand PM 2.5 chemical compositions, including mineral dust (MD), trace element oxides (TEO), OM, EC, SNA, Cl − and unidentified matters (UM) [51][52][53]. Figure 7a-e presents the chemical mass closure for PM 2.5 at different pollution levels at Xiantao site. As shown in Figure 7a, SNA occupied the largest proportion (58.6%) of PM 2.5 , followed by OM (19.9%), MD (4.31%), EC (4.21%), TEO (4.21%), and Cl − (2.76%). The contributions of MD, TEO, OM, EC, and Cl − to PM 2.5 decreased as the pollution levels increased, whereas the contribution of SNA to PM 2.5 increased as pollution levels increased (Figure 7b-e). The results may indicate the contribution variations of emission sources to PM 2.5 chemical compositions. On average, only a small proportion of PM 2.5 (UM, 6.02%) cannot be identified in this study as well as each pollution level, which indicated that the chemical masses of PM 2.5 aerosol samples were balanced during the winter haze episode in Xiantao.
the IGeo values of them ranged from 4 to 5. What calls for special attention is the extreme contamination for Zn, As, Ag, Pb, Co, Bi, Cd, Sn, Se, and Sb because the IGeo values of these metals were higher than 5, which may have caused population health carcinogenic or non-carcinogenic risk to some extent. It should be noted that EF values were positively correlated with IGeo values at the four pollution levels (r = 0.99, p < 0.01) (Supplementary Figure S4b), which indicated that the dynamic variation trends of EF and IGeo values remained consistent during the winter haze episode.

Chemical Mass Closure
The chemical mass closure method (Supplementary Text S3) was used to better understand PM2.5 chemical compositions, including mineral dust (MD), trace element oxides (TEO), OM, EC, SNA, Cl − and unidentified matters (UM) [51][52][53]. Figure 7a-e presents the chemical mass closure for PM2.5 at different pollution levels at Xiantao site. As shown in Figure 7a, SNA occupied the largest proportion (58.6%) of PM2.5, followed by OM (19.9%), MD (4.31%), EC (4.21%), TEO (4.21%), and Cl -(2.76%). The contributions of MD, TEO, OM, EC, and Cl − to PM2.5 decreased as the pollution levels increased, whereas the contribution of SNA to PM2.5 increased as pollution levels increased (Figure 7b-e). The   Table 2 presents ADPED values of PM2.5-bound metal(loid)s through ingestion, inhalation, and dermal contact pathways for children and adults during the winter haze episode. Sb showed the maximum ADPED values of PM2.5-bound metal(loid)s through these three different exposure pathways for children (3.66 × 10 −1 ) and adults (5.96 × 10 −2 ), whereas U showed the minimum ADPED values of PM2.5-bound metal(loid)s through these three different exposure pathways for children (3.64 × 10 −5 ) and adults (5.92 × 10 −6 ), respectively. Ingestion is found to be the dominant exposure pathway of PM2.5-bound metal(loid)s for children and adults in Xiantao, followed by inhalation and dermal contact, which agreed with previous studies in Kanpur, India [54] and Xiangyang, central China [18]. As shown in Supplementary Table S3, the ADPED values of PM2.5-bound metal(loid)s through ingestion pathway were 1-4 orders of magnitude higher for children and adults at the four pollution levels compared to inhalation and dermal contact pathways. It should be noted that ADPED values of PM2.5-bound metal(loid)s through these three exposure pathways for children were 5.74-fold higher than the values for adults, which indicated that children were more likely to be exposed to PM2.5-bound metal(loid)s than adults during the winter haze episode in Xiantao.   Table 2 presents ADPED values of PM 2.5 -bound metal(loid)s through ingestion, inhalation, and dermal contact pathways for children and adults during the winter haze episode. Sb showed the maximum ADPED values of PM 2.5 -bound metal(loid)s through these three different exposure pathways for children (3.66 × 10 −1 ) and adults (5.96 × 10 −2 ), whereas U showed the minimum ADPED values of PM 2.5 -bound metal(loid)s through these three different exposure pathways for children (3.64 × 10 −5 ) and adults (5.92 × 10 −6 ), respectively. Ingestion is found to be the dominant exposure pathway of PM 2.5 -bound metal(loid)s for children and adults in Xiantao, followed by inhalation and dermal contact, which agreed with previous studies in Kanpur, India [54] and Xiangyang, central China [18]. As shown in Supplementary Table S3, the ADPED values of PM 2.5 -bound metal(loid)s through ingestion pathway were 1-4 orders of magnitude higher for children and adults at the four pollution levels compared to inhalation and dermal contact pathways. It should be noted that ADPED values of PM 2.5 -bound metal(loid)s through these three exposure pathways for children were 5.74-fold higher than the values for adults, which indicated that children were more likely to be exposed to PM 2.5 -bound metal(loid)s than adults during the winter haze episode in Xiantao.  Figure 6b and Supplementary Figure S6a-d presents the population health carcinogenic risk and non-carcinogenic risk due to personal exposure to PM 2.5 -bound metal(loid)s at different pollution levels for children and adults, respectively. The highest carcinogenic risk for children and adults during the winter haze episode was As (1.49 × 10 −3 and 2.59 × 10 −3 ), followed by Pb (7.40 × 10 −4 and 3.17 × 10 −4 ), Cr(VI) (6.37 × 10 −4 and 2.72 × 10 −4 ), Co (1.91 × 10 −6 and 6.22 × 10 −7 ), Cd (4.35 × 10 −8 and 1.42 × 10 −8 ), and Ni (4.22 × 10 −8 and 1.37 × 10 −8 ). The CR values through ingestion, inhalation, and dermal contact exposure pathways widely ranged from 4.22 × 10 −8 to 1.49 × 10 −3 for children and 1.37 × 10 −8 to 2.59 × 10 −3 for adults. Moreover, the TCR values were approximately 1.12-fold higher for adults (3.22 × 10 −3 ) than for children (2.87 × 10 −3 ), which indicated that PM 2.5 -bound metal(loid)s may pose much higher carcinogenic risk for adults compared to children during the winter haze episode in Xiantao (Figure 6b). The CRs value for As is 1~5 orders of magnitude higher than the values for other PM 2.5 -bound metal(loid)s for children and adults. The TCR value is the sum of CR values for As, Cd, Co, Cr(VI), Ni, and Pb, which mainly depended on the CR value for As. The CR value for As for children (1.49 × 10 −3 ) is lower than the value for adults (2.63 × 10 −3 ), although the CR values for other PM 2.5 -bound metal(loid)s for children were higher than the values for adults. Therefore, the TCR values were approximately 1.12-fold higher for adults than for children. The TCR values through the three exposure pathways for children and adults during moderate pollution period (1.86 × 10 −3 and 3.28 × 10 −3 ) were the highest, followed by mild pollution (1.55 × 10 −3 and 2.73 × 10 −3 ), heavy pollution (1.20 × 10 −3 and 2.11 × 10 −3 ), and severe pollution periods (3.62 × 10 −4 and 6.38 × 10 −4 ) (Supplementary Figure S6a-d). The TCR values through the three exposure pathways for children and adults in Xiantao were lower than the values during the four pollution periods (mild pollution: 1.64 × 10 −1 , 2.21 × 10 −1 ; moderate pollution: 2.71 × 10 −3 , 3.77 × 10 −3 ; heavy pollution: 2.27 × 10 −3 , 2.30 × 10 −3 ; severe pollution: 1.48 × 10 −3 , 1.65 × 10 −3 ) in Xiangyang [18]. Among all the PM 2.5 -bound metal(loid)s, the TCR of As, Cr(VI), and Pb for children and adults were at the moderate or high levels, whereas the TCR of Cd, Ni and Co for children and adults were below moderate levels, except for the low carcinogenic risk level of Co for children.

Population Health Risks
The highest non-carcinogenic risk during the winter haze episode for children and adults was Sb (2.43 × 10 3 and 2.80 × 10 3 ), followed by As (38.7 and 17.1), Cr(III) (13.6 and 15.6), Pb (10.9 and 4.54), and V (1.34 and 1.85). In addition, Co (2.01) and Cd (1.48) may only pose non-carcinogenic risks for children during the winter haze episode. The HQ values through the three exposure pathways widely ranged from 4.13 × 10 −3 to 2.43 × 10 3 for children and 1.45 × 10 −3 to 2.80 × 10 3 for adults. Moreover, the THI values were approximately 1.14-fold higher for adults (2.84 × 10 3 ) than for children (2.50 × 10 3 ), which indicated that PM 2.5 -bound metal(loid)s may pose much higher non-carcinogenic risk for adults compared to children during the winter haze episode in Xiantao (Figure 6b). The THI values through the three exposure pathways for children and adults decreased as pollution levels increased: mild pollution (2.42 × 10 3 and 2.74 × 10 3 ) > moderate pollution period (2.07 × 10 3 and 2.32 × 10 3 ) > heavy pollution (1.87 × 10 3 and 2.12 × 10 3 ) > severe pollution (1.13 × 10 3 and 1.29 × 10 3 ) (Supplementary Figure S6a-d). The THI values through the three exposure pathways for children and adults in Xiantao were lower than the values during mild (4.22 × 10 3 , 4.84 × 10 3 ) and moderate (2.49 × 10 3 , 2.81 × 10 3 ) pollution periods in Xiangyang, whereas the THI values for children and adults were higher than the values during heavy (1.65 × 10 3 , 1.86 × 10 3 ) and severe (1.05 × 10 3 , 1.19 × 10 3 ) pollution periods in Xiangyang [18]. The total non-carcinogenic risk of Sb for children and adults through the three exposure pathways were 3-6 orders of magnitude higher at the four pollution levels compared to other PM 2.5 -bound metal(loid)s. It should be noted that As, Cr, and Pb may pose carcinogenic and non-carcinogenic risks for children and adults, whereas Sb and V may only pose non-carcinogenic risks for children and adults. Moreover, the population health risks of PM 2.5 -bound metal(loid)s may not depend on the pollution levels but depend on the PM 2.5 -bound metal(loid)s concentrations. Other CAs, such as polycyclic aromatic hydrocarbons (PAHs) could also pose population health risks and needed to be addressed in the future. The government and the public should pay more attention to the population health risks posed by PM 2.5 -bound metal(loid)s during the winter haze episode in JHP, central China.

Conclusions
In this study, a total of 43 pairs of PM 2.5 samples were effectively collected to determine the pollution characteristics, chemical compositions, and population health risks during 13-23 January 2018 in Xiantao in JHP, central China. During the sampling period, Xiantao experienced an eleven-day long-lasting haze episode with an average daily PM 2.5 mass concentration of 132 ± 56.0 µg m −3 . The level was higher than the levels (123 µg m −3 ) in January from 2012 to 2017, and approximately 1.76 and 2.49 times than the Grade II of NAAQS the "normal" levels (53 µg m −3 ) without pollution events there in January 2018, respectively, indicating a heavy PM 2.5 pollution during wintertime in JHP, central China. The PM 2.5 typical chemical compositions for the average WSIIs, OC, EC, and TE concentrations were 83.6 ± 46.7, 16.4 ± 5.75, 5.56 ± 1.78, and 7.97 ± 2.36 µg m −3 , respectively. The higher PM 2.5 levels during severe pollution period were dominated by the air masses from the WNW-NNE direction, whereas the lower PM 2.5 concentrations during other pollution periods were mainly affected by the air masses from the NE, S, and NW directions.
The anions/cations ratios indicated that the PM 2.5 aerosols were acidic during the winter haze episode in Xiantao. The NO 3 − /SO 4 2− ratio indicated that the mobile sources contributed more to PM 2.5 compared to the stationary sources. The OC/EC ratios indicated a mixed contribution of intensive vehicle exhaust and secondary formations. The Pb/Cd and ratios indicated that Pb and Cd may be emitted from anthropogenic emissions. The V/Ni ratios indicated that V and Ni may come from the industrial emissions during mild and severe pollution periods. The EF values were positively correlated with the I Geo values at the four pollution levels. Ingestion is found to be the dominant exposure pathway of PM 2.5 -bound metal(loid)s for children and adults in Xiantao, followed by inhalation and dermal contact. As, Cr, and Pb may pose carcinogenic and non-carcinogenic risks for children and adults, whereas Sb and V may only pose non-carcinogenic risks for children and adults. PM 2.5 -bound metal(loid)s may pose much higher population health risks for adults compared to children during the winter haze episode in Xiantao. Moreover, the population health risks of PM 2.5 -bound metal(loid)s may not depend on the pollution levels but depend on the PM 2.5 -bound metal(loid)s concentrations.
Other CAs, such as polycyclic aromatic hydrocarbons (PAHs) could also pose population health risks and need to be addressed in the future. The government and the public should pay more attention to the population health risks posed by PM 2.5 -bound metal(loid)s during a winter haze episode in JHP, central China.  Table S3: The ADPED (mg kg −1 day −1 ) of PM 2.5 -bound metal(loid)s through ingestion, inhalation, and dermal contact pathways for children and adults at different pollution levels, Figure S1: Time series of hourly meteorological parameter, including wind direction (WD), wind speed (WS), temperature (Temp.) relative humidity (RH), precipitation (Prec.), and visibility (Vis.) at Xiantao site in January 2018, Figure S2: Time series of trace gases (SO 2 and NO 2 ), SOR, and NOR during the sampling time at Xiantao site, Figure S3: The OC/EC ratios during the winter haze episode at Xiantao site, Figure S4: The (a) TE concentration and (b) correlations between EF and I Geo at different pollution levels, Figure S3: The contamination levels for PM 2.5 -bound metal(loid)s at different pollution levels: (a) mild pollution, (b) moderate pollution, (c) heavy pollution, and (d) severe pollution, Figure  S4: The population health carcinogenic risk and non-carcinogenic risk due to personal exposure to PM 2.5 -bound metal(loid)s for children and adults, respectively, at different pollution levels: