PM 2.5 -Bound Heavy Metals in Southwestern China: Characterization, Sources, and Health Risks

: The health risks of PM 2.5 -bound heavy metals have attracted extensive attention recently. In order to evaluate those deleterious effects on human health more accurately, and to propose proper measures to reduce health risks of air pollution, the conduction of a source-speciﬁc health risk assessment is necessary. Based on daily collected PM 2.5 samples at different functional sites during winter 2019 in a megacity Chongqing, China, combining source apportionment results from PMF and health risk assessment from the U.S. EPA, the source-speciﬁc health risks from PM 2.5 -bound heavy metals were given. Six types of PM 2.5 sources have been identiﬁed, coal burning (25.5%), motor vehicles (22.8%), industrial emissions (20.5%), biomass burning (15.9%), dust (7.8%), and ship emissions (7.5%). Results showed that the total hazard quotient (HQ) was 0.32 and the total carcinogenic risks (CR) were 2.09 × 10 − 6 for children and 8.36 × 10 − 6 for adults, implying certain risks for local residents. Industrial emissions related with Cr posed both the highest carcinogenic risk and noncarcinogenic risk (contributing 25% CR and 36% HQ). Coal combustion (associated with Cr, As, and Mn) contributed 15.46% CR and 20.64% HQ, while biomass burning and motor vehicles shared 19.99% and 19.05% of the total CR, respectively. This work indicated that health risks of air pollution sources were the combined effects of the source contribution and chemical components. In order to control the health risks of PM 2.5 to the local residents, the priority of targeted emission sources should be adopted for industrial emissions, biomass burning, vehicle emissions, and coal combustion sources. 6 January 2019 to 28 January 2019, and collected continuously for 23 h each time from 10:00 to 09:00 the next day. A total of 114 PM 2.5 samples and 15 blank samples were obtained. The collected samples were stored in a clean refrigerator environment of − 18 ◦ C to avoid contamination and volatilization.


Introduction
The components of atmospheric fine particulate matters (PM 2.5 ) are very complicated, including carbonaceous components (organic carbon, OC; elemental carbon, EC), major water-soluble inorganic ions, and trace elements [1,2]. Inhaled PM 2.5 can settle in the lungs and blood, causing harmful effects on human health, respiratory diseases such as bronchitis and asthma, as well as heart disease, which are more likely to occur or worsen in areas with high levels of PM 2.5 pollution [3][4][5]. Some PM 2.5 components can be highly deleterious at low concentrations, such as heavy metals (HMs), which have been shown to cause dysfunctions and carcinogenicity even though only at the level of nanograms per cubic meter in the air [6,7]. The inhaled PM 2.5 is an important approach for HMs get into which is a large city in southwest China and has a population of 32 million (National Bureau of Statistics, China. 1 November 2020). Jiangjin belongs to the eastern part of the Sichuan Basin (SCB; 25 •~3 5 • N, 95 •~1 10 • E); it is obviously influenced by pollution in the basin range and can represent the corresponding regional air quality. In this work, the characteristics and sources of PM 2.5 -bound HMs (Cr, As, Pb, Cu, Zn, Ni, V, Mn) were studied and their health risks were evaluated. Although As is a nonmetal element, it has similar health effects with HMs; so, As is discussed in this work. The research conclusions can provide scientific support for environmental administration of the government to establish relevant policies for reducing atmospheric pollution emissions with high health risks.

PM 2.5 Sampling and Chemistry Analysis
Jiangjin District is located southwest of Chongqing and is a developed industrial area about 40 km away from the city center. This project examined 5 sampling sites, among which there were two central urban sites, Jiangjin Middle school (S1) and Yuanshuai Square (S2); one suburban site, Binjiang New City (S3); and two industrial sites, De-Gan industrial zone (S4) and Shuangfu industrial zone (S5). The Yangtze River flows through Jiangjin District for 127 km and includes 135 ferry piers, there are large cargo terminals and ports in the vicinities of four industrial zones (Degan, Luohuang, Baisha, and Shuangfu). The sampling sites' information is shown in Figure 1. BGI OMNI samplers (5 L/min) were used in the urban and suburban sites (S1, S2, S3), Dandong Bettersize automatic sampler (16.67 L/min) was employed in the two industrial sites (S4, S5). All PM 2.5 samplers were placed on the roof of the building, which was about 9-15 m above the ground. PM 2.5 samples were collected by Teflon filter membrane (Whatman Corp., 47 mm) from 6 January 2019 to 28 January 2019, and collected continuously for 23 h each time from 10:00 to 09:00 the next day. A total of 114 PM 2.5 samples and 15 blank samples were obtained. The collected samples were stored in a clean refrigerator environment of −18 • C to avoid contamination and volatilization. pollution emission sources. The research area, Jiangjin District, is located in Chongqing, which is a large city in southwest China and has a population of 32 million (National Bureau of Statistics, China. November 1st, 2020). Jiangjin belongs to the eastern part of the Sichuan Basin (SCB; 25°~35° N, 95°~110° E); it is obviously influenced by pollution in the basin range and can represent the corresponding regional air quality. In this work, the characteristics and sources of PM2.5-bound HMs (Cr, As, Pb, Cu, Zn, Ni, V, Mn) were studied and their health risks were evaluated. Although As is a nonmetal element, it has similar health effects with HMs; so, As is discussed in this work. The research conclusions can provide scientific support for environmental administration of the government to establish relevant policies for reducing atmospheric pollution emissions with high health risks.

PM2.5 Sampling and Chemistry Analysis
Jiangjin District is located southwest of Chongqing and is a developed industrial area about 40 km away from the city center. This project examined 5 sampling sites, among which there were two central urban sites, Jiangjin Middle school (S1) and Yuanshuai Square (S2); one suburban site, Binjiang New City (S3); and two industrial sites, De-Gan industrial zone (S4) and Shuangfu industrial zone (S5). The Yangtze River flows through Jiangjin District for 127 km and includes 135 ferry piers, there are large cargo terminals and ports in the vicinities of four industrial zones (Degan, Luohuang, Baisha, and Shuangfu). The sampling sites' information is shown in Figure 1. BGI OMNI samplers (5 L/min) were used in the urban and suburban sites (S1, S2, S3), Dandong Bettersize automatic sampler (16.67 L/min) was employed in the two industrial sites (S4, S5). All PM2.5 samplers were placed on the roof of the building, which was about 9-15 m above the ground. PM2.5 samples were collected by Teflon filter membrane (Whatman Corp., 47 mm) from 6 January 2019 to 28 January 2019, and collected continuously for 23 h each time from 10:00 to 09:00 the next day. A total of 114 PM2.5 samples and 15 blank samples were obtained. The collected samples were stored in a clean refrigerator environment of −18 °C to avoid contamination and volatilization. In this study, a weighing method was adopted to measure the mass of PM2.5 samples. Before and after sampling, the Teflon filter membranes were placed in a relatively constant temperature (20-23 °C) and humidity (45-50%) environment for 48 h to reach equilibrium. Then, a microbalance (Sartorius, ME5-F, Goettingen, Germany) was used to weigh the filer membrane. The atmospheric PM2.5 concentration can be obtained In this study, a weighing method was adopted to measure the mass of PM 2.5 samples. Before and after sampling, the Teflon filter membranes were placed in a relatively constant temperature (20-23 • C) and humidity (45-50%) environment for 48 h to reach equilibrium. Then, a microbalance (Sartorius, ME5-F, Goettingen, Germany) was used to weigh the filer membrane. The atmospheric PM 2.5 concentration can be obtained according to the

Health Risk Estimation of PM 2.5 -Bound Heavy Metals
The carcinogenic and noncarcinogenic risks by PM 2.5 -bound HMs via inhalation were calculated by adopting the U.S. Environmental Protection Agency (US EPA) human health risk assessment models (US EPA, 1989) [27]. The adverse effects of heavy metals were determined by their bioavailable toxicity, which were influenced by chemical form, solubility, active organic location, aerosol surface property, and so on. The method included calculation of the deposition fraction of PM 2.5 that can penetrate in lungs [28], the bioavailable concentration of each HM during the monitoring periods [29], the exposure concentration of PM 2.5 -bound HMs (EC), hazard quotient (HQ, noncarcinogenic health risk), and carcinogenic risk (CR). This study referred to the bioavailable ratios of PM 2.5 -bound HMs in another research in China-the detail can be seen in Huang et al. [29]. The sensitive residents were divided into two groups, adults and children, the inhalation reference concentration (RfC, mg m −3 ) and inhalation unit risk (IUR, (µg m −3 ) −1 ) were obtained from the Regional Screening Level (RSL) Resident Ambient Air Table (US EPA, 2021) [30]. The 8 HMs that induce noncarcinogenic risks are V, Cr, Mn, Ni, and As, those showing carcinogenic risks are V, Cr, Ni, As, and Pb. The carcinogenic health risks of each air pollution source were assessed by the sum of CR of carcinogenic HMs contributed by that source (CR s ), as were the noncarcinogenic health risks of that source (HQ s ). The CR s and HQ s are calculated by Equations (4) and (5), respectively.
CR i = IUR i × EC (i = V, Cr( VI ), Ni, As, Pb) HQ i = EC RfC × 1000 µg mg −1 (i = V, Cr( VI ), Mn, Ni, As) HQ sj = ∑ (HQ i × RC ij ) (i = V, Cr( VI ), Mn, Ni, As) where C b is the average bioavailable concentrations of particulate elements (referring to the bioavailable index of heavy metals in Huang et al. [29]); E i is the deposition fraction of PM 2.5 that can penetrate in lungs [28] (Volckens and Leith, 2003); ET is the exposure time (24 h/day); EF is the exposure frequency (180 day/year); ED is the exposure duration (6 years for children and 24 years for adults); ATn is the averaging time (for noncarcinogens ATn = ED × 365 day/year × 24 h/day; for carcinogens ATn = 70 year × 365 day/year × 24 h/day); RC ij is the contribution rate of air pollution source j to element i in PM 2.5 , which was provided by PM 2.5 source apportionment results; CR sj and HQ sj are the CR and HQ of source j.

Pollution Characteristics of Heavy Metals in PM 2.5
The mean concentration of PM 2.5 and HMs in PM 2.5 during the observation period were shown in Table 1. The average PM 2.5 concentration was 97.06 µg/m 3 , varied at each site from 84.56 ± 28.39 to 106.60 ± 36.11 µg/m 3 ; of the 114 samples, 83 had PM 2.5 concentrations above 75 µg/m 3 (the second level of Chinese Ambient Air Quality Standard, GB3095-2012). The highest PM 2.5 concentration was 168.77 µg/m 3 , 2.25 times the secondlevel limit. The total concentration of 8 HMs was 220.46 ng/m 3 ranged from 163.30 to 329.53 ng/m 3 at five sites on average, which accounted for 0.19% to 0.33% of PM 2.5 . Zn, Mn, and Pb were the three HMs with high concentrations, accounting for 42.74%, 26.60%, and 17.21% of the total HMs, respectively. The PM 2.5 and its chemical compositions of urban areas in Chongqing and Chengdu have been reported by Chen et al. [31] and Tao et al. [32], respectively. In their research, the concentrations of Zn, Mn, and Pb were also higher than those of Cu, As, Cr, Ni, and V. It has also been reported that Zn and Pb in PM 2.5 were higher than other HMs during winter in northern China such as Baoding and Beijing [10,11,33]. When compared with the results in 2012, obtained at an urban area of Chongqing by Chen et al. [31], the HMs' concentrations have decreased, except for Mn and As, which somehow indicate that the air quality in this region has been getting better.
The PM 2.5 concentrations generally showed urban > industrial zone > suburban, while PM 2.5 -bound HMs were industrial zone > urban > suburban. HMs at the southern industrial site (S4, De-Gan Industrial zone) were the highest, accounting for 0.33% of PM 2.5 , especially Mn, Cu, Ni, and V, maximally 4.63, 2.16, 3.37, and 3.33 times of that at other sampling sites, respectively. However, HM concentrations at the northern industrial site (S5, Shuangfu industrial zone) were similar to those at urban sites. A possible reason was that the Shuangfu industrial zone was only about 40 km southwest of the downtown area of Chongqing. Under the dominant northeast wind, S5 was significantly influenced by the air pollution from the upwind downtown area.
As shown in Figure 2, among all heavy metals, the time trend of Pb and PM 2.5 has the best correlation, and the correlation coefficients of the five sites are 0.8~0.91. According to the PMF results of PM 2.5 and PM 2.5 -bound HMs, the proportions of Pb in source profiles of coal burning, motor vehicles, and industrial emission were quite close to each other, and these three sources contributed about 69% PM 2.5 and 58% of total Pb in PM 2.5 , which can explain the high consistency between PM 2.5 and Pb. Except for S4, the correlation coefficients between Cu and PM 2.5 were 0.66~0.80. The correlation between as and PM 2.5 showed obvious differences between sites, poorly correlated in the central urban sites (0.37 and 0.34 at S1 and S2, respectively), while good in the suburbs and industrial sites (0.64, 0.70, and 0.77 in S3, S4, and S5, respectively). The correlation between Zn and PM 2.5 was quite similar to that of As. The concentrations of V at the five sites had relatively small changes, indicating that it may have stable local sources like industrial emissions and ship was quite similar to that of As. The concentrations of V at the five sites had relatively small changes, indicating that it may have stable local sources like industrial emissions and ship emissions. Generally, the different variation trend of PM2.5-bound HMs among the five sites suggested differences in PM2.5 sources.  Table 2 lists the standard limits for heavy metals defined by China, WHO, the EU, and other countries. On average, As and Cr were the two HMs exceeding the standard limits of GB3095-2012 (As was 26% higher and Cr was 170 times higher) and WHO (As was 15% higher and Cr was 16 times higher). The concentration of PM2.5-bound Mn at the De-Gan industrial zone was higher than GB3095-2012 and WHO limits in nearly half of the sampling time with the highest value of 344.17 ng/m 3 . The HMs' pollution in this area was worthy of attention.   Table 2 lists the standard limits for heavy metals defined by China, WHO, the EU, and other countries. On average, As and Cr were the two HMs exceeding the standard limits of GB3095-2012 (As was 26% higher and Cr was 170 times higher) and WHO (As was 15% higher and Cr was 16 times higher). The concentration of PM 2.5 -bound Mn at the De-Gan industrial zone was higher than GB3095-2012 and WHO limits in nearly half of the sampling time with the highest value of 344.17 ng/m 3 . The HMs' pollution in this area was worthy of attention. Positive Matrix Factorization (PMF) was used to identify the pollution sources of PM 2.5 and the related heavy metals in the study area (Q true /Q exp = 1.003; detailed method description is available in the literatures reported by Amato and Hopke [34] and Polissar et al. [35]. The other chemical components of PM 2.5 , including OC, EC, major water-soluble inorganic ions (F − , Cl − , NO 3 − , SO 4 2− , Na + , K + , Mg 2+ , Ca 2+ , NH 4 + ), and other trace elements (Al, Si, Fe, Ti), have also been added into PMF to trace the possible sources. For the five sites on average, we identified the following 6 PM 2.5 sources in descending order: coal burning (25.5%), motor vehicles (22.8%), industrial emissions (20.5%), biomass burning (15.9%), dust (7.8%), ship emissions (7.5%) Figure 3. The first factor with high levels of SO 4 2− , NO 3 − , OC, EC, K + , and NH 4 + was recognized as coal burning [1,29]. Coal combustion plays a major role in Chinese energy consumption and air pollution, former studies have pointed out that coal combustion contributes over 1/3 of PM 2.5 [21] and was the highest air pollution source in Sichuan Basin. The second factor was characterized by high OC, EC, NO 3 − , and NH 4 + , this factor represented PM 2.5 from motor vehicles [1,36], which was correlated to the motor vehicle emissions from the vicinity of urban areas such as Chengdu and Chongqing. The sampling sites (29.25 •~2 9.40 • N, 106.21 •~1 06.29 • E) were located at the southeast Sichuan Basin and only about 40~60 km away from the central downtown of Chongqing, as shown in Figure 1. SCB is a lowland region with more than 11.3 million vehicles, the west and the north of Jiangjin District are generally plain cities, the south includes the Yunnan-Guizhou Plateau, and Wushan Mountains is to the east. under the combined action of disadvantageous diffusion conditions and high vehicle numbers, vehicle emissions have become the important PM 2.5 source of the study region. The third factor was dominated by metal elements such as Al, Fe, Mn, Ti, Ni, Cu, Cr, and Zn; OC, EC and NO 3 − also had significant proportions, which have been considered to be industrial emissions [37,38]. To the northwest of the sampling locations, the adjacent Sichuan province has about 15 steel industries. Chongqing also has a developed automobile manufacturing industry and iron and steel industry. In these factories, it often needs to be accompanied by high energy consumption such as coal combustion sources, so both metallic particulate matter and burning particles existed in this factor. The fourth factor was characterized by high levels of SO 4 2− , NO 3 − , NH 4 + , OC, K + , and Cl − and considered to be biomass burning [39,40]. The SCB is one of the most important grain-producing areas in China, as many local rural areas still rely on biomass as their main energy source. Early studies also have widely stated that biomass burning was one of the main sources of air pollution in this area, e.g., 12.8% in Wang et al. [19], 11.7% in Kong et al. [20], and even reached up to 56~65% in Zhou et al. [41,42] and Song et al. [43]. The last two factors were identified as dust and ship emissions, contributing roughly the same amount (less than 8%) to PM 2.5 . The dust factor was characterized by Na, Ca, Mg, Si, and Al [44,45]. Due to the occlusive terrain, the dust was mostly from local sources such as road dust caused by vehicle movement, construction dust, and soil dust due to urban construction [29]. With the continuous advancement of urbanization process in China, there have been steady construction activities in Chinese cities, including many muck and slag trucks shuttling between the urban area and suburbs, which were thought to be the main dust origin of the sampling sites. The proportions of V, Ni, and Mg in the last factor's source profile from PMF results were the highest among all the recognized sources. Ni and V have been regarded as trace markers for heavy diesel burning in ship emissions [11,15], then they were identified as ship emissions, which were also characterized by EC and NO 3 − due to cargo ships generally running on diesel in this area. The sampling sites are located near the Jiangjin Port area along the Yangtze River, which is the main transit port for goods from western Chongqing, northern Guizhou, and southern Sichuan. Therefore, ship emissions would contribute a proportion of air pollution in Jiangjin.

Source Apportionment of PM2.5-Bound Heavy Metals
According to the PMF results, the relative contribution of the above sources to the 8 mentioned HMs was given in Figure 4. The contributors of heavy metals were different from PM2.5, the order of source contributions to the 8 PM2.5-bound HMs was industrial emissions (29.2%) > biomass burning (18.7%) > coal burning (17.7%) > dust (16.4%) > vehicle emissions (10.4%) > ship emissions (7.6%). Although coal burning was the largest PM2.5 contributor, in total, industrial emissions contributed the most of the 8 HMs, and the sources of different metallic elements also varied. V was mainly contributed by coal burning and biomass burning, while motor vehicles, ship emissions, and dust also contributed certain amounts of V. The proportion of V in the ship emissions' PM2.5 profile from PMF was highest and always used as a marker of heavy oil burning [46][47][48]. Cr was mainly contributed by industrial emissions and biomass burning; motor vehicles also took account of 18% of the total PM2.5-bound Cr. Mn mainly originated from industrial emissions, such as the manganese steel production process [46,49]. The sources of Mn were dust, coal burning, and ship emissions. Ni, Cu, and Zn were mainly emitted from industrial emissions such as metal smelting, while the contribution of biomass burning should not be ignored. Ni can also be emitted from diesel exhausts [46,47,50]; the proportion of ship emissions and motor vehicles contributed about 21% of the total PM2.5-bound Ni. Dust, biomass burning, and coal burning shared similar contribution proportions to PM2.5-bound Zn. It has been reported that Zn accounted for about 1% in the PM2.5 source profile of coal combustion [51]. As was mainly contributed by coal burning, which is used as a tracer for coal combustion [7,52]. Motor vehicles, biomass burning, and coal combustion were three major Pb sources in this study-this is consistent with the results of [11,53].

Source Apportionment of PM 2.5 -Bound Heavy Metals
According to the PMF results, the relative contribution of the above sources to the 8 mentioned HMs was given in Figure 4. The contributors of heavy metals were different from PM 2.5 , the order of source contributions to the 8 PM 2.5 -bound HMs was industrial emissions (29.2%) > biomass burning (18.7%) > coal burning (17.7%) > dust (16.4%) > vehicle emissions (10.4%) > ship emissions (7.6%). Although coal burning was the largest PM 2.5 contributor, in total, industrial emissions contributed the most of the 8 HMs, and the sources of different metallic elements also varied. V was mainly contributed by coal burning and biomass burning, while motor vehicles, ship emissions, and dust also contributed certain amounts of V. The proportion of V in the ship emissions' PM 2.5 profile from PMF was highest and always used as a marker of heavy oil burning [46][47][48]. Cr was mainly contributed by industrial emissions and biomass burning; motor vehicles also took account of 18% of the total PM 2.5 -bound Cr. Mn mainly originated from industrial emissions, such as the manganese steel production process [46,49]. The sources of Mn were dust, coal burning, and ship emissions. Ni, Cu, and Zn were mainly emitted from industrial emissions such as metal smelting, while the contribution of biomass burning should not be ignored. Ni can also be emitted from diesel exhausts [46,47,50]; the proportion of ship emissions and motor vehicles contributed about 21% of the total PM 2.5 -bound Ni. Dust, biomass burning, and coal burning shared similar contribution proportions to PM 2.5 -bound Zn. It has been reported that Zn accounted for about 1% in the PM 2.5 source profile of coal combustion [51]. As was mainly contributed by coal burning, which is used as a tracer for coal combustion [7,52]. Motor vehicles, biomass burning, and coal combustion were three major Pb sources in this study-this is consistent with the results of [11,53].

Health Risks of Heavy Metals from Pollution Sources
Since the components of PM2.5 varied among different air pollution sources, so do the health risks from each source. Therefore, in order to control the health risk of PM2.5 to the local residents, source apportionment should be integrated with health risk evaluations to estimate source-specific health risks. The corresponding calculation parameters (which have been introduced in 1.2) and health risks of PM2.5-bound HMs were shown in Tables  3 and 4, respectively. The carcinogenic risk (CR) of PM2.5-bound Cr was the highest among the studied metals in the wintertime of Chongqing due to its high IUR, contributing 72% CR originating from PM2.5-bound HMs. Noncarcinogenic risk (HQ) of PM2.5-bound Mn was the highest in all listed HMs because of its high exposure concentration (44.06 ng/m 3 ), accounting for 65% of the total HMs' HQ. PM2.5-bound As was the second-highest CR and HQ element, which contributed 22% of the total CR and 26% of the total HQ. The noncarcinogenic risks resulted from PM2.5-bound HMs were the same for adults and children, but PM2.5-bound HMs were four times more likely to cause cancer in adults than in children, which might be due to HMs' bioaccumulation property in the human body. Ei-the deposition fraction of PM2.5 that can penetrate in lung; C-the inhaled concentration of each HM during the monitoring periods; RfC-the inhalation reference concentration; IUR-inhalation unit risk.
Through combining the results of HMs' source apportionment and health risks assessment, the health risks of HMs from each PM2.5 source were given in Table 5. Industrial emissions were the riskiest heavy metal sources at the study sites, contributing 25% CR and 36% HQ, which resulted from PM2.5-bound HMs, even though industrial emissions are only the third contributors to PM2.5. Biomass burning and vehicle emissions contributed about one fifth of CR, which is the same for industrial emissions; biomass burning

Health Risks of Heavy Metals from Pollution Sources
Since the components of PM 2.5 varied among different air pollution sources, so do the health risks from each source. Therefore, in order to control the health risk of PM 2.5 to the local residents, source apportionment should be integrated with health risk evaluations to estimate source-specific health risks. The corresponding calculation parameters (which have been introduced in 1.2) and health risks of PM 2.5 -bound HMs were shown in Tables 3 and 4, respectively. The carcinogenic risk (CR) of PM 2.5 -bound Cr was the highest among the studied metals in the wintertime of Chongqing due to its high IUR, contributing 72% CR originating from PM 2.5 -bound HMs. Noncarcinogenic risk (HQ) of PM 2.5 -bound Mn was the highest in all listed HMs because of its high exposure concentration (44.06 ng/m 3 ), accounting for 65% of the total HMs' HQ. PM 2.5 -bound As was the second-highest CR and HQ element, which contributed 22% of the total CR and 26% of the total HQ. The noncarcinogenic risks resulted from PM 2.5 -bound HMs were the same for adults and children, but PM 2.5 -bound HMs were four times more likely to cause cancer in adults than in children, which might be due to HMs' bioaccumulation property in the human body.  Through combining the results of HMs' source apportionment and health risks assessment, the health risks of HMs from each PM 2.5 source were given in Table 5. Industrial emissions were the riskiest heavy metal sources at the study sites, contributing 25% CR and 36% HQ, which resulted from PM 2.5 -bound HMs, even though industrial emissions are only the third contributors to PM 2.5 . Biomass burning and vehicle emissions contributed about one fifth of CR, which is the same for industrial emissions; biomass burning was the fourth PM 2.5 source, but held similar health risks as the second PM 2.5 source. Coal combustion, as the biggest PM 2.5 source, only accounted for 15% CR and 21% HQ. All these phenomena indicate that health risks of air pollution sources are the combination effects of the source contribution factor and chemical components' characteristics. In order to control the health risk of PM 2.5 to the local residents, the pollutants from industrial emissions, biomass burning, vehicle emissions, and coal combustion sources should be given more attention in Sichuan Basin and similar regions.

Conclusions
In this study, PM 2.5 and associated heavy metals were measured from 6 January 2019 to 28 January 2019 in Chongqing, southwest China. The PMF model was adopted to conduct source apportionment of PM 2.5 -bound HMs, and the health risks carried by 8 listed heavy metals in PM 2.5 were assessed with the U.S. EPA method. Finally, this work gave the source-specific health risks from PM 2.5 -bound HMs by coupling the PMF source apportionment with the risk assessment results. The mean concentrations of PM 2.5 and HMs in PM 2.5 during the observation period were 97.06 µg/m 3 and 220.46 ng/m 3 , respectively. The downwind industrial site had higher concentrations than the upwind industrial site and urban sites, while suburban was the lowest site. The concentration of Zn was the highest among the 8 heavy metals, followed by Mn and Pb. As and Cr were the two HMs that exceeded the standard limits of GB3095-2012 (As was 26% higher and Cr was 170 times higher) and WHO (As was 15% higher and Cr was 16 times higher). For the five sites on average, six PM 2.5 sources have been recognized: coal burning (25.5%), motor vehicles (22.8%), industrial emissions (20.5%), biomass burning (15.9%), dust (7.8%), ship emissions (7.5%). The larger contributors of PM 2.5 or PM 2.5 -bound heavy metals might not be the higher health risk sources; the study results show that health risks of air pollution sources are the combination effects of the source contribution factor and chemical components' characteristics. Industrial emissions were the riskiest heavy metal sources at the study sites, contributing 25% CR and 36% HQ, which resulted from PM 2.5 -bound HMs; biomass burning and vehicle emissions contributed about one fifth of CR; coal combustion only accounted for 15% CR and 21% HQ. In order to control the health risk of PM 2.5 to the local residents, the pollutants from industrial emissions, biomass burning, vehicle emissions, and coal combustion sources should be given more attention in Sichuan Basin and the similar regions.