Characteristics , Sources , and Health Risk Assessment of Trace Elements in PM 10 at an Urban Site in Chengdu , Southwest China

To investigate trace element pollution of PM10 in urban Chengdu, a Southwest megacity of China, as well as to assess the human health risks caused by exposure to PM10, we analyzed ambient PM10 samples collected at an urban site in Chengdu from November, 2014 to October, 2015. The annual mean concentration of PM10 in the sampling period was 173.6 ± 77.9 μg m, which is 2.5 times higher than the national recommended standard of 70 μg m. The mean metals content in PM10 was in the following order: iron (Fe) > zinc (Zn) > titanium (Ti) > copper (Cu) > manganese (Mn) > lead (Pb) > barium (Ba) > chromium (Cr) > strontium (Si) > nickel (Ni) > arsenic (As) > vanadium (V) > antimony (Sb) > cobalt (Co) > cadmium (Cd) > bismuth (Bi). The concentration of As in PM10 was observed to be 2.9 times higher than the secondary level recommended by the China National Ambient Air Quality Standard (GB3095-2012), whereas the concentrations of other metals were within the limit. Enrichment factor analysis indicated that As, Pb, Zn, Cu, Cd, and Sb mainly originated from anthropogenic sources. Five factors were identified by positive matric factorization (PMF) model. The sources identified were soil dust (48.4%), road dust (19.4%), fossil fuel combustion (14.9%), electroplating industry (13.8%), and metallurgy industry (3.5%). Particle morphology and chemical composition analysis revealed six major particle types, namely aluminosilicate, sulfur-containing, carbon-containing, chlorine-containing, biomass burning, and metal particles. The results of the health risk assessment indicated that Cr can be carcinogenic to both children and adults, and other trace elements were determined to be below the legislation threshold (Environmental Protection Agency limit values), except for As, which was observed to be above the threshold.


INTRODUCTION
Atmospheric inhalable particulate matter (PM 10 ) originating from high-intensity anthropogenic activities, such as combustion, industrial processes, traffic emissions, and urban construction activities, as well as natural sources, is one of the most crucial environmental concerns of the megacities in China (Cheng et al., 2013), because of its impact on air quality, ecosystem health, and human health (Schmale et al., 2014).In particular, these particles contain toxic metals and exhibit higher possibilities of causing cancer (Dockery et al., 1993).Over the past decade, special attention has been paid to investigate the concentrations, chemical characteristics, sources, and adverse health effects of PM pollution in megacities of China (Chen et al., 2013;Li et al., 2015;Lu et al., 2015;Niu et al., 2015;Hu et al., 2016;Lyu et al., 2016;Ma et al., 2016;Madaniyazi et al., 2016;Chen et al., 2017;Peng et al., 2017), and high concentrations of toxic metals in PM 10 have been reported in many cities (Kan et al., 2012;Lu et al., 2015).
A comprehensive understanding of the sources of metal aerosols is essential in considering the health and environmental impacts of these aerosols on the atmosphere.The particulate matter source apportionment studies have been performed to evaluate the contribution of different sources by using "bulk" analyses with multivariate statistical receptor modeling (Contini et al., 2012).Among which, positive matric factorization (PMF) is one of the most used approach in source apportionment works around the world nowadays (Belis et al., 2013), which can quantify the contribution of sources to samples based on the composition.In fact, particles possess different shapes, sizes, and chemical composition in relation to emission sources (Samara et al., 2016).Scanning electron microscopy (SEM) coupled with energy dispersive X-ray spectrometry (EDX) is another powerful tool for the characterization of particles in terms of morphology and chemical composition, which can provide a clearer insight into the origin of particles (Ebert et al., 2002(Ebert et al., , 2004)).In recent years, several studies have employed an integrated approach based on statistical analysis and individual particle micro-analyses to characterize single particles in order to identify the possible emission sources, and assess the common health and environmental effects (Yue et al., 2006;Beatrice et al., 2012;Genga et al., 2012;Samara et al., 2016).
Chengdu (located in the Sichuan Basin) is one of the most polluted megacities in China; this city has a population of 14.4 million and registered more than 4 million motor vehicles by the end of 2014.The city has been affected by atmospheric pollution caused by various anthropogenic sources, as well as unfavorable geographical and meteorological conditions.High aerosol loading in Chengdu not only considerably impairs visibility (Zhang et al., 2012), but also causes public health damage (Zeng et al., 2011;Li et al., 2016).Studies on atmospheric particles in Chengdu have characterized in terms of chemical composition (Tao et al., 2013;Tian et al., 2013;Wang et al., 2013;Tao et al., 2014;Zhang et al., 2014a); however, few studies on the morphology and health implication of PM 10 in this region have been reported.
In this study, the PM 10 levels, sources, and human health risk associated with trace elements in PM 10 were investigated at an urban site of Chengdu for a 1-year period (between November 2014 and October 2015).

Study Area
Chengdu (the capital of Sichuan province) is the center of political, economic, cultural, and transportation in Southwest China.It has a subtropical, humid, monsoon climate, the annual mean temperature is 16°C, wet precipitation is 900-1300 mm, and wind speed is 1.2 m s -1 .Chengdu is surrounded by high mountains on three sides (i.e., Longquan Mountain to the east, Muma Mountain to the south and Qionglai Mountain to the west of the city); static meteorological conditions are not conducive to the dispersion of locally produced pollutants.Air pollution is one of the most pressing problems of the city, severe haze pollution frequently occurrence especially in wintertime.

Sampling
PM 10 samples were collected from the roof (15 m above the ground) of an office building at Chengdu University of Technology (30°40′N, 104°08′E) (Fig. 1).The site selected for the study is an urban area located in the eastern district of Chengdu.No major stationary air pollution sources were present at the sampling site.However, other sources of PM pollution including heavy traffic (from diesel and gasoline exhaust), demolition of old buildings and construction work emissions, and industrial pollution were present outside the third-ring road.Thus, the site was considered to be representative of a typical living environment in the eastern district of Chengdu.
PM 10 samples were collected using an intelligent mediumflow sampler (Wuhan Tianhong Co. TH-150C, Wuhan, China) with a flow rate of 100 L min -1 on quartz filters with a diameter of 90 mm (QM-A, Whatman).Before sampling, quartz filters were baked at 450°C for 4 hours.Then, filters were put in a chamber at 25°C and 50% of relatively humidity before and after sampling for 48 h.Sampling was performed for 24 h at a frequency of 10 samples a month.The sampled filters were placed in a sealed plastic filter box and very strict operation practices are required to avoid the loss of filter due to improper operation.A total of 120 samples were collected during the periods of 20 th November, 2014 to 20 th February, 2015 (representative of winter), 20 th March to 20 th May (spring), 20 th June to 20 th August (summer), 20 th September to 20 th October (autumn) in 2015.The quartz filters was measured gravimetrically for particle mass concentration using a Sartorius electronic microbalance at controlled humidity (45 ± 5%) and temperature (20 ± 5°C).To guarantee the accuracy and reliability of sampling, three blank filters were preserved under the same environment.

Chemical Analysis
The concentrations of 17 trace elements (including As, Ba, Bi, Cd, Co, Cr, Cu, Mn, Ni, Pb, Sr, Sb, Sc, Ti, V, Fe and Zn) were analyzed by a PE 6000 inductively coupled plasma mass spectrometer (Perkin Elmer Corp., Norwalk, USA).The quality assurance (QA) and quality control (QC) of chemical analysis were conducted.The blank contribution (i.e., reagent + blank filter) was evaluated and taken into account.The detailed description of chemical analysis, the values of QA/QC, and blanks contribution were provided in our previous studies (Cheng et al., 2017 a, b).

Microscopic Studies
Ten PM 10 samples with a high mass load were selected and analyzed using an SEM instrument with an FEI Quanta 250 FEG operated at 20 kV and equipped with an X-ray energy dispersive spectrometer (EDX-Oxford INCAx-max20).Single portions (10 × 10 mm) of the quartz filters were cut from the central part of the membranes and mounted on SEM aluminum stubs by using double-sided carbon tape.The portions were coated with a thin film of gold and observed within randomly selected fields of view.For each sample, 10-20 particles were randomly selected.The chemical composition was determined using EDX (spot size 5, working distance 10 mm); the measurement time for EDX analysis of each particle was 60 s.

PMF Model
PMF is a multivariate factor analysis approach that decomposes a matrix of speciated sample data into two matrices: factor contributions (G) and factor profiles (F).These factor profiles need to be interpreted by the user to identify sources types that may be contributing to the speciated sample data (Paatero and Tapper, 1994;Paatero, 1997).The goal of PMF model is to solve the chemical mass balance between measured species concentrations and source profiles, which can be described as follow: where x ij is the concentration of i number of samples and j chemical species, g ik is the contribution of the kth factor to the ith sample, f kj is the jth species fraction from the kth source, e ij is the residual related with the jth species concentration measured in the ith sample, and p is the number of factors or sources.PMF model is non-negative constraints and each data point individually weighted.Factor contributions and profiles are derived by the PMF model minimizing the objective function Q: where u ij is the uncertainty of jth species in the ith sample, which is applied to weight the observations that contain sampling errors, detection limits, missing data and outliers (Paatero and Hopke, 2003).In this study, USEPA PMF5.0 was used to identify the potential sources of PM 10 .Seventeen chemical components were used for the PMF model, including As, Ba, Bi, Cd, Co, Cr, Cu, Mn, Ni, Pb, Sr, Sb, Sc, Ti, V, Fe and Zn.To estimate the optimal number of sources, three to seven factors were tested for each dataset.The Q values, the resulting source profiles, and the scaled residuals were examined.The Q values are examined as a function of the number of factor to identify any possible sharp decrease in values indicating the correct choice.In the five factor model, a value of F peak = -0.1 provided the most physically reasonable source profiles.

Human Health Risk Assessment
A health risk assessment including non-carcinogenic and carcinogenic risks was performed according to standard Environmental Protection Agency (EPA) methods (U.S. EPA, 2011).The adults and children living in this area are potential receptors.The crucial trace elements considered in the risk assessment include Cu, Pb, Zn, As, Cd, Cr, Co, Ni, V, Sb, and Mn.Among them, As, Cd, Cr, and Ni are carcinogenic elements, whereas Cu, Co, Pb, Zn, Sb, V, and Mn are non-carcinogenic metals (U.S. EPA, 2011).Risk assessment was performed using the mean concentrations of elements in a PM 10 particle, and inhalation was assumed to be the only pathway of exposure to PM-bound trace elements.The average daily dose (ADD, in mg kg -1 d -1 ) of noncarcinogenic elements and the lifetime average daily dose (LADD, in mg kg -1 d -1 ) of carcinogenic elements were calculated as follows: where C is the concentration of pollutants, mg m -3 ; IR is the inhalation rate (15.2 m 3 d -1 for men, 11.3 m 3 d -1 for women, and 8.7 m 3 d -1 for children, according to the U.S. EPA model); BW is the body weight (70 kg for men, 60 kg for women, and 36 kg for children); ED is the duration of exposure (30 years for adults and 18 years for children), d; and AT is the averaging time (for carcinogens, AT = 70 × 365 = 25,550 days, and for non-carcinogens, AT = 30 × 365 = 10,950 days for adults and 6570 days for children) for residents.Risk characterization was quantified separately for carcinogenic and non-carcinogenic effects.For carcinogenic elements, the cancer risk for individuals is calculated as follows: where R is the average annual excessive risk of cancer for an individual, dimensionless; SF is the slop factor (20.7 for As, 56.0 for Cr, 8.4 for Cd, and 1.2 for Ni), mg kg -1 d -1 ; 70 is the number of an individual's average lifetime, years.

PM 10 Mass Concentration
The annual mean concentration of PM 10 was 173.6 ± 51.1 µg m -3 (Table 1) (34.2-574.2µg m -3 ) in Chengdu during the study period, which is 2.5 times higher than the National Ambient Air Quality Standard (NAAQS) annual PM 10 guideline value of 70 µg m -3 in China and is 8.7 times higher than the WHO ambient air quality standard (20 µg m -3 ).Compared with the NAAQS daily standard PM 10 concentration (150 µg m -3 ), approximately 46% of PM 10 samples were observed to exceed the daily limit value.Compared with the reported data in other cities in China (Table 1), the PM 10 level in Chengdu was observed to be significantly lower than those in Guiyang and Xi'an and was determined to be comparable to those of Chongqing and Beijing.However, PM 10 in Chengdu was observed to be significantly higher than those observed in Changsha and Guangzhou.Moreover, the annual average PM 10 level in Chengdu was determined to be higher than those in most developed counties listed in Table 1 (Contini et al., 2010;Pateraki et al., 2010;Tony et al., 2010), which suggests a relatively high level of PM 10 pollution in the Chengdu urban atmosphere.
A seasonal variation of PM 10 mass concentration was observed (Table 2): higher concentrations were observed in winter (210.5 ± 97.8 µg m -3 ) and spring (193.3 ± 56.8 µg m -3 ) and lower concentrations were observed in autumn (156.6 ± 56.5 µg m -3 ) and summer (134.1 ± 42.5 µg m -3 ).The observed patterns are likely to reflect the effects of meteorological conditions, sources, and transportation.During the winter months, temperature inversions are generated, and the winter months are mainly characterized by low wind speed and high relative humidity, which favor the accumulation of particles in the atmosphere.Coal combustion increases in winter.Dust storms frequently occur in spring in northern China; long-range transported particles and fugitive dust were reported to substantially contribute to the increased PM 10 level (Tao et al., 2013).Agricultural waste burning is usually most severe in the harvest season-autumn in Chengdu (Chen et al., 2016).The mass concentrations of PM 10 in summer were the lowest, which can be explained in part by the strong scavenging action because of most frequent rainfall during this period.

Elemental Concentrations
The minimum, maximum, and mean values and the standard deviations of elemental concentrations among the four seasons are presented in Table 2.The crustal element Fe (Lim et al., 2010a) was the most abundant metal in the analyzed component in PM 10 , and its concentration was substantially higher than that of any other toxic metal.In addition, the anthropogenic metals Zn, Pb, Cu, and Mn were present in the highest concentrations among the measured trace metals in PM 10 .The As concentration was observed to be 2.9 times higher than level 2 of the national standard (6.0 ng m -3 ) (GB3095-2012, 2012), indicating the potential threat to human health.Seasonal variations of most elements were similar to that of PM 10 ; the lowest concentrations of most elements, except for Cu, were observed in summer.Wet deposition by rainout and washout scavenging effectively removes metal-bearing particles from the atmosphere.Related research demonstrated that approximately 40%-90% of the heavy metals are removed through wet deposition (Yang et al., 2009).Similarly, in Qingdao, approximately 99.37% of Pb and 17.19% of Cu were reported to be removed from PM 10 through rainfall scavenging (Li et al., 2003).
Compared with other urban areas in China (Table 3), the particulate heavy metal levels in Chengdu were comparable to those in Shanghai and Taiyuan.However, the concentrations of heavy metals were lower in Chengdu than those in Chongqing, except for As and Cu.In particular, the concentrations of As and Cd in Beijing were 2.1 and 2.6 times high than those in Chengdu, and the concentrations of Pb and Zn in Tianjin were 5.7 and 2.3 times high than   (Tao et al., 2014).Compared with foreign cities, the levels of most heavy metals in Chengdu were substantially higher than those in Costa, whereas they were lower than those in Delhi, except for Pb.These results indicate that the particulate toxic heavy metals in Chengdu were at a moderate level compared with other urban areas in the world.

Enrichment Factor for Trace Elements
Enrichment factors (EF) for trace elements relative to the earth's upper crust composition were calculated by recognizing natural and anthropogenic sources of aerosol components.EF is defined as follows: EF = (X/R) aerosol/(X/R) crust (Caroli et al., 1996) (6) where X is the mass concentration of element X and R is a reference element.In this study, scandium (Sc) was selected as the reference element because of its low volatility, lack of anthropogenic sources, and reliable quantitative determination (Bilos et al., 2001).The average composition of the earth's crust was obtained using the local soil background values (Zhu et al., 2004).The predominant sources are considered as anthropogenic activities, mixed sources, and natural sources when EF > 10, EF < 1, and 1 < EF < 10, respectively.As shown in Fig. 2, EF values of V, Ti, Co, Mn, Fe, barium (Ba), Sr, Ni, Cr, and bismuth (Bi) in PM 10 were less than 10, indicating that these elements originate from natural sources, as well as anthropogenic sources.The EF values of As, Pb, Zn, and Cu ranged from 10 to 100, and those for Cd and Sb were higher than 100 (Fig. 2), indicating that these elements pollution is mainly a results of human activities.The EF values of the carcinogenic heavy metals As, Cd, Co, Cr, Ni, Pb and Sb were 23.9, 106.6, 1.9, 6.9, 6.0, 40.4,and 153.9, respectively, indicating that these heavy metals were of anthropogenic origins, except for Co, Cr, and Ni, which originated from mixed sources.Seasonal, strong enrichment occurred in winter and spring for most of the elements, whereas low EF values were often observed in summer.Notably, Sb was considerably enriched in PM 10 , similar results were obtained in a previous study in Chengdu (Tao et al., 2014).Sb is considered a priority pollutant by the U.S. EPA and European Union (Daniel et al., 2017) because of its potential toxicity to human health.Numerous studies have investigated Sb content in atmospheric samples (PM and fly ash), which have indicated that Sb is released to the atmosphere mainly through traffic, urban incineration of plastic wastes (Filella et al., 2009;Daniel et al., 2017), and smelters.

Source Apportionment by PMF Model
Based on the PMF modeling results, five solved factors assigned as metallurgy industry, fossil fuel combustion, electroplating industry, soil dust and road dust were identified.Modeled source profiles together with the relative contributions of individuals to each analyzed species are shown in Fig. 3.The first factor comprises metallurgy industry, characterized by high factor loading of Cd, Cu and Fe (Allen et al., 2001).From EF analysis results, the main source of Cd is anthropogenic, such as metal smelting, fossil fuel combustion, and waste incineration.This factor contributed 3.5% to PM 10 .The second factor exhibited high loading for the anthropogenic-derived metals Cu, Pb, V, As, Ni and Cr, which might be identified as fossil fuel combustion source.Previous studies have suggested that As, Pb and V mainly derived from coal and oil combustion (Li et al., 2010;Xu et al., 2012), and Cr and Ni can be considered as indicators of emission from fuel combustion and vehicle emission (Tian et al., 2012).This source contributed 13.8% to PM 10 .The third factor was characterized by high factor loading of Zn, Bi, As, Sb, Cr and Ni, which may represent a contribution of electroplating industry.Electroplating industry is a pillar industry in Chengdu, Bi and Sb were used to configure alloys in electroplating industry.The forth factor had the high factor loading of Fe, Sc, Sb, Mn, Pb and Ba, which may represent the soil dust.Ba, Mn, and Fe are considered as markers for soil and re-suspension dust, Mn and Fe are markers for construction sources (Lim et al., 2010a).This source contributed 48.4% to PM 10 .The five factors is road dust, characterized by high factor loading of Ti, Sr, Co and Ba.This factor contributed 19.4% to PM 10 .The metal elements Ba, Sr, and Ti have been identified as markers for vehicles emission (Pakkanen et al., 2001;Lim et al., 2010a).Ba, Mn, and Ti are considered as markers for soil and re-suspension dust (Lim et al., 2010b).Thus, PM 10 in Chengdu was mainly influenced by five source categories: metallurgy industry, fossil fuel combustion, electroplating industry, soil dust and road dust.

Scanning Electron Microscopy
Based on SEM-EDX analysis of the particles for determining their morphology and elemental composition, six cluster types were identified: aluminosilicate, sculpturerich, carbon-rich, chlorine-rich, biomass burning, and metal particles (Figs.4-9).The effects of elements were analyzed using a blank filter.The SEM micrograph and corresponding EDX spectrum of one blank filter are presented in Fig. 5(a).The dates were then normalized to match the weight percentages of each element detected in the individual particles.
Sulphur-rich particles (Fig. 5): S and Ca particles, possibly calcium sulfate, were prevalent in the air in Chengdu, and they possibly originated from crustal re-suspensions or were formed as secondary particles generated from SO 2 oxidation (Murari et al., 2016) or from deterioration of building surfaces (e.g., the reaction of marble and limestone with sulfur compounds in the atmosphere) (Genga et al., 2012).Calcium sulfate is used in cement production and is a secondary product of flue gas desulphurization (Genga et al., 2012).In most analyzed cases, S particles were ephedra (flaky and rod) crystals (Figs. 5(a), 5(b), and 5(d)) and were observed to grow diagenetically.These particles exert notable effects on the regional climate by scattering the entire spectrum of solar radiation.EDX demonstrated that these particles were predominately associated with K, Al, and Si.In addition, S particles were evident with irregular morphologies in the form of chains and large aggregated clusters (Fig. 5(c)).These finest particles are formed by homogeneous or heterogeneous condensation of atmospheric SO 2 in fossil fuel combustion.
Carbon-rich particles (Fig. 6): These are particles with high carbon and oxygen content and are sometimes characterized by other trace combustion particulates such as Si and K and in certain cases are with crustal species (Ca, Al, and Fe).They are characterized by regular and symmetrical shapes, ranging from spherical (Figs. 6(b) and 6(d)) to elliptical (Figs. 6(a) and 6(c)) aggregated shapes and porous surfaces.The spongiform spherical particles found on aerosol filters result from burning in thermal power plants, motor engines, and petrochemical industries and are the most abundant type of spherical particles in the area.Smooth spherical particles are also emitted from burning processes.
Chlorine-rich particles (Fig. 7): These particles are present only in the coarse group.Cl and K can originate from coal combustion and biomass burning and incineration (Samek et al., 2017).Chengdu is one of the few inland megacities worldwide located far from the ocean.In this study, the contribution of sea-salt particles was not considered.Thus, coal and biomass burning was the main contributor of Cl in Chengdu.Fig. 7(a) presents coal combustion-related Cl particles, whereas Fig. 7(b) presents particles originating from biomass burning.Abundance of Cl spheroids with Si, K, and S invariably suggests their farming and vegetation/coal burning origin.
Biomass burning particles (Fig. 8): These particles are usually symmetrical and very structured and have porous surfaces.The particle size and number and distribution of the voids vary considerably and are helpful in determining their source.Particles originating from biomass are characterized by a carbonaceous matrix, and those originating from a vegetable source contain considerable quantities of Cl.
Metal particles (Fig. 9): These particles have a high content of metals such as Ti (Fig. 9    and Ba (Fig. 9(h)).Metal particles with an irregular shape were observed in coarse particles, which may be mainly derived from land sources such as soil dust, resuspension of dust from roads, and crust, but they may also originate from human activities such as industrial processes, abrasion of metallic materials and traffic-related sources, and construction works.Particles mainly containing Fe may be of crustal origin, but they may also originate from human activities such as industrial processes.Particles containing S and Ba can be considered as originating from road dust and soil re-suspension.Particles containing Mo can originate from Mo-related industries (Tao et al., 2014).Moreover, particles containing La, Ce, and Nd were assumed to be imported from other areas.Smooth spherical particles mainly containing Ti originate from pigment manufacturing processes.Biomass burning particles (Fig. 8): These particles are usually symmetrical and very structured, and porous surfaces.The particle size and number and distribution of the voids vary considerably and are helpful in determining their source.Particles originating from biomass are characterized by a carbonaceous matrix, and if they are from a vegetable source they will contain appreciable quantities of chlorine.
Metal particles (Fig. 9): These are particles with a high metal content such as Ti (Fig. 9 and Ba (Fig. 9(h)).Metal particles with irregular shape were observed in coarse particle which may be mainly derived from land such as soil dust, resuspension of dust from road, crust, but may also come from human activities such as industrial processes, abrasion of metallic materials and traffic-related sources and construction works.Particles containing mainly Fe, can be of crustal origin, but may also come from human activities such as industrial processes.Particles containing S and Ba can be explained as result of road dust and soil resuspension.Particles containing Mo can originate in the Mo-related industries (Tao et al., 2014).There were also particles containing La, Ce, and Nd, were assumed to be imported from other area.Smooth spherical particles consisting mainly of Ti originate from the manufacturing of pigments.

Human Health Risks Assessment
Health risk assessment is particularly useful in understanding the potential health hazard associated with inhalation exposure to airborne metals.Based on the average annual concentrations of airborne trace metals determined during the study periods, the carcinogenic and noncarcinogenic risks of metals to human health were calculated at the sampling site (Figs.10(a) and 10(b)).
The risk of exposure to carcinogenic heavy metals through the respiratory system was between 1.25 × 10 -8 and 3.08 × 10 -6 and the risk of exposure to Cr is slightly higher than the average risk acceptance level of 10 -6 year -1 (U.S. EPA, 1989) (Fig. 10(a)), which warrants attention.The risk levels of exposure to carcinogenic heavy metals were observed in the following order: Cr > As > Ni > Cd.In addition, carcinogenic substances pose the greatest cancer risk to men, followed by women and children.
The risk of exposure to non-carcinogenic heavy metals through the respiratory system was between 5.28 × 10 -11 and 1.11 × 10 -9 , which is within the acceptable levels of 10 -6 year -1 (U.S. EPA 1989) (Fig. 10(b)).The risk levels of non-carcinogenic metals in adults and children are similar.The risk levels of exposure to non-carcinogenic metals were observed in the following order: Mn > V > Pb > Cu > Zn > Sb.Non-carcinogenic metals can cause harm to children more easily than to men and women.
The total risk levels posed by the 10 metals were determined to be 3.59 × 10 -6 , 3.10 × 10 -6 , and 2.40 × 10 -6 for men, women, and children, respectively, which are higher than the average risk acceptance level of 10 -6 year -1 (U.S. EPA, 1989).The risk posed by carcinogenic metals is shown to be significantly higher than that posed by noncarcinogenic metals, implying the potential carcinogenic effects on local inhabitants.Because of the difference in body weight, respiration rate, and outdoor exposure of individuals, the carcinogenic and non-carcinogenic risks to men, women, and children are different.

CONCLUSIONS
The average mass concentration of PM 10 in the eastern part of downtown Chengdu was 173.6 ± 77.9 µg m -3 , which is approximately 2.5 times higher than the NAAQS guideline for PM 10 in China.Seasonal variations were in the following order: winter > spring > autumn > summer.The crustal elements Ti and Fe were the dominant components in PM 10 , and the anthropogenic metals Zn, Pb, Cu, and Mn had the highest concentrations among the measured trace metals.The levels of heavy metals in PM 10 in Chengdu were moderate compared with those in other cities in China; however, they were determined to be substantially higher than those in most cities in other countries.The elements As, Pb, Zn, Cu, Cd, and Sb had EF values higher than 10; therefore, anthropogenic emission was considered to be their main source.PMF modeling results indicated that the main sources of the trace elements are soil dust, road dust, fossil fuel combustion, electro-plating industry, and metallurgy industry.Through microstructure measurements, six major types of particles were identified.The results of the health risk assessment indicate that Cr could pose carcinogenic risks and other toxic heavy metals were present at a safe level.However, the total risk levels posed by the ten metals are higher than the average risk acceptance level of 10 -6 year -1 .The complex composition of anthropogenic particles included S, O, C, Si, F, K, Ba, Sb, Rb, Ca, AI, La, Ce, Nd, Na, Mg, Mo, Cl, Cr, Ti, Fe, V, Mn, and Zn.
Fig. 3. Factor profiles for the resolved factor of PM 10 by PMF.

Fig. 10 .
Fig. 10.The risks associated with carcinogenic metal (a) and non-carcinogenic metal (b) components of PM 10 for different demographic groups.

Table 1 .
Statistics of annual PM 10 mass concentrations in Chengdu and other cities in the world.

Table 2 .
Seasonal variations of PM 10 -bound metal concentration at sampling sites (ng m -3

Table 3 .
Comparison of trace metal mean levels in Chengdu PM 10 (ng m -3 ) with other studies.in Chengdu, indicating that the pollution of heavy metals in Chengdu was not the highest in China.Notably, relatively high levels of Cu were observed in Chengdu compared with other urban sites in China.It is a result of huge amounts of coal consumption in Chengdu those