Concentrations, Source Characteristics, and Health Risk Assessment of Toxic Heavy Metals in PM2.5 in a Plateau City (Kunming) in Southwest China

To explore the mass concentration levels and health risks of heavy metals in the air in dense traffic environments, PM2.5 samples were collected at three sites in the city of Kunming in April and October 2013, and January and May 2014. Ten heavy metals––V, Cr, Mn, Co, Ni, Cu, Zn, As, Cd and Pb––were analyzed by ICP–MS, and the results showed PM2.5 concentrations significantly higher in spring and winter than in summer and autumn, especially for Zn and Pb. The concentration of heavy metals on working days is significantly higher, indicating that vehicle emissions are significant contributors. An enrichment factor analysis showed that Cr, Mn, Ni, Cu, Zn, As, Cd and Pb come mainly from anthropogenic sources, while V and Co may be both anthropogenic and natural. The correlation and principal component analysis (PCA) showed that Ni, Cu, Zn, Cd and Pb mainly come from vehicles emissions and metallurgical industries; Cr and Mn, from vehicles emissions and road dust; and As, mainly from coal combustion. The health risk assessment shows that the non-carcinogenic risk thresholds of the heavy metals in PM2.5 to children and adult men and women are all less than 1. The carcinogenic risk of Cr for men and women in traffic-intensive areas exceeds 10−4, reaching 1.64 × 10−4 and 1.4 × 10−4, respectively.


Introduction
In recent decades, the number of motor vehicles and total energy consumption have increased in China, as a result of rapid economic development and urbanization. Today, atmospheric particulate matter (PM) is one of the most significant air contaminants [1][2][3]. PM less than 2.5 µm (PM 2.5 ) comprises heavy metals, bacteria, toxins and carcinogens due to its small particle size and large specific surface area [4,5]; consequently, it has received extensive attention from researchers. Studies have shown the adverse health effects from exposure to PM and heavy metals [6][7][8][9], and epidemiological studies over the past few decades identified a close relationship between increased atmospheric heavy metals and increased mortality and morbidity [10][11][12].
The heavy metals come from natural and anthropogenic sources, including coal combustion, traffic emissions, and smelting. Because of their high toxicity, persistence and bioaccumulation, research into the health effects of long or short-term exposure to traffic-related air pollutants has become the focus of attention [13][14][15][16]. Fang et al. [17] proposed that heavy metals in road dust are not easy to remove, and may diffuse into the atmosphere and persist in the environment for a long time. He et al. [18] believe that large and medium-sized cities show obvious effects of vehicle pollution, and a large number of studies found high concentrations of Cr, Cu, Zn, Cd and Pb. These metals can cause and analysis. All filters were analyzed within 2 weeks. Table 1 lists the detailed sampling time and weather conditions at the three sites. Wind speed and temperature data are derived from https://www.wunderground.com/ (accessed on 23 August 2021).

Heavy Metal Analysis
Inductively Coupled Plasma Mass Spectrometry (ICP-MS, Agilent 7500a, Agilent, Palo Alto, CA, USA) analysis was employed to determine concentrations of 10 kinds of heavy metal elements: V, Cr, Mn, Co, Ni, Cu, Zn, As, Cd and Pb. Aluminum (Al) was analyzed to calculate the enrichment factors as the reference element. The scrap sampled filters were immersed in mixture of HCl and HNO 3 (HCl: 16.7 mL, HNO 3 : 5.5 mL, ultrapure water: 150 mL) in a Teflon beaker. Then the beaker was covered by a lid, and the solution was placed on a 220 • C temperature-controlled electric hot plate for 2.5 h of circumfluence. As the liquid cooled, we washed out the inner wall of the beaker with ultrapure water, stewing for 0.5 h for extraction, filtration, then at constant volume to 50.0 mL under test.
Blanks (including filters) and duplicate sample analyses were performed for approximately 10% of all the samples. Certified reference materials (CRM) were used to ensure accuracy and precision (National Research Center of CRM, China). The field blank samples were analyzed using these same procedures. The results of the blank analyses were used to correct the corresponding samples.

Enrichment Factor (EF)
EF is an important indicator of the extent of disturbance to the natural environment caused by human activity. By comparing measured values with soil background values, we can see the influence of human activity on particulates [37]. EF is calculated by the following equation: where C i represents the concentration of heavy metals; C n represents the concentration of the reference element (we choose Al as the reference element in this study); (C i /C n ) sample is the concentration ratio of the samples; (C i /C n ) soil background is the concentration ratio of the corresponding element in the crust. If EF is close to 1, the element can be considered to originate mainly from soil particles; if EF > 10, the element mainly originates from human activity [38].

Principal Component Analysis (PCA)
PCA is a resource allocation method approved and recommended by the U.S. Environmental Protection Agency (EPA). It is an important multivariate statistical tool that can reduce the dimensionality of large datasets and extract the number of principal components needed to explain all the variances of such datasets, which is much less than the original number of variables [39,40]. This method establishes the orthogonal distribution between the components. No matter how many variables were included in this study [41], the regression adjustment results for each factor were simple and stable. During the analysis, the factors were determined by selecting principal components with eigenvalues greater than 1 according to previous studies [42,43]. They effectively explained and represented the source characteristics of each heavy metal in PM 2.5 , guided the understanding and control of heavy metal pollution emissions in the study area, and helped improve the health of local populations.

Health Risk Assessment (HRA)
This research used the health risk assessment model provided by the U.S. EPA as the basic framework, which was used for assessing the health risk of heavy metals in PM 2.5 [44]. The target population of this study was children, and adult men and women. Then, based on the concentration levels of heavy metals in the study area, we calculated and assessed the health effects according to the health risk assessment model [45,46]. Because the heavy metals in PM 2.5 enter the body through the respiratory system, this study focused on assessing the health risks of inhalation exposure, and the dose of this route was calculated by the following formula: where C (ng·m −3 ) represents the concentration of heavy metal elements in PM 2.5 ; the variable ADD (mg·(kg·d) −1 ) is normal average daily dose for non-carcinogenic elements; and LADD (mg·(kg·d) −1 ) is lifetime average daily dose for carcinogenic elements. The exposure parameters are mainly based on the exposure parameter manual of the population in China [47] as shown in Table 2. According to the EPA's Integrated Risk Information System (IRIS) and the International Agency for Research on Cancer (IARC), pollutants can be divided into carcinogens and non-carcinogens [48]. In this study, V, Mn, Cu, Zn and Pb are non-carcinogens; Cr, Co, Ni, As and Cd are carcinogens. The non-carcinogenic risk quotient (HQ) and carcinogenic risk (CR) of heavy metals were calculated as follows: In the formula, Rfd (mg·(kg·d) −1 ) is the reference dose of each heavy metal; SF ((kg·d)·mg −1 ) is the carcinogenic slope factor. The specific parameters were shown in Table 3 [49][50][51]. If HQ were less than 1, it meant that the non-carcinogenic risk was small or negligible; if it were greater than or equal to 1, it meant that there was a non-carcinogenic risk, which increased as the HQ value increased; CR represents the carcinogenic risk of heavy metals: when CR was between 10 −6 and 10 −4 , it was within the acceptable range. If CR ≥ 10 −4 , there was a significant risk [52]. Table 3. Response parameters of elements entering through the respiratory system [49][50][51].

Element
Risk

HYSPLIT4 Model
The HYSPLIT4 model is a professional model jointly developed by the National Oceanic and Atmospheric Administration (NOAA) Air Resources Laboratory (ARL) and the Australian Bureau of Meteorology (BOM) over the past 20 years to calculate and analyze the transport and diffusion of atmospheric pollutants. The model has a relatively complete transport, diffusion and sedimentation model that can handle multiple meteorological element input fields and physical processes, and different types of pollutant emission sources. It is widely used in the study of the transmission and diffusion of multiple pollutants in various regions. In this study, the independent version of the backward trajectory model and the auxiliary software package were used (GUI; Ghostscript; ImageMagick). Meteorological data were obtained through NCEP (National Centers for Environmental Prediction) and GDAS (Global Data Assimilation System). During the four sampling campaigns, the average mass concentrations of PM 2.5 at the three Kunming sites (DR, JDM, WM) were 125.16 ± 71.27 µg·m −3 , 170.10 ± 104.10 µg·m −3 and 114.00 ± 73.99 µg·m −3 , respectively. The overall concentration of the 10 heavy metals accounted for about 1.26% of PM 2.5 . Pb and Zn accounted for the highest proportions, 0.32 and 0.38%, respectively. Because most large industries in Kunming had moved to industrial parks far from urban areas, only a few industries were located around the JDM site. These industries were in a continuous operation, which shows that the concentrations of discharged heavy metals may not have fluctuated much between working days and rest days. Heavy-duty diesel vehicles are common around the site and emit a large amount of PM 2.5 heavy metals on workdays. Figure 2a shows the concentration variation of 10 heavy metals on working days and rest days. The total concentrations ranged from 481.70 to 2052.65 ng·m −3 on working days, and 199.97 to 1105.21 ng·m −3 on rest days. The concentration levels of heavy metal markers (Cr, Mn, Cu, Zn, and Pb) from traffic sources are significantly higher on working days, especially in spring and winter. Figure 2b shows the concentration variation of heavy metals at the three sampling sites. It was found that almost all of the concentrations fluctuated wildly on working and rest days, especially at the DR and JDM sites, which suggest that heavy metals emitted by the traffic sources make an important contribution to ambient air.

Seasonal and Spatial Variations of Heavy Metals in Kunming
Seasonal and spatial variations in PM 2.5 heavy metals in Kunming are shown in Figure 3, which also shows the concentration levels of all 10 heavy metals and their standard variance. The concentrations followed the order Zn > Pb > Mn > As > Cu > V > Cr > Ni > Cd > Co. The concentration of Cr, As and Cd exceeded the secondary level of National Ambient Air Quality Standard of China (GB 3095-2012, the limit concentrations of As, Cr and Cd are 6.00, 0.025 and 5.00 ng·m −3 ). The highest concentration was found in autumn at JDM with 996.6 ± 791.56 ng·m −3 (Zn), and the lowest concentration was found in autumn at WM with 0.52 ± 0.75 ng·m −3 (Co). Overall, the concentrations of Zn and Pb were much higher than other heavy metals, and the concentrations of most were consistent with the seasonal distribution of PM 2.5 , showing a pattern of high concentrations in winter and spring, and low concentrations in summer and autumn. Due to the intensive traffic flow and frequent industrial activity at the two sampling sites DR and JDM, concentrations of Ni, Cu, Zn, Cd and Pb were much higher than that at WM. Concentrations of Cr reached the highest value at DR because of heavy summer traffic (147.80 ± 247.85 ng·m −3 ), but summer concentrations of As and Pb were significantly higher at WM than at DR and JDM, possibly because WM is downwind from the petroleum, steel, phosphate, and salt industries at the Anning Industrial Park. However, there are many high mountains around the WM site that block the transmission of pollutants to Kunming. Therefore, the pollutants from Anning Industrial Park may not have had a significant impact on the DR and JDM sampling sites.

The Transport of Heavy Metals
To understand the transport of heavy metals in PM 2.5 from distant sources, 72 h backward trajectories starting at 1000 m at the center of Kunming during the four sampling periods were calculated every 6 h using the Hybrid Single Particle Lagrangian Integrated Trajectory 4.0 (HYSPLIT4) model. Meteorological data came from the Global Data Assimilation System (GDAS). Figure 4 shows the clustering of air mass trajectories at 1000 m in Kunming. It was found that the air masses mainly originated in Myanmar in the southwest during the spring, summer and winter, and 47.02% of autumn air masses originated in Guizhou.   Figure 4, we can see that high concentrations occur in spring, possibly caused by the long-distance transmission of dense biomass burning from Myanmar. In summer, due to rainfall, concentrations were relatively low, and in autumn, clean air from Guizhou did not increase concentrations. In winter, however, concentrations of total heavy metals were the highest, which may be the result of coal burning by local and regional residents or of the long-distance transmission of pollution from India.  Figure 5b shows that the concentration variation of total heavy metals and PM 2.5 at WM during all sampling periods followed the order spring > summer > winter > autumn.
In summer, the high concentrations of PM 2.5 may have been caused by forest fires in the region of Anning during early sampling, or the valley wind that brought pollution up to the WM site on the top of the mountain.

Comparisons with the Concentration of Heavy Metals in Other Cities
As shown in Table 4, the concentrations of most heavy metals in PM 2.5 were generally higher than for most other cities, but they were lower than in Xi'an [53], which is an important industrial base and comprehensive transportation hub in Northwest China. The concentrations of some carcinogenic heavy metals (Cr, Ni and As) were comparable to those reported in other cities. Kunming's As and Pb levels were 2.68 and 1.88 times higher, respectively, than those of Beijing [54]. Nanjing, Chengdu, Chongqing and Guangzhou [55][56][57] are all provincial capitals or municipalities with a dense transportation network, but the concentrations of heavy metals are much lower than for Kunming. Compared with Barcelona [58], all heavy metal concentrations for Kunming were higher: Cu, Zn and Pb were more than twice as high and Cr was 4.85 times higher. In Istanbul [59], the level of Cr was 4.18 times higher, but other heavy metals were lower. Similarly, as a plateau mountain city, the city of Lanzhou [60] also has very high concentrations of heavy metals in PM 2.5 . These results may explain to a certain extent the reason why the death rate due to problems with the respiratory system in Yunnan is higher than the national average. However, some cities have higher concentrations of heavy metals and lower death rates, which implies that there are other atmospheric pollutants in the environment that affect the respiratory system, such as persistent organic matter, etc.

Enrichment Factor (EF) and Correlation Analysis
In this study we used an enrichment factor (EF) and PCA to analyze possible sources of the 10 heavy metals in PM 2.5 in Kunming. Figure 6 shows that the EFs of Cr, Mn, Ni, Cu, Zn, As, Cd and Pb were higher than 10, which suggested that these heavy metals were significantly affected by anthropogenic sources. The EFs of V and Co were lower than 10 but higher than 1, which suggests that the two elements may have been affected by natural and anthropogenic sources.  The correlation among the elements can indicate the sources of the heavy metals. In this paper, we used SPSS17 software to calculate the correlation coefficient between elements. The coefficient matrix is shown in Table 5.  Table 5 shows that Cr and Mn had the highest correlation coefficient of 0.915. Previous studies showed that Cr, Mn and Ni were the main pollutants in road dust [61,62]. Manoli et al. [63] also reported that road dust had high levels of Cr and Mn, indicating that this might be the source.
V, Ni, Cu and Pb were strongly correlated, and some studies showed that they are usually related to the metallurgical industries and vehicle exhaust emissions [64][65][66]. Cu and Pb may also come from brake and tire wear [67]. V may be influenced by the type of pavement or industrial dust [64,68]. Therefore, V, Ni, Cu and Pb may mainly come from vehicle emissions and metallurgical industries.
There was no significant correlation between As and the other heavy metals since it was commonly used as a coal combustion indicator [69]. The table shows that As in Kunming PM 2.5 came mainly from coal combustion and did not highly correlate with vehicle emissions.

PCA of Heavy Metals
We use PCA to analyze heavy metal concentration data to identify possible sources of heavy metals in Kunming. The heavy metals data of all samples were input to the PCA model and analysis was performed. As shown in Table 6, three factors loading for heavy metals from PCA were listed, with varimax rotation of the data matrix. Factors 1-3 explained 49.43%, 18.73% and 12.61% of the total variance in the data set, respectively. In Factor 1, the heavy metals Ni, Cu, Zn, Cd and Pb have higher scores, which are 0.892, 0.862, 0.724, 0.912 and 0.880, respectively. The high load values of Ni, Cu, Zn, Cd and Pb indicate that Factor 1 is mainly from vehicle emissions and metallurgical industry sources [70,71]. The heavy metals Cr, Mn and Zn in Factor 2 are markers of traffic and road dust emission sources, so it is inferred that Factor 2 represents vehicle emissions and road dust [72]. In the Factor 3, heavy metal As has the highest score of 0.995. As is a marker component of coal, so it is inferred that Factor 3 is from coal combustion sources [48,73]. According to these results, we can know that vehicle emissions and metallurgical industry sources make the biggest important contribution to heavy metals in Kunming.

Health Risk Assessment of Heavy Metal Elements Exposure
In Kunming, the non-carcinogenic risk of heavy metals in PM 2.5 at three sampling sites was analyzed through respiration. The results during the entire sampling period are shown in Table 7. Among the three sampling sites, JDM had the highest non-carcinogenic risk of heavy metals, followed by DR and WM, but the trends for all three were the same. The risk levels of non-carcinogenic heavy metals for children, and men and women were Mn > Pb > Cu > V > Zn; and the order is children > men > women. The non-carcinogenic risk for each heavy metal was less than 1, indicating that it was within the acceptable range of respiratory exposure. However, most were affected by Mn and Pb. The carcinogenic risk of each heavy metal to each groups through respiration is shown in Table 8. At the three sampling sites, the carcinogenic risk of heavy metals for children was DR > WM > JDM; the carcinogenic risk of heavy metals for adults was DR > JDM > WM. The risk levels of carcinogenic heavy metals were Cr > As > Co > Cd > Ni. The carcinogenic risk of Cr to men and women in the DR sampling site exceeded 10 −4 . The carcinogenic risk of other heavy metals in all populations at the three sampling sites was between 10 −6 and 10 −4 , which was within the acceptable range. Cr posed the highest carcinogenic risk to all people through respiration, and it should receive more attention.

Discussion
We conducted this study to find out whether heavy metals from PM 2.5 had an impact on the death rate of the respiratory system in Yunnan. In the past few years, the number of deaths due to respiratory diseases in Yunnan Province is much higher than for the whole country, which implies that the environmental pollution has a significant impact on people in Yunnan. Heavy metals emitted from transportation sources may cause harm to the health of urban populations. Car ownership in Kunming has risen sharply in recent years, which is comparable to that of other medium and large cities in China. Therefore, we collected PM 2.5 samples in Kunming to explore the contribution of heavy metals to environmental pollution and human health. The results are in line with the original assumptions. First, we analyzed the spatial and temporal distribution characteristics of heavy metal concentrations, and found that at sampling site where traffic source emissions are more obvious, the changes in heavy metal concentrations between working days and rest days fluctuate significantly. At the same time, we used PCA to analyze the source of heavy metals. The results show that Cr, Mn, Cu, Zn and Pb are all related to traffic source emissions. In the health risk assessment of heavy metal exposure, Cr has a significant carcinogenic risk for people in Kunming, which may explain to a certain extent the reason why the death rate of the respiratory system in Yunnan is higher than the national average.
There are some limitations to this study. Firstly, there is inadequate explanation regarding the death rate due to problems with the respiratory system in Yunnan, indicating that more variables need considering, for example, the level of persistent organics in the atmosphere. Secondly, the data is a little old, so health risks may not reflect current reality well. Thirdly, due to the complexity of the terrain of the plateau city, we only studied the characteristics of heavy metal pollution in Kunming, which is far from being able to clarify the entire mechanism of atmospheric heavy metal pollution in the plateau area.

Conclusions
The concentrations of PM 2.5 at three sites (DR, JDM, WM) were 125.16 ± 71.27, 170.10 ± 104.10 and 114.00 ± 73.99 µg·m −3 , respectively, and as for the concentrations of the 10 heavy metals (Zn > Pb > Mn > As > Cu > V > Cr > Ni > Cd > Co), Cr, As and Cd exceeded the second level of China's national standard for ambient air quality (GB 3095-2012).
An EF analysis showed that Cr, Mn, Ni, Cu, Zn, As, Cd and Pb came mainly from anthropogenic sources, and that V and Co may have come from a mixture of natural and anthropogenic sources. For external sources, the clustering of air masses in Kunming showed that they mainly originated in Myanmar during the sampling periods, and the long-distance transmission of dense biomass burning in Myanmar in the spring affected the concentration levels of heavy metals.
We analyzed the concentration of heavy metals on working and rest days. Almost all concentrations fluctuated intensely on working and rest days at all three sampling sites, especially DR and JDM, which suggested that heavy metals from traffic made an important contribution to the ambient air. At the same time, the PCA results show that Ni, Cu, Zn, Cd and Pb came mainly from vehicle and metallurgical industry emissions (49.43% of the total variance), and Cr, Mn and Zn came from vehicle emissions and road dust (18.73% of the total variance). These results all showed that motor vehicles were already a critical air pollution source of heavy metals.
The total non-carcinogenic risk of heavy metals was less than 1, which was in the acceptable range of respiratory exposure. The risk levels of carcinogenic heavy metals were in the order: Cr > As > Co > Cd > Ni. The carcinogenic risk of Cr to men and women at the DR sampling site was greater than 10 −4 .
To reduce urban traffic emissions and protect people's health, it is recommended that preferential policies be enacted to encourage consumers to use clean energy vehicles on the Yunnan-Guizhou Plateau.  Data Availability Statement: The data used in this paper can be provided by Jianwu Shi (shijianwu@kust.edu.cn).

Conflicts of Interest:
The authors declare that there are no competing financial interests that could inappropriately influence the contents of this manuscript.