Chemical Composition, Sources, and Health Risk Assessment of PM2.5 and PM10 in Urban Sites of Bangkok, Thailand

Of late, air pollution in Asia has increased, particularly in built-up areas due to rapid industrialization and urbanization. The present study sets out to examine the impact that pollution can have on the health of people living in the inner city of Bangkok, Thailand. Consequently, in 2021, fine particulate matter (PM2.5) and coarse particulate matter (PM10) chemical composition and sources are evaluated at three locations in Bangkok. To identify the possible sources of such particulates, therefore, the principal component analysis (PCA) technique is duly carried out. As determined via PCA, the major sources of air pollution in Bangkok are local emission sources and sea salt. The most significant local sources of PM2.5 and PM10 in Bangkok include primary combustion, such as vehicle emissions, coal combustion, biomass burning, secondary aerosol formation, industrial emissions, and dust sources. Except for the hazard quotient (HQ) of Ni and Mn of PM2.5 for adults, the HQ values of As, Cd, Cr, Mn, and Ni of both PM2.5 and PM10 were below the safe level (HQ = 1) for adults and children. This indicates that exposure to these metals would have non-carcinogenic health effects. Except for the carcinogenic risk (HI) value of Cr of PM2.5 and PM10, which can cause cancer in adults, at Bangna and Din Daeng, the HI values of Cd, Ni, As, and Pb of PM2.5 and PM10 are below the limit set by the U.S. Environmental Protection Agency (U.S. EPA). Ni and Mn pose non-carcinogenic risks, whereas Cr poses carcinogenic risks to adults via inhalation, a serious threat to the residents of Bangkok.


Introduction
Globally, particulate matter (PM) levels are increasing rapidly, with most cases occurring in the developing world and causing significant health and environmental implications [1]. PM pollution is one of the significant global risks to human health. Current global air quality regulations aim to decrease the concentrations of PM 2.5 and PM 10 , as well as the chemical composition associated with adverse health implications [2]. The population in Asia has increased due to rapid industrialization, and, thus, energy consumption has also increased with population expansion. Major air pollutants have been emitted in urban 2 of 17 areas, intensified by climatic and topographical conditions that frequently inhibit the dispersion of pollutants in metropolitan regions [3,4]. Identifying the chemical composition, sources, and toxicity of PM has become essential.
The chemical composition of PM can vary significantly due to the influence of emission sources, weather conditions, and the possibility of dispersion [5,6]. The carbonaceous components of PM primarily consist of elemental carbon (EC) and organic carbon (OC), which account for approximately 10 and 45% of the mass of PM 2.5 and PM 10 , respectively [7,8]. One of the common components of atmospheric particles is water-soluble inorganic species (WSIS). Meteorological and geographical conditions, industrial particle emissions, transportation, agricultural practices, and natural sources significantly influenced WSIS [9]. Heavy metals are poorly biodegradable, carcinogenic, and contribute less to the mass of PM 2.5 and PM 10 . When humans are exposed to high doses of heavy metals for a long period, the heavy metals become toxic [10].
Human health and well-being are considered to be dependent on clean air. Air quality standards have been set in many countries to protect the public's health. These standards have become more important to environmental policies and national risk management [11]. Due to the high concentration of atmospheric PM 2.5 and PM 10 , epidemiological studies have demonstrated that humans are exposed to a serious risk of respiratory and cardiovascular diseases [12]. Several studies have found strong evidence that exposure to PM has directly caused cardiopulmonary and ischemic heart disease-related mortality [13]. The size and composition of particulate matter determine its toxicity. Researchers have found that particle composition might have a greater impact than particle size in some cases [14,15]. In addition to its adverse impacts on physical health that have been extensively studied, PM pollution also significantly impacts the economy [16,17].
In this paper, the carcinogenic and non-carcinogenic effects of metals and metalloids on adults and children have been estimated using U.S. EPA methodologies. During the wintertime, the high concentration of toxic metals in PM 2.5 poses serious carcinogenic (HI) and non-carcinogenic (HQ) risks via inhalation exposure to the residents of Lahore and Peshawar [18]. In urban parks in Beijing, the non-carcinogenic (HQ) risk of heavy metals via inhalation exposure was below the safe level (HQ = 1) in road dust; however, the HI values of heavy metals for children were higher than for adults [19]. The study aims to chemically characterize PM 2.5 and PM 10 at the urban sites of Bangkok (Ari, Din Daeng, and Bangna) in order to identify possible sources. The principal component analysis (PCA) method has been duly carried out. The carcinogenic and non-carcinogenic risks of heavy metals via inhalation exposure have been evaluated in the study area. Thus, the results of this study will provide health professionals and the general public with a better understanding of air quality, eliciting a sense of awareness about the toxicity of metal-bounded particles and how to control them.

Study Area and Sampling
For the monitoring of PM 2.5 and PM 10 , three sampling locations in Bangkok were selected: (1) Ari, (2) Din Daeng, and (3) Bangna. In Figure 1, the map of the sampling locations is shown.
2.1.1. Ari (13 • 46 59.6 N 100 • 32 25.8 E) This station is in the Pollution Control Department area. The height of air sampling equipment is 5 m above ground level. This area is surrounded by government office buildings, such as the Treasury Department, Ministry of Finance, Revenue Department, Ministry of Natural Resources and Environment, Department of Water Resources, Department of Environmental Quality Promotion, and Institute of Public Relations. In addition, this station is in an area with dormitories and condominiums. Ari was designated as the representative of the residential area in the city. This station is in the Pollution Control Department area. The height of air sampling equipment is 5 m above ground level. This area is surrounded by government office buildings, such as the Treasury Department, Ministry of Finance, Revenue Department,  The height of air sampling equipment is 5 m above ground level. This station is in the Thai Meteorological Department area. According to the Department of Industrial Works website, many factories are located within a 2 km radius, including car paint, metals, plastic packaging production, car repairs, garment factories, wood processing industries, and producing publications.
The samples of PM 2.5 and PM 10 were collected every 24 h, at each sampling site, from 8:00 am to 7:00 am (local time) the next day. The samples were collected on quartz fiber filter (PallFlex) by drawing air at a rate of 40 L/min using a Nano-sampler equipped with PM 2.5 and PM 10 impactors. At each sampling site, 30 samples of PM 2.5 and PM 10 were collected. Before and after sampling, the filters were weighed on an analytical balance and stored for 24 h at a temperature of 25 • C and relative humidity of 50%. The sampled filters were stored in a refrigerator at −40 • C for further analysis.

Chemical Analysis
The carbonaceous species, water-soluble ions, and metal elements were evaluated in PM 2.5 and PM 10 samples. The organic carbon (OC) and elemental carbon (EC) were measured using the thermal/optical reflectance (TOR) method of a DRI THERMAL OPTICAL Carbon Analyzer. To measure OC and EC, the filter punch size of 0.523 cm 2 was heated in a totally oxygen-free helium (He) atmosphere. All OC fractions were released on a filter at four stepwise temperatures of 120 • C (OC1), 250 • C (OC2), 450 • C (OC3), and 550 • C (OC4). EC was measured at 550 • C (EC1), 700 • C (EC2), and 800 • C (EC3), and the pure He was changed to an O 2 /He 2% mixture. A laser-monitored optical pyrolyzed carbon (OP). OC was obtained by OC1 + OC2 + OC3 + OC4 + OP, whereas EC was formed by EC1 + EC2 + EC3 − OP.
The water-soluble ions were measured by ultrasonically extracting one quarter of the filter with 10 mL of ultrapure water for 60 min. A Teflon filter (PTFE) with a pore size of 0.45 µm was used to remove the insoluble species of the extract. Next, an electrical conductivity detector examined the water-soluble ions using ion chromatography (ICS-1100). An Ion Pac AS11 connected to an Ion Pac AG23 pre-column was used to separate the anions, Cl − , NO 3 − , and SO 4 2− . The cations, NH 4 + , Na + , K + , Mg 2+ , and Ca 2+ , were separated with an Ion Pac CS12A analytical column connected to an Ion Pac CG12A pre-column.
Among the metals investigated are Na, Mg, Al, K, Ca, Sc, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Se, Cd, Ba, Ce, Pt, and Pb. An acid solution of 1.5 mL of HCl, 1.0 mL of HNO 3 , and 1.0 mL of HBF 4 , diluted with 10 mL of ultrapure water, was prepared. A 10 mL acid solution was used to digest one quarter of each filter. Using a microwave digestion unit, the temperature of the digestion solution was maintained at 170 • C for 10 min (ramping time, 15 min) and then increased to 180 • C for 15 min (ramping time, 20 min). After cooling, the extracted samples were filtered through a cellulose filter with a pore size of 6 µm and diluted with 10 mL of ultrapure water. Inductively coupled plasma-mass spectrometry (Agilent Technologies 7900 ICP-MS, Santa Clara, CA, USA) was used to measure the metals in the solution. To correct the background concentrations of carbonaceous species, ions, and metals, blank filters were measured using the same method as sample filters. The standard chemicals were applied to check the performance of three chemical analysis methods. Glucose standard was used for calibrating the carbon analysis technique and calibration curves of ion, and metal standards were also used for ion and elemental analysis, respectively.

Estimation of Primary Organic Carbon and Secondary Organic Carbon
The concentration of primary organic carbon (POC) and secondary organic carbon (SOC) was estimated following the EC-tracer method. The minimum OC/EC ratio was used to estimate POC and SOC in aerosol [20]. The following equations were used to estimate POC and SOC: where OC represents the measured OC concentration; OC EC min represents the estimated minimum OC/EC ratio.

Principal Component Analysis (PCA)
Principal component analysis (PCA), one of the receptor models, was used to identify the potential sources of PM 2.5 and PM 10 [21]. PCA considers the concentration of a species at a receptor site to represent the sum of the contributions from several sources/factors, as shown as follows: where x ij represented the concentration of jth species in the ith sample, g ik is contribution of the kth factor/source to the ith sample, and f kj represented the relative quantity of the jth species in the kth source. In an ideal situation, e ij should only contain an experimental error. However, in practice, some unexplained and unknown errors may be included. PCA was performed using a statistical package for the social sciences (SPSS). The PCA analysis used the concentrations of the carbonaceous species, ions, and metals as variables.

Health Risk Assessment of Heavy Metals
The U.S. EPA health risk assessment models were used to estimate heavy metal carcinogenic and non-carcinogenic health risk assessments [22]. The following subsections are the three steps in the process of evaluating health risk.

Estimating Exposure Concentrations
The following equation was used to determine the exposure concentration (ExC) via inhalation [22]: where ExC represents the exposure concentration (µg/m 3 ); CA is the concentration of studied heavy metals (µg/m 3 ) [23]; ET represents exposure time, which is considered to be 24 h/day; EF represents exposure frequency, and is supposed to be 350 days/year; and ED is exposure duration, and it is assumed to be 24 years for adults and 6 years for children. AT represents the average time, which is ED × 365 days/year × 24 h/day for non-carcinogens, whereas for carcinogens, it is 70 years × 365 days/year × 24 h/day.

Non-Carcinogenic Risk Assessment
After calculating the ExC values, each metal's hazard quotient (HQ) was determined to assess its non-carcinogenic risk: where the hazard quotient is HQ, the exposure concentration is ExC (µg/m 3 ), and the inhalation reference concentration is RƒC (mg/m 3 ). An HQ value < 1 indicates that there are no non-carcinogenic effects. Conversely, there is a high chance of non-carcinogenic effects when the HQ values > 1 [22,24].

Carcinogenic Risk Assessment
The following equation can be used to calculate the carcinogenic risk (CR) via inhalation exposure [22]: where IUR represents the inhalation unit risk (µg/m 3 ) -1 . For regulatory purposes, the range of permissible risk is from 10 -6 (1 in 1,000,000) to 10 -4 (1 in 100,000) [25]. Cd, As, Cr, and Ni are identified as human carcinogens via inhalation, as reported by the International Agency for Research on Cancer (IARC) [26]. Lead (Pb) is also a toxic metal that increases cancer risk. The RƒC, IUR, and exposure variables for these models were acquired from the Integrated Risk Information System (IRIS) and the U.S. EPA Regional Screening Levels (RSL).

Meteorological Parameters
The average data of meteorological parameters such as temperature, wind speed, relative humidity, wind direction, and rainfall were obtained from http://www.wunderground. com (accessed on 21 May 2022). During sampling periods, no rainfall event occurred; thus, the rainfall data were not included. In Figure 2, the weather conditions are depicted. During the sampling period, the temperature ranged from 19.0 to 29.0 • C, with an average value of 25.0 ± 2.2 • C. The relative humidity ranged from 50.6 to 92.0%, with an average value of 69.6 ± 12.9%. Similarly, the windspeed data were in the range of 0.1 to 0.6 m/s, with average value of 0.2 ± 0.1 m/s. The wind directions during the sampling time were north and north-eastern.
no non-carcinogenic effects. Conversely, there is a high chance of non-carcinogenic effects when the HQ values > 1 [22,24].

Carcinogenic Risk Assessment
The following equation can be used to calculate the carcinogenic risk (CR) via inhalation exposure [22]: CR = IUR × ExC (6) where IUR represents the inhalation unit risk (µg/m 3 ) -1 . For regulatory purposes, the range of permissible risk is from 10 -6 (1 in 1,000,000) to 10 -4 (1 in 100,000) [25]. Cd, As, Cr, and Ni are identified as human carcinogens via inhalation, as reported by the International Agency for Research on Cancer (IARC) [26]. Lead (Pb) is also a toxic metal that increases cancer risk. The RƒC, IUR, and exposure variables for these models were acquired from the Integrated Risk Information System (IRIS) and the U.S. EPA Regional Screening Levels (RSL).

Meteorological Parameters
The average data of meteorological parameters such as temperature, wind speed, relative humidity, wind direction, and rainfall were obtained from http://www.wunderground.com (accessed on 21 May 2022). During sampling periods, no rainfall event occurred; thus, the rainfall data were not included. In Figure 2, the weather conditions are depicted. During the sampling period, the temperature ranged from 19.0 to 29.0 °C, with an average value of 25.0 ± 2.2 °C. The relative humidity ranged from 50.6 to 92.0%, with an average value of 69.6 ± 12.9%. Similarly, the windspeed data were in the range of 0.1 to 0.6 m/s, with average value of 0.2 ± 0.1 m/s. The wind directions during the sampling time were north and north-eastern. Figure 2. Meteorological parameters, such as temperature, wind speed, relative humidity, and wind direction, during the sampling period in Bangkok metropolitan area, Thailand. Figure 2. Meteorological parameters, such as temperature, wind speed, relative humidity, and wind direction, during the sampling period in Bangkok metropolitan area, Thailand. found to be higher than the current study [27]. Another study carried out in Bangkok from 2003 to 2007 also found that the annual concentration of PM 10 was higher than in the current study [28]. The average concentration of PM 2.5 was reported in the residential area (50.3 µg/m 3 ), university area (32.8 µg/m 3 ), industrial zone (44.5 µg/m 3 ), and agricultural zone (29.5 µg/m 3 ) of Khon Kaen province in Thailand. These results are slightly higher than in the current study [29].   Compared to Ari and Bangna, the average concentration of PM 2.5 and PM 10 in Din Daeng was higher, since the Din Daeng site is close to an industrial zone. Furthermore, PM 2.5 and PM 10 mass concentrations exhibited similar variations at all sampling sites. The average concentration of PM 2.5 measured in Bangkok (23.6 µg/m 3 ) was slightly lower than all sites of the current study, whereas the PM 10 (53.4 µg/m 3 ) measured in the previous study was higher than in the current study [30]. In this region, the mass of PM may have been influenced by biomass combustion, industrial emissions, heavy traffic, the open burning of agricultural residues, and construction and demolition [31,32]. The concentration of PM 2.5 in the current study is reported to be higher than the permissible limit set by WHO, and can cause serious health problems. Therefore, this study highly recommends that in urban areas, the aerosol study should focus more on particles of 2.5 µm or less. As shown in Table 2, according to Spearman's rank correlation, the meteorological parameters, such as relative humidity, show strong correlation with PM 2.5 at Ari and Din Daeng, and a moderate correlation at Bangna. Likewise, windspeed shows a negative correlation with the mass concentration of PM 2.5 at Ari and Din Daeng. In contrast, there is no correlation between weather conditions and the mass of PM 10 .  Figure 3, OC is more dominant than EC at all sampling sites. Higher OC was observed at Din Daeng in both PM 2.5 (38.0 µg/m 3 ) and PM 10 (22.3 µg/m 3 ), followed by Ari: PM 2.5 (28.2 µg/m 3 ) and Bangna: PM 10 (11.2 µg/m 3 ). EC exhibited similar variation characteristics at all sites for PM 2.5 and PM 10 . These results are similar to previous studies conducted in various cities of China [33,34]. The strong correlation between OC and EC suggests that both are emitted from the same sources. As shown in Figure 4, a strong correlation between OC and EC is observed in both PM 2.5 and PM 10 at Ari and Din Daeng. At the same time, there is a moderate/weak correlation between OC and EC of PM 2.5 and PM 10 at Bangna. The weak OC and EC correlations indicate the secondary aerosol formation under favourable conditions for the gas-to-particle conversion of volatile organic compounds (VOCs) through photochemical reactions [35]. Possible sources of the high concentrations of OC and EC include increased coal combustion, biomass burning, and unfavorable meteorological conditions [18]. OC consists of both primary organic carbon (POC) and secondary organic carbon (SOC). The contribution of SOC to OC in PM 2.5 was 69.5% (Ari), 78.1% (Din Daeng), and 75.7% (Bangna). In contrast, the SOC contribution to OC in PM 10 was 57.0% (Ari), 58.3% (Din Daeng), and 63.3% (Bangna). Due to the role of high photochemical activity in its formation, SOC significantly contributes to OC [36].

Health Risk Assessment
The serious carcinogenic and non-carcinogenic toxicity of PM2.5-and PM10-bound metals in humans depends on the frequency, exposure duration, and concentration of metals. A health risk assessment of heavy metals for adults and children via inhalation pathways was determined. Table 3 presents the hazard quotient (HQ) and cancer risk (CR) values of PM2.5 and PM10 for both adults and children. At all sites, the non-carcinogenic risks of As, Cd, Cr, Mn, and Ni of PM2.5 and PM10 were below the safe level (HQ = 1) for children. In contrast, at Ari and Din Daeng, the HQ values of Ni and Mn of PM2.5 for adults exceeded the safe level. Moreover, the HQ value of Ni for adults in PM2.5 at Bangna also exceeded the safe level, indicating that the inhalation of these metals can cause non-carcinogenic health effects. NI and Mn posed no non-carcinogenic risks to both adults and children in Nanjing, China [50], and Mexico [51]. As illustrated in Table 3, the non-carcinogenic risks of PM2.5 and PM10 for adults and children were in decreasing order: Cd > Cr > As > Mn > Ni at Ari, Din Daeng, and Bangna. Normally, coarse particles tend to deposit and may be expelled in the nasal-pharyngeal region. In contrast, PM2.5 can enter the respiratory tract, penetrate deep into the lungs, and cause serious health problems. The non-carcinogenic risk posed by the toxic elements of PM2.5 for adults was significantly higher in Bangkok.

Chemical Species
In Table 1a,b, the mean concentrations of water-soluble inorganic species (WSIS) in fine and coarse particles at Ari, Din Daeng, and Bangna are shown. At Ari, the WSIS concentrations in PM 2.5 were reported to be in descending order of Cl − > NH 4 + > K + > Na + > Ca 2+ > SO 4 2− > NO 3 − > Mg 2+ ; the trend in the concentration at Din Daeng and Bangna followed the order of Ari. As for PM 10 , the concentration of WSIS at Ari, Din Daeng, and Bangna was as follows: SO 4 2− > Cl − > Ca 2+ > NO 3 − > Na + > K + > Mg 2+ > NH 4 + . At all three sites, the most abundant cations with high concentrations were ammonium (NH 4 + ), potassium (K + ), calcium (Ca 2+ ), sodium (Na + ), and magnesium (Mg 2+ ), whereas the anions with the highest concentrations were chloride (Cl − ), sulfate (SO 4 2− ), and nitrate (NO 3 − ). Of the anions, Cl − , and in cations, Ca 2+ , was found dominant at all sites. Crustal inputs from the exposed surface soil or re-suspension of soil dust can contribute to the higher concentration of Ca 2+ . At all sites, the higher concentration of Ca 2+ in PM 2.5 and PM 10 can also be attributed to the drier weather conditions [37]. During the secondary gas phase, SO 4 2− and NO 3 − are mainly formed [38]. NOx is a well-known precursor of NO 3 − , mainly released by vehicle exhaust (mobile sources). However, SO 4 2− is emitted mainly by coal-fired processes (stationary sources). The higher contributions of SO 4 2− and NO 3 − among anions to the mass of PM at all sites can be attributed to vehicular and industrial emissions [39]. The formation of sulfate-rich particles is mainly due to a change in equilibrium between the particle and gaseous phases caused by lower temperatures and higher humidity [40]. The HNO 3 and NH 3 gas-phase reaction formed NO 3 − ions in PM 2.5 [41]. In PM 10 , sea salt or soil dust particles react with HNO 3 gas to form NO 3 − [42]. The major source of ammonium salt is the formation of secondary particles by aqueous or gas-phase reactions of NH 3 with H 2 SO 4 , HNO 3 , and HCl. Potassium in PM 2.5 is derived mainly from biomass burning, crop residues, and coal combustion [43]. Minerals of crustal origin are the major sources of potassium in the coarse mode [44]. Potential sources of Cl − include the burning of waste plastic as a fuel at the residential and industrial levels, and its open burning [45]. Incinerators and power plants also contributed to Cl − emissions [46]. Na + and Cl − ions also contributed to the mass of aerosol due to the long-range transport of marine dust [44].
In Table 1a,b, the mean concentration of metal elements in PM 2.5 and PM 10 at all sites are listed. The metal contribution to the mass of PM 2.5 at Ari, Din Daeng, and Bangna was 29.0%, 23.2%, and 35.1%, respectively, whereas for PM 10 , the metal contribution at all sampling sites was 21.1% (Ari), 20.0% (Din Daeng), and 26.2% (Bangna). A significant contribution of metals to the mass of PM 2.5 and PM 10 was observed at all sampling sites. Anthropogenic emission sources, such as biomass burning, coal combustion for barbecues on streets and roadsides, and industrial emissions, are possible sources of high contributions of metals to the mass of PM. In addition, unfavorable climatic conditions, such as temperature inversions and low wind speeds at all sampling sites, inhibit the dispersion of pollutants. K is used as a marker of biomass and wood burning in several source apportionment studies conducted in various regions [47]. Therefore, the high concentration of K is due to biomass burning and coal combustion in current study. Gasoline additives, brake pads, and road dust re-suspension contribute to the high concentration of Pb that can cause human health problems [48]. Fuel combustion in cars, buses, trucks, motorcycles, rickshaws, etc., and tyre wear are the possible sources of the higher zinc (Zn) concentration at all sites [49].

Health Risk Assessment
The serious carcinogenic and non-carcinogenic toxicity of PM 2.5 -and PM 10 -bound metals in humans depends on the frequency, exposure duration, and concentration of metals. A health risk assessment of heavy metals for adults and children via inhalation pathways was determined. Table 3 presents the hazard quotient (HQ) and cancer risk (CR) values of PM 2.5 and PM 10 for both adults and children. At all sites, the non-carcinogenic risks of As, Cd, Cr, Mn, and Ni of PM 2.5 and PM 10 were below the safe level (HQ = 1) for children. In contrast, at Ari and Din Daeng, the HQ values of Ni and Mn of PM 2.5 for adults exceeded the safe level. Moreover, the HQ value of Ni for adults in PM 2.5 at Bangna also exceeded the safe level, indicating that the inhalation of these metals can cause non-carcinogenic health effects. NI and Mn posed no non-carcinogenic risks to both adults and children in Nanjing, China [50], and Mexico [51]. As illustrated in Table 3, the non-carcinogenic risks of PM 2.5 and PM 10 for adults and children were in decreasing order: Cd > Cr > As > Mn > Ni at Ari, Din Daeng, and Bangna. Normally, coarse particles tend to deposit and may be expelled in the nasal-pharyngeal region. In contrast, PM 2.5 can enter the respiratory tract, penetrate deep into the lungs, and cause serious health problems. The non-carcinogenic risk posed by the toxic elements of PM 2.5 for adults was significantly higher in Bangkok.
The health risk assessment of heavy metals indicated that the Cr of PM 2.5 and PM 10 pose a carcinogenic risk to adults at Bangna and Din Daeng. Similar results for the carcinogenic risk value of Cr exceeded the permissible limits obtained in India [52] and Tianjin, China [53]. In contrast, at all sites, the carcinogenic risk values, f, of Cd, Ni, As, and Pb were below the safe level for both adults and children. The total carcinogenic risk exceeded the permissible limit for adults at all sites, exposing adults in Bangkok to high carcinogenic risk. The non-carcinogenic risk posed by Ni and Mn and the carcinogenic risk by Cr to adults via inhalation are the most serious threats to the residents of Bangkok.

Source Apportionment Using Principal Component Analysis (PCA)
PCA was used to identify sources associated with carbonaceous species, inorganic ions, and major elements in PM 2.5 and PM 10 . The PCA with the varimax rotational factor analysis method was conducted using the Statistical Package for Social Sciences (SPSS v20.0) to identify the sources of PM 2.5 and PM 10 at Ari, Din Daeng, and Bangna. Factor 1 contributes 35.6%, 26.8%, and 31.8% to the total variance at Ari, Din Daeng, and Bangna, respectively. Factor 1 of PM 2.5 was interpreted as industrial emissions and dust/crustal re-suspension at Ari having high Mg 2+ , Ca 2+ , Al, Ti, Cr, Mn, As, and Ni loadings, as shown in Table 4a. At Din Daeng, factor 1 represents industrial emissions and dust re-suspension with high loadings of Mg 2+ , Ca 2+ , Ti, Mn, Ni, and As. In contrast, at Bangna, factor 1 is highly loaded with OC, EC, NH 4 + , K + , NO 3 − , SO 4 2− , Cl − , and Fe, which represent primary combustion sources, including vehicle emissions, biomass burning, coal combustion, and secondary aerosol formation. Factor 2 contributed 20.8% (Ari), 18.7% (Din Daeng), and 25.2% (Bangna) of the total variance. Factor 2 of PM 2.5 at Ari, Din Daeng, and Bangna represents secondary aerosol formation plus sea salt, primary combustion sources (biomass burning and vehicle emission), and industrial emissions. High loadings of Na + , NH 4 + , NO 3 − , SO 4 2− , Cl − , and Fe are found at Ari. At Din Daeng, high loadings of OC, EC, NH 4 + , Cl − , K + , and Fe are found. Ti, As, Ni, and Mn are highly loaded at Bangna. At Ari, factor 3 is considered the primary combustion source (vehicle emission and biomass burning), contributing 16.0% to the total variance, and is highly loaded with OC, EC, and K + . Factor 3 contributes 17.7% of the variance in Din Daeng, loaded with Na + , NO 3 − , and SO 4 2− . Therefore, it is categorized as sea salt and secondary aerosol formation. The third factor in Bangna is rich in Na + and moderate in Al, indicating a sea salt source, contributing 12.0% of the variance. The use of some markers is perplexing and can be affected by various sources, specifically K + from wood and biomass burning, Cl − from sea salt and coal combustion, and Mg from crustal emissions. The presence of NO 3 − indicates the probability of a marine origin through HNO 3 condensation [54,55]. Factor 4 at Ari is rich in Cu and Zn, accounting for 9.2% of the variance. The traffic source profile is also associated with Cu and Zn [56]. As a result, factor 4 at Ari is considered traffic/road dust. Factor 4, rich in Al and Ti at Din Daeng, contributed 12.7% of the variance. Al, Ti, and other trace elements have been used in different studies as a source of resuspension of soil/crustal dust [57]. As a result, at Din Daeng, factor 4 is classified as crustal/dust re-suspension. At Bangna, factor 4 contributed 11.0% to the total variance and is rich in Cu and Zn, indicating traffic/road dust. According to PCA results, the dominant sources of PM 2.5 at Ari, Din Daeng, and Bangna are primary combustion sources (biomass burning, vehicle emissions, coal combustion), road dust re-suspension, secondary aerosol formation, and sea salt contribution. The species with a high loading in each factor is marked as underline.
As shown in Table 4b, the sources of PM 10 at Ari, Din Daeng, and Bangna identified in the same way as PM 2.5 . At Ari, Din Daeng, and Bangna, factor 1 is seen to contribute 29.9%, 27.3%, and 23.8% to the total variance. At Ari, factor 1 of PM 10 was loaded with NO 3 − , SO 4 2− , Ti, Mn, Ni, and As, which are considered secondary aerosol formation and industrial emissions. Moreover, gaseous precursors such as NOx and SOx emitted from vehicles, fossil fuels, coal combustion, and industrial processes contribute to the formation of secondary nitrate and sulfate sources [58]. At Din Daeng, factor 1 is loaded with OC, EC, NH 4 + , K + , Cl − , Cu, and Fe. Factor 1 represented primary combustion sources at Ding Daeng. At Bangna, the same factor is significantly loaded with NH 4 + , K + , Al, and Fe, representing biomass burning. K + is used as a marker of biomass burning in various apportionment studies [59]. In several studies, the combination of NH 4 + and K + has been used as a marker of combustion emissions [60]. At Ari, Din Daeng, and Bangna, factor 2 of PM 10 contributed 23.9%, 25.6%, and 21.4% to the total variance, respectively. Factor 2 at Ari was loaded with OC, EC, NH 4 + , Cl − , Fe, and Cu, and represented primary combustion sources such as vehicle emissions. At Din Daeng and Bangna, factor 2 represents crustal/road dust re-suspension and vehicle emissions, which are the possible sources of PM 10 . At Din Daeng, factor 2 is loaded with Cl − , Al, Ti, Cr, Cu, and Zn, whereas at Bangna, OC, EC, Cr, Cu, and Zn contributed highly to factor 2. Two-stroke engines emit zinc, which is used as a fuel additive to reduce tyre wear [56]. Factor 3 accounted for 16.2% (Ari), 17.9% (Din Daeng), and 20.0% (Bangna) of the total variance. Biomass combustion, sea salt/secondary aerosol formation, and industrial emissions are possible sources of PM 10 at Ari, Din Daeng, and Bangna, represented by factor 3. At Ari, factor 3 is loaded with Na + , K + , and Zn, which represent biomass combustion. Factor 3 at Din Daeng is loaded with Na + , Mg 2+ , NO 3 − , SO 4 2− , and Ni, indicating sea salt/secondary aerosol formation, and at Bangna, factor 3 is loaded with Ti, Mn, Ni, and As, representing industrial emissions.
Factor 4 indicated crustal re-suspension, industrial emissions/sea salt, and secondary aerosol formation. At Ari, Din Daeng, and Bangna, the possible sources of PM 10 in factor 4 account for 9.5%, 17.6%, and 17.1% of the total variance. According to the results of PCA, the major combustion sources, such as vehicle emissions, coal combustion, and biomass burning, and secondary aerosol formation, industrial emissions, sea salt contribution, and dust sources contributed to PM 2.5 and PM 10 mass during the study period.

Conclusions
This paper investigated the PM 2.5 and PM 10 mass concentrations at three sampling locations in Bangkok. At all sampling sites, the daily concentration of PM 2.5 exceeded the permissible limit of WHO guidelines. However, PM 10 concentrations were found to be within the permissible limits. At Din Daeng, the average concentrations of PM 2.5 and PM 10 were comparatively higher than at Ari and Bangna. The high PM 2.5 concentrations indicated serious fine particle pollution in Bangkok. At all sites, OC and EC were the dominant species reported in PM 2.5 and PM 10 . At Din Daeng, the OC and EC concentrations were higher than at Ari and Bangna. The strong OC and EC correlation at Ari and Din Daeng indicated that they are emitted from the same sources. At Bangna, a weak correlation between OC and EC in both PM 2.5 and PM 10 indicated secondary aerosol formation through photochemical reactions under favorable conditions. The SOC contribution to OC in PM 2.5 at Ari, Din Daeng, and Bangna was 69.5%, 78.1%, and 75.7%, respectively. Likewise, in PM 10 , the SOC contribution to OC was 57.0% (Ari), 58.3% (Din Daeng), and 63.3% (Bangna). The contribution of SOC is due to the high photochemical activity.
A health risk assessment of heavy metals in PM 2.5 and PM 10 through the inhalation pathway for adults and children was determined at Ari, Din Daeng, and Bangna. The noncarcinogenic risks of As, Cd, Cr, Ni, and Mn of PM 2.5 and PM 10 were below the permissible limits for children, except Mn and Ni of PM 2.5 for adults, which exceeded the safe level at Ari and Din Daeng. In contrast, the Ni value of PM 2.5 at Bangna for adults was also higher than the safe level. The non-carcinogenic risk of PM 10 for both adults and children at all sites was below the safe level. The health risk assessment of heavy metals shows that Cr in PM 2.5 and PM 10 poses a carcinogenic risk to adults at Bangna and Din Daeng. The total carcinogenic risk for adults exceeded the permissible limits, suggesting a carcinogenic risk in Bangkok. The possible sources of PM 2.5 and PM 10 at Ari, Din Daeng, and Bangna are dust re-suspension, primary combustion sources such as biomass burning and vehicle emissions, coal combustion, secondary aerosol formation, and sea salt. Thus, the influence of marine and continental air masses contributed to the mass of PM 2.5 and PM 10 during the sampling period in Bangkok. Legislators can use our findings to formulate efficient