2.1 Sampling site

The National Institute of Development Administration (NIDA) Air Quality Observatory Site (NAQOS) is conveniently positioned on the NIDA campus in Bangkok, Thailand (see Fig 1). The campus has a mixture of urban and suburban areas, making it suitable for monitoring air quality from a variety of potential sources. The NAQOS is positioned on the rooftop of the Navamindradhiraj Building at an approximate height of 60 m. This observatory site was established to continuously monitor and assess air quality in the Bangkok metropolitan area, which is notorious for air quality issues owing to urbanisation, traffic congestion, and industrial operations. The information gathered at this site is used for environmental studies and policymaking targeted at improving air quality and public health.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. A map of NIDA air quality observatory site (NAQOS) in Bangkok, Thailand.

https://doi.org/10.1371/journal.pone.0320405.g001

PM samples across six size fractions (i.e., PM_0.65–1.1, PM_1.1–2.1, PM_2.1–3.3, PM_3.3–4.7, PM_4.7–7.0, and PM_{7.0 and above}) were collected using a six-stage cascade impactor setup from January 2^nd to April 7^th, 2023. Sampling took place at the NAQOS, which is equipped to capture ambient air quality data under standard environmental conditions. A total of 66 air samples (i.e., 6×11) were collected continuously from 9:00 AM Monday to 9:00 AM Saturday over an 11-week period, amounting to a total sampling duration of 120 hours per week, to ensure consistency and capture potential variations in particulate composition across weekdays. The sampling frequency and time frame were chosen to provide a comprehensive dataset of PM distribution and PAH concentration in relation to environmental and anthropogenic factors. Each PM fraction was collected onto pre-conditioned filters at designated flow rates to allow subsequent chemical analysis of particulate-bound compounds. In addition, it is also crucial to underline that all meteorological parameters were obtained from the weather station adjacent to NAQOS, which has been operated by Thai Meteorological Department Automatic Weather System (http://www.aws-observation.tmd.go.th/rprt/weatherDays).

2.2 Sampling protocol

To accumulate airborne aerosols with different particle sizes, Tisch Environmental employed ambient viable Andersen cascade impactors. The cascade impactor (TE-10–800) provided six-stage configurations (i.e., PM_0.65–1.1, PM_1.1–2.1, PM_2.1–3.3, PM_3.3–4.7, PM_4.7–7.0, and PM_{7.0 and above}) using a pump for sampling at 28.3 l m⁻¹. These ambient viable impactors were precision-machined from high-quality aluminium plates and contained holes of decreasing diameter. The relevant particulates strike the stage when the ambient air cascades through the stages, whereas smaller particulates continue to flow through the stages until they are entrapped on the associated plate. The process of sampling aerosols using a 6-stage cascade impactor entails a methodological approach for gathering particles of varying sizes onto substrates in preparation for subsequent analysis. The general methodology and principle for sampling aerosols using the 6-stage cascade impactor is described in Appendix I (see Supporting Material).

Size-segregated particle sampling tools, such as Tisch Environmental Andersen Cascade Impactors, MOUDI, Dekati Impactors, low-volume samplers like the Harvard Impactor, and high-volume samplers utilizing cyclones or virtual impactors, provide distinct benefits based on research goals [33–36]. Andersen Cascade Impactors, which were employed in this study, are ideal for multi-stage particle size analysis and viable particle sampling, enabling researchers to collect bioaerosol data throughout coarse and fine particle spectra [36]. These impactors necessitate meticulous maintenance, regulation of particle rebound, and function at reduced flow rates, thereby constraining sample collection efficacy, particularly in low-concentration settings [35].

MOUDI, recognized for its high-resolution collection of ultrafine particles, reduces particle bounce and excels in detailed fine-particle analysis, but with increased operational expenses and complexity [33]. Dekati Impactors encompass a wide particle size spectrum and offer simplified maintenance, rendering them advantageous for general particle investigations; nevertheless, the main disadvantages of Dekati Impactors are particle bouncing coupled with multiple stage deposition [34]. Low-volume samplers, such as the Harvard Impactor, are economical and portable for regulatory compliance monitoring of specified particulate matter sizes, whereas high-volume samplers combined with cyclones or virtual impactors provide extensive sample collecting, suitable for chemical analysis [37]. These high-volume samplers, however, require reliable power supplies and may volatilize semi-volatile compounds due to heating, rendering them less appropriate for some organic studies.

The Tisch Environmental Andersen Cascade Impactors have distinct benefits compared to the MOUDI for the collection of particle PAHs across various aerosol sizes, especially in urban environments. The Andersen Impactors facilitate efficient and size-specific collection of PAHs across various aerosol fractions due to their extensive range of cut-off points and elevated flow rates. This is especially advantageous in variable urban settings, where PAHs may be dispersed across fine, ultrafine, and coarse particles. Furthermore, the compatibility of Andersen Impactors with quartz and Teflon filters renders them ideal for PAH analysis, facilitating extraction and minimizing volatilization loss, which is crucial for accurate PAH measurements.

Moreover, Andersen Impactors are recognized for their robustness, economic efficiency, and straightforward maintenance in field circumstances, rendering them ideal for prolonged utilization in contaminated areas. Conversely, MOUDI machines, although exceptionally accurate for ultrafine particles, necessitate increased maintenance and meticulous handling. The Andersen design reduces temperature increase during sampling, therefore maintaining PAH integrity and averting volatilization, a concern associated with the finer stages of MOUDI. The Tisch Andersen Impactors are used for detailed, size-resolved PAH investigations in intricate urban and industrial environments.

2.2 Chemical analysis of PAHs

Soxhlet extraction, Accelerated Solvent Extraction (ASE), and Ultrasonication are prevalent techniques for the extraction of PAHs from aerosol particles, each presenting unique benefits. Soxhlet extraction, a conventional technique, is efficient for intricate matrices and yields high recovery rates; but, it is labor-intensive and necessitates substantial quantities of organic solvents, which can be expensive and raise environmental issues [38]. It functions at the boiling point of the solvent, hence constraining temperature regulation and potentially impacting the recovery of more volatile PAHs.

ASE is a more rapid and efficient alternative that utilizes less solvent than Soxhlet, rendering it suitable for high-throughput investigations. Its capacity to function at elevated temperatures and pressures improves extraction efficiency and permits condition regulation, while it poses a danger of thermal deterioration for some chemicals. ASE equipment is costly and necessitates maintenance, thereby restricting its accessibility. Ultrasonication is a readily available and economical method that employs moderate solvent usage and mild conditions to reduce thermal breakdown. Nonetheless, it may exhibit reduced efficiency for intricate samples, and outcomes can fluctuate based on handling, potentially affecting repeatability. The selection of the extraction process is contingent upon sample complexity, required throughput, and resource availability. Although Soxhlet extraction entails increased solvent consumption, which heightens environmental and financial concerns, it remains advantageous for comprehensive analyses owing to its simple configuration and capacity to operate without sophisticated equipment. Although ASE can diminish solvent usage, it necessitates considerable initial capital and ongoing maintenance due to its high-pressure operation, which may negate environmental and financial advantages in the long run. Ultrasonication, while cost-effective and minimizing solvent usage, frequently yields uneven results for complicated matrices and may lack the requisite consistency for high-quality data. Soxhlet extraction, despite being resource-intensive, is particularly advantageous for this study that prioritize data quality and seek to reduce extraction variability in complicated aerosol samples, while balancing cost, environmental impact, and data reliability. After sample collection, the quartz fibre filters were removed and subject to Soxhlet Extraction using dichloromethane (HPLC grade, Fisher Scientific). Before extraction, a known quantity of deuterated internal standard was added to the filters. The organic extract was nearly dried after vacuum concentration and evaporation under a stream of nitrogen. Before the fractionation procedure, the solution was changed to hexane and evaporated once more with nitrogen, resulting in a final volume of 2 mL. Flash chromatography was used to fractionate the hexane extract using silica gel and several solvents of varying polarities. Pongpiachan et al. (2013) and Tipmanee et al. (2012) provided full descriptions of the PAHs fractionation approach [39,40].

The fractionation and clean-up process adhered to the methodology outlined by Gogou et al. (1996) [41]. Following extraction, the dichloromethane (DCM) solvent was concentrated to dryness through a combination of rotary evaporation and gentle nitrogen stream blowing. The concentrated extract was then diluted in 10 mL of n-hexane and applied to the top of a disposable silica gel column for fractionation. The extract was fractionated into distinct compound classes using flash chromatography on silica gel. Specifically, the concentrated sample was introduced to the top of a 30 × 0.7 cm silica gel column containing 1.5 g of silica gel, which had been activated at 150°C for 3 h. Nitrogen pressure was applied to achieve a flow rate of 1.4 mL min�¹ at the column’s base. The fractionation process utilized the following eluents: (i) 15 mL of n-hexane to elute light molecular weight PAHs (Fraction 1); and (ii) 15 mL of a toluene-n-hexane mixture (5.6:9.4) to elute middle and heavy molecular weight PAHs (Fraction 2). Due to the toxicity of some solvents and the solubility properties of PAHs, toluene was selected as the preferred solvent. After fractionation, the eluates were further concentrated using a rotary evaporator followed by evaporation under a gentle nitrogen stream (flow rate set at 1.0 mbar). To address the low volatility of toluene, acetone (5–25%) was added to enhance dissipation. The concentrated sample was subsequently reduced to near-dryness and reconstituted to a final volume of exactly 100 µL using cyclohexane. This reconstituted sample was then transferred into a GC/MS vial insert for analysis.

Gas chromatography-mass spectrometry was used to identify and quantify the fractions (Shimadzu). The following 12 PAHs were observed: phenanthrene (Phe), anthracene (An), fluoranthene (Fluo), pyrene (Pyr), benz[a]anthracene (B[a]A), chrysene (Chry), benzo[b]fluoranthene (B[b]F), benzo[k]fluoranthene (B[k]F), benzo[a]pyrene (B[a]P), indeno[1,2,3-cd]pyrene (Ind), dibenz[a,h]anthracene (D[a,h]A], and benzo[g,h,i]perylene (B[g,h,i]P). Every solvent was of HPLC quality and was acquired from Fisher Scientific. The following 12 native PAHs were included in a mix standard solution (NS 9815: S-4008–100-T): Phe, An, Fluo, Pyr, B[a]A, Chry, B[b]F, B[k]F, B[a]P, Ind, D[a,h]A, and B[g,h,i]P. Two deuterated internal standards, d₁₂-perylene (d₁₂-Per) and d₁₀-fluorene (d₁₀-Fl), were acquired from Chiron AS (Stiklestadveine 1, N-7041 Trondheim, Norway). Nonane was used to prepare standard stock solutions of native and deuterated PAHs. Working solutions were obtained by diluting the mixture in cyclohexane.

A mass spectrometer is a sophisticated analytical instrument designed to separate ions based on their mass-to-charge ratio (m/z) and determine their relative abundances. The calibration of the instrument is performed using ions with known m/z values. Mass spectrometers universally operate by isolating gas-phase ions in a low-pressure environment, utilizing the influence of magnetic or electric fields on charged particles. In this study, a Shimadzu GCMS-QP2010 Ultra system was employed for analysis. This advanced instrument features a high-speed performance system with an Advanced Scan Speed Protocol (ASSP) function, enabling a maximum scan speed of 20,000 µs�¹ and ultra-fast data acquisition capabilities for comprehensive two-dimensional gas chromatography (GC×GC). The system’s rod bias voltage is automatically optimized during high-speed data acquisition, minimizing the sensitivity loss typically associated with scan speeds exceeding 10,000 µs�¹. With a sensitivity five times greater than conventional systems, the GCMS-QP2010 Ultra is highly effective for applications involving fast GC/MS and comprehensive GC/MS techniques (Patent: US6610979). The separation of target compounds was achieved using a capillary column with a 60 m length × 0.25 mm internal diameter, coated with a stationary phase of 0.25 µm film thickness. The stationary phase consisted of cross-linked/surface-bonded 5% phenyl and 95% methylpolysiloxane, in accordance with specifications from various EPA methods (e.g., 207, 508, 515, 524.2, 525, 8081, 8270). A helium carrier gas (purity: 99.999%) was used at a constant flow rate of 1.0 mL min⁻¹ with a pressure pulse of 25 psi applied for 0.50 minutes. All injections (1 µL) were performed in splitless mode through a universal injector, with standards introduced using a 10 µL Hamilton syringe. The gas chromatographic oven was programmed to maintain an initial temperature of 40°C for 1 minute, followed by heating at a rate of 8°C min⁻¹ to 300°C, which was then held for 45 minutes. These chromatographic conditions enabled the qualitative resolution of several pairs of commonly co-eluting isomers, including phenanthrene/anthracene (Ph/An) and benzo[a]anthracene/chrysene (B[a]A/Chry).

2.3 Quantification and QA/QC

To identify the molecular markers, the quantification ions were compared to the retention times of authentic standards within a range of ± 0.2 min. Chemicals were quantified using an Internal Standard (IS) method (Pongpiachan et al., 2012). To be used effectively, an IS must have comparable physicochemical characteristics or the same kind of substitution as the analytes. A relative response factor was calculated for every native analyte. Because the relative response between the native analyte and IS should always be the same, this was used for quantification. This technique is convenient as the recovery losses of the compound during extraction and analysis are assumed to be equal to those of the IS. All QA/QC analyses were conducted using SRM1941b, which is a reference material used in many scientific and industrial applications for quality assurance and control. They were employed to guarantee the precision and dependability of the analytical data, calibrate instruments, and validate the measurement techniques. The average percentage accuracy of all PAH congeners was 89 ± 24% with the application of SRM1941b.

2.4 Risk assessment analysis

The total toxic equivalent quotient (TEQ) is a widely employed factor in toxicology and risk assessment. This study evaluated the cumulative toxic impacts of various PAHs; however, particular emphasis was on B[a]P as the TEQ for each PAH was calculated by multiplying its concentration in a given sample using the toxic equivalence factor (TEF), which quantifies its relative toxicity compared to B[a]P. This method was used because, given the significant variation in toxicity levels across different PAH congeners, it is crucial to assess the toxicity of PAHs by considering the most potent form, B[a]P. The establishment of TEF values for various PAHs has been undertaken by the World Health Organization (WHO) and other regulatory bodies [42–44]. These values were determined based on the carcinogenicity and other hazardous qualities associated with PAHs. The equation used to compute the cumulative B[a]P-TEQ is

(1)

where C_i and TEF_i are the concentrations of specific PAHs and the corresponding TEFs, respectively.

The TEQ was derived by multiplying the concentration of individual TEF values of each PAH, as outlined in Eq. (1). Three TEQ equations, namely Eq. (2) by Nisbet and Lagoy (1992), Eq. (3) of the US EPA (1993), and Eq. (4) by Cecinato (1997) were calculated [42, 43]. The three TEQ formulas use the acronyms of the PAH congeners to denote their respective concentrations.

(2)

(3)

(4)

The sum of three and four-ring PAHs (Phe, An, Fluo, Pyr, B[a]A, and Chry) is denoted as ΣPAH(3,4) and for five and six-ring PAHs (B[b+k]F, B[e]P, B[a]P, Ind, D[a,h]A, and B[g,h,i]P) is ΣPAH(5,6). The Lifetime Lung Cancer Risk (LLCR) for the detected PAH concentrations in six different particle sizes was calculated using Eq. (5) [45].

(5)

where UR_BaP is the inhalation cancer unit risk (8.7×10⁻⁵ m³ ng⁻¹).

2.5 Multivariate statistical techniques

This study employed Principal Component Analysis (PCA) to identify the origins of aerosols based on the chemical composition of particulate PAH congeners in six-stage configurations collected from the NAQOS. The dataset underwent preprocessing, which involved the elimination of outliers and the normalization of concentrations to guarantee equal contribution of all variables to the study [46].

PCA was subsequently employed on the standardized dataset to diminish its dimensionality and ascertain principle components (PCs) that encapsulate variable combinations elucidating the variation within the dataset [47]. Principal components with eigenvalues over 1 were preserved, and varimax rotation was employed to enhance the interpretability of the source contributions [48]. The loadings of chemical species on each principal component were analyzed to ascertain potential aerosol sources, including automotive emissions, industrial operations, biomass combustions, and transboundary pollutions [5–7,9–11].

The scores of each principal component were computed to assess the contribution of each source to the aerosol mass. The results were corroborated by supplementary methods such as Positive Matrix Factorization (PMF). This method facilitated a thorough comprehension of the principal aerosol sources in the research area, aiding efforts for air quality management and pollution reduction.

• The PMF model, created by Paatero and Tapper in 1993 and 1994, is a sophisticated multivariate statistical instrument utilized for source apportionment of aerosols in ambient air [49,50]. Source apportionment involves identifying and quantifying the contributions of diverse pollution sources, including vehicular traffic, industrial emissions, and biomass combustion, to atmospheric particle matter (PM) levels [51–53]. PMF is particularly advantageous for the analysis of intricate environmental data, as it yields robust and interpretable outcomes, essential for formulating successful air quality control policies.
• To implement the PMF model, two critical types of input data are necessary: concentration data and uncertainty estimations. In this study, the data is structured in a matrix format, where each row denotes a monitoring date and each column signifies a particular chemical species (e.g., PAH congeners). These observations document temporal and spatial fluctuations in aerosol composition, essential for recognizing the distinct characteristics of diverse pollution sources. Furthermore, uncertainty estimates, which indicate the confidence level of each measurement, are integrated into the model to address measurement mistakes [54,55].
• The PMF model then decomposes the input data matrix into two principal matrices: a source profile matrix (f) and a source contribution matrix (g). A speciation data set can be represented as a data matrix x with dimensions i by j, where i denotes the number of samples and j indicates the observed chemical species, accompanied by uncertainty. The objective of receptor models is to resolve the chemical mass balance (CMB) between observed species concentrations and source profiles, as delineated in Equation 6, incorporating the number of factors p, the species profile f of each source, and the mass contribution g from each factor to each individual sample (refer to Eq. (6)):

(6)

The source profile matrix encompasses the chemical compositions or “signatures” of each pollution source, whereas the source contribution matrix delineates the extent to which each source contributes to the observed concentrations at various periods or locations. By reducing residuals and integrating uncertainty, PMF enables researchers to precisely quantify the contributions of various pollution sources without necessitating prior knowledge of source attributes, rendering it a versatile and broadly applicable instrument in air quality research. The “EPA PMF 5.0” was employed in this investigation (https://www.epa.gov/sites/default/files/2015-03/epa_pmf_5.0_setup.exe). Details on the PMF model and the method detection limit (MDL) employed in this research can be found in Supporting Materials. This study tested 3–8 factors to achieve the best solution. 100 runs were conducted for each computation to ensure that the estimated output correlates well with the detections. Since the content obtained from this study was greater than the MDL provided, the computation is based on the recommendation of EPA PMF5.0 User Guide as displayed in Eq. (7).

(7)

The bootstrap uncertainty prediction method was employed to confirm the robustness of the statistics [49,50]. The factor mapping rates of the selected schemes all exceed 90%, suggesting that the analytical results are comparatively stable. More details on the quality control of PMF results were clearly written in Table S2–S3 (see Supporting Material).

3.1 Meteorological conditions and atmospheric concentrations of PAHs in six different particle fractions

The meteorological data gathered at the weather station adjacent to NAQOS from January to April 2023 provides insight into the potential impact of weather conditions on the distribution and amounts of PM (see Table 1). The wind data indicates a predominant southeast direction with moderate velocities, suggesting generally tranquil conditions. Reduced wind speeds (WS) may result in the buildup of pollutants in the vicinity due to diminished dispersion. Conversely, increased wind variability, as evidenced by the standard deviation in direction (i.e., ± 68°), can intermittently facilitate the dispersion of pollutants from the monitoring location.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Statistical description of meteorological parameters obtained from this study.

https://doi.org/10.1371/journal.pone.0320405.t001

The air temperature (AT) and relative humidity (RH) were moderate throughout this period with the average values of 28 ± 2 °C and 67 ± 15%, respectively. Elevated RH levels can result in particle development, when water vapor condenses on particulate matter, augmenting particle size. This directly affects the various PM fractions observed, as RH may determine which particles remain suspended in the air and which settle. AT indirectly influences pollution levels by affecting the activity of pollution sources, such as automobile emissions, which can fluctuate with weather conditions.

The heat index, which accounts for both AT and RH, indicates comparatively moderate perceived temperatures. Consistent atmospheric pressure readings were recorded, signifying stable weather conditions devoid of substantial fluctuations that can affect air mixing layers. This stability indicates restricted vertical dispersion of contaminants, potentially causing them to remain near the surface, particularly on days with minimal wind.

Insignificant precipitation and elevated visibility indicate primarily arid and clear weather. In the absence of precipitation, there is restricted “washout” or elimination of airborne particles, resulting in extended particle residence durations. The generally strong visibility suggests that significant particulate accumulation was not constantly problematic, although intermittent reductions in visibility may correspond with days of heightened local pollution. Collectively, these meteorological variables establish a framework for examining pollutant dynamics at NAQOS and highlight the direct influence of AT, RH, and WS on particulate matter concentrations and dispersion in the area.

The arithmetic mean concentrations of ΣPAH₁₂ collected at PM_0.65–1.1, PM_1.1–2.1, PM_2.1–3.3, PM_3.3–4.7, PM_4.7–7.0, and PM_{7.0 and above} were 13.6 ± 7.44 pg m⁻³, 19.2 ± 8.37 pg m⁻³, 24.3 ± 15.0 pg m⁻³, 17.8 ± 7.65 pg m⁻³, 13.0 ± 6.90 pg m⁻³, 5.71 ± 2.41 pg m⁻³, and respectively (see Table 2). This study examined fluctuations in the arithmetic mean concentrations of ΣPAH₁₂ gathered at six different PM size fractions. Further examination of the spatial distribution of the concentrations of particulate PAHs was conducted at the given locations, revealing substantial disparities in the degree of pollution, as shown in Table 3. These regions can be classified into several concentration groups based on a quantitative analysis of their spatial distribution:

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Statistical description of PAH contents (pg m⁻³) in PM_{7.0 and above}, PM_4.7–7.0, PM_3.3–4.7, PM_2.1–3.3, PM_1.1–2.1, and PM_0.65–1.1.

https://doi.org/10.1371/journal.pone.0320405.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Comparison of Total PAHs contents (ng m⁻³) with previous studies.

https://doi.org/10.1371/journal.pone.0320405.t003

3.2 Percentage contribution of PAHs in six different particle fractions

The correlation between PAHs and the coarse fraction of airborne particles can arise from two mechanisms: particle coagulation or the volatilisation of fine particles, followed by condensation on coarse particles [71]. The morphology, chemical composition, electric charge, and hygroscopicity of particles, along with their relative humidity, are crucial elements that govern atmospheric coagulation processes [72,73]. If particle coagulation was the primary cause of PAHs’ interaction with coarse particles, the concentration of PAHs (ng m⁻³) in fine and coarse particles would be distributed equally. However, Fig 2 shows that this is not the case. As illustrated in Fig 2, the percentage contributions of Low Molecular Weight (LMW) PAHs, Middle Molecular Weight (MMW) PAHs, and High Molecular Weight (HMW) PAHs across six particle size fractions, ranging from PM_0.65–1.1 to PM_{7.0 and above}. These results highlight the varying distribution of PAHs based on their molecular weight across different particle size ranges, providing insights into their likely sources and atmospheric behavior. The percentage contribution of HMW PAHs (5–6 ring PAHs) is dominant across all particle size fractions, consistently contributing over 50% of the total PAH composition. This trend is particularly prominent in finer particles (PM_0.65–1.1 and PM_1.1–2.1), where HMW PAHs exceed 60%. The dominance of HMW PAHs in smaller particle sizes can be attributed to their strong association with combustion sources such as diesel exhaust and other high-temperature processes. Their preferential partitioning into finer particles also suggests a potential for long-range transport and significant health impacts, as smaller particles can penetrate deeper into the respiratory system.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Percentage contributions of

∑PAHs(3,4) and ∑PAHs(5,6) in PM_{7.0 and above}, PM_4.7–7.0, PM_3.3–4.7, PM_2.1–3.3, PM_1.1–2.1, and PM_0.65–1.1.

https://doi.org/10.1371/journal.pone.0320405.g002

MMW PAHs (4-ring PAHs), represented by species such as pyrene and chrysene, contribute significantly across all particle sizes, accounting for approximately 30–40% of the total PAHs. Their higher percentages in coarser fractions (e.g., PM_{7.0 and above}, PM_4.7–7.0) suggest contributions from biomass burning and mechanical re-suspension processes, which often generate coarser particles. The consistent presence of MMW PAHs across all fractions also reflects their role as intermediates between light and heavy PAHs, likely arising from mixed sources. LMW PAHs (3-ring PAHs) exhibit the lowest contributions, consistently below 10% across all particle size fractions. Their relatively higher proportion in larger particles (e.g., PM_{7.0 and above}) suggests an association with volatilized emissions that have re-condensed onto coarser particles or lower-temperature processes. The lower percentages of LMW PAHs in finer fractions highlight their volatility and reduced stability in atmospheric conditions dominated by combustion. The distribution of PAHs across particle sizes indicates that HMW PAHs dominate in finer particles, posing a significant risk for human health due to their persistence and toxicity. MMW PAHs are evenly distributed across particle sizes, reflecting their association with multiple sources, while LMW PAHs contribute minimally and are more associated with larger particle sizes. This distribution highlights the need for targeted mitigation strategies based on particle size and associated PAH sources to reduce environmental and health impacts.

Six particle size fractions (PM_{7.0 and above}, PM_4.7–7.0, PM_3.3–4.7, PM_2.1–3.3, PM_1.1–2.1, and PM_0.65–1.1) show different trends in particulate PAH concentration (ng g⁻¹) that guide both source characteristics and possible health consequences (see Fig 3). PAH congeners like An and B[b]F show the largest contributions in coarser particles (PM_{7.0 and higher}). This pattern implies that, where these PAHs condense or adsorb, bigger particles most likely come from sources like soil dust or coarse combustion particles. Conversely, less abundant in these fractions are toxic congeners include B[a]P, Ind, and B[g,h,i]P, which indicate their inclination to combine more with ultrafine particles [74]. This finding also reflects the pattern that high molecular weight PAHs (HMW-PAHs) are greatly connected with finer particles [75].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Percentage contributions of particulate PAH congeners (ng g⁻¹) in PM_{7.0 and above}, PM_4.7–7.0, PM_3.3–4.7, PM_2.1–3.3, PM_1.1–2.1, and PM_0.65–1.1.

https://doi.org/10.1371/journal.pone.0320405.g003

B [b]F keeps showing dominance when particle size drops to PM_4.7–7.0 and PM_3.3–4.7; B[k]F and Pyr also start to show prominence. Given that these mid-sized particles can reach deeper lung areas than coarser particles, which would cause possible health hazards, B[a]P is particularly noteworthy in them. The steady contribution of B[b]F over these fractions highlights its connection with mid-sized combustion-related particles, probably from vehicle or industrial emissions [76,77].

Whereas heavier, semi-volatile PAHs like Ind and B[g,h,i]P rise, the percentage contributions of B[k]F and An remain high in finer fractions (PM_2.1–3.3 and PM_1.1–2.1). For health effects, this size range is important as, especially those chemicals with carcinogenic qualities, particles tiny enough to reach the bronchioles are particularly hazardous due of high PAH content. The prominent presence of An in these fine fractions and Chry points to partial volatilization and re-condensation onto finer particles occurring in these compounds.

B[b]F, B[k]F, Ind and B[g,h,i]P have the largest concentrations in the finest fraction (PM_0.65–1.1). These hazardous substances have more health consequences since they can reach deeply into lung tissues. Consistent presence of B[b]F throughout all fractions, but particularly in finer particles, suggests major contributions from combustion sources, including automobile emissions. The way PAHs are distributed emphasizes overall the possible health risks connected to fine and ultrafine particles, which concentrate more hazardous and carcinogenic PAHs that the human body can inhale and absorb.

3.3 Diagnostic binary ratios of PAHs

The use of diagnostic binary ratios of PAHs serves as a valuable methodology for identifying potential sources of PAH contamination in the environment [9–11,21]. These ratios entail the comparison of concentrations of specific PAH components in relation to each other. Various sources of PAHs, including combustion processes, industrial operations, and natural sources, frequently provide the characteristic ratios of PAH constituents [78]. The influence of various factors, including meteorological conditions such as temperature, relative humidity, and solar radiation, combined with both heterogeneous and homogeneous reactions with trace gaseous species, can significantly affect PAH levels. Therefore, careful discretion is essential when applying diagnostic binary ratios. Notably, PAH diagnostic binary ratios exhibit significant diversity within a single source and similarity across different sources [79]. The identification of problems is particularly significant when categorising coal combustion types as it becomes challenging to differentiate between them when utilising the B[a]A/(B[a] A + Chry) ratio [80]. Nevertheless, it might be argued that these limitations are of minimal significance, as the primary source of PAHs in the ambient air of Bangkok is contamination resulting from vehicle release [9]. Furthermore, it is imperative to emphasize that coal combustion is not the main source of PAHs in Bangkok. For simplicity, it was postulated that the influence of meteorological conditions on the modification of PAH diagnostic binary ratios is minimal because of the presence of generally steady weather conditions during sample preservation.

As illustrated in Table 4, the diagnostic binary ratios of An/(An+Phe), Fl/(Fl+Pyr), and Ind/(Ind+B[g,h,i]P) detected in PM_{7.0 and above} obtained from this study, were in good agreement with those of PM₁₀ collected inside the tunnel [81], indicating the importance of traffic emissions as one of the main contributors of PAHs in the coarse particle mode. It is also interesting to note that the diagnostic binary ratios of An/(An+Phe), Fl/(Fl+Pyr), B[a]A/(B[a]A+Chry), Ind/(Ind+B[g,h,i]P), and B[a]P/(B[a]P+B[g,h,i]P) detected in PM_2.1–3.3 collected at NAQOS were 0.77, 0.44, 0.54, 0.29, and 0.28, respectively. These values differed significantly from those of PM_2.5 released from woody fuel combustion, suggesting that the impact of biomass burning as a potential source is of minor importance for ultrafine particles. Furthermore, the diagnostic binary ratios of An/(An+Phe) and Fl/(Fl+Pyr) in PM_2.5 collected from road tunnels were almost two times lower than those obtained in this study. These results indicate that a complex mixture of PM_2.1–3.3 bounded PAHs is released from various types of potential sources. As the NAQOS is adjacent to a canal boat terminal, it is reasonable to mention that the impact of canal boat emissions is inevitable.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Diagnostic binary ratios of PAH contents (pg m⁻³) in six different particle sizes obtained from this study in comparison with previous studies.

https://doi.org/10.1371/journal.pone.0320405.t004

3.4 Pearson correlation coefficients (PCCs)

Pearson Correlation Coefficients (PCCs) are a statistical technique that can be employed to detect possible sources of PAHs in various PM size fractions. It is worth mentioning that all size fractions, including PM_0.65–1.1, PM_1.1–2.1, PM_2.1–3.3, PM_3.3–4.7, PM_4.7–7.0, and PM_{7.0 and above} were applied for PCC analysis. Correlation analysis is a valuable tool for elucidating the association between the PAH concentrations in different fractions. This analytical approach provides valuable insights into the potential sources and contributions of PAHs. As illustrated in Table 5, exceedingly strong positive PCCs were detected for Fluo vs. Pyr (0.982) and B[a]A vs. Chry (0.919). This can be attributed to several factors. A plausible explanation for the robust positive correlations observed between these pairs of PAHs is the presence of shared emission sources. Fluo and Pyr, along with B[a]A and Chry, can be emitted concurrently from similar activities or combustion processes such as traffic emissions. This interpretation is in good agreement with a previous study that reported the highest concentrations of Fluo, Pyr, and Chry in PM₁₀ at the entrance and exit sites of the tunnel [82].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Pearson correlation analysis of PAH concentrations (pg m⁻³) in six different particle sizes collected at NAQOS.

https://doi.org/10.1371/journal.pone.0320405.t005

Moderately strong positive PCCs were also observed for Ind vs. B[g,h,i]P, B[b]F vs. B[g,h,i]P, B[b]F vs. Pyr, and B[b]F vs. Fluo, with values of 0.801, 0.896, 0.892, and 0.856, respectively (Table 5). These findings are in good agreement with an earlier study indicating strong positive correlations of Ind, B[g,h,i]P, and B[b]F with other high-molecular-weight PAHs in PM₁₀ and PM_2.5 samples collected at a road tunnel inlet and outlet in Nanjing, China [83]. This underlines the importance of vehicular release as one of the main contributors to particulate PAHs at the study site. PAHs with comparable molecular weights frequently exhibit analogous chemical characteristics, resulting in comparable environmental behaviours. Similarities can be observed in terms of solubility, vapour pressure, and propensity to partition into particles. PAHs with comparable molecular weights can demonstrate analogous chemical stabilities in an environmental context. This implies that these PAH congeners (i.e., Fluo, B[b]F, Pyr, B[g,h,i]P, and Ind) can remain in the environment for an extended period and engage in comparable chemical reactions, thereby strengthening their association. It is also imperative to consider the probable presence of sampling and measurement artefacts. If the sampling and analytical procedures employed for the collection and measurement of these PAHs exhibit a bias that consistently leads to their simultaneous detection, they could be responsible for significant positive associations.

3.5 Hierarchical cluster analysis (HCA) and principal component analysis (PCA)

Hierarchical cluster analysis (HCA) is a robust statistical methodology that can be effectively utilized to discern the potential origins of PAHs within ultrafine particles. The utilisation of HCA proved to be highly advantageous for the analysis of intricate datasets that encompass numerous factors, such as various types of PAH congeners and samples with multiple particle sizes originating from diverse sources. Previous studies have used HCA for the source identification of aerosols, terrestrial soils, and agricultural products [7,9,10,78]. In this study, an agglomerative clustering method coupled with Euclidean distance as a dissimilarity measure, in conjunction with an average linkage between groups, was applied to visualise the dendrogram, as displayed in Fig 4. The dendrogram structure depicted two main clusters of hierarchical arrangements of various PAH congeners according to their concentration patterns. Cluster I consists of B[b]F, while cluster II contains Phe, An, Fluo, Pyr, B[a]A, Chry, B[a]P, B[k]F, Ind, D[a,h]A, and B[g,h,i]P. This suggests that the PAH congeners detected in cluster II have some similarities in their concentration patterns and are not as independent as B[b]F. Prior research has established a correlation between the emissions of B[a]A, Chry, B[k]F, and B[g,h,i]P and the combustion of fuels such as diesel and petroleum. Pyr, Phe, and Fluo are tracers of the combustion of natural gas and coal [80,84,85]. Subcluster I of cluster II contained two PAH compounds: fluoranthene (Fluo) and pyrene (Pyr). The concentration profiles of these two compounds are extremely similar, which may indicate a shared origin of traffic release [86]. Subcluster II of cluster II contains Phe, An, and D[a,h]A. These compounds were categorised into this subcluster based on the similarity in their concentration profiles. Previous studies have highlighted the importance of diesel emissions as the main contributor to Phe and An [87,88]. The existence of sub-clusters within the main cluster II implies that these PAH compounds exhibit differing levels of similarity, potentially indicating the presence of common origins of traffic contributions. Visualising and interpreting the interrelationships among PAHs in a dataset can be facilitated using this practical tool.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Hierarchical cluster analysis (HCA) of PAHs in PM_{7.0 and above}, PM_4.7–7.0, PM_3.3–4.7, PM_2.1–3.3, PM_1.1–2.1, and PM_0.65–1.1.

https://doi.org/10.1371/journal.pone.0320405.g004

PCA is an advanced statistical technique used to transform original variables into a new collection of uncorrelated variables, known as principal components (PCs). PCA was employed for the source apportionment of PAHs in ultrafine particles to ascertain and measure the contributions of various emission sources to the observed PAH concentrations. To avoid the dominance of variables with larger concentration ranges, all PAH data were normalised on the same scale. Examining the eigenvalues is crucial for assessing the significance of each PC in elucidating the variability in the dataset. PCs with higher eigenvalues can capture a larger proportion of the variation in the data [7,9,10,78].

As shown in Table 6, the varimax-rotated components of the six distinct particulate fractions were represented by principal component patterns. These patterns consisted of 12 parameters that were further divided into five PCs. PC1, PC2, PC3, PC4, and PC5 explained 57.0%, 11.9%, 9.35%, 7.39%, and 4.80% of the cumulative variance, respectively. PC1 explained significantly more variance than the other PCs. PC1 primarily consisted of Fluo, Pyr, B[b]F, and B[g,h,i]P, with relatively high correlation values of 0.90, 0.89, 0.85, and 0.77, respectively. Based on the presence of these four congeners, which are commonly associated with gasoline burning, it can be inferred that the majority (57%) of particulate PAHs originated from light-duty cars [89]. It is widely recognized that Phe, B[a]A, Chry, and B[a]P are often used biomarkers for gasohol exhausts [90]. Therefore, the elevated levels of these four PAH congeners in PC2 can be attributed to traffic emissions associated with gasohol, accounting for approximately 11.9% of gasohol emissions. Notably, the relatively high positive loadings of B[k]F and Ind observed in PC3 indicate that heavy oil combustion and electric arc furnaces in the industrial sector accounted for 10% of the emissions [91]. A study conducted by Froehner et al. (2012) found that PC4 exhibits a significant positive loading of D[a,h]A [92]. This suggests that the presence of asphalt and coal tar pavements in the construction of highways may have contributed approximately 7.4% of the overall source contribution. Given that Ph and An have been identified as tracers of diesel combustion, it is reasonable to infer that the fairly significant positive correlations observed between these two congeners in PC5 can be attributed to the contribution of canal boat diesel emissions, accounting for 4.8%, as reported by Tavares et al. (2004) [88].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 6. Principal component analysis of PAH concentrations (pg m⁻³) in six different particle sizes collected at NAQOS.

https://doi.org/10.1371/journal.pone.0320405.t006

3.6 Positive matrix factorization (PMF)

PCA, despite its benefits, has constraints in source apportionment applications. Due to its emphasis on variation rather than actual source limitations, it might generate factors that are challenging to evaluate in practical contexts. The absence of non-negativity requirements may result in negative values, which are implausible in an environmental setting. Furthermore, predominant variables can distort PCA outcomes, and associating PCA-derived factors with actual sources frequently necessitates subjective interpretation, hence constraining its accuracy and relevance for comprehensive source analysis. On the contrary, PMF offers significant benefits in managing environmental datasets by integrating measurement uncertainties, especially when data exhibit varying quality or are incomplete [49,50]. This technique also mandates non-negativity, guaranteeing that the source contributions are physically significant, a crucial aspect for realistic apportionment. Furthermore, PMF can generate adaptable source profiles without the necessity of predetermined templates, enabling it to respond to local conditions and emerging sources [93,94].

The PMF analysis of 12 PAH congeners across six particle size fractions at the NAQOS site revealed three distinct emission sources: light vehicle exhaust, biomass combustion, and diesel emissions (see Fig 5). The primary factor (Factor 1), characterized by significant loadings of Fluo, Pyr, B[a]A, Chry, and B[k]F (all exceeding 80%), is ascribed to emissions from light vehicles, corroborated by analogous results in prior research [86,90]. This statistic indicates that PAHs are frequently emitted by gasoline-powered automobiles, possibly playing a substantial role in urban pollution from traffic sources. The second factor (Factor 2) includes D[a,h]A, Chry, B[b]F, B[k]F, Phe, An, Fluo, and Pyr, and is linked to biomass combustion, consistent with established emission profiles from open fires and agricultural waste incineration [21,95,96]. Biomass combustion significantly contributed to diverse particle sizes, especially in coarser fractions, highlighting its extensive influence on both fine and coarse particulate pollution. The third factor (Factor 3) highlights diesel emissions, characterized by elevated loadings of Ind, B[g,h,i]P, Phenanthrene (Phe), Anthracene (An), and B[a]P. These PAHs are generally released during diesel combustion and were predominantly identified in finer particles, hence corroborating the link between diesel exhaust and smaller particle fractions [77,88,97].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Comparison of factor profile extracted by the PMF model.

A total of three factors were finally selected as the best solutions, including the light vehicle releases, biomass burnings, and diesel emissions.

https://doi.org/10.1371/journal.pone.0320405.g005

Further examination of the relative percentage contributions of each factor reveals valuable insights into their significance in the total PAH burden. Factor 3, accounting for 45.4% of the total PAH contributions, underscores the dominance of diesel emissions in the pollution profile at NAQOS. This is particularly evident in finer particle fractions, emphasizing the need to prioritize diesel-related sources in urban emission control strategies. Factor 1 contributes 34.2%, highlighting the substantial role of light vehicle exhaust, particularly for species like B[a]A, Chry, and B[k]F. These findings corroborate previous studies that associate urban traffic emissions with PAH contamination. In contrast, Factor 2 contributes a relatively smaller share of 20.4%, suggesting biomass combustion as a significant, yet secondary, source of PAHs. Despite its lower contribution, the influence of biomass combustion spans both fine and coarse particles, underscoring its importance in localized air quality management.

Subsequent analysis reveals that biomass combustion is a primary source of PAHs including An, D[a,h]A, Phe, Fluo, Pyr, Chry, and B[k]F across various particle sizes, underscoring its influence on both fine and coarse particles (see Fig 6). Light vehicle emissions predominantly contribute to B[a]A, Chry, B[k]F, and D[a,h]A, whereas diesel emissions are accountable for B[g,h,i]P, B[a]P, Ind, B[b]F, Phe, Fluo, Pyr, and B[k]F, particularly in finer particulate fractions. These relative percentage contributions emphasize the importance of tailoring mitigation strategies to target the most significant emission sources, particularly diesel exhaust and light vehicles, while not overlooking the notable contribution of biomass combustion to PAH levels. In summary, PMF analysis delineates the distinct contributions of light vehicles, biomass combustion, and diesel emissions to PAH profiles across particle size distributions at NAQOS. The results underscore the critical role of diesel emissions in driving urban PAH pollution, with light vehicles and biomass combustion serving as complementary contributors. These findings have significant ramifications for pollution management tactics and public health policies aimed at mitigating PAH exposure in urban settings. Integrating insights from the relative contributions of each factor enhances the specificity of source apportionment efforts, providing a robust foundation for crafting effective intervention strategies.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Source contributions to 12 PAH congeners in PM_{7.0 and above}, PM_4.7–7.0, PM_3.3–4.7, PM_2.1–3.3, PM_1.1–2.1, and PM_0.65–1.1.

https://doi.org/10.1371/journal.pone.0320405.g006

3.7 Health risk assessment of PAHs in six different particle fractions

The potential negative impacts on human health resulting from exposure to PM containing B[a]P TEQs can differ depending on the size of the particles and the TEQs. The TEQs for the detected PAH contents in the six different particle fractions are listed in Table 7. The TEQs calculated using the three different methods in descending order were PM_2.1–3.3> PM_3.3–4.7> PM_4.7–7.0> PM_1.1–2.1> PM_0.65–1.1> PM_{7.0 and above}. Thus, the middle fractions (i.e., PM_2.1–3.3 and PM_3.3–4.7) showed the greatest toxicity. In contrast, the largest and finest fractions (i.e., PM_{7.0 and above} and PM_0.65–1.1) showed the lowest toxicity with TEQ_US-EPA values of 1.25 pg m⁻³ and 1.27 pg m⁻³, respectively.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 7. B[a]P toxic equivalent quotient (TEQ) for the measured PAH contents (pg m⁻³) in six different particle sizes collected at NAQOS.

https://doi.org/10.1371/journal.pone.0320405.t007

LLCR is an essential instrument for evaluating the detrimental health consequences associated with exposure to PAHs. This analysis offers a numerical assessment of the likelihood of acquiring lung cancer as a result of exposure during an individual’s lifespan. By comparing the calculated LLCRs with established benchmarks and effectively conveying the outcomes, public health workers and policymakers can make well-informed decisions aimed at safeguarding the health of both individuals and communities. In this study, all LLCR values in six different particulate fractions were below the benchmark level of 1×10⁻⁶ (see Table 8). The present study’s LLCR evaluation reveals that all six particle fractions are below the threshold benchmark of 1×10⁻⁶, signifying a negligible cancer risk from PAH exposure for residents in proximity to the sampling site in Bangkok. This research offers essential insights, indicating a reduced cancer risk associated with PAHs relative to other studies performed in more industrialized or extensively polluted areas. This risk assessment underscores the efficacy of LLCR as a significant instrument for public health evaluation, providing actionable data to guide policies focused on preserving air quality and reducing public health concerns.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 8. Lifetime lung cancer risk (LLCR) for the detected PAH concentrations in six different particle sizes collected at NAQOS.

https://doi.org/10.1371/journal.pone.0320405.t008

A prior study evaluated PAH exposure in a Mediterranean industrial region and found LLCR values between 3.2×10⁻⁵ and 4.3×10⁻⁵, significantly surpassing the benchmark level [98]. The predominant presence of PAHs was seen in the gaseous phase, with petroleum burning and vehicular emissions identified as the principal sources. This markedly elevated risk highlights the influence of industrial pollutants, particularly from volatile PAHs, on cancer susceptibility. In contrast to NAQOS’s relatively low LLCR values, the heightened levels in the Mediterranean context demonstrate how proximity to industrial activity can significantly boost cancer risk linked to PAH exposure. Another study similarly investigated PAH exposure across several urban and rural locales in the Czech Republic, revealing heightened LLCR values, especially in the industrial city of Ostrava, where risks were estimated at 1545 instances per million [99]. In rural areas, the risk of cancer persisted above the established standard. The Czech study emphasizes the impact of seasonal pollution and the role of industrial emissions in increasing PAH exposure risks, which sharply contrasts with the low cancer probability suggested by this research. This comparative investigation highlights how seasonal, localized pollution sources can exacerbate cancer risks in colder places, particularly during periods of elevated pollution.

Two prior studies offer additional insight concerning the disparities in industrial and urban exposure [100,101]. A prior study evaluated occupational PAH risks in French industries, indicating LLCR values surpassing 1×10⁻⁴ in high-risk sectors such as coke production [101]. This diversity among industries further emphasizes the elevated cancer risks linked to particular occupational exposures. In contrast, a separate study conducted in Los Angeles, Milan, and Thessaloniki revealed that LLCR values in Mediterranean cities exceeded 1×10⁻⁶, although Los Angeles demonstrated values below this threshold owing to stringent air quality standards [100]. These findings correspond with NAQOS’s low LLCR values, highlighting the efficacy of air quality control in mitigating PAH exposure hazards in urban environments with diminished industrial contributions. These comparisons underscore the significance of emission control strategies and air quality standards in mitigating health hazards associated with PAHs.

3.8 Practical recommendation for policy makers

LLCR values are compared against a standard threshold of 1×10⁻⁶, which represents a one-in-a-million cancer risk. Values below this benchmark are generally considered negligible in terms of health risks, while values above it indicate a higher probability of adverse health outcomes. In this study, all LLCR values for the six particle size fractions were below the 1×10⁻⁶ threshold, suggesting minimal cancer risk for residents in the studied area. The findings indicate that PAH exposure at the site poses relatively low health risks compared to more industrialized or polluted urban areas. For example, studies in heavily polluted regions, such as Mediterranean industrial zones or certain Chinese cities, have reported LLCR values significantly exceeding the threshold, highlighting the influence of local emission sources like heavy industry and coal combustion. The comparatively low LLCR values detected in this study suggest that while PAHs are present, effective regulatory measures or reduced industrial activity in Bangkok may contribute to lower risk levels. Nonetheless, finer particle fractions, which carry higher concentrations of carcinogenic PAHs, remain a concern due to their ability to penetrate deep into the respiratory system, emphasizing the importance of ongoing monitoring and mitigation strategies.

This study’s findings reveal essential insights into the distribution of particle-bound PAHs and their origins, emphasizing the necessity for focused public health policies and initiatives to reduce exposure. The prevalence of diesel emissions in smaller particle fractions (45.4% of total PAHs) underscores the substantial contribution of automobile emissions to urban air pollution. These smaller particles not only contain elevated levels of carcinogenic PAHs but also possess the ability to infiltrate deeply into the respiratory system, hence increasing health risks. Policymakers should prioritize the implementation of more stringent emission rules for diesel vehicles, especially in densely populated metropolitan areas. The establishment of low-emission zones (LEZs), as effectively exemplified in cities such as London, might markedly diminish automobile contributions to PAH pollution. Promoting the adoption of electric and hybrid vehicles while systematically eliminating outdated diesel engines will significantly improve air quality.

The contributions of light vehicle exhaust (34.2%) and biomass combustion (20.4%) highlight the necessity for a comprehensive strategy for emission regulation. Public health policies ought to advocate for the adoption of cleaner fuels, such as gasohol blends with diminished PAH emission profiles, while concurrently endorsing the utilization of catalytic converters to mitigate automotive PAH emissions. Interventions for biomass combustion should target the reduction of open burning practices, especially in peri-urban regions, while promoting cleaner, energy-efficient options for cooking and heating. Public awareness campaigns may be crucial in informing communities about the health hazards associated with biomass burning and promoting safer practices.

Moreover, canal boats, although accounting for about 5% of the total PAHs, persist as a limited yet substantial source of pollution in proximity to waterways and residential zones. Regulations requiring the adoption of cleaner fuels or electric propulsion systems for canal boats might significantly diminish their emissions. Regular maintenance plans to guarantee optimal performance of boat engines and the implementation of particulate filters may also reduce their environmental impact. In addition to targeted interventions, comprehensive urban design measures, such as enhancing green infrastructure, may serve as a natural barrier to capture airborne contaminants and diminish human exposure. Investments in real-time air quality monitoring and predictive modeling could facilitate data-driven policy decisions, permitting targeted interventions in high-exposure regions. These approaches would not only reduce PAH exposure but also enhance urban air quality and public health effects.

Highlights

• Higher PAH toxicities are detected in ultrafine particles.
• Traffic emissions are the main contributors of particulate PAHs.
• Canal boats account for only 5% of total PM emissions.