Seasonal variability and source diagnostics of ambient PAHs in Agra, India, using the CBPF and their health risk evaluation

Abstract

This study addresses the critical issue of polycyclic aromatic hydrocarbons (PAHs) bound to total suspended particulate (TSP) in urban-industrial environments, focusing on an understudied residential area in Agra, India—a city within the heavily polluted Indo-Gangetic Plain (IGP). The research aimed to investigate seasonal variations in PAH concentrations, identify their emission sources, and assess the associated health risks. TSP samples were collected during cold weather months (CWM; January and February 2023) and hot weather months (HWM; March–May 2023) and analysed for 16 priority PAHs. The results showed notably higher TSP (349.4 ± 56.2 µg m⁻³) and PAH (1857.3 ng m⁻³) levels in CWM compared to HWM (266.5 ± 33.1 µg m⁻³ and 721.7 ng m⁻³), with high-molecular-weight PAHs dominating in CWM and 3-ring PAHs prevailing in HWM. Source apportionment using diagnostic ratios and Positive Matrix Factorization (PMF) indicated vehicular emissions, fossil fuel combustion, and industrial activities as primary contributors. Conditional Bivariate Probability Function (CBPF) analysis revealed seasonal shifts in dominant source regions—southwest in HWM and northeast in CWM—correlating with local wind patterns. Health risk assessments based on benzo(a)pyrene toxicity equivalent (BaPeq-TEQ), benzo(a)pyrene mutagenic equivalent (BaPeq-MEQ), and Incremental Lifetime Cancer Risk (ILCR) highlighted carcinogenic and mutagenic risks from BaP, BbF, and DbA via dermal and ingestion pathways. These findings underscore the need for season-specific air pollution mitigation strategies, cleaner fuels, and stricter emissions controls. The study contributes to environmental chemistry by enhancing the understanding of TSP-bound PAH behaviour, exposure pathways, and health risks in urban residential zones, thereby supporting evidence-based policymaking aligned with Sustainable Development Goals 3 and 11.

---

Environmental significance

Despite policies to curb air pollution, lax enforcement in residential areas sustains emissions, elevating levels of toxic organic contaminants like polycyclic aromatic hydrocarbons (PAHs). Prolonged exposure to total suspended particulate-bound PAHs, including fine and coarse particles, endangers vulnerable residents. This study reveals higher PAH levels in Agra during winter, dominated by high-molecular-weight compounds like benzo(b)fluoranthene and benzo(ghi)perylene. Using CPF, CBPF, and PMF analyses, key sources—vehicular traffic, fossil fuel combustion, and local industries—were identified, with notable seasonal source shifts. Health risks specific to different age groups were also identified, providing a scientific foundation for age-targeted mitigation strategies. These findings urge stricter, season-specific air quality measures and provide a model for other urban areas to enhance policy and safeguard public health.

1 Introduction

Rapid urbanisation and socioeconomic growth within densely populated agglomerations have profoundly reshaped urban environments, creating intricate interactions between development, environmental quality, and public health. These dynamics have made it increasingly difficult to predict and mitigate the impacts of urbanisation on air quality, particularly as ambient particulate matter (PM) continues to pose a serious public health threat.^1–4 Epidemiological evidence consistently links PM exposure to a range of adverse health outcomes, including diminished lung function, asthma exacerbation, increased susceptibility to respiratory infections, premature births, and elevated neonatal mortality rates.^5,6

PM is chemically diverse, comprising both inorganic and organic constituents. While non-toxic organics such as alkanes and fatty acids account for 80–85% of the organic fraction, the remaining 15–20% includes hazardous compounds like polycyclic aromatic hydrocarbons (PAHs), which are of particular concern due to their toxic, mutagenic, and carcinogenic properties.^7,8 PAHs are primarily produced through incomplete combustion processes associated with domestic heating, transportation, biomass burning, industrial operations, and waste incineration.^9–12 PAHs are commonly classified by molecular weight into two groups: low molecular weight (LMW) PAHs, comprising 2–3 aromatic rings (molecular weight ranging from 128 to 178 Da), and high molecular weight (HMW) PAHs, containing 4 or more rings (molecular weight ranging from 252 to 278 Da).¹³ This classification is both chemical and functional, as it reflects the differences in volatility, phase distribution, and source attribution. LMW-PAHs are generally more volatile and predominantly found in the gas phase, while HMW-PAHs tend to adsorb onto particulate matter due to their lower volatility. However, LMW-PAHs can also associate with particles under specific meteorological conditions, such as low ambient temperatures, high relative humidity, and stagnant air masses, which complicates phase partitioning and underscores the need to account for seasonal and atmospheric influences when interpreting PAH behaviour. Source-wise, these two groups exhibit distinct emission profiles. LMW-PAHs are typically linked to petrogenic sources, including gasoline combustion and vehicular exhaust, whereas HMW-PAHs are more characteristic of pyrogenic activities such as biomass burning, coal combustion, and industrial processes. Accurately identifying the sources of atmospheric PAHs is critical for developing effective mitigation strategies. Diagnostic ratios (DRs) offer qualitative insights into emission origins; for instance, K. Ravindra et al.¹⁴ linked fossil fuel combustion to PAH emissions in Belgium, while M. Wang et al.¹⁵ attributed major contributions in Dalian, China, to coal combustion and vehicular traffic. For quantitative source apportionment, Positive Matrix Factorization (PMF) has been widely used. Y. He et al.¹⁶ reported that coal and biomass burning (32.2%) and traffic emissions (28.1%) dominated PAH sources in China, while M. Davoudi et al.¹⁷ found diesel (56.3%), petrol (15.5%), and biomass burning (13%) to be primary contributors in Iran. Combining DR and PMF enhances source identification by cross-validating emission estimates.

Additionally, the Conditional Probability Function (CPF) and Conditional Bivariate Probability Function (CBPF) are used to link pollutant concentrations with wind patterns, revealing probable source directions. Long-range atmospheric transport further complicates pollution profiles. For example, T. Hu et al.¹⁸ showed that CWM air masses from northern China significantly influenced PAH levels in Shanghai, and D. Wu et al.¹⁹ demonstrated seasonal variations in Bengbu, China, where HWM emissions were dominated by industrial and fuel combustion, while coal and biomass burning prevailed in CWM.

Given their semi-volatile nature, PAHs partition between gas and particulate phases in response to factors such as temperature, humidity, solar radiation, and PM composition.^9,20–22 This behaviour highlights the need to analyse PAHs within an appropriate PM fraction to capture their true atmospheric burden and associated risks.

TSP represents an especially important fraction for such analysis. Unlike PM₁₀ or PM₂.₅, which focus on specific aerodynamic diameters, TSP encompasses a broad spectrum of particle sizes, making it more inclusive of both fine and coarse particles. This wider particle range enhances the ability to detect and quantify HMW-PAHs, which tend to bind to larger particles and are more environmentally persistent. Additionally, TSP-bound PAHs are often more stable, contributing to long-range atmospheric transport and prolonged exposure risk.²³ Though coarse particles in TSP may not penetrate deeply into the respiratory tract, chronic exposure to TSP—especially its toxic organic and elemental constituents—has been associated with cytotoxic effects and increased risks of airway inflammation.^24,25

Despite the growing recognition of PAHs as public health threats, there remains a significant research gap in India's rapidly expanding urban environments, particularly in cities like Agra situated within the heavily polluted Indo-Gangetic Plain (IGP). Existing studies in Agra have largely centered on urban and industrial zones,^{8,13,22,26,27} with minimal focus on urban residential areas where people experience direct exposure to local sources such as traffic emissions, household heating, and nearby industrial activities. The selected urban residential site in Agra represents a unique microenvironment where both localised domestic emissions and peripheral industrial influences converge. While the area is primarily residential, characterised by household cooking, biomass burning, and waste incineration, it is also geographically surrounded by small-to-medium-scale industrial units and major traffic corridors. This spatial juxtaposition provides a valuable opportunity to assess the interplay between low-intensity, continuous domestic sources and intermittent or high-intensity industrial emissions. Such a setting allows for a more comprehensive understanding of the source complexity and seasonal variability of ambient PAHs in mixed-use urban landscapes. This dual exposure scenario is particularly relevant for health risk assessments, as residents are subjected to a broader spectrum of pollutants than those in purely residential or industrial areas.

This study seeks to fill that gap by investigating PAH concentrations in the TSP fraction at an urban residential site in Agra. TSP was chosen for its ability to capture a comprehensive profile of particle-bound PAHs across a full range of particle sizes, thereby improving the characterisation of both exposure and toxicity. The key objectives are: (i) to examine seasonal variation in PAH concentrations during HWM and CWM, (ii) to identify emission sources using various mathematical receptor model (CPF, CBPF, and PMF) analysis and (iii) to assess associated health risks. Through this integrated approach, the research enhances the understanding of both local and regional PAH pollution dynamics, supporting targeted interventions for air quality improvement and public health protection in the IGP.

2 Materials and methods

2.1 Description of the sampling site

Sulabh Puram residential Colony, which is close to Sikandra Tomb in Agra, India (27° 09′ 24″ N, 78° 03′ 02″ E; 178 MSL), was the study's sampling site, as shown in Fig. 1. The district has a total area of 4041 km² with a population of 5.8 million according to the Agra Population Report 2022. It is surrounded by diverse industrial establishments, including leather, chemical, packaging material, tyre, battery, and shoe factories, contributing to a complex industrial environment (https://www.agraonline.in/city-guide/industries-in-agra). It is located 1.5 km from National Highway 19 and is heavily influenced by vehicular emissions. The city of Agra itself, with an area of 188 km² and nearly 2 million inhabitants, faces significant air pollution challenges from both vehicular and industrial sources. The city's vehicle count increased from 640 000 to 670 000 between 2010–2011 and 2022–2023, out of which 600 000 were run on gasoline (petrol), while around 65 000 were on diesel and 5000 on green fuels like compressed natural gas. Additionally, small-scale industries (127 units), such as petha, cupola, rubber, and chemical producing units, consume substantial amounts of coal and wood (500 Kg of wood and 4.7 tons of coal per day by all units), further degrading air quality.² Notably, the study site is positioned 40 km north of the Mathura oil refinery, located upwind. In CWM, prevailing winds primarily originate from the west or northwest, significantly influencing the atmospheric conditions at the sampling location. Overall, the sampling site represents a complex urban setting influenced by industrial activities, vehicular emissions, high population density, and meteorological factors, necessitating comprehensive assessment and monitoring of air quality parameters.

Fig. 1 Sampling site.

2.2 Sample collection

TSP samples were collected on pre-baked QFFs (at 750 °C for 12 hours) using a high-volume air sampler, Tisch Environmental-1000X, operated at a flow rate of 0.2 m³min⁻¹ for 24 hours, following the methodology described in P. K. Verma et al.^8,22 The filters were weighed before and after each sampling, using a microbalance (Mettler Toledo, ME 204, Switzerland) under controlled relative humidity (40–45%) and temperature (22–24 °C) conditions, for the measurement of mass concentrations. All loaded filters were wrapped in pre-baked aluminium foil and stored in the freezer (−20 °C) until analysis. An operative blank was also collected in parallel. The particulate samples were collected for two periods, i.e., CWM (January and February) and HWM (March–May) in the year 2023.

The meteorological parameters, namely, ambient temperature, RH, WS, and solar radiation (SR), were obtained from the online portal of Continuous Ambient Air Quality Monitoring Stations (CAAQMS) (https://app.cpcbccr.com/ccr/) of Shashtripuram. Hourly data were collected from January 13, 2023 to May 27, 2023. Missing records, incorrect data, and outliers in the observation were excluded. Details of meteorological parameters, including temperature, relative humidity, wind speed and direction, and solar intensity over the study site, are summarised in Supplementary Table S1.

2.3 Extraction

PAHs were extracted with dichloromethane (DCM) and n-hexane (4 : 1 in volume, Merck). The extracts were cleaned using a silica gel column. Target analytes were eluted with 20 mL of DCM and n-hexane mixture (1 : 1 in volume). After that, the eluates were concentrated on a rotary evaporator to about 2 mL. Final extracts were transferred into glass ampoules and stored in a freezer at −20 °C for chromatographic analysis.

2.4 Analysis of organic carbon (OC) and elemental carbon (EC) content

The concentrations of OC and EC in aerosol samples were estimated using a thermogravimetric analyser (TGA 4000, PerkinElmer). A 1.0 cm² section of each loaded filter was carefully extracted and placed in the analyser's sample holder. The analysis was conducted by heating the filter incrementally to 850 °C at a constant rate of 20 °C per minute under a controlled N₂–O₂ atmosphere, with a gas flow rate maintained at 20 mL min⁻¹. Set temperatures during the analysis were 350 °C, 550 °C, 700 °C, and 850 °C, and the process was managed using PYRIS software. This procedure was conducted following the method outlined by D. C. Parashar et al.²⁸ OC content was determined by the weight loss observed up to 350 °C in the presence of oxygen. The EC concentration was subsequently quantified by the weight loss occurring between 350 °C and 550 °C, during which the filter underwent visible decolorisation, indicating the combustion of the carbonaceous material. Beyond 550 °C, no further weight loss was detected up to the final temperature of 850 °C, confirming the complete oxidation of the EC fraction.

2.5 Analysis of PAHs

The analysis of 16 priority PAHs, i.e., naphthalene (NaP), acenaphthylene (Acy), acenapthene (Ace), fluorene (Fluo), phenanthrene (Phen), anthracene (Anth), fluoranthene (Fla), pyrene (Pyr), benz(α)anthracene (BaA), chrysene (Chy), benzo(b)fluoranthene (BbF), benzo(k)fluoranthene (BkF), benzo(a)pyrene (BaP), indeno(1,2,3-cd)pyrene (IP), dibenzoanthracene (DbA), and benzo(ghi)pyrene (BghiP), was done using the method described in our previous publication.²² Briefly, 16-PAHs were analysed according to their m/z ratio using a Gas Chromatograph-Mass Spectrometer (GC-MS, BRUKER SCION SQ) in selected ion monitoring mode (Table S2). The separation was performed in an Rtx-5 MS capillary column (30 m length, 0.25 mm internal diameter, and 0.25 µm df) containing cross-bond dimethylpolysiloxane (Cat no. 12623). High-purity helium (99.999%) was used as the carrier gas at a constant flow rate of 1 mL min⁻¹. One microliter of either the PAH standard mixture (at a fixed concentration) or the extracted air sample was injected into the GC column at 270 °C in splitless mode. The oven temperature was set at 90 °C and held for 2 min. It was heated up to 290 °C at a rate of 6 °C min⁻¹ and finally held for 20 min. The GC/MS interface and ion source temperatures were kept at 280 °C and 250 °C, respectively. The MS analysis was performed in electronic impact mode (EI) in positive mode with an energy of 70 eV. The MS was operated in selected ion monitoring (SIM) mode with mass scanning ranging between 50 and 500.

2.6 Quality assurance/quality control

Identification of 16 PAHs was accomplished by analysing characteristic ions and comparing the retention times of chromatographic peaks with those of certified PAH standards. PAHs were quantified using external calibration with a certified 16-PAH standard mixture RESTEK (Centre County, PA, USA, cat.no-75834-700, USA). Calibration curves were established in the range of 1–200 ppm with seven concentration points and exhibited excellent linearity (R² > 0.99 for all compounds). Although isotopically labelled internal standards were not employed, a non-labelled internal standard was incorporated at 25 ppm near the mid-point of the calibration curve to monitor instrument performance.²⁹ Recovery tests were performed by spiking procedural blank filters with 100 µL of 50 ppm PAH standard solution. Average recoveries ranged from 95 to 105%, with relative standard deviations below 5%, demonstrating good extraction efficiency and reproducibility under blank conditions.

The method detection limits (LODs) for each PAH were determined as three times the standard deviation of replicate analyses of filter blank solutions (n = 7) and are presented in Supplementary Table S3. The LOD values (0.02–1.45 ng m⁻³) were well below the ambient concentrations measured in this study, confirming the adequacy of the method for atmospheric PAH determination. Although internal standards were not applied, the combination of stringent QA/QC protocols, consistent recoveries, and low detection limits ensured reliable quantification. The absence of isotopically labelled standards is acknowledged as a limitation, and their inclusion is recommended for future analyses to further minimise matrix-related uncertainties. The TGA 4000 was calibrated using 15 mg of calcium oxalate with a nitrogen purge gas at a scanning rate of 20 °C minute⁻¹, showing weight loss events that closely matched theoretical values.

Before sampling, quartz fiber filters (QFFs) were pre-heated at 550 °C to eliminate organic contaminants and desiccated for 24 hours to remove moisture. Filter weights were recorded before and after sampling using a microbalance with a sensitivity of ±0.0001 mg, with filters conditioned at 25 ± 3 °C and 22 ± 3% relative humidity for 72 hours. After sampling, filters were reweighed and stored at −20 °C to prevent the loss of semi-volatiles. Sampling flow rates were calibrated and monitored with a flow meter, and replicate analyses were conducted at a rate of one per ten samples. Ten blank filters were analysed to correct sample results based on average blank concentrations, with differences in replicate analyses less than 5% for total carbon (TC) and 10% for OC and EC.

2.7 Trajectory analysis

Backward air mass trajectories were calculated using the NOAA Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT-4) model (https://www.ready.noaa.gov/HYSPLIT.php) with meteorological input from the Global Data Assimilation System (GDAS). Trajectories were computed at 500 m above ground level (AGL) to minimise surface friction effects and represent transport within the lower boundary layer. Simulations were performed for each experimental day of the study period, with 72 hour trajectories generated to capture potential source regions and transport pathways influencing TSP at the receptor site.

2.8 Source apportionment

2.8.1 Diagnostic ratio. Several PAH ratios, such as Fla/(Fla + Pyr), IP/(IP + BghiP), BaA/(BaA + Chr), Flu/(Flu + Pyr), and Anth/(Anth + Phen), were used to trace the sources of PAHs.

2.8.2 Positive matrix factorisation. Positive Matrix Factorization (PMF), a multivariate factor analysis technique introduced by Paatero and Tapper (1993),³⁰ has been widely employed to identify sources of polycyclic aromatic hydrocarbons (PAHs) in diverse environmental matrices, including soil,^31,32 atmosphere,¹⁹ and street dust.^15,33 PMF decomposes the input data matrix into two components: the factor contribution matrix and the factor profile matrix, which are then used for source identification.¹⁹

In this study, EPA PMF 5.0, developed by the U.S. Environmental Protection Agency (EPA) (https://www.epa.gov/air-research/positive-matrix-factorization-pmf-model-environmental-data-analyses), was applied to perform source apportionment of PAHs. The input data consisted of the PAH concentration matrix and the associated uncertainty matrix. Concentrations for each sample (without standardization) were used directly, and uncertainties were generally set as 5% of the corresponding concentrations. For missing data, the quadruple mean concentration method was applied.³² Non-detectable values were replaced with half of the method detection limit (MDL),^34–37 and their uncertainties were assigned as 5/6 of the MDL. These processed matrices were then loaded into PMF 5.0, and the model was executed following the operational guidelines provided by the EPA Office of Research and Development.

A total of 24 PM samples, representing both CWM and HWM, were used in the PMF analysis. Given the limited sample size, uncertainty in the model results was rigorously evaluated using bootstrap (BS) and displacement (DISP) analyses. The bootstrap analysis was performed with 100 resampling runs, and the variability in species contributions for each factor is illustrated in Fig. S1. The base-run factor profiles consistently fall within the interquartile ranges of bootstrap distributions, indicating good agreement between the base solution and the resampled datasets. Furthermore, all resolved factors were successfully mapped to the base solution with mapping percentages exceeding 90%, demonstrating strong reproducibility and stability of the factor profiles despite the limited dataset.

To further assess rotational ambiguity, DISP analysis was applied by systematically perturbing the factor elements while monitoring changes in the objective function (ΔQ). The DISP run completed successfully (error code = 0) with no factor swaps, and no significant decrease in Q was observed within the acceptable uncertainty limits (ΔQ < 1%; Table S4). These results indicate that the PMF solution is stable and minimally affected by rotational freedom. Together, the bootstrap and DISP diagnostics confirm that the identified source factors are statistically robust, physically interpretable, and not unduly influenced by individual samples or data noise, thereby addressing the uncertainty concerns associated with the limited number of observations.

2.9 Conditional bivariate probability function

The CPF is a statistical tool used to estimate the likelihood that a measured pollutant concentration exceeds a specified threshold, based on the wind direction. Originally, the CPF was designed to identify which wind directions are most associated with high pollutant concentrations, offering insight into potential upwind sources of emissions.³⁸

The CPF is calculated using eqn (1):

where mΔθ = number of samples in the wind sector θ, C = pollutant concentration, x = threshold value of high percentile of concentration, e.g., 75th or 95th, and nΔθ = total number of samples from wind sector Δθ.

To enhance the identification of emission sources, particularly in complex environments, the CBPF incorporates both wind direction and wind speed (2). By adding wind speed as a second conditioning variable, the CBPF increases the reliability of source identification compared to the CPF, which considers only the direction.^38,39

where mΔθ,Δu = number of samples in the wind sector Δθ,Δu = wind speed interval; C = pollutant concentration, x = threshold value of high percentile of concentration, e.g., 95th, and nΔθ,Δu = total number of samples in that wind direction-speed interval.³⁹

The CBPF has proven to be more effective in areas with high source complexity, enabling the detection of a wider range of pollutant sources than CPF alone.³⁹ The OpenAir R package was utilised to apply the CBPF methodology due to its robustness and efficiency.

2.10 Health risk assessment

In developing regions like India, highly localised high-emission sources such as cooking, heating,^40,41 urban runoff, and industrial activities can significantly elevate PAH levels in PM and the associated risk.^42,43 To quantify human exposure to PAH, carcinogenic and mutagenic risks were evaluated following the USEPA (1989) guidelines, introducing PAH exposure through ingestion, dermal contact, and inhalation for both children and adults. The health risk assessment of individual PAHs was calculated relative to BaP in terms of Toxicity Equivalent (BaP_eq-TEQ) and Mutagenic Equivalent (BaP_eq-MEQ). The BaP_eq-TEQ (ng m⁻³) and BaP_eq-MEQ (ng m⁻³) of individual PAHs were computed as the product of concentration of each PAH (C) in ng m⁻³ and its toxic equivalent factor (TEF) and mutagenic equivalent factor (MEF), respectively, as shown in eqn (3) and (4).

BaP_eq-TEQ = Ci × TEFi (3)

BaP_eq-MEQ = Ci × MEFi (4)

The BaP-TEQ and BaP-MEQ for the mixture of PAHs in ambient samples were determined from the sum of 16-PAHs' BaP_eq. (eqn (5) and (6)).

BaP-TEQ = Σ BaP_eq-TEQ (5)

BaP-MEQ = Σ BaP_eq-MEQ (6)

2.10.1 Incremental Lifetime Cancer Risk (ILCR). Cancer risk assessments were done using an ILCR model (USEPA, 2004). It considers the average daily dose of PAHs in children and adults due to exposure to particulate-phase PAHs through ingestion, inhalation, and dermal contact. They were calculated using eqn (7)–(9) as mentioned below.where BaPeq is the total PAH toxicity potential estimated using the TEF approach. Cancer slope factor (CSF) is expressed as (mg kg⁻¹day⁻¹)⁻¹, and its values for ingestion, inhalation, and dermal exposure of BaP were 7.3, 3.85, and 25 (mg kg⁻¹day⁻¹)⁻¹, respectively. The remaining exposure parameters employed in the ILCR model for infants, children, and adults via ingestion, inhalation, and dermal pathways were derived from the United States EPA Risk Assessment Guidance and associated literature, as summarised in Table S5.

3 Results and discussion

3.1 Variation of TSP concentrations

The temporal variation of TSP concentrations and PAHs bound to TSP, along with the meteorological parameters—temperature, WS, and RH—during the study period is depicted in Fig. 2. During the study period, 24 h TSP concentrations varied from 214.8 to 435.4 µg m⁻³, with an average concentration of 301.1 ± 60.1 µg m⁻³. The TSP concentrations measured during the study period consistently exceeded the Indian National Ambient Air Quality Standard limits (150 µg m⁻³),⁴⁴ with every individual sampling event recording value above the prescribed threshold. The TSP concentrations were found to be approximately 44–50% lower than in previous years and compared to other urban and industrial sites in Agra.^8,22,45 This decrease in concentration was mainly due to the implementation of government policies to reduce air pollution, such as the introduction of e-rikshaws and electric vehicles, the use of water sprinklers for misting, etc. The mean TSP concentration in Agra was comparable to values reported at other urban sites across the Indo-Gangetic Plain (IGP) and northern India but exceeded those observed in several Indian cities such as Delhi, Kanpur, and Hisar. Similarly, the mean PAH concentration (1194.9 ± 351.2 ng m⁻³) was approximately 50–55% lower than previous measurements at urban and industrial sites in Agra,⁸ yet substantially higher than those reported for Delhi and other Asian and global urban locations. These results highlight the persistently elevated burden of particulate matter and PAHs in Agra's urban atmosphere. A more detailed comparison of concentrations across different regions and time periods is provided in the SI (Table S6).

Fig. 2 Daily concentration of ∑16 PAHs as functions of TSP and meteorological parameters (RH, WS, and temperature). The grey colour bars represent the daily TSP concentration, the colour bars within the grey bars represent the daily ∑16-PAHs, while the colour shading of the ∑16-PAHs bars indicate the corresponding daily temperature range.

The daily concentrations of TSP and PAHs generally fluctuated around their average values throughout the sampling period. However, elevated levels were consistently observed from late January to the end of February (Fig. 2), indicating a sustained period of increased pollutant load. This trend may be attributed to seasonal factors such as lower atmospheric dispersion, increased biomass burning for heating, and regional emission sources.

Additionally, a distinct spike on March 8, 2023 (Fig. 2) coincided with the celebration of Holika Dahan and Holi, major festivals in Northern India. These events are known to trigger short-term high-pollution episodes due to cultural practices and intensified human activity.^46–50 During the Holi celebration, colored powders composed of a mixture of inorganic compounds such as metal salts, silica, and chalk powder are dispersed in air.⁵⁰ The night before Holi, “Holika Dahan,” involves significant biomass burning, contributing directly to increased aerosol loading. Additionally, traffic density typically surges during these festivals due to people commuting to their homes or to the market and their native places, further straining the local atmosphere's carrying capacity. All this biomass combustion and traffic emissions during these events notably produce substantial amounts of pollutants, as also documented in previous studies.^46–49

3.2 Variation of TSP-bound PAHs during cold weather months and hot weather months

The temporal distribution of individual PAHs, HMW-PAHs, and LMW-PAHs is depicted in Fig. 3a and b. During the CWM, the TSP concentrations ranged between 254.3 and 435.6 µg m⁻³ with an average of 349.4 ± 56.2 µg m⁻³, and the corresponding PAH concentrations varied between 1103.6 and 2437.4 ng m⁻³ with an average of 1857.3 ± 72.2 ng m⁻³. The concentration of LMW-PAHs was 446.1 ± 57.5 ng m⁻³, which contributed to 24% of total PAHs, while the concentration of HMW-PAHs was 1411.3 ± 70.5 ng m⁻³, which accounted for 76% of total PAH concentrations. In contrast, during the HWM, the TSP concentrations ranged between 214.8 and 324.1 µg m⁻³ with an average of 266.5 ± 33.7 µg m⁻³. The total concentration of 16-PAHs ranged between 335.8 and 1360.1 ng m⁻³ (average = 721.8 ± 14.1 ng m⁻³). The concentration of LMW-PAHs was 291.5 ± 22.3 ng m⁻³, making up 40% of the total PAH concentration, and the HMW-PAH concentration was 430.3 ± 6.6 ng m⁻³, accounting for 60% of the total PAH concentration, as shown in Fig. 3b. The results indicate a predominant contribution of HMW-PAHs to total PAHs in TSP fractions, with slightly higher contributions observed during the CWM (76%) than the HWM (60%). Conversely, the LMW-PAH fraction was more prominent during the HWM, constituting 40% of the total PAHs, likely due to increased volatilisation of LMW-PAHs under high temperature and high vapor pressure conditions than HMW-PAHs.²² The PAH concentrations bound to TSP were 2.6 times higher during the CWM than the HWM, while the mass concentration of TSP was only 1.3 times higher during the CWM than the HWM. This is also evident from Fig. 2, which shows the impact of temperature, relative humidity, and wind speed on PAH concentration. The temperature variation is depicted using a color gradient from navy blue (representing lower temperatures, 11.11–17.11 °C) to red (representing higher temperatures, 32.55–36.11 °C). The data show an inverse correlation between PAH and PM concentrations with both temperatures (r = −0.73, −0.81), while relative humidity exhibited a positive correlation (r = 0.60, 0.71). Lower temperatures in the CWM may reduce the volatilisation of semi-volatile organic compounds, leading to increased PAH adsorption onto PM. This phenomenon is further intensified by reduced wind speed, which limits dispersion and enhances pollutant accumulation. Conversely, higher temperatures in the HWM facilitate the volatilisation of LMW-PAHs, explaining their decrease during this period. The positive association with relative humidity suggests that higher humidity can promote atmospheric stability, thereby enhancing particle-bound PAH concentrations due to reduced dispersion.

Fig. 3 Temporal distribution of (a) individual PAHs and (b) HMW and LMW-PAHs. (c) PAH ring contribution during the CWM and HWM in TSP.

PAH concentrations in this study were consistently higher during both CWM (2.4 to 191 times) and HWM (6.6 to 160 times) compared to levels reported in other regions, such as China, Iran, Mexico, Mongolia, Korea, Turkey, and Delhi (Table 1). These results suggest that the PAH levels in this area may be influenced by local factors or specific sources of emissions, which differ from those in other regions. The consistently high concentrations across both seasons highlight the need for further research to identify the potential causes, such as industrial activities, vehicle emissions, or local environmental conditions, that could contribute to the higher PAH levels (as discussed).

Table 1 Comparison of TSP-bound PAH concentrations in Agra with other global and national studies

City/Region	Total PAHs	Period	CWM mean ± standard deviation (ng m^−3)	HWM mean ± standard deviation (ng m^−3)	Ref.
Agra, India	∑16PAHs	January–May 2023	1857.3 ± 147.3	721.7 ± 102.3	Present study
Bengbu, China	∑16PAHs	October 2021–September 2022	19.56 ± 7.80	3.31 ± 1.05	19
South Pars Industrial Region, Iran	∑16PAHs Urban	CWM, 2019	8.77–13.48		51
	Industrial		9.7–21.8
Monterrey Metropolitan Area, Mexico	∑13PAHs	January 21–March 8 2016	1.34–8.76		1
Mexico	∑13PAHs Urban	CWM spring	32 ± 18 to 116 ± 71	1.0 ± 0.5 to 5.9 ± 2.7	52
	Rural		3 ± 2	1.3 ± 1.9
Ulaanbaatar city, Mongolia	∑15PAHs	January–March 2017	22.2–773.0		53
Ulsan, South Korea	∑13PAHs	June 2013–May 2014	2.55–11.07	5.12	54
Erzurum, Turkey	∑18PAHs	HWM 2008; CWM 2008–2009	212	109	55
Delhi	∑16PAHs (TSP)	January and February 2019	122–625		56

---

3.3 Individual ring contribution of PAHs in TSP during the cold weather months and hot weather months

The ring distribution and levels of PAHs in TSP during the CWM and HWM are depicted in Fig. 3c and 4a and b. The ring-wise distribution of PAHs during the CWM followed the trend: 5-ring > 4-ring > 3-ring > 6-ring > 2-ring, contributing 34.5, 24.2, 21.1, 17.3, and 2.9% to Σ₁₆PAHs, respectively (Fig. 3c). Among the 5 rings, the most abundant PAHs was BghiP with concentrations ranging from 97.8 to 445.1 ng m³ (average of 259.1 ng m³), contributing 8.2–20.4% (average of 13.4%) to Σ16PAHs, followed by BbF (79.4–358.2 ngm³, average of 202.9 ng m³), contributing 3.9–21.4% (average of 10.9%). Among the 4 rings, the most abundant was BaA (100.4–277.3 ng m³, average of 160.4 ng m³), which contributed 5.1–13.9% (average of 8.8%). Though the overall percentage of 6-ring PAHs was the least to the total PAH concentration, IP was the second most abundant PAH, ranging from 55.7 to 525.3 ng m³ (average of 212.4 ng m³), and contributing 4.3–21.5% (average of 11.1%) to Σ₁₆PAHs. The presence of BghiP, IP, BaA, and BbF indicated combustion sources such as diesel emissions and coal burning. This seasonal increase is consistent with previous studies,^57–61 which reported higher concentrations of HMW-PAHs during CWM in the Indo-Gangetic Plain (IGP) and other regions. In cities like Agra, Prayagraj, and Kanpur, HMW-PAHs, particularly IP and BghiP, were linked to increased diesel and traffic emissions under stable cold weather conditions. These observations underscore the complex interplay of local emissions, meteorological conditions, and seasonal variations affecting PAH profiles. In contrast, during HWM, the trend shifted to 3-ring > 5-ring > 4-ring > 6-ring > 2-ring, with contributions of 37, 25, 22.5, 12.4, and 3.1%, respectively. Notably, 3-ring PAHs dominated the distribution, Ace emerged as the most prevalent PAH, with concentrations ranging from 33.5 to 194.4 ng m³ (average of 82.7 ng m³), contributing 6.1–16.7% (average of 10.9%) to Σ₁₆PAHs. This was followed by Fluo (32.0–153.5 ng m³, average of 64.0 ng m³), contributing 3.1–16.9% (average of 9.3%), and Anth (7.4–124.2 ng m³, average of 51.7 ng m³), contributing 0.8–14.8% (average of 6.9%). The increased concentrations of 3-ring PAHs during the HWM reflect a seasonal shift in PAH profiles and emission sources. These PAHs, including Ace, Fluo, and Anth, indicate local petrogenic sources such as gasoline/petrol combustion and petrochemical emissions. The proximity of the study site to multiple vehicle refueling stations (within 300 meters),^62,63 heavy traffic on nearby highways, over 1500 petha production units primarily reliant on coal, diesel generators (150), and manufacturing facilities contributed significantly to the PAH emission. A.D. Gupta et al.⁶⁴ also highlighted similar trends in urban environments.

Fig. 4 Box plot of individual PAHs during the (a) CWM and (b) HWM.

The CWM/HWM individual PAH ratio followed the trend: Ace (0.4) < Fluo (0.5) < Fla (0.9) < Acy (1.1) < BaP (1.3) < Pyr (2.0) < DbA (2.2) < NaP (2.4) < BkF (2.6) < Anth (2.6) < Phen (4.0) < BaA (4.0) < Chy (4.8) < BbF (4.9) < IP (5.1) < BghiP (6.0), as shown in Fig. S2. Except for Ace, Fluo, and Fla, all other PAHs showed a high CWM/HWM (>1), indicating their predominance during the CWM due to enhanced combustion activities and meteorological stability, whereas Ace, Fluo, and Fla were relatively more abundant in the HWM, reflecting contributions from fresh traffic emissions and petrochemical activities. The ring-wise PAH ratio showed high values (>3) for HMW-PAHs during the CWM (Fig. S3). Though the 5–6 ring PAHs had a high ratio in the case of 4 rings, it was not very prominent, probably because of their semi-volatile nature. High cold-weather months/hot-weather months ratios of PAHs like BghiP, IP, BbF, Chy, BaA, Phen, and Anth suggested that vehicular emissions and fossil fuel combustion are the major contributors during the CWM, as these PAHs are key indicators of these sources. In the published studies, a similar pattern for seasonal variation of PAHs has been reported.^65–68 C. Cheng et al.⁶⁹ also reported the highest concentrations of PAHs in the CWM. Our findings were also consistent with the results of other studies.^9,70–72 The results of J. Yang et al.⁷³ showed that the average total PAH concentrations varied from 176.94 in the HWM to 458.95 ng m⁻³ in the CWM, with an annual mean (SD) value of 300.35 (176.60 ng m⁻³). Studies^65,66,68 highlighted several reasons for such higher concentrations of PAHs in the CWM than in the HWM, which include reduced photo-thermal and chemical oxidation during low temperature and low solar radiation, high use of heating systems in the homes, increase in the emission sources, and thermal inversion causing poor dispersion of pollutants during the CWM. During CWM, the overall PAH concentrations increase due to reduced dispersion and enhanced combustion sources, with HMW-PAHs contributing significantly at traffic-dominated sites such as Bhagwan Talkies in Agra.⁸ Similarly, R. Jangirh et al.⁵⁶ observed elevated PAH concentrations during CWM across eight urban sites in Delhi, attributing this increase to heightened vehicular activity and decreased pollutant dispersion. S. K. Sharma et al.⁷⁴ documented an average PAH concentration of 667.7 ± 399.4 ng m⁻³ in TSP, with higher levels observed during CWM compared to HWM and monsoon periods. Furthermore, H. Sharma et al.⁷⁵ recorded elevated PAH concentrations (672 ng m⁻³) in CWM in Delhi. These results are consistent with other studies that reported increased PAH levels during CWM, highlighting the impact of seasonal variations on PAH concentrations.^19,76–78

The summation of nine combustion-derived PAHs, i.e., Fla, Pyr, Chy, BbA, BbF, BkF, BaP, IP, and BghiP (Σ₉COMPAHs^20,79,80), in the CWM was found to be 1302.9 ± 73.8 ng m⁻³, constituting ∼70% of total PAHs, while in the HWM, it was 381.5 ± 6.7 ng m⁻³ with ∼52% of total PAHs. This high concentration, especially during CWM, highlights enhanced combustion activities. Over the entire sampling period, the Σ₉COMPAHs accounted for 64% of the total PAH concentration, with values ranging from 39.1 to 138.1 ng m⁻³. The highest concentration of COMPAHs was observed on January 21, 2023, while a similarly high contribution (64% of total PAHs) was noted during Holika Dahan, attributable to intense biomass combustion.^46,48 The LMW/HMW ratio is commonly used to differentiate between pyrogenic and petrogenic sources of PAHs. A higher concentration of LMW-PAHs indicates a predominance of non-combusted petroleum products, while elevated HMW-PAHs are associated with pyrogenic sources, primarily originating from the combustion of fossil fuels.⁸¹ The low LMW-/HMW-PAH ratio (0.2–0.7) indicated a dominant pyrogenic source from fossil fuel combustion, supported by the negative correlation between the LMW/HMW ratio and COMPAH concentrations. Furthermore, a strong positive correlation between COMPAHs and total concentrations of PAHs (T-PAHs) (r = 0.8, Fig. S4) reinforced the predominance of combustion sources at the study site.

3.4 Diagnostic ratios

PAHs such as NaP, Acy, Anth, Phen, Fluo, Pyr, and Ace were consistently detected throughout the sampling period. These PAHs are commonly reported in emissions from multiple combustion sources, including wood burning, vehicular exhaust, and coke production. Their widespread occurrence indicates that they are not source-specific; instead, their diagnostic relevance lies in their relative abundance and co-occurrence with more source-distinct PAHs. Among these, Pyr and Fluo are widely recognised as indicators of coal combustion.^{19,20,82–84} HMW-PAHs such as BbF, BkF, DbA, IP, BaA, Chy, BaP, and BghiP are predominantly associated with vehicular emissions. Due to their low vapour pressure, these compounds preferentially partition into the particulate phase, making them more stable and source-resolvable in ambient air samples.⁸⁵ PAH diagnostic ratios, particularly those with the same molecular weight, serve as effective fingerprints for source identification. The ratio of COMPAHs/T-PAHs >0.1 indicates combustion sources, while ratios <0.1 suggest petrogenic sources, and the IP/IP + BghiP ratios of 0.2–0.5, >0.5, and >0.7 suggest fossil fuel emission, biomass/coal combustion, and gasoline emission, respectively.^19,85 To differentiate petrogenic from pyrogenic (high-temperature) sources, the Anth/(Anth + Phen) ratio is used—values <0.1 imply petrogenic sources, while values >0.1 indicate pyrogenic origins.⁹ Additionally, the BaP/BghiP ratio distinguishes between traffic (>0.6) and non-traffic (<0.6) sources.⁸⁶ The BaA/(BaA + Chy) ratio is used to infer specific combustion sources: values <0.2 suggest volatilisation of petroleum, 0.2–0.35 indicate biomass/coal combustion, and >0.35 point to vehicular emissions.^19,87 This study employed diagnostic ratios such as COMPAHs/T-PAHs, Anth/(Anth + Phen), BaA/(BaA + Chy), BaP/BghiP, Fla/(Fla + Pyr), and IP/(IP + BghiP) to trace PAH sources. These are summarised in Table S7, and the corresponding scatter plots are presented in Fig. 5a–d.

As shown in Fig. 5a, COMPAH/TPAH ratios in all samples were greater than 0.1, indicating combustion sources. A ratio of 0.5 is mainly observed in HWM, which might be from catalyst-equipped automobiles; however, the ratios in CWM vary between 0.62 and 0.81 might indicate a complex source. Similarly, Fig. 5b shows Anth/(Anth + Phen) ratios exceeding 0.1 in all samples, suggesting pyrogenic origins. The LMW-/HMW-PAH ratio of 0.54 further supports a pyrogenic source profile. The IP/(IP + BghiP) ratio revealed that 63% of the samples fell within the 0.2–0.5 range, indicative of fossil fuel combustion, while 38% exhibited values >0.5, pointing to biomass/coal combustion—findings corroborated by BaA/(BaA + Chy) ratios.

Fig. 5 Cross plots of (a) IP/IP + BghiP vs. COMPAHs/T-PAHs, (b) Anth/Anth + Phen vs. BaA/BaA + Chy, (c) IP/IP + BghiP vs. BaP/BghiP, and (d) IP/IP + BghiP vs. Fla/Fla + Pyr.

Cross-plots of IP/(IP + BghiP) versus BaP/BghiP suggest that PAHs at the study site originate from both traffic and non-traffic sources, with a predominance of non-traffic fuel combustion, as depicted in Fig. 5c. This is further supported by the Fla/(Fla + Pyr) ratio, a known indicator of emission sources: 0.4–0.5 implies fossil fuel/vehicular emissions, >0.5 indicates biomass/coal combustion, and >0.35 reflects petroleum emissions. The Fla/(Fla + Pyr) ratio during the study was 0.41 ± 0.14, pointing toward vehicular/fossil fuel emissions, as illustrated in Fig. 5d. Previous studies in Agra^88,89 have also identified vehicular exhaust from diesel combustion as a significant source of PAHs, with IP/(IP + BghiP) and BaA/(BaA + Chy) ratios of 0.62 and 0.66, respectively. Other northern Indian cities provide similar insights, as evident from Table S8, which provides a comparative summary of PAH sources across Indian cities based on diagnostic ratios. In Delhi, PAH concentrations ranged from 22.1 to 258.4 ng m⁻³, with primary sources being diesel and gasoline exhausts, as well as coal and wood combustion.^78,90 The IP/(IP + BghiP) ratio in Delhi ranged from 0.51 to 0.62, reflecting increased diesel vehicle usage.⁷⁸ In cities like Prayagraj, Jamshedpur, Jharsuguda, and Kanpur, diesel engine and gasoline combustion were identified as dominant sources.^91–93 Prayagraj, due to heavy tourist traffic and reliance on diesel public transport, shows strong diesel signatures in PAH profiles. Similarly, in Kolkata, vehicular emissions, coal, and wood combustion were identified as major contributors,⁹⁴ with diagnostic ratios of IP/(IP + BghiP), BaA/(BaA + Chy), and Anth/(Anth + Phen) reported as 0.44, 0.32, and 0.12, respectively. In Amritsar, a Fla/(Fla + Pyr) ratio of 0.47 and a BaP/BghiP ratio of 5.9 indicated significant non-traffic sources such as fossil fuel and coal combustion. However, IP/(IP + BghiP) ratios of 0.52 and 0.68 in Jamshedpur and Amritsar also pointed to diesel vehicle emissions. In Odisha, PAH sources were primarily diesel and gasoline emissions, reflected by IP/(IP + BghiP), BaA/(BaA + Chy), and Fla/(Fla + Pyr) ratios of 0.5, <0.3, and >0.5, respectively. B. Ambade et al.⁹² reported Flu/(Flu + Pyr) ratios of approximately 0.50–0.52 in Bokaro, Ghamhari, and Ranchi (Jharkhand), indicating biomass and coal combustion. In Jamshedpur, A. Kumar et al.⁶⁷ found pyrogenic dominance through elevated Anth/(Anth + Phen) (>0.1), IP/(IP + BghiP) (>0.5), BaA/(BaA + Chy) (>0.2), and COMPAHs/T-PAHs (0.8–0.9), all indicative of coal combustion, non-catalyst vehicle emissions, and diesel exhaust. In northeast India, traffic emission and petroleum combustion were identified as significant PAH sources, as evidenced by the IP/(IP + BghiP) and BaP/BghiP ratios.⁹⁵

3.5 PMF analysis

To quantitatively identify the potential sources of PAHs, Positive Matrix Factorization (PMF) version 5.0 was applied, providing a receptor modelling framework that incorporates measurement uncertainties and accounts for co-varying pollutant behaviour.³⁰ While diagnostic ratios offered preliminary qualitative insights into potential sources, PMF enabled a more robust and quantitative source apportionment of PAHs in the urban air samples from Agra.

All analysed PAHs exhibited signal-to-noise ratios greater than 7.6, confirming their suitability for inclusion in the PMF model. The model performance was evaluated using Q values: Q(robust) = 297.2 (excluding extreme outliers) and Q(true) = 275.4 (including all data points). These values were close to the theoretical Q value of 264, calculated using Q_theoretical = n × m − p × (m + n), where n is the number of samples, m is the number of species, and p is the number of resolved factors. The closeness of the observed Q values to the theoretical value, along with diagnostic parameters such as regression slopes near 1.0, intercepts approaching 0, and coefficients of determination R² >0.9, confirms the robustness of the model. Additionally, the absolute scaled residuals remained within acceptable limits, indicating that the PMF model provided a reliable fit and that the input uncertainties were appropriately estimated. Through PMF analysis, three distinct factors contributing to PAH concentrations were identified (Fig. 6).

Fig. 6 PMF model of PAHs in TSP.

Factor 1 was dominated by Ace, Fluo, and Acy, with contributions exceeding 75%, and moderately influenced by Anth (>55%). Ace, Acy, Fluo, and Anth compounds are associated with the volatilisation of petroleum products, crude oil spills, and natural gas combustion. Fluo is considered a characteristic compound of coal smelting. Additionally, Ace and Anth are principal pollutants from the coking industry. Consequently, Factor 1 was identified as a petroleum and natural gas combustion source.

Factor 2 exhibited high contributions from BbF, Phen, BkF, DbA, and BghiP, with levels exceeding 70%. These compounds are recognised as typical pollutants from petrol and diesel combustion, indicating that Factor 2 was attributable to traffic emissions.

Factor 3 was predominantly characterised by Pyr and Fla, with contributions exceeding 60%, along with moderate contributions from NaP (>55%). Pyr and Fla compounds are key markers of combustion sources, particularly wood and coal burning. Hence, Factor 3 was attributed to emissions from wood and coal combustion.

Overall, the total contributions of Factor 1, Factor 2, and Factor 3 demonstrated that PAHs in the urban air of Agra predominantly originated from mobile sources such as traffic emissions, particularly diesel exhaust, along with stationary sources such as coal combustion and petroleum-related sources, with contributions of 31.6%, 40.7%, and 27.7%. While diagnostic ratios and PMF results showed slight variations, they consistently highlighted traffic emissions and coal combustion as major sources of PAHs in the region.

3.6 Variation of OC and EC content and the OC/EC ratio

Fig. S5 illustrates the temporal variation of OC and EC concentrations in TSP, with higher concentrations observed during the CWM. During CWM, OC concentrations in TSP ranged from 4.7 to 19.9 µg m⁻³, while EC concentrations ranged from 1.4 to 7.1 µg m⁻³. During the HWM, OC concentrations ranged from 2.6 to 13.2 µg m⁻³, and EC concentrations ranged from 1.5 to 6.9 µg m⁻³. The average concentrations of OC and EC during the CWM were 15.1 ± 5.7 µg m⁻³ and 4.1 ± 1.7 µg m⁻³, respectively, compared to 6.2 ± 2.9 µg m⁻³ and 3.5 ± 1.8 µg m⁻³ during the HWM. OC constituted 68.1 ± 9.1% of TC, while EC contributed 38.9 ± 8.4%, indicating a higher relative abundance of OC in TSP. The OC concentrations in TSP were 1.6–3.8 times higher than EC concentrations across seasons, with OC showing stronger seasonal variability than EC (Fig. S5). This can be attributed to enhanced emissions from biomass burning and coal combustion, combined with lower boundary layer heights and effective aerosol trapping during the CWM.⁹⁶

The OC/EC ratio in TSP ranged from 1.6 to 3.1, suggesting contributions from multiple sources, including vehicular emissions, coal combustion, and biomass burning. These ratios align with source-specific ranges reported in the literature, such as those for diesel vehicles (0.2–0.8), gasoline vehicles (1.0–2.0), and biomass burning (3.8–13.2),^97–99 as shown in Table S9. Compared to other urban centres like Delhi (4.38), Kanpur (6.55), and Prayagraj (7.20),^100–102 the lower OC/EC ratios in Agra indicate a substantial influence of fossil fuel combustion, particularly vehicular emissions and coal burning.

The observed OC/EC ratios and high levels of OC and EC highlight the significant role of anthropogenic activities, particularly combustion processes, in carbonaceous aerosol emissions at the study site, which is consistent with previous studies at different sites of Agra.^103–105

3.7 Conditional probability function and bivariate polar plots

This section focuses on TSP concentrations at the study site. Fig. 7 presents a conventional CPF plot for TSP, identifying source directions where concentrations exceed the 75th percentile—71.2 µg m⁻³ in HWM and 51.5 µg m⁻³ in CWM, as shown in Fig. 7a and c. These plots indicate a strong probability of higher TSP levels originating from the northeast, an area associated with industrial activities and traffic emissions. These regions correspond to areas with high vehicular activity and fossil fuel combustion, including nearby highways, refueling stations, and residential areas using diesel generators and solid fuels for cooking.

Fig. 7 (a) CPF plot of TSP during CWM for concentrations exceeding the 75th percentile (73.2 µg m⁻³), (b) bivariate polar plot of TSP concentrations during CWM, with wind speed (m s⁻¹) on the radial axis and TSP concentrations (µg m⁻³) indicated by the color scale, (c) CPF plot of TSP during HWM for concentrations exceeding the 75th percentile (51.5 µg m⁻³), (d) bivariate polar plot of TSP concentrations during HWM, (e) CBPF plot for HWM, for TSP concentrations between the 75th and 95th percentiles, and (f) CBPF plot for CWM, in the same percentile range.

In contrast, Fig. 7b displays a bivariate polar plot of the same TSP data, revealing additional insights. It shows that while the northeast and southwest directions are the main contributors to TSP loading, the highest concentrations in CWM occur at low wind speeds (<1.2 m s⁻¹) from the northeast. In HWM, CPF results still point to the northeast as the dominant source direction. However, the bivariate plot reveals the highest concentrations at moderate wind speeds (1.2–1.8 m s⁻¹) from the southwest, along with moderate concentrations (>60 µg m⁻³) from the east and northeast.

The northeast region is home to coking, leather, glass, and petha industries, while the southwest includes busy market areas with heavy traffic and domestic activities like coal and biomass burning. During CWM, high TSP concentrations are linked to stable atmospheric conditions and ground-level sources. In contrast, HWM patterns suggest pollutant transport from more distant sources, facilitated by higher wind speeds.

Fig. 7b and d thus offer valuable insights beyond the CPF plots alone. They not only identify key source directions but also reveal how pollutants disperse under varying conditions. Together, the CPF and bivariate polar plots help distinguish at least two types of sources with distinct dispersion patterns and provide a clearer picture of the dominant influences on air quality at the monitoring site.

Fig. 7e and f illustrate the directional dependence of elevated TSP concentrations, highlighting key areas where concentrations exceed the 75th percentile and aid in the identification of potential source regions during CWM (Fig. 7e) and HWM (Fig. 7f). Analysing a range of percentile (75th to 95th) thresholds helps provide a clearer picture of pollutant source contributions. In CWM, two main zones stand out: one to the southwest at wind speeds of 1.2–1.6 m s⁻¹ and another to the east at speeds above 1.8 m s⁻¹, with concentrations ranging from 52 to 476 µg m⁻³. In HWM, these high-concentration zones expand to three areas, all in the northeast, at varying wind speeds (<1.2, 1.2–1.4, and >1.6 m s⁻¹), with concentrations between 73 and 254 µg m⁻³.

To further substantiate these directional patterns, 72 hour backward trajectory analyses were conducted using the NOAA HYSPLIT model for representative months in CWM (January) and HWM (April), as shown in Fig. S6. The trajectories during CWM indicated that air parcels arriving at the receptor site predominantly originated from nearby regions within ∼700 km, with low-altitude transport and limited vertical mixing. This supports the interpretation that local sources are the dominant contributors under cold weather conditions, consistent with suppressed boundary layer heights and stagnant meteorology. In contrast, trajectories during HWM extended up to ∼4000 km from the northeast and exhibited higher altitudes and more complex vertical profiles, reflecting enhanced potential for regional and long-range transport under warmer, convective conditions.

3.8 Health risks

The BaP-TEQ and BaP-MEQ of PAHs are presented in Fig. 8a–c. The BaP-TEQ and BaP-MEQ values during the CWM were 259.9 and 318.1 ng m⁻³, and 125.6 and 112.1 ng m⁻³ during the HWM, indicating significant carcinogenic and mutagenic risks. The BaP-TEQ was found to be 2.1 times higher and BaP-MEQ was 2.8 times higher in CWM than in the HWM. These values exceeded regulatory thresholds and values reported in other global urban settings, such as China (9.8–4.3 ng m⁻³ BaP-TEQ) and New York (0.1–1.9 ng m⁻³ BaP-TEQ; 0.062–2.394 ng m⁻³ BaP-MEQ).²²

Fig. 8 (a) Σ₁₆BaP-TEQ and Σ₁₆BaP-MEQ values; (b) BaP_eq-TEQ values of individual PAHs and (c) BaP_eq-MEQ values of 8 carcinogenic PAHs in TSP during CWM and HWM periods; (d) ILCR via ingestion, inhalation, and dermal exposure pathways for TSP-bound PAHs; (e) percentage contribution of individual PAHs to ILCR during CWM, HWM, and the entire sampling duration.

The BaP-TEQ values for individual PAHs ranged from 0.03 ng m⁻³ (Acy, Ace, and Fluo) to 108.4 ng m⁻³ (DbA) during CWM, and from 0.02 ng m⁻³ (NaP) to 59.14 ng m⁻³ (BaP) in the HWM. Similarly, BaP-MEQ values ranged from 2.86 ng m⁻³ (Chy) to 80.32 ng m⁻³ (IP) during CWM, and from 0.6 ng m⁻³ (Chy) to 59.04 ng m⁻³ (BaP) in the HWM. Among the PAHs analysed, BaA, BbF, BkF, BaP, DbA, and IP were the primary contributors to both BaP-TEQ and BaP-MEQ. During the CWM, BaP and DbA accounted for 71.8% of the total BaP-TEQ, highlighting their dominant role in carcinogenic risk. Conversely, during the HWM, BaP and IP together contributed 84.4% to the total BaP-TEQ, reflecting seasonal shifts in the influence of specific PAH species. For BaP-MEQ, the CWM saw a substantial contribution from IP, BaP, DbA, and BbF, which collectively accounted for 84.9% of the total mutagenic potential. In contrast, BaP emerged as the sole significant contributor during HWM, representing 52.7% of the total BaP-MEQ. This seasonal disparity emphasised the dynamic behavior of PAH species, influenced by emission sources, atmospheric conditions, and chemical transformations. These findings emphasise the substantial carcinogenic and mutagenic risks associated with TSP-bound PAHs, particularly during CWM, driven by specific high-emission compounds such as BaP, DbA, and IP. Addressing these risks necessitates focused mitigation strategies targeting dominant sources, such as biomass burning, vehicular exhaust, and coal combustion, and seasonal factors contributing to the elevated toxicity of PAHs.

During the CWM, the HMW-PAHs like BghiP, IP, BbF, Chy, BaA, and Phen contributed over 60% to the BaP-TEQ (Fig. S7). In contrast, during the HWM, mostly the LMW-PAHs, including Fluo, Ace, and Acy, along with BaP and Fla, were the dominant contributors to the BaP-TEQ. These results indicate a seasonal variation in the contribution of PAHs to the overall toxicity. The higher proportion of HMW-PAHs during the CWM suggests their persistence and elevated toxicity under these conditions, likely due to reduced atmospheric dispersion and increased particle-bound retention. Conversely, the prevalence of LMW-PAHs in the HWM points to their enhanced volatility and emissions from sources more active under HWM conditions, such as certain combustion processes. To mitigate PAH-associated toxicity, targeted measures should be tailored to seasonal emissions. During HWM, reducing emissions from LMW-PAH sources, such as industrial and residential combustion, should be prioritised. During CWM, control strategies should focus on HMW-PAH sources, particularly vehicular emissions, to decrease their significant impact on toxicity.

3.9 Cancer risk assessment

Humans are exposed to PAHs via three exposure pathways, i.e., inhalation, ingestion, and dermal absorption. ILCR levels derived for the particulate phase PAHs via different exposure routes for children and adults in TSP are depicted in Fig. 8d and e. The ILCR values for children and adults were higher during the CWM, indicating a significant cancer risk (Fig. 8d). The data show that adults and children have similar levels of susceptibility to cancer risk during both periods. Exposure via dermal contact was the dominant route for cancer risk, followed by ingestion. The inhalation risk for both adults and children was within the maximum permissible cancer risk limit (1 × 10⁻⁴). In contrast, the ILCR values for dermal and ingestion pathways surpassed the maximum permissible cancer risk limit. The primary contributors to this risk were identified as DbA and BaP, with moderate contributions from BbF, IP, BkF, and BaA, which the USEPA recognises as highly carcinogenic PAHs (Fig. 8e). Among these, DbA was the largest contributor to the ILCR, accounting for 46%, followed closely by BaP at 43%. The BaPeq contributions from DbA and BaP were significant, making up 40–50% of TSP-bound PAHs. DbA and BaP accounted for 85% of the ILCR during the HWM and 71% during the CWM. The minimal variation in these percentages across different seasons suggests that the cancer risk remains consistently high, particularly associated with DbA and BaP, which are key indicators of traffic-related emissions.¹⁰⁶

4 Conclusion

This study highlights the significant influence of seasonal dynamics on the concentration, composition, and toxicity of TSP-bound PAHs in urban residential environments. The findings reveal that CWM conditions exacerbate pollutant accumulation due to unfavourable meteorological factors and intensified combustion-related emissions. The dominance of high molecular weight PAHs (BaP, DbA, and IP) and elevated toxicity indicators during colder months highlights the compounded health risks posed to urban populations, particularly through dermal and ingestion pathways.

Source apportionment and directional analyses provide robust evidence linking PAH levels to vehicular traffic, industrial operations, and domestic heating, with clear spatial and temporal patterns. These insights emphasise the importance of targeted mitigation strategies that account for seasonal variability and localised emission sources. To effectively reduce PAH-related health risks and improve air quality, a multifaceted approach is essential. This includes enforcing stricter emission regulations, upgrading fuel standards, promoting cleaner energy alternatives, and enhancing public awareness. Prioritising interventions during high-risk periods, especially CWM, will be crucial for protecting vulnerable populations and facilitating sustainable urban living.

Author contributions

Simran Bamola: conceptualization, methodology, writing – original draft, writing – review & editing, visualization. Gunjan Goswami: methodology. Muskan Agarwal: methodology. Anita Lakhani: writing – review & editing, funding acquisition, supervision, final approval.

Conflicts of interest

There are no conflicts of interest.

Data availability

The authors confirm that the data supporting the findings of this study are available within the article. Data will be made available on request.

Supplementary information (SI) is available. See DOI: https://doi.org/10.1039/d5em00933b.

Acknowledgements

This work was supported by the Department of Science & Technology Fund for Improvement of S&T Infrastructure programme, Govt. of India (CS-II/2017/38), Indian Space Research Organisation-Geosphere Biosphere Programme under Atmospheric Trace Gases-Chemistry, Transport and Modelling project. Simran Bamola received funding from the Department of Science & Technology-Innovation in Science Pursuit for Inspired Research, Govt. of India (DST/INSPIRE/03/2022/000101), for the fellowship.

References

1. F. E. Longoria-Rodríguez, L. T. González, A. Mendoza, C. Leyva-Porras, A. Arizpe-Zapata, M. Esneider-Alcalá, K. Acuña-Askar, O. Gaspar-Ramírez, O. López-Ayala, J. M. Alfaro-Barbosa and O. V. Kharissova, Arch. Environ. Contam. Toxicol., 2020, 78, 377–391,  DOI:10.1007/s00244-019-00701-1.
2. S. Bamola, G. Goswami, S. Dewan, I. Goyal, M. Agarwal, A. Dhir and A. Lakhani, Urban Clim., 2024, 53, 101763,  DOI:10.1016/j.uclim.2023.101763.
3. K. J. Maji, A. K. Dikshit and A. Deshpande, Cogent Environ. Sci., 2016, 2, 1193110,  DOI:10.1080/23311843.2016.1193110.
4. J. Fan, X. Liu, W. Pan, M. W. Douglas and S. Bao, Emerg. Infect. Dis., 2020, 26, 1257,  DOI:10.3201/eid2606.200251.
5. IQAir, World Air Quality Report: Region & City PM2.5 Ranking, IQAir, 2023, pp. 1–41.
6. J. K. Mutuku, W. C. Hou and W. H. Chen, Aerosol Air Qual. Res., 2020, 20, 1094–1110.
7. F. Jahedi, H. Dehdari Rad, G. Goudarzi, Y. Tahmasebi Birgani, A. A. Babaei and K. Ahmadi Angali, J. Environ. Health Sci. Eng., 2021, 19, 851–866,  DOI:10.1007/s40201-021-00652-7.
8. P. K. Verma, D. Sah, R. Satish, N. Rastogi, K. M. Kumari and A. Lakhani, J. Environ. Manage., 2022, 317, 115456,  DOI:10.1016/j.jenvman.2022.115456.
9. A. Nadali, M. Leili, A. Bahrami, M. Karami and A. Afkhami, Ecotoxicol. Environ. Saf., 2021, 209, 111807,  DOI:10.1016/j.ecoenv.2020.111807.
10. A. Pandey, M. Brauer, M. L. Cropper, K. Balakrishnan, P. Mathur, S. Dey, B. Turkgulu, G. A. Kumar, M. Khare, G. Beig and T. Gupta, Lancet Planet. Health, 2021, 5, e25–e38.
11. V. P. Kanawade, A. K. Srivastava, K. Ram, E. Asmi, V. Vakkari, V. K. Soni, V. Varaprasad and C. Sarangi, Atmos. Environ., 2020, 222, 117125,  DOI:10.1016/j.atmosenv.2019.117125.
12. G. Pehnec and I. Jakovljević, Int. J. Environ. Res. Publ. Health, 2018, 15, 2485,  DOI:10.3390/ijerph15112485.
13. P. K. Verma, D. Sah, J. Dubey, K. M. Kumari and A. Lakhani, Chem. Res. Toxicol., 2021, 34, 743–753,  DOI:10.1021/acs.chemrestox.0c00378.
14. K. Ravindra, E. Wauters and R. Van Grieken, Sci. Total Environ., 2008, 396, 100–110.
15. M. Wang, S. Jia, S. H. Lee, A. Chow and M. Fang, J. Hazard. Mater., 2021, 409, 124531,  DOI:10.1016/j.jhazmat.2020.124531.
16. Y. He, W. He, C. Yang, W. Liu and F. Xu, Sci. Total Environ., 2020, 727, 138690.
17. M. Davoudi, A. Esmaili-Sari and N. Bahramifar, et al. , Environ. Sci. Pollut. Res., 2021, 28, 9276–9289.
18. T. Hu, Y. Mao, Y. Ke, W. Liu, C. Cheng, M. Shi, Z. Zhang, J. Zhang, S. Qi and X. Xing, Environ. Pollut., 2021, 291, 118139.
19. D. Wu, L. Chen, Z. Ma, D. Zhou, L. Fu, M. Liu, T. Zhang, J. Yang and Q. Zhen, Sci. Rep., 2024, 14, 5080,  DOI:10.1038/s41598-024-55695-1.
20. S. Naydenova, A. Veli, Z. Mustafa, S. Hudai, E. Hristova and L. Gonsalvesh-Musakova, J. Environ. Sci. Health, Part A, 2022, 57, 306–317,  DOI:10.1080/10934529.2022.2060669.
21. F. J. Zhu, W. L. Ma, L. Y. Liu, Z. F. Zhang, W. W. Song, P. T. Hu, W. L. Li, L. N. Qiao and H. Z. Fan, Atmos. Environ., 2021, 261, 118595,  DOI:10.1016/j.atmosenv.2021.118595.
22. P. K. Verma, D. Sah, K. M. Kumari and A. Lakhani, Environ. Sci.: Processes Impacts, 2017, 19, 1051–1060,  10.1039/C7EM00168A.
23. H. C. Kung, B. W. Huang, J. K. Mutuku and G. P. Chang-Chien, Aerosol Air Qual. Res., 2024, 24, 230270,  DOI:10.4209/aaqr.230270.
24. I. Shim, W. Kim, H. Kim, Y. M. Lim, H. Shin, K. S. Park, S. M. Yu, Y. H. Kim, H. K. Sung, I. C. Eom and P. Kim, Toxics, 2021, 9, 167.
25. M. Amin, A. A. Ramadhani, R. Zakaria, Z. A. Hanami, R. M. Putri, W. Phairuang, M. Hata and M. Furuuchi, Urban Sci., 2024, 8, 37,  DOI:10.3390/urbansci8020037.
26. P. Agarwal, L. Singh, M. Anand and A. Taneja, Arch. Environ. Contam. Toxicol., 2018, 74, 218–227.
27. N. Rajput and A. Lakhani, Polycyclic Aromat. Compd., 2012, 32, 48–60,  DOI:10.1080/10406630903299520.
28. D. C. Parashar, R. Gadi, T. K. Mandal and A. P. Mitra, Atmos. Environ., 2005, 39, 7861–7871,  DOI:10.1016/j.atmosenv.2005.08.034.
29. S. Bamola, P. K. Verma and A. Lakhani, Environ. Pollut., 2025, 127523,  DOI:10.1016/j.envpol.2025.127523.
30. P. Paatero and U. Tapper, Chemom. Intell. Lab. Syst., 1993, 18, 183–194,  DOI:10.1016/0169-7439(93)80055-M.
31. H. Wang, Y. Yang, T. R. Walker, Y. Wang, H. Wu, X. Wang and Q. Luo, Environ. Sci. Pollut. Res., 2022, 29, 73401–73413,  DOI:10.1007/s11356-022-21025-z.
32. K. Luo, Y. Zeng, M. Li, Y. Man, L. Zeng, Q. Zhang, J. Luo and Y. Kang, Sci. Total Environ., 2021, 774, 145770.
33. X. Feng, Y. Feng, Y. Chen, J. Cai, Q. Li and J. Chen, J. Clean. Prod., 2022, 330, 129850,  DOI:10.1016/j.jclepro.2021.129850.
34. N. D. MacAskill, T. R. Walker, K. Oakes and M. Walsh, Environ. Pollut., 2016, 212, 166–177,  DOI:10.1016/j.envpol.2016.01.060.
35. E. Davis, T. R. Walker, M. Adams and R. Willis, Mar. Pollut. Bull., 2018, 137, 285–294,  DOI:10.1016/j.marpolbul.2018.10.043.
36. E. Davis, T. R. Walker, M. Adams, R. Willis, G. A. Norris and R. C. Henry, Sci. Total Environ., 2019, 691, 528–537,  DOI:10.1016/j.scitotenv.2019.07.114.
37. E. Davis, T. R. Walker, M. Adams and R. Willis, Remediation J., 2019, 29, 51–62,  DOI:10.1002/rem.21600.
38. D. C. Carslaw and K. Ropkins, Environ. Model. Softw., 2012, 27, 52–61,  DOI:10.1016/j.envsoft.2011.09.008.
39. I. Uria-Tellaetxe and D. C. Carslaw, Environ. Model. Softw., 2014, 59, 1–9,  DOI:10.1016/j.envsoft.2014.05.002.
40. A. B. Patel, S. Shaikh, K. R. Jain, C. Desai and D. Madamwar, Front. Microbiol., 2020, 11, 562813,  DOI:10.3389/fmicb.2020.562813.
41. H. I. Abdel-Shafy and M. S. Mansour, Egypt. J. Pet., 2016, 25, 107–123,  DOI:10.1016/j.ejpe.2015.03.011.
42. S. Sahin, H. I. Ulusoy, S. Alemdar, S. Erdogan and S. Agaoglu, Food Sci. Anim. Resour., 2020, 40, 675–686,  DOI:10.5851/kosfa.2020.e43.
43. F. Kafilzadeh, Egypt. J. Aquat. Res., 2015, 41, 227–231,  DOI:10.1016/j.ejar.2015.06.004.
44. D. Assanov, A. Kerimray, B. Batkeyev and Z. Kapsalyamova, Aerosol Air Qual. Res., 2021, 21, 200663,  DOI:10.4209/aaqr.200663.
45. R. Kumar and K. M. Kumari, Urban Clim., 2015, 12, 11–20,  DOI:10.1016/j.uclim.2014.12.001.
46. A. Khajuria, P. Saxena, S. Sonwani and A. Srivastava, Inhalation Toxicol., 2024, 36, 228–239,  DOI:10.1080/08958378.2024.2341247.
47. R. P. Singh and A. Chauhan, in Asian Atmospheric Pollution, Elsevier, 2022, pp. 1–37,  DOI:10.1016/B978-0-12-816693-2.00029-9.
48. R. Yadav, G. Beig, V. Anand, R. Kalbande and S. Maji, Environ. Res., 2022, 205, 112465,  DOI:10.1016/j.envres.2021.112465.
49. N. Verma, A. Lakhani and K. M. Kumari, Atmos. Res., 2017, 197, 232–243,  DOI:10.1016/j.atmosres.2017.07.014.
50. A. Agrawal, V. K. Upadhyay and K. Sachdeva, Atmos. Environ., 2011, 45, 3640–3644,  DOI:10.1016/j.atmosenv.2011.04.006.
51. A. Ghadrshenas, T. Tabatabaie, F. Amiri and A. R. Pazira, Environ. Geochem. Health, 2023, 45, 1635–1653,  DOI:10.1007/s10653-022-01286-w.
52. O. López-Ayala, L. T. González-Hernández, V. M. Alcantar-Rosales, D. Elizarragaz-de la Rosa, M. E. Heras-Ramírez, L. G. Silva-Vidaurri, J. M. Alfaro-Barbosa and O. Gaspar-Ramírez, Atmos. Pollut. Res., 2019, 10, 1655–1662,  DOI:10.1016/j.apr.2019.06.006.
53. B. Byambaa, L. Yang, A. Matsuki, E. G. Nagato, K. Gankhuyag, B. Chuluunpurev, L. Banzragch, S. Chonokhuu, N. Tang and K. Hayakawa, Int. J. Environ. Res. Publ. Health, 2019, 16, 442,  DOI:10.3390/ijerph16030442.
54. T. N. Nguyen, K. S. Jung, J. M. Son, H. O. Kwon and S. D. Choi, Environ. Pollut., 2018, 236, 529–539,  DOI:10.1016/j.envpol.2018.01.080.
55. C. Paloluoğlu, H. Bayraktar, M. Aktan, F. S. Turalioglu and E. E. Gaga, Stoch. Environ. Res. Risk Assess., 2016, 30, 1223–1234,  DOI:10.1007/s00477-015-1110-9.
56. R. Jangirh, S. Ahlawat, R. Arya, A. Mondal, L. Yadav, G. Kotnala, P. Yadav, N. Choudhary, M. Rani, R. Banoo and A. Rai, Environ. Sci. Pollut. Res., 2022, 29, 17892–17918,  DOI:10.1007/s11356-021-16572-w.
57. D. M. Agudelo-Castañeda and E. C. Teixeira, Atmos. Environ., 2014, 96, 186–200,  DOI:10.1016/j.atmosenv.2014.07.030.
58. M. Akyüz and H. Çabuk, J. Hazard. Mater., 2009, 170, 13–21,  DOI:10.1016/j.jhazmat.2009.05.029.
59. P. Chen, S. Kang, C. Li, M. Rupakheti, F. Yan, Q. Li, Z. Ji, Q. Zhang, W. Luo and M. Sillanpää, Sci. Total Environ., 2015, 538, 86–92,  DOI:10.1016/j.scitotenv.2015.08.006.
60. M. Masiol, E. Centanni, S. Squizzato, A. Hofer, E. Pecorari, G. Rampazzo and B. Pavoni, Environ. Sci. Pollut. Res., 2012, 19, 3142–3151,  DOI:10.1007/s11356-012-0858-4.
61. M. Villar-Vidal, A. Lertxundi, M. M. De Dicastillo, J. I. Alvarez, L. Santa Marina, M. Ayerdi, M. Basterrechea and J. Ibarluzea, Chemosphere, 2014, 99, 233–238,  DOI:10.1016/j.chemosphere.2013.11.006.
62. C. V. de Souza and S. M. Corrêa, Fuel, 2016, 185, 925–931,  DOI:10.1016/j.fuel.2016.08.054.
63. J. N. Paim, A. G. Santos, R. G. Araujo, M. M. Nascimento, J. B. De Andrade and L. L. Guarieiro, Atmosphere, 2023, 14, 656,  DOI:10.3390/atmos14040656.
64. A. D. Gupta, A. Soni and T. Gupta, Sustain. Chem. Environ., 2023, 1, 100002,  DOI:10.1016/j.scenv.2023.100002.
65. G. Goudarzi, E. Idani, N. Alavi, S. Salmanzadeh, A. A. Babaei, S. Geravandi, M. J. Mohammadi, M. Mahboubi and M. Moradi, Toxin Rev., 2017, 36, 282–289,  DOI:10.1080/15569543.2017.1304422.
66. M. Hoseini, M. Yunesian, R. Nabizadeh, K. Yaghmaeian, R. Ahmadkhaniha, N. Rastkari, S. Parmy, S. Faridi, A. Rafiee and K. Naddafi, Environ. Sci. Pollut. Res., 2016, 23, 1820–1832,  DOI:10.1007/s11356-015-5355-0.
67. A. Kumar, B. Ambade, T. K. Sankar, S. S. Sethi and S. Kurwadkar, Sustain. Cities Soc., 2020, 52, 101801,  DOI:10.1016/j.scs.2019.101801.
68. E. Manoli, A. Kouras, O. Karagkiozidou, G. Argyropoulos, D. Voutsa and C. Samara, Environ. Sci. Pollut. Res., 2016, 23, 3556–3568,  DOI:10.1007/s11356-015-5573-5.
69. C. Cheng, C. Bi, D. Wang, Z. Yu and Z. Chen, Front. Earth Sci., 2018, 12, 63–71,  DOI:10.1007/s11707-016-0613-0.
70. M. Shams Solari, K. Ashrafi, A. Pardakhti, M. S. Hassanvand and M. Arhami, J. Environ. Health Sci. Eng., 2022, 20, 813–826,  DOI:10.1007/s40201-022-00821-2.
71. I. Jakovljević, Z. Sever Štrukil, R. Godec, I. Bešlić, S. Davila, M. Lovrić and G. Pehnec, Int. J. Environ. Res. Publ. Health, 2020, 17, 9587,  DOI:10.3390/ijerph17249587.
72. V. Ojeda-Castillo, A. López-López, L. Hernández-Mena, M. A. Murillo-Tovar, J. D. Díaz-Torres, I. Y. Hernández-Paniagua, J. del Real-Olvera and E. León-Becerril, Atmosphere, 2018, 9, 137,  DOI:10.3390/atmos9040137.
73. J. Yang, W. Xu and H. Cheng, Atmosphere, 2018, 9, 63,  DOI:10.3390/atmos9020063.
74. S. K. Sharma, T. K. Mandal, S. Jain, Saraswati, A. Sharma and M. Saxena, Bull. Environ. Contam. Toxicol., 2016, 97, 286–293,  DOI:10.1007/s00128-016-1836-1.
75. H. Sharma, V. K. Jain and Z. H. Khan, Chemosphere, 2007, 66, 302–310.
76. D. K. Singh, S. Sharma, G. Habib and T. Gupta, Environ. Sci. Pollut. Res., 2015, 22, 12458–12468,  DOI:10.1007/s11356-015-4413-y.
77. Y. Ren, B. Zhou, J. Tao, J. Cao, Z. Zhang, C. Wu, J. Wang, J. Li, L. Zhang, Y. Han and L. Liu, Atmos. Res., 2017, 183, 322–330,  DOI:10.1016/j.atmosres.2016.09.015.
78. D. P. Singh, R. Gadi, T. K. Mandal, C. K. Dixit, K. Singh, T. Saud, N. Singh and P. K. Gupta, Environ. Monit. Assess., 2010, 169, 1–13,  DOI:10.1007/s10661-009-1145-9.
79. X. Wen, H. Zhang, H. Zhang, G. Liang, B. Ren, W. Peng, Y. Yao and M. Zhu, PLoS One, 2025, 20, e0334592,  DOI:10.1371/journal.pone.0334592.
80. K. Hayakawa, N. Tang, W. Xing, P. K. Oanh, A. Hara and H. Nakamura, Atmosphere, 2021, 12, 256.
81. M. A. Hassanien and N. M. Abdel-Latif, J. Hazard. Mater., 2008, 151, 247–254,  DOI:10.1016/j.jhazmat.2007.05.079.
82. P. Vishwakarma, P. Rajeev, S. Rabha, N. Islam, B. K. Saikia and T. Gupta, J. Atmos. Chem., 2023, 80, 251–269,  DOI:10.1007/s10874-023-09450-x.
83. R. Sun, Y. Sun, Q. X. Li, X. Zheng, X. Luo and B. Mai, Sci. Total Environ., 2018, 640, 264–272,  DOI:10.1016/j.scitotenv.2018.05.320.
84. G. Shen, S. Tao, W. Wang, Y. Yang, J. Ding, M. Xue, Y. Min, C. Zhu, H. Shen, W. Li and B. Wang, Environ. Sci. Technol., 2011, 45, 3459–3465,  DOI:10.1021/es104364t.
85. R. Cao, H. Zhang, N. Geng, Q. Fu, M. Teng, L. Zou, Y. Gao and J. Chen, Sci. Total Environ., 2018, 625, 1486–1493,  DOI:10.1016/j.scitotenv.2017.12.335.
86. G. Gbeddy, P. Egodawatta, A. Goonetilleke, E. Akortia and E. T. Glover, Environ. Pollut., 2021, 269, 116103,  DOI:10.1016/j.envpol.2020.116103.
87. J. Hu, C. Q. Liu, G. P. Zhang and Y. L. Zhang, Atmos. Res., 2012, 118, 271–279,  DOI:10.1016/j.atmosres.2012.07.015.
88. N. Rajput and A. Lakhani, Atmosfera, 2010, 23, 165–183.
89. A. Lakhani, Sci. World J., 2012, 2012, 781291,  DOI:10.1100/2012/781291.
90. S. Sarkar and P. S. Khillare, J. Atmos. Chem., 2011, 68, 107–126,  DOI:10.1007/s10874-012-9212-y.
91. P. Rajeev, A. K. Singh, G. K. Singh, R. C. Vaishya and T. Gupta, Urban Clim., 2021, 35, 100755,  DOI:10.1016/j.uclim.2020.100755.
92. B. Ambade, T. K. Sankar, L. K. Sahu and U. C. Dumka, Urban Sci., 2022, 6, 60,  DOI:10.3390/urbansci6030060.
93. S. Ekka, S. K. Sahu, S. Dwivedi, S. N. Khuman, S. Das, O. Gaonkar and P. Chakraborty, Environ. Geochem. Health, 2022, 44, 3991–4005.
94. M. Saha, D. Maharana, R. Kurumisawa, H. Takada, B. G. Yeo, A. C. Rodrigues, B. Bhattacharya, H. Kumata, T. Okuda, K. He and Y. Ma, Aerosol Air Qual. Res., 2017, 17, 2247–2262,  DOI:10.4209/aaqr.2017.05.0163.
95. M. Gope, R. E. Masto, J. George and S. Balachandran, Sustain. Cities Soc., 2018, 38, 616–626,  DOI:10.1016/j.scs.2018.01.006.
96. J. K. Buch, A. U. Lewandowska, M. Staniszewska, K. A. Wiśniewska and K. V. Bartkowski, Atmosphere, 2021, 12, 1005,  DOI:10.3390/atmos12081005.
97. E. H. Park, J. Heo, S. Hirakura, M. Hashizume, F. Deng, H. Kim and S. M. Yi, Air Qual. Atmos. Health, 2018, 11, 1167–1178.
98. H. H. Yang, N. B. Dhital, L. C. Wang, Y. S. Hsieh, K. T. Lee, Y. T. Hsu and S. C. Huang, Aerosol Air Qual. Res., 2019, 19, 1439–1449.
99. J. Tian, H. Ni, J. Cao, Y. Han, Q. Wang, X. Wang, L. W. Chen, J. C. Chow, J. G. Watson, C. Wei and J. Sun, Atmos. Pollut. Res., 2017, 8, 521–527.
100. K. Ram and M. M. Sarin, Atmos. Environ., 2011, 45, 460–468,  DOI:10.1016/j.atmosenv.2010.09.055.
101. K. Ram and M. M. Sarin, J. Aerosol Sci., 2010, 41, 88–98,  DOI:10.1016/j.jaerosci.2009.11.004.
102. S. K. Sharma, T. K. Mandal, M. Saxena, A. Sharma, A. Datta and T. Saud, J. Atmos. Sol. Terr. Phys., 2014, 113, 10–22,  DOI:10.1016/j.jastp.2014.02.008.
103. A. Satsangi, Environ. Sci. Technol., 2016, 1, 279–287.
104. T. Pachauri, A. Satsangi, V. Singla, A. Lakhani and K. M. Kumari, Aerosol Air Qual. Res., 2013, 13, 977–991,  DOI:10.4209/aaqr.2012.10.0263.
105. P. Tripti, R. K. Saraswat, S. Vyoma and A. Lakhani, Res. J. Recent Sci., 2013, 2, 1–7.
106. A. Hanedar, K. Alp, B. Kaynak and E. Avşar, Sci. Total Environ., 2014, 488, 437–446,  DOI:10.1016/j.scitotenv.2013.11.123.

---