Introduction

Accelerated urbanization and economic growth in India have led to rapid increase in vehicle activity, making vehicular emissions a major source of urban air pollution1. Urban-scale simulations suggest that fine particulate (PM2.5) pollution originating from vehicle exhaust and on-road resuspended dust, contributes to 25–50%, with hotspots concentrated along major road corridors of Indian cities2. Increasing evidence indicates atmospheric pollution, influenced by particle size and chemical composition of carbonaceous aerosols, is associated with climate impacts and significant health risks3. A recent study revealed 2.18 million deaths per year are attributable to ambient air pollution in India4. Carbonaceous aerosols include organic carbon (OC), such as polycyclic aromatic hydrocarbons (PAHs) along with various other hydrocarbons, elemental carbon (EC), and inorganic carbon compounds, primarily carbonates5. Contribution of total carbonaceous aerosol in ambient PM2.5 mass measured across different seasons in India is reported to be in the range of ~20–85%6,7,8.

Atmospheric PAHs are of particular concern as they are known to have mutagenic and carcinogenic properties9,10. These compounds are a group of semi-volatile organics with at least two or more fused carbon and hydrogen aromatic rings, fused together in linear, angular or cluster arrangements11. PAHs undergo transformation pathways, leading to the formation of more polar derivatives like nitro- oxy-, and hydroxy PAHs, which can be even more toxic and mutagenic than their parent PAHs12. Depending on their molecular weight, PAHs exist in different physical states: 2- and 3-ringed PAHs are typically found in the gaseous phase, while PAHs with more than 5 rings tend to associate with particulate matter, 4 ringed-PAHs can exist in both gaseous and particulate phases13. Higher molecular weight and enhanced polarity leads to increase in the toxicity of PAHs14. Based on their toxicity and abundance, the United States Environmental Protection Agency (USEPA) has identified sixteen priority PAHs11. Primarily, the effect of PAHs on human health is based on the duration and route of exposure (through inhalation, ingestion, and dermal contact), concentration of PAHs to which one is exposed to, and the relative toxicity of PAHs10.

Growing evidence from several studies has indicated that PAHs can influence the optical properties and radiative effects of brown carbon (BrC), which is the light-absorbing component of OC15,16. BrC absorbs light in the ultraviolet-to-visible spectrum, with its absorption exhibiting a strong dependence on wavelength17. PAHs have been identified as important BrC chromophores due to their large conjugated polycyclic structures, which impart strong light-absorbing properties in the near-UV range (300–400 nm)18,19. Many studies20,21,22 have shown that water-soluble organic carbon (WSOC) fractions, especially humic-like substances (HULIS) and other individual WSOC species, contribute to the light absorption of organic aerosols. However, while the water-soluble fraction of BrC has been extensively studied23,24,25, the contribution of water-insoluble species like PAHs to light absorption properties of atmospheric pollutants remains largely unquantified, in particular, from vehicular emissions.

PAHs from on-road vehicles originate from exhaust as well as non-exhaust sources such as brake wear, tyre wear, asphalt, and resuspended dust26,27. The characteristics and emission factors of PAHs emitted from vehicles are significantly influenced by factors such as engine type, operating conditions, fuel composition, vehicle mileage, and the presence of catalytic converters28. Sampling in roadway tunnels provide a more realistic traffic profile than laboratory-based dynamometer tests29, as they allow to investigate the characteristics of pollutants discharged from heterogenous mix of vehicle (or fleet composition) operated at controlled conditions like fixed vehicle speed and least influence of meteorological parameters inside the tunnel. Many recent studies in India have reported that on-road exhaust emission is a significant source of PAHs in urban areas14,30,31,32,33. However, so far no attempt has been made to characterize, and estimate emission factors of PAHs from real-world vehicular fleet. To the best of our knowledge, this is the first study, to report the profile and emission factors of real-world vehicular fleet PAHs using road tunnel measurement in India.

In this study, the primary objectives were to: (1) characterize on-road vehicular PAHs and assess the influence of traffic parameters on PAHs, (2) investigate the association of PAHs with bulk chemical composition and light absorption properties of BrC, (3) assess the leading health risks, and (4) estimate vehicular PAHs emission factors.

Results and Discussion

Traffic flow characteristics

The hourly average traffic of 1560(±410) vehicles consisted 80% LDVs and 20% HDVs, driving at an average speed of 78(±4) km h−1. Peak traffic observed during morning rush hours was 21% higher than afternoon non-rush hours. There was no significant difference (p > 0.05) in traffic volume and composition between North-and South-Bore of the Kamshet-I Tunnel (KT). Higher HDV fraction was observed during the early morning hours which was due to the movement restriction of HDVs in Mumbai (restriction time: 8:00 am–11:00 am and 5:00 pm–9:00 pm) and in Pune (between 7:00 am and 9:00 pm). The weekend traffic was observed to be 27% higher than the weekday traffic indicating inter-city traffic movement. Diesel vehicles across all vehicle class dominated the fleet (51%) followed by gasoline vehicles (41%), and compressed natural gas (CNG) vehicles comprised 8% of the fleet. While the share of gasoline vehicles was higher during the morning hours, the share of diesel vehicles were observed to higher during the afternoon hours. A significant share (14%) of older vehicles (>10 years: pre-2009 vehicles) was observed in the fleet. Across all vehicle class, aged vehicles were observed to be highest in trucks (23%), followed by light commercial goods vehicles (LCGVs) and buses (22%) and cars (11%). In terms of emission standards, vehicles that comply with poorer and obsolete emission technologies: Bharat Stage (BS) III, II, and I were observed to be most prevalent among trucks (44%), followed by buses (40%), LCGVs (37%), and cars (18%). Super-emitters (SE), characterized as older, poorly maintained, and overloaded heavy-duty diesel vehicles, accounted for 21%(±3%) of the total fleet34.

PAHs levels and composition

The average (±standard deviation) mass concentration of all sixteen PAHs at the entry and exit of the tunnel are shown below in Fig. 1. Concentration of all PAHs at the tunnel exit were significantly higher (1.4 to 3.8 times) than at entry of the tunnel (p < 0.05). The average concentrations of all sixteen PAH compounds (ΣPAH16) at the tunnel entry and tunnel exit were 107(± 46) and 203( ± 72) ng m−3, respectively. The molecular weights of 2–3 rings PAHs are 128 to 178 g mol−1, 4 rings: 202 to 228 g mol−1 and 5–6 rings: 252 to 276 g mol−1. The PAHs were classified into (1) low-molecular weight PAHs (ΣLMWPAHs) which include compounds of 2 and 3-rings (i.e. NAP, ACY, ACE, FLU, PHE, and ANT), and (2) high-molecular weight PAHs (ΣHMWPAHs) which include 4–6 rings compounds (i.e. FLA, PYR, BaA, CHR, BbjF, BkF, BaP, DahA, IcdP, and BghiP. The percentage mass fraction of each individual PAH to the total PAH concentration at tunnel entry and tunnel exit has been provided in Supplementary Fig. 1.

Fig. 1: Average PAHs concentration in ng m−3 at the entry and exit of Kamshet-I tunnel.
Fig. 1: Average PAHs concentration in ng m−3 at the entry and exit of Kamshet-I tunnel.The alternative text for this image may have been generated using AI.
Full size image

[Note: Number of samples, N = 14 (Tunnel entry), N = 14 (Tunnel exit)].

ΣHMWPAHs concentrations were significantly (p < 0.05) higher than ΣLMWPAHs at both the tunnel entry and tunnel exit. The ΣHMWPAHs accounted for 75% to 89% of the total PAHs concentrations and they were significantly higher during the morning hours when the traffic volume was higher. These findings were similar to previous road tunnel study13 that reported high molecular weight PAHs collectively accounted for about 88% of the total PAHs at the tunnel entry and exit. Also, as Zhao et al.35 reported that most of the PAHs emissions from traffic sources were high-rings PAHs, it can be implied that the PAHs emissions were driven by vehicles at the tunnel, especially at the exit. Among the high molecular weight PAHs too, it was observed that the 4-rings PAHs were higher than the 5-6 rings PAHs, this can be due to higher fraction of diesel vehicles in the study tunnel. The 4-ring PAHs constituted the highest proportion of 46(±8)% of the total PAH concentration. Previous road tunnel study in China too reported the highest contribution of 4-ring PAHs to be around 58%36. The 4-rings PAHs mainly derive from diesel vehicle emissions, while 5–6 rings PAHs were reported to be linked with gasoline vehicle emissions37. While the exit/entry ratio in HMWPAHs were upto 3 times higher as compared to the exit/entry ratio in LMWPAHs this too indicated the levels at the tunnel exit were true signature of vehicular exhausts and the tunnel entry levels were influenced by other ambient sources. Rajeev et al. 38 highlighted HMWPAHs as molecular precursor for soot particles. The presence of high fraction of diesel vehicles in both LDVs and HDVs led to the high contribution of HMWPAHs to total PAHs at the study tunnel.

Diagnostic ratios used for distinguishing traffic and non-traffic emissions, diesel and gasoline-related emissions, calculated from the concentration differences obtained in this study are listed and compared with previous studies35,39,40,41 in Supplementary Table 1. These ratios also show the intra-source variability and inter-source similarities. BaP/BghiP ratio of 0.57(±0.03) at the tunnel entry suggested the contribution of non-traffic sources and influence of other ambient sources including biomass burning from the neighbouring village and photochemical activities. While at the tunnel exit higher BaP/BghiP ratio of 0.84(±0.14) indicated significant contribution of traffic sources. ΣLMWPAH/ΣHMWPAH ratio <1 and ΣCOMB/ΣPAHs ratio close to 1 indicated pyrogenic sources and combustion signature, respectively at both tunnel entry and tunnel exit. The IcdP/(IcdP+BghiP) ratio of 0.48(±0.07) and 0.63(±0.11) at tunnel entry and tunnel exit, respectively, indicated the contribution of diesel exhaust emissions. Overall, ratios of FLA/(FLA + PYR), BaA/(BaA+CHR), and IcdP/(IcdP+BghiP) revealed higher diesel exhaust signature which can be attributed to the higher fraction of diesel vehicles and HDVs in the fleet.

Association of PAHs with bulk chemical species and traffic parameters

Scatter matrix plot of Pearson correlation at 95% confidence interval (CI) were developed to examine the association between the concentrations of the PAHs compounds, carbonaceous fractions and traffic parameters as shown in Fig. 2. Collectively, all the higher molecular weight (ΣHMWPAH) significantly (p < 0.05) correlated with all traffic parameters, indicating that the HMWPAHs were predominantly influenced by the vehicles driving inside the tunnel. ΣHMWPAH was strongly influenced by the traffic volume (R = 0.88), super-emitters (R = 0.84) and HDVs (R = 0.77). Diesel and gasoline vehicles also significantly (p < 0.05) influenced the ΣHMWPAH concentrations. Similar findings were reported by previous roadway tunnel studies that have reported >3-rings PAHs from vehicular sources13,42,43,44,45. Even in Indian cities of Allahabad and Kanpur, the higher percentage contribution of ΣHMWPAH to the total PAHs, mainly the 4-ring PAHs (FLA, PYR, BaA, CHR) was contributed to diesel exhaust emissions14. The ΣLMWPAH concentrations did not show any significant (p > 0.05) association with any of the traffic parameters, indicating they were driven by other petrogenic sources.

Fig. 2: Scatter matrix plot of Pearson correlation of PAHs concentrations, carbonaceous fractions and traffic parameters.
Fig. 2: Scatter matrix plot of Pearson correlation of PAHs concentrations, carbonaceous fractions and traffic parameters.The alternative text for this image may have been generated using AI.
Full size image

All the pollutant concentrations here are of at the tunnel exit (Number of samples, N = 14). [Note: Correlations between the parameters are denoted by the coloured circles in the figure, red: positive correlation and blue: negative. Association between the parameters is represented by the size of the circles, bigger the circles show stronger the association, and smaller circles show weaker association].

The bulk chemical composition: OC and EC profiles have been presented in our previous study46. While the ΣLMWPAH showed strong association with OC1 and OC2 (R > 0.91), ΣHMWPAH showed strong association with OC3, OC4, EC1 and EC2 (R > 0.75). The associations between the molecular weights of the PAHs and the volatility of OC fractions were examined and are presented in Supplementary Fig. 2. An inverse relationship between molecular weight and volatility was established. A strong and significant (p < 0.05) association between ΣHMWPAH and low volatile OC (LVOC) (R = 0.96) revealed that high molecular weight PAHs are less volatile. Conversely, the association between ΣLMWPAH and high volatile OC (HVOC) (R = 0.92) indicated that lower molecular weight PAHs exhibit higher volatility. Weak association was observed between HVOC and ΣHMWPAH (R = 0.18), as well as between LVOC and ΣLMWPAH (R = 0.2). This also indicate the difference in sources that originate HVOC and ΣHMWPAH. Vehicular exhausts, particularly diesel vehicles release more LVOC at higher temperature than HVOC46, and they emit more of HMWPAH13. Diesel vehicles and HDVs led to higher concentration levels of LVOC than HVOC from the measured fleet. Consequently, the higher abundance of LVOCs resulted in the increased levels of HMWPAH relative to LMWPAH. This relationship underscores the significant role of diesel emissions in shaping the chemical profile of atmospheric organic pollutants particularly in urban environments where diesel-powered vehicles are prevalent.

Correlation of PAHs with light absorption properties of BrC

Light absorption properties of the measured aerosols have been presented in our previous study46. The correlations between PAHs and other light absorption properties like the mass absorption coefficient (MAC) and absorption Ångström exponent (AAE) of BrC were examined (shown in Supplementary Fig. 3 and Supplementary Fig. 4, respectively). A strong correlation (R = 0.93) was observed between low volatile ΣHMWPAH and MACMS-BrC at 370 nm, indicating that with increasing molecular weight, the light absorbing property may increase. It also suggests ΣHMWPAH are associated with other compounds (LVOCs) emitted from the same source, that are light absorbing. Strong association between LVOC and MACMS-BrC (R = 0.87) was observed, as shown in Supplementary Fig. 5. This highlighted the contribution of hydrophobic OC fraction, which is dominated by HMWPAH, to the light-absorbing properties of BrC. Also reported by a previous study that PAHs are likely contributors to BrC chromophores, originating from common sources, including vehicular exhaust47. The high volatile ΣLMWPAH showed weak correlation with both water-soluble mass absorption coefficient (MACws-BrC) as well as methanol-soluble MAC (MACMS-BrC) at 370 nm. Similar findings were reported by Saleh et al. 48 where inverse correlation of BrC light absorption with volatility and direct correlation of light absorption with molecular weight was observed. Strong correlation between ΣHMWPAH and MACMS-BrC potentially may be due to methanol’s extraction efficiency of around 85%49.

Methanol extracts a broader spectrum of organic compounds, including both hydrophilic and hydrophobic compounds, unlike water, which extracts only hydrophilic compounds. These hydrophobic compounds include higher molecular weight PAHs, mostly derived from fossil fuel combustion50. Saleh et al.48 highlighted that with decreasing solubility in water as well as in organic solvents like methanol, the light absorbing property of the hydrophobic compounds increases. MS-BrC had higher absorption than WS-BrC across all wavelengths between 300 and 800 nm, which is consistent with the previous studies as well50,51. The wavelength dependence of light absorption, indicated by absorption Ångström exponent (AAE), between 370 and 550 nm for both water- and methanol-extracts was investigated. We observed strong correlation between LMWPAHs and AAEMS-BrC (R = 0.91), while no correlation (R = 0.05) between LMWPAHs and MACMS-BrC was observed, which indicates LMWPAHs has higher AAE, and lower MAC compared to HMWPAHs. This corroborates well with the BrC-BC light absorption continuum, where MAC is inversely linked with AAE as reported by Navinya et al. 52, and directly proportional with the molecular weight48. These findings can be used to represent the BrC absorption variability from real-world vehicular emissions in climate models. While in this study, light-absorbing HMWPAHs and their relationship with the bulk species and the light absorption properties were studied, nitro-PAHs that are also known to be highly light absorbing53 have not been included, which is a limitation of the study.

Health risk characterization

Based on the concentrations of each PAH compound and their corresponding toxic equivalent factor (TEF) values, benzo[a]pyrene-equivalent (BaPeq) concentrations,the cancer risks were estimated. The average ∑BaPeq concentration was 13.5(±5.6) ng m−3 and 25.2(±4.9) ng m−3 at the tunnel entry and tunnel exit, respectively, these values exceed the standard. The current regulatory limit for BaP in India is established by the National Ambient Air Quality Standards (NAAQS) at a daily average concentration of 1 ng m−3. Table 1 presents the values of BaPeq, lifetime average daily dose (LADD), and incremental lifetime cancer risk (ILCR) measured at the tunnel entry. The values at tunnel exit have been presented in Supplementary Table 2. Although previous tunnel studies36 have reported ILCR values at both tunnel exit and tunnel entry, the ILCR values at the tunnel exit represent extreme exposure scenarios. Since the probability of continuous exposure inside a tunnel environment is low, these values are less relevant for general population risk assessment. While ILCR values at the tunnel entry reflect the risk for populations with high exposure, such as roadside residents, inter-city travellers, and freight vehicle drivers within the study area. These values provide insight into the potential health risks from vehicular emissions in high-exposure environments, such as near roadways or inter-city expressways. BaPeq concentration for the seven carcinogenic PAHs (BaA, CHR, BbjF, BkF, BaP, DahA, and IcdP) identified by WHO International Agency for Research on Cancer (IARC) accounted for 98.9% of the total PAHs.

Table 1 Lifetime average daily dose (LADD) and Incremental lifetime cancer risk (ILCR) from PM2.5-bound 16 priority PAHs to humans through inhalation measured at the entry of Kamshet-I tunnel
Full size table

The USEPA uses ILCR to evaluate cancer risks from long-term carcinogen exposure, ILCR below 10−6 is acceptable, 10−6 to 10−4 indicates moderate risk, and above 10-4 signals high risk54. The ILCR values follow BaP>DahA>IcdP>BbjF for both adult and children at the tunnel exit, while at the tunnel entry it was DahA>BaP>BkF>IcdP for both adult and children at the tunnel entry. In this study, the observed ILCR values of 8.9 in one million and 3.7 in one million for adult and children, respectively at the tunnel exit are approximately double the ILCR values observed for both adult and children at the tunnel entry. This increase highlights the significant health risks associated with exposure to elevated levels of vehicular emissions. These values are similar to the ones reported in ambient PM2.5 over central Indo-Gangetic Pain with significant contribution of vehicular sources: 4.5 to 8.1 in a million for adult and 2.6 to 4.8 in a million for children14. However, the ILCR values from this study are notably higher than those reported for vehicular emission exposure at a national highway (NH) site (1.8E−10 for adult, 5E−11 for children) and a busy traffic site (1.7E−10 for adult, 4.8E−11 for children) in Asansol city, India32. Additionally, another road tunnel study in Turkey reported estimated cancer risks of urban population due to inhalation of PAH: 1.1 to 4.9 in one million55. While the observed ILCR values in this study fall under moderate risk, their magnitude highlights the pressing need for implementing effective mitigation measures, such as stricter vehicular emission standards (especially for diesel vehicles and HDVs), and a transition to cleaner energy vehicles, which collectively would help minimize pollutant exposure and associated health risks.

Estimated emission factor (EF) of PAHs

PAHs EFs (in μg veh−1 km−1) were estimated using the distance based approach, and they are presented in Fig. 3. The total average PM2.5-bound PAH (ΣPAH16) EF in this study was 52.7( ± 4.5) μg veh−1 km−1. ΣHMWPAH EF of 42.1(±8.6) μg veh−1 km−1 contributed to ∼80% of the total EF. Similar EFs were reported by previous tunnel studies with HDV fleet: 50.3 μg veh−1 km−1 by Alves et al.56 in Liberdade Avenue tunnel, Portugal, and 61 μg veh−1 km−1 by Zhao et al.35 in Yangkou Tunnel, China. However, the EFs of 4-5 rings PAHs that are driven by diesel vehicles and HDVs were lower in this study as compared to the ones reported by other studies with HDV fleet, this could be because there were only 20% HDV in our study tunnel whereas, the tunnel in Portugal had all HDV fleet and the tunnel in China had all HDV fleet. The EF from this study was about 2 times higher than the EFs reported from LDV and gasoline dominated tunnels13,57. This could be because of the presence of HDVs, super-emitters and higher diesel vehicles in the fleet as compared to the all LDV or gasoline dominated fleet. Chen et al.43 reported that the PAHs EF from diesel vehicles are 3.37 times higher compared with that from gasoline vehicles. Ho et al.42 observed that diesel vehicles emitted 5.45 times PAHs compared to gasoline vehicles.

Fig. 3: Estimated PAHs emission factors and comparison with other previous tunnel studies.
Fig. 3: Estimated PAHs emission factors and comparison with other previous tunnel studies.The alternative text for this image may have been generated using AI.
Full size image

1This study; 2He et al.57; 3Alves et al.56; 4Fang et al.13; 5Zhao et al.35.

Methods

Roadway tunnel sampling

The field campaign was conducted in the Kamshet-I roadway tunnel (18°44′15″N, 73°31′56″E), located on one of India’s busiest inter-city arterial corridors in terms of both passenger and goods movements, Mumbai-Pune Expressway in December 2019. The tunnel has two unidirectional bores, the North Bore (NB): Mumbai to Pune (949 m in length) and the South Bore (SB): Pune to Mumbai (969 m in length), the detailed description of the tunnel is given in our previous study34. The ventilation system inside the tunnel consisting of jet fans were all closed during the sampling period so as to ensure that the accumulation ensure the measured aerosols inside the tunnel were solely attributed to emissions from the vehicles traversed.

Filter-based aerosol measurements were carried out concurrently at the entry and exit of the tunnel for consecutive two weeks (NB: 6th to 12th December, 2019 and SB: 13th to 19th December, 2019). PM2.5 gravimetric samples were obtained using MiniVol samplers (Airmetrics, USA35 operated at an airflow rate of 5 L min⁻¹, along with a multi-stream PM sampler58 with an airflow of ∼20 L min−1. The tunnel entry samples provided an ambient background concentration which was subtracted from the exit concentrations to calculate the elevated vehicle-induced concentration. The sampling stations were located 50 m away from the entrance and about ∼100 m away from the exit of the tunnel, and the instruments were set up on the sidewalks at a height of 1.5 m above the road surface. Gravimetric samples were collected over five-hour intervals during two distinct periods: the morning peak traffic hours (7:00 AM to 11:59 AM) and the afternoon non-peak traffic hours (12:00 PM to 5:00 PM). These sampling periods were determined based on preliminary trial measurements conducted beforehand to assess the prevailing traffic conditions.

A high-resolution video camera was used to examine the traffic fleet based on its volume and composition every hour during the measurement period. From the recorded videos, the vehicles were counted manually and were classified into: (1) light-duty vehicles (LDV): passenger cars and light commercial goods vehicles (LCGVs) that weigh ≤3.5 tonnes, (2) heavy-duty vehicles (HDV): buses and trucks that weigh >3.5 tonnes. Around 2100 license plate numbers were manually collected randomly from all the vehicle categories. Two-wheelers and three-wheelers were not allowed to ply inside the tunnel due to restrictive movements on the high-speed expressway. Individual vehicle information like vehicle makes and model, fuel type, age of the vehicle, and emission standard were obtained from the license plate numbers using a mobile application-based software, “RTO vehicle information”34,59. The spot speed of the vehicles was captured using a velocity speed gun (Bushnell, USA60). The wind speed inside the tunnel was measured using a hot-wire anemometer (model GM8903, BENETECH, China) and was positioned inside the tunnel to capture representative airflow conditions. Due to the confined boundary conditions of the tunnel, the movement of vehicles created a piston effect, driving the wind in the same direction as the vehicles. The wind speed recorded by the anemometer corresponds to the airflow predominantly parallel to the tunnel axis. This measurement captures the vector component of wind speed aligned with the tunnel, which is the primary airflow direction influenced by vehicle movement.

Chemical analysis

Carbonaceous content (OC and EC) in the measured PM2.5 samples were analysed using the Desert Research Institute (DRI) Model 2015 thermal and optical carbon analyzer. Under the Interagency Monitoring of Protected Visual Environments (IMPROVE_A) Protocol, eight fractions of carbon were derived in two phases of heating: OC1, OC2, OC3, and OC4 (volatile) evolved in a non-oxidizing Helium (He) atmosphere; residual OC (OP), a pyrolyzed carbon fraction and EC1, EC2, and EC3 (non-volatile) evolved in a 2% O2 and 98% He atmosphere61,62. In this study, OC fractions were categorized into two groups: (1) high volatile organic carbon (HVOC), referred to as the sum of OC1 and OC2, emitted at 180 °C and 280 °C, respectively, and (2) low volatile organic carbon (LVOC), referred to as the sum of OC3 and OC4, emitted at 480 °C and 580 °C, respectively. The light absorption properties of BrC were analysed using the Ultraviolet-Visible spectrophotometer (UV-VIS Evolution 220, Thermo-Fisher) in water and methanol extracts. The light absorption spectra of the extracted samples were measured from 300 to 800 nm. Light absorption properties such as mass absorption coefficient (MAC) at 370 nm and the spectral dependence of absorption or absorption Ångström exponent (AAE) between 370 and 550 nm of water-soluble brown carbon (WS-BrC) and methanol soluble brown carbon (MS-BrC) were estimated using empirical Eqs. (1) and (2), respectively. A detailed explanation of the methods is provided in our previous study46.

$${{\rm{MAC}}}_{{\rm{\lambda }}}\left({{\rm{m}}}^{2}{{\rm{g}}}^{-1}\right)=\frac{{{\rm{b}}}_{{\rm{abs}}}}{{\rm{C}}}$$
(1)
$${{\rm{AAE}}}_{{{\rm{\lambda }}}_{1-{{\rm{\lambda }}}_{2}}}=-\frac{\mathrm{ln}({{\rm{b}}}_{{\rm{abs}}}\left({{\rm{\lambda }}}_{1}\right)/{{\rm{b}}}_{{\rm{abs}}}{({\rm{\lambda }}}_{2}))}{\mathrm{ln}({{\rm{\lambda }}}_{1}/{{\rm{\lambda }}}_{2})}$$
(2)

where babs is the absorption coefficient at λ = 370 nm, C represents the measured mass concentration of WSOC (C) for water soluble and OC (C) for methanol soluble extract. λ1 and λ2 used for AAE calculation are 370 and 550 nm, respectively.

PAHs analysis was carried out on the remaining PM2.5 samples collected on quartz fibre filters after OC-EC analysis. In this study, the sixteen priority USEPA PAHs, which include 2-ringed PAH: Naphthalene (NAP); 3-ringed PAH: Acenaphthylene (ACY), Acenaphthene (ACE), Fluorene (FLU), Phenanthrene (PHE), Anthracene (ANT); 4-ringed PAH: Fluoranthene (FLA), Pyrene (PYR), Benzo[a] Anthracene (BaA), Chrysene (CHR); 5-ringed PAH: Benzo[b,j] Fluoranthene (BbjF), Benzo[k] Fluoranthene (BkF), Benzo[a] Pyrene (BaP), Dibenzo[a,h] Anthracene (DahA); 6-ringed PAH: Benzo[g,h,i] Perylene (BghiP), and Indeno[1,2,3-c,d] Pyrene (IP) were analysed. Our study focussed on particle-phase PAHs only and gas-particle phase partitioning was not investigated. As lower molecular weight PAHs (in particular those < 4-rings) can have substantial gas-particle phase partitioning, our results are a conservative estimate of the PAHs profiles and emission factors. Gas chromatography coupled with mass spectrometry (GC-MS) was used to separate, identify and quantify the target PAHs by solvent extraction method. For sample extraction and preparation, the solvents dichloromethane (DCM), acetone and toluene of HPLC grade (purity ≥99.8%; Merck) were used for sample extraction and preparation (detailed explanation given in Rajeev et al.14. Quantification of all target compounds was done on GC-MS (Agilent GC: 7890B; MSD: 5977B) equipped with a capillary column (DB-5MS column) and helium was used as a carrier gas14.

Quality control and quality assurance (QA/QC)

Mini volume (MiniVol) samplers and multi-stream PM sampler underwent routine flow rate calibration to ensure precise sampling. Quality control measures were meticulously implemented to ensure the reliability and accuracy of the data, including the preparation and analysis of both laboratory and field blank samples (N = 2 for each time period, at the tunnel entry and tunnel exit). Some contamination was observed in laboratory blanks, with higher than 25% of mean tunnel sample concentrations were excluded. The analytical field blank values for each target compound were significantly lower than the concentrations detected in the actual samples, validating that contamination during the process was minimal. The concentrations of all the PAH compounds reported in this study are field blank corrected mass concentration. Additionally, as a quality control measure, the variability in internal standards (phenanthrene d-10 and perylene d-12 of 150 ng each was spiked to standards and samples) response across the analysed samples was found to be small (within ±5%) between subsequent runs which ensured the consistency and accuracy of the analytical procedure.

Human health risk assessment

Inhalation of PM2.5 bounded PAHs was regarded as risk on human health estimated in this study. Benzo[a]pyrene (BaP) is recognized as one of the most potent carcinogens among PAHs63. The potential health risks associated with PAH exposure through inhalation pathway were assessed using the BaP equivalent concentration (BaPeq) and incremental lifetime cancer risk (ILCR) methodologies recommended by the US Environmental Protection Agency (EPA)54.

Carcinogenic risk

To evaluate the cumulative carcinogenic risk associated with different PAHs, a toxicity equivalent factor (TEF) is assigned to each PAH compound relative to BaP, which can be used as the reference compound for health risk assessment64. It allows each PAH’s concentration to be adjusted based on its carcinogenic potency compared to BaP32. BaPeq of individual PAHs and total carcinogenic equivalent concentration was calculated using Eq. (3).

$${{\rm{TBaP}}}_{{\rm{eq}}}=\mathop{\sum}\limits_{{\rm{i}}=1}^{{\rm{n}}}({{\rm{C}}}_{{\rm{i}}}\,{\rm{x}}\,{{\rm{TEF}}}_{{\rm{i}}})$$
(3)

where, TBaPeq represents the total carcinogenic BaP equivalent concentration (in ng m−3); Ci represents the concentration of individual PAH compound and TEFi represents corresponding toxicity equivalent factor of the PAHs. The TEF values of NAP, ACY, ACE, FLU, PHE, FLA, and PYR are 0.001; ANT, CHR, and BghiP are 0.01; BaA, BbjF, BkF, and IcdP are 0.1; BaP and DahA are 1.

Incremental lifetime cancer risk

The increased risk of cancer over a lifetime as a result of continuous exposure to a carcinogen is represented by incremental lifetime cancer risk (ILCR). ILCR due to PAHs from respiratory exposure was calculated using lifetime average daily dose (LADD) and cancer slope factor (CSF). LADD refers to the intake amount (mg) of a chemical species per unit of body weight (kg) over an extended period, typically accounting for chronic or lifetime exposure14, used to evaluate the potential health impacts of long-term exposure to hazardous environmental species. Equations (4) and (5) are used to calculate LADD and ILCR, respectively.

$${\rm{ILCR}}={\rm{LADD\; x\; CSF}}$$
(4)
$${\rm{LADD}}({\rm{mg}}\,{{\rm{kg}}}^{-1}\,{\cdot}\,{{\rm{day}}}^{-1})=\frac{\,{{\rm{C}}}_{{\rm{p}}}{{\times}}\; {\rm{AIR}}\; \times\; {\rm{FE}}\; \times\; {\rm{LED}}\; \times\; {\rm{UCF}}}{{\rm{BW}}\; \times\; {\rm{AT}}}$$
(5)

where, Cp represents BaPeq concentration of individual PAHs (in ng m−3); AIR is the air inhalation rate: 20 m3 day−1 for adults and 10 m3 day−1 for children65; FE indicates frequency of exposure in days: 365 days for both adults and children32; LED is the lifetime exposure duration: based on US EPA (USEPA, 2011) guidelines, 24 years for adults to represent typical long-term exposure, and 6 years for children to capture early-life exposure, which is critical because children are more vulnerable to the effects of carcinogens are considered in this study. AT is the average exposure time to carcinogens in days: for both adults and children, the US EPA assumes that health risks from carcinogens are cumulative over a lifetime, and thus considers a lifetime averaging period of 70 years, equivalent to 70 × 365 days66; UCF refers to unit conversion factor (from ng to mg); BW represents body weight: 60 kg for adults and 18 years for children (ICMR 2010); CSF (per m kg−1 day−1) for all PAHs was considered using the CSF of BaP, and it was taken as 3.164.

Emission factor (EF) estimation

EF of PAHs from vehicles passing through the study roadway tunnel during a time interval was calculated using the distance-based equation67,68 as shown in Eq. (6).

$${{\rm{EF}}}_{{\rm{PAH}},{\rm{veh}}}=\frac{\left({{\rm{C}}}_{{\rm{PAH}},{\rm{exit}}}-{{\rm{C}}}_{{\rm{PAH}},{\rm{entry}}}\right){\rm{x}}\; {\rm{U}}\; {\rm{x}}\; {\rm{T}}\; {\rm{x}}\; {\rm{A}}}{{\rm{V}}\; {\rm{x}}\; {\rm{L}}}$$
(6)

where EFPAH,veh (mg veh−1 km−1) is the average fleet emission factor of PAH; CPAH,exit (µg m−³) represents the concentration of PAH compounds measured 100 m inside the tunnel near the exit, while CPAH,entry (µg m−³) corresponds to the concentration measured 50 m inside the tunnel near the entrance; U (m s−1) is the wind speed parallel to the tunnel sensed by the 3-D sonic anemometer; A (m2) is the tunnel cross-sectional area, which is 157 m2 in this study; V is the total number vehicles that travelled through the tunnel during the time interval T (s); and L is the length of the tunnel between the two monitoring stations (at entry and exit) (L is 879 m in South-Bore and 767 m in North-Bore).