Characteristics and Source-specific Health Risks of Ambient PM 2.5 -bound PAHs in an Urban City of Northern Taiwan

Polycyclic aromatic hydrocarbons (PAHs) with highly toxic compounds mainly exist in small-sized particles and can induce considerable human health risks. Studies on PM 2.5 -bound PAHs and their source-specific human health risks still remain scarce. Daily PM 2.5 samples ( n = 119) were collected every three days from 2016 to 2017 in Taipei city, Taiwan. Fifteen PAHs in PM 2.5 were analyzed via gas chromatography tandem mass spectrometry (GC/MS-MS). We utilized a positive matrix factorization (PMF) model, diagnostic ratios, and potential source contribution function (PSCF) to identify the origins of PM 2.5 -bound PAHs. The annual concentration of total PAHs (TPAH) was 0.79 ± 0.67 ng m – 3 (range = 0.11 – 3.27 ng m – 3 ). The highest and lowest values of TPAH appeared in winter and autumn with a mean of 1.36 ng m – 3 and 0.43 ng m – 3 , respectively. The contributions of high-molecular-weight PAHs (HMW PAHs) to TPAH were notably higher than those of low-molecular-weight PAHs (LMW PAHs) during the sampling period. Benzo[ghi]perylene (BghiP) accounted for the highest percentage (23.9%) of TPAH among selected congeners. Traffic emissions (31.3%) were identified as the predominant contributor to ambient PM 2.5 -bound PAHs, followed by industrial emissions (29.2%), evaporated/unburned oil (22.3%), and biomass/coal combustion (17.1%). Apart from the local sources, PSCF-derived results showed that emissions from industrial activities in northeast China and shipping around the Yellow Sea and East China Sea could affect the PAHs in the study area. Traffic emissions were the strongest contributor to human health risk, thus pointing to the significance of control over vehicle exhaust. This study suggests that it is necessary to distinguish the sources of the PM 2.5 -bound PAHs in order to underpin preventive and mitigative strategies for protecting environmental and public health.


INTRODUCTION
Particulate matter (PM) in the atmosphere has received extensive public and research attention because of its adverse impact on air quality and human health (Anderson et al., 2012;Pope and Dockery, 2006;Rohr and Wyzga, 2012), climate change (Bonasoni et al., 2010;Park et al., 2010;Strawa et al., 2010), and visibility (Watson, 2002) in the last decades.Among suspended particles, fine particles (PM2.5)have come to the forefront as their physical and chemical properties pose a greater threat to air quality, human health, and climate change than those of coarse particles (PM2.5-10)(Pope and Dockery, 2006;Squizzato et al., 2016).PM2.5 is referred to as a human carcinogen (Group 1) by the International Agency for Research on Cancer (IARC) (Loomis et al., 2013).
Polycyclic aromatic hydrocarbons (PAHs) are a large group of organic compounds consisting of two or more benzene rings that can be derived from both natural and anthropogenic sources, such as biomass burning, incomplete combustion, and vehicle emissions (Wang et al., 2013;Lin et al., 2019).PAHs in ambient air exist in the vapor phase and are adsorbed on airborne particulate matter, depending on the atmospheric conditions (e.g., temperature and relative humidity) and the properties of aerosols and individual PAHs (Zhou et al., 2014;Zhang et al., 2017).In general, high-molecular-weight PAHs (HMW PAHs, 5-6 rings) exist in the particulate phase, whereas lowmolecular-weight PAHs (LMW PAHs, 2-4 rings) exist in the vapor phase because of their vapor pressures (Wei et al., 2015;Tomaz et al., 2016;Zhang et al., 2017).Previous studies have shown that HMW PAHs are released from biomass burning (Chiu et al., 2011;Pham et al., 2019) and traffic emissions (Guarieiro et al., 2014;Shen et al., 2013), whereas LMW PAHs are emitted from evaporated oil and biomass burning (Liu et al., 2015b;Bragato et al., 2012;Sun et al., 2018).
Globally, PAHs have attracted research interest owing to their wide distribution and adverse effects on human health and the environment (Hong et al., 2015;Mehmood et al., 2020;Pongpiachan et al., 2015).With their different toxicity levels and ambient concentrations, PAHs have been classified as human mutagens and carcinogens by the United States Environmental Protection Agency (U.S. EPA) and World Health Organization (WHO) (WHO, 2000; U.S. EPA, 2017).The European Union (EU) has also recommended ambient air quality standards for benzo[a]pyrene (BaP) (EU, 2005).Given high population densities and heavy traffic flows, exposure to PAHs via inhalation has become an important issue in the urban areas.Studies conducted earlier have demonstrated that urban areas may have an exceptionally high concentration of PM2.5-bound PAHs (He et al., 2014;Yu et al., 2018;Mehmood et al., 2020).PAHs occur in higher concentrations in cold seasons than in warm seasons (Teixeira et al., 2012;Chen et al., 2021).This can be attributable to the meteorological conditions, such as reduced wind speed, lower atmospheric mixing heights, and decreased photochemical oxidation of PAHs due to reduced solar radiation exposure.High PAH concentrations in winter associated with coal combustion for heating and elevated emissions from gasoline vehicles with cold starting and idling at low temperatures were also reported (Wu et al., 2014).The dominant sources of PAHs in the urban areas include traffic emissions and biomass/coal combustion (Callén et al., 2014;Manoli et al., 2016;Chao et al., 2019).
Taipei, the biggest city in Taiwan, has a population of approximately 2.7 million and an extremely high population density (9314 people per km 2 ).With the heavy traffic flow across Taipei, PAH concentrations and characteristics in ambient aerosols have attracted increasing attention in recent decades (Lee et al., 2015;Chen et al., 2018), in particular, with regard to human health risks (Wang et al., 2021;Chuang et al., 2020).Hsu et al. (2019) reported several methods to quantify the impact on air quality and health risk of long-range transport (LRT) in the Taipei city and pointed to coal combustion-and traffic-related emissions as the primary PAH sources.Exposure to air pollution from cooking (APC) is a non-negligible source of PAHs (Yu et al., 2015).Huang et al. (2012) found that exposure to traffic-emitted fine particles and PM2.5-bound PAH could cause DNA damage.Zhao et al. (2011) used the excess lifetime cancer risk (ESCR) to assess inhalation exposure to APC at night markets in Taiwan and estimated that ESCR was above the acceptable risk range of 10 −6 -10 −4 for workers, as suggested by the U.S. EPA.Room et al. (2023) investigated the PM2.5-boundPAHs in an urban area and city tunnel in Taipei, and it indicated that coal-fired power plants and traffic emission were the main sources.In addition, Ngo et al. (2022) found that incineration, fossil fuel and biomass burning were the main contributors of PAHs in Taiwan.Previous studies in Taipei have identified several sources and estimated the human health risks of PAHs from receptor models.However, studies about the physicochemical properties and source-specific health impacts of PM2.5-bound PAHs in Taipei city remain limited.
In this study, a year-round sampling campaign of PM2.5 was conducted in Taipei city.The seasonal variations in PM2.5 and PM2.5-bound PAHs were investigated.Sources of PM2.5-bound PAHs were identified via a receptor model of positive matrix factorization (PMF) and ancillary information regarding the diagnostic ratios of congeners.A potential source contribution function (PSCF) was also performed to determine the regional contributions of PM2.5-bound PAHs sources.The identified sources were further utilized to assess the health risks of PM2.5-bound PAHs.The source-oriented health assessment of PM2.5-bound PAHs is a novel approach that offers valuable insights into developing mitigation strategies for air pollution.

Description of the Study Area and Sampling
From December 2016 to November 2017, daily PM2.5 samples (24-hour integrated) were gathered every three days at the Air Quality Monitoring Station (AQMS) situated in Guting (25°1′15′′N, 121°31′49′′E), which is located in Taipei city, Taiwan.The sampling site was located in the residential and commercial areas surrounded by the main traffic road, as shown in Fig. 1.A total of 119 samples were collected and analyzed in this study.A sampler (PQ200, BGI Inc., MA, USA) at a flow rate of 16.7 L min -1 was used to collect the PM2.5 samples on the 47 mm quartz fiber filters for 24 h in the ambient environment.The total volume of air that passed through the filter for each sample was recorded, and the sampler flow rate for each sample was readjusted via a calibration rotameter (Defender 520, MesaLabs, NJ, USA).The mass of PM2.5 on the filters was weighed via an electronic microbalance (MX5, Mettler-Toledo, Switzerland, ± 0.002 mg).Before and after sampling, the samples were kept in a desiccator shielded from light at a temperature of 23 ± 1°C and relative humidity of 40 ± 5% for 24 h.

PAH Analysis
The quartz filters were extracted with 25 mL of a mixture of dichloromethane (DCM) and hexane (2:1) for 15 mins at 100°C via a MARS Xpress microwave digestion instrument.After extraction, an automated evaporation and concentration system (XcelVap, Horizon Technology Inc., Salem, USA) was used to concentrate the extract to about 0.8 mL, and the remaining extract was purified by using solid-phase extraction (ProElutTM Bap extraction cartridge, CA, USA).The extracted solution was then reconstituted to 1 mL with hexane, and 100 µL of the internal standard mixture was added (naphthalene-d8, acenaphthene-d10, phenanthrene-d10, chrysene-d12, perylene-d12).A gas chromatographer (Agilent Technologies 7890 B) coupled with mass spectrometric detection (Agilent Technologies 7000C MS-MS Triple Quad) and multiple reaction monitoring (MRM) modes with a capillary column (Agilent 122-9632 DB-EUPAH) with dimensions of 0.25 mm (internal diameter) × 30 m (length) × 0.25 um (film thickness) was used to analyze PAHs.A constant flow rate (1.2 mL min -1 ) of helium was used as the carrier gas.Gas chromatography tandem mass spectrometry (GC-MS/MS) was used according to the following procedure: the oven temperature was held at 80°C for 1 min, heated to 200°C at a rate of 25°C min -1 , and elevated to 335°C at a rate of 8°C min -1 (hold for 17.325 min).Each compound was positively identified via the retention time in GC-MS/MS in MRM mode.Table S1 lists the optimized MRM transitions with collision energies.The method detection limit (MDL) of all target compounds ranged from 0.63 to 1.19 ng mL -1 (Table S1).A mixture of PAH standards was added to the blank filters by using the same analysis protocol to quantify the recovery rate.The recovery rates of the target PAHs ranged from 79.2% to 98.0% (Table S1).Fifteen PAHs, listed as priority pollutants by U.S. EPA, were determined: acenaphthylene (AcPy), acenaphthene (Acp), fluorene (Flu), phenanthrene (PA), anthracene (Ant), fluoranthene (FL), pyrene (Pyr), benzo

Source Identification
Positive matrix factorization (PMF) 5.0 model was used to identify and quantify the contributions of PAH sources.Details of the PMF model can be found in the U.S. EPA PMF 5.0 Fundamentals & User Guide (Norris et al., 2014).Briefly, a modified dataset can be regarded as a data matrix X of i × j dimension as Eq. ( 1): where i is the number of samples; j is the number of species; p is the number of factors; fkj is the species profile of each source; gik is the amount of mass contributed by each source; and eij represents the residuals.Uncertainty of the PMF model was estimated by using the method detection limit (MDL) and error fraction.Eqs. ( 2) and (3) were used when the concentrations were lower and higher than MDL, respectively, as follows: Potential source contribution function (PSCF) analysis (Polissar et al., 2001;Pongkiatkul and Kim Oanh, 2007) was performed via Zefir, an Igor-based package (Petit et al., 2017), to determine the origin of air pollutants according to Eq. ( 4).The spatial resolution of each cell in the PSCF results was 0.3° × 0.3° with the 75 th percentile threshold when the total number of trajectory endpoints (mij) was calculated.The weighting function was derived from Eq. ( 5) as follows: where nij is the total number of trajectory endpoints in the ij th cell; mij is the total number of trajectory endpoints in the ij th cell related to PAH concentrations above the defined threshold; and wij is an arbitrary weighting function used to reduce uncertainty in PSCF results.The 120-h back trajectories at an altitude of 200 m above sea level were obtained via the HYbrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT, version 4.9) model (Stein et al., 2015) developed by the National Oceanic and Atmospheric Administration Air Resources Laboratory.Global Data Assimilation System (GDAS) data were used for the calculations.

Health Risk Assessment
The Toxic Equivalent Quantity (TEQ) and Mutagenic Equivalent Quantity (MEQ) values were used to estimate the toxicity and mutagenicity of individual PAHs regulated by U.S. EPA, each of which has a corresponding toxic and mutagenic equivalency factor, respectively.Among PAHs, BaP, a reference compound, is the most suitable indicator to assess the carcinogenic potential of PAHs in the human body through different exposure pathways (Nisbet and LaGoy, 1992).The BaP equivalent concentration (BaPeq) was estimated from Eqs. ( 6) and ( 7).The concentration of the target PAH (ng m -3 ) and its toxic and mutagenic equivalence factors were obtained from previous studies in related literature (Nisbet and LaGoy, 1992;Schoeny and Poirier, 1993;Durant et al., 1999), as shown in Tables S5 and S6.
where Ci is the concentration of the target PAH (ng m -3 ); TEFi and MEFi are the toxic and mutagenic equivalence factors of the target PAH, respectively; and BaPTEQ and BaPMEQ are BaP equivalencies resulting from the product of the target PAH concentration and its individual TEF or MEF, respectively.The lifetime lung cancer risk (LLCR) associated with inhalation exposure to PAHs was estimated from Eq. ( 8) (Bari et al., 2011;Collins et al., 1998): where IURBaP is the inhalation unit risk, defined as 8.7  10 -5 per ng m -3 via inhalation exposure (WHO, 2000).

Mass Concentrations of PM2.5 and PAHs
During the sampling period, the PM2.5 concentration ranged from 2.28 to 40.94 µg m -3 with a mean of 16.56 ± 7.93 µg m -3 (Fig. 2), higher than the annual standard (15 µg m -3 ) of the Taiwan EPA (MOENV, 2020) and the annual guideline limit (5 µg m -3 ) of the WHO (WHO, 2022).Because of seasonal variations, the mean PM2.5 concentration was highest in spring (20.15 ± 7.73 µg m -3 ), followed by that in winter (18.73 ± 8.31 µg m -3 ), summer (13.87 ± 6.60 µg m -3 ), and autumn (13.72 ± 7.26 µg m -3 ).The highest concentration of PM2.5 in spring could be attributed to frequent transboundary air pollutants from the northeast monsoon and fewer rainy days in the spring, as compared with other seasons.The reason for the significant difference between the cold and warm seasons was probably due to several factors.For instance, the higher mixing layer in the warm period than in the cold period may have enhanced atmospheric dispersion, thus decreasing ambient PM2.5; the contribution of the long-range transport of the northeast monsoon (from October to April) to ambient aerosol concentration from the coastal cities in China and East Asia to northern Taiwan in the cold season could increase PM2.5 concentration (Hsu et al., 2019;Hung et al., 2019;Luo et al., 2018;Kishcha et al., 2018;Wang et al., 2016c).
The annual concentration of the total PAHs (TPAH) ranged from 0.11 to 3.27 ng m -3 with an average of 0.79 ± 0.67 ng m -3 , as shown in Table S2.The annual TPAH concentration was significantly lower in the study area than in Guangzhou, China (33.89 ng m -3 ) (Liu et al., 2015a); Tehran, Iran (12.2 ng m -3 ) (Taghvaee et al., 2018); Beijing, China (92.62 ng m -3 ) (Yu et al., 2018); and Islamabad, Pakistan (25.69 ng m -3 ) (Mehmood et al., 20200) but higher in the study area than in Kanazawa, Japan (0.53 ng m -3 ) (Kalisa et al., 2019) and Ciudad Real, Spain (0.54 ng m -3 in PM10-bound PAHs) (Lara et al., 2022).In comparison to previous studies on PAHs (3.84 ng m -3 ) conducted in Taipei (Hsu et al., 2019), this study exhibits a lower mean concentration of total PAHs.This can be attributed to improved air quality and a reduction in the number of observed PAH congeners.The seasonal variations of the annual TPAH concentration exhibited the following pattern: winter (1.36 ng m -3 ) > spring (0.86 ng m -3 ) > summer (0.60 ng m -3 ) > autumn (0.43 ng m -3 ), similar to those in Islamabad, Pakistan (Mehmood et al., 2020); Shizuoka, Japan (Ohura et al., 2004); and Ontario, Canada (Anastasopoulos et al., 2012).The average concentration of the PM2.5-boundTPAH in winter was approximately two and three times higher than that in summer and autumn, respectively, probably because of the effect of ambient temperature on gas-particle phase partitioning.The higher TPAH concentration in the cold seasons might be due to long-range transport from northeast China, where biomass/coal combustion for heating is prevalent in the cold seasons (Hsu et al., 2019;Lai, 2015), whereas the lower TPAH concentration in the warm seasons could be attributed to the increased volatilization of the PAHs (Chen et al., 2020;Vlachou et al., 2019;Tan et al., 2006).
In the sampling period, the concentrations of PM2.5-bound PAHs reduced in the given order: According to the annual and seasonal proportions of each PAH to TPAH in Fig. S1, BghiP accounted for the highest percentage (16-24%) of TPAH among the four seasons.The average concentrations of BghiP, BbF, IND, and CHR were the predominant PAHs, whereas Flu, Ant, AcPy, and Acp often showed a low average concentration in the four seasons.Previous studies have also reported that BghiP, BbF, IND, and CHR were the dominant PAH species in other urban areas, such as Bangkok, Beijing, and Shanghai (ChooChuay et al., 2020;Chao et al., 2019;Wang et al., 2016a), implying that these PAH species might be originated from the general sources in urban area.The further investigation of sources would be discussed in Section 3.2.
According to their molecular weights, PAHs were classified into the following two groups: lowmolecular-weight (LMW) (AcPy, Acp, Flu, PA, and Ant) and high-molecular-weight (HMW) (FL, Pyr, BaA, CHR, BbF, BkF, BaP, DBA, IND, and BghiP), as shown in Table S2.The annual HMW PAH concentration ranged from 0.07 to 3.00 ng m -3 with the highest and lowest mean values of 1.27 ± 0.67 ng m -3 and 0.38 ± 0.18 ng m -3 in winter and autumn, respectively.The annual LMW PAH concentration ranged from 0.03 to 0.27 ng m -3 with the highest and lowest mean values of 0.09 ± 0.07 ng m -3 and 0.04 ± 0.01 ng m -3 in winter and summer, respectively.The mean HMW PAH concentration was higher than the mean LMW PAH over the sampling period, probably owing to the low molecular weight and high volatility of the PM-bound PAHs, resulting in a tendency to exist in the gaseous phase.Both HMW and LMW PAHs showed the highest concentrations in winter, consistent with those in Thessaloniki, Greece (Manoli et al., 2016) and Isfahan, Iran (Soleimani et al., 2022).A significant difference in the HMW PAH concentration was found between winter and autumn, probably due to the influence of meteorological factors on the PM2.5-boundPAHs (Manoli et al., 2016).
According to the number of benzene rings, PAHs can be divided into 3 rings (i.e., AcPy, Acp, Flu, PA, and Ant), 4-rings (i.e., FL, Pyr, BaA, and CHR), 5-rings (i.e., BbF, BkF, BaP, and DBA), and 6-rings (i.e., IND and BghiP).As shown in Fig. 3, the 6-ring PAHs were the largest contributor (34%) to the annual average TPAH concentration, followed by 5-(29%), 4-(28%), and 3-ring PAHs (8%).This pattern was also found in spring, summer, and autumn but not in winter, which might be due to the chemical characteristics of the PAHs.Because the PAHs are semi-volatile organic compounds, temperature has been regarded as a critical regulator of the partitioning of the PAHs into the particle and gas phases (Chetwittayachan et al., 2002), in particular for the 4-and 5-ring PAHs.As shown in Table S2, the mean 4-ring PAH concentration in winter (0.44 ng m -3 ) was approximately five times that in summer (0.09 ng m -3 ), while the mean 5-ring PAH concentration in winter (0.40 ng m -3 ) was over three times that in summer (0.13 ng m -3 ).However, the average 6-ring PAH concentration in winter (0.38 ng m -3 ) was only approximately two times that in summer (0.18 ng m -3 ).Similar seasonal patterns of the PAHs have also been observed in Islamabad and Hsinchu (Mehmood et al., 2020;Yang et al., 2017), which might be due to the high volatility and vapor pressure of the LMW PAHs (Bandowe et al., 2014).

Positive matrix factorization (PMF)
The sources and their contribution to the individual PAHs identified via the PMF 5.0 model during the study period are illustrated in Fig. 4. The following four main sources contributing to TPAH were identified: evaporated/unburned oil (22.3%), industrial emissions (29.2%), traffic emissions (31.3%), and biomass/coal combustion (17.1%).
Factor 1 was identified as evaporated/unburned oil with the high levels of Ant (84%), Acpy (60%), Acp (61%), and Flu (52%).Previous studies have regarded these PAHs as the typical markers for the volatilization of petroleum and crude oil (Wang et al., 2016b;Liu et al., 2015a;Marr et al., 1999).The LMW PAHs, such as Ant, Flu, and PA, appeared to originate from petroleum spills (Hu et al., 2013).As shown in Fig. 5, the proportion of TPAH in factor 1 was notably higher in summer (30.9%) and autumn (44.4%) than in spring (15.5%) and winter (17.4%), which could be attributed to the effect of the high ambient temperature (Tomaz et al., 2016;Van Drooge et al., 2010).Taghvaee et al. ( 2018) indicated that gas stations could release the LMW PAHs, which could be a major source of evaporated oil in urban areas.In addition, heavy oil used for shipping has been suggested as one of the emission sources of Ant (Sugiyama et al., 2017;Cooper, 2003).AcPy and Acp serve as markers for the volatilization of crude oil or oil-contaminated surface water (Bamford et al., 1999;Marr et al., 1999).Flu and BghiP also act as indicators for oil spills or evaporation (Elzein et al., 2020).Furthermore, Tai and Chang (2022) pointed out that emissions from shipping activities can impact air quality in Taipei.
Factor 2 was attributed to industrial emissions with the relatively high proportions of PA (67%), FL (69%), and Pyr (70%), which might be produced by the coke-making process (Ciaparra et al., 2009).The HMW PAHs, such as CHR (41%), BbF, and BkF, can arise from the sintering process (Ciaparra et al., 2009;Lin et al., 2011).Factor 2 was the highest contributor to TPAH in winter (38.5%), which might be due to the emissions from the local steel and iron industries around northern Taiwan (Chen et al., 2016;Dat et al., 2018) and the long-range transport in the monsoon season (Li et al., 2017;Lai, 2015).This factor appeared to be determined by industrial activities from Hebei, Shandong, and other northeast China provinces and coastal cities and transported to northern Taiwan by the northeast monsoon in the cold seasons (Song et al., 2015;Han et al., 2019b).
Factor 3 was identified as traffic emissions mainly composed of the HMW PAHs, including BbF (31%), BkF (31%), BaP (47%), IND (42%), DBA (74%), and BghiP (48%).This factor contributed to the highest proportion of TPAH (31.3%), as was also reported for Taiwan by previous studies (Hsu et al., 2019;Chen et al., 2016).The HMW PAHs were previously reported to be mainly produced by the high-temperature combustion processes, such as emissions of gasoline-powered vehicles (Duval and Friedlander, 1981;Dachs et al., 2002).In addition, BghiP and IND have been recognized as indicators of vehicle exhaust and gasoline emissions (Miguel et al., 1998;Duval and Friedlander, 1981), while BaP, IND, and BghiP have been shown to be strongly correlated with traffic emissions in East Asia (Chao et al., 2019;Dat et al., 2018).This factor contributed a higher proportion of TPAH in spring (39.0%) and summer (33.3%) than did the other factors in the study period.Traffic emissions were reported as the dominant sources of PM2.5 in northern Taipei (Ho et al., 2018;Liang et al., 2013).Previous studies in Taipei also denoted that vehicular emission was the major source of ambient PM2.5-bound PAHs (Hsu et al., 2019;Room et al., 2023).
Factor 4 was considered biomass/coal combustion with high BaA (78%) and several species of HMW PAHs, such as BkF (32%) and BaP (35%).Ravindra et al. (2008), Caricchia et al. (1999), Mai et al. (2003), and Larsen and Baker (2003) have indicated BbF, CHR, and BaP as the markers of biomass/coal combustion.The contribution of factor 4 to TPAH was higher in winter (18.3%) and spring (18.2%) than in summer (5.1%) and autumn (9.6%), which might be due to the emissions of coal-fired power plants in Linkou or long-range transport of domestic heating emissions in China (Li et al., 2021;Horak et al., 2017).It was also reported that coal combustion was one of the major sources of PM2.5-bound PAHs in the Taipei urban area (Hsu et al., 2019;Room et al., 2023).

Diagnostic ratio
To further distinguish the potential sources of the PAHs, the diagnostic ratios of individual PAHs, such as Ant/Ant+PA, FL/FL+Pyr, BaA/BaA+CHR, and IND/IND+BghiP, were used as indicators for source identification (Gschwend and Hites, 1981;Budzinski et al., 1997;Sicre et al., 1987;Soclo et al., 2000;Yunker et al., 2002), as shown in Fig. 6.The average ratios for the study period are listed in Table S3.The ratio of Ant/Ant+PA indicated a pyrogenic source when > 0.10 and a combustion source when < 0.10.This ratio in this study was greater than 0.10 in all the seasons  (0.16, 0.23, 0.40, and 0.14, respectively), indicating that biomass combustion emissions might be one of the main PAH sources.In fact, the PAHs could result from a mixture of combustion and petroleum emissions, as observed in Rome (Finardi et al., 2017), Shanghai (Yang et al., 2021), and Ciudad Real (Lara et al., 2022).The ratio of FL/FL+Pyr pointed to the petroleum source when < 0.40 (Yunker et al., 2002;Ravindra et al., 2006), to diesel and gasoline emissions when between 0.40 and 0.50, and to biomass combustion when > 0.50 (De La Torre-Roche et al., 2009).In this study, this ratio varied between 0.40 and 0.50 in spring (0.46) and summer (0.49), but was above 0.50 in autumn (0.60) and winter (0.52), indicating that biomass burning as well as diesel and gasoline emissions might be the possible sources, as was also found in the urban areas by previous studies (Varea et al., 2011;ChooChuay et al., 2020).The ratio of BaA/BaA+CHR indicated a petroleum source when < 0.20, petroleum combustion (e.g., crude oil and liquid fossil fuel combustions) when between 0.20 and 0.35, and diesel traffic emissions when > 0.35 (Yunker et al., 2002).This ratio in all the four seasons in the present study varied between 0.20 and 0.35 (0.25, 0.24, 0.21, and 0.20, respectively), thus indicating the importance of petroleum combustion emission to the air quality in Taipei.The ratio of IND/IND+BghiP refers to a petrogenic source when < 0.20, to biomass/coal combustions when > 0.50 (Kavouras et al., 2001), and to petroleum combustion emission when between 0.20 and 0.50.In this study, the majority of this ratio in the four seasons varied between 0.20 and 0.50, which verified the significant contribution of petroleum emissions, as was also found in the urban areas by previous studies (Soleimani et al., 2022;Chao et al., 2019).Thus, liquid fossil fuels, diesel emissions, and petroleum combustion were the main sources of the PM2.5-boundPAHs according to the diagnostic ratio results.The contributions of biomass/coal combustion were relatively low but should not be ignored, in particular, in spring (Lin et al., 2014).
The diagnostic ratios for four sources identified by PMF were shown in Table S3.FL/FL+Pyr for biomass/coal combustion was over 0.5, and BaA/BaA+CHR for biomass/coal combustion was above 0.35.Ind/Ind+BghiP for traffic emission was between 0.2 and 0.5.These results reinforced the reliability of source apportioned by PMF.

PSCF analysis
To better understand the PMF-derived emission sources in Section 3.2.1,PSCF analysis was used to further clarify the regional contributions.Fig. 7 shows the regional contributions of each source to TPAH concentration in the study period.Compared with the other factors, evaporated/unburned oil in Fig. 7(a) exhibited a high PSCF value distributed mainly over the Yellow Sea and East China Sea, which might originate from shipping and harbor activities in the region (Bie et al., 2021;Yu et al., 2021).Evaporated or unburned oil might arise from petroleum spills, as discussed in Section 3.2.1.The origin of industrial emissions in Fig. 7(b) was concentrated in northeast China, a highly industrialized region (Zhou and Yang, 2016;Song et al., 2015).The contributor to TPAH in this area could be transportation from Hebei, Shandong, and other northeast China provinces and coastal cities (Dat et al., 2018;Han et al., 2019b), as was discussed in Section 3.2.1.Fig. 7(c) shows that the distribution of traffic emissions was more divergent than that of the other factors and was mostly from local and coastal urban cities near northern Taiwan, such as Taipei, Hangzhou,Shanghai,and Nanjing. Fig. 7(d) shows that northeast China most contributed to biomass/coal combustion, as was also indicated in previous studies (Zhang et al., 2007;Shen et al., 2013).
Emissions of industrial activities and biomass/coal combustion were higher from the northeastern and coastal cities of China than from the local area.However, traffic emissions as the highest contributor originated from the local area.To further distinguish the contributions of different vehicles, an emission inventory should be established in the future.On the other hand, traffic emissions were consistently centered around northern Taiwan throughout all four seasons, suggesting that local vehicular emissions play a significant role in this source.

Health Risk Assessment
To assess the potential hazards of the PAHs for human health, the TEQ, MEQ, and LLCR values of the PAHs were evaluated, as shown in Table S4.The values of BaPTEQ and BaPMEQ ranged from 0.02 to 0.45 ng m -3 and 0.02 to 0.50 ng m -3 with an annual average of 0.09 ng m -3 and 0.16 ng m -3 , respectively.The BaPTEQ and BaPMEQ values in this study were notably lower than the annual target value of 1 ng m -3 recommended by the EU and those in Thessaloniki, Greece (Manoli et al., 2016); Beijing, China (Yu et al., 2018); and Tak, Thailand (Janta et al., 2020).However, the mean BaPMEQ concentration in this study exceeded the annual reference level of the WHO (0.12 ng m -3 ).Throughout the sampling periods, a PM2.5 pollution episode (PM2.5 > 35 µg m -3 ) occurred, during which the BaPTEQ value was nearly three times higher than the annual average.This increase was attributed to a heightened contribution from traffic emissions.The seasonal variations of BaPTEQ and BaPMEQ were consistent with those of the TPAH concentration and higher in winter (0.15 and 0.25, respectively) and lower in autumn (0.04 and 0.08, respectively).Although the mean annual LLCR value (1.0  10 -5 ) was lower in the study area than in other urban studies (Yu et al., 2018;Soleimani et al., 2022), it remains essential to quantify the source-specific cancer risks for developing effective mitigation strategies in the future.
As shown in Table S5, BaP was the dominant contributor among the PAHs, responsible for 53.63% of the total BaPTEQ level in the four seasons, consistent with the results of previous studies (Gao et al., 2015;Yu et al., 2018).In addition, BbF, BaP, IND, and DBA contributed 11.58%, 10.71%, and 13.67%, respectively, to the total BaPTEQ concentration.The sum of these HMW PAHs amounted to more than 99% of the total BaPTEQ value, similar to particulate PAHs in the urban areas reported by previous studies (Teixeira et al., 2012;Chen et al., 2021).The HMW PAHs, in particular, BaP, a reliable tracer of the toxic PAHs, exhibited higher TEF values and concentrations than did the LMW PAHs (Belis et al., 2011).The BaPMEQ value of the different species, as shown in Table S6, was mainly dominated by BaP (31.75%),BghiP (22.14%),DBA (20.16%), and BbF (19.09%).There was no significant difference in the individual PAHs among the seasons; however, differences in the sources of the pollutants were detected.Based on the sources identified in Section 3.2.1, the HMW PAHs mainly originated from traffic emissions and biomass/coal combustion, thus pointing to them as the main contributors to human health risks.In addition, the sources contributing to the annual and seasonal BaPTEQ and BaPMEQ values, as shown in Figs. 8 and S2, respectively, demonstrated that traffic emissions and biomass/coal combustion were the dominant contributors throughout the year.Previous studies in urban areas have shown consistent results, exhibiting the significant adverse effects of these two sources on human health (Taghvaee et al., 2018;Han et al., 2019a;Zhang et al., 2019).From the perspective of the seasonal variations, the contributions of traffic emission and biomass/coal combustion exhibited a significant difference between the cold and warm seasons, whereas those of evaporated/unburned oil did not show a significant difference.Not only could these source-specific health risk results be used to quantify the relationship between the sources and their impacts on human health, but they might facilitate in predicting human health risk in different areas based on the source profiles.

CONCLUSIONS
The annual average concentrations of PM2.5 and TPAH in Taipei city were 16.56 ± 7.93 µg m -3 and 0.79 ± 0.67 ng m -3 , respectively.Both PM2.5 and TPAH concentrations showed the same seasonal variation, with higher values noted in the cold season and lower values in the warm season.A significant difference between the cold and warm seasons indicated that the seasonally varying emission sources of PM2.5 and TPAH concentrations need to be further studied.Overall, the mean concentration of the HMW PAHs was notably higher than that of the LMW PAHs both annually and seasonally.Benzo[ghi]perylene (BghiP) accounted for the highest percentage of TPAH among all the HMW PAHs in all the four seasons.The traffic emission was estimated to contribute approximately 31.3% to the annual TPAH concentration, indicating the importance of controls over the traffic-related sources.Diagnostic analysis also showed that petroleum and fuel combustions were the major sources of the PM-bound PAHs.Apart from the local sources, the PAHs appeared to be partially controlled by the long-range transport of emissions from industrial activities in northeast China and shipping around the Yellow Sea and East China Sea.Unlike previous studies in Taipei urban area, we concluded traffic emissions were the dominant contributors to higher cancer risks compared to the other three sources through the assessment of source-oriented health risks.The seasonally varying contributions of traffic emissions and

Fig. 3 .
Fig. 3. Annual and seasonal contributions of the PAHs with different numbers of benzene rings to TPAH concentration.

Fig. 4 .
Fig. 4. Profiles of the four sources (in percentage) identified by the PMF 5.0 model for 15 individual PAHs over the year.

Fig. 5 .
Fig. 5. Box plots of the contributions of the four sources to TPAH concentration in the four seasons, respectively.The boxes enclose the 25 th , 50 th , and 75 th percentiles and the whiskers represent the 5 th and 95 th percentiles, respectively.
Fig. S3 illustrates the distribution of industrial emissions during the cold season (Figs.S3(a) and S3(d)), primarily concentrated around coastal and eastern cities in China.Conversely, in the warm season (Figs.S3(b) and S3(c)), dominant sources were in proximity to Taiwan, signifying a contribution of local industrial emissions.

Fig. 7 .
Fig. 7. Annual PSCF values and spatial distributions for (a) evaporated/unburned oil, (b) industrial emission, (c) traffic emission, and (d) biomass/coal combustion.The green star marks the location of the sampling area.

Fig. 8 .
Fig. 8. (a) The annual concentrations of the individual PAHs contributing to BaPTEQ, and (b) the annual and seasonal contributions of BaPTEQ and lifetime lung cancer risk (LLCR) according to the four main sources.