Distribution and Sources of Atmospheric Polycyclic Aromatic Hydrocarbons at an Industrial Region in Kaohsiung , Taiwan

The chemical mass balance model was applied to estimate the major sources of atmospheric polycyclic aromatic hydrocarbons (PAHs) at an Industrial Region in Kaohsiung, Taiwan. The gaseous and particulate phases of 16 individual compounds were analyzed between March 2012 and August 2012. The mean total concentrations and total BaPeq were higher during the cold season and lower during the warm summer, with gaseous PAHs predominant at all sites. Low weight-PAHs and median weight-PAHs were found predominantly in the gaseous phase, while high weight-PAHs were predominant in the particle phase. Results from the receptor model revealed that the average contributions were 38.2%, 27.2%, 20.7%, 6.8%, 5.2%, and 2.0% from vehicles, heavy oil combustion, natural gas combustion, incinerator, tetrabromobisphenol A production, and diesel combustion at the seven receptors, respectively. Vehicle emissions appear to be the significant source of PAHs in the investigated area, although other industrial sources, as described above, also have an impact on the total PAHs.


INTRODUCTION
Polycyclic aromatic hydrocarbons (PAHs) consist of two or more benzene rings are generated from anthropogenic activities such as incomplete combustion and/or pyrosynthesis of organic compounds, and partly from natural combustion (Lai et al., 2007;Wu et al., 2014;Cheruiyot et al., 2015;Tiwari et al., 2015).The major sources of PAHs are heating (coal, oil, gas and wood), petroleum refinery, coke production, fossil fuel combustion, industrial processes, steel and iron furnaces, incinerators, and vehicles (Chen et al., 2014;Li et al., 2014b;Huang et al., 2015;Mwangi et al., 2015a;b;Qin et al., 2015).Since several PAHs have been classified as human carcinogens (groups 2B, 2A and 1) by the International Agency for Research on Cancer (IARC, 1987;Kamal et al., 2016), public concern with the release of these carcinogenic pollutants in the global environment has recently increased.
PAHs are widely distributed in the atmosphere, and the can then be transported over long distances before atmospheric dry and wet deposition onto soils, vegetation, and waters.Atmospheric PAHs can occur between particle and gas phases, depending on the ambient temperature, properties of PAH compounds, the interactions between PAH compounds and aerosols, and the fate of PAHs in the environment (Pankow, 1994).It is well known that both particle and gas phases of PAHs are regarded as significant hazards to human health through inhalation and/or ingestion (Chen et al., 2009;Cai et al., 2014).As such, investigating the occurrence and contributions of the various sources of atmospheric PAHs is important for reducing human exposure to these toxic chemicals.
The PAH profiles of individual emission sources have successfully been applied as tracers to evaluate the relationship between emissions of PAHs and ambient PAH concentrations (Li et al., 2003).As a tool for source apportionment, the chemical mass balance (CMB) model has been developed to identify the possible PAHs sources in the air, soil, and sediment (Li et al., 2003;Yang and Chen, 2004;Hanedar et al., 2011;Chen et al., 2012;Jung et al., 2015).Few studies have applied CMB to quantify contributions of ambient PAHs in Taiwan, showing that traffic and industrial combustions were the dominant sources for PAH emissions (Yang and Chen, 2004;Chen et al., 2012).In the CMB calculation, the source fingerprints and ambient measurements are needed for the input data (Watson et al., 1984).The contributions of sources can then be estimated by determining the best-fit combination of the chemical profiles of emission sources and chemical composition of the ambient samples (Pipalatkar et al., 2014).
Kaohsiung City, located in southern Taiwan, is a highly industrialized and heavily densely populated city with more than 2.7 million residents and 2.8 million motor vehicles.The city is surrounded by six large industrial complexes, the major industrials are steel plants, oil refineries, power plants, cement kilns, plastic and chemical factories, metalmaking plants, petrochemical plants, and incinerators.The air quality in Kaohsiung City is the worst in Taiwan due to emissions from these industrial complexes and traffic.Moreover, the unfavorable meteorological conditions (e.g., poor atmospheric mixing, declining visibility, and the sealand breeze circulation) also have a significant impact on the air quality (Wang et al., 2010).Although several studies have reported the spatial distribution and source apportionment of PAHs in urban areas, industrial areas, and sediments in Taiwan (Fang et al., 2004;Yang and Chen, 2004;Chen et al., 2009;Chen et al., 2012), those of atmospheric PAHs from petrochemical industrial complexes have not yet been adequately characterized.
The main objectives of this study were thus to characterize the 16 PAH species in both the gas and particulate phases in a petrochemical area in Kaohsiung, Taiwan, and to assess the source contributions to ambient PAH concentrations using a receptor model.

Ambient Sampling Sites
The seven sampling sites are located in northern Kaohsiung City in Taiwan, which is surrounded by Kaohsiung Oil Refinery and two major petrochemical industrial complexes of Dashe and Renwu (Fig. 1).Dashe industrial complex mainly contains petrochemical factories, while Renwu complex is a mix of petrochemical, chemical, and plastic factories.Site A is located on the northwestern side about 1 km from the Dashe Industrial Park, and 0.5 km from the west side of the Highway No. 1. Site B is located to the north (about 0.4 km) of Dashe Industrial Park, while site in C is located east of Dashe Industrial Park and 0.5 km from Highway No.10.Site D is located east (about 0.2 km) of Renwu Industrial Park and 0.5 km from Highway No.10.Site E is located between Dashe and Renwu Industrial Park and 1km east of Kaohsiung Oil Refinery.Site F is located south of Renwu Industrial Park and 0.5 km from Highway No.1.Site G is located south of Renwu Industrial Park, which is next to Highway No.1, and hence is subject to relatively high vehicle emissions.

Sampling and Analysis
Ambient air samples were obtained in March/April 2012 and July/August 2012 for PAH analyses in the spring and summer seasons, respectively.The meteorological information of the sampling sites obtained from the Taiwan Central Weather Bureau is summarized in Table 1.
A total of fourteen ambient air samples for the gas and particulate phases of PAHs were collected for 48 hours using a PS-1 sampler (Graseby Andersen, GA).The particleand gas-phase compounds were collected on glass fiber filters and glass PUF cartridges by PS-1 samplers, respectively.All the sampling devices were installed on the rooftops (about 6-10 m high) of elementary schools.The glass fiber filter was first put in a desiccator for 24 h.Recoveries were calculated using the known amount of surrogate standards.One field blank was completed to ensure that the collected samples were free of contamination.All particulate and gaseous samples transported to and from the sampling field were covered with aluminum foil to avoid photo dissociation.
Analyses of all samples were conducted by the Super Micro Mass Research and Technology Center in Cheng Shiu University.Each collected sample was extracted in a Soxhlet extractor with a v:v = 1:1 of n-hexane and dichloromethane mixture solution (total 700 mL) for 24 h.The extract was then concentrated, cleaned-up and reconcentrated to exactly 1 mL (Lai et al., 2007).A gas chromatograph (Agilent 6890 N) with a capillary column (HP Ultra 2-50 m × 0.32 mm × 0.17 µm) coupled to a mass selective detector (Agilent 5973 N) was used for the PAH analysis.The analytical recoveries and MDL of 16 individual PAH compounds are given in Table 2, with the full and abbreviated names of 16 PAHs.

Chemical Mass Balance
The Chemical Mass Balance (CMB) Model version 8.2 (US EPA, 2004) was adopted in this study to determine the source apportionment of atmospheric PAHs.The source profiles and ambient concentrations, with estimates of the uncertainty of both sets of data, serve as input for the CMB modelling (Watson et al., 2001).The output consists of a least squares solution to a set of mass balance equations that presents each PAH concentration at a receptor site as a linear sum of products of sources profiles and the associated concentrations.
An uncertainty of 20% was adopted for all ambient and source measurements of individual PAH compounds during modeling.To remove negative source contribution estimates from the CMB results, the "source elimination" option of CMB8.2 software was used in this study.Five performance measures were suggested by the EPA, including R-squared (R 2 ≥ 0.8), Chi-squared (χ 2 ≤ 4.0), mass% (percentage of mass accounted for 80-120%), the degrees of freedom (DF > 5), and the ratio of residual to its uncertainty (-2 < R/U < 2) (Watson et al., 2004).

Source Fingerprints
Fingerprints are the species composition of emissions from a source category, expressed as the mass fraction of total PAH emissions.The source fingerprints of PAHs play a key role in the CMB modelling.The PAHs source profiles used by the CMB model in this study are detailed in Fig. 2. Profiles for heavy oil combustion, natural gas combustion, incinerator, coal combustion, and vehicle traffic (diesel and   (Li et al., 1999;Mi et al., 2001;Yang and Chen, 2004;Wang et al., 2007).Those for tetrabromobisphenol A (TBBPA) production and diesel boilers were established by sampling and analyses in this study.Since PAHs are distributed between the gas and particulate phases in the atmosphere, the total (gaseous and particulate phase) PAHs were used in model fitting for better results.
As shown in Fig. 2, the source profiles used for this study were normalized to 1 for the 15 fitted species.Nap is excluded in the modelling, because of its significantly higher concentration than that of the other PAHs for many emission sources.Moreover, Nap is bicyclic and sometimes regarded as a volatile organic compound (Yang and Chen, 2004).

Atmospheric PAHs Concentrations
A total of 14 samples were collected and 16 US EPA priority PAH species were identified.The total-PAH (gas phase + particulate phase) concentration was the sum of the 16 individual PAH concentrations in each sample (Table 3).Total ∑16-PAHs concentrations (particle + gas) in the spring ranged from 101 to 233 ng m -3 , with an average of 167 ng m -3 .In summer, the total PAHs were between 69.3 and 325 ng m -3 , with an average of 149 ng m -3 .The concentrations of total PAHs generally followed the seasonal variations of TSP (121 µg m -3 and 59.6 µg m -3 in spring and summer, respectively), consistent with a previous study which showed that worse air quality during the winter or early spring was to be expected (Chen et al., 2009).A comparison was made between total PAHs in spring and summer using the t-test.The results revealed that the total PAHs detected during the two seasons were not significantly different (P > 0.05).The prevailing wind directions for spring and summer were NW and N, respectively (Table 1).The highest concentration of total PAHs was found in downwind sites D (in spring) and F (in summer), while the lowest level was found in upwind sites A (in spring) and C (in summer).This difference is probably due to the prevailing wind direction, physical dispersion/transportation of pollutants, and the emission intensities from PAHs emission sources during the sampling period.
Comparisons of the total PAHs concentrations found in the atmosphere with the results from other studies are given in Table 4.The mean concentrations of PAHs in this study (158 ng m -3 ) were higher than those for urban areas reported in Kaohsiung City, Taiwan (Chen et al., 2009), but were significantly lower than that have been reported in industrial/urban/rural sites (Fang et al., 2004) and a suburban area (Yang and Chen, 2004), both in Taichung, Taiwan.Since traffic sources are a significant contributor to PAHs emissions, the significantly higher PAHs concentrations of up to 1672 ng m -3 were obtained from Beijing (China) during summer (Liu et al., 2007) and Bursa (Turkey) in previous studies (Tasdemir and Esen, 2007;Esen et al., 2008).This reveals that atmospheric PAHs could be significantly affected by the location of the sampling site and its proximity to emission sources.

Characteristics of PAHs
Sixteen individual PAHs were divided into three categories to estimate the PAH homologue distribution, including six low molecular weight PAHs (LM-PAHs, 2-3-ring), four median molecular weight PAHs (MM-PAHs, 4-ring), and   This study five high molecular weight PAHs (HM-PAHs, 5-7-ring).Compared with the three categories of PAHs, the ƩLM-PAHs exhibits the highest partition (> 93%) in the total PAHs for all sampling sites (Table 3).The concentrations of Nap were at the highest levels (up to 304 ng m -3 determined at site F in summer) of all the 16 PAHs in all samples, followed by PA (4.33-10.5 m -3 ) and Flu (1.33-5.23 m -3 ).Results are similar to previous finding showed that atmospheric NaP was dominated over PAHs, ranged from 184 to 412 ng m -3 (Chen and Yang, 2004;Fang et al., 2004;Tasdemir and Esen, 2007).Moreover, it has been found that approximately 50% of the global atmospheric PAH emissions were NAP, the source of Nap was mainly from fuel combustion, biomass burning, and consumer products (Zhang and Tao, 2004).The low molecular weight PAHs, particularly NAP, primarily occurred in the gaseous phase.Significantly lower concentrations were measured for HM-PAHs (five or more aromatic rings).
Due to the highly carcinogenic property of BaP, it has been widely studied as an indicator for total PAHs concentrations in the atmosphere.Concentrations of BaP ranged from 0.147 to 0.419 ng m -3 and 0.007 to 0.023 ng m -3 in spring and summer, respectively.The carcinogenic activity of total PAHs was then calculated with the toxic equivalent factor (TEF) of individual PAHs and their concentrations.The results showed that the total BaP equivalent concentrations (i.e., total BaP eq ) for carcinogenic PAHs were higher in spring (0.469-1.51 ng m -3 ) than in summer (0.084-0.344 ng m -3 ).

Phase Distribution of PAHs
When PAHs are emitted into the air they can be partitioned between the gas and particulate phases based on the chemical/physical properties of PAH species and meteorological conditions.The distributions of PAHs in the gas and particulate phases at seven sampling sites in both seasons are shown in Fig. 3.At all sites, LM-PAHs (2-3 rings) were the most abundant in the gaseous phase during spring and summer; notably, Nap accounted for more than 80% in the gaseous phase.PAHs with higher molecular weights (> 4 rings) were mainly adsorbed in the particulate phase.Additionally, the amount off PAHs with 2-3 rings bound to particles increased with increasing the temperature from 20.9°C and 29.4°C, and higher contributions of LM-PAHs (2-3 rings) to total PAHs were found in summer than those in spring.Therefore, HM-PAHs with stronger carcinogenic properties could be more effectively removed by dry deposition than LM and MM-PAHs.
Table 5 summarizes the mean distributions of gaseous and particulate PAHs in both seasons.The LM-PAHs (about 97.9% in spring and 99.5% in summer) and MM-PAHs (about 64.6% in spring and 83.4% in summer) were found   (Chen et al., 2009).

Results of CMB Modeling
The results of the CMB model calculations were examined by several performance parameters, such as R 2 , χ 2 , percent mass, and degrees of freedom.An estimate was considered acceptable if the value met the criteria specified by the US EPA (USEPA, 2004).The average R 2 , χ 2 , percent mass, and degrees of freedom of all sampling sites for spring and summer were 0.81 and 0.76, 3.4 and 3.3, 89.4% and 80.6%, and 10 and 11, respectively (Table 6).This reveals that there were two samples with slightly lower R 2 (< 0.8) but with sufficient mass explained (> 80%), as calculated for summer.The values of χ 2 were all within target ranges (≤ 4.0), indicating that the source apportionment with the source fingerprints has a good fit with the measurements.It is noted that the best performance parameters were those for spring.
Mobile sources appeared to be the major contributors at all sampling sites, and a slightly higher contribution from diesel engines than that of gasoline engines was obtained.Due to the presence of two heavy traffic highways around the sampling sites, traffic emissions are expected to be a significant source of atmospheric PAHs in the investigated area.The finding is consistent with the studies found in the literature, showing that traffic and industrial combustions were the dominant sources for PAH emissions (Yang and Chen, 2004;Chen et al., 2009;Hanedar et al., 2011).Moreover, heavy oil combustion was the second most important contributor.The findings from our study are similar to those of Yang et al. (1998) and Li et al. (1999), showing a significantly higher emissions factor of total PAHs for the heavy oil boiler among the various industrial facilities (blast furnace, basic oxygen furnace, coke oven, electric arc furnace, heavy oil plant, power plant, and cement plant).
Since PAHs and PCDD/Fs are semi-volatile organic compounds with comparable properties, both pollutants may originate from similar emission sources such as waste incineration, metallurgical processes, power plants, biomass burning, and vehicle exhaust (Cheruiyot et al., 2015).With the installation of advanced pollution control techniques and enactment of stringent regulations on emissions of stationary sources, public concerns on the possible adverse effects of exposure from non-industrial sources (e.g., biomass   open burning, domestic heating, and vehicles) were raised recently (Cheruiyot et al., 2016).Although emissions from these sources are lower than stationary sources, their proximity to human living environment is considered significant.Therefore, more research is needed on PAHs emissions from these non-industrial sources such as vehicle exhausts, cooking, joss paper burning, and biomass open burning to protect human health.

CONCLUSIONS
Atmospheric PAHs collected at an industrial region in Kaohsiung City, Taiwan, were investigated and the source apportionment of atmospheric PAHs was estimated using the CMB receptor model.Most PAH species were LM-PAHs (> 93%), followed by MM-PAHs and HM-PAHs.Moreover, LM-PAHs and MM-PAHs were found predominantly in the gaseous phase; notably, Nap accounted for more than 80% in gaseous phase.As for HM-PAHs, they were found mainly in the particle phase.The results from the receptor model revealed that mobile sources (34%-43%) contributed the most to total PAH concentrations at all sampling sites, followed by heavy oil combustion (26%-29%) and natural gas combustion (20%-22%).Therefore, the results obtained from this study are very useful for the source control and management of PAHs.

Fig. 1 .
Fig. 1.Locations of the seven sampling sites in Kaohsiung City.
Comparison of total PAHs concentrations in ambient air with literature studies.

Fig. 3 .
Fig. 3. Distribution of PAHs in the gas and particulate phases at seven sampling sites in two seasons.

Fig. 4 .
Fig. 4. Average source contributions for Ʃ15PAHs at seven sampling sites in two seasons.

Table 1 .
Meteorological information during the sampling period.

Table 2 .
Analytical recoveries and MDL of 16 individual PAH compounds.

Table 3 .
Concentrations of total PAHs in the ambient air of seven sampling sites (ng m -3

Table 5 .
Phase distribution (%) of 16 individual PAHs in two seasons.

Table 6 .
CMB target values and average modeling results for two seasons.