Atmospheric Dry Deposition of Polychlorinated Dibenzo-p-Dioxins / Dibenzofurans ( PCDD / Fs ) and Polychlorinated Biphenyls ( PCBs ) in Southern Taiwan

Dry deposition is one of the major routes by which air pollutants enter the ecosystem, and thus this study investigated the dry deposition characteristics of polychlorinated dibenzo-p-dioxins/dibenzofurans (PCDD/Fs) and biphenyls (PCBs) in the ambient air of industrial, urban, and rural areas in southern Taiwan from November 2010 to May 2011. Average dry deposition fluxes of total PCDD/Fs and PCBs in the ambient air of four sites were 157–544 pg/m-day (8.30–27.5 pg ITEQ/m-day) and 289–1010 pg/m-day (0.540–1.94 pg WHO-TEQ/m-day), respectively. The results showed that particle phase depositions dominated the dry deposition processes for the removal of PCDD/Fs and PCBs from the atmosphere, and the atmospheric deposition flux in the cold season tended to be higher than that during the warm season. The dry deposition velocity of individual PCDD/Fs (0.031–0.546 cm/s) increased as the number of chlorinated substitutes increased, which were similar to those measured or predicted in other Asian countries. Similar patterns of dry deposition velocities were observed for individual PCBs (0.069–3.38 cm/s), due to the fact that low chlorinated PCBs are predominant in gas phase and have lower deposition velocities.


Polychlorinated
dibenzo-p-dioxins/dibenzofurans (PCDD/Fs) and biphenyls (PCBs) are persistent, lipophilic, and bioaccumulative semivolatile organic compounds (SOCs) that are classified as hazardous air pollutants (HAPs) by the USEPA.Since PCBs are chemically similar to PCDD/Fs, the formation mechanisms of PCBs are similar to those of PCDD/Fs via de novo synthesis and/or the thermolysis of precursor compounds which interacted with metallic catalysts (Chin et al., 2012;Wang et al., 2012).The main sources of PCDD/Fs and PCBs were human activities.PCDD/Fs have been detected in the emissions from waste combustion, chemical plants, thermal sources, metal smelting processes, and vehicles (Chung et al., 2010;Chen et al., 2011;Chiu et al., 2011;Kao et al., 2011;Kuo et al., 2011).As for PCBs, they have been released into the environment by primary (e.g., vaporization or burning of products containing PCBs) and secondary emission sources (e.g., revolatilization from soil, vegetation, water, and air) (Breivik et al., 2002;Yeo et al., 2004).
The atmospheric deposition is an important transport source of SOCs for entering terrestrial and aquatic environments.Deposition of SOCs in air can be divided into dry and wet deposition; both processes contribute significantly to the removal of these SOCs (Jurado et al., 2004).When SOCs are emitted into the atmosphere, they can be partitioned between the gas and particulate phases based on their concentrations, vapor pressures, the atmospheric temperature, and the concentration of particles within it (Hoff et al., 1996).Previous studies have shown that gaseous SOCs are depleted through photochemical degradation, while particle-bound pollutants account for the major source of the atmospheric fluxes into the ecosystem (Lohmann and Jones, 1998;Welsch-Pausch and McLachlan, 1998).Since the dry deposition is contributed significantly to the environment, knowledge of the characteristics of PCDD/Fs and PCBs is essential for observing fate of these airborne toxics.
Dry deposition fluxes of PCDD/Fs and PCBs have been investigated by direct measurement or modeled estimates by multiplication of atmospheric concentrations with dry deposition velocities.Shih et al. (2006) measured the atmospheric dry deposition flux of total PCDD/Fs in a rural area of Taiwan using the smooth plate with a sharp leading edge, and the averaged flux of total PCDD/Fs was 50 pg/m 2 -day.The impact of dry/wet PCDD/F depositions from the ambient air and the electric arc furnace dust treatment plant in Taiwan has been presented (Lee et al., 2009).The estimated dry deposition flux of PCDD/Fs was approximately 4 times the wet deposition.Lee et al. (1996b) reported the modeled and measured dry deposition flux of PCBs in an urban site of Taiwan.It was found that all the modeled and measures dry deposition data were within 41% of difference, revealing a good prediction of the adopted models in the authors study.Additionally, previous studies have shown that deposition velocities vary depending on the source type, particle size distribution, and micrometeorological conditions (Tasdemir and Holsen, 2005).Due to the fact that the direct measurement of dry deposition velocity is difficult and expected uncertainties associated with model estimation, little information is available on atmospheric deposition of PCDD/Fs and PCBs.
To address this gap in the literature, the characteristics of PCDD/Fs and PCBs in the ambient air of industrial, urban, and rural areas in southern Taiwan were investigated seasonally.Comparison between measurement and simulation results of the gas-particle partitioning of PCDD/Fs and PCBs were obtained.In addition, dry deposition fluxes and velocities of PCDD/Fs and PCBs were then determined by model calculations.

Ambient Sampling Sites
Sampling was performed at four areas, including one industrial area, two urban areas, and one rural area located in Tainan city, Taiwan (Fig. 1).Sampling sites around these areas were allocated by the results of the Industrial Source Short Term model (ISCST), and six samples in each area were taken during November 2010 and May 2011.Site I is on the rooftop of a senior high school located in an industrial area and is on northeastern side about 1 km from the Southern Taiwan Science Park (STSP).The park covers an area of about 1043 hectares and attracts companies producing semiconductors, wireless telecommunications, computer, micro electronic precision machinery, optoelectronics, and agricultural biotechnology products (STSP, 2012).Site A is on the rooftop of university located in an urban area and is close to the Freeway No. 1 (Sun Yat-sen highway, about 0.5 km west).Site B is on the east side about 2 km from the Tainan train station, and is on the rooftop of an elementary school located in an urban area.It is approximately 2 m from the west side of the Freeway No. 1. Site R is on the grassland of elementary school located in a rural area and is on eastern side about 2 km from the highway (Formosa No. 3).

Sampling and Analysis
A total of twenty-four ambient air samples were collected by using a PS-1 sampler (Graseby Andersen, GA) according to the revised EPA Reference Method T09A.In first season (fall), each sample was collected continuously on seven consecutive days, yielding a sampling volume of about 4000 m 3 .However, sampling period was doubled to get the higher sampling volume (~9000 m 3 ) in second season (spring).The PS-1 sampler was equipped with a quartz fiber filter for sampling particle-phase compounds, and a glass cartridge that contained PUF for sampling gas-phase ones.A known amount of surrogate standard was spiked to check the collection efficiency of the sampling train before the sampling was conducted.To ensure that the collected samples were free of contamination, one field blank was completed.
All meteorological information for the sampling sites during the investigation periods was obtained from the Meteorological Bureau in Tainan City, as summarized in Table 1.The maximum and minimum temperatures at sampling areas were 26.8°C (in May) and 21.3°C (in November).The wind speed ranged from 1.91 to 3.17 m/s, with the highest value found in November (i.e., in the fall).The total TSP concentrations were found to vary in the range of 81-126 μg/m 3 during the sampling periods.
Analyses of ambient air samples were performed in the Super Micro Mass Research and Technology Center in Cheng Shiu University, certified by the Taiwan EPA for analyzing PCDD/Fs and PCBs.Each sample was spiked with a known standard and extracted for 24 h.Then, the extract was concentrated and treated with sulfuric acid, followed by a series of cleanup and fraction procedures.A high resolution gas chromatography with a mass spectrometer (HRGC/HRMS) was used to determine the concentrations of seventeen individual PCDD/Fs and twelve PCBs.The HRGC (Hewlett-Packard 6970 Series gas, CA) was equipped with a silica capillary column (J&W Scientific, CA) and with a splitless injector, and the HRMS (Micromass Autospec Ultima, Manchester, UK) had a positive electron impact (EI+) source.Detailed analytical procedures and instrumental parameters of PCDD/Fs and PCBs are given in our previous work (Wang et al., 2010a).
where K p (m 3 /μg) is a temperature-dependent partitioning constant, TSP (μg/m 3 ) is the total suspended particulate concentration, and F (pg/m 3 ) and A (pg/m 3 ) are the associated particulate and gaseous concentrations of PCDD/Fs and PCBs, respectively.When the log K p is regressed against the logarithm of the subcooled liquid vapor pressure log P L o , the partitioning constant can be calculated as follows (Yamassaki et al., 1982): where m r is the slope and b r is the y-intercept of the trend line.
Eitzer and Hites (1989) have correlated P L o of PCDD/Fs with gas chromatographic retention indexes (GC-RI) on a non-polar (DB-5) GC-column using p,p'-DDT as a reference standard, and the correlation has been redeveloped by Hung et al. (2002): where RI is the gas chromatographic retention indexes derived by Donnelly et al. (1987) and Hale et al. (1985), and T is ambient temperature (K).
The parameters reported by Falconer and Bidleman (1994) have been used for simulating the P L o of PCBs.Values of slopes (m L ) and intercepts (b L ) of the equation were obtained from RI data for 32 PCBs.
Complete datasets on the gas-particle partitioning of PCDD/Fs in Taiwan have been reported by Chao et al. (2004), giving values for m r = -1.29 and b r = -7.2 with R 2 = 0.94.In this study, same parameters are also used for estimating the partitioning constant (K p ) of PCBs.

Atmospheric Dry Deposition
The atmospheric total deposition flux of PCDD/Fs and PCBs is a combination of both gas-and particle-phase fluxes, which is given by where F T is the total PCDD/F and PCB deposition fluxes contributed from both gas and particle phases, F g and F p are the deposition fluxes of PCDD/Fs and PCBs contributed by the gas phase and particle phase, respectively, C T is the measured concentrations of total PCDD/Fs and PCBs in air, V d,T is the dry deposition velocities of total PCDD/Fs and PCBs, C g and C p are the calculated PCDD/F and PCB concentrations in the gas phase and particle phase, respectively, and V d,g and V d,p are the dry deposition velocities of PCDD/Fs and PCBs in the gas phase and particle phase, respectively.Due to the lack of measured data for the dry deposition velocities of total PCDD/Fs and PCBs, average deposition velocities of total PCDD/Fs and PCBs in the ambient air of southern Taiwan proposed by Shih et al. (2006) and Lee et al. (1996b), which were 0.42 and 0.28 (cm/s), respectively, were adopted herein.Values were also adopted in this study for the calculation of deposition fluxes of total PCDD/Fs and PCBs.Dry deposition of gaseous PCDD/Fs and PCBs is mainly by diffusion.A selected value (0.010 cm/s) for gaseous polycyclic aromatic hydrocarbon (PAH) dry deposition velocity, reported by Sheu and Lee (1996) and used by Lee et al. (1996b), is also adopted here for the calculation of PCDD/F and PCB deposition fluxes contributed by their gas phase.

Ambient Air PCDD/F and PCB Concentrations
The average PCDD/F I-TEQ concentrations ranged from 0.0362-0.0720pg I-TEQ/Nm 3 and 0.0229-0.0393pg I-TEQ/Nm 3 in fall and spring, respectively (Table 2).The highest concentration was found in an urban area (site B), while the lowest level was found in a rural area (site R).The values were much lower than the Japanese ambient air quality standard (JAQS) of 0.6 pg I-TEQ/Nm 3 for PCDD/Fs (MEJ, 1999).The above results revealed that the atmospheric concentrations were similar to the results of Wang et al. (2003), which indicated that total PCDD/F concentrations in ambient air samples collected from a background area, a rural area, a residential area, an urban area, an industrial area, and the vicinity of a crematory in Taiwan were 0.006, 0.023, 0.050, 0.093, 0.190, and 0.521 pg I-TEQ/Nm 3 , respectively.As noted in the previous section, sampling site I was the one nearest to the science-based industrial park, if the transient emissions of the industrial activities were the predominant source for the study areas, a much higher level for site I than those in other areas could be expected.However, this was not shown in this study.Since traffic sources were already seen as significant contributor to PCDD/F emissions (Lee et al., 2004;Chung et al., 2010), the relatively high PCDD/F concentration for urban site (located in traffic center) may be due to the influence of vehicles.
When comparing atmospheric concentrations between the two seasons, the total I-TEQ concentration in fall was higher by a factor of 1.6-1.8than in spring, probably  (Shih et al., 2006;Lin et al., 2010;Huang et al., 2011).OCDD was the most abundant congener found in all atmospheric samples with an average contribution of 21-27% of the total PCDD/F concentration at sites I, A, and R, and for 42% at site B followed by 1,2,3,4,6,7,8-HpCDF, OCDF, and 1,2,3,4,6,7,8-HpCDD.Note that the total PCDF concentration was significantly higher than PCDDs except at site B in the fall, probably because of its high proportion of OCDD (42%).
Twelve WHO toxic congeners of PCBs with IUPAC Nos.77, 81, 105, 114, 118, 123, 126, 156, 157, 167, 169, and 189 were identified.The average concentrations of total PCBs detected in this study ranged from 1.21-4.17and 1.19-3.55pg/Nm 3 in fall and spring, respectively, and their corresponding toxic equivalent (TEQ) levels ranged from 0.00234-0.00802and 0.00223-0.00517pg WHO-TEQ/Nm 3 (Table 3).A similar pattern was seen in ambient concentrations of PCBs, showing that the highest and lowest levels were found in an urban area (site B) and a rural area (site R).Higher concentrations of PCBs in urban areas than those of others have been reported in many studies, it is due to the fact that PCB-containing products were widely used in the urban areas (Lee et al., 1996a;Tasdemir et al., 2004;Yeo et al., 2004).It was also found that atmospheric concentrations of PCBs decreased with increasing chlorine substitution numbers, PCB#118 (46.6%, 5Cl) was the dominant species measured at all sampling sites, followed by PCB#105 (19.4%, 5Cl) and PCB#77 (18.0%, 4Cl).
Atmospheric PCB concentrations presented in this study (sum of 12 PCBs) were significantly lower than those observed at urban and rural areas of Korea (sum of 20 PCBs, 5.61-193 pg/m 3 ) (Yeo et al., 2004) and the mixed institutional, residential, and commercial area in Chicago, USA (sum of 20 PCBs, 0.4-8.3ng/m 3 ) (Tasdemir et al., 2004b).However, the TEQ levels presented in this study were comparable to those observed in north Taiwan (2.23-4.49fg WHO-TEQ/m 3 ) (Chi et al., 2008) and southern Taiwan near a municipal solid waste incinerator (4.66-8.81fg WHO-TEQ/m 3 ) (Wang et al., 2010a).The difference can be attributed to the discrepancy in sampling sites, sampling techniques, and number of PCB congeners determined.Additionally, no significant seasonal difference was found in PCB concentrations during the sampling periods.

Gas-Particle Partitioning of PCDD/Fs and PCBs
Gas-particle partitioning of SOCs in the atmosphere is an important factor that controls their transport, degradation, and rate of removal by dry/wet deposition (Bidleman and Harner, 2000).Based on the regression model as stated above, the subcooled liquid vapour pressure (P L o ) and gasparticle partitioning constant (K p ) for individual PCDD/F and PCB congeners in the ambient air were calculated as well as gas-particle partitioning.It was found that the measured and simulated particulate fractions of PCDD/Fs increased as the number of chlorinated substitutes increased; the higher chlorinated congeners (particularly HCDD/F and OCDD/F) were predominant in the particle phase (Tables 4  and 5).The total PCDD was found predominantly associated with particles at all sites of both measured and simulated results, probably due to lower vapor pressures for the PCDDs (Rordorf, 1989).Additionally, the total PCDD/Fs bound to particles increased slightly with decreasing temperature, the highest level was observed in fall (average of 60.2-72.5%),while the lowest one in spring for all sites (average of 42.8-58.6%).The relatively higher PCDD/Fs in the particle phase during cold season was observed, which is similar to that obtained in many studies (Xu et al., 2009;Wang et al., 2010b;Huang et al., 2011).
Results of partitioning fractions of particulate PCBs in the ambient air, as shown in Tables 6 and 7, contrast with those obtained from PCDD/Fs.It was found that almost all PCB congeners were predominant in the gas phase, expect for higher chlorinated congeners (PCB-169 and PCB-189) which were obtained from model predictions.It is mainly caused by the larger analytical error due to the lower concentrations of the higher chlorinated congeners.An increase in the gas phase fraction of PCBs during spring (warm season) was observed in this study, it was caused mainly because of the atmospheric variables including temperature, wind speed, and wind direction (Brunciak et al., 2001;Tasdemir et al., 2004b).
Comparison of measured and modeled fractions of particulate PCDD/Fs and PCBs was also listed in Tables 4-7.The percentages of difference of PCDD/Fs between two methods ranged from -27.9 to 23.0%, those of PCBs ranged from -59.0% to 4.78%.The positive values resulted from the adsorption of gas-phase SVOCs onto the filter surface, whereas negative ones were calculated for volatilization (blow-off) of SVOCs from particles on the filter (Bidleman and Harner, 2000) or breakthrough of the compounds from : measured partitioning percentages of particle-phase PCDD/Fs P s : simulated partitioning percentages of particle-phase PCDD/Fs D: percentages of difference evaluated by P m -P s the gas-phase sorbent (Lee and Jones, 1999).Results reveal that the influence of gas adsorption was observed mostly from the lower chlorinated PCDD/F congeners (4-6 Cl) in fall, whereas the blow-off mechanism dominated in higher chlorinated PCDD/F congeners (≥ 6 Cl) in spring.The blow-off mechanism is also predominant for 6-7Cl PCB congeners in both season.Possible artifacts during air sampling, as well as the uncertainties in models may bias the results of both approaches (Mandalakis et al., 2007).Accordingly, the influence of sampling artifact due to the adsorption or blow-off is considerable and it cannot be neglected.

Dry Deposition Fluxes of PCDD/Fs and PCBs
Dry deposition fluxes of PCDD/Fs and PCBs were calculated based on Eqs. ( 5) and ( 6).The terms C g and C p were determined based on the gas-particle partitioning shown in Tables 4-7, V d,T and V d,g were assumed (shown in previous section), and then the unknown V d,p can be determined.The mean dry deposition fluxes of total PCDD/Fs in the ambient air of four sites ranged from 242-544 pg/m 2 -day (13.1-26.1 pg I-TEQ/m 2 -day) and 157-250 pg/m 2 -day (8.30-14.3pg I-TEQ/m 2 -day) in fall and spring, respectively.Those of total PCBs ranged from 292-1010 pg/m 2 -day (0.570-1.94 pg WHO-TEQ/m 2 -day) and 289-859 pg/m 2 -day (0.546-1.26 pg WHO-TEQ/m 2 -day) in fall and spring, respectively (data not shown).Results reveal that the particle bond deposition fluxes of PCDD/Fs (157- : measured partitioning percentages of particle-phase PCDD/Fs P s : simulated partitioning percentages of particle-phase PCDD/Fs D: percentages of difference evaluated by P m -P s Table 6.Mean partitioning fractions of particle-phase PCBs in the ambient air of four different areas in fall.
The estimated monthly fluctuations of dry deposition fluxes of PCDD/Fs and PCBs in the ambient air of four sampling sites were shown in Figs. 2 and 3, respectively.The total dry deposition fluxes and WHO-TEQ values for all sampling sites reached the highest level in cold season (December) and the lowest level in warm season (May), both values were found to decrease as the temperature increased.The same trend in atmospheric deposition fluxes of PCBs was also observed.The fluctuation of PCDD/F and PCB fluxes may be related to different ambient concentrations and the meteorological conditions (i.e., temperature, relative humidity, wind speed, and atmospheric stability) (Tasdemir and Holsen, 2005;Bozlaker et al., 2008).

Dry Deposition Velocities of Individual PCDD/Fs and PCBs
Dry deposition velocities (V d,T ) for total PCDD/Fs and PCBs were calculated by dividing the total fluxes of PCDD/Fs and PCBs (F T ) by simulated ambient total concentrations (C T ).In order to better understand the dry deposition process, the total deposition velocities of individual PCDD/Fs were calculated as shown in Fig. 4(a).The estimated deposition velocities of individual PCDD/Fs ranged from 0.039-0.511cm/s (mean = 0.319 cm/s) and 0.031-0.546cm/s (mean = 0.307 cm/s) in fall and spring,   respectively.The total deposition velocity increased as the number of chlorinated substitutes increased, and the highest level of OCDD/F was observed.However, no significant seasonal variation was found in the total deposition velocity of PCDD/Fs.The average deposition velocities of total PCDD/Fs were similar to that for the ambient air near two MSWIs in Taiwan (0.44-0.68 cm/s) (Wu et al., 2009) and urban site of Korea (0.49 cm/s) as reported by Moon et al. (2005).
The total deposition velocities for individual PCBs varied between 0.069 and 2.59 cm/s (mean = 0.809 cm/s) in fall and those of spring ranged from 0.069 to 3.38 cm/s (mean = 0.940 cm/s) (Fig. 4(b)).An increasing trend with increasing chlorinated PCBs was observed in this study, it is due to the fact that low chlorinated PCBs are predominant in gas phase which are deposited mainly by diffusion and thus resulted in lower deposition velocities (Lee et al., 1996a).The calculated average deposition velocity for PCBs in this study is slightly lower than the previously reported values ranged from 1.26-5.9cm/s (Tasdemir et al., 2004a;Tasdemir and Holsen, 2005;Cindoruk and Tasdemir, 2007;Bozlaker et al., 2008).The variations in deposition velocities are probably due to particle size distribution, the meteorological conditions, and different measurement techniques (Lee et al., 1996a;Franz et al., 1998).

CONCLUSIONS
This study investigated the deposition characteristics of PCDD/Fs and PCBs at four sampling sites located in southern Taiwan.The total PCDD/Fs was found predominantly associated with particles at all sites of both measured and simulated results, while those of low chlorinated PCBs were dominant in the gas phase.Comparison between measured and predicted particulate PCDD/Fs and PCBs demonstrated that the influence of sampling artifact due to the adsorption or blow-off cannot be neglected.Average particulate deposition contributed 96.6%-99.8% of total deposition fluxes of PCDD/Fs and PCBs at all sites, indicating that the dominant mechanism of dry deposition was particle phase deposition.Calculated deposition velocities of individual PCDD/Fs (0.031-0.546 cm/s) increased as the number of chlorinated substitutes increased.Similar patterns of velocities were also observed for PCBs (0.069-3.38 cm/s), due to the fact that low chlorinated PCBs are predominant in gas phase and have lower deposition velocities.

Fig. 1 .
Fig. 1.Sampling sites of the ambient air in southern Taiwan.

Fig. 2 .
Fig. 2. Estimated monthly dry deposition fluxes of PCDD/Fs in the ambient air of four different areas.

Fig. 3 .
Fig. 3.Estimated monthly dry deposition fluxes of PCBs in the ambient air of four different areas.

Table 1 .
Meteorological information and total suspended particulate (TSP) concentration during the sampling periods.

Table 2 .
Mean concentrations of PCDD/Fs in the ambient air of four different areas.

Table 3 .
Mean concentrations of 12 PCB species (pg/Nm 3 ) in the ambient air of four different areas.

Table 4 .
Mean partitioning fractions of particle-phase PCDD/Fs in the ambient air of four different areas in fall.

Table 5 .
Mean partitioning fractions of particle-phase PCDD/Fs in the ambient air of four different areas in spring.

Table 7 .
Mean partitioning fractions of particle-phase PCBs in the ambient air of four different areas in spring.

Table 8 .
Dry deposition fluxes of total PCDD/Fs published by previous studies.

Table 9 .
Dry deposition fluxes of total PCBs published by previous studies.