Atmospheric Deposition Modeling of Polychlorinated Dibenzo-p-dioxins , Dibenzofurans and Polychlorinated Biphenyls in the Ambient Air of Southern Taiwan . Part I . Dry Depositions

Atmospheric deposition, including dry and wet deposition, is a primary pathway for the transfer of POPs to terrestrial and aquatic ecosystems. In this study (that is, the part I.), the characteristics of PCDD/Fs and PCBs in the ambient air of Tainan City were simulated by the PM10 versus PCDD/Fs concentration regression analysis, gas-particle partition modeling, and the simulation of dry deposition. Dry deposition fluxes are obtained from the combination of the PCDD/F and PCB concentrations, meteorological information, dry deposition velocities, and scavenging ratios. The dry deposition fluxes of PCDD/F-TEQ2005 increase with decreasing temperature, while increase with a higher degree of chlorine numbers on PCDD/F homologues. In this study (that is, the part I.), the average PCDD/F dry deposition fluxes in spring, summer, fall and winter were 69.3, 28.2, 129 and 246 pg WHO-TEQ/m-month during 2012, respectively. As for 2013, the average PCDD/F dry deposition fluxes in spring, summer, fall and winter were 67.0, 29.8, 102 and 377 pg WHO-TEQ/m-month, respectively. The average PCB dry deposition fluxes in spring, summer, fall and winter were 2.16, 1.99, 5.70 and 11.9 pg WHO-TEQ/m-month during 2012, respectively. As for 2013, the average PCB dry deposition fluxes in spring, summer, fall and winter were 2.11, 1.27, 4.49 and 8.88 pg WHO-TEQ/m-month, respectively. The minimum simulated value occurred in summer, while the maximum dry deposition fluxes, which were about 4–5 times higher than the minimum values, occurred in winter. The lower values observed in summer may be caused by the atmospheric diffusion of SVOCs and high rainfall intensity.


INTRODUCTION
Concentration levels of persistent organic pollutant (POPs) in different environmental matrices had become a global concern due to their environmental characteristics of persistence, bioaccumulation, toxicity and global dispersion.
Polychlorinated dibenzo-p-dioxin (PCDD), polychlorinated dibenzofuran (PCDF) and polychlorinated biphenyls (PCB) are semi-volatile compounds.Both of PCDD/Fs and PCBs have high boiling points, high degree of chemical stability (Aristizábal et al., 2011;Wang et al., 2003).PCDD/Fs are almost formed as undesired byproducts from anthropogenic and nature processes such as thermal processing of waste, internal combustion engines of vehicles when chlorine, oxygen, hydrogen and carbon are present (Quaß et al., 2004;Altarawneh et al., 2009;Relvas et al., 2013).PCBs have been used in closed systems, such as hydraulic fluids and insulating fluid in electrical transformers and capacitors (Heinzow et al., 2007).Approximately two-third of the PCBs was used for various open applications where PCBs functioned either as softener, surfactant, flame retardant, lubricator or dispersant in different materials and products (Berg et al., 1998;Kerst et al., 2003;Anderson et al., 2004;Heinzow et al., 2007).Since some pollution incidents broke out in Taiwan during 20 and 21 century, PCDD/Fs as well as PCBs became the most important issues owing to their fat-soluble characteristic which could put threat to human health (Lee et al., 1996a;Lee et al., 2009).
The cycle of PCDD/Fs and PCBs contamination lasted for many years and people who were affected by it presumably throughout their whole life or for a relevant part of it.Consequently, we need to pay close attention to the concentration levels of PCDD, PCDF and PCB in the ambient air, including dry and wet depositions, which were thought to be major path way for the transfer of persistent organic pollutants (Agrell et al., 2002;Lee et al., 1996b;Tasdemir et al., 2005).Since PCDD/Fs and PCBs are both semi-volatile organic pollutants, they demonstrate two different phases in the ambient air, one is gaseous phase and the other is compounds bound to particles (Aristizábal et al., 2011).Once PCDD/Fs and PCBs are emitted into the ambient air, they could be transported by dry and wet depositions, especially particulate phase, into terrestrial environments, water systems, and even food chains.
A recent surge of research on PCDD/Fs and PCBs has given us new information and challenges in Asia (Mi et al., 2012;Jen et al., 2013;Sun et al., 2013a, b;Watcharavitoon et al., 2013).This study (the part I.) focuses on modeling the characteristics of PCDD/F and PCB in the atmosphere of southern Taiwan.By using the data collected seasonally in different types of areas including one industrial area, two urban areas as well as one rural area during 2010 and 2011 (Mi et al., 2012), the concentrations of PCDD/F and PCB could be determined through PM 10 concentrations.Gaseous and particulate concentrations were both obtained by means of simulation.Total fluxes (dry + wet) of PCDD/Fs and PCBs were developed by using meteorological information, gasparticle partitioning, scavenging ratios, etc.The simulated results from this study could provide helpful information to build up more complete inventory for understanding the level of PCDD/Fs and PCBs near Tainan city, or formulating stricter regulations about PCDD/Fs and PCBs.

Sampling of PCDD/Fs and PCBs from the Atmosphere
The ambient air in Tainan City of four different sites were sampled for two seasons.Each ambient air sample was collected using PS-1 sampler (Graseby Anderson, GA, USA) according to the revised U.S. EPA Reference Method T09A.The sampling flow rate was about 0.225 m 3 /min.In the first season, each sample was collected continuously on seven consecutive days yielding a sampling volume about 4000 m 3 .Nevertheless, in order to obtain more accurate samples, sampling period was doubled to get higher sampling volume (sampling volume ~ 9000 m 3 ).The PS-1 sampler was equipped with quartz-fiber filter for sampling particlephase SVOCs, and followed by a glass cartridge containing PUF for sampling gas-phase SVOCs, respectively.
In this study, total four atmospheric sampling sites (Fig. 1) and sampling periods in Tainan were presented in Mi et al. (2012).Based on the data given by Mi et al. (2012) combined with meteorological information provided by Air Quality Monitoring Stations, the concentration of PCDD/Fs and PCBs could be simulated.

Gas-Particle Partitioning Simulation Model
PCDD/Fs and PCBs are groups of semi-volatile organic compounds (SVOCs) as well as persistent organic pollutants (POPs) (Atkinson, 1996).They are stable in the environment, undergo long range transport over regional or even global scales in the atmosphere and can be accumulated in organisms via food chain (Bidleman, 1988).This means atmosphere is a major pathway to transport PCDD/Fs and PCBs.Atmospheric deposition is a significant process affecting the global sinks of natural and anthropogenic POPs (Quaß et al., 2004), thus gas-particle partitioning of PCDD/Fs and PCBs in the atmosphere will affect such deposition process (Lohmann and Jones, 1998).Gas-phase POPs are depleted from the atmosphere owing to photochemical degradation reactions (Pankow, 1994), while particle-bound deposition accounts for most of the PCDD/F and PCB flux to the ecosystem (Lohmann and Jones, 1998).As a result, knowledge of the partitioning of SVOCs in the atmosphere is essential for comprehending with their subsequent fate in the environment (Lohmann and Jones, 1998).
Gaseous and particulate concentrations are estimated by multiplying gas-particle partitioning and the total concentration of SVOCs together.The gas-particle partitioning is simulated by Eq. (1).The Eq. (1) has been used to describe the gasparticle partitioning constant (Yamasaki et al., 1982;Pankow, 1991;Pankow andBidleman, 1991, 1992;Pankow, 1994), and as follows: (1) K p : gas-particle partitioning constant (pg/m 3 ) -1 , TSP: concentration of total suspended particulate material (pg/m 3 ), TSP was calculated by TSP/PM 10 ratio 1.24 (Sheu et al., 1996) F: particulate concentration of SVOCs (pg/m 3 ), A: gaseous concentration of SVOCs (pg/m 3 ).Plotting log K p against the logarithm of the subcooled liquid vapor pressure (PL o ), gives: P L o : subcooled liquid vapor pressure (Torr), m r : slope of a plot of log Kp versus log P L o , b r : y-intercept in a plot of log K p versus log P L o (Lohmann and Jones, 1998).Chao et al. (2004) reported a complete datasets on the gas-particle partitioning of PCDD/Fs in Taiwan.The study gave parameters for Eq (2), m r = -1.29 and b r = -7.2 with R 2 = 0.94 (Chao et al., 2004).Those parameters are used for evaluating both PCDD/F and PCB gas-particle partitioning constant (K p ) in this study.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 (Eitzer and Hites, 1989), and the correlation has been redeveloped by Hung et al. (2002): Log (P L o ) = -3.14× (RI/T) + 1.67 × 10 -3 × (RI) -(1320/T) + 8.087 (3) RI: gas chromatographic retention indexes (GC-RI), referred to Donnelly and Hale (Hale et al., 1985;Donnelly et al., 1987), T: ambient temperature (K).Furthermore, parameters for simulating the subcooled liquid vapor pressure (P L o , Pa) of 180 PCB congeners as a function of temperature and ortho-chlorine substitution were given by Falconer and Bidleman (1994), which present in equation below.

Log (P
Parameters such as m L and b L were determined from gas chromatographic retention data for 32 PCB congeners.The slopes (m L ) altered regularly with homolog (same numbers of total substituted chlorines) and with the number of ortho-chlorines.From the data and information given by Falconer and Bidleman (1994), m L and b L values of other 148 PCB congeners whose vapor pressures had been reported at a fixed temperature were evaluated.

Dry Deposition Process
There are three principal steps of the dry deposition process, aerodynamic transport, boundary layer transport and interactions with the receptor surface (Pankow, 1987).The aerodynamic transport process involves pollutant transport from the free atmosphere down to the quasi-laminar sublayer immediately adjacent to the surface.Movement across the sublayer comprises boundary layer transport as well as physical and chemical interactions between pollutant and the surface material compose the boundary layer transport process (Davidson and Wu, 1987).
The atmospheric dry deposition flux of SVOCs is a combination of both gas-and particle-phase flux, which is given by F d,T : the dry deposition flux of SVOCs from both gaseous and particulate phases, F g : the dry deposition flux contributed by the gas-phase SVOCs, F p : the dry deposition flux contributed by the particlephase SVOCs, C T : the concentration of total SVOCs in the ambient air, V d,T : the dry deposition velocity of total SVOCs, C g : the simulated concentration of gas-phase SVOCs in the ambient air, V d,g : the dry deposition velocity of gas-phase SVOCs, C p : the simulated concentration of particle-phase SVOCs in the ambient air, V d,p : the dry deposition velocity of particle-phase SVOCs.Shih et al. (2006) and Lee et al. (1996a) reported the mean dry deposition velocities of PCDD/Fs and PCBs, which were 0.42 and 0.28 (cm/sec), respectively (Lee et al., 1996a;Shih et al., 2006).Dry deposition of gas-phase PCDD/Fs and PCBs are predominantly by diffusion.Owing to the lack of measured data for PCDD/Fs and PCBs, 0.010 (cm/s) is selected to simulate PCDD/F and PCB dry deposition flux contributed by their gaseous phase.The value was reported by Sheu et al. (1996) used as the dry deposition velocity of gaseous polycyclic aromatic hydrocarbon (PAH), applied by Lee et al. (1996b).
Dry deposition of particle-phase PCDD/Fs and PCBs are predominantly by the gravitational settling.Therefore, the PCDD/F total dry deposition velocity 0.42 (cm/s) combine with gaseous phase deposition velocity 0.01 (cm/s) to calculate deposition velocity of particulate phase.Then use the PCDD/F simulated deposition velocity of particulate phase as the PCB simulated deposition velocity of particulate phase, because we assumed PCDD/Fs and PCBs bounded to particle that have the same average settling velocity, which can be simulated by Eq. ( 6).

Meteorological Information
The pollutant transmission in the atmosphere would be affected by the meteorological conditions, such as wind speed, rainfall intensity, PM 10 concentration and the atmospheric stability.The meteorological information for Tainan City during 2012 to 2013 was obtained from the nearby Air Quality Monitoring Stations, which were Tainan, Annan, and Shanhua Air Quality Monitoring Stations.
The required meteorological information is verified and all applied meteorological information is the averaging of data provided by a fore-stated three Air Quality Monitoring Stations.The average temperature monthly during 2012 and 2013 were 17.6-29.5°Cand 17.2-29.3°C,respectively.The maximum rainfall intensity per month during 2012 and 2013 were 869 mm in June and 916 mm in August, respectively.By month, the average PM 10 concentrations and wind speed were 34.8-105 μg/m 3 and 1.74-2.77m/sec during 2012 and 2013.

Simulated Concentrations of PCDD/Fs and PCBs in the Ambient Air
The detail simulated PCDD/F concentrations in the atmosphere of Tainan City, Taiwan are presented in Tseng (2014).The main results indicate that the maximum and minimum total PCDD/F concentrations in 2012 were 1.30 pg/Nm 3 (March) and 0.532 pg/Nm 3 (July), respectively.The maximum and minimum concentrations in 2013 were 1.49 pg/Nm 3 (November) and 0.528 pg/Nm 3 (August), respectively.The maximum and minimum total PCDD/F WHO-TEQ 2005 concentrations in 2012 were 0.0645 pg WHO-TEQ/Nm 3 (March) and 0.0253 pg WHO-TEQ/Nm 3 (July), respectively.The maximum and minimum concentrations in 2013 were 0.0709 pg WHO-TEQ/Nm 3 (November) and 0.0251 pg WHO-TEQ/Nm 3 (August), respectively.
The PCDDs/PCDFs mass ratios (0.576-0.745) during 2012 and 2013 were all lower than 1, which indicated that PCDFs were the dominant contribution of the PCDD/F atmospheric concentrations.On the other hand, the PCDDs/PCDFs WHO-TEQ 2005 ratios (0.383-0.387) during 2012 and 2013 were all less than 1, which indicated that PCDFs also dominated the total toxicity.All simulated concentrations were much lower than the PCDD/F regulated standard of air quality in Japan (0.6 pg WHO-TEQ/Nm 3 ).The PCDD/F WHO-TEQ 2005 concentrations (0.0251-0.0709WHO-TEQ/Nm 3 ) in this study were in low range of the concentrations in comparison with other countries, such as Maderira, Portugal (0.130 pg I-TEQ/Nm 3 ) (Oh et al., 2006) and Bucheon, Korea (0.22-1.16 pg I-TEQ/Nm 3 ) (Coutinho et al., 2007).As can be seen from the atmospheric concentrations of summer and winter, the total mean WHO-TEQ 2005 concentration in winter (0.0508 and 0.0626 pg WHO-TEQ/Nm 3 in 2012 and 2013, respectively) was approximately 2 times the magnitude in summer (0.0278 and 0.0283 pg I-TEQ/Nm 3 in 2012 and 2013, respectively), which has been found in previous studies (Shih et al., 2006;Lee et al., 2009;Wang et al., 2010).
The maximum and minimum total PCB concentrations in 2012 were 3.43 pg/Nm 3 (March) and 1.40 pg/Nm 3 (July), respectively.The maximum and minimum concentrations in 2013 were 3.94 pg/Nm 3 (November) and 1.39 pg/Nm 3 (August), respectively.
The maximum and minimum total PCB WHO-TEQ 2005 concentrations in 2012 were 0.00580 pg WHO-TEQ/Nm 3 (March) and 0.00307 pg WHO-TEQ/Nm 3 (July), respectively.The maximum and minimum concentrations in 2013 were 0.00861 pg WHO-TEQ/Nm 3 (November) and 0.00305 pg WHO-TEQ/Nm 3 (August), respectively.When comparing with the PCDD/F concentrations, the atmospheric PCB WHO-TEQ concentrations were 10 times lower than the PCDD/Fs, which indicated the toxicity in the ambient air was mainly dominated by PCDD/Fs.In this study, the result shows the simulated total (PCDD/Fs plus PCBs) atmospheric concentrations of Tainan city.The maximum total concentrations in the ambient air during 2012 and 2013 are 0.0703 pg/Nm 3 (March) and 0.0795 pg/Nm 3 (November), respectively.

Gas-Particle Partitioning
Atmosphere is a significant transport pathway influencing the global deposition of natural and anthropogenic POPs (Pacyna et al., 2003;Quaß et al., 2004), thus gas-particle partitioning of PCDD/Fs and PCBs in the atmosphere will affect such process (Lohmann and Jones, 1998).The methods to evaluate gas-particle partitioning, the subcooled liquid vapor pressure and gas-particle partitioning constant were presented in foregoing section.
The required meteorological information for the subcooled liquid vapor pressure (P L o ) and gas-particle partitioning constant (K p ) are presented in Tseng (2014).The total suspended particulate (TSP) concentrations were evaluated by the fraction, TSP: PM 10 = 1.24:1, which was cited from Sheu et al. (1996).On the basis of the meteorological information, the evaluated P L o and K p of PCDD/Fs and PCBs in the atmosphere during 2012 and 2013 can be simulated.From the data of both PCDD/Fs and PCBs, the simulated value of P L o decline with the increase of substituted chlorine numbers; nevertheless, the value of K p shows increase trend, which is opposed to the former.Consequently, the PCDD/F and PCB congeners that possess more chlorine atoms will have higher fraction of particlephase at identical temperature and TSP concentration.
The gas-particle partitioning in the ambient air of PCDD/Fs and PCBs are also be calculated.According to the results of contribution fraction of particle phase, the PCDD/F and PCB congeners with higher chlorine numbers occupied a higher particle fraction.Compare the particulate fraction of PCDDs and PCDFs, PCDD homologues is apt to attaching to particles than the equivalent PCDFs, which might owing to the vapor pressure of PCDFs are slightly less than PCDDs (Rordorf, 1989).Comparable results can also be found in previous studies (Lohmann and Jones, 1998;Chao et al., 2004;Lin et al., 2010a).Different from the PCDD/F gasparticle partitioning, the predominance of gaseous PCBs in the atmosphere is well-known and it has been pointed out in several other studies (Mandalakis andStephanou, 2004 andTasdemir et al., 2005).
Furthermore, the particulate fractions of PCDD/Fs and PCBs decline with the increase of ambient temperature.Previous study reported that vapor pressure is the predominant factor which affect the partition between gas and particle phase of semi-volatile organic compounds (Pankow, 1987).The study also reported the ambient temperature would be a significant factor influencing the gas-particle partitioning.Take PCB-189 for example, the particulate fraction in January 2012 (temperature 12.2-27.3°C)is 88.5%; however, the fraction decrease to 31.6% in July 2012 (temperature 32.4-25.6°C).The result also reveals PCB-189 dominated by gas phase in summer (from June to Aug.), but particulate phase dominant in a lower temperature season, such as in winter (from Dec. to Feb.).

Dry Deposition of PCDD/Fs and PCBs
The atmospheric dry deposition fluxes of SVOCs are the combination of both gas-and particle-phase flux, and the method to calculate dry deposition fluxes of SVOCs were demonstrated in previous section of the article.Dry depositions of gas-phase PCDD/Fs and PCBs are predominantly by diffusion.Owing to the lack of measured data for PCDD/Fs and PCBs, 0.01 (cm/s) is selected as the dry deposition velocity for both gaseous phase PCDD/Fs and PCBs.The above value was reported by Sheu et al. (1996) used as the dry deposition velocity of gaseous polycyclic aromatic hydrocarbon (PAH), applied by Lee et al. (1996b).Shih et al. (2006) and Lee et al. (1996a) reported the mean dry deposition velocities of total PCDD/Fs and total PCBs, which were 0.420 and 0.280 (cm/sec), respectively (Lee et al., 1996a;Shih et al., 2006).
Dry deposition of particle-phase PCDD/Fs and PCBs are predominantly by the gravitational settling.Therefore, the dry deposition velocity (0.42 cm/s) of total PCDD/Fs combine with gaseous phase deposition velocity 0.01 (cm/s) to calculate deposition velocity of particulate phase.Then, by using the calculated total PCDD/F dry deposition velocity in particulate phase (Eq.( 6)) as that is particulate-phase total PCBs (V d,p of total-PCBs).This can be explained that both PCDD/Fs and PCBs were bounded to the same particulates that have the same average settling velocity.
The required information for evaluating PCDD/F and PCB dry deposition fluxes is presented in Tseng (2014).The maximum and minimum simulated dry deposition velocities of particle-phase PCDD/Fs and PCBs during 2012 are 0.718 and 0.489 cm/s in July and January, respectively.For 2013, the maximum and minimum simulated dry deposition velocities of particle-phase PCDD/Fs and PCBs are 0.705 and 0.479 cm/s in August and January, respectively.

CONCLUSIONS
Average PCDD/F dry deposition fluxes in spring, summer, fall and winter are 69.3,28.2, 129 and 246 pg WHO-TEQ/m 2 -month during 2012, respectively.As for 2013, the average PCDD/F dry deposition fluxes in spring, summer, fall and winter are 67.0,29.8, 102 and 377 pg WHO-TEQ/m 2month, respectively.The maximum PCDD/F dry deposition fluxes occurred in winter, while the minimum value were in summer.Average PCB dry deposition fluxes in spring, summer, fall and winter are 2.16, 1.99, 5.70 and 11.9 pg WHO-TEQ/m 2 -month during 2012, respectively.As for 2013, the average PCB dry deposition fluxes in spring, summer, fall and winter are 2.11, 1.27, 4.49 and 8.88 pg WHO-TEQ/m 2 -month, respectively.The maximum PCB dry deposition fluxes occurred in winter, while the minimum value were in summer.
The results also demonstrate that the dry deposition fluxes of PCDD/Fs and PCBs are dominated by particulate phase, which is owing to a higher value of particulate deposition velocity of both PCDD/Fs and PCBs (0.479-0.718 cm/s) than the gaseous phase deposition (0.010 cm/s), respectively.

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

Fig. 2
present total (PCDD/Fs plus PCBs) dry deposition fluxes.From Fig. 2, the maximum total dry deposition during 2012 and 2013 are 415 and 435pg WHO-TEQ/m 2 -month in March and January, respectively.The minimum monthly total dry deposition during 2012 and 2013 are 29.4 and 31.1 pg WHO-TEQ/m 2 -month both in August, respectively.Fig. 3 shows the fraction of WHO-TEQ contributed by PCB dry deposition fluxes.The contribution fractions from 2.28% to 5.87%, which means the dry PCDD/F deposition fluxes dominate the total dry deposition fluxes.The simulated results demonstrate the dry deposition fluxes of PCDD/Fs

Table 5 .
Simulated PCB dry deposition fluxes in the ambient air of Tainan City during January 2012 to June 2012 (pg/m 2

Table 6 .
Simulated PCB dry deposition fluxes in the ambient air of Tainan City during July 2012 to December 2012 (pg/m 2

Table 7 .
Simulated PCB dry deposition fluxes in the ambient air of Tainan City during January 2013 to June 2013 (pg/m 2

Table 8 .
Simulated PCB dry deposition fluxes in the ambient air of Tainan City during July 2013 to December 2013 (pg/m 2 The maximum and minimum total PCB dry deposition fluxes in winter during 2013 are 21.7 (96.2% contributed by particulate phase) and 8.88 (88.1% contributed by particulate phase) pg WHO-TEQ/m 2 -month, respectively.