Atmospheric PM 2 . 5 and Depositions of Polychlorinated Dibenzo-p-dioxins and Dibenzofurans in Kaohsiung Area , Southern Taiwan

Kaohsiung County, in southern Taiwan, has both highly industrial and rural areas. In this study, the characteristics of PCDD/Fs in the ambient air of Kaohsiung (urban), Meinong (rural) and Xiaogang (heavy industrial zone) in 2014 and 2015 were modeled based on the prevailing meteorological conditions and the measured ambient air concentrations of PM2.5, PM10 and TSP. The yearly average PM2.5 concentrations in the ambient air of the three areas were in the range of 23 to 31 μg m, all above the National Air Quality Standard of Taiwan (15 μg m). The simulated average concentrations of PCDD/Fs in the whole of Kaohsiung area in terms of toxicity equivalent were in the range of 0.034–0.053 pg WHO2005TEQ m. The average total deposition fluxes of total-PCDD/Fs-WHO2005-TEQ ranged between 155.4 and 276.6 pg WHO2005-TEQ m month with 1,2,3,7,8-PeCCD and 2,3,4,7,8-PeCDF being the dominant congeners in terms of PCDD/Fs WHO2005-TEQ. Xiaogang area with highly industrial activities had the highest concentrations of PM2.5 and PCDD/Fs and corresponding total-PCDD/Fs WHO2005-TEQ deposition fluxes, while the Meinong in the rural site recorded the lowest. Average dry deposition velocities of total PCDD/Fs WHO2005-TEQ for both 2014 and 2015 were 0.162, 0.148 and 0.161 cm s for Kaohsiung, Meinong and Xiaogang, respectively, while the average scavenging ratios of total PCDD/Fs WHO2005-TEQ were 6232, 4701 and 6802, for Kaohsiung, Meinong and Xiaogang, respectively. The information provided in this work is useful for both further studies and environmental control strategies concerning atmospheric aerosols and dioxins.


INTRODUCTION
Polychlorinated dibenzo-p-dioxins (PCDDs) and polychlorinated dibenzofurans (PCDFs) have received much public concern due to their ubiquitous nature caused by their stable, persistent, lipophilic and bio-accumulative properties as well as their ability to undergo global dispersion via long range transport (Wu et al., 2009b;Chen et al., 2010).PCDD/Fs are semi-volatile organic compounds (SOCs) and persistent organic pollutants (POPs), which are toxic to coming from the emissions of waste combustion, chemical plants, thermal sources, metal smelting process, and vehicles (Schuhmacher et al., 2000;Wang et al., 2003;Lin et al., 2007;Hsieh et al., 2009;Chuang et al., 2010Chuang et al., , 2011)).
PCDD/Fs are emitted in the atmosphere where they are transformed, degraded as well as transported from source to receptor sites (Chi et al., 2009;Xu et al., 2009;Fang et al., 2011).In the atmosphere, the PCDD/Fs are partitioned between gas and particle phases a process which is dependent on their vapor pressures, ambient temperatures and other parameters (Wu et al., 2009a;Wang et al., 2010;Cheruiyot et al., 2015).The PCDD/Fs can be degraded by chemical reactions controlled by the OH radicals as well as by the photochemical reactions (Chi et al., 2009).Removal of PCDD/Fs from the atmosphere occurs via both dry and wet deposition (Giorgi, 1988;Chi et al., 2009;Wu et al., 2009a;Huang et al., 2011a;Mi et al., 2012).
Particulate matter is a kind of aerosol, which is defined as a suspension of a solid or liquid particle in a gas (Ghosh et al., 2014).The composition of aerosols includes haze, smoke, fumes, mist and fog, dust and smog.Aerosols can be classified into three categories according to their aerodynamic diameters: TSP, PM 10 , and PM 2.5 (Lu et al., 2016).Aerodynamic diameters of TSP ranges from ~0 to 100 µm, while that of PM 10 ranges from ~0 to 10 µm, and that of PM 2.5 are from ~0 to 2.5 µm (Chow et al., 2015).TSP, PM 10 and PM 2.5 have been found to contribute to low visibility and poor air quality (Chen et al., 2014).The content of ambient particulate matters is conglomerate of many pollutant subclasses and it also comprises organic and inorganic species.The sources of particulate matter can be natural or anthropogenic (Kong et al., 2014;Alghamdi et al., 2015).Natural sources include volcanic eruptions, wood burning, and sea sprays, while anthropogenic sources include industries, automobiles and construction activities.In addition to the primary sources, there is the secondary aerosol formation in the environment.After gaseous pollutants through photochemical reaction produces freshly nucleated particles which called secondary aerosol formation.The studies show that the formation of secondary aerosol depending on their saturation ratio and environmental conditions.Ambient particulate matters can be removed from the atmosphere by either dry or wet deposition.Dry deposition occurs when the particles are affected by gravity and when inertial forces impact on surfaces, such as trees, buildings, ground and bodies of water, while wet deposition means that aerosols are incorporated into cloud droplets and removed with precipitation and also occurs when the particles located below a precipitation cloud become scavenged by impacting droplets.
The objectives of this study were to investigate concentrations of PM 2.5 , PM 10 , and the concentrations of PCDD/Fs.Additionally, the gas-particle partitioning and dry and wet deposition fluxes of atmospheric PCDD/Fs in southern Taiwan were also evaluated.This study was divided into three areas: one was industrial area, Xiaogang; the second one was nearby mountain area, Meinong; the other part was the rest of Kaohsiung City.To achieve these goals, samples were collected near municipal solid waste incinerators (MSWIs), electric arc furnaces and sinter plants in Xiaogang area for a period of four years.Collected samples were analyzed for specific 17 PCDD/Fs congeners to establish atmospheric concentration level, and then gaseous and particulate concentration could be calculated.Dry and wet deposition fluxes were simulated by model calculation during 2014 and 2015.Results of this study provides long-term data of PCDD/Fs in Taiwan and display helpful information of building up more comprehensive inventory for understanding the level of PCDD/Fs in Kaohsiung city as well as giving some useful data for further investigation.

Sampling Sites
Samples used in this experiment were obtained from three areas in Kaohsiung County in South of Taiwan (Fig. 1).The sampling sites were chosen since there are few studies focusing on the difference in atmospheric deposition patterns in Kaohsiung County of Taiwan.Sampling period was from February 2010 to April 2013.This data was used to simulate the deposition and partition characteristics of PCDD/Fs for the year 2014 and 2015.

Meteorological Conditions and PM Concentration during the Sampling Periods
The pollutant transmissions and deposition in the atmosphere are affected by the meteorological conditions, such as wind speed, rainfall intensity, PM 2.5 , PM 10 , and TSP concentrations and the atmospheric stability.In this study, the pertinent meteorological information and PM 10 concentrations for Kaohsiung, Meinong, and Xiaogang, respectively, for the year 2014 and 2015 were obtained from the nearby air quality monitoring stations.The meteorological conditions prevailing in the sampling areas over the whole simulated period of 2014-2015 are summarized in the Tables 1-3.

Sampling Procedures and Analysis
Samples were collected, for a period of five days, using PS-1 sampler (Graseby Andersen, GA) according to the T09A method which was referred by the United States Environmental Protection Agency (US-EPA).The PS-1 sampler was used to collect both gas and particle-phase compounds.For all the samples the average volume of ambient air sampled was in the range of 630 to 1370 m 3 .Particle-phase compounds were collected by quartz fiber filter, whereas the gas-phase was collected using polyurethane foam.To evaluate contamination during sampling, one field blank was taken during the individual sampling events for quality assurance purposes.Field blanks were loaded into the sampling system, but no air was drawn through them.They experienced the same handling, storage, and analysis procedures as the actual samples.
All the chemical analyses in this study were carried out in an accredited laboratory, Super Micro Mass Research and Technology Centre, in Cheng Shiu University which is certified by Taiwan EPA to analyze PCDD/Fs in Taiwan.High-resolution gas chromatograph/high-resolution mass Table 1.Meteorological data at Kaohsiung.spectrometry (HRGC/HRMS) (Hewlett Packard 6970 Series, CA, USA) was used for PCDD/F analysis.Detailed analytical procedures and instrumental parameters of PCDD/Fs given in the previous work (Wang et al., 2010).In summary, seventeen 2,3,7,8-substituted PCDD/F congeners were analyzed and one field blank was incorporated for each sample.Additionally, 13 C 12 -2,3,7,8-substituted PCDD/Fs internal standards were spiked into samples to quantify for recoveries during analysis.

Gas-Particle Partitioning
Gaseous and particulate concentrations of PCDD/Fs were evaluated by the gas-particle partitioning multiplying the total concentrations of PCDD/Fs.partitioning was simulated by an equation, proposed by several researchers, that successfully describes gas-particle partitioning constant (Yamasaki et al., 1982;Pankow, 1987;Pankow andBidleman, 1991, 1992): K p : gas-particle partitioning constant (m 3 µg -1 ), TSP: concentration of total suspended particulate material (µg m -3 ), F: particle phase concentration of PCDD/Fs (pg m -3 ), A: gaseous phase concentration of PCDD/Fs (pg m -3 ).Plotting log K p against the logarithm of the subcooled liquid vapor pressure (P L o ), gives: P L o : subcooled liquid vapor pressure (Torr), m r : slope of a plot of log K p vs. log P L o b r : y-intercept in a plot of log K p vs. log P L o (Lohmann and Jones, 1998).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 re-developed by (Hung et al., 2002) (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) (Hung et al., 2002) A complete datasets on the gas-particle partitioning of PCDD/Fs in Taiwan have been reported by (Chao et al., 2004).From their data, parameters for Eq.(1) were determined as m r = -1.29 and b r = -7.2 with R 2 = 0.94.In this study, those parameters are also used for estimating the partitioning constant (K p ) of PCDD/Fs.

Dry Deposition Fluxes of PCDD/Fs
The dry deposition fluxes of PCDD/Fs in the atmosphere is a combination of both gas-phase and the particle-phase fluxes, which is given by: F T : the summation of PCDD/F deposition fluxes from both gas and particle phases, F g : the PCDD/F deposition flux contributed by the gas phase (Wang et al., 2010) C T : the measured concentration of total PCDD/Fs in the ambient air, V d,T : the dry deposition velocity of total PCDD/Fs, C g : the calculated concentration of PCDD/Fs in the gas phase, V d,g : the dry deposition velocity of the gas-phase PCDD/Fs, C p : the calculated concentration of PCDD/Fs in the particle phase, V d,p : the dry deposition velocity of the particle-phase PCDD/Fs.Dry deposition of particle-phase PCDD/Fs occurs mainly via the gravitational settling.The dry deposition velocities of particle-phase PCDD/Fs (V d,p ) can be simulated by Eqs. ( 5) and ( 7).

Theory of Scavenging Ratios
The wet deposition flux of PCDD/Fs is a combination of both vapor dissolution into the rain and the removal of suspended particulates by precipitation.The gas scavenging ratio, S g , can be estimated by S g : the gas scavenging ratio of PCDD/Fs (dimensionless), R: the universal gas constant (82.06 × 10 -6 m 3 atm mol -1 K -1 ), T: ambient temperature (K), H: Henry constant (m 3 atm mol -1 ).
C rain, dis : the dissolved-phase concentration of PCDD/Fs in the raindrop, C g : the concentration of PCDD/Fs in the gas phase.
The particle scavenging ratio, Sp, on the other hand, can be calculated by: S p = C rain, particle /C p (10) S p : the particle scavenging ratio of PCDD/Fs (dimensionless), C rain,particle : the particle-phase concentration of PCDD/Fs in the raindrop, C p : the concentration of PCDD/Fs in the particle phase.Total scavenging of precipitation (S tot ) is the sum of gas and particle scavenging, which can be calculated by: S tot : the total scavenging ratio of PCDD/Fs (dimensionless), Φ: the fraction of PCDD/Fs bound to particles.Due to the lack of real measured data for the particle scavenging ratios of PCDD/Fs, the values used in this study were referenced to those in Eitzer and Hites (1989) work.

Determination of Wet Deposition Fluxes of PCDD/Fs
Wet deposition is the removal of particles in the atmosphere by precipitation (rainfall and cloud droplets) and precipitation scavenging accounts for the majority of removing SVOCs from the atmosphere by wet deposition (Huang et al., 2011b).Wet deposition flux of SVOCs is a combination of both vapor dissolution into rain and removal of suspended particulates by precipitation (Bidleman, 1988;Koester and Hites, 1992).
The wet deposition flux of SVOCs can be evaluated as follows:

QA/QC
The protocol for quality assurance/quality control (QA/QC) was rigorously followed.In this study, the recovery efficiency of all samples conformed to the relevant quality control requirements and the blank tests also show that no significant contamination.The field blank samples showed nondetection for most congeners except OCDF and OCDD which averaged at 0.00193 and 0.00206 pg m -3 , respectively.In this study the mean recoveries of standards for all 13 C 12 -2,3,7,8-substituted PCDD/Fs were in the range of 25-130% in comparison to the criteria recoveries of 70-130%.The MDLs for individual congeners ranged from 0.000438 to 0.00602 pg m -3 while the MDLs for the total PCDD/Fs the range was from 0.0147 to 0.0406 pg m -3 .

Meteorological Conditions during Sampling Period
The prevailing meteorological conditions influence the deposition and partitioning of PCDD/Fs in the atmosphere.The corresponding values for temperature, PM 2.5 , PM 10 , and TSP are as shown in Tables 1-3.During the sampling period the monthly average atmospheric temperatures at Kaohsiung, Meinong, and Xiaogang were in the range of 19.5-30.3°C, 18.5-30.1°C and 19.8-30.8°Cfor 2014, while for 2015 it was 19.9-30.6°C, 19.6-30.7°C and 19.8-30.7°Cfor 2015, respectively.The yearly average temperatures in 2014 were 25.8, 25.6 and 25.2°C for Kaohsiung, Meinong, and Xiaogang while for 2015 they were 26.1, 25.7 and 26.0°C, respectively with the highest temperature's being observed in June.
As for the prevailing wind speeds, the 2014 yearly average wind speeds at Kaohsiung, Meinong and Xiaogang were 8, 1.2 and 2 m s -1 while for 2015 they were 8, 1.2 and 2.1 m s -1 respectively.At Kaohsiung the annual average PM 2.5 , PM 10 and TSP levels were 31.0,66.0 and 81.5 µg m -3 in 2014 while for 2015 the levels were 26.0, 61.0 and 75.5 µg m -3 respectively.In 2014, the average annual PM 2.5 , PM 10 and TSP levels at Meinong were 25.0, 52.0 and 63.9 µg m -3 , respectively while in 2015 the levels were 23.0, 47.0 and 57.9 µg m -3 respectively.As for Xiaogang, the PM 2.5 , PM 10 and TSP annual average levels were 31.0,73.0 and 90.4 for 2014 and 29.0, 65.0 and 80.8 µg m -3 for 2015, respectively.The yearly average of PM 2.5 concentrations were all above the National Air Quality Standard of 15.0 µg m -3 for Taiwan.
In the whole period, January was the driest month while august was the month with highest rainfall intensity for all sampling sites.In 2014 Kaohsiung recorded the highest amount of annual rainfall (1940 mm) while in 2015 the highest rainfall amount was recorded at Meinong (1900 mm).

Simulated Ambient air PCDD/F Concentrations
Fig. 2 shows regression line showing the correlation between PM 10 and the total simulated PCDD/Fs mass concentrations at Xiaogang.The correlation coefficient in this study was 0.98 which is similar to the study of Chandra Suryani et al. (2015) and Huang et al. (2011a) who reported correlation coefficients of 0.99 and 0.94, respectively.Fig. 3 shows the regression line of the correlation of PM 2.5 with total simulated PCDD/Fs mass concentrations.Currently in the world, PM 2.5 levels have attracted the attention of many.In this study, the correlation coefficient of PM 2.5 and total PCDD/Fs mass was 0.92 which was lower than that of PM 10 vs. total simulated PCDD/Fs.During sample collection, the mass of PM 10 is greater than that of PM 2.5 which makes it more precise and reliable to do regression with the total mass concentration of PCDD/Fs.The strong correlations support the simulation of PCDD/Fs concentration data for the period 2014 and 2015.
Fig. 4 shows the simulated PCDD/F-WHO 2005 -TEQ concentrations and corresponding PM 10 and PM 2.5 .The simulated ambient air mean total PCDD/F mass concentrations at Kaohsiung during 2014 and 2015 were in the range of 0.33-1.20,0.33-1.13pg m -3 , respectively, and averaged 0.75 and 0.69 pg m -3 , respectively, while in terms of concentrations of toxicity equivalent quantity were in the range of 0.021-0.077and 0.021-0.072pg WHO 2005 -TEQ m -3 , respectively, and averaged 0.048 and 0.044 pg WHO 2005 -TEQ m -3 , respectively for 2014 and 2015.
During 2014 and 2015, simulated ambient air mean total PCDD/F mass concentrations at Xiaogang were in the Fig. 2. Regression between PM 10 (µg m -3 ) and total PCDD/F mass concentration (pg m -3 ) during sampling period in Xiaogang.
Comparing the simulated monthly PCDD/F-WHO 2005 -TEQ with the PM 10 and PM 2.5 levels in the ambient air shows that the concentrations were very much dependent on the levels of particulate matter.Therefore control of PM from the sources will subsequently lead to reductions in ambient dioxin levels.

Gas-Particle Partitioning of PCDD/Fs
The average seasonal gas-particle partitioning of PCDD/Fs for the whole of Kaohsiung are presented in the Fig. 5.A closer look at the seasonal gas partition profiles for Kaohsiung, Meinong and Xiaogang shows that higher chlorinated PCDD/Fs are primarily in the particle phase every season.Because of larger molecular weight, higher chlorinated PCDD/Fs have a lower vapor pressure, making which tend to be associated with particles in the ambient air (Wu et al., 2009a;Lin et al., 2010;Huang et al., 2011a, b).Additionally, gas phase of PCDD/Fs in winter shows higher partition comparing to the summer season, which is due to lower ambient air temperature in winter than in summer.From a previous study, particle phase of PCDD/Fs was found to increase with the reduction of temperature (Huang et al., 2011a) and as temperature rose, particle phase of PCDD/Fs would probably evaporate to the gas phase.These phenomenon caused by vapor pressure and ambient air temperature has been reported as the main factor influencing the partition of SVOCs (Pankow, 1987).2, 3, 4, 7, 8-P eC D F 1, 2, 3, 4, 7, 8-H xC D F 1, 2, 3, 6, 7, 8-H xC D F 1, 2, 3, 7, 8, 9-H xC D F 2, 3, 4, 6, 7, 8-H xC D F 1, 2, 3, 4, 6, 7, 8-H pC D F 1, 2, 3, 4, 7, 8, 9 2, 3, 4, 7, 8-P eC D F 1, 2, 3, 4, 7, 8-H xC D F 1, 2, 3, 6, 7, 8-H xC D F 1, 2, 3, 7, 8, 9-H xC D F 2, 3, 4, 6, 7, 8-H xC D F 1, 2, 3, 4, 6, 7, 8-H pC D F 1, 2, 3, 4, 7, 8, 9 The highest dry deposition fluxes at Xiaogang was in January in both years as well as Kaohsiung and Meinong, while the lowest dry deposition fluxes were in August in both years at three areas.Dry deposition fluxes strongly depend on the days without rainfall, the more days without rainfall, higher dry deposition are, due to no scavenging of rainfall from the ambient air.Previous studies mentioned that the total dry deposition flux was found to increase as the ambient air temperature decreased (Shih et al., 2006;Huang et al., 2011a).When ambient air temperature decreased in the winter season, the particle amount of PCDD/Fs would increase and thus the total dry deposition flux was larger in January than in August.In June 2014 and 2015, the number of days without rain was the highest in the respective years.This meant that the dry deposition dominated the deposition in the month of June and contributed a higher fraction to the total deposition.On the other hand, the particulate concentration in the month of June was low which meant the amount of PCDD/Fs available in particulate phase was lower.Additionally, the higher temperatures encountered in the month of June possibly encouraged volatilization of dioxins from particulate surfaces.Therefore, dry deposition, which is largely dependent on particulate phase, was lowest in June compared to other months.When comparing the annual dry deposition flux of total PCDD/Fs WHO 2005 -TEQ, the highest dry deposition was recorded at Xiaogang, followed by Kaohsiung, and then Meinong.Higher PM 10 concentrations at Xiaogang compared to Kaohsiung and Meinong were largely responsible for this observation.Table 4 further tabulates more values from other studies.
The dry deposition velocities were determined by using the annual total dry deposition fluxes divided by the average annual air concentrations.The resulting dry deposition velocities at Kaohsiung were determined as 0.167 cm s -1 and 0.157 cm s -1 for 2014 and 2015, respectively, and averaged at 0.162 cm s -1 .On the other hand, at Meinong, the calculated dry deposition velocities for 2014 and 2015 were 0.152 cm s -1 and 0.144 cm s -1 and averaged at 0.148 cm s -1 .At Xiaogang, the determined dry deposition velocities were 0.168 cm s -1 and 0.154 cm s -1 for 2014 and 2015, respectively and averaged at 0.161 cm s -1 .These values were lower than those reported for Hengchun (0.280 cm s -1 ) and Lulin (0.220 cm s -1 ) in our previous study.The dry deposition depends majorly on particle phase but for Kaohsiung, Meinong, and Xioagang greater fraction of PCDD/Fs WHO 2005 TEQ was in the more toxic lower molecular congeners as shown in the later section.

Monthly Wet Deposition Flux of Total PCDD/Fs-WHO 2005 -TEQ
Wet deposition of PCDD/Fs is the removal of both particle- The highest wet deposition fluxes were recorded in August for all the sampling areas since summer season has more rainfall amount compared to winter season which has less rainy days.By considering the annual total wet deposition and the annual total rainfall the concentrations in the rainfall were determined in terms of pg WHO 2005 -TEQ L -1 .At Kaohsiung, the corresponding concentrations in the rainfall were 0.263 and 0.307 pg WHO 2005 -TEQ L -1 for 2014 and 2015, respectively.As for Meinong, the concentrations in the rainfall were 0.166 and 0.167 pg WHO 2005 -TEQ L -1 for 2014 and 2015 respectively.For 2014 and 2015 the concentrations of PCDD/Fs in the rainfall at Xiaogang were the highest at 0.308 and 0.369 pg WHO 2005 -TEQ L -1 respectively.These values were one order of magnitude higher than the average concentrations previously recorded for Hengchun (0.064 pg WHO 2005 -TEQ L -1 ) and Lulin (0.027 pg WHO 2005 -TEQ L -1 ) (Chandra Suryani et al., 2015).Values of wet deposition from previous studies are tabulated in Table 5.
The corresponding scavenging ratios were estimated by dividing the total PCDD/Fs WHO 2005 -TEQ concentrations in the rainfall by the concentration of total PCDD/Fs WHO 2005 -TEQ in the ambient air.The corresponding scavenging ratios at Kaohsiung were 5490, and 6980 for 2014 and 2015, respectively and averaged at 6232.For Meinong the scavenging ratios were determined as 4490 and 4910 for 2014 and 2015 respectively and averaged at 4700.As for Xiaogang, the corresponding scavenging ratios calculate were 5820 and 7860 for 2014 and 2015 respectively with an average of 6840.The scavenging ratios at Kaohsiung Meinong, and Xiaogang were much lower than those reported for Hengchung (mean 8015) and Lulin (mean 13450) in our previous study (Chandra Suryani et al., 2015).For all the sites, the highest value occurred in winter season, especially in January, while the lowest one was probably in June during both years.These maximum and minimum values are similar to dry deposition fluxes but are opposite to wet deposition fluxes, which demonstrated the contributions of dry deposition fluxes, were much more than wet deposition fluxes during both years.
F w,p = C rain, particle × Rainfall (14) F w,T : the wet deposition flux of SVOCs from both vapor dissolution into the rain and removal of suspended particulates by precipitation, F w,dis : the wet deposition flux contributed by vapor dissolution into rain, F w,p : the wet deposition flux contributed by removal of suspended particulates by precipitation, Rainfall: monthly rainfall (m)

Fig. 4 .
Fig. 4. Simulated PCDD/F-WHO 2005 -TEQ concentration and corresponding PM 10 and PM 2.5 levels in the ambient air of Kaohsiung, Meinong and Xiaogang, respectively.
F p : the PCDD/F deposition flux contributed by the particle phase,

Table 5 .
Comparison for the wet deposition from previous studies and this study.