Behavior and source characteristic of PCBS in urban ambient air of Yokohama, Japan

Environmental Pollution 138 (2005) 290e298 www.elsevier.com/locate/envpol

Behavior and source characteristic of PCBS in urban ambient air of Yokohama, Japan Kyoung-Soo Kim a,b,*, Shigeki Masunaga b a

Center for Chemical Safety Management, Chonbuk National University, 664-14, Duckjin-dong, Chonju, 561-756, Korea b Graduate School of Environment and Information Sciences, Yokohama National University, 79-7, Tokiwadai, Hodogaya-ku, Yokohama, 240-8501, Japan Received 4 October 2004; accepted 17 March 2005

The relationship between the gas-particle partition coefficient (Kp) and sub-cooled liquid vapor pressure was estimated using gaseous and particle phase concentration in ambient air, and was estimated source apportionment of PCBs. Abstract To understand the behavior and sources of polychlorinated biphenyls (PCBs) in ambient air, gaseous and particulate phase concentrations were measured at Yokohama City, Japan, during March 2002 and February 2003. The concentration of total PCB and TEQ ranged from 62 to 250 pg/m3 and from 2 to 14 fgTEQ/m3, respectively. The gas-particle partition coefficient (Kp) was obtained as a function of temperature. The relationship between the partition coefficient and the sub-cooled liquid vapor pressure (PL) was also established (coefficients of determination for log Kp versus log PL plot were O0.76, except for three samples). As a result, the partition ratio of gaseous and particulate phase PCBs can be estimated for an arbitrary temperature. Principal component analysis (PCA) was applied to the source identification of PCBs in ambient air. The concentrations of 122 congeners between tetra-CBs and deca-CB were used as input variables, and three PCs with eigenvalue more than 10 were obtained. The principal component 1 (PC 1) accounted for 43.4% of the total variance, and was interpreted as volatilization from PCB products and/or sites polluted by PCBs. The concentrations of PCB congeners were strongly related with PC 1 which showed high correlation with temperature. PC 2 accounted for 22.3%, and was interpreted as PCBs from incineration sources, while PC 3 accounted for 10.8%, but could not be interpreted. Ó 2005 Elsevier Ltd. All rights reserved. Keywords: Polychlorinated biphenyls; Air; Gas-particle partitioning; Source identification; Principal component analysis

1. Introduction In Japan, about 59 thousand tons of polychlorinated biphenyls (PCBs) such as Kanechlor have been produced since 1954. These compounds were widely used for transformers, capacitors, printing inks and many * Corresponding author. Center for Chemical Safety Management, Chonbuk National University, 664-14, Duckjin-dong, Chonju, 561-756, Korea. Tel.: C82 63 270 2448; fax: C82 63 270 2449. E-mail address: [email protected] (K.-S. Kim). 0269-7491/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.envpol.2005.03.011

other applications. Their production was due to their thermal and chemical stability, and their electrical insulation properties. The production and new use of PCBs was banned in 1972. Presently, in spite of the ban on production and use of PCBs in many countries, PCBs are still found ubiquitously in the environment. Some studies have tried to identify the sources of PCBs pollution and have indicated that major sources of PCBs in the atmospheric environment are released by commercial PCB products and emitted by combustion processes (Brown et al., 1995; Simcik et al., 1997).

291

K.-S. Kim, S. Masunaga / Environmental Pollution 138 (2005) 290e298

Semi-volatile organic compounds (SOCs) such as PCBs are transported from sources to sinks primarily through the atmosphere. Therefore, ambient air is an important media for understanding the behavior of these compounds. The concentration of SOCs in the atmosphere is influenced by the partition equilibrium between the air and other media (soil, water, etc.), depending on the temperature and chemical properties of SOCs. Thus, partition is an important step in determining the fate of these chemicals in the environment (Pankow and Bidleman, 1992; Cotham and Bidleman, 1995). Several previous studies have reported the partition of SOCs; such as dioxins, PAHs, PCBs and pesticides (Junge, 1977; Yamassaki et al., 1982; Pankow, 1987; Finzio et al., 1997; Harner and Bidleman, 1998; Falconer and Harner., 2000; Mandalakis et al., 2002). Other factors such as relative humidity, wind speed and wind direction have also been found to play roles in determining ambient concentrations of PCBs (Burgoyne and Hites, 1993; Hornbuckle and Eisenreich, 1996). As mentioned above, temperature is a potent factor in PCB partitioning. Although some reports have studied the gas-particle partitioning of PCBs, these studies focus on dioxin like PCB congeners or selected PCB congeners. Moreover, some studies seldom report on particle fraction change dependent on temperature in individual PCB congeners. In this study, we researched the gasparticle partition coefficient as it effects temperature change in mono- to deca-chlorinated biphenyls (CBs). Since PCB has as many as 209 congeners, it is difficult to understand their behavior clearly as a whole from the obtained data. Thus, multivariate data analysis was used to solve this problem. Multivariate statistical techniques such as factor analysis (FA) and principal component analysis (PCA) have been used as an effective way to summarize and characterize the large amounts of data such as congener based PCB and dioxin concentrations. These methods have been used in source analysis and/or source apportioning in environmental media (Ogura et al., 2001; Sakurai et al., 2002; Imamoglu and Christensen, 2002). All the detected PCB congener concentrations in ambient air samples were subjected to multivariate statistical techniques in this study. The objectives of this study were: (1) to study season variation of PCBs in urban ambient air; (2) to grasp the behavior of PCBs in the atmosphere through gas/ particle partitioning, and (3) to identify the sources of PCBs in ambient air.

2. Experiment 2.1. Sampling Air samples were collected once a month for 7 days using a low volume air sampler (100 L/min) between

March 2002 and February 2003 in Yokohama City, Japan. Glass fiber filter (GFF) and two polyurethane foam plugs (PUF) were used to collect both the particulate and gaseous phase of PCBs, respectively. GFF and PUF were pre-cleaned by baking at 450
C for 4.5 h and by Soxhlet extraction with acetone for 8 h, respectively. Sampling details are given in Table 1. In addition, we conducted tests for possibility of breakthrough through separate analysis of PUF in July. The concentration ratio of the under PUF to the total PUF concentration was between 0.03 and 0.12. The results revealed that a possibility of breakthrough is slight.

2.2. Analytical method The analysis of the samples was performed according to US EPA method 1613 and 1668. Cleanup was conducted using sulfuric acid-impregnated silicagel column (sulfuric acid 44% w/w) and activated carbon column (Kanto Chemical Co., Inc., Japan). Quantification of the PCB congeners was conducted using a high-resolution gas chromatograph (Agilent HP6890) equipped with a high-resolution mass spectrometer (Micromass Autospec Ultima) with a DB-5 column (60 m ! 0.25 mm, i.d. ! 0.25 mm film thickness, J&W). The detailed analytical procedure is described in a previous paper (Kim et al., 2003). The average recoveries for PCB congeners (from tetra-CBs to deca-CBs) ranged from 50% to 85%. The detection limit was defined as three times the average mass measured in the blank test. In the data of PCB congeners, if the mass of the sample was above the detection limit, the mass of the blank was subtracted from the sample. The detection limits for individual PCB congeners in PUF and GFF ranged from 0.0008 to 0.26 ng and ranged from 0.0002 to 0.43 ng, respectively. Table 1 Ambient air sampling conditions Sample Sampling ID period

Mean Rainfall Mean TSP temperature (mg/m3)a (mm)b wind (
C) speed (m/s)

A1 A2 A3 A4 A5 A6 A7 A8 A9 A10 A11

13.8 18.4 18.5 22.2 29.7 29.1 22.5 14.5 11.6 7.3 7.9

a b

25 16 14 13 12 12 18 28 29 20 30

Mare1 Apr 02 Apre22 Apr 02 Maye21 May 02 June19 Jun 02 Jule22 Jul 02 Auge19 Aug 02 Sepe25 Sep 02 Octe5 Nov 02 Nove6 Dec 02 Dece27 Dec 02 Jane6 Feb 03

91.4 141.1 47.6 59.3 35.4 43.5 48.5 48.4 54.8 77.7 79.6

TSP is total suspended particles. Total precipitation during the sampling period.

71 38.5 40.5 116.5 85.5 124 9 8 25.5 29.5 0

1.78 3.58 1.38 1.36 3.78 2.29 1.19 1.36 1.08 1.28 1.35

292

K.-S. Kim, S. Masunaga / Environmental Pollution 138 (2005) 290e298

2.3. Statistical analysis Principal component analysis (PCA) was performed using Statistica for Windows 5.0J (StatSoft, Inc.). The PCB congeners with below-detection-limit concentration in more than half of the samples were removed from statistical analysis. As a result, a total of 122 individual and groups of congeners were subjected to PCA as input variables. The eigenvectors obtained were normal-varimax-rotated for better interpretation of the results.

3. Results and discussion 3.1. PCB concentrations in air PCB congener concentrations are given in Table 2. PCB congeners were expressed as their International Union of Pure and Applied Chemistry (IUPAC) numbers and TEQ concentrations were calculated using World Health Organization Toxic Equivalent Factors (WHO-TEFs) (Van den Berg et al., 1998). The concentration of total PCBs (from tetra-CBs to deca-CBs) and TEQ ranged from 62 to 250 pg/m3 and from 2 to 14 fgTEQ/m3, respectively. The monthly variation of PCB concentration is shown in Fig. 1. Many prior studies have shown the increase of PCB concentration with higher temperature (Manchester-Neesvig and Andren, 1989; Hoff et al., 1992; Halsall et al., 1995), however, the total PCB concentrations in July and August in this study were not

highest when temperature was highest. The two samples were collected under strong winds and heavy rainfall during a typhoon over 3 days. This may be the reason for the low PCB concentration in summer samples. Offenberg and Baker (1997) suggest that the rain is not an efficient scavenger of gas phase PCBs but does efficiently remove particulate phase PCBs. The lack of relationship between concentration and temperature has also been reported in some studies (Stern et al., 1997; Oehme et al., 1996). These studies were conducted in the Arctic region at a relatively low temperature (below 10
C). Thus, revolatilization of PCB was not substantial. In our study, however, average temperature was more than 10
C during the sampling period except for samples A10 and A11 (Table 1). We presume that meteorological parameters, namely, wind direction, wind speed and rainfall as well as a relatively long sampling period, are the cause of the weak relationship between total PCB concentration and temperature. The compositions of PCB homologue groups in air samples are shown Fig. 2. The ratios of tetra-CBs and penta-CBs to the total PCB concentration were largest, followed by hexa-CBsOhepta-CBsOocta-CBs. The ratio of the tetra-CBs in our study ranged from 53% to 76% of the total concentration. Bruncial et al. (2001) reported that the ratio of tri-CBs and tetra-CBs ranged from 70% to 90% in coastal New Jersey in 1997e1999. Halsall et al. (1995) reported that the ratio of tri- and tetra-CB was 50% in urban ambient air in the UK. The PCB homologue pattern was similar among samples (Fig. 2); the PCB isomer patterns in each

Table 2 Sum of gaseous and particulate PCB congener concentrations (pg/m3) IUPAC number

A1

A2

A3

A4

A5

A6

81 0.042 0.025 0.028 0.035 0.019 0.030 77 0.36 0.29 0.31 0.37 0.32 0.48 123 0.0051 0.057 0.049 0.19 0.10 0.12 118 1.4 2.0 2.6 2.5 2.0 3.5 114 0.074 0.084 0.074 0.095 0.11 0.13 105 0.58 0.78 0.69 0.88 0.87 1.6 126 0.051 0.022 0.025 0.027 0.015 0.026 167 0.056 0.054 0.052 0.059 0.066 0.073 156 0.098 0.099 0.093 0.11 0.097 0.15 157 0.035 0.029 0.033 0.029 0.035 0.029 169 0.011 0.0070 0.013 0.012 0.0034 0.0042 189 0.030 0.013 0.013 0.017 0.0069 0.0085 Tetra-CBs 53 67 67 67 55 76 Penta-CBs 17 31 32 32 24 48 Hexa-CBs 6.5 11 6.6 9.2 10 15 Hepta-CBs 2.1 1.8 1.4 1.9 2.5 3.1 Octa-CBs 0.31 0.28 0.26 0.40 0.45 0.33 Nona-CBs 0.10 0.084 0.094 0.080 0.053 0.037 Deca-CBs 0.074 0.058 0.058 0.047 0.03 0.026 Dioxin-like PCB 2.7 3.5 4.0 4.3 3.7 6.2 ) Total conc. 79 111 107 110 93 143 WHO-TEQ conc. 0.0055 0.0027 0.0031 0.0034 0.0020 0.0034 ) Total concentration means sum of concentration between tetra-CBs and deca-CBs.

A7

A8

A9

A10

A11

0.075 1.0 0.68 5.9 0.24 2.7 0.078 0.17 0.24 0.062 0.053 0.043 150 69 22 4.4 0.87 0.19 0.29 11 250 0.0097

0.095 0.72 0.16 2.4 0.079 0.83 0.059 0.11 0.15 0.053 0.055 0.051 82 30 10 1.8 0.62 0.21 0.20 4.7 125 0.0070

0.13 1.1 0.19 3.7 0.13 1.5 0.13 0.18 0.24 0.089 0.030 0.065 85 47 22 2.6 0.81 0.29 0.25 7.5 158 0.014

0.078 0.60 0.12 1.9 0.054 0.74 0.13 0.10 0.17 0.054 0.041 0.054 71 10 9.9 2.0 0.66 0.14 0.21 4.0 94 0.014

0.049 0.38 0.057 1.1 0.042 0.19 0.076 0.074 0.11 0.035 0.024 0.035 37 16 6.4 1.8 0.64 0.12 0.17 2.1 62 0.0081

293

K.-S. Kim, S. Masunaga / Environmental Pollution 138 (2005) 290e298 300

100

35 Total Conc

250

30

Temp

80 25

60

20

150 15

°C

pg/m3

200

40

100

10

50

5

0

0

Mar. Apr. May June July Aug. Sep. Oct. Nov. Dec. Jan. (A1) (A2) (A3) (A4) (A5) (A6) (A7) (A8) (A9) (A10) (A11)

Fig. 1. Monthly variation of total PCB concentration in air and mean temperature during the sampling period.

homologue were quite similar in all samples regardless of difference of sampling month (not shown in figures). This indicated that aerial PCBs at the sampling site were influenced either by one large emission source and/ or plural emission sources that have similar isomer compositions. As for dioxin-like PCBs, the levels of mono-ortho PCBs were higher than those of the non-ortho PCBs. PCB-118 was the predominant dioxin-like congener, followed by PCB-105 and PCB-77 in terms of concentration, while PCB-126 had the highest TEQ value among all the dioxin-like congeners (over 75% of total TEQ). 3.2. Gas/particle partition Fig. 3 represents the percentage of particulate phase PCB in each homologue at different temperature ranges. The ratio of the particulate phase increased with the increase in the number of substituted chlorines and with decrease of temperature during sampling. The tetra-CBs existed mostly in the gaseous phase regardless of temperature. Laboratory and field studies have shown that adsorption of gaseous phase to filter has been shown to occur (Foreman and Bidleman, 1987; Ligocki and Pankow, 1989). In this study, the amount by process that revolatilization of PCB captured to PUF and

20 0

4CB

5CB

6CB 5-15°C

7CB 15-20°C

8CB 20-25°C

9CB

10CB

25-30°C

Fig. 3. Percentage of particulate fraction (V) for each homologue at different temperature range. (V)ZKp!TSP/(1CKp!TSP).

adsorption of gaseous phase PCB to GFF were not considered. The gas-particle partition coefficient, Kp, was calculated using the following equation: ð1Þ

Kp ZðF=TSPÞ=A

where F (ng/m3) and A (ng/m3) are the particleassociated and gaseous concentrations of PCBs, respectively, and TSP (mg/m3) is the concentration of total suspended particulate matter. Yamasaki et al. (1982) suggested the above equation based on the assumption that the TSP concentration has a linear relationship with the surface area of particulates. The studies done by Pankow (1991) and Pankow and Bidleman (1992) have confirmed the validity of the above relationship through experimental and theoretical research. However, no agreement has yet been reached on what process (i.e. adsorption onto the surface, absorption into organic matter or a combination of both) is actually governing the process. The studies for these processes are reported now. The relationship between gas-particle partition coefficient and temperature is summarized by the regression in Table 3 for each homologue. This shows that the gasparticle partition of each homologue for an arbitrary temperature can be estimated from these results.

100 80

Table 3 Relationship between log Kp vs. 1000/T

60

Homologue

40 20 0

A1

A2 4CB

A3

A4

5CB

A5 6CB

A6

A7

A8

A9

7CB

8CB

9CB

10CB

A10

A11

Fig. 2. Composition of PCB homologues in air samples.

4CB 5CB 6CB 7CB 8CB 9CB 10CB

log KpZm(1000/T)Cb

P value

Slope (m)

Intercept (b)

Coefficient of determination (R2)

5.7 6.1 5.7 6.9 7.6 7.2 7.5

23.8 24.3 22.5 26.2 28.2 26.3 27.1

0.57 0.79 0.82 0.85 0.82 0.82 0.78

0.0075 0.0002 0.0001 0.0005 0.0001 0.0001 0.0003

294

K.-S. Kim, S. Masunaga / Environmental Pollution 138 (2005) 290e298

The relationship between gas-particle partitioning coefficient and sub-cooled liquid vapor pressure shown as Eq. (2) has been reported in previous studies (Pankow, 1987; Pankow and Bidleman, 1992; Lohmann et al., 2000): ð2Þ

log Kp Zm log PL Cb

where m and b are constants and PL is the sub-cooled liquid vapor pressure (Pa). This slope of m indicates the degree of the gas-particle partition equilibrium (Pankow and Bidleman, 1992), and the change of m value affects the value of the intercept b. The values of PL were calculated using data from Falconer and Bidleman (1994), and considering the detection limit and separation of congener cluster, the calculation of PL was conducted for the 41 congeners as follows: PCB-77, 81, 105, 110, 114, 118, 119, 122, 124, 126, 128, 129, 130, 137, 141, 153, 156, 157, 167, 169, 170, 174, 177, 179, 180, 183, 185, 189, 190, 193, 194, 195, 198, 199, 200, 201, 202, 206, 207, 208 and 209. Table 4 shows the relationship between log Kp observed in each ambient air sample and log PL and the values of m and b in Eq. (2) were calculated and are shown in Table 4 for each sample. The regressions of field data based on Eq. (2) gave a wide variance of m (0.48 to 0.97) although the sampling site and compound were the same in this study. This variance of slopes could be due to changes in the ambient temperature and TSP concentration (Lee and Jones, 1999; Oh et al., 2001). In this study, the range of ambient temperature and TSP concentration fluctuated over the sampling month from 7
C to 30
C and from 35 mg/m3 to 141 mg/m3, respectively. The values of m not close to 1 have been observed in other studies and are considered as deviation from non-equilibrium either due to sampling artifacts or analytical errors (Finizio et al., 1997; Lohmann et al., 2000; Pankow, 1994). Simcik et al. (1998) argued that slopes in the Table 4 Relationship between log Kp vs. log PL Sample ID

log KpZm log PLCb Slope (m)

Intercept (b)

Coefficient of determination (R2)

P value

A1 A2 A3 A4 A5 A6 A7 A8 A9 A10 A11

0.67 0.64 0.79 0.73 0.48 0.56 0.68 0.88 0.90 0.83 0.97

5.44 5.36 5.61 5.74 4.91 4.88 4.99 5.59 5.81 5.36 5.94

0.76 0.81 0.79 0.62 0.36 0.48 0.81 0.85 0.80 0.82 0.87

!0.00001 !0.00001 !0.00001 !0.00001 !0.00001 !0.00001 !0.00001 !0.00001 !0.00001 !0.00001 !0.00001

regression of log Kp vs log PL can be different from 1 in states of equilibrium and suggest that differences in particulate matter may be responsible for the shallow slopes. The coefficients of determination (R2) were low for samples collected during high temperature and heavy rainfall (A4, A5 and A6 in Table 3), but excellent R2 values more than 0.75 were obtained for other samples. Thus, temperature and TSP concentration, as well as rainfall, may have influenced partitioning. 3.3. Principal component analysis (PCA) and regression analysis PCA was used to study the sources of PCBs in ambient air. The plot of loading factors for the first and second principal components (PCs) is shown in Fig. 4. PC-1 and PC-2 accounted for 43.4% and 23.3% of the total variance, respectively. The congeners related to PC-2 with factor loadings greater than 0.8 were PCB198, 205, 189, 126, 81, 194, 169, 77, 162, 167, 157, 197, 203 and 196, 195, 208, 207, 206 and 209. The majority of these congeners were founded as specific congeners in the flue gas samples (Kim et al., 2004). Therefore, PC-2 could be interpreted as PCBs from combustion sources. Many PCB congeners that had a strong positive correlation with PC-1, particularly PCB-114, 149 and 139, 105, 118, 141, 70 and 76, 89 and 101 and 90, 90 and 95, 182 and 187, 47 and 75 and 48, 128, and 129, etc. showed factor loadings higher than 0.8. These were the same congeners that are known to be present characteristically in Kanechlor (Kim et al., 2004). The congeners related to PC-3 with loadings more than 0.6 were PCB-92, 125 and 116, 134, 56 and 60 and 140. However, PC-3 could not be interpreted. We interpreted PC-1 and PC-2 as representing Kanechlor and combustion sources, respectively. How different are these PCB congeners from two different sources in terms of environmental behavior? To answer this question, the relationship between the concentrations of these congeners located near the semicircle in Fig. 4 and environmental factors were examined. At first, these congeners were divided into three groups: group 1 congeners that had a high correlation (above 0.9) with PC-2 (PCB-126, 198, 206, 194, 203 and 196, 169, 81 and 157), group 2 congeners that were located in the upper right quadrant along the semicircle (PCB-77, 191, 167, 190, 172 and 192, 173, 170 and 156), and group 3 congeners that had high correlation (above 0.9) with PC-1 (PCB-171, 129, 128, 149 and 139, 118, 123, 114, and 105, etc.) (Fig. 4). The congeners belonging to group 2 were related to both the PC-1 and -2 (above 0.9). To understand the influence of environmental factors on this study, the relative concentration of each congener against PCB-189 was calculated. PCB-189

295

K.-S. Kim, S. Masunaga / Environmental Pollution 138 (2005) 290e298 1.0 189 126

81 208

206

157

167

207

203/196

169

Group 1

78

77 199

Group 2

156

138/164/163

0.6 132/168

Group 3

129

180 123 70/76

PC - 2 23.3

64/41/68

52/73

0.2

153

110

56/60 80/66 44

118/106 105/127

151 136 114 149/139 177 174 89/101/90 93/95 42 179 182/187

-0.2 Combustion KC300 and KC400 KC500 and KC600 Dioxin-like PCBs

-0.6 -0.6

-0.2

0.2

0.6

1.0

PC-1 43.4 Fig. 4. Loading factor plot for PC-1 and PC-2 after normal-varimax rotation. The square, triangle and cross mark represent congeners relate to combustion, KC300-400 and KC500-600, respectively (Kin et al., 2004).

had a high molecular weight (less volatile), had a high correlation with PC-2 (combustion) and was contained in Kanechlor only at a trace level. Therefore, it effectively minimizes the influence of fluctuation of combustion sources as well as environmental dilution due to the meteorological conditions at each sampling period. The relationship between the logarithm of each PCB congener concentration divided by the PCB-189 concentration (log CPCB-i/CPCB-189) and the reciprocal of absolute temperature (1/T ) was examined. The slopes and intercepts for the regression equations: log CPCB-i/ CPCB-189Za(1/T )Cb for three select groups of congeners are summarized in Table 5. As Table 6 indicates, the group 3 congeners had a steep absolute slope. This means that the concentrations of congeners belonging to group 3 were influenced by temperature more than those belonging to groups 1 and 2. The small absolute slopes for PCB-169 and PCB-126 indicated that concentrations of

those congeners were not affected by temperature. These results indicated that group 3 congeners originated from volatilization of PCB products (Kanechlor), while group 1 congeners were not influenced by volatilization and may have come mainly from combustion sources. The group 2 congeners had a moderate absolute slope and may have originated from both Kanechlor and combustion. For some congeners, the ratios of log CPCB-i/CPCB-189 with temperature change are shown in Fig. 5. The effect of temperature on the aerial concentration of dioxin-like PCB congeners was simulated using the regression equation (log (CPCB-i/CPCB-189)Za/TCb) results. The results of this simulation for dioxin-like PCB congeners in terms of both amounts and TEQ are shown in Fig. 6. Fig. 6 was drawn based on the following two assumptions: (1) the contribution from volatilization could be negligible at an atmospheric temperature of below 0
C, and (2) the increase of concentration with

Table 5 Slopes and regression coefficients of regressions between log (Cair/Cref) and 1/T for selected PCBs congeners. [log (Cair/Cref)Za/TCb]a Group 1

IUPAC no. Slope

169 126

126 184

206 728

198 1080

81 1087

194 1242

203/196 2005

157 2145

Group 2

IUPAC no. Slope

77 2341

191 2343

167 2358

190 2426

172/192 2518

173 2524

170 2555

156 2556

Group 3

IUPAC no. Slope

171 3294

129 3327

128 3377

149/139 3878

118 3904

123 4238

114 4264

105 4625

Group 1 congeners had a high correlation with PC-2. Group 3 congeners had a high correlation with PC-1 and group 2 congeners had a correlation with both PCs. a a and b are constants. PCB-189 was used as reference congener.

296

K.-S. Kim, S. Masunaga / Environmental Pollution 138 (2005) 290e298 2

1

Group 1 194

0.6 0.4 206

0.2

81

0 -0.2 -0.4

77

1

156

0.5

190

0 -0.5

173

-1

198

-0.6

0.00325 0.0033 0.00335 0.0034 0.00345 0.0035 0.00355 0.0036 0.00365

-1.5 0.00325 0.0033 0.00335 0.0034 0.00345 0.0035 0.00355 0.0036 0.00365

1/T

1/T

3

0.2

Group 3

2 1.5

1186

1 105

0.5 171 114

0

0.1

log(CPCB-i / CPCB-189)

2.5

log(CPCB-i / CPCB-189)

Group 2

1.5

log(CPCB-i / CPCB-189)

log(CPCB-i / CPCB-189)

0.8

0 169

-0.1 -0.2 -0.3 -0.4

-0.5 0.00325 0.0033 0.00335 0.0034 0.00345 0.0035 0.00355 0.0036 0.00365

-0.5 0.00325 0.0033 0.00335 0.0034 0.00345 0.0035 0.00355 0.0036 0.00365

1/T

1/T

Fig. 5. The relation between log(CPCB-i/CPCB-189) and temperature for the some congeners that belong to each group.

higher temperatures was caused solely by volatilization. The contribution from volatilization at 20
C was estimated to be about 7.5 times higher than that at 0
C in the case of PCB concentration. This increase of concentration was mainly caused by PCB-118 and PCB105 (Fig. 6a). On the other hand, the contributions from PCB-126 and PCB-118 were higher than other congeners in terms of TEQ concentration, however the contribution of PCB-169 was nearly constant in the temperature range of 1
C and 30
C (Fig. 6b). The

simulated trend of the contributions of volatilization for dioxin-like PCBs from this study is in agreement with the results of a previous study using the dioxin congener (2,3,4,7,8-penta-CDF) as a reference chemical (Ogura et al., 2002). In view of the results so far, it turned out that temperature was an important factor that ruled the aerial concentration of PCB congeners. The scatter locations of congeners that had high correlation either with PC-1 or PC-2 in factor loading plot (Fig. 4) were

Fig. 6. Simulation on the aerial concentration of dioxin-like PCB congeners by change of temperature (estimated using the regression equations given in Table 5). Others include PCB-81, 123, 114, 126, 167, 156, 157, 169 in 6-a, and PCB-81, 77, 123, 114, 167, 156, 157 in 6-b.

K.-S. Kim, S. Masunaga / Environmental Pollution 138 (2005) 290e298

decided by their sensitivity to temperature. Thus, PC-1 can be reinterpreted as the difference of volatilization cause by Kanechlor products subjected to changes of temperature. 4. Conclusion In this study, PCB congener concentration in the atmosphere of Yokohama City was studied with reference to meteorological parameters. It was found that partitioning behavior between gaseous and particulate phase was principally ruled by temperature except for lowly-chlorinated PCBs. The slope and intercept in the regression equation of log Kp versus log PL, however, varied depending on each sample, which could be influenced by other factors such as wind and rainfall. Although the relative composition of PCB isomers in ambient air was similar among monthly samples over 1 year, multivariate analysis successfully extracted the difference among samples. The results of PCA and regression analysis indicated that concentration of some PCB congeners is related to temperature due to volatilization from the Kanechlor product and/or media polluted by Kanechlor product. Other congeners were not related to temperature and may be caused by combustion sources. Some congeners were influenced by both. These results enabled us to simulate the effect of temperature on the concentration of each PCB congener. Acknowledgments This study was supported by Nippon Life Insurance Foundation and The 21st Century COE Program, Ministry of Education, Culture, Sports, Science and Technology, Japan. References Brown Jr., J.F., Frame, G.M., Olson, D.R., Webb, J.L., 1995. The sources of coplanar PCBs. Organohalogen Compounds 26, 427e430. Bruncial, P.A., Dachs, J., Gigliotti, C.L., Nelson, E.D., Eisenreich, S.J., 2001. Atmospheric polychlorinated biphenyl concentrations and apparent degradation in coastal New Jersey. Atmospheric Environment 35, 3325e3339. Burgoyne, T.W., Hites, R.A., 1993. Effects of temperature and wind direction on the atmospheric concentrations of a-Endosulfan. Environmental Science and Technology 27, 910e914. Cotham, W.E., Bidleman, T., 1995. Polycyclic aromatic hydrocarbons and polychlorinated biphenyls in air at an urban and a rural site near lake Michigan. Environmental Science and Technology 29, 2782e2789. Falconer, R.L., Bidleman, T.F., 1994. Vapor pressures and predicted particle/gas distributions of polychlorinated biphenyl congeners as functions of temperature and ortho-chlorine substitution. Atmospheric Environment 28, 547e554. Falconer, R.L., Harner, T., 2000. Comparison of the octanol-air partition coefficient and liquid-phase vapor pressure as descriptors

297

for particle/gas partitioning using laboratory and field data for PCBs and PCNs. Atmospheric Environment 34, 4043e4046. Finzio, A., Mackay, D., Bidleman, T., Harner, T., 1997. Octanol-air partition coefficient as a predictor of partitioning of semi-volatile organic chemicals to aerosols. Atmospheric Environment 31, 2289e2296. Foreman, W.T., Bidleman, T.F., 1987. An experimental system for investigating vapor-particle partitioning of trace organic pollutants. Environmental Science and Technology 21, 869e875. Halsall, C.J., Lee, R.G.M., Coleman, P.J., Burnett, V., HardingJones, P., Jones, K.C., 1995. PCBs in UK urban air. Environmental Science and Technology 29, 2368e2376. Harner, T., Bidleman, T.F., 1998. Octanol-air partition coefficient for describing particle/gas partitioning of aromatic compounds in urban air. Environmental Science and Technology 32, 1494e1502. Hoff, R.M., Muir, D.C.G., Grift, N.P., 1992. Annual cycle of polychlorinated biphenyls and organohalogen pesticides in air in southern Ontario. Atmospheric transport and sources. Environmental Science and Technology 26, 266e275 and 276e283. Hornbuckle, K.C., Eisenreich, S.J., 1996. Dynamics of gaseous semivolatile organic compounds in terrestrial ecosystem effects of diurnal and seasonal climate variations. Atmospheric Environment 30, 3935e3945. Imamoglu, I., Christensen, E.R., 2002. PCB sources, transformations, and contributions in recent Fox River, Wisconsin sediments determined from receptor modeling. Water Research 36, 3449e3462. Jung, C.E., 1997. In: Suffet, I.H. (Ed.), Fate of Pollutants in the Air and Water Environments, Part I. Wiley, New York, pp. 7e26. Kim, K.S., Hirai, Y., Masunaga, S., Nakanishi, J., Urano, K., Kato, M., 2004. Detailed PCB congener patterns in incinerator flue gas and commercial PCB formulations (Kanechlor). Chemosphere 55, 539e553. Lee, R.G.M., Jones, K.C., 1999. The influence of meteorology and air masses on daily atmospheric PCB and PAH concentrations at a UK location. Environmental Science and Technology 33, 705e712. Lee, R.G.M., Jones, K.C., 1999. Gas-particle partitioning of atmospheric PCDD/Fs: measurement and observation on modeling. Environmental Science and Technology 33, 3596e3604. Ligocki, M.P., Pankow, J.F., 1989. Measurements of the gas/particle distributions of atmospheric organic compounds. Environmental Science and Technology 23, 75e83. Lohmann, R., Lee, R.G.M., Green, N.J.L., Jones, K.C., 2000. Gasparticle partitioning of PCDD/Fs in daily air samples. Atmospheric Environment 34, 2529e2537. Manchester-Neesvlg, J.B., Andren, A.W., 1989. Seasonal variation in the atmospheric concentration of polychlorinated biphenyl congeners. Environmental Science and Technology 23, 1138e1148. Mandalakis, M., Tsapakis, M., Tsoga, A., Stephanou, E.G., 2002. Gas-particle concentration and distribution of aliphatic hydrocarbons, PAHs, PCBs and PCDD/Fs in the atmosphere of Athens (Greece). Atmospheric Environment 36, 4023e4035. Oehme, M., Haugen, J.E., Schlabach, M., 1996. Seasonal changes and relations between levels of organochlorines in Arctic ambient air: first results of an all-year-round monitoring program at Ny-A˚lesund, Svalbard, Norway. Environmental Science and Technology 30, 2294e2304. Offenberg, J.H., Baker, J.E., 1997. Polychlorinated biphenyls in Chicago precipitation: enhanced wet deposition to near-shore lake Michigan. Environmental Science and Technology 31, 1534e1538. Ogura, I., Masunaga, S., Nakanishi, J., 2001. Congener-specific characterization of PCDDs/PCDFs in atmospheric deposition: comparison of profiles among deposition, source, and environmental sink. Chemosphere 45, 173e183. Ogura, I., Masunaga, S., Yoshida, K., Nakanishi, J., 2002. Sources of PCBs in urban air and factors affecting their concentrations. Organohalogen Compounds 56, 197e200.

298

K.-S. Kim, S. Masunaga / Environmental Pollution 138 (2005) 290e298

Oh, J.E., Choi, J.S., Chang, Y.S., 2001. Gas/particle partitioning of polychlorinated dibenzo-p-dioxins and dibenzofurans in atmosphere; evaluation of predicting models. Atmospheric Environment 35, 4125e4134. Pankow, J.F., 1987. An absorption model of the gas/aerosol partitioning involved in the formation secondary organic aerosol. Atmospheric Environment 21, 2275e2283. Pankow, J.F., 1991. Common y-intercept and single compound regressions of gas-particles partitioning data vs 1/T. Atmospheric Environment 25A, 2229e2239. Pankow, J.F., Bidleman, T.F., 1992. Interdependence of the slopes and intercepts from log-log correlations of measured gas-particle partitioning and vapor pressure-1. Theory and analysis of available data. Atmospheric Environment 26A, 1071e1080. Pankow, J.F., 1994. An absorption model of gas/particle partitioning in the atmosphere. Atmospheric Environment 28, 185e188. Sakurai, T., Suzuki, N., Morita, M., 2002. Examination of dioxin fluxes recorded in dated aquatic-sediment cores in the Kanto region of Japan using multivariate data analysis. Chemosphere 46, 1359e1365. Simcik, M.F., Zhang, H., Eisenreich, St.J., Franz, Th.P., 1997. Urban contamination of Chicago/Coastal Lake Michigan atmosphere by

PCBs and PAHs during AEOLOS. Environmental Science and Technology 31, 2141e2147. Simcik, M.F., Franz, T.P., Zhang, H., Eisenreich, S.J., 1998. Gasparticle partitioning of PCBs and PAHs in the Chicago urban and adjacent coastal atmosphere: state of equilibrium. Environmental Science and Technology 32, 251e257. Stern, G.A., Halsall, C.J., Barrie, L.A., Muir, D.C.G., Fellin, P., Rosenberg, B., Rovinsky, F.Y.A., Kononov, E.Y.A., Pastuhov, B., 1997. Polychlorinated biphenyls in Arctic air. 1. Temporal and spatial trends: 1992e1994. Environmental Science and Technology 31, 3619e3628. Van den Berg, M., Birnbaum, L., Bosveld, A.T.C., Brunstro¨m, B., Cook, P., Feeley, M., Giesy, J.P., Hanberg, A., Hasegawa, R., Kennedy, S.W., Kubiak, T., Larsen, J.C., van Leeuwen, F.X.R., Liem, A.K.D., Nolt, C., Peterson, R.E., Poellinger, L., Safe, S., Schrenk, D., Tillitt, D., Tysklind, M., Younse, M., Waern, F., Zacharewski, T., 1998. Toxic equivalency factors (TEFs) for PCBs, PCDDs, PCDFs for humans and wildlife. Environmental Health Perspectives 106 (12), 775e792. Yamassaki, H., Kuwata, K., Miyamoto, H., 1982. Effects of ambient temperature on aspects of airborne polycyclic aromatic hydrocarbons. Environmental Science and Technology 16, 189e194.