Current Combustion-Related Sources Contribute to Polychlorinated

University of Toronto, 200 College Street,. Toronto, Ontario, Canada, M5S 3E5, and Meteorological. Service of Canada, ARQP, 4905 Dufferin Street,. Dow...
63 downloads 12 Views 240KB Size
Environ. Sci. Technol. 2003, 37, 1075-1082

Current Combustion-Related Sources Contribute to Polychlorinated Naphthalene and Dioxin-Like Polychlorinated Biphenyl Levels and Profiles in Air in Toronto, Canada P A U L A . H E L M * ,† A N D TERRY F. BIDLEMAN‡ Chemical Engineering and Applied Chemistry, University of Toronto, 200 College Street, Toronto, Ontario, Canada, M5S 3E5, and Meteorological Service of Canada, ARQP, 4905 Dufferin Street, Downsview, Ontario,Canada, M3H 5T4

Polychlorinated naphthalenes (PCNs) and mono- and non-ortho substituted PCBs were analyzed in air from two sites in Toronto, Ontario, Canada to determine whether current combustion-related sources contribute to the levels and profiles of PCNs found in urban air. High-volume air samples were collected periodically at the University of Toronto (UT, a downtown site) and in north Toronto at the Meteorological Service of Canada (MSC). ΣPCN concentrations ranged from 31 to 78 pg m-3 at UT and from 7 to 84 pg m-3 at MSC with concentrations lower at MSC than UT for paired samples. Ambient air congener profiles contrasted between the two sites with MSC profiles indicating inputs from combustion-related sources when compared to combustion fly ash and technical PCN and PCB mixture profiles. Combustion markers, including CN-44, -29, and -54, the more toxic CN-66 and -67 congeners, and non-ortho PCBs, were enriched in air at MSC on a mass percent basis in several samples. As a result, CN-66/67 contributed proportionally more to dioxin toxic equivalents at MSC than at UT. Downtown air PCN profiles resembled those of technical PCN and PCB mixtures, reflecting evaporative emissions from past uses, while PCN levels and profiles at MSC, a more industrialized location, are also influenced by current combustion sources, contributing as much as an estimated 54% of ΣPCN in samples collected.

Introduction Polychlorinated naphthalenes (PCNs) were manufactured as complex mixtures for use in the electrical industry as dielectrics for flame resistance and insulation in cables, transformers, and capacitors, similar to the polychlorinated biphenyls (PCBs). Other uses included lubricants, wood preservatives, plasticizers, and binding agents, with total global production estimated to be 10% of polychlorinated biphenyl (PCB) production or 150 000 metric tons (1, 2). Although production and use are thought to have ceased in * Corresponding author phone: 204-983-7113; fax: 204-984-2403; e-mail: [email protected]. Current address: Freshwater Institute, Fisheries and Oceans Canada, 501 University Crescent, Winnipeg, Manitoba, Canada, R3T 2N6. † University of Toronto. ‡ Meteorological Service of Canada. 10.1021/es020860a CCC: $25.00 Published on Web 02/12/2003

Published 2003 by the Am. Chem. Soc.

most countries, other sources contribute to environmental PCN contamination. PCNs are trace contaminants in PCB mixtures (3, 4) and are produced in thermal processes such as municipal solid waste incineration (5, 6) and metals refining (7, 8). PCNs are of concern because of their toxicity, persistence, and bioaccumulation potential. They are proposed additions to the United Nations - Economic Commission for Europe Convention on Long-Range Transboundary Air Pollution Protocol on Persistent Organic Pollutants (UN-ECE LRTAP - POPs Protocol) list of banned or restricted chemicals (9). PCNs exhibit dioxin-like toxicity with potencies similar to those of non-ortho substituted PCBs (10-13). They are sufficiently persistent to reach remote regions via long-range transport as they have been measured in arctic air (14) and marine mammals (15, 16), and they have been found to bioaccumulate in the Baltic Sea food web (17). In the Great Lakes region, PCNs are found in Detroit River sediment (18), in fishes (19) and fish-eating birds (20), and in human adipose tissue (21). ΣPCN concentrations measured in ambient air are elevated in urban areas relative to those found at background sites. Levels averaged 68 pg m-3 in Chicago, IL (22), 138 and 160 pg m-3 for two sampling periods in Manchester, U.K. (23), and 60 pg m-3 in Augsburg, Germany (24), whereas concentrations ranged from 0.3 to 8 pg m-3 at remote arctic stations (14) and 1-10 pg m-3 at two sites in Sweden (25). In this study, two sites in urban Toronto, Canada were chosen to represent a dense commercial/residential sector built during peak usage of PCN and PCB formulations and a sprawling industrial/commercial/residential area likely to contain point emission sources. This paper reports concentrations of PCNs and non-/mono-ortho substituted PCBs found in air at these sites. The congener profiles of PCNs found in the air are compared to those of technical mixtures typically used and to combustion fly ashes from industries found within the Great Lakes region to determine which sources, current or past, predominate. The relative contributions of current combustion sources to PCN levels in air are estimated, and the influence of these sources on toxic equivalent contributions of PCNs and non-/mono-ortho PCBs is examined.

Methods The city of Toronto (43°28’N; 79°23’W) is located on the northwest shore of Lake Ontario. High-volume air samples (24-48 h; 700-1800 m3) were collected on the roof of the Gage Institute, University of Toronto (UT) in the downtown area and at the Meteorological Service of Canada (MSC) in the north part of the city. The sampling train consisted of a glass-fiber filter (GFF; 20.3 × 25.4 cm) followed by two cylindrical polyurethane foam plugs (PUF; 8.0 × 7.5 cm). Source-related samples included combustion fly ashes and technical PCB and PCN formulations. The fly ashes were from a municipal waste incinerator, a medical waste incinerator, a cement kiln, and an iron sintering plant. PCB formulations were Monsanto Aroclors 1254 and 1260. Technical PCN mixtures of varying degrees of chlorination were obtained from the U.S. EPA (Research Triangle Park, NC) and included Halowaxes 1014, 1013, 1099, 1000, and 1051. Extraction and Fractionation. Front PUF, back PUF, and GFF were Soxhlet extracted individually overnight using petroleum ether (PE) and dichloromethane (DCM), respectively, with 13C12-labeled PCB recovery surrogates (77, 81, 105, 126, and 169; Cambridge Isotope Laboratories Inc., VOL. 37, NO. 6, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1075

TABLE 1. Air Sample Information, ΣPCN and ΣNon-/Mono-Ortho PCB Concentrations, Toxic Equivalents, and CN-52,60 and CN-66,67 Isomer Fractions for Individual Samples Collected in Toronto, Ontario sample

site

sampling dates

G1 G2 G3 G4 G5 G6

UT UT UT UT UT UT

Mar. 19-20, 2000 Jul. 4-5, 2000 Jan. 12-13, 2001 Mar. 2-3, 2001 May 14-15, 2001 Jul. 24-25, 2001

D1 D2 D3 D4 D5 D6 D7 D8 D9 D10 D11 D12

MSC MSC MSC MSC MSC MSC MSC MSC MSC MSC MSC MSC

Sept. 30-Oct. 2, 1998 Sept. 30-Oct. 2, 1998 Dec. 1-3, 1999 Dec. 1-3, 1999 Dec. 21-23, 1999 Dec. 21-23, 1999 Jul. 6-7, 2000 Dec. 13-15, 2000 Jan. 13-15, 2001 Mar. 1-2, 2001 May 14-15, 2001 Jul. 23-24, 2001

volume (m3)

Temperature (°C) mean min./max.

Concentration (pg m-3) ΣPCN ΣN-/m-o-CB

TEQ (fg m-3)b ΣPCN ΣN-/m-o-CB

Isomer Fractions IF52,60 IF66,67

726 793 742 708 816 817

4.5 22.0 -0.6 -2.5 13.4 24.4

-0.8/9.2 16.7/26.1 -3.6/2.4 -6.1/0.8 8.0/17.7 18.7/32.6

30.9 52.4a 41.1 52.7 51.0 78.4a

5.96 29.5 6.92 6.43 12.6 29.4

0.50 1.73 1.18 1.34 1.16 2.20

0.28 2.57 0.16 0.32 0.76 2.09

0.175 0.165 0.153 0.152 0.154 0.149

0.586 0.560 0.581 0.613 0.594 0.582

1569 1477 1647 1818 1466 1598 879 1588 1702 744 909 777

12.5 12.5 1.7 1.7 -6.4 -6.4 17.3 -6.6 -0.3 -4.4 11.3 26.6

5.0/22.0 5.0/22.0 -10.0/12.0 -10.0/12.0 -4.0/-9.0 -4.0/-9.0 10.0/23.0 -17.0/1.0 -7.0/2.5 -8.0/-1.0 3.5/19.5 19.5/32.0

84.5 80.6 25.6 27.7 10.6 8.1 12.3a 28.0 20.6 7.3 14.3 51.0a

10.6 12.1 3.67 3.63 2.30 2.04 6.27 2.82 1.50 1.16 2.84 15.8

9.18 9.24 0.72 0.77 0.34 0.36 0.08 2.86 0.50 0.14 0.26 0.53

21.26 23.97 0.44 0.42 0.39 0.32 0.24 3.59 0.36 0.18 0.30 0.98

0.276 0.270 0.216 0.220 0.252 0.274 0.189 0.280 0.255 0.245 0.240 0.207

0.524 0.522 0.564 0.571 0.609 0.574 -c 0.509 0.566 0.523 0.541 0.591

a July ΣPCN values should be considered a minimum value as breakthrough of some triCNs (P2/P1 > 0.33) occurred at the warmer temperatures. TEFs used in TEQ calculations are from refs 10-12 for PCNs and ref 13 for PCBs. c CN-66/67 was detected but levels were too low to obtain a reliable IF66,67. b

Andover, MA) added prior to extraction. The extracts were reduced in volume by rotary evaporation, transferred to test tubes for reduction to 1 mL under a gentle nitrogen stream, and exchanged into isooctane. All samples were then fractionated on pre-rinsed silicic acid (SA) columns (3.0 g, 3% deactivated) topped with neutral alumina (2.0 g, 6% water) and anhydrous Na2SO4 (1 cm). PCNs and PCBs eluted in fraction 1 (F1; 30 mL PE) while organochlorine pesticides were in fraction 2 (F2; 30 mL DCM; archived). F1 was reduced in volume then fractionated on mini-columns containing 50 mg of SA, 100 mg of 20:1 SA/AX-21 activated carbon, 50 mg of SA, and topped with Na2SO4 (22). The multi- and some mono-ortho PCBs eluted in fraction 1 (F1-1; 5 mL of 30% DCM in cyclohexane), while the PCNs, non-ortho, and remaining mono-ortho PCBs eluted in fraction 2 (F1-2; 6 mL of toluene). F1-2 fractions for each of the front and back PUF and GFF were reduced in volume to approximately 200 µL for analysis. F1-1 for front PUF was also analyzed for monoortho PCBs (118, 114, 105, 156, and 13C12-105) and the amounts found were summed with those in F1-2. PCNs were separated from solutions (216 ng 1254 µL-1, 1096 ng 1260 µL-1) of the Aroclor mixtures in isooctane by adding 20-1000 µL to mini-carbon columns but eluting F1 with 10 mL of 40% DCM in cyclohexane and eluting F2 with 6 mL of toluene. The greater volume and higher percentage of DCM in the elution solvent relative to the air sample fractionation was required to adequately elute the greater amounts of PCBs in these formulations. Ash samples (3-5 g, 0.2), were weaker. This provides further evidence that non-ortho PCB sources differ between the two sites and indicates that nonevaporative sources influence non-ortho PCB levels at MSC. PCN correlations were inconclusive with insignificant slopes at UT and MSC (p > 0.1). Concentrations and homologue distributions of PCNs in source-related samples are listed in Table 2. ΣPCN concentrations ranged from 1.8 to 2.7 ng g-1 in 3 of the combustion fly ashes but were more than 1000 times higher (5.4 µg g-1) in the medical waste incinerator fly ash. The ΣPCN content of the municipal waste incinerator ash in this study is lower than values of 269 ng g-1 (tri- to octaCNs) (31) and 28 ng g-1 (tetra-heptaCNs) (32) found in other municipal waste incinerator fly ashes. TetraCNs were the dominant homologue in each fly ash (47-51%) but the more chlorinated pentaCNs (14-22%), hexaCNs (3.5-9.3%), and heptaCNs (0.24-6.1%), which contain several congeners having dioxinlike toxicity, were also abundant. The ΣPCN content in PCB mixtures averaged 197 µg g-1 of Aroclor 1254 and 156 µg g-1 of Aroclor 1260 (Table 2). These values are an order of magnitude lower than the 3500 and 2700 µg g-1 reported for Aroclors 1254 and 1260, respectively, by Haglund et al. (3). Yamashita et al. (4) found 47.2, 171, and 155 µg g-1 ΣPCN in Aroclor 1254 in three separate lots, and 67.2 µg g-1 in Aroclor 1260. Concentrations and the mass distribution (Table 2) of PCN homologues in Aroclors in this study agree well with those of Yamashita et al. (4). Congener Profiles. Mass percent distributions of ΣPCN differed between the UT and MSC sites. Figure 2a depicts examples of these profiles for penta- and hexaCN congeners. Similar profiles were found in UT samples throughout the VOL. 37, NO. 6, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1077

TABLE 2. Average ΣPCN Concentrations, Isomer Fractions, and Homologue Distribution in Combustion-Related and Technical Sources Isomer Fractions source sample

codea

ΣPCNb

IF52,60

IF66,67

Homologue Mass % Distribution triCN

tetraCN

pentaCN

53.5 55.8 59.1 47.4

22.0 15.1 13.7 16.5

hexaCN

heptaCN

octaCN

9.29 5.79 6.55 3.48

1.47 6.13 0.794 0.237

ND 8.56 ND 0.022

g-1)

municipal solid waste incinerator cement kiln iron scintering plant medical waste incinerator

MWI CK IS MED

1254 1260

Ar54 Ar60

1014 1014d 1013 1000 1099 1051 1051d

H1014 H1013 H1000 H1099 H1051

1.82 2.09 2.66 5439 197 156

Fly Ashes (ng 0.329 0.484 0.279 0.495 0.223 0.624 0.358 0.468

13.8 8.59 19.8 32.3

Aroclor Mixtures (µg g-1) 0.122 0.620 0.013 0.311 0.512 0.073 Halowax Mixturesc 0.120 0.656 0.066 0.743 0.241 0.594 0.313 0.325 0.108

0.587 0.618 0.935

9.75 8.20 44.1 41.8 0.004

0.146 0.198

8.21 0.234

13.4

37.2

48.9 47.4 48.1 0.004

38.0 8.44 9.64 0.052

51.2 0.655 35.3 4.75 ND 0.40 0.282

35.6 14.6

4.80 84.2

4.24

0.079

0.154 ND 0.027 9.64

0.005 ND 0.025 90.0

a Code used in PCA plots. b Concentrations given are averages of 2-5 replicate analyses of the same mixture or fly ash. c Mass % distribution of ΣPCN analyzed in this study (tri- to octaCNs). Lower chlorinated homologues not included. d Additional Halowaxes courtesy of U. Jarnberg (Stockholm University).

only 0.06 and 0.08% to ΣPCN in July 2000 and 2001. Similarly, CN-66/67 contributed 0.16-0.43% to ΣPCN at UT compared to 0.39-3.7% at MSC and only 0.13 and 0.15% in the two July MSC samples. Penta- and hexaCN congener distributions in representative source samples are illustrated in Figure 2b, and complete profiles may be found in the Supporting Information. PentaCNs 52/60, 51, and 54, and hexaCNs 66/67, are the most abundant congeners in their homologue group in medical and municipal waste incinerator fly ashes, and are relatively enhanced in the iron sintering and cement kiln ashes. These congeners were predominant within their homologue group in other fly ash studies (31, 33) and may be considered combustion indicators. Furthermore, the profiles in the municipal and medical waste incinerator ashes were typical of Stoker-type incinerators (33). Of the PCN and PCB mixtures examined, only in Aroclor 1260 was CN-66/67 the most abundant hexaCN, while congeners 69 and 71/72 were the dominant hexaCNs in the other technical mixtures. Site and source-specific differences were also apparent in the tri- and tetraCN homologue groups. TriCNs 20, 21, 26, and 13 were enhanced in the MSC samples and in combustion fly ashes, as were tetraCNs 44, 29, 27/30, and 39. Congeners 26, 13, 44, 29, and pentaCN-54 are particularly appropriate as combustion markers as they are absent or present in trace amounts in technical PCN and PCB mixtures analyzed in this study and reported previously (4, 29). The pentaCN distributions in UT samples resembled those of Halowax 1014 and Aroclor 1254, as did the hexaCN profiles, but with lower absolute mass percent values. FIGURE 2. Penta- and hexaCN mass percent distributions in (a) representative air samples from UT and MSC, and (b) selected source samples. Combustion-related congeners are represented by lightly shaded bars. year, but MSC profiles varied. In all UT and July MSC samples, congeners 52/60, 61, 57, 62, 53, and 59 were the most abundant pentaCNs, and CN-69 and 71/72 were the dominant hexaCNs. However, MSC samples collected at times other than during summer exhibited a relative enrichment of pentaCN-54 and hexaCN-66/67 (highlighted in Figure 2). The mass contribution of CN-54 at UT ranged from 0.06 to 0.22% compared to 0.21-2.6% at MSC. CN-54 contributed 1078

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 6, 2003

The normalized congener distribution (DREL), defined as the mass distribution (individual PCN concentrations/ΣPCN) at MSC (DMSC) divided by the UT mass distribution (DUT) for samples collected during the same periods, illustrates the enrichment of many of the combustion-related congeners in air at the MSC site (Figure 3). During July, the warmest month (up to 32 °C), only lighter components and not combustion-related congeners were enriched at MSC relative to UT. This indicates that evaporative PCN sources dominate during warm periods and implies that downtown Toronto is the source and MSC is more removed from that source, or that evaporative sources yield a lighter PCN distribution in the MSC area. Hepta- and octaCNs were enriched on filters at MSC compared to those at UT, possibly reflecting

FIGURE 3. Normalized mass percent distributions (DREL; MSC mass percent/UT mass percent in samples collected during the same time period) for samples collected in July and March 2001. Congeners with bars above the 1 line are enriched (have a greater mass contribution to ΣPCN) at the MSC site relative to that at UT, whereas bars below 1 are depleted. Asterisks indicate congeners included in regressions with log p°L to estimate percent of ΣPCN due to combustion (see text). Combustion-related congeners are depicted by the lightly shaded bars. differences in particulate matter loading and character at the two sites rather than between sources. However, not enough information was collected to consider particle-gas partitioning and aerosol characteristics. The DREL for non-ortho PCBs revealed enrichment at MSC relative to UT during the cooler months. DREL values for paired samples were 1.8-12.9 in May, March, and January, but were less than 1 in July for CB-126. For the mono-ortho PCBs, this effect was minimized as most congeners had a DREL near 1 (CB-118, 114, 105) or were depleted (CB-156) at MSC relative to UT. This indicates that similar sources may be contributing non-ortho PCBs to air in the MSC region as for PCNs. Nonortho PCBs are also produced in thermal processes such as municipal waste incineration (34). Comparison of PCB deposition fluxes to estimates of emission fluxes from waste incinerators showed similar fluxes for non-ortho PCBs, but mono-ortho PCB depositional fluxes were much higher (35). Levels of non-ortho PCBs in air have been found to correlate with polychlorinated dibenzofuran (PCDF) concentrations (36) which the authors suggest is due to both compounds having similar sources such as municipal waste incineration. ΣPCN also correlates with PCDF in fly ash from municipal waste incinerators (33). Regressions of the sum of combustion marker PCN (CN-20, 26, 13, 44, 29, and 54) concentrations with those of ΣNon- and ΣMono-ortho PCBs revealed moderate correlations at UT (r2 ) 0.40 and 0.43, respectively). At MSC however, combustion marker PCNs were strongly correlated with ΣNon-ortho PCBs (r2 ) 0.96) and only weakly correlated with ΣMono-ortho PCBs (r2 ) 0.16), indicating that non-ortho PCB levels at MSC result predominantly from combustion sources. Comparison of the source and air profiles shows contributions of combustion sources to PCNs in ambient air near MSC, while evaporative emissions from technical PCN or PCB mixtures are largely responsible for observed PCN levels in downtown Toronto. The downtown was previously more industrialized but is now dominated by high-density commercial and residential buildings built during peak PCN/ PCB usage. The outer areas of Toronto, where MSC is located, are characterized by low-density commercial and residential developments, but also industrial sectors which contain combustion point sources. Contributions from combustionrelated PCNs have been noted in ambient air in Japan (37)

and are increasing on a mass percent basis in recent soils relative to historical soils in the U.K. (38). The fraction of PCNs in air samples collected at MSC resulting from combustion emissions may be estimated from DREL for paired UT and MSC samples if evaporation from old-use PCN sources and in-use PCB mixtures are assumed to be the only PCN sources at UT. This assumption is supported by the lack of combustion markers in UT sample profiles. Samples collected at MSC that did not have a corresponding UT sample were paired with UT samples that were collected when temperatures were similar (e.g., October 1998 at MSC with May 2001 at UT) for calculation of DREL. The decline of DREL during the summer, when combustion sources have little influence on air at MSC, is related to the vapor pressures (p°L ) of the individual congeners. During cooler months, an underlying decline with decreasing vapor pressure also occurred if the combustion-related congeners are not considered. A linear regression of DREL with the temperature-dependent log p°L (39) for congeners seemingly unaffected by combustion emissions (marked with a * in Figure 3) allows the evaporation component (DE,MSC) of DMSC to be estimated from the slope for combustion-related congeners (DE,MSC ) DREL × DUT). The combustion component, DC,MSC, is the difference between the observed DMSC and the estimated DE,MSC for the combustion-related congeners. Hepta- and octaCNs were excluded because of their enrichment on particles at MSC, and other unmarked congeners (Figure 3) were excluded from the regression and estimates because they were near or below the MDL. The estimated percent of PCNs in air at MSC due to combustion ranged from -5 and -3% (assumed to equal zero) for July 2000 and 2001 to 54% in October 1998. On average, 15% of PCNs in samples from MSC were from combustion sources, although contributions may be episodic. For example, in addition to the 54% contribution in October 1998, the December 2000 sample showed 38% from combustion, whereas two sample sets in December 1999 showed 5 and 11%. A January 2001 sample had 6% of PCNs from combustion while March and May 2001 samples had 14 and 11%. The negative values for July provide some indication of the error in the estimate and suggest that the estimate may be biased low (conservative). VOL. 37, NO. 6, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1079

FIGURE 4. Isomer fractions (IF) of pentaCNs-52 and 60 and hexaCNs-66 and 67 in air (standard deviation indicated) and individual source samples. The IF is defined as the amount of CN-52 (or 66) divided by the amount of CN-52/60 (or 66/67). Isomer Fractions. PentaCNs-52/60 and hexaCNs-66/67 will become enriched within their homologue groups as they are atmospherically transported away from evaporative sources because they have higher vapor pressures than other congeners in their homologue groups, possibly confounding observations of combustion inputs. The CN-52/60 and -66/ 67 congeners coelute by most GC methods and are estimated to have the same vapor pressure (39). Separation of these congeners by GC (26) allows the use of isomer fractions (IF52,60 [amounts CN-52/CN-52+60] and IF66,67 [amounts CN-66/ CN-66+67]) to provide additional indicators of sources. IF52,60 and IF66,67 values are listed in Table 1 for air samples and Table 2 for source samples and compared in Figure 4. IF52,60 values were significantly higher at MSC than UT (one-tail t test; p < 0.00001), whereas IF66,67 values were higher at UT but with less significance (p < 0.02). Statistical significance could not be considered for source samples as only single batches were analyzed. Although there are clear differences in IF52,60 and IF66,67 between the sites, interpreting these results based on IF values in source samples is not conclusive. For instance, Halowax 1099 was the more commonly used PCN mixture (40) but its IF52,60 is closer to the MSC value than for UT. The Aroclor 1254 IF52,60 is closer to UT than MSC values and, as a commonly used PCB mixture, it may have more influence on the evaporative signature. The fly ashes tend to be closer to the MSC value, consistent with observations of combustion markers in these samples. Similar arguments can be made for IF66,67 with combustion values generally approaching MSC values, whereas Aroclor 1254 and all Halowaxes were more similar to UT values. Variation of IF52,60 and IF66,67 values between batches may also be substantial, further confounding interpretations. There is a large difference between the two Halowax 1051 batches for IF52,60 with intermediate values in air at MSC and UT. Variation of isomer fractions in several types and batches of source samples needs further examination to provide statistical validity to their use as source indicators. Principal Component Analysis. Complex profiles in source and environmental samples, and the influence of particular factors on observed differences, were examined by principal component analysis (PCA). Analyses were performed using Statistica for Windows (Statsoft, Inc., Tulsa, OK). Missing values were replaced with 2/3 the LOD for air samples and 2/3 blank values in source samples. Individual 1080

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 6, 2003

congeners or congener groups were normalized to mass percent contributions to ΣPCN (tetra- to octaCN due to breakthrough of triCNs in some samples) and ΣMono-/nonortho PCB. The number of principal components (PC) to retain was determined by the Scree test. Figure 5 presents the results (score and loading plots) from two analyses after varimax rotation. The first analysis included ΣPCN in source and ambient air samples (Figure 5a). Four PCs explaining 89% of the total variance were retained. PC 1 can be considered a combustion/ noncombustion component with combustion-related congeners having significant loadings (>0.7) along the positive axis. PCs 2-4 are related to the degree of chlorination, with PC 2 having significant loadings on noncombustion tetraCNs, PC 3 for pentaCNs, and PC 4 for hexaCNs. A plot of PC 1 versus PC 2 explains differences between the various source samples and begins to illustrate differences between air samples. The municipal and medical waste incinerator ashes are separated along PC 1 by the combustion congeners, while the cement kiln and iron sintering source samples exhibit the influence of the relatively higher abundance of noncombustion hexaCN congeners (CN-69, 71/72). MSC samples D1, D2, and D8, separated from other air samples along PC 1, show a strong influence of combustion, possibly from municipal waste incineration which is the closest source sample, confirming estimates of percent PCN due to combustion in these samples. A municipal waste incinerator, located to the west of the MSC sampling site, may have been influencing levels and profiles observed at MSC as prevailing winds are from the west. The technical mixtures are grouped along PC2 according to their degree of chlorination, with Aroclor 1260 and Halowax 1051 affected by loadings on the hepta- and octaCNs; Halowax 1014 and Aroclor 1254 affected by the hexaCNs, and Halowaxes 1013, 1000, and 1099 affected by the tetraCNs. Of the source samples included, this analysis indicates that the Halowax mixtures (1000, 1099, and 1013) are the predominant contributors to PCNs in ambient urban air. The chlorination degree appears to influence the groupings of the air samples with the higher-vapor-pressure tetraCNs (eg. CN-42, 33/34/37, 47) separating most MSC samples from the UT samples which are affected by the noncombustion pentaCNs-59, 61, 62, 57, and 53. Factors influencing groupings of air samples were further considered by excluding the source samples but including mono- and non-ortho PCBs. The 3 retained PCs explained

FIGURE 5. Score and loadings plots from principal component analysis of (a) source and ambient air samples, and (b) ambient air samples only. The score plots show groupings of individual samples whereas the corresponding loading plots indicate which congeners influence the scores. (Note: For the loading plot in (b), compounds within the ellipse are combustion-related congeners and include CNs-44, 36/45, 28, 29, 27/30, 39, 35, 52, 50, 51, 54, 49, 66, and 67 and non-ortho PCBs-81, 77, 126, and 169.) 85% of the variance, and a plot of PC 1 versus PC 2 further elucidated site-specific differences (Figure 5b). PC 1 is influenced by the relative abundance of combustion-related congeners, similarly to the previous analysis, clearly showing the negative correlation between combustion marker congeners and the tetraCNs especially, but also the penta- and hexaCNs most abundant in technical mixtures. Interestingly, the 4 non-ortho PCBs congeners all had similar loadings to marker PCNs (e.g., CN-29, 44, 51, 54, 66, and 67), illustrating their source as combustion. MSC samples D1, D2, and D8 were separated from all other air samples along PC 1. D7 and D12, both July samples, are removed from the remaining MSC samples along PC 1 due to greater evaporative inputs in the warmer months (i.e., loadings on tetraCNs-42, 33/ 34/37) and virtually no influence of combustion sources. UT air samples were grouped into the lower left quadrant along PC2 with loadings by pentaCNs-59, 61, 62, 57, and 53 and hexaCNs-69, 71/72, and 65 (Figure 5b), which are typical of evaporative sources such as Halowax 1014 and Aroclor 1254. Loadings on the most abundant tetraCNs (42, 33/34/37), CB105, and tetraCN-32 grouped most MSC samples in the upper left quadrant along PC 2. This analysis confirms that evaporative emissions dominate levels at both sites, but that combustion exhibits varying degrees of influence at MSC, and that the UT site, with a greater abundance of highermolecular-weight and lower-vapor-pressure PCNs found in technical mixtures, is closer to evaporative sources than MSC. Relative Toxicity. PCNs, non-ortho, and several monoortho substituted PCBs exhibit dioxin-like toxicity with the ability to bind to and activate the aryl hydrocarbon receptor. Toxic equivalency factors (TEFs) have been determined for several PCNs (10-12) and non-/mono-ortho PCBs (13) via similar H4IIE enzyme induction assays that allow for comparisons among these compounds on a toxic equivalent (TEQ) basis. TEQ contributions (fg m-3) of ΣPCNs and ΣNon-/mono-ortho PCBs in air at the two sites are summarized in Table 1. The geometric mean total PCN and PCB TEQ (PCN + PCB TEQ) contributions were 2.0 fg m-3 at UT

and 1.3 fg m-3 at MSC. On average, PCNs, mono-ortho, and non-ortho PCBs contributed 64%, 3%, and 33%, respectively, to PCN + PCB TEQ at UT, while their contributions were 48%, 2%, and 50% at MSC. Using H4IIE-TEFs to recalculate PCN + PCB TEQ reported in Chicago air (14), PCNs contributed 68% compared to 2% and 30% for the monoand non-ortho PCBs, respectively, similar to the UT site. PCNs also contributed as much as, or more than, PCBs to PCN + PCB TEQ in arctic air, up to 67% in the Eastern Arctic (14), and in fishes from the Detroit River (53%) (20). Chlorinated dioxin and furan (PCDD/F) TEQ, determined using H4IIE TEFs (13) and PCDD/F concentrations measured in air at UT (T. Dann, unpublished), ranged 13-165 fg m-3 for time periods similar to those of PCN/PCB sampling. The PCN + PCB TEQ accounts for 0.5-14% of total TEQ when PCDD/F contributions are included. Differences in congener distributions between UT and MSC influenced PCN TEQ. HexaCNs contributed >90% to average PCN TEQ in air at both UT and MSC, but CN-69 was more abundant at the UT site and contributed 37(5% of PCN TEQ compared to a contribution of 15(9% at MSC. CN-66/67 were the most abundant hexaCNs in most MSC samples, contributing 48(10% and 10(3% when separated into individual CN-66 and --67 congeners, respectively, whereas contributions to PCN TEQ by these congeners were significantly lower (23 ( 3% and 4 ( 1%) in air at UT (pe 0.001). PentaCN-54 was a minor TEQ component but contributed significantly more at MSC than at UT (p ) 0.002). Contributions of the more toxic CN-66 congener were enhanced when influences of combustion were greater. The PCB TEQ was dominated by CB-126 which contributed 84 ( 8% at UT and 90 ( 5% at MSC. CB-81 contributed 6% at each site, while CB-105 and -156 contributed 3.5 ( 1.9% and 4.5 ( 2.2%, respectively, at UT, and 1.5 ( 1.4% and 1.4 ( 1.0%, respectively, at MSC. Samples collected at MSC in October 1998 (D1, D2) and December 2000 (D8) had much more PCN + PCB TEQ than all other MSC samples and were higher than in UT air. CB-126/118 concentration ratios ranged VOL. 37, NO. 6, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1081

from 0.0011 to 0.0049 in UT samples, but were 0.20 and 0.10 in the D1/D2 and D8 samples, respectively. CB-126/118 ratios ranged from 0.0035 to 0.016 in the remaining MSC samples. The PCA analysis illustrated that samples D1, D2, and D8 were most influenced by combustion sources which emit CB-126. In summary, PCN and non-/mono-ortho PCB levels are higher in the downtown area of Toronto as a result of evaporative emissions of PCB and PCN mixtures from past uses and disposal. However, combustion sources also contribute to levels at MSC, which is located in the north part of Toronto where numerous industries and possible point sources are located, as indicated by the presence of combustion-related congeners in profiles and the influence of these congeners highlighted through PCA. MSC is indirectly influenced by evaporative sources by advective transport of air from downtown. TEQ contributions were higher at UT but were enhanced at MSC when combustion sources added much greater amounts of PCN and non-ortho PCBs. Additional studies of the profiles of source mixtures, particularly variation within a type of source, are required to further examine source contributions in ambient air.

Acknowledgments We thank C. Butt, B. Bahavar (UT), A. Leone, and T. Harner (MSC) for assistance with sampling. T. Dann and C. Chiu (Environmental Technology Centre, Environment Canada) provided fly ash samples and PCDD/F data for TEQ calculations. Additional Halowax samples were provided courtesy of U. Ja¨rnberg (Stockholm University, Sweden). Funding was provided in part by the Toxic Substances Research Initiative (Health Canada, Environment Canada), project 227 (M.L. Diamond, UT).

(12) (13)

(14) (15) (16) (17) (18) (19) (20) (21) (22) (23) (24) (25) (26) (27)

Supporting Information Available

(28)

Complete tri- to octaCN mass percent distribution profiles of source samples, which include fly ashes and technical PCN and PCB mixtures, are provided as Supporting Information. This material is available free of charge via the Internet at http://pubs.acs.org.

(29)

Literature Cited (1) Falandysz, J. Environ. Pollut. 1998, 101, 77-90. (2) Jakobsson, E.; Asplund, L. In The Handbook of Environmental Chemistry Vol. 3, Part K: New Types of Persistent Halogenated Compounds; Paasivirta, J., Ed.; Springer-Verlag: New York, 1999; pp 97-126. (3) Haglund, P.; Jakobsson, E.; Asplund, L.; Athanasiadou, M.; Bergman, A° . J. Chromatogr. 1993, 634, 79-86. (4) Yamashita, N.; Kannan, K.; Imagawa, T.; Miyazaki, A.; Giesy, J. P. Environ. Sci. Technol. 2000, 34, 4236-4241. (5) Oehme, M.; Manø, S.; Mikalsen, A. Chemosphere 1987, 16, 143153. (6) Abad, E.; Caixach, J.; Rivera, J. Chemosphere 1999, 38, 109-120. (7) Theisen, J.; Maulshagen, A.; Fuchs, J. Chemosphere 1993, 26, 881-896. (8) Aittola, J.-P.; Paasivirta, J.; Vattulainen, A.; Sinkkonen, S.; Koistinen, J.; Tarhanen, J. Chemosphere 1996, 32, 99-108. (9) van de Plassche, E. J.; Schwegler, A. M. G. R.; Balk, F. Preliminary Risk Profile - Polychlorinated Naphthalenes. Ministry of VROM/ DGM. Prepared for the UN-ECE Expert Group meeting on POPs, October 2001. (10) Hanberg, A.; Ståhlberg, M.; Georgellis, A.; de Wit, C.; Ahlborg, U. G. Pharmacol. Toxicol. 1991, 69, 442-449. (11) Blankenship, A. L.; Kannan, K.; Villalobos, S. A.; Villeneuve, D.

1082

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 6, 2003

(30) (31) (32) (33) (34) (35) (36) (37) (38) (39) (40)

L.; Falandysz, J.; Imagawa, T.; Jakobsson, E.; Giesy, J. P. Environ. Sci. Technol. 2000, 34, 3153-3158. Villeneuve, D. L.; Kannan, K.; Khim, J. S.; Falandysz, J.; Nikiforov, V. A.; Blankenship, A. L.; Giesy, J. P. Arch. Environ. Contamin. Toxicol. 2000, 39, 273-281. Giesy, J. P.; Jude, D. J.; Tillitt, D. E.; Gale, R. W.; Meadows, J. C.; Zajieck, J. L.; Peterman, P. H.; Verbrugge, D. A.; Sanderson, J. T.; Schwartz, T. R.; Tuchman, M. L. Environ. Toxicol. Chem. 1997, 16, 713-724. Harner, T.; Kylin, H.; Bidleman, T. F.; Halsall, C.; Strachan, W. M. J.; Barrie, L. A.; Fellin, P. Environ. Sci. Technol. 1998, 32, 3257-3265. Jansson, B.; Andersson, R.; Asplund, L.; Litze´n, K.; Nylund, K.; Sellstro¨m, U.; Uvemo, U.-B.; Wahlberg, C.; Wideqvist, U.; Odsjo ¨, T.; Olsson, M. Environ. Toxicol. Chem. 1993, 12, 1163-1174. Helm, P. A.; Bidleman, T. F.; Stern, G. A.; Koczanski, K. Environ. Pollut. 2002, 119, 69-78. Falandysz, J.; Rappe, C. Environ. Sci. Technol. 1996, 30, 33623370. Furlong, E. T.; Carter, D. S.; Hites, R. A. J. Great Lakes Res. 1988, 14, 489-501. Kannan, K.; Yamashita, N.; Imagawa, T.; Decoen, W.; Khim, J. S.; Day, R. M.; Summer, C. L.; Giesy, J. P. Environ. Sci. Technol. 2000, 34, 566-572. Kannan, K.; Hilscherova, K.; Yamashita, N.; Williams, L. L.; Giesy, J. P. Environ. Sci. Technol. 2001, 35, 441-447. Williams, D. T.; Kennedy, B.; LeBel, G. L. Chemosphere 1993, 27, 795-806. Harner, T.; Bidleman, T. F. Atmos. Environ. 1997, 31, 40094016. Harner, T.; Lee, R. G. M.; Jones, K. C. Environ. Sci. Technol. 2000, 34, 3137-3142. Do¨rr, G.; Hippelein, M.; Hutzinger, O. Chemosphere 1996, 33, 1563-1568. Wideqvist, U. Ph.D. Thesis, Stockholm University, Sweden, 1999. ISBN 91-7153-987-5. Helm, P. A.; Jantunen, L. M. M.; Bidleman, T. F.; Dorman, F. L. J. High Resolut. Chromatogr. 1999, 22, 639-643. Wiedmann, T.; Ballschmiter, K. Fresenius’ J. Anal. Chem. 1993, 346, 800-804. Kannan, K.; Imagawa, T.; Blankenship, A. L.; Giesy, J. P. Environ. Sci. Technol. 1998, 32, 2507-2514. Falandysz, J.; Kawano, M.; Ueda, M.; Matsuda, M.; Kannan, K.; Giesy, J. P.; Wakimoto, T. J. Environ. Sci. Health 2000, A35, 281298. Lopez Garcia, A.; den Boer, A. C.; de Jong, A. P. J. M. Environ. Sci. Technol. 1996, 30, 1032-1037. Schneider, M.; Stieglitz, L.; Will, R.; Zwick, G. Chemosphere 1998, 37, 2055-2070. Ja¨rnberg, U.; Asplund, L.; de Wit, C.; Egeba¨ck, A.-L.; Wideqvist, U.; Jakobsson, E. Arch. Environ. Contam. Toxicol. 1997, 32, 232245. Imagawa, T.; Lee, C. W. Chemosphere 2001, 44, 1511-1520. Sakai, S.; Hiraoka, M.; Takeda, N.; Shiozaki, K. Chemosphere 1993, 27, 233-240. Sakai, S.-I.; Hayakawa, K.; Takatsuki, H.; Kawakami, I. Environ. Sci. Technol. 2001, 35, 3601-3607. Kurokawa, Y.; Matsueda, T.; Nakamura, M.; Takada, S.; Fukamachi, K. Chemosphere 1996, 32, 491-500. Nakano, T.; Matsumura, C.; Fujimori, K. Organohalogen Compds. 2000, 47, 178-181. Meijer, S. N.; Harner, T.; Helm, P. A.; Halsall, C. J.; Johnston, A. E.; Jones, K. C. Environ. Sci. Technol. 2001, 35, 4205-4213. Lei, Y.-D.; Wania, F.; Shiu, W.-Y. J. Chem. Eng. Data 1999, 44, 577-582. Crookes, M. J.; Howe, P. D. Environmental Hazard Assessment: Halogenated Naphthalenes; Report TSD/13; Department of the Environment: London, 1993.

Received for review August 1, 2002. Accepted November 26, 2002. ES020860A