Discontinued and Alternative Brominated Flame Retardants in the

Sep 26, 2011 - Seasonality and indoor/outdoor relationships of flame retardants and PCBs ... Statewide surveillance of halogenated flame retardants in...
1 downloads 0 Views 3MB Size
ARTICLE pubs.acs.org/est

Discontinued and Alternative Brominated Flame Retardants in the Atmosphere and Precipitation from the Great Lakes Basin Amina Salamova and Ronald A. Hites* School of Public and Environmental Affairs, Indiana University, Bloomington, Indiana 47405, United States

bS Supporting Information ABSTRACT: Air (vapor and particle) and precipitation samples were collected at five sites (two urban, one rural, and two remote) around the Great Lakes during 20052009 as a part of the Integrated Atmospheric Deposition Network (IADN). The concentrations of polybrominated diphenyl ethers (PBDEs), decabromodiphenylethane (DBDPE), hexabromobenzene (HBB), pentabromoethylbenzene (PBEB), and 1,2-bis(2,4,6-tribromophenoxy)ethane (BTBPE) were measured in these samples. The highest concentrations of these compounds were generally observed at the two urban sites—Chicago and Cleveland—with a few exceptions: The remote site at Eagle Harbor had particularly high levels of PBEB in all three phases, and the rural Sturgeon Point site had the highest HBB concentrations in the vapor phase. The sources of HBB and PBEB to these sites are unknown. A multiple linear regression model was applied to the concentrations of these compounds in the vapor phase, particle phase, precipitation, and the three phases combined. This regression resulted in overall (three phases combined) halving times for total PBDE concentrations of 6.3 ( 1.1 years. The overall halving times for HBB and BTBPE were 9.5 ( 4.6 years and 9.8 ( 2.8 years, respectively. For PBEB and DBDPE, the regression was not statistically significant for the combined phases, indicating that the atmospheric concentrations of these compounds have not changed between 2005 and 2009.

’ INTRODUCTION The rapid development of the polymer industry over the last 100 years eventually led to requirements that some of these materials be flame resistant. For example, polyurethane foam used in furniture must be able to withstand an open flame for 12 s.1 These flame resistant properties are imparted to the polymeric materials by adding percent level amounts of flame retardants. Over the last several decades, this has been a growth market for the chemical industry, and there are now more than 175 chemicals classified as flame retardants. Of these, at least 75 are brominated flame retardants (BRFs).2 Among them, the polybrominated diphenyl ethers (PBDEs) have been widely used and are persistent and accumulate in the environment.3,4 As a result, the use of the penta-, octa-, and deca-BDE commercial mixtures was restricted in the European Union,5,6 the production and use of the penta- and octa-mixtures in the United States was voluntarily phased out in 2004, and the production, import, and sale of the deca-BDE mixture in the United States will be discontinued by the end of 2013.7 These restrictions have led to an increased market demand for nonregulated flame retardants—or at least, flame retardants that are not in the news. These include decabromodiphenylethane (DBDPE) and 1,2-bis(2,4,6-tribromophenoxy)ethane (BTBPE), both of which have been marketed as alternatives to various PBDE r 2011 American Chemical Society

formulations. In addition, “older” chemicals are sometimes being reintroduced to the market. These include hexabromobenzene (HBB) and pentabromoethylbenzene (PBEB), compounds that have apparently been manufactured for several decades. Although little is known about their uses or production volumes, the environmental presence of several alternative flame retardants has been reported recently.813 Interestingly, de Wit et al.10 have detected several alternative flame retardants in the Arctic, which suggests long-range atmospheric transport of some of these chemicals. The goal of this study is to provide insights on the long-term spatial and temporal distribution patterns of PBDEs, DBDPE, HBB, PBEB, and BTBPE in the atmosphere (vapor and particle phases) and in precipitation at the five United States Integrated Atmospheric Deposition Network (IADN) sites located in the Great Lakes basin. To elucidate temporal trends in these chemicals’ concentrations, we used a harmonic regression approach14 that includes local population and distance from an assumed source to model the measured concentrations in each Received: June 15, 2011 Accepted: September 7, 2011 Revised: August 18, 2011 Published: September 26, 2011 8698

dx.doi.org/10.1021/es2020378 | Environ. Sci. Technol. 2011, 45, 8698–8706

Environmental Science & Technology

ARTICLE

phase (vapor, particle, and precipitation) separately and in all the phases combined at all sites.

instrumental analysis and quality control and quality assurance procedures are provided in the Supporting Information.

’ EXPERIMENTAL SECTION The locations of the United States IADN sampling sites are shown in the Supporting Information (SI) (Figure S1). The two urban sites are in Chicago, IL (41.8344°N, 87.6247°W) and Cleveland, OH (41.4921°N, 81.6785°W). A rural site is located at Sturgeon Point, NY (42.6931°N, 79.0550°W). The two remote sites are at Sleeping Bear Dunes, MI (44.7611°N, 86.0586°W) and Eagle Harbor, MI (47.4631°N, 88.1497°W). The IADN Web site provides detailed information on air sampling procedures and site operations (www.msc.ec.gc.ca/iadn). The samples discussed here were collected during the period of January 1, 2005 to December 31, 2009. A modified Anderson high-volume air sampler (General Metal Works, model GS2310) was used to collect air samples for 24 h every 12 days at a flow rate giving a total sample volume of about 820 m3. The vapor phase was collected on Amberlite XAD-2 resin (Supelco, Bellefonte, PA; 2060 mesh) held in a stainless steel cartridge, and particles were collected on Whatman quartz fiber filters (QM-A, 20.3  25.4 cm). Details of the sampling procedures and site operations can be found elsewhere.15 Precipitation samples were collected using MIC automated wet-only samplers (MIC Co., Thornhill, ON). Each sampler consists of a 46  46 cm stainless steel funnel connected to a 30 cm long by 1.5 cm i.d. glass column (ACE Glass, Vineland, NJ) packed with XAD-2 resin. The sampler is normally covered; it opens when a precipitation event is sensed by a conductivity grid located outside of the sampler. Precipitation flows through the funnel and the XAD-2 column into a large carboy used to measure the total precipitation volume. Both particulate and dissolved organic phase compounds are collected by the XAD-2 column. The funnel and the interior of the sampler are kept at 15 ( 5 °C to melt snow collected in the sampler and to prevent the XAD column from freezing. Details on the performance of these samplers are provided elsewhere.16 Precipitation events are integrated over each calendar month. A detailed description of the sample treatment and chemical analysis procedures for the air and precipitation samples has been given elsewhere.17 In summary, the samples were Soxhlet extracted for 24 h with a 1:1 acetone hexane mixture. Prior to extraction, a recovery standard was spiked into the sample; this standard included known amounts of BDE-77, BDE-166, and 12 13C -BDE-209. The extract was reduced in volume by rotary evaporation, the solvent was exchanged to hexane, and fractionated on a column containing 3.5% w/w water deactivated silica gel (3.0% w/w for precipitation samples). This column was eluted with 25 mL of hexane (fraction 1) and 25 mL of a 1:1 hexane dichloromethane mixture (fraction 2). After N2 blow down, the samples were spiked with the quantitation internal standard, consisting of a known amount of BDE-118. For chemical analysis, the samples were further concentrated by N2 blow down to ∼100 μL. The samples were analyzed for 34 PBDE congeners (7, 10, 15, 17, 28, 30, 47, 66, 85, 99, 100, 119, 126, 138140, 153, 154, 156, 169, 180, 183, 184, 191, 196, 197, 201, and 203209), and for decabromodiphenylethane (DBDPE), hexabromobenzene (HBB), pentabromoethylbenzene (PBEB), and 1,2-bis(2,4,6-tribromophenoxy)ethane (BTBPE) on an Agilent 6890 series gas chromatograph coupled to an Agilent 5973 mass spectrometer using helium as the carrier gas. Details on the

’ RESULTS AND DISCUSSION Polybrominated Diphenyl Ethers. PBDEs were found in the gas, particle, and precipitation phases. Figure 1 shows the spatial distributions of total PBDE (∑PBDE), BDE-47, 99, and 209 concentrations in the vapor, particle, and precipitation phases at the five U.S. IADN sites. Table 1 gives the arithmetic mean and standard errors of these concentrations along with ANOVA results (applied to log-transformed concentrations) and percent detected. The highest ∑PBDE concentrations were detected at the two urban sites (Chicago and Cleveland) in all three phases. The mean ∑PBDE concentrations in the vapor and particle phases at the rural Sturgeon Point (SP) site and at the remote Sleeping Bear Dunes (SB) site were lower but statistically indistinguishable from each another. The lowest ∑PBDE concentrations in the vapor and particle phases were measured at the remote Eagle Harbor (EH) site. In precipitation, there were no statistical differences among the ∑PBDE concentrations measured at the three rural and remote sites. Among the studies that have measured PBDE concentrations in the atmosphere only a few have reported concentrations for vapor and particle phases separately; this limits direct comparison with our results. In addition, not all of these previous studies have included BDE-209 in the reported ∑PBDE concentrations. Nevertheless, the spatial distribution pattern in ∑PBDE concentrations observed here was similar to the one measured in a previous study from our laboratory, where the ∑PBDE concentrations (sum of vapor plus particle phase concentrations) were shown to be strongly correlated with the human population density at a given site.18 In this previous study, ∑PBDE concentrations varied from a low of 5.8 ( 0.4 pg/m3 at the remote Eagle Harbor site to a high of 87 ( 8 pg/m3 at the urban Cleveland site. The levels of ∑PBDE concentrations (sum of vapor plus particle phase concentrations) in air samples collected in 19971999 at the same Great Lakes sites, except Cleveland, were ∼50 pg/m3 for Chicago and 515 pg/m3 at the other rural and remote sites.19 All of these levels were similar to our results. Salamova and Hites20 reported average ∑PBDE concentrations of 34 ( 3.5 pg/m3 and 23 ( 2.3 pg/m3 in Chicago and 26 ( 2.6 pg/m3 and 58 ( 6.9 pg/m3 in Cleveland in the vapor and particle phases, respectively, in air samples collected in 20032007 from the same Great Lakes sites. The levels at the rural and remote sites were in the range of 24 pg/m3 for the vapor phase and 25 pg/m3 for particles. ∑PBDE concentrations in precipitation ranged from ∼65 ng/L at the urban sites to ∼1 ng/L at the rural and remote sites. These levels were similar to our values. Our current values are also in the range of reported ∑PBDE concentrations in precipitation samples collected at the same Great Lakes sites in 20052006.21 On the global scale, our vapor and particle ∑PBDE concentrations for urban sites are similar to those reported at urban and industrial sites in Izmir, Turkey (929 pg/m3 in the gas phase and 2762 pg/m3 in particles),22 at a site near a sanitary landfill in Ottawa, Canada (∼20 pg/m3),23 and after a dust storm event in Kuwait (4754 pg/m3).24 The atmospheric ∑PBDE concentrations reported for rural sites in China are significantly higher than our values (210690 pg/m 3 ); 25 however, the values reported in Chinese precipitation samples are within the concentration range measured in this study (962 ng/L).25 8699

dx.doi.org/10.1021/es2020378 |Environ. Sci. Technol. 2011, 45, 8698–8706

Environmental Science & Technology

ARTICLE

Figure 1. Concentrations of ∑PBDE, BDE-47, BDE-99, and BDE-209 in the atmospheric vapor phase, particle phase, and precipitation; the thin black lines represent the median, and the thick red lines represent the arithmetic mean; the boxes represent the 25th and 75th percentiles; the whiskers represent the 5th and 95th percentiles. Site abbreviations: EH Eagle Harbor, CH Chicago, SB Sleeping Bear Dunes, CL Cleveland, SP Sturgeon Point.

Table 1. Arithmetic Mean ( Standard Error of ∑PBDE, BDE-47, BDE-99, and BDE-209 Concentrations in Vapor Phase (pg/m3), Particle Phase (pg/m3), and Precipitation (ng/L) at the Five U.S. IADN Sitesa Eagle Harbor conc.

*

Chicago

% det

conc.

*

Sleeping Bear Dunes % det

conc.

*

% det

Cleveland conc.

*

Sturgeon Point

% det

conc.

*

% det

∑PBDE

vapor

2.6 ( 0.4

c

100

35 ( 4

a

100

4.8 ( 0.7

b

100

25 ( 2

a

100

7.2 ( 1.6

b

100

BDE-47

vapor

1.1 ( 0.2

c

64

18 ( 1.8

a

94

1.7 ( 0.3

bc

83

14 ( 1

a

98

1.5 ( 0.2

b

73

BDE-99 BDE-209

vapor vapor

1.2 ( 0.3 0.5 ( 0.1

b c

59 65

6.4 ( 1.2 3.4 ( 0.9

a a

95 51

1.4 ( 0.3 0.8 ( 0.3

b c

80 66

4.8 ( 0.6 1.8 ( 0.5

a b

86 34

1.2 ( 0.4 0.7 ( 0.3

b c

72 76

∑PBDE

particle

3.2 ( 0.6

d

100

25 ( 3

a

100

4.3 ( 0.7

c

100

62 ( 13

a

100

8.2 ( 0.7

bc

100

BDE-47

particle

0.7 ( 0.2

b

42

4.5 ( 0.5

a

88

0.5 ( 0.1

b

68

4.5 ( 0.7

a

58

0.7 ( 0.1

b

74

BDE-99

particle

1.4 ( 0.5

b

23

3.9 ( 0.8

a

88

1.0 ( 0.1

b

80

7.3 ( 1.5

a

50

1.2 ( 0.2

b

70

BDE-209

particle

1.3 ( 0.4

d

64

13 ( 2

b

96

2.5 ( 0.9

c

69

56 ( 15

a

86

1.9 ( 0.2

c

81

∑PBDE BDE-47

precip precip

2.1 ( 0.4 0.9 ( 0.4

cd bc

100 100

42 ( 8 20 ( 4

a a

100 100

5.2 ( 1.2 2.5 ( 0.6

c b

100 100

6.4 ( 1.5 0.9 ( 0.2

b b

100 100

1.5 ( 0.3 1.3 ( 0.8

d c

100 100

BDE-99

precip

0.2 ( 0.02

b

100

8.7 ( 1.8

a

100

1.0 ( 0.4

b

100

0.7 ( 0.2

b

100

0.1 ( 0.02

c

100

BDE-209

precip

0.4 ( 0.1

b

97

2.1 ( 0.3

a

100

0.5 ( 0.1

b

100

4.1 ( 1.3

a

98

0.6 ( 0.1

b

100

a

ANOVA results for the log-transformed concentrations are given in the columns headed by an asterisk; the concentrations are not significantly different for those locations sharing the same letter. Percent of detects for each compound in each phase is also shown.

The ∑PBDE concentrations (excluding BDE-209) measured in the atmosphere over the Southern and Atlantic Oceans are similar to those reported here for the remote sites (1.1 pg/m3 for the gas phase and 0.3 pg/m3 for particles).8 ∑PBDE concentrations

reported in precipitation samples from the Baltic Sea26 and from Sweden27,28 are also similar to our values. BDE-47, -99, and -209 were the main components of the heavily used PBDE commercial mixtures, and as a result, these 8700

dx.doi.org/10.1021/es2020378 |Environ. Sci. Technol. 2011, 45, 8698–8706

Environmental Science & Technology

ARTICLE

congeners comprise up to 99% of the measured ∑PBDE concentrations (see SI Figure S2). Similar to the ∑PBDE spatial distribution pattern, the highest levels of these individual congeners were detected at the urban sites of Chicago and Cleveland, and the levels at the other three sites were generally indistinguishable from one another. Interestingly, the concentrations of BDE-209 in the particle phase were particularly high at Cleveland (56 ( 15 pg/m3). These elevated levels of BDE-209 in the particle phase at Cleveland have been shown to be associated with local sources.18 The levels of BDE-47, 99, and 209 detected in this study are in the range of the PBDE congener concentrations measured in air and precipitation samples collected in 20032007 at the same Great Lakes sites.20 Overall, BDE-47 is the relatively most abundant congener in the vapor phase, and BDE-209 is the relatively most abundant congener in the particle phase. This is true at all five sites except Sturgeon Point (see SI Figure S2). In precipitation, BDE-47 is the most abundant PBDE congener at all the sites, except Cleveland, where BDE-209 is the most abundant PBDE congener. This pattern is different from the pattern observed in precipitation samples in previous studies from our laboratory.20,21 In these previous studies, BDE-209 was the relatively most abundant congener in precipitation at all the sites except Chicago, where BDE-47 was the most abundant congener. PBDE Temporal Trends. Following the approach of Venier and Hites,14 the concentration data for all the sites were combined together for each phase. The natural logarithms of these concentrations were fitted using the following harmonic regression equation: lnðCÞ ¼ a0 þ a1 sinðztÞ þ a2 cosðztÞ þ a3 log2 ðpopÞ þ a4 logðdistÞ þ a5 t þ ε

ð1Þ

where C is the concentration in the gas phase, particle phase, or precipitation, t is time expressed in Julian days starting from January 1, 2005, z = 2π/365.25 (which fixes the periodicity at one year), pop is the number of people within a 25 km radius of the sampling site,14 dist is the distance (in km) of the sampling site from Cleveland, a0 is an intercept which rationalizes the units, a1 and a2 describe the seasonal variations in the concentration with time, a3 describes the change in concentration as a function of population, a4 is the coefficient describing the change of concentration as a function of distance from Cleveland, a5 is a first-order rate constant (in days1), and ε is the regression residual. Cleveland was assumed to be the source location because of relatively high BDE209 concentrations previously measured at this site.18 The calculations for this multiple regression were done using Minitab 16, which returned the coefficients, their standard errors, and the sum-ofsquares associated with each term. The details on the derivation of this expression and the calculations of the maximum concentration date and halving time can be found elsewhere.14 This regression was applied to ∑PBDE, BDE-47, -99, and -209 concentrations, and the coefficients and their standard errors from the regression for these compounds, as well as the dates of maximum concentration and halving times are reported in SI Table S1. We then normalized the data to the same scale by subtracting the variations caused by local human population, seasonal factors, distance from an assumed source, and the intercept. This correction gives what are known as “partial residuals” lnðCÞ  a0  a1 sinðztÞ  a2 cosðztÞ  a3 log2 ðpopÞ  a4 logðdistÞ ¼ ε0

ð2Þ

Table 2. Halving and Doubling (Noted with a Negative Sign) Times (In Years) and Their Standard Errors for ∑PBDE, BDE-47, BDE-99, BDE-209, DBDPE, HBB, PBEB, and BTBPE Concentrations in the Vapor, Particle, And Precipitation Phases, And in the Three Phases Combined compound

a

vapor

particles

precipitation

combined phases

∑PBDE

9.9 ( 4.2

6.6 ( 1.8

4.7 ( 2.4

6.3 ( 1.1

BDE-47 BDE-99

6.3 (1.8 9.8 ( 4.6

6.9 ( 2.2 8.8 ( 3.7

4.8 ( 2.2 1.9 ( 0.3

6.1 ( 0.3 5.1 ( 0.9

BDE-209

NSa

NS

7.2 ( 3.4

NS

DBDPE

NS

NS

NS

NS

HBB

5.9 ( 3.0

3.2 ( 0.6

2.2 ( 0.5

9.5 ( 4.6

PBEB

7.7 ( 3.5

4.3 ( 1.2

5.5 ( 2.5

NS

BTBPE

NS

13 ( 6.5

3.8 ( 1.0

9.8 ( 2.8

NS: not significant at P < 0.05.

In this equation, ε0 is the partial residual, and a0, a1, a2, a3, and a4 are the coefficients from eq 1. This step was applied to each chemical in each phase separately; the results for all three phases were combined together; and a new time-dependent linear regression was fitted to these partial residuals ε0 ¼ a00 þ a05 t 0

ð3Þ 0

where a5 is an overall first-order rate constant, and a0 is an intercept. This process yields rate constants that can be used to calculate the halving times of each chemical in each phase separately and in all three phases combined. The additional step of combining the data for all three phases together allows us to calculate a weighted average of the halving times using the number of data points in each phase as weights. Table 2 (top) presents the halving times and their standard errors for each phase and for the three phases combined using this approach. The combined phase regressions are shown in Figure 2. The regressions for ∑PBDE, BDE-47, and BDE-99 concentrations are all highly significant (P < 0.0001). Overall, ∑PBDE concentrations are declining in the atmosphere of the Great Lakes with halving times of ∼6 years. These results are generally slower than halving times reported previously; Venier and Hites reported declining trends for ∑PBDE atmospheric concentrations (sum of vapor plus particle phase concentrations) at each IADN site with halving times of ∼4 years.18 The overall halving time of ∑PBDE concentrations is shorter than that of ∑PCBs (∼17 years), ∑PAHs (∼10 years), and ∑DDTs (∼9 years) reported recently for the same sites around the Great Lakes and using the same harmonic regression approach used here.14 This may indicate that the 2004 production restrictions of penta- and octa-PBDEs are having an effect and that the concentrations of PBDEs in the atmosphere are declining faster than those of other persistent organic pollutants. The concentrations of BDE-47 and -99 are also declining with halving times of ∼6 years. Overall, with the exception of the low value of ∼2 years for BDE-99 in precipitation, the average halving times for these individual congeners are the same as for ∑PBDE concentrations. The values reported here for BDE-47 and -99 are somewhat higher than the halving times reported for these compounds in a previous study from our laboratory (∼13 years).18 The previous study used IADN PBDE data from only 20052006, which was right after production of the penta-BDE mixture was discontinued in 2004. Significantly faster halving times 8701

dx.doi.org/10.1021/es2020378 |Environ. Sci. Technol. 2011, 45, 8698–8706

Environmental Science & Technology

Figure 2. Regression of partial residuals versus sampling date for the three phases combined (vapor, red squares; particles, green triangles; and precipitation, yellow circles) for (A) ∑PBDEs (B) BDE-47 (C) BDE-99, and (D) BDE-209.

were reported by Venier and Hites for chlordane concentrations in precipitation in comparison to chlordane’s halving times in vapor and particle phases,14 which is similar to our observation for BDE99. We are not able to explain this finding. The regressions for the BDE-209 concentrations versus time were generally not significant, suggesting that the concentrations of this congener in the environment are not yet decreasing. Perhaps when the restrictions on the production and use of this compound go into effect after 2013, these atmospheric levels will start to decrease. DBDPE, HBB, PBEB, and BTBPE. Figure 3 shows the concentrations of these compounds as box-plots in the vapor, particle, and precipitation phases at the five U.S. IADN sites. Table 3 gives the arithmetic mean and standard errors of these concentrations along with ANOVA results (applied to log-transformed concentrations) and percent detected. DBDPE has been produced and used as an additive flame retardant for more than 20 years.29 Due to its structural resemblance to BDE-209, DBDPE seems to have the same applications as the deca-BDE commercial product and has recently become an alternative to this restricted formulation.30 DBDPE has been detected in indoor29,31 and outdoor air,18,20 precipitation,20 sewage sludge and sediments,29,32 tree bark,20,33,34 household dust,31,35 and biota.13,36,37 A recent international survey on DBDPE in sludge samples found DBDPE in samples from each of the 12 countries included in the survey.38 Clearly, DBDPE is a worldwide pollutant.

ARTICLE

In this study, DBDPE was detected in all three phases, but not as often as most of the other BFRs; see Table 3. The average percent detection for DBDPE was 6% in the vapor phase, 23% in the particle phase, and 61% in precipitation samples. Overall, the highest DBDPE concentrations were detected at the urban Chicago and Cleveland sites. These concentrations were not statistically different among the three nonurban sites in any phase; see Table 3. DBDPE concentrations are in general agreement with those reported in previous studies from our laboratory,18,20 but they are low relative to those of Kierkegaard et al.,29 who reported much higher concentrations in ambient air near an electronics dismantling facility in Stockholm, Sweden (700 pg/m3). SI Figure S3 shows the percentage of DBDPE concentrations relative to those of ∑PBDE at the five U.S. IADN sites in the vapor, particle, and precipitation phases. Interestingly, the ratio of DBDPE concentrations to ∑PBDE concentrations was the highest at the three rural and remote sites in the particle phase with the ratios of 0.75 at Sturgeon Point, 0.37 at Eagle Harbor, and 0.35 at Sleeping Bear Dunes. HBB and PBEB are also flame retardants, but not much is known about their production, use, or environmental fate. In the United States, both of these flame retardants were produced and used in 1970s and 1980s; however, there seems to be no current information on the U.S. production of these compounds. The last list from the EPA Inventory Update Report indicates HBB and PBEB production amounts in 1998 and 1986, respectively; in both cases, the amounts produced were 5250 tons.39 HBB seems to be produced in Japan and China, and PBEB seems to be produced by Albemarle Chemical Corporation in France.30 HBB was used as an additive flame retardant in wood, textiles, electronics, and plastics; PBEB was used in polyurethane foam, wire and cable coatings, and polyester resins.30 PBEB is included in the OSPAR List of Chemicals for Priority Action as a persistent, potentially bioaccumulative, and toxic chemical.40 HBB was first “spotted” in the environment several decades ago in sediment and human samples from Japan, which was probably a result of Japan’s heavy use of HBB at that time.41,42 Similarly, PBEB was identified in Canadian lake trout caught in 1979.43 More recent studies on HBB and PBEB are scarce and scattered. These compounds were found in air samples from the United States,44 Canada,45 the United Kingdom,46 and the Atlantic and Southern oceans;8 sediment and sludge samples from Norway;9 biota samples from the Great Lakes and Arctic;10,13,47 and human blood samples from China.48 In this study, we observed surprising trends in the spatial distributions of both HBB and PBEB concentrations; see Table 3. Both of these compounds were detected in all three phases ∼70% of the time, but their concentrations were generally higher in the vapor phase. Unlike ∑PBDE concentrations, the highest average HBB concentrations (6.4 ( 0.9 pg/m3) were observed at the rural Sturgeon Point site in the vapor phase. The HBB levels were similar to those measured in air samples from Drammen, Norway9 and in the atmosphere over the Atlantic Ocean (0.0411 pg/m3).8 The concentration of HBB at the Sturgeon Point site is ∼610 times higher than the HBB concentrations in Cleveland and Chicago. The average vapor HBB concentration at this site was 80% of that of the ∑PBDE concentration in the vapor phase; see SI Figure S3. This finding is remarkable given that there are no known HBB sources near Sturgeon Point. It has been suggested that possible HBB sources can include release from polymeric flame retardants; however, the estimated release of HBB from these sources was shown to be insufficient to explain 8702

dx.doi.org/10.1021/es2020378 |Environ. Sci. Technol. 2011, 45, 8698–8706

Environmental Science & Technology

ARTICLE

Figure 3. Concentrations of DBDPE, HBB, PBEB, and BTBPE in the atmospheric vapor phase, particle phase, and precipitation; the thin black lines represent the median, and the thick red lines represent the arithmetic mean; the boxes represent the 25th and 75th percentiles; the whiskers represent the 5th and 95th percentiles. Site abbreviations: EH Eagle Harbor, CH Chicago, SB Sleeping Bear Dunes, CL Cleveland, SP Sturgeon Point.

Table 3. Arithmetic Mean ( Standard Error of DBDPE, HBB, PBEB, and BTBPE Concentrations in Vapor Phase (pg/m3), Particle Phase (pg/m3), and Precipitation (ng/L) at the Five U.S. IADN Sitesa Eagle Harbor conc.

*

Chicago

% det

conc.

*

Sleeping Bear Dunes % det

conc.

*

Cleveland

% det

conc.

DBDPE

vapor

0.02 ( 0.01

c

4

4.7 ( 2.3

a

11

0.7 ( 0.2

ab

10

0.10 ( 0.04

HBB

vapor

0.2 ( 0.1

d

40

0.7 ( 0.1

b

75

0.2 ( 0.1

c

67

1.1 ( 0.2

*

Sturgeon Point % det

conc.

*

% det

abc

4

0.02 ( 0.01

bc

2

b

68

6.4 ( 0.9

a

81

PBEB

vapor

1.2 ( 0.2

c

75

0.8 ( 0.1

b

94

0.05 ( 0.01

e

66

1.5 ( 0.1

a

92

0.11 ( 0.01

d

87

BTBPE

vapor

0.2 ( 0.1

c

14

0.8 ( 0.2

a

60

0.3 ( 0.1

bc

40

0.5 ( 0.1

ab

13

0.2 ( 0.02

bc

43

DBDPE

particle

1.2 ( 0.5

b

7

3.6 ( 0.8

b

34

3.2 ( 1.0

b

13

14 ( 7

a

47

2.9 ( 1.1

b

14

HBB

particle

0.2 ( 0.1

c

55

0.4 ( 0.1

a

68

0.2 ( 0.1

c

57

0.3 ( 0.1

a

67

0.2 ( 0.04

b

62

PBEB

particle

0.7 ( 0.1

a

39

0.1 ( 0.02

c

75

0.1 ( 0.03

d

21

0.14 ( 0.03

b

88

0.1 ( 0.03

d

21

BTBPE

particle

0.2 ( 0.03

c

29

1.0 ( 0.2

a

91

0.9 ( 0.3

b

83

0.9 ( 0.1

a

93

0.5 ( 0.1

c

82

DBDPE

precip

0.3 ( 0.1

ab

31

0.8 ( 0.2

a

74

0.4 ( 0.1

ab

63

0.6 ( 0.2

a

71

0.5 ( 0.2

b

65

HBB PBEB

precip precip

0.3 ( 0.1 0.03 ( 0.01

a a

85 79

0.2 ( 0.04 0.03 ( 0.01

ab bc

72 67

0.6 ( 0.2 0.01 ( 0.01

a bc

93 62

0.1 ( 0.02 0.01 ( 0.01

b b

80 87

0.3 ( 0.1 0.01 ( 0.01

ab c

78 74

BTBPE

precip

0.01 ( 0.01

b

51

0.1 ( 0.02

a

94

0.04 ( 0.01

b

80

0.1 ( 0.01

a

89

0.05 ( 0.01

b

76

a

ANOVA results for the log-transformed concentrations are given in the columns headed by an asterisk; the concentrations are not significantly different for those locations sharing the same letter. Percent of detects for each compound in each phase is also shown.

prevailing HBB concentrations in air.45 Other possible sources of HBB include degradation of higher brominated PBDE congeners

or DBDPE in the environment or during e-waste recycling or burning.9,49 The concentrations of HBB in precipitation were 8703

dx.doi.org/10.1021/es2020378 |Environ. Sci. Technol. 2011, 45, 8698–8706

Environmental Science & Technology similar at all five sites, and it was detected in ∼80% of the samples. In the particle phase, the highest HBB concentrations in particle phase were detected at Chicago and Cleveland. The mean concentrations of PBEB in the vapor phase were highest at Cleveland and lowest at Sturgeon Point and Sleeping Bear Dunes. Surprisingly, PBEB concentrations in particles and precipitation were highest at Eagle Harbor—our most remote site. In fact, the PBEB concentrations at Eagle Harbor were ∼45% of those of ∑PBDE (see SI Figure S3). The generally low levels of PBEB in precipitation suggest that this compound is not scavenged by precipitation as effectively as PBDEs. The PBEB levels we observe here are much lower than the single day PBEB measurement in Chicago air (520 pg/m3 in the vapor phase and 29 pg/m3 in particles) reported previously;44 however, these concentrations in Great Lakes air are higher than those measured in air samples in Drammen, Norway, and Egbert, Canada.9,45 With the exception of the one sample in Chicago,44 the vapor and particle phase PBEB concentrations at Eagle Harbor reported in this study are the highest atmospheric concentrations yet reported for this compound. Considering that Eagle Harbor is a very remote site and few people live or work there, this observation is hard to explain. BTBPE (in some papers abbreviated as TBE) is an additive flame retardant produced since the 1970s by Great Lakes Chemical (now a part of Chemtura) and used as a replacement for the discontinued octa-BDE mixture.44 It was a high production volume chemical in the United States with an estimated production of 5005000 tons/year in 1998,44 but its estimated worldwide usage was ∼20 tons in 2000.30 In our samples, BTBPE was present in all three phases, but it was primarily detected in the particle phase and in precipitation. The highest levels of BTBPE were measured at Chicago and Cleveland in all three phases. The levels at the other three sites were, in general, not statistically different from one another in any phase. The average BTBPE concentration at Sleeping Bear Dunes was ∼20% of that of ∑PBDE concentration at this site in the particle phase (see SI Figure S3). The levels of BTBPE measured in this study were similar to those reported in a previous study from our laboratory (0.51.2 pg/m3)18 and to levels measured in Chicago (4.0 pg/m3),44 Bloomington, Indiana (2.8 pg/m3),44 and at a suburban area close to Stockholm (