Received for review August 15, 2007

Revised manuscript received October 10, 2007

Accepted October 18, 2007

Introduction

Flame retardants are widely used in commercial and household products to minimize property fires. Some of these flame retardants, for example the polybrominated diphenyl ethers (PBDEs), are persistent in the environment, bioaccumulate in biota, and are atmospherically transported long distances from their sources (1, 2). Presumably because of their ubiquity, two commercial PBDE mixtures (penta- and octa-BDE) have been regulated by the European Union and by several U.S. states, and the major manufacturer of these two products in the United States stopped their production in 2004 (3).

There are more than 175 different types of flame retardants (4), and those taken off the market are likely to be replaced by nonregulated ones. For example, Dechlorane Plus (DP, C₁₈H₁₂Cl₁₂) was used as a substitute for the highly chlorinated flame retardant Dechlorane (also known as the insecticide Mirex) when its use was restricted in the 1970s because of its environmental effects (5). However, the substitute might not always be as safe as hoped. For example, DP is now present in air, sediment, fishes, and herring gull eggs from the Great Lakes (6–9) and in fishes and sediment from Lake Winnipeg (Canada) (7), and DP’s concentration was high in sediment from Lake Ontario relative to the other Great Lakes (9). All these observations suggested that DP might be an environmental problem, especially in the lower Great Lakes region.

To investigate the spatial distribution of DP in a given region, it would be helpful to have several atmospheric samples taken over the area to be studied. Rather than set up active (pump and trap) samplers at a few selected sites, we decided to use samples collected passively. This allowed us to collect many more samples in a shorter time and to cover a wider spatial range. Although polyurethane foam (PUF) passive samplers have been shown to be effective (10), they have the disadvantage of requiring deployment for several months. Tree bark, on the other hand, is an easy and inexpensive monitor of persistent organic pollutants. Pervious work has shown that tree bark is a good passive sampler for semivolatile organic compounds with high K_oa values. Such analytes include organochlorine pesticides (11, 12), polychlorinated biphenyls (PCBs) (13–15), PBDEs (16), polycyclic aromatic hydrocarbons (PAHs) (17), and polychlorinated dibenzo-p-dioxins and dibenzofurans (PCDD/Fs) (18).

The objectives of this study were 3-fold: (a) to investigate the transport and distribution of DP in the northeastern United States by measuring its concentration in tree bark samples, especially near its potential source, the OxyChem manufacturing facility located in Niagara Falls, New York; (b) to fit the observed concentrations with a one-dimensional, Gaussian diffusion model so that the location of the potential source could be identified; and (c) to measure the concentrations of several other brominated flame retardants (BFRs), including the PBDEs, 1,2-bis(2,4,6-tribromophenoxy)ethane (TBE), and decabromodiphenyl ethane (DBDPE), in these samples and in others from Canada, Europe, and Asia.

Experimental Section

Sample Information. All the samples from the United Sates were collected in the autumn of 2006. The sampling locations are given in Table 1 and shown in Figure 1. These sampling locations were chosen primarily because of ease of accessibility but also because of distance from the potential source in western New York state. Geographical coordinates were obtained at each location using an Eagle Explorer (Catoosa, OK) global positioning system (GPS). In this study, only pine bark samples were collected, which were coarse and high in lipids. The average tree diameter was 42 cm (see the Supporting Information). Bark samples were collected from three individual trees at each location; the distance between the three trees was less than 50 m. A sample of bark (from 50 to 100 g) was removed from different sides of each tree using a hammer and chisel (cleaned with ethanol prior to each use) at a height of ~1.5 m above ground. To avoid long-term damage to the tree, the cambium was not sampled. The bark was wrapped in aluminum foil, sealed in plastic bags, placed in a cooler, and kept at ambient temperature until the samples were shipped to the laboratory, where they were kept at −20 °C until extraction.

Figure 1. Tree bark sampling sites and the corresponding DP concentration (ng g−1 bark). The star represents the DP manufacturing plant located in Niagara Falls, NY. The triangle represents the calculated source location; see eq 10. The abbreviations of IN, MA, MD, NY, OH, PA, and VA indicate the states of Indiana, Massachusetts, Maryland, New York, Ohio, Pennsylvania, and Virginia, respectively. Click to Enlarge

Organic Chemicals. DP was obtained from OxyChem (Dallas, TX) and used to quantitate all samples. 1,2-Bis(2,4,6- tribromophenoxy)ethane (TBE) and decabromodiphenyl ethane (DBDPE) were purchased from Wellington Laboratories (Guelph, ON). All the PBDE congeners, including BDE-28, 47, 99, 100, 153, 154, 183, 196, 197, 198, 201, 203, 204, 206, 207, 208, and 209 were from AccuStandard (New Haven, CT). The surrogate and internal standards were from AccuStandard [BDE-77, 118, 166, and decabromobiphenyl (BB-209)] and Cambridge Isotope Laboratories (Andover, MA) (¹³C₁₂-BDE-209). DP from OxyChem was calibrated with separated standards of syn- and anti-DP from Wellington Laboratories (50 µg mL⁻¹ in toluene, purity >98%). BDE-209 from AccuStandard was certified with samples from both the National Institute of Standards and Technology (NIST) and Wellington Laboratories. All the solvents used for the extraction and cleanup procedures were residue-analysis grade.

Sample Preparation. A previous method (12) was modified for this research. In brief, approximately 20–30 g of bark sample was cut into pieces of <1 cm using pruning shears and placed into a Soxhlet extraction thimble bedded with 30 g of granular anhydrous Na₂SO₄, and then covered with 20 g of Na₂SO₄ to keep the bark from floating. After spiking with known amounts of BDE-77, BDE-166, and ¹³C-BDE-209 as surrogate standards, the samples were extracted with 400 mL of hexane/acetone (1:1, v/v) for 45 h.

To measure the percent of lipophilic material in the sample, after Soxhlet extraction, ~10% of the extract was transferred into a preweighed aluminum weighing dish and evaporated at room temperature to a constant weight. The rest of the extract was evaporated to ~5 mL, and the turbid solution was transferred to a 15-mL centrifuge tube. After blow down and solvent exchange into hexane, the sample tubes were put into a cold bath (dry ice/acetone, −77 °C) for several minutes; this process reduced the solubility of the polymeric bark lipids in hexane. After a quick centrifugation, the upper solvent layer was transferred into a new tube. The separated solid was extracted twice more, each time with 3 mL of hexane, and treated in the same way. The combined extracts were blown down to 4 mL and treated with 5 mL of concentrated H₂SO₄ (EM Science, Gibbstown, NJ). After centrifugation, the organic layer was transferred into a new tube. The H₂SO₄ residue was washed twice with hexane, and the combined organic extracts were blown down to 3 mL. At this point, the solution was clear and colorless. Nevertheless, another H₂SO₄ treatment was performed to ensure that the polymeric bark lipids were completely removed. The extract was blown down to ~0.5 mL under a gentle N₂ flow and then loaded on an alumina (MP Biomedicals GmbH, Eschwege, Germany) column (0.6 cm i.d. × 6 cm, with 0.5 cm anhydrous Na₂SO₄ on the top). The column was eluted with 8 mL of hexane followed by 8 mL of 2:3 (v/v) dichloromethane/hexane. DP and the other target compounds eluted in the second fraction. After blow down, BDE-118 and decabromobiphenyl (BB-209) were added as quantitation standards, and the samples were analyzed by gas chromatographic mass spectrometry (GC−MS).

In each batch of six samples, one blank sample and one matrix spiked sample were included. In both cases, 70 g of clean Na₂SO₄ was used as the blank and as the spiked sample matrix. Throughout the extraction and analysis procedure, the analytes were protected from light by wrapping the containers with aluminum foil or by using amber glassware.

Instrumental Analysis. The samples were analyzed on an Agilent 6890 series gas chromatograph coupled to an Agilent 5973 mass spectrometer (GC−MS) with highly purified helium as the carrier gas. Injections (2 µL) were made in the pulse splitless mode with a purge time of 2.0 min. The injection port and the GC to MS transfer line were held at 285 °C. A 15-m long Rxi-5 ms column (250 µm i.d.; 0.25 µm film thickness; Restek Corporation, Bellefonte, PA) was used for all the target compounds. The GC oven temperature program was as follows: held at 100 °C for 2 min; 25 °C/min to 250 °C; 3 °C/min to 270 °C; 25 °C/min to 325 °C, and held for 8 min. The mass spectrometer was operated in the electron capture negative ionization mode using methane as the reagent gas, and the ion source temperature was 200 °C. The following ions were monitored: m/z 653.8 and 651.8 for DP, m/z 486.6 and 488.6 for BDE-209, m/z 494.6 and 496.6 for ¹³C₁₂-BDE-209, and m/z 81 and 79 for the other brominated compounds. The response factors for all compounds were determined using a calibration standard solution. BDE-118 was used for tri- through hepta-BDE and TBE, and BB-209 was used for DP, octa- through deca-BDEs, and DBDPE. For those samples with high concentrations of DP, a diluted duplicate sample was run to make sure the response of DP was in the linear range of the instrument. The DP concentration data >30 ng g⁻¹ bark (see Table 1) are from diluted samples.

Quality Control. Several quality control criteria were used to ensure the correct identification and quantitation of the target compounds: (a) The GC retention times matched those of the standard compounds within ± 0.1 min. (b) The signal-to-noise ratio was greater than 5:1. (c) The isotopic ratios for selected ion pairs were within ± 15% of the theoretical values. The recovery of surrogate standards (mean ± standard deviation) was 73 ± 11%, 69 ± 16%, and 64 ± 19% for BDE-77, BDE-166, and ¹³C₁₂-BDE-209, respectively. The recovery of matrix spiked samples was 88 ± 13%, 95 ± 5%, 100 ± 4%, 94 ± 8%, and 92 ± 16% for DP, BDE-47, BDE-99, BDE-209, and TBE, respectively. One procedural blank was also run with each batch of samples. All the compounds in the blank samples were undetected or less than 1% of the average value measured in tree bark samples, except BDE-47 and BDE-99, which were 4% and 6% of average concentration in tree bark samples, respectively (see Supporting Information). In this paper, concentrations have not been blank or recovery corrected.

Results and Discussion

Concentrations of DP and BFRs in Tree Bark Samples. The concentrations of DP (reported as the sum of the syn- and anti-DP isomers), PBDEs (reported as the sum of the 17 congeners listed above), TBE, and DBDPE are given in Table 1. The concentrations were normalized to grams of tree bark; the percent lipid is also given in Table 1. These reported average concentrations and standard deviations for each target compound are from the triplicate bark samples collected at each site. Given that none of these flame retardants are directly applied to trees, it seems likely that all of these compounds accumulated in the tree bark by way of atmospheric transport.

Figure 1 shows the location of the sampling sites and the corresponding concentrations of DP measured at each site. In general, the higher concentrations (>4 ng g⁻¹ bark) were found in or around the Niagara Falls and Buffalo areas of New York state. The highest concentration (115 ng g⁻¹ bark) was found in the Hyde Park neighborhood, a location which is ~2 km away from OxyChem’s DP manufacturing facility. With increasing distance from this potential source location, the concentrations of DP in tree bark decreased rapidly, but DP was still detected at low levels (0.03–0.04 ng g⁻¹ bark) in samples from Virginia, Maryland, and Indiana.

Gaussian Diffusion Model for DP. It has been shown that concentrations of PCBs in tree bark decreased rapidly with increasing distance from Superfund sources; concentrations dropped by a factor of about 10 within 11 km (14) and by a factor of 40 within 14 km (13). In another study, PCB concentrations dropped by a factor of >1,000 within ~7 km of a major PCB source (15). Safe et al. also found a rapid decrease in total PCDD/F levels in pine needles with increasing distance from a source, and the level of octachlorodibenzo-p-dioxin dropped by a factor of 6 within 1 km of the site’s perimeter (19). All of these studies indicated that vegetation could be used to generally locate pollutant sources. To describe the spatial distribution of toxaphene in tree bark more exactly and to pinpoint its source location, McDonald and Hites created a simple dilution model (12). This model also successfully explained the spatial distribution of PBDEs in North America (16). However, this model neglected the advective transport process of a pollutant as it moves through the atmosphere.

We have improved on this simple dilution model by including advective transport. In this field, Gaussian diffusion models are widely used to describe the regional distribution of inert gaseous pollutants and aerosols (20–22). The theory of this model is based on the normal (or Gaussian) statistical distribution function. Briefly, for a continuous-point source of a certain pollutant, the spatial distribution can be described by (23)

In this study, we assume that the source is near ground level, and thus, h = 0. In fact, there is not much difference in the concentration−distance profile between a ground level source and a 100 m high source, when the distances from the source were >2 km (21). Thus, for a downwind ground site, eq 1 can be simplified to

The diffusion coefficients, σ_y and σ_z, are a function of the downwind distance, x, and are influenced by many factors, including atmospheric stability. Approximately, these coefficients can be expressed by

In this study, the DP in tree bark was accumulated over the 5–10 years the bark sample was on the tree, and the annualized wind roses in the source region are relatively symmetrical (25). This suggests that, over a relatively long time period, DP is likely to be carried by air movement in all directions. Thus, it is reasonable to assume that all of our sampling sites are located downwind from the source, and we can replace the downwind distance, x, with the distance from the source to sampling site i, which is given by

During a long period, the source strength, Q, the average wind speed, u, and γ₁ and γ₂ in eq 5 can be assumed to be constant. Thus, the concentrations of DP in tree bark as a function of distance from its source are given by

Taking logarithms of both sides, we get

We have implemented this calculation in two ways: First, we assumed that the DP source was, in fact, the OxyChem manufacturing facility located in Niagara Falls at 43.077° N, 79.009° W. By fitting the log-transformed concentrations with log-transformed distances, we could determine the values of K′ and (a₁ + a₂). The results were K′ = 10^2.5 and (a₁ + a₂) = 1.33 with a correlation coefficient (r²) between the observed and expected values of 0.853; see Figure 2 (top). Second, we assumed that we did not know the location of the source, and we let the values of lat_sor and lon_sor float or vary in eq 6 to calculate the various values of d_i. For each sample, we had measured C_i values (the concentrations of DP in the tree bark sample taken at sampling site i), and we had calculated d_i. We then minimized ξ_d as given by

Figure 2. Concentrations (in ng g−1 bark) of DP in tree bark as a function of spherical Euclidian distance (in km) from the OxyChem plant in Niagara Falls, NY (top); concentrations of DP as a function of distance from the calculated source using eq 10 (middle); fanti value at each sampling site (using the distance from the OxyChem plant). The red dotted line indicates the isomeric composition of the commercial DP product. The errors shown are one standard deviation for triplicate samples. Click to Enlarge

This is the approach used previously (12, 16). With the assistance of the Solver feature of Excel, the fitted variables for the DP data set were obtained. The results were K′ = 10^2.7, (a₁ + a₂) = 1.42, lat_sor = 43.124° N, and lon_sor = 78.953° W, with an r² between the observed and expected values of 0.872; the result is shown in Figure 2 (middle). This fitted location is less than 7 km northeast of the OxyChem plant. Despite the uncertainties of this approach, we can surely conclude that DP’s source is located in Niagara Falls, New York.

In this research, the calculated exponent (α₁+α₂) was ~1.4, which suggests an average atmospheric stability between neutral conditions [(α₁+α₂) = 1.49] and slightly stable conditions [(α₁+α₂) = 1.30] (24) over a long time period. This value was less than that observed for the distribution of PBDEs in tree bark (1.73), which was based on a point source located in southern Arkansas (16). This difference might suggest, on average, that the atmospheric conditions are less stable in the central United States than in the northeastern United States, although other factors, such as additional sources and different topographic conditions, cannot be excluded.

Technical DP has two conformational isomers: syn (U-shaped) and anti (chair-shaped) (6). The fractional abundance (f_anti) of the anti isomer (defined as the concentration of the anti isomer divided by the sum of the concentration of syn and anti isomers) is 0.75 in the technical product as measured in our laboratory. Incidentally, although there are three industrial formulations of the technical DP product, they differ only in the particle size and not in composition (28). The spatial distribution of f_anti in tree bark is shown in Figure 2 (bottom) as a function of distance from the source. There is no obvious trend of f_anti values with the increasing distance. The average f_anti in all the tree bark samples is 0.76, which is close to that in the technical product and suggests that the two isomers have the same atmospheric persistence. This observation is different from the fate of these two isomers in sediment, where the anti isomer was found to be the more persistent (9).

Other Brominated Flame Retardants. PBDEs, TBE, and DBDPE are brominated flame retardants, which are manufactured in the southcentral United States (16). Thus if these compounds are present in tree bark collected in the northeastern United States, they should have arrived there by long-range atmospheric transport from the manufacturing plants and by local use. Table 1 shows that the average concentrations of PBDEs, TBE, and DBDPE in our samples are 1.9, 0.11, and 0.28 ng g⁻¹ bark, respectively (or 55, 3.2, and 8.5 ng g⁻¹ lipid, respectively). These concentrations were much lower than those observed near the manufacturing plants in Arkansas (16), suggesting that−unlike DP−there is no strong point source for these BFRs in the northeastern United States.

Within our data set, relatively high concentrations of BFRs were observed in tree bark from urban/suburban areas near Cleveland, Ohio and near Buffalo−Niagara Falls, New York. For instance, the highest concentration of TBE was found in bark from Cleveland, and the highest concentration of PBDEs was found in bark from Niagara Falls. Nevertheless, at most sites in the northeastern United States, the concentrations of DP were from 1 to 2 orders of magnitude higher than all of the BFRs combined; this is especially true in Niagara Falls. This observation is consistent with the relatively high concentrations of DP and relatively low concentrations of all other BFRs that we observed in a sediment core from Lake Ontario (9).

DP and BFRs in Tree Bark Samples from Other Countries. To evaluate the global impact of DP and BFRs, we obtained and analyzed tree bark samples from other countries; the sampling information and results are shown in Table 2. DP was not detected in tree bark from the southern Northwest Territories in Canada; however, DP was detected in all the other samples from Europe and Asia. In Germany and Italy, DP concentrations were similar to the lowest concentrations we measured in bark from the northeastern United States; in China and Korea, DP concentrations were nearly 1 order of magnitude higher than the lowest concentrations in the northeastern United States. This latter observation suggests that there may be Asia-specific sources of DP to the environment. In all these “foreign” samples, the f_anti was 0.73 on average, which is close to that measured in technical DP (0.75). This indicates a similar fate of the two DP isomers in the atmosphere.

PBDEs, TBE, and DBDPE were also detected in most of the samples from Europe and Asia. It is notable that the concentrations of PBDEs (especially BDE-209) and DBDPE were high in tree bark samples from China, especially in those from Shenzheng (see Table 2). For instance, BDE-209 was found in tree bark from Shenzheng at a concentration of 43 ng g⁻¹ bark (or ~1500 ng g⁻¹ lipid), which was similar to the BDE-209 concentration measured in tree bark collected near its manufacturing facility in Arkansas (16). BDE-209 is now produced in China; in fact, between the years 2000 and 2005, BDE-209 production increased from 10,000 to ~30,000 t (29). It is interesting to note that Shenzheng is located in the Pearl River Delta, which is the largest electronic and telecommunication equipment manufacturing region in China. It is perhaps not too surprising that high concentrations of BDE-209 are detected in tree bark and in other environmental media, such as sediment (30), from this region.

Supporting Information Available

Tables of analytical data by sample. This material is available free of charge via the Internet at http://pubs.acs.org.

* Corresponding author e-mail: HitesR@Indiana.edu.