Environ. Sci. Technol. 2008, 42, 386–391
Identification and Determination of HexachlorocyclopentadienylDibromocyclooctane (HCDBCO) in Residential Indoor Air and Dust: A Previously Unreported Halogenated Flame Retardant in the Environment J I P I N G Z H U , * ,† Y U Q I N G H O U , ‡ YONG-LAI FENG,† MAHIBA SHOEIB,§ AND TOM HARNER§ Chemistry Research Division, Health Canada, Ottawa, Canada, K1A 0L2, Meyers Institute for Interdisciplinary Research in Organic and Medicinal Chemistry and Department of Chemistry and Biochemistry, Southern Illinois University, Carbondale, Illinois 62901, and Air Quality Research Division, Science and Technology Branch, Environment Canada, Toronto, Ontario, Canada M3H 5T4
Received September 10, 2007. Revised manuscript received November 7, 2007. Accepted November 7, 2007.
Hexachlorocyclopentadienyl-dibromocyclooctane (HCDBCO, CAS 51936-55-1) has been detected in residential indoor air and indoor dust in Ottawa, Canada. The positive identification of the chemical was based on the interpretation of the mass spectra of the chemical obtained under both electron impact and negative chemical ionization operation modes, as well as through the synthesis of this chemical. This is the first report on the presence of HCDBCO in the environment. Although the levels of HCDBCO in indoor dust, with a geometric mean of 2.7 ng g-1 and a median of 2.0 ng g-1 respectively, are generally low compared to those of polybrominated diphenyl ethers (PBDEs) and dechlorane plus, another recently detected flame retardant, high levels of HCDBCO were detected in several dust samples with a maximum level of 93,000 ng g-1 which is 16 times higher than the maximum level of the structurally related dechlorane plus. On the other hand, levels of HCDBCO in indoor air, with a geometric mean of 70 pg m-3 and a median of 92 pg m-3, were higher than those of the major PBDE congeners. The maximum level of HCDBCO found in indoor air was 3000 pg m-3. Structurally, HCDBCO belongs to a group of norbornane based halogenated flame retardants. The presence of HCDBCO in the indoor environment may raise awareness of the potential release of this and related flame retardants into the environment during the production and usage of products that contain them, and the potential implications of human exposure to these chemicals as people spend the majority of their time indoors in modern society.
* Corresponding author e-mail:
[email protected]; fax: (613) 946-3573. † Health Canada. ‡ Southern Illinois University. § Environment Canada. 386
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 2, 2008
Introduction Flame retardants (FRs) are used to inhibit or resist the spread of fire. They are added to plastic materials, electronic components, clothing, and furniture to improve their resistance to fire (1). While FRs have been playing a significant role in reducing human fatality and property damage related to fires in modern society, some of the FRs such as those that are halogenated exhibit toxicities and may pose potential health risks to the human population (2) (Scheme 1). Organic halogenated flame retardant (HFR) is a general term for flame retardants that contain halogens (usually chlorines and bromines) attached to their carbon structure. The United Kingdom Environmental Agency’s Science Group 2003 report divided organic HFRs into aromatic HFRs, such as polybrominated diphenyl ethers (PBDEs), alicyclic HFRs, such as hexabromocyclododecane (HBCD), and aliphatic HFRs (3). Human exposure to PBDEs has been extensively monitored through the measurement of PBDE levels in both indoor and outdoor environments (4, 5), where potential human exposure to these chemicals occur, and through the direct detection of PBDEs in human body fluids such as breast milk and blood (6–8). The levels of two other brominated FRs, HBCD and tetrabromobisphenol A (TBBPA), in the environment have also been reported (9, 10). To date, many of the monitoring activities of halogenated fire retardants have been focused on PBDEs, HBCD, and TBBPA (11–13). However, there are a large number of other HFRs with various applications (3, 14). Their possible release and accumulation in the environment are not well understood. So far, only a few of them have been identified as being present in the environment. For example, 1,2-bis(2,4,6tribromophenoxy)ethane and decabromodiphenyl ethane were detected in tree bark samples collected in Arkansas, where two nearby manufacturing facilities were suspected of being the sources of these two compounds (15). Decabromodiphenyl ethane was also detected in the environment in Sweden and The Netherlands (16). Another HFR, dechlorane plus, has been detected recently in the air and sediments of remote locations in the Great Lakes area (17) and in indoor dust of residential homes in Ottawa, Canada (18). The detection of these HFRs illustrates the possible presence of many other unreported HFRs in the environment and challenges facing the scientific community in its effort to better monitor HFRs in the environment and in the human body (19). In our ongoing research efforts to identify and measure environmental contaminants as part of the characterization of human exposure to these substances, we have conducted a screening analysis of a pooled sample from extracts of indoor dust samples collected from residential homes in Ottawa, Canada. One previously unreported peak on the gas chromatogram was observed. In this paper, we are reporting the identification of this unknown peak and the determination of the levels of this chemical in the individual indoor air and indoor dust samples.
Experimental Section Chemicals and Materials. Hexachlorocyclopentadienyldibromocyclooctane (HCDBCO, CAS 51936-55-1) was synthesized at Southern Illinois University (see section below for the synthesis). Dichloromethane (DCM) (GC Resolv grade) and hexane were obtained from Omnisolv (EMD Chemical Inc., Gibbstown, NJ). N-octane was purchased from Caledon laboratories Ltd. (Georgetown, ON). Deionized (DI) water (18.3 Ω) was generated in-house using a Super-Q water 10.1021/es702272s CCC: $40.75
2008 American Chemical Society
Published on Web 12/19/2007
SCHEME 1. Synthesis route of hexachlorocyclopentadienyl-dibromocyclooctane (4)
generation system (Fisher Scientific, Ottawa, Canada). Sodium sulfate was purchased from Fisher Scientific (Ottawa, Canada). Synthesis of HCDBCO. Hexachlorocyclopentadiene 1 (8.5 g) and 1,5-cyclooctadiene 2 (28 g) were placed in a 100-mL round-bottom flask. The solution was heated to reflux and maintained at that temperature for 1 h. The extra 1,5cyclooctadiene was evaporated under vacuum to leave a mixture of thick oil and a small amount of white solid. The mixture was mixed with dichloromethane (25 mL) and filtered. The filtrate was evaporated under vacuum and distilled (155–170 °C/0.4 mmHg) to yield 7 g of the intermediate (1,10,11,12,13,13-hexachlorotricyclo[8.2.1.02.9]trideca5,11-diene, 3) in the form of thick oil. The structure of 3 was confirmed by 1H NMR (CDCl3): δ 1.50–1.62 (m, 2 H), 2.02–2.20 (m, 4 H), 2.33–2.44 (m, 2 H), 2.85–2.94 (m, 2 H), 5.72–5.82 (m, 2 H). The intermediate 3 (7.0 g) was then dissolved in DCM (25 mL) and cooled in an ice–water bath. To this solution Br2 was added dropwise until the bromine color did not disappear. The amber-colored solution was stirred and gradually turned into a white suspension. The solvent was removed under vacuum and the white solid was rinsed with a 1:1 mixture of hexane/dichloromethane (25 mL × 2). The white solid was filtered and dried in the oven at 70 °C for 3 h to produce 8.4 g of the final product HCDBCO 4; mp 198–200 °C (literature: 200–201 °C) (20). The product was confirmed to be the pure desired chemical by 1H NMR. 1H NMR (CDCl3): δ 1.50–1.74 (m, 2H), 1.92–2.02 (m, 2H), 2.15–2.34 (m, 2H), 2.54–2.65 (m, 1H), 2.69–2.79 (m, 1H), 3.09 (m, 1H), 3.35 (m, 1 H), 4.68–4.73 (m, 1H), 4.79–4.84 (m, 1 H). Analysis of Dust Samples. Screening analysis for the identification of the target chemical was carried out on pooled dust extracts of 2002–2003 samples prepared for the measurements of phthalates and other semivolatile chemicals (22). Random house selection process in Ottawa in the 2002–2003 study as well as the collection and sieving procedure for the dust samples are described elsewhere (5, 21). The sample preparation and instrument measurement (gas chromatograph coupled with negative chemical ionization mass spectrometer (GC/NCI-MS) operated in selected ion monitoring (SIM) mode) of the target chemical in the individual dust samples were the same as those for the measurement of dechlorane plus in dust samples (18). Analysis of Air Samples. Air samples were collected using a passive air sampler containing a disk of polyurethane foam (PUF) as the collection medium. These air samples were solvent extracted for the analysis of PBDEs (4). The same extracts were retro-analyzed in this study for the measurement of HCDBCO. The same analytical instrument (GC/NCIMS, SIM mode) used for the dust samples was employed for the air samples. The GC column for the air samples was a 30 m × 0.25 mm i.d. × 0.25 µm film thickness DB-5MS column (J&W Scientific, Folsom, CA). The oven temperature was modified to 70 °C (1.0 min), 15 °C min-1 to 210 °C, 4 °C min-1 to 250, and 10 °C min-1 to 290 °C (15 min). QA/QC. All calibration standards and spiking solutions were prepared by serial dilution in n-octane using volumetric flasks. All laboratory glassware used in the preparation of
dust samples was thoroughly cleaned and the sample jars, gloves, and polyethylene bags were checked to be free of target compounds. Daily multilevel (0.001-1.0 ng µL-1 for dust samples and 0.0008-0.5 ng µL-1 for air samples) calibration was conducted. For the dust samples, one laboratory control sample containing 1 g of solvent-washed dust spiked with 50 ng of target chemical and one laboratory blank were included in each analytical batch of six samples (18). The average blank level (standard deviation) of HCDBCO was 0.048 (0.060) ng g-1, resulting in a method detection limit (MDL) of 0.24 ng g-1 based on 1 g of dust sample. One duplicate sample analysis was performed every second batch (about 10% of all dust samples in total). Of the eight duplicates, six had a relative percent difference (RPD) between 12 and 18. The other two duplicate analyses had RPD values of 97 and 113, respectively. For the air samples, no recoveries of the target chemical were available as this was a retro-analysis of the extracts prepared for PBDEs (4). A 100% recovery was therefore assumed. The air volume was assumed to be 52.5 m3 based on the uptake rate (2.5 m3 day-1) for PBDEs and the deployment time of 21 days (4). The instrument detection limit (IDL), based on the analysis of seven replicates of the low-concentration standard solutions and an air volume of 52.5 m3, was estimated to be 0.14 pg uL-1 or 1.3 pg m-3. Extracts of laboratory blanks and field blanks of air samples were also analyzed along with the extracts of air samples to access the potential contamination in these air samples. The target chemical was not detected in the laboratory blanks. However, the seven field blanks associated with the air samples did contain trace HCDBCO at levels of ND, ND, 3.1, 3.3, 3.4, 24, and 41 pg m-3. The impact of HCDBCO in the field blanks on the interpretation of the air sample data will be discussed in the Results and Discussion section. Two duplicate samples associated with the air samples had RPDs of 54 and 120 respectively.
Results and Discussion Identification of HCDBCO in Pooled Indoor Dust Extract. During the screening analysis of pooled dust extracts that had been analyzed for other environmental contaminants (22), one peak with the presence of both chlorine (m/z 35/ 37) and bromine (m/z 79/81) ions were observed in the negative chemical ionization (NCI) GC/MS (Figure 1B). The molecular ion (M ) 536) was evident in the NCI spectrum, which produced several characteristic fragments under the NCI conditions. For example, the loss of a combination of bromine and chlorine atoms led to several fragments being observed in the NCI spectrum (m/z 500, m/z 344, and m/z 308) (see Supporting Information, Scheme S1). The retro Diels–Alder reaction occurring in the MS ionization source produced an intermediate structure hexachlorocyclopentadiene (C5Cl6), of which the fragment ion of C5Cl5- (m/z 235) was formed by further loss of one chlorine [C5Cl5- (Scheme S1)]. The presence of both chlorine and bromine atoms in the molecule can be confirmed by the presence of MS signals at m/z 35/37 (Cl-) as well as m/z 79/81 (Br-) and m/z 158 (Br2-). VOL. 42, NO. 2, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
387
FIGURE 1. NCI mass spectrum of hexachlorocyclopentadienyl-dibromocyclooctane standard (A) and its corresponding peak from pooled dust samples (B). To obtain further structural information, the sample was then analyzed under electron impact (EI) mode of GC/MS (Figure 2B). Again, a fragmentation through retro Diels–Alder reaction and further loss of a chlorine led to one of the predominant ions (m/z 235) in the mass spectrum (see Supporting Information, Scheme S2). This confirms that the molecule has a hexachlorocyclopentadiene substructure similar to many of the chlorinated pesticides containing a norbornane structure such as heptachlor, chlordane and nonachlor. A loss of a combination of chlorine and bromine atoms can also be observed in an EI mass spectrum (m/z 501 (M-Cl), m/z 457 (M-Br), and m/z 421 (M-HBr-Cl)). The formation of ions at m/z 373 (M-2HBr-3H), 341, (M-2HBrCl) and m/z 307 (M-2HBr-Cl-HCl) was more likely through a M-2HBr intermediate structure (Scheme S2). Although a high-resolution mass spectrometer was not available for this study, based on the assumption that there were two bromine atoms and a hexachlorocyclopentadiene substructure in the molecule, the remaining mass of 108 amu (M-C5Cl6-2Br) would most likely correspond to one the following two structures: cyclooctadiene (C8H12) or 1,4dioxo-cyclohexane (C6H4O2). In the case of the former structure, the peak observed on GC/MS spectrum could represent the structure of hexachlorocyclopentadienyldibromocyclooctane (HCDBCO, CAS 51936-55-1), which is a flame retardant on the Canadian Nondomestic Substances List under the Canadian Environmental Protection Act (23). 388
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 2, 2008
To further confirm that the observed unknown peak is indeed HCDBCO, synthesis of this compound was carried out by our research team as there was no commercial source available to us to obtain this chemical. HCDBCO was synthesized through a classical Diels–Alder reaction of hexachlorocyclopentadiene 1 and 1,5-cyclooctadiene 2, followed by the bromination of the double bond of the intermediate 3 (Scheme 1). The structure of the intermediate 3 was confirmed by its 1H NMR spectrum and that of 4 was confirmed by the reported melting point and its 1H NMR spectrum (20). In the case of the intermediate 3, a symmetrical structure was evident in the 1H NMR spectrum in which the shifts of protons were all in pairs (2H on each shift) to a total of 12 hydrogen atoms in the molecule. The 12 hydrogen atoms were retained in the final product 4, but the symmetry of the molecule was lost as the dibromination of the double bond led to the two bromine atoms being added in the opposite (anti-) position. Both NCI (Figure 1A) and EI (Figure 2A) mass spectra of the final synthesized product matched that of the peak from the pooled dust extract. While the EI spectra showed identical relative intensity of the ions, the relative intensity of the ions in the NCI mass spectra were different, likely due to the NCI ionization conditions and different matrices of the standard and the pooled dust extract. Levels of HCDBCO in Individual Dust Samples. Following the positive identification of HCDBCO in the pooled dust
FIGURE 2. EI mass spectrum of hexachlorocyclopentadienyl-dibromocyclooctane standard (A) and its corresponding peak from pooled dust samples (B).
FIGURE 3. Levels (ng g-1) of hexachlorocyclopentadienyl-dibromocyclooctane in indoor dust (
