Environ. Sci. Technol. 2004, 38, 3119-3125
Photochemical Decomposition of 15 Polybrominated Diphenyl Ether Congeners in Methanol/Water JOHAN ERIKSSON,* NICHOLAS GREEN, GO ¨ R A N M A R S H , A N D A° K E B E R G M A N Department of Environmental Chemistry, Stockholm University, SE-106 91 Stockholm, Sweden
Among all brominated flame retardants in use, the polybrominated diphenyl ethers (PBDEs) have been identified as being of particular environmental concern due to their global distribution and bioaccumulating properties, as observed in humans and wildlife worldwide. Still there is a need for more data on the basic characteristics of PBDEs to better understand and describe their environmental fate. Hence, the aim of this study was to investigate the photochemical degradation of PBDEs with different degrees of bromination. The photochemical degradation of 15 individual PBDEs substituted with 4-10 bromine atoms was studied in methanol/water (8:2) by UV light in the sunlight region. Nine of these were also studied in pure methanol, and four of the nine PBDEs were studied in tetrahydrofuran. The photochemical reaction rate decreased with decreasing number of bromine substituents in the molecule but also in some cases influenced by the PBDE substitution pattern. The reaction rate was dependent on the solvent in such a way that the reaction rate in a methanol/water solution was consistently around 1.7 times lower than in pure methanol and 2-3 times lower than in THF. The UV degradation half-life of decaBDE (T1/2 ) 0.5 h) was more than 500 times shorter than the environmentally abundant congener 2,2′,4,4′-tetraBDE (T1/2 ) 12 d) in methanol/water. The quantum yields in the methanol/water solution ranged from 0.1 to 0.3. The photochemical reaction of decaBDE is a consecutive debromination from ten- down to six-bromine-substituted PBDEs. Products with less than six bromines were tentatively identified as brominated dibenzofurans and traces of what was indicated as methoxylated brominated dibenzofurans.
Introduction Whereas physicochemical properties, long-range transport, bioaccumulation, and toxicity of potential environmental contaminants are described, at least to some extent, the persistence of chemicals is in general less well understood. Since a chemical is defined by its intrinsic physical properties and its reactivity, data on both are needed to assess the environmental fate of the compound. Depending on the chemical structure, it is possible to identify major potential reaction pathways. For a large number of environmentally relevant chemicals, photochemical degradations may occur when they are subjected to UV light, as may be exemplified * Corresponding author phone: +46-8-163686; fax: +46-8-163979; e-mail:
[email protected]. 10.1021/es049830t CCC: $27.50 Published on Web 05/01/2004
2004 American Chemical Society
by PCBs (1-3) and DDT (4, 5). In the present study polybrominated diphenyl ethers (PBDEs) are studied in relation to their susceptibility to UV light. PBDEs are a class of brominated flame retardants (BFRs) that are mainly used in polymers for electric and electronic goods and in textiles (6). The commercially available PBDEs can be subdivided into three groups of low, intermediate, and high bromination degrees, i.e., pentaBDE, octaBDE, and decaBDE. The two former products contain 10-20 PBDE congeners each, while decaBDE is mainly made up of one compound, the decabrominated diphenyl ether itself (BDE209) (7). The estimated worldwide demand of these products in 2001 was 7500 t for pentaBDE, 3790 t for octaBDE, and 56 100 t for decaBDE (8). However, PBDE production figures will be changed soon due to the European Community decision to ban the use of pentaBDE and octaBDE from August 2004 (9) and the decision by one of the major U.S. producers to cease the production of these two products from the end of 2004 (10). The PBDEs are additive flame retardants, which means that they are mixed with the preproduced polymer, rather than being chemically bound to it, and the PBDEs can therefore more easily migrate to the environment. PBDEs in general, and decaBDE in particular, have very low vapor pressures at normal temperatures and extremely low water solubilities. They can be considered as very lipophilic since their log Kow values are 5.9-6.2 for tetraBDE, 6.5-7.0 for pentaBDE, 8.4-8.9 for octaBDE, and 10 for decaBDE (11). Environmental monitoring data and specific analytical studies indicate that PBDEs released during manufacture and use and during the recycling of flameretarded goods constitute a major source of PBDEs to the environment (12). All PBDE congeners seem to be bioavailable, including decaBDE (13, 14), and several of the congeners show a high potential for bioaccumulation in mammals, fish, and birds (13, 15) and in humans worldwide (16). Watanabe and co-workers presented a study on the photolysis of decaBDE dissolved in a mixture of hexane, benzene, and acetone (8:1:1) in both sunlight and artificial UV light (17). In their study, using a mercury lamp (254 nm dominating), BDE-209 was debrominated down to triBDE and to relatively large amounts of polybrominated dibenzofurans (PBDFs) substituted with 1-5 bromines. In their sunlight experiments BDE-209 was debrominated to tetraBDE, and PBDFs found were tetra- to monobrominated compounds. Ohta and co-workers investigated BDE-209 in toluene and toluene/ethanol/water (1:3:6) both with an artificial light source and in sunlight and found debromination at least down to heptaBDE (18). In this case they found two major peaks of heptaBDE, one of which was identified as BDE-183. In a recently published study the photolytic debromination of BDE-209 was performed in toluene, on silica gel, on sand, on soil, and on sediment both with an artificial light source and in sunlight (19). Stepwise debromination occurred to form at least hexaBDEs and higher brominated PBDE congeners in toluene, while degradation down to heptaBDE was reported on sand. A study by Hua and co-workers (20) on photochemical degradation of BDE209 adsorbed to hydrated surfaces, under artificial and natural UV light, again showed similar results of photolytic debromination. Photochemical degradation studies have hitherto rarely been performed on other PBDE congeners than BDE-209. Recently, however, Peterman et al. (21) briefly reported on sunlight photolysis of a mixture of 39 PBDE congeners in triolein. Palm and co-workers have performed an in-depth study on photochemical degradation of BDE-209, BDE-153, VOL. 38, NO. 11, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3119
FIGURE 1. Structures of the PBDEs studied giving the substitution patterns of each PBDE congener. x and y are the positions of the bromine substituents in each phenyl ring. and BDE-47 (22). Both these studies are discussed in further detail below in relation to the results presented from this study. However, the clearest result from these studies is that of a slower rate of photochemical degradation with decreasing bromine content of the PBDE congener. The present study was aimed for examination of photolytical degradation rates of individual polybrominated diphenyl ethers, covering different bromination levels, in a solvent mixture of methanol/water. We have also examined the reaction rates for some selected PBDE congeners in pure methanol and in tetrahydrofuran. For several of the congeners, absorption spectra were obtained and quantum yields were calculated. While less emphasis was put on the identification of degradation products, this has been attempted for some of the PBDEs, and especially for BDE-209. The photolysis of BDE-209 has also been examined in pure water and in water containing dissolved humic substances. The 15 PBDEs examined were selected to be representative of PBDE congeners observed in the environment, to provide information on the influence of the bromination pattern on the rate of photolysis and to provide information on the possible mechanism of debromination.
Materials and Methods Chemicals. Decabromodiphenyl ether was purchased from Fluka Chemie AG, while the other 14 polybrominated diphenyl ethers were all synthesized in house. The purities of the used compounds (synthesized and purchased) were in all cases >98% according to GC-FID and GC-MS. The structures of the PBDE congeners with references to their route of synthesis are given in Figure 1. The PBDE congeners are numbered in analogy to the PCB numbering system as suggested by Ballschmiter et al. (23). Acetonitrile, methanol, toluene, and n-hexane were from Merck (Darmstadt, Germany). Tetrahydrofuran (THF) was purchased from Acros Organics (New Jersey) and water from Scharlau (Barcelona, Spain). All solvents were of HPLC grade. The humic substances used were the Nordic reference fulvic acid, available from the International Humic Substances Society (IHSS). Instruments. The UV lamp was a fluorescent tube TL 20W/09N from Philips (Holland). 3120
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 11, 2004
High-performance liquid chromatography (HPLC) was performed on a Shimadzu LC-9A (Kyoto, Japan) equipped with a UV detector UV100 from Spectra-Physics (Fremont, CA) and a C18 reversed-phase column (Ace 5 C18, 250 × 4.6 mm, 5 µm particles) from Advanced Chromatography Technologies (Aberdeen, Scotland). The detection wavelength was set at 235 nm, and the mobile phase was 8% or 10% water in acetonitrile. The flow rate was 2 mL min-1. Gas chromatography-mass spectrometry (GC-MS) was performed on a GCQ instrument from Finnigan Mat (San Jose, CA). The gas chromatograph was equipped with a DB-5 fused silica capillary column (15 m × 0.25 mm, 0.25 µm film thickness) from J&W Scientific (Folsom, CA). The column temperature was programmed: 80 °C, which was kept for 5 min, and then increased at 20 °C min-1 to 300 °C, which was held for 15 min. The injections were made in the splitless mode using an injection temperature of 260 °C. Helium was used as carrier gas. Mass spectra were recorded in electron ionization mode at an ion source temperature of 140 °C and electron energy of 70 eV. The scan range was 100-1000 amu. The UV spectrophotometer was a double-beam instrument U3000 from Hitachi Instruments (Tokyo, Japan). The scanning rate was 1 nm s-1 with a slit width of 1 nm. UV spectra were recorded in THF for the determination of quantum yields. UV Irradiation Experiments. The photochemical experiments were performed in a cylindrical vessel with a fluorescent tube placed longitudinally through the middle. Both the apparatus and the method for the quantum yield calculation have been presented in detail previously (24). Individual solutions of each of the 15 PBDE congeners were prepared by dissolving the pure solid compound in THF. A 1 mL portion of the THF solution was taken and made up to 2 L with a mixture of methanol/water (80:20) prior to illumination. A selection of nine and four of the individual PBDE congeners were also studied when dissolved in pure methanol and THF, respectively. The initial concentrations of the PBDE congeners were always less than 1 µM in the experiments. An experiment was also performed with BDE209 sorbed to humic substances (cf. below) in water. For the experiment with BDE-209 in water and in water containing humic substances the procedure was as follows:
FIGURE 2. Absorption spectra of seven PBDE congeners and the fluorescent tube emission. The line numbers are as follows: 1, BDE-209; 2, BDE-206; 3, BDE-203; 4, BDE-183; 5, BDE-155; 6, BDE-85; 7, BDE-77; line 8 represents the UV light source emission. A saturated solution (20 mL) of BDE-209 in ethanol was transferred to a conical flask. At this point in the experiment with humic substances, a solution (10 mL) of ethanol containing humic substances (50 mg) was added to the flask, and approximately 10 mL of the ethanol was evaporated using a stream of nitrogen. The flask was then filled with water (2 L). The solution was heated to 80 °C, kept there for 1 h under a constant flow of nitrogen, and protected from UV light. The solution was cooled to room temperature and thereafter transferred to the reaction vessel. In experiments using methanol or methanol/water, samples were taken directly from the reaction vessel with an HPLC syringe. For the experiments with THF as solvent, samples (5 mL) were taken with a volumetric pipet and then transferred to a test tube containing a mixture (10 mL) of acetonitrile/water (75:25). The samples were analyzed by HPLC, and the amount of unreacted compound was determined on the basis of peak area quantification. All calculations have been made on the assumption of first-order reaction.
Results Absorption spectra were measured in THF for 11 of the studied PBDE congeners, for which sufficient quantities were available. Seven of these spectra, representing the homologous series of tetraBDEs to decaBDE, are presented in Figure 2 together with the emission spectra of the fluorescent tube for wavelengths between 290 and 350 nm. Values of λmax and extinction coefficients at λmax are recorded in Table 1. The spectra for BDE-208 and BDE-139 had no local maxima at wavelengths longer than 290 nm, and extinction coefficients at 300 nm are presented in lieu of this. BDE-47 was sufficiently soluble to permit measurement of its absorption spectrum in methanol, presented together with its spectrum in THF in Figure 3. The rates of photodegradation and corresponding halflives for BDEs-209, -208, -207, -206, -203, -190, -183, -181, -155, -154, -139, -138, -99, -77, and -47 dissolved in methanol/ water (80:20) are presented in Table 1. Half-lives ranged over 3 orders of magnitude. Quantum yields were determined for BDEs-209, -208, -207, -206, -203, -183, -181, -155, -139, -77,
FIGURE 3. Comparison of absorbance spectra of 2,2′,4,4′-tetrabromodiphenyl ether (BDE-47) in methanol and THF. and -47, for which absorption data were available, and are presented in Table 1. The rates of photodecomposition of the nine PBDE congeners in pure methanol and of four congeners in THF are also presented in Table 1. All photolysis experiments were performed in at least duplicate (up to five times), and rate constants varied by less than 10% between experiments. Figure 4 shows the first-order decay of BDE209 in an aqueous solution of humic substances (k ) 3 × 10-5 s-1, R 2 ) 0.999), together with the results for an experiment performed in pure water (k ) 5 × 10-6 s-1, R 2 ) 0.904). Photolysis of BDE-209 in methanol/water (80:20) produced decomposition products of lower PBDEs and PBDFs. All three nonaBDEs were formed and at least seven octaBDEs, five of which were major products (Figure 5). Eight heptaBDEs, two of which were major peaks, were formed, as well as small amounts of hexaBDEs. No mono- to pentaBDEs were detected after 100 min of irradiation. Instead, all the major peaks eluting prior to the hexaBDEs (cf. Figure 5) had mass spectra consistent with those of mono- to pentaPBDFs. In addition, some minor peaks had mass spectra that may VOL. 38, NO. 11, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3121
TABLE 1. Rate of Decomposition of 15 PBDEs in Methanol/Water (80:20), 9 PBDE Congeners in Pure Methanol, and 4 PBDE Congeners in THF and Half-Lives, Spectral Data, and Quantum Yields for the UV Irradiation Experiments test compd BDE no.
rate of decomp in MeOH/H2O, k × 105 (s-1)
rate of decomp in MeOH, k × 105 (s-1)
209 208 207 206 203 190 183 181 155 154 139 138 99 77 47
40 17 19 8.3 3.7 3.0 0.67 3.2 0.41 0.33 0.40 0.62 0.3 0.06 0.07
65 30 32 13 5.9
a
1.1 5.6
rate of decomp in THF, k × 105 (s-1) 83 46
9.4
0.67
0.12
0.20
spectral absorption data half-life in MeOH/H2O (h)
λmax (nm)
(cm-1 M-1)
dissappearance quantum yield in MeOH/H2O
0.5 1.1 1.0 2.3 5.0 6.4 29 6.0 47 58 48 31 64 340 290
306 300a 300 300 297
2450 1730 2520 2500 2850
0.14 0.10 0.09 0.17 0.12
297 300 295
2640 910 2160
0.16 0.10 0.14
300a
650
0.14
284 291
2330 1850
0.29 0.22
E
No absorption maximum found in the range 280-360 nm.
rise to an almost identical set of products, although in the latter experiment a higher proportion of pentaBDFs was observed. Each of the three nonaBDEs produced a number of octaBDEs, although the major products were different for each nonaBDE congener. Hepta- and hexaBDEs and monoto pentaBDFs were also formed. The UV degradation products of two heptaBDEs (BDE-190 and BDE-183) and three hexaBDEs (BDE-155, BDE-154, and BDE-139) were also examined by GC-MS. These substances follow the same trend of consecutive debromination with the exception that tri- and tetraBDEs were also observed as products from these latter reactions.
Discussion FIGURE 4. Photochemical degradation of decabromodiphenyl ether (BDE-209) in water with (1) and without (2) humic substances. be cautiously interpreted as deriving from brominated methoxylated dibenzofurans (Figure 6). Photolysis of decaBDE in methanol/water, pure methanol, THF, and water containing dissolved humic substances gave
The rate of degradation of brominated diphenyl ethers by UV light in the sunlight region is dependent on the degree of bromination. Hence, lower brominated diphenyl ethers degrade slower than highly brominated congeners. The observed rate difference is up to 700 times between the slowest reacting PBDE studied here (BDE-77) and the fastest (BDE-209). Much of these differences can be explained by their absorbance behavior since the higher brominated
FIGURE 5. Total ion MS chromatogram of BDE-209 after UV irradiation for 100 min in methanol/water (80:20). The PBDE congener pattern, congeners identified, and PBDFs are indicated in the chromatogram. 3122
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 11, 2004
FIGURE 6. Mass spectrum of one of the photochemical degradation products of decabromodiphenyl ether (BDE-209) tentatively identified as a methoxylated tetrabromodibenzofuran (MeO-PBDF). diphenyl ethers absorb at longer wavelengths (Figure 2). Some more subtle differences may be apparent within groups with the same number of bromine substituents such as the three heptaBDEs analyzed here (Table 1, Figure 1). BDE-190 and BDE-181 both have one fully brominated phenyl ring, and their rates of phototransformation were similar to each other, while being around 5 times higher than the rate for the other heptabrominated isomer, BDE-183, which lacks a fully brominated ring. Unfortunately we did not have enough BDE190 to record its UV spectrum. Otherwise, the impact of structural parameters on the rate is not implicit from our results. For example, the rate of transformation of BDE-139 (with one ring substituted with four and the other with two bromine atoms) is lower than that of BDE-138 (both rings tribrominated), while that for BDE-138 (a hexaBDE) is very similar to that for BDE-183 (a heptaBDE with the same substitution pattern as BDE-138 but with an additional bromine at one of the two unsubstituted ortho positions (Figure 1)). Further, it may be worth mentioning that there is not a major degradation rate difference between the nonortho-substituted BDE-77 and the di-ortho-substituted BDE47 (Table 1). The relative rates of photolysis of the tetraBDE through heptaBDE congeners are very much in keeping with the results of Peterman et al. (21). After 2 h of exposure to sunlight of 39 congeners in triolein, they, too, observed fast degradation of those congeners with a fully brominated ring; BDE183 decayed 3 times slower than BDEs-190 and -181 but at a rate similar to that of BDE-138, and BDE-138 decayed faster than BDEs-155 and -154 (cf. Table 1). There is also a difference in the overall reaction rate depending on the solvent used. For instance, BDE-207 reacts 1.7 times faster in pure methanol, and 2.4 times faster in THF, as compared to the methanol/water mixture. While the difference in rate between the methanol and the methanol/ water mixture is almost constant for all congeners, the difference in rate between THF and methanol/water seems to increase with decreasing rate (a ratio of 2.1 for BDE-209 compared to 3.0 for BDE-47). The differences in rate between the different solvents cannot be explained by any difference in absorption behavior since the absorption spectra of BDE-47
in methanol and in THF (Figure 3) are very similar. As such, calculation of the quantum yield based on the absorption spectrum from THF gives a value of 0.24, while the same calculation made from the methanol spectrum gives a value of 0.22. The difference in reaction rate depends therefore on different quantum yields. This might be explained by pure methanol being a better “hydrogen donor” than methanol/ water and THF being even better than methanol. The decomposition of decaBDE generates a number of decomposition products substituted with fewer bromine atoms than the starting material. Several large peaks of nona-, octa-, and heptaBDEs were formed as shown by GC-MS (Figure 5), but also small amounts of hexaBDEs are observed. Whereas we can identify the absolute structures for all three nonaBDE products, the structures of the majority of the octaand heptaBDEs cannot yet be specified. Compounds formed photolytically from BDE-209, with less than six bromines, seem to be PBDFs. The product distribution exemplified in Figure 5 is observed after an illumination period of less than 2 h. If it is assumed that debromination takes place by loss of one bromine substituent at a time, it is perhaps not surprising that hexaBDE compounds are barely detectable after 100 min, given the photolysis rates of the nona- and heptaBDEs studied here (cf. Table 1). The presence of diand triBDFs on this time scale, at relatively high levels, presents an interesting question about their mechanism of formation. We cannot answer this with certainty, but one possible explanation is that the formation of the furan ring is possible for PBDEs of any level of bromination as long as one ortho position is nonbrominated. The subsequent photolytic debromination of the PBDFs formed in this way would need to be very fast for this mechanism to account for the occurrence of di- and triBDFs at much higher levels than tetra and higher brominated BDFs. This possible explanation is at least supported by the high decomposition rate and quantum yields of PBDFs (25, 26), and PBDF congeners are expected to have stronger absorbance at longer wavelengths than PBDEs. The mechanism for the formation of PCDFs from polychlorinated diphenyl ethers has previously been discussed by Norstro¨m et al. (27). We believe that a similar mechanism is operating when PBDFs are formed. VOL. 38, NO. 11, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3123
Due to its extremely low water solubility, photolysis experiments of BDE-209 in the aqueous phase are practically impossible. The results of our attempt to perform the photolysis in almost pure water (
