Comparative evaluation of liquid chromatography–mass spectrometry versus gas chromatography–mass spectrometry for the determination of hexabromocyclododecanes and their degradation products in indoor dust

Journal of Chromatography A, 1190 (2008) 333–341

Contents lists available at ScienceDirect

Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma

Comparative evaluation of liquid chromatography–mass spectrometry versus gas chromatography–mass spectrometry for the determination of hexabromocyclododecanes and their degradation products in indoor dust Mohamed Abou-Elwafa Abdallah a , Catalina Ibarra a , Hugo Neels b , Stuart Harrad a , Adrian Covaci b,∗ a b

Division of Environmental Health and Risk Management, School of Geography, Earth and Environmental Sciences, University of Birmingham, Birmingham B15 2TT, UK Toxicological Centre, University of Antwerp, Universiteitsplein 1, 2610 Wilrijk, Belgium

a r t i c l e

i n f o

Article history: Received 21 January 2008 Received in revised form 2 March 2008 Accepted 4 March 2008 Available online 7 March 2008 Keywords: Liquid chromatography Gas chromatography Mass spectrometry Hexabromocyclododecanes HBCDs Degradation products Indoor dust

a b s t r a c t Domestic and office dust samples (n = 37) were analyzed for hexabromocyclododecanes (HBCDs) using gas chromatography–electron-capture negative ionization–mass spectrometry (GC–ECNI/MS) and liquid chromatography–electrospray tandem mass spectrometry (LC–ESI/MS/MS). To determine the best method to quantify HBCDs using GC–ECNI/MS, BDE 128 was used as internal standard (I.S.) in all samples, while 13 C-labeled ␣-HBCD was used as I.S. in some samples. Total HBCD concentrations (sum of ␣-, ␤-, and ␥-HBCD diastereomers) were calculated using response factors (RFs) for ␣- and ␥-HBCD as individual diastereomers and using an average RF for both diastereomers. Statistical comparison showed that concentrations obtained via GC–ECNI/MS were statistically indistinguishable (p > 0.05) from those obtained using LC–ESI/MS/MS. The closest match between the two techniques was obtained using [13 C]␣-HBCD as I.S. and the average RF for ␣- and ␥-HBCDs. Excellent linear correlations (Pearson coefficient values r > 0.9) were obtained between the GC–ECNI/MS and LC–ESI/MS/MS results, with slopes ranging from 0.76 to 1.36. Pentabromocyclododecenes (four isomers) and tetrabromocyclododecadienes (two isomers) were detected in the studied samples and were identified as degradation products of HBCDs after separation from the parent compound on the basis of both retention time and mass spectrum. This finding suggests that the elimination of HBr is the major degradation pathway for HBCDs in dust. © 2008 Elsevier B.V. All rights reserved.

1. Introduction Hexabromocyclododecane (HBCD) is a high-volume brominated flame retardant (BFR) with a total world market demand of 16.7 × 106 kg in 2001, making it the second most widely used BFR in Europe [1]. It is used as an additive to expanded and extruded polystyrene foams for thermal insulation and to a lesser extent for back-coating of fabrics for furniture [1]. HBCD is an aliphatic, brominated cyclic alkane produced by bromination of cyclododecatri-1,5,9-ene, a process which leads theoretically to the formation of a mixture of 16 stereoisomers. The commercial mixtures consist mainly of the ␣-, ␤-, and ␥-HBCD diastereomers, with the latter being predominant [2]. HBCD’s low vapor pressure (6.27 × 10−5 Pa), very low water solubility (3.4 ␮g/L), fairly high log Kow value of 5.6 combined with its persistence, indicate that it may bioaccumulate in fatty tissues [3]. This is compounded by the fact that oral exposure to HBCDs may induce hepatic cytochrome P450 enzymes in rats [4] and can alter the normal uptake of neu-

∗ Corresponding author. Tel.: +32 3 820 2704; fax: +32 3 820 2722. E-mail address: [email protected] (A. Covaci). 0021-9673/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2008.03.006

rotransmitters in rat brain [5]. There are also indications that it can disrupt the thyroid hormone system [6] and induce cancer through a non-mutagenic mechanism in humans [7]. Following its first detection in fish and sediment samples from the River Viskan in Sweden [8], HBCD has attracted increasing attention, with its levels being reported in a wide variety of biotic and abiotic samples [9]. Traditionally, HBCD has been analyzed by gas chromatography–mass spectrometry (GC–MS) operated in electron-capture negative ionization (ECNI) mode. Bromide ion detection (m/z = 79 and 81) is used commonly because of its high sensitivity [3]. However, the GC technique has a number of serious drawbacks, specifically the impossibility of separating the HBCD diastereomers, the thermal rearrangement of HBCD diastereomers above 160 ◦ C, the decomposition of HBCDs at temperatures above 240 ◦ C, and the partial or complete degradation of HBCDs in dirty GC systems [10]. Furthermore, isotopically-labeled (e.g. [13 C]-) HBCD standards cannot be used when only the bromide ion is monitored. In addition, other brominated compounds of potential use as internal standards, such as polybrominated diphenyl ethers (PBDEs), have a better thermal stability than HBCDs and therefore cannot compensate for the decomposition that occurs in the GC system [3].

334

M.A.-E. Abdallah et al. / J. Chromatogr. A 1190 (2008) 333–341

In 2003, Budakowski and Tomy [11] introduced liquid chromatography coupled tandem mass spectrometry (LC–MS/MS) for the chromatographic resolution and determination of the major HBCD diastereomers. Since then, several refinements and modifications to enhance resolution and/or sensitivity have been reported for the LC–MS/MS analysis of HBCDs [10,12,13]. Presently, LC–MS/MS is considered the method of choice for the analysis of HBCDs, because, in addition to its ability to chromatographically resolve the HBCD diastereomers, no thermal degradation or analyte composition transmutation occurs during analysis and it is compatible with the use of isotopically-labeled HBCD diastereomers as internal standards. These labeled standards can compensate for both instrumental fluctuations of the MS system and for matrixrelated ion suppression or enhancement effects that can occur in the ion source [10,12]. Despite these clear advantages of LC–MS/MS, the higher capital cost of such instrumentation, means that it is possible that the use of GC–MS, at least for screening purposes, has been too early discounted. Currently, very little is known about the comparative performance of GC–MS and LC–MS/MS techniques for the determination of HBCDs. Recently, an interlaboratory comparison for HBCDs in fish and standard solutions [14] has shown good agreement between results obtained by GC–MS or LC–MS. However, none of the reporting laboratories presented results using both techniques. If such a comparison revealed GC–MS to offer acceptable performance, then it would open up a simple, more widely available and sensitive technique for the determination of HBCDs, that could for example be used to screen samples for HBCDs and PBDEs, in the same GC–ECNI/MS run. The principal aim of this study is to compare the performance of GC–ECNI/MS and LC–ESI/MS/MS for the determination of HBCDs in indoor dust. In addition, strategies to enhance the selectivity and accuracy of the GC–ECNI/MS method for HBCD determination are discussed. The present study also reports for the first time the identification by LC–MS/MS of degradation products of HBCDs, and their detection in indoor dust. Elucidation of such products offers potential insights into the pathways of HBCD degradation. 2. Experimental 2.1. Chemicals All solvents used in the extraction and analysis procedures for LC–MS/MS were of HPLC-grade quality (Fisher Scientific, Loughborough, UK). Pesticide-grade Florisil (60–100 mesh) and anhydrous sodium sulfate were obtained from Acros Organics (Geel, Belgium). Native ␣-, ␤-, and ␥-HBCD standards were obtained from Cambridge Isotope Labs. (Andover, MA, USA), while isotopicallylabeled [13 C]␣-, ␤-, and ␥-HBCDs and [2 H18 ]␥-HBCD (␥-HBCD-d18 ) were purchased from Wellington Labs. (Guelph, Canada). For the preparation of samples subjected to GC–MS analysis, silica (60–100 mesh), concentrated sulfuric acid and anhydrous sodium sulfate were obtained from Merck (Darmstadt, Germany). All solvents were of GC grade quality (Merck). BDE 128 was purchased from Accustandard (New Haven, CT, USA), while the polychlorinated biphenyl (PCB) 206 was obtained from Dr. Ehrenstorfer (Augsburg, Germany). 2.2. Sample collection Dust samples (n = 37) were collected from randomly chosen houses and offices in Birmingham, UK using a Nilfisk Sprint Plus 1600 W vacuum cleaner. Sampling was conducted in March–June 2007 according to a standard protocol, where 1 m2 of carpet was vacuumed for 2 min in each location and in case of bare floors 4 m2

for 4 min [15,16]. Samples were collected using nylon sample socks (25 ␮m pore size) that were mounted in the furniture attachment tube of the vacuum cleaner. After sampling, socks were closed with a twist tie, sealed in an air-tight plastic bag and stored at −20 ◦ C. Before and after sampling, the furniture attachment was cleaned thoroughly using an isopropanol-impregnated disposable wipe. 2.3. Analytical protocols 2.3.1. LC–MS/MS analysis The analytical procedure used for the sample preparation and analysis of HBCDs in dust via LC–MS/MS has been described previously [16]. A brief summary follows. 2.3.1.1. Sample preparation and extraction. Samples were sieved through a 500 ␮m mesh size sieve, weighed accurately and then extracted using pressurized liquid extraction (ASE 300 system; Dionex, Oakville, Canada). Dust samples (typically between 100 and 250 mg) were loaded into pre-cleaned 66 mL cells containing 1.5 g Florisil and Hydromatrix (Varian, Palo Alto, CA, USA) to fill the void volume of the cells, spiked with 25 ng of each of 13 C-labeled ␣-, ␤, and ␥-HBCD as internal standards and extracted with hexane:dichloromethane (1:1, v/v) at 90 ◦ C and 1500 psi. The heating time was 6 min, static time 5 min, purge time 100 s, flush volume 50%, and the number of static cycles was 3. 2.3.1.2. Clean up. The crude extracts were concentrated to 0.5 mL using a Zymark Turbovap II (Hopkinton, MA, USA), then washed with 98% sulfuric acid. After phase separation, the hexane layer was transferred onto a Florisil column topped with sodium sulfate and eluted with 30 mL of hexane:dichloromethane (1:1, v/v). The eluate was evaporated to dryness under a gentle stream of N2 and the dried extract was reconstituted in 100 ␮L of ␥-HBCD-d18 (25 pg/␮L in methanol) used as recovery determination (or syringe) standard, used to determine the recoveries of internal standards for quality assurance/quality control (QA/QC) purposes. 2.3.1.3. Analysis. Separation of ␣-, ␤-, and ␥-HBCDs was achieved using a dual pump Shimadzu LC-20AB Prominence liquid chromatograph (Kyoto, Japan) equipped with SIL-20A autosampler, a DGU-20A3 vacuum degasser and a Varian Pursuit XRS3 C18 analytical column (150 mm × 2 mm I.D., 3 ␮m particle size). A mobile phase of (a) 1:1 methanol/water with 2 mM ammonium acetate and (b) methanol at a flow rate of 120 ␮L/min was applied for elution of the target compounds; starting at 50% (b) then increased linearly to 100% (b) over 3 min, held for 5 min, followed by a linear decrease to 65% (b) over 2.5 min and held for 5.5 min. The three diastereomers were baseline separated on the C18 column with retention times of 12.3, 12.9 and 13.3 min for ␣-, ␤- and ␥-HBCD, respectively. Mass spectrometric analysis was performed using a Sciex API 2000 triple quadrupole mass spectrometer (Applied Biosystems, Foster City, CA, USA) operated in electrospray negative ionization (ESI) mode. Direct infusion of the target compounds (␣-, ␤- and ␥-HBCDs, native and 13 C-labeled isomers, 2 ng/␮L each in MeOH) into the MS/MS system was performed with a built-in Harvard syringe pump at a flow rate of 10 ␮L/min. The infusion experiments served for the tuning and adjusting the source and the compound-specific parameters during the method development. MS/MS detection operated in the MRM mode was used for quantitative determination of the HBCD isomers based on m/z 640.6 → m/z 79, m/z 652.4 → m/z 79 and m/z 657.7 → m/z 79 for the native, 13 C-labeled and 2 H -labeled diastereomers, respectively. The opti18 mized MS/MS parameters for the selected MRM transitions are given in Table 1. The optimum collision energy, declustering and collision cell potentials were found to be the same for all the target compounds.

M.A.-E. Abdallah et al. / J. Chromatogr. A 1190 (2008) 333–341 Table 1 Optimized MS/MS parameters for the analysis of HBCDs Parameter (units)

Value

Curtain gas (a.u.) Turbo gas temperature (◦ C) Ionspray voltage (V) Declustering potential (V) Focusing potential (V) Collision gas (a.u.) Collision energy (eV) Cell entrance potential (V) Collision cell exit potential (V)

35 500 −4500 −5 −365 5 40 −6 −10

These parameters were obtained through infusion experiments using native and 13 C-labeled ␣-, ␤- and ␥-HBCDs (2 ng/␮L each in MeOH). a.u., arbitrary units

2.3.2. GC–MS analysis 2.3.2.1. Sample preparation and extraction. Dust samples (typically between 150 and 250 mg) were spiked with internal standards (50 ng BDE 128 and 100 ng [13 C]␣-HBCD) and were then extracted with 100 mL n-hexane:acetone (3:1, v/v) in hot Soxhlet extraction mode for 2 h as described previously [17]. 2.3.2.2. Clean up. The crude extract was cleaned up by 8 g of acidified silica (44%, w/w, concentrated sulfuric acid). HBCDs were eluted with 20 mL of hexane, followed by 15 mL of dichloromethane. The eluate was concentrated by rotaryevaporation and evaporated to dryness by a gentle nitrogen stream. The extract was reconstituted in 200 ␮L of isooctane (containing 100 pg/␮L PCB 206) used as recovery determination standard. 2.3.2.3. Analysis. The determination of HBCDs was performed with an Agilent 6890GC-5973MS system equipped with a 15 m × 0.25 mm I.D., 0.10 ␮m DB-5 capillary column (J&W Scientific) and operated in the ECNI mode. The ion source, quadrupole and interface temperatures were 230, 150 and 300 ◦ C, respectively. Helium was used as carrier gas at constant flow (1.5 mL/min) and

335

methane as reagent gas. The MS was operated in the selected ion monitoring (SIM) mode and the electron multiplier voltage was set at 2100 V. A 1-␮L volume of the extract was injected in solvent vent mode (injector temperature at 90 ◦ C, kept for 0.05 min, then increased with 700 ◦ C/min to 295 ◦ C, vent time 0.03 min, vent flow 75 mL/min). The splitless time was 1.25 min. The oven temperatures was programmed from 90 ◦ C, kept for 1.25 min, then increased with 10 ◦ C/min to 200 ◦ C, then increased with 20 ◦ C/min to 300 ◦ C, kept for 10 min. Dwell times were 35 ms. Ions m/z 79 and 81, together with ions m/z = 561 and 573 deriving from the fragment [M − Br]− for HBCD and [13 C]HBCD, respectively, were monitored for the entire run, while the recovery standard PCB 206 was measured using ions m/z = 464 and 466. 2.4. Quality assurance/quality control Analyte identification was based on its retention time relative to the internal standard used for quantification, ion chromatograms and intensity ratios of the monitored ions. A deviation of the ion intensity ratios within 20% of the mean values obtained for the calibration standards was considered acceptable. For LC–MS/MS, individual five-point calibration plots (peak area ratios − analyte/internal standard − plotted against concentration ratios − analyte/internal standard) were created for the ␣-, ␤-, and ␥-HBCD diastereomers using the corresponding 13 C-labeled diastereomer as internal standard. The response factors (RFs) for each diastereomer were calculated at each of the five concentrations used for the calibration plot and averaged. With each batch of samples, a calibration standard containing the three HBCD diastereomers and their 13 C-labeled analogues was analyzed. For data from that batch to be deemed acceptable, the RFs obtained from this standard had to fall within 25% of the average of the RFs obtained for the five concentration levels of the calibration plot. For GC–MS, the RFs for ␣- and ␥-HBCDs were calculated relative to [13 C]␣-HBCD. This approach most closely parallels the

Fig. 1. LC–MS/MS chromatogram of standard solution (200 pg/␮L) of [12 C]- and [13 C]HBCDs and ␥-HBCD-d18 showing baseline separation of the three HBCD diastereomers and the absence of degradation products (MRM 480.4 to 79 for TBCDs and MRM 560.8 to 79 for PBCDs).

336

M.A.-E. Abdallah et al. / J. Chromatogr. A 1190 (2008) 333–341

degradation pattern of the native compounds during GC analysis. Interestingly, the two diastereomers had different RF values of 1.02 and 0.74, respectively. This finding is in agreement with the fact that HBCD diastereomers exhibit different thermal stability and thus, different thermal degradation during GC analysis [18,19]. The RFs for ␣- and ␥-HBCDs (monitored ions m/z 561 or 79) were also calculated based on BDE 128 as internal standard. The ␣- and ␥-HBCD diastereomers again generated two different RF values of 0.006 and 0.004 for m/z 561 and of 0.606 and 0.430 for m/z 79. Despite these diastereomer-specific RFs, the inability of GC to quantify the relative abundance of the different diastereomers in samples emphasized that an RF based on the average of those derived for ␣- and ␥-HBCD should also be derived. Concentrations of HBCDs were calculated based on each of these diastereomer-specific and averaged RFs, in an effort to identify the most appropriate computation method. For every five samples, one method blank was conducted. This consisted of 1 g anhydrous sodium sulfate treated in identical fashion to samples. Concentrations of HBCDs in blanks never exceeded 1% of the concentration in a sample from the same batch and data were thus not blank-corrected. Field blanks (n = 6) were also conducted. These comprised of sodium sulfate (1 g) “sampled” using the vacuum cleaner and sock according to the standard protocol and treated as a sample. For LC–MS/MS, method detection limits for individual diastereomers were governed by field blanks and were typically 0.2 ng/g. In comparison, method detection limits for HBCDs via GC–MS were 5 and 50 ng/g calculated based on m/z = 79 and 561, respectively. In the absence of an appropriate standard reference material for which certified concentrations of HBCDs exist, method performance was assessed through the analysis of five replicates of SRM 2585 (house dust) which has indicative values for HBCDs [20]. Ongoing monitoring of method accuracy was performed through the analysis of one SRM sample for every 20 dust samples. For samples where particularly elevated concentrations of HBCDs were found (i.e. such that the internal standard “spiking” levels were inappropriately low), a fresh aliquot of the dust sample in question was analyzed using a smaller quantity of dust and a higher amount of internal standards.

3. Results and discussion

Fig. 2. GC–ECNI/MS total ion chromatogram and full scan mass spectra of 13 C␣-HBCD standard (concentration 1 ng/␮L in isooctane) (A) and ␣-HBCD standard (concentration 1 ng/␮L in isooctane) (B). The retention time of HBCD peak was 16.6 min. The standard solutions were analyzed using conditions described in Section 2.

3.1. Chromatographic properties of HBCDs The major three diastereomers of HBCD were baseline separated via the LC–ESI/MS/MS upon monitoring the ion transition m/z 640.6 [M − H]− → 79 in the negative ion mode (Fig. 1). In contrast to GC, the absence of high temperatures in the LC system means that no degradation was observed for the native, 2 H18 - or 13 C-labeled HBCD standards (Fig. 1). Within the tandem MS, the turbo gas temperature of 500 ◦ C is used to maintain a clean ion source, but it has no direct effect because of the “uniform cushioning” of the analyte with the solvent in the ES source which, together with the very short transit time, allows the HBCDs to be ionized and enter the MS system without being thermally decomposed. In GC–ECNI/MS, only one peak is obtained corresponding to the total HBCDs due to the co-elution of the three isomers. Several ions were monitored and evaluated for their ability to quantify HBCDs. BDE 128 was used as internal standard in all analyzed samples; however, BDE 128 is more thermally stable compared to HBCD and hence cannot compensate for the thermal decomposition of HBCDs during GC elution. In an effort to minimize underestimation of HBCDs in real samples caused by thermal decomposition at temperatures >240 ◦ C, 13 C-␣-HBCD was used as internal standard in some samples, on the premise that 13 C-labeled and native HBCDs will behave identically in the GC system. Hence, quantifying

relative to [13 C]HBCD should compensate for within-GC degradation, as well as for instrumental variations in sensitivity during and between sample runs. In GC–ECNI/MS, [13 C]␣-HBCD was monitored using m/z 573 (Fig. 2A), while the native HBCDs were monitored at m/z 561 (Fig. 2B) corresponding to the ion [M − Br]¯ . This enhances method selectivity and allows better structural confirmation of HBCDs. The present paper reports for the first time the use of [13 C]␣-HBCD for GC–ECNI/MS using specific ions such as [M − Br]− . However, this enhancement in method selectivity is accompanied by a decrease in sensitivity as the peak at m/z 561 has much less intensity than the “traditionally” monitored peak at m/z 79 (Fig. 2A and B). Therefore, in samples with low concentrations of HBCDs, where the peak corresponding to m/z 561 cannot clearly be measured, only m/z 79 was used for quantitation. Concentrations of HBCD in dust samples were calculated based on RFs obtained using [13 C]␣-HBCD and BDE 128 as internal standards. The obtained GC–ECNI/MS results were classified into three sets according to the internal standard and ions used for quantitation (Table 2). The data in each set were compared statistically to the LC–MS/MS results obtained for the corresponding samples using a paired t-test for means and F-test for variances (Table 3). No

M.A.-E. Abdallah et al. / J. Chromatogr. A 1190 (2008) 333–341

337

Table 2 Total HBCD concentrations (ng/g) in dust samples using LC–MS/MS and GC–ECNI/MS Sample

LC–ESI/MS/MS

GC–ECNI/MS [13 C]HBCDs internal standard

BDE 128 internal standard m/z 561

S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 S12 S13 S14 S15 S16

12,200 2,500 2,400 880 3,500 780 4,000 1,500 1,050 1,600 8,500 890 390 6,400 15,700 4,700

S17 S18 S19 S20 S21 S22 S23 S24 S25 S26 S27 S28 S29 S30 S31

2,100 111,000 960 1,300 13,700 2,000 1,700 1,400 11,900 16,100 14,000 14,400 6,800 7,400 18,500

S32 S33 S34 S35 S36 S37

50 170 730 320 180 290

a

m/z 79

RF of ␣-HBCD

RF of ␥-HBCD

Average

RF of ␣-HBCD

RF of ␥-HBCD

Average

12,900 1,500 1,500 640 4,200 1,100 3,900 1,300 1,200 1,400 9,700 610 80 4,800 9,400 3,900

17,700 2,100 2,000 880 5,800 1,500 5,300 1,800 1,200 1,900 13,400 830 110 6,600 13,000 5,300

15,300 1,800 1,800 750 5,000 1,300 4,600 1,300 1,200 1,600 11,600 720 90 5,700 11,200 4,600

11,700 5,300 1,600 770 7,000 1,200 5,600 1,500 1,200 1,300 11,300 650 80 4,570 11,320 4,200

17,600 7,900 2,300 1,200 10,500 1,800 8,400 2,200 1,800 1,900 16,900 970 120 6,900 17,000 6,300

14,600 6,600 1,900 960 8,700 1,500 7,000 1,800 1,500 1,600 14,100 810 100 5,700 14,200 5,300

1,200 96,000 690 860 14,000 1,900 2,800 1,400 8,100 13,200 7,200 8,800 6,200 4,900 15,500

1,800 144,000 1,000 1,300 20,900 2,800 4,200 2,100 12,200 19,800 10,800 13,200 9,300 7,300 23,200

1,540 120,000 860 1,100 17,400 2,400 3,500 1,800 10,200 16,500 9,000 11,000 7,800 6,100 19,300

–a –a –a –a –a –a

–a –a –a –a –a –a

–a –a –a –a –a –a

RF of ␣-HBCD

RF of ␥-HBCD

Average

1,200 81,000 700 870 13,200 1,900 2,600 1,400 8,100 13,000 7,100 8,700 6,200 4,800 14,500

1,700 114,000 970 1,200 18,400 2,700 3,600 2,000 12,200 18,100 10,000 12,200 8,600 6,800 20,300

1,500 98,000 840 1,000 15,800 2,300 3,100 1,700 10,200 15,500 8,600 10,400 7,400 5,800 17,400

30 90 540 100 110 220

30 120 750 140 150 300

30 100 640 120 130 260

Concentrations based on m/z could not be calculated due to low ion intensities.

statistically significant differences (p > 0.05) were found between the three sets of GC/MS data and the concentrations obtained via LC–MS/MS. The closest match between the two techniques was obtained when [13 C]␣-HBCD was used as internal standard and the RF deployed was the average of those obtained for ␣- and ␥HBCD. An important point is that the superior performance of this averaged RF may be due to the roughly equal abundance of the ␣- and ␥-HBCD diastereomers in most dust samples revealed by the LC–MS/MS results. For matrices such as biota, where the ␣HBCD diastereomer predominates, then a RF derived solely for the ␣-HBCD may be more appropriate. The analyzed dust samples were subjected to a thorough clean up process prior to analysis which reduces substantially the occur-

rence of matrix effects during MS analysis. This, together with the deployment of an RF based on the average of the RFs for the ␣- and ␥-HBCDs, are likely to account for the good correlations obtained between the GC–MS and LC–MS/MS results (Table 3). The R-values obtained here exceed those reported by van Leeuwen and de Boer [21] for the correlation between GC–MS and LC–MS results of HBCD analysis in fish samples. In our study, the slopes of the linear correlations between the GC–MS and LC–MS/MS results ranged from 0.76 to 1.36 indicating a close match between the results obtained by the two techniques. This also differs from the results of van Leeuwen et al. who reported GC–MS results to be five times higher than LC–MS results in eels [22] and a slope of 0.23 for the linear correlation between

Table 3 Statistical comparison (t-test, F-test and Pearson correlation coefficients) between values of total HBCDs issued by LC–MS/MS and GC–ECNI/MS [13 C]HBCDs internal standard (n = 16)*

BDE 128 internal standard m/z 561 (n = 31)**

t-test paired for means F-test for variances Pearson correlation (r) Slope GC versus LC

m/z 79 (n = 21)***

RF of ␣-HBCD

RF of ␥-HBCD

Average

RF of ␣-HBCD

RF of ␥-HBCD

Average

RF of ␣-HBCD

RF of ␥-HBCD

Average

1.39 1.34 0.92 1.05

1.51 1.42 0.92 0.76

0.20 1.06 0.92 0.89

2.15 1.34 0.99 1.15

2.43 0.60 0.99 0.77

1.46 0.86 0.99 0.92

1.95 1.09 0.99 1.36

1.15 0.94 0.99 0.97

1.65 1.29 0.99 1.13

The confidence level was set at 95%. * t-critical = 2.13, F-critical = 2.40; ** t-critical = 2.73, F-critical = 1.84; *** t-critical = 2.08, F-critical = 2.12.

338

M.A.-E. Abdallah et al. / J. Chromatogr. A 1190 (2008) 333–341

Table 4 Concentrations of HBCDs (ng/g dust) in SRM 2585 compared to indicative values reported by Keller et al. [20] Average concentrations ± SD

Indicative values [19] LC–MS/MS (n = 5)

␣-HBCD

␤-HBCD

␥-HBCD

HBCDs

19 ± 3.7 19 ± 3.6

4.3 ± 1.1 4.4 ± 0.4

120 ± 22 125 ± 20

143 ± 27 150 ± 20

[13 C]HBCDs internal standard (n = 1) RF of ␣-HBCD RF of ␥-HBCD Average

105 145 125

BDE 128 internal standard m/z 561 (n = 5) RF of ␣-HBCD RF of ␥-HBCD Average

95 ± 9 140 ± 15 120 ± 10

m/z 79 (n = 4) RF of ␣-HBCD RF of ␥-HBCD Average

105 ± 10 150 ± 15 130 ± 10

LC–ESI/MS/MS and GC–ECNI/MS results of HBCDs analysis in fish [21]. 3.2. Method performance The accuracy of the applied analytical methods was checked via replicate analysis (n = 5) of SRM 2585 (house dust) which has indicative values for HBCDs ([20], Table 4). There was a good agreement between the LC–MS/MS results and the indicative values, while for the GC–ECNI/MS results, only the concentrations estimated using the RF of ␥-HBCD are closely matching the indicative values. The most probable explanation for this is that the ␥-HBCD isomer is the predominant isomer in the SRM 2585 material (>83% of total HBCDs as determined by LC–MS/MS). Furthermore, using the RF of the thermally stable ␣-HBCD isomer may underestimate substantially the total HBCD concentration as it decomposes in the GC system to a lesser extent than the ␥-HBCD isomer does. Method precision was calculated as the relative standard deviation of concentrations obtained from five replicates of SRM 2585. Method precision (RSD) was <15% for both LC–MS/MS and GC–ECNI/MS methods, indicating good precision for both methods.

Fig. 3. GC–ECNI/MS chromatograms (selected ion monitoring mode) of ions m/z 79 (A) and m/z 561 (B) for a dust sample (S26) showing enhanced selectivity and decreased sensitivity (see absolute value of the abundance) obtained when monitoring the ion m/z 561 ion. Degradation products, PBCDs and TBCDs, are also indicated. The sample was analyzed using conditions described in Section 2.

Average recoveries ± standard deviations for the [13 C]HBCD diastereomers as determined by LC–MS/MS were: ␣-HBCD = 82 ± 8, ␤-HBCD 88 ± 5, and ␥-HBCD 83 ± 8%. No HBCDs could be detected in method blanks (n = 6) for the dust analysis procedures consisting of sodium sulfate (1 g). Detectable, but very low concentrations (typically 0.1–0.5 ng/g HBCDs) were obtained for field blanks (n = 6). These consisted of sodium sulfate (1 g) “sampled” using the vacuum cleaner and sock according to the standard protocol and treated as a sample. Concentrations in samples were thus corrected for the contamination detected in the associated field blank.

Fig. 4. GC–EI/MS chromatogram of a dust sample (S18) showing the target compounds (HBCDs) and the degradation products (PBCDs and TBCDs). Analytical conditions are similar to those described in Section 2 except for the analytical column (HT-8, 25 m × 0.22 mm I.D., 0.25 ␮m) and temperature of the ion source (230 ◦ C).

M.A.-E. Abdallah et al. / J. Chromatogr. A 1190 (2008) 333–341

3.3. Degradation products of HBCD in dust The thermal isomerization and degradation of HBCDs at the high temperatures used during the GC/MS analysis have been reported previously [18]. These degradation products are observed

339

in the GC–ECNI/MS chromatograms for both samples and standards (Fig. 3), which makes it very difficult to determine if these degradation products are present originally in the samples or produced as a result of thermal degradation in the GC system. The identity of the degradation products was investigated by GC–MS which separated

Fig. 5. Full scan EI–MS spectra of HBCDs (A), PBCDs (B) and TBCDs (C) for dust sample S18.

340

M.A.-E. Abdallah et al. / J. Chromatogr. A 1190 (2008) 333–341

the degradation products from HBCDs (Fig. 4) and provided their full scan mass spectra (Fig. 5). The issue of whether these products are merely formed in the GC system may be resolved by the use of LC–MS/MS, during which no thermal degradation occurs and therefore, no degradation products are observed in the chromatograms of standards (Fig. 1). Two different degradation products were observed in almost all dust samples and in SRM 2585. The products thus identified are pentabromocyclododecenes (PBCDs) and tetrabromocyclododecadienes (TBCDs), which were monitored at transitions m/z 560.8 → 79 and m/z 480.4 → 79, respectively. The mass spectra of HBCDs, PBCDs and TBCDs are very similar suggesting that these compounds have a similar structure (Fig. 5). Furthermore, the low-mass fragment ions at m/z 53, 65/67, 79, 91, 105, and 117 deepens the assumption of an aliphatic backbone for the degradation products. At higher masses, the fragment ions at m/z 317 (2 Br) and m/z 399 (3 Br) are also in agreement with the mass spectrum of HBCDs. Unfortunately, GC–MS does not allow the correct identification of the number of PBCD or TBCD isomers, since, similar to HBCDs, these compounds also co-elute. No peaks were detected for lower brominated (1–3 Br atoms) degradation products of HBCDs in the dust samples. The LC–MS/MS chromatograms of PBCDs (Fig. 6) showed four well-resolved peaks indicating the presence of at least four major isomers. The PBCDs were well separated from the parent HBCDs on retention time basis (Figs. 1 and 6), with the PBCDs eluting between 10.13 and 11.90 min and the HBCDs between 12.32 and 13.32 min. The MS/MS technique enabled further structure confirmation through the monitoring of the PBCDs at m/z 560.8 → 79.0, which differs from that of the parent HBCDs monitored at m/z 640.9 → 79.0. These peaks were assigned the names of PBCD1 to PBCD4, with the first eluting isomer being PBCD1. PBCDs were reported previously by Barontini et al. [23,24] as thermal degradation products of HBCDs having a molecular weight of 561 g/mol and a total number of seven isomers detected and identified via GC–MS. Hiebl and Vetter [25] also detected PBCD (one isomer) as a metabolite/degradation product of HBCDs in chicken egg and in fish via GC–EI/MS. However, in the present study, only four PBCD isomers were detected and identified via LC–MS/MS indicating that the degradation pathways of HBCDs under ambient environmental conditions are different from those induced under the harsher thermal conditions (300–400 ◦ C) employed by Barontini et al. [23,24]. This can also explain the difference in the profiles of degradation products obtained by GC–MS and LC–MS/MS analysis of dust samples. During the GC run, the sample is exposed to high temperatures

Fig. 6. LC–MS/MS chromatogram of a dust sample (S26) showing PBCD isomers (MRM 560.8 to 79).

Fig. 7. LC–MS/MS chromatogram of SRM 2585 (house dust) showing TBCD isomers (MRM 480.4 to 79).

and the degradation products may be different from those originally present in the sample and detected as such by the LC–MS/MS. Efforts were made to baseline resolve PBCD3 from PBCD4 on the LC column by slightly changing the solvent gradients and flow rate. However, since this was not possible, no further trials were made to avoid alterations of the method parameters that might affect the separation and retention times of the main target compounds, in this case the ␣-, ␤-, and ␥-HBCD isomers. Only two peaks were observed for TBCDs in chromatograms indicating the presence of two congeners (Fig. 7). However, the first eluted peak (TBCD1) always was less intense than the second peak (TBCD2), which is also observed much more frequently in the dust samples. The detection of these two TBCD isomers in dust samples indicates that the degradation pathways of HBCDs in indoor dust are different from those in waste water sludge and freshwater sediments in which Davis et al. [26] reported only one TBCD isomer as a degradation product of HBCDs (via LC–APPI/MS). They concluded that HBCDs degrade via a dibromoelimination reaction resulting in the loss of two bromines from vicinal carbons with the subsequent formation of a double bond between the adjacent carbon atoms. However, the detection and identification of the four PBCDs together with the two TBCDs in the present study rules out vicinal debromination as the main degradation pathway of HBCDs in indoor dust. The detection of degradation products, PBCDs and TBCDs, and their relationship to the parent HBCDs in the analyzed dust samples, favors the elimination of HBr. This results in the formation of a double bond (adjacent to the initial carbon atoms containing Br) and it is suggested to be a significant degradation pathway of HBCDs in indoor dust. This hypothesis is in agreement with the degradation pathway suggested by Hiebl and Vetter [25] for HBCDs in chicken egg and white fish. Accurate quantitation of the detected degradation products was not possible due to the lack of either native or labeled standards for these compounds which are (to the authors’ knowledge) reported here in indoor dust for the first time. Furthermore, similar to HBCDs, the LC–MS/MS method allows the detection of different PBCD and TBCD diastereomers. Although LC–MS/MS remains the technique of choice for the isomer-specific analysis of HBCDs, the present study indicates that GC–ECNI/MS can be used as a screening tool to estimate the total HBCD concentrations with acceptable accuracy. This is facilitated by the use of [13 C]HBCDs as internal standards and by deployment of an RF based on the average of those obtained for the ␣- and ␥-HBCD diastereomers. However, this enhancement in method accuracy and specificity using [13 C]HBCDs as internal standards (monitoring

M.A.-E. Abdallah et al. / J. Chromatogr. A 1190 (2008) 333–341

of ion m/z = 573) is accompanied by a decrease in the method sensitivity. As an alternative, BDE 128 (or another PBDE congener which is not present in the PBDE technical mixture) can be used as internal standard (monitoring of ion m/z = 79) with lower method detection limits, but less selectivity. Nevertheless, the need for monitoring of total HBCDs in food and feed has recently been underlined by the European Food Safety Agency, which has included HBCDs (as sum of isomers) on the list with priority BFRs [27]. PBCDs and TBCDs, identified in the present study as the degradation products of HBCDs, were observed in the GC–MS chromatograms of dust samples. Using LC–MS/MS, it could be demonstrated that they are not only the result of thermal degradation in the GC, but are originally present in the dust samples. Acknowledgements The authors gratefully acknowledge the provision of studentships from the Egyptian government and Egyptian Ministry of Higher Education (M.A.A.), the National Council of Science and Technology – Mexico (CONACYT) (C.I.), and of a postdoctoral fellowship from the Research Scientific Foundation of Flanders (FWO) (A.C.). The support of the UK POPs network training fund for the visit of M.A.A. and C.I. to the University of Antwerp is also gratefully acknowledged. References [1] Bromine Science and Environmental Forum, Fact Sheet on HBCD, Brussels, Belgium, 2003; www.bsef.com, accessed on 8 December 2007. [2] N.V. Heeb, W.B. Schweizer, M. Kohler, A.C. Gerecke, Chemosphere 61 (2005) 65. [3] R.J. Law, M. Kohler, N.V. Heeb, A.C. Gerecke, P. Schmid, S. Voorspoels, A. Covaci, G. Becker, K. Janak, C. Thomsen, Environ. Sci. Technol. 39 (2005) 281A. [4] S. Germer, A.H. Piersma, L. van der Ven, A. Kamyschnikow, Y. Fery, H.J. Schmitz, D. Schrenk, Toxicology 218 (2006) 229.

341

[5] E. Mariussen, F. Fonnum, Neurochem. Int. 43 (2003) 533. [6] T. Yamada-Okabe, H. Sakai, Y. Kashima, H. Yamada-Okabe, Toxicol. Lett. 155 (2005) 127. [7] T. Helleday, K.L. Tuominen, A. Bergman, D. Jenssen, Mutat. Res. 439 (1999) 137. [8] U. Sellstrom, A. Kierkegaard, C. de Wit, B. Jansson, Environ. Toxicol. Chem. 17 (1998) 1065. [9] A. Covaci, A.C. Gerecke, R.J. Law, S. Voorspoels, M. Kohler, N.V. Heeb, H. Leslie, C.R. Allchin, J. de Boer, Environ. Sci. Technol. 40 (2006) 3679. [10] A. Covaci, S. Voorspoels, L. Ramos, H. Neels, R. Blust, J. Chromatogr. A 1153 (2007) 145. [11] W. Budakowski, G. Tomy, Rapid Commun. Mass Spectrom. 17 (2003) 1399. [12] G.T. Tomy, T. Halldorson, R. Danell, K. Law, G. Arsenault, M. Alaee, G. MacInnis, C.H. Marvin, Rapid Commun. Mass Spectrom. 19 (2005) 2819. [13] N.G. Dodder, A.M. Peck, J.R. Kucklick, L.C. Sander, J. Chromatogr. A 1135 (2006) 36. [14] L.S. Haug, C. Thomsen, V.H. Liane, G. Becher, Chemosphere 71 (2008) 1087. [15] S. Harrad, C. Ibarra, M. Diamond, L. Melymuk, M. Robson, J. Douwes, L. Roosens, A.C. Dirtu, A. Covaci, Environ. Int. 34 (2008) 232. [16] M. Abdallah, S. Harrad, C. Ibarra, M. Diamond, L. Melymuk, M. Robson, A. Covaci, Environ. Sci. Technol. 42 (2008) 459. [17] A.C. Dirtu, R. Khaiwal, L. Roosens, R. Van Grieken, H. Neels, A. Covaci, J. Chromatogr. A 1186 (2008) 295. [18] M. Peled, R. Scharia, D. Sondack, in: J.R. Desmurs, B. Gerard, M.J. Goldstein (Eds.), Advances in Organobromine Chemistry II (Industrial Chemistry Library, vol. 7), Elsevier, New York, 1995, p. 92. ¨ [19] R. Koppen, R. Becker, C. Jung, I. Nehls, Chemosphere 71 (2008) 656. [20] J.M. Keller, H.M. Stapleton, R. Heltsley, A. Peck, J.R. Kucklick, S. Schantz, S.A. Wise, poster presented at the 4th International Workshop on Brominated Flame Retardants (BFR 2007), Amsterdam, 24–27 April 2007, poster P007. [21] S. van Leeuwen, J. de Boer, Mol. Nutrit. Food Res. 52 (2008) 194. [22] S. van Leeuwen, P. Leonards, S.B. Brandsma, J. de Boer, Poster presented at Recent Advances in Food Analysis Conference, Prague, 7–9 November 2007. [23] F. Barontini, V. Cozzani, A. Cuzzola, L. Petarca, Rapid Commun. Mass Spectrom. 15 (2001) 690. [24] F. Barontini, V. Cozzani, L. Petarca, Ind. Eng. Chem. Res. 40 (2001) 3270. [25] J. Hiebl, W. Vetter, J. Agric. Food Chem. 55 (2007) 3319. [26] J. Davis, S.J. Gonsior, D. Markham, U. Friederich, R. Hunziker, J. Ariano, Environ. Sci. Technol. 40 (2006) 5395. [27] EFSA, European Food Safety Agency, Question no. EFSA-Q-2005-244, EFSA J. 328 (2006) 1.