Determination of Cyclic Volatile Methylsiloxanes in ... - ACS Publications

Oct 18, 2010 - Department of Applied Environmental Science (ITM), Stockholm University, SE-106 91 Stockholm, ... with n-hexane and analyzed with GC/MS...
0 downloads 0 Views 600KB Size
Anal. Chem. 2010, 82, 9573–9578

Determination of Cyclic Volatile Methylsiloxanes in Biota with a Purge and Trap Method Amelie Kierkegaard,* Margaretha Adolfsson-Erici, and Michael S. McLachlan Department of Applied Environmental Science (ITM), Stockholm University, SE-106 91 Stockholm, Sweden The three cyclic volatile methylsiloxanes (cVMS), octamethylcyclotetrasiloxane (D4), decamethylcyclopentasiloxane (D5), and dodecamethylcyclohexasiloxane (D6), are recently identified environmental contaminants. Methods for the trace analysis of these chemicals in environmental matrices are required. A purge and trap method to prepare highly purified sample extracts with a low risk of sample contamination is presented. Without prior homogenization, the sample is heated in water, and the cVMS are purged from the slurry and trapped on an Isolute ENV+ cartridge. They are subsequently eluted with n-hexane and analyzed with GC/MS. The method was tested for eight different matrices including ragworms, muscle tissue from lean and lipid-rich fish, cod liver, and seal blubber. Analyte recoveries were consistent within and between matrices, averaging 79%, 68%, and 56% for D4, D5, and D6, respectively. Good control of blank levels resulted in limits of quantification of 1.5, 0.6, and 0.6 ng/g wet weight. The repeatability was 12% (D5) and 15% (D6) at concentrations 9 and 2 times above the LOQ. The method was applied to analyze cVMS in fish from Swedish lakes, demonstrating that contamination in fish as a result of long-range atmospheric transport is low as compared to contamination from local sources. Cyclic volatile methylsiloxanes, cVMS, are clear, low-viscosity fluids used as precursors in the production of silicone polymers and as solvents or fragrance carriers in personal care/household products and in cleaning agents.1-4 Three cVMS employed for these applications are octamethylcyclotetrasiloxane (D4), decamethylcyclopentasiloxane (D5), and dodecamethylcyclohexasiloxane (D6) (Figure S-1). In a recent screening assessment of * To whom correspondence should be addressed. E-mail: amelie. [email protected]. (1) Horii, Y.; Kannan, K. Arch. Environ. Contam. Toxicol. 2008, 55, 701–710. (2) Brooke, D. N.; Crookes, M. J.; Gray, D.; Robertson, S. Environmental Risk Assessment Report: Octamethylcyclotetrasiloxane; Environment Agency of England and Wales: Bristol, 2009; available at http://publications. environment-agency.gov.uk/pdf/SCHO0309BPQZ-e-e.pdf (accessed on 6 Sept., 2010). (3) Brooke, D. N.; Crookes, M. J.; Gray, D.; Robertson, S. Environmental Risk Assessment Report: Decamethylcyclopentasiloxane; Environment Agency of England and Wales: Bristol, 2009; available at http://publications. environment-agency.gov.uk/pdf/SCHO0309BPQX-e-e.pdf (accessed on 6 Sept., 2010). (4) Brooke, D. N.; Crookes, M. J.; Gray, D.; Robertson, S. Environmental Risk Assessment Report: Dodecamethylcyclohexasiloxane; Environment Agency of England and Wales: Bristol, 2009; available at http://publications. environment-agency.gov.uk/pdf/SCHO0309BPQY-e-e.pdf (accessed on 6 Sept., 2010). 10.1021/ac102406a  2010 American Chemical Society Published on Web 10/18/2010

industrial chemicals, these substances were identified as priority chemicals for environmental risk assessment due to their large volume of usage, their persistence in the environment, and their bioaccumulative properties.5 Risk assessments prepared by the United Kingdom indicate that, following their release to the environment in sewage treatment plant effluent, these chemicals will accumulate and persist in sediment, rendering fish particularly susceptible to exposure.2-4 D4, D5, and D6 possess a high octanol-water partition coefficient (log KOW of 6.49, 8.03, and 9.06, respectively2-4), which makes them prone to bioaccumulation in fish. Measuring bioaccumulation in aquatic organisms is thus one of the priorities in the risk assessment of these chemicals. D4, D5, and D6 have been identified in a variety of freshwater and marine fish and benthic organisms, marine mammals, and seabirds from northern Europe and the North Atlantic. Concentrations of the order of 100 ng/g (D4, D6) and 1000 ng/g (D5) wet weight (ww) were measured in cod liver from two contaminated areas (Oslo Fjord and the Rhine River at the German/Dutch border).2-4 More recently, these chemicals have been detected in cod from the waters around Spitsbergen at concentrations in the range of 2-20 ng/g ww.6 To date, there have been no methods published in the peerreviewed literature for the trace analysis of cVMS in biota. However, brief descriptions of methods are available in several of the laboratory reports on cVMS levels in biota. These methods employ homogenization of the sample in a nonpolar/semipolar solvent followed by direct injection on a GC and detection using high resolution6,7 or low resolution8 mass spectrometry. The robustness of these methods is limited by the large quantity of matrix components in the extracts, which leads to deterioration of GC/MS system performance. The sample preparation has been designed in this simple manner to minimize the risk of sample contamination. This is known to be a major obstacle in the trace analysis of this class of compounds. Because of to their volatility, their use in personal care products, and their presence in silicones (5) Howard, P. H.; Muir, D. C. G. Environ. Sci. Technol. 2010, 44, 2277–2285. (6) Evenset, A.; Leknes, H.; Christensen, G. N.; Warner, N.; Remberger, M.; Gabrielsen, G. W. Screening of new contaminants in samples from the Norwegian Arctic. Norwegian Pollution Control Authority, SPFO-report: 1049/2009, TA-2510/2009, ISBN 978-82-449-0065-2, 2009. (7) Kaj, L.; Schlabach, M.; Andersson, J.; Palm Cousins, A.; Schmidbauer, N.; Brorstro ¨m-Lunde´n, E. Siloxanes in the Nordic Environment; TemaNord 2005: 593; Nordic Council of Ministers: Copenhagen, 2005; available at http:// www.norden.org/sv/publikationer/publikationer/2005-593/at_download/ publicationfile (accessed on 14 Oct., 2009). (8) Boehmer, T.; Gerhards, R.; Koerner, M.; Unthan, R. Cyclic Volatile Methyl Siloxanes in Mussels - Screening of Mussels from some Intertidal Areas of the Southern North Sea; Centre Europe´en des Silicones: Brussels, 2007.

Analytical Chemistry, Vol. 82, No. 22, November 15, 2010

9573

used in building materials and laboratory equipment, cVMS are ubiquitous contaminants in the laboratory.9 In this work, we present a novel method for the analysis of D4, D5, and D6 in fish tissue. A closed system is employed for sample homogenization, extraction, cleanup, and concentration to minimize sample contamination with the analytes and thus achieve low limits of quantification. The method exploits the physical chemical properties of the chemicals; the comparatively low octanol/air partition coefficients (log KOA ) 4.34, 5.07, and 5.76 for D4, D5, and D6, respectively)2-4 and high air/water partition coefficients (log KAW ) 2.69, 3.13, and 3.30 for D4, D5, and D6, respectively)2-4 allow effective extraction via purging of the headspace of an aqueous slurry of the sample. Building on an existing method for the trace analysis of these substances in ambient air,10 Isolute ENV+ cartridges are used to capture the analytes from the nitrogen stream and concentrate them. Following evaluation of the method with a number of fish species, it was applied to assess the spatial distribution of D5 contamination in arctic char from Swedish lakes. EXPERIMENTAL SECTION Materials. Native octamethylcyclotetrasiloxane (D4, Fluka), decamethylcyclopentasiloxane (D5, Fluka), and dodecamethylcyclohexasiloxane (D6, Fluorochem) were purchased from SigmaAldrich Sweden AB and Fluorochem, UK, respectively. 13C-labeled D4, D5, and D6 were used as surrogate standards. 13C4-labeled D4 and 13C5-labeled D5 (purity >99%) were obtained from Moravek Biochemicals Inc. (Brea, CA), while 13C6-D6 was a gift from Dow Corning Corp., MI. The Isolute ENV+ sorbent (hydroxylated polystyrene-divinylbenzene copolymer) and 1 mL cartridges with 10 mg of ENV+ sorbent were obtained from Biotage AB (Uppsala, Sweden). Aldrin (Analytical Standards, Sweden) was used as a volumetric standard for recovery calculations. Dichloromethane (DCM) and n-hexane (both Lichrosolv quality) were purchased from Merck (Darmstadt, Germany). Ethyl acetate (pestiscan quality) was from Lab-scan Ltd. (Dublin, Ireland), and concentrated sulfuric acid (98%) was from BDH AnalaR (Poole, England). The water was filtered and deionized (Milli-Q, Millipore Corp., MA). Extraction Apparatus. A schematic drawing of the experimental setup is shown in Figure 1. It consisted of a 250 mL glass Erlenmeyer flask with an extra side port resting on a heated magnetic stirring plate (5 positions, IKAMAG, Germany). The side port was closed with a ground stopper, while the top port was closed with a modified gas-washing bottle stopper. Nitrogen was supplied to the inlet of the gas-washing bottle stopper via PTFE tubings, a 50 mg ENV+ cartridge (to preclean the inflowing gas), and a glass adapter with luer fitting. The outlet of the gas-washing bottle stopper was modified to a vertical position. A glass column (inner i.d. ) 10 mm, 30 cm long) was attached with a ball and socket joint. The column passed through the bottom and the lid of a Styrofoam box. A sample trap (initially a 1 mL cartridge prepacked with 10 mg of ENV+, later the barrel of a 1 mL glass syringe (Dosys, Socorex, Switzerland) hand-packed with 20 mg of ENV+, see below) was mounted on top of the column via a (9) Varaprath, S.; Stutts, D.; Kozerski, G. Silicon Chem. 2006, 3, 79–102. (10) Kierkegaard, A.; McLachlan, M. S. J. Chromatogr., A 2010, 1217, 3557– 3560.

9574

Analytical Chemistry, Vol. 82, No. 22, November 15, 2010

Figure 1. Schematic setup for the extraction procedure.

glass adaptor with a luer fitting. A stopcock to control the flow was connected via a PTFE tube to a plug on the top of the ENV+ cartridge. The outlet of the stopcock was directed via a cannula into a 1.5 mL Eppendorf test tube filled with water to serve as a flow indicator. Procedure. Surrogate standards dosing capsules were prepared by adding 50 µL of a solution of 13C-labeled D4, D5, and D6 in ethyl acetate (1.6, 5.0, and 3.5 ng/µL, respectively) to 300 µL glass vial inserts. A drop of preboiled Milli-Q water was added to the top of the insert to serve as a plug, leaving an air space between the water and the standard solution. The insert was subsequently sealed and frozen (-17 °C) overnight in a preparative glass tube filled with 6 mL of milli-Q water. The sample traps (both prepacked and hand-packed) were rinsed with 3 mL of DCM and 6 mL of n-hexane, dried with ENVfiltered nitrogen (99.996%, AGA), and sealed at both ends until use (within 2 h). Hand-packed sample traps were prepared by packing the barrels of a 1 mL glass syringes with 20 mg of ENV+, sealed by plugs of fine glass wool. The extraction glassware was assembled and, together with the stir bar, rinsed with acetone before the addition of 190 mL of preboiled Milli-Q water to the Erlenmeyer flask. While stirring, the system was purged with precleaned nitrogen for at least 15 min. Ice packs were packed into the styrofoam box around the glass columns. The sample trap was mounted on the top of the glass column. After the introduction of the frozen sample (1-12 g ww) and the frozen surrogate standard dosing capsule through the side port, it was immediately closed. The water/sample/ standard mixture was heated to on average 73 °C (coefficient of variation (CV) ) 4%, n ) 60: 6 extraction rounds, each with 10 extractions) and stirred vigorously with a PTFE coated magnetic stir bar, while precleaned nitrogen was purged through the system at 75-200 mL min-1. The gas flowed up the glass column where

the steam condensed and recirculated into the flask. The stopcock served to regulate the gas flow, and the cannula/ bubble tube served to monitor it. Extraction was allowed to continue for 24-72 h while maintaining the temperature and nitrogen gas flow. It was terminated by removing the sample trap, which was immediately eluted with 0.5 mL (prepacked 10 mg ENV+ cartridges) or 0.8 mL (hand-packed 20 mg syringe barrels) of n-hexane (by gravity flow) directly into a GC vial for subsequent GC/MS analysis. Instrumental Method. Quantification was performed on a Trace GC Ultra (Thermo Electron Corp.) coupled to a MD800 MS detector (Fisons Instruments SpA) using electron ionization (EI). The GC was equipped with a large volume splitless injector (Thermo Electron Corp.) with a Merlin microseal septum. Five microliters of the extract was injected at an injector temperature of 220 °C. A 5 m retention gap of deactivated fused silica (0.32 mm i.d., Agilent Technologies) was connected to the analytical column, a 30 m DB5-MS (0.25 mm i.d., 0.25 µm film thickness, J&W Scientific). The carrier gas was helium (99.996%). The oven temperature program was 60 °C; hold 1 min; split valve closed for 1 min; 10 °C/min to 150 °C; 30 °C/min to 300 °C, hold 2 min. The transfer line was kept at 250 °C, and the ion source was kept at 200 °C. The ions monitored and the time windows applied are listed in Table S-1. The dwell time was 0.04 s. Details on the measures taken to optimize the instrumental analysis, including the minimization of instrumental blanks, can be found in the Supporting Information of the previously reported air method.10 QA/QC. To reduce contamination, the preparation of surrogate standard dosing capsules and the sample traps as well as the extractions were performed in a clean air cabinet under a laminar flow of charcoal-filtered air (Gigapleat, Camfil International AB) and particle-filtered air (HEPA H14, Trox Technik GmbH). The n-hexane used in the preparation and extraction of the cartridges was treated with concentrated sulfuric acid. Pasteur glass pipettes were heated to 450 °C overnight. Calibration standard mixtures, kept sealed in 1 mL glass ampules, were newly prepared at least every month. The calibration curve comprised 10 standard solutions with concentrations from 0.4-400 pg/µL. Test tubes and vials were sealed with aluminum foil under the PTFE lined caps. With every set of eight samples, a procedural blank and a control sample were included. METHOD EVALUATION AND APPLICATION Blanks. Because the LOD of the method was determined by the cVMS levels in the blanks, various tests to identify the sources of the cVMS contamination were conducted. This included instrument blanks, solvent blanks, material blanks, and blanks of different portions of the procedure. Extraction Efficiency and Recovery. The extraction efficiency was examined by doing sequential extractions for 24, 48, and, for a few samples, 72 h. The species tested were perch (Perca fluviatilis), European flounder (Platichthys flesus), ragworm (Hediste diversicolor), herring (Clupea harengus) muscle and muscle homogenate, arctic char (Salvelinus alpinus (L) sp. Complex), cod (Gadus morhua), and gray seal (Halichoerus grypus); their lipid and cVMS contents are summarized in Table S-2. The recovery of the 13C-labeled surrogate standards relative to the volumetric standard (aldrin) was calculated. Breakthrough of analyte

through the extraction apparatus was tested by mounting a second ENV+ syringe in series with the original ENV+ trap and analyzing it separately. Repeatability. The repeatability was determined by repeated extractions of control samples over a period of 2 years. During the first year, the analyses were performed using prepacked 1 mL ENV+ cartridges containing 10 mg of sorbent. During the second year, manually packed glass syringes were applied. The control samples were prepared from a herring homogenate. Two herring homogenates were prepared: one for the repeatability tests and one for the extraction efficiency and the method comparison studies. Herring fillets were ground repeatedly in a meat mill and mixed thoroughly. Aliquots were weighed, sealed, and frozen at -17 °C. Triplicate muscle samples from three individual herring were analyzed on separate occasions to determine the repeatability when using intact fish muscle tissue. LOD/LOQ. The method limit of detection (LOD) and limit of quantification (LOQ) were determined from the cVMS content in 34 procedural blanks. The LOD and LOQ were defined as the mean content in the blanks plus 3 or 10 times its standard deviation, respectively. Comparison with Cold Solvent Extraction. Samples of arctic char muscle, cod liver, and gray seal blubber as well as the herring control sample were also analyzed with a traditional cold solvent extraction method. About 1 g of tissue was homogenized using Ultra turrax in a mixture of 4 mL of acetone and 2 mL of n-hexane containing the labeled surrogate standards. The solvents were transferred to a 50 mL centrifuge tube, and the extraction was repeated with fresh solvents. The sample and solvent were then combined with the first extract. After the addition of 20 mL of Milli-Q water and 0.1-0.3 g of sodium chloride, the tube was gently shaken and centrifuged at 2000 rpm for 10 min. An aliquot of the hexane phase representing up to 0.06% of the sample was injected into the GC/MS for analysis. Method Application. Arctic char from Lake Va¨ttern, Abiskojaure, Tjultra¨sk, and Stor-Bjo¨rsjo¨n (see Figure S-2) were supplied by the Swedish Museum of Natural History. The characteristics of the samples are presented in Table S-3. Skin-free dorsal muscle tissues (5-10 g) were prepared in a clean air sample preparation room with a counter-flow of particle filtered air. The samples were immediately wrapped in aluminum foil, vacuum-sealed in polyethylene bags, and frozen (-17 °C). Fat content was determined on subsamples by the Oekometric GmbH using a gravimetric method based on ASE extraction with n-hexane/dichloromethane (1:1 v:v). RESULTS AND DISCUSSION Blanks. D5 was the focus of the blank evaluation work, as this is the cVMS with the largest use in personal care products. The D5 signal resulting from the injection of the surrogate and volumetric standards (n ) 40 over a 5 month period) was 0.2 pg. This corresponded to ∼40 pg per sample, and it served as the baseline for the blank assessment. The overall method blank obtained when the procedure was conducted with addition of the surrogate standard dosing capsule but without the addition of sample amounted initially to 1.3-3.2 ng per sample (n ) 7, mean 2.5 ng), ∼50 times more than the instrument blank. Method blanks for the cartridge elution step were similar to the injection blanks. This suggested that the primary D5 source in the overall Analytical Chemistry, Vol. 82, No. 22, November 15, 2010

9575

Figure 2. Extraction efficiency of D4, D5, and D6 for a range of different biota tissues. The quantity of analyte in sequential 24 h extracts of the same sample is shown. The quantity in the second and third extracts is expressed as a percentage of the quantity in the first extract. The mean and 95% confidence interval (where n g 3) of this percentage are plotted. The number of samples for which sequential extracts were conducted is noted above each bar. The upper number refers to the number of samples for which there were three sequential extracts. For D6 in perch and for D4 and D6 72 h in the herring homogenate, an insufficient number of data were above the LOQ.

method blanks was the extraction apparatus, the Milli-Q water, the internal standard addition, or the gas supply. However, selective elimination of these factors had no marked influence on the blank levels. Eventually the ENV+ sorbent was identified as the primary source of contamination. The contamination was not seen in the blanks for the cartridge elution step because the procedure included prewashing of the cartridge before use. Immediately eluting the cartridge again yielded low blanks, but if the cartridge was allowed to stand for 24 h (the length of the extraction procedure), the blank levels following elution were in the same range as the overall method blank. Repeating the procedure after another 24 h also gave high blanks, but somewhat lower than after the first 24 h. It was not possible to completely eliminate this blank by a more extensive cleaning program. Because the variation in the level of D5 contamination between different batches of ENV+ cartridges was considerable, it was decided to hand pack the cartridges using a single source of ENV+ with a comparatively low level of contamination. The mean content (n ) 34) of D5 was lowered by a factor of 4 to 0.7 ng per sample. Extraction Efficiency. The second 24 h extract of the biota samples contained between 0% and 77% of the cVMS present in the first 24 h extract (Figure 2). The lowest values (