On-site polymer-coated hollow fiber membrane microextraction and gas chromatography–mass spectrometry of polychlorinated biphenyls and polybrominated diphenyl ethers

Journal of Chromatography A, 1139 (2007) 157–164

On-site polymer-coated hollow fiber membrane microextraction and gas chromatography–mass spectrometry of polychlorinated biphenyls and polybrominated diphenyl ethers Chanbasha Basheer a , Muthalagu Vetrichelvan b , Suresh Valiyaveettil a,b,∗ , Hian Kee Lee a,∗∗ b

a Department of Chemistry, National University of Singapore, Singapore 117543, Singapore Singapore-MIT Alliance, National University of Singapore, 3 Science Drive 3, Singapore 117543, Singapore

Received 11 April 2006; received in revised form 2 November 2006; accepted 3 November 2006 Available online 21 November 2006

Abstract Porous polypropylene hollow fiber membrane coated with a conjugated polymer was used as an on-site sampling device for the extraction of polychlorinated biphenyls and polybrominated biphenyl ethers from coastal sea water samples. The coated hollow fiber membrane was placed in a vial containing the sample, and the target compounds extracted via manual shaking of the vials at the site of sample collection. For each extraction, two fibers were used. After extraction, the fibers with the adsorbed analytes were brought back to the laboratory for further processing. Care was taken to preserve the integrity of the analytes and to avoid contamination during transport; after extraction, the fibers were carefully removed and placed in air-tight crimper vials which were stored in an ice-box. The analytes were desorbed by solvent in the laboratory and analyses were carried out using gas chromatography/mass spectrometry. This method was highly reproducible with relative standard deviations in the range of 1–9%. Recoveries from spiked water samples ranged from 83% to 98%. Low limits of detections between 0.04 and 0.21 ng l−1 were achieved. The extraction efficiency was compared with solid-phase microextraction. © 2006 Elsevier B.V. All rights reserved. Keywords: On-site sampling; Trace level analysis; Persisting organic pollutants and microextraction

1. Introduction Persistent organic pollutants (POPs) including polychlorinated biphenyls (PCBs) and polybrominated diphenyl ethers (PBDEs) are threats to the environment due to their high resistance to physical, chemical and biological degradation [1]. PCBs and PBDEs are readily accumulated in the food chain, especially in meat, fish, and dairy products due to their physical properties such as low water solubility and high affinity to lipids [2,3] Trace amounts of PCB congeners have also been associated with endocrine disruption and incidence of fetal miscarriage [4,5]. ∗ Corresponding author at: Department of Chemistry, National University of Singapore, Singapore 117543, Singapore. Tel.: +65 6516 2995; fax: +65 6779 1691. ∗∗ Corresponding author. Tel.: +65 6516 2995; fax: +65 6779 1691. E-mail addresses: [email protected] (S. Valiyaveettil), [email protected] (H.K. Lee).

0021-9673/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2006.11.008

Analytical methodologies for quantitative analysis of trace level concentrations of PCB and PBDE from environmental samples are especially difficult due to the complexity of the mixtures of congeners, matrix interferences and to the low detection limits required, usually sub-parts per billions (ppb) to parts per trillion (ppt) [6]. Extraction of PCBs and PBDEs in environmental samples including seawater is a significant challenge demanding an effective sample preparation procedure prior to analysis. Usually, analyses of organic pollutants from seawater require that the samples be collected, transported back to the laboratory, and preserved before laboratory analysis [7]. If there is a need to monitor environmental pollutants away from the laboratory, then simplified but yet rigorous on-site procedures are advantageous. Current on-site or field-sampling techniques are usually multi-step procedures, require large volumes of sample size and additionally specialized equipment is needed for analysis. Several large-volume sampling strategies, such as on-board

158

C. Basheer et al. / J. Chromatogr. A 1139 (2007) 157–164

filtration/resin extraction [8,9], on-board filtration/liquid–liquid extraction [10,11], on-board centrifuging [12] and in situ water pumping [13–15], have been employed in field studies. These sampling methods, however, are either labor-intensive or costly to implement and are not feasible as a simultaneous sampling technique for most large-scale sampling programs. Recently, integrative semi-permeable membrane devices (SPMD) and solid-phase microextraction (SPME) procedures have been used for field-sampling and have overcome many of the drawbacks of the large-volume sampling approaches [16–19]. G´orecki and Pawliszyn [20] have evaluated the feasibility of on-site SPME (coupled to a portable GC), and Smith et al. [21] have used a GC–MS system for trace level analysis in the field. Koziel et al. [22] and Jia et al. [23] have reported the application of organic analytes in the field. In SPMDs, calibration, sample processing, and data interpretation are still involved [24,25]. SPME has been used for extracting PCBs directly from seawater-here; the fibers were immersed for several days in the sample [26]. Although convenient to use, SPME fibers are, however, still relatively expensive. In this work, for the first time polymer-coated hollow fiber microextraction (PC-HFME) [27,28] was evaluated as an on-site sample preparation approach for seawater samples. The on-site PC-HFME sampling procedure described here is a cost effective and convenient sampling technique and perhaps may be considered as a complementary technique to field SPME described previously [26]. The manual sample preparation and enrichment by PC-HFME allows a unique strategy of performing the sample preparation where the sample is taken in the remote environment and partially processed. It does not have to be brought back to the laboratory. Only the fiber and the extracted analytes are transported back to the laboratory for final processing before analysis.

give a salinity of 33 parts per thousand (‰), conductivity 48.8 mS and pH 8.1, was prepared for method optimization experiments. The SPME fiber holder and fibers for manual sampling were purchased from Supelco and used without modification. Poly(dimethylsiloxane) (PDMS, 100 ␮m) and poly(dimethylsiloxane)-divinylbenzene (PDMS-DVB, 65 ␮m) and polyacrylate (PA, 85-␮m) coated fibers were used for PCB and PBDE extraction in the comparative studies. Solvent desorption of analytes was carried out in Chrompack (Palo Alto, CA, USA) crimper vials.

2. Experimental

Three different polymers, i.e. P1–P3 (Fig. 1a) were synthesized, and evaluated as adsorbents. The general synthetic route toward the monomers and polymer P1 are outlined in Fig. 1b. Polymers P2 and P3 were also prepared using a similar scheme by using the 2,6-and 3,5-dibromopyridine, respectively. All polymerizations were done using Suzuki polycondensation reactions [29]. Polymerizations were carried out in a mixture (3:2, v/v) of toluene and aqueous potassium carbonate solution (2 M) containing 3.0 mol% tetrakis triphenylphosphine palladium(0) under vigorous stirring at 85–90 ◦ C for 72 h under a nitrogen atmosphere. After completion of the reactions, the polymers were precipitated from methanol. The derived polymers P1–P3 were purified by the reprecipitation of the polymers by dissolving in chloroform followed by adding methanol. The obtained polymers P1–P3 were soluble in common organic solvents, such as tetrahydrofuran, chloroform, toluene and trifluroacetic acid and insoluble in acetone, methanol, isooctane, nonane and hexane.

2.1. Reagents and materials A PCB mixture of six congeners (i.e. 2,6-dichlorobiphenyl (PCB-10), 2,4,4 -trichlorobiphenyl (PCB-28), 2,2 ,5,5 tetrachlorobiphenyl (PCB-52), 2,2 ,4,4 ,5,5 -hexachlorobiphenyl (PCB-138), 2,2 ,3,4,4 ,5 -hexachlorobiphenyl (PCB-153), 2,2 ,3,4,4 ,5,5 -heptachlorobiphenyl (PCB-180) (500 ␮g ml−1 congener)) were purchased from Supelco (Bellefonte, PA, USA) and individual PBDE congeners (i.e. 2,2 4,4 tetrabromodiphenyl ether (PBDE-47) and 2,2 4,4 ,5-pentabromodiphenyl ether (PBDE-99)) were purchased from Accu Standard (New Haven, CT, USA), and used as supplied. HPLC-grade solvents were purchased from Merck (Darmstadt, Germany), and ultrapure water was obtained from a Milli-Q system (Millipore, Milford, MA, USA). The mixture of PCB and PBDE was prepared and diluted in acetone. Accurel Q3/2 polypropylene hollow fiber membrane (Membrana, Wuppertal, Germany) with an inner diameter of 600 ␮m, wall thickness of 200 ␮m and wall pore size of 0.2 ␮m was used in the experiments. An artificial seawater sample using natural sea salt (Coral Reef Red Sea salt, obtained from Red Sea Fish Pharm (P), Eilat, Israel) dissolved in ultrapure water to

2.2. Instrumental parameters Sample analyses were carried out using a Shimadzu (Tokyo, Japan) QP2010 GC–MS system equipped with a Shimadzu AOC-20i auto sampler and a DB-5 fused silica capillary column 30 m × 0.32 mm i.d., film thickness 0.25 ␮m (J&W Scientific, Folsom, CA, USA). Helium (purity 99.9999%) was used as the carrier gas at a flow rate of 1.5 ml min−1 . Samples (5 ␮l) were injected in splitless mode. The injection temperature was set at 250 ◦ C, and the interface temperature at 280 ◦ C. The GC–MS temperature program used was as follows: initial temperature 60 ◦ C, held for 2 min, then increased by 30 ◦ C/min to 280 ◦ C and held for 2 min. PCB and PBDE standards and samples were analysed in selective ion monitoring (SIM) mode with a detector voltage of 1.5 kV. A specific ion was selected for each PCB congener and the most abundant ion was selected as the quantitative ion, while two other ions were used for confirmation of individual PCB and PBDE congeners. Both quantitative and confirmation ions with relative intensities are shown in Table 1. 2.3. Synthesis of the conjugated polymers

2.4. Preparation and characterization of the polymer-coated HFM The derived polymers are the copolymers consisting of phenylene with the long alkyl substitution (R = C12 H25 ) and

C. Basheer et al. / J. Chromatogr. A 1139 (2007) 157–164

159

Table 1 GC–MS–SIM conditions for PCBs and PBDEs, aqueous solubility, and octanol–water partition coefficients (log Kow ) of the analytes Congeners

Retention time

Primary ion (m/z)

Confirmation ion (m/z) (relative intensity)

2,6-Dichlorobiphenyl (PCB-10) 2,4,4 -Trichlorobiphenyl (PCB-28) 2,2 ,5,5 -Tetrachlorobiphenyl (PCB-52) 2,2 ,4,4 ,5,5 -Hexachlorobiphenyl (PCB-138) 2,2 ,3,4,4 ,5 -Hexachlorobiphenyl (PCB-153) 2,2 ,3,4,4 ,5,5 -Heptachlorobiphenyl (PCB-180) 2,2 4,4 -Tetrabromodiphenyl ether (PBDE-47) 2,2 4,4 ,5-Pentabromodiphenyl ether (PBDE-99)

9.5 12.5 13.2 20.4 22.3 28.5 31.7 33.3

222 (100) 256 (100) 292 (100) 360 (100) 360 (100) 394 (100) 406 (100) 484 (100)

224, 152 (64, 69) 258, 186 (98, 57) 290, 220 (99, 90) 362, 290 (80, 60) 362, 290 (79, 56) 396, 326 (83, 89) 404, 74 (98, 63) 482, 74 (85, 30)

WL a (mol m3 )

log Kow a

4.5 × 10−3 5.8 × 10−4 6.6 × 10−6 8.2 × 10−6 7.9 × 10−7 2.0 × 10−4 6.9 × 10−5

5.6 6.1 6.7 6.9 7.3 6.4 6.8

WL : aqueous solubility; Kow : octanol–water partition coefficient. a Reference [28,30].

OCH2 C6 H5 groups on opposite sides (para) of the benzene ring (see Fig. 1). The 2,5-(P1) 2,6-(P2), 3,5-substituted (P3) pyridine attached to the HFM, contains a nitrogen atom that can be expected to have stronger electrostatic interactions with chlorinated and brominated compounds such as PCBs and PBDEs. The hollow fiber membrane (HFM) was cut into ∼1.2-cm lengths and immersed in the respective polymer solutions (dissolved in tetrahydrofuran, concentration of ca. 1 g l−1 ) and kept at room temperature (25 ◦ C) for 1 day. The fibers were then removed and dried in an oven at 50 ◦ C for 30 min to evaporate the solvent completely. A micrograph of polymer-coated HFM shows a uniformly thin layer of polymer coating on the inner and outer surface of the fiber. The thickness of the coating on

the HFM was measured using atomic force microscopy and it varied between 1.2 and 1.3 ␮m. Compared with conventional SPME fibers, the porosity of the coated HFM affords a large surface area. In the Attenuated Total Reflection Fourier Transform Infrared spectrum of P1 on the fiber, appearance of a peak at 2923 cm−1 corresponds to the pyridine group. The P2 and P3 polymers show similar spectra although their colors vary (P1, burgundy; P2, grey and P3, dark grey). By varying the polymer concentration in the coating solution, the thickness of the polymeric film on HFM could be controlled. Although not studied in the present work, it is conceivable to tune the selectivity of the active extraction medium by changing the functional groups on the polymer.

Fig. 1. (a) Structures of the derived polymers and (b) polymer synthetic scheme (P1): (i) Br2 /AcOH, 80%; (ii) NaOH in abs. EtOH, CH3 (CH2 )11 Br, 45–50 ◦ C, 10 h, 65%; (iii) K2 CO3 in abs. EtOH, C6 H5 CH2 Br, 50 ◦ C, 10 h, 90%; (iv) 1.6 M solution of n-BuLi in hexane, THF at −78 ◦ C, tri-isopropylborate, RT (28 ◦ C) for 10 h, 60%; (v) 2 M K2 CO3 solution, toluene, 3.0 mol% Pd(PPh3 )4 , refluxed for 3 days.

160

C. Basheer et al. / J. Chromatogr. A 1139 (2007) 157–164

2.5. PDMS-coated hollow fibers PDMS polymer was prepared by using a mixture (30:1) of Sylgard 184 and curing agent (Dow Corning, Midland, MI, USA). A short length (1.2 cm) of HFM was cut and dipped into the viscous mixture for 5 min. The fibers were removed and dried in an oven at 60 ◦ C for 16 h. The PDMS coating thickness varies between 1 and 1.1 mm. 2.6. On-Site PC-HFME Two P1-coated HFMs were placed in a 10-ml sample vial containing PCBs and PBDEs-spiked seawater (sample pH and salt (seawater salinity, 3%, w/v) were not adjusted). The sample solution was then manually shaken for 10 min in the on-site. The coated-fibers tumbled freely and vigorously during manual shaking. The fibers were then removed with a pair of tweezers and placed in a crimper vial (100 ␮l capacity). The crimper vials which were stored in an ice-box were brought back to the laboratory. The extracted analytes on the fibers were desorbed in 100 ␮l of hexane (the coated polymer is insoluble in hexane) in a crimper vial via ultrasonication for 20 min. To increase the sensitivity of the quantitative analysis, the extract volume was reduced to dryness using nitrogen gas and made up to 20 ␮l with hexane. Finally, 5 ␮l was injected into the GC–MS. This accounts for the high sensitivity of the method. The analyte retention time in the GC injector port (sampling time) was maintained for 2.5 min. The extracted fibers were tested for carryover. Based on the negative results, the fibers were reused for further extractions. 2.7. Laboratory-based PC-HFME This extraction was carried out in its entirety in the laboratory. A 10-ml artificial seawater spiked with 1 ␮g l−1 of each PCB and PBDE sample was used. Two P1-coated fibers were placed in the sample and agitated at 105 rad s−1 (1000 rpm; 1 rpm = 0.1047 rad s−1 ) for 10 min on a Vibramax 100 (Heidolph, Kelheim, Germany) magnetic stirrer. After extraction, the fibers were subjected to solvent desorption and processed as mentioned above. Five microliters of the extract was injected into the GC–MS.

Blanks were run periodically to confirm the absence of contaminants. 3. Results and discussion 3.1. Selection of polymer coating The three different conjugated polymer coatings, P1–P3 (Fig. 1a) and commercial PDMS were initially evaluated. Fig. 2 shows the comparison of the performance of these fibers under identical conditions. P1- and P3-coated fibers showed the highest extraction efficiencies (based on peak area) for all compounds, compared to the PDMS coating. PDMS is generally not suitable for extraction that involves solvent desorption due to its solubility in various organic solvents including hexane, isooctane and toluene. It is not soluble in methanol and ethanol. Thus, methanol was selected for solvent desorption The P3coated fiber exhibited slightly better extraction efficiencies than its P2 counterpart for all analytes (Fig. 2), but P1 showed the best extraction efficiency. This may be due to the size of the binding site formed by the pyridinyl nitrogen with the adjacent OCH2 C6 H5 ring that is a better fit for the halogenated congeners. In addition, the stability of the bonds thus formed may also be higher than the other substituted polymers. No further study was attempted to determine the basis of this speculation, since this was outside the scope of the present work. Based on the extraction data, the P1-coated fiber was used for subsequent extractions. Optimization experiments were carried out with artificial seawater after it had been ascertained from blank runs that this sample did not contain PCBs and PBDEs. Initially, conventional (i.e. laboratory-based) PC-HFME conditions were optimized with P1-coated fibers. The procedure included stirring during extraction, a step that was excluded in on-site PC-HFME. All other conditions used in conventional PC-HFME were applicable to on-site PC-HFME. In another preliminary study, use of multiple (two to four) PCHFME fibers (P1-coated) placed in individual sample solutions to compare extraction efficiency was performed. A single fiber used for extraction of 2,6-dichlorobiphenyl gave a peak area

2.8. Conventional SPME Initially, PDMS, PDMS-DVB and PA SPME fibers were tested under optimized conditions, in comparative studies (results not shown), and PDMS gave the best results. Thus, it was selected for the evaluation of this microextraction procedure. A 10-ml sample of artificial seawater spiked with PCBs and PBDEs (at 1 ␮g l−1 of each analyte) (pH and salt concentration were not adjusted) was extracted by direct immersion of the SPME fiber with stirring (at 105 rad s−1 ). Equilibrium was established after 30 min. After extraction, the fiber was desorbed in the injection-port of the GC for 3 min at 250 ◦ C. Possible carryover was minimized by keeping the fiber in the injector for an additional 3 min before it was used for the next extraction.

Fig. 2. PC-HFME efficiency of different coatings. Extraction time 10 min at stirring speed of 105 rad s−1 ; sample pH and ionic strength were not adjusted. The extracted fibers were desorbed (ultrasonicated) for 20 min in hexane. GC–MS conditions are as given in the text.

C. Basheer et al. / J. Chromatogr. A 1139 (2007) 157–164

161

SPME efficiency from 5 to 50 min was also investigated. The observation that maximum extraction was achieved only after 30 min indicated that slow partitioning with respect to SPMEPDMS was observed compared to PC-HFME. This could be due to the non-porous nature of PDMS fibers and higher coating thickness (100 ␮m, compared to polymer-coated HFMs, ∼1.2 ␮m). Additionally, the coated functional polymer has a higher number of active extraction sites, which leads to shorter (10 min) extraction time. 3.3. Static versus stirring on extraction Fig. 3. Extraction time profile of PCBs and PBDEs during PC-HFME of artificial seawater samples spiked at 1 ␮g l−1 levels using P1-coated HFM.

The effect of sample agitation on extraction was evaluated (with artificial seawater sample spiked at 1 ␮g l−1 of each analyte). In the absence of stirring (static mode) P1-coated fibers were held in the sample solution for various durations, ranging from 10 to 120 min. Table 2 shows extraction data on various static extraction times compared with when the sample was stirred. Clearly, static PC-HFME gave poorer analyte extractability, as expected. To improve static (non-magnetically stirred) extraction, samples were shaken manually for different times (i.e. 5, 10 and 15 min) to simulate on-site sample processing. This would have an impact on the on-site extraction since a magnetic stirrer would not be available. After extraction, fibers were desorbed by a 20-min of ultrasonication, the results are shown in Table 2. They show that 10-min manual shaking gave comparable results with magnetically stirred samples (at 105 rad s−1 ). This supports the idea that on-site extraction was a viable approach.

of 1,001,264 arbitrary units; with two fibers, the peak area was 1,532,906; with four fibers, the peak area was 1,519,490. The last two values are similar. Therefore, two fibers were selected for method optimization. However, when more than two fibers were used, no significant additional analyte enrichment was obtained. Other factors affecting the PC-HFME procedure including effect of stirring on extraction time, sample pH, salt content of the sample, type of desorption solvent and desorption time were evaluated and optimized. 3.2. Effect of extraction time To investigate the sorption of PCB and PBDEs by the polymer-coated fiber, different extraction times, i.e. 5, 10, 15, 20, 25, and 30 min were applied to artificial seawater samples spiked at 1 ␮g l−1 of each PCB and PBDE congener. Samples were stirred at 105 rad s−1 , the maximum speed at which the stirring bar did not begin to tumble irregularly. The PCB and PBDE congeners analysed are hydrophobic compounds with log Kow values ranging from 4.3 to 8.2 and water solubilities ranging from 4.5 × 10−3 to 8.2 × 10−6 mol m−3 (see Table 1) [30,31]. Therefore, a rapid partition between the hydrophobic functional groups on the polymer-coated fiber and PCB and PBDE congeners was expected. Fig. 3 shows greater peak areas for both PCB and PBDE congeners over an extraction time of 10 min. As such, 10 min was selected as the optimum extraction time as equilibrium was obtained for all analytes within this period.

3.4. pH and salt The influence of the sample pH on extraction efficiency was investigated. An acidic range of pH was achieved by adding dilute 6 M hydrochloric acid, and a basic pH range by adding 2 M NaOH solution to the samples. Lower amount of the analytes were extracted at pH 2 and 12. Most of the PCBs and PBDEs exhibited better adsorption at pH 8 (real seawater pH). Experiments were conducted to determine the effect of ionic strength of the sample solution in the recoveries of the PCBs and PBDEs. Addition of salt did not have any influence in the extraction efficiencies of PCBs and PBDEs (results not shown).

Table 2 Relative response (with respect to response obtained with magnetic stirring) of PCBs and PBDEs congeners extracted under static mode and manual shaking by conditions on-site PC-HFME device Static PC-HFME (n = 2)

PCB-10 PCB-28 PCB-52 PCB-138 PCB-153 PCB-180 PBDE-47 PBDE-99

On-site PC-HFME by manual shaking

10 min

30 min

60 min

120 min

5 min

10 min

15 min

0.50 0.25 0.30 0.11 0.13 0.13 0.26 0.25

0.51 0.45 0.44 0.37 0.36 0.37 0.48 0.66

0.51 0.49 0.47 0.46 0.42 0.43 0.51 0.77

0.54 0.51 0.52 0.51 0.47 0.52 0.56 0.86

0.87 0.85 0.78 0.87 0.80 0.81 0.71 0.72

0.97 0.91 0.94 0.86 0.84 0.83 0.98 0.84

1.02 0.90 0.95 0.90 0.97 0.85 1.01 0.92

162

C. Basheer et al. / J. Chromatogr. A 1139 (2007) 157–164

Fig. 4. Desorption profile of PCBs and PBDEs with different solvents. Extraction time 10 min, P1-coated HFM and sample pH and salt were not adjusted.

As such, no adjustment was made to seawater samples in this respect (note: seawater contains 3%, w/v salt concentration).

Fig. 5. Desorption profile of PCBs and PBDEs at different desorption times. Extraction time 10 min, P1-coated HFM and sample pH and salt were not adjusted.

3.6. Performance evaluation 3.5. Selection of desorption solvent and desorption time The selection of the most appropriate solvent (related to analyte solubility and polymer insolubility) is a prerequisite in order to achieve the highest analyte desorption factor. The P1–P3 coatings used in this work were earlier determined to be unaffected by organic solvents such as acetone, methanol, isooctane, nonane and hexane. Fig. 4 shows that the desorption efficiency of various organic solvents for the PCBs and PBDEs. Hexane gave higher responses (i.e. peak areas) for all congeners. In contrast, the more polar solvents, i.e. methanol and acetone resulted in relatively lower peak area responses. Thus, hexane was selected as the most suitable solvent for analyte desorption. After extraction, the PCB and PBDE desorption efficiency was investigated by evaluating different desorption times (2, 5, 10, 15, 20 and 25 min) based on ultrasonication with 100 ␮l of hexane. As can be seen in Fig. 5, 20-min ultrasonication was enough to effect total desorption. The coated-HFM was immediately desorbed again; and the extract analysed, under the same conditions to determine carry-over; no peaks appeared in the resulting GC–MS chromatogram confirming that the PCBs and PBDEs were completely removed. This also confirmed that hexane was a suitable desorption solvent.

After evaluating the experimental results, the following conditions were selected to assess the extraction performance of the present method: HFM coated with P1; 10-ml sample volume; sample stirring at 105 rad s−1 ; 10 min extraction time, and desorption with 100 ␮l of hexane by 20-min ultrasonication. Parallel PDMS-coated SPME was carried out to determine its optimum conditions. These were as follows: PDMS fiber (100 ␮m) was selected for extraction for 30 min at sample pH 8 (seawater pH) and desorption temperature 250 ◦ C for 5 min. Linearity, repeatability, congener extraction precision, and limits of detection (LODs) were determined for both procedures. The calibration study was performed using spiked artificial seawater samples. The concentration range is shown in Table 3. Each level of concentration was analyzed in duplicate or triplicate. Calibration curves were linear for both PC-HFME and SPME in the range studied for each compound. As can be seen in Table 3, the correlation coefficient (r) values were ≥0.996 for all the compounds using magnetically stirred-PC-HFME, ≥0.991 for on-site PC-HFME and ≥0.984 for SPME, so a directly proportional relationship between the extracted amount of compounds and the initial concentration in the sample was demonstrated for all three procedures. LODs were calculated by

Table 3 Linearity, precision, and LODs of both P1-coated PC-HFME modes and conventional SPME (PDMS fiber)

PCB-10 PCB-28 PCB-52 PCB-138 PCB-153 PCB-180 PBDE-47 PBDE-99 a b c

LODs (S/N = 3, ng l−1 )

Repeatability (%RSD)a

Correlation coefficient (r) On-site PC-HFMEb

Laboratory-based PC-HFMEb

SPMEc

On-site PC-HFME

Laboratory-based PC-HFME

SPME

PC-HFME

SPME

0.999 0.998 0.999 0.996 0.998 0.998 0.991 0.993

0.996 0.996 0.999 0.999 0.999 0.998 0.997 0.999

0.996 0.997 0.999 0.990 0.998 0.984 0.999 0.993

1.6 5.1 5.5 6.9 6.9 6.4 4.5 8.6

1.3 2.6 9.0 6.6 5.2 5.1 6.0 5.0

4.4 4.1 10.1 5.9 5.0 4.5 5.6 15.7

0.07 0.04 0.04 0.15 0.11 0.21 0.06 0.09

0.34 0.14 0.13 4.26 1.62 6.74 1.39 4.63

1 ␮g l−1 . 0.05–5 ␮g l−1 . 0.5–5 ␮g l−1 .

C. Basheer et al. / J. Chromatogr. A 1139 (2007) 157–164

163

conventional SPME (PDMS-coated fiber), of artificial seawater samples spiked at 1 ␮g l−1 of individual PCBs and PBDEs. The respective extraction conditions use are as described above. Clear and well resolved total ion chromatograms (without co-elution) were obtained. A usual examination of the chromatograms suggests that both of PC-HFME methods showed good extraction capacity. 4. Real sample analysis

Fig. 6. GC–MS chromatogram of artificial seawater sample spiked with 1 ␮g l−1 of each PCB and PBDE congener, after (a) on-site PC-HFME (P1-coated HFM), (b) laboratory-based PC-HFME (P1-coated HFM), and (c) SPME (PDMS coating). [1] PCB-10 [2] PCB-28 [3] PCB-52 [4] PCB-138 [5] PCB-153 [6] PCB-180 [7] PBDE-47 [8] PBDE-99.

progressively decreasing the analyte concentration in the spiked sample such that GC–MS–SIM signals were clearly discerned at S/N of 3 at the final lowest concentration. LODs were from 0.04 to 0.21 ng l−1 for both PC-HFME approaches, and 0.13 and 6.74 ng l−1 for SPME (Table 3). The precision of the procedures were also evaluated at 1 ␮g l−1 spiked concentration levels by calculating the relative standard deviations (RSDs) (three replicates). The RSD values were between the range of 1% and 9% for PC-HFME, and between 4% and 15.7% for SPME analysis (Table 3). Fig. 6 shows typical chromatograms of extracts after on-site PC-HFME by manual shaking (with desorption in the laboratory), laboratory-based PC-HFME (with P1 coating) and

To assess the matrix effects on PC-HFME, tap water samples were extracted using the optimized laboratory-based PC-HFME conditions; no PCBs and PBDEs were detected. Therefore, the samples were spiked at two different concentrations (0.05 and 1 ␮g l−1 ) and extracted by laboratory-based PC-HFME. Table 4 lists the relative recoveries of the PCBs and PBDEs, between 89% and 111% with an RSD of less than 15%. No matrix effects were indicated by these results. Seawater sample matrix effects were also evaluated. PCBs and PBDEs were detected in all coastal seawater samples collected in Singapore. Therefore, we selected a Bedok Jetty (one of the sampling locations considered in this work) seawater sample as an example to test for matrix effects. PCBs and PBDEs concentrations detected in the Bedok Jetty were PCB10 (1.5 ng l−1 ), PCB-52 (7.2 ng l−1 ) and PBDE-99 (1.7 ng l−1 ). Laboratory-based PC-HFME (with magnetic stirring), on-site PC-HFME (with manual shaking) and SPME procedures were evaluated with Bedok Jetty seawater samples spiked with 1.5 ␮g l−1 of individual PCBs and PBDEs. Relative recoveries defined as the recoveries obtained by laboratory-based PCHFME with respect to those obtained by on-site PC-HFME were calculated using standard addition. The results are given in the Table 4. Relative recoveries were between 83% and 106% with RSDs less than 15%. The on-site PC-HFME and laboratory-based PC-HFME techniques were further applied to determine the PCBs and PBDEs in seawater samples collected from several of Singapore’s coastal locations. The results are presented in Table 5. Generally, on-site PC-HFME performed satisfactorily in comparison to laboratorybased PC-HFME. This suggests that our on-site PC-HFME is a viable approach to field sampling and extraction.

Table 4 PCB and PBDE recoveries (and precision of analysis) from tap water using laboratory-based PC-HFME and on-site PC-HFME (both using P1-coated HFMs)

PCB-10 PCB-28 PCB-52 PCB-138 PCB-153 PCB-180 PBDE-47 PBDE-99

Spiked tap water laboratory-base PC-HFME

Bedok Jetty

Bedok Jetty spiked seawater (1.5 ␮g l−1 )a

0.05 ␮g l−1

1 ␮g l−1

ng l−1

Laboratory-based PC-HFME

SPME

On-site PC-HFME

96.8 (7.1) 90.5 (7.3) 88.9 (10.8) 100.2 (13.7) 113.9 (9.1) 97.9 (14.8) 96.3 (8.5) 88.5 (9.0)

100.9 (2.7) 99.4 (5.4) 99.5 (7.8) 98.8 (5.3) 107.8 (5.3) 99.2 (2.6) 95.6 (8.7) 98.8 (4.2)

1.5

97.7 (2.3) 99.5 (6.7) 104.8 (7.9) 100.9 (2.2) 100.9 (9.9) 99.3 (1.5) 91.5 (2.7) 93.7 (8.1)

101.7 (2.4) 95.0 (7.4) 105.6 (7.5) 93.2 (10.3) 99.2 (4.1) 92.9 (10.8) 91.2 (13.6) 91.3 (13.5)

97.3 (1.9) 90.8 (7.2) 93.8 (4.7) 85.9 (11.6) 84.4 (13.1) 82.5 (15.0) 97.5 (1.8) 84.3 (13.2)

b

7.2 b b b b

1.7

Genuine Bedok Jetty spiked seawater samples were evaluated for comparison of both PC-HFME and SPME (PDMS coating). a Standard addition method. b Below LOD.

164

C. Basheer et al. / J. Chromatogr. A 1139 (2007) 157–164

Table 5 PCB and PBDE concentrations (ng l−1 ) in Singapore coastal seawater using laboratory-based PC-HFME and on-site PC-HFME (both P1-coated HFMs) Labrador Park

PCB-10 PCB-28 PCB-52 PCB-138 PCB-153 PCB-180 PBDE-47 PBDE-99 a

Sembawang Park

Laboratory-based PC-HFME

On-site PC-HFME

1.26

1.17

a

a

4.4 a

0.29 0.81 5.46 6.02

Laboratory-based PC-HFME

Kranji On-site PC-HFME

Laboratory-based PC-HFME

East Coast Park On-site PC-HFME

a

a

a

a

a

a

a

0.27 1.29 6.78 5.87

a

a

0.37

0.37

a

a

a

2.37

3.02

0.98 1.93 2.62

3.54

a

a

0.86 1.16 3.48

2.64 2.72 4.63

1.29 10.27

0.98

a

On-site PC-HFME

2.39 4.23 5.62

1.47 12.79

0.71

Laboratory-based PC-HFME

a

0.75 26.58

a

1.89 21.84

a

a

a

a

a

a

Not detected.

5. Conclusions For the fist time, polymer-coated hollow fiber membrane extraction (PC-HFME) in combined with GC–MS has been developed and demonstrated to have on-site sampling capability. Both on-site PC-HFME, and laboratory-based PC-HFME as comparison were applied to seawater to determine the concentrations of polychlorinated biphenyls and polybrominated biphenyl ethers. The on-site approach exhibited good linearity with correlation coefficients ≥0.991, high relative recoveries ranging from 83% to 98%, relative standard deviation of 9% (n = 5), and low limits of detection ranging from 0.04 to 0.21 ng l−1 ; these values compare favorably with laboratory-based PC-HFME, and laboratory-based SPME. The PC-HFME device when used onsite and can be easily stored and transported to preserve the integrity of the analytes extracted. Field sampling is advantageous because only the analytes adsorbed on the fibers are brought back to the laboratory. Transportation of large volumes of water samples is avoided, and no sampling accessories such as pumps and filters (as required in on-site SPE, for example) are needed. The fibers can be used several times and as demonstrated, provide good extraction recoveries. This is a rugged and convenient on-site sampling procedure that can be applied to other micropollutants found in aqueous environmental samples. Acknowledgements The authors gratefully acknowledge the financial support of this research by the Agency for Science, Technology and Research of Singapore, and the National University of Singapore. References [1] A. Covaci, S. Voorspoels, J. de Boer, Environ. Int. 29 (2003) 735. [2] P.O. Darnerud, G.S. Eriksen, T. Johannesson, P.B. Larsen, M. Viluksela, Environ. Health Perspect. 109 (2001) 49.

[3] C. de Wit, Chemosphere 46 (2000) 583. [4] J.J. Amaral Mendes, Food Chem. Toxicol. 40 (2002) 781. [5] Hormonally Active Agents in the Environment, National Academy Press, Washington, DC, 2000. [6] M. Polo, G. Gomez-Noya, J.B. Quintana, M. Llompart, C. Garcia-Jares, R. Cela, Anal. Chem. 76 (2004) 1054. [7] S.L. Simonich, W.M. Begley, G. Debaere, W.S. Eckoff, Environ. Sci. Technol. 34 (2000) 959. [8] D. Prats, F. Ruiz, D. Zarzo, Mar. Pollut. Bull. 24 (1992) 441. [9] N. Kannan, N. Yamashita, G. Petrick, J.C. Duinker, Environ. Sci. Technol. 32 (1998) 1747. [10] B.W. de Lappe, R.W. Risebrough, W. Walker II, Can. J. Fish. Aquat. Sci. 40 (1983) 322. [11] R.F. Pearson, K.C. Hornbuckle, S.J. Eisenreich, D.L. Swackhamer, Environ. Sci. Technol. 30 (1996) 1429. [12] J.L. Domagalski, K.M. Kuivila, Estuaries 16 (1993) 416. [13] D.R. Green, J.K. Stull, T.C. Heesen, Mar. Pollut. Bull. 17 (1986) 324. [14] E.Y. Zeng, C.C. Yu, K. Tan, Environ. Toxicol. Chem. 21 (2002) 1600. [15] E.Y. Zeng, J. Peng, D. Tsukada, T.L. Ku, Environ. Sci. Technol. 36 (2002) 4975. [16] Y. Lu, Z. Wang, J. Huckins, Aquat. Toxicol. 60 (2002) 139. [17] G.L. Hook, G.L. Kimm, T. Hall, P.A. Smith, Trends Anal. Chem. 21 (2002) 534. [18] Y. Chen, J. Pawliszyn, Anal. Chem. 75 (2003) 2004. [19] Y. Chen, J. Pawliszyn, Anal. Chem. 76 (2004) 5807. [20] T. Gorecki, J. Pawliszyn, J. Field Anal. Chem. Technol. 1 (1997) 227. [21] P. Smith, T.A. Kluchinsky, P.B. Savage, R.P. Erickson, A.P. Lee, K. Williams, M. Stevens, R.J. Thomas, AIHA J. 63 (2002) 284. [22] J.A. Koziel, J. Noah, J. Pawliszyn, Environ. Sci. Technol. 35 (2001) 1481. [23] M.Y. Jia, J. Koziel, J. Pawliszyn, J. Field Anal. Chem. Technol. 4 (2000) 73. [24] D.R. Luellen, D. Shea, Environ. Sci. Technol. 36 (2002) 1791. [25] J.D. Petty, C.E. Orazio, J.N. Huckins, R.W. Gale, J.A. Lebo, J.C. Meadows, K.R. Echols, W.L. Granor, J. Chromatogr. A 879 (2000) 83. [26] E. Zeng, D. Tsukada, D. Diehl, Environ. Sci. Technol. 38 (2004) 5737. [27] C. Basheer, V. Suresh, R. Renu, H.K. Lee, J. Chromatogr. A 1033 (2004) 213. [28] C. Basheer, S. Valiyaveettil, A. Partiban, J. Akhila, H.K. Lee, J. Chromatogr. A 1100 (2005) 137. [29] M. Vetrichelvan, S. Valiyaveettil, Chem. Eur. J. 11 (2005) 5889. ¨ [30] N. Li, F. Wania, aO.Y.D. Lei, G.L. Daly, J. Phys. Chem. Ref. Data 32 (2003) 1545. [31] Y. Marcus, Ion Solvation, Wiley, New York, 1985, p. 306.