Ion-trap tandem mass spectrometry-based analytical methodology for the determination of polychlorinated biphenyls in fish and shellfish

Ion-trap tandem mass spectrometry-based analytical methodology for the determination of polychlorinated biphenyls in fish and shellfish

Journal of Chromatography A, 1142 (2007) 199–208 Ion-trap tandem mass spectrometry-based analytical methodology for the determination of polychlorina...

632KB Sizes 0 Downloads 3 Views

Journal of Chromatography A, 1142 (2007) 199–208

Ion-trap tandem mass spectrometry-based analytical methodology for the determination of polychlorinated biphenyls in fish and shellfish Performance comparison against electron-capture detection and high-resolution mass spectrometry detection Sergei S. Verenitch a,∗ , Adrian M.H. deBruyn a , Michael G. Ikonomou b , Asit Mazumder a a

Water and Watershed Research Program, University of Victoria, P.O. Box 3020 STN CSC, Victoria, British Columbia V8W 3N5, Canada b Fisheries and Oceans Canada, Institute of Ocean Sciences, 9860 West Saanich Road, Sidney, British Columbia V8L 4B2, Canada Received 15 September 2006; received in revised form 10 December 2006; accepted 13 December 2006 Available online 23 December 2006

Abstract Optimization of the Varian Saturn 2200 ion-trap tandem mass spectrometry (IT-MS/MS) system and comparison of its data quality with two other detection methods [electron-capture detection (ECD) and high-resolution mass spectrometry (HRMS)] was pursued by measuring polychlorinated biphenyls (PCBs) levels in fish and shellfish samples. IT-MS/MS methodology provided limits of detection (LOD) comparable to those obtained by ECD but superior specificity for the detection of a selected number of 39 PCB native congeners and 9 13 C-labelled PCB standards. The method detection limits (MDLs) established for IT-MS/MS ranged between 1.0 and 5.0 pg/g on a wet weight basis while those obtained by ECD and HRMS were 1.0–4.0 pg/g and 0.1–2.0 pg/g, respectively. Overall, the results obtained in the study demonstrate that gas chromatography (GC) combined with IT-MS/MS provide higher data quality than those achievable by GC–ECD. For this particular set of target analytes the specificity achievable with IT-MS/MS was comparable to that obtained by HRMS and both techniques provided comparable data in terms of accuracy and precision. © 2006 Elsevier B.V. All rights reserved. Keywords: Polychlorinate biphenyls; GC–IT-MS/MS; ECD; HRMS; Fish; Shellfish

1. Introduction High-resolution capillary gas chromatography (GC) coupled with electron-capture detection (ECD) or high-resolution mass spectrometry (HRMS) systems have been the reference methods for determination of trace level of polychlorinated biphenyls (PCBs) in various environmental matrices for decades [1–12]. ECD was specifically designed to have high sensitivity for halogenated compounds, and is easy to operate and maintain. These advantages and its low cost have been the reasons for its wide use as the detection method of choice for the analysis of low levels of halogenated contaminants. The main disadvantage of this technique is that identification of analytes relies solely on their retention time (tR ). As a result, any compound interfering



Corresponding author. Tel.: +1 250 472 4833; fax: +1 250 721 7120. E-mail address: [email protected] (S.S. Verenitch).

0021-9673/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2006.12.068

or co-eluting with the target analytes can prevent accurate quantification of those analytes [13,14]. Additionally, Cochran and Frame [15] have shown that ECD has non-linear response behavior in a relatively narrow concentration range and it also gives a wide variation in response factors within a PCB homologue group. As a result, analysts are forced to use non-linear calibration functions or reanalyze samples when using ECD systems for determination of PCB congeners [16]. The application of low-resolution mass spectrometry (LRMS) detection in the positive ion electron impact (EI) ionization mode provides higher specificity than ECD as it adds qualitative information for analyte identification along with GC retention time. However, the limitation in sensitivity of conventional EI-MS techniques, even in the widely used selected-ion monitoring (SIM) mode, restricts its applications to samples with relatively high concentrations of PCBs, 2–10 pg [17–19]. On the other hand, negative-ion chemical ionization mass spectrometry (NCI-MS) has allowed analysts to extend the scope of

200

S.S. Verenitch et al. / J. Chromatogr. A 1142 (2007) 199–208

applications of LRMS detection methods to analysis of PCBs at trace level, 0.1–0.3 pg [20–23]. But the complexity of a use of NCI-MS methodology restricts its application. A powerful alternative to the techniques discussed above for the determination of such persistent organic pollutants (POPs) as PCBs, organochlorinated pesticides (OCPs), and polychlorodibenzo-p-dioxins and furans (PCDDs/DFs) is highresolution mass spectrometry. Very high sensitivity and powerful identification capability of HRMS have made this technique the reference method for determination of many POPs at subpg/g concentrations. However, HRMS systems are relatively expensive and require specialized laboratory infrastructure to run effectively. For applications where sub-pg/g detection capabilities are not required, researchers have explored alternative MS-based analytical techniques. Recently, as an alternative to HRMS, ion-trap tandem mass spectrometry (IT-MS/MS) systems have been applied for the determination of ultra-low levels of PCDD/DFs in food samples [24–27], PCBs in air and atmospheric aerosols [28–30] and other POPs in environmental matrices. A few papers have also been published [31,32] reporting the use of GC–IT-MS/MS for detection of PCB congeners in environmental samples. Malavia et al. [32] examined the performance of IT-MS/MS and HRMS by measuring only four non-ortho-PCBs (77, 81, 126 and 169) in fish samples. Gomara et al. [31] provided an optimized ITMS/MS-based method for the determination of a wider range (n = 22) of mono- and non-ortho-PCB congeners in environmental samples. In the later study the researchers also compared the IT-MS/MS obtained results, for three non-ortho-PCB congeners (77, 126 and 169) detected in cod liver and milk samples with those obtained by HRMS. Overall, GC with IT-MS/MS detection provides high confidence in identification of target analytes, based on a selected parent ion and a whole mass spectrum of its daughter ions, high sensitivity and selectivity, as well as low cost and ease in operation and maintenance. These characteristics have made MS/MS analysis by ion-trap systems a very competitive and widely used technique. The overall objective of this study was to establish a robust, reliable and cost-effective method which can be routinely applied to a determination of ultra-low level of PCB congeners in biological samples such as fish and shellfish. Of particular interest was the applicability of the method for the accurate and precise determination of mono-, di/tri/tetra-ortho- and nonortho-PCB congeners in the matrices of interest. A mix of 39 PCB congeners covering the range of PCBs of interest was used for this study. A set of critical instrumental parameters for the IT-MS/MS Varian system in the MS/MS mode was established. Based on literature data as well as optimization work conducted in this study, these parameters were optimized to achieve maximum sensitivity. The performance of the optimized method was compared to that obtained from ECD and HRMS techniques for the determination of PCBs in the matrices of interest. The quality of data with an identification of possible causes of the discrepancies between the three detection methods has been assessed. Methods of isotope dilution, internal recovery standard, as

well as matrix spike have been used in the quality assurance/quality control (QA/QC) procedures to properly evaluate the accuracy and precision of the methodologies examined in this work. 2. Experimental 2.1. Materials All organic solvents used in this work [acetone, dichloromethane (DCM) and n-hexane] were HPLC grade and were purchased from Fisher Scientific (Ottawa, Canada). Standard stock solutions of a mixture of 39 PCBs congeners (17, 18, 28, 31, 33, 44, 49, 52, 74, 95, 99, 101, 87, 110, 151, 82, 149, 118, 105, 132, 153, 138, 183, 187, 128, 177, 171, 156, 180, 191, 169, 170, 201, 195, 208, 194, 205, 206 and 209), a mixture of eight 13 C-labelled PCBs surrogate internal standards (SIS) (28* , 52* , 118* , 153* , 180* , 208* , 194* and 209* ) and a 13 C-labelled PCB111 recovery standard (RS) were purchased from Cambridge Isotope Labs. (Andover, MA, USA). All native and 13 C-labelled PCBs used in this study are also listed in Table 1. Anhydrous sodium sulphate (Na2 SO4 ) purchased from Fisher and used to remove moisture from a sample extract was baked at 400 ◦ C for 4 h prior to use. Solid-phase extraction (SPE) cartridges LC-Florisil, 2 g from Supelco (Bellefonte, PA, USA) were used in a clean-up procedure. 2.2. Sampling and sample preparation Various species of clams, rockfish and prawns were collected in three broadly separated regions in British Columbia, Canada. Fish samples were analyzed individually. Clams and prawns were pooled within each size class into composite samples containing enough tissue for all planned analyses. All samples were wrapped in hexane-rinsed aluminum foil, double-bagged in new zip-loc bags with unique identifier labels and stored on ice until the end of sampling day, when they were transferred to a −20 ◦ C freezer. All samples were processed usually within 3 months after their collection. Samples were kept frozen at −20 ◦ C until the day before analysis and then thawed overnight at 4 ◦ C. Samples were rinsed thoroughly with deionized water to remove any sediment. Approximately 20 g of untreated sample tissue was ground with diatomaceous earth prior to a solvent extraction. Each sample, including replicates, matrix spike and method blank, was spiked with 50 ␮L of 200 pg/␮L of 13 C-labelled SIS. Taking into account that only 1 mL of 5 mL of the extract was processed further the concentration of SIS in the final volume of 25 ␮L was 80 pg/␮L. Samples were extracted using 80 mL of DCM on a Dionex (Sunnyvale, CA, USA) ASE 100 pressurized liquid extraction (PLE) system. The extraction process was carried out under nitrogen at 100 ◦ C and 1600 psi for 32 min. Baked and ground anhydrous Na2 SO4 was then added to the sample bottle with the extract to remove moisture. The extract was evaporated until almost dry on a rotary evaporator, re-dissolved in hexane, quan-

S.S. Verenitch et al. / J. Chromatogr. A 1142 (2007) 199–208

201

Table 1 Optimized parameters for ion preparation mode (IPM) of PCB congeners and 13 C-labelled PCBs Compound

tR (min)

Segment start-end (min)

Parent ion (m/z)

Excitation storage level (m/z)

CID (V)

Product ions (m/z)

Quantification mode

Ion ratios

TrCB18 TrCB17 TrCB28a [13 C]TrCB28 TrCB31 TrCB33 TeCB52a [13 C]TeCB52 TeCB49a TeCB44a TeCB74a TeCB70a PeCB95 PeCB101 PeCB99 PeCB87a [13 C]PeCB111 PeCB110a PeCB82 HxCB151 HxCB149a PeCB118 [13 C]PeCB118 HxCB132 HxCB153 [13 C]HxCB153 PeCB105 HxCB138 HpCB183a HpCB187 HxCB128 HpCB177 HpCB171 HxCB156 HpCB180a [13 C]HpCB180 HpCB191 HxCB169a HpCB170a OcCB201a NoCB208a [13 C]NoCB208 OcCB195 OcCB194a [13 C]OcCB194 OcCB205 NoCB206a DeCB209a [13 C]DeCB209

13.82 13.89 15.46 15.46 15.46 15.75 16.62 16.62 16.76 17.31 18.61 18.77 18.80 19.67 19.82 20.59 20.52 20.91 21.28 21.28 21.71 21.89 21.89 22.64 22.72 22.72 22.80 23.58 24.09 24.26 24.51 25.09 25.26 25.39 25.92 25.92 26.08 26.73 26.89 27.12 28.24 28.24 28.34 29.06 29.06 29.19 30.23 31.22 31.22

13.6–14.0 13.6–14.0 15.3–15.6 15.3–15.6 15.3–15.6 15.6–15.85 16.45–16.90 16.45–16.90 16.45–16.90 17.05–17.45 18.45–19.00 18.45–19.00 18.45–19.00 19.55–20.00 19.55–20.00 20.40–20.75 20.40–20.75 20.75–21.10 21.10–21.50 21.10–21.50 21.50–22.20 21.50–22.20 21.50–22.20 22.40–23.00 22.40–23.00 22.40–23.00 22.40–23.00 23.30–23.80 23.80–24.75 23.80–24.75 23.80–24.75 24.90–25.50 24.90–25.50 24.90–25.50 25.75–26.20 25.75–26.20 25.75–26.20 26.55–27.00 26.55–27.00 27.00–27.20 28.00–28.50 28.00–28.50 28.00–28.50 28.85–29.40 28.85–29.40 28.85–29.40 30.00–30.40 31.00–31.40 31.00–31.40

(M+2) 258 (M+2) 258 (M+2) 258 (M+2) 270 (M+2) 258 (M+2) 258 (M+2) 292 (M+2) 304 (M+2) 292 (M+2) 292 (M+2) 292 (M+2) 292 (M+2) 326 (M+2) 326 (M+2) 326 (M+2) 326 (M+2) 338 (M+2) 326 (M+2) 326 (M) 360 (M) 360 (M+2) 326 (M+2) 338 (M) 360 (M) 360 (M) 372 (M+2) 326 (M) 360 (M+2) 396 (M+2) 396 (M) 360 (M+2) 396 (M+2) 396 (M) 360 (M+2) 396 (M+2) 406 (M+2) 396 (M) 360 (M+2) 396 (M+4) 430 (M+4) 464 (M+4) 476 (M+4) 430 (M+4) 430 (M+4) 442 (M+4) 430 (M+4) 464 (M+4) 498 (M+4) 510

113.7 113.7 113.7 119 113.7 113.7 128.8 134.1 128.8 128.8 128.8 128.8 143.8 143.8 143.8 143.8 149.1 143.8 143.8 158.8 158.8 143.8 149.1 158.8 158.8 164.1 143.8 158.8 174.8 174.8 158.8 174.8 174.8 158.8 174.8 179.2 174.8 158.8 174.8 189.8 205.3 210.1 189.8 189.8 195 189.8 204.8 220.8 225.2

1.20 1.20 2.10 1.70 2.10 1.80 1.20 1.30 1.20 1.10 2.10 2.10 0.90 1.30 1.30 1.20 1.40 1.80 1.20 0.80 1.30 2.10 2.40 1.50 1.50 1.70 2.10 1.50 1.10 1.30 1.50 0.90 1.20 2.40 1.30 1.60 1.80 2.40 1.20 0.90 1.00 1.50 1.20 1.40 1.95 2.40 1.20 2.10 2.10

186 + 221 + 256 186 + 221 + 223 + 258 186 + 188 + 221 + 258 198 + 233 + 268 186 + 188 + 221 + 258 186 + 221 + 256 222 + 255 + 290 234 + 267 + 302 220 + 255 + 290 220 + 255 + 290 220 + 222 + 255 + 290 220 + 222 + 256 + 290 254 + 256 + 289 + 291 256 + 289 + 324 256 + 289 + 324 256 + 289 + 324 268 + 266 + 303 + 336 254 + 256 + 289 + 324 256 + 291 + 324 290 + 325 + 358 290 + 323 + 325 + 358 254 + 256 + 289 + 324 266 + 268 + 301 + 336 290 + 321 + 323 + 358 269 + 290 + 323 + 358 335 + 358 + 370 254 + 256 + 289 + 290 + 324 290 + 323 + 325 + 358 326 + 359 + 394 324 + 326 + 359 + 394 290 + 323 + 325 + 358 324 + 326 + 359 + 397 324 + 357 + 359 + 392 288 + 290 + 323 + 358 324 + 326 + 359 + 394 334 + 336 + 369 + 404 322 + 324 + 357 + 392 288 + 290 + 323 + 358 324 + 359 + 394 360 + 393 + 395 + 428 392 + 394 + 429 + 462 404 + 406 + 439 + 474 358 + 360 + 393 + 395 + 428 358 + 360 + 391 + 393 + 428 370 + 372 + 405 + 407 358 + 360 + 393 + 395 + 428 392 + 394 + 427 + 429 + 462 426 + 428 + 461 + 496 438 + 440 + 473 + 508

MS/MS MS/MS MRM MRM MRM MS/MS MRM MRM MRM MS/MS MRM MRM MRM MS/MS MS/MS MRM MRM MS/MS MRM MRM MRM MRM MRM MRM MRM MRM MRM MS/MS MRM MRM MRM MRM MRM MRM MRM MRM MRM MRM MRM MS/MS MRM MRM MRM MRM MRM MRM MS/MS MRM MRM

13.8/53.9/32.5 12.8/48.9/14.6/23.4 51.0/15.8/18.7/14.9 49.8/18.5/28.6 54.2/11.6/22.4/11.1 62.0/26.1/11.7 27.7/56.7/12.6 24.3/42.1/17.1 20.2/51.7/28.1 17.5/57.9/24.6 52.1/31.7/8.5/7.6 51.1/31.8/7.5/8.6 12.8/35.7/42.0/24.9 35.7/42.0/22.7 33.6/45.1/18.9 28.6/48.7/22.1 33.6/36.1/8.7/12.3 35.6/37.8/11.4/10.5 26.8/49.2/18.9 33.2/51.4/14.3 15.5/20.4/50.1/13.6 35.2/40.2/5.2/12.3 44.9/39.3/7.0/8.2 34.6/27.5/24.1/11.4 8.4/32.8/40.6/12.2 53.6/22.3/13.6 36.8/39.6/8.6/9.4/5.1 24.2/18.5/45.7/12.7 10.2/65.4/24.1 13.1/15.4/50.6/20.3 35.6/52.6/6.2/3.4 6.3/8.6/62.5/12.3 21.2/27.2/50.0/1.8 15.4/22.3/47.2/13.4 18.6/20.4/38.9/21.2 20.4/43.6/26.5/6.2 24.0/53.6/2.2/14.4 31.7/50.8/6.6/11.4 16.8/68.9/8.7 1.9/14.0/58.3/18.6 16.4/22.0/21.9/40.4 20.8/34.1/12.8/32.3 6.6/7.8/27.2/40.8/15.6 16.8/17.2/28.6/30.9/6.2 8.0/22.6/44.3/18.2 24.6/38.4/13.2/11.6/10.2 6.4/8.3/18.8/29.8/33.8 30.9/40.6/16.6/12.4 32.4/47.4/16.1/3.3

a

Congeners analyzed on HRMS.

titatively transferred into a 10 mL glass vial and blown down to 5 mL using N2 stream. Since the lipid content in the sample extract did not exceed 100 mg, the cleaning procedure was simplified to the following. Clean up of each sample was performed using 1 mL of the extract and 12 mL (2 g) Florisil SPE cartridge [33]. All target PCBs were eluted from the SPE cartridges using 10 mL of

hexane. The extract was then evaporated under N2 to 0.1 mL and transferred to a GC vial. All sample extracts were analyzed within 14 days in all three systems, GC–IT-MS/MS, GC–HRMS and GC–ECD. For certain GC–MS/MS and HRMS experiments the final extract was evaporated under N2 almost to dryness. Twenty-five microlitres of 41 pg/␮L solution of recovery standard, [13 C]PCB111 was added to each vial prior to analysis.

202

S.S. Verenitch et al. / J. Chromatogr. A 1142 (2007) 199–208

2.3. Instrumentation 2.3.1. ECD and IT-MS/MS chromatographic conditions The samples analyzed using ECD and IT-MS/MS detection techniques were run at the identical chromatographic setup: Varian CP 3800 gas chromatograph equipped with a 30 m × 0.25 mm i.d., 0.25 ␮m film thickness CP-SIL 8CB-MS capillary column, a split/splitless injector (split ratio = 100 was on at 0.5 min) and Varian CP-8200 autosampler. The oven was operated under the following conditions: injection port temperature 225 ◦ C; injection volume 2 ␮L; carrier gas (He) at the constant flow rate of 1 mL/min; the initial oven temperature was held at 100 ◦ C for 2 min, then raised to 160 ◦ C at the rate of 15 ◦ C/min, then to 300 ◦ C at 5 ◦ C/min, with the final hold time of 10 min. 2.3.2. ECD The ECD system was operated at 310 ◦ C, contact potential at −250 mV and make up flow 25 mL/min. Twenty of 36 samples were analyzed on GC–ECD. These samples were simultaneously analyzed by IT-MS/MS. Identification of PCB congeners was solely based on their tR which were established from the analysis of authentic PCB standard solutions. Verification of correct tR assignment for samples analyzed by GC–ECD was established from the analysis of a replicate sample spiked with known amounts of the 39 native target PCB congeners. Such a spiked replicate sample was included in each batch of samples prepared for analysis. 2.3.3. HRMS detection The analyses of PCB congeners by GC–HRMS were performed according to the procedures described in detail previously [34]. The system consisted of a HP 5890 (HewlettPackard, Palo Alto, CA, USA) gas chromatograph coupled to a triple sector HRMS VG-AutoSpec-S (VG-Analytical, Manchester, UK) and also equipped with an CTC A200S (CTC Analytics, Zwingen, Switzerland) autosampler. The injector was operated in splitless mode at 282 ◦ C and was purged at 2 min. The chromatographic column used was DB-5 (60 m × 0.25 mm i.d., 0.1 ␮m film thickness; J&W Scientific, Folsom, CA, USA). Ultra-high-purity helium was used as carrier gas at a constant head pressure of 25 psi. The temperature conditions for the chromatographic column were the following: the initial temperature was held at 80 ◦ C for 2 min, then raised to 150 ◦ C at the rate of 8 ◦ C/min, then to 285 ◦ C at 4 ◦ C/min, with the total run time of 44.5 min. The mass spectrometer source temperature was set at 305 ◦ C, with the interface line at 260 ◦ C. All PCB analyses were carried out under positive ion electron impact ionization conditions with the filament in the trap stabilization mode at 600 ␮A and electron energy of 28–35 eV. The instrument was routinely resolving at 10,000 resolution power (10 K RP) and data were acquired in the single ion resolution mode (SIR) for achieving maximum sensitivity possible. The two most abundant isotope ions, M+ and (M+2)+ or (M+2)+ and (M+4)+ (see Table 1), of known relative abundance were monitored for each homologue series and 13 C-labelled surrogate standards.

Only 16 out of the 36 samples extracted were analyzed by GC–HRMS. The results for only 18 out of the total 39 native PCB congeners (28, 52, 49, 44, 74/61, 70, 87/115, 110, 149, 169, 183, 180, 170/190, 201, 194, 208, 206 and 209) were selected for the performance comparison between GC–HRMS and GC–ITMS/MS. 2.3.4. IT-MS/MS optimization The methodology of IT-MS/MS in MS/MS mode has been actively studied for a number of years now. It has been optimized and applied to several organic environmental contaminants in various matrices. The instrumental parameters which determine the performance efficiency of IT-MS/MS and require optimization are instrument specific. For the ion-trap mass spectrometer from Varian, the instrument used in this study, such parameters were narrowed down to: parent isolation window, collision-induced dissociation (CID) amplitude, CID time, acquisition mass range, isolation time, broadband amplitude, CID bandwidth, modulation range, filament current and iontrap temperature. According to Kuchler and Brzezinski [24] and the Varian GC–MS manual [35], some of these parameters as well as other stable tandem mass spectrometry conditions were set as follows. For ionization mode—electron impact at 70 eV, multiplier offset ±300 V, target total ion counts (TIC) at 2000, maximum ionization time 25,000 ␮s, pre-scan ionization time 1500 ␮s, background mass 45 m/z, RF dump value 650 m/z, ejection amplitude 20 volts; for isolation mode—low edge offset 6 steps, high edge offset 2 steps, high edge amplitude 30.0 V; for ion preparation mode (IPM)—isolation window 3 m/z, waveform type: resonant, modulation range 2 steps, modulation rate 3000 ␮s/step, number of frequencies 1, CID frequency offset 0 Hz. The other six instrumental parameters of different stages of IT-MS/MS technique that required fine optimization were: scan time (ST), emission current (EC), isolation time (IT), excitation time (ET), excitation amplitude (EA) or CID amplitude and excitation storage level (ESL). In this study four instrumental parameters were set at fixed values. The optimized values for these parameters were based on the results found by Mandalakis et al. [30] and Brochu et al. [36]. They were 0.5 s/scan for ST [36], 80 ␮A for EC [36], 5 ms for IT [30], 20 ms for ET [30]. The excitation storage level was calculated for each congener individually based on its molecular weight and “qz ” value equal to 0.400 [30,36]. The only parameter that required optimization was CID amplitude. To be able to identify retention time of selected PCB congeners, a mixture of the 39 PCB congeners and a mixture of 9 isotope-labelled PCBs were analyzed separately on the GC–ITMS operating in full scan mode. The data base of retention times of PCB congeners relative to PCB209 obtained on different GC columns [37] was referenced to confirm proper identification. A molecular ion of each congener was selected as the precursor/parent ion for the sequential application of MS/MS conditions, Table 1. The remaining MS/MS parameter required optimization, CID amplitude was determined using the automated method development (AMD) option built into the Varian Saturn GC–MS/MS software. This parameter refers to radio frequency voltage applied to the end-cap electrodes of the ion-trap

S.S. Verenitch et al. / J. Chromatogr. A 1142 (2007) 199–208

mass analyzer in MS/MS mode and it is used to control the dissociation of the selected precursor ions. AMD uses up to 10 different CID voltages/amplitudes for the same precursor ion. The optimization was performed at the resonant conditions in two steps. For the first step, the CID voltage was incrementally raised with the interval 0.3 V by using the AMD option from 0 to 3.0 V and full mass spectra of daughter ions were acquired at each CID amplitude. Fig. 1a and b presents the chromatograms and the product ion mass spectra obtained in MS/MS mode for PCB congener 52 and [13 C]52, respectively. The mass spectra shown in Fig. 1a and b were produced at different dissociation voltages (CID) using the AMD function. The best CID was the one that gave the highest yield of the sum of daughter ions (m/z = 255, 222 for PCB52 and 267, 234 for [13 C]PCB52) and very little of the precursor ions M/M+2 (m/z = 290/292 and 302/304 for PCB52 and [13 C]PCB52, respectively). Once a rough estimate of the most suitable CID amplitude was determined, the voltage was optimized using the AMD function at lower increments, 0.05 V. The changes in precursor and fragment ion abundances as a function of CID amplitude are plotted in Fig. 2a for PCB52 and b for [13 C]PCB52. The optimum CID voltages resulting from these experiments were 1.20 V for

203

PCB52 and 1.30 V for [13 C]PCB52. Otimization of CID parameters was performed for all targeted PCBs and the optimum values established are reported in Table 1. A precursor ion and a selected number of the most abundant daughter ions were chosen for the identification and quantification of each PCB congener. Their m/z values of the product ions selected are listed in Table 1. One of the criteria of analyte identification in real samples under MS/MS conditions was that the ratio of the product ions versus the precursor ions had to be within ±20% of that established from the analysis of the corresponding authentic standards. 2.4. Quality assurance quality control (QA/QC) To insure ultimate instrument performance, for all three systems (GC–ECD, GC–HRMS and GC–IT-MS/MS), a number of tests were performed on a routine basis. The tests were designed to assess: GC column performance, retention time windows, ultimate sensitivity, multipoint calibration and linearity, instrument detection limits, sample carryover, continuous calibration verification. The same criteria were applied to all detection systems for each of these seven tests. The HRMS and MS/MS systems were also optimized for resolution, transmission and

Fig. 1. Daughter ion full mass spectra of PCB52 (a) and [13 C]PCB52 (b) obtained in AMD mode at different CID amplitudes. The intensities of both chromatograms are presented as total ion counts (TIC), m/z.

204

S.S. Verenitch et al. / J. Chromatogr. A 1142 (2007) 199–208

Fig. 1. (Continued ).

mass calibration using perfluorokerosene (PFK) for the HRMS and perfluorotributylamine (PFTBA) for the IT-MS/MS. The identification of target analytes using GC–ECD was based on a match of retention time of an analyte with that obtained from the analysis of the corresponding standard. The criteria for analyte identification using GC–HRMS were those specified in the EPA 1668 method [38]. For GC–IT-MS/MS methodology, compounds were identified as PCB (either native or surrogate) congeners, only when the chromatographic peaks obtained satisfied all of the following criteria. (1) The retention time of a specific congener must be within ±0.2 min window of that obtained during analysis of the authentic compounds in the calibration standards. In the case of a congener for which a labelled analogue is present in the surrogate spiking solution, all native and surrogate ion peak maxima must be coincident within 3 s. (2) The threshold of the spectral match of a specific PCB congener in a sample and its analogue in a standard mixture was set at 700 out of max 1000. Only the ions with intensities higher than 50% of the reference spectrum base peak were considered when calculating the spectral match. (3) All product ions of each specific PCB congener (see Table 1) must be present, and must be detected at their exact m/z. (4) The signal-to-noise

ratio in each congener GC–MS/MS segment must be >3 for a sample extract, and >10 for a calibration standard. (5) The ratios between the integrated isotope signals of the product ions chosen for each congener must be within ±20% of the values outlined in Table 1. Each batch of samples consisted of nine samples, plus three QA/QC samples: a sample duplicate, a replicate sample spiked with known amounts of target analytes and a procedural blank. The replicate samples were spiked with known amounts of the mixture of the 39 PCB congeners and were used as a laboratory reference material (LRM). Data obtained from the spiked replicate sample were also used in ECD analyses to correct for recoveries of the native PCB congeners. For HRMS and MS/MS systems a mix of 13 C-labelled PCB congeners (Table 1) was used to determine the recoveries efficiency. These were added to each sample prior to extraction and their concentration in the final extract was adjusted to 80 pg/␮L. Corrections for recoveries of the native PCB congeners were made against the corresponding 13 C-labelled PCB surrogate internal standards. The 13 C-labelled PCB111 was used as a recovery standard. It was added to a dry sample extract at the final concentration of 41 pg/␮L. In order for the data to be acceptable the following QA/QC criteria had to be met. The reproducibility of duplicate analysis

S.S. Verenitch et al. / J. Chromatogr. A 1142 (2007) 199–208

205

Fig. 3. MS/MS chromatograms and “daughter ions” spectra of the co-eluting PCB70 and 95 congeners at the concentration levels of 10 and 5 pg, respectively. Fig. 2. Ion abundance curves obtained by plotting precursor ion and daughter ion abundances (expressed as the response signal) as a function of the CID amplitudes. (a) Precursor ions m/z = 292 (), m/z = 290 (♦) and daughter ions m/z = 255 (), m/z = 222 (×) for PCB52. (b) Precursor ions m/z = 304 (), m/z/ = 302 (♦) and its daughter ions with m/z = 269 () and m/z = 234 (×) for [13 C]PCB52.

had to be between 60 and 130% and the percentage recoveries of all surrogate internal standards (HRMS or MS/MS system) used or native congeners in a replicate sample (ECD) had to be between 30 and 140%. 3. Results and discussion 3.1. Chromatographic resolution All samples were prepared once (unless done as a duplicate for QA). The extraction and purification procedures were briefly described above. Sixteen out of 36 samples were analyzed by HRMS and IT-MS/MS to establish the comparability of the techniques in determination of trace level of PCB congeners in such sample matrices as rockfish and shellfish. Twenty other samples were analyzed by ECD and IT-MS/MS to compare another industry standard method with the novel methodology of tandem mass spectrometry. The chromatographic conditions exploited in the ECD and ITMS/MS detection methods were described above. Under these conditions the retention times of the trichlorinated biphenyls PCB18 and PCB17, were found to be very close but a resolution of more than 35% was achieved between the peaks.

PCB congeners 28 and 31, both trichlorinated biphenyls, due to their co-elution (see Table 1) and identical mass spectra had to be quantified as a sum. PCBs 70 (tetraCB) and 95 (pentaCB) were co-eluting (Fig. 3) yet could be accurately quantified by IT-MS/MS technique due to the differences in their precursor ions (M+2)+ , m/z = 292 and 326, respectively (see Table 1). PCBs 82 (pentaCB) and 151 (hexaCB) were coeluting as well but could also be accurately quantified when in MS/MS mode due to the differences in their precursor ions, 326 (M+2)+ and 360 (M). These groups of co-eluting PCB congeners (28/31, 70/95 and 82/151) were fully resolved in GC–HRMS analysis as different GC conditions were used (see Section 2.3.3). As far as a comparison of the results between ECD and IT-MS/MS detection techniques is concerned, the results for PCBs 70 and 95, as well as 82 and 151, obtained on the ECD detection system, were presented as the sum of 70/95 and 82/151. PCBs 28 and 31 could not be separated at the GC conditions chosen for GC–ECD and GC–IT-MS/MS systems. Thus, these two congeners were quantified as the sum of two peaks for both ECD and MS/MS detection methods. Since in actual fish samples there is a possibility of a presence of other non-mass resolved PCB congeners which are not chromatographically separated on 5-ms phase column, 74/61, 87/115, 170/190 and 138/160/163/164, these analytes were quantified by all detection systems as a sum based on the assumption that the co-eluting pairs produce similar product spectra. All other PCB congeners from the group of 39 were quantified individually on each of the three detection systems, ECD, HRMS and IT-MS/MS.

206

S.S. Verenitch et al. / J. Chromatogr. A 1142 (2007) 199–208

3.2. Detection limits for PCB congeners on HRMS, MS/MS and ECD The instrumental limits of detection (LOD) for HRMS, MS/MS and ECD for all PCB congeners analyzed in this study ranged between 0.02 and 0.9, 0.2 and 1.2 and 0.2 and 0.8 pg, respectively. These instrument LODs translated to the method detection limits (MDLs) of 0.1–2.0 (HRMS), 1.0–5.0 (IT-MS/MS) and 1.0–4.0 pg/g (ECD) for a 4 g of wet weight tissue sample. It should be noted here that 20 g of tissue sample was extracted but only 1 mL of the 5 mL extract was purified and analyzed. The method detection limits for HRMS and MS/MS detection systems were calculated for each specific PCB congener. The calculation steps involved the measurements of the minimum detectable area of a PCB homologue (based upon three times of the maximum height of the noise) and the recovery of the corresponding isotope-labelled surrogate internal standard, SIS. The final MDL values determined for each individual PCB congener were corrected against the recovery of its isotope labelled SIS. For ECD, the MDLs were determined using LRM sample and the criteria that the area of a minimum detectable amount for each PCB homologue must be three times higher than the background noise in that region. All results were reported in pg/g of wet weight of sample tissue. 3.3. Comparison of ECD and MS/MS results Sixteen marine biota samples and four blanks have been comparatively studied using ECD and MS/MS. The results of all samples were corrected against the corresponding blanks. Since some PCB congeners had co-eluting interferences and could not be individually quantified by ECD, their results were presented as a sum. For this reason, and for the sake of comparison between the two methods, the co-eluting peaks such as PCBs 70 and 95, and 82 and 151, although easily quantifiable individually by MS/MS, were reported as a sum of 70/95 and 82/151. PCB congeners 28 and 31 could not be resolved by both detection systems and were quantified as a sum. Since ECD does not distinguish between native and 13 C-labelled PCBs, the comparison of the results for native PCBs between ECD and MS/MS was performed with the exclusion of the following native PCB congeners for which isotope labelled isomers were used as a part of QA/QC procedure: 28, 52, 118, 153, 180, 194, 208 and 209. In real samples, peaks identified by ECD as PCB congeners 74, 87, 105, 128, 132 and 171 were not possible to confirm by MS/MS experiments. The mass spectra obtained did not match those of the authentic standards. As such, these peaks were considered as interferences skewing the accurate identification of PCBs 74, 87, 105, 132 and 171 under GC–ECD. To avoid data misinterpretation these PCBs were excluded from the performance comparison between ECD and MS/MS. The rest of the PCB congeners analyzed in this study were subdivided onto two groups, PCB congeners eluting at earlier retention times (17, 18, 33, 44, 49, 70/95, 82/151, 99, 101, 110, 138, 149 and 183) and those eluting at later tR s (128, 156, 170, 177, 187, 191, 195, 201, 205 and 206). The results for both groups of the light and

Fig. 4. Comparison of the results obtained for the group of light PCB congeners, () 17, () 18, () 33, (*) 44, (+) 49, () 70/95, (䊉) 82/151, (−) 99, (♦) 101, (×) 110, () 138, () 149 and (–) 183 analyzed in marine biota samples using ECD and MS/MS.

heavy PCB congeners obtained by ECD and MS/MS detectors are compared in Figs. 4 and 5. The dashed line represents the ideal situation when both detection methods produce absolutely identical results. In this case the parameters of the regression equation described by the slope and y-intercept are equal to 1 and 0, respectively. The resulting data bases obtained by ECD and MS/MS can be described with the regression equations y = 1.02x (r2 = 0.80) and y = 1.14x (r2 = 0.69) with the confidence interval at 95%, for the light and heavy groups, respectively. The results of analysis of surrogate internal standards spiked in the samples at known concentrations were also compared between ECD and MS/MS (Fig. 6). The regression equation for these results is y = 1.25x (r2 = 0.40) with the confidence interval at 95%. Higher residual scatter in this case may be explained by the impact of native PCB congeners on the quantification of 13 C-labelled PCB surrogate internal standards by ECD. Despite obvious advantages such as high sensitivity, low cost and ease of maintenance and operation, ECD is prone to interferences compromising the identification and accurate quantification of target PCB congeners. Sample or blank interferences, co-eluting peaks and the use of isotope-labelled internal standards may all negatively affect the quality of the results obtained by ECD. These factors become significant espe-

Fig. 5. Comparison of the results obtained for the group of heavy PCB congeners, () 128, (×) 156, (䊉) 170, () 177, () 187, (♦) 191, () 195, (*) 201, () 205 and () 206 analyzed in marine biota samples using ECD and MS/MS.

S.S. Verenitch et al. / J. Chromatogr. A 1142 (2007) 199–208

Fig. 6. Comparison of the results obtained for the surrogate internal standards 13 C-labelled PCBs, () 52* , () 118* , () 153* , (×) 180* , (*) 194* , (䊉) 208* and () 209* spiked in marine biota samples using ECD and MS/MS.

207

Fig. 8. Comparison of the results obtained for the group of heavy PCB congeners, () 149, (♦) 169, (×) 180, () 183, () 170, () 194, (䊉) 201, () 206 and () 208 analyzed in marine biota samples using MS/MS and HRMS.

cially when complex matrices are to be examined and/or when target analytes are present at trace levels. 3.4. Comparison of MS/MS and HRMS results Sixteen samples of marine biota were processed and simultaneously analyzed on both HRMS and MS/MS. The method of isotope dilution was not applied in this part of the study. PCB209 was used as a surrogate internal standard and was spiked to each sample before extraction at a known concentration. All PCB congeners studied in this part of the work were subdivided to two groups. The results obtained on both systems for the group of lighter PCB congeners 28, 52, 49, 44, 74, 70, 87 and 110 are compared in Fig. 7. The results for heavier PCB congeners 149, 169, 183, 180, 170, 201, 194, 208 and 206 are compared in Fig. 8. The performance of the two systems for the detection of PCB209 spiked at different concentrations in biota samples is compared in Fig. 9. The regression lines describing these data obtained on both detection systems show excellent correspondence for all PCB congeners: y = 1.01x and r2 = 0.94, Fig. 7; y = 0.96x and r2 = 0.94, Fig. 8; y = 1.08x and r2 = 0.99, Fig. 9 for light, heavy and PCB209 congeners,

Fig. 7. Comparison of the results obtained for the group of light PCB congeners, () 28, () 52, (×) 44, () 49, (䊉) 70, (*) 74, () 87 and () 110 analyzed in marine biota samples using MS/MS and HRMS.

Fig. 9. Comparison of the results obtained for the surrogate internal standard PCB209 spiked in marine biota samples using MS/MS and HRMS.

respectively. The confidence interval for all the regressions is 95%. 4. Conclusions Based on the comparative analyses of 32 samples of rockfish and shellfish tissue, as well as four procedural blanks this study has demonstrated that ECD provides less confidence of identification than mass spectrometry. As a result, co-eluting peaks or interferences become an issue affecting ECD’s data quality. It was shown that IT-MS/MS methodology in MS/MS mode can provide limits of detection comparable to those obtained by ECD. The method detection limits established for 39 PCB congeners and 9 13 C-labelled PCB surrogate internal standards determined using MS/MS varied slightly depending on the PCB congener and ranged between 1.0 and 5.0 pg/g on a wet weight basis using a 4 g tissue sample. The MDLs obtained from HRMS ranged between 0.1 and 2 pg/g for the same size sample. For the specific analytes examined in this study, the specificity achievable with IT-MS/MS was comparable to that obtained by HRMS and both techniques provided comparable data in terms of accuracy and precision, however, lower MDLs were achievable by HRMS. This study demonstrated that GC with IT-MS/MS instrumentation in MS/MS mode permits a higher data quality than GC–ECD.

208

S.S. Verenitch et al. / J. Chromatogr. A 1142 (2007) 199–208

Acknowledgements This research was supported by the Natural Sciences and Engineering Research Council (NSERC) Industry Research Chair Grant, Capital Region District (CRD) Environmental Services Department, CRD Water Services Department and the Ministry of Environment. Our particular thanks are to Rob Christian from Varian Inc. for technical support and valuable recommendations and to Maike Fischer from the Institute of Ocean Sciences for performing the GC–HRMS analyses. References [1] Method 8082, US Environmental Protection Agency (EPA), Revision 0, December 1996. [2] M.D. Mullin, C.M. Pochini, S. McCrindle, M. Romkes, S.H. Safe, L.M. Safe, Environ. Sci. Technol. 18 (1984) 468. [3] M.S. Rahman, S. Bowadt, B. Larsen, J. High Resolut. Chromatogr. 16 (1993) 731. [4] M. Morosini, J. Schreitmuller, U. Reuter, K. Ballschmitter, Environ. Sci. Technol. 27 (1993) 1517. [5] C.Y. Chen, Y.C. Ling, J. High Resolut. Chromatogr. 17 (1994) 784. [6] C. Porte, J. Albaiges, Arch. Environ. Contam. Toxicol. 26 (1993) 273. [7] A. Ruus, K.I. Ugland, O. Espeland, J.U. Skaare, Mar. Environ. Res. 48 (1999) 131. [8] Y. Sapozhnikova, O. Bawardi, D. Schlenk, Chemosphere 55 (2004) 797. [9] M.D.L. Easton, D. Luszniak, E. Von der Geest, Chemosphere 46 (2002) 1053. [10] S. Burreau, Y. Zebuhr, D. Broman, R. Ishaq, Chemosphere 55 (2004) 1043. [11] J.C. Colombo, C. Brochu, C. Bilos, S. Moore, Environ. Sci. Technol. 31 (1997) 3551. [12] L.R. Bordajandi, G. Gomez, M.A. Fernandez, E. Abad, J. Rivera, M.J. Gonzalez, Chemosphere 53 (2003) 163. [13] K.C. Jones, V. Burnett, R. Duarte-Davidson, K.S. Waterhouse, Chem. Br. 27 (1991) 435. [14] S.J. Harrad, A.S. Stewart, R. Boumphrey, R. Duarte-Davidson, K.C. Jones, Chemosphere 24 (1992) 1147. [15] J.W. Cochran, G.M. Frame, J. Chromatogr. A 843 (1999) 323.

[16] E. Storr-Hansen, J. Chromatogr. 558 (1991) 375. [17] M.D. Erickson, Analytical Chemistry of PCBs, second ed., Lewis, New York, 1997, p. 235. [18] M. Perugini, M. Cavaliere, A. Giammarino, P. Mazzone, V. Olivieri, M. Amorena, Chemosphere 57 (2004) 391. [19] S. Bayarri, L.T. Baldassarri, N. Iacovella, F. Ferrara, A. di Domenico, Chemosphere 43 (2001) 601. [20] H. Bagheri, P.E.G. Leonards, R.T. Ghijsen, U.A.Th. Brinkman, Int. J. Environ. Anal. Chem. 50 (1993) 257. [21] R. Chaler, R. Vilanova, M. Santiago-Silva, P. Fernandez, J.O. Grimalt, J. Chromatogr. A 823 (1998) 73. [22] B. Rothweiler, J.D. Berset, Chemosphere 34 (1997) 2451. [23] B. Rothweiler, J.D. Berset, Chemosphere 38 (1999) 1517. [24] Th. Kuchler, H. Brzezinski, Chemosphere 40 (2000) 213. [25] D.G. Hayward, Chemosphere 34 (1997) 929. [26] D.G. Hayward, J. Holcomb, R. Glidden, P. Wilson, M. Harris, V. Spencer, Chemosphere 43 (2001) 407. [27] C. Helen, M. Lemasle, A. Laplanche, E. Genin, J. Mass Spectrom. 36 (2001) 546. [28] P.E.G. Leonards, U.A.Th. Brinkman, W.P. Cofino, Chemosphere 32 (1996) 2381. [29] R. Barro, S. Ares, C. Garcia-Jares, M. Llompart, R. Cela, J. Chromatogr. A 1072 (2005) 99. [30] M. Mandalakis, M. Tsapakis, E.G. Stephanou, J. Chromatogr. A 925 (2001) 183. [31] B. Gomara, M.A. Fernandez, M.-J. Gonzalez, L. Ramos, J. Sep. Sci. 29 (2006) 123. [32] J. Malavia, F.J. Santos, M.T. Galceran, J. Chromatogr. A 1056 (2004) 171. [33] Method 3620B, US Environmental Protection Agency (EPA), Revision 0, December 1996. [34] M.G. Ikonomou, T.L. Fraser, N.F. Crewe, M.B. Fischer, I.H. Rogers, T. He, P.J. Sather, R.F. Lamb, Can. Tech. Rep. Fish. Aquat. Sci. 2389 (2001) 95. [35] GC–MS Manual, Varian, Walnut Creek, CA, 1999. [36] C. Brochu, S. Moore, G. Hamelin, Varian Application Note 64, Varian, Walnut Creek, CA, 2000. [37] E. Rogers, M. Petreas, J.S. Park, G. Zhao, M.J. Charles, J. Chromatogr. B 813 (2004) 269. [38] Method 1668, US Environmental Protection Agency (EPA), Revision A, December 1999.