Journal of Chromatography A, 1391 (2015) 18–30

Contents lists available at ScienceDirect

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

Multiscreening determination of organic pollutants in molluscs using matrix solid phase dispersion H. Ziarrusta ∗ , M. Olivares, A. Delgado, O. Posada-Ureta, O. Zuloaga, N. Etxebarria Department of Analytical Chemistry, Faculty of Science and Technology, University of the Basque Country (UPV/EHU), P.O. Box 644, 48080 Bilbao, Spain

a r t i c l e

i n f o

Article history: Received 16 December 2014 Received in revised form 25 February 2015 Accepted 26 February 2015 Available online 6 March 2015 Keywords: Organic pollutants Matrix solid phase dispersion Multiresidue extraction Molluscs Gas chromatography–tandem mass spectrometry

a b s t r a c t This work describes the optimisation, validation and application of matrix solid-phase dispersion (MSPD) coupled to gas chromatography mass spectrometry, both single quadrupole (GC–MS) and tandem (GC–MS/MS), for the quantification in molluscs of up to 40 different analytes belonging to several families of priority and emerging organic contaminants, including polycyclic aromatic hydrocarbons (PAHs), polychlorinated biphenyls (PCBs), polybrominated diphenyl ethers (PBDEs), organochlorine pesticides (OCPs), organophosphorus pesticides (OPPs) and musk fragrances. The MSPD procedure was fully optimised with a special focus on the clean-up strategy. The best recoveries were obtained using glass syringes, 0.30 g of freeze-dried sample, 0.30 g of Florisil as solid support, 4.00 g of activated silica and 25 mL of dichloromethane as elution solvent. Using GC–MS/MS the method afforded good linearities (r2 , between 0.980 and 0.9996), adequate repeatability and reproducibility (RSD < 17% and 33%, respectively) and low instrumental limits of detection (between 0.010 and 2.74 ng mL−1 ). The accuracy of the method was evaluated using different approaches, i.e. assessment of spiked fish hatchery samples, laboratory reference material and standard reference material (SRM 2977). Satisfactory apparent recoveries were obtained for all the target analytes after correction with the corresponding labelled surrogate, except for PAHs in the case of SRM 2977, which required the use of the standard addition method. Finally, MSPD was applied to mollusc species collected in Colombia and Nicaragua, where PAHs, PCBs, musks and pesticides were detected at low ng g−1 levels. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Different environmental compartments tend to end up accumulating a huge variety of organic pollutants, primarily resulting from anthropogenic sources. Although these organic pollutants may be present at trace levels, their adverse effects on aquatic life, on animals and even on humans are increasing the concern of society and the scientific community [1,2]. In consequence, the European Union has adopted specific regulations such as the Water Framewok Directive (2013/39/EU), in which the environmental quality standards (EQSs) for 45 priority substances have been included for water and biota samples. The use of aquatic organisms has been pointed as an integrative target sample because they are extensively exposed to hydrophobic contaminants including polycyclic aromatic hydrocarbons (PAHs) [3,4], polychlorinated biphenyls (PCBs) [5], polybrominated diphenyl ethers (PBDEs) [5–8], organochlorine pesticides (OCPs) [8–10], organophosporous pesticides (OPPs) [11] and musk fragances [12]. However, the

∗ Corresponding author. Tel.: +34 94 601 55 51; fax: +34 94 601 35 00. E-mail address: [email protected] (H. Ziarrusta). http://dx.doi.org/10.1016/j.chroma.2015.02.072 0021-9673/© 2015 Elsevier B.V. All rights reserved.

high fat content that provides the bioaccumulation potential of lipophilic pollutants is one of the major challenges in the analysis of pollutants in biota [13]. In this scenario, where the number of pollutants to be monitorised is getting larger, multi-residual analytical methodologies become an excellent strategy to get this goal [14,15]. In this context, gas chromatography (GC) coupled to single mass spectrometry (MS) [7,16] and, specially, GC coupled to tandem mass spectrometry (MS/MS) [1,17–19] has been the technique of choice to analyse semivolatile and volatile organic compounds, due to its favourable combination of high selectivity and resolution, good accuracy and precision, wide dynamic concentration range and high sensitivity [20]. The use of large volume injection (LVI) in a programmable temperature vaporiser (PTV) can render better limits of detection, but few studies have used this set-up when molluscs have to be analysed [21]. Notwithstanding the advantages of such instrumental techniques, the whole analytical process can be wasted if an unsuitable sample preparation is employed [22]. In the biomonitorisation of coastal environments many extraction techniques have been applied to mollusc samples, such as the conventional Soxhlet extraction [23,24] or sonication-assisted extraction [25]

H. Ziarrusta et al. / J. Chromatogr. A 1391 (2015) 18–30

which have been replaced by supercritical fluid extraction [26], microwave assisted extraction [27,28], pressurised liquid extraction [16,29,30] or focused ultrasound solid–liquid extraction [31]. In order to solve some operational drawbacks, minimise time, solvent consumption and cost shown by some of the mentioned extraction techniques, matrix solid-phase dispersion (MSPD) has been introduced during the last years [32]. MSPD has been reported as a reliable and simple procedure to extract organic pollutants from several solid samples [4,7,8,18,33–36]. The key factors for the success of MSPD are its feasibility, flexibility, versatility, low cost and rapidity [37,38]. Furthermore, MSPD extraction can isolate and preconcentrate the target compounds and clean up the extract in-line in a single step, which eases the method throughput. Therefore, this work was focused on the development of a MSPD extraction method followed by GC–MS or GC–MS/MS analysis to determine up to 41 non-polar or slightly non-polar organic pollutants (PAHs, PCBs, PBDEs, OCPs, OPPs and musk fragrances) in molluscs. The suitability of both GC–MS and GC–MS/MS for the multiscreening of such contaminants was also assessed previous to its validation and application to mollusc samples.

19

acquired from Supelco (Bellefonte, PA, USA) and Rutherford Vintage (Ruthe, Portugal), respectively. 6 mL and 12 mL polyethylene frits were purchased from Supelco. For dispersion and clean-up purposes, different solid sorbents were used. Diatomaceous earth (acid washed not further calcined, 95% purity), Florisil and silica (high purity grade, 70–230 mesh) were provided by Sigma–Aldrich. Envi-Carb (Supelclean 120/400) was acquired from Supelco, zeolite from Zeolyst International (Conshohocken, USA) and Plexa and octadecyl-functionalized silica (Bondesil-C18) from Agilent Technologies (Lake Forest, USA). Silica and Florisil were activated at 130 ◦ C overnight and maintained in a dry atmosphere until their use for analysis purposes. When necessary, both phases were deactivated with controlled percentages of sulphuric acid (H2 SO4 ; 95–97%) purchased from Merck (Darmstadt, Germany). A Cryodos-50 laboratory freeze-dryer from Telstar Instrument (Sant Cugat del Valles, Barcelona, Spain) was used to freeze-dry the mollusc samples. Extracted fractions were evaporated at 20 ◦ C in a Turbovap LV Evaporator (Zymark, Hopkinton, MA, USA) using a gentle N2 (99.999%, Messer, Vilaseca, Spain) blowdown. 2.2. Material cleaning

2. Experimental 2.1. Reagents and materials The names of the target analytes and the isotopically labelled standards used as surrogates, the abbreviations and the purity of the standards are included in Table 1. PCB Mix-3 (CB 28, CB 52, CB 101, CB 118, CB 138, CB 153 and CB 180) was supplied by Dr. Ehrenstorfer GmbH (Augsburg, Germany) and the individual standard of CB 52 by Sigma–Aldrich (St. Louis, MO, USA). Bromodiphenyl Ethers Lake Michigan Study mix (BDE 28, BDE 47, BDE 66, BDE 85, BDE 99, BDE 100, BDE 138, BDE 153 and BDE 154) was purchased from Isostandards Materials (Madrid, Spain) and the individual standard of BDE 100 from Sigma–Aldrich (St. Louis, MO, USA). SS TCL PAH Mix containing EPAs 16 priority PAHs was obtained from Supelco (Walton-on-Thames, UK). The surrogate standard PAH deut 5 containing 5 deuterated PAHs ([2 H8 ]-naphthalene, [2 H10 ]-acenaphthene, [2 H10 ]-phenanthrene, [2 H12 ]-chrysene and [2 H12 ]-perylene) was provided by Dr. Ehrenstorfer GmbH. The two polycyclic musks, tonalide (AHTN) and galaxolide (HHCB), were obtained from LGC Standards GmbH (Wesel, Germany), whereas [2 H15 ]-musk xylene was supplied by Dr. Ehrenstorfer GmbH. The four OCPs (o,p -dichlorodiphenyldichloroethane, p,p -dichlorodiphenyl-dichloroethylene, o,p -dichlorodiphenyl trichloro-ethane, p,p -dichlorodiphenyl-trichloroethane) and four HCH isomers were supplied by Dr. Ehrenstorfer GmbH. The two OPPs, chlorpyriphos and chlorfenvinphos, as well as the deuterated [2 H8 ]-1,1,1-trichloro-2,2-bis(4-chlorophenyl)ethane, analogue were provided by Sigma–Aldrich. [2 H66 ]-n-dotriacontane and [2 H46 ]-n-docosane were acquired from CDS Isotopes Inc. (Sainte-Foy-La-Grande, France). Individual stock solutions from each solid standard were dissolved to prepare ∼1000 ␮g g−1 stock solutions in 2-propanol (HPLC-grade, 99.8%, LabScan, Dublin, Ireland). These solutions were stored in amber vials at −20 ◦ C. 100 mg L−1 dilutions were prepared in 2-propanol monthly and more diluted stocks were prepared daily according to the experimentation. The solvents n-hexane (95%), dichloromethane (DCM; 99.8%), ethyl acetate (EtOAc; 99.8%) and acetone (99.8%) used for elution were provided by LabScan. Empty polypropylene cartridges (10 mL capacity) were purchased from HSW Norm-Jet (Keltenstrasse, Germany), BD Discardit II (Huesca, Spain) and Omnifix-F (B. Braun Melsungen AG, Germany). Empty glass syringes of 6 mL and 10 mL capacity were

All the laboratory material was cleaned with abundant pure water ( 159.0 (20; 25)

314.0 > 258.0 (10; 30) 323.1 > 267.0 (10;10)

181.0 > 145.0 (20; 40) 181.0 > 145.0 (20; 40) 181.0 > 145.0 (20; 40) 181.0 > 145.0 (20; 70) 246.3 > 176.1 (35; 70)

MSPD–GC–MS/MS

Instrumental

Procedural

Instrumental

Procedural

LOD (ng mL−1 )

LOQ (ng mL−1 )

LODproc (ng g−1 )

LOD (ng mL−1 )

LOQ (ng mL−1 )

LODproc (ng g−1 )

13

26

22

1.0

2.0

2.7

nd

nd

nd

nd

nd

nd

219.0 > 183.0 (15; 20) 219.0 > 183.0 (15; 20) 219.0 > 183.0 (15; 20) 219.0 > 183.0 (15; 30) 318.3 > 248.1 (20; 30)

0.64

0.83

18

0.012

0.018

0.26

1.0

1.8

68

0.010

0.017

0.19

0.86

1.5

19

0.16

0.18

0.27

1.1

2.0

7.7

0.067

0.12

0.18

0.74

2.7

2.1

0.37

0.68

0.83

20.44

199/314

97.3

22.81

267/269

12.64

181/183

14.92

181/183

14.18

181/183

15.42

181/183

25.49

246/248

25.87

235/237

235.1 > 165.1 (22; 70)

237.0 > 165.1 (20; 30)

0.87

4.2

2.2

0.010

0.018

0.74

27.61

235/237

235.0 > 165.1 (20; 70)

237.3 > 165.1 (22; 30)

1.0

1.9

7.1

0.024

0.026

0.32

29.18

235/237

235.3 > 165.1 (20; 20)

237.0 > 165.1 (20; 10)

2.2

8.0

2.9

1.1

2.0

0.94

53.5*

16.65

243/258

29

14

0.89

1.5

6.3

16.98

243/258

243.2 > 171.2 (20; 10) 243.2 > 187.5 (20; 10)

8.8

97.9

243.3 > 213.2 (22; 20) 243.3 > 159.2 (22; 20)

10

22

29

0.69

1.1

4.1

99.0

17.13

256/186

1.2

8.2

0.34

0.58

3.0

19.01

220/292

0.86

1.3

7.9

0.43

0.73

2.4

99.0

23.88

326/256

258.0 > 186.1 (40; 10) 292.1 > 222.1 (22; 30) 325.9 > 254.0 (26; 30)

0.68

99.0

256.1 > 186.1 (35; 20) 292.1 > 220.0 (24; 70) 326.0 > 256.0 (22; 70)

1.0

1.9

6.9

0.37

0.62

3.4

99.5

27.07

326/256

326.0 > 254.0 (40; 40)

326.0 > 256.0 (40; 10)

1.1

2.2

2.5

0.20

0.30

1.1

99.5

28.13

360/290

360.0 > 290.0 (35; 70)

360.0 > 288.0 (35; 30)

1.1

1.9

5.0

0.28

0.45

2.1

H. Ziarrusta et al. / J. Chromatogr. A 1391 (2015) 18–30

Polychlorinated byphenyls CB 28 2,4,4 -Trichlorobiphenyl CB 52 2,2 ,5,5 -Tetrachlorobiphenyl CB 101 2,2 ,4,5,5 Pentachlorobiphenyl CB 118 2,3 ,4,4 ,5Pentachlorobiphenyl CB 138 2,2 ,3,4,4 ,5 Hexaachlorobiphenyl

SIM

99.0

Organochlorine pesticides ␣-HCH 98 ␣-Hexachlorocyclohexane ␤-HCH 100 ␤-Hexachlorocyclohexane ␥-HCH 98.6 ␥-Hexachlorocyclohexane ␦-HCH 99 ␦-Hexachlorocyclohexane 4,4 -DDE 99.7 p,p -Dichlorodiphenyldichloroethylene 99.5 2,4 -DDD o,p Dichlorodiphenyldichloroethane 99.5 2,4 -DDT o,p -Dichlorodiphenyltrichloro-ethane 99.0 4,4 -DDT p,p -Dichlorodiphenyltrichloroethane Polycyclic musks HHCB Galaxolide AHTN Tonalide

tR (min)

CB 153 2,2 ,4,4 ,5,5 Hexachlorobiphenyl CB 180 2,2 ,3,4,4 ,5,5 Heptachlorobiphenyl

Polycyclic aromatic hydrocarbons Acy Acenaphthylene Ace Acenaphthene Flu Fluorene Phe Phenanthrene

29.33

360/290

360.0 > 290.0 (33; 20)

360.0 > 288.0 (33; 10)

1.2

2.4

4.5

0.24

0.41

1.9

99.5

32.07

394

393.9 > 322.0 (33; 40)

393.9 > 323.8 (33;10)

1.0

2.7

1.1

0.17

0.18

0.40

97.9

27.08

406/408

405.9 > 246.1 (20; 40)

407.8 > 248.0 (25; 10)

7.2

14

8.0

0.016

0.030

0.24

32.12

486/484

486.0 > 326.0 (30; 40)

488.0 > 327.9 (30; 10)

1.3

5.9

8.7

0.30

0.51

0.77

97.0

32.54

486/484

486.0 > 326.0 (25; 40)

484.0 > 325.6 (25; 10)

0.94

7.1

4.6

0.013

0.022

0.44

97.3

33.79

404/564

403.9 > 296.7 (50; 40)

566.0 > 406.0 (30; 20)

7.5

16

15

0.030

0.052

0.76

98.5

34.14

404/564

404.0 > 297.0 (50; 70)

566.0 > 405.7 (15; 30)

6.6

14

20

0.18

0.28

6.9

34.85

404/564

403.7 > 297.0 (40; 30)

564.0 > 403.8 (10; 10)

8.5

14

17

1.1

1.7

3.5

97.0

35.23

484/644

484.0 > 375.0 (50; 35)

644.0 > 483.1 (10; 10)

8.2

18

22

0.84

1.3

2.8

99.3

35.82

484/644

484.0 > 323.9 (50; 35)

644.0 > 483.1 (10; 10)

13

20

25

2.7

3.9

8.1

99.9

7.77

152/153

1.5

3.1

0.16

0.23

0.89

8.33

153/154

0.86

1.3

7.6

0.40

0.56

3.3

98.5

10.11

165/166

0.92

1.4

35

0.29

0.45

10

99.1

14.58

178/179

152.1 > 126.1 (35; 30) 154.1 > 152.1 (25; 20) 166.1 > 164.1 (28; 30) 178.1 > 152.1 (48; 10)

0.93

99.9

152.1 > 150.1 (40; 70) 154.1 > 153.1 (25; 40) 166.1 > 165.1 (28; 70) 178.1 > 176.1 (48; 40)

1.1

2.1

44

0.21

0.39

35

100

100

H. Ziarrusta et al. / J. Chromatogr. A 1391 (2015) 18–30

Polybrominated diphenyl ethers BDE 28 2,4,4 -Tribromodiphenyl ether BDE 47 2,2 ,4,4 -Tetrabromodiphenyl ether BDE 66 2,3 ,4,4 -Tetrabromodiphenyl ether BDE 100 2,2 ,4,4 ,5Pentabromodiphenyl ether BDE 99 2,2 ,3,4,4 Pentabromodiphenyl ether BDE 85 2,2 ,4,4 ,6Pentabromodiphenyl ether BDE 154 2,2 ,4,4 ,5,6 Hexabromodiphenyl ether BDE 153 2,2 ,4,4 ,5,5 Hexabromodiphenyl ether

99.5

21

22

Table 1 (Continued) Compound

Isotopically labelled compounds [2 H10 ]-Ace [2 H10 ]-Acenaphthene [2 H10 ]-Phe [2 H10 ]-Phenanthrene [2 H10 ]-Chr [2 H10 ]-Chrysene [2 H12 ]-Per [2 H12 ]-Perylene [2 H15 ]-MX [2 H15 ]-musk xylene [2 H8 ]-4,4 -DDT 1,1,1-Trichloro-2,2-bis(4chlorophenyl)ethane-d8 [2 H46 ]-n-docosane [2 H66 ]-n-dotriacontane *

tR (min)

SIM

SRM

Quantifier/ Qualifier

Quantifier (collision energy, eV; dwell, ms)

Qualifier (collision energy, eV; dwell, ms)

178.1 > 176.1 (48; 40) 202.2 > 200.1 (52; 25) 202.2 > 200.1 (52; 25) 228.2 > 226.2 (52; 70) 228.2 > 226.2 (52; 70) 252.2 > 250.1 (52; 40) 252.2 > 250.1 (52; 40) 252.2 > 250.1 (46; 30) 276.2 > 274.2 (56; 40) 278.2 > 276.2 (50; 40) 276.2 > 274.2 (56; 40)

99.5

14.83

178/179

99.5

21.99

202/203

97.5

23.38

202/203

99.9

30.86

228/229

98.4

31.06

228/229

98.1

34.04

252/253

99.9

34.09

252/253

99.9

34.59

252/253

99.9

36.89

276/277

99.9

36.97

276/277

99.6

37.48

276/277

98

8.23

164

98

14.48

188

98

30.93

243

98

34.7

264

100

16.58

295/313

98

29.08

243

98

25.24

66

98

36.3

66

Galaxolide HHCB contains approx. 25% of diethylphtalate; nd: not determined.

MSPD–GC–MS

MSPD–GC–MS/MS

Instrumental

Procedural

Instrumental

Procedural

LOD (ng mL−1 )

LOQ (ng mL−1 )

LODproc (ng g−1 )

LOD (ng mL−1 )

LOQ (ng mL−1 )

LODproc (ng g−1 )

178.1 > 152.1 (48; 10) 202.2 > 150.1 (52; 10) 202.2 > 151.1 (52; 10) 228.2 > 225.1 (52; 30) 228.2 > 225.1 (52; 30) 252.2 > 224.1 (52; 10) 252.2 > 224.1 (52; 10) 252.2 > 224.1 (46; 10) 276.2 > 274.1 (54; 10) 278.2 > 276.1 (40; 10) 276.2 > 274.1 (54; 10)

1.3

2.5

9.1

0.14

0.16

5.2

1.5

3.2

34

0.14

0.24

19

1.6

3.4

17

0.16

0.29

8.7

0.34

5.1

6.6

0.19

0.23

0.18

2.8

3.5

5.4

0.42

0.56

1.2

6.5

13

14

0.22

0.28

7.5

2.9

7.6

19

0.099

0.17

3.3

0.40

10

6.8

0.28

0.42

0.37

0.22

0.31

0.40

0.38

9.6

8.4

0.29

0.42

0.41

164.2 > 162.2 (20; 40) 188 > 188 (0; –)

–

nd

nd

nd

nd

nd

nd

–

nd

nd

nd

nd

nd

nd

240.3 > 240.3 (0; –) 264.2 > 264.2 (0; –) 295.0 > 277.0 (12; 20) 243.1 > 173.2 (20; 20)

–

nd

nd

nd

nd

nd

nd

–

nd

nd

nd

nd

nd

nd

294.2 > 276.0 (12; 10) 246.1 > 174.2 (20; 10)

nd

nd

nd

nd

nd

nd

nd

nd

nd

nd

nd

nd

–

nd

nd

nd

nd

nd

nd

–

nd

nd

nd

nd

nd

nd

66.8 > 66 (20; 20) 66.8 > 66 (20; 20)

H. Ziarrusta et al. / J. Chromatogr. A 1391 (2015) 18–30

Ant Anthracene Flr Fluoranthene Pyr Pyrene B[a]A Benzo[a]anthracene Chr Chrysene B[b]F Benzo[b]fluoranthene B[k]F Benzo[k]fluoranthene B[a]P Benzo[a]pyrene Ind Indene[1,2,3-cd]pyrene D[ah]A Dibenzo[a,h]anthracene B[ghi]P Benzo[g,h,i]perylene

Purity (%)

H. Ziarrusta et al. / J. Chromatogr. A 1391 (2015) 18–30

were spiked at 50 ng g−1 in the blended material and a second frit was placed over. Subsequently, analytes were eluted with 25 mL of DCM (measured at collection) and the eluate obtained was evaporated to dryness using a gentle N2 blowdown and reconstituted to a final volume of 140 ␮L of n-hexane. The extracts were analysed by means of GC–MS or GC–MS/MS. 2.5. GC–MS and GC–MS/MS analysis The MSPD extracts were analysed in an Agilent 7890A gas chromatograph coupled to an Agilent 7000 triple quadrupole mass spectrometer and using an Agilent 7693 autosampler (Agilent Technologies). 2 ␮L of the extract was injected in the splitless mode (1.5 min) at 300 ◦ C in an Agilent HP-5 ms capillary column (30 m × 0.25 mm, 0.25 ␮m) with helium (99.9995%, Carburos Metálicos, Barcelona, Spain) as carrier gas at a constant flow of 1.3 mL min−1 . The following oven temperature programme was used for the separation of the 41 target compounds: 60 ◦ C (held 1 min) to 140 ◦ C at 30 ◦ C min−1 , to 200 ◦ C at 3 ◦ C min−1 , to 240 ◦ C at 5 ◦ C min−1 and a final increase of 30 ◦ C min−1 up to 300 ◦ C, where it was held for 10 min. When LVI-PTV injection was considered, 10 ␮L of the sample extract was injected at 50 ◦ C while the vent valve was opened for 3 min at a flow rate of 75 mL min−1 and a vent pressure of 2.9 psi. Subsequently, the analytes were focused to the column in splitless mode for 1.5 min while the temperature of the PTV injection port was increased at 12 ◦ C min−1 to 300 ◦ C and held for 5 min. Finally, the inlet was further cleaned at a purge flow of 50 mL min−1 before further injections. The mass spectrometer worked in the electron impact (EI) mode with electron energy of 70 eV. The temperature of the interface was kept at 310 ◦ C, while the temperature of the ionisation source and the detector were maintained at 230 ◦ C and 150 ◦ C, respectively. The measurements were performed both in the selected ion monitoring (SIM) and in the selected reaction monitoring (SRM) modes (see Table 1). In the latter, N2 (99.9999%; Air Liquide, Spain) was required as collision gas at a flow of 1.5 mL min−1 . In the case of the SIM mode, the first ion was used as quantifier while the second ion was considered as qualifier. In the case of the SRM mode, one precursor ion and two product ions (one used as quantifier and the other one as qualifier) were monitored, as recommended by the identification requirements of the European guidelines (Commission Decision, 2002/657/EC and SANCO/10684/2009 Guideline). The sum of the dwell times within a window was maintained in 100 ms and longer dwell times were used for the target analytes compared to the corresponding labelled standards in order to improve the signal of the former (see Table 1). The MassHunter WorkStation Acquisition Software (Version B.05.02/Build 5.2.365.0, Agilent Technologies, 2008) was used for data acquisition and automatic integration and quantification of the results. 3. Results and discussion 3.1. Optimisation of the MSPD procedure Among all the parameters that can affect the MSPD extraction procedure the nature of the solid support, the nature of the sorbent used for sample clean-up, the use of sulphuric acid in the clean-up step, the dimensions and material of the MSPD column, the nature and volume of the elution solvent and the sample amount were studied. Sample extracts obtained after MSPD procedure during optimisation were analysed by GC–MS. 3.1.1. Selection of the spiking procedure Previous to method optimisation, the fortification of the unpolluted mollusc samples was considered. In a first approach, slurries

23

of the freeze-dried mussel tissues were prepared using different organic solvents (acetone, n-hexane and 2-propanol) but all the solvents induced matrix modification. Consequently, the optimisation assays were carried out using two types of mollusc matrices: a naturally polluted LRM [39] and spiked fish hatchery mussels. Although spiking is not always the best approach to simulate polluted samples, it has already been successfully used in the literature [7] when fortification via slurry is not viable. Two different approaches are followed in the literature when spiked samples are used for MSPD optimisation purposes: spiking in the mortar [4,40] and spiking onto the blended sample (spiking in the syringe) [7]. Both strategies were checked, and comparable results (p-value > 0.05) were obtained for all the target compounds. Hence, spiking in the syringe was chosen for simplicity and practical reasons. The samples were spiked using external stock solutions of each compound in n-hexane (i.e., PAHs, OPPs, OCPs and musk compounds), except for PCBs and PBDEs for which CB 52 and BDE 100 were used during the optimisation. 3.1.2. MSPD syringes Firstly, both polypropylene (HSW Norm-Jet, BD Discardit II and Omnifix-F) and glass syringes were considered as MSPD column support. With this purpose, 0.30 g of freeze-dried mussel sample and 0.30 g of C18 solid support were homogenised in a glass mortar. A syringe, containing a polyethylene frit at the bottom, was filled with 0.30 g of silica followed by 1.00 g of activated silica and 1.00 g of Florisil (deactivated with 5% (w/w) Milli-Q water), and the sample/solid support mixture was placed over the clean-up sorbents. The analytes were recovered with 12 mL of n-hexane:DCM mixture (25:75, v/v), the extracts were concentrated to dryness using a gentle N2 blowdown and reconstituted in 140 ␮L with n-hexane for GC–MS analysis. Although plastic syringes are often used in the literature [4,7,33,41] because they are disposable, dirty chromatograms, with mainly hydrocarbon interferences, were obtained. The dirtiness made impossible the quantification of target analytes no matter which plastic material was used. Therefore, glass syringes were used in upcoming experiments, which were not discarded but properly cleaned after use as indicated in Section 2.2. 3.1.3. Selection of the solid support Different solid support materials, including reverse-phase (C18, Plexa, Envi-Carb), normal phase (Florisil, silica) and inert (diatomaceous earth and zeolite) materials, were considered. In these experiments, 0.30 g of freeze-dried mussel tissue and 0.30 g of solid support were homogenised in a glass mortar. A 6-mL glass syringe, containing a polyethylene frit at the bottom, was filled with 0.30 g silica followed by 1.00 g of activated silica and 1.00 g of Florisil (deactivated with 5% (w/w) Milli-Q water), and the sample/solid support mixture was placed over the clean-up sorbents. Samples were spiked at 0.50 ␮g g−1 in the syringe and a second frit was placed over (n = 3). The compounds were eluted with 12 mL of a n-hexane:DCM (25:75, v/v) mixture and concentrated in 140 ␮L of n-hexane to be analysed by GC–MS. Fig. 1 compares the recoveries of the target analytes obtained after the dispersion of the spiked mussel samples with different materials. The results for a compound per each family are only shown because comparable results were obtained for the rest of the compounds belonging to the same family. Overall, extraction recoveries higher than 100% were obtained except for the most volatile ones, indicative of a positive matrix effect regardless of the solid support used. In the case of chlorfenvinphos, the results were not meaningful due to the weak elution solvent used in this set of experiments. Anyhow, several conclusions could be withdrawn. On the one hand, Envi-Carb was discarded since the high molecular weight PAHs (i.e., Flr, Pyr, B[a]A, Chr, B[b]F, B[k]F, B[a]P, D[ah]A,

24

H. Ziarrusta et al. / J. Chromatogr. A 1391 (2015) 18–30

Fig. 1. Absolute recoveries and standard deviations obtained with MSPD extraction using different solid supports for the most representative target compounds (n = 3). The recoveries were calculated by comparison with a standard solution at 1 ␮g mL−1 after subtraction of the blank signal.

B[ghi]P and Ind) could not be recovered. On the contrary, much higher absolute recoveries were obtained when reverse-phase solid supports (C18 and Plexa) were used, mainly attributed to their efficiency to disperse other interfering compounds from the matrix, such as hydrocarbon compounds and fatty acids. Consequently, despite of the higher efficiencies obtained with these materials, reverse phases were discarded. Among the rest of the dispersants, since statistically comparable results were obtained for both inert and normal-phase solid supports, Florisil was chosen. The results obtained were consistent with other published studies focused on the analysis of molluscs by MSPD, where Florisil [5,29] or diatomaceous earth [33] were used but other works using C18 have been also published [8]. The importance of the first disruption and dispersion step in MSPD has been reported in the literature based on scanningelectron microscopy results [32]. However, from the results above, it was not very clear whether it aided the extraction of the target analytes or was only significant for the retention of interferences of the matrix. In order to ensure this key point, a LRM [39] was processed with and without this first blending step. In terms of recovery, both approaches rendered comparable (according to the analysis of variance, test Fcalc = 3.38 < Fcrit = 3.95) results and, therefore, it could be assumed that extraction of the analytes did not occur during blending but during the column elution step. Further research should be carried out in order to verify this hypothesis. However, the blending step was maintained in terms of sample homogenisation during dispersion and trapping of interferences. 3.1.4. Optimisation of the clean-up step The positive matrix effect observed was induced by the high percentage of fat in the extract. Therefore, and based on the literature [3,5,7,8,18,41–43], three different strategies were studied in order to minimise this matrix effect: (i) selection of the clean-up sorbent; (ii) study of the sulphuric acid effect and; (iii) optimisation of the amount of sorbent. In the first step, several combinations of normal-phase (activated Florisil and activated silica) and reverse-phase (C18 and Envi-Carb) clean-up sorbents were assayed in triplicate. The different assays consisted in the combination of the so called lower and upper clean-up sorbents. From the literature [4,7,8,33,34,44], it was concluded that normal-phase sorbents work better for fatty matrices, so the lower sorbent was fixed in the normal-phase. Hence, the considered sorbent combinations were the following: activated Florisil, activated silica, Envi-Carb and C18 as upper clean-up sorbents, and as lower sorbent activated silica for all the cases. Concurrently, another couple of tests were performed fixing activated Florisil as lower and upper sorbent. With this purpose, a 6-mL glass syringe was filled with 0.30 g of silica followed by 1.00 g of lower

sorbent (silica) and 1.00 g of the corresponding upper sorbent. Then, the sample/Florisil dispersed mixture (0.30 g: 0.30 g) was placed over the clean-up sorbents. Thereupon, analytes were collected in 12 mL of a n-hexane:DCM mixture (25:75, v/v), concentrated in 140 ␮L of n-hexane and analysed by GC–MS. In this experiment, as the objective was to reduce the signal enhancement by reducing the fat content, samples were spiked after the elution, but before the evaporation step. Absolute recoveries were calculated by comparison with a standard solution at 1 ␮g mL−1 after subtraction of the blank signal (non-spiked freeze-dried mussel sample). According to the results obtained (see Fig. 1S in the supplementary material), none of the sorbent combinations (i.e., combinations of silica with florisil, C18 and Envicarb as well as pure silica and florisil) tested was able to eliminate completely the existing matrix effect. On the one hand, when C18 and Envi-Carb were used as top clean-up layer, large interfering chromatographic peaks, which prevent the quantification of HCHs, were observed (data not shown). Therefore, the reverse-phase sorbents were discarded in order to avoid the introduction of dirtiness in the chromatographic system. On the other hand, the use of Florisil as lower and upper sorbent rendered higher absolute recoveries (around 200% for the late eluting analytes) in comparison with the rest of combinations. Indeed, the chromatograms obtained in the scan mode for the former extracts showed a high presence of interfering peaks at the end of the chromatogram as well as enhanced baseline increase; thus, the use of this set-up was discarded. Since similar matrix effect was observed for the rest of normal phase sorbent combinations, the repeatability, in terms of relative standard deviation (RSD), was taken into account to fit it. While the average RSDs were 16% for silica-Florisil (lower–upper), a 5% was obtained for silica–silica (lower–upper) combination. This better repeatability using activated silica could be attributed to a lower fat content in the final extract, which showed the cleanest chromatograms and robust retention times. Thus, it was selected for further assays. Fatty acids and hydrocarbons were still the main interferences in the analysis and thus, a destructive lipid removal method compatible with MSPD was tested [3,5,7,18,42]. This approach consists on an oxidative dehydration of the lipidic fraction with sulphuric acid, which is easily introduced in the MSPD procedure by impregnation of the silica sorbent with sulphuric acid. Before assessing the efficiency of this second strategy, the stability of the analytes in the presence of a strong acid, some experiments were carried out considering different percentages of H2 SO4 (i.e., 0%, 5%, 10% and 20% (w/w)). All the assays were performed using the previously selected optimum clean-up sorbent (activated silica) and without changing any of the conditions previously established. The lower clean-up sorbent was only impregnated with

H. Ziarrusta et al. / J. Chromatogr. A 1391 (2015) 18–30

25

Fig. 2. Absolute recoveries and standard deviations for non-stable analytes in the presence of different percentages of sulphuric acid (n = 3).

H2 SO4 (acidified silica), which guaranteed that the matrix was not in direct contact with the acid. In this set of experiments, samples were spiked (500 ng g−1 for all the target analytes, except for chlorpyriphos and chlorfenvinphos, which were spiked at 5 ␮g g−1 ) before and after extraction, in order to determinate the matrix effect during the extraction and detection steps, respectively. Nonspiked samples were also processed in order to subtract the blank signals. Since some of the target analytes (chlorfenvinphos and musks) were not eluted quantitatively in the previous experiments, this time, the elution was carried out collecting two consecutive fractions of 12 mL of n-hexane:DCM (25:75, v/v) and 10 mL of pure DCM. Overall, cleaner extracts were obtained when the percentage of sulphuric acid in the clean-up phase increased, which was supported also with chromatographic analyses performed in the scan mode (the higher the acid percentage in the silica sorbent, the lower the chromatogram baseline and interferences). However, two conclusions could be drawn using the results obtained for samples spiked before MSPD extraction (see Fig. 2). On the one hand, in the presence of sulphuric acid at percentages of 10% or higher, a decrease in the recoveries was observed for certain PAHs (Acy, Ace, Flu, Ant, Pyr, B[a]A, Ind, D[ah]A and B[ghi]P) and musks (AHTN and HHCB) (see Fig. 2), probably due to compound decomposition [3]. On the other hand, this behaviour was not observed for less chemically labile analytes, such as CB 52, BDE 100, 4,4 DDE, 2,4-DDD, 2,4-DDT and some PAHs (Phe, Flr, Chr and B[b]F, B[k]F). Thus, although the use of sulphuric acid rendered cleaner extracts, it was discarded due to the degradation of part of the target analytes. Regarding the results obtained from the samples spiked after extraction, chromatographic signal enhancement with recoveries over 100% was still observed (data not shown) even in the presence of sulphuric acid. The third clean-up strategy consisted on the increase of the amount of clean-up sorbent. In the next set of experiments the amount of silica was doubled so, a 10-mL glass syringe was filled with 0.60 g of silica followed by 4.00 g of activated silica and the sample/solid support mixture (0.30 g:0.30 g) was placed over the clean-up sorbent. As the MSPD column volume was increased, the elution volume was also doubled accordingly and the elution was carried out collecting two consecutive fractions of 24 mL of a (25:75, v/v) n-hexane:DCM mixture and 20 mL of pure DCM that after evaporation were analysed separately by means of GC–MS. Using the large scale clean-up the fat content in the extract was lower, and thus, the absolute recoveries were closer to 100%, at least for the early eluting analytes, whereas a positive matrix effect was still observed for those compounds (the heavier ones) that

co-eluted with the remaining fat. Hence, large scale clean-up was selected as the best option. 3.1.5. Optimisation of the elution step In all the previous experiments, elution solvents were limited to n-hexane:DCM mixtures and pure DCM because they are known to be compatible with the use of sulphuric acid [45]. However, since the use of sulphuric acid was discarded, five different elution solvents and mixtures were tested: DCM, DCM:EtOAc (90:10, v/v), DCM:EtOAc (80:20, v/v), EtOAc and acetone. The solvents were chosen based on the literature [41,46–48] and taking into account their eluotropic strength to eluate quantitatively the target analytes, while trying to minimise the co-extraction of interferences. Elution profiles were performed using the different solvents under the previous MSPD extraction conditions (0.30 g of mussel tissue spiked at 500 ng g−1 was dispersed with 0.30 g of Florisil and loaded onto 4.00 g of activated silica). 6 consecutive 5-mL fractions of the solvents were collected in separate vials to get the elution profile corresponding to a total volume of 30 mL. All the fractions were evaporated to dryness, reconstituted in 140 ␮L of n-hexane and analysed separately by means of GC–MS. Although the use of a stronger elution solvent, such as acetone or EtOAc, allowed the quantitative elution of all the compounds with minimum solvent volume (see Fig. S2 in the supplementary material), very dirty extracts were obtained. On the contrary, weaker organic solvents provided the cleanest extracts but larger elution volume was necessary (see Fig. S2). Consequently, two different alternatives were assessed in parallel in order to get compromise conditions to obtain fat free extracts but using the minimum solvent volume: (i) a small volume of a strong solvent, such as 10 mL of DCM:EtOAc (80:20, v/v) and (ii) a large volume of a weak solvent, such as 25 mL of pure DCM. When using a solvent with higher eluotropic strength, a high amount of fat that prevented the quantification of last eluting compounds was still introduced to the chromatographic system, even when using small elution volumes. Hence, although chlorfenvinphos was not recovered, elution was limited to 25 mL of DCM. Under these conditions, the rest of the 40 target analytes were quantitatively recovered except AHTN, but it was still included in the validation of the method. 3.1.6. Selection of the sample amount In order to obtain better limits of detection, extraction of higher sample amount was lastly considered. To this end, 0.30 g and 0.45 g were tested and the results showed that, although higher chromatographic signals were obtained using higher sample amount, a much noisier chromatogram was obtained because more co-eluting

26

H. Ziarrusta et al. / J. Chromatogr. A 1391 (2015) 18–30

interferences were also extracted. Thus, in order to prolong the lifetime of the chromatographic column, 0.30 g of mussel sample were chosen as optimum. 3.2. Method validation At the validation step, besides GC–MS, the use of GC–MS/MS was also considered with the purpose of increasing selectivity and sensitivity. All analyses were performed following the transitions detailed in Table 1 and in duplicate, using both detection techniques. Regarding SRM, optimum precursor ions, product ions and collision energies were optimised using a standard solution of each target analyte at 1 ␮g mL−1 in n-hexane (see Table 1). Furthermore, apart from CB 52 and BDE 100 used in the optimisation, method validation was carried out considering 7 PCBs (CB 28, CB 52, CB 101, CB 118, CB 138, CB 153 and CB 180) and 8 PBDEs (BDE 28, BDE 47, BDE 66, BDE 100, BDE 99, BDE 85, BDE 154, BDE 153). 3.2.1. Linearity and limits of detection Calibration curves were built with standard solutions in nhexane and were analysed by GC–MS and GC–MS/MS at 8 concentration levels in the calibration range of 0.60–200 ng mL−1 for all the target analytes, except for 4,4 -DDT (1.8–600 ng mL−1 ), HHCB and AHTN (6.0–2000 ng mL−1 ) and BDE 154, BDE 153, Ind and D[ah]A (4.0–200 ng mL−1 ). Good linearity was found out over the wide range of the tested concentrations and the determination coefficients (r2 ) were between 0.980 and 0.9996 for all the target analytes in both approaches. Instrumental limits of detection (LODs) and quantification (LOQs) were estimated and defined as the average response (n = 3) of the lowest concentration level of the calibration curve for each analyte plus three and ten times the standard deviation [49], respectively (see Table 1). Slightly better results were obtained for GC–MS/MS. In order to calculate the procedural limits of detection (LODproc ) procedural blanks (i.e., the whole MSPD procedure performed without the mussel tissue) were analysed (n = 6) and LODproc were estimated as the average response for each analyte plus three times the standard deviation (see Table 1). Other approaches for the determination of the limits of the method that include the presence of the matrix (i.e. the definition suggested by the US EPA, http://www.epa.gov/waterscience/methods/det/rad.pdf) were not possible due to the presence of the target analytes at very different concentration ranges in real mussel tissues, even in fish hatchery mussels. Significantly better procedural LODproc values were obtained for GC–MS/MS (see Table 1). This improvement in sensitivity was mainly attributed to the selectivity of the GC–MS/MS technique. Similar results were reported by Sánchez-Avila et al., who analysed different non-polar organic pollutant families by SPE–GC–MS/MS in mussels [1]. The sensitivity obtained using GC–MS/MS was good enough to reach the EQSs values for biota adopted by the Directive 2013/39/EU in the case of Flr (30 ng g−1 ) and B[a]P (5 ng g−1 ). In the case of PBDEs, the established value is much lower (0.0085 ng g−1 ) and, although quantification would not be possible by this analytical method, the procedure developed is good enough for a first screening of the target analytes. In order to increase further on the sensitivity of this method, the LVI-PTV injection was considered. However, when increasing the injection volume from 2 ␮L to 10 ␮L the chromatographic system got very dirty (see the chromatograms for n-hexane injections performed after the injection in the splitless injector and the PTV injector in Fig. S3). Consequently, the LVI-PTV was no further studied and the whole validation was performed using 2 ␮L splitless injection.

3.2.2. Recoveries and apparent recoveries Different strategies were tried to assess the accuracy of the method. On the one hand, fish hatchery mussels spiked at 50 ng g−1 were analysed. On the other hand, two reference materials (SRM2977 and LRM mussel tissue [39]) were analysed. In the case of fish hatchery mussels spiked at 50 ng g−1 , nonspiked samples were processed in parallel. The results obtained with the analysis of both GC–MS and GC–MS/MS (see the chromatograms in Fig. S4 in the supplementary material), with (apparent recovery) and without (recovery) correction with the corresponding labelled surrogate, are included in Table 2. The separation of Ind and D[ah]A was not possible by GC–MS so the results are expressed as the sum of both. Without any correction, matrix enhancement was observed for most of the target analytes no matter the analysis was performed by GC–MS or GC–MS/MS (see Table 2). The high recoveries obtained also using GC–MS/MS indicates that the signal enhancement due to matrix effect does not occur in the detection, but may occur in the injection port of the chromatographic system [50] or during the extraction step. This phenomenon of signal enhancement has been also observed in other works in the literature for some compounds determined in solid samples [50]. The use of labelled surrogates was in this work, however, capable of correcting this matrix effect (see Table 2). For AHTN, which was not quantitatively recovered, none of the isotopically labelled standard worked, not even with [2 H15 ]MX. Although its analogue [2 H3 ]-AHTN is commercially available, its use was discarded due to the proton exchange that gives the original undeuterated product forging the accuracy of the results [51]. Therefore, for this analyte, the concentrations found in real mussel samples were corrected applying the recovery factor obtained. In order to assess the developed method with non-spiked mussel samples, the method was also applied to SRM 2977 reference material. Due to the lower concentration expected in the SRM 2977 material, analysis was only performed by GC–MS/MS. During the processing of the SRM 2977 material, notable differences in the method performance were observed, i.e. the elution step was much slower (150 min compared to 45 min for the rest of the samples), a white powder appeared after sample evaporation that was not present in any of the other samples processed and the evaporation step also took longer (90 min compared to 50 min for the rest of the samples). These differences could be related to physical differences between SRM 2977 and the rest of the freeze-dried and homogenised materials processed in the present work. While particle size of SRM 2977 was below 50 ␮m, larger particle sizes were obtained during the sample preparation of the rest of the samples (50–500 ␮m). Consequently, the packaging of the SRM 2977 was better than those obtained with the rest of samples, the cartridge was blocked and the elution process was much slower. Table 3 summarises the recoveries and apparent recoveries obtained for the extraction in triplicate of the SRM 2977. Good apparent recoveries were obtained in most of the cases, after subtraction of procedural blanks and corrected with deuterated surrogates, except for 4,4 -DDT and PAHs such as, Ace, Flu, Phe, Chr, B[a]P, D[ah]A, B[ghi]P and Ind, whose matrix effect could not be compensated by the use of surrogates. In the case of 4,4 -DDT, the poor recovery results obtained could be attributed to a possible degradation of the target analyte [52]. With the aim to check why unsatisfactory apparent recoveries were obtained for certain PAHs, two different validation approaches were carried out. On the one hand, standard additions were performed over the extract solutions of SRM 2977 after the reconstitution step. This quantification strategy is often applied in the literature to overcome matrix effect problems, especially for the quantification of PAHs using this extraction method [4]. To this aim, the final extract was divided in two aliquots of 60 ␮L each and

H. Ziarrusta et al. / J. Chromatogr. A 1391 (2015) 18–30

27

Table 2 Recoveries and apparent recoveries for GC–MS and GC–MS/MS, calculated with the MSPD extraction of fish hatchery mussels spiked at 50 ng g−1 . Repeatability (n = 4) and reproducibility (n = 8) for MSPD coupled to GC–MS/MS. Compound

MSPD–GC–MS

MSPD–GC–MS/MS

Recovery (%)

Recovery (%)

Precision

Apparent recovery

Absolute recovery

Apparent recovery

Organophosphorous compounds 185 Chlorpyrifos

100b

177

104b

4

14

Organochlorine pesticides 116 ␣-HCH 79 ␤-HCH 137 ␥-HCH ␦-HCH 134 2,4 -DDD 201 4,4 -DDE 205 251 2,4 -DDT 261 4,4 -DDT

69b 57b 81b 80b 113d 115d 88e 92e

140 160 142 162 204 213 236 247

82b 94b 83b 95b 116d 122d 91e 95e

5 4 5 3 6 7 3 1

6 4 6 6 7 10 5 4

6 16

11 18

Absolute recovery

Polycyclic musks HHCB AHTN

116 45

Polychlorinated byphenyls 161 CB 28 160 CB 52 CB 101 176 224 CB 118 CB 138 195 206 CB 153 212 CB 180 Polybrominated diphenyl ethers BDE 28 242 263 BDE 47 158 BDE 66 204 BDE 100 BDE 99 224 178 BDE 85 159 BDE 154 BDE 153 185 Polycyclic aromatic hydrocarbons 32 Acy Ace 38 Flu 79 Phe 153 Ant 190 Flr 266 Pyr 145 B[a]A 330 289 Chr 292 B[b]F + B[k]F 274 B[a]P 325 Ind D[ah]A 287 B[ghi]P * a b c d e f g

* *

123 45

* *

Repeatability (%RSD)

Reproducibility (%RSD)

96b 95b 105b 79e 68e 72e 74e

160 164 186 223 207 205 204

94b 96b 109b 86e 80e 79e 79e

3 3 3 4 3 3 7

21 24 10 10 9 7 10

86c 92c 55c 71f 79f 62f 56f 65f

234 236 204 198 198 172 162 181

89c 93c 81c 78f 78f 68f 64f 71f

3 5 4 4 5 12 15 17

18 9 7 8 11 29 28 33

69a 82a

36 39 90 175 202 264 259 296 271 266 280 264 260 238

73a 80a

2 4 11 16 5 11 4 1 5 4 2 8 12 14

30 21 13 19 7 27 22 7 5 14 9 12 12 18

*

91b 113b 95c 87c 117c 103c 102f 96f 92g 81g

*

102b 119b 100c 98c 112c 102c 105f 111f 82g 81g 74g

No surrogate correction needed. Corrected with [2 H10 ]-Ace. Corrected with[2 H10 ]-Phe. Corrected with [2 H10 ]-Chr. Corrected with [2 H15 ]-MX. Corrected with [2 H8 ]—4,4 -DDT. Corrected with [2 H12 ]-Per-d. Corrected with [2 H46 ]-n-dotriacontane.

the target analytes were added at 20 ng g−1 in one of the aliquots. On the other hand, the MSPD method developed in the present work was applied to the mussel LRM available in the laboratory where PAHs were analysed by means of MAE followed by a Florisil clean-up and GC–MS analysis [39]. In the case of the results obtained by means of standard additions on SRM 2977, satisfactory results were obtained for B[a]P, D[ah]A, B[ghi]P and Ind, with recoveries in the 50–94% range, while recoveries higher than 100% were still obtained for Ace, Flu, Phe

and Chr. The second alternative to evaluate the suitability of the developed method for the analysis of PAHs rendered better results as can be observed in Table 4. In fact, the apparent recoveries obtained after the correction of the concentrations with the corresponding labelled surrogate were in the range of 68–149% for all the PAHs. Since the validation for PAHs was satisfactory for both spiked mussel samples and the LRM and only odd results were obtained for SRM 2977, it could be concluded that MSPD is very dependent on the physical state of the sample. Further research on

28

H. Ziarrusta et al. / J. Chromatogr. A 1391 (2015) 18–30

Table 3 Absolute and apparent recoveries for SRM-2977. Compound

Recovery (%)

Certified conc. at 95% of confidence level (ng g−1 )

Organochlorine pesticides 2,4 -DDD 4,4 -DDE 4,4 -DDT

3.15 ± 0.25 11.8 ± 1.2 1.32 ± 0.16

Polychlrorinated byphenyls CB 28 CB 52 CB 101 CB 118 CB 138 CB 153 CB 180

5.17 8.02 10.6 10.0 7.94 14.1 6.32

± ± ± ± ± ± ±

0.36 0.56 0.9 0.41 0.63 1.3 0.72

Absolute recovery

Apparent recovery

87 66 35

92d 71d 13e

230 193 142 212 272 170 185

166b 139b 102b 80e 102e 64e 69

Polybrominated diphenyl ethers BDE 28 BDE 47 BDE 66 BDE 100 BDE 99 BDE 154 BDE 153

2.54 ± 0.40 36.5 ± 4.0 0.453 ± 0.046 1.82 ± 0.64* 4.68 ± 0.92* 0.20 ± 0.09 0.16 ± 0.04*

274 355 338 324 335