Fp

IAD

a

ARR2AA

KPFVFAM

1

ihihbsoflth

0h

Journal of Chromatography A, 1331 (2014) 27– 37

Contents lists available at ScienceDirect

Journal of Chromatography A

jou rn al hom epage: www.elsev ier .com/ locate /chroma

ocused ultrasound solid–liquid extraction for the determination oferfluorinated compounds in fish, vegetables and amended soil

tsaso Zabaleta, Ekhine Bizkarguenaga, Arantza Iparragirre, Patricia Navarro,ilette Prieto, Luis Ángel Fernández, Olatz Zuloaga ∗

epartment of Analytical Chemistry, Faculty of Science and Technology, University of the Basque Country (UPV/EHU), P.O. Box 644, 48080 Bilbao, Spain

r t i c l e i n f o

rticle history:eceived 23 October 2013eceived in revised form3 December 2013ccepted 12 January 2014vailable online 22 January 2014

eywords:erfluorinated compoundsocused ultrasound solid–liquid extractionegetablesishmended soilatrix effect

a b s t r a c t

In the present work a method was developed for the determination of different perfluorinated compounds(PFCs), including three perfluorinated sulfonic acids (PFSAs), seven perfluorocarboxylic acids (PFCAs),three perfluorophosphonic acids (PFPAs) and perfluorooctanesulfonamide (PFOSA) in fish, vegetables andamended soil samples based on focused ultrasound solid–liquid extraction (FUSLE) followed by solid-phase extraction (SPE) clean-up and liquid chromatography–tandem mass spectrometry (LC–MS/MS).Different variables affecting the chromatographic separation (column type and pH of the mobile phase),the electrospray ionization (capillary voltage, nebulizer pressure and drying gas flow) and mass spec-trometric detection (fragmentor voltage and collision energy) were optimized in order to improve thesensitivity of the separation and detection steps. In the case of FUSLE variables such as the solvent type,the solvent volume, the extraction temperature, the sonication and extraction time and the percent-age of applied irradiation power were studied. Under optimized conditions, sonication of 2.5 min withpulse times on of 0.8 s and pulse times off of 0.2 s in 7 mL of (9:1) acetonitrile (ACN): water mixture induplicate guaranteed exhaustive extraction of the matrices analyzed. Due to the non-selective extractionusing FUSLE, different SPE cartridges (200-mg Waters Oasis-HLB, 150-mg Waters Oasis-WAX and 150-mgWaters Oasis-MAX) were tested in terms of extraction efficiency and matrix effect both in the extractionand detection steps. Mix mode SPE using Waters Oasis-WAX provided the best extraction efficiencies

with the lowest matrix effect. The final method was validated in terms of recovery at two fortificationlevels (in the 80–120% for most of the analytes and matrices), precision (relative standard deviation inthe 2–15% range) and method detection limits (MDLs, 0.3–12.4 ng/g for vegetables, 0.2–12.5 ng/g for fishand 1–22 ng/g for amended soil). Finally the method was applied for the determination of the 14 PFCsin different vegetables and fish samples from a local supermarket and in a soil amended with a compostfrom a local wastewater treatment plant (WWTP).

. Introduction

An emerging contaminant is a chemical or a material thats characterized by a perceived, potential or real threat touman health or the environment. Among the different emerg-

ng compounds defined in the recent years (pharmaceutics, certainormones, etc.), per- and polyfluorinated compounds (PFCs) haveecome of emerging concern due to their potential toxicity, per-istence and bioaccumulation [1]. PFCs represent a large group ofrganic compounds that are characterized by a fully or partially

uorinated hydrophobic and lipophobic carbon chain attachedo one or more different hydrophilic functional groups [1]. Theydrophilic end group can be neutral, or positively or negatively

∗ Corresponding author. Tel.: +34 946013269; fax: +34 946013500.E-mail address: olatz.zuloaga@ehu.es (O. Zuloaga).

021-9673/$ – see front matter © 2014 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.chroma.2014.01.025

© 2014 Elsevier B.V. All rights reserved.

charged. The resulting compounds are non-ionic, cationic or anionicsurface active agents due to their amphiphilic character [2]. Thehighly chemical and biological stability of PFCs is conferred by thecarbon–fluorine bond. This is a covalent bond (one of the strongestfound in organic chemistry) and is resistant to hydrolysis, photoly-sis, metabolism, and biodegradation [3]. This resistance confers toPFCs rigidity, low chemical reactivity and environmentally persis-tence; therefore, they have the potential to be bioaccumulative.

PFCs are widely used due to their special properties, such aschemical and thermal stability, acid resistance and water, dirt andgrease repellency [4]. Among the principal applications they canbe used as surface protectors in carpets, leather, cookware, sportsclothing, paper, food containers, fabric and upholstery and as per-

formance chemicals in products such as fire-fighting foams, floorpolishes, shampoos, paints and inks [5,6]. Furthermore, PFCs arealso used in industrial applications as surfactants, emulsifiers, wet-ting agents, additives and coatings [7].

2 matog

tafaecawaafi0Arc2tdcpi[iw1mo2Pao

g

apcbfHvrsMl

Ktpaapior

cascwrvBa

8 I. Zabaleta et al. / J. Chro

Due to the growing concern about this class of chemicals, inhe year 2000 the largest producer of PFCs, the 3M Company,nnounced the phase out of the production of perfluorooctanesul-onic acid (PFOS). Since then, new shorter-chained PFCs (C4–C7)nd their precursors are being introduced as replacements consid-ring that these are less persistent or toxic in humans [8]. However,ontinued manufacturing of PFC precursors may result in furtherccumulation of PFOS and other PFC residues to the environment,ildlife, and humans [4,9]. In 2004, Environment Canada initi-

ted a temporary ban on fluoropolymers containing fluorotelomerlcohols (FTOHs) [9] and Norway banned the use of PFOS in fireghting foams, textiles and impregnation agents (max. content.005%) [6]. Moreover, in 2006 the US Environmental Protectiongency (US EPA) announced a voluntary stewardship program toeduce by 95% perfluorooctanoic acid (PFOA) and related chemi-als in the environment by 2010 and to eliminate all of them by015 [7]. Furthermore, the European Union (EU) issued a Direc-ive that prohibited from June 2008 the general use of PFOS anderivates [10]. In May 2009, PFOS was listed as “restricted use”ompounds under the Stockholm Convention on persistent organicollutants (POPs) [11]. However, PFOA and the homologous chem-

cals of PFOS, which may degrade to PFOS, are not regulated yet12]. Finally, PFCs have been announced as emerging contaminantsn the food chain by the European Food Safety Authority (EFSA),

hich have recently established the tolerable daily intakes (TDI) of50 ng/kg/day for PFOS and 1500 ng/kg/day for PFOA [13]. Further-ore EFSA recommended that an additional monitoring focused

n PFCs is needed. On this account, Commission Recommendation010/161/EU invited the Member States to monitor the presence ofFOS and PFOA, different chain length (C4–C15) PFCs similar to PFOSnd PFOA, and their precursors, in order to estimate the relevancef their presence in food [14].

Due to the concern on exposure to PFCs, a special interest hasrown to develop robust analytical methods in the last years [15].

According to the literature [3,13–26], three main approachesre used for the extraction of PFCs from solid samples: (i) ion-airing, (ii) alkaline digestion and (iii) the use of an organic solventombined with an energy source. In the case of ion-pairing, tetra-utylammoium (TBA) is used as ion-pair reagent, while the neutralorms generated are extracted into methyl tert-butyl ether (MTBE).owever, the robustness of this extraction is questionable, sinceery variable recoveries are obtained, that vary from the lowecoveries (<50%) obtained for short- or long-chain PFCs to thetrong matrix effects that causes high recoveries (>200%) [18].atrix effect can be related to the co-extraction of lipids and other

ipophilic matrix components [14].Another extraction strategy is the alkaline digestion using

OH/methanol (MeOH) or NaOH/MeOH. The use of alkaline diges-ion helps to extract bound PFC residues by removing lipids androteins before extraction. Different studies [13,16] presented

comparison of the two extraction methods (ion-pairing andlkaline digestion) and both reported that the alkaline digestionrovided three-to-five higher concentration levels of several PFCs

n liver samples than ion pairing. They attributed these differencesf concentrations to the effective digestion of the matrix and theelease of these compounds from the sample.

An alternative to the use of ion-pairing and alkaline digestiononsists on the use of different organic solvents, together withqueous solutions in many cases, in combination with an energyource, such as heating or pressurization. For instance Llorca ando-workers [13] used water and pressurized liquid extraction (PLE)ith this purpose. The recoveries obtained were in the >85–89%

ange for liver and muscle samples. This extraction method pro-ided better recoveries than alkaline digestion and ion-pairing.esides, PLE was much more rapid than the alkaline digestionnd provided cleaner extracts than that based on ion pairing. In

r. A 1331 (2014) 27– 37

addition focused ultrasound solid–liquid extraction (FUSLE) usingACN has been recently applied for the determination of PFCs insewage sludge. This method provided faster analysis than the otherpreviously reported for the determination of these compounds insewage sludge. Recovery values between 69 and 104% were foundfor the target analytes [24]. However this extraction techniquehas not been reported in food samples yet. In addition, Tittlemieret al. [17] used centrifugation with MeOH as extraction tech-nique. The recoveries for food samples including fish were generallygreater than 80%. However, the mean recoveries of the longer chainperfluorocarboxylates were between 64 and 74%. Other authors[19] employed extraction with ACN/water combined with ultra-sonication for vegetables, meat and fish samples. The recoveriesranged between 59 and 98% for all the analytes.

The extraction methods described above usually need a clean-upstep. This is the case of alkaline digestion or most extraction per-formed by an organic solvent combined with an energy source. Themost usual clean-up process is solid phase extraction (SPE), whichrepresents the option for isolation and/or pre-concentration of PFCsin biotic samples. Widely used cartridges are WAX (mix-modeweak anion exchanger), MAX (mix-mode strong anion exchanger)and HLB (hydrophilic–lipophilic balanced sorbents). A compari-son of a variety of SPE adsorbents such as WAX, MAX and WCX(mix-mode weak cation exchanger) after ultrasonic extraction wasperformed for mono-, di- and tri-substituted polyfluoroalkyl phos-phates (mono-PAPs, di-PAPs and tri-PAPs) and PFPAs in sewagesludge. WAX provided the most satisfactory recoveries for mostanalytes with recoveries ranging from 71 to 119% [27]. In the workby Ullah et al. [19] a mix mode SPE containing a quaternary aminewas used for the determination of PFCAs, PFSAs and PFPAs in veg-etables, meat, and fish samples. The recoveries ranged between 59and 98% for all target analytes [19].

The drawbacks of the use of GC–MS (need for derivatization)have made LC coupled to MS the most widely used technique forthe analysis of PFCs. The most common MS instrumental set-upused for PFC analysis is the triple-quadrupole mass spectrometer(QqQ), which is one of the best suited for quantification of PFCs.Nowadays, the performance of ion trap (IT), quadrupole-linear iontrap (QqLIT), and time of flight (TOF) have also been exploited fortrace quantification of PFCs [28,29].

The aim of the present work was to develop a method for theaccurate and precise determination of four families of PFCs, includ-ing perfluorinated carboxylic acids (PFCAs), perfluorinated sulfonicacids (PFSAs), perfluorinated phosphonic acids (PFPAs) and perflu-orooctanesulfonamide (PFOSA) in food samples such as vegetables(lettuce, pepper and carrot) and fish, as well as in amended-soilused for the growing of different crops. FUSLE was tested for theextraction step, while different clean-up approaches of the extractsusing different SPE cartridges (reverse and mix-mode) were tested.Matrix effect was thoroughly studied both in the clean-up andLC–MS/MS (triple quadrupole) analysis steps.

2. Experimental

2.1. Reagents and materials

The names of the target analytes, the abbreviations, the chemicalstructure, the supplier of the standards, the purity of the standards,the octanol–water partition coefficient (as log Kow) and pKa valuesare included in Table 1. In the case of the surrogate standards theinformation has been included in Table 2.

Stock solution for PFOS, PFOA and PFOSA were dissolved indi-vidually in MeOH in order to prepare approximately 5000 mg/Lsolutions. 100 mg/L dilutions were prepared in MeOH everymonth and dilutions at lower concentrations were prepared daily.

I. Zabaleta et al. / J. Chromatogr. A 1331 (2014) 27– 37 29

Table 1Structures, suppliers, purity, log Kow and pKa values of the target analytes.

Analyte Abbreviation Structure Supplier Purity % Log Kow pKa

Potassiumperfluoro-1-butanesulfonate

L-PFBS Wellington >98 2.41d 0.14b

Sodiumperfluoro-1-hexanesulfonate

L-PFHxS Wellington >98 4.34d 0.14b

Potassiumperfluoro-1-octanesulfonate

L-PFOS Wellington >98 6.28d 0.14b

Perfluorooctylphosphonicacid

PFOPA Wellington >98 5.85c 2.4/4.5c

Perfluorohexylphosphonicacid

PFHxPA Wellington >98 3.55c 2.1/4.4c

Perfluorodecylphosphonicacid

PFDPA Wellington >98 8.27c 3.4/5.6c

Perfluoro-n-butanoicacid

PFBA Wellington >98 2.43d 0.2-0.4a

Perfluoro-n-pentanoicacid

PFPeA Wellington >98 3.40d 0.5a

Perfluoro-n-hexanoicacid

PFHxA Wellington >98 4.37d 0.9a

Perfluoro-n-heptanoicacid

PFHpA Wellington >98 5.33d –

30 I. Zabaleta et al. / J. Chromatogr. A 1331 (2014) 27– 37

Table 1 (Continued)

Analyte Abbreviation Structure Supplier Purity % Log Kow pKa

Perfluoro-n-octanoicacid

PFOA Wellington >98 6.3d 2.8a

Perfluoro-n-nonanoicacid

PFNA Wellington >98 7.27d 2.57e

Perfluoro-n-decanoicacid

PFDA Wellington >98 7.90d 2.6a

Perfluorosulfonamide PFOSA Dr. Ehren-storfer(Germany)

97.5 7.58d 6.52b

a [30].b [31].

Pp(a1(wp[ap[dMo(rMi

TS

c [32].d [33].e [34].

erfluorobutanoic acid (PFBA), perfluoropentanoic acid (PFPeA),erfluorohexanoic acid (PFHxA), perfluoroheptanoic acidPFHpA), perfluorooctanoic acid (PFOA), perfluorononanoiccid (PFNA), perfluorodecanoic acid (PFDA), potassium perfluoro--butanesulfonate (L-PFBS), sodium perfluoro-1-hexanesulfonateL-PFHxS) and potassium perfluoro-1-octanesulfonate (L-PFOS)ere obtained at 5 mg/L in MeOH and the surrogate mix (sodiumerfluoro-1-hexane [18O2] sulfonate, MPFHxS, sodium perfluoro-1-1,2,3,4-13C4] octanesulfonate, MPFOS, perfluoro-n-[13C4] butanoiccid, MPFBA, perfluoro-n-[1,2-13C2] hexanoic acid, MPFHxA,erfluoro-n-[1,2,3,4-13C4] octanoic acid, MPFOA, perfluoro-n-1,2,3,4,5-13C5] nonanoic acid, MPFNA, perfluoro-n-[1,2-13C2]ecanoic acid, MPFDA, perfluoro-n-[1,2-13C2] undecanoic acid,PFUdA, perfluoro-n-[1,2-13C2] dodecanoic acid, MPFDoA) was

btained at 2 mg/L in MeOH. Perfluorooctanephosphonic acidPFOPA), perfluorohexanephosphonic acid (PFHxPA) and perfluo-odecane phosphonic acid (PFDPA) were obtained individually in

eOH at 50 mg/L. All the chemicals standards were stored at 4 ◦C

n the dark and the stock solutions were stored at −20 ◦C.

able 2upplier, abbreviations and purities for surrogates standards, as well as which target ana

Surrogate Abbreviation

Sodium perfluoro-1-hexane [18O2] sulfonate MPFHxS

Sodium perfluoro-1-[1,2,3,4-13C4] octanesulfonate MPFOS

Perfluoro-n-[13C4] butanoic acid MPFBA

Perfluoro-n-[1,2-13C2] hexanoic acid MPFHxA

Perfluoro-n-[1,2,3,4-13C4] octanoic acid MPFOA

Perfluoro-n-[1,2,3,4,5-13C5] nonanoic acid MPFNA

Perfluoro-n-[1,2-13C2] decanoic acid MPFDA

Perfluoro-n-[1,2-13C2] undecanoic acid MPFUdA

Perfluoro-n-[1,2-13C2] dodecanoic acid MPFDoA

MeOH (HPLC grade, 99.9%) and acetone (HPLC grade, 99.8%)were supplied by LabScan (Dublin, Ireland), acetonitrile (ACN,HPLC grade, 99.9%) by Sigma–Aldrich (Steinheim, Germany), aceticacid (HOAc, 100%), hydrochloric acid (HCl, 36%) and potassiumhydroxide (KOH, 85%) by Merck (Darmstadt, Germany), formic acid(HCOOH, 98–100%) by Scharlau (Barcelona, Spain) and ammoniumhydroxide (NH4OH, 25%) by Panreac (Barcelona, Spain). Ultra-purewater was obtained using a Milli-Q water purification system(<0.05 �S/cm, Milli-Q model 185, Millipore, Bedford, MA, USA).

Waters Oasis-HLB (poly(divinylbenzene-co-N-vinylpirrilidonepolymer, 200 mg), Waters Oasis-MAX (poly(divinylbenzene-co-N-vinylpirrilidone + quaternary amine polymer, 150 mg) andWaters Oasis-WAX (poly(divinylbenzene-co-N-vinylpirrilidone +secondary amine polymer, 150 mg) SPE cartridges were purchasedfrom Waters Corporation (Milford, USA).

For the mobile phase composition MeOH and ACN (Romil-UpS, Waterbeach, Cambridge, UK) were used. 1-Methyl-piperidine

(1-MP, ≥98%) was obtained from Merck and ammonium acetate(NH4OAc) was purchased from Sigma–Aldrich.

lyte is corrected with each isotopic analog.

Corrected compounds Purity % Supplier

L-PFHxS, L-PFBS >98% WellingtonL-PFOS, PFOPA >98% WellingtonPFBA >98% WellingtonPFHxA, PFHpA, PFHxPA >98% WellingtonPFOA, PFPeA >98% WellingtonPFNA, PFOSA >98% WellingtonPFDA >98% Wellington– >98% Wellington– >98% Wellington

matog

(tuEmEotae

c

2e

toaa

wsMampwseT

2

2

mtacooptssga

2

mo2olvwuvaLt

phosphonate and sulfonamide derivatives of PFCs using a mobile

I. Zabaleta et al. / J. Chro

A Cryodos-50 laboratory freeze-dryer from Telstar InstrumentSant Cugat del Valles, Barcelona, Spain) was used to freeze-dryhe samples. For extraction, a Bandelin Sonopuls HD 3100 sonifierltrasonic cell disruptor/homogenizer (100 W, 20 kHz; Bandelinlectronic, Berlin, Germany) equipped with a 3-mm titaniumicrotip was used. Fractions were evaporated in a Turbovap LV

vaporator (Zymark, Hopkinton, MA, USA) using a gentle streamf nitrogen. After the extraction step, the supernatant was filteredhrough PTFE filters (0.45 �m, 25 mm, Macherey-Nagel, Germany)nd nylon microfilters (0.2 �m, 13 mm, Pall, USA) were used to filterxtracts before LC–MS/MS analysis.

Fish (hake, prawn and tuna) samples and vegetables (lettuce,arrot and pepper) were obtained from a local market.

.2. Sample treatment and focused ultrasound solid–liquidxtraction

Vegetable and fish samples were frozen and freeze-dried beforehe extraction step. For optimization experiments a known amountf matrix was weight, covered with acetone, spiked with targetnalytes and stirred during 24 h. After that, acetone was evaporatednd the sample was aged for one week.

Under optimal conditions 0.5 g of sample were placed togetherith 7 mL of an ACN: Milli Q water (9:1) mixture in a 40-mL ves-

el and surrogate standards (MPFHxS, MPFOS, MPFBA, MPFHxA,PFOA, MPFNA, MPFDA, MPFUdA, MPFDoA) were added (25 �L of

0.5 ng/�L solution). The FUSLE step was performed in the pulsedode for 2.5 min in duplicate, with a pulsed time on of 0.8 s and

ulsed time off of 0.2 s and at 10% of irradiation power. Extractionsere carried out at 0 ◦C in an ice-water bath. After the extraction

tep, the supernatant was filtered through PTFE filters and FUSLExtract was evaporated to ∼1 mL under a nitrogen stream using aurbovap LV Evaporator depending on the clean-up selected.

.3. Clean-up

.3.1. Waters Oasis-HLBThis clean-up approach was a modification performed to the

ethod published by Loos et al. [35]. Briefly, the extract evaporatedo ∼1 mL was diluted in 6 mL of Milli-Q water previously adjustedt pH 1 with HCl. The 200-mg WATERS Oasis-HLB cartridges wereonditioned with 5 mL of MeOH and 5 mL of Milli Q water previ-usly adjusted at pH 1. After the sample was loaded (pH = 1), 5 mLf a (95:5) Milli-Q: MeOH mixture was added with cleaning pur-oses and the cartridges were dried for 1 h under vacuum. Then,he analytes were eluted using 8 mL of MeOH and collected in aingle vial. The eluate was concentrated to dryness under a gentletream of nitrogen at 35 ◦C and reconstituted in 250 �L of LC–MSrade MeOH. Finally, the reconstituted extract was filtered through

0.2 �m nylon filter before LC–MS/MS analysis.

.3.2. Waters Oasis-WAXThis clean-up approach was a modification performed to the

ethod published by Chu and Letcher [36]. Briefly, the extract evap-rated to ∼1 mL was diluted in 6 mL of Milli-Q water at pH 7. The00-mg Waters Oasis-WAX cartridges were conditioned with 5 mLf MeOH and 5 mL of Milli-Q water at pH 7. After the sample wasoaded, 1 mL of formic acid (2%) and 1 mL of Milli-Q:MeOH (95:5,/v) mixture were added with cleaning purposes and the cartridgesere dried for 1 h under vacuum. Then, the analytes were elutedsing 4 mL of acetone with 2.5% NH4OH and collected in a singleial. After elution, the extract was concentrated to dryness under

gentle stream of nitrogen at 35 ◦C and reconstituted in 250 �L ofC–MS grade MeOH. Finally, the reconstituted extract was filteredhrough a 0.2 �m nylon filter before the LC–MS/MS analysis.

r. A 1331 (2014) 27– 37 31

2.3.3. Waters Oasis-MAXThis clean-up approach was performed according to the stan-

dardized method published by Waters [37]. Briefly, the extractevaporated to ∼1 mL was diluted in 6 mL of Milli-Q Waters. The 150-mg Waters Oasis-MAX cartridge was conditioned with 5 mL MeOHand 5 mL water, respectively. The concentrated sample extract wasloaded, and the cartridge was rinsed with 2 mL of 5 mol/L NH4OHin 5% MeOH followed by 4 mL of MeOH. The analytes were subse-quently eluted with 8 mL of 2% formic acid in MeOH. The extractwas evaporated to dryness under nitrogen at 35 ◦C and reconsti-tuted in 250 �L of LC–MS grade MeOH. Finally, the reconstitutedextract was filtered through a 0.2 �m nylon filter before LC–MS/MSanalysis.

2.4. LC–MS/MS analysis

An Agilent 1260 series HPLC chromatograph equipped with adegasser, binary pump, autosampler and column oven coupledto an Agilent 6430 triple quadrupole (QqQ) mass spectrometerequipped with both ESI and APCI sources (Agilent Technologies,Palo Alto, CA, USA) was employed for the separation and quan-tification of PFCs. Under optimized conditions, mobile phase Aconsisted of a Water/MeOH (95:5, v/v) mixture and mobile phase Bof MeOH/water (95:5, v/v), and both contained 2 mmol/L NH4OAcand 5 mmol/L 1-MP. The gradient profile started with 90% A (holdtime 0.3 min) and continued with a linear change to 80% A up to1 min, to 50% A up to 1.5 min and to 20% A up to 5 min (hold time5 min) followed with a linear change to 0% A up to 13 min and ahold time until 16 min. Initial conditions were regained at 17 minfollowed by equilibration until 26 min. The flow rate was set at0.2 mL/min and the volume injected was 5 �L.

Two chromatographic columns were tested for analyte separa-tion. An ultra high performance liquid chromatographic (UHPLC)Agilent Zorbax Extend-C18 (2.1 mm, 50 mm, 1.8 �m) column (pHrange 2.0–11.5) and an Agilent Zorbax SB-C18 (2.1 mm, 50 mm,1.8 �m) column (pH range 1–8). In all the cases an UHPLC ZorbaxEclipse XDB-C18 pre-column (2.1 mm, 5 mm, 1.8 �m) was used. Thecolumn temperature was set to 35 ◦C for Agilent Zorbax Extend-C18column and at 40 ◦C in the case of Agilent Zorbax SB-C18 column.

Quantification was performed in the multiple reaction moni-toring (MRM) acquisition mode. Nitrogen was used as nebulizer,drying, and collision gas. ESI in negative mode was carried outusing a capillary voltage of 3000 V, a drying flow rate of 10 L/min,a nebulizer pressure of 50 psi (1 psi = 6.8948 kPa) and drying gastemperature of 350 ◦C.

Fragmentor electric voltage and collision energy were optimizedfor ESI in the 60–240 V and 5–45 eV ranges, respectively, by injec-tion of individual compounds. Optimized values are included inTable 4 (Section 3).

Instrumental operations, data acquisition and peak integrationwere performed with the Masshunter Workstation Software (Qual-itative Analysis, Version B.06.00, Agilent Technologies).

3. Results and discussion

3.1. Optimization of LC–MS/MS

3.1.1. Optimization of chromatographic column and the mobilephase

In a first approach, Zorbax SB C-18 column was tested for theseparation of up to 14 analytes, including carboxylic, sulfonate,

phase A consisting of 95:5 water: MeOH and a mobile phase B con-sisting of 95:5 MeOH: water, with 5 mmol/L ammonium acetatein both A and B. However, the chromatographic signal, especially

32 I. Zabaleta et al. / J. Chromatogr. A 1331 (2014) 27– 37

Table 3Comparison of calibration slopes to study the influence of 1-MP in the mobile phase.

Analyte With 1-MPSlope ± s(ng/mL)

Without 1-MPSlope ± s(ng/mL)

PFOS 182.56 ± 19.23 158.94 ± 2.06PFOA 25.54 ± 1.32 8.54 ± 0.10

oogaltMctwTfcPP

tco5wFo

3

d[i

Fig. 1. Chromatograms of a 25 ng/g fortified carrot sample extracted by FUSLE andOasis WAX clean-up. (1) PFBA, (2) PFHxPA, (3) PFPeA, (4) PFBS, (5) PFHxA, (6) PFOPA,

TPr

PFOSA 47.25 ± 4.30 66.19 ± 0.72PFOPA 4.96 ± 0.09 1.33 ± 0.06

f phosphonated PFCs, was very poor. According to the resultsbtained by Ullah et al. [38], 1-MP can improve the chromato-raphic behavior of PFCs since 1-MP behaves as an ion-pairinggent that masks the negative charges of the phosphonate group,eading to an increase in the retention on a C-18 stationary phasehrough hydrophobic interactions. In order to test the use of 1-

P in the mobile phase, the chromatographic column had to behanged since a chromatographic column able to support pHs upo 11 was necessary. In this sense Zorbax Extend-C18 columnhich stands pHs up to 11.5 was chosen. As can be observed in

able 3 for the calibrations curves (see calibration ranges in Table 4)or PFOPA, PFOS, PFOSA and PFOA, the addition of 1-MP signifi-antly improved the slope of the calibration curve for PFOPA andFOA and, in a less extent, of PFOS. No improve was observed forFOSA.

Furthermore, different compositions of the mobile phase con-aining MeOH, ACN and water were tested. Mobile phase Aonsisting of 95:5 water: MeOH and mobile phase B consistingf 95: 5 MeOH: water with 2 mmol/L ammonium acetate and

mmol/L 1-MP in both A and B was selected since, when ACNas added, the sensitivity obtained was worse (data not shown).

ig. 1 shows a chromatogram for a spiked carrot sample (25 ng/g)btained under optimized conditions.

.1.2. Optimization of the electrospray ionization

According to the literature [15], ESI has been mostly used for the

etermination of PFCs using LC–MS. Only in the case of Esparza et al.32], APCI showed better sensitivity when PFPAs and PFOS werenvestigated, but since the simultaneous determination of up to 14

able 4recursor and product ions (first ion was used as quantifier and the second as qualifier) atanges, the correlation coefficients, the instrumental LODs and LOQs for target analytes.

Analytes Precursor ion Product ion Fragmentor (V) Collisionenergy (eV)

PFBA 213 169 60 5

PFHxPA 399 79 100 10

PFPeA 263 219/175 60 5

PFBS 299 99/80 100 30

PFHxA 313 269/119 60 5

PFOPA 499 79 150 20

PFHpA 363 319/169 60 10

PFHxS 399 99/80 150 20

PFOA 413 369/169 60 5

PFDPA 599 79 100 5

PFNA 463 419/169 60 5

PFOS 499 99/80 150 45

PFOSA 498 498/78 220 5

PFDA 513 469/269 100 5

MPFBA 217 172 60 5MPFHxA 315 270 60 5MPFHxS 403 103 150 30MPFOA 417 372 60 5MPFOS 503 99 60 45MPFNA 468 423 60 5MPFDA 515 470 100 5MPFUdA 565 520 60 5MPFDoA 615 570 100 5

(7) PFHpA, (8) PFHxS, (9) PFOA, (10) PFDPA, (11) PFNA, (12) PFOS, (13) PFOSA, (14)PFDA.

PFCs was aimed in the present work, only ESI was optimized. Duringoptimization of ESI PFOS, PFOA, PFOSA and PFOPA were studied.Three variables were studied: the capillary voltage (3–6 kV), thenebulizer pressure (30–50 psi) and the drying gas (nitrogen) flow(8–12 L/min). Drying gas temperature was fixed at 350 ◦C accordingto the manufacturer.

A central composite design (CCD) was built using the Statgraph-ics program (Statgraphics centurion XV). The CCD consisted of a 23

factorial design with a six star points located at ± from the cen-ter of the experimental domain and three replicates of the centralpoint. An axial distance of 1.68 was selected in order to guaranteethe rotatability (data not shown).

Fig. 2(a–d) showed the response surfaces obtained using onlythe significant (p < 0.05) parameters. As it can be observed inFig. 2(a) for PFOA, capillary voltage had a negative effect and asimilar behavior was observed for PFOS and PFOSA (Figs.2(c) and(d), respectively), except for PFOPA, which showed no effect forthis parameter (Fig. 2(b)). According to these results, the capillary

voltage was fixed at 3000 V for the rest of experiments.

The drying gas flow was significant only for PFOA and PFOSA (seeFigs. 2(a) and (d), respectively). While PFOA showed the highest

optimum fragmentor (V) and collision energy (eV) values, as well as the calibration

Calibrationrange (ng/mL)

Correlationcoefficient

LOD (ng/mL) LOQ (ng/mL)

3.7–207 0.993 2.29 3.731.7–207 0.995 1.28 1.734.9–207 0.995 2.32 4.924.0–207 0.995 2.01 3.993.3–207 0.996 1.97 3.332.6–207 0.996 1.54 2.583.5–207 0.994 1.94 3.472.7–207 0.993 1.47 2.734.2–207 0.995 2.47 4.222.5–207 0.992 1.41 2.465.7–179 0.992 2.47 5.650.7–194 0.994 0.46 0.734.1–179 0.994 1.91 4.093.6–179 0.978 1.81 3.61

I. Zabaleta et al. / J. Chromatogr. A 1331 (2014) 27– 37 33

F psi, (bg 10 L/

rtm

PtsA

cl

icvofaa

lTamp

3i

uMTwrcmecab2

hake and carrot were extracted with 7 mL of the different solventsmentioned above for 2.5 min. Figs. 3(a and b) show the responsesobtained (normalized to the highest signal). In the case of hake (seeFig. 3(a)), the responses obtained were significantly higher when

ig. 2. Response surfaces for (a) PFOA when the nebulizer pressure was fixed at 50as flow was fixed at 10 L/min and (d) PFOSA when the drying gas flow was fixed at

esponses at a low value of this parameter, 8 L/min, PFOSA showedhe highest responses at a high value of this parameter. An inter-

ediate value was fixed for drying gas flow, 10 L/min.Finally, the nebulizer pressure was significant for PFOSA and

FOPA (see Figs. 2(d) and (b), respectively). While PFOPA showedhe highest signals at a low value of this parameter, 30 psi, PFOSAhowed the highest signals at a high value of this parameter, 50 psi.

high value was fixed for drying gas flow, 50 psi.In summary the optimized parameters were fixed as follows:

apillary voltage at 3000 V, drying gas flow at 10 L/min and nebu-izer pressure at 50 psi.

Parameters related to the mass spectrometry were also stud-ed, thus, fragmentor voltage and collision energy were optimizedonsidering all the target analytes and surrogates. The fragmentoroltage (60, 100, 150, 220 and 240 V) was optimized in order tobtain the highest signal of the precursor ion, while minimizing itsragmentation. Optimization was performed in the MS2 Scan modend Table 4 summarizes optimum fragmentor values for each targetnalyte and surrogates.

In order to obtain the best signals for the product ions, the col-ision energy was studied in the 5–45 eV range at 5 eV increments.he most intense product ions were selected as the quantifiersnd, when possible, qualifier ions were also selected. Table 4 sum-arizes optimum collision energies, as well as, the precursor and

roduct ions for each target analyte and surrogates.

.1.3. Calibration ranges, correlation coefficients andnstrumental limits of detection

Under optimized chromatographic and mass spectrometric val-es, calibration curves were built with standard solutions (ineOH) and at 8 concentration levels (see calibration ranges in

able 4). As can be seen in Table 4, squared correlation coefficientsithout correction with the corresponding internal standard in the

ange of 0.992–0.996 were obtained, except for PFDPA, in whichase the correlation coefficient value obtained was 0.978. Instru-ental limit of detection (LODs) and quantification (LOQs) were

stimated and defined as the average response (n = 3) of the low

oncentration level of the calibration curve (1 ng/mL) plus threend ten times the standard deviation, respectively [39]. As cane observed in Table 4, the LODs and LOQs obtained were below.47 ng/mL and 5.65 ng/mL, respectively.

) PFOPA when the drying gas flow was fixed at 10 L/min, (c) PFOS when the dryingmin.

3.2. Optimization of FUSLE

For the optimization of FUSLE extraction PFOS, PFOA and PFOSAwere chosen as target analytes. In addition, hake and carrot sampleswere used during the optimization.

Six extraction solvents were tested according to the literature:MeOH, acetone, acetic acid, 9:1 MeOH: acetic acid, 9:1 ACN: Milli-Q water and 10 mmol/L KOH in MeOH. The experiments wereperformed in triplicate. Aliquots of 0.5 g (dry weight) of spiked

Fig. 3. Influence of solvent type during FUSLE extraction in (a) hake and (b) carrotsamples. Signals were normalized to the highest chromatographic response. Averageresponses (n = 3) and standard deviations were used.

3 matogr. A 1331 (2014) 27– 37

9Hwao1rbwtswMeTa

speSM

ad

eddttrtArf

nhu(vf

arei

ati

fa

ouUsitqsrt

4 I. Zabaleta et al. / J. Chro

:1 ACN: Milli-Q water mixture was used for all target analytes.owever, in the case of carrot samples (Fig. 3(b)), this evidenceas not so clear. Although 9:1 ACN: Milli-Q water mixture was

lso the best extractant for PFOSA, the same results were notbtained for PFOS and PFOA. In the case of PFOS acetone, MeOH,0 mM KOH in MeOH and 9:1 ACN Milli-Q water provided similarecoveries. In the case of PFOA 10 mM KOH in MeOH provided theest results, but statistically no difference was found if comparedith 9:1 ACN:Milli-Q water (95% of confidence level). According

o the results mentioned above, 9:1 ACN: Milli-Q water was cho-en as extraction solvent for further experiments. Similar resultsere obtained by Ullah et al. [19] for food samples. Furthermore,artínez-Moral and Tena [24] reported that ACN was the best

xtraction solvent for sewage sludge samples, while Moreta andena [26] used ethanol for the extraction of six perfluorocarboxyliccids and PFOS from packaging material.

In order to improve FUSLE extraction efficiency three extractionolvent volumes were tested: 4, 7 and 10 mL. The experiments wereerformed in triplicate. 7 mL (data not shown) provided the high-st recoveries, as well as the lowest relative standard deviations.imilar results in terms of extraction volumes were obtained byartínez-Moral and Tena [24] for sewage sludge.Extraction efficiency was also tested at room temperature and

t 0 ◦C and no significant differences were observed (95% of confi-ence level).

A CCD was carried out using Statgraphics in order to optimizextraction time (0.5–5 min), pulsed time on (0.2–0.8 s) and irra-iation power (10–56%). In pulsed sonication, extraction time isivided in different cycles. A cycle is a sum of the period of timehat pulsed time is on and the period of time that pulse is off. Inhis work cycles of 1 s were used. The CCD consisted of a 23 facto-ial design with a six star points located at ± from the center ofhe experimental domain and three replicates of the central point.n axial distance of 1.68 was selected in order to guarantee theotatability. The responses obtained were scaled in the logarithmicorm.

Fig. 4 shows the response surfaces obtained using only the sig-ificant (p < 0.1) parameters. As can be observed, the pulsed timead a positive effect for PFOS (see Fig. 4(a)), showing the highest val-es at the highest value of this parameter, 0.8 s. In the case of PFOSAsee Fig. 4(b)) the highest response was obtained at an intermediatealue of this parameter, 0.5 s. The sonication time was fixed at 0.8 sor the rest of the experiments.

The amplitude was significant for PFOS and PFOA (see Fig. 4(a)nd (c), respectively) and both analytes showed the highestesponses at a low value of this parameter, 10%. Therefore, the low-st value was chosen for amplitude, 10%. Besides, low amplitudesncrease the life of the titanium tips.

Finally, the extraction time was significant for PFOS and PFOAnd they showed the highest values at an intermediate value ofhis parameter (see Fig. 4(c) for PFOA). According to this result, anntermediate value was fixed for extraction time, 2.5 min.

In summary, optimum extraction conditions were fitted asollows: extraction time at 2.5 min, pulsed time on of 0.8 s andmplitude at 10%.

In the absence of a certified reference material (CRM) and inrder to determinate whether exhaustive extraction was carriednder optimized condition, repeated extractions were performed.p to three successive extractions were performed on the same

amples. Each experiment was carried out in triplicate. Results arencluded in Fig. 5(a) and (b) for carrot and hake, respectively. Inhe case of hake samples a unique extraction was sufficient for

uantitative extraction. In the case of carrot samples, two succes-ive extractions were necessary for quantitative extraction, whileecoveries lower than 20% were obtained in the third extraction. Ahird extraction was not considered in order to avoid increasing the

Fig. 4. Response surfaces obtained during the FUSLE optimization for (a) PFOS whenextraction time was fixed at 2.5 min, (b) PFOSA when amplitude was fixed at 10%and (c) PFOA when sonication time was fixed at 0.8 s.

solvent volume (7 mL × 3) submitted to the evaporation step. Sim-ilar results were obtained by Martínez-Moral and Tena [24] for thedetermination of this target analytes in sewage sludge where a sec-ond FUSLE step was necessary. In the case of the extraction of PFCsfrom packaging material a single FUSLE extraction was necessaryaccording to Moreta and Tena [26].

Although the CCD provided the highest responses when anintermediate value of extraction time was used (2.5 min), since suc-cessive extractions showed that a second extraction was needed fora quantitative extraction, 5 min extraction was tested. For this rea-son two consecutives extractions of 2.5 min were compared with aunique extraction of 5 min for carrot samples. As it can be observedin Fig. 6 and in concordance with the results obtained in the CCD,a single 5 min extraction did not guarantee quantitative extractionand, finally, 2 × 2.5 min extraction was chosen.

3.3. Optimization of the clean-up step

3.3.1. Extraction efficiency of the different clean-up proceduresAs mentioned in the experimental section, different clean-

up approaches were performed in order to determine the

I. Zabaleta et al. / J. Chromatog

Fig. 5. Influence of the number of repeated extractions in (a) carrot and (b) hakesamples.

Fe

sbaO

TE

ig. 6. Comparison of extraction yield at different extraction time values. (a) 1stxtraction of 2.5 min, (b) 2nd extraction of 2.5 min and (c) 5 min extraction.

uitability of each of them. The extraction efficiency was calculated

y comparing the responses obtained when the sample was spikedt 1.5 ng/�L before and after clean-up step (see Table 5). Watersasis MAX was only tested for carrot samples.

able 5fficiencies (%) for different clean-up approaches for carrot and hake samples.

Analyte Oasis HLB Oasis WAX Oasis MAX

CarrotPFOS 75 90 3PFOA 98 93 3PFOSA 59 77 33PFOPA 63 82 1

HakePFOS 88 91 –a

PFOA 92 83 –PFOSA 87 88 –PFOPA 54 98 –

a Not performed for hake samples.

r. A 1331 (2014) 27– 37 35

As shown in Table 5 Waters Oasis-HLB and Waters Oasis-WAXshowed the best efficiencies for all the target analytes. In the case ofWaters Oasis HLB a modification of the method published by Looset al. [35] was performed. In order to increase extraction efficien-cies, the analytes must be in their non-ionic form. Since PFCs arevery acidic analytes, acidification of the sample (pH = 1) was carriedout in our work compared to pH = 7 used in the referenced work.Waters Oasis WAX approach was a modification performed to themethod published by Chu and Letcher [36]. The retention mech-anism was mixed mode (both ion exchange and reverse phase),which improves retention for strong acidic compounds. While Chuet al. loaded the sample at pH = 4, different pH values (4 and 7)were tested in the present work. While comparable results wereobtained for PFOS, PFOA and PFOSA, higher extraction efficiencieswere obtained for PFOPA at pH 7 (data not shown).

In the case of Waters Oasis MAX the clean-up approach was per-formed according to the standardized method published by Waters[37], where the retention mechanism was also mixed mode (bothanion exchange and reverse phase). Recoveries lower than 3% wereobtained for all the target analytes except for PFOSA, 33% (Table 5).So, this clean-up approach was discarded. Similar results wereobtained by Liu et al. [27] for the determination of PFPAs in sewagesludge. However, Ullah et al. [19] obtained satisfactory results whena similar cartridge, CUQAX256 (C18+ quaternary amine, UnitedChemical Technologies, UCT, Bristol, PA), was used to determinateperfluoroalkyl carboxylic, sulfonic, and phosphonic acids in food.So, further studies should be carried out in order to improve theresults obtained with Waters Oasis-MAX cartridge.

3.3.2. Matrix effect for the different clean-up approachesThe extraction efficiency can be affected by the composition of

the sample matrix since high levels of matrix compounds may com-pete with the sorptive material or can lead to matrix effects duringLC–MS/MS determination due to changes of the ESI ionization effi-ciency.

Therefore, matrix effects occurring at LC–MS/MS detection wereevaluated by comparing the responses obtained for carrot and hakesamples which were spiked with 1.5 ng/�L after clean-up step anda standard solution in MeOH with the same concentration. Non-spiked blank samples were also analyzed and their response wasconsidered in matrix effects calculations. The results are included inFig. 7(a) and (b) for carrot and hake, respectively, where values closeto 100% indicate a lack of matrix effect. As shown in Fig. 7(a) and (b),only extracts cleaned up using Waters Oasis HLB showed significantmatrix effect during the detection step (signal enhancement forPFOPA and signal suppression for PFOA). Therefore, Waters Oasis-HLB clean-up was discarded from method validation.

3.4. Method validation and application to real samples

Method validation was only performed for FUSLE extractionwith a posterior clean-up with Waters Oasis WAX. Apparentrecovery, defined as the recovery obtained after correction withthe corresponding surrogate, was calculated using carrot andhake samples spiked at 12.5 ng/g and 25 ng/g and at 25 ng/gand 50 ng/g for pepper, lettuce and amended soil. Furthermore,matrix-matched calibration was also performed for carrot sampleswith samples spiked at the same concentrations [19]. Recover-ies obtained are included in Table 6. As can be observed for theresults obtained for carrot samples, matrix-matched calibrationwas unnecessary and good apparent recoveries were obtained aftercorrection of the concentrations obtained with an external calibra-

tion with the corresponding surrogates.

Apparent recoveries in the 80–120% range were obtained inmost of the cases. In the case of hake samples, PFBA was notdetected at the lowest concentration. It should be mentioned that

36 I. Zabaleta et al. / J. Chromatogr. A 1331 (2014) 27– 37

Table 6Apparent recoveries at two different levels for carrot (12.5 ng/g and 25 ng/g), pepper (25 ng/g and 50 ng/g), lettuce (25 ng/g and 50 ng/g), hake (12.5 ng/g and 25 ng/g) andamended soil (25 ng/g and 50 ng/g). In the case of carrot, samples apparent recoveries were calculated compared to an external calibration and compared to matrix matchedcalibration. For the rest of the matrices external calibration was only used.

Analyte Carrot

Apparent recovery withexternal calibration12.5 ng/g

Apparent recovery withexternal calibration25 ng/g

Recovery with matrix-matchedcalibration12.5 ng/g

Recovery with matrix-matchedcalibration25 ng/g

MDL (ng/g)

PFBA 113 118 169 94 3.2PFPeA 80 94 76 93 1.9PFHxA 75 81 87 94 0.9PFHpA 79 86 83 92 0.5PFOA 69 74 78 85 0.7PFNA 65 69 81 93 0.3PFDA 65 70 77 85 0.5L-PFBS 92 100 86 98 0.9L-PFHxS 68 73 81 86 0.8L-PFOS 65 69 78 86 1.0PFOSA 116 106 81 76 1.3PFHxPA 101 104 36 126 0.8PFOPA 125 134 88 90 1.6PFDPA 129 136 84 89 1.5

Pepper Lettuce

25 ng/g 50 ng/g MDL (ng/g) 25 ng/g 50 ng/g MDL (ng/g)

PFBA 91 74 6.9 94 87 8.7PFPeA 92 73 12.0 71 64 7.8PFHxA 90 74 8.2 75 75 6.8PFHpA 77 68 5.6 84 84 7.3PFOA 93 66 7.5 77 76 5.3PFNA 88 70 6.4 75 77 5.3PFDA 86 67 6.7 78 78 6.6L-PFBS 94 68 10.1 58 76 8.7L-PFHxS 87 62 9.3 75 76 2.4L-PFOS 90 69 6.3 83 85 8.3PFOSA 97 77 8.5 98 95 12.4PFHxPA 96 86 10.1 96 86 11.1PFOPA 95 96 2.1 85 111 3.2PFDPA 80 111 11.5 105 111 8.2

Hake Amended soil

12.5 ng/g 25 ng/g MDL (ng/g) 25 ng/g 50 ng/g MDL (ng/g)

PFBA –a 117 12.5 101 98 1.0PFPeA 75 77 1.2 88 90 2.1PFHxA 103 102 0.5 91 91 2.4PFHpA 79 93 0.4 88 98 3.7PFOA 85 96 0.4 83 93 3.2PFNA 86 85 0.4 92 89 2.6PFDA 82 86 0.2 92 89 3.4L-PFBS 105 94 0.4 98 100 1.8L-PFHxS 84 94 0.6 78 77 1.2L-PFOS 83 94 0.8 90 90 1.5PFOSA 104 88 0.4 55 56 7.0PFHxPA 96 96 0.5 123 105 7.0

Rsfl

baTpvwHfu

PFOPA 29 87 1.7PFDPA 80 99 1.9

a Not detected.

P columns are not suitable for the analysis of short-chain PFCAsince broad peaks are obtained. Better results might be obtainedor PFBA using and ion exchange column [40]. PFOSA showed theowest recoveries (approx. 55%) for amended soil samples.

Method detection limit (MDL) of each analyte was determinedy spiking five replicates of each blank matrix with each analytet the lowest concentration used in the validation (see Table 6).he lowest MDL values were obtained for hake and carrot sam-les, always lower than 1.89 ng/g (except for PFBA). Similar MDLalues obtained were reported by Naile et al. (MDL 0.1–2 ng/g) [3]

hen alkaline digestion with a posterior clean-up by Waters Oasis-LB was performed or by Moreta and Tena (LOD 0.5–2.2 ng/g) [26]

or packaging material using FUSLE. Furthermore, similar MDL val-es were reported by Bossi et al. [21] when ion-pair extraction

111 119 14.0112 103 22.0

was performed (MDL 3–7 ng/g). However, better MDL values werealso reported, for instance, Ullah et al. [19] reported MDL valuesbetween 0.002 and 0.02 ng/g when extraction with ACN/water andclean-up on a mixed-mode co-polymeric sorbent (C8+ quaternaryamine) were used in food samples. For the rest of the matrices MDLvalues were in the 1–12 ng/g level.

The precision of the method, expressed as relative standarddeviation (RSD), was evaluated at the two concentration levelsmentioned above and five replicates were performed at each level.Similar RSD values were obtained after correction with the corre-

sponding surrogate for both fortification levels, in the 2–15%, exceptfor PFPeA and PFOPA in hake (23% and 38%, respectively). Similarresults were reported when SPE clean-up approaches were used.For instance, Liu et al. [27] obtained RSD values between 1 and 14%

I. Zabaleta et al. / J. Chromatog

w[W

4

cwttpttubq2SOomfcP

[

[

[

[[

[[

[

[

[[[

[

[

[[[[

[[[[[

[[

[

[[

Fig. 7. Matrix effect in the detection for (a) carrot and (b) hake samples.

hen Waters Oasis WAX approach was used. Moreover, Llorca et al.13] who optimized PLE extraction with a posterior Waters Oasis

AX clean-up approach obtained RSD values between 5 and 17%.

. Conclusions

Different steps for the analysis of up to 14 PFCs, includingarboxylic, sulfonate, phosphonic and sulfonamide derivatives,ere successfully optimized in the present work. Next, some of

he major conclusions obtained in the present work are cited. Ahorough optimization of the LC–MS/MS analysis of the target com-ounds was carried out, including the chromatographic column,he mobile phase, the ionization conditions and the mass spec-rometric variables. It should be underlined that mobile phasesing 1-MP as ion-pair reagent increased the sensitivity of car-oxylic, sulfonate and phosphonic PFCs. FUSLE extraction rendereduantitative extraction of the target analytes in two successive.5 min extractions using 7 mL of a (9:1) ACN: Milli-Q mixture. ForPE clean-up Waters Oasis-HLB, Waters Oasis-WAX and Watersasis-MAX cartridges were evaluated. The low extraction efficiencybtained with the Waters Oasis-MAX cartridges and the strong

atrix effect observed for Waters Oasis-HLB discarded them from

urther validation and finally FUSLE coupled to Waters Oasis-WAXlean-up was chosen for method validation of the four families ofFCs studied in the present work.

[[[

r. A 1331 (2014) 27– 37 37

Acknowledgements

This work was financially supported by the MICINN through theCTM2011-24094 project. E. Bizkarguenaga and A. Iparragirre aregrateful to the Basque Government for their pre-doctoral fellow-ships.

References

[1] S.D. Richardson, Anal. Chem. 84 (2012) 747.[2] X. Trier, K. Granby, J.H. Christensen, J. Chromatogr. A 1218 (2011) 7094.[3] J.E. Naile, J.S. Khim, T. Wang, C. Chen, W. Luo, B. Kwond, J. Park, C. Koh, P.D.

Jones, Y. Lu, J.P. Giesy, Environ. Pollut. 158 (2010) 1237.[4] T. Wang, Y. Lu, C. Chen, J.E. Naile, J.S. Khim, J. Park, W. Luo, W. Jiao, W. Hua, J.P.

Giesy, Mar. Pollut. Bull. 62 (2011) 1905.[5] T. Wang, Y. Lu, C. Chen, J.E. Naile, J.S. Khim, J.P. Giesy, Environ. Geochem. Health

34 (2012) 301.[6] D. Herzke, E. Olsson, S. Posner, Chemosphere 88 (2012) 980.[7] G.B. Post, P.D. Cohn, K.R. Cooper, Environ. Res. 116 (2012) 93.[8] M. Wilhelm, S. Bergmann, H.H. Dieter, Int. J. Hyg. Environ. Health 213 (2010)

224.[9] B.C. Kelly, M.G. Ikonomou, J.D. Blair, B. Surridge, D. Hoover, R. Grace, F.A.P.C.

Gobas, Environ. Sci. Technol. 43 (2009) 4037.10] European Directive 2006/122/EC relating to restrictions on the marketing and

use of certain dangerous substances and preparations (perfluorooctane sul-fonates).

11] Governments Unite to Step-up Reduction on Global DDT Reliance and Add NineNew Chemicals Under International Treaty, Stockholm Convention Secretariat,Geneva, 2008.

12] S. Poothong, S.K. Boontanon, N. Boontanon, J. Hazard. Mater. 205 (2012)139.

13] M. Llorca, M. Farré, Y. Picó, D. Barceló, J. Chromatogr. A 1216 (2009) 7195.14] O. Lacina, P. Hradkova, J. Pulkrabova, J. Hajslova, J. Chromatogr. A 1218 (2011)

4312.15] Y. Picó, M. Farré, M. Llorca, D. Barceló, Crit. Rev. Food Sci. Nutr. 51 (2011) 605.16] S. Taniyasu, K. Kannan, M.K. So, A. Gulkowska, E. Sinclair, T. Okazawa, N.

Yamashita, J. Chromatogr. A 1093 (2005) 89.17] S.A. Tittlemier, K. Pepper, C. Seymour, J. Moisey, R. Bronson, X.L. Cao, R.W.

Dabeka, J. Agric. Food Chem. 55 (2007) 3203.18] N. Luque, A. Ballesteros-Gómez, S. Van Leeuwen, S. Rubio, J. Chromatogr. A 1217

(2010) 3774.19] S. Ullah, T. Alsberg, R. Vestergren, Anal. Bioanal. Chem. 404 (2012) 2193.20] R. Vestergren, S. Ullah, I. Cousins, U. Berger, J. Chromatogr. A 1237 (2012) 64.21] R. Bossi, F. Riget, R. Dietz, C. Sonne, P. Fauser, M. Dam, K. Vorkamp, Environ.

Pollut. 136 (2005) 323.22] T. Stahl, J. Heyn, H. Thiele, J. Hüther, K. Failing, S. Georgii, H. Brunn, Arch. Environ.

Contam. Toxicol. 57 (2009) 289.23] H. Zhao, Y. Guan, G. Zhang, Z. Zhang, F. Tan, X. Quan, J. Chen, Chemosphere 91

(2013) 139.24] M.P. Martínez-Moral, M.T. Tena, Talanta 109 (2013) 197.25] M.P. Martínez-Moral, M.T. Tena, Talanta 101 (2012) 104.26] C. Moreta, M.T. Tena, J. Chromatogr. A 1302 (2013) 88.27] R. Liu, T. Ruan, T. Wang, S. Song, M. Yu, Y. Gao, J. Shao, G. Jiang, Talanta 111

(2013) 170.28] U. Berger, M. Haukas, J. Chromatogr. A 1081 (2005) 210.29] M. Llorca, M. Farré, Y. Picó, D. Barceló, Anal. Bioanal. Chem. 398 (2010) 1145.30] S. Rayne, K. Forest, J. Environ. Sci. Health A 44 (2009) 317.31] E. Steinle-Darling, M. Reinhard, Environ. Sci. Technol. 42 (2008) 5292.32] X. Esparza, E. Moyano, J. de Boer, M. Galceran, S. Van Leeuwen, Talanta 86 (2011)

329.33] The Free Chemical Data Base: https://www.chemspider.com34] Y. Moroi, H. Yano, O. Shibata, T. Yonemitsu, Bull. Chem. Soc. Jpn. 74 (2001)

667.35] R. Loos, G. Locoro, T. Huber, J. Wollgast, E. Christoph, A. Jager, B. Gawlik, G.

Hanke, G. Umlauf, J.M. Zaldívar, Chemosphere 71 (2008) 306.36] S. Chu, R. Letcher, Anal. Chem. 81 (2009) 4256.37] Waters Oasis Mix-Mode Sample Extraction Products, Waters Corporation, 2001

https://www.waters.com38] S. Ullah, T. Alsberger, U. Berger, J. Chromatogr. A 1218 (2011) 6388.39] M. Moeder, S. Shrader, U. Winkler, R. Rodil, J. Chromatogr. A 1217 (2010) 2925.40] S. Taniyasu, K. Kannan, L. Yeung, K. Kwok, P. Lam, N. Yamashita, Anal. Chim.

Acta 619 (2008) 221.