Semiautomated solid-phase extraction followed by derivatisation and gas chromatography–mass spectrometry for determination of perfluoroalkyl acids in water

Journal of Chromatography A, 1318 (2013) 65–71

Contents lists available at ScienceDirect

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

Semiautomated solid-phase extraction followed by derivatisation and gas chromatography–mass spectrometry for determination of perfluoroalkyl acids in water Beatriz Jurado-Sánchez a , Evaristo Ballesteros b,∗ , Mercedes Gallego a,∗ a b

Department of Analytical Chemistry, Campus of Rabanales, University of Córdoba, E-14071 Córdoba, Spain Department of Physical and Analytical Chemistry, E.P.S. of Linares, University of Jaén, E-23700 Linares, Jaén, Spain

a r t i c l e

i n f o

Article history: Received 12 July 2013 Received in revised form 8 October 2013 Accepted 9 October 2013 Available online 17 October 2013 Keywords: Perfluoroalkyl acids Water Solid-phase extraction Derivatisation Gas chromatography–mass spectrometry

a b s t r a c t This paper describes a sensitive approach for the determination of 6 perfluoroalkyl carboxylic acids and perfluorooctane sulfonic acid in water. Samples were preconcentrated using an automatic solidphase extraction module and then manually derivatised and determined by gas chromatography–mass spectrometry. The analytes were derivatised with a isobutyl chloroformate/isobutanol mixture, using 3% N,N-dicyclohexylcarbodiimide in pyridine as the catalyst. From a systematic comparison of several reversed-phase and anion-exchange sorbent materials for the retention of perfluoroalkyl acids, the highest retention efficiencies (∼100%) were achieved with LiChrolut EN and Discovery DSC-SAX columns. LiChrolut EN was the sorbent selected due to several advantages (sample pH ∼1; sample flow rate, 5.5 mL/min; breakthrough volume, 300 mL) over Discovery DSC-SAX (sample pH ∼6; sample flow rate, 3.0 mL/min; breakthrough volume, 45 mL), for the retention of the studied compounds. Detection and quantification limits within the range of 0.1–0.5 ng/L and 0.4–1.7 ng/L, respectively, were obtained for a sorbent column of 70 mg of LiChrolut EN and 250 mL of sample, the relative standard deviation being lower than 7%. The method was applied both to the analysis of water collected at the intake (raw) and at the exit (treated) of two drinking water treatment plants, as well as to various types of water. Few samples were positive for perfluoroalkyl acids and only one acid (perfluoroheptanoic or perfluorooctanoic) was found in each treatment plant. The highest number and concentration of analytes (perfluoroheptanoic, perfluorooctanoic and perfluorodecanoic acid) were found in one wastewater. © 2013 Elsevier B.V. All rights reserved.

1. Introduction Perfluoroalkyl acids (PFAAs) are a group of man-made chemicals widely employed in numerous industrial and domestic applications [1]. The most studied compounds in this group are perfluoroalkyl carboxylic acids (PFCAs) and perfluorooctane sulfonic acid (PFOS). Important sources of these persistent, ubiquitous environmental contaminants are wastewater treatment plants [2], urban watersheds, industrial discharges [3], places of fire fighting foam usage and landfills [4]. Due to the strength of the carbon-fluorine bound, perfluoroalkyl acids are exceptionally stable with respect to metabolic or environmental degradation. Several reports associate these compounds with adverse health effects in mammals [5]; thus restrictions on their production and

∗ Corresponding authors. Tel.: +34 957 211 066; fax: +34 957 218 614. E-mail addresses: [email protected] (E. Ballesteros), [email protected], [email protected] (M. Gallego). 0021-9673/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.chroma.2013.10.034

use have been introduced globally [6–8]. The European Commission proposed an Environmental Quality Standard for inland surface water of 0.65 ng/L for PFOS and its derivatives [9]. In addition, the Environmental Protection Agency (EPA) established Provisional Health Advisory values for PFOS and perfluorooctanoic acid (PFOA) in drinking water at 200 and 400 ng/L, respectively [10]. The European Food Safety Authority fixed tolerable daily intake at 150 and 1500 ng/Kg/bw/day for PFOS and PFOA [11]. Many studies have proven that the above-mentioned pollutants are poorly removed during water treatment [12–14], thus the estimated human daily intake of PFOS or PFOA via drinking water can vary from 1.5% [15] to 55% [16]. This is mainly since their concentration in drinking water supplies exhibit a wide range of variability. For instance, concentrations of PFCAs and PFOS ranged from 0.2 to 9 ng/L [13] and 0.4 to 6 ng/L [14], respectively, in contaminated water from different European countries. In regions near highly industrialised zones in China or in Europe, Japan, India and the USA, the values were in the range of 2–120 ng/L [12] or 0.4–45 ng/L

66

B. Jurado-Sánchez et al. / J. Chromatogr. A 1318 (2013) 65–71

[17], respectively. In this scenario, there is an increasing demand for novel, rapid, efficient and selective analytical strategies to monitor these chemicals at low levels. The EPA method 537 [18] and other liquid chromatography coupled with tandem mass-spectrometry (LC–MS/MS) based procedures suffer from background contamination arising from fluoropolymers materials in the equipment [19–21]. Several strategies have been adopted in an effort to address this problem, such as the use of polyether ether ketone tubing [22] or the inclusion of a ‘guard’ column prior to the injector [23]. Another interesting method relies on the conversion of PFCAs into their corresponding phenacyl esters prior to LC–MS/MS, but requires a derivatisation time of 30 min [24]. With regard to gas chromatography (GC), PFAAs fulfil key requirements for direct measurements – volatility and thermal stability – but tailing peaks are obtained in the direct analysis of these compounds [25–28]. As a result, the analytes should be derivatised into i.e. methyl or butyl esters, by using methyl iodide [29] or ion-pairing reagents [25,27]. A less common approach is based on the derivatisation of PFCAs with 2,4-difluoroaniline, using N,N-dicyclohexylcarbodiimide as the catalyst, to produce acid anilides [30]. Fast derivatisation procedures by means of chloroformates have also been developed [28,31]. PFAAs are present in water samples at low levels and analyte enrichment is commonly required before chromatographic determination. Solid-phase microextraction (SPME) followed by in-port derivatisation has been employed for the determination of PFCAs by GC–MS [25]. Two problems are associated with this method: high detection limits (LODs), 20–750 ng/L, and a low lifetime of SPME fibres at the high temperatures. Two recently developed GC–MS approaches partly address these limitations. In the first one, PFCAs are isolated by dispersive liquid–liquid microextraction after ion-pair formation (tetrabutylammonium hydrogen sulphate) [27], with LODs ranging from 40 to 50 ng/L. The second one, which also uses SPME after fast analyte derivatisation with propylchloroformate, provides lower LOD values (0.1–7 ng/L), but does not avoid the problems associated with the deterioration of fibres [28]. As an alternative, the use of solid-phase extraction (SPE) for sample treatment is bound to substantially improve analytical sensitivity and throughput while minimising sample manipulation. In this context, the choice of the sorbent is a key point since it can control parameters such as selectivity, affinity and capacity. The most popular materials for the SPE of the target analytes from water are polymeric [22,26,30,32,33] followed by anion-exchange sorbents [29] and hemimicelles [34]. In addition, a comparison of reversed-phase and ion exchange sorbent for the isolation of a wide range of perfluoroalkyl and polyfluoroalkyl substances in water has been reported elsewhere [35]. Automated SPE methods for these compounds are based on LC instruments which incorporate the column inside the injection valve [36,37] or SPE automatic modules off-line with LC [38]; all of them suffer from the background contamination associated with the equipment. From these premises, the first aim of this study was to develop the first continuous tailor-made SPE module for the preconcentration and derivatisation of 6 PFCAs (C6 –C11 ) and PFOS, and determination of their isobutyl esters, at low parts-per-trillion levels, in water. Potential sources of blank contamination were avoided by constructing the SPE system with polyether ether ketone tubing and polyethylene materials [22]. The second aim was to study several sorbents including polymeric, silica-reversed phase and anion-exchange materials in order to establish, for the first time, the advantages/disadvantages of these materials for perfluoroalkyl acids using the same method. The third purpose was orientated to establish the possibilities of chloroformates as fast and efficient reagents for the simultaneous derivatisation of PFCAs (C6 –C11 ) and PFOS.

Table 1 Abbreviations, linear formula and pKa values of the 7 perfluoroalkyl acids. Compound

Abbreviation

Linear formula

pKa

Perfluorohexanoic acid Perfluoroheptanoic acid Perfluorooctanoic acid Perfluorononanoic acid Perfluorodecanoic acid Perfluoroundecanoic acid Perfluorooctane sulfonate

PFHxA PFHpA PFOA PFNA PFDA PFUA PFOS

CF3 (CF2 )4 COOH CF3 (CF2 )5 COOH CF3 (CF2 )6 COOH CF3 (CF2 )7 COOH CF3 (CF2 )8 COOH CF3 (CF2 )9 COOH CF3 (CF2 )7 HSO3

3.4 3.6 3.8 3.9 4.0 4.0 3.9

See reference [31].

2. Materials and methods 2.1. Standards, reagents and samples Standards of the 6 perfluoroalkyl carboxylic acids and perfluorooctane sulfonic acid were supplied from Sigma–Aldrich (Madrid, Spain), at the highest purity available. The analytes and their abbreviations are listed in Table 1. Chromatographic grade solvents (isobutanol, ethyl acetate, acetonitrile, methanol, n-hexane and acetone), acetyl chloride and acetic acid were purchased from Merck (Darmstadt, Germany). Purified water was obtained in the laboratory from a Milli-Q system (Millipore, Bedford, MA, USA). Tetradecane (internal standard, IS) and the derivatising reagents isobutyl chloroformate (IBCF), pyridine and N,N-dicyclohexylcarbodiimide (DCC) were supplied by Fluka (Madrid, Spain). All products were handled with care, using efficient fume hoods and wearing protective gloves. Polymeric sorbents Oasis HLB and LiChrolut EN were supplied by Waters (Madrid, Spain) and Merck, respectively. Amberlites XAD-2 and XAD-7, silica-reversed phase with octadecyl functional groups (Supelclean ENVI-18) and the anion-exchange sorbents Discovery DCS-NH2 , Discovery DCS-SAX, Amberlites IRA-93 and IRA-400 were purchased from Supelco (Madrid, Spain) and Lewatit MP-600 from Lanxess (Barcelona, Spain). The surface area, particle size, moisture and exchange capacity available data of sorbent materials are listed in Table 2. Stock solutions of the individual acids (5 g/L) were prepared in acetonitrile and stored in pre-cleaned polypropylene tubes at 4 ◦ C. Working solutions were prepared daily from these stocks by appropriate dilution with purified water. The eluent consisted of a mixture of ethyl acetate/isobutanol (9:1), containing 7.5% (v/v) of IBCF (derivatising reagent) and 60 ␮g/L of tetradecade (IS). A mixture of 3% DCC in pyridine was used as catalyst. All these solutions were also prepared on a daily basis. All samples were collected in 1 L polypropylene containers which were pre-cleaned with methanol following reported recommendations [18]. Before sampling, each bottle was thoroughly rinsed with sample three times. After collection the samples were stored at 4 ◦ C up to 14 days before analysis or at–20 ◦ C for a month. 2.2. Equipment Gas chromatography–mass spectrometric analysis was performed using a Focus gas chromatograph coupled to a DSQ II quadrupole mass spectrometer detector (Thermo Electron, Madrid, Spain). The GC column was a 30 m × 0.25 mm i.d. DB-5 MS capillary column with a film thickness of 0.25 ␮m (J & W, Folson, CA, USA). Helium (purity 6.0) was used as the carrier gas at a constant flow of 1 mL/min. Samples (1 ␮L) were injected in the split mode (1:20 ratio). The oven temperature was programmed as follows: 50 ◦ C, held for 3 min, ramped at 10 ◦ C/min to 170 ◦ C, held for 1 min, then at 45 ◦ C/min to 300 ◦ C (total run time, 18.8 min). The injector, transfer line and ion source temperature were 250, 300 and 200 ◦ C, respectively; and the time for solvent delay was set at 4 min.

B. Jurado-Sánchez et al. / J. Chromatogr. A 1318 (2013) 65–71

67

Table 2 Sorption efficiency (%) of the 7 perfluoroalkyl acids on different materialsa . Reversed-phase sorbent

Ion-exchange sorbent

Oasis HLB

LiChrolut EN

Amberlite XAD-2

Amberlite XAD-7

Supelclean ENVI-18

Discovery DSC-NH2

Discovery DSC-SAX

Lewatit MP-600

Amberlite IRA-93

Amberlite IRA-400

800 30–60 – –

1200 40–120 – –

800 20–60 – –

450 20–60 – –

500 60–80 – –

480 50 – –

480 50 – –

– 600 55 1.2

– – 60 1.2

– 600–750 45 1.4

Compound PFHxA PFHpA PFOA PFNA PFDA PFUA PFOS

89.5 ± 5.5 90.5 ± 5.7 91.5 ± 6.0 99.5 ± 5.5 91.9 ± 5.0 94.8 ± 5.0 95.7 ± 5.0

95.5 ± 6.0 98.6 ± 6.2 99.0 ± 6.5 99.9 ± 5.5 98.3 ± 5.5 99.8 ± 5.0 99.9 ± 5.0

68.5 ± 4.5 86.7 ± 5.5 59.7 ± 4.0 70.7 ± 4.0 85.3 ± 4.5 79.3 ± 4.0 43.2 ± 2.5

53.0 ± 3.0 62.3 ± 4.0 42.7 ± 3.0 55.6 ± 3.0 65.3 ± 3.5 69.3 ± 3.5 23.5 ± 1.5

30.0 ± 2.0 34.8 ± 2.5 84.6 ± 5.5 89.1 ± 4.5 85.1 ± 4.6 84.1 ± 4.4 86.3 ± 4.4

90.7 ± 5.4 89.6 ± 5.6 92.9 ± 6.0 91.0 ± 5.0 89.1 ± 4.8 89.8 ± 4.7 88.3 ± 4.5

98.7 ± 6.0 98.2 ± 6.2 97.1 ± 6.3 98.9 ± 5.2 98.7 ± 5.3 97.9 ± 5.1 98.9 ± 5.0

89.1 ± 5.5 89.3 ± 5.6 88.9 ± 5.8 90.9 ± 4.8 87.9 ± 4.7 90.2 ± 4.7 95.2 ± 4.9

88.4 ± 5.5 88.7 ± 5.6 88.9 ± 5.8 87.9 ± 4.7 90.6 ± 4.9 88.2 ± 4.6 91.3 ± 4.7

90.1 ± 5.5 88.9 ± 5.6 89.5 ± 5.8 90.9 ± 4.8 87.3 ± 4.7 91.4 ± 4.8 92.3 ± 4.7

Average

93.3 ± 5.3

99.0 ± 5.6

70.4 ± 4.1

53.1 ± 3.0

70.5 ± 4.0

90.2 ± 5.1

98.3 ± 5.6

90.2 ± 5.1

89.1 ± 5.1

90.1 ± 5.1

Surface area (m2 /g) Particle size (␮m) Moisture (%) Exchange capacity (meq/mL)

a

For 80 mg of each sorbent.

Selected-ion monitoring mode (ionisation energy, 70 eV) was used for quantitative analysis, and the ions of m/z 69, 131, 169, 181, in addition to the [M–CH3 –CO]+ (for perfluoroalkyl carboxylic acids) and [M–CH3 –SO2 ]+ ions (for perfluorooctane sulfonic acid) were monitored. For tetradecane, the mass values 43, 57 and 198 were used for detection. The continuous solid-phase extraction system consisted of a peristaltic pump (Gilson Minipuls-3, Villiers-le-Bel, France) fitted with poly(vinylchloride) tubes and two modified Rheodyne 5041 injection valves (Cotati, CA, USA), in which all PTFE tubing was replaced with polyether ether ketone tubing (1/16 × 0.5 mm i.d., VICI AG International, Switzerland). All teflon-based materials were also replaced with polyethylene. Sorbent columns were prepared by packing commercial deactivated-glass columns of different lengths (3 mm i.d, Omnifit, UK) with each sorbent material. The sorbent columns were conditioned as follows: 0.5 mL of methanol and then 0.5 mL of acetone for polymeric sorbents and Supelclean ENVI-18 or 1 mL of 2% acetic acid in methanol for all ionexchange materials. Finally the remaining solvents were eliminated with 1 mL of purified water for all sorbents. 2.3. Continuous SPE and derivatisation procedure The manifold employed for the preconcentration/clean-up of C6 –C11 perfluoroalkyl carboxylic acids and perfluorooctanesulfonate is depicted in Fig. 1. Initially 250 mL of water or standard solutions containing 0.4–100 ng/L of each analyte at pH ∼1.0 (adjusted with 2 mL of 12 M HCl), were passed through 70 mg LiChrolut EN column. After retention, any residual water from the system was removed by an air stream, which were

Fig. 1. Scheme of the continuous flow unit employed for the SPE extraction of the 6 perfluoroalkyl carboxylic acids and perfluorooctane sulfonic acid and manual derivatisation and evaporation. IV, injection valve; W, waste; IS, internal standard; IBCF, isobutyl chloroformate; DCC, N,N-dicyclohexylcarbodiimide; GC–MS, gas chromatograph–mass spectrometer.

subsequently used as a carrier of the eluent [loop IV2 , 200 ␮L of ethyl acetate/isobutanol (9:1) containing 7.5% (v/v) of IBCF and the IS]. The whole organic extract was collected in a 0.5-mL screw-top polypropylene vial and mixed with 4 ␮L of 3% DCC in pyridine; the vial was placed in test tube rack, which was introduced into an ultrasound-bath for 1 min. The organic extract containing the isobutyl derivatives was evaporated to ∼50 ␮L under a gentle stream of ultrapure N2 and 1 ␮L aliquot of the extract injected into the GC for analysis. Potential errors in measuring the final volume of the evaporation step were corrected with the internal standard. 3. Results and discussions 3.1. Development of the derivatisation procedure Esterification of perfluoroalkyl carboxylic acids with chloroformates has proved to be a simple and efficient strategy for the preparation of derivatives suited to GC analysis [26,31]. Apart from perfluoroalkyl carboxylic acids, this study focused on the determination of perfluorooctane sulfonic acid, which was selected because: (i) several countries set regulatory limits for this acid in tap water; (ii) while the EPA and European Union have banned the use of PFOS, this chemical is still permitted in certain applications for which no feasible alternatives are currently available, such as photo imaging, photo resist and semi conductors, aviation hydraulic fluids; (iii) it is a toxic and persistent organic pollutant widely distributed in the environment. Isobutyl chloroformate, in the presence of pyridine and isobutanol, can react with carboxylic and sulfonate groups to yield a mixed anhydride (step 1) which decarboxylates to produce the corresponding ester (step 2) [39]. The reaction scheme for the derivatisation of the perfluoroalkyl carboxylic acids and perfluorooctane sulfonic acid can be found as supplementary data (Fig. S1). In the present work a similar procedure was employed [31], but as we also included perfluorooctane sulfonic acid, a study of the optimal conditions for derivatisation was mandatory. Chromatographic relative response, calculated by dividing the peak area of each derivatised acid by the peak area of the IS, were used to study the influence of a change in the experimental parameters (amount of isobutyl chloroformate, type and amount of catalyst(s), reaction temperature, reaction time and solvent medium). Tetradecane was chosen as IS due to its inertness during derivatisation and the virtual constancy of its peak area. Thus, due to the constant concentration of the target analytes and IS, at the same chromatographic conditions, the increase in relative response is related to the derivatization yield.

68

B. Jurado-Sánchez et al. / J. Chromatogr. A 1318 (2013) 65–71

To study the influence of the amount of isobutyl chloroformate (IBCF), 20 ␮L of standard solutions in acetonitrile, containing 2.5 mg/L of each fluorinated compound and 0.5 mg/L of the IS, were derivatised by adding 0–20 ␮L of IBCF (derivatising reagent 1), 0–30 ␮L of isobutanol (derivatising reagent 2) and 4 ␮L of pyridine (catalyst). The final volume of the mixture was normalised to 200 ␮L with acetonitrile. After 3 min ultrasonication, isobutyl derivatives were extracted with 200 ␮L of n-hexane. Then, 200 ␮L of a 1 M NaCl aqueous solution were added to aid phase separation and 1 ␮L aliquots of the n-hexane extract were injected into the gas chromatograph–mass spectrometer for analysis. The derivatisation yield, in terms of relative response, increased with the amount of IBCF and isobutanol up to 15 ␮L and 20 ␮L, respectively, remaining constant above this value. Therefore both volumes were selected as optimal. On the other hand, it has been proved that the use of N,N-dicyclohexylcarbodiimide (DCC) as co-catalyst (along with pyridine) produced a significant increase in the derivatisation yield of carboxylic and sulfonic acids with chloroformates [39]. This is mainly because DCC can act as a coupling agent, playing an active role in the elimination of small molecules such as the HCl liberated after the reaction of IBCF with the target analytes (step 1). In this work, the effect and amount of two catalysts were studied by using variable volumes of individual pyridine (0–10 ␮L) or DCC (0–10 ␮L) and mixtures of DCC (1–10%) in pyridine. The highest relative response for all compounds were achieved with 4 ␮L of 3% DCC in pyridine. Neither the temperature of the derivatising reaction (20–60 ◦ C) nor the ultrasonic shaking time (0–10 min) has an effect on the derivatisation yield. Nevertheless, the sample was homogenised in an ultrasonic bath for 1 min at room temperature to increase the precision of the results. After choosing the derivatising reagents and optimal reaction conditions, several solvents were evaluated as reaction medium. For this purpose, 20 ␮L of the standard solutions containing the 7 perfluoroalkyl acids and 0.5 mg/L of the IS were prepared in different solvents (acetonitrile, ethyl acetate, acetone, methanol, purified water at pH 1, 3, 5 and 7, aqueous NH4 OH and aqueous NaCl solutions) and derivatised by adding the derivatising reagents (15 ␮L of IBCF and 20 ␮L of isobutanol, and 4 ␮L of 3% DCC in pyridine). The final volume of each sample was normalised to 200 ␮L with the appropriate solvent in each case. Satisfactory derivatisation yields (see Fig. S2) were obtained for all solvents assayed omitting acetone and methanol which provide relative responses 10–15 times lower For the selection of solvent reaction medium, other variables such as the most suitable sorbent material and eluent for the preconcentration/elution of the studied perfluoroalkyl acids should be taken into account (see next section). 3.2. Optimisation of the sorption and elution processes Firstly, a systematic comparison was performed regarding the efficiency of several sorbent materials with different modes of action (reversed-phase and ion-exchange) for the retention of the perfluoroalkyl acids. To this end, we used the solid-phase extraction system depicted in Fig. 1 and columns packed with 80 mg of each sorbent listed in Table 2. Blank contamination was eliminated by avoiding the use of fluoropolymer materials during sample preparation and by rigorously cleaning all material with methanol before use. In the literature, there is no consensus on the true pKa values for PFCAs and PFOS at environmentally relevant concentrations. Thus, the pKa of PFOA has been estimated to be 3.8 [40], whereas similar values has been proposed for C6 –C11 PFCAs and PFOS [31,40]. Nevertheless, Goss [41] and Cheng et al. [42] have both calculated the pKa ’s of PFOA and PFOS to be lower than 1. best approach of the pKa values for the 7 perfluoroalkyl acids is included in Table 1. When the analyte is in neutral form, it becomes more hydrophobic and retention strengthens under reversed-phase conditions. In order

to increase sorption efficiency, the aqueous sample was adjusted at pH ∼1 by adding 80 ␮L of 12 M HCl per 10 mL of sample containing 25 ␮g/L of each fluorinated compound. The mechanism of isolation by ion-exchange involves the exchange of anionic PFAAs from water onto the oppositely charged anionic resin. In this case, the pH of the sample matrix must be 2 pH units above the pKa of the acid to ensure its anionic charge, thus the aqueous sample was maintained at pH ∼6. The sorption efficiency was assessed by comparing the amount of each acid present in 10 mL of aqueous standard solutions before and after passing through the sorbent column. To this end, aliquots of 160 ␮L of each aqueous solution (before and after preconcentration) were derivatised with 15 ␮L of IBCF and 20 ␮L of isobutanol, and 4 ␮L of the mixture of catalysts, following the developed procedure. Then, isobutyl ester derivatives were extracted with 200 ␮L of n-hexane and analysed by GC–MS. The average sorption efficiency (n = 3) for the 7 perfluorinated species was dependent on the individual sorbent as listed in Table 2. For reversed-phase sorbents, the maximum analyte sorption (as mean sorption efficiency) was obtained for LiChrolut EN (99%) and Oasis HLB (93%). For Amberlites the sorption was lower (53–70%), which can be ascribed mainly to the differences in surface areas and polarity. In addition, LiChrolut EN and Oasis HLB are doped with hydrophilic groups (sulfonic acid or N-vinylpyrrolidone, respectively), which imparts a somewhat polar character to the matrix and allows more effective mass transfer and better sorption characteristics. In other words, the hydrophilic groups in the stationary phase of LiChrolut EN and Oasis HLB are interacting with the polar headgroups of PFAAs. For the silica reversed-phase with octadecyl groups (SupelClean ENVI-18), the mean sorption efficiency was over 86% for C8 –C11 acids and PFOS, whereas, only 30% of the other two acids (PFHxA and PFHpA) were retained. Supelclean ENVI-18 is a silica reverse-phase sorbent with trifunctional groups. Unlike common monofunctional silicas, the above mentioned sorbent is more stable in acidic medium because organosilane is attached to the silica surface at several locations. Thus, when the sample pH is lower than 2, it is critical to use trifunctional phases in order to prevent hydrolysis of the hydrocarbon group from the surface of the sorbent. For anion-exchange materials, the mean average percent sorption for all the acids was similar, ∼90%, omitting 98% for Discovery DSC-SAX. It can be concluded that both LiChrolut EN and Discovery DSC-SAX were the most adequate materials for retaining the 7 PFAAs with the highest efficiency, but with different retention mechanisms. The main difference between the two materials so far was the sample pH, but further experiments were conducted in order to establish other behaviours. The SPE system was optimised by using two columns packed with LiChrolut EN or Discovery DSC-SAX materials. The variables studied in the first place were the amount of sorbent and type of eluent. The elution efficiency was tested by using columns packed with 80 mg of each material and 300 ␮L of eluent. The most suitable solvents for derivatisation (acetonitrile, ethyl acetate and aqueous solutions) were evaluated as eluents. For Discovery DSC-SAX, taking into account the exchange properties of the resin, elution was carried out in alkaline medium (NH4 OH) or by anionic change (NaCl solutions), mixed with acetonitrile. Ethyl acetate or 0.5 M aqueous NaCl/acetonitrile (1:1) was found to be the best eluent for LiChrolut EN or Discovery DSC-SAX. The other eluents were ∼3 to 5 times less efficient. In order to perform the simultaneous elution and derivatisation of the target analytes, the derivatising reagents (isobutanol and IBCF) plus 3% DCC in pyridine (catalyst) were added to the eluents. Moreover, it was not possible to add the derivatising reagents and the catalyst at the same time, since in the presence of pyridine a precipitate was formed. Thus, the derivatising reagents were mixed with the appropriate eluent and pyridine was added after elution. As a result, the elution process was as follows: a solution of IBCF (7.5% v/v) prepared in

B. Jurado-Sánchez et al. / J. Chromatogr. A 1318 (2013) 65–71 Table 3 Chemical and flow variables of sorption and elution of the 7 perfluoroalkyl acids. Sorbent

LiChrolut EN Studied range (selected value)

Eluent

Amount of sorbent (mg) Eluent volume (␮L) Sample pH Sample flow rate (mL/min) Eluent flow rate (mL/min) Breakthrough volume (mL)a a

Ethyl acetate/isobutanol (9:1) 25–150 (70) 50–250 (200) 1.0–8.0 (1.0) 0.5–5.5 (5.0) 3.0 10–300 (250)

Discovery DSC-SAX Studied range (selected value) 0.5 M aqueous NaCl/acetonitrile (1:1) 25–150 (70) 50–250 (200) 1.0–8.0 (6.0) 0.5–4.0 (2.5) 3.0 10–100 (40)

For 70 mg of each sorbent.

ethyl acetate/isobutanol (9:1) for LiChrolut EN or in 0.5 M aqueous NaCl/acetonitrile (1:1) for Discovery DSC-SAX. Table 3 lists the studied range and the selected value of all variables related with LiChrolut EN and Discovery DSC-SAX sorbents. The best analytical signals for all compounds were obtained above 50 and 60 mg LiChrolut EN and Discovery DSC-SAX or 175 ␮L of eluent (for additional information see Figs. S3 and S4). The sample pH was studied over the range 1–8 (adjusted with diluted HCl or NaOH) to 10 mL of aqueous standard solutions of the perfluoroalkyl acids. For LiChrolut EN, as expected, the highest sorption efficiency for all compounds was obtained in acidic medium (pH 1.0–1.3), whereas for Discovery DSC-SAX the acids were retained above pH 5.5 or 5.9 for PFHxA and PFHpA or the other species, respectively (see Fig. S5). Flow variables are also listed in Table 3. The sample flow rate can be increased for LiChrolut EN (5.5 mL/min) with regard to Discovery DSC-SAX, which is an important advantage in order to preconcentrate large volumes of sample. This is mainly due to the particular mechanism of anion-exchange, which requires some time for the diffusion of the anion to the layer immediately surrounding the anion-exchange-bead, thus the sample flow rate should be considerably lower than with the reversed-phase mechanism [37]. The last variable evaluated in the optimised system was the breakthrough volume, which can be defined as the largest sample volume that can be processed to obtain the highest possible amount of isolated analyte (thus the highest enrichment factors). This crucial parameter was examined by using aqueous standard solutions at the optimum pH of each sorbent containing 10 ng of each analyte at different volumes (10–300 mL), for insertion into the SPE system. Sorption efficiency of ∼100% was obtained with volumes up to 300 and 45 mL for LiChrolut EN and Discovery DSCSAX, respectively (see Fig. S5). Overall, it can be concluded that LiChrolut EN performed better for the isolation of the 7 PFAAs from water than Discovery DSC-SAX, namely: the sample pH required for analyte retention (pH ∼1) minimised/avoided interferences caused by metals (which are not precipitated) and natural organic matter present in the water and higher sample flow rate and breakthrough volume, which allowed for greater sensitivity without compromising analysis time. Finally, in order to increase the sensitivity of the method, the extract (200 ␮L) containing derivatised analytes was evaporated to ∼50 ␮L with a stream of N2 . 3.3. Analytical features Under optimised SPE-GC–MS conditions, analytical parameters such as detection limits (LODs), quantification limits (LOQs), and intra and inter-day precision were examined in order to validate the method developed for the 7 perfluoroalkyl acids and PFOS (Table 4) for 250 mL of sample.

69

Table 4 Analytical figures of merit of the determination of the 7 perfluoroalkyl acids using the developed SPE and GC–MS method. Compound

PFHxA PFHpA PFOA PFNA PFDA PFUA PFOS

Linear range (ng/L)a 1.7–100 0.8–100 1.7–100 0.4–100 1.7–100 0.7–100 0.4–100

LODs (ng/L)

0.5 0.3 0.5 0.1 0.5 0.2 0.1

RSD (%) (n = 12)b Intra-day

Inter-day

6.0 6.3 6.5 5.3 5.4 5.2 5.1

7.0 6.9 7.0 6.0 6.2 6.4 6.3

a

LOQ values are the lower concentration of the linear range. Relative standard deviation. Values obtained for samples fortified with 5 ng/L of each compound. b

In this case, a blank signal is of special importance, taking into account the problems associated with the contamination coming from the fluoropolymer materials employed in the laboratory. Possible contamination was checked by using blank samples, consisting of 250 mL of ultrapure water acidified at pH 1, which were analysed every 3 sample injections. No contamination of PFAAs was observed at the detectable levels of the method. The linearity of the calibration graphs was studied by spiking ultrapure water with the analytes over a concentration range of 0.4–100 ng/L. Correlation coefficients were over 0.995 in all cases, demonstrating a good linearity in the range studied. The LODs, calculated at a signal-to-noise ratio of three, based on peak-to-peak noise under selected-ion monitoring mode, ranged between 0.1 and 0.5 ng/L. The LOQs for each compound were taken as the lowest concentration of the linear range (see Table 4). The method developed provided higher LODs than those obtained by the most sensitive LC methods described in the literature using manual SPE (250–500 mL of water) and sophisticated instruments such as LC coupled with a quadrupole-time-of-flight mass spectrometer (LODs from 0.02 to 0.14 ng/L) [13] or an ultrahigh LC–tandem mass spectrometer (LODs from 0.002 to 2 ng/L) [22]. Nevertheless, higher LODs are obtained with SPE automatic methods based on LC instruments, for the determination of PFOS (LOD, 5 ng/L) [36] or several perfluoroalkyl acids (3–15 ng/L) [37] and (100–3000 ng/L) [38]. A comparison with other GC alternatives revealed that the LODs here obtained (0.1 and 0.5 ng/L) are lower than a method based on manual SPE and GC–MS with negative chemical ionisation (LODs from 0.1 and 6 ng/L; 250 mL of sample) [26] or SPME (LODs from 0.1 and 7 ng/L) [28]. The precision of the method (as relative standard deviation, RSD) was evaluated by analysing twelve spiked ultrapure water samples containing 5 ng/L of the 7 PFAAs on the same day (intra-day) or by performing four consecutive extractions each day over a period of 3 working days (inter-day precision). As can be seen in Table 4, the RSDs obtained were satisfactory in all instances, with average values of 5.7% and 6.5% intra and inter-day. 3.4. Real water sample analysis In order to study the effect of the sample matrix upon the SPE-GC–MS method for its application to real sample analysis, a recovery study was conducted. For this purpose, 250 mL of various types of uncontaminated water including tap, swimming pool, well, rain, river and wastewater were fortified at two different concentrations (2 and 10 ng/L) of each compound and analysed in triplicate (n = 3). Average recoveries (94–98%) were acceptable for all samples, which testified to the applicability of our method in different water. Shown in Fig. 2 is the GC–MS chromatogram for a blank sample. As can be seen, none of the 7 perfluoroalkyl acids studied were found at detectable levels with the method. Then, the applicability of the method was tested on 22 water samples with the following

70

B. Jurado-Sánchez et al. / J. Chromatogr. A 1318 (2013) 65–71

Table 5 Analysis of the water samples from two DWTPs before (raw) and after treatment (treated) by SPE-GC–MS (±RSD, %, n = 3). Compound

PFHxA PFHpA PFOA PFNA PFDA PFUA PFOS

DWTP 1 (chloramination)

DWTP 2 (ozonation/chlorination)

Raw 1,2

Treated 1,2

Raw 3

Treated 3

Raw 4

Treated 4

Raw 5

Treated 5

Raw 6

Treated 6

<0.5 <0.3 <0.5 <0.1 <0.5 <0.2 <0.1

<0.5 <0.3 <0.5 <0.1 <0.5 <0.2 <0.1

<0.5 1.1 (±7.2%) <0.5 <0.1 <0.5 <0.2 <0.1

<0.5 1.3 (±6.8%) <0.5 <0.1 <0.5 <0.2 <0.1

<0.5 <0.3 4.0 (±6.8%) <0.1 <0.5 <0.2 <0.1

<0.5 <0.3 2.9 (±7.3%) <0.1 <0.5 <0.2 <0.1

<0.5 <0.3 3.6 (±7.0%) <0.1 <0.5 <0.2 <0.1

<0.5 <0.3 2.4 (±7.4%) <0.1 <0.5 <0.2 <0.1

<0.5 <0.3 3.8 (±6.9%) <0.1 <0.5 <0.2 <0.1

<0.5 <0.3 2.6 (±7.3%) <0.1 <0.5 <0.2 <0.1

aims: (i) to study both the occurrence of perfluoroalkyl acids in raw water (taken from a reservoir) used for tap water production and (ii) to determine the occurrence of the 7 target analytes in some untreated and treated (swimming pool) water. Method blank samples consisting of ultrapure water (250 mL) were also analysed in parallel to control possible contamination. The results obtained are listed in Tables 5 and 6. The first aim was carried out by collecting water samples from two DWTPs located in SE Spain. DWTP 1 employed chloramines as the disinfectant and provides a population of ∼60,000 inhabitants. DWTP 2 uses ozone and chlorine as disinfectants and is located in a city of ∼700,000 inhabitants. At each plant, raw and finished treated water samples were collected once a week in 6 different months, in triplicate (n = 6). None of the 7 perfluoroalkyl acids were found at detectable levels (
of the compound to any surfaces that may be present during the treatment process. Because of the absence of fluorochemical manufacturing industries in the regions under study, the presence of PFHpA or PFOA in each drinking water supply can be ascribed to discharges from urban watersheds and wastewater treatment plants. Similar findings were obtained in a study about the occurrence of 6 PFCAs in 7 DWTPs across the US employing either chlorine or ozone and chlorine for disinfection. Thus, the total concentration of all the compounds was higher in raw and treated water samples with higher degrees of wastewater impact (from <5 to 50% or more, total concentration <1 to 107 ng/L). PFOA was detected in ∼60% of the utilities at an average concentration of 15 ng/L [36]. In Europe, a more recent report conducted in a DWTP using ozonation revealed the poor efficiency of these treatments for the removal of 5 PFCAs. Among the compounds detected, PFHpA and PFOA were found at average concentrations of 1.4 and 5.7 ng/L, respectively [14]. The method developed was also tested in several environmental water samples (well, rain, river) as well as in two swimming pools treated with different disinfectants, and in two wastewaters. Fortunately, only PFHpA (10 ng/L), PFOA (13 ng/L) and PFDA (12 ng/L) were detected in one wastewater (see Fig. 2B). The presence of PFHpA, PFOA and PFDA in the wastewater can be ascribed to several sources such as their direct uses in domestic applications (i.e. cosmetics, lubricants, paints, etc.) and indirectly such as their presence as residual impurities in commercial products [43–45] or the breakdown of their fluorinated precursors present in commercial products.

4. Conclusions

Fig. 2. GC–MS chromatograms (SIM mode) obtained in the analysis of (A) a blank (250 mL of acidified ultrapure water) sample and (B) 250 mL of the wastewater 1. For peaks identification see Table 1.

A sensitive method has been developed for the continuous solid-phase extraction, rapid derivatisation (1 min) with chloroformates and off-line GC–MS determination of 6 perfluoroalkyl carboxylic acids and perfluorooctane sulfonic acid (a representative compound of the group of perfluoroalkyl sulfonates) in water. The automated SPE system was assembled in the laboratory using commonly available low pressure peristaltic pump and injection valves, polyether ether ketone tubing and polyethylene materials. Thus, this new approach successfully addresses the current trend in the development of analytical methods for measuring fluorinated compounds, namely: minimisation of sample manipulation and elimination of sample contamination. In the absence of systematic studies in the literature, an exhaustive comparison was performed on the advantages and disadvantages of several sorbent materials (polymeric sorbents, silica reversed-phase with octadecyl functional groups and silica based and polystyrene anionexchange resins) on the retention of the compounds studied. Under the same experimental conditions, the highest retention efficiencies and enrichment factors (highest breakthrough volume) for the 7 perfluoroalkyl acids were obtained for LiChrolut EN sorbent, which can interact with analytes through various types of mechanisms including polar interactions. Therefore, it was concluded that LiChrolut EN can be proposed as a good sorbent for the retention

B. Jurado-Sánchez et al. / J. Chromatogr. A 1318 (2013) 65–71

of the fluorinated acids studied since it provides such remarkable advantages as low sample pH and high breakthrough volume. An overall evaluation in terms of linearity, precision and sensitivity shows that the SPE method developed is suitable for monitoring these compounds in several types of water at very low concentrations. Acknowledgements This work was funded in the frameworks of Projects CTQ 2010-17008 (Spain’s Ministry of Education) and P09-FQM-4732DGI (Andalusian Regional Government). FEDER also provided additional funding. Beatriz Jurado is grateful for the award of a contract from Project CTQ 2010-17008. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.chroma.2013. 10.034. References [1] A.B. Lindstrom, M.J. Strynar, E.L. Libelo, Environ. Sci. Technol. 45 (2011) 7954. [2] C.P. Higgins, J.A. Field, C.S. Criddle, R.G. Luthy, Environ. Sci. Technol. 39 (2005) 3946. [3] K.J. Hansen, H.O. Johnson, J.S. Eldridge, J.L. Butenhoff, L.A. Dick, Environ. Sci. Technol. 36 (2002) 1681. [4] M.M. Schultz, D.F. Barofsky, J.A. Field, Environ. Sci. Technol. 38 (2004) 1828. [5] C. Lau, K. Anitole, C. Hodes, D. Lai, A. Pfahles-Hutchens, J. Seed. Toxicol. Sci. 99 (2007) 366. [6] Directive 2006/122/EC of the European Parliament and of the Council, Official J. Eur. Commun. L 372 (2006) 32. [7] EPA (Environmental Protection Agency), 2006, PFOA Stewardship Program, available at http://www.epa.gov/oppt/pfoa/pubs/stewardship/index.html (accessed 27.06.13). [8] The new POPs under the Stockholm Convention, available at http://chm. pops.int/Programmes/New%20POPs/The%209%20new%20POPs/tabid/672/ language/en-US/Default.aspx (accessed 27.06.13). [9] Directive of the European Parliament and of the Council amending directives 2000/60/EC and 2008/105/EC as regards priority substances in the field of water policy, available at http://ec.europa.eu/environment/ water/water-dangersub/pdf/com 2011 876.pdf (accessed 27.06.13). [10] EPA, 2009, Provisional Health Advisory for perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS), available at http://water.epa.gov/action/ advisories/drinking/upload/2009 01 15 criteria drinking pha-PFOA PFOS.pdf (accessed 27.06.13). [11] European Food Safety Authority, EFSA J. 653 (2008) 1. [12] Y.L. Mak, S. Taniyasu, L.W.Y. Yeung, G. Lu, L. Jin, Y. Yang, P.K.S. Lam, K. Kannan, N. Yamashita, Environ. Sci. Technol. 43 (2009) 4824. [13] S. Ullah, T. Alsberg, U. Berger, J. Chromatogr. A 1218 (2011) 6388. [14] C. Eschauzier, E. Beerendonk, P. Scholte-Veenendaal, P. De Voogt, Environ. Sci. Technol. 46 (2012) 1708.

71

[15] H. Fromme, S.A. Tittlemier, W. Völkel, M. Wilhelm, D. Twardella, Int. J. Hyg. Environ. Health 212 (2009) 239. [16] C.W. Noorlander, S.P.J. van Leeuwen, J.D. Biesebeek, M.J.B. Mengelers, M.J. Zeilmaker, J. Agric. Food Chem. 59 (2011) 7496. [17] M. Llorca, M. Farré, Y. Picó, J. Müller, T.P. Knepper, D. Barceló, Sci. Total Environ. 431 (2012) 139. [18] EPA Method 537, Determination of selected perfluorinated alkyl acids in drinking water by solid-phase extraction and liquid chromatography/tandem mass spectrometry (LC/MS/MS), EPA/600/R-08/092, National Exposure Research Laboratory, Office of Research and Development, US Environmental Protection Agency, Cincinnati, OH, 2009. [19] V.I. Furdui, P.W. Crozier, E.J. Reiner, S.A. Mabury, Chemosphere 73 (2008) S24. [20] N. Luque, A. Ballesteros-Gómez, S. Van Leeuwen, S. Rubio, J. Chromatogr. A 1217 (2010) 3774. [21] N. Marchetti, L. Caciolli, A. Laganà, F. Gasparrini, L. Pasti, F. Dondi, A. Cavazzini, Anal. Chem. 84 (2012) 7138. [22] M. Onghena, Y. Moliner-Martínez, Y. Picó, P. Campíns-Falcó, D. Barceló, J. Chromatogr. A 1244 (2012) 88. [23] N. Luque, A. Van Leeuwen, S. Ballesteros-Gómez, S. Rubio, J. Chromatogr. A 1235 (2012) 84. [24] J. Qiu, C. Wu, Y. Fang, C. Yang, X. Li, X. Piao, D. Li, J. Chromatogr. A 1235 (2012) 132. [25] R. Alzaga, J.M. Bayona, J. Chromatogr. A 1042 (2004) 155. [26] V. Dufková, R. Cabala, V. Sevcík, Chemosphere 87 (2012) 463. [27] W.L. Liu, Y.C. Ko, B.H. Hwang, Z.G. Li, T.C.C. Yang, M.R. Lee, Anal. Chim. Acta 726 (2012) 28. [28] M. Monteleone, A. Naccarato, G. Sindona, A. Tagarelli, J. Chromatogr. A 1251 (2012) 160. [29] C.A. Moody, G.N. Hebert, S.H. Strauss, J.A. Field, J. Environ. Monitor. 5 (2003) 341. [30] B.F. Scott, C.A. Moody, C. Spencer, J.M. Small, D.C. Muir, S.A. Mabury, Environ. Sci. Technol. 40 (2006) 6405. [31] V. Dufková, R. Cabala, D. Maradová, M. Stícha, J. Chromatogr. A 1216 (2009) 8659. [32] Y.C. Chang, W.L. Chen, F.Y. Bai, P.C. Chen, G.S. Wang, C.Y. Chen, Anal. Bioanal. Chem. 402 (2012) 1315. ˜ [33] O. Quinones, S.A. Snyder, Environ. Sci. Technol. 43 (2009) 9089. [34] X. Zhao, J. Li, Y. Shi, Y. Cai, S. Mou, G. Jiang, J. Chromatogr. A 1154 (2007) 52. [35] S. Taniyasu, K. Kannan, M.K. So, A. Gulkowska, E. Sinclair, T. Okazawa, N. Yamashit, J. Chromatogr. A 1093 (2005) 89. [36] M. Takino, S. Daishima, T. Nakahara, Rapid Commun. Mass Spectrom. 17 (2003) 383. [37] F. Gosetti, U. Chiuminatto, D. Zampieri, E. Mazzucco, E. Robotti, G. Calabrese, M.C. Gennaro, E. Marengo, J. Chromatogr. A 1217 (2010) 7864. [38] Z. Kuklenyik, J.A. Reich, J.S. Tully, L.L. Needham, A.M. Calafat, Environ. Sci. Technol. 38 (2004) 3698. [39] S. Angelino, V. Maurino, C. Minero, E. Pelizzetti, M. Vincenti, J. Chromatogr. A 793 (1998) 307. [40] J.M. Armitage, M. Macleod, I.T. Cousins, Environ. Sci. Technol. 43 (2009) 5830. [41] K.U. Goss, Environ. Sci. Technol. 42 (2008) 456. [42] J. Cheng, E. Psillakis, M.R. Hoffmann, A.J. Colussi, J. Phys. Chem. A 113 (2009) 8152. [43] S.T. Washburn, T.S. Bingman, S.K. Braithwaite, R.C. Buck, L.W. Buxton, H.J. Clewell, L.A. Haroun, J.E. Kester, R.W. Rickard, A.M. Shipp, Environ. Sci. Technol. 39 (2005) 3904. [44] X. Liu, K. Krebs, Z. Guo, N. Roache, J. Chromatogr. A 1216 (2009) 3910. [45] S. Fiedler, G. Pfister, K.W. Schramm, Toxicol. Environ. Chem. 92 (2010) 1801.