Analytica Chimica Acta 804 (2013) 126–134

Contents lists available at ScienceDirect

Analytica Chimica Acta journal homepage: www.elsevier.com/locate/aca

Determination of benzotrifluoride derivative compounds in groundwater Roberto Lava a,b,∗ , Emilia Aimo a , Luciana Menegus a , Giulio Pojana b,1 , Antonio Marcomini b a b

Laboratory Department, ARPAV – Regional Agency for Environmental Protection and Prevention of Veneto, Via Lissa 6, 30174 Mestre Venice, Italy Department of Environmental Sciences, Informatics and Statistics, University Ca’ Foscari, Dorsoduro 2137, Calle Larga Santa Marta, 30123, Venice, Italy

h i g h l i g h t s

g r a p h i c a l

a b s t r a c t

• Two methods for the simultaneous

• • • •

determination of nine benzotrifluoride derivatives (BTFS) in groundwater are proposed. The method for the more volatile BTFS is based on Purge-and-Trap and GC-MS techniques. The method for the semivolatile BTFS is based on SPE and GC-MS techniques. The methods are fit-for purpose validated and tested on samples collected from a polluted area. Most of the compounds investigated and detected have never been reported as environmental pollutants.

a r t i c l e

i n f o

Article history: Received 22 May 2013 Received in revised form 29 September 2013 Accepted 3 October 2013 Available online 14 October 2013 Keywords: Benzotrifluoride derivative Purge-and-Trap Solid Phase Extraction Gas chromatography–mass spectrometry Groundwater

a b s t r a c t Two simple analytical methods for the simultaneous determination and quantification of benzotrifluoride and eight chlorinated, amino and nitro benzotrifluoride derivatives in groundwater are proposed. Benzotrifluoride, 4-chlorobenzotrifluoride, 2,4-dichlorobenzotrifluoride and 3,4-dichlorobenzotrifluoride, were extracted by Purge-and-Trap on the basis of their volatile properties, while 3-aminobenzotrifluoride, 4-nitrobenzotrifluoride, 3-amino-4-chlorobenzotrifluoride, 3-nitro-4-chlorobenzotrifluoride and 4-chloro-3,5-dinitrobenzotrifluoride extractions were done with an automated SPE system. The analytical separations and detections were performed with two different GC systems, both equipped with single quadrupole mass spectrometer as detector. The LOD ranges for the two methods were 0.002–0.005 ␮g L−1 and 0.01–0.07 ␮g L−1 , respectively. Both extraction methods were developed using spiked Milli-Q water and were then demonstrated with groundwater samples collected during autumn 2008. The areas of groundwater collection were polluted due to an episode of improper industrial soil disposal and consequent leakage of aliphatic and aromatic, fluorinated chemicals into the groundwater. This work eventually revealed the presence of several benzotrifluoride compounds most of them, like dichloro- and amino-derivatives, never been reported as environmental contaminants. © 2013 Elsevier B.V. All rights reserved.

∗ Corresponding author. Present address: Institute for Reference Materials and Measurements (IRMM), European Commission Joint Research Centre (JRC), Retieseweg 111, 2440, Geel, Belgium. Tel.: +32 (0)14 571907. E-mail address: [email protected] (R. Lava). 1 Present address: Department of Philosophy and Cultural Heritage, University Ca’ Foscari, Dorsoduro 2137, Calle Larga Santa Marta, 30123, Venice, Italy. 0003-2670/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.aca.2013.10.006

R. Lava et al. / Analytica Chimica Acta 804 (2013) 126–134

1. Introduction A relatively inert trifluoride group ( CF3) bonded to an aromatic ring characterizes benzotrifluoride (BTF) and its synthetic derivatives (BTFS). It provides more stability and hydrophobic character as compared to similar, not-substituted molecules. Although trifluoride group does not exert a resonance effect, it withdraws an electron which deactives the ring and leads to aromatic electrophilic substitutions primarily directed towards the meta-position. Furthermore, a wide range of compounds with commercial significance can be obtained through aromatic nucleophilic substitutions of chlorine and nitrogen atoms in benzotrifluoride derivatives [1]. For many years, BTF has been used as solvent, dielectric fluid, additive for coating formulation and vulcanizing agent in several industrial sectors [2]. It is inert to very basic conditions and also reasonably stable in acid. This considerable stability of BTF in non-reactive applications suggests nowadays its sustainable use as an efficient alternative solvent for chemical organic reactions and processes currently conducted in dichloromethane or similar chlorinated solvents [3]. Moreover, BTFS are widely synthesized as intermediates for different industrial purposes [4] like diphenyl ethers, phenylurea, and dinitroaniline herbicides [1,5]. More than half of the BTFS produced is used in the agrochemical field as a first intermediate step for the synthesis of crop protection agents and biocides [6], in particular insecticides and herbicides like trifluralin. The ability of fluorine atom to become a more effective target than hydrogen or other halogens means that a much smaller quantity of product is required to achieve effective results, which in turn has obvious benefits for the environment. Although trifluralin is one of the most frequently used herbicides worldwide, many new generation herbicide derivatives are continuously developed from BTFS in order to obtain more efficient and environmental friendly products. These are all characterized by nitro groups in both orto-positions to the CF3 and by a different functional group that substitute the chlorine atom in 4-chloro-3,5-dinitrobenzotrifluoride. Therefore, trifluralin appears to have a low toxicity for mammals and birds and

127

thus may not be as acutely harmful to terrestrial wildlife as many other herbicides [7]. However, aquatic organisms are highly susceptible to these compounds, especially to trifluralin. Because of this relevant toxicity it was listed in Annex II of Directive 2008/105/EC [8] as a priority substance in the European water policy [9]. Its use in agricultural crop protection is currently prohibited in all EU territories and it was widely considered in the EU strategy for the implementation of the Water Framework Directive [10]. Other considerable amounts of BTFS are used as processing aids or intermediates for the production of paint additives, dyes, in the chrome plating, and as drug intermediate in pharmaceutical. For example, phenothiazine and pyridazinone BTFS derivatives are used for the production of antidepressant, antihypertensive, diuretics, antimalarials and antibiotics [1,11–13]. The target compounds (Fig. 1) considered in this work are: benzotrifluoride (BTF), 4-chlorobenzotrifluoride (4CBTF), 2,4dichlorobenzotrifluoride (24DCBTF), 3,4-dichlorobenzotrifluoride (34DCBTF), 3-aminobenzotrifluoride (3ABTF), 4-nitrobenzotrifluoride (4NBTF), 3-amino-4-chlorobenzotrifluoride (3A4CBTF), 3-nitro-4-chlorobenzotrifluoride (3N4CBTF) and 4-chloro-3,5dinitrobenzotrifluoride (35DN4CBTF). Their most important physicochemical properties are listed in Table 1 [14]. Presently, there is no adequate data published in literature to assess the environmental fate of the target compounds mentioned in this work. When BTFS are released in the environment, they will primarily partition in the air with atmospheric lifetimes in the order of a few months. Consequently, BTFS are not implicated in ozone depletion and global warming [1]. When released into soil, BTFS have low to moderate mobility. However, volatilization from moist and dry soil surfaces (estimations of KOC based on KOW experimental values) may be considerable [15]. The adsorption into soil is expected to attenuate volatilization, which is the principal route of dispersion. When released into water, BTFS may adsorb into suspended solids and sediments. Experimental bioconcentration factor (BCF) values ranges from 26 to 58 in different organisms for BTF, up to 1500 in rainbow trout for 34DCBTF. This underlines that bioconcentration

Fig. 1. Benzotrifluoride and its derivatives considered into this study.

128

R. Lava et al. / Analytica Chimica Acta 804 (2013) 126–134

Table 1 Physicochemical properties of BTF and BTFS.

CAS # MW [amu]  [g mL−1 ] mp [◦ C] bp [◦ C] Vp [mm Hg] Log Kow S [mg L−1 ] KH [atm m3 mol−1 ]

BTF

4CBTF

24DCBTF

34DCBTF

3ABTF

4NBTF

3A4CBTF

3N4CBTF

35DN4CBTF

98-08-8 146.11 1.19 −29.1 103.1 38.83 3.01 140 0.017

98-56-6 180.60 1.35 −33 139.5 7.63 3.60 84.5 0.035

320-60-5 215.00 1.48 10.0 117.5 2.00 nd 11.6 nd

328-84-7 215.00 1.48 −12.5 173.5 2.36 2.39 15 0.026

98-16-8 161.12 1.30 5.5 187.5 2.00 2.15 5 nd

98-46-4 191.12 1.44 −5.0 202.8 0.21 2.81 nd nd

121-50-6 195.57 1.42 10.0 82.0 0.11 2.71 nd nd

121-17-5 225.55 1.51 −2.0 222.0 0.04 3.43 29 nd

393-75-9 270.55 1.60 57.0 >250.0 nd nd nd nd

Source of data: eChemPortal OECD database [14]. Notes: nd, not determined; CAS #, CAS number; mp, melting point; bp, boiling point; , density; Vp , vapour pressure calculated at 25 ◦ C; Log KOW , logarithm octanol/water partition coefficient; S, water solubility; KH , Henry law constant.

in different aquatic organisms is a moderate but not irrelevant fate process for substituted BTFS [16,17]. According to a classification scheme [18], BCF values for BTFS suggest a potential bioconcentration in aquatic organisms that increases with the number of substituted halogenated atoms. Biodegradation is expected to be slow, as reported for other compounds containing a large number of halogens. Furthermore, biodegradation and hydrolysis are not considered significant environmental removal processes [19]. The dinitro-functional group in dinitro herbicides dramatically decreases the molecules’ solubility in water since it forms hydrogen bonds with alkyl groups of surrounding molecules. Therefore, for example trifluralin concentration can generally be two to three orders of magnitude higher in sediments than in the surrounding environmental waters [20]. To the best of our knowledge, no investigations have been conducted so far about environmental behaviour of BTFS despite of the relatively wide industrial and commercial applications of these chemicals. Only one analytical method as been proposed for the determination of few BTFS in environmental samples to support a monitoring survey following a local pollution episode and it is later discussed [21]. In detail, this study presents two simple and reliable analytical methods based on GC–MS techniques, useful for monitoring the environmental impact on groundwater from BTF and the eight mentioned chlorinated, amino- and nitro-derivatives shown in Fig. 1. On the basis of the physicochemical properties [14], the extractions of the more volatile BTFS were carried out by Purge-and-Trap technique and the semi-volatile BTFS were preconcentrated with an automated Solid Phase Extraction (SPE) system. The quantitative separation and determination were both carried out with two different GC systems equipped with single quadrupole mass spectrometer detectors. The methods were developed using spiked Milli-Q water and then demonstrated using groundwater samples. Hence, the feasibility of the two procedures was tested on real samples collected during autumn 2008 in the Vicenza Province (Regione Veneto, Italy), which is an area polluted by event of improper industrial soil disposal. This event occurred in Valleagno area during the summer of 1977 and it is the only one reported involving an industry that produced BTFS, among other different fluorinated intermediates.

2. Experimental

in dichloromethane as internal standard and surrogate for the SPE method, respectively. Both were purchased from Supelco (Sigma–Aldrich). Moreover, the internal standard for the Purgeand-Trap method, 1,2,3-trifluorobenzene with purity 99%, was purchased from Aldrich (Sigma–Aldrich). Stock primary solutions for each compound were prepared in 2-propanol. Multiresidue working solutions were prepared by dilution of stock solutions with methanol (Purge-and-Trap extraction) and with a mixture of methanol/ethylacetate 1:1 (SPE). Stock and working solutions were stored at 4 ◦ C, except aniline-d5 solutions stored at −20 ◦ C. Aqueous solutions were prepared by spiking water with appropriate amount of working solution. Water used for chromatographic purpose was purified by a Milli-Q system (Millipore® , Billerica, MS, US) and all organic solvents (methanol, ethylacetate and 2propanol) were of pesticide grade (VWR Milan, Italy). The pH 9 Fixanal grade buffer solution (sodium tetraborate/hydrochloric acid) used was purchased from Fluka (Sigma–Aldrich). The sodium hydroxide, which was used to prepare a solution of 10% w w−1 , was purchased from Merck (Readington, NJ, US). 2.2. Purge-and-Trap extraction and concentration A commercial Velocity XPT sample concentrator (TekmarDohrmann Mason, OH, US) was used as Purge-and-Trap system, equipped with a Teledyne Aquatek liquid autosampler purchased from the same company. Approximately 20 mL of sample were transferred into specific 40 mL glass vials and closed with a PTFEfaced silicone septum screw-caps. Therefore, from the autosampler, a precise amount of sample was directly transferred into a 25 mL glass sparging vessel (purge chamber), together with the internal standard 1,2,3-trifluorobenzene properly diluted, and purged with high purity nitrogen. The purge flow was 40 mL min−1 and the total extraction time of 25 min. Then, target compounds were adsorbed at room temperature onto a Supelco trap K VOCARB 3000 (Sigma–Aldrich, St. Louis, MO, US), filled with 10 cm of CarbopackTM B, 6 cm of CarboxenTM -1000 and 1 cm of CarboxenTM -1001. After adsorption on the trap, the volatile organic compounds were desorbed by rapid heating and transferred via heated transfer line for the consequent GC injection. The trapped sample components were thermodesorbed at 250 ◦ C for 5 min with a flow of 100 mL min−1 . After each sample, the autosampler and the extractor were automatically cleaned with hot water and the trap was baked for 10 min to decrease to a negligible level of carry over.

2.1. Reagents and standards 2.3. Solid Phase Extraction and cleanup All tested benzotrifluoride compounds were purchased from Aldrich (Sigma–Aldrich, St. Louis, MO, US) as technical reagents for organic synthesis (purity ≥ 97%) because there are no environmental standards available on the market. Aniline-d5 and nitrobenzene-d5 were used in concentration of 2000 ␮g mL−1

For the determination of the semi-volatile BTFS fraction, an amount of 200 mL of sample at room temperature, with or without spiking, was prepared by adding 20 mL of buffer and 5 mL of methanol. Environmental samples were previously filtered

R. Lava et al. / Analytica Chimica Acta 804 (2013) 126–134

129

Table 2 MS conditions, linearity ranges, LOD, LOQ, coefficients of determination of calibration curves and recoveries for the two proposed methods. Retention time [min] Volatile BTFS fraction – Purge and Trap 10.79 1,2,3-Trifluorobenzene (IS) BTF 11.15 4CBTF 15.27 20.46 34DCBTF 21.15 24DCBTF

MS ion (quant)

MS ion (qual)

132 146 180 214 214

134 96 127 161 182 195 216 179 216

Semi-volatile BTFS extraction – Solid Phase Extraction 5.38 98 Aniline-d5 (IS) 34DCBTF 5.61 214 24DCBTF 6.41 214 Nitrobenzene-d5 (surrogate) 6.42 82 7.42 145 3NBTF 7.52 161 3ABTF 11.59 195 3A4CBTF 3N4CBTF 11.89 179 18.4 270 35DN4CBTF

70 71 179 216 179 216 54 128 95 191 114 142 176 197 144 225 30 143

a

Linearity range [␮g L−1 ]

LOD [␮g L−1 ]

LOQ [␮g L−1 ]

R2

Recoverya [%]

5.12 0.005–158.90 0.005–164.15 0.01–190.70 0.01–177.48

0.002 0.002 0.004 0.005

0.006 0.005 0.013 0.018

0.9972 0.9998 0.9986 0.9995

98 88 82 88

± ± ± ±

3 2 3 4

0.01 0.01

0.03 0.03

0.01 0.01 0.01 0.02 0.07

0.03 0.03 0.02 0.05 0.24

0.9980 0.9982 0.9997 0.9996 0.9999 0.9991 0.9997 0.9983

52 61 77 82 85 79 83 86

± ± ± ± ± ± ± ±

5 4 3 9 10 3 4 6

0.03–4.13 0.03–4.18 0.40 0.02–3.99 0.02–3.77 0.02–3.93 0.03–8.62 0.10–3.36

Recoveries based on triplicate injections.

at 0.45 ␮m (Sartorius, Goettingen, Germany) and the pH was eventually adjusted to 9 adding NaOH 10% (w/v) drop wise and controlled with a Crison pH-meter GLP 21 (Barcelona, Spain). 50 ␮L of surrogate nitrobenzene-d5 (1.6 ␮g L−1 ) were added to each sample. The extraction was carried out by a Gilson Aspec XL Solid Phase Extraction Station equipped with a Gilson ValvemateTM (Middleton, MA, US), using 200 mg/6 mL Oasis HLBTM cartridges (Waters, Milford MS, US). Each cartridge was conditioned with 5 mL of ethylacetate and equilibrated with 5 mL of methanol followed by 5 mL drop-wise at 1 mL min−1 of Milli-Q buffered water. The total amount of sample was loaded onto the conditioned cartridge and percolated at a flow-rate of 5 mL min−1 , avoiding letting the cartridge run dry. After the extraction, the cartridge was rinsed with 3 mL of Milli-Q water to remove possible impurities and co-retained interferences. Afterwards, each cartridge was dried under vacuum for about 1 h. The elution was performed with 6 mL (2 × 3 mL aliquots) of ethyacetate at 1 mL min−1 . A gentle stream of nitrogen was applied using a Turbovap II (Zymark, Basel, Switzerland) to concentrate the solution until a final volume of 0.5 mL. Before the analysis, an additional volume of 50 ␮L of internal standard aniline-d5 was added to each sample obtaining a final volume of 0.55 mL into 2-mL Teflon-capped screw cap vials. The vials were stored at 4 ◦ C before injection into the chromatographic system. 2.4. Chromatographic and MS conditions A Perkin Elmer Clarus 500 (Waltham, MA, US) gas chromatograph, equipped with a single quadrupole mass spectrometer detector, was used for the volatile BTFS separation. The GC–MS system was connected to the Purge-and-Trap system to permit a direct thermal desorption from the concentrator. A Restek RTXTM -624 (Bellafonte, PA, US) column was used (30 m, 0.25 mm I.D., 0.25 ␮m film) injecting 1 ␮L of sample in splitless mode. The injection temperature was 250 ◦ C and separation was carried out with the following oven programme: 40 ◦ C isothermal for 7 min, ramp of 10 ◦ C min−1 to 110 ◦ C, 5 min isothermal, ramp of 20 ◦ C min−1 to 250 ◦ C and final isothermal of 3 min; for a total run of 29 min. The instrumental and processing methods were managed by TurboMass® Perkin Elmer software. For semi-volatile BTFS, GC–MS analysis was carried out using a Thermo Finnigan Trace GC Ultra equipped with a DSQ single quadrupole mass detector (Waltham, MA, US). A Varian (30 m, 0.25 mm I.D., 0.25 ␮m film) VF-5MS column (Palo Alto, CA, US) was installed and a PTV injector set at constant temperature in splitless mode was used for the investigations. The injector was maintained at 270 ◦ C with a splitless time of 0.8 min and 2.0 mm

metal SiltekTM deactivated inlet liner from Thermo was used; 1 ␮L of sample was injected. The following oven conditions were used: an initial temperature of 75 ◦ C for 3 min, a ramp of 2.5 ◦ C min−1 from 75 ◦ C to 110 ◦ C, a ramp of 32 ◦ C min−1 until 300 ◦ C and a final isothermal of 8 min at 300 ◦ C. Helium was used as carrier gas at constant flow of 1 mL min−1 for a total run of about 24 min. The ion source and transfer line were kept at 250 ◦ C and 300 ◦ C, respectively. The instrumental and processing methods were both managed by Xcalibur® Thermo software. For both the methods, the simultaneous acquisition in full scan (FS) and selected ion monitoring (SIM) were performed. One ion of each peak signal (not always the molecular ion) was selected to quantify the response from the spectrum in SIM mode, while two other ions were chosen for the spectral characterization, as shown in Table 2. 2.5. Sample collection Three environmental samples of groundwater, collected in the Valleagno area (Vicenza Province, Italy) during autumn 2008, were analyzed in order to check the performance of the two methods described. For the semi-volatile BTFS determination, 1000 mL of samples were collected in a borosilicate amber glass bottle and analysis was performed within 96 h after sampling. In the meantime the samples were stored in a fridge at 4 ◦ C. For the volatile BTFS determination, a 250 mL borosilicate amber glass bottle was completely filled without headspace air interface and samples were extracted the same day of the sampling. 3. Results and discussion The volatile BTFS compounds were extracted by a Purge-andTrap system, while the semi-volatile BTFS were extracted, purified and concentrated with an automated SPE system. The selection of the extraction technique was based on BTFS physicochemical properties (Table 1), for which data are available in the literature, namely vapour pressure, boiling point and Henry’s law constant [14]. The determination of 34DCBTF and 24DCBTF, both with intermediate physicochemical properties, was directly compared between the two methods. 3.1. Purge-and-Trap extraction and GC–MS detection From the multiresidue working solutions, a calibration curve for the volatile BTFS with six data points over the range 0.005–20 ␮g L−1 was prepared in 40 mL vials at each batch of

130

R. Lava et al. / Analytica Chimica Acta 804 (2013) 126–134

Fig. 2. Chromatogram of a standard level on Milli-Q water spiked with volatile BTFS (final concentration: 0.016 ␮g L−1 for BTF and 4CBTF and 0.018 ␮g L−1 for 24DCBTF and 34DCBTF).

RT: 5.00 - 20.00 5.61 6.38 100

RT: 4.51 - 19.84 6.40 100 5.59 34DCBTF 24DCBTF

50 0 100

0 100

7.42 3NBTF

50 0 100

7.42

50

3NBTF

0 100

7.49 3ABTF

50

34DCBTF 24DCBTF

50

0 100

0 100 3A4CBTF

50

11.59 3A4CBTF

50 0 100

11.89

11.89 3N4CBTF

50

3N4CBTF

50

3ABTF

50

11.61

0 100

7.52

11.59 0 100

0 100 35DN4CBTF

50

0 8

10

12 14 Time (min)

a) lowest level calibration curve

16

35DN4CBTF

50

0 6

18.40

18

6

8

10

12 14 Time (min)

16

18

2

b) highest level calibration curve

Fig. 3. Chromatograms of the (a) lowest and (b) highest standard levels for semi-volatile BTFS with values of concentration in the range 13–15 ␮g L−1 and 1220–1500 ␮g L−1 , respectively. For 35DN4CBTF the signal on the first chromatogram is ≤LOD because the lowest calibration point is set at ∼60 ␮g L−1 .

R. Lava et al. / Analytica Chimica Acta 804 (2013) 126–134

34DCBTF

100

131

20.43

%

24DCBTF

3: SIR of 10 Channels EI+ 214.00 1.01e6

21.12

0

15.27

100

3: SIR of 10 Channels EI+ 180.00 3.70e6

4CBTF

% 0

13.55

100

2: SIR of 8 Channels EI+ 166.00 1.34e5

TeCE

% 0

2: SIR of 8 Channels EI+ 130.00 3.66e4

10.52

TrCE

100 % 0

11.16

100

BTF

2: SIR of 8 Channels EI+ 146.00 2.14e4

INT STD

2: SIR of 8 Channels EI+ 132.00 2.87e6

% 0

10.80

100 % 0

4.00

6.00

8.00

10.00

12.00

14.00

16.00

18.00

20.00

22.00

24.00

26.00

28.00

Time

Fig. 4. Chromatograms of the volatile BTFS from the groundwater sample collected in the Trissino municipality (TeCE = tetrachloroethylene, TrCE = trichloroethylene).

analysis. Since there are no suitable certified reference materials available, spiked samples have been used for the development and validation of the method. Hence, Milli-Q water was used and spiked with native BTFS solution. The spiked and environmental samples were directly transferred into the vials in order to avoid loss of volatile compounds by immediate cap closure. The linearity was demonstrated from 0.005–0.01 ␮g L−1 over six orders of magnitude, for a range of concentrations wider than the working range of calibration curves and not applicable for trace determination. Calibration was based on least square analysis and peak area ratios relative to internal standard were used for all quantification calculations, obtaining values of R2 ≥ 0.9972 for all the analytes considered. Two main retention time windows from 0 to 14 min (for BTF and the internal standard) and from 14 to 22 min (for 4CBTF, 24DCBTF and 34DCBTF) were set. The Purge-and-Trap technique offers no selectivity in removing analytes from the sample by gas purging. Trap selection, GC column, and MS conditions determined the level of selectivity. Thus, a careful choice of mass fragments (m z−1 ) derived from ionization of BTFS and similar molecules with a bonded CF3 group, limited interferences and increased the selectivity of target compounds. LODs and LOQs were studied in order to determine the sensitivity of the method for each analyte. LODs were calculated by comparison of the signal-to-noise ratio (S/N = 3) of the lowest detectable concentration with the average value of noise (n = 7), while LOQs were calculated considering a S/N = 10. LODs values ranged from 0.002 to 0.005 ␮g L−1 and LOQs were between 0.005 and 0.018 ␮g L−1 , as shown in Table 2. Preliminary recovery experiments were carried out by spiking groundwater at 1.59–1.91 ␮g L−1 and recoveries were all above 82%. However the calculation of extraction efficiency was not so relevant because calibration plots were obtained by adding different levels of multiresidue standard to real sample. All concentrations from the standard solutions were directly compared with the real samples and correction of recovery was automatically included. Furthermore, the use of 1,2,3-trifluorobenzene as internal standard with a final concentration of 5.120 ␮g L−1 attributed good stability and

ruggedness to the method also at very low concentration. Good intra-day repeatability (six replicates) calculated on Milli-Q water was obtained with RSDs values of 8–16% at 0.016–0.018 ␮g L−1 and 4–6% at 15.89–19.07 ␮g L−1 spiked concentrations, respectively. Fig. 2 shows a chromatogram of a standard level spiked at 0.016 ␮g L−1 of BTF and 4CBTF, and 0.018 ␮g L−1 of 24DCBTF and 34DCBTF. The main advantages of the Purge-andTrap technique, followed by focalization on a chemical absorber, are that it is completely solvent-free and is a fully automated procedure. The pre-concentration and extraction technique, which is also suggested by official and standardized protocols (i.e. EPA methods for drinking water and wastewater analysis [22]), was lately considered for the determination of organic contaminants in water samples at trace level [23,24]. The application to BTFS compounds revealed good sensitivity and especially satisfactory repeatability, considering that the technique is based on a liquid-phase purging system. 3.2. Solid Phase extraction and GC–MS detection For the determination of the semi-volatile BTFS, a linear range was calculated from five multiresidue levels prepared and stored at 4 ◦ C. Acceptable linearity in the 8–1500 ␮g L−1 range was achieved for all tested compounds and calibration was based on the least square method. The values of R2 were 0.9983–0.9999. Fig. 3 shows the chromatogram of a standard solution at the lowest and the highest levels. Four main SIM windows were chosen and set at 5–7, 7–10, 10–13 and 17–19 min. The overlapping between 3NBTF and 3ABTF peaks and between 3N4CBTF and 3A4CBTF peaks in the chromatograms were overcome considering the different masses in the SIM mode used for the quantification and qualification of area peaks, as described in Table 2. No relevant interfering compounds were detected considering the specificity of the ions selected for the experiments and the clean matrix. The ratios between the quantification and qualification ions were always used as confirmation criteria comparing the obtained spectra with

132

R. Lava et al. / Analytica Chimica Acta 804 (2013) 126–134 RT:

5.05 - 18.67 5.64

100

34DCBTF 50

24DCBTF 6.43 0

7.42

100

3NBTF 50

0

7.54

100

3ABTF 50

0

11.59

100

3A4CBTF 50

0 11.92

100

3N4CBTF 50 11.59 0 100

35DN4CBTF 50

0 6

8

10

12

14

16

18

Time (min)

Fig. 5. Chromatograms of the semi-volatile BTFS from the groundwater sample collected in the Trissino municipality.

the one obtained by injection of high concentrated standards. Six injections for each level were made for the calculation of the RSDs of spiked Milli-Q water solutions. The extraction step was optimized to achieve the trueness of the method obtaining acceptable levels of recovery, except for the dichloro-derivatives. The recoveries of 200 mL of Milli-Q water, spiked with concentrations in the range 0.11–0.20 ␮g L−1 , were: 61 ± 4% for 24DCBTF, 52 ± 5% for 34DCBTF, 82 ± 9% for 3NBTF, 85 ± 10% for 3ABTF, 79 ± 3% for 3A4CBTF, 83 ± 4% for 3N4CBTF and 86 ± 6% for 35DN4CBTF. Nitrobenzene-d5 , was used as a surrogate for controlling the entire process with an average recovery of 77 ± 3%. Aniline-d5 was used as internal standard and the area ratio for quantification, after its normalization with the area of analytes. The extraction of the semi-volatile fraction of BTFS was achieved using SPE with 6 mL Waters Oasis HLBTM cartridges filled with 200 mg of a copolymeric poly(divinylbenzene/N-vinylpyrrolidone) resin. This copolymeric sorbent allows hydrophilic–lipophilic balance, enhancing the ability of the resin to retain non-polar and polar compounds simultaneously and presented acceptable performance even when the solid phase accidentally ran dry. Conditions for

solid phase, elution solvent and pH extraction were chosen on the basis of previous in-house experience on determination of similar compounds [25]. This previous work presented an in-depth investigation of the best SPE condition for simultaneous detection of chloroaniline and chloronitrobenzene derivatives in water samples. Therefore, the choice to work at pH 9 enhanced the recoveries of amino-derivatives with weak basic properties among the target BTFS compounds. Lower recoveries were obtained for the dichloroderivatives (Table 2). Consequently, the extraction of 24DCBTF and 34DCBTF was directly compared between the two methods. More reliable results were achieved by Purge-and-Trap extraction because of the intermediate physicohemical properties concerning the volatility of 24DCBTF and 34DCBTF. According to the SPE procedure developed, a partial loss of all analytes due to evaporation cannot be prevented during the final concentration step, especially for 24DCBTF and 34DCBTF. LODs (range 0.01–0.07 ␮g L−1 ) and LOQs (range 0.02–0.24 ␮g L−1 ) were calculated in the same way as for the more volatile BTFS (Table 2). The LOQs roughly corresponded to the lowest point of the considered calibration curves.

R. Lava et al. / Analytica Chimica Acta 804 (2013) 126–134 Table 3 Concentrations (␮g L−1 ) of BTFS recorded in the samples from Valleagno area analyzed in triplicate. Trissino groundwater Volatile BTFS fraction – Purge and Trap 0.098 ± 0.005 BTF 4CBTF 11.9 ± 0.7 6.4 ± 0.4 34DCBTF 0.22 ± 0.04 24DCBTF

Brendola groundwater ≤LOD 0.047 ± 0.005 0.025 ± 0.006