Microchim Acta DOI 10.1007/s00604-014-1290-x

ORIGINAL PAPER

Determination of eight pesticides of varying polarity in surface waters using solid phase extraction with multiwalled carbon nanotubes and liquid chromatography-linear ion trap mass spectrometry Soraya Dahane & María Dolores Gil García & Ana Uclés Moreno & María Martínez Galera & María del Mar Socías Viciana & Aicha Derdour

Received: 18 March 2014 / Accepted: 13 May 2014 # Springer-Verlag Wien 2014

Abstract We describe a MWCNT-based method for the solid-phase extraction of eight pesticides from environmental water samples. The analytes are extracted from 100 mL samples at pH 5.0 (containing 5 mmol L−1 of KCl) by passing the solution through a column filled with 20 mg of multiwalled carbon nanotubes. Following elution, the pesticides were determined by LC and electrospray ionization hybrid quadrupole linear ion trap MS. Two selected reaction monitoring transitions were monitored per compound, the most intense one being used for quantification and the second one for confirmation. In addition, an information-dependent acquisition experiment was performed for unequivocal confirmation of positive findings. Matrix effect was not found in real waters and therefore the quantitation was carried out with calibration

graphs built with solvent based standards. Except for cymoxanil, the detection and quantitation limits in surface waters are in the range from 0.3 to 9.5 ng L−1 and 1.6 to 45.2 ng L−1, respectively. Recoveries from spiked ultrapure water are ~100 %, except for the most polar pesticides methomyl and cymoxanil. The same behavior is found for real water samples (except for phosalone). The relative standard deviation is <10 % in all cases. Keywords Pesticides . Solid-phase extraction . Multiwalled carbon nanotubes . Liquid-chromatography-mass spectrometry

Introduction Electronic supplementary material The online version of this article (doi:10.1007/s00604-014-1290-x) contains supplementary material, which is available to authorized users. M. D. Gil García : A. Uclés Moreno : M. Martínez Galera (*) : M. M. Socías Viciana Department of Chemistry and Physics, Area of Analytical Chemistry, University of Almería, E-04120 La Cañada de San Urbano, Almería, Spain e-mail: [email protected] M. D. Gil García : A. Uclés Moreno : M. Martínez Galera : M. M. Socías Viciana Centro de Investigación en Biotecnología Agroalimentaria, BITAL, La Cañada de San Urbano, E-04120 Almería, Spain M. D. Gil García : A. Uclés Moreno : M. Martínez Galera : M. M. Socías Viciana Campus de Excelencia Internacional Agroalimentario CeiA3, La Cañada de San Urbano, Almería, Spain S. Dahane : A. Derdour Department of Chemistry, University of Oran, Oran, Algeria

Since removal of organo chlorine insecticides from use, organophosphate insecticides have become the most widely used insecticides available today. More than forty of them are currently registered for use and all run the risk of acute and subacute toxicity. All of them apparently share a common mechanism of cholinesterase inhibition by disrupting the enzyme that regulates acetylcholine, a neurotransmitter, and can cause similar symptoms. Therefore, exposure to one or multiple organophosphates can lead to serious additive toxicity [1]. N-methyl carbamate insecticides are widely used and they share with organophosphates the capacity to inhibit cholinesterase enzymes and, therefore, they share similar symptomatology during acute and chronic exposures [2]. Fungicides also have wide use in agriculture and, although the acute toxicity of fungicides to humans is generally considered to be low, they must also be controlled in water

S. Dahane et al.

sources, following the European Union Directive on drinking water quality (98/83/EC) [3]. Before determining contaminants in water samples, a preliminary sample pre-concentration and clean step must be carried out, the most popular technique being solid phase extraction (SPE). In environmental analysis, particularly in the isolation and pre-concentration of pesticides and their transformation products from aqueous samples, the use of hydrophobic non-selective sorbents, such as C18-bonded silica and styrene-divinylbenzene co-polymers has been considered the best option [4]. However, new advances have been developed, related to research on novel sorbent materials for SPE. Thus, F Augusto et al. [5] published a recent review devoted to new materials and trends in sorbents for SPE and K Pyrzynska [6] has reviewed the use of nanomaterials (metallic, silica and carbon-based) in sample preparation. Carbon nanotubes (CNTs) are among the most used nanomaterials for SPE [6, 7]. They interact strongly with organic molecules via non-covalent forces such as hydrophobic, electrostatic and van der Waals interactions, hydrogen bonding and π stacking and these interactions, along with theirs hollow and layered structures which involve large surface-to-volume ratios, makes them good candidates for use as sorbents, specially multi wallet carbon nanotubes (MWCNTs) [6]. CNTs have been applied for the extraction of both organic and inorganic compounds. As for the first ones, CNTs have been widely used for extraction of pesticides, as can be seen from the information gathered by LM Ravelo-Pérez et al. [7], K Pyrzynska [8] and BT Zhang et al. [9]. Thus, pesticides with different chemical properties were pre-concentrated from water samples [10–16], soils [17, 18], fruit juices [19], virgin olive oils [20] or vegetables and fruits [21]. In response to environmental problems, due to the bioaccumulation and global transport of volatile apolar pesticides, the modern ones are more polar, thermo-labile and less volatile compounds [22]. Therefore, the analytical approaches to determine pesticides have been adapted to these properties and, even though gas chromatography (GC) with mass spectrometry (MS) detection is still used [23, 24], the coupling of liquid chromatography (LC) with mass spectrometry (MS) via atmospheric pressure ionization (API) is the best choice to determine current pesticides. Although LC-tandem MS (LC-MS2) techniques, such as triple quadrupole (QqQ) and ion trap (IT) are routinely used in environmental analysis, more recent approaches in LC-MS2 such as linear ion trap (LIT), new generation QqQs and hybrid instruments such as quadrupole-time of flight (Qq-TOF) and Q-linear trap (Qq-LIT), which offers advantages such as high scanning speed, accurate mass measurement (Qq-TOF) and increased sensitivity (LIT and new-generation QqQ), are being increasingly used.

The hybrid quadrupole-linear ion trap (Qq-LIT) is a QqQ in which the third quadrupole can be operated as either a quadrupole or an ion trap. In this way, the same analyzer can be run in two different modes, retaining the classical QqQ scan functions, such as selected reaction monitoring (SRM), product ion, neutral loss and precursor ion, while providing access to sensitive IT experiments. This allows very powerful scan runs which can be combined in one single experiment when performing information-dependent data acquisition (IDA). The QqLIT enables extraction of nearly an order of magnitude more ions into the linear ion trap compared with the conventional 3D ion traps, when both are operated at the same ion density. This is because the LIT, in contrast to the 3D ion trap, omits the quadrupolar electric field in the axial direction which means that fewer ions are lost during the process of filling and emptying the trap [25]. As a consequence, it enables higher sensitivity for product-ion scan and offers some extra scans possibilities such as enhanced full scan (EMS), enhanced product ion (EPI) and MS3 [26]. However, in spite of these advances, determination and correct quantification of a large number of pesticides of different chemical classes at low concentrations and/or in complex matrices still remains a challenge. The main goal of the present paper was to investigate the potential of MWCNTs as sorbents for extracting eight pesticides of different polarity including organophosphate insecticides (methidation, parathion-methyl, malathion, phosalone and diazinon), fungicides (cymoxanil and penconazole) and one carbamate (methomyl), from environmental water samples as well as to exploit the improvements in sensitivity and specificity provided by the LC-hybrid MS QqLIT detection for quantitating and confirming the above mentioned pesticides.

Experimental Reagents Pesticide standards of methomyl (MET), cymoxanil (CYM), methidathion (MET), parathion-methyl (PMT), malathion (MAT), penconazole (PEN), phosalone (PHO) and diazinon (DIA) with >99 % purity, were purchased from Riedel-de Haën (Seelze, Germany, www.riedeldehaen.com). Chemical structures and principal physicochemical properties of selected pesticides are included in the Electronic Supplementary Material (ESM) (See Table S1). Individual stock standard solutions of the target analytes (400 mg L−1) were prepared by weighing and dissolving the corresponding compounds in acetonitrile and stored in the dark at −20 °C, being stable for at least 3 months. Working standard solutions were prepared daily by appropriate dilution of individual stock standard solutions in acetonitrile-ultrapure water (10:90, v/v).

Determination of pesticides in environmental waters using SPE with MWCNTs and LC-QqLIT

Real water sample collection Two types of surface waters (river and natural dam) were used to evaluate the ability of MWCNTs as sorbent for preconcentration of the eight pesticides in environmental waters. Two water samples were collected from rivers located in Almería (Spain), one sample from Nacimiento River (sample NCR) and the other from Andarax River (sample AXR). Nacimiento River flows through a scarcely populated area and discharges into the Andarax River, which flows through a populated area. The sample NCR was picked up in the Nacimiento River at the top of the stream, whereas the sample AXR was picked up in the Andarax River near a wastewater treatment plant which discharges its effluent in the river flowing. In addition, two water samples were collected from two natural dams (samples BND and GND), both located in Almería (Spain). The sample BND was picked up in a dam whose water comes from the Andarax River and is used to irrigate crops, whereas the sample GND was picked from a dam whose water comes from the rainfall and is used for drinking by animals living in the area. All real water samples were collected using 1 L amber bottles, with Teflon lined caps, which were previously rinsed with the sample water on site, and were immediately carried to the laboratory. Both, river and natural dam water samples were filtered through 0.45 μm cellulose filters and stored at 4ºC in refrigerator. Solid-phase extraction The selected pesticides were extracted from environmental water samples by SPE using MWCNTs as sorbent. A Büchi Vac V-500 (Flawil, Switzerland, www.buchi.com) vacuum pump, connected to an extraction manifold of Waters (Milford, MA, USA, www.waters.com) was used in this process. MWCNTs with an average external diameter (O.D.) of 6– 13 nm, length (L) of 2.5–20 μm and 98 % carbon basic were acquired from Sigma–Aldrich (Madrid, Spain, www. sigmaaldrich.com/spain.html). The SPE cartridges packed with 20 mg of MWCNTs sorbent were prepared in the laboratory using 1 mL polypropylene body syringes and two polyethylene frits, both from Supelco (Madrid, Spain, www. sigmaaldrich.com/spain.html), to retain inside the sorbent material. To activate the sorbent, the MWCNT-SPE cartridge was firstly conditioned with 5 mL of a mixture acetone:n-hexane (50:50 v/v) and 5 mL of ultrapure water adjusted at pH 5.0±0.1 at a flow rate of 1 mL min−1. Then, 100 mL of water sample, previously filtered, was adjusted to pH 5.0 ± 0.1 and both potassium chloride 0.005 mol L−1 and 1 % methanol (organic modifier) were added to the sample prior pre-concentration.

The water sample was pre-concentrated onto the MWCNT cartridge at a flow rate of 1 mL min−1 and the sorbent was dried by successively passing air through it for 5 min and then N2 for another 5 min. The retained analytes were eluted with 3 mL of a mixture acetone:n-hexano (50:50, v/v) and the eluate was evaporated to dryness using a N2 stream. Furthermore, the obtained residue was dissolved in 1 mL of acetonitrile:water (10:90, v/v) and finally, 10 μL of the filtered extract were injected. LC-QqLIT-MS/MS procedure The chromatographic separation of the pesticides was achieved in an Agilent 1200 LC (Agilent Technologies, CA, USA www.agilent.com) provided with a binary pump. The analytical column was a ZORBAX Eclipse XDB C8 (150 mm×4.6 mm, 5 μm particle size) from Agilent. The LC mobile phase consisted of a mixture of acetonitrile (solvent A) and LC-grade water containing 0.1 % formic acid (solvent B). The elution gradient program was as follows: initially 0.5 min of 10 % A, 9 min linear gradient to 100 % A, 5 min of 100 % A, 0.1 min to return to initial conditions and finally initial conditions were reached in 10 min. The flow rate of the mobile phase was set constant at 0.4 mL min−1 during the whole process and the injection volume was 10 μL. The LC was connected to a hybrid triple quadrupole-linear ion trap (QqLIT) mass spectrometer 5500 QTRAP® (AB Sciex Instruments, CA, USA, www.absciex.com) with an electrospray interface (ESI) which was operated in positive ionization mode (PI). The Turbo Ion Spray source settings were: Ion Spray Voltage (IS) 5,500 V, Source Temperature (TEM) 500 °C, Curtain Gas (CUR) 30 psi, Collision Gas (CAD) Medium, both Ion Source Gas 1 (GS1) and Ion Source Gas 2 (GS2) 50 psi. Nitrogen was used as nebulizer gas and collision gas. For the SRM mode, the optimization was performed by direct injection of individual standard solutions of each pesticide in methanol at 0.1 mg L−1. The operational MS/MS parameters for each pesticide such as declustering potential (DP), entrance potential (EP) for precursor ions, collision energy (CE) and collision cell exit potential (CXP) for product ions, for each pesticide, are included in ESM (See Table S2). In the IDA method, only the main SRM transition for each substance is recorded and, therefore, 8 transitions were monitored, all of them in positive ion mode. This SRM experiment was recorded by using the Scheduled MRM option, setting the target scan time at 1 s with a MRM detection window of 40s. Q1 was set at low resolution and Q3 at unit. Regarding the IDA criteria, the intensity threshold was set to 500 cps and without exclusion after dynamic background subtraction of the survey scan. EPI scans were recorded at DP=100 V at one collision energy value CE=25 V, but using the collision energy spread (CES=5 V) and the LIT scanning from 80 to

S. Dahane et al.

450 amu at a scan rate of 10,000 amu.s−1. The dynamic filltime option was selected on the ion trap and EPI scans were monitored in each SRM-IDA-EPI cycle, being the complete SRM-IDA-EPI cycle time of 1.58 s. The ABSciex Analyst software was used for data acquisition and processing.

the reference spectrum of the library and the spectrum obtained in the samples being given by the fit, reverse fit (RevFit) and purity values provided by the software. The confirmation criteria for identification of the target compounds were a fit value or purity higher than 70 %. Optimization of the SPE procedure

Results and discussions Optimization of LC-QqLIT-MS/MS method The optimization of the chromatographic separation of the eight pesticides was performed by modification of a previously published method [27]. For all the pesticides, positive electrospray ionization (PI) mode showed higher response and, thus the protonated molecule [M + H+] was selected as the precursor ion. In order to comply with the EU requirements the two most intense transitions (SRM1 and SRM2) were selected for each pesticide [28] (See Table S2). The SRM1 was used for quantitation purposes, whereas the two SRM1 and SRM were used for identification. In addition, in order to avoid overestimations or false positive findings in quantitative analysis, other criteria were used for identification: (i) the pesticide tR in the real sample must be within ±2 % the pesticide tR in the standards and (ii) the relative abundances of the two selected SRM transitions (SRM2/SRM1) in the real samples must be between±20 % of the SRM2/SRM1 ratios in the analytical standards [29]. When working with standards, the abundances of the two selected fragments (SRM1 and SRM2) were similar for most analytes, with SRM2/SRM1 between 0.84 and 0.70, except for cymoxanil and phosalone with SRM2/SRM1 <0.3. In these cases, the SRM2 transition cannot be used for identification at low concentration levels, a drawback that can be overcome by the IDA approach, as EPI spectra can be obtained at very high sensitivity levels providing confirmation information at low concentrations [30]. Thus, working in the IDA mode provided further structural information for confirmation of cymoxanil and phosalone, as well as extra confirmation for the rest of the analytes. The IDA approach used combined the SRM acquisition mode with an enhanced product ion scan (EPI). EPI spectra were acquired using the collision energy spread (CES) with CE=25 V and CES of 5 V, in such a way that the trap was automatically filled with fragment ions obtained at three different energies (CE-CES, CE, CE + CES) [29], which were 20, 25 and 30 V, and enough fragmentation was obtained for all pesticides (See Figure S1, ESM). The EPI spectra generated for each pesticide were matched against a mass spectral library, the good agreement between

Selection of type and volume of eluent solvent CNTs have shown high adsorption capacity for organic compounds, the complete elution of the retained analytes from the CNT sorbent being very difficult in some cases [7, 31]. Therefore, in order to choose the best elution solvent, preliminary studies were carried out using methanol, acetonitrile, acetone, n-hexane, dichloromethane and ethyl acetate. In all these experiments, 10 mL of ultrapure water, spiked with 20 ng of the pesticides, were passed through the SPE cartridge containing 20 mg of MWCNTs sorbent, previously conditioned with 5 mL of the selected elution solvent followed by 5 mL of ultrapure water. The elution of the retained analytes was performed with 2 mL of the elution solvent. It was found that all pure solvents assayed gave poor elution efficiency, except acetone that showed recoveries higher than 60 % for most compounds except for phosalone, whose recovery was lower than 30 %. This behaviour can be due to the low polarity of this pesticide (log Kow =4.01), which remains retained onto the sorbent. With the aim of improving the elution of phosalone, a mixture acetone:n-hexane (50:50, v/v) was assayed and a recovery of 70 % was achieved for this pesticide and also for the most polar methomyl. Therefore, acetone and n-hexane (50:50, v/v) was chosen as the elution solvent in the following experiments (Fig. 1a). Next, the elution volume of the selected solvent (acetone:nhexane 50:50, v/v) was studied between 2 and 7 mL. The obtained results (Fig. 1b) showed that a volume of 3 mL of acetone/n-hexane (50:50 v/v) allowed the complete elution of all pesticides from the MWCNTs sorbent, with recoveries of 80 % for methomyl and phosalone and about 100 % for the other ones. Optimization of sample pH, organic modifier and ionic strength The pH of the water sample exerts an important effect on the SPE retention because it defines the molecular state of the target compounds (ionic or neutral state) and therefore, their adsorption onto the MWCNTs sorbent. Most selected pesticides do not show acidic-basic properties, except penconazole, diazinon and cymoxanil with pKa values of 1.5, 2.6 and 9.3, respectively. Therefore, the effect of sample pH was evaluated in the range 3.0–9.0. Extreme pH

Determination of pesticides in environmental waters using SPE with MWCNTs and LC-QqLIT Fig. 1 Optimization of SPE parameters: a) sample pH, b) salt content (KCl), c) percentage of organic modifier (Methanol) and (d) water sample volume (breakthrough) on the retention into MWCNTs cartridges of: (1) Methomyl, (2) Cymoxanil, (3) Methidathion, (4) Malathion, (5) Penconazole, (6) Diazinon, (7) Parathion-methyl and (8) Phosalone

values were not considered due to the possible hydrolysis of some pesticides under strong acidic or basic conditions. Experiments were performed using 50 mL of ultrapure water containing 20 ng of each pesticide, adjusted at different pH values with hydrochloric acid 0.1 mol L−1 or sodium hydroxide 0.1 mol L−1. In general, recoveries were similar between pH 3 and 7, except for methomyl which showed better recoveries at pH 5 and, therefore, in the light of these results (Fig. 2a), pH 5 was considered the optimum value for the water sample. The influence of the salt content in the water sample on the extraction efficiency, was also investigated. The addition of salt to the water sample usually improves the retention of analytes into SPE sorbents due to the salting out effect [32]. The increase in the pesticide extraction can be explained by the engagement of water molecules in the hydration spheres around the ionic salt and hence in the reduction of the water concentration available to dissolve the analytes [33]. However, at higher salt concentration the interaction between the salt ions and the analytes predominates, thus reducing their capacity to be retained in the sorbent [33]. The effect of the salt content in the water sample was examined by analyzing ultrapure water samples containing different concentrations of KCl (0–0.2 mol L−1). The extraction efficiencies increased or remained constant for most pesticides up to concentration of 0.005 mol L−1 KCl and higher concentrations of salt in the water sample decreased the retention for most of them (Fig. 2b). Thus, in all cases, 0.005 mol L−1 KCl was added to the water sample before SPE extraction. Finally, the addition of organic modifier was evaluated in order to improve pesticide recoveries. With this aim, 50 mL of ultrapure water samples spiked with 20 ng of each pesticide (adjusted at pH 5.0±0.1 and containing 0.005 mol L−1 KCl) were modified by the addition of different amounts of methanol (0–3 % v/v) and 1 % was selected as a compromise between the optimal conditions for the most polar pesticides and the less polar ones (Fig. 2c). MWCNTs amounts and flow rate The effect of the amount of sorbent on the pesticide retention into SPE cartridge was investigated using 20 and 50 mg of

MWCNTs. Higher amounts of sorbent were not checked because the small particle size of MWCNTs we used results in compactness of the sorbent, which difficult the pass of the water sample through it, in addition to the volume of organic solvent required for elution increases significantly [31]. 20 mg of MWCNTs were selected as a compromise as the best conditions for methomyl (50 mg with recovery 91 %) were the worse ones for phosalone (50 mg with recovery 0 %). On the other hand, in a previous work we found that the small particle size of MWCNTs reduces the sample flow rate through the SPE-sorbent and this fact leads to a significantly increasing of the pre-concentration time [31]. The above mentioned study showed that flow rates higher than 1 mL min−1 led to low recoveries and, therefore, in this case the flow rate was restricted to this value. Sample loading volume (Breakthrough) The breakthrough volume was estimated by analyzing different volumes of ultrapure water, all of them containing 20 ng of each pesticide. The retention of the analytes in the sorbent remained constant up to a volume of 500 mL (Fig. 2d), except for methomyl, which eluted completely when more than 250 mL of water were passed through the cartridge. Although 250 mL could be considered as the breakthrough volume, due to the high sensitivity of the LC-QqLIT method and with the aim of reducing the time spent in this step, 100 mL of water were chosen for further studies. Method validation Detection and quantitation limits The instrumental detection (IDL) and quantitation (IQL) limits were estimated by direct injection of mixtures of pesticide standards in solvent (acetonitrile:ultrapure water, 1:9, v/v) at low concentrations. The IDLs were determined as the lowest pesticide concentration whose SRM2 transition showed a S/N=3 whereas the IQLs were determined as the lowest pesticide concentration whose SRM1 transition showed a S/N=10 (See Table S3, ESM). The IDLs achieved

S. Dahane et al. Fig. 2 Optimization of SPE parameters: a) sample pH, b) salt content (KCL), c) percentage of organic modifier (Methanol, (2) Cymoxanil, (3) Methidathion, (4) Malathion, (5) Penconazole, (6) Diazinon, (7) Parathion-methyl and (8) Phosalone

for all pesticides were ranged from 0.02 to 0.25 μg L−1 and the IQL were ranged from 0.07 to 0.93 μg L−1. In addition, detection (LOD) and quantitation (LOQ) limits were calculated using the criterion described above but analyzing SPE blank extracts of four different environmental water samples (two river and two dump waters) spiked at low concentrations of pesticides (See Table S3, ESM). The LOD and LOQ were similar for all environmental waters but slightly higher than in solvent for most pesticides (LOD between 0.03 and 0.95 μg L−1 and LOQ between 0.16 and 4.52 μg L−1), except for cymoxanil whose LOD and LOQ calculated in real waters were significantly higher than in solvent. Taking into account that the pre-concentration factor achieved by SPE-MWCNTs was 100, the LODs and LOQs of the method in the water samples were 121.0 and 541.4 ng L−1 for cymoxanil and between 0.3 and 9.5 ng L−1 and between 1.6 and 45.2 ng L−1, respectively for the other pesticides. The SPE-MWCNT method allowed the quantitation of the selected pesticides in all types of environmental waters at concentration levels lower than the ones established by the EU, except for cymoxanil, as can be deduced from their corresponding LOQ values. Linearity and matrix effect The linearity in the response was studied by using standard solutions of the target analytes prepared in solvent, ranging from the IQL of each analyte up to 200 μg L−1. The response was linear with the concentration between the corresponding

IQL and 10 μg L−1 for diazinon, 50 μg L−1 for cymoxanil, malathion and methomyl, 100 μg L−1 for methidathion, parathion-methyl and phosalone and 200 μg L −1 for penconazole, with R2 >0.99 (See Table S4, ESM). On the other hand, the effect of the co-eluting residual matrix components present in the real water samples was evaluated by comparing the slopes of calibration graphs obtained using standards prepared in solvent and blank extracts of environmental water (river and dump waters) by means of a t-test [34]. Significant differences in the slopes of calibration graphs were not found for any pesticide and we stated that matrix effect was absent for all pesticides in the four environmental waters assayed. Therefore, quantitation of pesticides in real water samples was performed using external calibration curves with standards prepared in solvent. Accuracy and precision studies The recovery of the pesticides was evaluated by spiking, in triplicate, ultrapure and environmental waters at two concentration levels (0.05 and 0.10 μg L−1 for diazinon and 0.25 and 0.50 μg L−1 for the other pesticides). Recoveries were about 100 %, except for methomyl and cymoxanil (Table 1), whose low recoveries can be explained by their high water solubility (See Table S1, ESM), being only partially retained. The precision of the overall method was calculated as the relative standard deviation (RSD %) of three replicate samples giving satisfactory results (Table 1), with values ranging from 3.0 to 9.3 %. To demonstrate the applicability of SPE-MWCNT-LCQqLIT-MS/MS method, the two river waters and the two

Determination of pesticides in environmental waters using SPE with MWCNTs and LC-QqLIT Table 1 Mean recovery (%) and RSD (%) obtained in the pre-concentration of pesticides in ultrapure and environmental waters, using MWCNTs as sorbent Pesticides

Ultrapure water

Sample NCR

Sample ANX

Sample BNT

Sample GUA

0.25 (μg L−1)

0.50 (μg L−1)

0.25 (μg L−1)

0.50 (μg L−1)

0.25 (μg L−1)

0.50 (μg L−1)

0.25 (μg L−1)

0.50 (μg L−1)

0.25 (μg L−1)

0.50 (μg L−1)

Methomyl Cymoxanil Methidathion Malathion Penconazole Diazinona

51 (7) 80 (6) 100 (5) 110 (5) 100 (4) 106 (3)

54 (6) 59 (4) 95 (5) 122 (6) 86 (4) 121 (4)

43 (5) 86 (10) 98 (8) 86 (8) 66 (9) 91 (6)

50 (4) 57 (7) 98 (6) 97 (6) 69 (7) 58 (4)

49 (6) 79 (8) 84 (7) 77 (6) 87 (6) 111 (3)

52 (6) 71 (8) 86 (5) 107 (4) 102 (4) 96 (4)

60 (5) 71 (7) 93 (4) 89 (5) 86 (7) 94 (7)

53 (6) 68 (5) 86 (4) 107 (4) 73 (5) 70 (5)

56 (7) 74 (6) 84 (6) 109 (6) 87 (6) 71 (5)

58 (7) 68 (4) 90 (5) 76 (3) 91 (6) 72 (5)

Parathion-methyl Phosalone

89 (6) 115 (5)

81 (5) 81 (3)

81 (8) 98 (7)

83 (6) 89 (4)

79 (6) 65 (8)

72 (5) 64 (8)

95 (4) 57 (6)

79 (4) 57 (4)

84 (6) 58 (7)

72 (4) 60 (7)

a

0.05 and 0.1 μg L−1 RSD, relative standard deviation (n=3) between parentheses

dump waters, spiked at the same concentration levels that ultrapure water, were analyzed in triplicate. Firstly, retention times of the eight pesticides in the four real water samples were within ±2 % of the corresponding tR in the standards and the relative abundances of the two selected transitions were within ±20 % of the two SRM ratios in the analytical standards (See Table S2, ESM). In addition, the EPI spectra acquired during the analysis of real water samples were identified by the spectral library with a match quality of 70 %. As for recoveries, the target pesticides kept in real waters the same behaviour as in ultrapure water but, a significant decrease in the recoveries of the less polar pesticide phosalone was found in three of the four real water samples, which can be due to the matrix components existing in these water samples, according to the sampling sites. Humic acids (HAs), with 26.3 % of aromatic carbon (more than fulvic acids), can interact with electron-rich sites of MWCNTs surfaces via π-π interactions [35] and compete with organic molecules by direct site competition and pore blockage on MWCNTs. In addition, due to its high aromatic carbon percentage, HAs can interact with the more hydrophobic compounds, which would reduce their adsorption to MWCNTs. In the light of these results, we propose this last mechanism to explain the low recoveries obtained for phosalone in the more complex environmental water. Although it seems a contradiction between the low recoveries obtained for phosalone and the absence of matrix effect, this behaviour can be explained because most organic matter was removed in the SPE procedure (by interaction with electron-rich sites of MWCNTs surfaces via π-π interactions) and, therefore, it was absent in the blank extracts when the matrix matched calibration graphs were built. However, when the pesticides were added to the real water samples, they were preferably adsorbed onto the MWCNTs whereas the organic matter present in these samples interacted with the most apolar

pesticide phosalone reducing its adsorption to MWCNTs. This demonstrates that comparing the slopes of calibration graphs built in solvent and in blank extract of real samples is not always a safe procedure to check the effect of matrix components and the standard addition method should be used when there is a risk for any analyte. In this way, a deep study about the physico-chemical properties of analytes should be always undertaken as the first step when developing an analytical method.

Comparative study of the MWCNTs-LC-QqLIT method According to the literature, the pre-concentration of the target pesticides in aqueous samples was carried out with different conventional SPE sorbents, the most frequently used being Oasis HLB although another sorbents such as Chroma bond or even Sep-Pak C18 cartridges in tandem have been used (See Table S5, ESM). As can be seen the LODs obtained with our method (using 20 mg of MWCNTs) were in the same order or lower than those obtained using the above mentioned sorbents. The recoveries were also in the same order, except for the most polar pesticides methomyl and cymoxanil. In fact, our previous studies showed recoveries near 90 % for these two pesticides when using 50 mg of sorbent, but phosalone was not recoveried at all (0 % recovery). Therefore, a compromise between both situations was chosen. As for methods using MWCNTs, the high sensitivity provided by the QqLIT detection allowed to use less quantity of sorbent, unlike the methods using this sorbent and less sensitive detection, such as UV (300 mg) or MS (100 mg). Also, the highly sensitive QqLIT detection allowed working with low sample volumes, thus allowing low pre-concentration factors (100 in our work v.s. 250–1,000 in the other cases). An additional advantage provided by the QqLIT detection is

S. Dahane et al.

the absence of matrix effect, which is present in the methods using LC-MS. However, in addition to the analytical performance, it is also interesting to take into account the cost of the adsorbent. In this sense, the price of one Oasis HLB cartridge was about 7.5 euros, whereas the price of one MWCNTs cartridge used in this work was 4.9 euros, including the polypropylene syringe and frits, which involves a cost reduction of 30 %. Moreover, MWCNTs cartridges used to pre-concentrate solvent or clean water samples can be re-used up to ten times, after washing with 5 mL methanol, whereas conventional cartridges are usually discarded.

Conclusions In the present paper, SPE cartridges packed in the laboratory with 20 mg of MWCNTs were successfully used to pre-concentrate eight pesticides in surface water samples. High efficiency was obtained except for the most polar ones methomyl and cymoxanyl, whose recoveries were lowered to get acceptable recoveries for the most apolar phosalone. We can conclude that differences in the polarity of the analytes are not compatible with good recoveries using MWCNTs. The pesticides were analyzed by liquid chromatography coupled to a hybrid triple quadrupole-linear ion trap-mass spectrometer at trace levels working in SRM and IDA modes. The developed methodology was applied to the analysis of two river waters and two natural dam water samples, spiked with the target pesticides, and the two criteria to avoid overestimations (tR and SRM2/SRM1 in real samples) were accomplished. The IDA mode provided additional EPI spectra at low concentration levels, thus allowing reliable identification at the quantitation level for those pesticides with SRM2/SRM1 <0.3 (cymoxanyl and phosalone) and providing additional confirmation for the other ones. The interaction of HAs with the more hydrophobic compound (phosalone) reduced its adsorption to MWCNTs. Comparing the slopes of calibration graphs is not always a safe procedure to check matrix effect and the standard addition method should be used when there is risk for any analyte, which could be predicted from its physico-chemical properties.

Acknowledgments The authors grateful to Agencia Española de Cooperación Internacional (AECI, Acción Integrada A1/035959/11) for financial support. Soraya Dahane thanks to Agencia Española de Cooperación Internacional for a research fellowship associated to the Acción Integrada A1/035959/2011.

References 1. Roberts JR, Reigart JR (2013) Pyrethrins and Pyrethroids in : Recognition and Management of Pesticide Poisonings, 6th edn., U.S. Environmental Protection Agency, pp 38–42. http://npic.orst. edu/RMPP/rmpp_ch4.pdf 2. Roberts JR, Reigart JR (2013) Organophosphate Insecticides, in: Recognition and Management of Pesticide Poisonings, 6th edn., U.S. Environmental Protection Agency, pp 43–55. http://npic.orst. edu/RMPP/rmpp_ch5.pdf 3. Council EU (1998) Directive on the quality of water intended for human consumption, 98/83/CE. European Union, Brussels 4. Mourão Rodrigues A, Ferreira V, Vale Cardoso V, Ferreira E, Benoliel MJ (2007) Determination of several pesticides in water by solid-phase extraction, liquid chromatography and electrospray tandem mass spectrometry. J Chromatogr A 1150:267–278. doi:10. 1016/j.chroma.2006.09.083 5. Augusto F, Hantao LW, Mogollón NGS, Braga SCGN (2013) New materials and trends in sorbents for solid-phase extraction. TrAC Trends Anal Chem 43:14–23. doi:10.1016/j.trac.2012.08.012 6. Pyrzynska K (2013) Use of nanomaterials in sample preparation (2013). TrAC Trends Anal Chem 43:100–108. doi:10.1016/j.trac. 2012.09.022 7. Ravelo-Pérez LM, Herrera-Herrera AV, Hernández-Borges J, Rodríguez-Delgado MA (2010) Carbon nanotubes: solid-phase extraction. J Chromatogr A 1217:2618–2641. doi:10.1016/j.chroma. 2009.10.083 8. Pyrzynska K (2011) Carbon nanotubes as sorbents in the analysis of pesticides. Chemosphere 83:1407–1413. doi:10.1016/j. chemosphere.2011.01.057 9. Zhang BT, Zheng X, Li HF, Lin JM (2013) Application of carbon based nanomaterials in sample preparation: a review. Anal Chim Acta 784:1–17. doi:10.1016/j.aca.2013.03.054 10. Zhou Q, Wang W, Xiao J (2006) Preconcentration and determination of nicosulfuron, thifensulfuron-methyl and metsulfuron-methyl in water samples using carbon nanotubes packed cartridge in combination with high performance liquid chromatography. Anal Chim Acta 559:200–206. doi:10.1016/j.aca.2005.11.079 11. Zhou Q, Xiao J, Wang W (2007) Comparison of multiwalled carbon nanotubes and a conventional absorbent on the enrichment of sulfonylurea herbicides in water samples. Anal Sci 23:189–192 12. Niu H, Shi Y, Cai Y, Wei F, Jiang G (2009) Solid-phase extraction of sulfonylurea herbicides from water samples with single-walled carbon nanotubes disk. Microchim Acta 164:431–438. doi:10.1007/ s00604-008-0079-1 13. Zhou Q, Ding Y, Xiao J (2006) Sensitive determination of thiamethoxam, imidacloprid and acetamiprid in environmental water samples with solid-phase extraction packed with multiwalled carbon nanotubes prior to high-performance liquid chromatography. Anal Bioanal Chem 385:1520–1525. doi:10.1007/s00216-006-0554-7 14. Pyrzynska K, Stafiej A, Biesaga M (2007) Sorption behavior of acidic herbicides on carbon nanotubes. Microchim Acta 159:293– 298. doi:10.1007/s00604-007-0739-6 15. Wang S, Zhao P, Min G, Fang G (2007) Multi-residue determination of pesticides in water using multi-walled carbon nanotubes solidphase extraction and gas chromatography–mass spectrometry. J Chromatogr A 1165:166–171. doi:10.1016/j.chroma.2007.07.061 16. Asensio-Ramos M, Hernández-Borges J, Ravelo-Pérez LM, Rodríguez-Delgado MA (2008) Simultaneous determination of seven pesticides in waters using multi-walled carbon nanotube SPE and NACE. Electrophoresis 29:4412–4421. doi:10.1002/elps. 200800254 17. Asensio-Ramos M, Hernández-Borges J, Borges-Miquel TM, Rodríguez-Delgado MA (2009) Evaluation of multi-walled carbon nanotubes as solid-phase extraction adsorbents of pesticides from

Determination of pesticides in environmental waters using SPE with MWCNTs and LC-QqLIT

18.

19.

20.

21.

22.

23.

24.

25.

26.

agricultural, ornamental and forestal soils. Anal Chim Acta 647:167– 176. doi:10.1016/j.aca.2009.06.014 Min G, Wang S, Zhu H, Fang G, Zhang Y (2008) Multi-walled carbon nanotubes as solid-phase extraction adsorbents for determination of atrazine and its principal metabolites in water and soil samples by gas chromatography–mass spectrometry. Sci Total Environ 398:79–85. doi:10.1016/j.scitotenv.2008.02.016 Ravelo-Pérez LM, Hernández-Borges J, Rodríguez-Delgado MA (2008) Multi-walled carbon nanotubes as efficient solid-phase extraction materials of organophosphorus pesticides from apple, grape, orange and pineapple fruit juices. J Chromatogr A 1211:33–42. doi: 10.1016/j.chroma.2008.09.084 López-Feria S, Cárdenas S, Valcárcel M (2009) One step carbon nanotubes-based solid-phase extraction for the gas chromatographic-mass spectrometric multiclass pesticide control in virgin olive oils. J Chromatogr A 1216:7346–7350. doi:10.1016/j. chroma.2009.02.060 Zhao P, Wang L, Zhou L, Zhang F, Kang S, Pan C (2012) Multiwalled carbon nanotubes as alternative reversed-dispersive solid phase extraction materials in pesticide multi-residue analysis with QuEChERS method. J Chromatogr A 2225:17–25. doi:10.1016/j. chroma.2011.12.070 Kuster M, López de Alda M, Barceló D (2009) Liquid chromatography–tandem mass spectrometric analysis and regulatory issues of polar pesticides in natural and treated waters. J Chromatogr A 1216:520–529. doi:10.1016/j.chroma.2008.08.031 Liu L, Hashi Y, Qin Y, Zhou H, Lin JM (2007) Development of automated online gel permeation chromatography-gas chromatograph mass spectrometry for measuring multiresidual pesticides in agricultural products. J Chromatogr B 845:61–68. doi:10.1016/j. jchromb.2006.07.032 Li M, Jin Y, Li H, Hashi Y, Ma Y, Lin JM (2013) Rapid determination of residual pesticides in tobacco by QuEChERS sample pretreatment method coupled with liquid chromatography-mass spectrometry. J Sep Sci 36:2522–2529. doi:10.1002/jssc.201201091 Kosjet T, Heath E (2008) Applications of mass spectrometry to identifying pharmaceutical transformation products in water treatment. TrAC Trends Anal Chem 27:807–820. doi:10.1016/j.trac.2008.08.014 García Galán MJ, Villagrasa M, Díaz-Cruz MS, Barceló D (2010) LC-QqLIT MS analysis of nine sulfonamides and one of their acetylated metabolites in the llobregat river basin. Quantitative determination and qualitative evaluation by IDA experiments. Anal Bioanal Chem 397:1325–1334. doi:10.1007/s00216-010-3630-y

27. Pareja L, Martínez-Bueno MJ, Cesio V, Heinzen H, Fernández-Alba AR (2011) Trace analysis of pesticides in paddy field water by direct injection using liquid chromatography-quadrupole-linear ion trapmass spectrometry. J Chromatogr A 1218:4790–4798. doi:10.1016/ j.chroma.2011.02.044 28. Commission Decision 2002/657/EC of 12 August 2002 (2002). Off. J. Eur. Commun. 221:8–36. http://ec.europa.eu/food/food/ chemicalsafety/residues/lab_analysis_en.htm 29. Gros M, Rodríguez-Mozaz S, Barceló D (2012) Fast and comprehensive multi-residue analysis of a broad range of human and veterinary pharmaceuticals and some of their metabolites in surface and treated waters by ultra-high-performance liquid chromatography coupled to quadrupole-linear ion trap tandem mass spectrometry. J Chromatogr A 1248:104–121. doi:10.1016/j.chroma.2012.05.084 30. Martínez Bueno MJ, Agüera A, Hernando Guil MD, Gómez MJ, Fernández-Alba AR (2009) Evaluation of various liquid chromatography-quadrupole-linear ion trap-mass spectrometry operation modes applied to the analysis of organic pollutants in wastewaters. J Chromatogr A 1216:5995–6002. doi:10.1016/j.chroma.2009. 06.047 31. Dahane S, Gil García MD, Martínez Bueno MJ, Uclés Moreno A, Martínez Galera M, Derdour A (2013) Determination of drugs in river and wastewaters using solid-phase extraction by packed multiwalled carbon nanotubes and liquid chromatography–quadrupolelinear ion trap-mass spectrometry. J Chromatogr A 1297:17–28. doi:10.1016/j.chroma.2013.05.002 32. Zhang ZL, Zhou JL (2007) Simultaneous determination of various pharmaceutical compounds in water by solid-phase extraction–liquid chromatography–tandem mass spectrometry. J Chromatogr A 1154: 205–213. doi:10.1016/j.chroma.2007.03.105 33. Carabias- Martínez R, Rodríguez-Gonzalo E, Herrero-Hernández E, Sánchez-San Román FJ, Prado Flores MG (2002) Determination of herbicides and metabolites by solid-phase extraction and liquid chromatography: Evaluation of pollution due to herbicides in surface and groundwaters. J Chromatogr A 950:157–166. doi: 10.1016/S00219673(01)01613-2 DOI:10.1016/S0021-9673%2801%29016132#doilink 34. Massart DL, Vandeginste BGM, Buydens LMC, De Jong S, Lewi PJ, Smeyers-Verbeke J (1997) Handbook of Chemometrics and Qualimetrics Part A, Elsevier, Amsterdam, pp 190, 209, 436 35. Wang X, Tao S, Xing B (2009) Sorption and competition of aromatic compounds and humic acid on multiwalled carbon nanotubes. Environ Sci Technol 43:6214–6219. doi:10.1021/es901062t