ANALYTICAL SCIENCES OCTOBER 2011, VOL. 27
1031
2011 © The Japan Society for Analytical Chemistry
Determination of Glyphosate in Water Samples by Multi-pumping Flow System Coupled to a Liquid Waveguide Capillary Cell Aline S. SILVA,* Ildikó V. TÓTH,**† Leonardo PEZZA,* Helena R. PEZZA,* and José L. F. C. LIMA** *Instituto de Química, Universidade Estadual Paulista (UNESP), P. O. Box 355, Araraquara CEP 14801-970, SP, Brazil **REQUIMTE, Departamento de Química, Faculdade de Farmácia, Universidade do Porto, Rua Aníbal Cunha 164, Porto 4099-030, Portugal
A simple screening method was developed for the determination of glyphosate in water samples using a multi-pumping flow system. The proposed method is based on the reaction between glyphosate and p-dimethylaminocinnamaldehyde (p-DAC), in an acid medium where the reaction product can be measured spectrophotometrically at λmax = 495 nm. An experimental design methodology was used to optimize the measurement conditions. The proposed method was applied to the determination of glyphosate in water samples in a concentration range from 0.5 to 10 μg mL–1. The limit of detection and quantification were 0.17 and 0.53 μg mL–1, respectively. The results obtained (88.5 to 104.5%) in recovery studies for the determination of glyphosate in different water samples indicated good accuracy and no matrix effect for the developed method. Samples were also analyzed by a confirmatory HPLC method, and agreement within the two set of results was found. (Received June 26, 2011; Accepted August 17, 2011; Published October 10, 2011)
Introduction Glyphosate (N-(phosphonomethyl)glycine, C3H8NO5P) is a well-known herbicide. The commercial product, Roundup® marketed by Monsanto, is the most frequently used formulation all around the world due to its high weed and vegetation control in several culture types, while it is nonselective, post-emergent and has low toxicity.1 Although there is no established consensus about the toxicity degree of glyphosate in the environment, its frequent use in agricultural, industrial or domestic areas; strong retention by soil colloids and high solubility in water can result in significant environmental impact.2,3 According Songa et al.4 and Jan et al.5 its residue can be found in soil, the atmosphere, agricultural products and ground water. Recently, the risks associated with the use of glyphosate have been revised, and suggest that glyphosate may pose a significant toxic threat to some native amphibians species.6 Therefore, given the widespread use of glyphosate in agriculture, forestry, and home-use sectors, its monitoring and control is very important, especially in an aquatic environment.7–10 Analytical methods have been reported for the quantitative determination of glyphosate in water and other environmental matrices; these methods are based on thin-layer chromatography,11 capillary electrophoresis,3 gas chromatography,8,12,13 liquid chromatography,14–16 luminescence,17,18 visible spectrophotometry5,19 or use electrochemical detections4,7,20 or immunosensors.21 Because the glyphosate molecule does not have chromophore or fluorophore groups in its structure; pre- or postcolumn To whom correspondence should be addressed. E-mail:
[email protected]
†
derivatization has usually to be employed in liquid chromatographic determinations; however, direct detection at 195 nm is possible.22,23 The official method23 for glyphosate determination in water applies high-performance liquid chromatography (HPLC) with postcolumn derivatization, and fluorescence detection. In general, the derivatization procedures require several stages to improve the resolution and selectivity, but their major drawbacks are a laborious clean-up procedure, a lower reproducibility and problems arising from the stability of the generated products.7,8 Glyphosate is also a highly polar compound, resulting in additional difficulty in reversed-phase chromatographic separation procedures. However, there are effective solutions (e.g. HILIC mode, or multicomponent gradient elution) being developed to overcome these difficulties; these solutions often result in a more complex and a more costly experimental set up. In this context, sample screening methods present various advantages. These sample screening methods refer to expeditious analytical methodologies that are designed to select from an initial set of samples, those that are above a pre-set concentration level in a simple, fast and reliable way. Only those samples that provide a positive response should be further processed in a more complex and expensive analytical method. The flow-based analytical systems proposed since 1975 have reached large proportion in the environmental research area. The vast acceptance is probably due the possibility to develop analytical methods with increased precision, along with low equipment costs. Moreover, the utilization of a closed-flow circuit minimizes the risk of contamination with reduced reagent and sample consumption; these characteristics are important in the development of environmental-friendly analytical procedures.24 However, for the determination of glyphosate, flow-injection methods are uncommon. Adcock et al.18 developed
1032 a flow-injection reversed-flow chemiluminescence method based on the reaction between glycine and tris(2,2′-bipyridyl)ruthenium(II). The reagent was chemically generated using a lead dioxide recirculating system. The method was successfully applied to the determination of the analyte and its mono-isopropylamine salt in commercial formulations. Recently, Colombo and Masini25 published a sequential injection method for the fluorometric determination of glyphosate based on the adaptation of the postcolumn derivatization reaction between glycine and o-phthaldialdehyde in the presence of 2-mercaptoethanol, applicable to soil and sediment samples. Multi-pumping flow systems (MPFS)26 represent a new tendency in miniaturization within flow-based analytical methods. In MPFS micro-pumps are used and can be operated individually or in a combined form to perform the insertion and transport of solutions.27 Some of the most important characteristics of this procedure are the achievement of pulsed flow conditions in the turbulent flow regime and the straight-forward implementation of a binary sampling technique contributing to a faster and more efficient mixing between the solutions, as well as a consequent improvement of the method sensitivity due to a lower dispersion effect.28 In spectrophotometric measurements, an additional approach can also be exploited to increase the analytical sensitivity; the use of an increased optical path length flow cell enhances the sensitivity due to an increase in the number of absorbent species that can interact with a beam of radiation.29 Using this liquid waveguide capillary cell (LWCC) in combination with flow methods has proved to be advantageous in achieving the low detection limits required for trace analysis, that along with molecular spectrophotometric methods were difficult to reach before.30,31 The aim of the present work was to develop a simple and environmental friendly screening method for the determination of glyphosate in water samples using a multi-pumping flow system. The proposed method is based on the reaction between glyphosate and p-dimethylaminocinnamaldehyde (p-DAC), in an acid medium where the reaction product can be measured spectrophotometrically at λmax = 495 nm. Experimental design methodologies were used to optimize the measurement conditions. The proposed method was applied to the determination of glyphosate in environmental water samples.
Experimental Reagents and solutions All reagents employed in this study were of analytical grade, and were used without previous purification. In preparing the solutions, ultrapure (18 MΩ cm) water obtained in a Milli-Q purification system (Millipore, RO15) was used. The p-dimethylaminocinnamaldehyde (p-DAC) (Sigma-Aldrich) 600 μg mL–1 aqueous solution was prepared by dissolving 30.0 mg p-DAC in 50 mL of 0.02 mol L–1 hydrochloric acid (Panreac). The p-DAC solution 200 μg mL–1 was obtained by diluting a stock solution (600 μg L–1) in 0.001 mol L–1 hydrochloric acid. Glyphosate standard solutions were prepared from a solid purchased from Sigma-Aldrich (96% purity). Working standard solutions of glyphosate, in the range of 0.5 to 10 μg mL–1, were prepared by appropriate dilution of the stock solution (50 μg mL–1). These solutions were kept for no more than 1 week. The effect of foreign species was studied as follows: various concentrations of the compounds were added to a standard solution of glyphosate (4 μg mL–1). CuCl2 (BDH Reagents &
ANALYTICAL SCIENCES OCTOBER 2011, VOL. 27
Fig. 1 Flow diagram of the MPFS used for glyphosate determination: S, sample; R, reagent; C, carrier stream; Pi, micropumps; x, confluence point; r, reactor; LWCC, liquid waveguide capillary cell; W, waste. The arrow indicates the direction of the solutions in the system.
Chemicals, 1000 μg mL–1), MnCl2 (Sigma-Aldrich, 99%), Fe(NO3)3, Al(NO3)2 and Zn(NO3)2 (Fluka, 1000 μg mL–1) were used to study as cationic interferents in the concentration range 5 to 300 μg L–1. K2SO4 (Riedel-de Haën, 99%), NaHCO3 (Merck, 99.5%), Na2B4O7 (Merck, 99.5%), C6H5Na3O7 (Riedel-de Haën, 99.5%), KH2PO4 (Riedel-de Haën, 99.5%) and NaNO3 (Sigma-Aldrich, 99.5%) were used to study anionic interferents in the concentrations range 10 to 600 μg mL–1, while NaNO2 (Riedel-de Haën, 99%) was used in the concentration range 40 to 1000 μg L–1. Other interferents containing amine groups were studied in the form of ethylendiamine (Merck, 99%) and dibuthylamine (Merck, 99%) in a concentration range similar to that of the nitrite. The pesticide glufosinate (Fluka) and the metabolite aminomethylphosphonic acid (AMPA) (Aldrich, 99%) were studied in the concentration range 0.5 to 10 μg mL–1. Apparatus Three solenoid micro-pumps (one 120SP1210 and two 120SP1220, Bio-Chem Valve Inc.) were used in the manifold. Absorbance measurements at 495 nm were carried out with a spectrophotometer (USB 4000 UV-Vis, Ocean Optics) coupled via optical fibers (600 μm diameter) to a liquid waveguide capillary cell (LWCC) with a flow path of 100 cm (inner volume, 250 μL) and to a light source (halogen lamp, LS-1-LL, Ocean Optics). The flow set-up was built using polytetrafluoroethylene tubing (0.8 mm i.d.); lab-made X-type confluence and connectors were also used. System operation for processing of the involved solutions was computer controlled (PCL 711 Advantech interface card) using lab-developed Software in Microsoft Quick Basic 4.5. Data acquisition was performed using the Ocean Optics SpectraSuit Software. Procedure A multi-pumping flow system was designed with three micro-pumps (P1, P2, P3) for insertion and propulsion of the involved solutions (Fig. 1). The sequence of pump operation is summarized in Table 1. Initially, micro-pumps P1 and P3 were actuated to insert reagent and sample solutions, respectively. A binary sampling mode was used for the insertion of aliquots of these solutions. Therefore, when micro-pump P1 was maintained on (off), micro-pump P3 was switched off (on). After insertion
ANALYTICAL SCIENCES OCTOBER 2011, VOL. 27
1033
Table 1 Protocol sequence for the MPFS determination of glyphosate
Step
1 2 3
Pump functioning and volume delivered per stroke/μL P1
P2
P3
Number of pulses
10 — —
— — 20
20 — —
20 — 100
Pulse frequency/ min–1
Action
30 0 30
Binary sampling, introduction of the sample and reagent solutions Stop flow for 180 s Transport of the reaction mixture to the flow cell and signal registration
of the selected volume, the solutions were transported to the reactor where the flow was stopped during a predefined period of time. In the next step, while micro-pumps P1 and P3 were maintained off, micro-pump P2 was actuated to insert a carrier stream and transport the reaction mixture through the detector flow cell where the transient signal of absorbance was registered. The signal magnitude is proportional to the glyphosate concentration. After registering the absorbance maximum, the flow system is switched back to its initial position, starting another cycle. Reference/confirmational method An HPLC method with precolumn fluorogenic labeling was used as a reference (confirmational) procedure. This procedure involves a derivatization step with 9-methylchloroformate (FMOC-Cl) yielding fluorescent derivates and their separation on a revered-phase column. A derivatization procedure was carried out as described by Le Bot et al.,32 including an additional acidification step to release glyphosate, as suggested by Ibáñez et al.33 The HPLC (Jasco, manual injector, PU-2080 pump and FP-2020Plus fluorometric detector) conditions were as follows: isocratic separation on an amino-functionalized column (Phenomenex, Luna 5 μ NH2 100A, 250 × 4.6 mm), mobile phase 55:45 phosphate buffer (1.5%, pH 5.8):acetonitrile (Merck) 1 mL min–1, sample loop 20 μL. Detection of the eluted compounds was carried out at λex: 260 nm/λem: 310 nm. Under these experimental conditions, a linear calibration was obtained in the concentration range from 5.0 to 20 μg L–1 with a corresponding detection limit of 1 μg L–1. Experimental design After the identification of significant parameters, the variables were optimized by multivariate analysis using a fractionary factorial design (29–5) to obtain the optimum analysis conditions.34 The study of variables involved in the reaction were the reactor length, pulse number and time interval of solenoid-pump actuation (for sample, reagent and carrier solutions), p-DAC and HCl concentrations and stopped-flow period. The design of the matrix was performed using Minitab software, and optimization graphs were constructed using Statistica 8.0 software. Water samples Drinking water was purchased from a local supermarket. Natural water samples were collected from rivers and springs in the region of north of Portugal, filtered through 0.45 μm micro C in pore membrane, and stored in a refrigerator at 4° polyethylene flasks. For recovery studies, the water samples were spiked with 2, 4, 6 μg mL–1 of the glyphosate standard solution.
Results and Discussion There is a secondary amine group in the glyphosate molecule,
Fig. 2 Pareto chart for visualizing the effects produced by the variables studied using a 2(9–5) factorial design.
which can react with p-DAC in an acid medium. The reaction between primary or secondary amines and p-DAC is assumed to take place via condensation of the protonated amino group with the carbonyl group of the reagent to produce an imminium salt. The probable mechanism for this reaction is shown in Fig. S1 (Supporting Information), which is based on reactions suggested in the literature.35–37 Optimization of variables The optimization was developed by multivariate analysis where a factorial design (29–5) was employed to obtain the maximum response in sensitivity for the proposed method. The variable screening involved in the reaction was used to determine the influence for each variable effect on the analytical signal. In this design, the variables of interest were: reactor length, 100 and 190 cm; pulse interval of the carrier stream solution, affecting the flow rate of the carrier stream, 1 and 2 s; pulse number of the carrier stream, affecting the volume of the carrier stream introduced and the cleaning of the system, 40 and 80 pulses; pulse interval of the reagent and sample solutions, affecting the flow rate of the introduced solutions, 1 and 2 s; pulse number of the reagent and sample solutions, affecting the total volume of the reagent and sample solution, 10 and 20 pulses; concentration of p-DAC, 30 and 60 μg mL–1; concentration of HCl in the p-DAC solution, 0.001 and 0.01 mol L–1; concentration of HCl of the carrier stream solution, 0.001 and 0.01 mol L–1; stopped flow period, 90 and 180 s. The experimental matrix employed with variables and their respective levels examined, low (–1) and high (+1), are summarized in Table S1 (Supporting Information). For realization of this experimental design, 16 experiments were
1034 necessary, with three replicates, using a randomized matrix to eliminate any environmental variation. The Pareto chart (Fig. 2) showed graphical presentation for the effects produced by the studied variables. This indicated that the variables with no significant effects were the reactor length, pulse interval for the reagent and sample solutions, number pulse of the carrier stream and HCl concentrations, because their influences on the analytical signal magnitudes were smaller than 5%.38 Therefore, the combination of 80 pulses of the carrier stream, 2 s for the pulse interval of the reagent and the sample solutions, 190 cm of the reactor length, 0.001 mol L–1 of HCl concentration in the reagent solution (solvent) and in the carrier stream were used. In this MPFS set-up, the variables with more significant effects were the p-DAC concentration, the number of pulses for the reagent and sample solutions, the stopped time period and the pulse interval of the carrier stream. This indicates that any further development of the reaction is limited by the reaction kinetics, and not by insufficient mixing or a too-high dispersion. Regarding the pulse interval for the reagent and sample solutions and the stopped flow period, these were fixed at their level of +1 (20 pulses, 2 s and 180 s, respectively) so as to avoid a low analytical frequency of the system. It was observed that some combinations of levels between the studied variables promoted a base-line instability and high value of the standard deviation. Figure S2 (Supporting Information) shows a Pareto chart concerning the standard deviation obtained in each experiment (Table S1). According to Fig. S2, the variable with more influence on the standard deviation of the measurements is the HCl concentration in the carrier stream, probably due to inefficient cleaning of the reactor coil between measurements. Therefore, the HCl concentration as the carrier stream solution was fixed at its level of +1 (0.01 mol L–1), and the number of pulses of these solutions was fixed at 100. After the optimal condition of these variables involved in the system were established, the p-DAC concentration, as remaining variable, was studied in a new range (50 to 200 μg mL–1) by an univariate design with the aim to increase the method sensitivity. In this study, a better analytical signal was obtained using 200 μg mL–1 of p-DAC; a further increase in the reagent concentration will increase the blank reading (over an absorbance value of 0.600), and therefore shorten the linear response range of the developed method.
ANALYTICAL SCIENCES OCTOBER 2011, VOL. 27 Table 2 Results of recovery assays of glyphosate added to water samples Sample Bottled drinking water 1 Bottled drinking water 2 Spring water 1 Spring water 2 River water 1
River water 2
Glyphosate added/ μg mL–1
Glyphosate founda/ μg mL–1
Recovery, %
2 4 6 2 4 6 2 4 6 2 4 6 0 2 4 6 2 4 6
1.86 3.54 5.56 1.91 3.58 5.84 2.06 4.16 5.60 1.93 3.71 5.41 2.25 4.04 5.82 8.10 2.09 4.10 5.66
93.1 88.5 92.7 95.7 89.5 97.4 103.0 104.0 93.6 96.4 92.8 90.2 — 89.2 89.3 97.5 104.5 102.6 94.3
a. Values found are the average of three independent assays (n = 3).
where s is the standard deviation of the measures referring to the blank (n = 11) and S the slope of the linear dynamic range. The limit of detection and quantification were 0.17 and 0.53 μg mL–1, respectively. The LOD is consistent with the maximum contaminant level for drinking water in the United States of 0.7 μg mL–1,40 and also the concentration of 0.28 μg mL–1 recommended by Health and Welfare Canada in the Canadian Water Quality Guidelines.40
Analytical characteristics of the proposed method The developed analytical method was evaluated in terms of the linear dynamic range, limit of detection (LOD), limit of quantification (LOQ), precision and accuracy.34 The repeatability was evaluated by employing ten inter-day and intra-day replicate measurements from a 4 μg mL–1 standard solution, whose variation coefficients were 6.4 and 7.1%, respectively. These results of precision can be considered to be acceptable, according to González-Martinez et al.21
Interferences The possible influence of various cations (Al3+, Fe3+, Cu2+, Zn2+, Mn2+), anions (PO43–, SO42–, NO3–, NO2–, B4O72–, CO32–, C6H5O(COO)33–, glufosinate and the metabolite AMPA was evaluated. For this study, solutions containing 4 μg mL–1 for glyphosate and each one of the foreign species were taken separately in different ranges of the concentrations. These were analyzed under the same conditions as those described in the recommended procedure. The results showed that the ions Al3+, Zn2+, PO43–, B4O72–, CO32–, C6H5O(COO)33– can interfere in the reaction if present at concentrations higher than 6 μg mL–1. The ion NO2– was studied in the range of 40 to 1000 μg L–1, and no interferences were observed. Aminomethylphosphonic acid (AMPA) and glufosinate were also studied as possible interferences; within these, only AMPA was found to interfere, and only if present in concentrations over 10 μg mL–1.
Analytical curve, LOD and LOQ In this study, a linear relationship (Fig. S3, Supporting Information) was observed between the absorbance and the concentration (μg mL–1). By using a long optical path length flow cell in this MPFS it was possible to obtain an analytical curve in the concentration range from 0.5 to 10 μg mL–1 for glyphosate. A least-squares treatment of the calibration data (n = 6) yielded the regression equation y = 0.1177(±0.0065) + 0.05509(±0.0011)× C, with a correlation coefficient of 0.9992. The LOD and LOQ were determined according to an IUPAC39 recommendation: LOD = 3 × (s/S) and LOQ = 10 × (s/S),
Recovery and sample analysis To evaluate the efficiency of the proposed method in real samples, this method was applied in the determination of glyphosate in water samples (drinking and natural waters). The results obtained in the determination of glyphosate in the different water samples showed that only one sample presented positive results (2.25 μg mL–1) for glyphosate. The other samples studied presented glyphosate concentrations lower than the limit of detection. In order to evaluate the accuracy and to detect possible interferences from the matrix, recovery assays were carried out.
ANALYTICAL SCIENCES OCTOBER 2011, VOL. 27 Table 3 Results obtained by the developed MPFS and by the confirmatory (HPLC) method in the analysis of non-spiked and spiked (2.0 μg mL–1) water samples Sample Bottled water 1 Bottled water 2 Spring water 1 Spring water 2 River water 1 River water 2
(ns) (s) (ns) (s) (ns) (s) (ns) (s) (ns) (s) (ns) (s)
HPLC/μg mL–1
MPFS/μg mL–1
< 0.2 1.90 ± 0.03 < 0.2 1.16 ± 0.03 < 0.2 1.86 ± 0.06 < 0.2 2.10 ± 0.06 < 0.2 2.04 ± 0.06 < 0.2 2.08 ± 0.06
(–) (+) 1.70 ± 0.10 (–) (+) 1.88 ± 0.11 (–) (+) 1.70 ± 0.10 (–) (+) 2.04 ± 0.12 (+) 10), causing experimental difficulties when applied to samples of some degree of complexity. The approach presented here was based on a reaction occurring under acidic conditions. Combining the advantages of the multi-pumping flow method and the increased sensitivity spectrophotometric detector, it was possible to develop an analytical system for the determination of glyphosate. Good accuracy and no matrix effect was found for the developed procedure, indicating that the method can be a valuable tool as a fast screening method for the detection of glyphosate at eco-toxicological concentration levels.
Acknowledgements The authors would like to thank CAPES for a scholarship granted to A. S. da Silva. I. V. Tóth thanks FSE and MCTES
1035 (Ministério da Ciência, Tecnologia e Ensino Superior) for financial support through the POPH-QREN program.
Supporting Information The Supporting Information include: Fig. S1, Scheme of the probable reaction mechanism between glyphosate and p-DAC. Figure S2, Pareto chart for visualizing the effects produced on the standard deviations by each studied variable. Figure S3, Transient signals obtained for glyphosate standard solutions measured in triplicate. Table S1, Matrix used in the fractionary factorial design. This material is available free of charge on the Web at http://www.jsac.or.jp/analsci/.
References 1. G. M. Dill, C. A. CaJacob, and S. R. Padgette, Pest Manage. Sci., 2008, 64, 326. 2. S. B. Powles, D. F. Lorraine-Colwill, J. J. Dellow, and C. Preston, Weed Sci., 1998, 46, 604. 3. J. Jiang and C. A. Lucy, Talanta, 2007, 72, 113. 4. E. A. Songa, O. A. Arotiba, J. H. O. Owino, and J. Nazeem, Bioelectrochemistry, 2009, 75, 117. 5. M. R. Jan, J. Shah, M. Muhammad, and B. Ara, J. Harzard. Mater., 2009, 169, 742. 6. J. Struger, D. Thompson, B. Staznik, P. Martin, T. McDaniel, and C. Marvin, Bull. Environ. Contam. Toxicol., 2008, 80, 378. 7. J. Xiao, Q. Zhou, B. Yang, and Z. Wang, Anal. Sci., 2004, 20, 1415. 8. H. Kataoka, S. Ryu, N. Sakiyama, and M. Makita, J. Chromatogr., A, 1996, 726, 253. 9. A. J. Aubin and A. E. Smith, J. Agric. Food Chem., 1992, 40, 1163. 10. V. Subramaniam and P. E. Hoggard, J. Agric. Food Chem., 1988, 36, 1326. 11. J. C. Joung, S. U. Khan, and P. B. Marriage, J Agric. Food Chem., 1977, 25, 918. 12. J.-Y. Hu, C.-L. Chen, and J.-Z. Li, J. Anal. Chem., 2008, 63, 371. 13. P. L. Alferness and Y. Iwata, J. Agric. Food Chem., 1994, 42, 2751. 14. K. Qian, T. Tang, T. Shi, F. Wang, J. Li, and Y. Cao, Anal. Chim. Acta, 2009, 635, 222. 15. M. V. Khrolenko and P. P. Wieczorek, J. Chromatogr., A, 2005, 1093, 111. 16. I. Hanke, H. Singer, and J. Hollender, Anal. Bioanal. Chem., 2008, 391, 2265. 17. I. D. Merás, T. G. Díaz, and M. A. Franco, Talanta, 2005, 65, 7. 18. J. L. Adcock, N. W. Barnett, R. D. Gerardi, C. E. Lenehan, and S. W. Lewis, Talanta, 2004, 64, 534. 19. B. L. Bhaskara and P. Nagaraja, Helv. Chim. Acta, 2006, 89, 2686. 20. F. R. Simões, L. H. C. Mattoso, and C. M. P. Vaz, Sens. Lett., 2006, 4, 319. 21. M. Á. González-Martinez, E. M. Brun, R. Puchades, Á. Maquieira, K. Ramsey, and F. Rubio, Anal. Chem., 2005, 77, 4219. 22. J. You, M. Kaljurand, and J. A. Koropchak, Int. J. Environ. Anal. Chem., 2003, 83, 797. 23. W. Horwitz and G. W. Latimer Júnior, “Official Methods of Analysis of AOAC International”, 2005, AOAC International,
1036 Gaithersburg. 24. P. T. Anastas and T. C. Williamson, “Green Chemistry. Designing Chemistry for the Environment”, 1996, American Chemical Society, Washington. 25. S. M. Colombo and J. C. Masini, Microchem. J., 2011, 98, 260. 26. J. L. M. Santos, M. F. T. Ribeiro, A. C. B. Dias, and J. L. F. C. Lima, Anal. Chim. Acta, 2007, 600, 21. 27. A. C. B. Dias, J. L. M. Santos, J. L. F. C. Lima, C. M. Quintella, A. M. V. Lima, and E. A. G. Zagatto, Anal. Bioanal. Chem., 2007, 388, 1303. 28. P. R. Fortes, M. A. Feres, M. K. Sasaki, E. A. Alves, E. A. G. Zagatto, J. A. V. Prior, J. L. M. Santos, and J. L. F. C. Lima, Talanta, 2009, 79, 978. 29. K. Fuwa, W. Lei, and K. Fujiwara, Anal. Chem., 1984, 56, 1640. 30. J. Avivar, L. Ferrer, M. Casas, and V. Cerdà, Anal. Bioanal. Chem., 2010, 397, 871. 31. R. N. M. J. Páscoa, I. V. Tóth, A. A. Almeida, and A. O. S. S. Rangel, Sens. Actuators, B, 2011, 157, 51. 32. B. Le Bot, K. Colliaux, D. Pelle, C. Briens, R. Seux, and M. Cément, Chromatographia, 2002, 56, 161. 33. M. Ibáñez, O. J. Pozo, J. V. Sancho, F. J. López, and F. Hernández, J. Chromatogr., A, 2006, 1134, 51. 34. J. C. Miller and J. N. Miller, in “Statistics for Analytical
ANALYTICAL SCIENCES OCTOBER 2011, VOL. 27 Chemistry”, 1993, Ellis Horwood, New York. 35. A. S. Silva, F. C. B. Fernandes, J. O. Tognolli, L. Pezza, and H. R. Pezza, Spectrocim. Acta, Part A, 2011, 79, 1881. 36. N. H. Zawilla, M. Abdul Azim Mohammad, N. M. El Kousy, and S. M. El-Moghazy Aly, J. Pharm. Biomed. Anal., 2002, 27, 243. 37. A. Saeed, S. Haque, and S. Z. Quresh, Talanta, 1993, 40, 1867. 38. J. L. Rufino, H. R. Pezza, and L. Pezza, Anal. Sci., 2008, 24, 871. 39. G. L. Long and J. D. Winefordner, Anal. Chem., 1983, 55, 712A. 40. F. Rubio, L. J. Veldhuis, B. S. Clegg, J. R. Fleeker, and J. C. Hall, J. Agric. Food Chem., 2003, 51, 691. 41. CAC/GL 40-1993, “Guidelines on Good Laboratory Practice in Pesticide Residue Analysis”, Codex Committee on Pesticide Residues. 42. EU/CD 96/23/EC, Official Journal of European Communities L221/8, 17.8.2002. 43. L. Folmar, H. Sanders, and A. Julin, Arch. Environ. Contam. Toxicol., 1979, 8, 269. 44. M. S. Vera, L. Lagomarsino, M. Sylvester, G. L. Pérez, P. Rodríguez, H. Mugni, R. Sinistro, M. Ferraro, C. Bonnetto, H. Zagarese, and H. Pizarro, Ecotoxicology, 2010, 19, 710.
