Analysis of glyphosate and aminomethylphosphonic acid by capillary electrophoresis with electrochemiluminescence detection

Available online at www.sciencedirect.com

Journal of Chromatography A, 1177 (2008) 195–198

Short communication

Analysis of glyphosate and aminomethylphosphonic acid by capillary electrophoresis with electrochemiluminescence detection Hsien-Yi Chiu, Zhen-Yu Lin, Hsin-Ling Tu, Chen-Wen Whang ∗ Department of Chemistry, Tunghai University, Taichung 407, Taiwan Received 28 August 2007; received in revised form 13 November 2007; accepted 14 November 2007 Available online 19 November 2007

Abstract A capillary electrophoresis (CE) method coupled with electrochemiluminescence (ECL) detection for the analysis of glyphosate (GLY) and its major metabolite aminomethylphosphonic acid (AMPA) is presented. Complete separation of GLY and AMPA was achieved in 8 min using a background electrolyte of 20 mM sodium phosphate (pH 9.0) and a separation voltage of 21 kV. ECL detection was performed with an indium tin oxide (ITO) working electrode bias at 1.6 V (vs. a Pt-wire reference) in a 300 mM sodium phosphate buffer (pH 8.0) containing 3.5 mM Ru(bpy)3 2+ (where bpy = 2.2 -bipyridyl). Linear correlation (r ≥ 0.997) between ECL intensity and analyte concentration was obtained in the ranges 0.169–16.9 and 5.55–111 ␮g ml−1 for GLY and AMPA, respectively. The limits of detection (LODs) for GLY and AMPA in water were 0.06 ␮g ml−1 and 4.04 ␮g ml−1 , respectively. The developed method was applied to the analysis of GLY in soybeans. The LOD of GLY in soybean was 0.6 ␮g g−1 . Total analysis time including sample pretreatment was less than 1 h. © 2007 Elsevier B.V. All rights reserved. Keywords: Glyphosate; Aminomethylphosphonic acid; Capillary electrophoresis; Electrochemiluminescence detection; Soybean

1. Introduction Glyphosate (N-(phosphonomethyl)glycine; GLY) is a widely used broad-spectrum, foliar-applied herbicide for weed and vegetation control. Its degradation in the environment mainly occurs under biological conditions yielding aminomethylphosphonic acid (AMPA) as the major metabolite. Due to its relatively low mammal toxicity, GLY has become one of the most extensively used herbicides worldwide. This indiscriminate application generates some concerns regarding the possible health hazard and environmental impact caused by GLY. Therefore, the monitoring of GLY at residue levels in various agricultural products and environmental samples has attracted considerable attention [1]. GLY often represents an analytical challenge because of its relatively high solubility in water, insolubility in organic solvent, high polarity and low volatility. Derivatization prior to gas chromatography (GC) analysis is required to lower its polarity and enhance the volatility [2,3]. On the other hand, GLY shares

∗

Corresponding author. Tel.: +886 4 23590488x311; fax: +886 4 23506473. E-mail address: [email protected] (C.-W. Whang).

0021-9673/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2007.11.042

the structure with amino acids and possess no chromophore or fluorophore. Both pre- and postcolumn derivatization has been commonly employed to high-performance liquid chromatography (HPLC) [4,5] and capillary electrophoresis (CE) [6,7] of GLY and its metabolites with UV-absorption or fluorescence detection. However, the procedure of derivatization is always tedious and time-consuming, and sometimes generates unstable products. Non-derivatization methods, such as indirect UV detection [8–10], indirect laser-induced fluorescence (LIF) detection [11], electrospray ionization mass spectrometry (ESIMS) [12,13], flame photometric detection [14], and electrospray condensation nucleation light scattering detection [15] have also been reported for CE analysis of GLY. Among those non-derivatization methods, indirect detection generally suffers from low sensitivity while other detection modes require expensive and complicated apparatus. In recent years, tris(2,2 -bipyridyl)ruthenium(II) 2+ (Ru(bpy)3 )-based electrochemiluminescence (ECL) [16] has become an important and powerful detection method for CE due to its high sensitivity and simple instrumental setup. Although Ru(bpy)3 2+ -ECL detection using a glassy carbon working electrode has been employed to the HPLC of GLY and related compounds, no application to real sample was

196

H.-Y. Chiu et al. / J. Chromatogr. A 1177 (2008) 195–198

demonstrated [17]. Besides, CE with ECL detection has not been reported previously for GLY analysis. In this communication, we report a simple and economic CE method with Ru(bpy)3 2+ -ECL detection for the analysis of underivatized GLY and its major metabolite, AMPA. The applicability of this method was demonstrated by analyzing GLY in soybean samples. 2. Experimental 2.1. Apparatus Cyclic voltammetry (CV) and ECL studies were performed with a CHI model 635-electrochemical station (Austin, TX, USA). A three-electrode cell with an indium tin oxide (ITO) working electrode (Delta Technologies, Stillwater, MN, USA), an Ag/AgCl/saturated KCl reference and a Pt-wire auxiliary electrode was used for electrochemical measurements. Optical signals were captured using a Hamamatsu R928 photomultiplier tube (PMT; Hamamatsu city, Japan), positioned in front of the ITO electrode and biased at −900 V. The laboratory-assembled CE-ECL system was similar to that described previously [18]. Separation capillaries (Polymicro Technologies, Phoenix, AZ, USA) were of 55 cm total length × 360 ␮m O.D. × 50 ␮m I.D. An ITO electrode, situated at the capillary outlet and biased at 1.6 V (vs. a Pt-wire reference), was used for in situ generation of the active Ru(bpy)3 3+ . Sample injection was carried out hydrodynamically for 15 s at 15 cm height.

Fig. 1. (A) Cyclic voltammograms of 80 mM sodium phosphate, pH 8 (dotted line), 5 ␮M Ru(bpy)3 2+ in phosphate buffer (dashed line) and 0.1 M GLY in phosphate buffer containing 5 ␮M Ru(bpy)3 2+ (solid line). (B) Corresponding ECL-potential curves of (A). Conditions: working electrode, ITO (0.25 cm2 ); reference electrode, Ag/AgCl/saturated KCl; counter electrode, Pt-wire; potential scan rate, 60 mV s−1 ; PMT voltage, −900 V.

3. Results and discussion 2.2. Chemicals Glyphosate (GLY, 96%), aminomethylphosphonic acid (AMPA, 99%), iminodiacetic acid (IDA, 98%; used as an internal standard (I.S.)) and tris(2,2 -bipyridyl)ruthenium(II) chloride (Ru(bpy)3 Cl2 , 98%) were purchased from Aldrich (Milwaukee, WI, USA). All other chemicals were of analyticalreagent grade. The CE buffer was 20 mM aqueous sodium phosphate (pH 9.0). All solutions were filtered through a 0.45 ␮m pore-size membrane filter before use. 2.3. Soybean sample preparation Both transgenic and nontransgenic (organic) soybeans were purchased from a local supermarket. Pre-dried (90 ◦ C, ∼12 h) soybeans were milled to powder form with approximate particle diameter of 10–20 ␮m, as estimated under a microscope. Two grams of ground soybeans were extracted with 6 ml of water in a sonicating bath for 30 min. After resting for 1 min, 2 ml of supernatant was taken and transferred to a 10 ml vial. Two milliliters of acetonitrile was added to precipitate the proteins, followed by centrifugation at 500 × g for 15 min. The supernatant was filtered through a 0.45 ␮m pore-size membrane filter before CE-ECL analysis. GLY-spiked sample was prepared by adding appropriate amounts of GLY standard solution into the ground soybean, followed by water extraction and deproteinization.

3.1. Electrochemical and ECL behavior of GLY on an ITO electrode The electrochemical and Ru(bpy)3 2+ -based ECL responses of GLY on ITO electrodes were first studied by CV using a three-electrode cell. Fig. 1(A) illustrates the cyclic voltammogram obtained in an 80 mM sodium phosphate buffer at pH 8.0 (dotted line). This background voltammogram is featureless in the potential range scanned (−0.1 to 1.5 V). With the addition of 5 ␮M Ru(bpy)3 2+ , the well-known reversible Ru2+/3+ redox waves should appear at ∼1.1 V [19], but not very clear in the present case due to low Ru(bpy)3 2+ concentration added (dashed line). In the presence of 0.1 M GLY, the anodic wave for Ru(bpy)3 2+ oxidation at ∼1.1 V is enhanced dramatically and the cathodic wave disappears (solid line), as expected for a catalytic reaction route. Similar catalytic effect of some amino acids on the redox reaction of Ru(bpy)3 2+ at ITO electrode has also been reported recently [20]. It was found that the catalytic current enhancement was proportional to the concentration of GLY. The corresponding ECL-potential curves obtained during CV scan are illustrated in Fig. 1(B). In the absence of GLY, ECL emission emerges following the oxidation of Ru(bpy)3 2+ and reaches a maximum at ∼1.4 V (dashed line in Fig. 1(B)). With the addition of GLY, ECL emission appears at a less positive potential, ∼1.0 V, and increases sharply with increasing anodic voltage (solid line in Fig. 1(B)), which evidences the cat-

H.-Y. Chiu et al. / J. Chromatogr. A 1177 (2008) 195–198

197

alytic effect of GLY on the Ru(bpy)3 2+ -ECL. Electrochemical and ECL responses of AMPA at ITO electrode were examined too. However, no significant catalytic effect on the Ru2+/3+ redox reaction was observed. 3.2. Optimization of CE-ECL procedure In the present work, a simple two-electrode cell with an ITO working electrode and a Pt-wire pseudo-reference was used as the ECL detector for CE [18]. The optimal detection potential for the three species, viz. GLY, AMPA and IDA (I.S.), was investigated in the range 0.8–2.0 V. The optimal potential was found at 1.6 V. ECL intensity of AMPA at this voltage was about two orders of magnitude weaker than that of GLY at the same concentration level. This is attributed to the fact that GLY, a secondary amine, is more ECL active with Ru(bpy)3 2+ than AMPA, a primary amine [21]. The variation of ECL intensities on the concentrations of Ru(bpy)3 2+ in the range 1–5 mM was examined next. The intensity of ECL emission increased with increasing Ru(bpy)3 2+ concentration, but the background noise also increased with higher concentration of Ru(bpy)3 2+ in the detector. The optimum concentration of Ru(bpy)3 2+ was found at 3.5 mM. The effect of buffer pH on the separation of GLY, AMPA and IDA was examined in the pH range 8–10. With a 20 mM phosphate buffer, the optimal separation was obtained at pH 9.0. However, ECL intensity for the reaction of Ru(bpy)3 2+ with amines is also known to be dependent upon pH values [22]. We found that the optimum pH for ECL detection of GLY is 8.0. In order to offset the possible pH change in the detector caused by CE eluate during analysis, a 300 mM phosphate buffer at pH 8.0 was used in the ECL detection cell. Fig. 2 illustrates a typical electropherogram obtained with ECL detection under the optimal experimental conditions described above. Sharp and symmetric peaks could be observed for GLY, AMPA as well as the I.S. The number of theoretical plates, N, for GLY and AMPA were 2.3 × 104 and 1.7 × 104 , respectively. Complete separation of the three species was achieved in about 8 min. 3.3. Analytical figures of merit Calibration graphs were constructed for GLY and AMPA in the concentration ranges 0.169–16.9 ␮g ml−1 (1–100 ␮M) and 5.55–111 ␮g ml−1 (50–1000 ␮M), respectively. The linear regression equations for GLY and AMPA were: y = 0.595x − 0.689 (correlation coefficient, r = 0.999, n = 5) and y = 0.008x + 0.011 (r = 0.997, n = 5). Repeatability was evaluated by performing replicate injections of a solution containing 0.85 ␮g ml−1 GLY and 11.1 ␮g ml−1 AMPA. Based on seven replicate analyses, the relative standard deviation (RSD) values on the ratios of migration time of GLY/I.S. and AMPA/I.S. were 0.8% and 0.9%, respectively. The RSD values on the ratios of peak area of GLY/I.S. and AMPA/I.S. were 7.0% and 4.6%, respectively. The limits of detection (LOD; S/N = 3) for GLY and AMPA in water were 0.06 ␮g ml−1 and 4.04 ␮g ml−1 , respec-

Fig. 2. Electropherogram of (1) AMPA and (2) GLY obtained with ECL detection. Conditions: capillary size, 55 cm length × 360 ␮m O.D. × 50 ␮m I.D.; anodic buffer, 20 mM sodium phosphate (pH 9.0); cathodic buffer, 300 mM sodium phosphate (pH 8.0) containing 3.5 mM Ru(bpy)3 2+ ; CE voltage, 21 kV; hydrodynamic injection, 15 s at 15 cm height; ITO potential, 1.6 V; PMT voltage, −960 V. Analyte concentrations: [GLY] = 0.85 ␮g ml−1 ; [AMPA] = 11.1 ␮g ml−1 ; [I.S.] = 0.65 ␮g ml−1 .

tively. In comparison with other non-derivatization methods, the LOD of GLY with ECL detection is more than one order of magnitude lower than that with indirect UV detection [8–10], and is similar to those with indirect LIF [11] and ESIMS detection [12,13]. Considering the cost and complex of LIF and MS instruments, the advantage of ECL detection for GLY is evident. As for AMPA, its LOD with ECL detection is more than one order of magnitude higher than those with MS detection [12], but similar to that with indirect LIF detection [11]. 3.4. Applications The applicability of the developed CE-ECL method was examined by analyzing GLY in soybeans. Both transgenic GLYresistant (GR) and herbicides-free organic soybeans were used as the test samples. Extraction with water followed by deproteinization with acetonitrile was the only pretreatment performed. Fig. 3A–C shows the typical electropherograms of GR soybean, GR soybean spiked with 12 ␮g g−1 GLY, and organic soybean, respectively. The complex soybean matrix did not interfere with the analysis of GLY. No detectable amount of residual GLY was found in both GR and organic soybeans tested. Calibration graph was constructed using GLY-spiked organic soybean in the concentration range 1.5–100 ␮g GLY per gram soybean. The linear regression equation was y = 0.226x + 0.656 (r = 0.998, n = 7). The recovery of 12 ␮g g−1 spiked-GLY from soybean was 92.7 ± 3.5% (n = 3). Total analysis time including sample pretreatment was less than 1 h. The LOD for GLY in soybean was 0.6 ␮g g−1 . This value is 10-fold higher than that in water, which may be attributed to the complex soybean matrix. The maximal permissible level of GLY in soybean is 20 ␮g g−1 in the USA [23] and the European Union [24], and 10 ␮g g−1 in Taiwan [25]. The LOD obtained in the present method is

198

H.-Y. Chiu et al. / J. Chromatogr. A 1177 (2008) 195–198

Fig. 3. Electropherograms of (A) transgenic GLY-resistant (GR) soybean, (B) GR soybean spiked with 12 ␮g g−1 GLY and (C) organic soybean obtained with ECL detection. Conditions as in Fig. 2.

adequate for rapid screening of soybeans for residual GLY. Currently, there is no tolerance level for AMPA in soybeans set by most countries. Acknowledgement Financial support from the National Science Council of Taiwan is gratefully acknowledged. References [1] C.D. Stalikas, C.N. Konidari, J. Chromatogr. A 907 (2001) 1. [2] S.H. Tseng, Y.W. Lo, P.C. Chang, S.S. Chou, H.M. Chang, J. Agric. Food Chem. 52 (2004) 4057.

[3] C.D. Stalikas, G.A. Pilidis, J. Chromatogr. A 872 (2000) 215. [4] M.V. Khrolenko, P.P. Wieczorek, J. Chromatogr. A 1093 (2005) 111. [5] A. Ghanem, P. Bados, L. Kerhoas, J. Dubroca, J. Einhorm, Anal. Chem. 79 (2007) 3794. [6] I. Tetsuya, T. Kouichi, O. Haruki, O. Kousaburo, J. Foren. Toxicol. 22 (2004) 7. [7] S.Y. Chang, M.Y. Wei, J. Chin. Chem. Soc. 52 (2005) 785. [8] T. Soga, M. Imaizumi, Electrophoresis 22 (2001) 3418. [9] M. Corbera, M. Hidalgo, V. Salvado, P.P. Wieczorek, Anal. Chim. Acta 540 (2005) 3. [10] M.G. Gikalo, D.M. Goodall, W. Matthews, J. Chromatogr. A 745 (1996) 189. [11] S.Y. Chang, C.H. Liao, J. Chromatogr. A 959 (2002) 309. [12] L. Goodwin, J.R. Startin, B.J. Keely, D.M. Goodall, J. Chromatogr. A 1004 (2003) 107. [13] H. Safarpour, R. Asiaie, Electrophoresis 26 (2005) 1562. [14] E.W.J. Hooijschuur, Ch.E. Kientz, J. Dijksman, U.A.Th. Brinkman, Chromatographia 54 (2001) 295. [15] J. You, M. Kaljurand, J.A. Koropchak, Int. J. Environ. Anal. Chem. 83 (2003) 797. [16] M.M. Richter, Chem. Rev. 104 (2004) 3003 (and references therein). [17] J.S. Ridlen, G.J. Klopf, T.A. Nieman, Anal. Chim. Acta 341 (1997) 195. [18] M.T. Chiang, M.C. Lu, C.W. Whang, Electrophoresis 24 (2003) 3033. [19] I. Rubinstein, A.M. Bard, J. Am. Chem. Soc. 103 (1981) 512. [20] C.J. Fecenko, T.J. Meyer, H.H. Thorp, J. Am. Chem. Soc. 128 (2006) 11020. [21] A.W. Knight, G.M. Greenway, Analyst 121 (1996) 101R. [22] S.N. Brune, D.R. Bobbitt, Talanta 38 (1991) 419. [23] Code of Federal Regulations (CFR) Title 40, Volume 23, 180.364. US Government Printing Office, Washington, DC, 2005. [24] Informal coordination of MRLs established in Directives 76/895/EEC, 86/362/EEC, 86/363/EEC, and 90/642/EEC: MRLs sorted by crop group, Health and Consumer Protection, European Commission, Brussels, 2004. [25] Pesticide Residue Limits in Foods, DOH Food No. 0960406212 Amended, Department of Health, Executive Yuan, Taiwan, 2007.