Development of a Portable Immunoextraction-Reversed-Phase

through the use of miniaturized components, a portable power supply, and a laptop computer. ... A cart for transporting this system was built at t...
2 downloads 0 Views 246KB Size
Download PDF
Anal. Chem. 2004, 76, 805-813

Development of a Portable Immunoextraction-Reversed-Phase Liquid Chromatography System for Field Studies of Herbicide Residues Mary Anne Nelson, Arther Gates, Maud Dodlinger, and David S. Hage*

Chemistry Department, University of Nebraska, Lincoln, Nebraska 68588-0304

A portable system based on immunoextraction and reversed-phase HPLC was developed for the field analysis of herbicides in groundwater and surface water. Atrazine, simazine, and cyanazine were used as model analytes for this work. These were measured in water by using three coupled columns: an anti-atrazine antibody column for the selective extraction of these analytes, a reversed-phase precolumn for their reconcentration, and a reversedphase analytical column for their separation. Various factors were considered in the optimization of this system, including the binding properties of the immunoextraction column, the effect of flow rate on the performance of each column, the selection of sample volume, and the choice of mobile phases for the RPLC columns. A typical analysis with this system allowed the injection of one sample every 7.5 min and provided results for all three of the tested herbicides in less than 10 min. In the analysis of atrazine alone, samples could be injected every 4 min and results were obtained within 8 min. There was good correlation between this technique and a comparable benchtop system. The lower limits of detection for the given analytes were approximately 0.2-0.25 µg/L, with a linear range that extended to 20 µg/L and a dynamic range that went up to at least 100 µg/L. The use of this technique in the field was demonstrated through applications that involved the development of time and location profiles for triazine herbicides in environmental samples. Atrazine, or 2-chloro-4-(ethylamino)-6-(isopropyl)-s-triazine, is a herbicide used for the control of broadleaf weeds. This agent and related triazine compounds, such as simazine and cyanazine (see Figure 1), act through the inhibition of photosynthesis. Atrazine can be applied both pre- and postemergence by several methods, including ground spraying, aerial spraying, and spreading by tractor. An estimated 76.4 million tons of this herbicide are applied each year in the United States, with more than 80% being used on corn.1 Since atrazine has good water solubility and tends to persist in the environment, it has become a major * To whom correspondence should be addressed. [email protected]. (1) Overview of Atrazine Risk Assessment; U.S. Environmental Protection Agency: Washington, DC, 2002; pp 1-23. 10.1021/ac030298m CCC: $27.50 Published on Web 01/03/2004

© 2004 American Chemical Society

Figure 1. Structures of three common triazine herbicides: atrazine, simazine, and cyanazine.

contaminant in water supplies.2 It is also a suspected endocrine disruptor,3 raising concerns regarding its toxicological effects. The maximum allowable level for atrazine in drinking water is 3 µg/L, as set by the U.S. Environmental Protection Agency (EPA). This requires that atrazine be routinely monitored in public water supplies.4 Methods for analyzing atrazine include gas chromatography (GC) with mass spectrometric detection,5,6 highperformance liquid chromatography (HPLC),7-10 and enzymelinked immunosorbant assays (ELISAs).11-13 The U.S. EPA also has standard methods that use GC with mass spectrometry or alternative modes of detection for the trace determination of other triazine compounds in water.14,15 (2) Agricultural Chemicals in Groundwater: Proposed Pesticide Strategy; U.S. Environmental Protection Agency: Washington, DC, 1987; pp 1-150. (3) Timm, G. E.; Maciorowski, A. F. ACS Symp. Ser. 2000, 747, 1-10. (4) National Survey of Pesticides in Drinking Water Wells, Phase II Report; U.S. Environmental Protection Agency: Springfield, VA, 1992. (5) Beltran, J.; Lopez, F. J.; Hernandez, F. J. Chromatogr., A 2000, 885, 389404. (6) Scribner, E. A.; Thurman, E. M.; Zimmerman, L. Sci. Total Environ. 2000, 248, 157-167. (7) Rotich, H. K.; Li, J.-c.; Zhang, Z.-y.; Gu, X.-x. Huaxue Yanjiu 2002, 13, 5964. (8) Lee, W.-Y. Environmental Applications of Chiral HPLC and Development of New Chiral Stationary Phases. Thesis, University of Texas, 2000. (9) Ritter, J. P. Environmental and Biological Applications of High-Performance Chromatographic Techniques. Thesis, Clarkson University, 2000. (10) Lee, W. Y.; Salvador, J. M. Proc. Conf. Hazard. Waste Res., Denver, CO, May 23-25, 2000; pp 36-45. (11) Seiber, J. N.; Li, Q. S.; Van Emon, J. M. In Immunochemical Methods of Environmental Analysis; American Chemical Society: Washington, DC, 1990; pp 156-167. (12) Lee, N. A. K.; Kennedy, I. R. J. AOAC Int. 2001, 84, 1393-1406. (13) Linde, C. D.; Goh, K. S. Pestic. Outlook 1995, 6, 18-23.

Analytical Chemistry, Vol. 76, No. 3, February 1, 2004 805

In recent years, there has been growing interest in the creation of field-portable methods for measuring agents such as atrazine in the environment. However, current techniques have several limitations when considered for such an application. For instance, ELISAs have become popular for the on-site screening of environmental samples; however, these methods have only moderate accuracy and precision and give a result that is affected by any compound that binds to the antibodies they employ as reagents.16-18 GC and HPLC are more accurate and precise but generally involve laboratory-based methods that require the transport and storage of large quantities of sample. In addition, GC and HPLC often require several sample pretreatment steps (e.g., derivatization or solid-phase extraction) for environmental analytes that may result in additional products, coextracted interferences, or contaminants from the pretreatment step.19,20 Furthermore, the lag between the time of sample collection and analysis in these methods is of concern, since some herbicides can have a large and rapid change in concentration during high-use seasons (e.g., during or following their application).21 One possible solution to these problems is to employ a method that combines the selectivity of antibodies, as used for detection in ELISAs, with the accuracy, precision, and ease of automation of flow-based methods such as HPLC. Recent examples of such work in the environmental field include methods based on flow injection immunoanalysis22-25 and immunoextraction coupled to reversed-phase liquid chromatography (RPLC).26-29 Related examples include the use of molecularly imprinted polymers as antibody mimics for the recognition of analytes in environmental samples.30-32 Of these techniques, on-line immunoextraction with RPLC has been of particular interest. This has given rise to techniques for (14) U.S. Environmental Protection Agency. Fed. Regist. 1998, 63, 47098-47114. (15) Thoma, J. J.; Kraut, A.; George, J. E.; Day, R. S. Proc. Water Qual. Technol. Conf. 1992; pp 851-86. (16) Dunbar, B. D.; Niswender, G. D.; Hudson, J. M. U.S. Patent 4530786, 1985. (17) Maqbool, U.; Anwar-ul-Haq, Q.; Jamil, M.; Iqbal, M. Z.; Hock, B.; Kramer, K. J. Environ. Sci. Health, Part B 2002, B37, 307-322. (18) Goh, K.; Hernandez, J.; Powell, S. J.; Garretson, C.; Troiano, J.; Ray, M.; Greene, C. Bull. Environ. Contam. Toxicol. 1991, 46, 30-36. (19) Thurman, E. M.; Meyer, M.; Perry, C.; Schwab, P. Anal. Chem. 1990, 62, 2043-2048. (20) Junk, G. A.; Avery, M. A.; Richard, J. J. Anal. Chem. 1988, 60, 1347-1350. (21) Atrazine Environmental Fate and Effects; U.S. Environmental Protection Agency: Washington, DC, 2002; pp 1-99. (22) Bjarnason, B.; Bousios, N.; Eremin, S.; Johansson, G. Anal. Chim. Acta 1997, 347, 111-120. (23) Kramer, P. M.; Franke, A.; Standfuss-Gabisch, C. Anal. Chim. Acta 1999, 399, 89-97. (24) Lopez, M. A.; Dominguez, E.; Ortega, F. An. R. Acad. Farm. 1999, 65, 305-325. (25) Wortberg, M.; Middendorf, C.; Katerkamp, A.; Rump, T.; Krause, J.; Cammann, K. Anal. Chim. Acta 1994, 289, 177-86. (26) Carrasco, P. B.; Escola, R.; Marco, M. P.; Bayona, J. M. J. Chromatogr., A 2001, 909, 61-72. (27) Kim, B. B.; Vlasov, E. V.; Miethe, P.; Egorov, A. M. Anal. Chim. Acta 1993, 280, 191-6. (28) Rollag, J. G.; Beck-Westermeyer, M.; Hage, D. S. Anal. Chem. 1996, 68, 3631-3637. (29) Thomas, D.; Beck-Westermeyer, M.; Hage, D. S. Anal. Chem. 1994, 66, 3823-3829. (30) Ferrer, I.; Lanza, F.; Tolokan, A.; Horvath, V.; Sellergren, B.; Horvai, G.; Barcelo, D. Anal. Chem. 2000, 72, 3934-3941. (31) Pap, T.; Horvath, V.; Tolokan, A.; Horvai, G.; Sellergren, B. J. Chromatogr., A 2002, 973, 1-12. (32) Turiel, E.; Martin-Esteban, A.; Fernandez, P.; Perez-Conde, C.; Camara, C. Anal. Chem. 2001, 73, 5133-5141.

806

Analytical Chemistry, Vol. 76, No. 3, February 1, 2004

the measurement of triazine herbicides,26-29 carbofuran,33 carbendazim,34,35 phenylurea pesticides,36 and polycyclic aromatic hydrocarbons.37 However, all of these previous procedures have used laboratory-based systems rather than field-portable methods. The aim of this current report is to combine immunoextraction with RPLC in the development of a portable system for the onsite testing of herbicides in water samples. This will be illustrated by using atrazine, simazine, and cyanazine as model analytes. EXPERIMENTAL SECTION Reagents. Atrazine, simazine, and cyanazine were purchased from Sigma-Aldrich (St. Louis, MO). The anti-atrazine monoclonal antibodies (cell line AM7B2) were produced at the University of Nebraska Medical Center (Omaha, NE) from a cell line provided by the Monoclonal Antibody Production Facility at the University of CaliforniasBerkeley (Berkeley, CA). Rabbit immunoglobulin G (IgG) from Sigma-Aldrich was used as the standard in the bicinchoninic acid (BCA) protein assay. Reagents for the BCA assay were from Pierce (Rockfield, IL). The Optima grade 2-propanol, HPLC grade acetonitrile, and HPLC grade methanol were from Fisher Scientific (Plano, TX). The Nucleosil Si-1000 (7-µm particle, 1000-Å pore size) used for antibody immobilization and the Pt EPS C18 Nucleosil Si-300 (5-µm particle, 300-Å pore size) used in the RPLC precolumn were both from P. J. Cobert (St. Louis, MO). Other chemicals were of the highest grades available. The deionized water used for sample and mobile-phase preparation was generated by a Nanopure system (Barnstead, Dubuque, IA). Apparatus. The field-portable system developed in this study is shown in Figure 2. This was based on a design reported for a previous benchtop system,28,29 which was modified through the use of miniaturized components, a portable power supply, and a laptop computer. This system consisted of three Knauer MicroStar pumps from Sonntek (Upper Saddle River, NJ), 6-port and 10-port LabPro actuated valves from Rheodyne (Rhonert Park, CA), and a Knauer variable-wavelength K-2500 UV/visible detector from Sonntek. The system was controlled and data were collected through the use of a SCB-68 shielded interface box, a PCMCIA card, and LabView software from National Instruments (Austin, TX), along with a Solo 2500 laptop computer from Gateway (Poway, CA). When used in the field, this system was powered by a Yamaha 48 hp generator (Cypress, CA). A cart for transporting this system was built at the University of Nebraska machine shop (Lincoln, NE). A transportation case for the instrument was supplied by Danielle’s Case Co. (Littleton, CO). To help with stabilization during transport, this case was modified to incorporate two shelves with aluminum straps that secured each piece of equipment. Ventilation slots were cut in each side of the transport case and a small fan was attached to the side to help air flow within. To reduce the possibility of glass breakage and the overall weight of the instrument, polymer bags (intravenous delivery reservoirs) were obtained from Midland (33) Rule, G. S.; Mordehai, A. V.; Henion, J. Anal. Chem. 1994, 66, 230-5. (34) Bean, K. A.; Henion, J. D. J. Chromatogr., A 1997, 791, 119-126. (35) Thomas, D. H.; Lopez-Avila, V.; Betowski, L. D.; Van Emon, J. J. Chromatogr., A 1996, 724, 207-17. (36) Lawrence, J. F.; Menard, C.; Hennion, M.-C.; Pichon, V.; Goffic, F. L.; Durand, N. J. Chromatogr., A 1996, 732, 277-281. (37) Bouzige, M.; Pichon, V.; Hennion, M. C. J. Chromatogr., A 1998, 823, 197210.

Figure 2. (a) Schematic of the portable immunoextraction/RPLC system and (b) a photo of the instrument within its transportation case.

Medical Supply (Lincoln, NE) and used to hold and transport the mobile phases. This last item also minimized the appearance of air bubbles in the system, since the solvent reservoirs were collapsible and did not suffer from cavitation during use. The chromatographic system included a 2.1 mm i.d. × 10 mm immunoextraction column, a 2.1 mm i.d. × 10 mm RPLC precolumn, and a 4.6 mm i.d. × 150 mm or 7 mm i.d. × 33 mm RPLC analytical column. The immunoextraction column and RPLC

precolumn were packed according to a previously published technique38 at a pressure of 3500 psi. The immunoextraction column was packed with a silica support containing immobilized anti-atrazine antibodies. The packing solvent for this column was a pH 7.0, 0.1 M potassium phosphate buffer that had been filtered through a 0.45-µm cellulose acetate membrane. The RPLC (38) Clarke, W.; Hage, D. S. Anal. Chem. 2001, 73, 1366-1373.

Analytical Chemistry, Vol. 76, No. 3, February 1, 2004

807

precolumn was packed with Pt EPS C18 Nucleosil Si-300 silica using HPLC grade methanol that had been filtered through a 0.45µm nylon membrane. Two commercial columns were considered for use as the RPLC analytical column: a 4.6 mm i.d. × 150 mm Aqua column from Phenomenex (Torrance, CA) and a 33 mm × 7 mm i.d. Rocket column from Alltech (Deerfield, IL). The elution of all analytes was monitored at 223 nm. Methods. The immobilized antibody supports and diol-bonded silica were prepared according to previous methods,39,40 using Nucleosil Si-1000 silica as the starting material. The final coverage of the diol-bonded silica was 38 ((7) µmol of diol/g of silica ((1 SD). The amount of antibody on the immunoextraction support was measured by a BCA assay using rabbit IgG as the standard and diol-bonded silica as the blank. This gave a typical protein content of 38 ((1) mg of antibody/g of silica. The activity of each immunoextraction column was determined by frontal analysis.41 This involved the continuous application of each analyte at concentrations of 10-50 µg/L and at a flow rate of 0.5 mL/min in the presence of a pH 7.0, 0.1 M potassium phosphate buffer. All of these experiments were performed at 25 °C. The breakthrough times were determined from the resulting saturation curves. Corrections for the system void volume and nonspecific binding were made by performing identical studies with a control column that had no immobilized antibodies but contained the same support as the immunoextraction column. The retention of atrazine and other triazines on the reversedphase columns was examined at room temperature using an initial mobile phase that contained a 50:50 (v/v) mixture of acetonitrile in pH 3.0, 0.10 M potassium phosphate buffer. This was performed on the Aqua and Rocket columns using a 50-µL sample that contained 7.5 µg/L atrazine, simazine, or cyanazine. It was found under these conditions that the Rocket column gave the fastest separation for these analytes, so this column was used in all later studies. The composition of the mobile phase was then varied from 10:90 to 40:60 (v/v) acetonitrile in phosphate buffer at pH values ranging from 2.5 and 4.5. The use of 2-propanol in place of acetonitrile was also considered. Triplicate injections were made under each set of conditions, with the resulting retention data then being analyzed using DryLab software (LCResources, Walnut Creek, CA) for RPLC method optimization. To optimize the speed and throughput of the field-portable system, a 25 µg/L sample of atrazine was injected and analyzed at various flow rates. In these experiments, the application flow rate ranged from 0.5 to 1.0 mL/min, the flow rate for analyte elution from the immunoextraction column ranged from 0.2 to 0.75 mL/min, and the flow rate for the RPLC analytical column ranged from 0.5 to 1.0 mL/min. The resulting peak areas and elution times on the RPLC analytical column were then used to select the final combination of flow rates that gave the best recovery and total analysis time for the overall system. The final optimized system used pH 7.0, 0.1 M potassium phosphate as the application buffer, pH 2.5, 0.1 M potassium phosphate as the elution buffer for the immunoextraction column, and a 27:73 (v/v) mixture of 2-propanol with pH 3.0, 0.1 M potassium phosphate buffer as the RPLC mobile phase. These solvents were applied at flow rates of 0.75, 0.50, and 0.75 mL/ (39) Ruhn, P. G.; Garver, S.; Hage, D. S. J. Chromatogr. 1994, 669, 9-19. (40) Larsson, P.-O. Methods Enzymol. 1984, 104, 212-223. (41) Hage, D. S. J. Chromatogr., B 2002, 768, 3-30.

808

Analytical Chemistry, Vol. 76, No. 3, February 1, 2004

min, respectively. The analytical column selected for use in the final system was the Alltech Rocket column, and the amount of applied sample was 500 µL. A typical run on this system began with the application of the sample. At 1.5 min after injection, the elution buffer was passed through the immunoextraction column and RPLC precolumn. This was continued until 4.5 min after injection, when the RPLC precolumn was put in series with the RPLC analytical column and its mobile phase. As the analytes were separated and eluted from the RPLC columns, the immunoextraction column was regenerated by placing it back into the application buffer prior to the next sample injection, which typically occurred at 7.5 min from the beginning of the previous run. During this period, a few additional minutes were allowed for the elution of all desired analytes from the RPLC analytical column. The field studies described in this report were conducted at three locations: (1) five sites on or leading into the Platte River near Ashland, NE; (2) four wells in the Lincoln Well Facility near Ashland, NE; and (3) four sites along the Salt Creek near Lincoln, NE. The first two groups of sites were selected on the basis of their frequent use over the past decade in the monitoring of triazine herbicides.42 The sites in the last group were chosen on the basis of their general location and accessibility. RESULTS AND DISCUSSON Optimization of the Immunoextraction Column. Several components of the portable system were studied and optimized in this study. The immunoextraction column was examined first. For instance, the apparent binding capacity (mL,app) of each immunoextraction column was measured at several concentrations of applied analyte ([A]) by performing frontal analysis with the various triazines considered in this study. Plots of 1/mL,app versus 1/[A] were then prepared and used to estimate the binding capacity of the given column for these analytes, as described previously for related compounds.28 For instance, one such column gave a total binding capacity (mL) for atrazine of 3.6 ((0.34) × 10-11 mol, or 7.9 ((0.6) ng. The capacities obtained for simazine and cyanazine on the same column were 5.9 ((0.4) and 3.9 ((0.4) × 10-11 mol, respectively (i.e., 12 and 9.4 ng). Based on these binding capacities and the known antibody content of this column, it was estimated that 75-92% of the immobilized antibodies were able to bind atrazine, simazine, and cyanazine. Similar results were obtained for the other immunoextraction columns prepared during this study, with typical atrazine binding capacities of 7-10 ng. The plots of 1/mL,app versus 1/[A] gave linear behavior (r ) 0.89-0.96 for seven to eight data points) with slightly positive slopes. From the slopes and intercepts of these plots, it was possible to estimate the association equilibrium constant (Ka) for the binding of each analyte to the immobilized antibodies.28 Atrazine, simazine, and cyanazine were all found to have fairly tight binding to these antibodies, with association constants of 1.6 ((0.3) × 107, 1.6 ((0.2) × 107, and 2.3 ((0.6) × 107 M-1, respectively. The presence of these large equilibrium constants indicated that relatively strong retention was present for each of these compounds on the immunoextraction column. As will be seen later, this became an important feature when an increase in (42) Dormedy, D. G.; Rodriquez-Fuentes, R.; Carr, J. D. 217th ACS National Meeting, 1999; Abstract ENVR-167.

Table 1. Effect of Acetonitrile Concentration on the Retention and Resolution of Atrazine (Atz), Simazine (Sim), and Cyanazine (Cya) in Mobile Phases Containing pH 3.0, 0.10 M Phosphate Buffer % organic modifier in mobile phase acetonitrile 10 20 30 40

retention factor, k

resolution, Rs

critical paira

atrazine simazine cyanazine Atz-Sim Sim-Cya 50.1 11.0 4.2 2.8

23.5 5.8 2.5 1.7

26.3 5.6 2.2 1.8

5.98 8.48 6.57 4.66

1.54 0.55 1.30 1.08

Sim-Cya Sim-Cya Sim-Cya Sim-Cya

a The critical pair refers to the pair of neighboring peaks that have lowest resolution under a given set of separation conditions.

sample injection volume was explored as a way for decreasing the limit of detection for the field portable system. Optimization of the RPLC System. Other parameters optimized in this work included those dealing with the RPLC precolumn and analytical column. Previous studies on a related benchtop system used a 50:50 (v/v) mixture of acetonitrile with pH 7.0, 0.10 M phosphate buffer for the separation of atrazine and its degradation products on a C18 column.28 However, cyanazine and simazine were found to coelute in this mobile phase, making it necessary to find alternative conditions where they could be resolved. In another study, atrazine, simazine, and other triazines were resolved on a RPLC column with a 55:45 mixture of methanol in pH 2.5, 0.05 M phosphate buffer.29 Based on these results, a slightly acidic buffer containing acetonitrile was selected as the initial RPLC mobile phase for this present study. Lowering the amount of acetonitrile in a mobile phase containing pH 3.0, 0.10 M potassium phosphate buffer was one item considered for improving the separation of simazine from cyanazine, the most closely eluting set of analytes in this work. As shown in Table 1, decreasing the amount of acetonitrile from 40 to 10% did result in baseline resolution between simazine and cyanazine, but it also gave rise to significantly higher retention and longer analysis times. 2-Propanol was also considered for use as an organic modifier. Figure 3 shows the retention factors measured for atrazine, simazine, and cyanazine in mobile phases that contained this modifier in pH 3.0, 0.10 M phosphate buffer. It was found that changing from acetonitrile to 2-propanol gave better resolution at comparable organic modifier levels. For instance, the resolution between simazine and cyanazine was 2.0-3.6 when 10-30% 2-propanol was used but only 0.55-1.54 for equivalent amounts of acetonitrile. As a result, 2-propanol was employed as the organic modifier in all further studies. The effect of changing the mobile-phase pH was considered as well. In the presence of 30% 2-propanol, all pH values between 4.0 and 2.5 gave baseline resolution for the three triazines of interest in this study. Over this pH range, there was a small decrease in retention and resolution when going from pH 3.0 to 2.5 (e.g., a 6-11% decrease in k and 7-11% decrease in Rs). However, little change in retention or resolution was noted when going from pH 3.0 to 4.0. These observations, plus the better buffering capacity of phosphate at pH 3.0 versus 4.0, led to the

Figure 3. Plots of log k (i.e., the retention factor) versus the percent (v/v) content of organic modifier (2-propanol) in the mobile phase for the three model triazine compounds, atrazine (]), simazine (0), and cyanazine (4), on a RPLC column in the presence of various mixtures of pH 3.0, 0.10 M phosphate buffer and 2-propanol. These results were obtained using the Rocket column from Alltech. The best-fit lines were obtained with a quadratic equation that had the general form log k ) log kw + aΦ + bΦ2, where kw is the estimated retention factor in the presence of no organic modifier, Φ is the volume fraction of organic modifer in the mobile phase, and a and b are constants for the given system.43

choice of pH 3.0, 0.10 M phosphate buffer for use in the RPLC mobile phase in the field-portable system. By combining this information with the data in Table 1 and Figure 3 and analyzing these results with DryLab software, it was found that the best combination of resolution and retention would be obtained with a mobile phase that contained a 27% (v/v) mixture of 2-propanol in pH 3.0, 0.1 M potassium phosphate. The actual resolutions obtained under these conditions were 2.33 for the cyanazine and simazine peaks and 3.72 for the simazine and atrazine peaks, with retention factors in the range of 1.7-3.9. Thus, this mobile-phase composition was found to be satisfactory for further studies with the field-portable system. The data in Figure 3 also indicate how atrazine, simazine, and cyanazine would have behaved on the C18 RPLC precolumn in the presence of the immunoextraction elution buffer. For example, the intercepts in Figure 3 gave estimated retention factors of 213 for atrazine, 115 simazine, and 132 for cyanazine when a pH 3.0, 0.10 M phosphate buffer was used on the RPLC analytical column. As noted earlier, only slightly lower values (a decrease in k of 6-11%) would be expected when using a pH 2.5, 0.10 M phosphate buffer. Although these retention factors may have varied slightly from those seen on the C18 precolumn, which contained a support slightly different from the analytical column, these results do indicate that all three triazines had strong retention under these conditions. This was useful in that it allowed the analytes to be concentrated in a narrow band as they eluted from the immunoextraction column, as noted in previous reports.28,29 Optimization of Flow Rate. After the individual immunoextraction and RPLC columns had been studied, these columns were connected and used to examine the effect of flow rate on the performance of the overall portable system. The first flow rate considered was that used on the immunoextraction column for (43) Poole, C. F.; Poole, S. K. Chromatography Today; Elsevier: Amsterdam, 1991; p 396.

Analytical Chemistry, Vol. 76, No. 3, February 1, 2004

809

Figure 4. Analysis of a groundwater sample by the portable system with and without the use of the immunoextraction column. The concentration of atrazine in the sample was ∼2 µg/L. Details are given in the text.

sample application. This was examined by looking at the rate of analyte binding to such a column at several flow rates. For a 500µL sample of 25 µg/L atrazine applied to an immunoextraction column with a binding capacity of roughly 10 ng, a 42% decrease in binding was seen as the application flow rate was increased from 0.5 to 1.0 mL/min. This behavior was in agreement with previous kinetic studies of immunoaffinity supports.28 However, the use of a faster application flow rate also resulted in a shorter analysis time. Based on these results, an application flow rate of 0.75 mL/min was selected for all further work as a compromise between extraction efficiency and assay speed. The relative degree of capture for the analytes at this flow rate was ∼74%. Desorption of analytes from the immunoextraction column was also studied as a function of flow rate. Changing the flow rate of the elution buffer from 0.5 to 1.5 mL/min did not give any significant change in the final peak areas. For instance, elution flow rates of 0.75-1.5 mL/min gave relative areas that were 95103% of those seen at 0.5 mL/min. These results agree with other studies that have described the desorption of analytes from antibody columns as a first-order process that depends on time but not flow rate. In addition, these results confirmed that the desorbed atrazine, cyanazine, and simazine were effectively captured and retained by the RPLC precolumn under these conditions, since the use of slower flow rates and longer elution times did not result in any significant loss of analytes from the precolumn. The third flow rate considered was that used for analyte elution from the RPLC precolumn and analytical column. This was examined for a mobile phase containing 27% (v/v) 2-propanol in pH 3.0, 0.10 M potassium phosphate buffer, as discussed previously. As would be expected for a concentration-based detector such as a UV/visible absorbance monitor, the injection of a fixed amount of atrazine (500 µL of a 10 µg/L sample) gave a proportional decrease in peak area as the flow rate was increased from 0.5 to 1.5 mL/min. Under these conditions, the total analysis time for the system was reduced from 11.5 min at 0.5 mL/min to 8.5 min at 1.5 mL/min. As a compromise between detection limits and speed, a flow rate of 0.75 mL/min was used for the RPLC columns during the rest of this study. This gave a total analysis 810 Analytical Chemistry, Vol. 76, No. 3, February 1, 2004

Figure 5. Typical chromatogram obtained by the portable immunoextraction/RPLC system for the analysis of triazines in surface water samples. The atrazine concentration in these samples was ∼5 µg/L. Other details are given in the text. The small fluctuations in peak size between injections were attributed to variations in the detector response since all peaks in these chromatograms, including those for the solvent and analyte, gave the same proportional change in size over time.

time of less than 10 min for the elution of all analytes of interest from the portable system. System Evaluation. To determine the effectiveness of the immunoextraction column in removing contaminants from samples, groundwater was injected onto the portable system in both the presence and absence of this column. The results are shown in Figure 4. As can be seen from this result, the RPLC columns alone gave a high sample background and were not sufficient for the measurement of triazine compounds. But when the immunoextraction column was also employed, atrazine and related compounds could easily be detected at levels down to the low parts-per-billion range. Other items considered were the sample sizes and concentrations that were used with the portable device. In previous work, 250-µL samples were used with a similar benchtop system for the detection of atrazine at the low parts-per-billion level (µg/L).29 However, the miniaturized detector used in the current system was less sensitive than that used in the original benchtop device. To compensate for this, the sample volume was increased until an acceptable limit of detection was again obtained. As noted in earlier studies,28 this approach is possible because the tight binding of most antibody columns gives them a response related to the moles of an applied analyte rather than to its concentration. This, in turn, allows an increase in sample volume to be used to provide a decrease in the concentration-based lower limit of detection. The sample sizes examined for use with the immunoextraction column were 0.25, 0.50, 1.00, and 2.00 mL. The desired detection limit for this system was less than 1.0 µg/L atrazine, which was well below the 3 µg/L level allowed by the U.S. EPA in drinking water. This limit of detection could be obtained by using a 500µL sample, which gave a signal-to-noise ratio (S/N) of 5.1 for 0.5 µg/L atrazine and a lower limit of detection of 0.2 µg/L at S/N ) 2. Similar limits of detection (0.2-0.25 µg/L) were obtained for simazine and cyanazine under these conditions. Some early chromatograms generated with the portable system are shown in Figure 5. In this particular example, one water

Figure 6. Calibration curves obtained for atrazine (4), simazine (]), and cyanazine (0) on the portable HPIAC/RPLC system. The bestfit parameters were as follows: the slopes for atrazine, simazine, and cyanazine were 1.85 × 10-4, 1.22 × 10-4, and 7.18 × 10-5, respectively, and the intercepts were 1.64 × 10-3, 9.47 × 10-4, and 7.79 × 10-4, respectively. The experimental conditions are given in the text.

sample was injected every 10 min, with atrazine eluting within 13 min after injection. However, this analysis time could be further reduced if desired. For instance, it was found that the analysis of atrazine alone could be performed by injecting one sample every 4 min, giving a total analysis time of less than 8 min. For the analysis of all three triazine compounds, a total run time of 10 min was later used. The peak shown in Figure 5 for atrazine represents a concentration of ∼5 µg/L. The changes in signal at the valve switching events at 8, 18 and 28 min were due to passage

of the pH 3.0, 0.10 M phosphate elution buffer from the RPLC precolumn and through the absorbance detector.28,29 Figure 6 shows some calibration curves for the field-portable system. As already stated, the lower limit of detection for atrazine in a 500-µL sample was 0.2 µg/L at S/N ) 2. The lower limits of detection for simazine and cyanazine were 0.25 and 0.2 µg/L. The lower limits of quantitation for these analytes (defined here as the analyte concentration where S/N ) 5) were 0.5, 0.6, and 0.5 µg/L for atrazine, simazine, and cyanazine, respectively. The linear range for all these analytes extended up to 20 µg/L, and the dynamic range extended beyond 100 µg/L (see Figure 6). The correlation coefficients over the linear range (n ) 7 points) were 0.995, 0.979, and 0.975 for atrazine, simazine, and cyanazine, respectively. Previous studies have shown that the use of this immunoextraction/RPLC approach in a benchtop device gives good correlation versus reference methods such as GC/MS.28,29 In this current study, a comparison was made between the performance of this benchtop system under controlled conditions and the portable device when it was operated in the field under variable atmospheric conditions and with a transportable power supply. The results are shown in Figure 7. In this experiment, 16 split samples were tested that contained atrazine concentrations between 1 and 10 µg/L. The two systems gave good agreement, with a correlation coefficient of 0.94, a best-fit slope of 0.87 ((0.05), and a best-fit intercept of 1.6 ((1.0). From this, it was concluded that the portable system gave results equivalent to those obtained under more stable conditions in the laboratory.

Figure 7. Correlation of the results obtained by the immunoextraction/RPLC system in the field and in the laboratory. The solid line is the best-fit response to the results, which gave a slope of 0.87 ((0.05), an intercept of 1.6 ((1.0), and a correlation coefficient of 0.94. The dashed line is shown as a reference and is the response expected if there were a perfect correlation between these two sets of results.

Analytical Chemistry, Vol. 76, No. 3, February 1, 2004

811

Spiked recovery experiments were also performed with the portable device, using environmental water samples that contained various concentrations of atrazine. This involved the use of 20 samples spiked with 1-15 µg/L atrazine. These samples contained amounts of atrazine that ranged from low levels (i.e., below the U.S. EPA drinking water limit) to high concentrations (i.e., those above the regulatory limit). The correlation coefficient between the measured and actual amounts of atrazine in these samples was 0.95, with a best-fit slope of 0.92 (( 0.05) and an intercept of 0.9 (( 0.5). This indicated that the system had good accuracy when used to measure the amount of atrazine in such samples. The within-day precision of the portable system was examined by making 20 sequential injections of a spiked water sample. This gave a precision of (15.8% at an atrazine level of 10 µg/L, which was considered adequate for work with this device. The day-today precision was evaluated with a 10 µg/L atrazine sample analyzed over the course of 7 days. This gave a precision of (16.4%, which was also considered acceptable for this device. Previous studies with the same immobilized antibodies as used in this report have found that these can be used in laboratory systems for up to one year and over 700 injection cycles for aqueous standards and surface water samples if the antibodies are stored at 4 °C when not in use.28,29 For the portable system created in this work, it was found that the antibody lifetime was still relatively long, with the immunoextraction column being stable for six months if stored in pH 7.0, 0.1 M potassium phosphate buffer between testing cycles (typically overnight). There was a decrease in the maximum number of usable cycles for this column when it was used in field studies with surface water samples (i.e., 350-400 injections). This decrease was most likely due to the greater variation in temperature when such a column was used in an outside environment. For instance, the typical laboratory temperature during the course of this study was 70-75 °C, while the outside temperatures during the field studies varied from 40 to 90 °C. This not only affected the lifetime of the immunoaffinity column but also made it necessary to periodically inject a calibration standard to adjust for the resulting changes in retention on the RPLC columns. The long-term stability of the portable device was further examined by comparing the calibration curves generated with this instrument as it was moved from one location to the next. In this study, the triplicate analysis of standards containing 1-10 µg/L atrazine gave curves with slopes that varied by less than 5%. This indicated that the calibration curve did not have to be reevaluated at each site. Instead, the instrument could simply be calibrated once a day or as desired, even when it was used at multiple locations. During its use in field studies, it was found that the system could be easily transported and handled by a single person. The stabilization of the system within its case eliminated any problems regarding the movement or breakage of its components. The actual instrumentation took up a space that was 40 cm long, 36 cm high and 36 cm deep. The case that contained this instrumentation was 54 cm long, 38 cm high, and 50 cm deep. The total weight of the system plus its case and four-wheeled transportation cart was 35 lb, with the generator adding another 45 lb. Due to its self-containment, this device could be easily transported on 812 Analytical Chemistry, Vol. 76, No. 3, February 1, 2004

Figure 8. (a) Time course study of triazine herbicides at a single location near Ashland, NE, and (b) analysis of triazine herbicides at several locations near Lincoln, NE, during a single day. The largest peak in (a) was atrazine and the largest peak in (b) was cyanazine, with the smaller peaks representing other triazines or triazine degradation products.

its cart over a variety of terrain conditions, including soft earth and high grass areas. When arriving at a site, it took ∼3 min to place the instrument onto the cart. Once at the desired testing site, 2 min was required to start the generator, power up the system, and begin solvent flow through the columns. Since the solvent reservoirs were not disconnected from the pump during transport, it typically took less than 10 min to establish a baseline and begin analysis. The steps involved in setting up this system at each site involved opening the transportation case, mounting the solvent reservoirs, firing up the generator, and plugging in the instrument and computer. The reverse procedure was all that was required prior to moving the instrument to the next site. One application explored for this instrument was its use at a single site to monitor the changes in herbicide levels over time. This is shown in Figure 8a, in which this device was used to routinely measure triazine levels in a creek near Ashland, NE, over the course of seven months. This covered the application season for atrazine, which extends from April through the beginning of June. In this particular example, the recorded atrazine levels in the April and May samples were 17 ((6) and 30 ((4) µg/L, as determined by analyzing standards on the same dates as these samples (Note: the larger peak seen for April versus May was due to a small decrease in binding capacity for the

immunoextraction column between these two dates.). It can be seen from Figure 8a that atrazine could even be detected as late as October. However, in this latter case, the amount of atrazine was below the limit of quantitation for the system. In another study, the portable device was used to map the local changes in herbicide concentrations during a single day. This is illustrated in Figure 8b. This particular study was performed in June 2003 at four sites along the Salt Creek near Lincoln, NE. These sites were located along a 20-mi stretch and were examined within a time period of 4 h. As shown in the figure, these samples showed variable levels of triazine and triazine degradation products. On the basis of these data it was possible to identify areas with increased herbicide concentrations and to begin the location of any point sources for these herbicides along the path of the creek. CONCLUSIONS This work examined the development and evaluation of a portable immunoextraction/RPLC system for the analysis of herbicide residues in groundwater and surface water. The final device gave good correlation versus a previous benchtop system and was capable of simultaneously measuring several triazine compounds in a single water sample. The total analysis time for these compounds was less 10 min; however, this system could be used to analyze a specific herbicide (e.g., atrazine) with shorter analysis times. The limits of detection for 500-µL samples were 0.2-0.25 µg/L, with linear and dynamic ranges that extended up to 20 and 100 µg/L, respectively. This portable system has several advantages over traditional immunoassays for the screening of environmental samples. One major advantage is the ability of this method to differentiate

between structurally similar components, such as the degradation products of a herbicide.28,29 By interfacing the immunoextraction column with RPLC columns, it is also possible to obtain simultaneous information for herbicides within the same family (e.g., atrazine, simazine, and cyanazine). Other advantages of this method include its speed and precision versus standard immunoassays. The approach described here can be modified to analyze other pollutants of interest in water or soil samples. All that is necessary to do this is to employ antibodies against the desired agents and to adjust the mobile-phase conditions for their capture, elution, and separation. It was found in this report that the speed and precision of this method make it a viable option for the field analysis of environmental samples. With some modifications to allow for remote data access and control, this portable device could also be used for well monitoring or the control of water purification systems. ACKNOWLEDGMENT The authors thank Dr. James Carr and Stacy Stohl at the University of Nebraska for some of the samples examined in this work and for their help in the field studies. The authors also thank the Lincoln Water Treatment Facility in Ashland, NE, for access to the groundwater testing sites. This work was supported by an exploratory grant from the U.S. Environmental Protection Agency.

Received for review August 12, 2003. Accepted November 13, 2003. AC030298M

Analytical Chemistry, Vol. 76, No. 3, February 1, 2004

813