Anal. Chem. 1997, 69, 2640-2645
Renewable-Reagent Electrochemical Sensor for Monitoring Trace Metal Contaminants Joseph Wang,* Jianmin Lu, Dengbai Luo,† Jianyan Wang, Mian Jiang,‡ and Baomin Tian
Department of Chemistry & Biochemistry, New Mexico State University, Las Cruces, New Mexico 88003 Khris Olsen
Environmental Sciences Department, Batelle PNL, Richland, Washington 99352
This work extends the concept of in situ electrochemical stripping sensors to environmentally important metals that are not readily accumulated by amalgamation. A renewable-reagent sensor has thus been designed to accommodate the complex formation and adsorptive accumulation steps of adsorptive stripping protocols. Such flow probe relies on the delivery of a ligand solution through a microdialysis sampling tube and transport of the resulting complex to a downstream adsorptive stripping detector. The integrated membrane sampling/adsorptive stripping sensor is characterized, optimized, and tested in connection with the monitoring of trace uranium and nickel using the propyl gallate and dimethylglyoxime chelating agents, respectively. Experimental variables, including the reagent delivery rate and ligand concentration, are explored. The microdialysis sampling step minimizes the interference of surface-active macromolecules and extends the linear dynamic range compared to conventional adsorptive stripping measurements. Detection limits of 1.5 × 10-8 M nickel and 4.2 × 10-8 M uranium are obtained following 5- and 20-min adsorption times. A relative standard deviation of 1.7% is obtained for prolonged operations of 20 runs. The applicability to assays of river water and groundwater samples is demonstrated. The renewable-reagent adsorptive stripping sensor holds great promise for remote monitoring of various trace metals (via a judicious selection of the ligand). Current interest in metal contaminants has resulted in an ever increasing demand for their determination in a variety of environmental matrixes. Continuous in situ measurementss performed in the natural environmentsare preferred, as they provide an early detection of trace metal contaminants while minimizing errors, labor and cost associated in the collection, transport, and storage of individual samples for subsequent laboratory analysis. Electrochemical stripping analysis has long been recognized as a very useful technique for the determination of trace metals in environmental samples1-3 and is very suitable for the task of
on-site metal analysis.4 The growing needs for field monitoring of priority metal pollutants have led to the development of automated flow stripping systems5-8 and, more recently, to the introduction of hand-held metal analyzers.4,9 A more attractive in situ approach is to immerse the stripping electrode directly in the natural water matrix. The ability to deploy submersible electrodes for in situ monitoring of trace metals has been demonstrated recently in our laboratory.10,11 Such a remote stripping operation has relied on the coupling of gold microelectrodes with the potentiometric stripping mode and a novel probe design (involving a special electrode housing, environmentally sealed connectors, and a 50ft-long shielded cable). Real-time monitoring of copper and lead has thus been documented in various field trials. A submersible flow cell has also been developed for in-depth monitoring of heavy metals such as lead or cadmium.12 These concepts are suitable for the in situ detection of other heavy metals (e.g., mercury, cadmium, selenium, or arsenic) that can be plated electrolytically onto the gold or mercury working electrodes. The objective of the present paper is to extend the concept of remote metal monitoring toward additional environmentally significant metals that cannot be electrodeposited. Such extension is achieved by employing adsorptive stripping procedures,13-15 involving the formation and adsorptive accumulation of appropriate complexes of the target metal. The use of alternative (nonelectrolytic) accumulation schemes offers great promise for monitoring priority metal pollutants, including uranium, chromium, nickel, cobalt, aluminum, or iron. To achieve the goal of remote adsorptive stripping operation, we rely on renewable chemistry, involving controlled reagent (ligand) delivery, in a manner
† On leave from Department of Chemistry, South-Central University for Nationalities, Wuhan 430074, China. ‡ On leave from Department of Chemistry, Wuhan University, Wuhan 430072, China. (1) Nurnberg, H. W. Anal. Chim. Acta 1984, 164, 1. (2) Wang, J. Environ. Sci. Technol. 1982, 16, 104A.
(3) Bately, G. E. Mar. Chem. 1982, 12, 107. (4) Wang, J. Analyst 1994, 119, 763. (5) Zirino, A..; Lieberman, S. H.; Clavell, C. Environ. Sci. Technol. 1978, 12, 73. (6) Wang, J.; Ariel, M. Anal. Chim. Acta 1978, 99, 89. (7) Jagner, D. Trends Anal. Chem. 1983, 2, 53. (8) Luque de Castro, M. D.; Izquierdo, A. Electroanalysis 1991, 3, 457. (9) The Metalyzer 3000. ETG Inc., Baltimore, MD, 1995. (10) Wang, J.; Larson, D.; Foster, N.; Armalis, S.; Lu, J.; Rongrong, X.; Olsen, K.; Zirino, A. Anal. Chem. 1995, 67, 1481. (11) Wang, J.; Foster, N.; Armalis, S.; Larson, D.; Zirino, A..; Olsen, K. Anal. Chim. Acta 1995, 310, 223. (12) Tercier, M.; Buffle, J. Electroanalysis 1993, 5, 187. (13) Wang, J. Voltammetry Following Nonelectrolytic Preconcentration. In Electroanalytical Chemistry; Bard, A. J., Ed.; Dekker: New York, 1989; Vol. 16. (14) Paneli, M.; Voulgaropoulos, A. Electroanalysis 1993, 5, 355. (15) Kalvoda, R.; Kopanica, M. Pure Appl. Chem. 1989, 61, 97.
2640 Analytical Chemistry, Vol. 69, No. 14, July 15, 1997
S0003-2700(97)00022-X CCC: $14.00
© 1997 American Chemical Society
Figure 1. (A) Schematic diagram of the renewable-ligand adsorptive stripping electrochemical sensor: (a) Plexiglas body; (b) glassy carbon disk; (c) reagent inlet; (d) reagent outlet; (e) microdialysis sampling tubings (band of four cellulose fibers); (f) Teflon drainage tubing; (g) plastic cap; (h) Vycor disk. (B) The entire flow system.
analogous to that employed in renewable fiber-optic devices.16-18 A new electrochemical probe design, involving appropriate delivery and drainage capillaries, a microdialysis sampling, and mercury-coated glassy carbon electrode (Figure 1), has thus been constructed, characterized, optimized, and tested. Such new monitoring capability is illustrated in the following sections in connection with adsorptive stripping measurements of uranium and nickel. Besides its great potential for remote monitoring of additional metals, the integrated membrane sampling/stripping sensor prevents interferences from surface-active materials and greatly extends the linear range (compared to conventional adsoprtive stripping measurements). EXPERIMENTAL SECTION Probe Design. A schematic diagram of the adsorptivestripping flow probe is illustrated in Figure 1A. The probe is based on a Plexiglas cylindrical body [(a), 20.3-mm diameter, 50.8-mm length] that accommodates the glassy carbon disk working electrode and the reagent delivery and drainage capillaries. The glassy carbon electrode [(b) 3-mm diameter, Model MF-2012, BAS Inc.] was inserted through a 5.7-mm diameter hole drilled in the center of the Plexiglas cylinder. The electrode terminates in a thin-layer channel at the base of the sensor body. Such 80-µL channel was formed by fixing (with epoxy) a plastic end cap (g). A 3-mm-diameter Vycor disk (BAS Inc.) was fixed at the center of the end cap, below the thin-layer channel (h), to provide the (16) Berman, R.; Christian, G.; Burgess, L. Anal. Chem. 1990, 62, 2066 (17) Lin, Z.; Burgess, L. W. Anal. Chem. 1994, 66, 2544. (18) Kuhn, K. J.; Dyke, J. T. Anal. Chem. 1996, 68, 2890.
conductivity to the external reference (Ag/AgCl) and Pt wire counter electrodes (located in the sample solution). The dialysis sampling tube (e) and the Teflon drainage capillary (f) were fixed to holes in the Plexiglas body. A portion of the Plexiglas body was removed to accommodate these sampling and drainage tubes. A band of four regenerated cellulose (RC) hollow fibers [molecular weight cutoff (MWCO) 13 000; 200-µm i.d.; Spectrum Medical Industries Inc.] serves for the dialysis sampling. A Teflon tubing connects the microsyringe pump (2.5-mL volume, Model MD100/MF-5127, BAS Inc.) with the inlet of the dialysis tubing [(c) and Figure 1B). The outlet of the dialysis tubing (d) is connected to the thin-layer working electrode channel through a hole in the Plexiglas body. Apparatus. Potentiometric stripping analysis (PSA) was performed with a TraceLab system (PSU20, Radiometer Inc.), in connection with an IBM PS/55SX computer. Reagents. Stock solutions (1000 mg/L) of nickel and uranium (atomic absorption standard, Aldrich) were diluted as required. Dimethylglyoxime (DMG) and propyl gallate (PG) were received from Aldrich. A 0.02 M ammonia buffer solution (pH 9.2) containing 5 × 10-5 M DMG served as the receiving solution for the nickel system. A 0.05 M acetate (Aldrich) buffer solution (pH 4.5) containing 1 × 10-4 M PG was used for uranium measurements. Dodecyl sodium sulfate and Triton X-100 were obtained from J. T. Baker Chemical Co., and Arabic gum was received from Sigma. Rio Grande river samples were collected in Las Cruces, NM, while the groundwater sample was collected at the Handford site (Richland, WA). Tap water samples were obtained at the NMSU laboratory. All solutions were prepared from double-distilled water. All chemicals used were of analytical grade. Procedure. Monitoring of Nickel. The mercury film was preplated by immersing the polished glassy carbon disk electrode in a 10-mL cell containing a stirred 100 mg/L Hg2+ and 0.1 M HCl solution, for 5 min while the potential was held at -0.6 V. Following the mercury deposition, the glassy carbon electrode was incorporated into the flow probe body. The probe was then immersed into a 100-mL cell, containing 40 mL of a tap water solution. Adsorptive accumulation proceeded (usually for 2 min) while the potential was held at -0.7 V, and the DMG/ammonia buffer reagent solution was pumped at 10 µL/min. After the accumulation, the potentiogram was recorded by applying a suitable constant current. A “cleaning” step was followed by holding the sensor at -1.4 V for 10 s. The same procedure was repeated after spiking an aliquot of the nickel standard solution to the external tap water solution. Monitoring of Uranium. The glassy carbon-based mercury film electrode was prepared using the procedure reported above for the measurement of nickel prior to experiments by preplating mercury at -0.6 V for 5 min from a 100 ppm Hg2+/0.1 M HCl solution. The coated electrode was inserted into the flow probe body and was ready for use. The probe, along with the reference and counter electrodes, was then dipped in a 100-mL cell containing 40 mL of a 0.03 M NaCl solution. The reagent solution (1 × 10-4 M propyl gallate in 0.05 M acetate buffer solution, pH 4.5) was allowed to flow at 10 µL/min, while an accumulation potential of -0.05 V was applied for a selected time. After the accumulation step, a potentiogram was recorded by applying a constant negative current (-20 µA). The surface “cleaning” was carried out by holding the potential at -1.5 V for Analytical Chemistry, Vol. 69, No. 14, July 15, 1997
2641
Figure 2. Chronopotentiograms for 117 µg/L nickel (A) and 1000 µg/L uranium (B) following (a) 0; (b) 30; (c) 60; (d) 90; (e) 120; and (f) 150-s accumulation. Constant current -5 (A) and -20 µA (B); accumulation potential -0.7(A) and -0.05 V (B); “cleaning”, -1.4 V for 10 s (A) and -1.5 V for 1 min (B); flow rate, 10 µL/min.
60 s. Following each addition, a 3-min “waiting” period was used to allow for the microdialysis sampling and transport to the working electrode compartment. All experiments were carried out at room temperature. Conventional adsorptive stripping potentiomentric measurements, used for comparison, were carried out in a 10-mL cell (Model VC-2, BAS), using a stirred (∼500 rpm) sample solution, and accumulation periods, stripping currents, and cleaning steps similar to those used with flow probe. RESULTS AND DISCUSSION The renewable-reagent adsorptive stripping sensor (shown in Figure 1) relies on continuous delivery of the ligand, its complexation reaction with the metal “collected” in the dialysis sampling tube, transport of the complex to the working electrode compartment, and chronopotentiometric detection of the accumulated complex. Established adsorptive stripping protocols for trace uranium19 and nickel,20,21 based on complexation with propyl gallate and dimethylglyoxime, respectively, were used for characterizing and testing the new stripping probe. Figure 2 displays chronopotentiograms for nondeaerated solutions of 0.12 mg/L nickel (A) and 1.0 mg/L uranium (B), obtained with the flow probe, following different adsorption times (0-150 s, a-f). The nickel/DMG and uranium/PG chelates yield well-defined reduction peaks at -1.14 and -0.63 V, respectively. The longer the accumulation time, the more the amount of complex is adsorbed onto the working electrode, and the larger is the peak height. While the nickel peak rises rapidly with time at first and then more slowly, the uranium signal increases over the entire time scale tested. Convenient measurements at the milligram per liter level are thus feasible following very short (1-2 min) accumulation periods. No response is observed without the adsorptive accumulation. Such favorable background response, obtained for nondearated samples, eliminates the need for a time-consuming deaeration step, hence making the chronopotentiometric stripping (19) Wang, J.; Wang, J.; Lu, J.; Olsen, K. Anal. Chim. Acta 1994, 292, 91. (20) Pihlar, B.; Valenta, P.; Nurnberg, H. W. Fresenius Z. Anal. Chem. 1981, 307, 337. (21) Brett, C. M.; Oliveira Brett, A. M.; Pereira, J. L. Electroanalysis 1991, 3, 683.
2642 Analytical Chemistry, Vol. 69, No. 14, July 15, 1997
Figure 3. Effect of delivery flow rate (A) and ligand concentration (B) on the nickel (a) and uranium (b) signals: (a) 234 µg/L nickel and 5 × 10-5 M DMG; (b) 1000 µg/L uranium and 1 × 10-4 M PG. Flow rate (B) 10 µL/min; accumulation -0.7 V for 2 min (a) and -0.05 V for 3 min (b); cleaning -1.4 V for 10 s (a) and -1.5 V for 1 min (b); stripping current -10 (a) and -20 µA (b).
mode attractive for potential field applications. A short cleaning period [at -1.4 (A) or -1.5 V (B)] is sufficient for desorbing the complex prior to the next measurement. The influence of various experimental variables affecting the response of the renewable-reagent sensor was tested. Figure 3A shows the dependence of the adsorptive stripping response of nickel (a) and uranium (b) on the flow rate of the reagent solution. Different profiles are observed for these metals. While the nickel response increases rapidly upon raising the flow rate between 5 and 10 µL/min and then more slowly, the uranium signal decreases sharply upon increasing the flow rate. Such different profiles are attributed to the fact that the reagent-solution flow rate affects (in a different fashion) various steps of the sensor operation, including the metal sampling, complex formation, and adsorptive accumulation. Obviously, the target metal and the ligand used have a profound effect upon the transport rate through the membrane and upon the rate of the complex formation (leading to different flow rate effects for the different metals). Apparently, the slower sampling and complex formation of the uranyl ion dominate its flow rate profile. The nickel profile, in contrast, is dominated primarily by transport of the nickel/DMG chelate toward the surface (in view of its more facile collection and complexation). The influence of the ligand concentration (in the receiving solution) is shown in Figure 3B. For both nickel (a) and uranium (b), the response rises sharply (to a steady-state value) with the ligand concentration at first and then decreases slowly. Compared to conventional adsorptive stripping measurements, the ligand and
Figure 4. Effect of dodecyl sodium sulfate (A) and Triton X-100 (B) on the nickel stripping signal, and of Arabic gum (C) on the uranium response using a batch cell (a, b) and with the renewable sensor (c, d). Surfactant concentration, 0 (a, c) and 2.5 mg/L (A, b and d), 1.3 mg/L (B, b and d) and 2000 mg/L (C, b and d); Ni concentration 40 (a, b) and 250 µg/L (c, d); U concentration 80 (a, b) and 2000 µg/L (c, d); (A, B) accumulation -0.7 V for 120 (A, a-d; B, c and d) and 10 s (B, a and b); stripping current -10 µA; cleaning, -1.4 V for 10 s. (C) accumulation -0.05 V for 180 (C, c and d) and 60 s (C, a and b); stripping current -20 µA; cleaning -1.5 V for 1 min.
its level affect the response also through their influence on the microdialysis collection of the target metal. Adsorptive stripping measurements commonly suffer from interferences by surface-active materials present in environmental samples.13 Such substances have a marked effect on the adsorptive stripping response owing to competitive adsorption for surface sites. The microdialysis sampling of the renewable sensor greatly minimizes such matrix effects. Because the membrane does not favor permeation of large macromolecules, the flow probe offers good resistance to surfactant effects. Such resistance is illustrated in Figure 4. Conventional adsorptive stripping measurements (left) result in a substantial depression of the nickel (B) and uranium (C) peaks following the addition of Triton X-100 and Arabic gum [compare (a) and (b)]. In addition, the presence of dodecyl sodium sulfate causes a severe distortion of the nickel peak (A). In contrast, no change of the nickel or uranium is observed at the renewable-reagent sensor in the presence of similar levels of these surfactants (right). Such protective action is illustrated also in Figure 5. Using the conventional stripping protocol b, the nickel [A (O)] and uranium [B (O)] peaks decrease rapidly upon raising the surfactant concentration (with 90% depressions at 1.3 and 1000 mg/L Triton X-100 and Arabic gum, respectively). In contrast, the flow probe offers a highly stable uranium response up to 2000 mg/L Arabic gum [B (9)]. The nickel peak is also not affected by the Triton X-100 concentration up to 0.9 mg/L but displays a 20% loss at 1.3 mg/L Triton X-100 [A (9)]. Dialysis membrane-covered electrodes have been used for suppressing surfactant effects in
Figure 5. Effect of surfactant on nickel (A) and uranium (B) measurements in the remote sensor with dialysis membrane (9) or in the batch cell (O). For Ni system, Triton X-100 was used. Ni concentration 250 µg/L (9) and 40 µg/L (O); Constant current -10 (9) and -15 µA (O); accumulation, -0.7 V for 120 (9) and 10 s (O); cleaning -1.4 V for 10 s. For U system, Arabic gum was used for this experiment. Concentration 2000 (9) and 80 µg/L (O); constant current -20 µA (B); accumulation -0.05 V for 180 (9) and 60 s (O); cleaning -1.5 V for 1 min.
conventional stripping measurements,22 but not in connection with nonelectrolytic accumulation or remote monitoring. The renewable-reagent adsorptive stripping sensor displays a well-defined concentration dependence. For example, a series of 10 concentration increments of 200 µg/L uranium yielded a linear calibration plot over the entire range [slope 0.041 ms‚L/µg (10 ms/µM); correlation coefficient 0.997; 4-min adsoprtion; constant current -10 µA; other conditions, as in Figure 2B]. Similarly, a linear relationship was obtained between the nickel peak area and its bulk concentration over the 60-600 µg/L range (sensitivity, 1.025 ms‚L/µg (60 ms/µM); correlation coefficient 0.999; 2-min adsorption; constant current -20 µA; other conditions, as in Figure 2A). Apparently, conditions of low surface coverage (linear adsorption isotherm) exist. The very wide linear range (compared to conventional adsorption stripping measurements) is attributed to the “built-in” dilution action associated with the microdialysis sampling. An analogous conventional (batch) adsorptive stripping calibration experiment for uranium yielded a nonlinear calibration plot, with a leveling off at 600 µg/L; the sensitivity (slope of linear portion) was 20-fold higher than that of the flow probe (not shown). The different sensitivities observed with the renewablereagent sensor for uranium and nickel are attributed in part to their different transport rates through the microdialysis membrane (with a more facile collection of the smaller nickel ion). The use of sampling membranes with higher molecular weight cutoffs (22) Aldstadt, J. H.; King, D. F.; Dewald, H. D. Analyst 1994, 119, 1813.
Analytical Chemistry, Vol. 69, No. 14, July 15, 1997
2643
Figure 6. Adsorptive stripping response for 8 µg/L Ni (A, b) and 25 µg/L U (B, b) following 5- (A) and 20-min (B) accumulation, along with the corresponding blank signals (a). Accumulation potential -0.7 (A) and -0.05 V (B); stripping current -5 (A) and -20 µA (B); cleaning -1.4 V for 10 s (A) and -1.5 V for 1 min (B); flow rate 10 µL/min.
would increase the recovery of the target metals and would further enhance the sensitivity. In addition, the sensitivity and the dynamic range may be changed by adjusting the ligand delivery rate or the ligand concnetration. Despite the internal dilution, the renewable-flow probe results in extremely low detection limits. These were estimated from the adsorptive stripping response for 8 µg/L nickel and 25 µg/L uranium (Figure 6(b, A and B, respectively)). Well-defined peaks are observed for these low levels following 5- (A) and 20- (B) min accumulation. Detection limits of 0.9 µg/L (1.5 × 10-8 M) nickel and 10 µg/L (4.2 × 10-8 M) uranium were estimated from the signal-to-noise characteristics of these data (S/N ) 3). Even lower detection limits are expected in connection with longer accumulation times, use of more permeable membranes, or a stopped-flow operation. The renewable-ligand adsorptive stripping sensor yields a reproducible and stable response. High stability (RSD ) 1.7%) was observed in a prolonged (60 min) unbroken series involving 20 successive measurements of 400 µg/L nickel (2-min accumulation; not shown). Similarly, a stable uranium peak (RSD ) 4%) was obtained for 10 repetitive measurements of a 1 mg/L uranium solution (4-min adsoprtion; not shown). These data reflect the efficiency of the electrochemical cleaning step, i.e., regeneration of “metal-free” surface prior to each run. Figure 7 demonstrates the applicability of the renewablereagent sensor to measurements of trace metals in environmental samples. While no response is observed for the unpolluted river water (A) or groundwater (B) samples, subsequent additions of 0.11 mg/L nickel (A, b-d) or 0.5 mg/L uranium (B, b-d) yielded defined adsorptive stripping peaks in connection with short accumulation times of 2 (A) and 3 (B) min, respectively. No major interferences are indicated from the response of the unspiked sample. In conclusion, we have demonstrated that renewable-reagent flow probes are very suitable for adsorptive stripping electrochemical sensing of trace metals. Such integration of microdialysis sampling with adsorptive stripping detection extends the scope of remote stripping sensors of additional metals that are not readily electroplated. Near-real-time monitoring of numerous metals can thus be accomplished via a judicious choice of the complexing ligand. The remote monitoring capability is coupled to minimization of interferences from surface-active materials and extension of the linear dynamic range (compared to conventional adsorptive stripping measurements). We are currently integrating 2644 Analytical Chemistry, Vol. 69, No. 14, July 15, 1997
Figure 7. Stripping response for a river water sample (A) and for a groundwater solution (B) with increasing concentrations of nickel in steps of 117 µg/L (A, b-d) and of uranium in steps of 500 µg/L (B, b-d), along with the response for the unspiked sample (a). Flow rate 10 µL/min; accumulation -0.7 V for 2 min (A) and -0.05 V for 3 min (B); cleaning, -1.4 V for 10 s (A) and -1.5 V for 1 min (B); constant current -10 (A) and -20 µA (B).
Figure 8. Renewable-reagent electrochemical probe, integrating the three-electrode system with microdialysis sampling. (A) glassy carbon electrode; (B) reference eleectrode; (C, D) O-rings; (E) reagent inlet; (F) microdialysis sampling tubings; (G) reagent outlet; (H) counter electrode; (I) platinum wire.
the reference and counter electrodes within the body of the flow probe (Figure 8), coupling it to a long shielded cable, and designing multichannel sensor arrangements for the simultaneous monitoring of several metals (with each channel carrying the desired complexation reaction). Preliminary results with our three-electrode integrated probe (of Figure 8) indicate a similar
performance. Besides the integration advantage, such a probe obviates the need for a conductive sample matrix (in view of the conductivity of the reagent solution). Considering the versatility of the renewable-flow adsorptive stripping concept, we expect its adaptation to various environmental or industrial monitoring scenarios. Conventional stripping measurements, which also suffer from interference of surfactants,23 may also benefit from the isolation of the target metals from large macromolecules. Other electrochemical detection schemes (biosensing of pollutants) may also benefit from the versatility of the renewable reagent strategy. (23) Sahlin, E.; Jagner, D. Anal. Chim. Acta 1996, 333, 233.
ACKNOWLEDGMENT This work was supported by the U.S. Department of Energy (DOE) Grant DE-FG07-96ER62306, and by the DOE Waste Management Education and Research Consortium (WERC).
Received for review January 7, 1997. Accepted April 30, 1997.X AC970022C X
Abstract published in Advance ACS Abstracts, June 15, 1997.
Analytical Chemistry, Vol. 69, No. 14, July 15, 1997
2645
