Technical Notes Anal. Chem. 1995, 67, 1481-1485
Remote Electrochemical Sensor for Trace Metal Contaminants Joseph Wang,* David Larson, Nancy Foster, Saulius Armalis, Jianmin Lu, and Xu Rongrong Department ofChemistry and Bimhemistty, New Mexico State University, Las Cruces, New Mexico 88003
Khris Olsen Environmental Sciences Department, Batelle PNL, Richland, Washington 99352 Albert0 Zirino Marine Environment Branch, Code 522, Naval Ocean Systems Center, San Diego, Califomia 92152
An electrochemicalsensor technolo@, based on stripping potentiometry at a 100-pm gold fiber electrode, has been developed for remote detection of trace metal contaminants. The new probe circumvents technical difliculties, including the need for mercury surfaces, removal of oxygen, or forced convection, that previously prevented the adaptation of stripping analysis to in situ remote operations. The gold microelectrode, connected to a long communication cable, allows convenient measurements of trace copper, mercury, lead, or selenium at 3-4-min intervals and instrument/sample distances up to 32 m. The remarkable sensitivity of the probe is coupled with multielement capability, high stability, and simplicity. Field demonstrations involving both downhole groundwater monitoring and in situ shipboard seawater analysis were successful. The new probe should be extremely useful for the characterization of priority metal pollutants in hazardous waste sites and for the immediate detection of sudden metal contaminations.
Contamination of hazardous waste sites and groundwater with toxic heavy metals (e.g., Hg, Pb, U, As, Cr, and Al) represents a major national problem. Site monitoring and surveillance programs are required for a closer control of metal pollutants. The traditional use of atomic spectroscopy central laboratory measure ments of heavy metals is too expensive and time consuming.Also, samples often change composition during their collection, transport, and delay, ultimately producing unreliable results. Innovative field deployable methods are highly desired for the task of site characterization and remediation, as they minimize the huge labor or analytical costs and provide timely data for real-time emergencies and decision making. Chemical sensors are particularly attractive for providing real-time, remote monitoring of priority pollutants. While fiber-optic probes have been suggested for 0003-2700/95/0367-1481$9.00/0 0 1995 American Chemical Society
monitoring organic contaminants,' analogous remote monitoring of trace metals has not been demonstrated. Clearly, a costeffective metal-sensortechnology, capable of monitoring the metal both in time and in location, is needed to support the characterization and remediation of hazardous waste sites.2 In this paper we report on the design, characterization, optimization, and field deployment of a remote sensor for in situ monitoring of trace metals. 'The compact instrumentation and low power needs of electrochemical techniques satisfy many of the requirements for on-site metal analysis. Particularly attractive for in situ monitoring of metal contaminants is the remarkably sensitive technique of stripping anal~sis.~ 'The extremely low (subnanomolar) detection limits of stripping analysis are attributed to its "built-in" preconcentration step, during which the target metals are deposited onto the working electrode. The feasibility of using stripping analysis for field-based operations was demonstrated first by the US. Navy: which developed an automated flow system, based on a mercurycoated open tubular electrode, for continuous shipboard monitoring of trace metals in oceans. A useful and submersible flow-through cell (based on mercury drop or film electrodes and stripping voltammetry) was deployed by Buffle's group5for in situ metal monitoring in lakes and oceans. The present probe circumvents the need for solution pumping or mercury electrodes and offers greater simplification and miniaturization as needed for downhole groundwater monitoring. To accomplish this goal, the probe combines several advances in stripping methodologies. These include replacement of the traditional mercury electrodes with gold surfaces;6-9 the use of (1) Chudyk, W.; Carrabba, M.; Kenny, J. Anal. Cbem. 1985.57,1237. (2) Batiaans, G.; Haas, W. J.; Junk, G., Eds. Chemical Sensors: Technology Development Planning; US. Department of Commerce: Springfield, 1993. (3) Wang, J. Stripping Analysis: Principles, Instrumentation, and Applciations;
VCH Publishers: Deerfield Beach, FL, 1985. (4) Zirino, A; Lieberman, S. H.; Clavell, C. Enuiron. Sci. Tecbnol. 1978,12, 73.
(5) Tercier, M.; Buffle, J. Electroanalysis, 1993,5, 187. (6) Sipos, L;Golimowski, J.; Valenta, P.; Numberg, H. W. Fresenius'Z. Anal. Cbem. 1979,298,1. Analytical Chemistry, Vol. 67, No. 8, April 15, 1995 1481
Figure 1. Schematic diagram of the remote electrochemical sensor for trace metals: (a) reference electrode; (b) counter electrode; (c) gold wire working electrode; (d) male Teflon fittings; (e) female connectors; (f) threaded end of PVC housing; (9) PVC housing; (h) environmentally sealed connector; (i) PVC tubing; and (j) shielded cable.
microelectrodes, which offers efficient mass transport and minimizes natural convection effects, as well as allowing work in lowionic-strength natural waters;l0-l4 and incorporation of modern potentiometric stripping analysis (PSA), which eliminates the need for oxygen removal and minimizes the surfactant effects.l5JGSuch combination of gold fiber electrodes with PSA, in a specially designed sensing probe, thus facilitates remote operations by making solution stirring or deoxygenation, electrolyte addition, and mercury electrodes unnecessary. Direct immersion of the sensing electrode in the sample (in a manner analogousto fiberoptic devices) thus becomes feasisle. These electrochemical considerations have been coupled with the need for a compact and rugged probe, hence ensuring that the performance of the in situ sensor is comparable to that of established laboratory-based stripping instruments. Such realization of in situ, near-real-time detection of trace metals brings significant changes to the way by which these contaminants are monitored. We wish to report on these developments in the following sections. EXPERIMENTAL SECTION
Sensor Design. Figure 1shows a schematic diagram of the remote metal sensor. It consists of a three-electrode assembly (1Wpmdiameter, 5”-long gold microcylinder working electrode (c), silver-silver chloride reference (BASModel RE-4) (a), and a platinum wire counter (b)) in a WC housing which was (7) Wang, J.;Tan, B. Electroanalysis 1993,5,809. (8) Wang, J.; Sucman, E.; Tian, B. Anal. Chim. Acta 1994,286,189. (9) Gil, E. P.;Ostapczuk, P. Anal. Chim. Acta 1994,293,55. (10)Wang, J.; Zadeii, J. J. Electroanal. Chem. 1989,246,297. (11) Baranski, A. S.Anal. Chem. 1987,59,662. (12) Huiliang, J.; Hua, C.; Jagner, D.; Renman, L Anal. Chim. Acta. 1987,193, 61. (13) De Vitre, R R; Tercier, M. L; Buffle, J. A d . Proc. 1991,28,74. (14) Kounaves, S.P.;Deng, W.; Hallock, P.; Kovas, G.; Storment, C. Anal. Chem. 1994,66,423. (15)Jagner, D. Trends Anal. Chim. Acta 1983.2.53. (16)Ostapczuk, P.Anal. Chim. Acta 1993,273,35. 1482 Analytical Chemistry, Vol. 67, No. 8, April 15, 1995
connected to a long shielded cable via a Spin environmentally sealed rubber connector. The gold microelectrode was fabricated as previously described? Both the working and reference electrodes were sealed into Teflon fittings (d) and screwed into female coupling connectors (e), which were fixed with epoxy in the WC housing (f,g). Such arrangement allows easy replacement of these electrodes. The counter electrode was fixed permanently into the housing with epoxy. Electrical contact to the working and reference electrodeswas accomplished with the aid of brass screws and spring assemblies, contained inside 7 - m 0.d. copper tubes. The latter were placed within the female connectors and soldered to the copper wire contact. The other end of the copper wires was connected to the male environmentally sealed connector (h). Another copper wire was used to connect the platinum electrode to the environmentally sealed connector. The entire assembly was sealed into the WC housing. The 3-pin male connector (Newark Electronics) was connected to the receptable attached to the shielded cable 6). Such connectionpermits quick disconnectionof the electrode housing from the cable. The female connector was sealed in a WC tube (i) that provides additional stability. A 3cmdiameter WC tube was threaded onto the housing unit to protect the electrode assembly during field testings (not shown). Cables of different lengths, ranging from 8 to 32 m, were employed in accordance with the specific field application. Apparatus. Potentiometric stripping experiments were performed using the TraceLab potentiometric stripping unit (PSU 20 Radiometer, Copenhagen), in connection with an IBM PS/
55sx Reagents. All solutions were prepared with doubledistilled water. Stock metal solutions were prepared daily from the corresponding lo00 ppm atomic absorption standards (Aldrich). All experiments were carried out in unpreserved groundwater (from the Hanford Site, Richland, WA) or coastal seawater (from the mouth of San Diego Bay, CA; salinity,33%)solutions (without additional supporting electrolyte). Procedure. Laboratory experiments and field pre- and postcalibrations were performed in a 4WmL Erlenmeyer flask, containing relevant samples, into which the entire probe was immersed. Silicone grease was applied to all connectors (to provide sealing needed for preventing entry of solution into electrical contacts). Prior to the deposition step, the gold electrode underwent a “cleaning”procedure for 1-2 min at +0.5 to +1.0 V (depending on the target metals; see figure captions). Deposition proceeded in the quiescent solution for 1-2 min at potentials ranging from -0.3 to -0.7 V (depending on the soughtfor metal(s)). After the deposition period, the potentiogram was recorded by applying a constant oxidation current (in the 0.32.@@ range) and the BASE 3 or 4 commands of the TraceLab software for baseline treatment. A similar PSA protocol was employed when measurements were made in the field (for in situ monitoring of seawater or groundwater). RESULTS AND DISCUSSION
The probe design aims at addressingboth the electrochemical and environmentalrequirements, namely an optimized stripping performance, as well as compatibility with remote field work (e.g., downhole or shipboard monitoring). It thus consists of an electrode assembly connected (through environmentally sealed connectors) to a long, shielded communication cable. The latter assures also a negligible electrical noise even for large instrument/
A
b
C
‘5, W
7J v
0.6 0.4
.___. ._.__._ ___ ............... .-.........
B
-
P o T E N r m , (v)
Figure 2. PSA response of the remote probe: (A) seawater sample spiked with 5 pg/L lead, copper, and mercury; (8) groundwater sample spiked with 5 pg/L copper and 10 pgJL lead. Conditions: deposition for (A) 2 and (B) 1 min at -0.4 and -0.7 V, respectively; cleaning for (A) 1 and (B) 2 min at +0.8 and $0.5 V, respectively.
sample distances. The gold fiber electrode, where the deposition/ stripping reactions of interest take place, represents the heart of the probe. Such gold surfaces have been shown recently to offer convenient quantitation of environmentally relevant trace metals (such as copper, lead, selenium, and mer~ury).+~J~ The 1Wpm diameter electrode cylinder contiguration obviates the need for solution stirring during the deposition step and allows measurements in low-ionic-strengthmedia. Its coupling to a PSA operation also eliminatesthe need for a timeconsuming deoxygenation step. Such simplified operation thus meets many of the requirements for remote in situ chemical sensing. Detailed laboratory characterization of the probe (under simulated field conditions) and its successful field testings are described below. Figure 2 displays the stripping potentiometric response of the sensor to untreated seawater (A) and groundwater (€3) samples, spiked with 5-10 pg/L (ppb) levels of free lead, copper, and mercury. Well-definedand sharp peaks are observed, despite the use of nondeaerated samples, unstirred solutions or short (1-2 min) deposition periods. The well-resolved peaks (Up > 0.2 V) and the flat baseline allow convenient multielement determinations of low pg/L concentrations in these untreated samples. The exact peak potentials (around -0.2 (Pb), +0.2 (Cu), and $0.45 V (Hg)) depend on the extent of binding to complexing agents present in these samples. The copper peak (in the seawater medium) increased linearly with the preconcentration time (up to 10 min), while the lead one increased sigmoidally upon changing the deposition potential between -0.3 and -1.0 V (with a plateau above -0.5 V (not shown)). Overall, a deposition potential of -0.5 V allows simultaneous measurements of all three metals, while a lower one (-0.3 V) is preferred for the copper-mercury pair. (More positive deposition potentials extend the anodic potential window, as desired for measurements of metals with relatively positive stripping potentials.) Analogous stripping voltammetric measurements (using the rapid square-wave mode) yielded an inferior performance (with convenient quantitation only above the 10pg/L level) due to a higher baseline response associated with the gold surface reactions (not shown). The computerized PSA instrument is superior in addressing background contributions associated with the gold surface. Blank potentiograms (using prolonged 10-20-min deposition periods) indicated no apparent contamination from the PVC body or the connectors. The flat blank response indicates also that oxygen is not posing a background problem. Dissolved oxygen is participating, in part, in the metal strippingprocess (which is controlled mainly by the stripping current). (17) McLaughlin, K; Boyd, D.;Hua, C.;Smyth, M. Electroanalysis 1992,4, 689.
C
b
I
0.8
OB
POTENTIAL, (v)
Figure 3. Stripping potentiograms for groundwater solutions containing increasing concentrations of mercury (A) and selenium (B) in 20 pglL steps (a-d). Conditions: deposition for 1 min at (A) -0.3 and (B) +0.1 V; cleaning for 1 min at (A) +0.7 and (B) +1.0 V.
The analytical utility of the remote probe is based on the linear correlation between PSA response and the target metal. Figure 3 displays stripping potentiograms, following 1-mindeposition, for unstirred groundwater solutions containing increasing levels of (A)mercury(II) and (Ei) selenium 0 (in 20 pg/L steps). These peaks are part of a series of 10 concentration increments over the 10-100 pg/L range. The well-definedpeaks increase linearly with the metal concentration over the entire range. Linearity up to 100 pg/L was observed also in analogous calibration experiments for lead and copper (not shown). The dynamic range of the probe can be changed by adjusting the deposition time. Detection limits of 0.7 pg/L copper and 1.1pg/L lead were estimated from the response, following 1-min deposition, for a groundwater sample spiked with 5 and 10 pg/L of these metals (e.g., Figure 2B). Such detectability is very attractive for the monitoring of contaminated sites. Yet, application to unpolluted water (particularly open ocean) will require lower detection limits (which can be achieved by extending the deposition time or using a preconditioned gold film electrodeg). Overall, 3-4 min is required for the entire deposition/stripping/cleaning cycle (for quantitation at the 1-20 pg/L range), thus leading to near-realtime monitoring at a rate of 15-20 runs/h. High stability is another important requirement for a remote in situ probe. The short “cleaning”step ensures complete removal of the deposited metal at the end of each run and hence a “fresh” (analyte-free) gold surface prior to the next measurement. Hence, the depositiordstripping cycle leads to a reversible sensor behavior. Figure 4 displays the probe response during a long run of 20 successive measurements of copper (A) and lead @) in groundwater. A highly stable response is observed over these prolonged (50-70 min) operations. The relative standard deviations for these complete series were 2.1 (A) and 3.2% e).An analogous experiment using a lower level (10 pg/L) of copper in seawater yielded a relative standard deviation of 5.6%. High stability was observed also in the longer (2 days) field testings described below. The stable response in untreated natural water samples is attributed also to a minimal surfactant effect associated with the use of PSA and gold surfaces. The probe construction permits a fast and easy replacement of electrodes whenever needed. A proper sealing of all connectors (by coverage with silicone grease) is essential to achieve such stable response through prevention of liquid entry. Real-time monitoring capabilities will provide a valuable tool for probing dynamic processes in an aquifer or during remediation activities. The ability of the remote probe to follow sudden changes in the concentration of copper and lead is illustrated in Analytical Chemistry, Vol. 67, No. 8, April 15, 1995
1483
ao
760 o
1
I
I
A
'-
c
w
-u
v
n
v1 30
E
v
5
20
2
0 1 2 3 4 5 8 7 8 9 10 11 12 19 14 16 18 17 18 19 20 100 ,
-90 80
T
I
B
r
A I
o
1
z a
s
4 5
7 8
I
I
I
I
I
I
I
I
1
1
s i o 11 12 13 14 IS 16 17
1
18 19 20
RUN NUMBER Figure 4. Stability of 20 repetitive measurements of groundwater spiked with 50 pglL copper (A) and 40 pglL lead (B). Conditions: deposition for (A) 1 and (8) 2 min at -0.5 V; cleaning for 1 min at +0.6 V. Also shown are the first and last potentiograms of each series.
I
cu
I
cu
'-yP.-,.-+
0.6
10
e
/u -
A
0.2
POTENTIAL, (V) Figure 6. Field testings involving downhole groundwater monitoring (A) and shipboard seawater analysis (B). Deposition for (A) 1 and (B) 2 min at -0.3 and -0.4 V, respectively. Conditions: cleaning for (A) 2 and (B) 1 min at (A) +0.6 and (6)+0.8 V, respectively. See text for details.
hold great promise for possible precalibration and/or postcalibration using the same protocol employed during the in situ runs. The new sensor technology has moved beyond the laboratory environment to the field. Several on-site demonstrations were performed, including two groundwater monitoring wells and an in situ shipboard seawater analysis. F i e 6A displays a stripping potentiogram for a groundwater well at the Hanford site. Such an in situ PSA response was obtained at a depth of 13 m and is part of a continuous 3 h downhole operation at different depth intervals (11-18 m). A small copper peak (at ca. f0.25 V) was observed in these repetitive runs. Such response corresponds to 3 pg/L, as was confirmed in subsequent ICPMS measurements. No apparent electrical noise is observed despite the significant cable length. F w r e 6B illustrates the response of the in situ probe for seawater in San Diego Bay. Such measurement was made from a small boat, with the probe dangling on the side of the vessel just below the water surface. It is part of a 5 h in situ study at different locations in San Diego Bay. An on-boat precalibration experiment, employing a spiked seawater sample (two standard additions), was used for computing the concentration. Large copper and lead peaks, corresponding to 11 and 2 pg/L, respectively, are observed. Such relatively high levels are anticipated for this specific location (the south embayment "pockef' of Shelter Island), where discharge from pleasure boats and restricted circulation exist. Substantially lower copper and lead levels (of 0.9 and 0.7 pg/L, respectively) were detected at the mouth of the bay (around Point Loma), as expected for the metal distribution in San Diego Bay.4 Such values indicate that contaminationfrom the boat or the probe is not contributing to the sensor response. The copper values (0.9-11.0 pg/L) found are in relatively good agreement with those (1.0-7.7 pg/L) of an earlier laboratorybased stripping study.'* An on-boat postcalibration experiment yielded a sensitivity similar to the precalibration one, indicating high stability throughout this in situ operation. Indeed, a series of 60 in situ measurements of copper over a 3-h period yielded a relative standard deviation of 13%. These shipboard calibration procedures used the corresponding sample solutions, stirred slowly (at 300 rpm). Such procedures address the effect of seawater's natural convection. Despite its tiny dimension, the gold fiber displayed some convection effect (with 2.1- and 26fold enhancements of the response at 300 and 600 rpm, respectively,
:[I>',4,, 9" 30
0
U
2
4
6
8 10 I2 14 16 18 20 22 7%
r
2ol,, , , . I 0
0
2
4
6
81012141618
RUNNUMBER Figure 5. Response of the probe to sudden changes in the metal concentration of the groundwater sample: (A) 10 runs at 10 pg/L Cu, followed by 3 runs at 40 pg/L Cu and 10 more runs at 10 pg/L Cu;(6)3 runs at 20pglL Pb, followed by 10 replicates at lOpg/L Pb and 4 more runs at 20 pglL Pb. Conditions: deposition for 1 min at -0.70 V; cleaning for 2 rnin at f0.5V.
Figure 5. The temporal profile, shown in curve A, resembles a sharp contaminant release, with 10 runs in a 10 pg/L copper solution, followed by 3 replicates in a 40 pg/L solution, and 10 measurements at the original 10 pg/L level. The response rises rapidly from the 10 pg/L baseline upon immersion of the sensor in the 40 pg/L solution and decays sharply upon returning the sensor to the original low-concentration medium. Similarly, minimum carry-over effects are indicated from Figure 6B, involving 3 replicates at the 20 pg/L lead level, 10 runs in a 10 pg/L solution, and a return to the original 20 pg/L one. These data 1484 Analytical Chemisity, Vol. 67,No. 8,April 15, 1995
(18) Scarano, G.;Morelli, E.; Sentti, A; Zirino, A. Anal. Chem. 1990,62, 943.
compared to a quiescent sample). Similar effect was observed when a smaller (25 pm) gold fiber was used. While an ordinary reference electrode was used for these surface water analyses, a pressure-resistant one will be required for depth measurements. Specific details of this and other (longer, more systematic, and more extensive) marine surveys and the field calibration procedures will be reported elsewhere. In conclusion, the previous experiments illustrate the utility of stripping-based electrochemical sensors for remote monitoring of trace metals. The new probe offers a unique capability for in situ measurements of priority metal contaminants, including remarkable sensitivity, multielement and speciation capabilities, high selectivity, small size, and low cost. Extension to additional metals of environmental relevance is anticipated on the basis of the judicious choice of the transducer surface or the adaptation of adsorptive stripping procedures. The probe could be further miniaturized to permit adaptation to a cone penetrometer technol-
ogy. More sophisticated probes, employing in situ sample manipulation and calibration,are also expected. While such novel in situ probes are still at an early developmental stage, such an advance should have a substantialimpact on the characterization of contaminated sites or hostile environments, marine surveys, and industrial quality control. ACKNOWLEDGMENT This work was supported by a subcontract from Battelle PNL and a grant from the US.DOE through the Waste Management Education and Research Consortium (WERC). Battelle PNL is operated for the US.DOE by Battelle Memorial Institute under contract DE-AC0676RLO 1830. Received for review August 18, 1994. Accepted January 26, 1995.@ AC940825P @Abstractpublished in Advance ACS Abstracts, March 1, 1995.
Analytical Chemistry, Vol. 67, No. 8, April 15, 1995
1485
