Porphyrin-Modified Electrodes as Biomimetic Sensors for the

D. J. Dobson, and S. Saini*. Institute of .... Cyril Poriel , Yann Ferrand , Paul Le Maux , Christine Paul-Roth , Gérard Simonneaux , Joëlle Rault-B...
0 downloads 0 Views 310KB Size
Download PDF
Anal. Chem. 1997, 69, 3532-3538

Porphyrin-Modified Electrodes as Biomimetic Sensors for the Determination of Organohalide Pollutants in Aqueous Samples D. J. Dobson and S. Saini*

Institute of BioScience and Technology, Cranfield Biotechnology Centre, Cranfield University, Cranfield, Bedfordshire MK43 0AL, UK

A preliminary examination of a simple and rapid screening method for quantifying a range of toxic organohalides directly in aqueous solution based on their electrocatalytic reduction with a metalloporphyrin catalyst is described. Homogenous catalysis is described as well as heterogeneous catalysis using precipitated cobalt(II) tetraphenylporphine ((TPP)Co) at a graphite foil electrode which permitted the sensitive detection of a wide range of different organohalides, including a number of chemically diverse industrial pollutants such as 1,2,3,4,5,6-hexachlorocyclohexane (lindane) and carbon tetrachloride, representative of haloalkane compounds, haloalkenes such as perchloroethylene, and aromatics, such as 2,4-dichlorophenoxyacetic acid, pentachlorophenol, and the insecticide DDT. The coordinating effect of solvent on the thermodynamics of the Co(II)/(I) electrode reaction is used to practical advantage to build an amperometric detector that is insensitive to interference from oxygen, a parameter that varies considerably in environmental samples. Devices also appear relatively insensitive to the ionic composition of the analyte sample. The work provides the basis for developing a prototype sensor for screening toxic organohalogen pollutants for use in environmental monitoring situations. Organohalides, organic substances with one or (often) more halogens substituted into the molecule, are becoming increasingly recognized as important and recalcitrant environmental pollutants arising from various industrial as well as smaller scale sources;1,2 they may also arise from biological sources.3 Thus, halogenated methanes and ethanes are often used as solvents, while polyhalogenated aromatics number among the various biocides in agricultural or domestic use.4 Accidental or deliberate release of these materials into soil, water courses, and the like can exert long-term toxic effects, and the quantification of the problem, together with the needs of legislation, requires means of measuring their presence.5 (1) NSCA. Pollution Handbook; National Society for Clean Air and Environmental Protection; Brighton, UK, 1994. (2) Nicholson, S.; Blaine, L. M. A review of the properties of the red list substances in trade effluent. Department of the environment report, 1993. (3) Nightingale, P. D.; Malin, G.; Liss, P. S. Limnol. Oceanogr. 1995, 40, 680689. (4) Kroswich, J. I.; Howe-Grant, M. Kirk-Othmer Encyclopaedia of Chemical Technology, 4th ed.; Wiley & Sons; New York, 1994; Vol. 5, 1071. (5) Greenberg, A. E.; Clescerci, L. S.; Eaton, A. D. Standard Methods for the Examination of Water and Wastewater, 18th ed.; American Public Health Association: Washington DC, 1992; Chapter 5.

3532 Analytical Chemistry, Vol. 69, No. 17, September 1, 1997

Presently, analysis of organohalide contamination often relies on the chromatographic separation of individual compounds, followed by detection or analysis; an extraction protocol may also be needed.5-8 These are in essence laboratory-based techniques that not only imply centralized facilities and trained personnel but can also require extensive sample taking and introduce a time delay into the result, particularly if there is relatively little foreknowledge of the nature or spread of contamination. Moreover, such techniques are often confined to the analysis of individual or small groups of compounds; measurement of generic organohalide contamination still relies on wet chemical testing.5 A simple, general-purpose organohalide sensor, that could directly detect the carbon-halogen bond in a variety of sample matrixes without prior purification, could augment analysis considerably. Not only could it function as an in situ safety monitor but could also provide forewarning of incipient contamination and at-site environmental testing (assessing contaminant spread, for example). In this respect, an electrochemical sensor would appear ideal. A number of different organohalides may be detected in solution by their electrochemical reduction; this is, however, kinetically disfavored and involves, at least for noble metal electrodes, the use of extreme negative overpotentials, typically -1.3 to -1.6V (vs SCE).9-11 Such high potentials run the risk of considerable interference, not least from oxygen, thus limiting the usefulness of the technique for field sensing. It has been known for a number of years, however, that certain metal complexes, notably iron,12,13 nickel,9,14-16 and particularly cobalt14,17-25 porphyrins, corrins, and phthalocyanines,26,27 as well (6) Lepine, L.; Archambault, J. F. Anal. Chem. 1992, 64, 810-814. (7) Huhnerfluss, H.; Kallenborn, R. J. Chromatogr. Biomed. Appl. 1992, 580, 191-214. (8) Simmonds, P. G.; O’Doherty, S.; Nickless, G.; Sturrock, G. A.; Swaby, R.; Knight, P.; Ricketts, J.; Woffendin, G.; Smith, R. Anal. Chem. 1995, 67, 717-723. (9) Arai, T.; Kondo, H.; Sakaki, S. J. Chem. Soc., Dalton Trans. 1992, 18, 27532754. (10) Dahm, C. E.; Peters, D. G. Anal. Chem. 1994, 66, 3117-3123. (11) Tokoro, R.; Bilewicz, R.; Osteryoung, J. Anal. Chem. 1986, 58, 1964-1969. (12) Elliott, C. M.; Marrese, C. A. J Electroanal. Chem. 1981, 119, 395-401. (13) Rusling, J. F.; Nassar, A. E. F. J. Am. Chem. Soc. 1993, 115, 11891-11897. (14) Lewis, T. A.; Morra, M. J.; Habdas, J.; Czuchajowski, L.; Brown, P. D. J. Environ. Qual. 1995, 24, 56-61. (15) Kadish, K. M.; Franzen, M. M.; Han, B. C.; Araullo-McAdams, C.; Saxon, D. Inorg. Chem. 1992, 31, 4399-4403. (16) Lahiri, G. K.; Schussel, L. J.; Stolzenberg, A. N. Inorg. Chem. 1992, 31, 4991-5000. (17) Asaf-Anid, N.; Hayes, K. F. Vogel, T. M. Environ. Sci. Technol. 1994, 26, 246-252. (18) Che, G.; Dong, S. Electrochim. Acta 1993, 38, 1345-1349. (19) Fukuzumi, S.; Maruta, J. Inorg. Chim. Acta 1994, 226, 145-150. (20) Lin, X.-Q.; Liu.; D.-J.; Wang, E.-K. Sci. China, Ser. B 1994, 37, 257-264. S0003-2700(97)00353-3 CCC: $14.00

© 1997 American Chemical Society

as some other metals28 and ligands,10 can catalyze, via reduction of the central metal, the reduction of organohalides at much more moderate potentials. The overall reaction is

larly toward aqueous environments. Three simple, commercially available compounds, cobalt(II) tetraphenylporphine ((TPP)Co), the tetrakis(p-methoxyphenyl)porphyrin ((TMOP)Co), and cobalt octaethylporphyrin ((OEP)Co) were chosen for this work.

RX + H+ + 2e- f RH + X-

The source of electrons for these reactions may be either chemical9,14,17,19 or electrochemical; in the latter case, measurement of electron flow can, in principle, be used to furnish a sensor. There is some precedent for this: cyclic voltammetric studies at iron porphyrin-modified graphite electrodes showed the possibility of detection of some of the more reactive organohalides as early as 1981,12 while more recent studies have shown the potential for monitoring ethylene dibromide in petrol using cobalamins in microemulsions21 and the fabrication of detectors to halogenated pesticides following HPLC or other procedures.29-31 As of yet, however, no single system has been used as a general-purpose organohalide sensor. Originally, study of organometallic catalysts was motivated to a large extent by their use as model compounds32,33 with a view to understanding the mode of action of dehalogenase enzymes possessed by various microorganisms34,35 (hence the term “biomimetic”) as well as the possible mechanisms of organohalide toxicity in higher organisms. Latterly, though, it has become clear that they have potentiality in their own right as industrial catalysts, in particular for the destruction of organohalide wastes in situations where decontamination by microorganisms is less than ideal;14,24 emphasis has thus shifted toward the use of chemically robust materials on a larger scale. Factors affecting the choice of catalyst for a sensor are the potential at which reduction takes place and the ease of reaction (reaction rate and intermediate stability) with the organohalide; both are highly dependent upon the nature of metal and ligand. By far the best characterized ligands are porphyrins, together with their ring-reduced derivatives, corrins, a consequence of their appearance in natural systems. Iron and nickel porphyrins are catalytically active with the metal in its (I) oxidation state, but electrochemical generation of this usually requires negative potentials of -1 V or more. This can be modified by chemical alteration of the ring (“redox tuning”), but more positive redox potentials imply lessened reaction rates, albeit with some exceptions.15 By contrast, cobalt porphyrins offer relatively easy access to the (I) oxidation state, coupled with stability of the organometallic intermediate, particu(21) Rusling, J. F; Connors, T. F.; Owlia, A. Anal. Chem. 1987, 59, 2123-2127. (22) Rusling, J. F.; Miaw, C. L.; Couture, C. L Inorg. Chem. 1990, 29, 20252027. (23) Steiger, B.; Ruhe, A.; Walder, L. Anal. Chem. 1990, 62, 759-766. (24) Ukrainczyk, L.; Chibwe, M.; Pinnavaia, T. J.; Boyd, S. A. Environ. Sci. Technol. 1995, 29, 439-445. (25) Zhou, D.-L.; Walder, P.; Scheffold, R.; Walder, L. Helv. Chim. Acta 1992, 75, 995-1011. (26) Iwunze, M. O.; Hu, M.; Rusling, J. F. J. Electroanal. Chem. 1992, 333, 353361. (27) Zhang, H. P.; Rusling, J. F. Talanta 1993, 40, 741-747. (28) Anderson, J. E.; Yao, C.-L.; Kadish, K. M. Inorg. Chem. 1986, 25, 718719. (29) Root, D. P.; Pitz, G.; Priyantha, N. Electochim. Acta 1991, 36, 855-858. (30) Priyantha, N.; Tambalo, M. ACS Symp. Ser. 1992, No. 511, 41-47. (31) Priyantha, Y.; Weerabahau, D. Anal. Chim. Acta 1996, 320, 263-268. (32) Karlin, K. D. Science 1993, 261, 701-708. (33) Lexa, D.; Saveant, J.-M.; Wang, D. L. Organometallics 1986, 5, 1428-1434. (34) Fetzner, S.; Lingens, F. Microbiol. Rev. 1994, 58, 641-685. (35) Khindaria, A.; Grover, T. A.; Aust, S. D. Environ. Sci. Technol. 1995, 29, 719-725.

EXPERIMENTAL SECTION Materials. Acetonitrile (HPLC grade), and dimethylformamide (DMF, HPLC grade) were stored over 3A molecular sieves and tetraphenylporphyrinatocobalt(II), tetrakis(4-methoxyphenyl)porphyrinatocobalt(II), and octaethylporphyrinatocobalt(II) were used as received from Aldrich. We attempted to select organohalides for this study that were widely representative of the class as a whole, as well as being, in many cases, compounds of environmental concern in their own right. Among alkyl halides were HPLC grade carbon tetrachloride (CT), 1,2-dibromoethane (ethylene dibromide, or EDB, 99%), and 1,2,3,4,5,6-hexachlorocyclohexane, γ isomer, (BHC, lindane, 97%), while haloalkenes were represented by HPLC grade perchloroethylene (PCE), hexachloropropene (HCP, 97%), and hexachlorocyclopentadiene (HCC, 98%). In order of increasing halogen content, aryl halides were chlorobenzene (99%) bromobenzene (99+%), iodobenzene (PhI, 98%), 2,4-dichlorophenoxyacetic acid (DCPA, 98%), 2,4,5trichlorophenoxyacetic acid (TCPA, 97%), and 1,2,4,5-tetrachlorobenzene (TCB, 98%), all of which were obtained from Aldrich, as was 1,1-bis(4-chlorophenyl)-2,2,2-trichloroethane (DDT, 99%), which contains both alkyl and aryl carbon halogen bonds. On chemical grounds the alkyl chlorines may be expected to be more labile however; they are the first to be lost when DDT is degraded in nature.2,36 Aryl halide pentachlorophenol (PCP, 99%) was obtained from Eastman/Kodak, and the electrolyte tetrabutylammonium perchlorate (TBAP, electrochemical grade) from Fluka. Methods. Electrochemical measurements were made in a two-compartment cell (5 mL, with the capacity for the introduction of gas, oxygen, or nitrogen, into the solution). For homogenous solution measurements, a 1 mm diameter platinum disk working electrode (Bioanalytical Systems Ltd., West Lafayette, IN) surrounded by a coil of platinum wire, 6.5 mm in diameter, acting as a counter electrode was used. The other compartment, connected via a Luggin probe, contained a saturated calomel reference (SCE) electrode (Russel) against which all potentials (homogeneous and heterogeneous) are quoted. For aqueous solution measurements, a mixture of acetonitrile and 0.1 M KCl in water (1:4 by volume) was used. This maintains a predominantly aqueous environment, while allowing greater solubility of organic substrates (up to approximately 2-3 mM, depending on the compound concerned), as well as more reliable dissolution, than water alone. For aerated solutions, essentially identical results could be obtained by working in an open 10 mL beaker; the solution was saturated with oxygen and then left open to the atmosphere. This volume and ionic concentration were adopted as standard for all aqueous measurements. Substrates to be introduced were prepared as 1 M solutions in either DMF (for addition to homogeneous porphyrin solutions) or acetonitrile (for aqueous solution), and aliquots of these solutions were added to stirred solutions to achieve a desired total final concentration. Exceptions were TCB and TCPA, which were prepared as 100 mM solutions in acetonitrile, as a result of their lower solubility therein. The extra volume of solvent added by (36) Zoro, J. A.; Hunter, J. M.; Eglington, G.; Ware, G. C. Nature 1974, 247, 235-236.

Analytical Chemistry, Vol. 69, No. 17, September 1, 1997

3533

amperometric responses over a similar time scale. This is, however, distinct from the high concentration peaking noted in the results section. Electrochemical measurements were made with a PSTAT 10 Autolab electrochemical analyzer (Ecochemie), which was interfaced to a PC and operated within the General Purpose Electrochemical Software program (Ecochemie). Modified graphite electrodes were constructed from 4 mm diameter disks of graphite foil sealed at the end of glass tubes of the same diameter by epoxy cement and attached via silver-loaded epoxy cement to a connecting wire. This arrangement gave a geometric area of graphite of approximately 12.5 mm2 facing the solution. A platinum (4N, Johnson Matthey) or silver (Aldrich, 99.99%) wire loop wound around the tube end near this provided the counter electrode. Electrodes were alumina (0.3 µm) polished, ultrasonicated in distilled water (60 s), and air-dried; 20 µL of a solution (2 mM) of porphyrin in DMF was deposited on the surface of this and allowed to dry. Drying took about 1 h, during which time some soaking of the solution into the graphite foil could be seen. The amount used appeared to correspond to an empirical optimum, as a second coating did not yield substantially improved responses, while a decrease in response was seen if less than 10 µL was used.

Figure 1. Cyclic voltammograms of (TPP)Co in DMF (1 mM) alone (1 A) and showing the effect of adding 1 mM CT (B) or DCPA (C); scan rate 50 mV/s (B, 5 mV/s). For both organohalides, catalytic current is associated with the Co(II/I) couple. Supporting electrolyte is TBAP (0.1 M).

this procedure was low enough (up to 100 µL) to be ignored; control additions of solvent without substrate (200 µL) did not appear to elicit significant changes in voltammograms or amperometric responses. The only disadvantage is that homogeneity takes 10-20 s to establish, as visualized by introduction of a dye (Meldola’s blue), and thus coated electrodes see a “waft” of concentrated solution that can lead to an initial peaking of 3534

Analytical Chemistry, Vol. 69, No. 17, September 1, 1997

RESULTS AND DISCUSSION Cyclic Voltammetry in Homogeneous Solution and Reaction Mechanisms. Cyclic voltammograms of (TPP)Co in DMF solution are depicted in Figure 1A. Two couples may be seen: the (TPP)Co(III)+/(TPP)Co(II) couple at +500 mV, and the reduction from (TPP)Co(II) to (TPP)Co(I)- at -850 mV, in agreement with literature values.37 Addition of increasing concentrations of organohalide substrate, such as CT (Figure 1B) or DCPA (Figure 1C), to this solution caused an increase in current starting at the Co(II)/(I) couple, indicative of catalytic reduction. Similar results were seen with the other organohalides used, and in all cases, a catalytic current in excess of the diffusion current was seen in accord with catalysis at or near the Co(II)/(I) reduction potential. Similar catalytic efficiencies were seen with the other two porphyrins used, (TMOP)Co and (OEP)Co, though in these cases the critical Co(II)/(I) couple occurred at more negative potentials, namely, -900 and -1000 mV respectively. (TPP)Co was therefore chosen for further development. Organohalide reduction at these potentials is in line with previously proposed reaction mechanisms, though, despite much study, a comprehensive picture has yet to emerge regarding the precise mechanism of biomimetic organohalide reductive catalysis: side reactions occur, and different studies have suggested conflicting scenarios. Nonetheless, there appears sufficient similarity in the proposed mechanisms of many catalysts that the salient features of a general mechanism can be identified, operating in three principal stages. (1) The metal complex undergoes one (or more38 ) electron reduction:

(L)Mn + e- f (L)Mn-1 (2) The reduced form interacts with the organohalide to yield an organometallic intermediate: (37) Kadish, K. M.; Lin, X. Q.; Han, B. C. Inorg. Chem. 1987, 26, 4161-4167. (38) Schanke, C. A.; Wackett, L. P. . Environ. Sci. Technol. 1992, 26, 830-833.

Figure 2. (TPP)Co-modified graphite electrodes in aqueous acetonitrile. (A) Cyclic voltammogram (scan rate 50 mV/s), showing prominent couple identified with Co(II/I) transition; vestigial activity at +4-500 mV may be the (III/II) couple. (B) Square wave voltammogram (amplitude, 0.1 V; step, 1.22 mV; f ) 50Hz; direction, positive to negative) comparing traces of the electrode alone (Nil) and with 20 µM of CT or 10 µM DDT added to the solution. In both cases, the most marked change is the appearance of a new reduction peak centered around -400 mV (arrow). The effect of adding PCE (20 µM; C) appears more complex, but a secondary peak, now centered at -420 mV, is again visible.

(L)Mn-1 + RX f (L)MR + X(3) This is then reduced to reform the catalyst:38,39

(L)MR + e- f (L)Mn + RM being the active metal, L its ligand, and RX a generic organohalide. The species R- then undergoes protonation to form RH, elimination, or reaction with other components of the solution. The overall effect is two-electron reduction of the organohalide according to the equation in the introduction. Shown is a generic mechanism; in the case of this work, M ) Co, L is the porphyrin, and n ) 2. The ease of fission of the carbon-halogen bond in the second step, and thus the reactivity and ease of detection of the organohalide, depends on its chemical nature, decreasing in the order haloalkane > haloalkene > haloarene. Haloarenes in particular are extremely resistant to substitution, though catalytic dehalogenation via one-electron radical pathways (ArX f Ar• f Ar2) is still possible in principle, and iodobenzene has been observed to react with cobalt(I) porphyrins to give the cobalt(II) form.19 Moreover, evidence of radical pathways in porphine- and corrin-mediated dehalogenations has been found by other workers.25 In practice, however, most haloarene pollutants are polyhalogenated, and polyhalogenated arenes are generally more reactive toward substitution by a number of pathways,40 in addition to enhanced activity on statistical grounds. Cyclic Voltammetry in Heterogeneous Systems: Modified Electrodes. Cyclic voltammograms under similar conditions of (TPP)Co-coated graphite electrodes in aqueous acetonitrile showed only one prominent couple (Figure 2A), centered at around -30 ( 10mV. Peak separation was considerably wider (264 mV in this instance), suggesting slower electron transfer compared with the homogeneous case; the integrated charge under the reduction peak was 4.13 × 10-4 C which, for a one-electron-transfer process, corresponds to 10.7% of the porphyrin molecules present being electroactive. This is not unexpected, as electrodes prepared by simple evaporation probably contain many molecules of porphyrin inside solid crystals or absorbed into the graphite where electron (39) Kadish, K. M.; Han, B. C.; End, A. Inorg. Chem. 1991, 30, 4502-4506. (40) March, J. Advanced Organic Chemistry: reactions, mechanism and structure, 4th ed.; Wiley & Sons: New York, 1992; Chapter 13.

transfer to the solution is impaired. The exact redox potential is somewhat sensitive to the nature of the solvent and in 50% ethanol was -84 ( 4mV; also more of the porphyrin was active, up to 46%. However, the porphyrin coating was unstable in this solvent, so further studies used aqueous acetonitrile only. The couple in Figure 2A is assigned to the Co(II)/(I) transition; other workers have reported that this couple is shifted considerably positive on moving from organic to aqueous solution, an effect ascribed to the coordinating nature of the solvent.29 The influence of coordinating solvents on the thermodynamic parameter, E°, presents two advantages for this work. First, in principle, cosolvents can be selected in order to avoid interference from competing electroactive species, and second, the use of predominantly aqueous solvent systems permits a significantly lower detection potential to be used with respect to oxygen reduction. This is especially relevant to environmental applications since the concentration of oxygen in samples can vary widely according to system and the severity of contamination. Square Wave Voltammetry of Modified Electrodes. The electrochemical activity of the coated electrodes is far more readily seen by square wave voltammetry (Figure 2B,C). Voltammograms of these electrodes in acetonitrile/water show two couples, a broad one centered at +500 mV, and assigned here to the Co(III)/(II) transition, and a sharper one at -100 mV, corresponding to the Co(II)/(I) reduction. The addition of 20 µM carbon tetrachloride (Figure 2B) to the surrounding solution caused a number of changes in the appearance of the voltammogram, notably the appearance of an extra broad peak (arrowed) centered around -400 mV. A very similar peak was seen with the addition of 10 µM DDT (lower trace). These peaks, which are not seen in the absence of porphyrin or substrate, or on scanning from negative to positive, may correspond to the reduction of a bound organohalide intermediate. The corresponding picture with haloalkene PCE (Figure 2C) is slightly more complex, but a secondary peak still occurred, this time centered around -420 mV. Analogous peaks were not seen with either of the haloarenes; this may reflect a reduced rate of formation of the postulated intermediate. If the observed peak corresponds to a second reduction then, in principle, it should be possible at potentials of -500 mV or below to achieve catalytic reduction and build a simple amperometric organohalide sensor. This potential, while considerably negative of the Co(II)/(I) potential, is still small Analytical Chemistry, Vol. 69, No. 17, September 1, 1997

3535

Table 1. Comparative Responses of (TPP)Co/Graphite Electrodes to the Three Classes of Organohalide, in Terms of Lower Detection Limits and Initial (Linear) Sensitivitiesa substrate haloalkanes CT EDB BHC DDT haloalkenes PCE HCP HCC haloarenes PhI DCPA TCPA TCB PCP Figure 3. Typical amperometric trace of (TPP)Co/graphite electrode in aqueous acetonitrile (potential, -500 mV), showing the effect of adding three successive aliquots of CT (50 µM concentration steps, arrows) to the solution; the slower response at 300 s is caused by the addition of 20 µM CT. Trace noise is a result of air sparging: essentially identical results are obtained if nitrogen is substituted.

enough to escape interference by oxygen at carbon electrodes and represents a practical potential for use in an environmental monitoring sensor as alluded to earlier. Experiment confirmed an independence from oxygen interference (nitrogen purging over various periods) at this detection potential using amperometry in aqueous solution. Amperometric Studies. A typical amperometric trace is seen in Figure 3; it shows a clear response to three successive additions (arrows) of 50 µM CT. Response vs substrate concentration for different organohalides gave the response curves depicted in Figures 4 and 5. Responses shown are an average of five runs; bars shown represent absolute scatter of results. The lower detection limit was around 5 µM, which in the case of CT corresponds to 0.75 ppm by weight (Table 1). While the electrode responded to all the organohalides presented to it, the sensitivity of the system was noticeably dependent on the chemical character of the C-X bond. Among the highest responses was that given by the polyhaloalkane CT (Figures 4A and 5A), and this has proved a favorite substrate for other authors. Here the relative stability of the CCl3- ion probably assists its elimination in the third step of the catalytic scheme outlined above, and thus catalytic turnover as a whole.17,25 Another factor that may augment the apparent reactivity of CT is its relatively high reduction potential (-780mV vs SCE at a mercury electrode), so the decrease in overpotential for this organohalide is not as dramatic as some of the others. Uniquely among compounds tested, CT gave fugitive responses at a bare graphite electrode (at -500 mV). These were considerably smaller (200 µM) concentrations of some substrates, particularly CT, was a spiked response, followed by a fall in signal. It should be noted that the solution volume was maintained around the electrode, and the current levels were nowhere near enough to decompose an appreciable fraction of the analyte over the observed time scale, so some other explana(42) Coetzee, J. F.; Gardner, C. W. Anal. Chem. 1986, 58, 608-611. (43) Midgley, D. Ion-Sel. Electrode Rev. 1986, 8, 3-54. (44) Glazier, S. A.;Arnold, M. A. Abstracts Of Papers 193rd National Meeting of the American Chemical Society, Denver, CO, Sept 7-12, 1987; American Chemical Society, Washington, DC, 1987; p 119. (45) Alexander, P. W.; Koopetngarm, S. A. Anal. Chim. Acta 1987, 197, 353359. (46) Davey, D. E.; Mulcahy, D. E.; O’ Connell, G. R. Talanta 1990, 37, 683687 (47) Glazier, S. A.; Arnold, M. A.; Demeulenaere. Abstracts Of Papers 201st National Meeting of the American Chemical Society, American Chemical Society: Washington, DC, 1991; p 117. (48) Hara, H. ; Kusu, S. Anal. Chim. Acta 1992, 261, 411-417.

Analytical Chemistry, Vol. 69, No. 17, September 1, 1997

3537

tion must be sought. Inhibition of the catalyst is one possibility but is argued against by the observation that the addition of a second aliquot of substrate produced a response of the same order of magnitude. Physical explanations are also possible. Small droplets of substrate may adhere briefly to the electrode surface, giving a temporary local enhancement of concentration, or volatile substrates may be subsequently partially removed by gas sparging, though the latter is argued against by the observation that spiking may also be seen in experiments carried out without sparging. As stated in the introduction, the exact mechanism by which catalysis occurs is complex, and the possibility of time-dependent catalytic phenomena cannot be ruled out; studies are continuing into this. With low (0.5-20 µM) levels of some substrates (notably CT, DDT, and TCB, among the least polar compounds tested), electrode response was markedly slower, taking some minutes to grow, as opposed to the short (seconds) time scale of the response to submillimolar substrate levels. With time, the response grew to a level more typical of substrate levels 5-10 times as high and may be responsible for the relative sensitivity toward DDT. This effect could be seen most clearly on a square wave voltammogram (Figure 6). Here, a fresh (TPP)Co electrode was placed into acetonitrile/water, and 10 µM CT added; successive voltammograms taken over time showed a steady increase in response (descending traces). A possible reason for this was that the graphite material of the electrode gradually acted to absorb the relatively nonpolar substrate, providing a preconcentration effect. Alternatively, a slow reaction may have been taking place between the organohalide and electrogenerated Co(I) on the graphite surface, forming a pool of organocobalt species which was detected in following scans; a similar form of signal enhancement has been noted by other workers.23 CONCLUSIONS It has been shown possible, even with an elementary (and potentially disposable) electrode construction, to electrochemically detect a range of hazardous organohalides in predominantly aqueous solution, without the need for deoxygenation. Furthermore, devices appear relatively insensitive to other factors such as the ionic composition of the solution, though some interference

3538 Analytical Chemistry, Vol. 69, No. 17, September 1, 1997

Figure 6. Increase with time of the -400 mV peak in the square wave voltammogram (same conditions as Figure 2B) of (TPP)Comodified electrode in solution containing 10 µM CT. Traces are shown immediately (0 min) and 5, 10, and 30 min after addition of the organohalide.

is seen with phosphate. Sensitivities are presently in the partper-million, or just sub-part-per-million range, and while this does not yet match the part-per-billion levels of international environmental control standards, there are some applications, such as initial mapping of a spill site, where lesser sensitivity may suffice. Moreover, electrodes are still in a preliminary state of development, and there seems considerable scope for sensitivity improvement; deliberate augmentation of the analyte preconcentration effect noted above may provide a first step toward this. ACKNOWLEDGMENT D.J.D. is a recipient of a BBSRC-CASE award. The authors thank Hewlett-Packard (Europe) for an equipment grant. Received for review April 2, 1997. Accepted June 18, 1997.X AC970353P X

Abstract published in Advance ACS Abstracts, August 1, 1997.