ANALYST, NOVEMBER 1991, VOL. 116 1125 Examination of Ammonia-Poly( pyrrole) Interactions by Piezoelectric and Conductivity Measurements Jonathan M. Slater and Esther J. Watt Analytical Science Group, Birkbeck College, University of London, 20 Gordon Street, London WCI H OAJ, UK The conducting polymer poly(pyrrole), electrochemically prepared and doped with anions, has been found to be a responsive coating for a piezoelectric gas detection system. Polymers doped with bromide, nitrate and sulphate ions were tested. It was found that samples of ammonia gas cause a measurable frequency decrease, interpreted as adsorption by the polymer coating of the quartz crystal; the linear range was 0.051% for mixtures of the gas in nitrogen. These signals were found to correspond to simultaneous conductivity changes of a similarly prepared poly(pyrro1e) sample, showing analogies in the two sensing mechanisms.The duality of the poly(pyrro1e) response increases the possibilities of using it as a gas sensor. Keywords: Ammonia detector; piezoelectric gas detection system; gas sensor; poly(pyrrole) gas sensor Poly(pyrro1e) is a conducting polymer with conductivity ranging from 1 to 100 S cm-1 (Fig. 1).1 It may be conveniently prepared by the electrochemical oxidation of the pyrrole monomer in an electrolytic solution, onto gold or platinum electrodes .24 The reaction is initiated by the electrochemical generation of monomer radicals which combine with other units in solution to form the polymer chains, along which electrons may conduct. Excess oxidation of the polymer generates a net positive charge resulting in the uptake of counter ions from the electrolyte solution.These dopant anions render the polymer a p-type semiconductor, giving it a second mode of conduction."." However, its true structure, which is dependent on conditions such as pH, potential, solvent and dopant anion, has still not been fully charac- terized. Despite this apparent lack of characterization poly- (pyrrole) is a potentially useful material for the fabrication of sensors. It has been used as a sensing material in field effect transistor7 and ion-selective electrode8 devices and as a conducting matrix for enzyme entrapment electrodes.9 It shows interesting gas sensing possibilities which were first demonstrated by Nylander et al.10 in 1983. The polymer was prepared by chemical oxidation, the precipitated 'pyrrole black' being impregnated into filter-paper which was then shown to give a 30% change in conductivity on exposure to 1% More recently Miasik et af. I have reported a device utilizing electrochemically prepared polymer promising convenient and controllable preparation which should lead to a more stable and potentially reproducible sensor material. The mechanism of interactions was attributed to the p-type semiconducting nature of poly(pyrro1e). Exposure to elec- trophilic gases, such as NO,, tends to attract electrons out of the polymer matrix, causing an increase in conductivity, whereas nucleophilic gases, such as NH3, will have the opposite effect. This clear evidence of polymer-gas interac- tions renders poly(pyrro1e) a suitable coating for piezoelectric sensor crystals which should be a useful means of investigating such interactions.Furthermore piezoelectric crystals already NH3. H Fig. 1 NO3-, SO4'- or Br- Structure of doped poly(pyrrole), X- = doping anion, e.g., contain electrodes suitable for the electrochemical deposition of poly(pyrro1e) layers. The piezoelectric gas detection systems described to date work on the principle that a gas adsorbed onto a crystal coating changes the crystal mass resulting in a shift in its fundamental frequency. 11-12 The Sauerbrey relationship is commonly used to relate the observed frequency change to the adsorbed mass: 13 AF = -2.3 x 106 Fo2 AmlA where A m is the change in mass of the crystal (g), AF the related frequency change (Hz), A the gas-sensitive area (cm2) and Fo the initial frequency of the quartz crystal (MHz).It is well-known that moisture is a major interferent in piezoelectric measurement systems and other workers have reported methods of overcoming this problem.14 Poly- (pyrrole) readily adsorbs moisture; in fact poly(pyrro1e) tosylate is reported to be hygroscopic.15 However, this work was carried out using cylinder gases which are assumed to be of constant moisture. A possible application of poly(pyrro1e) gas sensors is in arrays and in this instance the moisture response would be treated as a mixed sensor response. Experimental Apparatus The piezoelectric crystals used were 5 MHz, AT-cut quartz crystals with gold electrodes (Webster Electronics, Ilminster, Somerset, UK).The measurement electronics were con- structed by British Gas. The sensor and reference crystals were built into matched oscillator circuits, driven by a 5 V d.c. supply, and a signal which related to the difference in frequency between the two crystals was extracted. This output could be displayed on a y-t recorder (Siemens Kompenso- graph X-T C1011) or transferred to an IBM PC via a suitable interface (Blue Chip Technology AIP-24, 12 bit analogue input card). Data logging software was written to provide a permanent record of results. A further output allowed the frequency of the sensor crystal to be monitored directly (Thandar TF-600 frequency counter). Measurements were made in two flow systems. The first, illustrated in Fig.2, is similar to those previously described.16 A continuous flow of air is pumped through activated charcoal and silica gel drying agent. The scrubbed, dry air is passed through an injection cell where the sample may be introduced. The gas system was split into two, one stream passed through a reference cell, containing an uncoated crystal, the other passed through an identical cell containing the poly(pyrro1e)1126 r Chart Double oscillator recorder Digital computer meter frequency - crystal or ANALYST, NOVEMBER 1991, VOL. 116 5 V d.c. - supply coated crystal. These cells were based on the double impinger cell design of Karmarkar and Guilbault17 wherein the gas is split into two streams impinging directly onto the two faces of the crystal. Exhaust gases were monitored by flow meters, a constant flow optimized at 40 ml min-1 being maintained.The second flow system utilized a British Gas multi-gas sensor test rig which allowed both crystals to be exposed to alternate 5min pulses of carrier gas (air mixture; Air Products) and sample gas. Similar flow conditions were used with the same double impinger flow cell. The test rig also has a facility for determining the moisture content (relative humidity) of test and carrier gases. Crystal Coating Crystal coatings were prepared by the electrochemical oxida- tion of the pyrrole monomer using a Princeton Applied Research 174A polarographic analyser in a three-cell poten- tiostatic assembly. The three-electrode cell consisted of a saturated calomel reference electrode, a glassy carbon auxiliary electrode and the gold electrodes of the piezoelectric crystal as the working electrode.The electrolysis solution was a combination of pyrrole (freshly distilled, Aldrich) and supporting electrolyte, either KBr, K2S04 (both Fisons), KCI or KN03 [both Merck (formerly BDH)]. Different polymer coatings were grown onto the gold electrodes of the crystals by sweeping aqueous solutions of the pyrrole and electrolyte between 0.0 and 0.9V. Table 1 shows the preparation conditions of different types and thicknesses of the poly- (pyrrole) coatings prepared. Poly(pyrro1e) coated crystal Uncoated crystal Perspex Perspex sample crystal reference crystal cell cell (through rubber septum) Fig. 2 Piezoelectric gas detector flow apparatus Conductivity Measurements A Degussa Dew Point Sensor E was used as the basis of a conductivity sensor.The device (4 x 6mm) consists of alumina, coated with a platinum film, into which a number of meanders have been cut to form finely spaced multi-interdigi- tated electrodes. The shape and size of the meanders are shown in Fig. 3, the total number being 24, each of dimensions 100 x 700 pm with an electrode gap of 10 pm. Poly(pyrro1e) may be deposited on the platinum surface by anodic oxidation of pyrrole producing a coating sufficiently thick to bridge the gap between the device electrodes (sensor I, Table 1). The conductivity of the polymer can therefore be measured at these points. A simple Wheatstone bridge was constructed so that changes in the polymer conductivity caused by different gases could be measured on a Thurlby 1503-HA digital multimeter and recorded on a chart recorder.Gas Samples The test gas samples were obtained from cylinders of 1% sample gas, in a nitrogen balance (Air Products). Gas samples were taken with a 10 ml glass syringe. Dilution of the gas was carried out by syringe dilution, a procedure previously described by Karmakar and Guilbault .17 Results and Discussion Poly(pyrro1e) Preparation A typical cyclic voltammogram obtained during poly(pyrro1e) preparation is shown in Fig. 4. It has similar characteristics to those previously reported for the electrochemical oxidation of pyrrole while cycling between 0.0 and 0.9 V.18 The polymer 700 pm t I I100 prn 7 Fig. 3 Geometry of the conductivity sensor (total number of meanders is 24) Table 1 Preparation conditions of poly(pyrro1e) coated onto piezoelectric crystals and the conductivity sensor Change in crystal parameters due to poly(pyrro1e) coating Sample A B C D E F G H Ill Electrolyte solution" K2S04 (0.5 rnol dm-3) K2S04 (0.5 mol dm-3) K2SO4 (0.5 mol dm-3) KBr (1 .O mol dm-3) KBr (1 .O mol dm-3) KBr (1.0 mol dm-3) K2SO4 (0.5 mol dm-3) KN03 (1 .O mol dm-3) KBr (1 .O mol dm-3) Preparation timetls 54 54 54 180 180 36 36 99 1800 Drying conditions 110 "C, 10 min 110 "C, 10 min llO"C, 10 min Ambient Ramp§ 110 "C, 10 min 110 "C, 10 min 110 "C, 10 min 110 "C, 10 min FrequencyIHz 4 322 3 967 2 754 29 861 22 347 4 396 17 782 11 742 MassSIyg 85.3 78.3 54.2 589.0 441 .o 86.7 350.0 232.0 * Solutions contain electrolyte and freshly distilled pyrrole (0.05 mol dm-3, aqueous).t Prepared by cyclic voltammetry, sweeping between 0.0 and 0.9 V, scan rate = 100 mV s-*. $ Mass calculated from the Sauerbrey equation. 5 Ramp-dried sample was heated in the oven from room temperature to 110 "C over a period of 10 min. '1[ Conductivity sensor.ANALYST, NOVEMBER 1991, VOL. 116 5 , I 1127 B CI 2 3 a 0 e a o 0.3 0.6 EN versusSCE 0.9 Fig. 4 Preparation of sensor G by the cyclic voltammetry of pyrrole (0.05 mol dm-3) in aqueous K2S04 (0.5 mol dm-3), by scanning between 0.0 and 0.9 V, scan rate = 100 mV s-*. A, First cycle; B, second cycle; and C, third and subsequent cycles 0 0 50 100 1 50 Reaction time/s Fig. 5 Preparation of poly(pyrro1e) onto piezoelectric crystals, mass of polymer coating versus reaction time, scan rate = 100mVs-1.Doping anion: 0, bromide; 'I, sulphate; and M, nitrate preparation is a three-dimensional nucleation and growth reaction; thus, it would be expected that growth is favoured on the polymer surface rather than on the bare electrode. This type of polymerization is characterized by voltammograms of increasing current which eventually reaches a steady-state limit.3 Fig. 4 exhibits the first stages of this polymerization process; further cycling, in other experiments, has shown small increases in current eventually reaching a maximum value. The growth profiles were similar in shape for the different electrolytes used. In all instances the cycling process was terminated at the more positive potential in order to ensure that the polymer was in the oxidized form.The mass of deposited poly(pyrro1e) increases with the reaction time (see Table l), Fig. 5. Samples A, B and C (sulphate-doped) were prepared under similar conditions (54 s polymerization time, equivalent to three cycles between 0.0 and 0.9V followed by a final, half-scan ending at 0.9V, 100 mV s-1 followed by drying for 10 min at 110 "C), and therefore have a comparable mass of polymer deposited, 85.3 x 10-6, 78.3 x 10-6 and 54.2 x 10-6g. Duplicate bromide- doped samples, D and E, were also prepared (180 s poly- merization time, equivalent to ten cycles between 0.0 and 0.9 V, 100 mV s-I), but while sample D was left to dry in the atmosphere sample E was ramp-dried in an oven from ambient temperature to 110 "C. The two samples have different total masses, the variation in which is attributed to the drying conditions.Thus sample E (oven dried) weighed 441 x 10-6g and sample D (air dried) weighed 589 X 10-6g, the additional mass probably being due to excess moisture trapped in the polymer structure. Further dryinglheating did not significantly change the polymer mass. Later it will be shown that this trapped moisture has a significant effect on the response of the piezoelectric system to gases. Flow Injection Experiments The sensors prepared by coating with doped poly(pyrro1e) (Table 1) were used as the gas-sensitive elements in the flow apparatus, Fig.2. A typical response to repeated 10ml 1% NH3 I 1 0.5% NH3 I 30 min Time - Fig. 6 Typical recorder trace showing the correlation of res onse of A, piezoelectric and B conductivity sensors coated with poly&yrrole) (Table 1, sensor D) to 10ml injections of NH3 into an air carrier stream 0 N -1oL \ g) -20 5 -30 - E -40 - $ -60 - k -70- c 0 $ -50- i -lo - g - 2 0 - -80 1 I 1 I I I I 0 100 200 300 400 500 600 Mass of polymer coating/vg Fig.7 Flow injection apparatus responses to 10 ml injections of 1% NH3, frequency change versus mass of polymer coating. Doping anion: a, bromide; 'I, sulphate; and ., nitrate injections of 1% ammonia in nitrogen can be seen in Fig. 6. The observed decrease in frequency is instantaneous (within the limitations of equipment response), as the gas interacts, but subsequent recovery takes approximately 10 min. The average frequency change observed for each of the crystals in responses to these injections versus the total mass of the crystal coating was plotted, Fig.7. The response increases with the mass of crystal coating; for the sulphate-doped poly(pyrro1es) (all dried at 100 "C for 10 min), this increase is almost linear, but for bromide-doped polymers, which were prepared under varying conditions, the response appears to be related to the drying conditions. Assuming that the two coatings contain the same amount of polymer and the mass difference is solely due to absorbed moisture, crystal D gives a response enhanced by the trapped water. This would be expected if the response was due to ammonia gas dissolving in the water; however, this is unlikely to result in the full recovery of the sensor. If mass was continually being added to the crystal, the baseline would drift dramatically in the direction of decreasing frequency.Additionally, the response would be expected to diminish as the surface of the poly- (pyrrole) became saturated with base, thus becoming unfav- ourable for interactions with ammonia. Moisture does, however, enhance the gas adsorption. The relative responses of the other, differently doped poly(pyrro1es) were also plotted (Fig. 7). Although individual responses vary the results are of the same magnitude. The nitrate-doped polymer gives the smallest response while the sulphate-doped polymers give the largest. Multi-gas Sensor Test Rig Experiments Typical responses to alternate pulses of air and 1% ammonia gas can be seen in Fig. 8. The response of sensor D (bromide-doped and dried at room temperature), Fig.8(a),1128 ANALYST, NOVEMBER 1991, VOL. 116 400 300 200 100 0 -100 - 200 - 300 -400 0 i p -loo c 2’ -200 S Q) CT -300 -400 0 10 20 30 1 1 I I 0 5 10 15 20 25 30 0 10 20 30 Time/min Fig. 8 Typical responses from gas test rig of ( a ) sensor D, ( b ) sensor E and (c) sensor A to alternate 5 rnin pulses of 1% NH3 and air shows an initial frequency decrease of over 300Hz on exposure to 1% NH3 for 5 min; no plateau is reached. The frequency then recovers to the baseline on subsequent exposure of the crystal to a stream of air. The following cycle of exposure to the sample gas causes a frequency decrease of less than 200Hz followed by a recovery of 400Hz. This pattern is repeated. Thus, if the frequency change is due to change in the mass of the sensor, after initial exposure to ammonia twice the amount of material is lost from the polymer surface as is originally adsorbed.This could simply be moisture lost from the polymer in the dry stream of air, or moisture depletion due to reaction with ammonia. However, this would be expected to result in a more rapid decrease in response as the experiment progresses due to a loss of moisture available for the ammonia reaction. Sensor E (bromide-doped and oven dried), Fig. 8(b), shows an initial frequency drop of about 250Hz but a reduced recovery, approximately 75 Hz. Subsequent exposure to ammonia causes a similar effect, in contrast to crystal D where the response diminished considerably. Also sensor E has a drifting baseline indicating increasing mass.This indicates the gas is not desorbed as efficiently as it is adsorbed; possibly Table 2 Response of three nominally identical sensors to 5 rnin pulses of ammonia gas Response? Mass of Standard polymer Frequency Weighted Mean deviation coating*/pg change/Hz response response (Yo) 55.9 99 1 985 886 882 825 56.5 838 1039 1069 1032 952 69.3 953 1068 1084 1239 1224 8 0.0142 0.0131 7 0.0141 0.0127 0.0126 0.0119 0.0180 0.01538 0.0147 0.0151 0.0156 0.0135 0.01 10 0.01284 10 0.0123 0.0125 0.0143 0.0141 * Mass calculated from the Sauerbrey equation. ? The frequency change caused by 5 rnin exposure to 1% ammonia. The weighted response is obtained by dividing the frequency change caused by the ammonia by that caused by the poly(pyrro1e) loading on the crystal. remaining water in the dried polymer is more strongly bound or deeply imbedded than in the ‘wet’ polymer.The response of sensor A (sulphate-doped and oven dried) to an analogous experiment was observed, Fig. 8(c). The results are similar to those for sensor E: a large initial decrease in frequency (100 Hz) followed by a recovery of approxi- mately 40Hz. Subsequently the size of the response is matched by the recovery. The sulphate response is marked by the virtual plateauing of the signal after 5min of exposure. This implies that all reaction sites have been filled. Recovery is very rapid suggesting that a response due to reaction with water vapour is not occurring. This pattern of response is most promising for an effective gas sensor. As shown in Fig. 8, the response to the first and subsequent exposures to ammonia was found to be different. The initial response was enhanced with the sulphate-doped material being most sensitive and the nitrate-doped material least sensitive.However, for subsequent measurements the responses were reduced and the order of sensitivity changed. Crystal D (air dried) gave a lower response than crystal E (oven dried). It appears that the wetter crystal has been desensitized by exposure to the ammonia, suggesting that there is a difference in the interaction between ammonia and either the water on the wet polymer or more strongly or deeply bound moisture on the dried polymer. It would appear that increasing the thickness of the poly(pyrro1e) coating will enhance the sensitivity of the sensor.There is however a limit to the mass of coating a crystal will allow, as large mass loadings will stop its steady oscillation.19 Additionally there is speculation as to the validity of the Sauerbrey equation at high crystal coatings. Table 2 shows the reproducibility of three nominally identical sensors to 5 min pulses of ammonia gas. Conductivity Measurements There is a strong correlation between the responses of the piezoelectric and conductivity sensors, Fig. 6. The bromide- doped poly(pyrro1e) coated conductivity sensor (Table 1, sensor I) was positioned downstream of the piezoelectric sensor, being impacted by the carrier gas and injected sample gas as they left the piezoelectric sensor cell. The initial resistance of the poly(pyrro1e) was 5800 kQ. A 10 ml injectionANALYST, NOVEMBER 1991, VOL.116 0 1 1 I 1 1 I I 1129 of 1% NH3 caused a resistance increase of 890 kQ, agreeing with results observed by Nylander et al. 10 They explained the change as a movement of electrons in the poly(pyrro1e) structure. Our piezoelectric experiments indicate that a response is caused by the reaction between ammonia gas and water; it is also expected that this would increase the poly(pyrro1e) conductivity. The implication is that two mechanisms may cause conductivity changes; however, the similarity between the results obtained for piezoelectric and conductivity measurements (Fig. 6) suggests that the sensing mechanisms are linked to those reported in the literature, with electrophilic gases giving increases in conductivity while nucleophilic gases give decreases in conductivity.Measurements with the piezoelectric sensors showed that of the gases investigated (NH3, CO, NO,, CH4 and C02) in the range 0.05-1%, only NH3 gave an initial sharp frequency decrease and could easily be discriminated. All other gases were characterized by an initial frequency increase. The shapes of these responses suggest a biphasic response mechan- ism and are reported in detail elsewhere.19Jo Calibration Graphs The Sauerbrey equation relates the mass adsorbed on the crystal surface to the frequency change. Beitnes and Schroder20 have questioned the validity of the Sauerbrey equation, at trace concentrations of gas, by showing that a plot of log frequency change versus log concentration should be a straight line with a slope of unity.Although our measurements are at higher concentrations a similar method of looking at the validity of the poly(pyrro1e) coated piezoelectric crystal response to ammonia was employed. A log-log plot of the NH3 response from 0.05 to 1% gas has a slope of 0.9 (Fig. 9), within the acceptable limits for such measurements. The exact mcchanism of interactions between the ammonia gas and the poly(pyrro1e) coated crystal is not clear although the response profiles, shown in Fig. 8, suggest a sorption related rather than diffusion related reaction. These profiles are in agreement with those of Bartlett and Ling-Chung21 who have used partial least squares curve fitting to indicate the profiles fit a sorption model better than either of two diffusion models considered.However, it is important to note that other schemes can yield a similar functional form and further work is required to elucidate the mechanism of response. The system responds linearly to gas concentration in the range 0.05-1% (Fig. 9). The conductivity mode is less sensitive to the gas concentration but can discriminate between different gases. Current work into evaluating a combined sensor may over- come some of the limitations experienced with separate devices. 2.5 r 1 Selectivity Poly(pyrro1e) is known to change in conductivity on exposure to a range of different gases, the exact response depending critically on the method of polymer preparation, particularly on the dopant ion.1.10 Other workers have reported the conductivity change of differently doped polymers to vapours and gases such as methanol and ethano1,21922 which cause a conductivity decrease, and N02,23 PC13 and S02,24 which give increases in conductivity.Our investigations have included NO2 and methanol for which conductivity increases and decreases, respectively, were also obtained; the response profiles are analogous to those for ammonia (Fig. 8). Conclusion This study has shown the potential of the conducting polymer poly(pyrro1e) as a gas sensor operating in a dual mode. Measuring mass and conductivity changes simultaneously is an important step in overcoming the ever present problem of moisture interference in the piezoelectric response. Other workers have employed several methods of overcoming this problem.14 Rigorous steps have been taken to dry gas samples.For example, tubing of the cation-exchange polymer Nafion (Perma Pure Products, Toms River, NJ, USA), a hygroscopic material which allows the transfer of water molecules through its walls, will facilitate the exchange of moisture from the gas travelling through the tube, to a desiccating agent packed around the tube. A second sensor may be used to obtain a measure of the moisture content which is then subtracted. The conductivity measurements used in this work are unaffected by moisture; thus the dual operation reported here provides a method of characterizing and compensating for the humidity response. The sensors reported have been used for periods of up to 1 month with no noticeable loss of sensitivity. Long-term stability trials are underway and will be reported elsewhere.25 Although some workers have reported that ammonia causes irreversible decreases in conductivity,26 we have not observed this effect and this is consistent with results reported by Miasik et al.1 and Nylander et al. 10 The authors thank British Gas for the funding and support of this work and in particular B. Price and D. Byrne of the British Gas Midlands Research Station for the development of the double crystal oscillator and the data acquisition software. We are grateful to Dr. M. J. Freeman, also of the Midlands Research Station, for assisting in the gas sensor test rig experiments and T. Cross for helpful and stimulating dis- cussions. 1 2 3 4 5 6 7 8 9 10 References Miasik, J. J., Hooper, A., and Tofield, B.C., J. Chem. SOC., Faraday Trans. I , 1986,82, 1117. Diaz, A. F., Castillo, J. I., Logan, J. A., and Lee, W., J. Electroanal. Chem., 1981, 129, 115. Asavapiriyanont, S., Chandler, G. K., and Gunawardena, G. A., J. Electroanal. Chem., 1984, 177, 229. Diaz, A. F., Kanazawa, K. K., and Gardini, G. P., J. Chem. SOC., Chem. Commun., 1979,635. Diaz, A. F., and Castillo, J. I., J. Chem. Soc., Chem. Commun., 1980, 397. Kanazawa, K. K., Diaz, A. F., Geiss, R. H., Gill, W. D., Kwak, J. F., Logan, J. A., Talbot, J. F., and Street, G. B., J. Chem. SOC., Chem. Commun., 1979, 854. Josowicz, M., and Janata, J., Anal. Chem., 1986, 58, 514. Dong, S., Sun, Z., and Lu, Z., J. Chem. SOC., Chem. Commun., 1988, 993. Faulds, N. C., and Lowe, C. R., J. Chem. SOC., Faraday Trans. I , 1986, 82, 1259. Nylander, C. Armgarth, M., and Lundstrom, I., Proceedings of the International Meeting on Chemical Sensors, Fukuoka, 1983, eds. Seiyama, T., Fueki, K., Shiokawa, J., and Suzuki, S., Elsevier, Amsterdam, 1983, pp. 203-207.1130 ANALYST, NOVEMBER 1991, VOL. 116 11 12 13 14 15 16 17 18 19 Alder, J. F., and McCallum, J. J., Analyst, 1983, 108, 1169. McCallum, J. J., Analyst, 1989, 114, 1173. Chemical Sensors, ed. Edmonds, T. E., Blackie, Glasgow, 1988, Ho, M. H., Guilbault, G. G., and Reitz, B., Anal. Chem., 1980, 52, 1489. Advances in Electrochemical Science and Engineering, eds. Gerischer, H., and Tobias, W., VCH, Weinheim, 1990, Lai, C. S. I., Moody, G. J., and Thomas, J. D. R., Analyst, 1986, 111, 511. Karmarkar, K. H., and Guilbault, G. G., Anal. Chim. Acta, 1974, 71, 419. Slater, J. M., and Watt, E. J., Anal. Proc., 1989, 26, 397. Lu, C., J. Vac. Sci. Technol., 1975, 12, 578. pp. 295-317. pp. 1-74. 20 Beitnes, H., and SchrGder, K., Anal. Chim. Actu, 1984,158,57. 21 Bartlett, P. N., and Ling-Chung, S . K., Sens. Actuators, 1989, 19, 141. 22 Gardner, J. W., Bartlett, P. N., Dodd, G. H., and Shurmer, H. V., NATO ASZ Ser., 1990, H39, 131. 23 Hanawa, T., Kunabata, S., and Yoneyama, H., J. Chem. SOC., Faraday Trans. I , 1988, 84, 1587. 24 Hanawa, T., and Yoneyama, H., Bull. Chem. SOC. Jpn., 1989, 62, 1710. 25 Slater, J. M., and Watt, E. J., unpublished results. 26 Gustafsson, G., and Lundstrom, I., Synth. Met., 1987,21,203. Paper 1 l00872B Received February 22nd, 1991 Accepted July 2nd, 1991