ANALYST, FEBRUARY 1991, VOL. 116 123 Voltammetric Behaviour of Screen-printed Carbon Electrodes, Chemically Modified With Selected Mediators, and Their Application as Sensors for the Determination of Reduced Glutathione Stephen A. Wring and John P. Hart* Science Department, Bristol Polytechnic, Coldharbour Lane, Frencha y, Bristol BS16 lax UK Brian J. Birch Unilever Research, Sensors Group, Colworth Laboratory, Colworth House, Sharnbrook, Bedford MK44 1 LO, UK The evaluation of screen-printed carbon electrodes, chemically modified with selected ferrocene, phthalo- cyanine and hexacyanoferrate(iii) derivative mediators for the determination of reduced glutathione (GSH), is described. Cyclic voltammetry was used t o investigate the effect of pH on the electrochemical behaviour of these mediators incorporated in the disposable electrodes.Values of the electron-transfer coefficient (an,) were calculated for the oxidation of the mediators in phosphate buffer solution and solutions containing 0.48 mmol dm-3 GSH. Amperometry in stirred solutions was used t o construct hydrodynamic voltammograms for each of the modified electrodes; these voltammograms were used t o elucidate their steady-state behaviour. Both electrochemical techniques were used t o calculate the reduction in overpotential for the oxidation of GSH; these calculations were performed for all of the chemically modified electrodes. Amperometry in stirred solutions was used as the technique for quantitative measurements, t o determine the calibration response factors (PA mmol-1 dm3) and limits of detection for selected mediators towards GSH.The appropriate applied potentials were selected by reference t o the hydrodynamic waves; a range of values were investigated t o find the potential that gave the maximum sensitivity. The most promising mediators were cobalt phthalocyanine, iron phthalocyanine and ferrocenecarbaldehyde. Keywords: Screen-printed carbon electrodes; chemically modified ferrocene, phthalocyanine and hexacyano- ferrate(ii1) mediators; reduced glutathione; cyclic voltammetry; amperometry Recently, there has been considerable interest in the use of disposable, chemically modified electrodes for the determina- tion of various biomolecules. 1-7 Indeed, careful selection of suitable electron mediators can significantly reduce the over- potential necessary for the determination of some analytes and enhance the selectivity of electroanalytical sens0rs.J-7 Frew and co-workers597 described this enhancement in a device for plasma glucose which uses the electrochemically generated ferricinium ion to act as an electron acceptor from the reduced flavoenzyme, glucose oxidase.Batchelor et aZ.6 also demonstrated that 4-methyl-o-benzoquinone could be incorporated in a disposable device for the determination of the ketone body 3-hydroxybutyrate. In a recent investigation,g we described a method of producing screen-printed graphite electrodes chemically modified with the important electrocatalyst cobalt phthalo- cyanine (CoPC); these devices successfully reduced the overpotential necessary for the determination of reduced glutathione (GSH), ascorbic acid and coenzyme A at plain graphite electrodes.This electrocatalyst has also been incor- porated in carbon-paste9.*" and re-usable carbon-poxy resin electrodes,lO which were used for the determination of GSH in human whole blood9 and plasma" by high-performance liquid chromatography with electrochemical detection. Our chemically modified screen-printed electrodes (SPEs) previously described8 were easily and reliably fabricated and could be employed, using differential-pulse voltammetry or in the amperometric mode with stirred solutions, for quantita- tive determinations in simple sample matrices. However, before analyses could be performed on complex biological samples their selectivity would need to be further enhanced.") Therefore, it was considered that the study of other mediators that could permit improved selectivity, through application of even lower applied potentials,'? was worthy of * To whom correspondence should be addressed.investigation. We set out to examine the possibility of using a variety of organometallic compounds as mediators for the determination of GSH. To our knowledge, apart from CoPC, none of these compounds has previously been used for this application with SPEs. This investigation involved three studies. The purpose of the first was to use cyclic voltammetry to examine the electrochemical behaviour of the selected mediators in plain phosphate buffer solutions, and those containing GSH, over a range of pH values. In the second part, hydrodynamic voltammograms were obtained by using simple amperometry in stirred solutions for each of the modified SPEs.Finally, amperometry in stirred solutions was used to compare the selectivity and sensitivity of the most promising electrodes for the determination of GSH. In these studies, GSH was used as the analyte of choice, because it is a very important cofactor in many physiological processes and also plays a key role in the detoxification of some common drugs;'3,14 in addition, changes in its circulating concentration can be used as a marker for certain dis- orders. 15-17 Experimental Chemicals and Reagents The CoPC was purchased from Kodak, all other mediators studied were obtained from Aldrich. Graphite (Ultra Carbon Ultra 'F' grade UCP-1M) and GSH were supplied by Johnson Matthey and Sigma, respectively.All the materials used for the production and cleaning of the screen-printing template were obtained from Sericol. The inert support used for the electrodes was semi-rigid, white poly(viny1 chloride) (PVC) marketed under the trade name Pentawhite and was obtained from ADP. The cellulose acetate and solvents used to prepare the graphite suspension for printing on to the Pentawhite were obtained from Aldrich.124 ANALYST, FEBRUARY 1991, VOL. 116 The supporting electrolyte used throughout all the investi- gations was phosphate buffer solution, prepared from 0.5 rnol dm-3 stock solutions of sodium dihydrogen orthophos- phate, disodium hydrogen orthophosphate and orthophos- phoric acid. These were mixed to yield buffer solutions of the required pH values (a pH meter was used).These were subsequently diluted to provide working solutions of 0.05 mol dm-3. All solutions were prepared with purified water (> 18 MQ cm) obtained using a Millipore Milli-Q purification system. All GSH solutions were prepared in the appropriate working buffer immediately prior to use and were protected from light during all investigations. Purified nitrogen was obtained from BOC. Apparatus Cyclic voltammetric and amperometric measurements were obtained using a Metrohm E612 VA-scanner in conjunction with a Metrohm E611-detector; these were used with a JJ Instruments PN4 x-y plotter to record voltammograms and amperograms. A three-electrode cell was used, incorporating the SPEs with a saturated calomel reference electrode (Russell Electrodes) and a laboratory-constructed platinum- wire counter electrode.Electrical contact to the SPEs was facilitated with a spade connector glued into a piece of glass tubing (15 x 0.3 cm i.d.) to form an electrode holder. For amperometric measurements in stirred solutions, a small circular stirring disc of 14 mm diameter (BDH) was placed in the bottom of the cell and rotated at a fixed rate by a Whatman Mini-MR stirrer. Elect rode Construct ion The SPEs consisted of a circular 3 mm working area with a 25 x 1 mm connecting strip. These were printed in parallel groups of six electrodes separated by a 7 mm space.* The SPEs were prepared by the method and template described previously.8 In brief, for the unmodified electrode this involved preparing a 1.5% m/m solution of cellulose acetate in a 1 + 1 (v/v) mixture of cyclohexanone and acetone; this solvent system permits the graphite film to adhere to the PVC support.Immediately prior to use, 1.1 g of this solution were added to 0.5 g of graphite in a small glass vial. These components were mixed to form an even suspension, which could then be printed through the screen on to the PVC support, which had previously been cleaned with ethanol. After use, the screen template was cleaned in a commercial thinner solution (Sericol XG) . For the modified electrodes, 5% m/m of the required mediator was added to the graphite. Once printed, the electrodes were left in the fume cupboard overnight to allow the solvents to evaporate. Immediately prior to use individual electrodes were cut from the piece of PVC, and the connecting strip was trimmed to 15 mm and covered with insulating tape (RS Components), leaving the 3 mm circular working area exposed, in addition to a 6 mm length at the opposite end to allow electrical contact with the spade connector in the electrode holder.Voltammetric Procedures Using SPEs Cyclic voltammetry Cyclic voltammetric measurements were obtained for blank solutions of 0.05 mol dm-3 phosphate buffer (pH 3, 5 and 7) and then for the same solutions containing 0.48 mmol dm-3 GSH, using both the unmodified and 5% m/m modified SPEs in order to investigate the effects of pH. Higher pH values were not studied because GSH is particularly unstable in alkaline media.18 The mediators used for this study were split into two groups; the first group consisted of molecules based on ferrocene, namely, ferrocene itself, dimethylferrocene, ferrocenedicarboxylic acid, ferrocenecarboxylic acid, ferro- cenecarbaldehyde and dimethylferrocenedicarboxylic acid.The second group consisted of CoPC, iron phthalocyanine (FePC), Prussian Blue and potassium hexacyanoferrate(n1). The voltammetric conditions were as follows: initial poten- tial, -0.5 V; scan rate, 20 mV s-1; and final potential, 1.2 V. All experiments were performed in triplicate, using a fresh electrode for each run; all the results quoted in later sections are the mean values for each of the parameters studied. Prior to each experiment the supporting electrolyte solution was de-aerated with purified nitrogen to eliminate the oxygen reduction waves.Hydrodynamic voltammetry Hydrodynamic voltammograms were obtained for both modi- fied and unmodified SPEs by amperometry in stirred solutions of 0.05 rnol dm-3 phosphate buffer (20 ml) and in similar buffer solutions containing 0.48 mmol dm-3 GSH. The applied potentials of the working electrodes were increased in steps; the resulting steady-state anodic-current responses were measured for each plateau and plotted versus applied poten- tial. Each experiment was performed in triplicate with fresh electrodes and the results quoted represent the mean current values. Calibration, Sensitivity and Selectivity By use of amperometry, in stirred solutions of 0.05 rnol dm-3 phosphate buffer (pH 7), the magnitude of the anodic current responses following additions of small volumes of stock GSH solutions was recorded over the final concentration range 1.48 x 10-7-2 x 10-3 mol dm-3 GSH (i.e., for solutions containing 1.48 X 10-7, 8.67 X 10-7, 4.76 x 10-6, 4.97 X 10-5, 4.76 X 10-4 and 2 x 10-3 rnol dm-3 GSH). In each instance, the stock GSH solution was added to 20 ml of plain buffer solution in the voltammetric cell, and the difference in the recorded current was measured; a fresh electrode was used for each individual determination .* The applied potential values selected for these investigations were taken, where possible, at different points along the hydrodynamic wave for each mediator studied. The sensitivity of the calibration response (calculated from the slope of the calibration graph) was then determined at every applied potential for each of the selected mediators; these results were plotted versus applied potential and could be used to establish the relationship between sensitivity and selectivity for each electrode type.Results and Discussion The voltammetric evaluation of the different mediators for the electrocatalytic determination of GSH at the modified screen- printed electrodes is described below. Cyclic Voltammetry Cyclic voltammograms were recorded, for both groups of the modified SPEs, in 0.05 rnol dm-3 phosphate buffer (pH 3, 5 and 7) and in similar solutions containing 0.48 mmol dm-3 GSH. Cyclic voltammetric studies on the ferrocene group of mediators For all but one mediator, the cyclic voltammograms obtained in plain buffer solutions, at the SPEs modified with the ferrocene compounds, showed a single quasi-reversible redox couple; the anodic (Ep,) and cathodic (Epc) peak potential values, and their separation (6Ep), are given in Table 1.The voltammograms obtained at the electrodes modified with dimethylferrocenedicarboxylic acid revealed only a singleANALYST, FEBRUARY 1991, VOL. 116 0.2 125 - +- ------ +------+ I I I I I Table 1 Values of peak potential recorded, using cyclic voltammetry in plain phosphate buffer solution, ferrocene group of mediators Buffer Mediator PH Ferrocene 3 5 7 Ferrocene- carbaldehyde 3 Dimethylferrocene 3 5 7 5 7 Ferrocene- carboxylic acid 3 5 7 Ferrocene- dicarboxylic acid 3 5 7 Dimethylferrocene- dicarboxylic acid 3 5 7 at the SPEs-modified EpJ V 0.330 0.347 0.372 0.540 0.540 0.530 0.200 0.203 0.223 0.428 0.315 0.315 0.640 0.460 0.450 0.810 0.810 0.800 EpJ V 0.146 0.137 0.150 0.250 0.240 0.200 0.063 0.063 0.070 0.310 0.237 0.233 0.490 0.370 0.360 - - - with -the 6Epl V 0.184 0.210 0.222 0.290 0.300 0.330 0.137 0.140 0.153 0.118 0.078 0.082 0.150 0.090 0.090 - - - -1 2 1 ' I I 1 I anodic wave; hence, the oxidation appeared to be irreversible over the potential range studied.For the determination of GSH involving mediated processes, anodic responses are of more importance than the cathodic responses because the magnitude of the former current is expected to increase during electrocatalysis. Bearing this in mind, the behaviour of the oxidation waves was investigated during the remainder of this study.The graph of E,, versus pH for the modified electrodes in plain buffer solution [Fig. l ( a ) ] indicates that the anodic wave 30 f 2o \ . -! 10 0 3 4 5 6 7 pH of the 0.05 mol dm-3 phosphate buffer Fig. 2 Effect of pH on ip, for the ferrocene group of mediators using ( a ) 0.05 mol dm-3 phosphate buffer and ( b ) solutions of the same buffer containing 0.48 mmol dm-3 GSH. Symbols as in Fig. 1 shifts to more negative potentials with increasing pH only for the electrodes modified with ferrocenecarboxylic and fer- rocenedicarboxylic acids; for the remainder of the mediators the electrochemical behaviour was independent of pH over the range studied. For the first two SPEs, a break in the graph line occurs at pH 5; this suggests the presence of a pK, (or pK') for the carboxylic acid groups at this approximate value.Similar redox behaviour was observed in the presence of 0.48 mmol dm-3 GSH [Fig. 1(b)]; however, in solutions of pH 7 an extra anodic wave (Epa, 0.65 V) was seen for the dimethylferrocenedicarboxylic acid-modified electrodes, and the peak for ferrocenedicarboxylic acid-modified devices had moved to a more positive potential. All of the electrodes, except those modified with dimethyl- ferrocene, afforded an enhanced anodic current response in the presence of GSH. Fig. 2(b) shows the difference in current measured between the plain buffer solution [Fig. 2(a)] and solutions containing GSH; a negative value indicates that the anodic current measured in the presence of GSH was less than that recorded in the plain buffer solution.Interestingly, the i,, versus pH graph reveals that the mediators with the most positive E,, values afforded the most enhanced current responses. Values of the electron-transfer coefficient (an,) were calculated,19 where possible, for the anodic waves obtained with plain buffer solution and with solutions containing 0.48 mmol dm-3 GSH (Fig. 3). This graph indicates clearly that the an, values determined at the modified SPEs, which demon- strate an electrocatalytic response for GSH, are reduced in the presence of the analyte. This observation is especially apparent for the SPEs modified with ferrocenedicarboxylic acid and ferrocenecarbaldehyde in pH 7 solutions. These electrodes were amongst those that exhibited the largest increase in current response in the presence of GSH.(Unfortunately, the an, values could not be determined for GSH at the other promising SPE containing dimethylferro- cenedicarboxylic acid, because the resulting new wave at 0.65 V was not sufficiently resolved.) The decrease in the observed an, values implies that, in the presence of GSH electrocataly- sis, the oxidation of the mediator becomes more irreversible.126 0.8 0.7 c" 0.6 0.5 ANALYST, FEBRUARY 1991, VOL. 116 - - - - 0.4 - 3 4 5 6 7 pH of the 0.05 mol dm-3 phosphate buffer Fig. 3 Effect of pH on the values of the electron transfer coefficient (an,), determined for the ferrocene group of mediators, in 0.05 mol dm-3 phosphate buffer (broken lines) and solutions of the same buffer containing 0.48 mmol d ~ r - ~ GSH (solid lines).Symbols as in Fig. 1 3 - However, as enhanced current responses were recorded at these devices, the reduction in the an, values could be reflecting the increased time taken during the ECE mechan- ism5 involved in the mediated oxidation of GSH compared with the simple E mechanism occurring in the plain buffer solutions. Therefore, the results suggest that the intermediate chemical reaction is the rate-determining step in the over-all oxidation process. Approximate values for the apparent experimental rate constant ( k , ) were calculated from the 6Ep values (at scan rates when 6Ep was greater than 200 mV), by the method of Laviron,20321 for two of the most promising mediators (ferro- cenecarbaldehyde and ferrocenedicarboxylic acid) in pH 7 buffer solutions.The mean value of k, for the ferrocene- dicarboxylic acid-modified electrodes, for scan rates between 150 and 250 mV s-1, was 0.357 s-1 [n = 3, relative standard deviation (RSD) = 13.9%]. This value agrees well with those determined by Laviron and Roullier21 for ferrocene-modified polymer electrodes. However, the calculated mean value of k , for the ferrocenecarbaldehyde-modified electrodes, under the same experimental conditions, was only 0.049 s-1 ( n = 3, RSD = 8.5%). This suggests that electron-transfer rates for the latter mediator in the present SPE matrix can appear very slow, particularly at high scan rates; this observation is confirmed by the large 6Ep values (Table 1) observed with this compound. . 0. ( a ) Cyclic voltammetry with the phthalocyanine and hexacyano- ferrate( 111) derivative mediators CoPC.The electrochemical behaviour of CoPC-modified electrodes has been described previously,g,10 and the observa- tions from the present study confirm these findings. In plain phosphate buffer solution, two irreversible anodic waves were recorded; e.g., in the pH 3 buffer solution these occurred at 0.48 and 0.80 V, respectively. Using the notation published previously,l" these were designated waves 2 and 3, respec- tively. When GSH was added to the supporting electrolyte solution, an additional irreversible anodic wave (wave 1) was observed at less positive potentials than waves 2 and 3 [Fig. 4(b)]. As peak 1 has been used successfully for the determina- tion of GSH in biological samples, it was again studied as the peak of interest for the remainder of this investigation.The cyclic voltammograms recorded for the FePC- modified SPEs in plain phosphate buffer solution (pH 3 ) revealed two anodic waves [Fig. 4(a)] and one broad cathodic (Epcl, -0.15 V) wave; the position of the second anodic wave was found to be dependent on the pH of the supporting electrolyte. In the pH 5 and 7 buffer solutions the broad cathodic wave was resolved into two smaller waves (pH 5: Epcl, -0.08V; E,,2, -0.23 V, and pH 7: Ep,l, -0.13 V; Epc2, -0.30 V). This suggests that the electrode reactions for FePC are reversible. FePC. I 0 . . . . . . . . o . . . . . . . o.2 t --------- 0 3 4 5 6 7 Fig. 4 Effect of pH on E,, for the phthalocyanine and hexacyanofer- rate(w) group of mediators using (a) 0.05 mol dm-3 phosphate buffer and ( b ) solutions of the same buffer containin 0.48 mmol dm-3 GSH.0, Prussian blue; A , FePC wave 1; #. FePC wave 2; 0, hexacyanoferrate(u1); and +, CoPC wave 1 pH of the 0.05 mol dm-3 phosphate buffer 1 1 I 1 2 t'b' -1 3 4 5 6 7 pH of the 0.05 mol dm-3 phosphate buffer Fig. 5 Effect of pH on ipa for the phthalocyanine and hexacyanofer- rate(m) group of mediators using ( a ) 0.05 mol dm-3 phosphate buffer and ( 6 ) solutions of the same buffer containing 0.48 mmol dm-3 GSH. Symbols as in Fig. 4 Similar voltammetric behaviour was observed for the electrodes in the buffer solutions containing 0.48 mmol dm-3 GSH [Fig. 4(b)]; however, an additional, poorly resolved, anodic wave was recorded at about 0.7 V for each of t h e pH values studied (not shown). The magnitude of the anodic waves 1 and 2 was also found to increase in the presence of GSH in pH 7 phosphate buffer solution [Fig.5(a) and (b)]. As the electrocatalytic response of waves 1 and 2 was the most promising in terms of sensor selectivity, their behaviour wasANALYST, FEBRUARY 1991, VOL. 116 f 1.0 0.8 . * I 2? 2 0.6 0 5 0.4 0 2 0.2 I 9 0 - 127 - - - - - investigated during the remainder of this study, and is summarized in Figs. 4 and 5 . HexucyanOferrate(iii). For each buffer pH studied, two anodic waves (Epal, about 0.30 V; Epa2, about 0.45 V) and one cathodic wave were recorded; the behaviour of the first anodic wave is summarized in Fig. 4(a) and ( 6 ) . The second anodic wave observed for this usually model redox system probably arises owing to some interaction of the hexacyanoferrate(111) with the electrode matrix or solvent system.This behaviour has been described previously for other modified polymer-based electrodes,2* and this process is considered probable because only a single, small and tailing cathodic wave was observed; in addition, at applied potentials below 0.0 V the current response became very erratic and noisy. Fig. 5(u) and (b) clearly indicates that, despite the presence of these matrix interactions, the magnitude of the first anodic wave increases in the presence of GSH, with the maximum response occurring in pH 3 buffer solutions. Prussian Blue [iron( 111) hexacyanoferrute(rr~)] . The cyclic voltammograms of the Prussian Blue-modified electrodes, obtained at each buffer pH studied in the absence of GSH, reveal an ideal quasi-reversible redox couple with single anodic and cathodic waves.As seen in Fig. 4(u) and ( b ) the position of the peak potential (Epa) is independent of pH over the range studied. In buffer solutions containing 0.48 mmol dm-3 GSH, the anodic current increases to give the maximum mediated response at pH 5 [Fig. 5(u) and ( b ) ] . The values of an, were calculated for the anodic responses shown in Fig. 5(a) and (6) for each of the second group of modified SPEs (Fig. 6). This graph confirms the observations made by using the ferrocene group of electrodes, where the electron-transfer coefficient is decreased in the presence of a mediated response to GSH. The mechanism for this process is undoubtedly similar to that mentioned previously and involves an ECE mechanism as the Fe2+ moiety is electro- chemically oxidized to Fe3+, whereupon it is chemically reduced by the GSH and subsequently re-oxidized electro- chemically.For the CoPC-modified SPEs, values of an, could only be determined in the presence of GSH because wave 1 is absent in plain buffer solution; the mechanistic behaviour of this mediator has been described previously.10 Hydrodynamic Voltammetry For biomedical sensor applications with use of enzymes to enhance selectivity, buffer solutions usually need to be at, or near, physiological pH values. In the second part of our current investigation, we studied the anodic current response of each of the modified SPEs to GSH in pH 7 phosphate buffer solution, using amperometric detection in stirred solutions.0.6 - 2 0.4 - 0.2 - I I 1 I I I 1 3 4 5 6 7 pH of the 0.05 mot dm-3 phosphate buffer Fig. 6 Effect of pH on the values of the electron transfer coefficient (an,), determined for the phthalocyanine and hexacyanoferrate(li1) group of mediators, in 0.05 mol dm-3 phosphate buffer (broken lines) and solutions of the same buffer containing 0.48 mmol dm-3 GSH (solid lines). 0, Prussian Blue; A , FePC wave 1; +, CoPC wave 1; and 0, hexacyanoferrate(iri) wave 1 This simple detection technique is particularly important for sensor applications and has been used successfully for the analysis of selected biomolecules in our work and that of other workers. 10,2244 Hydrodynamic voltummetry with the ferrocene group of mediators The hydrodynamic voltammograms [Fig.7 ( u ) ] indicate that, for solutions of GSH prepared in phosphate buffer solution (pH 7), enhanced anodic current responses are observed only at the SPEs modified with ferrocenedicarboxylic acid, ferro- cenecarbaldehyde and dimethylferrocenedicarboxylic acid. These findings confirm the observations carried out by cyclic voltammetry in the same buffer solutions. These voltammo- grams were constructed by plotting the difference in current measured between the response in plain phosphate buffer solution (pH 7) and the same solution containing 0.48 mmol dm-3 GSH; the current at the initial applied potential, under steady-state conditions, was used as the zero reference point. Hence, the voltammograms illustrate the current arising solely from the mediated oxidation of GSH at the electrode surface.1.2 1 .o 0.8 0.6 0.4 0.2 0 1 0 100 200 300 400 $00 600 700 800 900 1'2 I (b) - P- -0.2 I I i I I 1 I -200 0 200 400 600 800 25 -200 0 200 400 600 800 1000 Applied potential/mV versus SCE Fig. 7 Hydrodynamic voltammograms for ( a ) the fcrrocene group of mediators; (b) CoPC and the hexacyanoferrate(ii1) mediators; and ( c ) FePC. The anodic current values represent the difference in the measured current between 0.05 mol dm-3 phosphate buffer (pH 7) solutions and those containing 0.48 mmol dm-3 GSH, and hence illustrate the current arising from thc mediated oxidation of GSH at the modified electrodes. (a): x , Ferrocene; A, ferrocenecarb- aldehyde; + , dimethylferrocene; V , ferrocenecarboxylic acid; 0, ferrocenedicarboxylic acid; and 0, dimethylferrocenedicarboxylic acid.(6): 0, Prussian Blue; 0, hexacyanoferrate(ii1); V, CoPC; and A , unmodified128 1 x 10-3 ANALYST, FEBRUARY 1991, VOL. 116 - + I 1 I Hydrodynamic voltammetry with the phthalocyanine and hexacyanoferrate(m) group of mediators The hydrodynamic voltammograms for the four mediators confirm their electrocatalytic behaviour in stirred solutions [Fig. 7(b) and (c)]. The response for the CoPC-modified SPEs clearly reveals a ‘shoulder’ at 0.4 V on the main voltammo- gram owing to the first analytical wave (peak l), which is seen only in the presence of GSH. Fig. 7(b) also shows the current response recorded at the unmodified SPEs; as expected, negligible current values were recorded at the applied potentials, corresponding to the maximum steady-state currents for the mediated electrodes.Reduction of Overpotential With the Different Mediators The results from both of the voltammetric studies indicate clearly that the overpotential necessary for the electrochem- ical detection of GSH can be decreased by using suitable electron mediators. Final selection of the optimum mediator for a particular application will depend on its efficiency at reducing overpotential and its sensitivity, i.e., its current response factor (PA mmol-* dm3) during calibration. The reductions in overpotential for the electrocatalytic oxidation of GSH are given in Table 2 for each of the modified electrodes that demonstrated a favourable response. The values given are the differences between the E,, values for the oxidation of GSH at the unmodified SPEs, obtained by cyclic voltammetry (Epa, 1.26 V) and hydrodynamic voltammetry (>1.4 V, a plateau was not reached within the usable potential window of the electrodes), and the E,, values at the modified devices.From those data, and the magnitude of the amper- ometric current responses seen in Fig. 7(a-c), the most promising mediators for the selective determination of GSH at pH 7 are: CoPC, FePC, ferrocenedicarboxylic acid, ferro- cenecarbaldehyde and Prussian Blue. Large increases in the current response were also recorded for dimethylferrocene- dicarboxylic acid; however, for the last mediator the applied potentials necessary for oxidation were deemed to be too positive for practical sensor applications.Calibration In the final part of this investigation, amperograms were recorded, by using the method described previously,g for the most promising modified electrodes. The concentrations studied covered the range 1.48 x 10-7-2 x 10-3 mol dm-3 GSH, and the applied potential values were selected from different points on their hydrodynamic waves. The current Table 2 Comparison of the reduction in measured overpotential for the oxidation of GSH between the modified and unmodified SPEs in 0.05 mol dm-3 phosphate buffer (pH 7); where: AT]CV = ECVunmodrfied - ECVrnodified and AT]AMP = EAMPunrnodifled - EAMPrnodified Mediator Ferrocene Ferrocene- carbalde h yde Dimethylferrocene- dicarboxylic acid Ferrocene- dicarboxylic acid Prussian Blue FePC (wave 1) FePC (wave 2) Hexacyanoferrate(n1) CoPC (wave 1) Cyclic voltammetry 889 ( AT]cv)/mV * Indicates no significant response.730 610 720 1080 1180 1010 960 960 Amperometry (A%4MP)ImV * > 880 >670 > 1000 >1200 >lo10 >950 > 1000 * response factors were calculated from the slope of the calibration graph for each of the mediators by using the different applied potentials; these values were plotted versus potential to yield the current response profiles [Fig. 8(a) and (b)]. Fig. 8(a) indicates that FePC-modified electrodes offer the best selectivity during calibration owing to the low applied potentials required. However, for optimum sensitivity, the ferrocenecarbaldehyde-modified SPEs become the devices of choice, although for concentrations of <4.76 x 10-6 mol dm-3 GSH, high background currents and noise restric- ted the practical detection limits to this concentration when the applied potential was 400 mV and to 8.67 x 10-7 rnol dm-3 for Eapplied = 525 mV.By using the most sensitive potential (450 mV) the limit of detection was 1.48 x 10-7 mol dm-3 GSH as determined for the other devices. Fig. 8(a) also indicates that the optimum sensitivity cannot be obtained on the plateau of the hydrodynamic wave, but at a potential slightly greater than E0.5. Fig. 8(b) illustrates the current response factors plotted versus the position of the applied potential on the hydrodynamic wave. For the ferro- cenecarbaldehyde- and CoPC-modified electrodes the opti- mum response was achieved at approximately the E0.65 point.Unfortunately, owing to excessive baseline noise, calibration was not possible for the full range of potentials with the FePC- and ferrocenedicarboxylic acid-modified electrodes; however, the slopes of their response curves do increase towards the mid-point of the hydrodynamic wave. A possible explanation for this observation involves the redox cycling mechanism that was proposed earlier. At lower applied potentials the energy barrier for the chemical reduction of the oxidized electron mediator by the GSH will be at a minimum, allowing this reaction to proceed at a greater rate. This process would in turn replenish the reduced form of the mediator for further electro-oxidation and hence the generation of larger electro- catalytic currents.ANALYST.FEBRUARY 1991, VOL. 116 129 t c t ?! 3 u -0.4 0 0.4 0.8 -0.4 0 0.4 0.8 PotentialN versus SCE I I 1 -0.4 0 0.4 0.8 Fig. 9 GSH (solid lines) using: ( a ) ferrocenecarbaldehyde: ( b ) FePC; and ( c ) ferrocenedicarboxylic acid modified SPEs. Scan rate, 20 mV s-1 Cyclic voltammograms recorded in 0.05 mol dm-3 phosphate buffer (broken lines) and the same solutions containing 0.48 mmol dm-3 For comparison, Fig. 9(a)-(c) shows the cyclic voltammetric behaviour of these iron-containing mediators in 0.05 mol dm-3 phosphate buffer (pH 7) and for the same solutions containing 0.48 mmol dm-3 GSH; the cyclic voltammograms for CoPC have been described previously.8.") No response was observed with use of amperometry at constant potential for the Prussian Blue-modified electrodes during calibration.(The experiments were repeated and the results confirmed; at present we are unable to explain this phenomenon.) The highest current responses were obtained with the phthalocyanine-based sensors. This suggests that they should offer better sensitivity and selectivity when compared to other mediators. Conclusion The voltammetric data and calibration results reported here suggest that the screen-printed carbon electrodes can be used as a substrate suitable for modification with a variety of electron mediators. The systematic evaluation of these elec- trocatalysts has shown that, with careful selection of the potential, the oxidation process for GSH can be controlled; therefore, the selectivity and sensitivity can be altered to suit the particular application.This study has shown that, of the mediators selected initially, the most promising for the determination of GSH are: CoPC, FePC and ferrocenecarbaldehyde . The most selective media- tor, requiring the least positive applied potential, was FePC, while the most sensitive was ferrocenecarbaldehyde. However, both of these responses were superimposed on anodic background currents arising from an initial electrochemical oxidation of the Fez+ moieties. Conversely, the anodic current measured at the CoPC devices arises solely from the presence of the GSH in solution, which could permit faster response times for some applications. It is envisaged that the screen-printed carbon electrodes chemically modified with these mediators, and used in conjunction with selective enzymes, could form the basis of selective biosensors for the determination of GSH in biolog- ical matrices.The authors thank the National Advisory Board for financial support. They are also grateful to Apple Litho (Bristol) for their help and advice regarding the screen-printing of the carbon electrodes. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 References Turner, A . P. F., Senb. Actuar., 1989, 17, 433. Scheller, F., Schubert, F., Pfeiffer, D., Hintsche, R., Drans- feld, I., Renneberg, R.. Wollenberger, U., Riedel, K., Pavlova, M.. Kuhn, M., Muller, H.-G., Tan, P. M., Hoffmann. W., and Moritz, W.. Analyst, 1989. 114, 653. Frew, J . E., Bayliff, S. W., Gibbs, P. N . B., and Green, M. J . , Anal. Clzim. Actu, 1989,224, 39. Patriarche. G . J . , Kauffmann, J.-M., and Vire, J.-C., Redox Chem. Interfacial Behav. Biol. Mol., 1987, 3. 479. Frew, J . E., and Hill, H . A. O., Anal. Chem.. 1987, 59. 933A. Batchelor, M. J . , Green, M. J., and Sketch, C. L . . Anal. Clrim. Acra, 1989, 221, 289. Frew, J . E., and Green, M. J., Anal. Proc., 1988, 25, 276. Wring, S. A . , Hart, J. P., Bracey, L.. and Birch, B. J., Anal. Clrim. Acta, 1990, 231, 203. Halbert, M. K.. and Baldwin. R. P., J . Chromarogr., Biomed. Appl., 1985. 345, 43. Wring. S. A., Hart, J . P.. and Birch, B. J., Analyst, 1989, 114, 1563. Wring, S. A . , Hart, J. P., and Birch, B. J., Analyst, 1989, 114, 1571. Imisides, M. D., Wallace, G . G . , and Wilke, E. A . , TrAC, Trends Anal. Chem., Pers. Ed.. 1988, 7, 143. Meistcr, A., J . Biol. Cliem., 1988, 263, 17205. Black. M.. Ann. Rev. Med.. 1984, 35. 577. Curello. S., Ceconi, C.. Cargnoni, A , , Cornacchiari. A , , Ferrari, R., and Albertinia, A., Clin. Clzem., 1987, 33, 1448. Hall. R., and Malia, R. G., Medical Laboratory Haematology. Butterworths, London, p. 294. Eastman, K. D., Clinical Haematology. Wright, Bristol, 6th edn., 1984. p. 126. Perrett, D., and Rudge, S. R.. J. Plrarm. Biomed. Anal., 1985, 3, 3. Galus. Z., Fundamentals of Electrochemical Analysis. Ellis Horwood, Chichester. 1976, p. 237. Laviron, E., J. Electroanal. Chem., 1979, 101, 19. Laviron, E., and Roullier, L., J . Electrounal. Chem., 1980, 115, 65. Swain, A.. Int. Ind. Biotechnol.. 1988, 8. 11. Bennetto, H. P., Dekeyzer. D . R.. Delaney, G. M., Koshy, A., Mason, J . R., Razack, L. A . , Stirling, J . L., and Thurston, C. F., Int. Analyst, 1987, 8, 22. Wring, S. A., Hart, J . P.. and Birch, B. J., Anal. Chim. Acta, 1990, 229, 63. Paper 0103295F Received July 23rd, I990 Accepted September 27th, 1990