Anal. Chem. 1996, 68, 2635-2640

A Glucose Sensor Made of an Enzymatic Clay-Modified Electrode and Methyl Viologen Mediator Jyh-Myng Zen* and Chin-Wen Lo

Department of Chemistry, National Chung-Hsing University, Taichung, Taiwan 402, Republic of China

A novel glucose sensor has been contrived by immobilizing glucose oxidase between two nontronite clay coatings on glassy carbon electrode with methyl viologen as mediator. The sandwich configuration proved to be very effective in the determination of glucose. The response of the glucose sensor was determined by measuring cyclic voltammetric peak current values under aerobic solution conditions. The effects of the amount of enzyme immobilized, the operating pH, and the common interferences on the response of the glucose sensor were studied. The detection limit was 5 µM, with a linear range extending to about 6 mM, giving a dynamic range of over 3 orders of magnitude for 0.1 mM methyl viologen. When stored in pH 7 phosphate buffer at 4 °C, the sensor shows almost no change in performance after operating for at least 2 months. A mechanism for the operation of the glucose sensor is also proposed. The development of an electrochemical glucose sensor with an immobilized enzyme on a chemically modified electrode surface has received continuous interest.1-10 Traditionally, glucose sensors have relied on the immobilization of glucose oxidase (GOx) onto various metallic or carbon transducers and monitoring of the current associated with the oxidation of the liberated hydrogen peroxide. Various methods have been used for the immobilization of GOx enzyme in the fabrication of glucose sensors. These methods include cross-linking with glutaraldehyde, chemical immobilization on carbon carriers, and entrapment in polymer layers, metal matrices, or electrochemically grown polymer films on the electrode surface.11-16 Overall, no matter what kind of (1) Kulys, J. J. Analytical Systems Based on Immobilized Enzymes; Mokslas: Vilnius, 1981. (2) Mascini, L.; Guilbault, G. G. Sensors 1986, 2, 147. (3) Murray, R. W.; Ewing, A. G.; Durst, R. A. Anal. Chem. 1987, 59, 379A. (4) Rechnitz, G. A. Electroanalysis 1991, 3, 73. (5) Cass, A. E. G.; Davis, G.; Green, M. J.; Hill, H. A. O. J. Electroanal. Chem. 1985, 190, 117. (6) Frew, J. E.; Hill, H. A. O. Anal. Chem. 1987, 59, 933A. (7) Degani, Y.; Heller, A. J. Am. Chem. Soc. 1989, 111, 1357. (8) Coury, L. A.; Oliver, H. A. O.; Egekeze, J. O.; Sosnoff, C. S.; Brumfield, J. C.; Murray, R. W. Anal. Chem. 1990, 62, 452. (9) Garguilo, M. G.; Michael, A. C. Anal. Chem. 1994, 66, 2621. (10) Bartlett, P. N.; Birkin, P. R. Anal. Chem. 1994, 66, 1552. (11) Ikariyama, Y.; Yamauchi, S.; Yukiashi, T.; Ushioda, H. Anal. Lett. 1987, 20, 1791. (12) Chi, Q.; Dong, S. Anal. Chim. Acta 1991, 278, 17. (13) Bartlett, P. N.; Cooper, J. M. J. Electroanal. Chem. 1993, 362, 1. (14) Gulce, H.; Ozyoruk, H.; Celebi, S. S.; Yildiz, A. J. Electroanal. Chem. 1995, 394, 63. (15) Adbel-Hamid, I.; Atanssov, P.; Wilkins, E. Anal. Chim. Acta 1995, 313, 45. (16) Brown, R. S.; Luong, J. H. T. Anal. Chim. Acta 1995, 310, 419. S0003-2700(96)00090-X CCC: $12.00

© 1996 American Chemical Society

approach is used, the key parameters for optimization of enzyme stability and sensitivity are selection and modification of the immobilization matrix. In our previous study, excellent catalytic activity was observed for the electroreduction of hydrogen peroxide by nontronite (SWa-1, ferruginous smectite) clay coating on glassy carbon electrode (SWa-1-GCE) with incorporated methyl viologen as electron mediator.17 The present study further extends the previous observation to construct a novel glucose sensor by immobilizing GOx onto the SWa-1-GCE. As far as we know, no report on the use of clay film in sensor fabrication can be found up to now. Furthermore, because of colloidal clays’ appreciable surface area, intercalation properties, low cost, high stability, and high cation exchange capacity, their uses for fabricating sensors certainly deserve an extensive study. We report here the construction and characterization of a novel glucose sensor by immobilizing GOx between two nontronite clay coatings on GCE. Methyl viologen centers present in the clay coatings are expected to act as mediators for the reduction of the hydrogen peroxide produced during the enzymatic reaction. EXPERIMENTAL SECTION Chemicals and Reagents. Glucose oxidase (GOx, EC 1.1.3.4, 181 600 units/g, from Aspergillus niger), β-D-glucose, bovine serum albumin (BSA), and methyl viologen (MV2+) were bought from Sigma (St. Louis, MO). The interferents used, L-ascorbic acid, uric acid, cystine, galactose, and oxalate, were also from Sigma. Glutaraldehyde (GA) was obtained from Merck. Standard clay mineral, nontronite (SWa-1, ferruginous smectite), was purchased from the Source Clay Minerals Repository (University of Missouri, Columbia, MO). All the other compounds (ACS-certified reagent grade) used in this work were prepared without further purification in doubly distilled deionized water. Apparatus. Electrochemistry was performed on a Bioanalytical Systems (BAS, West Lafayette, IN) CV-50W electrochemical analyzer. A BAS Model VC-2 electrochemical cell was employed in these experiments. The three-electrode system consisted of a nontronite/methyl viologen-modified glucose sensor, a Ag/AgCl reference electrode (Model RE-5, BAS), and a platinum wire auxiliary electrode. The supporting electrolyte was 0.1 M, pH 7 phosphate buffer solution in most cases. Electrode Preparation. The GCE (3 mm diameter, BAS) was polished with a polishing kit (BAS), and clay colloids were prepared in the sodium form, generally according to the proce(17) Zen, J.-M.; Jeng, S.-H.; Chen, H.-J. J. Electroanal. Chem., in press.

Analytical Chemistry, Vol. 68, No. 15, August 1, 1996 2635

dures previously described.17,18 Clay films were prepared by dropping 4 µL of a clay colloid (0.5 g/L) onto a clean GCE and drying under ambient conditions, usually ∼0.5-1 h. Uniform films could be cast reproducibly from clays. They were subsequently coated with a GOx enzyme layer and finally another clay layer. A typical enzyme casting solution was prepared by dissolving 20 mg of GOx, 15 mg of BSA, and 100 µL of 5% GA in sequence in 1 mL of 0.1 M, pH 7 phosphate buffer solution. General Procedure. All solutions were used directly for detection under open air at room temperature. The glucose response of the electrode was measured as the peak current value corrected for the base line from cyclic voltammograms taken between 0.0 and -0.9 V vs Ag/AgCl at a scan rate of 100 mV/s. The measurements can be done in two ways: First, a suitable amount of MV2+ was directly added to the measuring system containing glucose. Second, the steady-state background current of MV2+ was measured first, and the enzyme electrode was then switched to a glucose solution for detection. Note that the measuring systems also contain the same concentration of MV2+ to prevent leaching problem. It takes about 2-3 min to reach the steady state for 1 mM MV2+, and a longer time is needed for a lower concentration of MV2+. Overall, as long as the steady state was reached, virtually the same results were obtained either way. For convenience, most of the experiments were performed in the first way described. The enzyme electrode was kept in a buffer solution (pH 7) at 4 °C when not in use. RESULTS AND DISCUSSION Electrochemical Characterization of the Glucose Sensors. The electrochemical behavior of MV2+ incorporated into the nontronite-modified GCE (SWa-1-GCE) was thoroughly studied previously.17,18 Among the standard clay minerals, kaolinite (KGa1), illite, bentonite, nontronite (SWa-1, ferruginous smectite), montmorillonite (SWy-1), and vermiculite (VTx-1), tested, the steady-state reduction currents for MV2+ at the clay-modified electrodes are in the following order: SWa-1 > illite > SWy-1 > bentonite > KGa-1 > VTx-1. The electrochemical behavior of paraquat with the cathodic peak at -0.70 V vs Ag/AgCl permits adequate quantification of the analyte. Figure 1A (dashed line) shows the typical steady-state cyclic voltammograms for MV2+ incorporated into the SWa-1-GCE. Excellent catalytic activity was observed for the electroreduction of H2O2 by the SWa-1-GCE with incorporated MV2+ as mediator, as shown in Figure 1A (solid line). There is a dramatic change in the cyclic voltammetry behavior of MV2+ when H2O2 is added into the solution. The reversible cyclic voltammograms of MV2+ are replaced by an S-shaped curve, showing characteristics of a chemical reaction following the electron transfer step that regenerates the electroactive material (i.e., a catalytic, or EC, mechanism).19 The role of the nontronite coating in the catalysis of the electroreduction of H2O2 and the mechanism of the reaction are proposed in our previous study as shown in Figure 1B.17 Since glucose sensors normally rely on the immobilization of GOx onto various transducers and monitoring of the current associated with the oxidation of the liberated hydrogen peroxide, three different sensor configurations were investigated. The configuration and extent of interaction of the nontronite and GOx were altered in each case, as shown in Figure 2. First, the sensor (18) Zen, J.-M.; Jeng, S.-H.; Chen, H.-J. Anal. Chem. 1996, 68, 498. (19) Bard, A. J.; Faulkner, L. R. Electrochemical Methods; Wiley: New York, 1980.

2636 Analytical Chemistry, Vol. 68, No. 15, August 1, 1996

Figure 1. (A) Cyclic voltammograms for the reduction of MV2+ with (solid line) and without (dashed line) the addition of 5 mM H2O2 into 0.5 mM MV2+ and 0.05 M, pH 8 phosphate buffer solution at the SWa1-GCE. Scan rate, 50 mV/s. The solution was deoxygenated with argon. (B) Proposed reaction mechanism for the reduction of H2O2 at the SWa-1-GCE with MV2+ as mediator.

Figure 2. Configurations for the three different electrode types (A-C) of nontronite/methyl viologen-modified glucose sensors.

was constructed by using GA and BSA to immobilize GOx onto the SWa-1-GCE and further coated by another layer of the SWa-1 clay (electrode type A). Second, the glucose sensor was constructed with an inner clay and outer enzyme layers (electrode type B). Third, the glucose sensor was constructed with an outer clay and inner enzyme layers (electrode type C). Cyclic voltammograms similar to that of Figure 1A were then obtained for the three different configurations of glucose sensors, as shown in Figure 3. As can be seen, the electrode type A shows the best performance (Figure 3A). Apparently, the presence of the clay layer is not detrimental to the enzyme. Methyl viologen centers,

Figure 3. Cyclic voltammograms for the detection of 5 mM glucose in 0.1 mM MV2+ and 0.1 M, pH 7 phosphate buffer solution with the electrode types A-C. Scan rate, 100 mV/s.

which are present between the inner and the outer clay layers, can then act as mediators for the oxidation of the hydrogen peroxide. On the other hand, it is interesting to see that almost no response was observed for the electrode type C at all (Figure 3C). This phenomenon is expected since the inside enzyme layer, directly coated on the GCE surface, can actually block the role of the nontronite coating and the MV2+ mediator in the catalysis of the electroreduction of H2O2. As mentioned earlier, the catalytic activity cannot operate without the existence of both the nontronite and the MV2+ mediator. Hence, the result observed is quite consistent with the proposed mechanism (Figure 1B). Finally, the electrode type B was found to improve the catalytic activity of the MV2+ mediator but with less effective catalytic performance than that of the electrode type A. This is because that part of the hydrogen peroxide produced by the enzymatic reaction can quickly diffuse into the solution and therefore reduce the sensitivity of the glucose sensor. Overall, the outer clay coating, as shown in Figure 3A, was proved to be critical in the preparation of the glucose sensor for the following three reasons: First, the outer clay coating can prevent the hydrogen peroxide from diffusing into the solution and improve the sensitivity of the glucose sensor. Second, it can also exclude electroactive interferents and keep the sensor surface free from fouling agents. Third, the long-term stability of the enzyme layer was found to be largely improved. Note that the last two points will be discussed in more detail and that the electrode type A is designated as SWa-1/MV-GS below. Catalytic oxidation of glucose by the proposed sensor was studied next. Figure 4 shows the typical cyclic voltammograms for the SWa-1/MV-GS with (A) and without (B) the presence of glucose under aerobic conditions. The oxygen reduction peak can be clearly seen at -0.55 V vs Ag/AgCl in the absence of glucose, while it completely disappears in the presence of glucose. Obviously, the reduction peak current of the MV2+ increases at the expense of the dissolved oxygen. This result further confirms the proper function of the sensor, since glucose oxidase enzyme catalyzes the aerobic oxidation of glucose as follows:

Figure 4. Typical cyclic voltammograms for the SWa-1/MV-GS in 0.1 mM MV2+ and 0.1 M pH 7 phosphate buffer solution with (A) and without (B) the presence of glucose under aerobic conditions. Scan rate, 100 mV/s. x

x

Figure 5. Proposed reaction mechanism for the detection of glucose at the SWa-1/MV-GS. GOx

glucose + O2 + H2O 98 gluconic acid + H2O2

Considering the above results, the mechanism of the reaction at the SWa-1/MV-GS is proposed to proceed as shown in Figure 5. Optimization of the Glucose Signal on the SWa-1/MV-GS. The effects of experimental conditions, such as the amount of enzyme immobilized, the operating pH, and the mediator concentration, were investigated to optimize testing performance. The results of the optimization of GOx loading in-between the SWa-1 clay layers are shown in Figure 6. The data show that the response increases with increasing enzyme concentration in the range of 2-10 mg/mL. When the enzyme concentration is above 10 mg/mL, the increase becomes slower. The response starts to decrease when the enzyme concentration is higher than 15 mg/ mL. If both sensitivity and economical factors are simultaneously considered, the optimum concentration of enzyme should be in the range of 7-12 mg/mL. An enzyme concentration of 10 mg/ mL was therefore selected in subsequent studies. Figure 7 shows the effect of solution pH on the cyclic voltammetric response at the SWa-1/MV-GS. GOx operates over a wide pH range (2.7-8.5), and the optimum is usually reported to lie between 4.8 and 6.0 for the free enzyme when dioxygen is used as the electron acceptor.20 The best response at pH 7.0 for the SWa-1/MV-GS indicates that the sensor is apparently also affected by other factors. Numerous earlier studies have de(20) Uhlig, H. In Enzymes in Industry; Gerhartz, W., Ed.; VCH: Weinheim, 1990; p 77.

Analytical Chemistry, Vol. 68, No. 15, August 1, 1996

2637

x

Figure 6. Effect of enzyme loading upon the response to 5 mM glucose at the SWa-1/MV-GS. Other conditions as in Figure 4. Data were obtained with a background subtraction of the signal without the addition of glucose.

Figure 7. Effect of solution pH on the cyclic voltammetric response at the SWa-1/MV-GS. Conditions as in Figure 4, except in different 0.1 M buffer solutions. Data were obtained with a background subtraction of the signal without the addition of glucose.

scribed a shift in the pH optimum exhibited due to the use of artificial electron acceptors.21 Immobilization is another factor that may alter the operating characteristics of the enzyme.22 Hence, the observation that the optimum value is 7.0 is not necessarily an indication that the behavior of the enzyme has been altered. A proper explanation for the result, however, can be provided by the pH effect on the cyclic voltammetric response at the SWa-1GCE, as discussed in our previous study.17,18 The mediator shows the best catalytic ability toward hydrogen peroxide at pH 8.0, and the clay starts to show an unstable behavior when the pH is higher than 9. The best response at pH 7.0 for the SWa-1/MV-GS is therefore a compromise result between the optimum pH values of the enzyme (∼pH 5.5) and the clay (pH 8.0). A buffer solution of pH 7.0 was therefore used in subsequent studies. The effect of MV2+ concentration, ranging from 0.01 to 0.15 mM, on the electrode behavior is illustrated in Figure 8, which (21) Wilson, R.; Turner, A. P. F. Biosens. Bioelectron. 1992, 7, 165. (22) Gorton, L.; Csoregi, E.; Dominguez, E.; Emmmneus, J.; Johnson-Petterson, G.; Marko-Varga, G.; Persson, B. Anal. Chem. 1991, 250, 203.

2638 Analytical Chemistry, Vol. 68, No. 15, August 1, 1996

Figure 8. Effect of MV2+ concentration, ranging from 0.01 to 0.15 mM, on the glucose response for (A) 5 and (B) 0.5 mM glucose at the same SWa-1/MV-GS. Conditions as in Figure 4.

shows 5 and 0.5 mM glucose responses for the same sensor. The optimum mediator concentration was found to be around 0.1 mM for 5 mM glucose (Figure 8A). When the mediator concentration was lower than 0.1 mM, the hydrogen peroxide produced during the enzymatic reaction apparently was more than MV2+ can mediate. On the other hand, at higher mediator concentrations, the large background signal renders the measurement difficult. When the glucose concentration was reduced to 0.5 mM, the measured peak current was lower by almost 10 times at the optimal mediator concentration of 0.05 mM (Figure 8B). This phenomenon again confirms the proper function of the sensor. Furthermore, it also indicates that, with a proper choice of mediator concentration, the linear range of the sensor for glucose detection can be easily varied. This point will be discussed in more detail in the following section. Analytical Characterizations of the SWa-1/MV-GS. Calibration data were obtained with the optimum experimental parameters mentioned above under aerobic conditions. Figure 9A presents cyclic voltammograms and calibration plot for the SWa-1/MV-GS with glucose concentrations of 0 (a), 2 (b), 4 (c), and 6 mM (d). In all cases, a catalytic reduction response of hydrogen peroxide was observed at a potential near -0.78 V vs Ag/AgCl. The observed reduction peak currents were then used for the construction of the calibration plot. For 0.1 mM methyl viologen, the plot shows a very linear behavior, with slope (µA/ mM), intercept (µA), and correlation coefficient of 3.7, 40.7, and 0.9993, respectively. The detection limit was 5 µM (S/N ) 3), with a linear range extending to about 6 mM, giving a dynamic range of over 3 orders of magnitude. This result is quite consistent with numerous studies showing that use of membranes is a convenient way of extending the linear range of a biosensor.23 The sensitivity started to decrease when the concentration of glucose was higher than 6 mM. An even lower detection limit could be achieved for glucose, provided that the mediator concentration was lower than 0.1 mM, as indicated in Figure 9B. For 0.05 mM methyl viologen, the plot also shows a very linear behavior, with slope (µA/mM), intercept (µA), and correlation coefficient of 1.2, 23.8, and 0.9980, respectively. The detection (23) McDonnell, M. B.; Vadgama, P. M. Ion-Sel. Electrode Rev. 1989, 11, 17.

Figure 9. Cyclic voltammograms and calibration plots in (A) 0.1 mM MV2+ with (a) 0, (b) 2, (c) 4, and (d) 6 mM glucose and (B) 0.05 mM MV2+ with (a) 0, (b) 0.05, (c) 0.07, (d) 0.09, and (e) 0.11 mM glucose at the SWa-1/MV-GS. Conditions as in Figure 4.

limit was 3 µM (S/N ) 3), with a linear range extending to about 0.11 mM. The mediator concentration exhibits a greater effect on the dynamic range of the sensor than on the sensitivity. This is consistent with the upper linearity limit, indicating the glucose concentration where the mediator is no longer able to maintain diffusion-limited substrate oxidation. The Hanes plot (not shown) provides a maximum current of 663 µA/cm2 and apparent Michaelis-Menten (Km) value of 0.028 mM for the 0.05 mM methyl viologen. The apparent Km value is much lower than those reported earlier in homogeneous solution reaction,16,24 indicating significant diffusional constraints due to the clay coatings. The clay coatings form a diffusion barrier, and substrate/product concentration profiles within this layer are, therefore, a function of both diffusion and reaction. If the diffusional effects are dominant, due to a low Km value, the limitation on the linear range becomes negligible. To characterize the reproducibility of the SWa-1/MV-GS, repetitive measurement cycles were carried out in 5 mM glucose (24) Luong, J. H. T.; Male, K. B.; Zhao, S. Anal. Biochem. 1993, 212, 269.

and 0.1 mM MV2+ solution. After the voltammogram was recorded in 5 mM glucose and 0.1 mM MV2+ solution, the electrode was then switched to a solution containing 0.1 mM MV2+ to double-check the reproducibility of the SWa-1/MV-GS before the next measurement. The results of 10 successive measurements showed relative standard deviations of 1.5% and 2.9% for glucose and MV2+, respectively. Thus, the SWa-1/MV-GS shows a good, reproducible surface and can be used for repetitive measurements. When the enzyme electrode was stored in a buffer solution (pH 7) at 4 °C, the long-term stability was examined by measuring the response to various concentrations of glucose for a period of 60 days. The results of measurements during this period showed a relative standard deviation of 1.9%, with a 0.95% decrease in signal at the end of the lifetime test. Since the activity remains virtually constant during this period of time, the result indicates excellent long-term stability of the electrode. Finally, various substances were examined concerning their interference in the determination of 0.1 mM glucose with the SWaAnalytical Chemistry, Vol. 68, No. 15, August 1, 1996

2639

1/MV-GS under aerobic conditions. The results showed that ascorbic acid (0.5 mM), uric acid (0.5 mM), oxalate (0.2 mM), galactose (0.1 mM), and cystein (0.1 mM) do not cause any observable interference in the determination of glucose. CONCLUSIONS The nontronite clay film shows good inherent chemical and physical stability and provides high electrocatalytic activity and sensitivity of electrochemical response to hydrogen peroxide. The sandwich configuration of the SWa-1/MV-GS was proved to be very effective in the determination of glucose under aerobic conditions. The method of the enzyme immobilization assures a great stability of the sensor and optimal molecular proximity between the GOx active sites and the electrode surface. The coverage of the enzyme layer with another clay film further improves the stability and antiinterference ability of the SWa-1/ MV-GS. Especially, there is no interference from ascorbic acid

2640

Analytical Chemistry, Vol. 68, No. 15, August 1, 1996

and uric acid, which are common components of biological fluids. Simplicity and long-term stability are further unique properties of this novel glucose sensor. Additional obvious applications of this electrode involve using different substrate and enzyme couples. Research along this direction is currently underway in our laboratories. ACKNOWLEDGMENT The authors gratefully acknowledge financial support from the National Science Council of the Republic of China under Grants NSC 85-2113-M-005-014. Received for review January 30, 1996. Accepted May 13, 1996.X AC960090J X

Abstract published in Advance ACS Abstracts, July 1, 1996.