671

Feature Article

Electrochemical Investigation of Glucose Sensor Fabricated at Copper-Plated Screen-Printed Carbon Electrodes Annamalai Senthil Kumar and Jyh-Myng Zen* Department of Chemistry, National Chung-Hsing University, Taichung 402, Taiwan; e-mail: [email protected] Received: June 11, 2001 Final version: August 3, 2001 Abstract Copper-plated screen-printed carbon electrode (CuSPE) provides a suitable catalytic surface for the amperometric detection of hydrogen peroxide. Glucose oxidase (GOD) is immobilized on the top of the CuSPE to form a glucose sensor. The interaction of copper oxide with GOD was found to be an important factor in the glucose detection. Preliminary investigation under hydrodynamic conditions showed a linear calibration plot up to 26.7 mM glucose with a slope and regression coefficient of 4.5 mA/mM and 0.9902, respectively. The Michaelis-Menten kinetics by nonlinear curve fitting yielded a Km value close to that in solution indicating the ideality and suitability of the present system. Classical mixed potential mechanism is for the first time applied to the enzyme-coated CuSPE to further understand the system. Keywords: Copper, Screen-printed electrode, Glucose, Sensor

1. Introduction Development of inexpensive, simple, and sensitive glucose biosensors is a continuous interest. Up to now, the detection of H2O2 with natural oxygen as a mediator for glucose oxidase (GOD) is still an effective approach in glucose determination [1 ± 3]. The H2O2 detection can be performed simply on Pt, Pd, Rh, and some alloy electrodes [3, 4 ± 8] or accompanied with horseradish peroxidase (HRP) and redox mediators [9 ± 14]. The metal electrodes are relatively stable and easy to prepare as disposable screen-printed electrodes (SPEs), and indeed, Pt and Pd are most often used for this purpose [15]. The main drawbacks, however, are high cost of the materials and significant interference effect at the detection potential (i.e., > 0.5 V vs. Ag/AgCl). Recently, our group has demonstrated a simple and inexpensive H2O2 detection scheme based on a copper-plated SPE (CuSPE) [16]. Since the electrocatalytic process proceeds at an electrode potential of < 0.0 V, it is possible to eliminate the discharge of interfering species which cause erratic contribution to the response current. In this article, we describe a preliminary application of the CuSPE for glucose detection by immobilizing GOD on the electrode surface. Electron transfer from the CuSPE to GOD is essential for a successful glucose sensor. Thus, we first investigated the basic electrode configuration and transducing behaviors of the CuSPE with GOD. The CuSPE with variable oxides like I CuIIO and Cu 2 O seems to have strong interaction with GOD and this phenomenon turns out to be an advantage in fabricating the sensor. This result is not surprising as several recent studies have clearly indicated that most proteins can induce strong adsorption on metal oxide surfaces [17 ± 21]. Furthermore, a mixed potential approximation was applied Electroanalysis 2002, 14, No. 10

for the first time to pursue some mechanistic aspects about the overall reaction [22 ± 28]. A basic kinetic aspect from Michaelis-Menten mechanism (in terms of LineweaverBurk (LB), Eadie-Hofstee (EH), and Hanes plots) was used for the analysis [29 ± 31]. All these experimental results proved the ideality of this modified electrode and we anticipate that this copper-based electrode will have a big impact in glucose assays.

2. Experimental 2.1. Chemicals and Reagents GOD (EC 1.1.3.4, Type VII-S, 132 000 units/g, from Aspergillus niger), b-d-glucose and bovine serum albumin (BSA) were bought from Sigma (St. Louis, MO, USA). Glutaraldehyde (GA) and Cu(NO3) ¥ 2.5 H2O were obtained from RDH. All the other compounds used in this work were ACScertified reagent grade. Distilled, deionized water was used for preparing the standard solutions. Unless otherwise mentioned, a pH 7.0 phosphate buffer solution (PBS, I ˆ 0.1 M) was used as the supporting electrolyte. Standard glucose solution (500 mM in pH 7.0 PBS) was prepared daily and used for all the electrochemical measurements.

2.2. Apparatus and Procedure Cyclic voltammetric (CV) measurements were performed using a CHI Model 660 electrochemical workstation (Austin, TX, USA). The three-electrode system consists of either a GOD-modified CuSPE (GOD-CuSPE) or the CuSPE

¹ WILEY-VCH Verlag GmbH, 69469 Weinheim, Germany, 2002 1040-0397/02/1005-0671 $ 17.50+.50/0

672 working electrode, an Ag/AgCl reference electrode (Model RE-5, BAS), and a platinum auxiliary electrode. A potential window of 0.8 to 0.5 V with a scan rate of 100 mV/s in pH 7 PBS was applied in most voltammograms. Hydrodynamic measurements were performed using stir mode 4 in the BAS cell stand. Surface charge (Q) was calculated by integrating the voltammetric cathodic peak area (n ˆ 100 mV/s) using built in BAS program. The Q ˆ nFAGCu equation was further used to calculate the surface coverage (GCu) of respective films.

2.3. Electrode Preparation A semi-automatic screen printer was used to prepare disposable SPEs [16, 32]. The working area and the average resistance of the SPE was 0.196 cm2 and 85.64  2.10 W/cm, respectively. The SPE was pretreated by CV in 0.1 M H2SO4 at n ˆ 100 mV/s for 20 cycles. The CuSPE was then prepared by plating at 0.7 V (vs. Ag/AgCl) for 300 s in (50 mM Cu(NO3)2, 0.1 M HNO3) solution using constant-potential chronoamperometric (CA) technique. The film thickness (d) was calculated from the charges (Q*) measured during the CA experiments. As to the GOD-CuSPE, 5 mL of GOD solution was spread on the surface of CuSPE and allowed for air-dried ( 30 ± 45 min). A typical enzyme casting solution with GE ˆ 4.83 mol/cm2 (calculated from casting volume of GOD on the working CuSPE geometric surface and the GOD molecular units) was prepared by dissolving in sequence of 25 mg GOD, 21.3 mg BSA, and 25 mL 25% GA in 1 mL water. The GOD-CuSPE was stored at 5 8C after experiments.

A. S. Kumar, J.-M. Zen

3. Results and Discussion 3.1. Electrochemical and Analytical Characterization Since glucose sensors rely on the immobilization of GOD onto various to monitor the oxidation current of hydrogen peroxide liberated, two different sensor configurations were evaluated (Fig. 1). Type I electrode was the GOD-CuSPE in which GOD was coated on top of the CuSPE. As to Type II electrode, GOD was coated on an SPE followed by Cu plating (designated as Cu-GOD-SPE). Note that the color of the copper surface turned blackish due to the interaction of GOD in both configurations. Type I electrode showed a clear response with glucose under stationary and hydro-

Fig. 1. Electrode configurations of Type I and Type II glucose sensors.

Fig. 2. A) CV response of the CuSPE with and without glucose. B) Various chemically modified electrodes under static and hydrodynamic voltammetric conditions. [G] ˆ 3.74 mM. n ˆ 100 mV/s for 2 cycles. Base electrolyte was pH 7.0 PBS (I ˆ 0.1 M). Electroanalysis 2002, 14, No. 10

Glucose Sensor Fabricated at Carbon Electrodes

Fig. 3. SEM images for the A) SPE and B) CuSPE, Cu-film thickness (d) ˆ 9.42  10 8 cm; GCu ˆ 1.82  10 8 mol/cm2.

dynamic conditions; whereas, almost no response was observed for Type II electrode. This behavior is as expected since the Cu coating in Type II electrode inhibits the interaction of glucose with GOD. The electrochemical mechanistic aspects of Type I electrode was studied in detail next. Figure 2A shows the typical CV response of the CuSPE in pH 7 PBS. A diffusioncontrolled cathodic peak corresponding to the reduction of I I Cu 2 O ! Cu0 and CuIIO ! Cu 2 O at 0.3 V (C1) together with an anodic peak for the oxide formation at 0.0 V (A1) were observed [16]. The fact that no marked alteration in the CV peak was observed in the presence of glucose indicates the absence of interaction between glucose and CuSPE. This is not the case, however, for the GOD-CuSPE due to the interaction of GOD with CuSPE. The clear change in CV response is quite stable after continuous scans under hydrodynamic conditions specifying strong adhesion of GOD on the copper oxide surface. In fact, similar behavior was reported earlier for Pt electrode modified with GOD

673 and alcohol hydrogenase [19, 20]. The marked decrease of C1 after enzyme coating implies the surface interaction between enzyme and copper oxide. Parallel experiments by coating with (BSA ‡ GA) and (GOD ‡ BSA ‡ GA) on the CuSPE cause 15% and 45% decrease in the cathodic response, respectively, support the above expectation. This behavior has something to do with the formation of O , OH, and OH2 groups in the interfacial structure of the metal oxide surface [33 ± 35] and the electronegative N and O group of GOD [36]. In fact, Baldwin and co-workers have reported strong complexing properties of certain amino acids like tyrosine, tryptophan, methionine, and cysteine with copper ions due to the chelating properties of the N terminal [37]. It needs extensive physicochemical experiments in terms of XPS and in situ-FTIR and Raman studies to prove the above behavior. To optimize the electrode fabrication, the thickness of plated copper to the sensor performance was evaluated first. Different deposition time of 1, 2, 3, and 5 min was taken as a variable parameter to alter the film thickness (d). Meanwhile, CV experiments with GOD-coated bare SPE showed huge cathodic signals in the H2 gas evolution region with no glucose sensing activity. This is reasonable since GOD (i.e., FADH) contains a considerable amount of exchangeable hydrogen inside its macrostructure [36]. To avoid the exposure of bare carbon surface, compact coating of Cu was thus preferred. Typical SEM pictures, as shown in Figure 3, demonstrate the formation of a compact copperlayer. A deposition time of 5 min was taken as the optimumlayer (i.e., with 5 min deposition time). Note that the above CuSPE also showed an optimum response according to the results obtained for the oxidation of 5 mM glucose (Fig. 4A). Based on d ˆ Q* M/n FA1[38], the value of d for the 5 min deposited film was calculated as 9.42  10 8 cm (GCu ˆ 1.82  10 8 mol/cm2, respectively). In the above equation, Q*, M, and 1 (1Cu ˆ 6.0 g/cm3) are deposition charge (from CA measurements), molecular weight, and density of the plating solution, respectively. Under the optimum conditions, linear scan voltammograms obtained with increasing concentration of glucose are shown in Figure 4B. As can be seen, the anodic peak current increases as the concentration of glucose increases. The obtained voltammetric responses were very stable upon repetitive cycles. The calibration plot was linear up to 26.7 mM and reached a plateau after that in pH 7 PBS. The obtained slope of 4.5 mA/mM is higher than some Pt- and ferrocene-based glucose biosensors [39 ± 41]. The extra peak at 0.3 V observed after saturation may attribute to the glucose trapped in the interface of the enzyme layer and actually can be used to further extend the detection range of the calibration plot. Basic mechanism for the glucose oxidation is illustrated in Scheme 1. Note that both the cathodic reduction and anodic oxidation current signals were found to increase with increasing concentration of glucose except with double in magnitude for anodic than cathodic process. This behavior is different from our previous H2O2 study, where no marked variation in anodic response was observed [16]. The increase in the anodic Electroanalysis 2002, 14, No. 10

674

A. S. Kumar, J.-M. Zen

Fig. 4. A) Hydrodynamic response of GOD-CuSPE (Type I) system with different copper film thickness (insert figure) obtained by varying the Cu deposition time as 1 (a), 2 (b), 3 (c), and 5 min (d). B) Calibration plot for glucose under hydrodynamic conditions at various concentrations in pH 7 PBS. Other conditions as in Figure 1.

oxidation obviously has something to do with the coated enzyme layer.

Eadie-Hofstee (EH) expression ipa/[G] ˆ nFAkcat GE/Km

ipa/Km ˆ IEH

IEH ˆ nFAkcat GE/Km and SEH ˆ

SEH ipa

1/Km

(4) (5)

3.2. Kinetics and Mechanistic Aspects Surface saturation effect at higher concentration of glucose is a typical example of the Michaelis-Menten (MM) type of key-lock mechanism for the specific oxidation of glucose [42, 43]. In fact, a slope of 0.96 obtained from the log (ipa) vs. log ([G]) plot confirms the MM kinetics on the GODa a CuSPE. Figure 5 shows the typical responses of ipa, Q p Q po , a a a and E p E po po observed upon increasing in glucose concentration, where p and po represent with and without glucose, respectively. Basic kinetic parameters of MichaelisMenten (MM) binding constant (Km) and catalytic rate constant (kcat) were calculated by curve fitting analysis using nonlinear least square regression program based on the Marquardt-Levenberg algorithm [43]. The kinetic parameters were also obtained from the analysis of LineweaverBurke (LB), Eadie-Hofstee (EH), and Hanes plots according to the following linearized equations.

Hanes expression [G]/ipa ˆ [G]/nFAkcat GE ‡ Km/nFAkcat GE

(6)

IHn ˆ Km/nFAkcat GE and SHn ˆ 1/nFAkcat GE

(7)

S and I denote the slope and intercept of the linearized 1 equations, respectively. The i pa vs. [G] 1 (LB), ipa/[G] vs. ipa (EH), and [G]/ipa vs. [G] (Hanes) plots all showed good linearity with regression coefficients of 0.990, 0.930, and 0.990, respectively (Fig. 6). The Km and kcat values together with other kinetic results obtained are listed in Table 1. As can be seen, the MM kinetic parameters that calculated from three different methods are all in the same order. The Km ˆ 12.82 mM for the GOD-CuSPE is fairly close to that of 9.6 mM for the solution phase biocatalytic system indicating the ideality of the present system [29].

Michaelis-Menten general form ipa ˆ nFAkcat GE [G]/(Km ‡ [G])

(1)

Lineweaver-Burke (LB) expression 1/ipa ˆ Km/nFAkcat GE [G] ‡ 1/nFAkcat GE ˆ SLB [G] ‡ ILB

(2)

SLB ˆ Km/nFAkcat GE and ILB ˆ 1/nFAkcat GE

(3)

Electroanalysis 2002, 14, No. 10

3.3. Mixed Potential Approximation The interaction between the GOD-CuSPE and glucose can be visualized from the mixed potential mechanism approach. Although wide application of this analysis can be found in electroless plating, mineral flotation, and photo-

675

Glucose Sensor Fabricated at Carbon Electrodes Table 1. Mechanistic features of glucose oxidation on the GODCuSPE ( Type I ). Parameters Anodic half wave potential ( E1/2) Current function, if (ipa/n [ G]) Sensitivity Transfer coefficient (aa ) Tafel slope (ba ) Order of the reaction (m) Surface enzyme concentration, GE [a] I. Nonlinear curve fitting analysis Km kcat II. LB plot analysis Km kcat III. EH plot analysis Km kcat IV. Hanes plot analysis Km kcat

Values 0.057 V 86.28 A V 1 s mol 4.5 mA/mM 1.94 30.3 mV/decade 0.96 4.83 mol/cm2

1

cm3

12.82 mM 3.14 s 1 20.17 mM 1.42 s 1 14.4 mM 1.18 s 1 10.7 mM 1.06 s-1

[a] Calculated from the casting volume and molecular-unit of GOD.

assisted water decomposition [22, 27], no such attempt was made on the biosensing system. The potential at which two partial electrochemical half-cell reactions occur simultaneously can be represented as mixed potential (Emix) [23, 24]. The steady state mixed potential condition can provide information about the kinetics of the electrochemical reactions. The classical Michaleis-Menten approach follows the catalytic applications with desired pathway in terms of limiting concentration of the analyte at a steady state potential to get Km and kcat . In fact, Michaleis-Menten oxidation at a particular catalyst surface is also under the category of mixed potential mechanism with successive reaction pathways. The comparison between these two mechanisms is still open to further analysis and discussions. In this work, the glucose biosensing can also be regarded as two partial electrochemical half-cell reactions of FAD ‡ O2 ! FADH2 ‡ H2O2 (R2) and glucose ! gluconolactone (R1) as shown in Scheme 1A. The H2O2 formed is taken as a direct qualification entity for glucose determination [15]. In the present work, since the H2O2 formed directly cause the

I

I

formation of Cu 2 O, the normalized anodic peak of Cu 2 O (R3) was assumed as an alternative Emix parameter (i.e., Epa Epao). Note that the H2O2 reduction reaction was I proved to have major effect on the Cu0/Cu 2 O redox couple as reported in our recent study [16]. Overall, the coupled oxidation reactions of R1 and R3 (which in turn to R2) are considered to occur simultaneously for the further investigation. Power and Ritche [24, 25] model is adopted in this work for evaluating the kinetics and reaction mechanism considering the hydrodynamic potentials at limiting regions of CV as their steady state values. In their model, Emix was related to angular rotation (w) and the concentration of analyte and from the sign of variation with these parameters the pathway of the two partial electrochemical half-cell reactions can be discriminated. Since Emix has positive sign with w and the concentration of glucose in this study (Fig. 2B), R1 can thus be assigned as activation-controlled and R2 as diffusioncontrolled according to the Power and Ritche×s diagnostic criteria. The above statement is acceptable since R1 is the prim electrochemical reaction and R2 is the consecutive reaction related to H2O2 diffusion [15]. Under these conditions the simplied view of the Emix can be written as [25]: Emix ˆ E 0a' (RT/aa na F) ln ksh (RT/aa na F) ln {Za FAe} ‡ (RT/aa na F) ln {0.62 nc FDox2/3 n-1/6} 2.303 (RT/2 aa na F) pH ‡ (RT/aa na F) ln [G] ‡ (RT/2 aa na F) ln w (8) In the above equation, ksh and na represent apparent rate constant and number of electrons, respectively, in rate determining step (rds) of the anodic reaction and other symbols have their own significance. The net slope for the first derivative form of Equation 8 under constant w and solution pH can be represented as, [(@Emix/@ln [G]) ˆ RT/aa na F]w,pH 2.203 RT/aa na F]w,pH

or

[(@Emix/@log [G]) ˆ (9)

Since the normalized anodic potential (i.e., Epa Epao) is fairly close to the surface charge variation and to the ipa value (Figure 5), it is taken as an indirect Emix parameter for the glucose oxidation and for further calculation. A plot of Emix

Scheme 1. A) Schematic representation of the transducing biosensor signals on the GOD-CuSPE. B) Reaction pathway in terms of Michaelis-Menten kinetics. Electroanalysis 2002, 14, No. 10

676

A. S. Kumar, J.-M. Zen

Fig. 5. Typical plots for the normalized anodic peak current (A), surface charge (B) and potential (C) parameters against [G] based on Figure 4B.

Fig. 6. Michaelis-Menten analysis in terms of LB (A), EH (B) and Hanes (C) plots based on the data of Figure 5A. Electroanalysis 2002, 14, No. 10

677

Glucose Sensor Fabricated at Carbon Electrodes

5. Acknowledgement The authors gratefully acknowledge final support from the National Science Council of the Republic of China, Taiwan.

6. References

Fig. 7. A plot of Emix vs. log [G] on the GOD-CuSPE.

vs. log [G] yielded a slope of 30.3 mV/decade (Figure 7). This value is in comparable with mixed potential systems on corrosion (40 mV/decade) and Mo(VI) deposition on precathodized surface (43 mV/decade) [26]. By assuming na ˆ 1 in the rds, the calculated aa value is 1.94 with a Tafel slope (ba) of 30.3 mV/decade. These values are markedly different from normal reversible system with aa ˆ 0.5 and ba ˆ 120 mV/decade. Most importantly, the results indicate the asymmetric and low potential energy barrier for the activation-controlled reaction (i.e., glucose oxidation). In other words, the glucose oxidation reaction can be catalyzed by GOD and leading to the suitability of the GOD-CuSPE for biosensing. Negligible interference in the presence of ascorbic acid and uric acid by CV assures its further extension to a single use biosensing system. The electrode is stable if stored in low temperature ( 5 8C) for several days. Nevertheless, the durability of the system is still needed to be improved to extend into real applications. Further work is in progress to combine oxygen rich fluorocarbon polymer like Nafion or Tosflex to improve the stability and durability and also to work with oxygen free solution by flow injection analysis [1, 44].

4. Conclusions We have successfully demonstrated a simple and inexpensive copper-based screen-printed biosensor for glucose determination. In addition to the operation at low potentials to avoid some interfering agent, the sensitivity is also better than some Pt-based systems. The calculated MichalisMenten rate constant (Km) is fairly close to that of solution phase supporting the ideality of the proposed system towards glucose oxidation. Applicability of the mixed potential mechanism was demonstrated for the first time for the biosensing system. The calculated transfer coefficient (aa) and Tafel slope (ba) indicates the favorable catalytic oxidation of glucose at the GOD-CuSPE.

[1] J. Wang, Anal. Chem. 1999, 71, 328R and references therein. [2] J.-M. Zen, H.-H. Chung, A. Senthil Kumar, Anal. Sci. 2001, 17, i287. [3] H. Gunasingham, C. B. Tan, Analyst 1989, 114, 695. [4] Q. Chi, S. Dong, Anal. Chim. Acta 1993, 278, 17. [5] J.-J. Xu, H.-Y. Chen, Anal. Biochem. 2000, 280, 221. [6] S. Cere, M. Vazquez, S. R. de Sanchez, D. J. Schiffrin, J. Electroanal. Chem. 1999, 470, 31. [7] J. Wang, J. Liu, L. Chen, F. Lu, Anal. Chem. 1994, 66, 3600. [8] M. Somasundrum, M. Tanticharoen, K. Kirtikara, J. Electroanal. Chem. 1996, 407, 247. [9] Q. Chi, S. Dong, Anal. Chim. Acta 1995, 310, 429. [10] A. A. Karyakin, E. E. Karyakina, L. Gorton, Anal. Chem. 2000, 72, 1720. [11] N. Oyama, F. C. Anson, J. Electroanal. Chem. 1986, 199, 467. [12] J.-M. Zen, S.-H. Jeng, H.-J. Chen, J. Electroanal. Chem. 1996, 408, 157. [13] J.-M. Zen, C.-W. Lo, Anal. Chem. 1996, 68, 2635. [14] L.-T. Cai, H.-Y. Chen, Sen. Actuat. B 1999, 55, 14. [15] P. N. Barlett, J. M. Cooper, J. Electroanal. Chem. 1993, 362, 1 and references therein. [16] J.-M. Zen, H.-H. Chung, A. Senthil Kumar, Analyst 2000, 125, 1633. [17] N. R. Cabilio, S. Omanovic, S. G. Roscoe, Langmuir 2000, 16, 8480. [18] D. R. Jackson, S. Omanovic, S. G. Roscoe, Langmuir 2000, 16, 5449. [19] R. K. R. Phillips, S. Omanovic, S. G. Roscoe, Electrochem. Comm. 2000, 2, 805. [20] C.-S. Kim, S.-M. Oh, Electrochim. Acta 1996, 41, 2433. [21] S. Omanovic, S. G. Roscoe, Langmuir 1999, 15, 8315. [22] A. Senthil Kumar, Ph.D. Thesis, Allagapa College of Technology, University of Madras, India, 1998. [23] A. Mills, Chem. Soc. Rev. 1989, 18, 285. [24] G. Power, I. M. Ritche, Electrochim. Acta 1981, 26, 1073. [25] G. Power, W. P. Staunon, I. M. Ritche, Electrochim. Acta 1982, 27, 165. [26] G. Ilangovan, K. Chandrasekara Pillai, J. Electroanal. Chem. 1997, 431, 11 and references therein. [27] M. Spiro in The Physical Chemistry of Solutions (Eds: D. V. Fenby, I. D. Watson), Massey University Press, New Zealand 1983 and references therein. [28] K. C. Pillai, J. O. Bockris, J. Electrochem. Soc. 1984, 131, 568. [29] K. Kojima, T. Yamauchi, M. Shimomura, S. Miyauchi, Polymer 1998, 39, 2079 and references therein. [30] P. D. Hale, L. I. Boguslavsky, T. Inagaki, H. I. Karan, H. S. Lee, T. A. Skotheim, Y. Okamoto, Anal. Chem. 1991, 63, 677. [31] B. F. Y. Yon-Hin, M. Smolander, T. Crompton, C. R. Lowe, Anal. Chem. 1993, 65, 2067. [32] J.-M. Zen, H.-H. Chung, A. Senthil Kumar, Anal. Chim. Acta 2000, 421, 189. [33] Electrochemistry of Novel Materials (Eds: S. Trasatti in J. Lipkowski, P. N. Ross), VCH, Weinheim/New York 1995. [34] A. Senthil Kumar, K. Chandrasekara Pillai, J. Solid State Electrochem. 2000, 4, 408. [35] L. D. Burke, M. A. Murphy, J. Solid State Electrochem. 2001, 5, 43 and references therein. [36] Q. Su, J. P. Klinman, Biochemistry 1999, 38, 8572. Electroanalysis 2002, 14, No. 10

678 [37] P. Luo, F. Zhang, R. P. Baldwin, Anal. Chem. 1991, 63, 1702. [38] Y.-Y. Su, Tribology International 1997, 30, 423. [39] J.-H. Cho, M.-C. Shin, H.-S. Kim, Sen. Actuat. B 1996, 30, 137. [40] J.-C. Vidal, E. Garcia, J.-R. Castillo, Biosen. Bioelectron. 1998, 13, 371. [41] D.-M. Zhou, H.-X. Ju, H.-Y. Chen, Sen. Actuat. B 1997, 40, 89.

Electroanalysis 2002, 14, No. 10

A. S. Kumar, J.-M. Zen [42] D. Voet, J. G. Voet, Biochemistry, 2nd ed., Wiley, New York 1995. [43] M. E. G. Lyons in Advances in Chem. Phys. Polymeric Systems (Eds: I. Pyrigogine, S. A. Rice), Wiley, New York 1996, p. 297. [44] J. Wang, F. Lu, J. Am. Chem. Soc. 1998, 120, 1048.