A tubular graphite-epoxy electrode incorporating horseradish peroxidase as a potentiometric sensor for hydrogen peroxide

722

A Tubular Graphite-Epoxy Electrode Incorporating Horseradish Peroxidase as a Potentiometric Sensor for Hydrogen Peroxide Zuljikar,' D. Brynn Hihhert,*' +

and Peter W . Alexander'i

Department of Analytical Chemistry, The Univixsity of New South Wales, Sydney, Australia 2052

'+ Department of Chemistry, University of Tasmania, Launceston, Tasmania, Australia Received: July 8, 1994 Final version: September 23, 1994

Abstract A novel potentiometric peroxidase enzyme sensor based on a tubular graphite-epoxy design is reported for the determination of hydrogen peroxide in an FIA system. The sensor was formed from horseradish peroxidase (40 mg), graphite (240 mg) and epoxy (780 mg). Hydrogen peroxide, injected into p H 7,0.05 M phosphate buffer at 26" C, was determined with a linear slope of 45.6mV/decade in the concentration to 5.0 x range 7.5 x M. Acceptable reproducibility of. 1.2 9'0 relative standard deviation was obtained in the flow injection analysis. The advantages of this sensor are low cost, simple electronic circuit design and better selectivity than amperometric sensors and a detection limit comparable with other methods. Keywords: Biosensor, Graphite epoxy, Peroxidase, Hydrogen peroxide, Flow injection analysis

1. Introduction Hydrogen peroxide is used extensively in industrial processes, such as for waste water treatment, sterilization and as a source of oxygen. In analytical chemistry, especially in amperometric methods, hydrogen peroxide has been studied extensively as an electroactive substance and as an intermediate in the indirect enzymatic determination of many different substrates. In one such case, the oxidation of an analyte is catalyzed by a peroxidase at a carbon black working electrode at 600niV (vs. SCE), to produce hydrogen peroxide which is then detected amperometrically [I]. Thomas and co-workers [2] reported an amperometric enzyme sensor for detecting hydrogen peroxide. The electrodes were constructed by chemical immobilization of horseradish peroxidase and catalase on nylon net with glutaraldehyde. Sanchez et al. [ 3 ] introduced a peroxidaseferrocene-modified carbon paste electrode and also [4] described a peroxidase electrode based on the passive adsorption of the enzyme on a glassy carbon electrode. Peterson [5] built an amperometric peroxidase enzyme sensor based on graphite epoxy without mediator. Rechnitz and co-workers [6] introduced an amperometric bioelectrode for hydrogen peroxide which was constructed by mixing pineapple tissue, graphite powder and mineral oil in an appropriate ratio and Tetsuma et al. [7] reported a peroxidase-incorporated polypyrrole membrane electrode. This electrode was constructed by electropolymerizing pyrrole in aqueous potassium chloride on a tin oxide electrode in the presence of horseradish peroxidase. The peroxidase electrodes are used not only for sensing hydrogen peroxide and small organic peroxides, but also in cornbination with hydrogen peroxide producing oxidase for serving as the substrate oxidase to detect glucose, alcohols, amino acids and xanthine. In most case soluble mediators or mediator modified electrodes have been used in conjunction with these sensors [l]. A graphite-epoxy electrode was first introduced by Anderson and Julman [8] and was used in anodic stripping voltammetry to detect heavy metals. Graphite rods were impregnated by immersing them into molten waxes [9] or into styrene [lo]. Wang et al. [l I ] demonstrated a polishable and robust graphiteepoxy-modified electrode with cobalt phthalocyanine for amperometric determination of hydrazine, L-cysteine, penicillamine, and oxalic acid. Graphite epoxy was also modified by

sulfonated styrene (a cation exchanger) for detecting copper and nickel. Potentiometric enzyme sensors have been reported for many types of substrate determinations. Most of them have been built on the basis of ion selective electrodes, glass electrodes and gas sensors. Alexander and Joseph [12] demonstrated a potentiometric enzyme sensor for urea which was constructed by immobilizing urease in polyvinyl chloride on an antimony and tungsten electrode. A similar type of potentiometric biosensor based on a redox electrode has also been reported with a metal transducer [13-151. In addition, it is known that the graphite absorbs HRP in the presence of a pasting liquid such as paraffin or silicone oil providing electrochemical properties [ 11 and the presence of a filler such as silver, copper. or graphite on epoxy increases the electrical conductivity significantly [ 161. This approach was used to construct glucose and mannose sensors by measuring the change in electrical conductivity [17]. it was also observed that a conducting polymer may amperometrically determine hydrogen peroxide in the presence of horseradish peroxidase without the addition of mediating substances [7, 18-21]. In the present study, the possibility of directly monitoring hydrogen peroxide was investigated with a potentiometric peroxidase enzyme sensor based on a tubular graphite-epoxy electrode in a flow system. The peroxidase on the graphiteepoxy surface catalyzes the oxidation of hydrogen peroxide and a Nernstian response is measured at the electrode. The construction of the electrode is simple, of low cost and has good mechanical and chemical stability with possibilities f o r mass production.

2. Experimental 2.1. Sensor Design The electrode (Fig.]) was constructed by filling a perspex tube (diameter 6 mm, length 7 mm) with a slurry containing 40 mg enzyme horseradish peroxidase (EC.1.1 1.1.7), 240 mg graphite (Aldrich) and 780 mg epoxy (K219-Ciba Geigy). After insertion of a copper connector the slurry was dried for 5 min. and a flow

Enzyme Sensor for H 2 0 2

723

4 , >

. E

1

'50

50.0 loo

1

3

2

Fig. 1. Design of the tubular peroxidase enzyme electrode. 1) Perspex tube (OD 6mm); 2) immobilized horseradish peroxidase on graphiteepoxy; 3) flow through path (ID 1 mm); 4) copper wire conductor.

through channel of 1 mm diameter was drilled through the tube. The electrode was stored in a refrigerator until required.

1000

0

2000

3000

Time Is

Fig. 2. Typical potentiometric response of the sensor in the flow system. Lower peaks: graphite-epoxy electrode. Upper peaks: graphite epoxy and horseradish peroxidase for duplicate injections of hydrogen peroxide of concentration 0.25-100 pM.

hydrogen peroxide [22] 2H202

2.2. Apparatus The carrier solution was pumped by a peristaltic pump (Desaga 131900) through a water bath for temperature control then through the flow-through electrode. A platinum wire reference electrode was situated down stream in the waste collection vessel. A standard solution of hydrogen peroxide was introduced by a four-way sample loop (Rheodyne). The cell potential was monitored by a mV meter (Orion SA 720) interfaced to a microcomputer (Apple IIe) via a 12 bit analogto-digital card.

2.3. Procedure A single channel flow system was used with a carrier solution of 0.05M phosphate buffer (Na2HP04 and NaH2P04). The flow rate of the carrier was 1.2 mL min- and temperature at 26" C. A stock solution of hydrogen peroxide (25 YO, Pacific Co. Ltd., Sydney, Australia) was diluted to give working solutions of appropriate concentrations. These were standardized by dichromate titration before use. 100 pL aliquots were introduced into the carrier stream and the resulting peaks recorded. Each experiment was performed in duplicate.

'

3. Results and Discussion 3.1. Response of the Sensor to Hydrogen Peroxide Preliminary experiments were undertaken first to illustrate that incorporated horseradish peroxidase could respond to the analyte within the graphite-epoxy matrix. Experiments were performed using electrodes with and without horseradish peroxidase. Figure 2 compares the potentiometric response of both electrodes to successive injections of hydrogen peroxide working solutions. In the absence of the enzyme the graphiteepoxy gave only a small response to the hydrogen peroxide injected. In discussing the mechanism of operation of the electrode it must be determined what the electrode is sensing and whether a sufficiently fast equilibrium has been established to allow use of the Nernst equation. We develop two approaches below. Horseradish peroxidase catalyzes the decomposition of

+

2H20

+0 2

(1)

but graphite is not an active electrocatalyst for the oxygen electrode reaction.

O2 + 2H'

+ 2e + H 2 0 2

(2)

We observed no significant change in the signal when hydrogen peroxide was introduced in a carrier stream saturated with oxygen, and then with the stream purged with nitrogen. Thus molecular oxygen in the carrier does not appear to take part in the redox process that is measured at this electrode. The graphite-epoxy electrode does give a response that is very sensitive to pH below about pH 5. The reaction of Equation 1 may be written as the sum of the following redox process

+ 2H' + 2e + 2 H 2 0 H 2 0 2+ O2 + 2Hf + 2e

H202

(3) (4)

The action of horseradish peroxidase is thought to catalyze the reduction of Equation 3 via two intermediates following an initial binding of hydrogen peroxide [22]. Mediators such as aniline, mesitol or p-toluidine are employed as proton acceptors that facilitate the decomposition of the second intermediate. The electrode described here does not require mediators, which implies that the kinetics, although slower without mediators, is still sufficient to allow a measurement. Indeed it may be that a mismatch in the overall rates of Equations 3 and 4 could lead to transient changes in pH that give rise to the observed signal. Alternatively, a mixed potential may be measured reflecting the relative kinetics of the forward reactions in Equations 3 and 4. The mixed potential depends on the logarithm of the ratio of anodic and cathodic exchange current densities which in turn will contain the concentrations of hydrogen peroxide, oxygen and the pH, each raised to powers of the transfer coefficients for the reactions. It is theref~reexpected that a Nernst-like relation will be followed, i.e., that the peak height will be proportional to the logarithm of the peroxide concentration, but that the slope of the plot will depend on the particular kinetic parameters of the reactions involved. At constant pH and noting the lack of effect of molecular oxygen (5) where Act is the difference in transfer coefficients for the anodic reaction in reaction 3 and cathodic reaction of 4. As the Electroanalysis 1995, 7, No. X

724

Zu@kar et al.

Table 1. Response of three electrodes of different enzyme, graphite and epoxy content. Enzynie :graphite :epoxy (mg]

Slope (nz V decade-']

r2

40 : 240 : 780 40 : 200 : 480 40 : 280 : 800

45.6 24.7 30.6

0.999 0.985 0.996

~

concentration of hydrogen peroxide increase,sthe peaks become more negative against the reference electrode. This would be consistent with an electrode responding as described above with the oxidation of hydrogen peroxide being more sensitive to the concentration of hydrogen peroxide. The kinetics of the enzyme-catalyzed reaction were found to follow the MichaelisMenton equation with an apparent K , of 15.3 pM.

3.2. Optimization of Experimental Parameters The effect of experimental variables were investigated to optimize the composition ratio of the sensor, enzyme loading, flow rate, temperature, buffer, pH and the buffer concentration. The ratio of enzyme to epoxy to graphite was studied by comparing the response of three different preparations of the sensors. Table 1 shows the slopes and correlation coefficients of the calibration plots of each sensor. The highest sensitivity was found for the sensor containing 40 mg horseradish peroxidase, 240mg graphite and 780mg epoxy. This composition allowed the maximum diffusion of analyte through the electrode. A higher concentration of epoxy increased the hardness of the electrode and reduced the diffusion of the analyte decreasing the sensitivity. At the lower concentration of epoxy, the response also decreased as the enzyme was removed by the carrier stream. The effect of the concentration of enzyme was investigated in electrodes having one quarter of the amounts of enzyme, graphite and epoxy to conserve enzyme. The slope of the calibration plot (Fig. 3) increased with increasing amount of enzyme until a level was reached where a further increase in the amount of enzyme in the electrode produced only small changes in the response. In addition, an increase in the enzyme concentration shifted the intercept of the calibration plot to higher potentials indicating a greater activity. Similar trends have also been observed by other groups [23-241. The effect of flow rate was studied by using three different flow rates: 1.6, 1.4 and 1.2 mLmin-' . A gradual increase in the electrode response occurred as the flow rate decreased. The best

IP0r 100

-

0 1-

-7

1

The optimized parameters as described above were then used for flow injection calibration by injecting in duplicate, different concentrations of hydrogen peroxide solution between 2.5 x 1OP7-5 x w 5 M . Figure 2 shows the outputs obtained and Figure 3 a calibration plot. The linear range of the calibration graph is between 7.5 x lo-' and 5.0 x M and corresponds to the equation, E[mV] = 45.6 log[H202] 297.8, with correlation coefficient 0.999. The slope is consistent with Equation 5 and Act = 1.5. The detection limit of the determination by this electrode (baseline plus 3 4 was found to be 3.8 x lO-7M. This value obtained is greater than that by chemical immobilization [2] and comparable to the peroxidase-ferrocene-modified carbon paste electrode [4] and to the electrode with peroxidase incorporated into a polypyrrole membrane 171. The reproducibility of this method was found to be 1.2% relative standard deviation (Fig. 4).

+

3.4. Interferences Some potential interfering substances that may affect the hydrogen peroxide measurement include cations (Fe2+, Fe3+, K+, Na') and anions (NO:, NO,, C1-, Br-, F-. I-, SO:-) which may alter the pH or inhibit the enzymatic reaction, and oxidizable compounds such as ascorbic acid, uric acid and lactic acid and proteins (bilirubin and albumin) that may bind to the electrode. Many of these substances are expected to be present M in a real sample. The data obtained from the injection of hydrogen peroxide, alone, and in the presence of interfering substances of various concentrations (lop6, and M) shows that none of the cations and anions mentioned above gave significant interferences. Of the electroxidizable

-

-5 -4 Log(Hydrogen Peroxide)

-3

Fig 3 The cdlibration of the sensor for different amounts of horseradish peroxidase 4 5 7 w t % , 0 3 8 w t % (40mg), H 3 8 w t % (lOmg), 0 1 9 wt%. and 0 none E~ectroanal.~sis. 1995, 7, No. 8

3.3. Electrode Calibration, Limit of Detection and Reproducibility

-

1

-6

response with the widest linear calibration range was at 1.2mL min-' . This flow rate was then used for further studies. The optimum pH was established by comparing the response of the sensor carrier streams of different pH. The optimum pH was 7.0 which agrees with the findings of Thomas and co-workers for this enzyme [2]. The concentration of phosphate buffer was varied between 0.001 and 0.1 M with an optimum response at 0.05 M. A plateau in the temperature dependence of the system was observed between 24 to 28 "C. Therefore a temperature of 26 "C was adopted for further studies.

01 0

.L--

~~

500

1000

-

1500

Time Is

Fig. 4. Reproducibility study. Analysis of nine aliquots containing 10- M hydrogen peroxide.

Enzyme Sensor for H 2 0 2

725

Table 2. Recovery studies of hydrogen peroxide from different matrices., Matrix sample

Peroxide Recovered injected [ p M ] [ p M ]

Normal serum 25.0 Elevated serum 7.5 Orange juice 25.0 7.5 Coca Cola

26.0 7.8 22.6 7.2

RSD [%] 3.15 3.33 0.75 0.98

% Recovery

104 104 90.4 96

substances only ascorbic acid at the high concentration of M increased the peak height significantly (12 %). The effect of ascorbic acid in this experiment was better than in the aniperometric method [23] and can be neglected at low concentration. An interesting result is of the effect of proteins on the response. The presence of albumin and bilirubin in the carrier stream shifted the base line, relatively reduced the response and gave a longer response time. Adsorption of large protein molecules would serve to hinder the action of the enzyme. cured this problem with Triton X at a concentration of an accompanying small base line shift.

3.5. Recovery Study of Hydrogen Peroxide in Matrix Sample The lack of major interferences implies that the peroxidase sensor could be used to determine hydrogen peroxide in matrix samples such as soft drink and serum. However, for the best accuracy and its precision, it is necessary to calibrate the sensor using hydrogen peroxide standard solutions containing the matrix. Recovery studies in orange juice, Coca Cola and normal serum demonstrated the accuracy of this method for the determination of hydrogen peroxide in consumer products and biological substances in a matrix containing high concentration of protein (Table 2). Before analysis a ten-fold dilution of the matrix sample was required using phosphate buffer at pH 7.0. The sensor was calibrated using hydrogen peroxide standard solutions containing the appropriate interferents and recovery were carried out with six replicates for each matrix. The results showed a good recovery in all matrices with the exception of orange juice (90.4 % recovery). The error in this experiment could be caused by reaction of the matrix with hydrogen peroxide thus reducing the response.

3.6. Life of the Sensor The life of a peroxidase enzyme sensor based on a tubular graphite-epoxy electrode is up to five months for intermittent use. This condition can be achieved by storage at 4°C in a refrigerator after rinsing in the buffer solution at pH 7.0 for 15min. The life of this sensor obtained is better than that reported by Sanchez et al. [4] and agrees well with the life of electrodes prepared by chemical immobilization [2].

4. Conclusion A new technique is reported for the construction of a potentiometric sensor by means of physical immobilization of horseradish peroxidase in graphite-epoxy. The potential of the sensor has a logarithmic relation to concentration in the range to 5.0 x 10-5M corresponding to the equation 7.5 x E[mV] = 45.6 log[H,O,] + 297.8. The detection limit of this sensor was 3.8 x 10-7M which is comparable with other methods. The sensor is selective and has a long operational life (five months). Only ascorbic acid at high concentration can interfere with the measurement of the analyte. This potentiometric peroxidase enzyme sensor could be developed as a bilayer enzyme electrode by coupling, for example, with glucose oxidase or amino acid oxidase for indirect determination of glucose or amino acids.

5. Acknowledgement The authors are grateful to the Indonesian government for a fellowship in support of Zulfikar.

6. References [l] L. Gorton, G.J. Peterson, E. Csoregi, K. Johansen, E. Dominguez, G.M. Varga, Analyst 1992, 117, 1235. [2] M. Cosgrove, G.J. Moddy and J.D.R. Thomas, Analyst 1988,113, 181 1. [3] P.D. Sanchez, P.T. Blanco, J.M.F. Alfarez, R. O’Kennedy, R. Smyth, Electroanalysis 1990, 2, 303. [4] P.D. Sanchez, J.A.M. Ordeires, C. Garcia, P.T. Blanco, Eleclroanalysis 1991, 3, 281. 151 G.J. Peterson, Electroanalysis 1991, 3, 741. [6] M.S. Lin, S.Y. Tham, G.A. Rechnitz, Elecfroanulysis 1990,2, 51 1 . [7] T. Tetsuma, M. Gondaira, T. Watanabe, Anal. Chem. 1992,64, 1183. [8] J. Anderson, D.E. Talman, Anal. Chem. 1976, 48, 209. [9] R.G. Clem, G. Litton, L.D. Ornellas, Anal. Chem. 1976, 46, 1307. [lo] R.G. Clem, A.F. Sciamanna, Anal. Chem. 1975,47, 1778. [l I] J. Wang, T. Golden, K Varghuse, I. El-Rayes, Anal. Chem. 1989,61, 508. [12] P.W. Alexander, J.P. Joseph, Anal. Chim. Acta. 1981, 131, 103. [13] J.F. Castner, L.B. Wingard Jr., Anal. Chem. 1984,. 56, 2891. [I41 A.K. Chen, J.A. Starzman, C.C. Liu, Biotechnol. Bioeng. 1982,24,971. [15] L.B. Wingard Jr., L.A. Cantin, J.F. Castner, Biochim. Biophys. Actu 1983, 748, 21. [16] B. EIPs, Chemistry and Technology of Epoxy Resin, Blackie. Glasgow 1993, p. 124. [17] D.T Hoa, T.N.S. Kumar, N.S. Punekar, R.S. Srinivasa, R. Lal, A.Q. Contractor, Anal. Chem. 1992,64, 2645. [18] S. Yabuki, H. Shinohara, M. Aizawa, J. Chem. SOC.Chem. Cornmun. 1989,945. [19] R.M. Paddock, E.F. Bowden, J. Electrounal. Chem. 1989,260, 487. [20] J.J. Kulys, R.D. Schmid, Bioelectrochem. Bioenerg. 1990, 24, 487. [21] U. Wollenberger, V. Bogdanovskaya, S . Borin, F. Scheler, M. Tarasevich, Anal. Lett. 1990, 23, 1795. [22] P. George, J. Biochem. 1953, 54, 267. [23] M. Mascini, A. Liberti, Anal. Chim. Acta. 1974, 68, 677. [24] P.W. Carr, Anal. Chem. 1977, 49, 799.

Electroanalysis 1995, 7, No. 8