- Email: [email protected]
A glucose sensor fabricated by the screen printing technique
Biosensors & Bioelectronics 10 (1995) 261-267
A glucose sensor fabricated by the screen printing technique Ryohei Nagata Central Research Institute, Dai Nippon Printing Co., Ltd., 250--1 Wakashiba, Kashiwa, Chiba 277, Japan
Kenji Yokoyama School of Materials Science, Japan Advanced Institute of Science and Technology, Hokuriku, Tatsunokuchi, Ishikawa 923-12, Japan
Susan Anne Clark & Isao Karube* Research Center for Advanced Science and Technology, Univ. of Tokyo, 4--6-1 Komaba, Meguro, Tokyo 153, Japan Tel: [81] (0)3 3481 4470 Fax: [81] (0)3 3481 4581 (Received 6 June 1994; revised 9 August 1994; accepted 16 August 1994)
Abstract: A novel glucose sensor employing ferrocene-modified glucose oxidase
is fabricated using the screen printing technique. Glucose oxidase is covalently bound to the electron mediator ferrocenecarboxylic acid in order to obtain higher enzyme activity. The ferrocene-glucose oxidase shows an increased catalytic current because the ferrocene acts as an electron transfer relay between the active centre of the enzyme and the gold electrode. Glucose sensors employing enzymes modified with ferrocene in various ways are successfully fabricated using the screen printing technique. The ink component containing the ferrocene-glucose oxidase is specially developed to be applicable to the printing machine. The printed glucose sensor chip offers a stable calibration profile and stable electrochemical properties. Keywords: sensor, ferrocene, ferrocenecarboxylic acid, glucose, glucose oxidase, printing, screen printing, metal mask, squeegee
*To whom all correspondence should be addressed.
biosensors (Matthews et al., 1987). Since the initial work by Clark & Lyons, (1%2) and Updike & Hicks, (1%7), glucose oxidase (GOD) catalyzed glucose biosensors have been actively investigated. Recent research has focused on the fabrication of the glucose sensor (Cass et al., 1984; Mascini & Selleri, 1989; Bartlett et al., 1991) and microminiaturized and intelligent sensors have been proposed (MannBuxbaum et al., 1990).
0956-5663/95/$07.00 © 1995 Elsevier Science Ltd
261
INTRODUCTION The need for biosensors has increased dramatically over the last few years (Wagner & Schmid, 1990). The detection of glucose in human blood is one of the most widespread and important applications of
R. Nagata et al. For electrodes that include immobilized enzymes and mediators, the stabilization of the mediator is important since its leakage from the electrode must be eliminated during in vivo measurements. A method has been designed that encapsulates the artificial mediator in an electrode thin film to prevent the mediator from becoming dissolved in the electrolyte (Back & Lennox, 1992). An alternative method has been proposed in which a paste, prepared by mixing the enzyme and ferrocene in paraffin, is pressed into an electrode pit (Amine et al., 1991). Turner and company proposed a unique ink-jet printing system (Newman & Turner, 1991) in which different enzymes were applied, but only the pattern or electrode base was designed by the screen printing technique (Wring et al., 1990; Bilitewski et al., 1991; Hart & Wring, 1991; Atanasov et al., 1992). In this paper, the fabrication of a ferrocenebound GOD (fc-GOD) glucose sensor using the screen printing technique is described. The screen printing technique is one of the most promising candidates for the rapid reproducible and economic production thick-film sensor chips. Its advantages include its capability in the areas of mass production and quality control, which are vital for the advent of the biosensor generation. This paper is organized as follows: (1) Modification of glucose oxidase by ferrocenecarboxylic acid (fc) is described. (2) The fc-GOD glucose sensor and electrode system is presented. (3) The development of a special ink component containing the fc-GOD enzyme and organic solvent is detailed. (4) The fabrication of a glucose sensor chip by the screen printing technique is described and measurement results are given.
MATERIALS AND METHODS
Chemicals All reagents were commercially available and of analytical-reagent grade chemicals. Glucose oxidase (type II, EC 1.1.3.4., Aspergillus niger) (Sigma Chemical Co.), ethylene glycol (Wako Chemicals Ltd.), sodium periodate (Wako), adipyl hydrazide (Tokyo Kasei Organic Chemicals [TCI]), urea (Wako), 1-ethyl-3-(3dimethylaminopropyl) carbodiimide (TCI), ferrocene carboxylic acid (Sigma), N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid (HEPES) 262
Biosensors & Bioelectronics (Wako), polyvinylpyrrolidone (PVP) (BASF), polyvinylbutyral (PVB) (Wako), n-pentanol (Wako), sodium dihydrogen phosphate dihydrate and disodium hydrogen phosphate.12H20 (Wako) were purchased, from the respective companies. All reagents were employed without further purification. Solutions of enzyme were prepared with distilled water. 0.1 M, Phosphate buffered saline (PBS), pH 6.0, was used for all measurements, unless stated otherwise.
Enzyme preparation Three types of ferrocene-modified GOD were prepared by the following procedures, which are illustrated in Figs. 1 (a), (b), and (c). The enzyme surface targets for the covalent bonding of ferrocene are symbolized to distinguish the modification processes in each procedure. In each procedure, G O D was modified with ferrocene to obtain a high sensitivity to glucose. The binding sites for ferrocene were assumed to be peripheral sugar chains and amino acid residues. Type (a), sugar chain modification Sodium periodate was dissolved in PBS. GOD (100 mg) was then dissolved in the buffered sodium periodate solution and slowly stirred at 4 °C for 1 h. Ethylene glycol (100 ~1) was added to terminate the oxidation. The mixture was slowly stirred at 4 °C for 30 min, then dialyzed over a period for 48 h or more against 0.1 M PBS, pH 6.0. During the dialysis, the external solution was exchanged at least four times with fresh solution. Adipyl hydrazide (100 mg, dried powder) was dissolved in the solution after the dialysis, and left to react at 4 °C for 12 h in the dark. Afterwards, the solution was dialyzed against PBS, the external solution again being replaced four times or more with fresh PBS, to remove the remaining unreacted hydrazide. 1ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC) (500 mg) and ferrocenecarboxylic acid (450 mg) were previously dissolved in 0.15 M HEPES buffer (25 ml) and the pH value controlled to 7.3. This ferrocene solution was added to the dialyzed enzyme solution and the mixture stirred at 4 °C for 12 h in the dark. The stirred enzyme solution was again dialyzed over a period of 48 h, during which the external solution was replaced four times or more with fresh solution, to remove unreacted ferrocene carboxylic acid and other low-molecular weight components from
Biosensors & Bioelectronics
Glucose sensor printing technique
+ I0~~ ~ C H O
(A)
CHO H2NHNCO(CH~4CONHNH~ ~ ' - N H 2 NH2 ~. ~--COOH +(CH3~2N(CH:~3N:C:NC_/2H. (EDC)
+ IO~ ~
(B)
(~CHO CHO HzNHNCO(CH~4CONHNH~,2~ ~ - - N H 2 NH2 ~'~COOH + (CH3)2N(CH~3N:C:NO2H5 I (EDC) urea
~
(c)
.COOH
(CH3)2N(CH2)'~:C:NO2H5 (EDC)
+
urea
Fig. 1. Three types of chemical modification processes for ferrocene modified glucose oxidase. Targets were: A) sugar chains, B) sugar chains and amino acid residues, C) amino acid residues.
the modified G O D , which led ferrocene introduced via its sugar chains. Type (b), sugar chain and amino acid residue modification GOD (100 mg) was dissolved in the PBS buffered sodium periodate solution and slowly stirred at 4 °C for 1 h. Ethylene glycol (100 Ixl) was added and slowly stirred with the mixture at 4 °C for 30 min to terminate the oxidation reaction, and the solution then dialyzed over a period for 48 h or more against PBS. Again the PBS was replaced at least four times with fresh PBS during dialysis. Adipyl hydrazide (100 mg, dried powder) was dissolved in the solution and left to react at 4 °C for 12 h in the dark. The solution was dialyzed using PBS, as described above,
to remove the remaining unreacted hydrazide. Urea (810 mg), EDC (500 mg) and ferrocenecarboxylic acid (450 rag) were dissolved in 0.15 M HEPES buffer (25 ml) and the pH value adjusted to 7.3. This ferrocene solution was mixed with the enzyme solution, after the dialysis described above, and then slowly stirred at 4 °C for 12 h in the dark, during which time the pH value was kept at 7.3. The stirred enzyme solution was again dialyzed using PBS to remove unreacted ferrocene carboxylic acid and other low-molecular weight components from the modified GOD, which, in this case, had ferrocene introduced via its sugar chains and lysyl residues.
263
R. Nagata et al. Type (c), amino acid residue modification GOD (100 mg) was dissolved in PBS. Urea (810 mg), EDC (500 mg) and ferrocenecarboxylic acid (450 mg), were first dissolved in 0.15 M HEPES buffer (25 ml) and the pH value adjusted to 7.3. This ferrocene solution was mixed with the enzyme solution and stirred at 4 °C for 12 h in the dark while the pH value was kept at 7.3. The stirred enzyme solution was dialyzed using PBS, as above, to remove unreacted ferrocenecarboxylic acid and other low-molecular weight reagents from the modified GOD, which had ferrocene introduced via its amino acid residues.
Biosensors & Bioelectronics patterned onto the polyimide flexible film. All parts were insulated with the same materials except for both terminals, which were coated with 150 Ixm thick gold for contact points. Each contact point was 1.0 mm wide, 10 mm long, and 2.0 mm pitch. A 10 mm-wide stripe of ink component was transferred to one side of the terminals through the metal mask using the screen printing process, and printed on the polyimide base film manually. After the ink was printed on the film, it was dried in a stream of cool air for 20 min. Once the sensor array was ready, sensor chips were cut out of the array and measured.
Instrumentation The electrochemical properties of each solutions were determined by cyclic voltammetry with potentiostat and constant potential techniques. The potentiostat was an HAB151 from Hokuto Denko Ltd. A gold disk (diameter 5 mm) working electrode (WE), platinum foil counter electrode, and silver/silver chloride (Toa Electronics Ltd. model HS-907) reference electrode (RE) were connected to the potentiostat. All measurements were performed at room temperature in pH 6.0 PBS solution, each sample being mixed with a magnetic stirrer, unless stated otherwise. The dissolved oxygen was purged by bubbling nitrogen gas into each solution for 15 min before measurement. All three electrodes were immersed in the electrochemical glass cell (3 ml) to evaluate the modified enzyme in solution. All potentials were measured with respect to the RE. The printed electrode was connected to the potentiostat, in place of the gold disk electrode (WE), with the RE and counter other two electrodes in order to test the electrode fabricated by the screen printing technique. Graphtec WX1200 and Toa Electronics EPR-100A recorders were used for the cyclic voltammogram and constant potential measurements, respectively.
Electrode preparation and printing apparatus A conventional metal mask (280 mm sq, 500 Ixm thick) and a polyurethane squeegee (sword type, 80 mm length) were utilized to transfer the ink to the base film. The polyimide film used in flexible printed wire circuits, originally designed for use in computer display, was adopted as the base film. At least 60 lines (500 Ixm wide, 180 mm long, 2.0 mm pitch) of thin copper foil were 264
RESULTS AND DISCUSSION GOD was modified to obtain a more functional enzyme for the glucose biosensor. Heller and others have reported the electrochemical properties of ferrocene bound GOD in their papers (Degani & Heller, 1987; Degani & Heller, 1988; Gregg & Heller, 1990). Badia et al., (1993) have shown that the ferrocene molecule bound to the surface of GOD acts as an electron transfer relay between the active centre of the enzyme and the electrode. The modified enzyme was shown to exhibit enhanced electron production when it reacted with the substrate. Chemical modification of GOD is illustrated in Fig. 1. GOD, ferrocene carboxylic acid, and hydrazide were the material reagents in the modification. The localization and the density of the ferrocene molecules were varied in an attempt to optimize the electrochemical properties of the enzyme. The localization of the ferrocene molecule is an important parameter for the electrochemical reaction. With regards to the molecular design, the density of ferrocene molecules was expected to determine the molecular sensitivity. For these reasons, the ferrocene binding site, the peripheral sugar chains and lysyl residues in the sequence of the GOD molecule (Frederick et al., 1990), were targeted. The peripheral sugar chain and/ or lysyl residues in GOD offered the preferred binding sites for ferrocenecarboxylic acid, in order to obtain an electroconductive enzyme. An advantageous feature in this modification process was that the reagents could be varied because ferrocenecarboxylic acid was added to the reaction solution only in the final stage. For example, the peripheral sugar chains were
Biosensors & Bioelectronics
oxidized to couple with hydrazide molecules. The ferrocenecarboxylic acid was bound to the terminal of hydrazide in the case of Fig. l(a). Amino groups of lysyl residues in the GOD sequence and the hydrazide terminals were simultaneously reacted with ferrocenecarboxylic acid in the case of Fig. l(b). Only lysyl residues in the GOD sequence were targeted by ferrocene molecules in the case of Fig. l(c). Urea was employed in the reaction to "loosen" the GOD molecule and expose the lysyl residues hidden in the inner of the enzyme to the outer water phase. When urea was added to the reaction in cases (b) and (c), ferrocenecarboxylic acid attacked the lysyl residues that were separated from water molecules. The number of ferrocene molecules bound to the GOD molecule was estimated and calculated by Inductively Coupled Plasma Atomic Emission Spectroscopy (ICP) analysis in each of the three cases. It was anticipated that the number of ferrocene molecules bound to GOD in case (b) would be the maximum among the three cases because the modification procedure allowed ferrocenecarboxylic acid to simultaneously encounter the oxidized sugar chains (hydrazide amine terminals), amino acids (lysyl residues), and the hidden lysyl residues. On the contrary, the ICP data revealed that type (a) fc-GOD contained the largest number of ferrocene molecules. The average number of ferrocene molecules was calculated to be 19.8, 14.3, and 9.2 for (a), (b), and (c), respectively. Cyclic voltammograms of the electrochemical cell for types (a), (b), and (c) are shown in Fig. 2. As expected, a response to glucose was observed for all three fc-GOD types. It was also expected that the binding location of the ferrocene molecule might play an important role in the transfer of electrons from the glucose molecules to the electrode. Catalytic current in fc-GOD types (a) and (b) showed a larger response than type (c) when glucose was added to the fc-GOD solutions. The ICP data suggested a correlation between the number of ferrocene molecules and the current increase shown in Fig. 2. The difference between fc-GOD (b) and (c) was the presence of the sugar chains. Based on these results, type (a) fc-GOD was adopted for the printing process. Type (a) fc-GOD was lyophilized to obtain a powder form, suitable for preparation of the ink. The requirements of a successful ink preparation
Glucose sensor printing technique
1.0
pAT (A)
(B)
(c)
glucose concentration (mg/ml) 0.66 / / ~ 0.33
I
•j•...•
0.66 0.33
~"
0.66 0.33
I
I 1
0
!
700 mV potential vs. Ag/AgCI
Fig. 2. Responses of catalytic current in three different types of ferrocene modified GOD solution, A) sugar chain- modified, B) sugar chain- and amino acid residues- modified, C) amino acid residue-modified. Scan rate was 2.0 mJ/sec.
included the prevention of enzyme denaturation, the correct resolution for glucose, and the correct physical properties required in the printing process. The ink consisted of only four components: modified GOD, two kinds of polymer resin as binder, and organic solvent in order to prepare the ink for the glucose sensor. Water-soluble resin, polyvinylpyrrolidone (PVP [DP ca. 9500]), and water-insoluble resin, polyvinylbutyral (PVB [DP ca. 700]), were employed as the polymer resins in the ink. The objective of the binder was to immobilize fc-GOD in the ink component, but at the same time allow glucose molecules in the water to encounter the fc-GOD, therefore a mixture of water-soluble and non-soluble polymers was used to provide a proper environment for the fc-GOD. The polymer resins were also designed for printing. The polymers were dissolved in n-pentanol. After the four components, described above, were agitated to form a homo265
R. Nagata et al.
Biosensors & Bioelectronics
geneous mixture: the ink was adjusted to an appropriate viscosity by the same solvent. Five mixing ratios of PVP and PVB in the ink were examined, and are shown in Table 1. Each ink was printed on the polyimide film and dried immediately, according to the conventional screen printing technique. The printing pattern was a 10 mm wide stripe on the array. The sensor chip was cut out of the printed array and electrically evaluated in solutions of varying glucose concentration. Five types of ink were characterized with approximate PVP:PVB ratios of 5:1, 3:1, 1:1, 1:2, and 1:3. The printed glucose sensors with the five types of ink all responded to the 0.33 and 0.66 mg/ml of glucose concentration in the 0.1 M PBS (pH 6.0). However, the (1:2) and (1:3) (= PVP:PVB) inks exhibited low sensitivity and stability. The (1:1) (= PVP:PVB) ink was the most satisfying because it showed the highest sensitivity and stability among the (5:1), (3:1) and (1:1) (= PVP:PVB) inks tested. The (1:1) ink was therefore employed to produce many printed glucose sensor arrays. A cyclic voltammogram of the printed chip with the (1:1) ink is shown in Fig. 3. The printed electrode exhibited the same response to glucose as the fc-GOD molecule in PBS (Fig. 2), proving that the printed sensor chip was also sensitive to glucose, especially as the output response showed the same current increase between 450 and 700 mV. Even though the current response at 450 mV in the cyclic voltammogram, which was derived from the oxidation of ferrocene, seemed stronger than the response at 700 mV, it was expected that the printed electrode would produce a stronger output for the higher applied constant potential. The applied potentials, 450 and 700 mV, were determined from Fig. 3. Fig. 4 shows the calibration curves obtained at applied TABLE 1 Ink recipe for printing process. ink(abbr.)
(5:1) (3:1) (1:1) (1:2) (1:3)
fc-GOD resin PVP PVB solvent (n-pentanol)
15.0 17.7 3.5 63.8
15.0 15.9 5.3 63.8
1 5 . 0 1 5 . 0 15.0 10.6 7.1 5.3 1 0 . 6 1 4 . 1 15.9 6 3 . 8 6 3 . 8 63.8
1.0 I,JA glucose concentration (mg/ml) ~
t-
0.66
0.33 0
o
I
I
0
700 mV potential
vs
Ag/AgCI
Fig. 3. Cyclic voltammograms of printed electrode including fc-GOD. Scan rate was 2.0 mJ/sec.
4
0
2
1
(700mY) !450mV)
0
0
1
2
3
4
Concentration (mg/ml) Fig. 4. Calibration profiles of printed electrode. I(450), 1(700) mean applied constant potentials were 450 m V and 700 m V vs. Ag/AgCl, respectively.
constant potentials of 450 and 700 mV. The output response was observed to be reproducible and stable for several hours. It was also demonstrated, in a preliminary experiment, that the dynamic range could be increased by a factor of approximately two by increasing the amount of the modified enzyme.
CONCLUSION mixing ratio (wt%) = (PVP:PVB) PVP; polyvinylpyrrolidone (DP. ca9500) PVB; polyvinylbutyral (DP.ca700) 266
A unique printed glucose sensor is demonstrated with an fc-GOD enzyme mixed with organic
Biosensors & Bioelectronics solvent to produce the ink component. This screen printing method provides a method for the rapid, reproducible and economic manufacture of sensor chips. The modification of the enzyme improved the response of the printed glucose sensor. An electrode of specified composition can be deposited onto a base material by a largescale printing technique such as screen printing. The screen printing technique used here enables the simple production of an enzyme sensor electrode with a stable and reproducible performance.
ACKNOWLEDGEMENT The authors thank Visiting Prof. T. Nishida for corrections and editorial comments.
REFERENCES Amine, A., Kauffmann, J.-M. & Patriarche, G.J. (1991). Amperometric biosensors for glucose based on carbon paste modified electrodes. Talanta, 38(1),107-110. Atanasov, P., Kaisheva, A., Iliev, I., Razumas, V. & Kulys, J. (1992). Glucose biosensor based on carbon black strips. Biosensors & Bioelect., 1, 361-365. Back, R. & Lennox, R.B. (1992). Electrochemical investigation of novel polymerizable thiophene/ ferrocene conjugates. Langmuir, 8, 959-964. Badia, A., Carlini, R., Fernandez, A., Battaglini, F., Mikkelsen, S.R. & English, A.M. (1993). Intramolecular electron-transfer rates in ferrocenederivatized glucose oxidase. J. Am. Chem. Soc., 115(15), 7053-7060. Bartlett, P.N., Bradford, V.Q. & Whitaker, R.G. (1991). Enzyme electrode studies of glucose oxidase modified with a redox mediator. Talanta, 38(1), 57-63. Bilitewski, U., Rueger, P. & Schmid, R.D. (1991). Glucose biosensors based on thick film technology. Biosensors & Bioelect., 6, 36%373. Cass, A.E.G., Davis, G., Francis, G.D., Hill, H.A.O., Aston, W.J., Higgins, I.J., Plotkin, E.V., Scott, L.D.L. & Turner, A.P.F. (1984). Ferrocenemediated enzyme electrode for amperometric
Glucose sensor printing technique determination of glucose. Anal. Chem, 56, 667~571. Clark, L.C, & Lyons, C. (1962). Electrode systems for continuous monitoring in cardiovascular surgery. Ann. N Y Acad. Sci., 102, 29--45. Degani, Y. & Heller, A. (1987). Direct electrical communication between chemically modified enzymes and metal electrodes.l.Electron transfer from glucose oxidase to metal electrodes via electron relays, bound covalently to the enzyme. J. Phys. Chem., 91, 1285-1289. Degani, Y. & Heller, A. (1988). Direct electrical communication between chemically modified enzymes and metal electrodes.2.Methods for bonding electron-transfer relays to glucose oxidase and D-amino-acid oxidase. J. Phys. Chem., 110, 2615-2620. Frederick, K.R., Tung, J., Emerick, R.S., Masiarz, F.R., Chamberlain, S.H., Vasavada, A., Rosenberg, S., Chakraborty, S., Schopter, L.M. & Massey, V. (1990). Glucose oxidase from aspergillus niger. J. Bio. Chem., 265(7), 3793-3802. Gregg, B.A. & Heller, A. (1990). Cross-linked redox gels containing glucose oxidase for amperometric biosensor applications. Anal. Chem., 62,258-263. Hart, J.P. & Wring, S.A. (1991). Carbon-based electrodes and their application as electrochemical sensors for selected biomolecules. Anal. Proc., 28(1), 4-7. Mann-Buxbaum, E., Pittner, F., Schalkhammer, T., Jachimowicz, A., Jobst, G., Olcaytug, F. & Urban, G. (1990). New microminiaturized glucose sensors using covalent immobilization techniques. Sensors & Actuators, B, 1, 518-522. Mascini, M. & Selleri, S. (1989). Glucose biosensor with extended linearity. Anal. Lett., 22(6), 1429-1449 Matthews, D.R., Holman, R.R., Bown, E., Steemson, J., Watson, A., Hughes, S. & Scott, D. (1987). Pen-sized digital 30-second blood glucose meter. Lancet, April 4, 778-779. Newman, J.D. & Turner, A.P.F. (1991). Ink-jet printing for the fabrication of amperometric glucose sensor. Anal. Chem. Acta., 262, 13-17. Updike, S.J. & Hicks, G.P. (1967). The enzyme electrode. Nature, 214, 986-988. Wagner, G. & Schmid, R.D. (1990). Biosensors for food analysis. Food Biotechnology, 4(1), 215-240. Wring, S.A., Hart, J.P., Bracey, L. & Birch, B.J. (1990). Development of screen-printed carbon electrodes, chemically modified with cobalt phthalocyanine, for electrochemical sensor applications. Anal. Chim. Acta., 213, 203-212.
267
A glucose sensor fabricated by the screen printing technique Ryohei Nagata Central Research Institute, Dai Nippon Printing Co., Ltd., 250--1 Wakashiba, Kashiwa, Chiba 277, Japan
Kenji Yokoyama School of Materials Science, Japan Advanced Institute of Science and Technology, Hokuriku, Tatsunokuchi, Ishikawa 923-12, Japan
Susan Anne Clark & Isao Karube* Research Center for Advanced Science and Technology, Univ. of Tokyo, 4--6-1 Komaba, Meguro, Tokyo 153, Japan Tel: [81] (0)3 3481 4470 Fax: [81] (0)3 3481 4581 (Received 6 June 1994; revised 9 August 1994; accepted 16 August 1994)
Abstract: A novel glucose sensor employing ferrocene-modified glucose oxidase
is fabricated using the screen printing technique. Glucose oxidase is covalently bound to the electron mediator ferrocenecarboxylic acid in order to obtain higher enzyme activity. The ferrocene-glucose oxidase shows an increased catalytic current because the ferrocene acts as an electron transfer relay between the active centre of the enzyme and the gold electrode. Glucose sensors employing enzymes modified with ferrocene in various ways are successfully fabricated using the screen printing technique. The ink component containing the ferrocene-glucose oxidase is specially developed to be applicable to the printing machine. The printed glucose sensor chip offers a stable calibration profile and stable electrochemical properties. Keywords: sensor, ferrocene, ferrocenecarboxylic acid, glucose, glucose oxidase, printing, screen printing, metal mask, squeegee
*To whom all correspondence should be addressed.
biosensors (Matthews et al., 1987). Since the initial work by Clark & Lyons, (1%2) and Updike & Hicks, (1%7), glucose oxidase (GOD) catalyzed glucose biosensors have been actively investigated. Recent research has focused on the fabrication of the glucose sensor (Cass et al., 1984; Mascini & Selleri, 1989; Bartlett et al., 1991) and microminiaturized and intelligent sensors have been proposed (MannBuxbaum et al., 1990).
0956-5663/95/$07.00 © 1995 Elsevier Science Ltd
261
INTRODUCTION The need for biosensors has increased dramatically over the last few years (Wagner & Schmid, 1990). The detection of glucose in human blood is one of the most widespread and important applications of
R. Nagata et al. For electrodes that include immobilized enzymes and mediators, the stabilization of the mediator is important since its leakage from the electrode must be eliminated during in vivo measurements. A method has been designed that encapsulates the artificial mediator in an electrode thin film to prevent the mediator from becoming dissolved in the electrolyte (Back & Lennox, 1992). An alternative method has been proposed in which a paste, prepared by mixing the enzyme and ferrocene in paraffin, is pressed into an electrode pit (Amine et al., 1991). Turner and company proposed a unique ink-jet printing system (Newman & Turner, 1991) in which different enzymes were applied, but only the pattern or electrode base was designed by the screen printing technique (Wring et al., 1990; Bilitewski et al., 1991; Hart & Wring, 1991; Atanasov et al., 1992). In this paper, the fabrication of a ferrocenebound GOD (fc-GOD) glucose sensor using the screen printing technique is described. The screen printing technique is one of the most promising candidates for the rapid reproducible and economic production thick-film sensor chips. Its advantages include its capability in the areas of mass production and quality control, which are vital for the advent of the biosensor generation. This paper is organized as follows: (1) Modification of glucose oxidase by ferrocenecarboxylic acid (fc) is described. (2) The fc-GOD glucose sensor and electrode system is presented. (3) The development of a special ink component containing the fc-GOD enzyme and organic solvent is detailed. (4) The fabrication of a glucose sensor chip by the screen printing technique is described and measurement results are given.
MATERIALS AND METHODS
Chemicals All reagents were commercially available and of analytical-reagent grade chemicals. Glucose oxidase (type II, EC 1.1.3.4., Aspergillus niger) (Sigma Chemical Co.), ethylene glycol (Wako Chemicals Ltd.), sodium periodate (Wako), adipyl hydrazide (Tokyo Kasei Organic Chemicals [TCI]), urea (Wako), 1-ethyl-3-(3dimethylaminopropyl) carbodiimide (TCI), ferrocene carboxylic acid (Sigma), N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid (HEPES) 262
Biosensors & Bioelectronics (Wako), polyvinylpyrrolidone (PVP) (BASF), polyvinylbutyral (PVB) (Wako), n-pentanol (Wako), sodium dihydrogen phosphate dihydrate and disodium hydrogen phosphate.12H20 (Wako) were purchased, from the respective companies. All reagents were employed without further purification. Solutions of enzyme were prepared with distilled water. 0.1 M, Phosphate buffered saline (PBS), pH 6.0, was used for all measurements, unless stated otherwise.
Enzyme preparation Three types of ferrocene-modified GOD were prepared by the following procedures, which are illustrated in Figs. 1 (a), (b), and (c). The enzyme surface targets for the covalent bonding of ferrocene are symbolized to distinguish the modification processes in each procedure. In each procedure, G O D was modified with ferrocene to obtain a high sensitivity to glucose. The binding sites for ferrocene were assumed to be peripheral sugar chains and amino acid residues. Type (a), sugar chain modification Sodium periodate was dissolved in PBS. GOD (100 mg) was then dissolved in the buffered sodium periodate solution and slowly stirred at 4 °C for 1 h. Ethylene glycol (100 ~1) was added to terminate the oxidation. The mixture was slowly stirred at 4 °C for 30 min, then dialyzed over a period for 48 h or more against 0.1 M PBS, pH 6.0. During the dialysis, the external solution was exchanged at least four times with fresh solution. Adipyl hydrazide (100 mg, dried powder) was dissolved in the solution after the dialysis, and left to react at 4 °C for 12 h in the dark. Afterwards, the solution was dialyzed against PBS, the external solution again being replaced four times or more with fresh PBS, to remove the remaining unreacted hydrazide. 1ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC) (500 mg) and ferrocenecarboxylic acid (450 mg) were previously dissolved in 0.15 M HEPES buffer (25 ml) and the pH value controlled to 7.3. This ferrocene solution was added to the dialyzed enzyme solution and the mixture stirred at 4 °C for 12 h in the dark. The stirred enzyme solution was again dialyzed over a period of 48 h, during which the external solution was replaced four times or more with fresh solution, to remove unreacted ferrocene carboxylic acid and other low-molecular weight components from
Biosensors & Bioelectronics
Glucose sensor printing technique
+ I0~~ ~ C H O
(A)
CHO H2NHNCO(CH~4CONHNH~ ~ ' - N H 2 NH2 ~. ~--COOH +(CH3~2N(CH:~3N:C:NC_/2H. (EDC)
+ IO~ ~
(B)
(~CHO CHO HzNHNCO(CH~4CONHNH~,2~ ~ - - N H 2 NH2 ~'~COOH + (CH3)2N(CH~3N:C:NO2H5 I (EDC) urea
~
(c)
.COOH
(CH3)2N(CH2)'~:C:NO2H5 (EDC)
+
urea
Fig. 1. Three types of chemical modification processes for ferrocene modified glucose oxidase. Targets were: A) sugar chains, B) sugar chains and amino acid residues, C) amino acid residues.
the modified G O D , which led ferrocene introduced via its sugar chains. Type (b), sugar chain and amino acid residue modification GOD (100 mg) was dissolved in the PBS buffered sodium periodate solution and slowly stirred at 4 °C for 1 h. Ethylene glycol (100 Ixl) was added and slowly stirred with the mixture at 4 °C for 30 min to terminate the oxidation reaction, and the solution then dialyzed over a period for 48 h or more against PBS. Again the PBS was replaced at least four times with fresh PBS during dialysis. Adipyl hydrazide (100 mg, dried powder) was dissolved in the solution and left to react at 4 °C for 12 h in the dark. The solution was dialyzed using PBS, as described above,
to remove the remaining unreacted hydrazide. Urea (810 mg), EDC (500 mg) and ferrocenecarboxylic acid (450 rag) were dissolved in 0.15 M HEPES buffer (25 ml) and the pH value adjusted to 7.3. This ferrocene solution was mixed with the enzyme solution, after the dialysis described above, and then slowly stirred at 4 °C for 12 h in the dark, during which time the pH value was kept at 7.3. The stirred enzyme solution was again dialyzed using PBS to remove unreacted ferrocene carboxylic acid and other low-molecular weight components from the modified GOD, which, in this case, had ferrocene introduced via its sugar chains and lysyl residues.
263
R. Nagata et al. Type (c), amino acid residue modification GOD (100 mg) was dissolved in PBS. Urea (810 mg), EDC (500 mg) and ferrocenecarboxylic acid (450 mg), were first dissolved in 0.15 M HEPES buffer (25 ml) and the pH value adjusted to 7.3. This ferrocene solution was mixed with the enzyme solution and stirred at 4 °C for 12 h in the dark while the pH value was kept at 7.3. The stirred enzyme solution was dialyzed using PBS, as above, to remove unreacted ferrocenecarboxylic acid and other low-molecular weight reagents from the modified GOD, which had ferrocene introduced via its amino acid residues.
Biosensors & Bioelectronics patterned onto the polyimide flexible film. All parts were insulated with the same materials except for both terminals, which were coated with 150 Ixm thick gold for contact points. Each contact point was 1.0 mm wide, 10 mm long, and 2.0 mm pitch. A 10 mm-wide stripe of ink component was transferred to one side of the terminals through the metal mask using the screen printing process, and printed on the polyimide base film manually. After the ink was printed on the film, it was dried in a stream of cool air for 20 min. Once the sensor array was ready, sensor chips were cut out of the array and measured.
Instrumentation The electrochemical properties of each solutions were determined by cyclic voltammetry with potentiostat and constant potential techniques. The potentiostat was an HAB151 from Hokuto Denko Ltd. A gold disk (diameter 5 mm) working electrode (WE), platinum foil counter electrode, and silver/silver chloride (Toa Electronics Ltd. model HS-907) reference electrode (RE) were connected to the potentiostat. All measurements were performed at room temperature in pH 6.0 PBS solution, each sample being mixed with a magnetic stirrer, unless stated otherwise. The dissolved oxygen was purged by bubbling nitrogen gas into each solution for 15 min before measurement. All three electrodes were immersed in the electrochemical glass cell (3 ml) to evaluate the modified enzyme in solution. All potentials were measured with respect to the RE. The printed electrode was connected to the potentiostat, in place of the gold disk electrode (WE), with the RE and counter other two electrodes in order to test the electrode fabricated by the screen printing technique. Graphtec WX1200 and Toa Electronics EPR-100A recorders were used for the cyclic voltammogram and constant potential measurements, respectively.
Electrode preparation and printing apparatus A conventional metal mask (280 mm sq, 500 Ixm thick) and a polyurethane squeegee (sword type, 80 mm length) were utilized to transfer the ink to the base film. The polyimide film used in flexible printed wire circuits, originally designed for use in computer display, was adopted as the base film. At least 60 lines (500 Ixm wide, 180 mm long, 2.0 mm pitch) of thin copper foil were 264
RESULTS AND DISCUSSION GOD was modified to obtain a more functional enzyme for the glucose biosensor. Heller and others have reported the electrochemical properties of ferrocene bound GOD in their papers (Degani & Heller, 1987; Degani & Heller, 1988; Gregg & Heller, 1990). Badia et al., (1993) have shown that the ferrocene molecule bound to the surface of GOD acts as an electron transfer relay between the active centre of the enzyme and the electrode. The modified enzyme was shown to exhibit enhanced electron production when it reacted with the substrate. Chemical modification of GOD is illustrated in Fig. 1. GOD, ferrocene carboxylic acid, and hydrazide were the material reagents in the modification. The localization and the density of the ferrocene molecules were varied in an attempt to optimize the electrochemical properties of the enzyme. The localization of the ferrocene molecule is an important parameter for the electrochemical reaction. With regards to the molecular design, the density of ferrocene molecules was expected to determine the molecular sensitivity. For these reasons, the ferrocene binding site, the peripheral sugar chains and lysyl residues in the sequence of the GOD molecule (Frederick et al., 1990), were targeted. The peripheral sugar chain and/ or lysyl residues in GOD offered the preferred binding sites for ferrocenecarboxylic acid, in order to obtain an electroconductive enzyme. An advantageous feature in this modification process was that the reagents could be varied because ferrocenecarboxylic acid was added to the reaction solution only in the final stage. For example, the peripheral sugar chains were
Biosensors & Bioelectronics
oxidized to couple with hydrazide molecules. The ferrocenecarboxylic acid was bound to the terminal of hydrazide in the case of Fig. l(a). Amino groups of lysyl residues in the GOD sequence and the hydrazide terminals were simultaneously reacted with ferrocenecarboxylic acid in the case of Fig. l(b). Only lysyl residues in the GOD sequence were targeted by ferrocene molecules in the case of Fig. l(c). Urea was employed in the reaction to "loosen" the GOD molecule and expose the lysyl residues hidden in the inner of the enzyme to the outer water phase. When urea was added to the reaction in cases (b) and (c), ferrocenecarboxylic acid attacked the lysyl residues that were separated from water molecules. The number of ferrocene molecules bound to the GOD molecule was estimated and calculated by Inductively Coupled Plasma Atomic Emission Spectroscopy (ICP) analysis in each of the three cases. It was anticipated that the number of ferrocene molecules bound to GOD in case (b) would be the maximum among the three cases because the modification procedure allowed ferrocenecarboxylic acid to simultaneously encounter the oxidized sugar chains (hydrazide amine terminals), amino acids (lysyl residues), and the hidden lysyl residues. On the contrary, the ICP data revealed that type (a) fc-GOD contained the largest number of ferrocene molecules. The average number of ferrocene molecules was calculated to be 19.8, 14.3, and 9.2 for (a), (b), and (c), respectively. Cyclic voltammograms of the electrochemical cell for types (a), (b), and (c) are shown in Fig. 2. As expected, a response to glucose was observed for all three fc-GOD types. It was also expected that the binding location of the ferrocene molecule might play an important role in the transfer of electrons from the glucose molecules to the electrode. Catalytic current in fc-GOD types (a) and (b) showed a larger response than type (c) when glucose was added to the fc-GOD solutions. The ICP data suggested a correlation between the number of ferrocene molecules and the current increase shown in Fig. 2. The difference between fc-GOD (b) and (c) was the presence of the sugar chains. Based on these results, type (a) fc-GOD was adopted for the printing process. Type (a) fc-GOD was lyophilized to obtain a powder form, suitable for preparation of the ink. The requirements of a successful ink preparation
Glucose sensor printing technique
1.0
pAT (A)
(B)
(c)
glucose concentration (mg/ml) 0.66 / / ~ 0.33
I
•j•...•
0.66 0.33
~"
0.66 0.33
I
I 1
0
!
700 mV potential vs. Ag/AgCI
Fig. 2. Responses of catalytic current in three different types of ferrocene modified GOD solution, A) sugar chain- modified, B) sugar chain- and amino acid residues- modified, C) amino acid residue-modified. Scan rate was 2.0 mJ/sec.
included the prevention of enzyme denaturation, the correct resolution for glucose, and the correct physical properties required in the printing process. The ink consisted of only four components: modified GOD, two kinds of polymer resin as binder, and organic solvent in order to prepare the ink for the glucose sensor. Water-soluble resin, polyvinylpyrrolidone (PVP [DP ca. 9500]), and water-insoluble resin, polyvinylbutyral (PVB [DP ca. 700]), were employed as the polymer resins in the ink. The objective of the binder was to immobilize fc-GOD in the ink component, but at the same time allow glucose molecules in the water to encounter the fc-GOD, therefore a mixture of water-soluble and non-soluble polymers was used to provide a proper environment for the fc-GOD. The polymer resins were also designed for printing. The polymers were dissolved in n-pentanol. After the four components, described above, were agitated to form a homo265
R. Nagata et al.
Biosensors & Bioelectronics
geneous mixture: the ink was adjusted to an appropriate viscosity by the same solvent. Five mixing ratios of PVP and PVB in the ink were examined, and are shown in Table 1. Each ink was printed on the polyimide film and dried immediately, according to the conventional screen printing technique. The printing pattern was a 10 mm wide stripe on the array. The sensor chip was cut out of the printed array and electrically evaluated in solutions of varying glucose concentration. Five types of ink were characterized with approximate PVP:PVB ratios of 5:1, 3:1, 1:1, 1:2, and 1:3. The printed glucose sensors with the five types of ink all responded to the 0.33 and 0.66 mg/ml of glucose concentration in the 0.1 M PBS (pH 6.0). However, the (1:2) and (1:3) (= PVP:PVB) inks exhibited low sensitivity and stability. The (1:1) (= PVP:PVB) ink was the most satisfying because it showed the highest sensitivity and stability among the (5:1), (3:1) and (1:1) (= PVP:PVB) inks tested. The (1:1) ink was therefore employed to produce many printed glucose sensor arrays. A cyclic voltammogram of the printed chip with the (1:1) ink is shown in Fig. 3. The printed electrode exhibited the same response to glucose as the fc-GOD molecule in PBS (Fig. 2), proving that the printed sensor chip was also sensitive to glucose, especially as the output response showed the same current increase between 450 and 700 mV. Even though the current response at 450 mV in the cyclic voltammogram, which was derived from the oxidation of ferrocene, seemed stronger than the response at 700 mV, it was expected that the printed electrode would produce a stronger output for the higher applied constant potential. The applied potentials, 450 and 700 mV, were determined from Fig. 3. Fig. 4 shows the calibration curves obtained at applied TABLE 1 Ink recipe for printing process. ink(abbr.)
(5:1) (3:1) (1:1) (1:2) (1:3)
fc-GOD resin PVP PVB solvent (n-pentanol)
15.0 17.7 3.5 63.8
15.0 15.9 5.3 63.8
1 5 . 0 1 5 . 0 15.0 10.6 7.1 5.3 1 0 . 6 1 4 . 1 15.9 6 3 . 8 6 3 . 8 63.8
1.0 I,JA glucose concentration (mg/ml) ~
t-
0.66
0.33 0
o
I
I
0
700 mV potential
vs
Ag/AgCI
Fig. 3. Cyclic voltammograms of printed electrode including fc-GOD. Scan rate was 2.0 mJ/sec.
4
0
2
1
(700mY) !450mV)
0
0
1
2
3
4
Concentration (mg/ml) Fig. 4. Calibration profiles of printed electrode. I(450), 1(700) mean applied constant potentials were 450 m V and 700 m V vs. Ag/AgCl, respectively.
constant potentials of 450 and 700 mV. The output response was observed to be reproducible and stable for several hours. It was also demonstrated, in a preliminary experiment, that the dynamic range could be increased by a factor of approximately two by increasing the amount of the modified enzyme.
CONCLUSION mixing ratio (wt%) = (PVP:PVB) PVP; polyvinylpyrrolidone (DP. ca9500) PVB; polyvinylbutyral (DP.ca700) 266
A unique printed glucose sensor is demonstrated with an fc-GOD enzyme mixed with organic
Biosensors & Bioelectronics solvent to produce the ink component. This screen printing method provides a method for the rapid, reproducible and economic manufacture of sensor chips. The modification of the enzyme improved the response of the printed glucose sensor. An electrode of specified composition can be deposited onto a base material by a largescale printing technique such as screen printing. The screen printing technique used here enables the simple production of an enzyme sensor electrode with a stable and reproducible performance.
ACKNOWLEDGEMENT The authors thank Visiting Prof. T. Nishida for corrections and editorial comments.
REFERENCES Amine, A., Kauffmann, J.-M. & Patriarche, G.J. (1991). Amperometric biosensors for glucose based on carbon paste modified electrodes. Talanta, 38(1),107-110. Atanasov, P., Kaisheva, A., Iliev, I., Razumas, V. & Kulys, J. (1992). Glucose biosensor based on carbon black strips. Biosensors & Bioelect., 1, 361-365. Back, R. & Lennox, R.B. (1992). Electrochemical investigation of novel polymerizable thiophene/ ferrocene conjugates. Langmuir, 8, 959-964. Badia, A., Carlini, R., Fernandez, A., Battaglini, F., Mikkelsen, S.R. & English, A.M. (1993). Intramolecular electron-transfer rates in ferrocenederivatized glucose oxidase. J. Am. Chem. Soc., 115(15), 7053-7060. Bartlett, P.N., Bradford, V.Q. & Whitaker, R.G. (1991). Enzyme electrode studies of glucose oxidase modified with a redox mediator. Talanta, 38(1), 57-63. Bilitewski, U., Rueger, P. & Schmid, R.D. (1991). Glucose biosensors based on thick film technology. Biosensors & Bioelect., 6, 36%373. Cass, A.E.G., Davis, G., Francis, G.D., Hill, H.A.O., Aston, W.J., Higgins, I.J., Plotkin, E.V., Scott, L.D.L. & Turner, A.P.F. (1984). Ferrocenemediated enzyme electrode for amperometric
Glucose sensor printing technique determination of glucose. Anal. Chem, 56, 667~571. Clark, L.C, & Lyons, C. (1962). Electrode systems for continuous monitoring in cardiovascular surgery. Ann. N Y Acad. Sci., 102, 29--45. Degani, Y. & Heller, A. (1987). Direct electrical communication between chemically modified enzymes and metal electrodes.l.Electron transfer from glucose oxidase to metal electrodes via electron relays, bound covalently to the enzyme. J. Phys. Chem., 91, 1285-1289. Degani, Y. & Heller, A. (1988). Direct electrical communication between chemically modified enzymes and metal electrodes.2.Methods for bonding electron-transfer relays to glucose oxidase and D-amino-acid oxidase. J. Phys. Chem., 110, 2615-2620. Frederick, K.R., Tung, J., Emerick, R.S., Masiarz, F.R., Chamberlain, S.H., Vasavada, A., Rosenberg, S., Chakraborty, S., Schopter, L.M. & Massey, V. (1990). Glucose oxidase from aspergillus niger. J. Bio. Chem., 265(7), 3793-3802. Gregg, B.A. & Heller, A. (1990). Cross-linked redox gels containing glucose oxidase for amperometric biosensor applications. Anal. Chem., 62,258-263. Hart, J.P. & Wring, S.A. (1991). Carbon-based electrodes and their application as electrochemical sensors for selected biomolecules. Anal. Proc., 28(1), 4-7. Mann-Buxbaum, E., Pittner, F., Schalkhammer, T., Jachimowicz, A., Jobst, G., Olcaytug, F. & Urban, G. (1990). New microminiaturized glucose sensors using covalent immobilization techniques. Sensors & Actuators, B, 1, 518-522. Mascini, M. & Selleri, S. (1989). Glucose biosensor with extended linearity. Anal. Lett., 22(6), 1429-1449 Matthews, D.R., Holman, R.R., Bown, E., Steemson, J., Watson, A., Hughes, S. & Scott, D. (1987). Pen-sized digital 30-second blood glucose meter. Lancet, April 4, 778-779. Newman, J.D. & Turner, A.P.F. (1991). Ink-jet printing for the fabrication of amperometric glucose sensor. Anal. Chem. Acta., 262, 13-17. Updike, S.J. & Hicks, G.P. (1967). The enzyme electrode. Nature, 214, 986-988. Wagner, G. & Schmid, R.D. (1990). Biosensors for food analysis. Food Biotechnology, 4(1), 215-240. Wring, S.A., Hart, J.P., Bracey, L. & Birch, B.J. (1990). Development of screen-printed carbon electrodes, chemically modified with cobalt phthalocyanine, for electrochemical sensor applications. Anal. Chim. Acta., 213, 203-212.
267