Anal. Chem. 1994,66, 1575-1577
Fractured Carbon Fiber-Based Biosensor for Glucose James W. Furbee, Jr.,t Theodore Kuwana,* and Richard S. Keliy'gt Department of Chemistry, Lake Forest College, 555 North Sheridan Road, Lake Forest, Illinois 60045, and Department of Chemistry, Malott Hall, University of Kansas, Lawrence, Kansas 66045 A microsensor for glucose was constructed using DuPont E120 fibers, which exhibit extensive fracturing upon severe anodic pretreatment. The fracturing is accompanied by an increase in the electrochemical surface area of several orders of magnitude, which is exploited here in the design of a microsensor. Following fracture, the fibers were platinized to enhance their ability to detect enzymatically producedhydrogen peroxide, and glucose oxidase was immobilized at the surface of the fiber in polypyrrole. When operated amperometrically in flow injection analysis,the treated electrodes showed a linear response to injected glucose concentration up to 10 mM, with an observed Km' near 20 mM. The sensors were found to be stable for up to 2 months when stored dry at 4 OC.
The development and characterization of biosensors are active areas of electrochemical research, with much recent effort in the miniaturization of these devices.1.2 One promising avenue toward this end involves the use of carbon fibers as a substrate for the immobilization of specific enzymes and related cofactors or mediator substance^.^" The advantages of ultramicroelectrodes in these applications are welldo~umented.~~~ It was recently reported that severe anodic pretreatment of certain carbon fibers (DuPont E120, 10-12-pm diameter) resulted in extensive fracturing along the principle axis of the fiber.9J0 This process is accompanied by an increase in the electrochemical surface area by several orders of magnitude, while the diameter increases only by 2&50%." These high surface area carbon fibers have certain characteristics that make them unique from unfractured fibers and other electrode materials for the design of biosensors. The inherent small size and enhanced current response are of particular interest both for measurements in limited volumes of solution and for potential use as direct in vivo probes. Further, the large void volume created by the fracturing may lead to an increased available internal area for immobilization of enzymes, cot Lake Forest College.
University of Kansas. (1) Abe, T.; Lau, Y. Y.; Ewing, A. G. Anal. Chem. 1992,64, 2160-2163. (2) Wang, J.; Angnes, L.Anal. Chem. 1992,64,456-459. ( 3 ) Csoregi, E.; Gorton, L.; Marko-Varga, G. Anal. Chim. Acta 1993,273, 5970. (4) Kawagoe, J. L.; Niehaus, D. E.; Wightman, R. M.Anal. Chem. 1991,63, 2961-2965. (5) Pantano, P.; Morton, T. H.; Kuhr, W. G. J. Am. Chem. Soc. 1991, 113, 1832-1833. (6) Wang, J.; Li, R.; Lin, M.4. Elecrrounalysis 1989, 1, 151-154. (7) Reynolds, E. R.; Geise, R. J.; Yacynych, A. M. In Biosensors d Chemical Sensors; Edelman, P. G., Wang, J., Eds.; ACS Symposium Series 487; American Chemical Society: Washington, DC,1992; Chapter 15. (8) Yokoyama,K.;Nakajima,K.;Uchiyama,S.;Suzuld,S.;Suzuki,M.;Takeuchi, T.; Tamiya, E.;Karube, I. E/ectrmna/ysis 1992, 4, 859-864. (9) Swain, G. M.; Kuwana, T. A n d . Chem. 1991,63, 517-519. (10) Swain, G. M.;Kuwana, T. Anal. Chem. 1992, 64, 565-568. (11) Chong, S. H.; Kuwana, T., submitted for publication in Anal. Chem. 0003-2700/94/036&1575$04.50/0 0 1994 American Chemical Socletv
factors, and mediators in applications involving a polymeric matrix. This is the first report which describes the use of fractured carbon fibers in the fabrication of a biosensor. We chose to test the viability of this approach using the well-characterized glucose/glucose oxidase system.12-ls The oxidase was immobilized at the surface of a platinized fractured fiber in a thin film of polypyrrole. Fibers treated in this way were incorporated into a flow-through cell, and the hydrogen peroxide produced in the presence of varying concentrations of glucose was detected amperometrically by flow injection analysis (FIA) .
EXPERI MENTAL SECTION Reagents and Materials. Carbon fibers (10-pm diameter) were pitch-based, high-modulus Type E l 20 (DuPont, Chattanooga, TN). Potassium nitrate, dibasic sodium phosphate, and phosphoric acid were all reagent grade. Pyrrole (99%, Aldrich, Milwaukee, WI) was freshly distilled before each use. Glucose oxidase (Type 11, from Aspergillus niger) was obtained from Sigma (St. Louis, MO) and used as received. Solutions of fl-D-glucose (Sigma, 97%) were prepared in pH 7.0 phosphate buffer and were allowed to equilibrate 24 h before use. L-Ascorbicacid was obtained from Sigma (reagent grade), and solutions were prepared in pH 7.0 phosphatebuffer. All water was HPLC grade and was filtered through a 0.2-pm membrane filter before use. Apparatus. A three-electrode system was used for all measurements. For solution work, potentials are reported vs a Ag/AgCl reference electrode (Bioanalytical Systems, West Lafayette, IN; 3 M NaC1). A platinum wire served as the auxiliary electrode. For flow injection measurements, a silver chloride-coated silver tube (99.95%, 1.6-mm id.; Goodfellow, Malvern, PA) was used as a quasi-reference electrode, and a platinum tube (99.9%, 1.6-mm i.d.; Johnson-Matthey, Ward Hill, MA) served as the auxiliary electrode. Cyclic voltammetry and chronoamperometry were performed with an EG&G PAR Model 273 potentiostat/ galvanostat running with PAR System 270 software. A Beckman Model 110Bpump was employed in the flow injection system, with a flow rate of 1.O mL/min for all measurements. A Rheodyne Model 9126 metal-free injection valve was used with a sample injection volume of 200 pL. The potential was controlled at the detector using an Omni Model 90 potentiostat (Cypress Systems, Lawrence, KS). The carrier phase was 0.1 M phosphate buffer (pH 7.0) in all cases. Data collection (12) (13) (14) (15)
Dong, S.; Kuwana, T. Electroanalysis 1991, 3, 485-491. Reynolds, E. R.; Yacynych, A. M. Elecrroanalysis 1993, 5, 405411. Wilson, R.; Turner, A. P. F. Biosens. Bioelecrron. 1992, 7, 165-185. Sun, Z.; Tachikawa, H. In ref 7, Chapter 11.
AnalyticalChemlstry, Vol. 66, No. 9,May 7, 1994
1575
was accomplished with a MacLab A/D converter (AD Instruments, Inc., Milford, MA) running in conjunction with a Macintosh Quadra 900 computer and MacLab Chart software (version 3.2.6). Data was collected at a rate of 4 points/s, saved to disk, and analyzed using the MacLab Peaks program (version 1.1.1). Electrode Preparation. Single carbon fibers were aspirated into glass capillary tubes containing an inner glass filament (diameter 2.0 mm, length 4 in.; WPI, Inc., Sarasota, FL), and the tubes pulled to a fine tip with a microelectrode puller (Narishige Model PE-2), The fiber was then cut in the middle with scissors, and the two electrodes which resulted were removed from the puller. The tip of the glass was sealed around the carbon fiber with Parafilm that had been heated to a liquid. Electrical contact was made to the fiber through a small amount of mercury inside the capillary into which a segment of conducting wire was inserted. The open end of the capillary was also sealed around the connecting wire with Parafilm to prevent mercury spillage. The protruding fiber was then cut with a scalpel to a length of 0.75 cm with the aid of a microscope. Electrodes were generally constructed in groups of 5-10. The carbon fibers were fractured in freshly prepared 0.1 M potassium nitrate solution by the application of constant potential (+2.5 Vvs Ag/AgCl) for 2 min. Capacitances were calculated before and after fracture from the current envelope obtained between440 and +0.60 V as described previously.1° Platinum microparticles were deposited onto the surface of the fractured fiber by a 204 potential step to -0.20 V in a 2 mM solution of KzPtC16 (1 .O M perchloric acid). The extent of activation of the carbon surface could be estimated from the anodic shift in the hydrogen reduction current limit in 1.0 M HzS04. Glucose oxidase was immobilized at the surface of the fractured fiber by copolymerization with pyrrole. The pyrrole (0.1 pmol/mL) was dissolved in pH 7.0 phosphate buffer by gently bubbling with nitrogen. After the pyrrole had completely dissolved, 133 unitsjmL glucose oxidase was added and dissolved with gentle stirring. The fiber was then placed in the solution and a constant potential of +0.65 V was applied for 3 min. After treatment, the electrode was stored at 4 OC until use in the fabrication of a flow cell detector. Construction of Detector. The detector was prepared from a small length (-2 cm) of Tygon tubing (l/16-in. i.d., '/*-in. 0.d.) in a fashion similar to that reported by Huiliang.16 The tubing was cut with a scalpel perpendicular to the flow axis to a depth of -'/I6 in. A 1-in. piece of PEEK tubing (0.01in. id., l/16-in. 0.d.) was then inserted into one end of the Tygon tube. The silver tubing which served as reference and the platinum tube which served as auxiliary were then attached in series with the Tygon. A small copper wire was wrapped around one end of the Tygon to allow for electrical connection to the treated fiber (vide infra). For stability, the cell components were secured to an insulated aluminum or wooden rod with transparent tape. Under a microscope, the 0.75cm-long fiber was laid across the cut in the Tygon so that 1-2 mm extended outside of the tube on both sides. The fiber was then broken at the tip of the glass capillary, and the tubing (16) Huiliang, H.; Hua, C.; Jagner, D.; Renman, L. Anal. Chim. Acta 1987,193, 61-69.
1576 AnaIyticalChemistty, vol. 66, No. 9, May 1, 1994
0
100
200
300
400
500
Time, seconds Figure 1. Flow injection peaks recorded for four injections of 10 mM glucose solutlon with fractured-fiber enzyme electrode. CondMns: flow rate, 1.O mL/min; operating potential, +0.70 Vvs Ag/AgCI; carrier phase, 0.1 M phosphate buffer, pH 7.0; injected volume, 200 pL.
sealed around the fiber using liquid Parafilm. The Parafilm was allowed to harden, and the exposed ends of the treated fiber were connected to the copper wire using colloidalgraphite (2-propanolsuspension; Ladd Research Industries, Burlington, VT).
RESULTS AND DISCUSSION The fibers used for construction of glucose sensors were subjected to a relativelymild fracture, with an observed average increase in capacitance from 3.5 f 2.3 to 2800 f 1200 pF/ cm2 ( N = 8). The platinization step resulted in a deposition of 183 f 36 pg/cm2 platinum based on the geometric area of the fiber. On average, 72 mC/cm2 (SD = 72) charge was passed during the polymerization of pyrrole, yielding a film thickness of -0.2 pm. This value for film thickness, along with the amount of glucose oxidase in the solution from which the polymer was formed, was chosen as optimal on the basis of the results of a previous report." Figure 1 shows a typical result for a series of injections of 10 mM glucose with the sensor operated in an amperometric mode at +0.70 V. The maximum response to glucoseoccurred 14 s (SD = 1) after the solution entered the cell. The average residence time in the cell was 57 s (SD = 6), with a calculated cell volume of -20 pL. The average peak area for the four injections in Figure 1 was 22.3 area units (AU) (SD = O S ) , with an RSD value of 2.4%. The total area under the peaks was calculated by summing the current values for each data point collected and multiplying the total by the distance along the time axis between consecutive points. The average peak height corresponded to 12.4 nA (SD = 0.7). A total of eight electrodes were included in this study, with a minimum limit of detection of less than 1 mM of injected (17) Belanger, D.; Nadreau, J.; Fortier, 0.Electroanalysis 1992, 4, 933-940.
50 40
-
30
-
20 10-
00
2
4
6
8
1
0
1
2
1
[Glucose], mM
0
20
40
60
80
0
100
Figure 2. Calibrationplot for injected glucose over the concentration range 1- 100 mM recordedwith fractured-fiber enzyme electrode. Inset shows only the range 1-10 mM injected glucose. Condklons same as Figure 1.
glucose. All of the electrodes showed a linear response to glucose up to at least 10 mM, with a typical calibration plot shown in Figure 2. The response to injected glucose begins to flatten out at high concentration, indicating that the electrodes are under the control of enzyme kinetics. Lowe has reported a value for the apparent Michaelis constant (Km') of 19 mM for glucose oxidase entrapped in polypyrrole, consistent with our observations.'* One set of experiments were conducted.to test whether the nonlinearity of response at higher concentrations was due to oxygen depletion in the buffer. It was observed that oxygenating the buffer did not change the response characteristics of our electrodes. The sensitivities of the electrodes to injected glucose varied considerably, ranging from 0.031 to 5.8 AU/mM, with an average value of 1.7 AU/mM (SD = 2.3). There may be several reasons for the lack of reproducibility in the response observed between electrodes. It is not uncommon for enzymes entrapped in polypyrrole to leach out over time. Further, at very positive potentials, oxidation of the polymer chain can occur, which may lead to its degradation.19 Also, the morphology of the fractured fiber is not well understood at present, and differences between electrodes following fracture may be responsible for enzyme deposition in diffusion domains that are in some cases inaccessible to injected glucose. In this preliminary study, no attempt was made to isolate the cause of the variability in response from electrode to electrode. The sensors were found to be stable for up to 2 months when stored dry at 4 OC. Generally, a maximal response to injected glucose was obtained after several hours of equilibration at open circuit in the flow stream. (18) Yon-Hin, B. F. Y.; Smolander, M.; Crompton, T.; Lowe, C. R. Anal. Chem. 1993.65, 2067-207 1. (19) Trojanowicz, M.; Matuszewski, W.; Podsiadla, M. Biosens. Bioelecfron. 1990, 5, 149-156.
4000
6000
Time, seconds
120
[Glucose], mM
2000
Figure 3. Response of fractured-flber enzyme electrode to 80 consecutive injectlons of 10 mM glucose. Condltlons same as Flgure 1.
Figure 3 shows the stability of the sensor to a series of 80 consecutive injections of 10 mM glucose. The response decreases significantly over the course of the first 30 injections, from 34.6 AU for the first injection to 27.8 AU for injection 30 (20%). Injections 61-80 yield an average peak area of 25.8 f 0.4 AU (RSD = 1.5%). The reduction in current initially may be due to the loss of enzyme not specifically bound within the polypyrrole film. The response of the electrodes to injected ascorbic acid was obtained in a single experiment to indicate the magnitude of the problem this common interferent presents. For three injections of 200 pM ascorbic acid, the response observed was 3 times greater than the response at the same electrode for 10 mM glucose.
-
CONCLUSION The fractured carbon fiber has been shown to be a promising substrate for the immobilization of enzyme in the design of a biosensor. Future work in this area will focus on the relationship between the extent of fracturing and the stability and longevity of the resulting sensor. We will also address the problems observed in the reproducibility of response between electrodes and the elimination of signals due to ascorbate and other interfering substances. ACKNOWLEDGMENT We gratefully acknowledge support for this project from the National Science Foundation (University of Kansas, MACRO-ROA Program; T.K., R.S.K.), a William and Flora Hewlett Foundation Award of Research Corp. (R.S.K.), and Lake Forest College (J.W.F.). A gift of carbon fibers from Dr. Roger Ross at DuPont, Chattanooga, TN, is also acknowledged. Received for review October 19, 1993. Accepted February 14, 1993." Abstract published in Advance A C S Abstracfs, March 15, 1994.
Analytical Chemistry, Vol. 66,No. 9, May 1, 1994
1577
