Anal. Chem. 2003, 75, 2075-2079
Carbon Nanotube/Teflon Composite Electrochemical Sensors and Biosensors Joseph Wang* and Mustafa Musameh
Department of Chemistry, New Mexico State University, Las Cruces, New Mexico 88003
The fabrication and attractive performance of carbon nanotube (CNT)/Teflon composite electrodes, based on the dispersion of CNT within a Teflon binder, are described. The resulting CNT/Teflon material brings new capabilities for electrochemical devices by combining the advantages of CNT and “bulk” composite electrodes. The electrocatalytic properties of CNT are not impaired by their association with the Teflon binder. The marked electrocatalytic activity toward hydrogen peroxide and NADH permits effective low-potential amperometric biosensing of glucose and ethanol, respectively, in connection with the incorporation of glucose oxidase and alcohol dehydrogenase/NAD+ within the three-dimensional CNT/ Teflon matrix. The accelerated electron transfer is coupled with minimization of surface fouling and surface renewability. These advantages of CNT-based composite devices are illustrated from comparison to their graphite/Teflon counterparts. The influence of the CNT loading upon the amperometric and voltammetric data, as well as the electrode resistance, is examined. SEM images offer insights into the nature of the CNT/Teflon surface. The preparation of CNT/Teflon composites overcomes a major obstacle for creating CNT-based biosensing devices and expands the scope of CNT-based electrochemical devices. Carbon nanotubes (CNT) represent an important group of nanomaterials with attractive electronic, chemical, and mechanical properties.1,2 The unique properties of carbon nanotubes make them extremely attractive for the task of chemical sensors, in general, and electrochemical detection, in particular.3 Recent studies demonstrated that CNT can impart strong electrocatalytic activity and minimization of surface fouling onto electrochemical devices. An improved electrochemical behavior of catecholamine neurotransmitters,4 cytochrome c,5 ascorbic acid,6 NADH, 7 and hydrazine compounds8 has thus been illustrated at carbon nanotube-modified electrodes. The ability of carbon nanotubes to * Corresponding author. E-mail:
[email protected]. (1) Rao, C. N. Satishkumar, B. C.; Govindaraj, A.; Nath, M. ChemPhysChem. 2001, 2, 78. (2) Baughman, R. H. Zakhidov, A.; de Heer, W. A. Science 2002, 297, 787. (3) Zhao, Q.; Gan, Z.; Zhuang, Q. Electroanalysis 2002, 14, 1609. (4) Wang, J.; Li, M.; Shi, Z.; Li, N.; Gu, Z. Electroanalysis 2002, 14, 225. (5) Wang, J.; Li, M.; Shi, Z.; Li, N.; Gu, Z. Anal. Chem. 2002, 74, 1993. (6) Wang, Z. H.; Liu, J.; Liang, Q. L.; Wang, Y. M.; Luo, G. Analyst 2002, 127, 653. (7) Musameh, M.; Wang, J.; Merkoci A.; Lin, Y. Electrochem. Commun. 2002, 4, 743. (8) Zhao, Y.; Zheng, W. D.; Chen, H.; Luo, Q. M. Talanta 2002, 58, 529. 10.1021/ac030007+ CCC: $25.00 Published on Web 03/28/2003
© 2003 American Chemical Society
promote the electron-transfer reactions of NADH and hydrogen peroxide suggests great promise for dehydrogenase- and oxidasebased amperometric biosensors. A major barrier for developing such practical CNT-based biosensing devices is the insolubility of CNT in most solvents. Previously reported CNT-modified electrodes have thus commonly relied on casting a CNT/sulfuric acid solution onto a glassy carbon surface,3 a procedure that is not compatible with the immobilization of biocomponents. New fabrication schemes are highly desired to broaden the application of CNT-based electrochemical sensors. This article reports on a new and simple avenue for preparing effective CNT-based electrochemical sensors and biosensors using CNT/Teflon composite materials. Carbon composites, based on the dispersion of graphite powder within an insulator, have received considerable attention.9,10 Such carbon composites offer convenient bulk modification for the preparation of reagentless and renewable biosensors. The use of Teflon as binder for graphite particles has shown to be extremely useful for various electrochemical sensing applications.11,12 Unlike early CNT-modified electrodes,3-8 the new composite devices rely on the use of CNT as the sole conductive component rather than utilizing it as the modifier in connection with another electrode surface. The bulk of the resulting CNT/Teflon electrodes serves as a “reservoir” of the enzyme, in a manner similar to their graphite-based counterparts. Such CNT/Teflon composites thus combine the major advantages of CNT with those of bulk composite electrode and open the door for wide-range sensing applications of CNT. For example, the CNT/Teflon composites display a marked electrocatalytic action of CNT toward hydrogen peroxide and NADH and, hence, an effective biosensing of glucose and ethanol (in connection with the corresponding oxidase and dehydrogenase enzymes). These and other attractive features of CNT/Teflon biocomposites, along with characterization of the new surfaces, are illustrated in the following sections. EXPERIMENTAL SECTION Apparatus. Amperometric experiments were performed with a Bioanalytical Systems (BAS) CV-27 voltammograph, in connection with a BAS X-Y recorder. The working electrode, the Ag/ AgCl reference electrode (model CHI111, CH Instruments, Austin, (9) Kalcher, K.; Kauffmann, J. M.; Wang, J.; Svacara, I.; Vytras, K.; Neuhold, C.; Yang, Z. Electroanalysis 1995, 7, 5. (10) Tallman, D. E.; Peterson, S. Electroanalysis 1990, 2, 499. (11) Wang, J.; Reviejo, A. J.; Angnes, L. Electroanalysis 1993, 5, 575. (12) Del Cerro, M. A.; Cayuela, G.; Reviejo, A. J.; Pingarron, J. M.; Wang, J. Electroanalysis 1997, 9, 1113.
Analytical Chemistry, Vol. 75, No. 9, May 1, 2003 2075
TX), and the platinum wire counter electrode were inserted into the 20-mL cell (BAS, model VC-2) through holes in its Teflon cover. A magnetic stirrer provided the convective transport during the amperometric measurement. The flow injection system consisted of a carrier solution reservoir, an injection valve with 250 µL of loop, interconnecting PTFE tubing, and a peristaltic pump (FIAlab, Alitea US, Medina, WA). SEM images were obtained using the Hitachi S-3200N unit. Reagents. All solutions were prepared from double-distilled water. NADH (β-nicotinamide adenine dinucleotide, reduced form), NAD+ (β-nicotinamide adenine dinucleotide, sodium salt), potassium dihydrogen phosphate, dipotassium hydrogen phosphate, glucose oxidase (GOx, EC 1.1.3.4, Type X-S from Aspergillus niger, 157 500 units/g of solid), alcohol dehydrogenease (ADH, EC 1.1.1.1 360 000 units/g of solid from baker’s yeast), and β-D(+)glucose were purchased from Sigma. Hydrogen peroxide (30 wt %) and potassium ferricyanide were purchased from Aldrich (Milwaukee, WI) The Teflon granules (type 7A; average particle size, 34 µm; density, 460 g L-1) were obtained from Dupont Inc. (Wilmington, DE), while the graphite powder (grade 38) was received from Fisher Scientific (Fair Lawn, NJ). Multiwall carbon nanotubes (MWCNT), with ∼95% purity, were obtained from NanoLab (Brighton, MA); further purification was accomplished by stirring the CNT in concentrated nitric acid at 25 °C for 24 h. Single-wall carbon nanaotubes (∼75-80% purity) were obtained from Mer Inc. (Tuscon, AZ). Electrode Preparation. CNT/Teflon composite electrodes were prepared in the dry state by hand-mixing (with a spatula) the desired amounts of the carbon nanotubes with granular Teflon for 10 min (essential for complete coverage of the granules). The composite biosensors were prepared by adding the desired amount of the enzyme (GOx or ADH) and of the NAD+ cofactor, to the 30/70 wt % CNT/Teflon composite. A portion of the resulting composite was packed firmly into the electrode cavity (2 mm diameter, 2 mm deep) of a glass sleeve. This was accomplished by multiple dipping of the glass sleeve onto the composite material (placed on a weigh paper). The electrical contact was established via a copper wire. The composite surface was smoothed on a weighing paper and rinsed carefully with double-distilled water prior to each measurement. Procedure. Measurements were carried out in a phosphate buffer (0.05 M, pH 7.4) supporting electrolyte medium. Amperometric detection proceeded under forced-convection batch and flow conditions. The desired working potential was applied, and transient currents were allowed to decay to a steady-state value. All measurements were performed at room temperature. RESULTS AND DISCUSSION The attractive behavior of the new CNT/Teflon composite electrodes has been illustrated in connection with the detection of hydrogen peroxide and NADH owing to the involvement of these compounds in a wide range of biosensing applications. Control experiments, using graphite-based Teflon composites, were used to demonstrate the various advantages of the new CNT/Teflon sensors. Figure 1 compares hydrodynamic voltammograms (HDV) for 1 mM hydrogen peroxide (A) and 1 mM NADH (B) at the graphite/Teflon (a) and CNT/Teflon (b) composite electrodes. No redox activity is observed for either analyte at the conventional graphite/Teflon electrode using 2076 Analytical Chemistry, Vol. 75, No. 9, May 1, 2003
Figure 1. Hydrodynamic voltammograms for 1 mM hydrogen peroxide (A) and 1 mM NADH (B) using the graphite/Teflon (a) and the MWCNT/Teflon (b) electrodes. Supporting electrolyte, phosphate buffer (0.05 M, pH 7.4); carbon/Teflon composition ratio, 60:40 wt %; stirring rate, ∼400 rpm.
Figure 2. Current-time recordings obtained at the graphite/Teflon (a) and MWCNT/Teflon (b) composite electrodes upon increasing the concentration of hydrogen peroxide (A) and NADH (B) in steps of 2 and 0.1 mM, respectively. Operating potential, +0.4 V; other conditions, as in Figure 1.
potentials lower than 0.6 (A) and 0.5 V (B). A small gradual increase of the response is observed at higher potentials. In contrast, the CNT/Teflon electrode responds favorably to both analytes over the entire (0.0-1.0 V) potential range. Significant oxidation and reduction currents, starting around +0.20 V, are thus observed for hydrogen peroxide. The anodic signal of NADH increases rapidly between 0.0 and 0.6 V and levels off at higher potentials. The substantial lowering of the detection potential observed for both analytes at the CNT-based composite is coupled to significantly larger current signals. Overall, the data of Figure 1 indicate that the association of CNT with the Teflon binder does not impair their strong electrocatalytic properties. Such electrocatalytic action facilitates low-potential amperometric measurements of hydrogen peroxide and NADH. Figure 2 compares the amperometric response (at +0.40 V) of the graphite/Teflon (a) and CNT/Teflon (b) electrodes to successive additions of 2 mM hydrogen peroxide (A) and 0.1 mM NADH (B). As expected from the voltammetric data, the graphite/Teflon electrode is not responsive to these concentration changes using this low-detection potential. The CNT/Teflon electrode, in contrast, responds very rapidly to the changes in the level of hydrogen peroxide and NADH, producing steady-state signals within 8-10 s. The favorable signals are accompanied by a low noise level. The hydrogen peroxide response was not affected by polishing (regenerating) the surface; such renewability was indicated from a series of six successive measurements, each recorded on a
Figure 3. Calibration plots for potassium ferricyanide using CNT/ Teflon electrodes of different compositions. MWCNT/Teflon loading ratios, 10:90 (a), 30:70 (b), 40:60 (c), 60:40 (d), and 80:20 (e). Operating potential, +0.1 V; other conditions, as in Figure 1.
Figure 4. Influence of the CNT loading upon the electrode resistance (a) and amperometric response to 1mM ferricyanide (b). Operating potential, +0.1 V; other conditions, as in Figure 1.
freshly polished surface, that yielded highly reproducible results (RSD ) 4%; not shown). Similar reproducibility was observed upon packing different surfaces from the same batch or for different batches with the same composition. The CNT content of the new composites has a profound effect upon their electrochemical behavior. Figure 3 compares calibration plots for potassium ferricyanide obtained at electrodes containing different MWCNT loadings. All electrode compositions yield highly linear calibration plots over the entire concentration range. The sensitivity increases with the CNT loading between 10 and 60 wt % and decreases thereafter. The influence of the CNT content was also examined using cyclic voltammetry (CV) experiments. Such experiments led to a well-defined peak-shaped
response for composites containing higher than 20 wt % CNT (scan rate, 10 mV s-1; 100 mM ferricyanide; not shown). The peakshaped CV response indicates overlap of the diffusion layers of adjacent CNT sites that leads to a linear diffusional flux. The CV ferricyanide cathodic and anodic peaks increased linearly with the CNT content between 30 and 70 wt %. Nearly identical peak separations (∆Ep) of ∼0.15 V were observed upon raising the CNT content between 30 and 70 wt %; a larger peak separation of 0.31 V was observed in connection with a CNT content of 80 wt %. Composites containing more than 75 wt % CNT were too dry and porous and had poor mechanical stability. A poorly defined and negligible CV response was observed on the other extreme using a 10 wt % CNT content. The above trend in the sensitivity can be attributed, in part, to the effect of the CNT content on the electrode resistance. Figure 4 shows the sensitivity-loading (b) and resistance-loading (a) profiles. As expected, “Teflon-rich” composites (containing less than 30 wt % CNT) lead to high resistance (i.e., a nearly insulating matrix) and low sensitivity. Increasing the CNT content beyond 30 wt % results in a low resistance (5-10 Ω) and higher sensitivity (except of a lower sensitivity at 80 wt % CNT). The latter reflects an operation below the mass-limiting plateau associated with the shift of the voltammetric signal (in accordance with the CV data). SEM microscopy was employed to gain insights into the nature of the new carbon/Teflon composites. Figure 5 compares the SEM images for graphite/Teflon (A) and MWCNT/Teflon (B) electrode surfaces. It can be seen that the two electrodes have distinctly different morphologies. The CNT/Teflon surface is characterized with bundles of MWCNT covering the Teflon granules. The carbon nanotube fibers have a diameter ranging from 30 to 50 nm. In contrast, distinct Teflon “mountains”, without such fibrous bundles, are observed at the graphite/Teflon electrode surface. Different morphologies are expected for different composite compositions. Similar to common carbon composite electrodes, the CNT/ Teflon composite can be used as effective detector for flowing streams. The attractive performance of such a detector was demonstrated for flow injection measurements of NADH. For example, Figure 6A displays the flow injection response for NADH solutions of increasing concentrations (0.2-1.0 mM, b-f). Well-
Figure 5. SEM images of the surfaces of the graphite/Teflon (A) and MWCNT/Teflon (B) electrodes. Accelerating voltage, 14 kV; carbon/ Teflon composition ratio, 80:20 wt %.
Analytical Chemistry, Vol. 75, No. 9, May 1, 2003
2077
Figure 6. (A) Flow injection amperometric response of the CNT/ Teflon detector to NADH solutions of increasing concentrations (0.21.0 mM, b-f), along with the response to the blank solution (a) and the resulting calibration plot (inset). (B, C) Flow injection response to 14 successive injections of a 1.0 mM NADH solution at the MWCNT/ Teflon and bare glassy-carbon electrodes, respectively. Operating potential, +0.4 V; flow rate, 1.5 mL min-1; carbon/Teflon composition ratio (A, B), 50:50 wt %. Other conditions, as in Figure 1.
defined peaks, proportional to the NADH concentration, are observed along with a low noise level. The resulting calibration plot (shown in the inset) is highly linear (slope, 4.375 µA/mM; correlation coefficient, 0.999). The CNT/Teflon composite detector greatly minimizes surface passivation effects (common to amperometric detection of NADH). Figure 6 compares the stability of the response for repetitive flow injection measurements of 1.0 mM NADH at the CNT/Teflon (B) and conventional glassy-carbon (C) detectors. The glassy-carbon detector displays a gradual decrease of the response (with a 70% decrease and a RSD of 39%; n ) 14). In contrast, a stable signal is observed over the entire operation using the CNT-coated electrode (RSD ) 1.2%). Notice also the significantly higher sensitivity of the CNT-based detector (from the 10-fold different current scales), although this comes with the cost of a slower response time (peak width of 70 vs 45 s for the GC detector). The attractive low-potential detection of hydrogen peroxide and NADH, along with the minimal surface fouling, makes the CNT extremely attractive for amperometric biosensing in connection with oxidase or dehydrogenase enzymes. The new CNT-based composites permit convenient and controlled incorporation of the desired enzyme (and its cofactor or mediator), in a manner analogous to other bulk-modified bioelectrodes.11-13 The bulk of these biosensors thus serves as a source for the biocatalytic activity. The favorable behavior of the resulting CNT/Teflon-based 2078 Analytical Chemistry, Vol. 75, No. 9, May 1, 2003
Figure 7. Current-time recordings for successive 2 mM additions of glucose at the graphite/Teflon/GOx (a) and the MWCNT/Teflon/ GOx (b) electrodes measured at +0.6 (A) and +0.1 V (B). Electrode composition, 30:69:1 wt % carbon/Teflon/GOx. Other conditions, as in Figure 1.
composites (in comparison to graphite/Teflon composites) was illustrated in connection with amperometric biosensing of glucose and ethanol through the incorporation of GOx or ADH/NAD+ within the three-dimensional electrode matrix. Figure 7 compares the amperometric response to successive additions of 2 mM glucose at the graphite/Teflon/GOx (a) and the MWCNT/Teflon/GOx (b) electrodes using operating potentials of +0.6 (A) and +0.1 (B) V, along with the resulting calibration plots (insets). Both electrodes respond to the glucose additions at +0.6 V. Yet, the CNT-based bioelectrode offers substantially larger signals reflecting the electrocatalytic activity of CNT (A, a vs b). Such electrocatalytic action is even more pronounced from comparison of the response at +0.1 V, where the conventional graphite/Teflon biosensor is not responding (B, a). Such low-potential operation of the CNT-based biosensor results in a highly linear response (over the entire 2-20 mM range) and a slower response time (∼1 min vs 25 s at +0.6 V). As expected from the HDV (of Figure 1A), higher sensitivity is observed at +0.6 V (note the different scales), along with currents in the opposite directions. The +0.6-V operation is coupled to good linearity up to 12 mM, with a slight curvature at higher levels. An analogous glucose biocomposite based on single-wall rather than multiwall CNT resulted in a more sensitive and slower response (not shown). The low-potential detection leads also to high selectivity, i.e., effective discrimination against coexisting elec(13) Pena, N.; Ruiz, G.; Reviejo, A. J.; Pingarron, J. M. Anal. Chem. 2001, 73, 1190.
additions using a low detection potential of +0.20 V (b). The response is relatively fast (∼60 s for steady state) and nonlinear. No response is observed in analogous measurements at the graphite-based biocomposite (a). The greatly enhanced biosensing of ethanol at the CNT-based device reflects the accelerated oxidation of NADH, in accordance with the HDV data of Figure 1B. Such low-potential detection of ethanol commonly requires a redox mediator (to shuttle the electrons from the NADH product to the surface).
Figure 8. Current-time recordings for successive 1 mM additions of ethanol at the graphite/Teflon/ADH/NAD+ (a) and the MWCNT/ Teflon/ADH/NAD+ (b). Operating potential, +0.2 V; electrode composition, 28.5:65:1.5:5 wt % carbon/Teflon/ADH/NAD+. Other conditions, as in Figure 1.
troactive species. Despite the absence of external (permselective) coating, the glucose response at +0.1 V was not affected by the addition of the common acetaminophen and uric acid interferences (at 0.2 mM; not shown). A similar addition of ascorbic acid resulted in a large interference, reflecting the accelerated oxidation of this compound at the CNT surface.6 Reproducible activity was observed after two weeks of dry storage at 4 °C (not shown), which is in good agreement with the high stability of enzymes in Teflon-based carbon composites.12 A sensitive low-potential detection is observed in Figure 8 for the biosensing of ethanol in connection with the coimmobilization of ADH and NAD+ within the CNT/Teflon matrix. The resulting reagentless biocomposite responds favorably to the 1 mM ethanol
CONCLUSIONS The experiments described above indicate that CNT can be used to prepare attractive composite electrodes for amperometric biosensing. The preparation of such CNT/Teflon composites overcomes a major obstacle for creating CNT-based biosensing devices. The resulting composite CNT/Teflon material brings new capabilities for electrochemical devices by combining the advantages of CNT and bulk composite electrodes. The electrocatalytic properties of CNT toward hydrogen peroxide and NADH are not impaired by their association with the Teflon binder. Various chemical and biological moieties could be readily incorporated into the bulk of the CNT/Teflon composite. While most of our data have been obtained with multiwall CNT, our preliminary results indicate similar improvements using single-wall CNT. By providing a useful avenue for preparing renewable CNT-based biosensors and CNT-modified electrodes, such fabrication scheme expands the scope of CNT-based electrochemical devices and holds great promise for routine biosensing applications. ACKNOWLEDGMENT This work was supported by grants from the U.S. EPA (Grant RD830900) National Science Foundation (Grant CHE 0209707). M.M. acknowledges a fellowship from the Islamic Development Bank (IDB) (Jeddah, SA). Received for review January 2, 2003. Accepted February 20, 2003. AC030007+
Analytical Chemistry, Vol. 75, No. 9, May 1, 2003
2079
