Optimized carbon nanotube fiber microelectrodes as potential analytical tools

Anal Bioanal Chem (2007) 389:499–505 DOI 10.1007/s00216-007-1467-9

ORIGINAL PAPER

Optimized carbon nanotube fiber microelectrodes as potential analytical tools Lucie Viry & Alain Derré & Patrick Garrigue & Neso Sojic & Philippe Poulin & Alexander Kuhn

Received: 6 June 2007 / Revised: 21 June 2007 / Accepted: 22 June 2007 / Published online: 25 July 2007 # Springer-Verlag 2007

Abstract The preparation and interesting electrochemical properties of carbon nanotube (CNT) fiber microelectrodes are reported. By combining the advantages of CNT with those of fiber electrodes, this type of microelectrode differs from CNT-modified or CNT-containing composite electrodes, because they are made solely of CNT without other components, for example additives or binders. The performance of these electrodes has been characterized with regard to, among others, the electrocatalytic oxidation of analytes via dehydrogenase-mediated reactions. In this context the reversible regeneration of the coenzyme NAD+ using a mediator is a key step in the development of new amperometric sensor devices and we have successfully immobilized mediator molecules that are very efficient for this purpose on the surface of the CNT fiber electrode. The microelectrodes thus obtained have been compared with classic carbon microelectrodes and have promising behavior in biosensing applications, especially after specific pretreatments such as CNT alignment inside the fiber or expansion of the specific surface by chemically induced swelling. Keywords Carbon nanotubes . Microelectrodes . Electrocatalysis . NADH oxidation . Glucose sensor Electronic supplementary material The online version of this article (doi:10.1007/s00216-007-1467-9) contains supplementary material, which is available to authorized users. L. Viry : A. Derré : P. Poulin Centre de Recherche Paul Pascal, CNRS, 115 Av. A. Schweiter, 33600 Pessac, France L. Viry : P. Garrigue : N. Sojic : A. Kuhn (*) Université Bordeaux 1, CNRS, ISM, ENSCPB, 16 Av. Pey Berland, 33607 Pessac, France e-mail: [email protected]

Introduction For many years there has been much interest in developing new types of electrode as biosensing tools. Several forms of carbon that are suitable for electroanalytical applications are available. Among others, glassy carbon and carbon paste are the most popular carbon electrode materials. The recent discovery of carbon nanotubes (CNT) [1, 2] has attracted much attention because of their dimensions, their high specific surface area, and the electronic properties which arise because of their curved-structure [3]. The electronic properties of CNT suggest that they might have the ability to efficiently mediate electron-transfer reactions with electroactive species in solution when used as electrodes [4]. A variety of publications have reported on the electrochemistry at CNT film-modified electrodes and CNT paste electrodes, for example with regard to the catalytic conversion of important biomolecules [4–8]. The electrodes employed were composed of randomly distributed tubes with no control over the alignment of the nanotubes inside CNT films or CNT composites. Moreover, binders and additives are usually present in these CNT electrode materials. To benefit fully from the attractive properties of the CNT and to better understand their influence on physicochemical behavior it would be preferable to study electrodes composed solely of CNT. Wang et al. recently used such electrodes for the first time for electrooxidation of nicotinamide adenine dinucleotide (NADH) and promising features have been demonstrated [9]. Even more recently, needle-like or forest-like CNT electrodes [10] have been successfully designed with better electrochemical properties than bare metal electrodes. A vertically aligned CNT array designed for neural stimulation [11] has been demonstrated as being safer and more reliable than traditional approaches.

500

Here, we report the fabrication, the general electrochemical characterization, the surface modification for biosensor applications, and the different ways of performance improvement of CNT fiber microelectrodes (CNTFMs). Microelectrodes have found widespread use for exploring microscopic domains and measuring local concentration profiles [12]. Using CNT fibers developed by Vigolo et al. [13], the alignment of CNTs inside the fiber can be controlled [14] and a 100% CNT electrode material is obtained. A procedure for CNTFM elaboration has been established leading reproducibly to microelectrodes of high quality. The performance of these electrodes has been characterized in comparison to classical carbon fiber microelectrodes (CFM), in the electrocatalytic oxidation of analytes via dehydrogenase mediated reactions. In this context the reversible regeneration of the coenzyme NAD+ using a mediator is a key step in the development of new amperometric sensor devices. We have immobilized mediator molecules that are very efficient for this purpose [15] on the surface of the CNTFM, leading finally to a miniaturized biosensor with promising and more attractive properties than classic CFMs. The active electrode surface can be easily regenerated and, in contrast to CFMs, the performance of CNTFM can be controlled and improved by several pretreatments such as CNT alignment inside the fiber and chemical enhancement of the active surface area. Experimental Apparatus Cyclic voltammetry (CV) experiments were carried out in a conventional one-compartment cell with an Autolab PGSTAT 10 potentiostat at ambient temperature (20±2 °C). Potentials were measured with respect to a commercial Ag∣AgCl∣3 mol L−1 KCl reference electrode (BAS) and the counter-electrode was a platinum wire. If not otherwise mentioned, scans were started at the positive end of the potential range for the study of the adsorbed species and at the negative end of the potential range for catalysis experiments.

Anal Bioanal Chem (2007) 389:499–505

prepared from ultra-pure water obtained from a purification train (Milli-Q Plus 185, Millipore). Elicarb carbon nanotubes (CNTs) were obtained from Thomas Swan. Cold solidification resin Araldite CY230 + hardener HY956 from ESCIL-France was used for the preparation of the microelectrodes. Procedures Fiber synthesis We spin fibers using a continuous method following a procedure described in the literature [13]. Briefly, CNTs (0.3 wt %) are dispersed in water using sodium dodecyl sulfate (SDS) as the dispersant (1 wt %). The dispersions are homogenized by sonication. The fibers are spun continuously by injection of the homogenous CNT dispersion in the coflowing stream of an aqueous solution of poly (vinyl alcohol) (PVA; 5 wt %, MW 150,000). The resulting fibers contain an equal weight fraction of PVA and CNTs. Subsequently the fibers are dried and then heat-treated at 600 °C for 6 h under argon atmosphere to fully eliminate the PVA. Fibers are electrically conductive and composed exclusively of CNTs. Classical carbon fibers for control experiments and comparison were purchased from Goodfellow. Electrode fabrication Pieces of annealed CNT fibers 2 cm long were inserted into glass capillaries of the same length then one end was dipped in an epoxy resin bath. Before the resin fills the capillaries, a copper wire is inserted with silver paint to connect the fiber. The resin coats and seals the fiber entirely. At this stage, the capillary section is polished to release the section of the fiber which will be the active surface of the CNTFM (Fig. 1). Classical carbon fiber microelectrodes (CFM) have been made by following the same fabrication procedure. Surface modification

Chemicals The mediator 2,4,7-trinitro-9-fluorenone and phosphomolybdic acid (Aldrich) was used as received. The reduced and oxidized form of β-nicotinamide adenine dinucleotide was purchased with 98% purity (Sigma). Glucose dehydrogenase (GDH, EC 1.1.1.47; Jülich Fine Chemicals) had an activity of 77.8 units mg−1. Tris buffer 0.1 mol L−1 was prepared by dissolving the adequate amount of compound (Merck) and adjusting the pH to 8 by addition of HNO3. All experiments were carried out at this pH, because it gives the best results in terms of catalysis. Solutions were

Before each experiment CNTFMs were polished with a diamond buffing wheel, rinsed thoroughly and then sonicated for 1 min in ultrapure water. The electrode was dried in an air stream before modification. After dipping for 20 min in a 6 mmol L−1 solution of the mediator in tetrahydrofuran (THF), the electrode was rinsed with water. THF being miscible with water, this step allowed the solvent to be removed from the electrode surface and a layer of the insoluble organic molecule remained adsorbed at the CNT surface. Subsequently, the electrode was transferred into the electrochemical cell. The electrochem-

Anal Bioanal Chem (2007) 389:499–505

501

Fig. 1 Optical microscope pictures showing a side view of the carbon nanotube fiber microelectrode (a) and the polished and cross-section (b). Typically, a 20-μm diameter disk section is obtained. The SEM image

(c) illustrates the active surface area of the CNT fiber (surrounded by a black circle) that is perfectly sealed by the resin after polishing

ical transformation of only one of the three nitro groups into hydroxylamine is sufficient for good catalysis, as shown in previous studies [15].

Results and discussion

Pretreatments

The surface preparation and, hence, the final surface structure is often found to be critical for the performance of electrodes, their stability, and the reproducibility of the results. The electrode reaction is also sensitive to surface-modification effects. Potassium ferrocyanide has served as a classic redox couple to investigate the electrochemistry at different . CNTFMs. The redox couple FeðCN Þ4 Fe ð CN Þ3 6 6 is close to an ideal system with quasi-reversibility. In electrochemistry this redox couple is commonly used in instrument calibration, determination of diffusion coefficients, and measurements of the electroactive area . of an electrode. The FeðCN Þ4 FeðCN Þ3 6 6 model system has been used to characterize CNTFMs in terms of reproducibility, electron transfer kinetics, and active surface area. Figure 3 shows a typical cyclic voltammetric curve (CV) of 1 mmol L−1 potassium ferrocyanide obtained with a CNTFM. For the low scan rates usually employed for microelectrodes to minimize hysteresis the shape of the CV

Active surface area enhancement Before surface modification, electrodes are dipped overnight in a 5 mmol L−1 solution of polyoxometalate H3PMo12O40 (POM) [16] in 0.5 mol L−1 H2SO4 leading to the formation of an adsorbed layer of these molecules [17]. This treatment has been shown to exfoliate graphite and carbon particles [18, 19] and was thus tested here to improve the active surface of CNTFM. Then, to hydrolyze and to desorb the POMs from the surface, electrodes are dipped for several minutes in pH 8 buffer solution. CVs of the CNTFM background before and after POM conditioning differ significantly (Fig. 2). An initial capacitive current of 7.4 mF cm−2 increases by a factor 3 after POM treatment to reach, finally, 22.2 mF cm−2 for this electrode. This procedure can be repeated several times leading to electrodes with a capacitive current up to ten times higher than initially. The capacitive current being directly proportional to the active surface of the electrode [20], we conclude from this experiment that the active surface has been increased. As the fiber is completely sealed by the resin the diameter remains constant. The increase of active surface area therefore seems to be because of an increase of the effective porosity of the CNT fiber. POM molecules generate, through strong chemisorption, new cavities in the network formed by the nanotubes. Studies are in progress to obtain direct evidence of this phenomenon.

30 20 10

I / nA

Hot stretching Some fibers were pulled to 500% in length in a flow of hot air at 180 °C, a temperature well above the PVA glass transition. These fibers were then annealed and used to produce CNTFM with improved properties in comparison with those made from raw unstretched fibers. Indeed it has been shown in previous studies that hot pulling of PVA/CNT composite fibers leads to a substantial increase of the CNT alignment along the axis of the fiber [14].

Electrochemical characterization

0 -10 -20 -30 0

100

200

300

400

500

600

E / mV

Fig. 2 Polyoxometalate treatment. Cyclic voltammograms of a bare CNTFM in 0.5 mol L−1 H2SO4. Background before (squares) and after (triangles) polyoxometalate treatment and subsequent hydrolysis. Scan rate 0.05 V s−1

502

Anal Bioanal Chem (2007) 389:499–505

indicates a fast electron transfer rate comparable with that observed for other carbon microelectrodes and the limiting steady state current is characteristic for hemispherical diffusion, well known for small electrode dimensions [12]. No hysteresis is observed, meaning that the section of the fiber is perfectly sealed by the resin, with the CNTFM active surface being well defined as a disk-type microelectrode. From the CV and the relationship ΔI=4nFDCr (n: number of electron, F: Faraday constant, D: diffusion coefficient (8× 10−6 cm2 s−1), C: concentration of redox species (1 mmol L−1), r: electrode radius) the CNTFM diameter can be determined as ∼20.4 μm. These surface measurements are in good agreement with the actual size measured using optical or electron microscopy. To test the reproducibility, several CNTFMs have been made following the fabrication procedure described above. CVs of 1 mmol L−1 potassium ferrocyanide using different electrodes are shown in the Electronic supplementary material. We observe a steady-state current for all electrodes and superposition of the CVs obtained for different microelectrodes allows estimation of a mean plateau current and standard deviation (ΔI=3.1±0.5 nA). The observed differences between the electrodes are mainly due to the polishing procedure and the fact that the spinning procedure used for fabricating the CNT fibers does not lead to completely identical fibers in terms of diameter. Comparable and reproducible CVs are also obtained for classical carbon fiber electrodes and therefore a better performance of CNTFMs is not evident at this stage. Mediator adsorption

the amount of adsorbed molecules, the formal potential of the involved redox couples, and the kinetics of electron transfer can be obtained. Here, a surface modification has been performed to characterize the stability and the catalytic activity of a molecule belonging to a family that is used as efficient mediators in bioelectrocatalysis [21]. The CV obtained with 2,4,7-trinitro-9-fluorenone adsorbed on the surface is shown in Fig. 4. In the first scan three irreversible reductions are observed at −315 mV, −450 mV, and −620 mV corresponding to the transformation of the three nitro groups present in the molecule into hydroxylamine. In the back scan the overlapping reoxidation of the hydroxylamine substituents into nitroso groups is observed between −100 mV and 0 mV. In the second and subsequent scans a reversible redox signal was observed in the same potential range corresponding to the three –NO/–NHOH couples. The adsorbed redox couples undergo individually a 2e− and 2H+ redox process and are well adapted to mediate the overall two-electron transfer involved in the electrooxidation of NADH. Although we show here the activation of the three nitro groups, the transformation of only one nitro group is sufficient and even better for efficient catalysis, because of intramolecular effects [15]. Therefore in the following work we activated only one nitro group by cycling the potential between +600 mV and −315 mV. Figure 5 shows that the redox wave of adsorbed mediator is clearly enhanced on stretched CNTFM and it is definitely better than on classical CFMs (only one out of ten classical carbon microelectrodes will give a small signal in this case). The exact origin of this improvement upon stretching and the superior performance of CNTFMs compared with classical carbon fiber microelectrodes is not yet completely

Cyclic voltammetry is also well suited to study surfaceconfined redox species. Valuable information concerning

60 40

0

20

I / nA

0

-1

∇

I / nA

I

-20 -40 -60

-2 -80 -800

-600

-400

-200

0

200

400

600

E / mV -100

0

100

200

300

400

500

E / mV

Fig. 3 Microelectrode behavior. Cyclic voltammogram of a bare CNTFM in a solution of 1 mmol L−1 K3[Fe(CN)6] in 0.5 mol L−1 KNO3. Scan rate 0.01 V s−1

Fig. 4 Mediator adsorption. Cyclic voltammogram of a CNTFM modified with a monolayer of 2,4,7-trinitro-9-fluorenone in 0.1 mol L−1 Tris buffer. Background (solid circles) of the unmodified electrode, first scan after modification (triangles) showing the activation of the three nitro groups and a second scan (open squares) revealing the presence of the nitroso/hydroxylamine redox couple. Scan rate 0.05 V s−1

Anal Bioanal Chem (2007) 389:499–505

503

1.5

than classical carbon fiber electrodes, moreover CNTFMs offer the possibility of relatively easy optimization, for example surface treatment and stretching. These advantages make CNTFMs potentially more versatile than other conventional microelectrodes.

j / (nA/mm2)

1.0

0.5

Microelectrode regeneration 0.0

-0.5

-1.0 -150

-100

-50

0

50

100

150

E / mV

Fig. 5 Optimization of mediator adsorption. Cyclic voltammogram of 2,4,7-trinitro-9-fluorenone modified CNTFMs in 0.1 mol L−1 Tris buffer. The unstretched electrode shows a small mediator signal (open squares), for the stretched one the adsorption is significantly enhanced (solid squares). Scan rate 5 mV s−1

clear. Nevertheless, on the basis of the different structures a qualitative explanation can be proposed. Because of their fabrication process classical carbon fibers are composed of disordered graphite sheets and amorphous carbon [22]. Graphite and also CNTs are known to be highly reactive on their edges (or ends) and defect domains. It can be suggested that the mediator will preferentially adsorb on a well organized interface such as that present in stretched CNT fibers. Figure 6 illustrates schematically the change of CNT orientation in the fiber upon stretching. CNTs are anisotropic and exhibit better electrical performance along their main axis. The possibility of increasing the average alignment of the tubes inside the fibers offers an opportunity to exploit more efficiently their electronic properties and to benefit from the preferential exposure of reactive open CNT ends at the microelectrode section. CNTFMs have much better adsorption properties

By simply polishing the microelectrode section with a diamond buffing wheel, we can regenerate the active surface and obtain, reproducibly, a new modified interface. This is shown in a series of cyclic voltammograms in the Electronic supplementary material. From this it is obvious that the repolished electrode has electrochemical behavior almost identical with that of the surface used in the first experiment. Electrocatalysis Electrocatalytic NADH oxidation The catalytic activity of the mediator modified microelectrode can be easily checked when a certain amount of NADH is added to the supporting electrolyte. Figure 7 shows a cyclic voltammogram after addition of 3 mmol L−1 NADH. It can be seen that the adsorbed molecules are catalytically active because NADH oxidation usually takes place only at potentials higher than +500 mV on a bare electrode surface. The calibration curves (Fig. 8) obtained for different NADH concentrations with stretched and unstretched CNTFMs demonstrate the influence of the pretreatment on the microelectrode performance. With unstretched fibers, CVs are difficult to exploit because of important noise and poor stability during the measurement. This is probably

6

4

I / nA

2

0

-2

-4 -200

Fig. 6 Nanotube alignment. Schematic representation of the orientation of CNTs in a fiber. Stretched fibers present more oriented CNTs and statistically more tube ends are orientated towards the polished fiber section

-150

-100

-50

0

50

100

150

E / mV

Fig. 7 Electrocatalysis. Mediator-modified CNTFM in the absence (open squares) and in the presence (solid triangles) of 3 mmol L−1 NADH in 0.1 mol L−1 Tris buffer. Scan rate 0.005 V s−1

504

Anal Bioanal Chem (2007) 389:499–505 1800 1600 1400

j (nA/mm2)

1200 1000 800 600 400 200 0 0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

NADH (mM)

Fig. 8 Calibration curve. The catalytic current density as a function of NADH concentration measured with a mediator modified stretched (solid squares) and unstretched (open squares) CNTFM in 0.1 mol L−1 Tris buffer

because of unstable adsorption of the mediator on the surface of the CNTFMs. Because of improved adsorption of the mediator on the surface of stretched fiber microelectrodes, CVs are more stable and less noisy during the measurement, electrocatalysis of NADH is more efficient, and the overall current response is improved by a factor of 2. Detection of glucose The model system chosen is an enzymatic assay with glucose dehydrogenase to detect glucose. In contrast with other reports using glucose oxidase in combination with CNTs or carbon nanofibers [23, 24] the present enzyme needs a cofactor (NAD+). The CV (Fig. 9) of a mediator modified CNT fiber microelectrode is recorded in a solution containing the oxidized form

of the coenzyme (NAD+), the enzyme (GDH), and 30 mmol L−1 glucose. The addition of glucose leads to a significant increase in oxidation current, indicating that the enzyme oxidizes glucose. The oxidized form of the enzyme is then recovered by reaction with the coenzyme, generating NADH, which in turn is recycled electrochemically into NAD+. We note that the present data have been obtained for a given coenzyme-to-enzyme ratio. The signal can still be optimized by varying this ratio, but this was not the main objective of this study. Calibration curves (Fig. 10) demonstrate the instability of the catalytic signal for unstretched CNTFMs, whereas for stretched CNTFMs the response is in good agreement with electrochemical Michaelis–Menten kinetics. A quasi-linear relationship between the catalytic current and the glucose concentration is obtained for concentrations below 15 mmol L−1. This fits well the range of concentrations important for medical applications. As mentioned above optimization of the mediator/NAD+/GDH ratios might enable access of a broader range of concentrations, however the intrinsic limit at high glucose concentrations is given by the kinetics of the enzymatic reaction. In particular the Michaelis–Menten constant, Km, is a good indication of upper detection limits. For the values reported in Fig. 10 we fitted electrochemical Michaelis–Menten kinetics and obtained for stretched CNTFMs Km =3 mmol L−1 and imax =0.94 μA mm−2, and for stretched and POM-treated CNTFMs Km =1 mmol L−1 and imax =2.37 μA mm−2. imax depends not only on the kinetics of the enzymatic reaction, but is also very sensitive to the design of the electrode/electrolyte interface and, therefore, direct comparison with literature values is difficult. Km values can be compared, however, and apparent Km values between 2500

6 2000

j (nA/mm2)

4

I / nA

2

1500

1000

0 500

-2 0

-4 -200

0

-150

-100

-50

0

50

100

150

E / mV

Fig. 9 Optimization. Cyclic voltammograms of a stretched and POMtreated CNTFM in pH 8 buffer solution containing 1 mmol L−1 NAD+ and 10 units of GDH, in the absence (open squares) and in the presence (solid circles) of glucose. Scan rate 5 mV s−1

5

10

15

20

25

30

35

glucose (mM)

Fig. 10 Optimized calibration curve. Calibration curve of the catalytic current density as a function of glucose concentration with a mediatormodified unpretreated (open circles), stretched (black squares), stretched and POM treated (black triangles) CNTFM in buffer solution containing NAD+ and GDH

Anal Bioanal Chem (2007) 389:499–505

1.5 mmol L−1 and 8 mmol L−1 can be found in the literature as a function of the pretreatment of the electrode [25]. It would have been impossible to perform such measurements with classical carbon microelectrodes, because of the absence of mediator adsorption and therefore the importance of this kind of new electrode material becomes very clear at this point.

Conclusion To summarize, a fabrication procedure has been established leading to microelectrodes composed only of CNTs. CNTFMs show a good electrochemical performance and can be elaborated with reproducible quality. This kind of new electrode material allows several optimizations, for example surface treatment and stretching. We built a glucose sensing electrode by adsorption of a mediator on the surface of the CNTFM. Electrocatalytic oxidation of analytes via a dehydrogenase works efficiently at 0 V, which is a key point in developing such bioanalytical tools. CNTFMs adsorb mediators on their surface much better than classical CFMs. It would have been impossible to perform the same experiments on classical CFMs. Hence, because of the intrinsic properties of CNTs and different treatments of the fibers, we have reached a better performance with regard to stability and sensitivity. We can thus conclude that CNTFMs, although not yet fully optimized, are already very competitive with the classic fiber microelectrodes in terms of electrocatalysis, and superior in terms of adsorption properties. These advantages make CNTFMs potential bioanalytical tools which are more versatile than conventional carbon microelectrodes.

505

References 1. Iijima S (1991) Nature 354:56–58 2. Ebbesen TW, Ajayan PM (1992) Nature 358:220–222 3. Hamada N, Oshiyama A, Sawada S-I (1992) Phys Rev Lett 68:1579–1581 4. Gooding JJ (2005) Electrochim Acta 50:3049–3060 5. Britto PJ, Santhanam KSV, Ajayan PM (1996) Bioelectrochem Bioenerg 41:121–125 6. Wang J, Musameh M (2003) Anal Chem 75:2075–2079 7. Valentini F, Amine A, Orlanducci S, Terranova ML, Palleschi G (2003) Anal Chem 75:5413–5421 8. Wang Y, Li Q, Hu S (2005) Bioelectrochemistry 65:135–142 9. Wang J, Deo RP, Poulin P, Mangey M (2003) J Am Chem Soc 125:14706–14707 10. Huang X-J, Im H-S, Yarimaga O, Kim J-H, Jang D-Y, Lee D-H, Kim H-S, Choi Y-K (2006) J Electroanal Chem 594:27–34 11. Wang K, Fishman HA, Dai H, Harris JS (2006) Nano Lett 6:2043–2048 12. Stulick K, Amatore C, Holub K, Marecek V, Kutner W (2000) Pure Appl Chem 72:1483–1492 13. Vigolo B, Penicaud A, Coulon C, Sauder C, Pailler R, Journet C, Bernier P, Poulin P (2000) Science 290:1331–1334 14. Miaudet P, Badaire S, Maugey M, Derre A, Pichot V, Launois P, Poulin P, Zakri C (2005) Nano Lett 5:2212–2215 15. Mano N, Kuhn A (1999) J Electroanal Chem 477:79–88 16. Special issue about polyoxometalates (1998) Chem Rev 98:1–390 17. Viry L, Derre A, Garrigue P, Sojic N, Poulin P, Kuhn A (2007) J Nanosci Nanotech 7:1–5 18. Garrigue P, Delville MH, Labrugere C, Cloutet E, Kulesza PJ, Morand JP, Kuhn A (2004) Chem Mater 16:2984–2986 19. Rohlfing DF, Kuhn A (2006) Carbon 44:1942 20. Bard J, Faulkner R (2000) Electrochemical methods—fundamentals and applications. Wiley, Somerset, NJ, USA 21. Mano N, Kuhn A (2001) Biosens Bioelectron 16:653–660 22. Donnet JB, Wang TK, Rebouillat S, Peng JCM (eds) (1998) Carbon fibers. Marcel Dekker, New York 23. Sotiropoulou S, Chaniotakis NA (2003) Anal Bioanal Chem 375:103–105 24. Vamvakaki V, Tsagaraki K, Chaniotakis N (2006) Anal Chem 78:5538–5542 25. D’Costa EJ, Higgins IJ, Turner PF (1986) Biosensors 2:71–87