Available online at www.sciencedirect.

com

Electrochimica Acta 53 (2008) 6732–6739

Development of electrochemical oxidase biosensors based on carbon

nanotube-modified carbon film electrodes for glucose and ethanol
Carla Gouveia-Caridade, Rasa Pauliukaite, Christopher M.A. Brett ∗
Departamento de Quı́mica, Faculdade de Ciências e Tecnologia, Universidade de Coimbra, 3004-535 Coimbra, Portugal
Received 11 December 2007; received in revised form 16 January 2008; accepted 17 January 2008
Available online 29 January 2008

Abstract
Functionalised multi-walled carbon nanotubes (MWCNTs) were cast on glassy carbon (GC) and carbon film electrodes (CFE), and were
characterised electrochemically and applied in a glucose-oxidase-based biosensor. MWCNT-modified carbon film electrodes were then used to
develop an alcohol oxidase (AlcOx) biosensor, in which AlcOx–BSA was cross-linked with glutaraldehyde and attached by drop-coating. The
experimental conditions, applied potential and pH, for ethanol monitoring were optimised, and ethanol was determined amperometrically at −0.3 V
vs. SCE at pH 7.5. Electrocatalytic effects of MWCNT were observed with respect to unmodified carbon film electrodes. The sensitivity obtained
was 20 times higher at carbon film/MWCNT-based biosensors than without MWCNT.
© 2008 Elsevier Ltd. All rights reserved.

Keywords: Multi-walled carbon nanotubes; Carbon film electrodes; Glucose oxidase; Alcohol oxidase; Electrochemical biosensor

1. Introduction electron transfer reaction of some important biomolecules, such

as cytochrome c [9–11], myoglobin [12–14], glucose [15–19],
The development of carbon nanotubes (CNTs) for electroan- catalase [20,21], NADH [22–26], and haemoglobin [27,28],
alytical applications is currently a very active multidisciplinary among others.
field. During the last 15 years CNTs have attracted enormous The insolubility of CNTs in all solvents can be a major
interest due to their unique structure, mechanical strength and drawback to their use in electrochemical sensors and biosen-
electronic properties [1–4]. CNTs consist of seamless cylindrical sors. Several strategies have been proposed to dissolve CNTs
graphite sheets, which can be formed as a single tube of graphite including oxidative treatment [29], polymer wrapping [30],
(single-walled carbon nanotubes (SWCNT)), and multi-walled and sidewall functionalisation [31]. Functionalisation of CNTs
carbon nanotubes (MWCNTs), which consist of several concen- improves the solubility and processability, giving the oppor-
tric tubes of graphite inside one other. The reactivity of carbon tunity to develop new types of nanotube-based materials
nanotubes has been shown, as would be predicted, to be due only [32].
to the edge plane sites and to structural defects on the cylindrical Wrapping CNTs in polymeric chains, besides improving the
surface of the nanotubes [5]. solubility, also maintains the physical properties of the CNTs.
The chemical stability of CNTs and affinity to biomolecules The perfluorosulfonated polymer Nafion® has been extensively
make them very promising for application in electrochemical used for the modification of electrode surfaces and for the
sensors and biosensors [6]. As an electrode material, carbon construction of sensors and biosensors. Wang et al. [16] demon-
nanotubes have the ability to promote electron transfer reac- strated the usefulness of Nafion® to solubilise CNTs and they
tions with electroactive species in solution [7], showing better reported an electrocatalytic effect toward hydrogen peroxide,
electrochemical behaviour than conventional carbon electrodes which is of much interest for the operation of oxidase-based
[8]. It has been shown that CNTs are able to improve the direct amperometric biosensors. Tsai et al. [33] cast MWCNT dis-
persed in Nafion® on a glassy carbon (GC) electrode to construct
an electrochemical sensor for the analytical determination of
∗ Corresponding author. Tel.: +351 239835295; fax: +351 239835295.
heavy metals in the presence of surfactants such as SDS and
E-mail address: [email protected] (C.M.A. Brett). Triton X-100.

0013-4686/$ – see front matter © 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.electacta.2008.01.040
 C. Gouveia-Caridade et al. / Electrochimica Acta 53 (2008) 6732–6739 6733

Various electrochemical biosensors have been prepared using nanotube structure [32,41]. After this, 2-mg MWCNT–COOH
CNTs [6]. Most of these have been for glucose, with and with- were dispersed in 1-mL 1% Nafion® in ethanol.
out redox mediators. With respect to alcohol biosensors, the
only reports, to our knowledge, all use alcohol dehydrogenase, 2.3. Electrode preparation and enzyme immobilisation
mixing with NAD+ cofactor in a MWCNT composite [34], incor-
poration in a CNT paste electrode [35] or self-assembly on the Carbon film cylindrical electrodes were made from carbon
surface of a glassy carbon electrode modified with SWCNT held film resistors (resistance ∼2 ), their preparation protocol is
in place by polycations and then covered by Nafion® [36]. shown in Fig. 1A and is described in detail elsewhere [38,42].
Another form of carbon also used as support electrode The exposed disc electrode geometric area was ∼0.020 cm2 .
material is that of carbon films coated on a ceramic sub- Before use electrodes were electrochemically pretreated by
strate by pyrolysis—these small and inexpensive electrodes cycling the applied potential between −1.5 and +1.5 V vs. SCE
have been electrochemically characterised [37,38] and success- in PBS solution for not less than 10 cycles, until stable cyclic
fully applied to the development of electrochemical sensors and voltammograms were obtained.
biosensors, e.g. Refs. [39,40]. A glassy carbon electrode with 7 mm diameter was used
The aim of this study is to extend the use of carbon film elec- to optimise the casting modification procedure. Before use the
trodes (CFE) as substrate electrode material by modifying them glassy carbon surface was polished with 1.00 and 0.05 ␮m alu-
with MWCNT to develop an alcohol oxidase (AlcOx)-based mina powder on a polishing cloth.
biosensor. A casting method for MWCNT was first investigated, A volume of 40 ␮L of MWCNT in Nafion® was used to
based on methods developed for glassy carbon electrodes [41]. modify the glassy carbon electrode, or MWCNT were attached
The MWCNTs were first functionalised in HNO3 and then dis- directly on the surface of the carbon film electrodes without
solved in Nafion® . The MWCNT/Nafion® dispersion was then any solvent (Fig. 1B), by gently rubbing the electrode on a
used to develop and optimise a glucose biosensor, using glucose filter paper with carbon nanotubes placed on its surface [43].
oxidase (GOx) enzyme, on a modified glassy carbon electrode The surface coverage of MWCNT in the case of attachment
prior to application in developing an alcohol oxidase-based with Nafion® at carbon film electrodes was calculated from the
ethanol biosensor on carbon film electrodes. peak current for hexaamineruthenium(III) reduction in cyclic
voltammograms and was found to be 27 ± 1 nmol cm−2 .
2. Experimental For attaching the enzyme layer, a volume of 4 ␮L enzyme–
BSA–GA mixture (5 ␮L 10% GOx solution in 0.1 M PBS, pH
2.1. Materials and reagents 7.0 + 5 ␮L 10% BSA solution in 0.1 M PBS, pH 7.0 + 1 ␮L glyc-
erol + 1 ␮L 2.3% GA in water or 2 ␮L 5% AlcOx in 0.1 M PBS,
Multi-walled carbon nanotubes were obtained from NanoLab pH 7.0 + 1 ␮L 10% BSA solution in 0.1 M PBS, pH 7.0 + 0.5 ␮L
(Newton, MA, USA). Nafion® 5% solution in ethanol was from glycerol + 0.5 ␮L 2.3% GA in water) was used to modify the
Aldrich (Germany). Glucose oxidase from Asperigilus niger GC/CNT or C film/CNT working electrodes (Fig. 1C). The elec-
EC1.1.3.4, alcohol oxidase from Hansenula sp. EC1.1.3.13, and trodes were left for an hour to dry at room temperature and then
bovine serum albumin (BSA) were purchased from Sigma (Ger- immersed in buffer solution for half an hour before the first mea-
many). 70% glutaraldehyde (GA) solution in water was from surement to dissolve the rest of the unreacted GA. All biosensors
Fluka (Switzerland). were stored in phosphate buffer at 4 ◦ C while not in use.
Phosphate buffer or phosphate buffer saline (PBS) solutions,
concentration 0.1 M, pH from 6.0 to 8.5 were prepared from 2.3.1. Optimisation of casting method
sodium dihydrogenphosphate and disodium hydrogenphosphate Optimisation of the casting mixture was performed on a GC
and 0.05 M NaCl (all these reagents from Riedel-de-Haën, electrode and using GOx. Carbon nanotubes were attached to the
Germany). Millipore Milli-Q nanopure water (resistivity GC electrode in different ways, and the amperometric response
>18 M cm) was used for preparation of all solutions. Experi- to glucose was considered as representing the biosensor effi-
ments were performed at room temperature, 25 ± 1 ◦ C. ciency. The three procedures tested were:

2.2. Pretreatment of carbon nanotubes • GC/MWCNT/GOx: GC was modified by drop-coating with

MWCNT dispersed in Nafion® and, after evaporation of the
MWCNTs were purified and functionalised by stirring in 2 M solvent, the GOx–BSA–GA mixture was placed on top of the
nitric acid solution for 20 h. The solid product was collected on MWCNT/Nafion® film.
a filter paper and washed several times with nanopure water • GC/MWCNT–GOx: GC was modified by drop-coating a mix-
until the filtrate pH became nearly neutral. The functionalised ture of MWCNT–GOx–BSA–GA.
MWCNTs obtained were then dried in an oven at ∼80 ◦ C for • GC/GOx/MWCNT: The enzyme layer was first placed on top
24 h. This procedure was performed to ensure complete removal of the GC electrode and, after drying at room temperature,
of transition metal ion catalyst, used in the production of nan- the MWCNTs were then attached to the enzyme layer.
otubes, as well as of amorphous carbon. Nitric acid also causes
significant destruction of carbon nanotubes and introduces A GC/GOx assembly, in which the enzyme layer was placed
–COOH groups at the ends of, or at the sidewall defects in the directly on top of the GC electrode without MWCNT, was also
6734 C. Gouveia-Caridade et al. / Electrochimica Acta 53 (2008) 6732–6739

Fig. 1. Scheme of fabrication of the ethanol biosensor. (A) Preparation of the carbon film electrode: removal of one tight-fitting metal cap (1), and protection of
the connecting wire and other metal cap by a plastic sheath and epoxy resin (2). (B) Modification of carbon film disc electrode with –COOH functionalised carbon
nanotubes without solvent. (C) Cross-linking of alcohol oxidase (AlcOx) and BSA mixture with glutaraldehyde (GA) on top of the carbon nanotubes.

prepared, to enable comparison with the MWCNT-modified for immobilising the carbon nanotubes and the enzyme layer on
electrodes. Amperometric measurements were performed at the carbon film electrode surface.
+0.70 V vs. SCE, where electrooxidation of H2 O2 formed in
the enzyme-catalysed reaction occurs. 3.1.1. Glassy carbon electrode support
The optimisation of the casting mixture was performed on
2.4. Instruments and methods a GC electrode with the well-known and stable GOx enzyme.
Carbon nanotubes were attached to the GC electrode in different
A three-electrode electrochemical cell of 15 mL volume was ways as described in Section 2. The amperometric response to
used for electrochemical measurements. It contained the car- electrooxidation of H2 O2 at +0.70 V vs. SCE produced in the
bon film or the glassy carbon working electrode, a platinum enzyme-catalysed oxidation of glucose was used to evaluate the
foil as counter electrode and a saturated calomel electrode biosensor efficiency.
(SCE) as reference. Measurements were performed using a The results of measurements carried out are presented in
computer-controlled ␮-Autolab Type II potentiostat/galvanostat Table 1, and they show that the best biosensor activity was
with GPES 4.9 software (Eco Chemie, Netherlands). obtained when the GOx–BSA mixture was deposited on top
of the MWCNT. The sensitivity to glucose at this biosen-
3. Results and discussion sor was 17 times higher than without any MWCNT. When
MWCNTs were attached on the top of the enzyme layer, the
3.1. MWCNT casting method for glucose biosensor sensitivity of the biosensor was the same as without CNT,
showing that the CNT completely covered the enzyme layer
A good casting method for MWCNT on carbon film elec- and, consequently, contact of analyte with the enzyme was
trodes was first investigated, based on methods developed for poor.
glassy carbon electrodes [41]. However due to the small volume When MWCNT were mixed with the GOx–BSA–GA mix-
required for electrode modification (1 ␮L) difficulties arose in ture and deposited together on the GC electrode, the response to
placing the solutions containing the dispersion of MWCNT on glucose was much lower than in the case of GOx drop-coated in
the top of the electrode, and it was necessary to develop strategies a separate layer on top of the CNT. This decrease in response was
 C. Gouveia-Caridade et al. / Electrochimica Acta 53 (2008) 6732–6739 6735

Table 1
Calibration data at different CNT–GOx biosensors
Biosensor assembly composition Linear range (mM) Sensitivity (␮A cm−2 mM−1 ) R2 Detection limit (␮M) KM (mM)

GC/GOx 0.05–1.10 0.52 0.998 8.1 2.2

GC/CNT/GOx 0.05–1.10 9.43 0.997 9.3 1.9
GC/CNT–GOx 0.05–1.00 2.39 0.997 14.0 2.5
GC/GOx/CNT 0.05–1.00 0.49 0.998 16.2 2.2

Applied potential +0.70 V vs. SCE; supporting electrolyte 0.1 M PBS, pH 7.0.

probably caused by partial blocking of the enzyme and poorer bly, calculated from the reduction peak current obtained in
contact between the enzyme and the carbon nanotubes. 3 mM hexaamineruthenium(III) solution was ∼0.060 cm2 , using
a diffusion coefficient of 9.1 × 10−6 cm2 s−1 [45], which was 3
3.1.2. Carbon film electrode support times higher than the electroactive/geometric area of the carbon
The next step was employing CFE as substrate material. The film electrode, ∼0.020 cm2 , and was 30 times higher than the
diameter of the CFE disc was 1.5-mm which made it difficult to electroactive area of CFE/Nafion® (1% Nafion® ), ∼0.002 cm2 .
place the CNT–Nafion® mixture accurately on the disc and con- Comparison of these data shows that Nafion® film blocks some
trol the amount CNT at the electrode surface experiments (such electroactive centres of the carbon film electrode.
as those described below) showed that the amount of MWCNT A glucose biosensor was then prepared by binding the
deposited was variable. Therefore it was decided to attach the solid functionalised MWCNT to the carbon film with the
functionalised MWCNT directly to carbon film electrode surface enzyme–BSA–GA mixture, in the same sequence as was found
without any solvent, as in Ref. [43], following the procedure in to be the best at GC electrode supports.
Section 2 and finally a volume of 1 ␮L of a 1% Nafion® solution It is known that CNT promote direct electron transfer
(as binder) or the enzyme–BSA–GA mixture was drop-coated in the case of GOx [17], so amperometric testing of the
on top. CFE/MWCNT–GOx–BSA–GA electrode was performed at 0 V
Cyclic voltammograms of the CFE/MWCNT/Nafion® vs. SCE. It was found that H2 O2 is reduced at this potential at
assembly were recorded in 3 mM hexaamineruthenium(III) in CFE/MWCNT–Nafion® (Fig. 3), but there is a positive change
order to evaluate the electrochemical behaviour of the modi- in current, corresponding to oxidation, at the biosensor after
fied electrode (see Fig. 2). A fully reversible redox behaviour addition of aliquots of glucose solution. Moreover, decreasing
of hexaamineruthenium(III) was observed, with the cathodic the applied potential to −0.45 V, the oxidation current increased
and anodic peaks situated at ∼−0.220 and −0.160 mV, respec- and the linear range of amperometric response decreased, as
tively, suggesting ideal reversibility at the electrode, as was seen in Fig. 3. This confirms that direct electron transfer is tak-
also observed with Fe(CN)6 3−/4− at MWCNT-modified elec- ing place at CNT as in Ref. [17], so that no redox mediator is
trodes [44]. A linear dependence of peak current on square required. At −0.45 V, regeneration of GOx–FAD takes place at
root of scan rate was found with a slope of 142 ␮A V−1/2 s1/2 . carbon substrates [19,46].
The electroactive area of the CFE/MWCNT/Nafion® assem- As seen from Fig. 3, hydrogen peroxide reduction occurs
at 0.0 V, showing the catalytic effect of CNT. However, the

Fig. 2. Cyclic voltammograms of 3 mM Ru(NH3 )6 3+ in 0.1 M KCl at Fig. 3. Amperometric response to H2 O2 at CFE/CNT–Nafion® at 0.0 V vs.
CFE/CNT/Nafion® at different scan rates from 0.025 to 0.05 V s−1 (a–d) after SCE (䊉) and to glucose at CFE/CNT–GOx–BSA–GA at 0.0 V (♦), −0.1 V (),
baseline subtraction. Inset: dependence of anodic peak current on the square −0.2 V (), and −0.45 V () vs. SCE supporting electrolyte 0.1 M PBS, pH
root of scan rate. 7.0.
6736 C. Gouveia-Caridade et al. / Electrochimica Acta 53 (2008) 6732–6739

Table 2
Calibration data at various CFE–CNT–GOx biosensors; at different applied potentials, supporting electrolyte 0.1 M PBS, pH 7.0
Potential (V) Analyte Linear range (mM) Sensitivity (␮A cm−2 mM−1 ) R2 Detection limit (␮M) KM (mM)

0.0 (CNT) H2 O2 – 106 0.998 15.5 –

0.0 (CNT/GOx) Glucose 0.05–1.2 75.0 0.998 7.1 1.7
−0.10 (CNT/GOx) Glucose 0.1–1.1 237 0.998 5.1 2.0
−0.20 (CNT/GOx) Glucose 0.1–0.9 364 0.998 2.8 1.5
−0.45 (CNT/GOx) Glucose 0.1–0.8 666 0.999 2.2 1.7

response at the biosensor with MWCNT was significantly lower biosensors based on alcohol oxidase [47,48]. However, the pH
than that to added H2 O2 but, nevertheless, the sensitivity to glu- values vary depending on the assay and increase with increase
cose at the biosensor was higher at −0.1 or −0.2 V than at +0.7 V, in aliphatic chain length of the alcohol [48].
where hydrogen peroxide oxidation takes place at conventional The response of the CFE/MWCNT/AlcOx biosensor was
carbon electrodes, see Tables 1 and 2, and the substrate–enzyme tested at pH 7.5. The dependence of the amperometric signal
reaction had typical Michaelis–Menten kinetics. A potential on the applied potential was examined in the range −0.5 to
value in the middle of the range of the best response, −0.15 V +0.8 V vs. SCE, Fig. 5. Higher currents were obtained at the
vs. SCE, was chosen for further investigations. The response to MWCNT-modified electrode, compared to the unmodified one,
glucose at separate biosensors prepared in the same way var- for all the potentials in Fig. 5. This can be also observed in
ied up to 10%, which was attributed to different surface areas
owing to variations in the mass fraction of MWCNT attached to
the electrode supports.
The linear range at −0.15 V vs. SCE at the CFE/MWCNT–
GOx–BSA–GA biosensor was from 0.05 to 0.9 mM, the sen-
sitivity was 289 ␮A cm−2 mM−1 , the limit of detection was
5.0 ␮M and the apparent Michaelis–Menten constant was
2.4 mM. These calibration parameters were more suitable for
glucose monitoring than at +0.7 V vs. SCE in terms of sensitivity
and detection limit.

3.2. Ethanol biosensor

3.2.1. Optimisation of operational conditions

Alcohol oxidase was used for the development of an elec-
trochemical ethanol biosensor. The enzyme-catalysed reaction
produces acetaldehyde and hydrogen peroxide:
AlcOx
C2 H5 OH + O2 −→ CH3 CHO + H2 O2 (1)
First, the CFE was used by itself, without MWCNT, to procure
the best experimental conditions for using AlcOx. Enzyme was
immobilised in the same way as GOx, cross-linked with GA
from a solution of AlcOx and BSA in a 1:1 ratio. The influence
of applied potential was studied for the most common experi-
mental conditions used for oxidases, in 0.1 M PBS solution, pH
7.0 (Fig. 4(a)). A good operating potential without any mediator
was found to be −0.45 V vs. SCE where regeneration of FAD
takes place, as in the case of GOx (see above), and the response
was much higher than that for electro-oxidation of H2 O2 at
+0.75 V, formed during the enzymatic reaction (Fig. 4(a), inset).
Therefore an applied potential of −0.45 V was used for pH
optimisation.
Phosphate buffer solutions of analytical concentration 0.1 M
with pH values from 6.0 to 8.5 were used for determination of
the best pH value, the pH giving the best response being between
Fig. 4. Optimisation of CFE/AlcOx–BSA–GA electrode (a) operating potential
7.5 and 8.0 (Fig. 4(b)). Thus, pH 7.5 was chosen for the further in 0.1 M PBS, pH 7.0 + 1 mM ethanol and (b) pH in 0.1 M PBS +0.3 mM ethanol
development of the ethanol biosensor with MWCNT. Simi- at −0.45 V vs. SCE. Inset of (a) shows calibration curves of ethanol in 0.1 M
lar optimum pH values have been reported for optical alcohol PBS, pH 7.0 at () −0.45 V and () +0.75 V vs. SCE.
 C. Gouveia-Caridade et al. / Electrochimica Acta 53 (2008) 6732–6739 6737

Fig. 5. Optimisation of operating potential at CFE/MWCNT/AlcOx–BSA–GA

electrode in 0.1 M PBS, pH 7.5 + 1.0 mM ethanol.

the cyclic voltammograms in Fig. 6 recorded at the modified

and unmodified MWCNT biosensor with 6 mM ethanol in solu-
tion. The reduction current at the MWCNT biosensor increases
after ethanol addition in the negative potential region, although
no clearly defined cyclic voltammetric peak appears as it does
in the case of the biosensor without carbon nanotubes. How-
ever, the MWCNT biosensor gives a well-defined response to
ethanol at −0.20 V, while at the CFE biosensor the response to
the presence of ethanol starts only at a more negative potential
of −0.39 V. Although the highest current response in ampero-
metric measurements to ethanol addition was at −0.45 V, partly
due to the reaction of FAD itself, it was decided to choose an
operating potential of −0.30 V since a good current response
to ethanol is still obtained and the background current is much
lower. Fig. 6. Cyclic voltammograms at carbon film biosensor: (a) unmodified and (b)
modified with MWCNT in 0.1 M PBS solution, pH 7.5 (solid line), and after
3.2.2. Amperometric response to ethanol addition of 6.0 mM of ethanol (dashed line).
The CFE/MWCNT/AlcOx biosensor was applied to the
amperometric determination of ethanol at −0.30 V vs. SCE.
After stabilisation of the baseline current, ethanol was injected
into the buffer solution. The calibration curve obtained for
ethanol is shown in Fig. 7, the linear part being described by
the equation:
j (␮A cm−2 ) = 1.88(±0.77) + 44.5(±0.99) [ethanol]
with a correlation coefficient of 0.998. The linear range was
up to 1.4 mM, and the detection limit was 86 ␮M. The appar-
ent Michaelis–Menten constant (KM ), determined from the
Lineweaver–Burk plot, was 2.2 mM. The data obtained with the
same enzyme at carbon film electrodes without MWCNT and
with poly(neutral red) redox mediator under identical conditions
lead to a linear range and apparent Michaelis–Menten constant
with lower values than at the biosensor with MWCNT—the lin-
ear range is up to 0.6 mM and KM is 2.1 mM [49]. The kinetics of
alcohol oxidase also depends on the immobilisation method, as
reported in Ref. [48]: KM for AlcOx increased from 1.5 mM (free Fig. 7. Calibration curve obtained with CFE/CNT/AlcOx biosensor in 0.1 M
AlcOx in solution) to 6.8 mM after its electrochemical immobili- PBS pH 7.5, at −0.30 V vs. SCE.
6738 C. Gouveia-Caridade et al. / Electrochimica Acta 53 (2008) 6732–6739

sation into polypyrrole. However, the critical dependence of the 4. Conclusions

kinetics is on the enzyme nature, i.e. which organism AlcOx
is taken from. For example, the apparent Michaelis–Menten The best casting method of functionalised multi-walled car-
constant of AlcOx from different mutants of yeast Hansenula bon nanotubes has been investigated on glassy carbon and on
polymorpha varied from 1.3 to 12.1 mM for ethanol [50]. carbon film electrodes. The most effective method for glassy
The calibration parameters for the CFE/AlcOx biosensor carbon was mixing MWCNTs with Nafion® , whereas at carbon
without MWCNT obtained at −0.45 V (best potential for the films, immobilisation of solid MWCNTs was best. Optimisation
unmodified biosensor), in Fig. 2, were a linear range up to of the operation of the biosensing system with carbon nanotubes
∼1.0 mM, sensitivity of 2.23 ± 0.05 ␮A cm−2 mM−1 and detec- was first carried out with GOx and the optimal conditions were
tion limit of 37 ␮M. Thus, the sensitivity obtained with the found to be a potential of −0.15 V vs. SCE and pH 7.0.
MWCNT biosensor was 20 times higher than without MWCNT, The MWCNT were then used to develop an AlcOx-based
even at an applied potential closer to zero, and was more than 50 ethanol biosensor. The optimal conditions for sensor operation
times higher than the biosensor with poly(neutral red) mediator were similar to those for glucose except that a more negative
[49]. potential of −0.30 V vs. SCE was used. Electrocatalytic effects
Reproducibility studies at three biosensors, prepared in the of MWCNT were observed in relation to the unmodified car-
same way, showed variations of up to 30%, which can be a bon film electrode. Although there was a higher detection limit
limitation to the use of carbon nanotube biosensors prepared by for ethanol of 86 ␮M at the CFE/MWCNT/AlcOx compared
attaching solid CNT to electrode surfaces. However, individual with 37 ␮M at CFE/AlcOx, there was a significant increase in
calibration of each sensor can easily be done together with use sensitivity by a factor of twenty, which augurs well for future
of the standard addition method for unknown samples, which application of these sensors.
surmounts this possible drawback.
Acknowledgements
3.2.3. Operational lifetime and biosensor selectivity
The stability of the enzyme biosensor was tested daily dur- Financial support from Fundação para a Ciência e Tec-
ing 3 weeks. The biosensors were stored in buffer at 4 ◦ C while nologia (FCT), project PTDC/QUI/65255/2006, POCI 2010
not in use. After this time there was a decrease of 70% of (co-financed by the European Community Fund FEDER) and
the initial sensitivity value. At biosensors without MWCNT ICEMS (Research Unit 103), is gratefully acknowledged. CGC
but with poly(neutral red) mediator [49], a similar reduction and RP thank FCT for a PhD grant (SFRH/BD/18659/2004) and
is seen. However, the lack of stability of alcohol oxidase is well a postdoctoral fellowship (SFRH/BPD/27075/2006), respec-
known [51] and both this study and the results in Ref. [49] repre- tively.
sent an improvement with respect to previous results regarding
stability. References
Selectivity is another important parameter to analyse when
developing an electrochemical biosensor. The relative amper- [1] C.N.R. Rao, B.C. Satishkumar, A. Govindaraj, M. Nath, Chemphyschem
ometric response of the CFE/MWCNT/AlcOx biosensor to 2 (2001) 78.
ethanol in the presence of some common interferents in food [2] R.H. Baughman, A.A. Zakhidov, W.A. de Heer, Science 297 (2002) 787.
and beverages is shown in Table 3. All the compounds tested [3] E. Katz, I. Willner, Chemphyschem 5 (2004) 1085.
[4] J.J. Gooding, Electrochim. Acta 50 (2005) 3049.
were present at concentrations of 0.8 mM with an ethanol con- [5] G.G. Wildgoose, C.E. Banks, H.C. Leventis, R.G. Compton, Microchim.
centration of 0.4 mM and showed a small interference, except Acta 152 (2006) 187.
in the case of ascorbic acid where an increase of ∼30% in the [6] G.A. Rivas, M.D. Rubianes, M.C. Rodrı́guez, N.F. Ferreyra, G.L. Luque,
response was obtained. Considering wine as a sample matrix, M.L. Pedano, S.A. Miscoria, C. Parrado, Talanta 74 (2007) 291.
this is not a problem since ascorbate is present in wine in a [7] P.J. Britto, K.S.V. Santhanam, P.M. Ajayan, Bioelectrochem. Bioenerg. 41
(1996) 121.
much lower concentration than ethanol. [8] Q. Zhao, Z.H. Gan, Q.K. Zhuang, Electroanalysis 14 (2002) 1609.
[9] J.X. Wang, M.X. Li, Z.J. Shi, N.Q. Li, Z.N. Gu, Anal. Chem. 74 (2002)
Table 3 1993.
Interference effect of various compounds on the assay of ethanol at CFE/ [10] G.C. Zhao, Z.Z. Yin, L. Zhang, X.W. Wei, Electrochem. Commun. 7 (2005)
MWCNT/AlcOx 256.
[11] Y.J. Yin, P. Wu, Y.F. Lu, P. Du, Y.M. Shi, C.X. Cai, J. Solid State Elec-
Interferent Relative response to the analyte in the trochem. 11 (2007) 390.
presence of the interferents (%) [12] G.C. Zhao, L. Zhang, X.W. Wei, Z.S. Yang, Electrochem. Commun. 5
Acetic acid 100 (2003) 825.
Ascorbic acid 131 [13] L. Zhang, G.C. Zhao, X.W. Wei, Z.S. Yang, Chem. Lett. 33 (2004) 86.
Citric acid 100 [14] Y.F. Lu, Y.J. Yin, P. Wu, C.X. Cai, Acta Phys. Chim. Sin. 23 (2007) 5.
Lactic acid 105 [15] A. Guiseppi-Elie, C.H. Lei, R.H. Baughman, Nanotechnology 13 (2002)
Malic acid 100 559.
Oxalic acid 111 [16] J. Wang, M. Musameh, Y.H. Lin, J. Am. Chem. Soc. 125 (2003) 2408.
Tartaric acid 107 [17] C.X. Cai, J. Chen, Anal. Biochem. 332 (2004) 75.
[18] F. Patolsky, Y. Weizmann, I. Willner, Angew. Chem. Int. Ed. 43 (2004)
Ratio 1:2 of ethanol to interferent. 2113.
 C. Gouveia-Caridade et al. / Electrochimica Acta 53 (2008) 6732–6739 6739

[19] X.L. Luo, A.J. Killard, M.R. Smyth, Electroanalysis 18 (2006) 1131. [33] Y.C. Tsai, J.M. Chen, F. Marken, Microchim. Acta 150 (2005) 269.
[20] L. Wang, J.X. Wang, F.M. Zhou, Electroanalysis 16 (2004) 627. [34] J. Wang, M. Musameh, Anal. Lett. 36 (2003) 2041.
[21] A. Salimi, A. Noorbakhsh, M. Ghadermarz, Anal. Biochem. 344 (2005) [35] M.D. Rubianes, G.A. Rivas, Electroanalysis 17 (2005) 73.
16. [36] S. Liu, C.H. Cai, J. Electroanal. Chem. 602 (2007) 103.
[22] M. Musameh, J. Wang, A. Merkoci, Y.H. Lin, Electrochem. Commun. 4 [37] O.M.S. Filipe, C.M.A. Brett, Electroanalysis 16 (2004) 994.
(2002) 743. [38] C. Gouveia-Caridade, C.M.A. Brett, Electroanalysis 17 (2005) 549.
[23] J. Chen, J.C. Bao, C.X. Cai, T.H. Lu, Chin. Chem. Lett. 14 (2003) 1171. [39] M.E. Ghica, C.M.A. Brett, Anal. Lett. 39 (2006) 1527.
[24] J. Chen, C.X. Cai, Chin. J. Chem. 22 (2004) 167. [40] C. Gouveia-Caridade, C.M.A. Brett, J. Electroanal. Chem. 592 (2006) 113.
[25] J. Wang, Electroanalysis 17 (2005) 7. [41] N.S. Lawrence, R.P. Deo, J. Wang, Electroanalysis 17 (2005) 65.
[26] K.S. Prasad, J.C. Chen, C. Ay, J.M. Zen, Sens. Actuator B: Chem. 123 [42] C.M.A. Brett, L. Angnes, H.D. Liess, Electroanalysis 13 (2001) 765.
(2007) 715. [43] A. Salimi, A. Noorbakhsh, S. Soltanian, Electroanalysis 18 (2006) 703.
[27] J. Chen, C.X. Cai, Chin. Chem. Lett. 15 (2004) 813. [44] J.M. Nugent, K.S.V. Santhanam, A. Rubio, P.M. Ajayan, Nano Lett. 1
[28] G.C. Zhao, M.Q. Xu, J. Ma, X.W. Wei, Electrochem. Commun. 9 (2007) (2001) 87.
920. [45] F. Marken, J.C. Eklund, R.G. Compton, J. Electroanal. Chem. 395 (1995)
[29] W. Zhao, C.H. Song, P.E. Pehrsson, J. Am. Chem. Soc. 124 (2002) 335.
12418. [46] G.D. Withey, A.D. Lazareck, M.B. Tzolov, A. Yin, P. Aich, J.I. Yeh, J.M.
[30] M.J. O’Connell, P. Boul, L.M. Ericson, C. Huffman, Y.H. Wang, E. Haroz, Xu, Biosens. Bioelectron. 21 (2006) 1560.
C. Kuper, J. Tour, K.D. Ausman, R.E. Smalley, Chem. Phys. Lett. 342 [47] R.C.W. Lau, M.H.F. Choi, J.Z. Lu, Talanta 48 (1999) 321.
(2001) 265. [48] H.B. Yildiz, L. Toppare, Biosens. Bioelectron. 21 (2006) 2306.
[31] R.K. Saini, I.W. Chiang, H.Q. Peng, R.E. Smalley, W.E. Billups, R.H. [49] M. Barsan, C.M.A. Brett, Talanta 74 (2008) 1505.
Hauge, J.L. Margrave, J. Am. Chem. Soc. 125 (2003) 3617. [50] K.V. Dmytruk, O.L. Smutok, O.B. Ryabova, G.Z. Gayda, V.A. Sibirny, W.
[32] P. Papakonstantinou, R. Kern, L. Robinson, H. Murphy, J. Irvine, E. Schuhmann, M.V. Gonchar, A.A. Sibirny, BMC Biotechnol. 7 (2007) 33.
McAdams, J. McLaughlin, T. McNally, Fullerene Nanotube Carbon Nanos- [51] N.G. Patel, S. Meier, K. Camman, G.C. Chemnitius, Sens. Actuator B:
truct. 13 (2005) 91. Chem. 75 (2001) 101.