Influence of surface adsorption on the interfacial electron transfer of flavin adenine dinucleotide and glucose oxidase at carbon nanotube and nitrogen-doped carbon nanotube electrodes

Article abstract of DOI:10.1021/ac3028036

The adsorption of flavin adenine dinucleotide (FAD) and glucose oxidase (GOx) onto carbon nanotube (CNT) and nitrogen-doped CNT (N-CNT) electrodes was investigated and found to obey Langmuir adsorption isotherm characteristics. The amount adsorbed and adsorption maximum are dependent on exposure time, the concentration of adsorbate, and the ionic strength of the solution. The formal potentials measured for FAD and GOx are identical, indicating that the observed electroactivity is from FAD, the redox reaction center of GOx. When glucose is added to GOx adsorbed onto CNT/N-CNT electrodes, direct electron transfer (DET) from enzyme-active FAD is not observed. However, efficient mediated electron transfer (MET) occurs if an appropriate electron mediator is placed in solution, or the natural electron mediator oxygen is used, indicating that GOx is adsorbed and active on CNT/N-CNT electrodes. The observed surface-confined redox reaction at both CNT and N-CNT electrodes is from FAD that either specifically adsorbs from solution or adsorbs from the holoprotein subsequently inactivating the enzyme. The splitting of cathodic and anodic peak potentials as a function of scan rate provides a way to measure the heterogeneous electron-transfer rate constant (k_s) using Laviron's method. However, the measured k _s was found to be under ohmic control, not under the kinetic control of an electron-transfer reaction, suggesting that k_s for FAD on CNTs is faster than the measured value of 7.6 s^-1.

Full text of DOI:10.1021/ac3028036

Article
pubs.acs.org/ac
Influence of Surface Adsorption on the Interfacial Electron Transfer
of Flavin Adenine Dinucleotide and Glucose Oxidase at Carbon
Nanotube and Nitrogen-Doped Carbon Nanotube Electrodes
Jacob M. Goran,† Sandra M. Mantilla,‡ and Keith J. Stevenson*,†
†Department of Chemistry and Biochemistry, Center for Electrochemistry, Center for Nano- and Molecular Science and Technology,
The University of Texas at Austin, 1 University Station, A5300, Austin, Texas 78712, United States
‡Department of Biomedical Engineering, Florida International University, 10555 West Flagler Street, EAS 2600, Miami, Florida
33174, United States
S
* Supporting Information
ABSTRACT: The adsorption of flavin adenine dinucleotide
(FAD) and glucose oxidase (GOx) onto carbon nanotube
(CNT) and nitrogen-doped CNT (N-CNT) electrodes was
investigated and found to obey Langmuir adsorption isotherm
characteristics. The amount adsorbed and adsorption max-
imum are dependent on exposure time, the concentration of
adsorbate, and the ionic strength of the solution. The formal
potentials measured for FAD and GOx are identical, indicating
that the observed electroactivity is from FAD, the redox
reaction center of GOx. When glucose is added to GOx
adsorbed onto CNT/N-CNT electrodes, direct electron transfer (DET) from enzyme-active FAD is not observed. However,
efficient mediated electron transfer (MET) occurs if an appropriate electron mediator is placed in solution, or the natural
electron mediator oxygen is used, indicating that GOx is adsorbed and active on CNT/N-CNT electrodes. The observed surface-
confined redox reaction at both CNT and N-CNT electrodes is from FAD that either specifically adsorbs from solution or
adsorbs from the holoprotein subsequently inactivating the enzyme. The splitting of cathodic and anodic peak potentials as a
function of scan rate provides a way to measure the heterogeneous electron-transfer rate constant (ks) using Laviron’s method.
However, the measured ks was found to be under ohmic control, not under the kinetic control of an electron-transfer reaction,
suggesting that ks for FAD on CNTs is faster than the measured value of 7.6 s−1.
arbon nanotubes (CNTs) have been explored for a vast
array of biosensing applications, either by themselves1−4
(GC),37,38 modified GC,39 gold,40 titanium,41 TiO2,42 and
graphite,43,44 but has not been adequately investigated on
CNTs. This study examines the spontaneous adsorption of
FAD and GOx on CNTs and nitrogen-doped CNTs (N-
CNTs), which offers a unique way to modify the surface and
extend our understanding of the surface-confined redox
reaction. Our study is unique in that oxidative acids such as
sulfuric and nitric18,22,23,29,30 are not used to create oxygen
functionalities and assist in the dispersion of the hydrophobic
CNTs into aqueous solutions. In addition, we do not employ
surfactants like cetyltrimethylammonium bromide (CTAB, a
cationic surfactant),19 Triton X-100,32 or 3-aminopropyltrie-
thoxysilane (APTES)20 to assist suspension, or binders such as
Nafion19,24,25,29 to ensure adhesion of the FAD or GOx to the
electrode. Furthermore, GOx or FAD was not covalently
attached to the CNTs with carbodiimide coupling,22,32
dispersed in an immobilizing film,27 or constructed from
layer-by-layer assembly with cationic films such as polyethyle-
C
or modified with enzymes such as glucose oxidase (GOx).5−8
The ideal enzymatic biosensor would be a so-called “third-
generation” biosensor, where the electrode would have direct
electrical access to the redox-active center of a functioning
enzyme. Given that the protein shell is designed to protect the
redox-active center and impart selectivity, only a few enzymes
naturally exhibit this behavior on traditional electrode
surfaces;9−11 however, novel methods have been employed to
facilitate electron transfer between electrode surfaces and other
enzymes.12−17 CNTs have been noted as an ideal material for
direct electron transfer (DET) due to their small size and
excellent electronic conductivity.18 The literature contains
many reports of DET between CNT electrodes and the
enzyme GOx.18−32 Flavin adenine dinucleotide (FAD), the
redox-active center of GOx, is tightly bound inside a deep
pocket of GOx, but is not covalently bound.33 A hypothesis of
the DET process between CNTs and GOx is that the small
diameter CNTs are able to penetrate into the protein or
glycoprotein shell deep enough for electron tunneling to
occur.18 The electrochemical behavior of FAD has been studied
on electrode materials such as mercury,34−36 glassy carbon
Received: September 26, 2012
Accepted: January 5, 2013
Published: January 5, 2013
© 2013 American Chemical Society
1571
dx.doi.org/10.1021/ac3028036 | Anal. Chem. 2013, 85, 1571−1581

Analytical Chemistry
Article
nimine (PEI).23 Our approach provides a more accurate
understanding of the natural physical processes behind GOx
and FAD adsorption and the influence of surface adsorption on
measured interfacial electron-transfer rates. Conclusively, the
electroactive FAD observed from the spontaneous adsorption
of GOx onto CNT and N-CNT electrodes is not associated
with active enzyme, as bioelectrocatalytic current was not
observed upon the addition of glucose, thereby demonstrating
that DET is not occurring with GOx. FAD does, however,
rapidly adsorb at CNT/N-CNT electrodes, even at low
concentrations, to produce a significant and reversible redox
response. The commonly used Laviron method, which takes
the anodic and cathodic potential peak-to-peak splitting as a
way to determine the heterogeneous electron-transfer rate
constant is also investigated, displaying the need to fully
determine whether the measured potential peak splitting is
under the kinetic control of an electron transfer, or ohmic
control due to uncompensated resistance.
surface was polished with 0.05 μm alumina slurry on microcloth
(Buehler) and sonicated in 18.2 MΩ cm water prior to CNT/
N-CNT application. After drop casting, CNT/N-CNT electro-
des were allowed to fully dry in air before “wetting” the
electrode surface in a mixture of sodium phosphate buffer
(SPB) and ethanol. CNT/N-CNT electrodes were cycled in
SPB (from 0.00 to −1.20 V vs Hg/Hg2SO4 100 times at 300
mV/s) to passivate electroactive iron prior to use.49 Electrodes
were then placed in an inverted position in SPB solutions
containing varying concentrations of FAD or GOx. CVs were
taken after specific exposure times to the FAD and GOx
solutions (adsorption was always performed at open circuit
potential). Electrodes were rinsed with the appropriate
concentration of SPB to remove loosely adsorbed FAD or
GOx prior to electrochemical measurements. Voltammograms
were obtained by initially sweeping the potential negative
(usually starting at 0.00 V vs Hg/Hg2SO4) before reversing
back in a positive direction. The amount of FAD adsorbed was
calculated from integration of the cathodic peak (coulometri-
cally). A five-neck 125 mL glass cell was outfitted with a Hg/
Hg2SO4 reference electrode (CH Instruments, +0.64 V vs SHE;
+0.40 vs SCE; +0.44 vs Ag/AgCl) and a coiled Au wire counter
electrode for all electrochemical experiments, in conjunction
with an Autolab PGSTAT30 potentiostat interfaced with
Autolab GPES software (version 4.9). Electron micrographs
of the CNT/N-CNTs electrodes were obtained using a FEI
Quanta 650 FEG with secondary electrons (electron energy of
30 keV).
EXPERIMENTAL SECTION
■
Chemicals. Sodium phosphate monobasic (NaH2PO4),
sodium phosphate dibasic (Na2HPO4), sodium hydroxide
(NaOH), potassium nitrate (KNO3), o-phosphoric acid
(85%), and pyridine were purchased from Fisher (all ACS
grade). Bis(cyclopentadienyl)iron (ferrocene, 99%) was
obtained from Alfa Aesar. Flavin adenine dinucleotide disodium
salt hydrate (≥95%), glucose oxidase (Type X-S from
Aspergillus niger 100 00−250 00 U/g), α-D-glucose, and m-
xylene were purchased from Sigma-Aldrich.
RESULTS AND DISCUSSION
Spontaneous Adsorption of GOx and FAD onto CNT/
N-CNT Electrodes. Figure 1 presents electron micrographs of
■
CNT, N-CNT Synthesis. CNTs and N-CNTs were prepared
by chemical vapor deposition (CVD) using ferrocene dissolved
in either m-xylene (nondoped) or pyridine (N-doped) at a
concentration of 20 mg mL−1 as described in previous
reports.45−48 Approximately 1 mL of the ferrocene solution
was fed into a single quartz tube (at 0.1 mL min−1) through a
programmable syringe pump (New Era Pump Systems NE-
1000) fitted with a glass syringe (Hamilton 81320). The quartz
tube was placed lengthwise across two separate tube furnaces
(Carbolite model HST 12/35/200/2416CG), the first of which
was set at a temperature to properly vaporize the organic
solvent (150 °C for m-xylene and 130 °C for pyridine), while
the second was set to ensure formation of multiwalled CNTs/
N-CNTs (700 °C m-xylene and 800 °C for pyridine) on the
inside lining of the quartz tube. Argon gas was used to carry the
vaporized ferrocene solution through the tube along with
hydrogen (for nondoped CNTs) or ammonia to increase the
amount of nitrogen incorporated into the carbon lattice. The
amount of incorporated nitrogen as a function of the flow rate
of ammonia was shown in a prior report.46
Figure 1. Scanning electron micrographs of the CNT/N-CNT
electrodes: (A) nondoped CNTs, (B) 4.0 atom % N-CNTs, (C) 7.4
atom % N-CNTs.
Electrochemistry and Electrode Preparation. Cyclic
voltammograms (CVs) and chronoamperograms (CAs) are
presented with cathodic current being assigned a positive value
and the anodic current given a negative value. CNT/N-CNT
electrodes were prepared by drop casting 24 μg of CNTs/N-
CNTs on a 0.5 cm diameter glassy carbon (PINE Instruments
AFE2M050GC) electrode surface, without the use of binders,
surfactants, or oxidizing agents. CNT/N-CNT solutions were
sonicated (Branson ultrasonic cleaner, model 2510R-MTH) for
2 h in absolute ethanol prior to drop casting to ensure a
homogeneous mixture. The mass-normalized 24 μg were drop
cast from a single 12 μL aliquot of a 2 mg mL−1 solution for N-
CNTs, while six aliquots of 10 μL were drop cast from a
nondoped CNT solution of 0.4 mg mL−1. The glassy carbon
the CNTs/N-CNTs after sonication and solution drop casting
on a GC electrode. All three morphologies are similar,
composed of a three-dimensional mesh of CNTs or N-
CNTs; however, the 7.4 atom % N-CNTs still contain a
substantial amount of bundled tubes. In addition to the many
attractive physicochemical properties of CNTs, three-dimen-
sional networks of CNTs can greatly enhance the spontaneous
adsorption of adsorbates, especially in the case of enzymes.50
Figure 2 shows CVs of CNT and N-CNT electrodes at
increasing exposure times to solutions containing 6.5 μM FAD
or 81.3 μM GOx in 0.1 M SPB at a pH of 7.00. Electrodes and
solutions were kept in the dark at all times, except during
1572
dx.doi.org/10.1021/ac3028036 | Anal. Chem. 2013, 85, 1571−1581

Analytical Chemistry
Article
Figure 2. CVs of 6.5 μM FAD and 81.3 μM GOx in 0.1 M SPB pH 7.00 adsorbing onto CNTs/N-CNTs over 1 h (black = background, red = 1 min,
orange = 5 min, yellow = 10 min, green = 15 min, aqua = 30 min, light blue = 45 min, blue = 60 min) at a constant scan rate (100 mV/s).
electrochemical measurements (∼3 min), to ensure that
photolytic reactions did not occur.51 The observed redox
reaction is due to a surface-confined electron transfer of FAD
for both the adsorbed FAD and GOx via eq 1:
functionalities, decreases hydrophobicity, and creates more
edge-plane sites for facile electron transfer.52
The background capacitive current increases with nitrogen
doping of CNTs, as seen in the increase of nonfaradaic current
in the CVs of Figure 2. Adsorption of GOx causes the
background capacitive current of N-CNTs to decrease with
increasing amounts of adsorbed GOx. For nondoped CNTs,
the initial adsorption of GOx appears to impart an increased
capacitive current, suggesting that GOx interacts differently at
the more hydrophobic CNT electrodes. In contrast, adsorbed
FAD does not appreciably change the background capacitive
current for either CNT or N-CNT electrodes. The rates of
adsorption were found to be linear during the first 15 min for
FAD, but not as monotonic for GOx. Initial adsorption rates
are shown in Table 1. These values display an increase in the
FAD + 2H+ + 2e− ↔ FADH2
(1)
The E1/2 or formal potential of this reaction resides at −0.85 V
for the N-CNT electrodes and −0.86 V for the CNT
electrodes. Figure S-1 in the Supporting Information displays
overlaid CVs of FAD adsorbed at nondoped CNT, 4.0 atom %
N-CNT, and 7.4 atom % N-CNT modified electrodes, showing
the 10 mV more negative shift in the E1/2 of FAD on nondoped
CNTs. The difference between E1/2 at CNTs and N-CNTs is
reflective of the nitrogen incorporated into the carbon lattice,
which subsequently produces positively charged nitrogen
1573
dx.doi.org/10.1021/ac3028036 | Anal. Chem. 2013, 85, 1571−1581

Analytical Chemistry
Article
GOx, whereby the electroactive amount adsorbed will change
over time.
Table 1. Initial Adsorption Rates of FAD and GOx onto
CNTs/N-CNTs
The FAD adsorption maximum increases concurrently with
concentration (shown in the Supporting Information as Figure
S-3B), indicating Langmuir adsorption behavior. This effect is
not large between the concentrations used, and deviations in
the amount of CNTs drop cast on the electrode can confound
the trend (shown in the Supporting Information as Figure S-
3A); however, if the same electrode is placed in increasing
concentrations of FAD, the adsorption maximum also increases.
Previous studies of FAD adsorbed onto TiO242 and GC37 have
also shown that the FAD adsorption process displays Langmuir
isotherm behavior.
Type
FAD (pmol/min)
GOx (pmol/min)
nondoped CNTs
5.0
6.6
7.6
1.1
2.9
3.6
4.0 atom % N-CNTs
7.4 atom % N-CNTs
adsorption rate as a function of nitrogen doping, which is
reflective of the increased surface area of the N-CNTs and the
presence of more edge-plane sites which can promote stronger
and more facile adsorption.47,52 Figure S-2 of the Supporting
Information presents the amount adsorbed from Figure 2
extended for 45 h. The adsorption of FAD on CNT and N-
CNT electrodes appears to reach a constant maximum around
25 h (at 6.5 μM FAD in 0.1 SPB). The amount of GOx
adsorbed also reaches a maximum, but begins to decrease from
that maximum within the measured time frame. We believe this
observed maximum for FAD and GOx is in fact a dynamic
equilibrium based on the concentration of adsorbing species in
solution and the available surface sites. In the case of FAD, the
equilibrium is stable since any desorbing FAD will be quickly
replaced by another FAD from solution. In the case of GOx, the
desorbing FAD molecules can be replaced by nonelectroactive
Effect of Ionic Strength. For a surface-confined redox
reaction, the peak current should scale linearly with the scan
rate according the equation (eq 2) below:53
ip = vΓA[n2F2/(4RT)]
(2)
where ip is the peak current (A), v is the scan rate (V s−1), Γ is
the surface concentration (mol cm−2), A is the electroactive
surface area of the electrode (cm2), n is the number of electrons
transferred in the redox reaction, F is Faraday’s constant (96
485 C), R is the gas constant (8.314 J mol−1 K−1), and T is the
temperature (K). The ionic strength of the supporting
Figure 3. Observed surface-confined redox reaction of FAD adsorbed onto 4.0 atom % N-CNTs as a function of scan rate and concentration of the
supporting electrolyte (SPB pH 7.00; red = 0.1 M, blue = 0.5 M, black = 1.0 M).
1574
dx.doi.org/10.1021/ac3028036 | Anal. Chem. 2013, 85, 1571−1581

Analytical Chemistry
Article
Figure 4. CVs of 6.5 μM FAD adsorbing onto 4.0 atom % N-CNTs in 0.1 M pH 7.00 (A) and 1.0 M pH 6.75 (C) SPB at increasing time points
(scan rate 100 mV/s). The coordinating amount adsorbed (as calculated from the integration of the cathodic peak) is shown for each CV time point
on the right (B and D).
electrolyte, however, plays a large role in determining the scan
rates at which a linear relationship is observed. A series of three
different ionic strengths are shown in Figure 3, at varying scan
rates.
strength solutions. When FAD is adsorbed in 0.1 M SPB and
transferred into higher ionic strength solutions, the amount
adsorbed stays relatively constant. When FAD is adsorbed in
1.0 M SPB and transferred to lower ionic strengths, the amount
adsorbed decreases. It should also be noted that the measured
ΔEp is significantly smaller in 1.0 M SPB in comparison to
values measured in 0.1 M SPB (Supporting Information Table
S-4).
At lower scan rates the ionic strength of the supporting
electrolyte plays less of a role in influencing the magnitude of
the peak current; however, a difference is still observed. At high
scan rates and low ionic strengths, the peak potential positions
are shifted (peak-to-peak potential splitting gets larger) and
peak broadening occurs. It should be noted that, although the
peak-to-peak cathodic and anodic potential splitting (ΔEp) gets
larger at higher scan rates and lower ionic strengths, the E1/2 or
formal potential remains essentially constant. Although depend-
ent on the amount adsorbed, the limit of the linear relationship
between the peak current and scan rate was determined to be
150 mV/s for 0.1 M, 250 mV/s for 0.5 M, and 400 mV/s for
1.0 M SPB.
The ionic strength also affects the maximum or equilibrium
amount of FAD adsorbed onto the CNT/N-CNT surface.
Analogous to a hydrophobic protein purification column, more
FAD is adsorbed at higher ionic strengths, indicating that
hydrophobic interactions play a role in the adsorption of FAD.
This effect can be seen in Figure 4, parts A and C, which display
CVs of 4.0 atom % N-CNT electrodes at increasing exposure
times to a 6.5 μM solution of FAD in 0.1 or 1.0 M SPB. Parts B
and D of Figure 4 show the coordinating amount adsorbed as
calculated from integration of the cathodic peak.
The full width of the FAD peak at half the maximum (fwhm)
peak current on nondoped CNTs during early adsorption times
in 1.0 M SPB and a scan rate of 100 mV/s is close to 45 mV.
The predicted fwhm for an ideal nerstian reaction under
Langmuir conditions (90.6 mV/n, where n is a two-electron
transfer process associated with eq 1) is also 45 mV.53 Although
deviations from ideal behavior such as lateral interactions,53−55
inhomogeneity of adsorbate populations or surface sites,54,56
and ion interactions57 can drastically affect peak shape, we note
that the ideal fwhm occurs at low surface coverages, where
intermolecular forces are minimal and initial adsorption sites
are most likely identical. Additionally, the fwhm increases
concurrently with increasing amounts of the adsorbing species
(from 45 mV at low surface coverages to 95 mV at the
maximum coverage), while the formal potential remains fairly
constant. We cannot rule out, however, that the ideal fwhm is a
composite of the aforementioned interactions. Adsorbed FAD
on N-CNTs exhibit larger than ideal fwhm values due to the
inhomogeneity of surface adsorption sites as nitrogen disrupts
the carbon lattice.
Although the concentration of FAD is the same in each case,
the maximum amount adsorbed is significantly higher in 1.0 M
SPB as opposed to 0.1 M SPB. This effect can be seen more
clearly when CVs are taken as an FAD-loaded CNT/N-CNT
electrode is transferred into increasing or decreasing ionic
Laviron’s Method for Determining the Heteroge-
neous Electron-Transfer Rate Constant and Uncompen-
sated Resistance. In almost every reported heterogeneous
electron-transfer rate constant (ks) between GOx and
1575
dx.doi.org/10.1021/ac3028036 | Anal. Chem. 2013, 85, 1571−1581

Analytical Chemistry
Article
Figure 5. CVs of FAD on 4.0 atom % N-CNTs at increasing scan rates (panel A, 25−1000 mV/s) and during adsorption at a constant scan rate
(panel C, 100 mV/s). The Ep,c vs ip,c plots (B and D) present the measured resistance as a function of scan rate (B) or amount adsorbed (D).
CNTs18,19,21,22,24,26,27,29−32 the authors have employed Lavir-
on’s method to determine ks.58 This method uses the cathodic
and anodic peak potential splitting, ΔEp, as a way to determine
ks, which should be a function of the scan rate. Although this
method is analytically sound, the uncompensated resistance
(Ru) is not taken into account, and this aspect can drastically
affect the outcome of observed measurements. In fact, Laviron’s
method is only valid if the electron-transfer reaction (i.e.,
kinetic) is controlling ΔEp. Figure 5 displays CVs of FAD
adsorbed onto 4.0 atom % N-CNTs at different scan rates
(Figure 5A) and during adsorption at a constant scan rate
(Figure 5C) in 1.0 M SPB pH 6.75. At a constant scan rate, the
ΔEp should be independent of the amount adsorbed and Ep,c
should concurrently remain constant.59 Figure 5D displays a
shift in the cathodic peak potential (Ep,c) as a function of the
cathodic peak current (ip,c), which is increasing concurrently
with the amount adsorbed. Using Ohm’s law (V = iR), the
slope of Figure 5D is a measure of the resistance due to the
increase in peak current as FAD is increasingly adsorbed. Figure
5B shows an identical plot, but of the Ep,c shifting as the scan
rate is changed. Interestingly, the slope or measured resistance
in both plots are identical (67 Ω), indicating that resistance is
controlling ΔEp, and not the kinetics of heterogeneous electron
transfer.
“nonkinetic” peak splitting contribution is subtracted out.31
Furthermore, our ks was measured in 1.0 M SPB, but the
majority of ks were determined in 0.1 M solutions, where the
ionic strength was already shown to play a large role in the
observed electrochemical behavior and peak splitting. Table S-4
in the Supporting Information presents the absorption time,
ΔEp, and calculated ks (using Laviron’s method) for each CV
time point shown in Figure 4.
CVs of FAD display a peak splitting (ΔEp) which is
independent of scan rate or amount adsorbed. This ΔEp is
approximately 10 mV, which may be explained by an N-shaped
free energy curve.60 In this case, FAD is expected to undergo a
structural change, where the oxidized form is planar and the
reduced form resembles a butterfly conformation.35 This
structural change is most likely the reason for a persistent
ΔEp, even though it should ideally be zero at slow scan rates.
The Ep,c versus ip,c plots at varying scan rates are dependent
on the ratio of adsorbed FAD current to the background
capacitive current. The Supporting Information displays
measured resistances as a function of the ratio of the FAD
current to the background current on 4.0 atom % N-CNTs in
0.1 SPB pH 7.00 (Supporting Information Table S-5).
Unusually high resistances are measured when the capacitive
current is dominating the overall current, but as more FAD is
adsorbed, the slope eventually matches the slope of Ep,c versus
ip,c plots where FAD is increasingly adsorbed at a constant scan
rate. Resistances converge to 56.6 13.5 Ω for 1.0 M SPB pH
6.75 and 266 52 Ω for 0.1 M SPB pH 7.00. Interestingly, the
measured resistance provides an excellent estimate of the
uncompensated resistance (Ru), which can be matched up to
the universal working curve presented by Feldberg.59 Feldberg’s
At nondoped CNTs in 1.0 M SPB and at low surface
coverages of FAD, we observe a rate constant of 7.6 s−1 using
Laviron’s method. Since this value is under ohmic control, this
is merely a lower limit, with the actual ks being much faster.
Given this measurement, it should be noted that every cited ks
is slower than this value, except when FAD is covalently
attached to the ends of aligned single walled CNTs,30 or the
1576
dx.doi.org/10.1021/ac3028036 | Anal. Chem. 2013, 85, 1571−1581

Analytical Chemistry
Article
Figure 6. CVs of FAD and GOx adsorbed onto (A) nondoped CNT and (B) 7.4 atom % N-CNT modified electrodes (scan rate 100 mV/s 0.1 M
SPB pH 7.00). Rotating disk chronoamperograms (1000 rpm) of FAD and GOx adsorbed onto nondoped CNT and 7.4 atom % N-CNT electrodes,
where 10 mM of Ar-purged glucose is injected into the solution at 300 s (Ar is continuously bubbled in the 0.1 M SPB) while the potential is held at
the open circuit potential (C and D) or the formal potential (E and F).
working curve models the ideal behavior of a surface-bound
redox system with finite Ru. Since our data aligns with
Feldberg’s working curve when our measured resistances are
used in place of Ru, we can reasonably assume that our
measured resistances are a good estimate of Ru. Furthermore, it
should also be noted that in 1.0 M SPB at high scan rates
(500−40 000 mV/s), FAD adsorbed onto a roughened GC
surface or an edge-plane pyrolytic graphite electrode (PINE
AFE2M050GE) display peak shifts whose Ep,c versus ip,c slopes
of 62.7 12.3 Ω correlate well with those measured on CNT/
N-CNT electrodes. This indicates that the measured resistances
from the Ep,c versus ip,c plots most likely represent solution
resistance, since neither film resistance (the CNT/N-CNT film
was removed) nor the electron-transfer resistance (the peak-to-
peak splitting should have remained constant at a constant scan
rate during adsorption) is playing a major role in the measured
resistance. The calculated ks for FAD on GC was measured as
high as 30.7 s−1 (Laviron’s method), suggesting that this value
may still be slower than the actual rate constant.
GOx Activity and Bioelectrocatalysis. In the presence of
GOx and oxygen, glucose is oxidized by GOx into D-glucono-
1,5-lactone while oxygen is reduced by GOx to H2O2. The
product D-glucono-1,5-lactone is subsequently hydrolyzed in
aqueous solutions into D-gluconic acid. Both reactions are
shown below (eqs 3 and 4):
β‐
D
‐glucose + O2 → ‐glucono‐1, 5‐lactone + H2O2
D
(3)
(4)
D‐glucono‐1, 5‐lactone + H2O ↔ D‐gluconic acid
The cofactor FAD is the redox reaction center of GOx, where
two protons and two electrons are transferred from glucose.
1577
dx.doi.org/10.1021/ac3028036 | Anal. Chem. 2013, 85, 1571−1581

Analytical Chemistry
Article
Figure 7. CVs of GOx adsorbed onto nondoped CNTs and 7.4 atom % N-CNTs with increasing concentrations of Ar-purged glucose (black = 0
mM, red = 10 mM, orange = 20 mM, green = 30 mM, blue = 50 mM, purple = 75 mM) in the absence (A and B) and presence (E and F) of 0.5 mM
1,4-hydroquinone and 0.5 mM 1,4-benzoquinone. FAD is used instead of GOx in panels C and D as a non-glucose-active control (also in the
presence of the redox mediator). Scan rate is 25 mV/s for all the CVs presented, 0.1 M SPB pH 7.00.
GOx(FADH2) → GOx(FAD) + 2H+ + 2e−
(anodic current at the electrode)
Under normal physiological conditions, oxygen would be
reduced by acquiring two protons and two electrons from
(6)
FADH2, reoxidizing the active center, thereby allowing GOx to
turnover another glucose molecule. In the absence of O2 or
another appropriate electron mediator, only the electrode
would be able oxidize FADH2 to FAD, subsequently producing
an anodic current at the electrode surface. The electrode
turnover scheme, known as bioelectrocatalysis, is shown below
(eqs 5 and 6):
The main test for DET between enzyme-active FAD and CNTs
is the appearance of anodic current when glucose is added in an
oxygen-free environment. The adsorbed GOx was tested for
bioelectrocatalysis by rotating disk chronoamperometry where,
in an Ar-saturated buffer (continuously bubbled Ar), the
potential was held at specific points that would sufficiently
oxidize FADH2 back to FAD, shown in Figure 6C−F. Figure 6,
parts A and B, display CVs of GOx and FAD adsorbed onto
CNT and 7.4 atom % N-CNT electrodes. The dotted blue lines
indicate the open circuit potential (−0.37 V for CNTs and
GOx(FAD) + β‐D‐glucose
→ GOx(FADH2) + D‐glucono‐1, 5‐lactone
(5)
1578
dx.doi.org/10.1021/ac3028036 | Anal. Chem. 2013, 85, 1571−1581

Analytical Chemistry
Article
−0.18 V for 7.4 atom % N-CNTs) and the formal potential
(−0.86 V for CNTs and −0.85 V for 7.4 atom % N-CNTs) for
adsorbed FAD/GOx. CNT and 7.4 atom % N-CNT electrodes
with adsorbed FAD were used as a non-glucose-active
background.
enzymatic activity as a function of the glucose concentration in
the presence of H2Q/Q.
A second more interesting way of verifying the activity of
adsorbed GOx on the surface of N-CNTs is by using the
natural electron mediator oxygen. The ORR at N-CNTs occurs
via a peroxide pathway, where H2O2 created from oxygen
reduction will rapidly disproportionate on the N-CNT surface
to create a local increase in oxygen.45,48 This unique pathway
allows for detection of glucose (through enzymatically
generated H2O2) at close to the open circuit potential of the
electrode, minimizing electroactive interferences and back-
ground processes. We have previously demonstrated that this
electrochemical detection scheme can be applied to GOx62 or
other enzymes, such as lactate oxidase,61 which also produce
H2O2 as a byproduct. We note that in our paper describing the
detection of L-lactate via lactate oxidase and N-CNTs,61 we
mistakenly assigned the observed FAD signal (FAD is also the
active center of lactate oxidase) to DET. The aforementioned
analysis has shown that the observed FAD signal was not DET,
but simply electroactive FAD derived from inactive enzyme.
Figure 8 shows rotating disk chronoamperometric data of a 7.4
Rotating disk chronoamperograms held at the open circuit
potential display lower noise and less background drift than
those held at the formal potential; however, anodic current is
not observed at either electrode, at either potential held.
Besides a small mixing artifact when glucose is added at 300 s,
there is virtually no change in the chronoamperograms. The
Supporting Information displays the exact same experiments,
but using a glucose solution that was not Ar-purged
(Supporting Information Figure S-6). Oxygen reduction plays
a major role, even in minute amounts, as seen by the large
reduction currents when glucose is injected into the solution at
the formal potential (Supporting Information Figure S-6, parts
E and F). The oxygen reduction reaction (ORR) occurs at
nondoped CNTs and 7.4 atom % N-CNTs at −0.81 and −0.30
V, respectively.61 Since the formal potential of FAD on
nondoped CNTs and 7.4 atom % N-CNTs occurs more
negative of the ORR, it is clear why such a pronounced oxygen
reduction current is observed. At nondoped CNTs, the ORR
potential is far enough away from the open circuit potential that
no change is observed when glucose is added (Supporting
Information Figure S-6C); however, at 7.4 atom % N-CNTs,
the open circuit potential is near the onset of the ORR
(Supporting Information Figure S-6D). In addition, the ORR
occurs via a different pathway on N-CNTs,45,48 so the small
amount of injected oxygen that is enzymatically transformed
into H2O2 subsequently disproportionates into more oxygen at
N-CNTs, increasing the reductive component of the back-
ground current.61,62
Figure 7 shows CVs of nondoped CNT and 7.4 atom % N-
CNT electrodes with adsorbed GOx as Ar-purged glucose is
added to the solution (Figure 7, parts A and B). If DET occurs,
the glucose should spontaneously reduce FAD, causing the
reduction peak to decrease, and the oxidation peak to increase.
No change is observed in the CVs as the glucose concentration
is increased (Figure 7, parts A and B), except for a slow
desorption of GOx from the nondoped CNTs during the time
frame of the measurement. Even though no change is observed,
GOx is still active and can be verified if an electron mediator is
present or if the natural electron mediator oxygen is present.
Figure 7 also displays CVs of GOx adsorbed onto nondoped
CNT and 7.4 atom % N-CNTs when an electron mediator, 1,4-
hydroquinone/1,4-benzoquinone (H2Q/Q), is added to the
SPB solution (Figure 7, parts E and F). When the H2Q/Q
redox couple is present, an increase in the anodic wave for the
H2Q/Q couple appears when glucose is added to the solution.
This anodic wave is the result of the enzymatic turnover of
glucose, whereby the reduced FADH2 is reoxidized by Q to
FAD and Q is reduced to H2Q. The cathodic wave of the
quinone couple goes down, while the anodic peak is increased
concurrently with the glucose concentration, indicative of the
enzyme turnover rate. Note that the FAD peak (potential or
current) does not change, thus giving more evidence that DET
is not occurring between GOx and CNTs. Wang and Yao have
also shown that the widely reported DET between GOx and
CNTs is without value in certain mediator-free applications.63
Parts C and D of Figure 7 display FAD adsorbed onto
nondoped CNTs and 7.4 atom % N-CNTs, which do not show
Figure 8. Rotating disk chronoamperogram of GOx adsorbed onto 7.4
atom % N-CNTs as 0.5 mM injections of glucose are added to an
oxygen-saturated solution or an argon-saturated solution (Ar or O2
constantly bubbled, 0.1 M SPB pH 7.00, rotating rate 1000 rpm).
atom % N-CNT electrode with GOx adsorbed on the surface in
an oxygen-saturated or an argon-saturated 0.1 M SPB pH 7.00.
When aliquots of 0.5 mM glucose are injected into the solution,
a clear reduction current response is observed, directly
proportional to the glucose concentration in the oxygen-
saturated solution, while no response is observed in the argon-
saturated solution. This glucose biosensor, without optimiza-
tion, displays a sensitivity of 0.011 A M−1 cm−2 (based upon the
geometric area of the GC electrode), a detection limit of 24
μM, and a linear range of 6.5 mM (based on R2 > 0.995) at
−0.23 V vs Hg/Hg2SO4 after GOx was allowed to adsorb onto
the 7.4 atom % N-CNTs at room temperature for 42 h.
CONCLUSION
■
FAD and GOx were spontaneously adsorbed onto CNTs and
N-CNTs without covalent attachment, oxidizing agents,
surfactants, or binders. The subsequent unimpeded physisorbed
interactions were investigated by cyclic voltammetry and
chronoamperomtery, which showed a dependence of the
amount adsorbed on the exposure time to the adsorbing
species, the concentration of adsorbate, and the ionic strength
of the adsorbate solution. CNT and N-CNT electrodes with
1579
dx.doi.org/10.1021/ac3028036 | Anal. Chem. 2013, 85, 1571−1581

Analytical Chemistry
Article
(16) Katz, E.; Willner, I. ChemPhysChem 2004, 5, 1084−1104.
adsorbed GOx were examined for DET, but anodic current
associated with the oxidation of glucose was not observed when
glucose was added to an oxygen-free solution. Adsorbed GOx is
active, clearly shown if an appropriate electron mediator is
present or the natural electron mediator oxygen is employed.
Since the surface-confined redox reaction is identical for
adsorbed FAD and GOx, the observed FAD peak from
adsorbed GOx is a result of FAD which is not enzymatically
active. The well-defined FAD redox peaks were used to
determine the heterogeneous electron-transfer rate constant
(ks) between FAD and CNTs/N-CNTs using Laviron’s
method. The measured value of 7.6 s−1 was determined to be
under ohmic control, thereby suggesting that the actual ks is
much faster.
(17) Patolsky, F.; Weizmann, Y.; Willner, I. Angew. Chem., Int. Ed.
2004, 43, 2113−2117.
(18) Guiseppi-Elie, A.; Lei, C.; Baughman, R. H. Nanotechnology
2002, 13, 559−564.
(19) Cai, C.; Chen, J. Anal. Biochem. 2004, 332, 75−83.
(20) Luong, J. H. T.; Hrapovic, S.; Wang, D.; Bensebaa, F.; Simard,
B. Electroanalysis 2004, 16, 132−139.
(21) Gutierrez, F.; Rubianes, M. D.; Rivas, G. A. Sens. Actuators, B
2012, 161, 191−197.
(22) Janegitz, B. C.; Pauliukaite, R.; Ghica, M. E.; Brett, C. M. A.;
Fatibello-Filho, O. Sens. Actuators, B 2011, 158, 411−417.
(23) Deng, C.; Chen, J.; Nie, Z.; Si, S. Biosens. Bioelectron. 2010, 26,
213−219.
(24) Ivnitski, D.; Branch, B.; Atanassov, P.; Apblett, C. Electrochem.
Commun. 2006, 8, 1204−1210.
(25) Deng, S.; Jian, G.; Lei, J.; Hu, Z.; Ju, H. Biosens. Bioelectron.
2009, 25, 373−377.
ASSOCIATED CONTENT
* Supporting Information
Four figures and two tables as noted in the text. This material is
■
S
(26) Deng, C.; Chen, J.; Chen, X.; Xiao, C.; Nie, L.; Yao, S. Biosens.
Bioelectron. 2008, 23, 1272−1277.
(27) Periasamy, A. P.; Chang, Y.-J.; Chen, S.-M. Bioelectrochem 2011,
80, 114−120.
available free of charge via the Internet at http://pubs.acs.org.
(28) Wang, Y.; Yuan, R.; Chaia, Y.; Li, W.; Zhuo, Y.; Yuan, Y.; Li, J. J.
Mol. Catal. B: Enzym. 2011, 71, 146−151.
AUTHOR INFORMATION
Corresponding Author
*E-mail: Stevenson@cm.utexas.edu.
■
(29) Zhao, H.-Z.; Sun, J.-J.; Song, J.; Yang, Q.-Z. Carbon 2010, 48,
1508−1514.
(30) Liu, J.; Chou, A.; Rahmat, W.; Paddon-Row, M. N.; Gooding, J.
J. Electroanalysis 2005, 17, 38−46.
Notes
The authors declare no competing financial interest.
(31) Vaze, A.; Hussain, N.; Tang, C.; Leech, D.; Rusling, J.
Electrochem. Commun. 2009, 11, 2004−2007.
(32) Jose, M. V.; Marx, S.; Murata, H.; Koepsel, R. R.; Russell, A. J.
Carbon 2012, 50, 4010−4020.
ACKNOWLEDGMENTS
■
Financial support of this work was provided by the R. A. Welch
Foundation (Grant F-1529). S.M.M. acknowledges support
from NSF-REU program (Grant CHE-1003947). We thank Dr.
Stephen Feldberg for helpful discussions on uncompensated
resistance and surface-confined redox reactions, as well as his
invaluable contribution to the scientific literature regarding
these matters. We also thank Professor Viola Birss for her
insight on the electrochemical behavior of FAD and her work
with FAD on mercury. Lastly, we thank Dr. Sankaran
Murugesan for helping us obtain SEM images of the CNT/
N-CNT electrodes.
(33) Wilson, R.; Turner, A. P. F. Biosens. Bioelectron. 1992, 7, 165−
185.
(34) Honeychurch, M. J.; Ridd, M. J. Electroanalysis 1996, 8, 362−
369.
(35) Birss, V. I.; Elzanowska, H.; Turner, R. A. Can. J. Chem. 1988,
66, 86−96.
(36) Kamal, M. M.; Elzanowska, H.; Gaur, M.; Kim, D.; Birss, V. I. J.
Electroanal. Chem. 1991, 318, 349−367.
(37) Wei, H.; Omanovic, S. Chem. Biodiversity 2008, 5, 1622−1639.
(38) Verghagen, M. F. J. M.; Hagen, W. R. J. Electroanal. Chem. 1992,
334, 339−350.
(39) Wei, H. Z.; Van de Ven, T. G. M.; Omanovic, S.; Zeng, Y. W.
Langmuir 2008, 24, 12375−12384.
REFERENCES
■
(40) Wang, Y.; Zhu, G.; Wang, E. Anal. Chim. Acta 1997, 97−101.
(41) Garjonyte, R.; Malinauskas, A.; Gorton, L. Bioelectrochemistry
2003, 61, 39−49.
(1) Musameh, M.; Wang, J.; Merkoci, A.; Lin, Y. Electrochem.
Commun. 2002, 4, 743−746.
(2) Wang, J.; Musameh, M. Anal. Chem. 2003, 75, 2075−2079.
(3) Jacobs, C. B.; Peairs, M. J.; Venton, B. J. Anal. Chim. Acta 2010,
662, 105−127.
(42) Shinohara, H.; Gratzel, M. Bioelectrochem. Bioenerg. 1991, 26,
307−320.
(43) Gorton, L.; Johansson, G. J. Electroanal. Chem. 1980, 113, 151−
158.
(4) Vashist, S. K.; Zheng, D.; Al-Rubeaan, K. Biotechnol. Adv. 2011,
29, 169−188.
(44) Miyawaki, O.; Wingard, L. B. J. Biotechnol. Bioeng. 1984, 26,
(5) Lin, Y.; Yantasee, W.; Wang, J. Front. Biosci. 2005, 2, 492−505.
1364−1371.
́ ́
(6) Yanez-Sedeno, P.; Pingarron, J. M.; Riu, J.; Rius, F. X. TrAC,
̃ ̃
(45) Maldonado, S.; Stevenson, K. J. J. Phys. Chem. B 2005, 109,
4707−4716.
Trends Anal. Chem. 2010, 29, 939−953.
(7) Lin, Y.; Lu, F.; Tu, Y.; Ren, Z. Nano Lett. 2004, 4, 191−195.
(8) Balasubramanian, K.; Burghard, M. Anal. Bioanal. Chem. 2006,
385, 452−468.
(9) Gorton, L.; Lindgren, A.; Larsson, T.; Munteanu, F. D.; Ruzgas,
T.; Gazaryan, I. Anal. Chim. Acta 1999, 400, 91−108.
(10) Yaropolov, A. I.; Tarasevich, M. R.; Varfolomeev, S. D.
Bioelectrochem. Bioenerg. 1978, 5, 18−24.
(11) Gorton, L.; Jonsson-Pettersson, G.; Csoregi, E.; Johansson, K.;
Dominguez, E.; Marko-Varga, G. Analyst 1992, 117, 1235.
(12) Heller, A. Acc. Chem. Res. 1990, 23, 128−134.
(13) Degani, Y.; Heller, A. J. Am. Chem. Soc. 1989, 111, 2357−2358.
(14) Heller, A. J. Phys. Chem. 1992, 96, 3579−3587.
(15) Zayats, M.; Willner, B.; Willner, I. Electroanalysis 2008, 20, 583−
601.
(46) Maldonado, S.; Morin, S.; Stevenson, K. J. Carbon 2006, 44,
1429−1437.
(47) Wiggins-Camacho, J. D.; Stevenson, K. J. J. Phys. Chem. C 2009,
113, 19082−19090.
(48) Wiggins-Camacho, J. D.; Stevenson, K. J. J. Phys. Chem. C 2011,
115, 20002−20010.
(49) Lyon, J. L.; Stevenson, K. J. Langmuir 2007, 23, 11311−11318.
(50) Nejadnik, M. R.; Deepak, F. L.; Garcia, C. D. Electroanalysis
2011, 23, 1462−1469.
(51) Birss, V. I; Guha-Thakurta, S.; McGarvey, C. E.; Quach, S.;
Vanysek, P. J. Electroanal. Chem. 1997, 423, 13−21.
(52) Maldonado, S.; Morin, S.; Stevenson, K. J. Analyst 2006, 131,
262−267.
1580
dx.doi.org/10.1021/ac3028036 | Anal. Chem. 2013, 85, 1571−1581

Analytical Chemistry
Article
(53) Bard, A. J.; Faulkner, L. R. Electrochemical Methods:
Fundamentals and Applications, 2nd ed.; John Wiley and Sons: New
York, 2001.
(54) Laviron, E. J. Electroanal. Chem. 1979, 100, 263−270.
(55) Laviron, E. J. Electroanal. Chem. 1974, 52, 395−402.
(56) Clark, R. A.; Bowden, E. F. Langmuir 1997, 13, 559−565.
(57) Stiles, R. L.; Balasubramanian, R.; Feldberg, S. W.; Murray, R.
W. J. Am. Chem. Soc. 2008, 130, 1856−1865.
(58) Laviron, E. J. Electroanal. Chem. 1979, 101, 19−28.
(59) Feldberg, S. W. J. Electroanal. Chem. 2008, 624, 45−51.
(60) Feldberg, S. W.; Rubinstein, I. J. Electroanal. Chem. 1988, 240,
1−15.
(61) Goran, J. M.; Lyon, J. L.; Stevenson, K. J. Anal. Chem. 2011, 83,
8123−8129.
(62) Lyon, J. L.; Stevenson, K. J. ECS Trans. 2009, 16, 1−12.
(63) Wang, Y.; Yao, Y. Microchim. Acta 2012, 176, 271−277.
NOTE ADDED AFTER ASAP PUBLICATION
■
This paper was published on the Web on January 18, 2013.
Revisions were made to Table S-4 in the Supporting
Information, and the corrected version was reposted on January
22, 2013.
1581
dx.doi.org/10.1021/ac3028036 | Anal. Chem. 2013, 85, 1571−1581