electrochemical characterization of multi-walled carbon nanotube coated electrodes for biological...

10
Electrochemical characterization of multi-walled carbon nanotube coated electrodes for biological applications Saugandhika Minnikanti a , Perry Skeath b , Nathalia Peixoto a, * a Neural Engineering Laboratory, Department of Electrical and Computer Engineering, Volgenau School for Information Technology and Engineering, George Mason University, Fairfax, VA 22030, United States b Naval Research Laboratories, Optical Sciences Division, Code 5675, 4555 Overlook Avenue SW, Washington, DC 20375, United States ARTICLE INFO Article history: Received 11 August 2008 Accepted 22 November 2008 Available online 7 December 2008 ABSTRACT Carbon nanotubes, if used as a coating film on a conductive substrate, can substantially raise the charge storage capacity and lower the impedance of electrodes without significant increases to the geometric area. This is especially interesting in the case of stimulation of nervous tissue. We design implantable electrodes targeted at wide frequency stimulation of deep brain structures. Here we report on results in vitro with multi-walled carbon nano- tubes coatings applied onto stainless steel substrates using direct current electrophoresis. We experimentally demonstrate, through electrochemical techniques such as cyclic vol- tammetry and impedance spectroscopy, the enhanced performance of multi-walled carbon nanotube (MWCNT) coatings for implantable electrodes by contrasting our experimental results against the more traditional stainless steel substrate characteristics. We also inves- tigate surface morphology of aged electrodes. The interest in aged electrodes is dual fold: implantable electrodes have to be mechanically stable and present high shelf life. On the other hand, chemical modifications of the surface should be characterized. The effect of superficial oxygen adsorption on the aged MWCNT electrodes is observed through a mod- ified cyclic voltammetric spectrum, but not through any changes in impedance spectroscopy. Ó 2008 Elsevier Ltd. All rights reserved. 1. Introduction Biological tissue response is triggered by the presence of any foreign body. Trackingof such response over time is an active area of research which would profit from better analysis methods over several temporal and spatial scales [1]. Elec- trodes which record from and stimulate excitable tissue are ubiquitous and necessary for the development of implantable prosthetics, as well as for the advancement of our under- standing of basic brain mechanisms [2]. In this paradigm, the interface between the electronics and the biological tissue is a conductive surface. Until a decade ago, the majority of such substrates were metallic [3–5]. Over the last years, issues with the processibility of conductive polymeric materials were overcome, and they became viable alternatives to metal electrodes [6–8]. Despite these advances, the reactive glial re- sponse is still present upon implantation [9–12] and has been repeatedly reported, independently of material [13–15]. Charge transfer between the conductive surface and the biological tissue determines the most important figure of merit for a stimulating electrode. Whether the process is reversible, and, if stable, for how long it maintains the characteristic charge transfer, are two other constraints which guarantee a well designed system. If one of these three requirements is not met, any long term stimulation system will fail. The inflam- matory response from the tissue undermines the charge 0008-6223/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.carbon.2008.11.045 * Corresponding author: Fax: +1 206 600 4261. E-mail address: [email protected] (N. Peixoto). CARBON 47 (2009) 884 893 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/carbon

Upload: independent

Post on 16-Nov-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

C A R B O N 4 7 ( 2 0 0 9 ) 8 8 4 – 8 9 3

. sc iencedi rec t .com

ava i lab le at www

journal homepage: www.elsevier .com/ locate /carbon

Electrochemical characterization of multi-walled carbonnanotube coated electrodes for biological applications

Saugandhika Minnikantia, Perry Skeathb, Nathalia Peixotoa,*

aNeural Engineering Laboratory, Department of Electrical and Computer Engineering,

Volgenau School for Information Technology and Engineering, George Mason University, Fairfax, VA 22030, United StatesbNaval Research Laboratories, Optical Sciences Division, Code 5675, 4555 Overlook Avenue SW, Washington, DC 20375, United States

A R T I C L E I N F O

Article history:

Received 11 August 2008

Accepted 22 November 2008

Available online 7 December 2008

0008-6223/$ - see front matter � 2008 Elsevidoi:10.1016/j.carbon.2008.11.045

* Corresponding author: Fax: +1 206 600 4261E-mail address: [email protected] (N. Pe

A B S T R A C T

Carbon nanotubes, if used as a coating film on a conductive substrate, can substantially

raise the charge storage capacity and lower the impedance of electrodes without significant

increases to the geometric area. This is especially interesting in the case of stimulation of

nervous tissue. We design implantable electrodes targeted at wide frequency stimulation of

deep brain structures. Here we report on results in vitro with multi-walled carbon nano-

tubes coatings applied onto stainless steel substrates using direct current electrophoresis.

We experimentally demonstrate, through electrochemical techniques such as cyclic vol-

tammetry and impedance spectroscopy, the enhanced performance of multi-walled carbon

nanotube (MWCNT) coatings for implantable electrodes by contrasting our experimental

results against the more traditional stainless steel substrate characteristics. We also inves-

tigate surface morphology of aged electrodes. The interest in aged electrodes is dual fold:

implantable electrodes have to be mechanically stable and present high shelf life. On the

other hand, chemical modifications of the surface should be characterized. The effect of

superficial oxygen adsorption on the aged MWCNT electrodes is observed through a mod-

ified cyclic voltammetric spectrum, but not through any changes in impedance

spectroscopy.

� 2008 Elsevier Ltd. All rights reserved.

1. Introduction

Biological tissue response is triggered by the presence of any

foreign body. Tracking of such response over time is an active

area of research which would profit from better analysis

methods over several temporal and spatial scales [1]. Elec-

trodes which record from and stimulate excitable tissue are

ubiquitous and necessary for the development of implantable

prosthetics, as well as for the advancement of our under-

standing of basic brain mechanisms [2]. In this paradigm,

the interface between the electronics and the biological tissue

is a conductive surface. Until a decade ago, the majority of

such substrates were metallic [3–5]. Over the last years, issues

er Ltd. All rights reserved

.ixoto).

with the processibility of conductive polymeric materials

were overcome, and they became viable alternatives to metal

electrodes [6–8]. Despite these advances, the reactive glial re-

sponse is still present upon implantation [9–12] and has been

repeatedly reported, independently of material [13–15].

Charge transfer between the conductive surface and the

biological tissue determines the most important figure of merit

for a stimulating electrode. Whether the process is reversible,

and, if stable, for how long it maintains the characteristic

charge transfer, are two other constraints which guarantee a

well designed system. If one of these three requirements is

not met, any long term stimulation system will fail. The inflam-

matory response from the tissue undermines the charge

.

C A R B O N 4 7 ( 2 0 0 9 ) 8 8 4 – 8 9 3 885

transfer in acute and chronic situations. Several attempts have

been reported to overcome or delay glial reaction and biofoul-

ing. They can be lumped into surface modification (mechanical

and chemical) and charge control (electrical and chemical).

Such strategies include the use of biomolecules patterned onto

a substrate: printing poly-L-lysine structures onto the surface

of microfabricated multielectrode arrays is a typical example

[16]. Surfaces with protein-imprinted nanocavities having a

shape complementary to the protein [17] and electrodes cast

with thin films of a biopolymer chitosan [18] are further exam-

ples of successful strategies under the ‘‘surface modification’’

model. Surface topography can influence the growth and orien-

tation of neurons in culture. Surfaces having nanofeatures

have been shown to influence cell attachment and can be used

to isolate cells to a particular area on the silicon substrate [19].

On the other hand, the ‘‘charge control’’strategy presents inter-

esting advantages: for example, activated iridium oxide films

(AIROF) [20] and sputtered iridium oxide films (SIROF) [21] offer

low impedance and high charge injection, and have been suc-

cessfully rejuvenated in long term implants [22]. We hypothe-

size that a combination of the two strategies, that is, coating

the electrodes with a nano-featured surface, along with inter-

esting electrical characteristics of the coating, may be the most

successful approach for long term interfaces with the nervous

tissue.

Carbon nanotubes (CNTs) have been used, in the last two

decades, in applications ranging from DNA detection [23] to

nanorobots [24]. The electroactive ends of the CNTs can be

exploited to controllably deliver charge out of a nano- or

micrometer-long channel, and the ballistic conductance guar-

antees there will be no shortage of electrons at that site. De-

spite the awareness of possible cytotoxicity driven by

bioavailable iron [25], many groups have deposited CNTs onto

substrates, and shown the advantages of such coated sub-

strates to either implants [26,27] or substrates for in vitro

use [28–30]. We hypothesize that as long as the electrode is

made up of nanotubes, and so long as it exhibits CNT-like

electrochemistry, it has stable electrical and mechanical char-

acteristics, in solution or in the biological tissue.

Iridium oxide films rely mostly on hydrogen for the Fara-

daic reactions at the electrolyte–electrode interface [31]. The

carrier ions depleted and then replenished during a full cycle

of stimulation limits the safe reversible current that can be

passed during each stimulation cycle. No matter how well

the electrode performs in vitro, the in vivo charge will depend

on the composition of the electrolyte (biological tissue). The

electrode–electrolyte interface can be mimicked in vitro, it

has been modeled experimentally with buffered solutions

such as phosphate buffered saline (PBS), artificial cerebro-

spinal fluid (aCSF), or agar. However, this interface cannot

be controlled in stimulation schemes for implants. The most

reliable way of guaranteeing charge delivery is to utilize elec-

trodes with surfaces where the Faradaic and non-Faradaic

reactions rely on the material on the interface, and not on

the ionic composition of the media.

While the principle of charge transfer in CNT-based elec-

trodes is different from traditional surfaces, the characteriza-

tion of the electrodes should allow for comparison against

other conductive surfaces. In order to evaluate the newly

developed electrodes, we assess their performance against

bare stainless steel under the same conditions, and using

the same instrumentation. We show here that we can suc-

cessfully deposit CNTs onto stainless steel, and that they out-

perform stainless steel substrates.

2. Experimental

2.1. Electrode preparation

2.1.1. MWCNT suspension preparationThe biggest challenge we face in making stable nanotube sus-

pension is the agglomeration of carbon nanotubes in solution.

The extended p-electron system present in the tube walls

leads to Van der Waals forces which are enhanced by the fact

that the tubes can interact over extended distances [32]. The

agglomeration of MWCNTs is overcome by leveraging the sur-

factant adsorption at the CNT surface, so that MWCNTs do

not bind to one another. In order to obtain good dispersion

and a surface charge on the MWCNTs we used a surfactant,

NanoSperse-AQTM, provided by Nanolab (Newton, MA).

MWCNTs were first dispersed in deionized water using the

surfactant; that solution was then sonicated from 1 to 2 h in

order to obtain a stable suspension. Deionized water was used

as dispersant as it is not cytotoxic. The surfactant separates

the CNT aggregates into individual tubes during sonication

in deionised water [33] and is adsorbed in a regular pattern

on the CNT surfaces [34]. As this particular surfactant is anio-

nic, it induces a negative charge on the nanotubes. Although

not described in this work, we have tried several MWCNT sus-

pensions, as reported in the literature. Solutions were pre-

pared for example from DMF (dimethylformamide) [35,36] or

triton X-100 as a surfactant, along with IPA (isopropyl alco-

hol), ethanol, or distilled water as solvent [37]. Here we report

the results on the two suspensions yielding best homogeneity

and highest charge carrying capacity electrodes.

In order to compare the performance of charge delivery in

CNT-coated electrodes, we selected two types of custom-

made suspensions. These suspensions were used to deposit

MWCNTs over stainless steel substrates. Either industrial

grade or research grade MWCNTs having a concentration ra-

tio of 1 ml:0.55 mg:18.8 mg (DiH20:MWCNT:Nanosperse-AQTM)

were used. The research grade MWCNT concentration of the

suspension 0.55 mg/ml was reported as optimal for high qual-

ity electrophoretic deposition [38]. We screened several com-

positions of the solution, and found that the same ratio of

MWCNT applies to both research grade and industrial grade

materials. The MWCNTs present nominal diameter of 15 nm

and are longer than 20 lm (industrial) or 1 lm (research).

We would like to point out that MWCNTs were not treated

with nitric acid during purification process. Nitric acid is

known to strongly oxidize the nanotubes [39]. The results re-

ported here on the electrochemical characterization for both

CNT grades were lumped, as there was no statistical differ-

ence in charge or impedance upon deposition.

2.1.2. Electrophoretic depositionSubstrates were made from 250 lm diameter medical grade

stainless steel (316 L) wire with polyimide insulation. We re-

move 3mm of the polyimide from the tip, leaving an exposed

886 C A R B O N 4 7 ( 2 0 0 9 ) 8 8 4 – 8 9 3

geometrical area of 2.5 mm2 for MWCNT deposition. Selection

of substrate is crucial in implantable systems. Here we intend

to stimulate deep brain structures, especially the hippocam-

pus and subthalamic nucleus, with electric fields [40]. The

cylindrical shape of the wire, along with the 3 mm of exposed

height, will allow us to optimally stimulate the whole extent

of the hippocampus or subthalamic nucleus.

Electrophoresis is the movement of charged particles in a

suspension upon application of an electric field. The com-

bined process of movement followed by deposition on a

charged surface is called electrophoretic deposition (EPD)

[41]. The suggested mechanism for EPD in the case of carbon

nanotubes is the particle double layer distortion due to in-

duced electric fields and particle coagulation on the substrate.

Fig. 1 – Schematic illustration of the setup used for

electrophoretic deposition. Silver electrode and bare

stainless steel electrode serve as cathode and anode,

respectively. The electrodes are immersed in a glass vial

containing MWCNT suspension in deionized water. 2.33 V

is applied between the electrodes with a DC power supply.

Inset: photograph of a bare stainless steel electrode (right)

and a MWCNT coated electrode (left).

Fig. 2 – (a) SEM image of a single bare stainless steel electrode a

electrode diameter is 250 lm. (b). SEM image at higher magnific

electrophoretic MWCNT deposition. The MWCNT nominal diam

random entangled arrangement of MWCNTover the metal subst

over the electrode increases the effective area of the electrode. T

as catalysts to grow MWCNTs.

The coagulation is based on the Derjaguin-Landau-Verwey-

Overbeek (DLWVO) theory and is governed by London-Van

der Waals forces [42]. Electrophoretic deposition of carbon

nanotubes has been successfully developed for field emission

arrays [43,44] for alignment and deposition of films over sur-

faces [45,46], and for high power density supercapacitors [47].

Electrophoresis was the method employed here for deposi-

tion of MWCNT [38] on as prepared stainless steel electrodes.

For each set of EPD experiments, a DC voltage value of 2.3 V

was applied between a stainless steel electrode (anode) and

bare gold or silver electrode (cathode). The distance between

the two electrodes was kept at 1 mm (Fig. 1). The field used

for in our experiments yields a stable cohesion of MWCNTs

onto metallic substrates [48]. MWCNTs can be clearly ob-

served as a black deposit over the 3mm uninsulated stainless

steel electrode tip acting as the anode (Fig. 1 inset). Deposition

times varied from 5 to 20 min. In the experiments reported

here, more than 70 electrodes were deposited. After deposi-

tion the electrodes were immediately characterized, and aged

in air.

2.2. Physiochemical characterization

The surface morphology was evaluated by a scanning elec-

tron microscope (SEM) operated at 5.00 kV. The resolution var-

ied from 100 nm to 200 lm. Fig. 2 shows the SEM images

obtained from a single MWCNT coated electrode aged in air

for 60 days. The images reveal homogeneous random entan-

gled arrangement of MWCNT over the metal substrate. The

nanotubes are oriented parallel to the surface of the stainless

steel substrate. The bright spots are assumed to be the metal

nano particles used as catalysts to grow MWCNTs. This par-

ticular result has raised several questions about the metal

impurities present on the surface of the electrode coatings.

In order to quantitatively evaluate impurities in our samples,

we have performed two analyses, inductively coupled

plasma atomic emission spectroscopy (ICP-AES) and X-ray

fter 10 min of electrophoretic MWCNT deposition. The bare

ation of single bare stainless steel electrode after a 10 min

eter is 15 nm and length is longer than 20 lm. Homogeneous

rate is observed. The observed mesoporous matted structure

he bright spots are assumed to be metal nanoparticles used

C A R B O N 4 7 ( 2 0 0 9 ) 8 8 4 – 8 9 3 887

photoelectron spectroscopy (XPS). While the ICP-AES screens

for metal catalysts, the XPS reveals the state of the surface of

the MWCNT used for deposition.

Elemental analysis for the industrial and research grade

suspensions used for EPD was achieved by inductively cou-

pled plasma atomic emission spectrometer (ICP-AES) (Per-

kin–Elmer Optima 3000DV). MWCNTs were dispersed in

HNO3 (1%) solution. The ICP-AES characterizations were cali-

brated with standard samples. All the elements detected are

listed in Fig. 3a. Industrial grade MWCNTs show higher per-

centage of elements such as Al, B, Ca, K, Na, and Si (Fig. 3a).

These can be associated with ceramic oxides, potential resid-

ual components in industrial grade MWCNTs. The elemental

iron content in industrial grade MWCNTs is 1.91 wt%. Surpris-

ingly, research grade MWCNTs show higher percentage of ele-

mental Fe (2.10 wt%).

XPS was performed using a custom-instrument predomi-

nantly composed of equipment from Perkin–Elmer (physical

electronics division). The X-ray source has a spot size of a

few millimeters and is operated at 15 kV, 17 mA (250 W) using

a dual Mg/Al anode. Non-monochromatized Al (Ka 945;

hv = 1486.6 eV) source was used for these experiments. Survey

spectra were collected by scanning from 1000 to 0 eV on the

binding energy scale at pass energy of 100 eV; a total of 10

scans were averaged. The operating pressure of the spectrom-

eter was typically 10�9 mbar and signal processing was per-

formed using AugerScanTM (RBD Instruments, Bend, OR). XPS

spectrum of both research grade (Fig. 3b) and industrial grade

(Fig. 3c) show clear peaks for carbon and oxygen, while peaks

for metal catalysts were not detected.

Fig. 3 – (a) ICP-AES elemental analysis of industrial and

research grade MWCNT suspensions used for electrode

coatings. ICP analysis realized in the Research Analytical

Laboratory (Univ. Minnesota, MN). The following elements

were present at quantities below the minimum detection

level: Ni, As, Be, Cd, Co, Cr, Mn, Rb, and V. (b). XPS of

research grade MWCNT. (c). XPS of industrial grade MWCNT.

The surface composition of samples shows in (b) and (c) is

indistinguishable, presenting 97.5% (industrial) and 98%

(research) carbon content. The other peak in each spectrum

refers to oxygen.

2.3. Electrochemical characterization of electrodes

Implantable electrodes may be characterized by a myriad of

techniques. Cyclic voltammetry (CV) and electrochemical

impedance spectroscopy (EIS) are two methods traditionally

employed in electrochemistry [49]. While in chemistry the cell

is formed by electrolyte and electrodes, and the particular

processes taking place at that interface are analyzed, here

we intend to gain a deeper understanding of the working elec-

trode, ideally independently of the electrolyte. The solid–li-

quid interface is still present, and charge transfer occurs,

but the electrode, rather than the interface, is the main target

for such characterizations. Voltammetry, if performed at slow

scan rates (less than 50 mV/s), constitutes a direct way to

measure the maximum charge the electrode can deliver: elec-

trode polarization is assumed, and the capacitance is maxi-

mized by forcing a slow charging and discharging of the

double layer. This method not only provides a straightforward

and well-known figure of merit for the electrode in one single

number, the ‘‘charge storage capacity’’, but it is also appealing

in the design of electrodes for in vivo applications. Cyclic vol-

tammetric spectra can be taken in vivo and compared to pre-

vious in vitro characterizations. Impedance spectroscopy (IS)

is a small signal technique especially useful for electrode sys-

tems involving interface polarization. The method relies on

sweeping a sinusoidal waveform across several decades of

frequencies, and evaluating the amplitude and phase of the

resulting impedance between two electrodes in solution. A

detailed physico-electrical model of the reactions occurring

at the electrode–electrolyte interface is unavailable [50]. In or-

der to interpret data from IS, a circuit with varying phase was

first suggested by Fricke, in 1931 [51]. A constant-phase ele-

ment, CPE, was added to the model in order to account for

the inhomogeneities on the surface of the solid capacitor as

seen experimentally in the electrochemical cell.

A commercial potentiostat (REF 600, Gamry Instruments,

Warminster, PA) was used to determine electrode impedance

as well as the charge storage capacity (CSC). A two electrode

setup with silver/silver chloride (Ag/AgCl) as the reference

electrode was used for CV and EIS. A two electrode setup

can be used if the flow of current does not significantly affect

the potential of the reference electrode [52]. In our case, the

current through the electrochemical cell is lower than 10 lA,

and the solution resistance (Rs) is approximately 58 X. This

causes a negligible change in voltage at the reference elec-

trode, always below 0.58 mV.

Unless otherwise noted, the electrolyte used in all charac-

terizations was phosphate buffered saline (PBS), adjusted to a

pH of 7.4. The rate of electron transfer at a solid electrode de-

pends on the present state of the surface as well as on the

previous history of the electrode [53]. An electrode surface

that is not clean usually will manifest, in voltage sweep

experiments, by a shift of the peak potential and by a change

in the peak current. For voltammetric characterizations in

aqueous solution, electrochemical cycling is commonly used

to activate or clean the surface of the electrode [53]. There-

fore, preconditioning experiments preceded all characteriza-

tion experiments. All electrodes were preconditioned for

about 50 cycles at the rate of 50 mV/s from �0.7 V to 0.7 V.

In the case of CV the excitation voltage is ramped from

Fig. 5 – Cyclic voltammogram (CV) of a bare stainless steel

electrode (s) against a MWCNT electrode. All CV

measurements were taken using a two electrode setup in

PBS (pH 7.4) solution with Ag/AgCl as the reference electrode

and a scan rate of 50 mV/s. Cathodic area under the CV

curve yields the CSCc value; bare stainless steel electrode

CSCc = 2.18 lC/mm2 and MWCNT coated stainless steel

electrode CSCc = 7.15 lC/mm2.

888 C A R B O N 4 7 ( 2 0 0 9 ) 8 8 4 – 8 9 3

�0.7 V to 0.7 V at a rate of 50 mV/s in reference to the open cir-

cuit potential. The cathodic charge storage capacity (CSCc) is

obtained by integrating the cathodic current over time for

one period of the triangular waveform, and can be visualized

from the lower half of the enclosed area under the CV curve.

For all EIS experiments reported here the applied frequency

ranged from 0.2 Hz to 100 kHz, with a DC bias set to the open

circuit potential (Eoc).

3. Results and discussion

3.1. Impedance lowers as result of MWCNT deposition

The EIS characterizations were performed immediately after

CV. A small amplitude (50 mVrms) sinusoidal signal was ap-

plied, and the modulus and phase plotted over a frequency

range of 0.2 Hz–100 kHz. The impedance spectrum of a

MWCNT coated electrode in comparison to bare stainless

steel electrode is represented in Bode plots (Fig. 4). The mod-

ulus of impedance of the bare stainless steel electrode de-

creases by two orders of magnitude over the first six

decades of frequencies tested. The modulus of impedance

for bare, MWCNT deposited, and aged electrodes were statis-

tically compared using Wilcoxon Signed Rank test (n = 8). The

impedance of bare electrode significantly lowered (P = 0.04, 2-

tailed) after MWCNT deposition. For example for a MWCNT

electrode (Fig. 4) the impedance lowered from 760 kX to

281 kX at 0.2 Hz and from 4.1 kX to 2.3 kX at 100 Hz. The data

on aged electrodes also revealed a significant lowering of

impedance when compared to bare electrodes. However, the

difference was no longer significant (P > 0.005, 2-tailed) when

aged electrodes were compared against MWCNT deposited

electrode.

The frequency dispersion in impedance values is modeled

as a CPE due to surface non-uniformity and roughness. The

impedance of the CPE is given by Z = 1/Ys(ix)a, where Ys is a

constant with dimension Fsa�1, x is angular frequency (2pf),

i =p�1, and 0 < a < 1, where a = 1 for an ideal capacitor [54].

The a values for bare, MWCNT deposited and aged electrodes

were calculated by a data fitting algorithm (simplex method).

The mean (n = 8) a values decreased from 0.82 ± 0.05 (bare) to

0.80 ± 0.05 (MWCNT coated). The a values of bare and MWCNT

Fig. 4 – Bode plots of impedance modulus (left) and phase (right)

All EIS measurements were taken using a two electrode setup i

electrode. The frequency was swept from 0.2 Hz to 100 kHz. Low

frequencies below 40 kHz.

deposited electrode were not significantly different (Wilcoxon

Signed Rank test P = 0.09), whereas the aged data revealed a

significant difference (Wilcoxon Signed Rank test P = 0.03) in

a when compared to bare electrodes. The mean a values of

electrodes decreased from 0.82 ± 0.05 (bare, n = 8) to

0.74 ± 0.07 (aged MWCNT, n = 8).

3.2. High charge delivery with MWCNT electrodes

MWCNT coated electrodes deliver higher charge than

stainless steel electrodes. This was one of the objectives of

our study, as these electrodes will be used in deep brain stim-

ulation scenarios. The cyclic voltammetric spectrum of a typ-

ical MWCNT coated electrode (Fig. 5) in comparison to bare

electrode shows an increase in the enclosed area of the curve.

The CSCc for bare, MWCNT deposited, and aged electrodes

were calculated and statistically compared using Wilcoxon

Signed Rank test (n = 8). The averaged results are presented

in Table 1. The CSCc of bare electrodes significantly increased

(P = 0.005, 1-tailed) after MWCNT deposition, for example for

of a stainless steel electrode (+) and a MWCNT electrode (s).

n PBS (pH 7.4) solution with Ag/AgCl as the reference

ering of impedance is observed, in this case, for all

Table 1 – Electrode performance is evaluated according to charge carrying capacity, a values, and Yo.a

Figure of merit Stainless steel MWCNT coated Aged MWCNT coated

CSCc (lC/mm2) 4.24 ± 3.74 (n = 8) 10.88 ± 8.87 (n = 8) 7.42 ± 3.36 (n = 4) low

12.89 ± 5.37 (n = 4) high

a 0.82 ± 0.05 (n = 8) 0.80 ± 0.05 (n = 8) 0.74 ± 0.07 (n = 8)

Yo (lS) 1.57 ± 0.87 (n = 8) 2.31 ± 1.07 (n = 8) 3.04 ± 1.37 (n = 8)

a Bare (stainless steel) electrodes, MWCNT coated, and MWCNT coated aged (45 days) electrodes are statistically compared against each other.

C A R B O N 4 7 ( 2 0 0 9 ) 8 8 4 – 8 9 3 889

the MWCNT electrode of Fig. 4 the CSCc increased from

2.18 lC/mm2 to 7.15 lC/mm2.

The aged electrodes also showed a significant (P = 0.005, 1-

tailed) increase in CSCc when compared to bare stainless steel

electrodes (Table 1). However, the difference was no longer

significant (P = 0.67, 1-tailed) for aged electrodes when com-

pared to MWCNT deposited electrode. Four of the eight aged

electrodes showed an increase in the CSCc if compared to

either bare or MWCNT coated electrodes. This group is iden-

tified as high in Table 1, in contrast to the low group, also

shown in the same row of that table. Fig. 6 shows an aged

electrode with higher CSC than the as prepared electrode,

with an increase from 5.08 lC/mm2 to 15.22 lC/mm2.

The electrochemical behavior of CNTs is dependent on its

impurities (metal catalysts) [55,56], structural defects [57],

pre-treatment, orientation, and the amount of available ad-

sorbed species on its sidewalls and open ends [58]. In our

experiments the CVs present noticeable reversible redox

peaks for the bare stainless steel electrodes which are en-

hanced after MWCNT deposition as shown in Figs. 4 and 5.

We consider these to be the associated with the redox

peaks of stainless steel [59]. The charge transfer can be asso-

ciated both with charging and discharging of the electrolyte–

electrode double layer.

CNTs with greater amount of metal catalysts show supe-

rior electrochemical behavior if compared to purified nano-

tubes [56]. However, in our experiments no significant

difference in the performance of research and industrial

grade was observed with respect to CSC and impedance. We

Fig. 6 – Cyclic voltammogram (CV) of as prepared MWCNT

electrode (+) against aged MWCNT electrode (*). All CV

measurements were taken using a two electrode setup in

PBS (pH 7.4) solution with Ag/AgCl as the reference electrode

and a scan rate of 50 mV/s. The CSCc of this particular

MWCNTelectrode is 5.08 lC/mm2 while the CSCc of the aged

MWCNT electrode is 15.22 lC/mm2.

hypothesize that the orientation of the nanotubes in relation

to the surface is one of the reasons for the industrial and re-

search grade nanotubes to behave similarly. We showed, with

the CV and EIS spectra, that the adsorbed species and metal

catalysts do not play a major role in charge transfer.

When left in air for 30 days or longer, stainless steel elec-

trodes present unstable cyclic voltammetric spectra. The oxi-

dizing surface, however, may be cleaned with a sequence of

cyclic voltammetry cycles. Eventually the CV stabilizes to a

typical spectrum, as shown in Fig. 5, with one peak in each

direction of the excitation waveform. Electronic properties

of nanotubes may change on exposure to air or oxygen. In

particular, semiconducting nanotubes are thought to convert

to metal on exposure to air (oxygen) [60]. As our CNTs are mu-

tli-walled, and these have been shown to be metallic [61,62];

we do not believe that the mechanism through which our

aged electrodes outperform as prepared electrodes relies on

metallic content.

We also conjectured that carbon nanotube coated elec-

trodes would be more stable than the stainless steel elec-

trodes. In order to test the stability of coated electrodes, as

well as the oxidizing behavior over time, we left the coated

electrodes to age, exposed to air. We then characterized

coated electrodes after 15, 30, 45, and 60 days. The difference

in charge carrying capacity and a value is conclusive after 45

and 60 days, and therefore we present the data here for the

45 day old electrodes. On the other hand, both modulus and

phase plots for the impedance spectroscopy indicated no sig-

nificant deviation from the coated data (Fig. 7).

Without long term stability, electrodes for biological appli-

cations cannot be used in vivo. In order to investigate the

chemical stability of electrodes in solution, 50 CV cycles were

applied at 50 mV/s scan rate to electrodes kept in phosphate

buffered saline. For consistency, all electrodes underwent

the same treatment. Every single electrode presented the

same behavior: Fig. 8 shows some of the 50 cycles for a typical

electrode. The first two cycles represent the highest deviation

from the mean. However, after the third cycle there is no sig-

nificant difference between any two subsequent cycles. The

same behavior is observed with the impedance spectra: at

any frequency, the impedance is stable for over 50 cycles in

solution (data not shown).

As summarized in Table 1, MWCNT coated electrodes sig-

nificantly increase their CSCc, while maintaining their capac-

itive behavior: a values are similar for bare and coated

electrodes. This indicates that the CNT coatings do not mod-

ify the actual interface characteristics of the surface; the

underlying capacitive behavior of the stainless steel surface

persists despite the CNT coating. This result is also supported

by the SEM image of the carbon nanotube deposition (Fig. 2),

Fig. 7 – Bode plots of impedance (modulus and phase) of as prepared electrode (s) against aged electrode (x). All EIS

measurements were taken using a two electrode setup in PBS (pH 7.4) solution with Ag/AgCl as reference. The frequency

ranged from 0.2 Hz to 100 kHz. No statistically significant difference in impedance is observed between these two electrodes.

Fig. 8 – Cyclic voltammogram (CV) of 50 cycles for a MWCNT

electrode. All CV measurements were taken using a two

electrode setup in PBS (pH 7.4) solution with Ag/AgCl as the

reference electrode and a scan rate of 50 mV/s. Over a period

of 50 cycles the electrode holds its charge delivery capacity.

CV does not change significantly over time, indicating long

term stability during charge delivery.

890 C A R B O N 4 7 ( 2 0 0 9 ) 8 8 4 – 8 9 3

forming an entangled porous mesh. Even though the geomet-

ric area is the same, there is an increase in the accessible area

for the ions at the electrode–electrolyte interface. The entan-

gled crosslinked nanotubes create connected pores in the

mesh. These pores provide a multitude of pathways [63] for

electrons to pass through randomly organized MWCNTs

either via the sidewalls or through the open ends of the nano-

tubes [58]. This is crucial for the charging of the electric dou-

ble layer.

4. Conclusion

After aging, we observe a bimodal distribution of electrodes

for the CSC: a high and a low group (Table 1, CSCs row, last col-

umn). In half of the aged electrodes characterized, the CSCc

was enhanced 45 days after coating. If compared to the bare

electrodes, the aged electrodes from the high group present

three times higher CSCc. The low group, however, did not

present any statistically significant enhancement in CSCc.

We are investigating the bimodal behavior of the aged elec-

trodes. The oxygen induced charge transfer is dependent on

tube defects, impurities such as metal catalysts and adsorbed

chemical species [59]. Even though the percentage of impuri-

ties (metal catalyst) is greater in the industrial grade solution,

we did not see any significant difference in aged electrodes

when compared industrial to research grade. We have at-

tempted to investigate variables which could have influenced

that result. The deposition solution, along with surfactants

and solvents, would be one major parameter. The grade of

the multi-walled nanotubes, research or industrial, would

be another variable, as well as the particular substrate used,

cleaning procedure, handling, and deposition time. Most of

these variables were controlled for: they were kept constant

throughout all the experiments. The MWCNT grade is the

only parameter which we explicitly modified (research versus

industrial). However, every set of electrodes coated with any

of the MWCNTs showed the same results as the pool of all

electrodes: half high, half low. Interestingly, the a value signif-

icantly lowers as electrodes age if the aged electrodes are

compared to bare electrodes. Even for the low group, a de-

creases, indicating a more resistive and less homogeneous

interface. As the middle and last rows of Table 1 show, the

highly capacitive characteristic of stainless steel and MWCNT

coated electrodes is modified with age. The constant-phase

element (CPE) is the best model so far proposed in the litera-

ture (suggested in 1931 by Fricke) to describe a non-isotropic

charge transfer mechanism [51]. As the traditional circuit ele-

ments (capacitor, resistor, inductors) cannot be used in the

case of an electrochemical cell, due to microscopic fractal-like

behavior, the modified equation for impedance incorporates

the a value as an exponent. The results described in the liter-

ature for a values state that carbon nanotube coating of sur-

face renders the substrate capacitive [28]. Even though

similar a values were obtained in our experiments, they were

not significantly different when compared with bare stainless

steel electrodes. In our work, the result showing lowering of a

is universal: all characterized electrodes presented the same

behavior; this may be a better measure of the electrode age

than the previously discussed charge storage capacity.

This is, to our knowledge, the first time an electrode is ob-

tained with high charge carrying capacity, without the respec-

tive decrease in impedance. For all traditional electrodes [64],

including some reports on carbon nanotube coatings [28],

C A R B O N 4 7 ( 2 0 0 9 ) 8 8 4 – 8 9 3 891

these two figures of merit are highly correlated. For all elec-

trodes reported here, we see no such link: we have manufac-

tured high CSCc electrodes, with very similar impedance

spectra to medical grade stainless steel. The decrease in

impedance is still observed in some cases, in aged electrodes,

or if additional MWCNT layers are deposited (data not

shown).

Conducting biocompatible polymers such as polypyrrole

[28] or PEDOTs [65] can be used as coatings for MWCNTs or

carbon nanofibers in order to enhance their electrochemical

behavior. Such electrodes are also applicable as neural stimu-

lation interfaces [8,29]. Reports in the literature show promis-

ing results with the electrochemical polymerization of

PEDOT–MWCNT enhancing capacitance and stability of the

electrodes [65]. On the other hand, such coatings can be used

to address any issues related to the chemicals used to process

nanotube suspensions. These coatings would also serve the

purpose of rendering the surface biocompatible, indepen-

dently of the chemicals used to aid the deposition of

MWCNTs. We have tested two grades of MWCNTs: research

and industrial. They present the same electrochemical char-

acteristics. Initially, we hypothesized that research grade

would present lower percentage of metal impurities, however

the ICP-AES (Fig. 3a) results show otherwise. Irrespective of

how pure the samples are, metal nanoparticles will still be

present. The accessibility of these metal nanoparticles to bio-

logical medium is of major concern as they generate reactive

oxygen species which are cytotoxic to the cells [66]. The ICP-

AES analyses reveal that the total iron content is around 2%.

This is comparable to the lowest level of iron content in car-

bon nanotube samples discussed in the literature [25]. While

this result is reassuring and provides a good indication that

we may use these MWCNTs for biological applications, sam-

ples to be implanted will undergo two further steps: atmo-

spheric aging (for periods longer than 170 days) and

incubation with chelators. Both of these have been shown

to reduce bioavailable iron, reducing the probability of oxida-

tive stress toxicity [67].

MWCNTs can be electrophoretically deposited over stain-

less steel electrodes for stimulation of biological tissue. The

coated electrodes demonstrate high charge storage capacity

and low impedance and maintain their original characteris-

tics after being aged. The surface area increases with the

MWCNT deposition, thus allowing for higher CSC, but the

underlying electrochemistry is preserved, and the redox

peaks observed in the metallic substrate are maintained after

MWCNT coating, as well as after aging.

We thus propose that the treated and purified MWCNT

coated electrodes designed and characterized here will be

evaluated in acute in vivo experiments. We expect them to

outperform stainless steel and confirm carbon nanotubes

well researched and explored characteristics for neural tissue

stimulation.

Acknowledgements

This project was supported by the Volgenau School for Infor-

mation Technology and Engineering, George Mason

University.

R E F E R E N C E S

[1] Polikov VS, Tresco PA, Reichert WM. Response of brain tissueto chronically implanted neural electrodes. J NeurosciMethods 2005;148(1):1–18.

[2] Williams JC, Hippensteel JA, Dilgen J, Shain W, Kipke DR.Complex impedance spectroscopy for monitoring tissueresponses to inserted neural implants. J Neural Eng2007;4:410–23.

[3] Rezai AR, Finelli D, Nyenhuis JA, Hrdlicka G, Tkach J, SharanA, et al. Neurostimulation systems for deep brainstimulation: in vitro evaluation of magnetic resonanceimaging-related heating at 1.5 tesla. J Magn Reson Imaging2002;15(3):241–50.

[4] Boockvar JA, Telfeian A, Baltuch GH, Skolnick B, Simuni T,Stern M, et al. Long-term deep brain stimulation in apatient with essential tremor: clinical response andpostmortem correlation with stimulator termination sitesin ventral thalamus. Case report. J Neurosurgery2000;93(1):140–4.

[5] Ashby P, Kim YJ, Kumar R, Lang AE, Lozano AM.Neurophysiological effects of stimulation through electrodesin the human subthalamic nucleus. Brain1999;122(10):1919–31.

[6] Ludwig KA, Uram JD, Yang J, Martin DC, Kipke DR. Chronicneural recordings using silicon microelectrode arrayselectrochemically deposited with a poly (3,4-ethylenedioxythiophene)(PEDOT) film. J Neural Eng2006;3(1):59–70.

[7] Cui X, Wiler J, Dzaman M, Altschuler RA, Martin DC. In vivostudies of polypyrrole/peptide coated neural probes. ProcNatal Acad Sci 2003;24(5):777–87.

[8] Cui X, Martin D. Electrochemical deposition andcharacterization of poly (3,4-ethylenedioxythiophene) onneural microelectrode arrays. Sens Actuators, B: Chem2003;89(1):92–102.

[9] Giordana MT, Attanasio A, Cavalla P, Migheli A, Vigliani MC,Schiffer D. Reactive cell proliferation and microglia followinginjury to the rat brain. Neuropath Appl Neurobiol1994;20(2):163–74.

[10] Fujita T, Yoshimine T, Maruno M, Hayakawa T. Cellulardynamics of macrophages and microglial cells in reaction tostab wounds in rat cerebral cortex. Acta Neurochir1998;140(3):275–9.

[11] Stensaas SS, Stensaas LJ. The reaction of the cerebral cortexto chronically implanted plastic needles. Acta Neuropathol(Berl) 1976;35(3):187–203.

[12] Szarowski DH, Andersen MD, Retterer S, Spence AJ, IsaacsonM, Craighead HG, et al. Brain responses to micro-machinedsilicon devices. Brain Res 2003;983(1–2):23–35.

[13] Nicolelis MAL, Dimitrov D, Carmena JM, Crist R, Lehew G,Kralik JD, et al. Chronic, multisite, multielectrode recordingsin macaque monkeys. Proc Natal Acad Sci2003;100(19):11041–6.

[14] Biran R, Martin DC, Tresco PA. Neuronal cell lossaccompanies the brain tissue response to chronicallyimplanted silicon microelectrode arrays. Exp Neurology2005;195(1):115–26.

[15] Yuen TGH, Agnew WF. Histological evaluation ofpolyesterimide-insulated gold wires in brain. Biomaterials1995;16(12):951–6.

[16] James CD, Davis R, Meyer M, Turner A, Turner S, Withers G,et al. Aligned microcontact printing of micrometer-scalepoly-L-lysinestructures for controlled growth of culturedneurons on planarmicroelectrode arrays. IEEE Trans BiomedEng 2000;47(1):17–21.

892 C A R B O N 4 7 ( 2 0 0 9 ) 8 8 4 – 8 9 3

[17] Shi H, Tsai W-B, Garrison MD, Ferrari S, Ratner BD. Template-imprinted nanostructured surfaces for protein recognition.Nature 1999;398(6728):593–7.

[18] Cruz J, Kawasaki M, Gorski W. Electrode coatings based onchitosan scaffolds. Anal Chem 2000;72(4):680–6.

[19] Craighead HG, Turner SW, Davis RC, James C, Perez AM, JohnSt PM, et al. Chemical and topographical surfacemodification for control of central nervous system celladhesion. Biomed Microdevices 1998;1(1):49–64.

[20] Cogan SF, Cogan SF, Troyk PR, Ehrlich J, Plante TDAPTD. Invitro comparison of the charge-injection limits of activatediridium oxide (AIROF) and platinum–iridium microelectrodes.IEEE Trans Biomed Eng 2005;52(9):1612–4.

[21] Slavcheva E, Vitushinsky R, Mokwa W, Schnakenberg U.Sputtered iridium oxide films as charge injection material forfunctional electrostimulation. J Electrochem Soc2004;151:E226–37.

[22] Johnson MD, Otto KJ, Kipke DR. Repeated voltage biasingimproves unit recordings by reducing resistive tissueimpedances. IEEE Trans Neural Syst Rehabilitation Eng2005;13(2):160–5.

[23] Kouklin NA, Kim WE, Lazareck AD, Xu JM. Carbon nanotubeprobes for single-cell experimentation and assays. Appl PhysLett 2005;87:173901–3.

[24] Baxendale M. Biomolecular applications of carbonnanotubes. IEEE Proc Nanobiotechnol 2003;150(1):3–8.

[25] Guo L, Morris DG, Liu X, Vaslet C, Hurt RH, Kane AB. Ironbioavailability and redox activity in diverse carbon nanotubesamples. Chem Mater 2007;19(14):3472–8.

[26] Keefer EW, Botterman BR, Romero MI, Rossi AF, Gross GW.Carbon nanotube coating improves neuronal recordings. NatNano 2008;3(7):434–9.

[27] Wang K, Fishman HA, Dai H, Harris JS. Neural stimulationwith a carbon nanotube microelectrode array. Nano Lett2006;6(9):2043–8.

[28] Nguyen-Vu T, Chen H, Cassell A, Andrews R, Meyyappan M, LiJ. Vertically aligned carbon nanofiber arrays: an advancetoward electrical-neural interfaces. Small 2006;2:89–94.

[29] Nguyen-Vu T, Chen H, Cassell A, Andrews R, Meyyappan M, LiJ. Vertically aligned carbon nanofiber architecture as amultifunctional 3-D neural electrical interface. IEEE TransBiomed Eng 2007;54(6):1121–8.

[30] Chlopek J, Czajkowska B, Szaraniec B, Frackowiak E, SzostakK, Beguin F. In vitro studies of carbon nanotubesbiocompatibility. Carbon 2006;44(6):1106–11.

[31] Merrill D, Bikson M, Jefferys J. Electrical stimulation ofexcitable tissue: design of efficacious and safe protocols.J Neurosci Methods 2005;141(2):171–98.

[32] Schaefer DW, Brown JM, Anderson DP, Zhao J, ChokalingamK, Tomlin D, et al. Structure and dispersion of carbonnanotubes. J Appl Crystallogr 2003;36(3):553–7.

[33] Zhang X, Zhang J, Wang R, Zhu T, Liu Z. Surfactant-directedpolypyrrole/CNT nanocables: synthesis, characterization,and enhanced electrical properties. Science2004;5(7):998–1002.

[34] Islam MF, Rojas E, Bergey DM, Johnson AT, Yodh AG. Highweight fraction surfactant solubilization of single-wallcarbon nanotubes in water. Nano Lett 2003;3(2):269–73.

[35] Du C, Yeh J, Pan N. Carbon nanotube thin films with orderedstructures. J Mater Chem 2005;15(5):548–50.

[36] Papakonstantinou P, Kern R, Robinson L, Murphy H, Irvine J,McAdams E, et al. Fundamental electrochemical propertiesof carbon nanotube electrodes. Fullerenes Nanotubes CarbonNanostructures 2005;13(2):91–108.

[37] Liu X, Spencer JL, Kaiser AB, Arnold WM. Electric-fieldoriented carbon nanotubes in different dielectric solvents.Curr Appl Phys 2004;4(2–4):125–8.

[38] Thomas BJC, Boccaccini AR, Shaffer MSP. Multi-walled carbonnanotube coating using electrophoretic deposition (EPD).J Am Ceram Soc 2005;88(4):980–2.

[39] Frackowiak E, Beguin F. Carbon materials for theelectrochemical storage of energy in capacitors. Carbon2001;39(6):937–50.

[40] Sunderam S, Chernyy N, Peixoto N, Mason JP, Weinstein SL,Schiff SJ, et al. Improved sleep–wake and behaviordiscrimination using MEMS accelerometers. J NeurosciMethods 2007;163(2):373–83.

[41] Van der Biest OO, Vandeperre LJ. Electrophoretic deposition ofmaterials. Ann Rev Mater Sci 1999;29(1):327–52.

[42] Sarkar P, Nicholson PS. Electrophoretic deposition (EPD):mechanisms, kinetics, and application to ceramics. J AmCeram Soc 1996;79(8):1987–2002.

[43] Choi WB. Electrophoresis deposition of carbon nanotubes fortriode-type field emission display. Appl Phys Lett2001;78(11):1547–9.

[44] Zhao H, Song H, Li Z, Yuan G, Jin Y. Electrophoretic depositionand field emission properties of patterned carbon nanotubes.Appl Surf Sci 2005;251(1–4):242–4.

[45] Kamat PV, Thomas KG, Barazzouk S, Girishkumar G,Vinodgopal K, Meisel D. Self-assembled linear bundles ofsingle-wall carbon nanotubes and their alignment anddeposition as a film in a DC field. J Am Chem Soc2004;126(34):10757–62.

[46] Senthil Kumar M, Lee SH, Kim TY, Kim TH, Song SM, Yang JW,et al. DC electric field assisted alignment of carbonnanotubes on metal electrodes. Solid-State Electron2003;47(11):2075–80.

[47] Du C, Pan N. High power density supercapacitor electrodes ofcarbon nanotube films by electrophoretic deposition.Nanotechnology 2006;17(21):5314–8.

[48] Boccaccini AR, Cho J, Roether JA, Thomas BJC, Jane Minay E,Shaffer MSP. Electrophoretic deposition of carbon nanotubes.Carbon 2006;44(15):3149–60.

[49] Hu CG, Wang WL, Liao KJ, Liu GB, Wang YT. Systematicinvestigation on the properties of carbon nanotubeelectrodes with different chemical treatments. J Phys ChemSolids 2004;65(10):1716–31.

[50] Barsoukov E, MacDonald RJ. In impedance spectroscopytheory, experiment and applications. 2nd ed. Wiley; 2005.

[51] Fricke H. The electric conductivity and capacity of dispersesystems. Physics 1931;1(2):106–15.

[52] Bard AJ, Faulkner LR. Electrochemical methods,fundamentals and applications, vol. 25. New York: JohnWiley & Sons; 1980. p. 25.

[53] Crow DR. 3rd ed. Chapman and Hall; 1998. p. 197–8.[54] Bockris JOM, White RE, Conway BE. Modern aspects of

electrochemistry, vol. 53. Springer; 1999. p. 204–5.[55] Banks CE, Crossley A, Salter C, Wilkins SJ, Compton RG.

Carbon nanotubes contain metal impurities which areresponsible for the ‘‘electrocatalysis’’ seen at some nanotube-modified electrodes. Angew Chem Int Ed 2006;45(16):2533–7.

[56] Niessen RAH, de Jonge J, Notten PHL. The electrochemistry ofcarbon nanotubes. J Electrochem Soc 2006;153:A1484–91.

[57] Portet C, Yushin G, Gogotsi Y. Electrochemical performance ofcarbon onions, nanodiamonds, carbon black and multiwallednanotubes in electrical double layer capacitors. Carbon2007;45(13):2511–8.

[58] Gooding JJ. Nanostructuring electrodes with carbonnanotubes: a review on electrochemistry and applications forsensing. Electrochim Acta 2005;50:3049–60.

[59] Da Silva S, Basseguy R, Bergel A. Electron transfer betweenhydrogenase and 316 L stainless steel: identification of ahydrogenase-catalyzed cathodic reaction in anaerobic mic.J Electroanal Chem 2004;561:93–102.

C A R B O N 4 7 ( 2 0 0 9 ) 8 8 4 – 8 9 3 893

[60] Collins PG. Extreme oxygen sensitivity of electronicproperties of carbon nanotubes. Science2000;287(5459):1801–4.

[61] Collins PG, Hersam M, Arnold M, Martel R, Avouris P. Currentsaturation and electrical breakdown in multiwalled carbonnanotubes. Phys Rev Lett 2001;86(14):3128–31.

[62] Kasumov AY, Bouchiat H, Reulet B, Stephan O, Khodos II,Gorbatov YB, et al. Conductivity and atomic structure ofisolated multiwalled carbon nanotubes. Europhys Lett1998;43(1):89–94.

[63] Frackowiak E, Beguin F. Electrochemical storage of energy incarbon nanotubes and nanostructured carbons. Carbon2002;40(10):1775–87.

[64] Peixoto N, Chernyy N, Mason JP, Sunderam S, Weinstein SL,Schiff SJ, et al. Electrochemical evaluation of implantediridium oxide electrodes for low frequency non-pulsatilestimulation. Epilepsia 2005;46(s8):297–297.

[65] Baibarac M, Gomez-Romero P. Nanocomposites based onconducting polymers and carbon nanotubes: from fancymaterials to functional applications. J Nanosci Nanotechnol2006;6:289–302.

[66] Papanikolaou G, Pantopoulos K. Iron metabolism andtoxicity. Toxicol Appl Pharmacol 2005;202(2):199–211.

[67] Liu X, Guo L, Morris D, Kane AB, Hurt RH. Targeted removal ofbioavailable metal as a detoxification strategy for carbonnanotubes. Carbon 2008;46(3):489–500.