Nano Res (2010) 3: 452–458
452
DOI 10.1007/s12274-010-0006-8
Research Article
Nano Res (2010) 3: 452–458
Aqueous Supercapacitors on Conductive Cotton
Mauro Pasta1,2, Fabio La Mantia2, Liangbing Hu2, Heather Dawn Deshazer2, and Yi Cui2 (
)
1
Dipartimento di Chimica Inorganica, Metallorganica e Analitica “Lamberto Malatesta”, Università degli Studi di Milano, Via Venezian 21,
20133 Milano, Italy
2
Department of Materials Science and Engineering, Stanford University, Stanford, CA 94305, USA
Received: 17 March 2010 / Revised: 27 April 2010 / Accepted: 4 May 2010
© The Author(s) 2010. This article is published with open access at Springerlink.com
ABSTRACT
Wearable electronics offer the combined advantages of both electronics and fabrics. In this article, we report the
fabrication of wearable supercapacitors using cotton fabric as an essential component. Carbon nanotubes are
conformally coated onto the cotton fibers, leading to a highly electrically conductive interconnecting network.
The porous carbon nanotube coating functions as both active material and current collector in the supercapacitor.
Aqueous lithium sulfate is used as the electrolyte in the devices, because it presents no safety concerns for
human use. The supercapacitor shows high specific capacitance (~70–80 F·g–1 at 0.1 A·g–1) and cycling stability
(negligible decay after 35,000 cycles). The extremely simple design and fabrication process make it applicable
for providing power in practical electronic devices.
KEYWORDS
Supercapacitor, wearable electronics, energy storage, carbon nanotubes
1. Introduction
Wearable electronics constitute a new class of devices
with an array of novel functionalities, which make
them ideal for emerging applications such as highperformance sportswear, wearable displays, new classes
of portable power, and embedded health monitoring
devices [1–3]. All these electronic applications require
light mass, wearable power conversion, and storage
devices. Textiles are flexible and porous materials
made by weaving or pressing natural or synthetic
fibers, which gives them important properties such
as flexibility, stretchability, and light mass. Thus, the
ideal wearable power device would incorporate a
textile as a component. Among textiles, cotton has the
advantages of being an inexpensive natural fiber, which
Address correspondence to yicui@stanford.edu
is highly hydrophilic and light.
In this work, we describe the conformal coating of
single-walled carbon nanotubes (SWNTs) onto cellulose
fibers to make porous conductors. The fabrication
process is simple and scalable, similar to those widely
used for dyeing fibers and fabrics in the textile
industry. The SWNT coating makes these textiles highly
conductive, with sheet resistance less than 1 Ω·sq–1.
The conductive textiles show outstanding mechanical
and chemical properties. A thorough characterization
of this material was recently reported [4], and can be
referenced for specific details regarding physical
properties. Specifically, the porous structure and high
surface area make this material particularly interesting
for supercapacitor applications. As an extension of
our previous study, here we report the performance
Nano Res (2010) 3: 452–458
453
of a symmetric SWNT/cotton supercapacitor in a
safe aqueous electrolyte, whose components are fully
compatible with wearable device applications.
2. Experimental
Sodium dodecylbenzene sulfonate (SDBS) and lithium
sulfate anhydrous ( ≥ 99.99% trace metals basis) were
purchased from Sigma Aldrich. Nitric acid (68%) was
purchased from EMD Chemicals. Fluffy cotton sheets
were purchased from Wal-Mart Stores, Inc.
Electrochemical characterization was carried out
using a BioLogic VMP3 potentiostat/galvanostat
multichannel equipped with an electrochemical
impedance spectroscopy (EIS) board. A double
junction Ag|AgCl|KCl (3.5 mol·L–1) reference electrode
(RE) was used in the measurements. The double
junction was employed to prevent OH– diffusion and
reaction at the Ag/AgCl interface. In addition, the
RE potential was monitored after each measurement
to confirm that no change had taken place. All the
measurements were performed under inert (nitrogen)
atmosphere and at room temperature.
2.1
SWNT ink fabrication
10 mg·mL–1 SDBS surfactant was dissolved in deionized
(DI) water with the help of bath sonication. Laser
ablation SWNTs were then dispersed in the surfactant solution to a concentration of 1.6 mg·mL–1.
After bath sonication for 5 min, the SWCNT dispersion
was probe-sonicated for 30 min at 200 W (VC 505,
Sonics Inc).
2.2
SWNT/cotton preparation procedure
A fluffy cotton sheet with thickness of ~1–2 mm was
dipped into the SWNT ink and immediately removed.
Due to the strong absorption, the ink quickly coated the
textile. The textile with SWNT ink was subsequently
dried in an oven at 120 °C for 10 min to remove water.
The mass of the SWNTs was obtained from the mass
difference before and after the dipping and drying of
the cotton sheet. This process was repeated to increase
the SWNT loading in the cotton. Figure 1 shows the
accumulated ink mass absorbed per cm2 of the cotton
after different numbers of dipping and drying cycles. A
Figure 1 The amount of ink absorbed per area by the cotton sheet
after dipping for different numbers of times
small variation (<5%) of absorbed ink mass per area
was observed when repeating the process.
The cotton fiber structure is comprised of multiple
individual cotton fibrils, which are in turn composed
of multiple microfibrils, bundled together. The microfibrils are made of poly-D-glucose chains, usually
arranged in crystalline, or partially crystalline, domains.
This structure allows the fibers to absorb large amounts
of water, or other polar solvents.
The scanning electron microscope (SEM) image in
Fig. 2(b) reveals the macroporous structure of a cotton
sheet and Fig. 2(c) shows the conformal coating of
SWNTs onto the fibers. This conformal coating is a
result of the mechanical flexibility of the individual
SWNTs and the strong binding energy between SWNTs
and cotton fibers [5], accounting also for the high
electrical conductivity of the textile. Previous studies
have shown that SWNT films have microscale porosity,
which is required to maximize the specific capacitance [6]. The microporous structure on top of the
macroporous textile leads to what we call a “double
porous structure” that facilitates the easy access of
electrolyte ions to the SWNTs, which is essential for
high power supercapacitor applications. Transmission
electron microscopy (TEM) images of the SWNT/cotton
fibers (Fig. 2(d)) show that the SWNTs are well
bonded to the fiber, forming a cross-linked network,
which provides conducting pathways.
454
Nano Res (2010) 3: 452–458
surface. However, this surfactant needs to be removed
to avoid side reactions, which would generate irreversible specific charge and self-discharge phenomena.
The surfactant removal is essential for the final device
performance, especially its cyclability and coulombic
efficiency. Washing with abundant DI water and pressing with a grid are sufficient to remove the surfactant.
A simple rule of thumb to evaluate whether all the
surfactants have been removed is to press the fabric
and see if bubbles are produced.
2.4
SWCNT/cotton pretreatment
Such porous textile conductors demonstrate excellent
electrical, mechanical, and chemical resistance performance [4] suggesting that the SWNTs adhere very
strongly to the cotton fibers, which is critical for
wearable electronic and power devices. Such strong
binding may be due to the following:
1) There are strong van der Waals forces existing
between SWNTs and the textile fibers [5].
2) Acid-treated SWNTs have carboxyl functional
groups on the surfaces and the ends, which can form
strong hydrogen bonds with the hydroxyl groups in
the cellulose fibers.
3) The flexibility of SWNTs allows them to conformally adhere to the surface of cotton fibers,
maximizing the surface contact area between SWNTs
and textile fibers [7].
The superior mechanical adhesion of SWNTs on
cotton (measured by the standard tape test and by
washing in water) is essential for high-speed roll-to-roll
fabrication and energy storage device stability.
After the SDBS removal, the textile exhibits a hydrophobic behavior which is incompatible with its
application in a water-based electrolyte. To improve
the surface hydrophilicity and remove the last traces
of surfactant, the textile was dipped into a 4 mol·L–1
HNO3 solution for about 6 h and then washed again
with DI water to remove the acid. Treating the nanotubes with nitric acid introduces a larger number of
oxygen-containing functional groups such as carboxyl,
lactones, and phenols and creates a more hydrophilic
surface structure [8]. The nitric acid treatment also has
two other advantages: it helps to remove impurities,
such as catalytic metal particles responsible for self
discharge problems [9] and induces hole doping, which
decreases the resistivity of the nanotubes [10–12].
Figure 2(a) shows the results of wettability tests of the
substrate after the acid treatment.
As mentioned above, the acid wash is also the final
step in the surfactant removal process. To confirm the
success of the washing process in removing all the
surfactants it is sufficient to observe the open circuit
voltage value against the RE. If the value lies around
0.0 V vs. RE, this means that there is still some
surfactant left, and it is necessary to repeat the
washing process. The observed open circuit voltage
is due to an unidentified secondary reaction involving
the surfactant. If the value lies around 0.4 V vs. RE, we
can be sure that all the surfactants have been eliminated
and proceed with the electrochemical tests.
2.3
2.5
Figure 2 (a) A 15 cm × 15 cm foldable textile conductor based
on a cotton sheet with sheet resistance of 2 Ω·sq–1. The water drop
test on the SWNT-coated cotton shows the highly hydrophilic
surface. (b) SEM image of the cotton sheet coated with SWNTs.
(c) Higher magnification image of the sample in (b) showing the
conformal coating of SWCNTs on the cotton surface. (d) TEM
image of SWNTs on a cotton fiber demonstrating that there is no
evidence of agglomeration of SWNTs
Surfactant removal
The material prepared as described above is highly
hydrophilic, due to the surfactant still present on the
Electrochemical tests
The performances of the prepared SWCNT/cotton
electrodes were tested in an aqueous 2 mol·L–1 Li2SO4
Nano Res (2010) 3: 452–458
solution. This electrolyte was chosen for the following
reasons:
1) Sulfates combine the advantages of electrochemical stability in the potential range investigated,
with low cost.
2) Among the sulfates, lithium sulfate has the
advantage of a higher solubility in water, as compared
to the less expensive sodium and potassium salts.
3) A 2 mol·L–1 solution of lithium sulfate has an
acidic pH (around 3.2); a slow H3O+ ion sorption has
been reported in the literature, giving an additional
capacity [13].
Before each measurement the electrolyte was purged
with nitrogen gas for 30 min to remove the dissolved
oxygen which could give rise to an irreversible
specific charge. All the measurements were performed
under nitrogen atmosphere.
3. Results and discussion
Figure 3 reports galvanic cycles (100 µA·cm–2) for
SWCNT/cotton materials with four different SWCNT
loadings (0.12, 0.24, 0.36, 0.47 mg·cm–2) corresponding
to one to four “dip and dry” cycles, referred to as cotton
#1, #2, #3, #4, respectively. As one can observe, the
specific charge (represented by the area underneath
the curves) increases with the SWCNT loading.
Figure 3 Galvanostatic cycling at apparent current density of
0.1 mA·cm–2 for working (continuous lines) and counter (dotted
lines) SWNT/cotton electrodes vs. Ag/AgCl 3.5 mol·L–1 in 2 mol·L–1
Li2SO4 electrolyte, for different mass loadings of SWNTs. Black
curve: cotton #1 (0.12 mg·cm–2); red curve: cotton #2 (0.24 mg·cm–2);
blue curve: cotton #3 (0.36 mg·cm–2); green curve: cotton #4
(0.47 mg·cm–2)
455
In the presence of an RE, the behavior of both
positive and negative electrodes can be monitored
during the cycling. With this approach some irreversible behavior in the cathodic side can be easily
identified during the first few cycles: the reason is the
low hydrogen evolution overpotential on SWCNTs,
as already reported in Ref. [13]. Moreover, the shape
of the curve—especially the one for high SWCNT
loadings—suggests the presence of a slow ion sorption,
leading to an additional specific charge at lower specific
currents. Ion sorption is a process that allows more
charge to accumulate in the inner Helmholtz layer, but
it is normally a slow process (time constant >10–6 s)
with respect to the accumulation of charge in the outer
Helmholtz layer (time constant <10–10 s). Therefore, it
can only be observed at low scan rates.
In order to thoroughly investigate this phenomenon, the differential curves, obtained by taking the
derivative of the charge Q with respect to the electrode
potential UE, are reported in Fig. 4 for a current
intensity of 100 µA. We want to stress that at higher
mass loading (lower specific current) a peak appears
in the positive electrode of the supercapacitor around
0.4 V vs. Ag/AgCl 3.5 mol·L–1. This peak is evidence
of an electrochemical adsorption or an ion sorption
process, as described above.
The electrochemical adsorption/ion sorption is even
more evident in the Nyquist plots (Fig. 5) of the EIS
performed on the electrodes. From the shape of the
Figure 4 Differential curves relative to the positive electrode for
different SWNT loadings. Black curve: cotton #1 (0.12 mg·cm–2);
red curve: cotton #2 (0.24 mg·cm–2); blue curve: cotton #3
(0.36 mg·cm–2); green curve: cotton #4 (0.47 mg·cm–2)
456
Figure 5 Nyquist plots for supercapacitors with different SWNT
loadings. Black curve: cotton #1 (0.12 mg·cm–2); red curve: cotton
#2 (0.24 mg·cm–2); blue curve: cotton #3 (0.36 mg·cm–2); green
curve: cotton #4 (0.47 mg·cm–2)
Nyquist plot and the value of the impedance at
different mass loadings, we can obtain the following
information:
1) The distorted capacitive semicircle at high
frequencies (see insert in Fig. 5) is evidence of a current
distribution along the thickness of the electrode and of
an electrochemical process which is different from
simple double layer charging/discharging [14–16].
2) The increase in the capacitive part of the
impedance after the high-frequency semicircle is due
to the blocking nature of the electrochemical process,
thus, indicating that it is an electrochemical adsorption
or an ion sorption.
3) The small difference in the value of the imaginary
part of the impedance at low frequencies between the
different mass loadings suggests that the electrochemical process is limited by the transport of the
adsorbed species in the pores. Also, the inflection
point (*) at low frequencies is indicative of such a
phenomenon.
4) After reaching a mass loading of 0.24 mg·cm–2,
the internal resistance of the supercapacitor is due to
the transport of ions in the electrolyte.
There are four main species in solution: Li+, SO42–,
H3O+, and OH–. The Li+ and SO42– ions are present in
Nano Res (2010) 3: 452–458
such high concentrations that their ion sorption should
not be transport limited. OH– is present in such a low
concentration that it is unable to account for the
specific charge under the peak in Fig. 4 (green curve).
Thus, it can be concluded that hydronium (H3O+) ion
sorption is responsible for the EIS behavior.
Figure 6 shows the specific capacitances at various
specific currents. The curve representing the lowest
SWCNT content (cotton #1) shows the worst performances in terms of both specific capacitance
(55 F·g–1) and capacitance retention at higher specific
currents: this behavior is due to the low electrical
conductivity of the substrate (as also indicated by the
Nyquist plot, Fig. 5). At this mass loading, the SWCNT
content is insufficient to form a good interconnecting
network. For cotton #2, the specific capacitance is
higher, with a maximum value of about 70 F·g–1, and it
is relatively constant in the current range from 10 µA·g–1
to 1 mA·g–1. Comparing cotton #2 and cotton #3, it is
found that the performances are quite similar, in terms
of maximum specific capacitance, but the specific
capacitance decreases more rapidly at higher current
values for cotton #3. For cotton #4 the maximum
specific capacitance is higher (around 80 F·g–1), but it
continuously decreases with increasing current. This
behavior can be interpreted by using the framework
of the theory of current density distribution in porous
Figure 6 Plots of specific capacitance versus specific current
for SWNT/cotton electrodes with different SWNT loadings.
Black curve: cotton #1 (0.12 mg·cm–2); red curve: cotton #2
(0.24 mg·cm–2); blue curve: cotton #3 (0.36 mg·cm–2); green
curve: cotton #4 (0.47 mg·cm–2)
457
Nano Res (2010) 3: 452–458
electrodes. This theory can be summarized by the
following equation [17]:
∂ 2 ΔΦ ρ s I
=
∂x 2
af
(1)
where ΔΦ is the potential drop at the solid/liquid
interface, x represents the position within the porous
electrode, ρs is the resistivity of the electrolyte solution,
a the characteristic length of the porous electrode,
f the void fraction, and I the current density at the
solid/liquid interface (which is dependent on the
position x). The characteristic length, a, increases
on increasing the ratio between the volume of the
electrode and the active surface area, V/A. Equation (1)
is valid when the electric resistance of the solid phase
in the porous electrode is negligible. The behavior of
the electrodes at different current densities is better
understood when the areal capacitance (capacitance
per geometrical area) with respect to the mass loading
is reported (see Fig. 7). When the apparent current
density (the current density per geometrical area) is
low (black curve), Eq. (1) reduces to a constant current
density distribution along the pores of the electrode,
and therefore, the areal capacitance increases linearly
with the mass loading. At higher apparent current
densities (the red, blue, and green curves in Fig. 7),
the electrodes with a larger a value (lower mass
Figure 7 Plots of areal capacitance versus mass loading for
different apparent current densities. Black curve; 0.01 mA·cm–2;
red curve: 0.2 mA·cm–2; blue curve: 2 mA·cm–2; green curve:
10 mA·cm–2
loading) still charge homogeneously, while ones with
lower a (higher mass loading) show a distribution of
current density, and consequently a distribution of
surface charge. As observed in Fig. 7 for the red, blue,
and green curves, the higher the apparent current
density, the stronger the deviation from linearity. The
deviation from linearity always leads to a reduction
in the areal capacitance. Depending on the application,
the optimal loading is easily predicted.
The cycling stability of the SWNT/cotton#2 device
was tested at room temperature, under a nitrogen
atmosphere with a three electrode configuration cell,
as shown in Fig. 8. The electrolyte used was 2 mol·L–1
Li2SO4 and the experiment was performed with an
apparent current density of 1 mA·cm–2. No fading was
observed after 35,000 cycles. The coulombic efficiency
after 35,000 cycles was higher than 99% of the initial
value, with the specific capacitance being around
45 F·g–1.
4. Conclusions
A supercapacitor has been fabricated using a conformally coated SWNT/cotton fabric as both active
material and current collector (resistance lower than
1 Ω·sq–1). The device has excellent cycling performance
(good capacity retention after 35,000 cycles) and high
Figure 8 Cycling stability of SWNT/cotton #2 supercapacitor
device in 2 mol·L–1 Li2SO4 aqueous electrolyte at 1 mA·cm–2
apparent current density
458
Nano Res (2010) 3: 452–458
specific capacitance (70–80 F·g–1 at 0.1 mA·cm–2). The
device as prepared is fully wearable since both the
textile (cotton) and the lithium sulfate electrolyte
are compatible with the human body. It can also
be integrated into wearable devices. By means of
impedance spectroscopy and differential curves, we
have highlighted an additional reversible capacitance
due to a slow ion sorption process.
Potential improvements in the future need to be
directed towards an asymmetric supercapacitor since
the hydrogen evolution overpotential limits the
operative voltage range, and therefore the high power
limit.
[6] An, K. H.; Kim, W. S.; Park, Y. S.; Choi, Y. C.; Lee, S. M.;
Acknowledgements
[10] Parekh, B. B.; Fanchini, G.; Eda, G.; Chhowalla, M. Improved
Chung, D. C.; Bae, D. J.; Lim, S. C.; Lee, Y. H. Supercapacitors using single-walled carbon nanotube electrodes.
Adv. Mater. 2001, 13, 497–500.
[7]
Iijima, S.; Brabec, C.; Maiti, A.; Bernholc, J. Structural
flexibility of carbon nanotubes. J. Chem. Phys. 1996, 104,
2089–2092.
[8] Yu, X.; Lin, B.; Gong, B.; Lin, J.; Wang, R.; Wei, K. Effect
of nitric acid treatment on carbon nanotubes (CNTs)-cordierite
monoliths supported ruthenium catalysts for ammonia
synthesis. Catal. Lett. 2008, 124, 168–173.
[9] Tohji, K.; Goto, T.; Takahashi, H.; Shinoda, Y.; Shimizu, N.;
Jeyadevan, B.; Matsuoka, I.; Saito, Y.; Kasuya, A.; Ohsuna, T.;
Hiraga, K.; Nisshina, Y. Purifying single-walled nanotubes.
Nature 1996, 383, 679.
conductivity of transparent single-wall carbon nanotube
Y. C. acknowledges support from the King Abdullah
University of Science and Technology (KAUST)
Investigator Award (No. KUS-l1-001-12).
Open Access: This article is distributed under the terms
of the Creative Commons Attribution Noncommercial
License which permits any noncommercial use,
distribution, and reproduction in any medium,
provided the original author(s) and source are credited.
thin films via stable postdeposition functionalization. Appl.
Phys. Lett. 2007, 90, 121913.
[11] Zhou, W.; Vavro, J.; Nemes, N. M.; Fischer, J. E.; Borondics,
F.; Kamaras, K.; Tanner, D. B. Charge transfer and Fermi
level shift in p-doped single-walled carbon nanotubes. Phys.
Rev. B 2005, 71, 205423.
[12] Beaudrouet, E.; Le Gal La Salle, A.; Guyomard, D.
Nanostructured manganese dioxides: Synthesis and properties
as supercapacitor electrode materials. Electrochim. Acta
2009, 54, 1240–1248.
References
[1] Gniotek, K.; Krucińska, I. The basic problems of textronics.
Fibres Text. East. Eur. 2004, 12, 13–16.
[2] Lukowicz, P.; Kirstein, T.; Troster, G. Wearable systems for
health care applications. Method. Inform. Med. 2004, 43,
232–238.
[3] Park, S.; Jayaraman, S. Smart textiles: Wearable electronic
systems. MRS Bull. 2003, 28, 585–591.
[4] Hu, L.; Pasta, M.; La Mantia, F.; Cui, L.; Jeong, S.; Deshazer,
H. D.; Choi, J. W.; Han, S. M.; Cui, Y. Stretchable, porous,
and conductive energy textiles. Nano Lett. 2010, 10, 708–714.
[5] Hertel, T.; Walkup, R. E.; Avouris, P. Deformation of carbon
nanotubes by surface van der Waals forces. Phys. Rev. B
1998, 58, 13870–13873.
[13] Prosini, P. P.; Pozio, A.; Botti, S.; Ciardi, R. Electrochemical
studies of hydrogen evolution, storage and oxidation on
carbon nanotube electrodes. J. Power Sources 2003, 118,
265–269.
[14] Newman, J. S.; Tobias, C. W. J Theoretical analysis of
current distribution in porous electrodes. Electrochem. Soc.
1962, 109, 1183–1191.
[15] de Levie, R. Porous electrodes in electrolyte solutions. Ⅳ.
Electrochim. Acta 1964, 9, 1231.
[16] Ng, S. H.; La Mantia, F.; Novak, P. A multiple working
electrode for electrochemical cells: A tool for current density
distribution studies. Angew. Chem. Int. Ed. 2009, 48, 528–532.
[17] Daniel-Bek, V. S. Polarization of porous electrodes. I.
Distribution of current and potential within an electrode.
Zh. Fiz. Khim. 1948, 22, 697–710.