ARTICLE

Effect of Temperature on the

Capacitance of Carbon Nanotube
Supercapacitors
Charan Masarapu, Hai Feng Zeng, Kai Hsuan Hung, and Bingqing Wei*
Department of Mechanical Engineering, University of Delaware, Newark, Delaware 19716

upercapacitors are becoming attrac-

S tive power sources in memory backup

devices, electric vehicles, military
weapons, space equipment, and in a num-
ABSTRACT The effect of temperature on the kinetics and the diffusion mechanism of the ions in a
supercapacitor assembled with single-walled carbon nanotube (SWNT) film electrodes and an organic electrolyte
were thoroughly investigated. An improved room temperature performance of the supercapacitor was observed
ber of day-to-day electronic equipment.1⫺4
due to the combined effects of an increase in the conductivity of the SWNT films and surface modifications on the
Especially for military, space, and electric ve-
SWNT films by repeatedly heating and cooling the supercapacitor between the temperatures of 25 and 100 °C.
hicle applications, supercapacitors with
high energy and power densities that can Modified Randles equivalent circuit was employed to carry out an extensive analysis of the Nyquist spectra
withstand harsh temperature environments measured at different temperatures between 25 and 100 °C in order to understand the fundamentals of the
are extremely desirable. To design a super- capacitive and resistive variations in the supercapacitor. The experimental results and their thorough analysis
capacitor for a specific application that re- will have significant impact not only on the fundamental understanding of the temperature-dependent electrode/
quires high energy density or high power electrolyte interfacial properties but also on supercapacitor design with appropriate electrode materials for
density or both, proper electrode materials
numerous industrial and consumer applications. The supercapacitor with SWNT film electrodes was capable of
and a suitable electrolyte are to be chosen.
withstanding current densities as high as 100 A/g, yielding eminent specific power density values of about 55 kW/
For this purpose, it is of utmost importance
to have the fundamental understanding of kg. Ultralong galvanostatic chargeⴚdischarge cycling over 200 000 cycles with a constant current density of 20 A/
the interface properties of the electrode/ g at 25 and 100 °C, respectively, showed excellent stability in capacitance with more than 80% efficiency. The
electrolyte, such as the interface structure, usage of such a supercapacitor potentially enables far-reaching advances in backup energy storage and high pulse
charge transfer, ion diffusion, etc., which power applications.
play a vital role in governing the rate capa-
bility and specific capacitance of the super- KEYWORDS: single-walled carbon nanotube · temperature
capacitor. Cyclic voltammetry (CV) and elec- effect · supercapacitor · impedance spectroscopy · ionic diffusion
trochemical impedance spectroscopy (EIS)
are some of the powerful characterization
capacitors assembled with conventional
techniques that have the capability to
carbonaceous materials in an organic elec-
throw light on such electrode/electrolyte
trolyte have been limited to around 70
properties in a supercapacitor assembly. °C.16,17 However, these studies did not ad-
Carbon, in its various forms, is currently dress the effects of rigorous temperature
the most extensively examined and widely conditions such as the repeated heating
utilized electrode material in and cooling, continuous high temperature,
supercapacitors.5⫺7 While extensive work etc. on the supercapacitor performance,
has been done in understanding the kinet- which are practically important for commer-
ics and diffusion mechanism of ions in the cial applications. Moreover, there has been
activated carbons8,9 and in nanosized car- no systematic work on the temperature ef-
bon materials10,11 at room temperature, very fects on the nanosized carbon electrodes in *Address correspondence to
little research has been realized on the weib@udel.edu.
supercapacitors compared with the con-
temperature-dependent studies. Organic ventional carbonaceous materials. Received for review May 14, 2009
electrolytes in general have wide voltage Carbon nanotubes (CNTs), with their and accepted July 01, 2009.
windows of operation and high decomposi- unique architecture,18 excellent conductiv-
Published online July 7, 2009.
tion temperatures, compared to aqueous ity,19 and high surface area20 have drawn 10.1021/nn900500n CCC: $40.75
electrolytes.12⫺15 Until recently, the significant attraction as nanosized superca-
temperature-dependent studies on super- pacitor electrodes.21,22 Rate capability © 2009 American Chemical Society

www.acsnano.org VOL. 3 ▪ NO. 8 ▪ 2199–2206 ▪ 2009 2199

ARTICLE were used to extensively analyze the electrode/electro-
lyte interface properties in the supercapacitor with re-
spect to changes in the temperature. Effects of re-
peated heating and cooling on the performance of the
supercapacitor was studied and compared to the re-
sults of the supercapacitor with activated carbon fab-
ric.33 In addition, ultralong cycle galvanostatic
charge⫺discharge has been carried out at tempera-
tures 25 and 100 °C with repeated heating and cooling
to demonstrate the long-term stability of the superca-
pacitor with the SWNT film electrodes.

RESULTS AND DISCUSSION

Figure 1 shows the electrode preparation from the
free-standing SWNT films and the schematics of a su-
percapacitor cell assembly (detailed information can be
found in the Methods section). The cyclic voltammo-
grams (CVs) of the cell were measured in the voltage
window of ⫺1.5 to ⫹1.5 V (actual potential of opera-
Figure 1. Photographs illustrating the free-standing SWNT electrode tion is 1.5 V) with scan rates of 50, 100, 200, 500, and
preparation process. (a) Purified SWNT film on water surface. (b) Free- 1000 mV/s at each set of temperatures, 25, 50, 75, and
standing SWNT film on the wire loop. (c) Electrodes with 1.25 cm diam- 100 °C (Supporting Information, Figure S1). Figure 2a,b
eter punched from the free-standing SWNT film. (d) SWNT film on the wire
loop after punching the electrodes. (e) Schematic representation of the su- displays the CVs measured at 50 and 1000 mV/s at 25,
percapacitor coin cell assembly with the free-standing SWNT films with- 100, and 25 °C after cooling back from 100 °C. The vol-
out any current collectors. A Teflon O-ring was used to seal the coin cell to tammograms were close to rectangular shape at 50
perform high-temperature measurements up to 100 °C.
mV/s as well as at a very high scan rate of 1000 mV/s, in-
studies determine the rate at which the supercapacitor dicating ideal capacitive behavior within this wide tem-
can be charged and discharged. Extremely high rate ca- perature window. Figure 2c shows the capacitance val-
pabilities can be achieved by utilizing CNT electrodes ues of the cell calculated from the voltammograms
in supercapacitors when compared to the conventional (Supporting Information S1) measured at different tem-
carbonaceous materials.23,24 Especially the single-walled peratures and plotted with respect to the scan rate.
carbon nanotubes (SWNTs) can be synthesized as long Interesting results can be observed from these fig-
tube structures25 that are favorable for preparing en- ures. An increasing pseudocapacitive behavior in the
tangled mats of free-standing electrodes with the filtra- voltammograms was observed with the increase in the
tion technique.26 The free-standing films can be used di- temperature from 25 to 100 °C, similar to perfectly re-
rectly as electrode materials without any complex versible Faradaic reactions taking place in the capaci-
nanotube coating procedures,27 and there is no require- tor. Generally, the Faradaic process involves the elec-
ment for the current collector. The supercapacitors as- tron transfer across the double layers with a consequent
sembled with such free-standing SWNT films have change of oxidation state, and the measurable capaci-
shown good capacitance stability and high rate capabil- tance arising due to this process is designated as a
ity and yielded reasonable specific capacitance values pseudocapacitance. In Figure 2a,b, the pseudocapaci-
at room temperature in both aqueous and organic tive behavior is indicated by the ovals. The capacitance
electrolytes.28⫺31 at all scan rates gradually increased with the increase
Here, we report the temperature effects of a coin in temperature (Figure 2c). Figure 2c also illustrates the
cell supercapacitor assembled with free-standing SWNT performance of the cell with repeated heating and cool-
film electrodes and tetraethylammonium ing. The value of the capacitance at all scan rates ob-
tetrafluoroborate⫺polypropylene carbonate (TEABF4/ tained at 25 °C after cooling back from 100 °C has a
PC) electrolyte under rigorous temperature conditions slight increase (about 10%) compared to the capaci-
(25⫺100 °C). Unlike the filtration technique, where the tance values obtained initially at the same tempera-
nanotubes are dispersed in a solution and filtered to ob- ture of 25 °C. In the present experiment, the maximum
tain the free-standing SWNT films,26 we utilized the potential was limited to 1.5 V during the CV measure-
SWNT films directly synthesized by a chemical vapor ments, but as the cell is being heated, even though the
deposition (CVD) method reported elsewhere32 with potential is within the safe decomposition limit of the
post-purification techniques that did not involve a com- PC electrolyte,12 some electrolyte will evaporate and in-
plex filtration process to prepare the electrodes. Electro- creases the vapor pressure inside the cell. The trace
chemical characterization techniques, cyclic voltamme- amount of evolved gases from the electrolyte
try, and the electrochemical impedance spectroscopy system34,35 modifies the surface of the SWNT electrode,
2200 VOL. 3 ▪ NO. 8 ▪ MASARAPU ET AL. www.acsnano.org
 ARTICLE
The principle objective of the EIS measurements is
to gain insight into the temperature dependence of
the resistive and capacitive elements and their effects
on the performance of the supercapacitor. Figure 3a
shows the Nyquist spectra in the frequency range of
100 kHz to 10 mHz measured at equilibrium open cir-
cuit potential (⬃0 V) at different temperatures from 25
to 100 °C with repeated heating and cooling. The
Nyquist spectrum can be represented by a modified
Randles circuit36 with a set of resistors and capacitors
in series and parallel as shown in Figure 3b. The first in-
tersection point on the real axis of the Nyquist spec-
trum in the high-frequency region provides the value
of the ohmic resistance of the electrolyte and the inter-
nal resistance of the electrode materials and is repre-
sented as RS. The semicircle in the high-frequency re-
gion to midfrequency is modeled by an interfacial
charge transfer resistance RCT and the double layer ca-
pacitance CDL connected parallel to each other. After
the semicircle, the Nyquist spectrum shows a long tail
in the low-frequency region pertaining to the diffusion
of ions into the bulk of the electrode. The transition
from the high-frequency semicircle to the midfre-
quency tail is represented by the Warburg element Wo,
which is expressed as A/(j␻)n, where A is called the War-
burg coefficient, ␻ is the angular frequency, and n is
an exponent. At very low frequencies, an ideally polar-
izable capacitance would give rise to a straight line par-
allel to the imaginary axis with a mass capacitance rep-
resented as CL. However, from Figure 3a, it can be
observed that the spectra obtained at low-frequency
Figure 2. Cyclic voltammograms of the cell at different tem- range (⬃1 Hz to 10 mHz) are inclined at an angle to the
peratures with various scan rates: (a) 50 and (b) 1000 mV/s.
The ovals represent pseudocapacitive behavior in the real impedance axis, suggesting that a resistive ele-
voltammograms with perfectly reversible Faradaic reac- ment is associated with CL. This resistance can be desig-
tions taking place in the capacitor. The pseudocapacitive be- nated as the low-frequency leakage resistance Rleak
havior is present in the CVs at 25 °C even after cooling from
100 °C. (c) Capacitance of the cell calculated from CVs at dif- and is placed parallel to CL in the equivalent circuit. Fig-
ferent scan rates and at different temperatures versus the ure 3c shows the Nyquist spectra measured at differ-
scan rate. Also illustrated is the capacitance of the cell with ent temperatures plotted on a shifted real axis empha-
repeated heating and cooling between 25 and 100 °C. Due to
the introduction of Faradaic reactions, an increase in the ca- sizing the high-frequency semicircle and the
pacitance is observed at high temperature. midfrequency spike (100 kHz to 1 Hz).
To obtain all of the equivalent circuit parameters,
which possibly leads to the physisorbtion of the electro-
the fitting of the impedance expression obtained from
lyte ions at high temperature, giving rise to the Fara-
the equivalent circuit to the experimental Nyquist spec-
daic currents in the CVs (Figure 2a,b) and causing an in-
tra in the whole frequency range from 100 kHz to 10
crease in the capacitance when heated from 25 to 100
mHz was performed using the EIS data fitting program
°C (Figure 2c). The surface modification of the SWNT
ZVIEW. All of the fitting parameters of the Nyquist spec-
films is partially irreversible, and the physisorbtion re-
tra at all temperatures are shown in the table format
mains even at 25 °C (refer to the ovals in Figure 2a,b) af- in Figure 3d. The value of RS is significantly decreased
ter cooling from 100 °C. This is evidenced by the CV with the increase in the temperature from 25 to 100 °C.
curve which shows the pseudocapacitive behavior with The decrease is caused by an increase in the conductiv-
a capacitance value obtained at 25 °C, slightly higher ity of the electrolyte due to the enhanced mobility of
than the previously obtained value at the same temper- the ions in the electrolyte with the increase in the tem-
ature of 25 °C (Figure 2c). The effects of this phenom- perature but is not contributed by the intrinsic temper-
enon on the values of the capacitance can be further ex- ature dependence of the SWNTs themselves. This is
plained in detail by the electrochemical impedance verified by measuring the variation in the resistance of
spectroscopy (EIS) measurements presented in the sub- the SWNT film from 25 to 100 °C without the involve-
sequent paragraphs. ment of the electrolyte. Similar to the SWNT bundles,
www.acsnano.org VOL. 3 ▪ NO. 8 ▪ 2199–2206 ▪ 2009 2201
ARTICLE cation is also increasing the conductivity of the SWNT
films. This is evident from the lower values of the RCT at
25 °C after cooling from 100 °C, compared to the val-
ues obtained at 25 °C before heating from the fitting pa-
rameters in Figure 3d.
Some of the important features of the CVs in Figure
2 can also be explained based on the fitting param-
eters in Figure 3d. Generally, the Warburg element rep-
resents the diffusion of ions into the pores on the sur-
face of the electrode during the transition from the
high-frequency semicircle to the midfrequency spike.
Porous electrodes such as activated carbons generate
an exponent value of n ⫽ 0.5, where the spike is at an
angle of 45° to the real axis at room temperature.38 For
the case of SWNTs in this study, it can be observed that
the value of n is 0.6 at 25 °C with the spike near verti-
cal to the real impedance axis yielding an ideal capaci-
tive behavior. The higher value of the exponent indi-
cates that the ion diffusion is taking place only on the
surface of the SWNT electrode and suggests that the
SWNTs are not behaving as typical porous electrodes,33
which is evident from the BET surface area measure-
ments on the SWNT films (see Supporting Information
S3 and Figure S2). Also, the transition from the high-
frequency semicircle to the midfrequency spike as
shown in Figure 3c is occurring at relatively higher fre-
Figure 3. Electrochemical impedance spectroscopy measurements on the su-
percapacitor at different temperatures with repeated heating and cooling. (a) quency of ⬃221 Hz in the case of SWNTs compared to
In the frequency range from 100 kHz to 10 mHz. (b) Randles equivalent cir- porous carbon electrodes.23,33 The high rate capability
cuit representing the circuit elements in the Nyquist spectra. The equivalent of the SWNT film supercapacitor compared to the po-
circuit is used to fit the Nyquist spectra using the software ZVIEW. (c) High-
frequency response of the Nyquist spectra of panel a showing the semicircles rous electrodes is the direct consequence of the surface
plotted on the shifted real axis for clarity. (d) Equivalent circuit parameters diffusion effects with high exponent values in the War-
obtained from the fitting results. The performance of the cell when cooled burg element and a higher transition frequency from
back to 25 °C improved due to betterment of RS, RCT, CDL, and Wo values com-
pared to the as-assembled cell at 25 °C. the semicircle to the vertical spike. Therefore, the CVs
do not deviate much from the rectangular shape even
which act as metallic conductors at ambient tempera- at a scan rate as high as 1000 mV/s in the case of the su-
tures,37 the resistance of the SWNT film increased by percapacitor with carbon nanotube electrodes. Even
about 8% at 100 °C compared to the value at 25 °C (con- though the rate capability is high for carbon nanotube
ductivity is about 0.2 S/cm at 25 °C). electrodes, the nonporous nature and the surface diffu-
The physisorbtion of the electrolyte ions on the sion effects limit the number of ions that form the
SWNT electrodes with the increase in the temperature double layer at the electrode surface and yield lower
can be explained based on the variations in the values gravimetric capacitance compared to the capacitor em-
of CDL and RCT. At high frequencies, the double layer ca- ploying an activated carbon material.33
pacitance CDL arises from the charges present close to With an increase in the temperature from 25 to 100
the surface of the SWNT film that overcome the energy °C, the value of n is increasing and the Warburg coeffi-
to migrate with the alternating potential. The surface cient A is decreasing. This leads to a better rate capabil-
modification of the SWNTs by the evolved gases from ity of the supercapacitor at high temperature and is evi-
the electrolyte presumably provides more accessible dent from the loop obtained at 100 °C, which is more
sites for the mobile charges near the surface of the rectangular than the loop obtained at 25 °C with a scan
SWNT electrode thus leading to an increase in the value rate of 1000 mV/s in Figure 2b. However, it is interest-
of CDL from 3 ␮F at 25 °C to 28 ␮F at 100 °C. Also, the ing to note from the impedance data of Figure 3c that
value of the CDL obtained at 25 °C after cooling from 100 the variation of temperature does not lead to any shift
°C is higher than the value obtained at 25 °C before in the frequency of the transition from the high-
heating. So it should be obvious that the physisorb- frequency semicircle to the midfrequency spike which
tion is taking place even at 25 °C after cooling back from is occurring at ⬃221 Hz. This behavior suggests that
100 °C. RCT actually represents the resistance of the there is no time dependence in the charging process
charges in the pores of the SWNT electrode. A reduc- on temperature variation, and the physisorbtion of the
tion in the value of RCT implies that the surface modifi- electrolyte ions on the surface of the SWNT films at high
2202 VOL. 3 ▪ NO. 8 ▪ MASARAPU ET AL. www.acsnano.org
 ARTICLE
Figure 4. Long cycle galvanostatic chargeⴚdischarge studies on the supercapacitor measured at 25 and 100 °C with differ-
ent current densities from 10 to 100 A/g. Chargeⴚdischarge curve obtained with a constant current density of 10 A/g at (a)
25, (b) 100, and (c) 25 °C after cooling from 100 °C. (d) Chargeⴚdischarge curve obtained with a constant current density of
100 A/g at 100 °C. (e) Long cycle discharge capacitance with a constant current density of 10 A/g at 25 and 100 °C with re-
peated heating and cooling. (f) Long cycle discharge capacitance with current densities of 50 and 100 A/g measured at 25 and
100 °C. (g) Ragone chart of the supercapacitor obtained from the discharge curves measured at different constant current
densities at 25 and 100 °C. (h) Displaying the ultralong cycling stability of the cell by running 246 700 and 224 000 cycles of
chargeⴚdischarge with a current density of 20 A/g at 25 and 100 °C. (i) Cell resistance variations in the coin cell at different
stages of the chargeⴚdischarge measurements.

temperature (Figure 2) is happening at a fast rate with- ity of the fitting parameters. The low-frequency leak-
out affecting the rate capability. From Figure 3d, the pa- age resistance of the cell is decreasing from 38 k⍀ at ini-
rameters are getting better at 25 °C with repeated heat- tial 25 °C to about 8 k⍀ at 100 °C. Lower leakage
ing and cooling thus leading to an improved resistance implies a higher leakage current that causes
performance at 25 °C after cooling from 100 °C com- the cell to self-discharge faster.16 Considering the varia-
pared to the performance of the as-assembled superca- tion of the overall parameters in Figure 3d, the perfor-
pacitor at 25 °C. mance of the supercapacitor coin cell with SWNT elec-
At very low frequency, the mass capacitances esti- trodes at 25 °C is improving with repeated heating and
mated from the fitting parameters in Figure 3d are com- cooling. Such kind of improvement in the performance
parable to the values obtained from CVs measured at was not observed in case of the supercapacitor with ac-
the scan rate of 50 mV/s (Figure 2c) showing the valid- tivated carbon fabric electrodes.33
www.acsnano.org VOL. 3 ▪ NO. 8 ▪ 2199–2206 ▪ 2009 2203
ARTICLE The improvement in the supercapacitor perfor-
mance with SWNTs with repeated heating and cooling
cell can be operated with a large current density toler-
ance in the temperature range from 25 to 100 °C.
is further elucidated by the constant current After all the long cycle measurements at different
charge⫺discharge cycling on the supercapacitor (see current densities with repeated heating and cooling,
Supporting Information S4 for a sequence of the cell was run at 25 °C for 246 700 and at 100 °C for ad-
charge⫺discharge measurements). Figure 4a⫺c shows ditional 224 000 charge⫺discharge cycles with a con-
typical charge⫺discharge curves of the supercapacitor stant current density of 20 A/g. Figure 4h shows the dis-
at initial 25 °C, when heated to 100 °C and after cooling charge capacitance of all the cycles at both
back to 25 °C, respectively, run at a constant current temperatures. The cell showed excellent ultralong cycle
density of 10 A/g. A significant reduction in the IR drop stability at both 25 and 100 °C with an efficiency of ap-
can be observed in the charge⫺discharge curve ob- proximately 82 and 84%, respectively. The efficiency
tained at 25 °C after cooling from 100 °C compared to was calculated based on the degradation in the value
the IR drop in the curve of the as-assembled sample at of discharge capacitance after all the charge⫺discharge
25 °C. cycles (246 700 cycles at 25 °C and 224 000 cycles at
To demonstrate the long-term stability of the super- 100 °C) from the initial discharge capacitance obtained
capacitor, several tens of thousands of charge⫺ from the first charge⫺discharge cycle.
discharge cycles were run at different constant current Since supercapacitors are power devices, the most
densities from 100 mA/g (81.46 mA/cm2) to 100 A/g important figure of merit to be considered is the stabil-
(81.46 A/cm2) at 25 and 100 °C by repeated heating and ity of their impedance during the charge⫺discharge cy-
cooling (see Supporting Information S4 for a sequence cling with the repeated heating and cooling. Figure 4i
of charge⫺discharge measurements and Figure S3 for shows the resistance RS of the supercapacitor cell ob-
tained from the EIS measurements at different stages of
some charge⫺discharge measurements at 100 mA/g
the charge⫺discharge measurements (see Supporting
and 1 A/g). Figure 4d shows the charge⫺ discharge
Information S4 for the sequence of charge⫺discharge
curves of the supercapacitor at 100 °C, measured with
measurements carried on the coin cell). There is no sig-
constant current density of 100 A/g. The long cycle dis-
nificant increase in the impedance of the cell even af-
charge capacitances of the cell calculated from the gal-
ter running several thousands of charge⫺discharge
vanostatic charge⫺discharge curves are displayed in
cycles at 100 °C, implying a stable power output from
Figure 4e,f. An impressive result of the supercapacitor
the supercapacitor. It took more than 3 months to per-
is obvious from these measurements. The charge⫺
form all of the charge⫺discharge measurements with
discharge curves represented ideal capacitive behavior
repeated heating and cooling on the coin cell, and the
even at a current density as high as 100 A/g. As the cur-
cell is still operational. Figure 4 provides strong evi-
rent density was very high, large overshoots were ob-
dence that the supercapacitor with the free-standing
served in the charge⫺discharge curves over 1.5 V. Since
SWNT film electrodes in TEABF4/PC electrolyte can be
the upper potential of operation for the TEABF4/PC elec-
conveniently operated at any current density from 100
trolyte system can be as high as 2.5 V,12 the overshoots
mA/g to 100 A/g in the temperature range from 25 to
did not cause any negative effects such as electrolyte 100 °C with no significant cell damage.
decomposition or cell failure. This is obvious from the There are several advantages of utilizing the free-
long cycle performance tested by running over 10 000 standing SWNT film synthesized from our method32 as
charge⫺discharge cycles at 25 and 100 °C both at 50 the supercapacitor electrodes. The foremost is that the
and 100 A/g (Figure 4f). The slight changes in the slope SWNT can be handled as an entire film during the whole
of the charge⫺discharge curves lead to variations in process of the synthesis, purification, and electrode
the calculated capacitances (Supporting Information preparation, which is extremely simple compared to
S5) at high current densities as shown in the Figure 4e,f the filtration technique.26 The process can be easily ex-
(for example, ⬃23 ⫾ 6 F/g at 100 A/g). panded to acquire SWNT film electrodes with uniform
Figure 4g shows the plot of specific energy density thickness and on a large scale. The complex mixing and
versus specific power density, also called a Ragone coating procedure27 involved with the powder samples
plot,39 for the supercapacitor at 25 and 100 °C, respec- is not required. The SWNT film itself acts as a good con-
tively. The total weight of the two SWNT film electrodes ductor, thus eliminating the usage of the current collec-
was considered in estimating the energy and power tors. By using the supercapacitors with the SWNT film
density (Supporting Information S5). The supercapaci- electrodes and the TEABF4/PC electrolyte, we demon-
tor has a maximum power density of about 55 kW/kg at strated that the operating temperature can be easily ex-
100 °C obtained with the discharge curve run at a con- panded to 100 °C with extremely high rate capability.
stant current density of 100 A/g. The maximum energy
density obtained from the discharge curve run with 100 CONCLUSIONS
mA/g at 100 °C is 4.2 Wh/kg. To obtain an optimum en- In conclusion, the electrochemical performance of
ergy and power density for different applications, the the coin cell supercapacitor assembled with SWNT film
2204 VOL. 3 ▪ NO. 8 ▪ MASARAPU ET AL. www.acsnano.org
 ARTICLE
electrodes using 1 M TEABF4/PC electrolyte was sys- mance of the supercapacitor. The long cycle gal-
tematically studied in a temperature window rang- vanostatic charge⫺discharge measurements
ing from 25 to 100 °C. A pseudocapacitive behavior showed excellent stability in capacitance at both 25
was observed in the cyclic voltammograms at high and 100 °C, even at high current densities of 50 and
temperature, which was attributed to the physisorb- 100 A/g. The stability of the cell was also reflected
tion of the electrolyte ions on the surface of the under rigorous temperature conditions and the su-
SWNTs. Fitting parameters of Nyquist spectra ob- per long galvanostatic charge⫺discharge cycles. The
tained from a modified Randles equivalent circuit re- usage of such a supercapacitor potentially enables
vealed that repeated heating and cooling between far-reaching advances in civilian, military, and aero-
25 to 100 °C enhanced the room temperature perfor- space applications.

METHODS measurements at current densities of 100 mA/g and 1 A/g. This

Fabrication of the SWNT Macrofilm Electrodes. The SWNT films were material is available free of charge via the Internet at http://
directly deposited on a copper foil by CVD method28 using fer- pubs.acs.org.
rocene as carbon feedstock/catalyst and sulfur as an additive to
promote SWNT growth to a high percentage. There is no addi- REFERENCES AND NOTES
tional carbon source (e.g., xylene, hexane, and methane) required 1. Miller, J. R.; Simon, P. Electrochemical Capacitors for
for the synthesis. The deposited films were peeled off from the Energy Management. Science 2008, 321, 651–652.
copper foil and were purified by first heating in air up to 450 °C 2. Burke, A. R&D Considerations for the Performance and
for 1 h to remove amorphous carbon and then treated in 9 M HCl Application of Electrochemical Capacitors. Electrochim.
solution for 0.5 h to remove the iron catalyst particles. After the Acta 2007, 53, 1083–1091.
acid treatment, the film was thoroughly washed with DI water. 3. Conway, B. E.; Pell, W. G. Double-Layer and
The purification process did not involve any filtration technique, Pseudocapacitance Types of Electrochemical Capacitors
and the SWNTs were handled as an entire film through the en- and Their Applications to the Development of Hybird
tire purification process. The complex entanglement of the long Devices. J. Solid State Electrochem. 2003, 7, 637–644.
SWNT strands provided strong mechanical stability for the film. 4. Kotz, R.; Carlen, M. Principles and Applications of
Figure 1a shows a purified SWNT film floating on the water sur- Electrochemical Capacitors. Electrochim. Acta 2000, 45,
face. The film was transferred directly from the beaker onto a 2483–2498.
wire mesh (loop), as shown in Figure 1b. After drying the film, 5. Simon, P.; Gogotsi, Y. Materials for Electrochemical
two 1.25 cm diameter electrodes were punched with an arch Capacitors. Nat. Mater. 2008, 7, 845–854.
punch. The photographs of the punched electrodes and the wire 6. Chmiola, J.; Yushin, G.; Gogotsi, Y.; Portet, C.; Simon, P.;
mesh after the electrodes were punched can be seen from Fig- Taberna, P. L. Anomalous Increase in Carbon Capacitance
ure 1c,d. The thickness as well as the weight of the electrodes can at Pore Sizes Less than 1 Nanometer. Science 2006, 313,
be controlled by varying the number of layers of the SWNT films 1760–1763.
deposited on the wire mesh. 7. Frackowiak, E. Carbon Materials for Supercapacitor
Supercapacitor Assembly. The supercapacitor was assembled in Application. Phys. Chem. Chem. Phys. 2007, 9, 1774–1785.
a 2032 coin cell with a pair of SWNT films weighing 0.7 mg each, 8. Frackowiak, E.; Beguin, F. Carbon Materials for the
without any current collectors. A 1 M TEABF4 (Alfa Aesar) dis- Electrochemical Storage of Energy in Capacitors. Carbon
solved in battery-graded PC (Alfa Aesar) solvent was used as the 2001, 39, 937.
electrolyte. A Wattman glass microfiber filter paper soaked in 9. Qu, D. Y. Studies of the Activated Carbons Used in Double-
the electrolyte was used as the separator. The coin cell was Layer Supercapacitors. J. Power Sources 2002, 109,
sealed using a Teflon gasket to withstand temperatures up to 403–411.
150 °C. The supercapacitor assembly was carried out in a glove- 10. Frackowiak, E.; Metenier, K.; Bertagna, V.; Beguin, F.
box (MBraun, Unilab) with oxygen and water vapor levels less Supercapacitor Electrodes from Multiwalled Carbon
than 0.1 ppm. For the temperature-dependent measurements Nanotubes. Appl. Phys. Lett. 2000, 77, 2421–2423.
from 25 up to 100 °C, the coin cell was placed in a box oven (LIN- 11. Liu, C. G.; Liu, M.; Li, F.; Cheng, H. M. Frequency Response
BERG/BLUE M) capable of maintaining a very stable set temper- Characteristic of Single-Walled Carbon Nanotubes as
ature up to one-tenth of a °C. Figure 1e shows the experimental Supercapacitor Electrode Material. Appl. Phys. Lett. 2008,
arrangement of the supercapacitor. 92, 143108.
Electrochemical Measurements. The electrochemical characteriza- 12. Ue, M.; Ida, K.; Mori, S. Electrochemical Properties of
tion at different temperatures was done by measuring the cy- Organic Liquid Electrolytes Based on Quaternary Onium
clic voltammograms, electrochemical impedance spectra, and Salts for Electrical Double-Layer Capacitors. J. Electrochem.
constant current charge⫺discharge curves. Cyclic voltammo- Soc. 1994, 141, 2989–2996.
grams and the electrochemical impedance spectra of the super- 13. Kurzweil, P.; Chwistek, M. Electrochemical Stability of
capacitor were measured with an EG&G Parstat 2273 poten- Organic Electrolytes in Supercapacitors: Spectroscopy and
tiostat/galvanostat. The galvanostatic charge⫺discharge Gas Analysis of Decomposition Products. J. Power Sources
examination was carried out by an Arbin battery testing system. 2008, 176, 555–567.
14. Macneil, D. D.; Dahn, J. R. Can an Electrolyte for Lithium-
Acknowledgment. This work was financially supported by Ion Batteries Be Too Stable. J. Electrochem. Soc. 2003, 150,
the National Science Foundation (NSF CMMI #0753462) and the A21–A28.
Delaware NASA EPSCoR Research Infrastructure Development 15. Li, X. H.; Meng, Y. Z.; Zhu, Q.; Tjong, S. C. Thermal
(RID) Award. Decomposition Characteristics of Poly(propylene
carbonate) Using TG/IR and Py-GC/MS Techniques. Polym.
Supporting Information Available: Cyclic voltammograms of Degrad. Stab. 2003, 81, 157–165.
the coin cell supercapacitor at different temperatures from 25 16. Kotz, R.; Hahn, M.; Gallay, R. Temperature Behavior and
to 100 °C measured at different scan rates from 50 to 1000 mV/s. Impedance Fundamentals of Supercapacitors. J. Power
Surface area measurements on purified SWNT macrofilms. Se- Sources 2006, 154, 550–555.
quence of charge⫺discharge measurements carried on the coin 17. Liu, P.; Verbrugge, M.; Soukiazian, S. Influence of
cell at different temperatures. Long cycle charge⫺discharge Temperature and Electrolyte on the Performance of

www.acsnano.org VOL. 3 ▪ NO. 8 ▪ 2199–2206 ▪ 2009 2205

ARTICLE Activated-Carbon Supercapacitors. J. Power Sources 2006,
156, 712–718.
39. Service, R. E. New ‘Supercapacitor’ Promises to Pack More
Electrical Punch. Science 2006, 313, 902–905.
18. Wei, B. Q.; Vajtai, R.; Jung, Y.; Ward, J.; Zhang, R.; Ramanath,
G.; Ajayan, P. M. Organized Assembly of Carbon
Nanotubes. Nature 2002, 416, 495–496.
19. Wei, B. Q.; Vajtai, R.; Ajayan, P. M. Reliability and Current
Carrying Capacity of Carbon Nanotubes. Appl. Phys. Lett.
2001, 79, 1172–1174.
20. Peigney, A; Laurent, Ch; Flahaut, E; Bacsa, R. R.; Rousset, A.
Specific Surface Area of Carbon Nanotubes and Bundles
of Carbon Nanotubes. Carbon 2001, 39, 507–514.
21. Niu, C; Sichel, E. K; Hoch, R; Moy, D.; Tennent, H. High
Power Electrochemical Capacitors Based on Carbon
Nanotube Electrodes. Appl. Phys. Lett. 1997, 70,
1480–1482.
22. Dalton, A. B.; Collins, S.; Munoz, E.; Razal, J. M.; Ebron, V. H.;
Ferraris, J. P.; Coleman, J. N.; Kim, B. G.; Baughman, R. H.
Super-Tough Carbon-Nanotube Fibres. Nature 2003, 423,
703.
23. Obreja, V. V. N. On the Performance of Supercapacitors
with Electrodes Based on Carbon Nanotubes and Carbon
Activated MaterialOA Review. Physica E 2008, 40,
2596–2605.
24. Futaba, D. N.; et al. Shape-Engineerable and Highly
Densely Packed Single-Walled Carbon Nanotubes and
Their Application as Super-Capacitor Electrodes. Nature
2006, 5, 987–994.
25. Zhu, H. W.; Xu, C. L.; Wu, D. H.; Wei, B. Q.; Vajtai, R.; Ajayan,
P. M. Directly Synthesis of Long Nanotube Strands. Science
2002, 296, 884–886.
26. Baughman, R. H.; et al. Carbon Nanotube Actuators.
Science 1999, 284, 1340–1344.
27. An, K. H.; Kim, W. S.; Park, Y. S.; Moon, J. M.; Bae, D. J.; Lim,
S. C.; Lee, Y. S.; Lee, Y. H. Electrochemical Properties of
High Power Supercapacitors Using Single-Walled Carbon
Nanotube Electrodes. Adv. Funct. Mater. 2001, 11, 387–392.
28. Shiraishi, S.; Kurihara, H.; Okabe, K.; Hulicova, D.; Oya, A.
Electrical Double Layer Capacitance of Highly Pure Single-
Walled Carbon Nanotubes (hipco Buckytubes) in
Propylene Carbonate Electrolytes. Electrochem. Commun.
2002, 4, 593–598.
29. Barisci, J. N.; Wallace, G. G.; Chattopadhyay, D.;
Papadimitrakopoulos, F.; Baughman, R. H. Electrochemical
Properties of Single-Wall Carbon Nanotube Electrodes. J.
Electrochem. Soc. 2003, 150, E409–E415.
30. Frackowiak, E.; Jurewicz, K.; Delpeux, S.; Beguin, F.
Nanotubular Materials for Supercapacitors. J. Power
Sources 2001, 97, 822–825.
31. Barisci, J. N.; Wallace, G. G.; Baughman, R. H.
Electrochemical Studies of Single-Wall Carbon Nanotubes
in Aqueous Solutions. J. Electroanal. Chem. 2000, 488,
92–98.
32. Zhu, H. W.; Wei, B. Q. Direct Fabrication of Single-Walled
Carbon Nanotube Macro-Films on Flexible Substrates.
Chem. Commun. 2007, 29, 3042–3044.
33. Hung, K.; Masarapu, C. ; Tsehao, K.; Wei, B. Q. Wide-
Temperature Range Operation Supercapacitors from
Nanostructured Activated Carbon Fabric. J. Power Sources
2009, 193, 944–949.
34. Hahn, M.; Wursig, A.; Gallay, R.; Novak, P.; Kotz, R. Gas
Evolution in Activated Carbon/Propylene Carbonate Based
Double-Layer Capacitors. Electrochem. Commun. 2005, 7,
925–930.
35. Hahn, M.; Kotz, R.; Gallay, R.; Siggel, A. Pressure Evolution
in Propylene Carbonate Based Electrochemical Double
Layer Capacitors. Electrochim. Acta 2006, 52, 1709–1712.
36. Randles, J. E. B. Kinetics of Rapid Electrode Reactions.
Discuss. Faraday Soc. 1947, 1, 11–19.
37. Zhu, H.; Zhao, G.-L.; Masarapu, C.; Young, D. P.; Wei, B. Q.
Super Small Energy Gaps of Single-Walled Carbon
Nanotube Strands. Appl. Phys. Lett. 2005, 86, 203107.
38. Wang, K. P.; Teng, H. Structural Feature and Double-Layer
Capacitive Performance of Porous Carbon Powder Derived
from Polyacrylonitrile-Based Carbon Fiber. J. Electrochem.
Soc. 2007, 154, A993–A998.

2206 VOL. 3 ▪ NO. 8 ▪ MASARAPU ET AL. www.acsnano.org