Hydrogen storage capacity of catalytically grown carbon nanofibers

Article abstract of DOI:10.1021/jp051371a

In 1996, R. T. K. Baker, and N. M. Rodriguez claimed to have synthesized a new type of carbon nanofiber material capable of storing large amounts of hydrogen at room temperature and pressures above 100 bar, thus making it a powerful candidate for a very efficient energy storage system in mobile applications. Consequently, many scientists all over the world tried to test and verify these findings, however, with partly inconsistent results. We present here for the first time independent hydrogen storage measurements for several types of nanofibers, both synthesized by our group following precisely the specifications given in the literature as well as original samples supplied by Rodriguez and Baker for this study. The hydrogen storage capacities at room temperature and pressures up to 140 bar were quantified independently by gravimetric and volumetric methods, respectively. No significant hydrogen storage capacity has been detected for all carbon nanofibers investigated. ? 2005 American Chemical Society.

Full text of DOI:10.1021/jp051371a

J. Phys. Chem. B 2005, 109, 14979-14989
14979
Hydrogen Storage Capacity of Catalytically Grown Carbon Nanofibers
Matthias Rzepka,*,† Erich Bauer,† Gudrun Reichenauer,‡ Thomas Schliermann,§
Babette Bernhardt,| Klaus Bohmhammel,| Eva Henneberg,| Uta Knoll,
Heinz-Eberhard Maneck, and Wolfgang Braue#
BaVarian Center for Applied Energy Research (ZAE Bayern e.V.), DiV. 1, 85748 Garching, Germany, BaVarian
Center for Applied Energy Research (ZAE Bayern e.V.), DiV. 2, 97074 Wu¨rzburg, Germany, Physics
Department, Wu¨rzburg UniVersity, Am Hubland, 97074 Wu¨rzburg, Germany, Technische UniVersita¨t
Bergakademie Freiberg, 09599 Freiberg, Germany, Federal Institute for Materials Research and Testing,
12200 Berlin, Germany, and Deutsches Zentrum fu¨r Luft und Raumfahrt (DLR), Materials Research Institute,
51147 Ko¨ln, Germany
ReceiVed: March 16, 2005; In Final Form: June 8, 2005
In 1996, R. T. K. Baker, and N. M. Rodriguez claimed to have synthesized a new type of carbon nanofiber
material capable of storing large amounts of hydrogen at room temperature and pressures above 100 bar, thus
making it a powerful candidate for a very efficient energy storage system in mobile applications. Consequently,
many scientists all over the world tried to test and verify these findings, however, with partly inconsistent
results. We present here for the first time independent hydrogen storage measurements for several types of
nanofibers, both synthesized by our group following precisely the specifications given in the literature as
well as original samples supplied by Rodriguez and Baker for this study. The hydrogen storage capacities at
room temperature and pressures up to 140 bar were quantified independently by gravimetric and volumetric
methods, respectively. No significant hydrogen storage capacity has been detected for all carbon nanofibers
investigated.
TABLE 1: Hydrogen Storage Capacities of Carbon
Nanofibers Exhibiting Different Fiber Architectures As
1. Introduction
Published by Rodriguez and Baker2
Fuel cells, which convert hydrogen directly and with high
efficiency to electric energy, may be one of the main energy
converters of the near future. However, the lack of suitable
hydrogen storage systems greatly hinders the introduction of
fuel cell systems, especially in mobile applications. Commonly
used hydrogen storage systems such as pressurized gas tanks,
cryo tanks, or metal hydride storage systems all suffer from
major drawbacks such as insufficient storage densities, high
costs, or short lifetimes.
nanofiber
structure
adsorbed H2
in wt %a
nanofiber
structure
adsorbed H2
in wt %a
tubular
herringbone
herringbone
11
58
68
herringbone
platelet
platelet
61
54
46
a The value of adsorbed H2 is given in units of weight fraction wt(H2)/
wt(H2 + C).
It was therefore of worldwide interest when, in December
1996, Rodriguez and Baker claimed that their carbon nanofibers
can store up to 70 wt % of hydrogen1-3 at room temperature
and 140 bar hydrogen pressure. Table 1 summarizes their
published values for different nanofiber samples.
Carbon nanofibers have been known for a long time.4-7 Their
lengths and diameters are typically in the range of 1-100 µm
and 10-200 nm, respectively. They can be grown in different
morphologies (see, e.g., Table 1) depending on the synthesis
parameters applied; details on the effect of the synthesis
parameters on the structural properties are given elsewhere.8-10
The unique microstructure of platelet and herringbone-type
nanofiberssconsisting of graphene sheets stacked in an angle
with respect to the fiber axissmakes them interesting candidates
for a variety of applications, for example, gas storage systems,11
electrochemical devices,12,13 or catalyst supports.14
Rodriguez and Baker concluded from their data that some
sort of intercalation of hydrogen atoms or molecules in the
carbon nanofibers might provide storage densities highly
exceeding the storage density expected for physical adsorp-
tion on the fiber surface. To date, the potential mechanism
for high-capacity hydrogen storage in carbon nanofibers
still remains unclear, but it obviously has to differ from the
storage mechanism in activated carbon or carbon nanotubes,
where the hydrogen molecules are physically adsorbed at the
surface of the micropores (<2 nm) at the inner tube walls or
stored in the interstices between adjacent nanotubes, respec-
tively.
In recent years, different groups tried to reproduce the results
of Rodriguez and Baker in similar or modified carbon nano-
structures. It was claimed15 that alkali-doped carbon nanotubes
can store up to 20 wt % of hydrogen, which later was
demonstrated16 to be only an effect of hydroxide formation as
a result of the moisture content in the hydrogen gas applied
in the storage tests. In 1999, Fan et al. and Cheng et al.17,18
* Corresponding author. Tel: +49-89-32944231. E-mail: Rzepka@
muc.zae-bayern.de.
† Bavarian Center for Applied Energy Research, Div. 1.
‡ Bavarian Center for Applied Energy Research, Div. 2.
§ Wu¨rzburg University.
| Technische Universita¨t Bergakademie Freiberg.
Federal Institute for Materials Research and Testing.
# Materials Research Institute.
10.1021/jp051371a CCC: $30.25 © 2005 American Chemical Society
Published on Web 07/14/2005

14980 J. Phys. Chem. B, Vol. 109, No. 31, 2005
Rzepka et al.
TABLE 2: Carbon Nanofibers Investigated in the Framework of This Study
synthesis parameters
catalyst pretreatment sequence
gas composition
(gas flow of components (Nml/min)),
tCVD (min), TCVD (°C),
gas composition
(gas flow of components (Nml/min)),
tpre (min), Tpre (°C)
supplier
catalyst
n.sp.a
amount of catalyst applied (mg)
#1
#2
#3
#4
#5
#6
#7
#8
#9
#10
#11
Catalytic Materials Inc.
F1-21
n.sp.
n.sp.
n.sp.
n.sp.
n.sp.
n.sp.
n.sp.
Catalytic Materials Inc.
F1-27
Catalytic Materials Inc.
F1-28
Catalytic Materials Inc.
F1-63
Catalytic Materials Inc.
F1-64
n.sp.
n.sp.
n.sp.
n.sp.
n.sp.
n.sp.
n.sp.
n.sp.
n.sp.
Catalytic Materials Inc.
G1
n.sp.
BAM, Berlin, synthesized
according to ref 9
BAM, Berlin, synthesized
according to refs 19 and 20
BAM, Berlin, synthesized
according to ref 21
ZAE Bayern,
Wu¨rzburg
ZAE Bayern,
Wu¨rzburg
Fe0.7Cu0.3
Ni0.98Cu0.02
He, 60 min, 25...500 °C
H2/He (10/90), 120 min, 500 °C
He, 60 min, 25...350 °C
H2/He (10/90), 120 min, 350 °C
He, 60 min, 25...500 °C
H2/He (10/90), 120 min, 500 °C
H2/He (9/91), 60 min, 25...600 °C H2/He
(9/91), 60 min, 600 °C He; 60 min, 600 °C;
H2/He (9/91),
C2H4/H2/He (10/60/30) 120 min,
600 °C, 50 mg
C2H4/H2 (20/5) 120 min,
600 °C, 120 mg
C2H4/H2/He (10/60/30)
120 min, 600 °C, 50 mg
C2H4/H2 (48/12)
90 min, 600 °C, 50 mg
C2H4/H2 (80/20) 60 min,
600 °C, 12 mg
Fe0.85Ni0.1
Cu0.05
Ni0.7Cu0.3
Ni0.7Cu0.3
120 min, 600 °C
a Not specified by supplier.
reported on up to 13 wt % hydrogen storage capacity for tubular-
shaped vapor-grown carbon nanofibers, whereas in graphitic
nanofibers Gupta and Srivastava measured up to 10 wt %
hydrogen;19,20 this range was confirmed by Browning et al.21
which found a hydrogen storage capacity of 7 wt % for carbon
nanofibers.
2. Materials
As published by Rodriguez and Baker et al.,2,3 some batches
of their carbon nanofibers showed hydrogen uptake capacities
in the range between 10 and 70 wt %. In the meantime, they
have established a standardized synthesis procedure yielding a
reproducible fiber composition and morphology. The hydrogen
storage capacity of these fibers, however, is lower, but it is still
in the range between 2 and 4 wt %30 depending on the
pretreatment (activation) procedure prior to hydrogen loading
applied. For an independent test of the hydrogen storage capacity
of these fibers, we have examined a total of six batches of
catalytically grown carbon nanofibers provided by Rodriguez
and Baker (Table 2).
Besides these optimistic publications, many other experi-
mental results showed no hydrogen storage capacity at all in
carbon nanofibers. Measurement of the hydrogen uptake with
a microbalance performed for vapor-grown nanofibers by
Stro¨bel et al. yielded values below 1.5 wt %;22 this tendency
was confirmed by Ahn et al.23 who measured values below 1
wt %. In the last few years, similar results have been published
for several nanostructures.24-26
In addition to these external samples, we also synthesized
carbon nanofibers in our laboratories using catalysts following
the specifications published for the materials with high hydrogen
storage capacity by Rodriguez and Baker,9,31 Gupta and Srivas-
tava,19,20 and Browning et al.,21 respectively (see also Table 2).
The fibers were synthesized according to these procedures with
various catalysts and by applying ethylene/hydrogen gas mix-
tures. The metal powder catalysts used were precipitated from
their nitrate solutions as carbonates, calcinated in air at 400 °C
for 4 h, and subsequently reduced with 10% H2 in He at the
temperatures given in the respective references (Table 2). The
synthesis of the carbon nanofibers was carried out in horizontal
quartz tube furnaces of different sizes (furnace at ZAE Bayern,
Wu¨rzburg: inner diameter 45 mm, length 100 cm; furnace at
BAM, Berlin: inner diameter 38 mm, length 40 cm) at 600 °C
with an ethylene/hydrogen ratio of 4:1 (samples #8, #10, and
#11) and 1:6 (samples #7 and #9), respectively.
Therefore, the issue of the hydrogen storage potential and
mechanism of carbon nanofibers is still unresolved. One possible
explanation for the great discrepancies between the published
values could be a strong dependence of the storage capacity on
details of the carbon fiber synthesis and the activation process
prior to the storage experiment. Therefore, no final statement
in terms of the hydrogen uptake of a certain type of fiber can
be made when the carbon nanofibers investigated have been
synthesized by different groups. For a conclusive storage test,
identical nanofiber batches from the same source have to be
employed.
For the study presented, we therefore investigated samples
provided by Rodriguez and Baker. In addition, we also
synthesized carbon nanofibers according to the specifications
given by Baker and Rodriguez,8,27 Gupta et al.,19,20 and
Browning et al.21 Special care has to be taken when measuring
the hydrogen uptake especially if only a small amount of sample
material (<100 mg) is available. For a detailed discussion of
the experimental difficulties and possible sources of experi-
mental errors, see, for example, refs 3, 28, and 29. To obtain
the most reliable results, we measured the hydrogen uptake with
three different experimental methods including volumetric
adsorption, volumetric desorption, and a gravimetric technique.
3. Characterization Techniques
Complementary to the measurements of the hydrogen storage
capacity, all CNF (carbon nanofiber) specimens have been fully
characterized in terms of structural, morphological, and chemical
properties.

Hydrogen Storage in Carbon Nanofibers
J. Phys. Chem. B, Vol. 109, No. 31, 2005 14981
3.1. Structure and Morphology. Scanning Electron Micros-
copy (SEM)/Transmission Electron Microscopy (TEM). The
morphological characterization of the CNF was performed with
a Hitachi 4100 SEM.
A Philips Tecnai F 30 transmission electron microscope
operating at an acceleration voltage of 300 kV has been
employed for the TEM investigations. The structural variants
of the carbon phase and the metallic particles retained from the
catalyst have been characterized via the combination of high-
resolution phase contrast imaging (HREM), selected and
convergent electron diffraction (SAD, CBED), X-ray mi-
croanalysis (EDS), and z-contrast imaging with a high-angle-
annular-dark-field (HAADF) detector in STEM mode. TEM
specimens were prepared via ultrasonic dispersion of as-received
CNF in isobutyl alcohol followed by the deposition of a drop
of the suspension on a holey carbon-film Pt-support grid. No
supplementary thinning technique was applied.
N2-Sorption. The nitrogen sorption measurements were carried
out using a commercial sorption apparatus (ASAP 2000 with
micropore option, Micromeritics, GA) equipped with 100, 10,
and 1 mbar pressure transducers.
Figure 1. Schematic setup of the volumetric apparatus. The reference
volume and the volume of the sample cell are about 13 and 8 mL,
respectively.
The samples were degassed at 350 °C in a vacuum for several
hours until a stable gas pressure was reached; the nanofibers
were stepwise exposed to an increasing nitrogen pressure at 77.4
K covering the relative pressure range from 0.001 to 1.
Typically, 100 mg of carbon material was used for each analysis.
The nitrogen sorption isotherms were evaluated to determine
the BET surface area in the relative pressure range between
0.04 and 0.24 using standard BET theory.32 The micropore
volume was determined from a Dubinin-Radushkevich (DR)
analysis of the sorption data in the relative pressure range of
0.001-0.1.32
XRD. X-ray diffraction (XRD) was carried out with a Philips
PW 1140 diffractometer in reflection geometry using Cu KR
radiation employing ground CNF material. The structural carbon
variant in CNFs were characterized by the interplanar spacing
d002 of the graphite (002) reflection.33
Raman Spectroscopy. The Raman spectra of the nanofibers
were taken with a micro-Raman setup (LabRam, Jobin-Yvon-
Horiba). The spectrometer has a focal length of 300 nm and is
equipped with a 1800-lines/mm grating. The spectral solution
was about 5.70 cm-1. The 514.532 nm line of an argon laser
with a power of about 80 mW was used as the excitation
wavelength. The scattered light was detected by a CCD camera
operating at 220 K. An Olympus Mplan 10 objective focused
the laser light onto the samples. To prevent the samples from
being modified, damaged, or heated, the intensity of the laser
light was reduced with filters.
Typically, the Raman intensity was collected from 800 to
1900 cm-1 covering the wavenumber regime of the characteristic
first-order Raman bands of graphitic carbons, the disorder
induced D-band (at approximately 1350 cm-1) and the G-band
(at approximately 1580 cm-1). The Raman spectra can be
evaluated in terms of width, position, and relative integrated
intensities of the first-order Raman bands to extract trends in
terms of the crystallinity34 of the carbon nanofibers.
evaluation, the software ECLIPSE provided with the instrument
was applied. For the analysis of the C 1s peak, the program
UNIFIT (version 32-33, 2001, University Leipzig) was used.
TDS. For further investigation of the active surface groups,
thermodesorption spectra have been collected with a custom-
built quadrupole mass spectrometer (Prisma, Pfeiffer Vacuum).
In the temperature range from -100 to +900 °C, the desorbing
gases are analyzed and recorded. In a typical experiment, 10
mg of the fibers were heated from room temperature to 900 °C
with a rate of 10 K/min while monitoring the characteristic
masses of released surface groups. This equipment was also
used to activate samples of about 200 mg for hydrogen uptake
experiments.
3.2. Hydrogen Storage. Special care has to be taken to avoid
artifacts during measuring hydrogen uptake from small amounts
(<100 mg) of carbon nanofiber samples. To achieve maximum
reliability, we have performed three independent measurements
with different equipment for each of the samples. In the
following sections, experimental setups are described in detail.
Volumetric Apparatus. The experimental setup for the volu-
metric measurements is shown schematically in Figure 1. The
central section of the apparatus consists of two high-pressure
cells interconnected by a valve. This section of the instrument
is thermally stabilized by a thermostat to (0.01 °C to avoid
thermal drifts which otherwise would dominate the experimental
error. In addition, the actual temperature inside each cell is
measured independently via thermocouples. The gas pressure
is measured by two piezoresistive pressure transducers (resolu-
tion 0.01 bar) connected to the central cells by thin, stainless-
steel capillaries. A difference between room temperature and
thermostat temperature is accounted for by corrections for the
amount of gas inside these capillaries. All temperature and
pressure sensor values are recorded automatically by a PC. To
remove residual water in the hydrogen (6.0 grade) used for the
storage analysis, a liquid nitrogen trap was inserted in the gas
supply line. High-pressure tests over several days showed no
detectable leakage rate for the two central cells (leak rate <0.01
bar/day).
X-ray Photon Emission Spectroscopy (XPS). The total amount
of oxygen present at the fibers surface as well as the chemical
environment of oxygen and carbon was investigated by XPS
using an ESCALAB 200X (VG Instruments, East Grinstead,
U.K.). For the experiment, ground CNF material was attached
to one side of a double-sided adhesive Scotch tape, and the
second side of the tape was glued to the sample holder. The
sample was exposed to Mg KR radiation (1253.6 eV). For data
Prior to each measurement, the sample was weighed and
inserted into the sample cell. Inside the sample cell, the carbon
fibers are enclosed in a gas-permeable microfilter element to
avoid the spreading of the fibers inside the apparatus especially
upon fast pressure changes. Thensas described latersoptional

14982 J. Phys. Chem. B, Vol. 109, No. 31, 2005
Rzepka et al.
TABLE 3: Summary of the Characterization Results for the
Pristine Samples
BET
micropore surface
density surface volume
oxygen d002
fiber morphology (g/cm3)a (m2/g)b (cm3/g)
(at.-%)c (nm)d
1
2
3
4
5
6
7
8
9
herringbone
herringbone
herringbone
herringbone
1.98
2.03
1.99
2.00
119
53
31
7
8
6
1
0.343
0.341
0.342
0.343
271
263
150
94
175
116
200
144
0.106
0.096
0.061
0.038
0.065
0.051
0.080
0.051
herringbone
herringbone
segmented
herringbone
2.14
2.17
2.05
2.13
3
1.7
0.339
0.337
0.341
Figure 2. Calibration experiments data (without carbon samples). The
apparent hydrogen uptake reflects the experimental inaccuracy of the
volumetric instrument of approximately (0.04 mg.
10 herringbone
11 herringbone
0.346
0.347
a The densities were determined via He pycnometry. b BET surface
areas and micropore volumes were measured with nitrogen sorption.
c XPS measurements provided the surface oxygen content. d The d002
values were derived from XRD measurements.
activation procedures can be applied. Subsequently, the whole
apparatus is evacuated for several hours by a turbo molecular
pump to a pressure level below 10-7 mbar. Finally, the valve
connecting the two high-pressure cells is closed.
The hydrogen uptake of the sample at about 50, 90, and 140
bar was measured in a three-stage process. At the beginning of
each stage, the reference volume is filled with hydrogen to a
certain pressure level (90, 140, and 200 bar, respectively). After
about 30 min, the whole system is in thermal equilibrium, and
the valve to the sample cell is opened. Due to the expansion of
the hydrogen gas into the sample cell volume, the pressure in
the sample cell increases while the pressure in the reference
volume decreases to a final value of about 50, 90, and 140 bar
(values determined by the volumes of the reference and the
sample cell). An uptake of hydrogen by the sample will cause
an additional pressure decrease. If the time constant for the
hydrogen absorption is high (e.g., several hours), then the
corresponding pressure drop can be recorded directly as the
difference between the gas pressure immediately after opening
the valve and the final pressure after several hours. However,
if the time constant is short (e.g., milliseconds), the additional
pressure drop has to be calculated. To achieve sufficient
accuracy we have applied the van der Waals equation
Figure 3. Schematic diagram of the volumetric desorption setup.
them to the experimental data. This procedure leads to an
experimental error of about 0.04 mg (Figure 2).
Desorption Measurement. The apparatus shown in Figure 3
was designed for measuring hydrogen desorption from a storage
material. In a typical experiment, 200-250 mg of the material
was loaded in the heatable steel vessel (total volume 46.7 cm3).
After evacuation at 10-7 mbar, hydrogen at room temperature
was dosed onto the sample in steps of 30-40 bar until a pressure
of 130 bar was reached. For the experiments, hydrogen with a
purity of 99.999% (5.0) was used; possible traces of water were
removed by a liquid nitrogen cold-trap. Twenty-four hours was
allowed for the hydrogen uptake and the equilibration of the
system.
The amount of hydrogen in the vessel was then calculated
from the pressure, the temperature, and the volume of the
apparatus (reduced by the sample volume). For this calculation,
the van der Waals equation was used. Then the hydrogen
pressure was slowly released to atmospheric pressure; the
amount of hydrogen released was quantified by the mass flow
controller (MFC) in the outlet line (Figure 3). The hydrogen
uptake by the material under investigation results in a difference
between the initially calculated amount of hydrogen dosed to
the vessel and the amount of hydrogen released. This method
is 10 times less sensitive as the volumetric method described
above, but is very simple; it is also less sensitive to leaks of
the apparatus. As confirmed by dry runs, a minimum storage
capacity of 3 wt % can be reliably detected.
(p + an2/V2)(V - bn) ) nRT
(1)
with p and T the actual pressure and temperature of the hydrogen
gas, R ) 8.314 J/(mol K), the gas constant, a and b the (gas
specific) van der Waals constants, and n the amount of hydrogen
gas (in moles). V is the volume of the sample cell, the reference
volume, orsif the valve is opensthe combined volume of both
cells. For each experiment, the volume of the sample cell is
corrected for the displaced volume of the carbon sample
assuming a bulk density of the sample of 2.1 g/cm3 (see Table
3). Following eq 1, the amount n1 of gas before opening the
valve is calculated with the appropriate pressure and temperature
p1 and T1. Similarily, the amount n2 after opening the valve is
calculated with p2 and T2. The difference n1 - n2 directly yields
the amount of gas adsorbed.
The values a and b can be taken from the literature, and the
volumes V can be determined experimentally. However, to
achieve the highest accuracy possible and to avoid any additional
systematic errors, we determined all values experimentally in
runs without any sample or with steel reference samples with
well-defined volumes. As in these runs no gas should be
adsorbed, the values a, b, and Vi can be determined by fitting
GraVimetric Apparatus. To confirm the reliability of the
volumetric data, we independently determined the hydrogen
storage capacity applying a further experimental technique. In
the volumetric experiments, the effect of the hydrogen adsorp-
tion on the gas pressure in the volume of the sample cell is
evaluated to calculate the storage capacity. The third experi-
mental technique records the mass increase of the sample due
to hydrogen adsorption with a high-accuracy balance. The
principle of the experimental setup is shown in Figure 4.

Hydrogen Storage in Carbon Nanofibers
J. Phys. Chem. B, Vol. 109, No. 31, 2005 14983
Figure 4. Scheme of the gravimetric apparatus.
We used a modified Sartorius Supermicro S3D-P with balance
chambers, which can be pressurized up to 150 bar. While the
weighing cell is kept at a constant temperature (T1 ) 40 °C),
each of the suspended cells can be thermostated at a temperature
between -40 and 500 °C. Buoyancy corrections of the sample
and the reference have been performed using
mMH
2 p
F
∆m )
(2)
a
RT
RT + b -
p
(
)
with m and F the mass and the density of the sample, MH the
2
Figure 5. Hydrogen uptake by an activated carbon sample at room
temperature (open circles, gravimetric measurement; dot, volumetric
measurement).
molar mass of hydrogen, R the gas constant, and T and p the
temperature and pressure of the hydrogen gas, repectively. As
van der Waals constants we used a ) 202 000 (cm6 bar) mol-2
and b ) 22.34 cm3 mol-1. An additional buoyancy correction
for the weighing system was determined by measurements with
empty crucibles under otherwise identical conditions. The
resulting overall relative error for the determination of the
sample mass was about 0.25% at a typical net weight of the
sample of 10 mg.
The experimental procedure was as follows: About 5-10
mg of carbon sample was weighed and placed into the sample
crucible. The same mass of NaCl (which has a similar density
as carbon nanofibers) was placed into the reference crucible.
Then the whole balance chamber was rinsed with hydrogen
(purity 6.0) for several hours and subsequently evacuated with
an oil pump until both pressure and mass were stable. Similar
to the volumetric setups, an optional pretreatment procedure
can be applied to the sample prior to the actual adsorption
experiments. Hydrogen adsorption was measured under iso-
thermal conditions at 25 °C in steps of maximum 40 bar up to
a maximum pressure of 140 bar.
Figure 6. Comparison of the H2 uptake measured for a commercial
metal hydrid at 19.3 bar (cross symbol) with the ad-/desorption
isotherms provided by the supplier.36
%. This is in very good agreement with the value from the data
sheet provided by the supplier of the sample.36
Reference Measurements. We measured the hydrogen storage
capacity of two reference samples to confirm the accuracy of
the volumetric and gravimetric techniques applied. Figure 5
shows the result for an activated carbon sample. Both experi-
ments are in good agreement. The measured adsorption isotherm
is also well predicted by the results of Monte Carlo simulations
for various shaped pore geometries.35
Figure 6 shows the result of the volumetric method for a
commercial metal hydride getter. At a final gas pressure of 19.3
bar, the volumetrically measured hydrogen uptake is 0.93 wt
4. Results and Discussion
4.1. Structure and Morphology of the Samples Investi-
gated. A series of six CNF batches (samples #1 through #6,
Table 2) has been supplied by Catalytic Materials Inc. SEM
images of five batches are compiled in Figure 7. On a
micrometer scale, the fibers show the characteristic ribbonlike
morphology. The fiber thickness distribution is rather broad
containing fibers between 200 and 600 nm in diameter. Coiled
fibers may also be present. On the microscopic scale, the

14984 J. Phys. Chem. B, Vol. 109, No. 31, 2005
Rzepka et al.
Figure 7. Ribbonlike morphologies (SE images) of different CNF samples supplied by Catalytic Materials Inc.: (a) #1 (F1-21), (b) #2 (F1-27),
(c) #3 (F1-28), (d) #4 (F1-63), and (e) #6 (G1).
one of the most effective catalyst systems for nanofiber
synthesis.8 The SEM images in Figure 9 reveal herringbone-
type nanofibers exhibiting a broad diameter distribution (espe-
cially batch #11). The sometimes very long fibers of batch #10
(<50 µm, Figure 9) show rather smooth surfaces, whereas in
batch #11, considerable amounts of twisted nanofibers together
with more irregularly shaped fibers are present (Figure 9).
The CNF batch #8 (Gupta’s19,20 formula) exhibits a segmented
fiber architecture (Figure 10) which should be favorable for gas
storage applications. As shown in Figure 10b, the graphene
layers are stacked indeed parallel to the base of the Ni/Cu
particle following a growth mechanism as suggested by Chen
et al.37
CNFs exhibiting a coiled morphology were obtained with
Browning’s synthesis parameters21 employing an Fe/Ni/Cu
catalyst (Figure 11).
Supplementary to high-spatial-resolution techniques (SEM,
TEM), bulk characterization integral methods such as nitrogen
sorption, X-ray diffraction, and Raman spectroscopy were
applied to obtain a complete dataset of the samples studied.
In Figure 12, the sorption isotherms of the nanofibers are
plotted. Remarkably different nitrogen uptakes are found in the
very low relative pressure regime, which is dominated by
micropore adsorption, at medium relative pressures (i.e., 0.1-
0.4) where multilayer adsorption takes place as well as at relative
pressures above 0.4 where capillary condensation in mesopores
is predominant. Hence, the quantitative evaluation of the data
(Table 3) exhibits a wide scatter of both the BET surface areas
(31-271 m2/g) and the micropore volumes (0.04-0.11 cm3/
g). The latter indicates differences in the microstructural
arrangement of the microcrystallites and/or the presence of
surface roughness.
Since the BET surface area represents the sum of the specific
surface located in the micropores within the fibers and the size
of the outer surface of the fibers, differences in this quantity
can be attributed to different fiber diameters or diameter distri-
butions within the batches and surface roughness (microporosity)
which includes the presence of more exotic nanofiber types.
For some samples, a slight hysteresis loop at large relative
pressures is present; the course of the sorption curve in this
regime is influenced by both the size and shape of mesopores
(<50 nm) within or in between the carbon nanofibers. In
the latter case, the mechanical characteristics of the loose
Figure 8. Herringbone-type CNF batch #7 (see Table 2) synthesized
on an Fe/Cu catalyst following the processing route by Rodriguez and
Baker:9 (a) SE image showing CNF size distribution within the batch,
(b) HREM image of wrinkled (002) basal planes of graphitic carbon,
(c) faceted Fe-Cu catalyst particle triggering herringbone-type growth
mechanism (TEM, BF), and (d) catalyst distribution within the batch
as imaged with a HAADF detector in STEM mode.
herringbone-type arrangement of graphene layers inclined with
respect to the fiber growth axis is dominating.
In addition to the samples provided by Rodriguez and Baker,
five different CNF batches (samples #7 through #11, Table 2)
have been synthesized in the framework of this study and pro-
cessed following the specifications given in the literature.8,9,19-21
A herringbone-type CNF batch (#7, Table 2) employing an
Fe/Cu catalyst has been synthesized according to the route given
by Rodriguez and Baker in ref 9. The microstructural charac-
teristics of this type of carbon nanofiber are displayed in Figure
8. Notably, the (002) basal planes are wrinkled in the HREM
image (Figure 8b) indicating a substantial amount of structural
disorder due to heteroatoms present in the graphitic structure.
This was confirmed as a general feature for all CNF batches
investigated in this study.
Two more batches of nanofibers were synthesized using a
Ni/Cu (7/3) powder catalyst (#10 and #11, Table 2) known as

Hydrogen Storage in Carbon Nanofibers
J. Phys. Chem. B, Vol. 109, No. 31, 2005 14985
Figure 9. Herringbone-type CNF batch #10 and #11 synthesized using a Ni/Cu (7/3) powder catalyst: (a) very long regular-shaped CNF (#10), (b)
fishbone-like CNF exhibiting bidirectional growth characteristic (#10), and (c) regular and irregular CNF, that is, twisted morphologies (#11).
removal of surface groups and a general cleaning of the surface,
a high-temperature activation may also lead to more distinctive
structural changes of the nanofibers. We applied the precise
activation procedure given by Rodriguez and Baker40 to the
pristine CNF specimens provided by them to ensure compara-
bility between our results and their data, that is: (a) heating
while purging with argon from room temperature to 925 °C in
35 min; (b) holding the temperature at fixed argon flow at 925
°C for 60 min; (c) cooling down to room temperature in about
60 min at continuous argon flow; (d) transferring the sample to
Figure 10. Segmented fiber architecture of batch #8 (see Table 2)
the hydrogen storage measurement cell, however, exposing it
following the process route by Gupta et al.:19,20 (a) coarse-scale fiber
to atmosphere for less than 10 min; (e) heating the sample under
argon flow inside the measurement cell at 200 °C for a minimum
of 10 h.
microstructure (SE image) and (b) typical location of Ni/Cu catalyst
within CNF (TEM bright field), see encircled area.
Compared to the characteristics of the pristine fibers, remark-
able microstructural changes involving the graphitic fiber
architecture, the distribution of catalyst particles, and the surface
topology of the fibers take place upon activation in argon
atmosphere for all CNF batches investigated. This effect is
highlighted for batch #2, sample F1-27 (Figure 14) from
Catalytic Materials Inc. and batch #8 (Figure 15), respectively.
Defining the front edge and the truncated cone of the pristine
herringbone-type morphology as tip and tail sections (Figure
14a) for the sake of simplicity, Figure 16, illustrates the
continuous loss of structural integrity in the fiber tip section
due to gaps formed between the graphene layers. It should be
noted that the formation of graphite loops at the fiber periphery
significantly increases the surface roughness. In addition, high-
temperature activitation of CNFs gives rise to a considerable
enrichment of catalyst particles at damaged tip or tail sections
(Figure 14c) and agglomerates of newly formed, extremely
fine-grained, nanotube-type CNF fuzz usually concentrated at
tail sections (Figure 14d) or fracture surfaces from individual
fibers.
Interestingly, high-temperature activation introduces quite
similar microstructural changes to the segmented fibers (batch
#8, parts a through c of Figure 15) suggesting that a general
damage mechanism is effectively independent of the different
CNF characteristics involved.
The formation of graphite loops at the fiber periphery has
been attributed to surface reconstruction upon the release of
heteroatoms (e.g., hydrogen) from highly active edge sites during
heat treatment at CNFs.41,42 However, with 1800-3000 °C, the
temperatures applied in these studies have been considerably
higher as compared to the temperature of 925 °C that we used
in our activation experiments. This aspect requires further
clarification.
nanofiber agglomerates can play a crucial role since upon
adsorption the capillary tension of the condensing nitrogen is
locally compressing the ground sample thus preventing a
straightforward interpretation of the data.
For quantification of fiber crystallinity, the interplanar spacing
and full width at half maximum (fwhm) of the (002) graphite
reflection were derived from X-ray diffraction patterns. The
results are summarized in Table 3. For all samples, the
interplanar distances are larger than the value for graphite
(0.3354 nm); this is consistent with the turbostratic microstruc-
ture of the fibers and the presence of heteroatoms as detected
by analytical TEM.
All samples show a similar crystallinity except for the samples
#6 and especially #7, where the crystallinity is significantly
enhanced resulting in smaller d002 values accompanied by larger
microcrystallite sizes. The behavior of sample #7 is in ac-
cordance with earlier findings, namely, that Fe/Cu catalyzed
nanofibers exhibit a considerably higher crystallinity than Ni/
Cu catalyzed fibers.9,39
The latter result is confirmed by Raman measurements. The
Raman spectra of all fibers synthesized in our laboratories
(except for sample #7) show a similar fingerprint (Figure 13)
that is characteristic for Ni/Cu catalyzed nanofibers: Two strong
bands at about 1340 cm-1 (D-band) and 1600 cm-1 (G-band)34
indicate the existence of a sp2 carbon arranged in small
microcrystallites and maybe some less organized carbon. It is
interesting to note that the Raman spectra of the fibers supplied
by Catalytic Materials Inc. show almost an identical pattern. In
contrast, the half widths of the spectrum for sample #7 are
significantly smaller, and the G-band shows a clearly visible
shoulder at higher wavenumbers in accordance with a higher
crystallinity of the nanofibers.
ActiVation Procedures. Besides the CNF processing paths,
hydrogen storage is largely dependent on the pretreatment of
the fiber prior to hydrogen loading. As described in the literature,
the activation of the carbon materials is an important step for
maximizing the hydrogen storage capacity.3,21,38 A different or
missing pretreatment of the samples may in part explain the
strongly varying results found by different groups. Besides
The very obvious redistribution of catalyst material upon high-
temperature activation of CNFs may be explained by the fact
that transition metals easily wet on carbon materials after heat
treatment in Ar atmosphere at 1000 °C for several hours if the
defect density is sufficiently high.43 This is certainly the case
for CNFs, and it may be anticipated that the well-pronounced

14986 J. Phys. Chem. B, Vol. 109, No. 31, 2005
Rzepka et al.
Figure 11. Coiled carbon nanofibers from batch #9 (Table 2) following the processing route by Browning et al.21 over a Fe/Ni/Cu catalyst: (a and
b) coarse-scale CNF microstructure (SE image vs. TEM bright field), (c) faceted catalyst location at the CNF tip.
Figure 12. Nitrogen sorption isotherms taken at 77 K for different
types of herringbone nanofibers.
Figure 14. Microstructural changes of CNF batch #2 (specimen F1-
27 from Catalytic Materials Inc.) upon activation in argon atmosphere
(TEM bright field images): (a) pristine fiber morphology defining tip
vs tail sections, (b) loss of herringbone architecture (initial stage, tail
section) and formation of graphitic carbon loops at fiber surface (see
boxed area), (c) loss of herringbone architecture (mature stage, tip
section) associated with notable enrichment of catalyst particles
(encircled areas), and (d) formation of secondary extremely fine-grained
nanotube-type fuzz at fiber tail section.
In addition to the procedure describe earlier, we also tested
some slightly modified activation procedures. For instance, in
some cases a custom-built transfer system was used to transfer
the fibers from the activation system to the hydrogen storage
vessel under anaerobic conditions.
4.2. Hydrogen Storage. Investigation of the different
nanofiber batches by thermodesorption (TDS) yields similar
results for all fibers investigated; Figure 17 shows a typical
Figure 13. Raman spectra of batches #1-3, #7, #10, and #11.
spectrum. The key fragments hydrogen (02), water (18), and
carbon monoxide (28) as monitored by a mass spectrometer as
dispersion effect of catalyst particles described for a transition
metal/multiwall nanotube mixture43 holds for CNFs too.
a function of time are presented; additional mass fragments
The formation of nanotube-type fiber fuzz (Figure 14d) may
necessary for the correct assignment are suppressed for a better
be related to pyrolytic carbon surface layers which stem from
clarity of the plot. No significant desorption of hydrogen- or
polymerization of C2H2 molecules during fiber processing. Their
carbon-containing fragments at moderate temperatures is de-
growth may be enhanced via redistribution/enrichment of fine-
grained catalyst particles as described above.
tected.
A hydrogen release starting at 600 °C is remarkable. This
high-temperature hydrogen is found for all fibers. It varies from
0.1 to 1 wt % as found after semiquantitative analysis with a
calibration volume and confirmed by elementary analysis. In
their publications,3,38 Rodriguez and Baker claimed this high-
temperature peak to be an indicator for the presence of a
hydrogen storing material. They interpret this effect as a small
part of the stored hydrogen, which can be released only at high
A TEM investigation of CNF batches exposed to high-
temperature activation followed by an uptake of hydrogen
virtually reveals the same microstructural pattern as previously
discussed solely for the effect of activation. This is displayed
for batch #3 (Table 2) in Figure 16a-c emphasizing that during
the activation stage already the fibers are prone to a continuous
loss of structural integrity.

Hydrogen Storage in Carbon Nanofibers
J. Phys. Chem. B, Vol. 109, No. 31, 2005 14987
Figure 15. Microstructural changes of CNF batch #8 (Table 2) upon activation in argon atmosphere (TEM bright field images): (a) segmented
fiber architecture in the as-received condition, (b and c) loss of structural integrity of the fiber and enrichment of catalyst particles.
Figure 16. Microstructural changes of CNF batch #3 (specimen F1-28, Catalytic Materials Inc.) upon both activation in argon atmosphere and
hydrogen uptake (TEM bright field images): (a and b) various stages of disintegration of graphic layers at CNF tip and tail sections associated with
enrichment of catalyst particles. Note the formation of very fine-grained tubular fiber fuzz at the fiber surface adjacent to disintegrated areas, (c)
complete loss of CNF herringbone architecture of graphene sheets.
Figure 18. Thermodesorption of hydrogen from sample #3, both for
pristine fiber (dashed line) and for fibers which have been exposed to
Figure 17. Thermodesorption spectrum of sample #4.
130 bar hydrogen (solid line).
temperatures because it is chemisorbed at the fibers. If this is
the case, the exposure of the fibers to hydrogen should reproduce
the peak. However, after the samples were exposed to hydrogen
at 1 bar, only a very small peak of desorbing hydrogen around
120 °C, corresponding to 0.02 wt %, is found (Figure 18). The
peak at 850 °C does not appear again; therefore it is unlikely
that it is related with a reversible adsorption-desorption property
of the material.
What remains unclear is the origin of the hydrogen observed
in the TDS experiment. Through the use of TDS, it is impossible
to distinguish between hydrogen located inside or outside of
the carbon nanofiber. Since at the applied temperatures the
cracking of covalent C-H-bonds is expected, the hydrogen may
cause hydrogen-saturated edges of the graphene sheets. Further
studies are needed to clarify these facts.
gravimetric apparatus. If available, we show also the results of
the volumetric desorption measurements. As in the desorption
experiments, all samples gave values below the limit of
detection; this gives an upper limit for the released hydrogen.
In addition to the measurements for pristine samples, Table 4
also contains the results for activated samples. Besides the
activation procedure described above (labeled “925 °C Ar”),
we also show measurements for samples, which are only heated
at 190 °C under argon flow (labeled “190 °C Ar”) and helium
flow (labeled “190 °C He”). In general, the measured hydrogen
uptake is low for all samples. Within the experimental error (in
the order of 0.1 wt % for most samples), only samples #1 and
#2 show a significant hydrogen uptake up to 0.4 wt % but far
below values of 2-4%.
Table 4 summarizes the results of the volumetric and
gravimetric hydrogen storage measurements of the samples
under investigation. The major part of these measurements has
been done with the volumetric absorption apparatus and the
Complementary data derived via thermal desorption measure-
ments (Table 5) after exposing the samples to hydrogen at a
pressure above 100 bar also show no uptake within the range
of the experimental error.

14988 J. Phys. Chem. B, Vol. 109, No. 31, 2005
Rzepka et al.
TABLE 4: Hydrogen Storage Capacity of the Nanofiber
Samples As Measured with the Volumetric Adsorption
Apparatus in wt %
5. Conclusion
Within the framework of the study presented, we investigated
carbon nanofibers provided by the Rodriguez and Baker group
as well as nanofibers synthesized in our labs following the
procedures published by different research groups working in
this field. The batches comprise different structural and mor-
phological characteristics. The fibers were analyzed with respect
to their hydrogen uptake at room temperature in the pressure
range from 0 to 140 bar using gravimetric and volumetric
techniques. In some cases, additional activation procedures were
applied to the pristine fibers prior to the hydrogen storage
experiments. Within the experimental error of 0.1 wt %, we
obtained good agreement between the two high-precision
experimental techniques applied for quantification of hydrogen
uptake. However, no significant hydrogen uptake could be
detected for any of the CNF specimens investigated. With a
maximum uptake of 0.4 wt %, all experimental results can be
adequately explained by pure physical adsorption on the CNF
surface; no evidence for any other storage mechanism has been
found. We therefore were unable to confirm the data published
by other groups reporting hydrogen storage capacities for CNFs
above 3 wt %.
sample #1
pristine
50 bara
0.14
95 bar
0.33
140 bar
0.39
190 °C Ar
925 °C Ar
0.12
0.11 (0.19)
0.19
0.24 (0.26)
0.24
0.22 (0.36)
sample #2
pristine
50 bara
0.10
95 bar
0.22
140 bar
0.29
190 °C Ar
925 °C Ar
0.05
0.11 (<0.10)
0.06
0.15 (<0.10)
0.25
0.14 (<0.10)
sample #3
pristine
190 °C Ar
925 °C Ar
50 bara
0.04 (<0.10)
0.07
95 bar
0.03 (<0.10)
0.05
140 bar
-0.09 (<0.10)
-0.05
-0.08 (<0.10)
-0.12 (<0.10)
-0.20 (<0.10)
sample #4
190 °C He
50 bara
0.03
95 bar
0.08
140 bar
0.09
sample #5
pristine
190 °C Ar
50 bara
0.00
-0.03
95 bar
0.00
-0.03
140 bar
-0.07
-0.04
sample #6
190 °C He
50 bara
0.03
95 bar
0.06
140 bar
0.20
sample #7
pristine
50 bara
0.13
95 bar
140 bar
Acknowledgment. The financial support from the German
Federal Ministry of Economics and Labor (BMWi) through
Grant 0327304 is kindly acknowledged. We would like to thank
R.T.K. Baker, and N.M. Rodriguez for providing us with
different carbon nanofiber material and the information about
the adequate fiber pretreatment procedure. The authors would
also like to acknowledge Prof. W. Kiefer and co-workers at
Physical Chemistry, Wu¨rzburg University, for providing the
Raman characterization.
250 °C
-0.14
50 bara
0.03
50 bara
0.13
sample #8
925 °C Ar
95 bar
0.04
140 bar
0.05
sample #9
pristine
95 bar
0.09
140 bar
0.02
a If available, the gravimetric value is given in parentheses. The
experimental error of all measurements is approximately (0.1 wt %.
TABLE 5: Hydrogen Storage Capacities Determined by
Desorption Measurements
References and Notes
sample
#1
activation
pressure
storage capacitya
(1) New Sci. 1996, 2061, 20.
(2) Chambers, A.; Park, C.; Baker, T. K.; Rodriguez, N. M. J. Phys.
Chem. B 1998, 102, 4253.
(3) Park, C.; Anderson, P. E.; Chambers, A.; Tan, C. D.; Hidalgo, R.;
Rodriguez, N. M. J. Phys. Chem. B 1999, 103, 10572.
(4) Boehm, H. P. Carbon 1973, 11, 583.
(5) Baker, R. T. K.; Barber, M. A.; Harris, P. S.; Feates, F. S.; Waite,
R. J. J. Catal. 1972, 26, 51.
(6) Yang, R. T.; Chen, J. P. J. Catal. 1989, 115, 52.
(7) Baker, R. T. K. Carbon 1989, 27, 315.
(8) Rodriguez, N. M. J. Mater. Res. 1993, 8, 3233.
(9) Krishnankutty, N.; Park, C.; Rodriguez, N. M.; Baker, R. T. K.
Catal. Today 1997, 37, 295.
150 °C.
vacuum 1000 °C.Ar/H2
150 °C.
vacuum 900 °C.
vacuum 1000 °C. Ar/H2
150 °C.
130 bar
120 bar
120 bar
100 bar
120 bar
100 bar
100 bar
130 bar
-0.2 wt %
0.8 wt %
-1.0 wt %
0.2 wt %
-1.0 wt %
0.5 wt %
-0.3 wt %
0.5 wt %
-0.3 wt %
0.7 wt %
0.4 wt %
0.1 wt %
0.1 wt %
-0.1 wt %
-1.0 wt %
-0.2 wt %
-0.8 wt %
#2
#3
#4
#6
#7
#8
vacuum 925 °C. Ar
150 °C. vacuum
150 °C. vacuum
130 bar
(10) Rodriguez, N. M.; Chambers, A.; Baker, R. T. K. Langmuir 1995,
11, 3862.
150 °C.
vacuum 700°C Ar/O2
150 °C.
vacuum, 150 °C.
vacuum 925 °C. Ar
1000 °C. Ar/H2
130 bar
130 bar
100 bar
120 bar
100 bar
120 bar
(11) Cheng, H.-M.; Yang, Q.-H.; Liu, C. Carbon 2001, 39, 1447.
(12) Baker, R. T. K.; Rodriguez, N. M. Graphite nanofiber catalyst
systems for use in fuel cell electrodes. U.S. Patent 6,485,858, 2002.
(13) Endo, M.; Kim, C.; Nishimura, K.; Fujino, T.; Miyashita, K. Carbon
2000, 38 183.
#9
a Note: The standard deviation of this procedure as derived from
dry runs was about 0.8 wt %.
(14) Rodriguez, N. M.; Kim, M.-S.; Baker, R. T. K. J. Phys. Chem.
1994, 98, 13108.
(15) Chen, P.; Wu, X.; Lin, J.; Tan, K. L. Science 1999, 285, 91.
(16) Yang, R. T. Carbon 2000, 38 623.
(17) Fan, Y.; Liao, B.; Liu, M.; Yei, Y.; Lu, M.; Cheng, H. Carbon
1999, 37, 1649.
(18) Cheng, F.; et al. Z. Metallkd. 2000 91, 306.
(19) Gupta, B. K.; Srivastava, O. N. Int. J. Hydrogen Energy 2000, 25,
825.
In addition, no increase of the hydrogen uptake due to
activation procedures could be detected. On the contrary, the
two pristine samples #1 and #2, which show a slight storage
capacity, lose this property by the activation. This is in
accordance with the TEM analysis of the fibers, which shows
a strong structural change due to the high-temperature treat-
ment.
In particular, we were not able to confirm the hydrogen
storage values of 1.5 and 4.2 wt % provided by Baker and
Rodriguez for their sample #5 in the pristine state and after
activation, respectively.
(20) Gupta, B. K.; Srivastava, O. N. Int. J. Hydrogen Energy 2001, 26,
857.
(21) Browning, D. J.; Gerrard, M. L.; Lakeman, J. B.; Mellor, I. M.;
Mortimer, R. J.; Turpin, M. C. Nano Lett. 2002, 3, 201.
(22) Stro¨bel, R.; Jo¨rissen, L.; Schliermann, T.; Trapp, V.; Schu¨tz, W.;
Bohmhammel, K.; Wolf, G.; Garche, J. J. Power Sources 1999, 84, 221.
(23) Ahn, C. C.; Ye, Y.; Ratnakumar, B. V.; Witham, C.; Bowman, R.
C.; Fultz, B. Appl. Phys. Lett. 1998, 73, 3378.

Hydrogen Storage in Carbon Nanofibers
J. Phys. Chem. B, Vol. 109, No. 31, 2005 14989
(24) Poirier, E.; Chahine, R.; Bose, T. K. Int. J. Hydrogen Energy 2001,
26, 831.
(36) Hycob GmbH, Moers. www.hycob.de.
(37) Chen, X. H.; Deng, F. M.; Lu, X. N.; Wu, G. T.; Wang, M.; Yang,
(25) Hirscher, M.; et al. J. Alloys Compd. 2002, 330, 654.
(26) Tibbetts, G. G.; Meisner, G. P.; Olk, C. H. Carbon 2001, 39, 2291.
(27) Baker, R. T. K.; Rodriguez, N. M. Private communication, 2002.
(28) Hirscher, M.; Becher, M. J. Nanosci. Nanotechnol. 2003, 3, 3.
(29) Ritschel, M.; Uhlemann, M.; Gutfleisch, O.; Leonhardt, A.; Graff,
A.; Ta¨schner, Ch.; Fink, J. Appl. Phys. Lett. 2002, 80, 2985.
(30) Rodriguez, N. M.; Baker, T. K. Private communication, 2002.
(31) Rodriguez, N. M.; Kim, M.; Baker, R. T. K. J. Catal. 1993, 140,
16.
(32) Gregg, S. J.; Sing, K. S. Adsorption, Surface Area and Porosity;
Academic Press: London, 1982.
(33) Kinoshita, K. Carbon: Electrochemical and Physicochemical
Properties; John Wiley & Sons: New York, 1988.
(34) Ferrari, A. C.; Robertson, J. Phys. ReV. B 2000, 61, 14095.
(35) Rzepka, M.; Lamp, P.; de la Casa-Lillo, M. A. J. Phys. Chem. B
1998, 102, 10894.
H. S.; Zhang, X. B. J. Cryst. Growth 2001, 222, 163.
(38) Rodriguez, N. M.; Baker, R. T. K. International Patent WO00/
76625 A1, 2000.
(39) Fricke; et al. Innovative Energiespeicher auf der Basis von
Kohlenstoffnanostrukturen, Abschlussbericht, FKZ 0327304D; Report E21-
0203-2, 2003.
(40) Rodriguez, N. M.; Baker, T. K. Private communication
(41) Endo, M.; Kim, Y. A.; Hayashi, T.; Yanagisawa, T.; Muramatsu,
H.; Ezaka, M.; Terrones, H.; Terrones, M.; Dresselhaus, M. S. Carbon 2003,
41, 1941.
(42) Zheng, G. B.; Sano, H.; Uchiyama, Y. Carbon 2003, 41, COI-
856.
(43) Zhong, Z.; Liu, B.; Sun, L.; Ding, J.; Lin, J.; Tan, K. I. Chem.
Phys. Lett. 2002, 362, 135.