Development of Ni-Cu-Mg-Al catalysts for the synthesis of carbon nanofibers by catalytic decomposition of methane

Article abstract of DOI:10.1016/j.jcat.2007.06.022

This work presents the kinetics and characterization results of Ni-Cu-Mg-Al catalysts used for the synthesis of carbon nanofibers (CNFs) by catalytic methane decomposition (CCVD). Ni-Cu-Mg-Al-based lamellar double hydroxides (LDHs) were synthesized by the coprecipitation method using either a regular maturation stage or a colloid mill process. After calcination at 1073 K and reduction at 1023 K, the formation of alloyed Ni-Cu nanoparticles over a mixed MgO-MgAl₂O₄-NiAl₂O₄ support was observed. The kinetic study of CNF synthesis by CCVD of methane was carried out in a thermobalance operated as a differential reactor. Small proportions of Cu (≤ 10 wt%) were found to improve the activity and mainly the stability over time of Ni-Mg-Al catalysts. On the other hand, large proportions of Cu dramatically decrease the activity of the catalysts in terms of CNF yield. The catalysts were able to produce high-quality CNFs with a graphitic or turbostratic structure. The size of metal particles and the diameter of carbon filaments obtained depend on the catalyst composition (i.e., Ni:Cu ratio). The application of a kinetic model based on the growing mechanism of CNFs allows us to determine the influence of the operating conditions on the kinetic parameters.

Full text of DOI:10.1016/j.jcat.2007.06.022

Journal of Catalysis 251 (2007) 223–232
www.elsevier.com/locate/jcat
Development of Ni–Cu–Mg–Al catalysts for the synthesis of carbon
nanofibers by catalytic decomposition of methane
L. Dussault a, J.C. Dupin a, C. Guimon a, M. Monthioux b, N. Latorre c, T. Ubieto c, E. Romeo c,
C. Royo c, A. Monzón c,∗
a
Institut Pluridisciplinaire de Recherche sur l’Environnement et les Matériaux, Equipe de Chimie, Physique (IPREM-ECP),
UMR #5254 CNRS, 2 Av. P. Angot, F-64053 Pau Cedex 9, France
b
CEMES, UPR-8011 CNRS, B.P. 94347, 31055 Toulouse Cedex 4, France
c
Department of Chemical and Environmental Engineering, Institute of Nanoscience of Aragon (INA), University of Zaragoza, 50009 Zaragoza, Spain
Received 10 February 2007; revised 23 May 2007; accepted 16 June 2007
Available online 21 August 2007
Abstract
This work presents the kinetics and characterization results of Ni–Cu–Mg–Al catalysts used for the synthesis of carbon nanofibers (CNFs)
by catalytic methane decomposition (CCVD). Ni–Cu–Mg–Al-based lamellar double hydroxides (LDHs) were synthesized by the coprecipitation
method using either a regular maturation stage or a colloid mill process. After calcination at 1073 K and reduction at 1023 K, the formation of
alloyed Ni–Cu nanoparticles over a mixed MgO–MgAl O –NiAl O support was observed. The kinetic study of CNF synthesis by CCVD of
2
4
2 4
methane was carried out in a thermobalance operated as a differential reactor. Small proportions of Cu (ꢀ 10 wt%) were found to improve the
activity and mainly the stability over time of Ni–Mg–Al catalysts. On the other hand, large proportions of Cu dramatically decrease the activity
of the catalysts in terms of CNF yield. The catalysts were able to produce high-quality CNFs with a graphitic or turbostratic structure. The size of
metal particles and the diameter of carbon filaments obtained depend on the catalyst composition (i.e., Ni:Cu ratio). The application of a kinetic
model based on the growing mechanism of CNFs allows us to determine the influence of the operating conditions on the kinetic parameters.
© 2007 Elsevier Inc. All rights reserved.
Keywords: Catalytic methane decomposition; Carbon nanofiber; Ni–Cu–Mg–Al catalysts; LDH precursors; Kinetic modeling
1. Introduction
ous forms, including nanosized filaments (i.e., CNFs), had to
be avoided because of their deactivation effects in catalytic
dehydrogenation reactions and the cracking of hydrocarbons
[16–21]. The overall mechanism of CNF growth over metal-
lic particles is now well established [1]. It consists of several
steps: (i) Methane is cracked on the metal surface; (ii) carbon
atoms, which can form subsurface carbides, dissolve and dif-
fuse through the bulk of the solid or through quasi-liquid metal
particles; (iii) due to the cooling down of the particles that lower
the carbon solubility or due to oversaturation, carbon is then
expelled and precipitates as more or less perfect polyaromatic
layers (graphenes). This continuous process explains why the
metal particles are most often located at the tips of the CNFs.
CNF diameter can range from 2 to 500 nm [1,21] and depends
mainly on the metal particle size. Two main types of carbon fil-
aments can be obtained: hollow, thereby forming nanotubes in
which graphenes are displayed parallel (i.e., concentric texture)
Over the last decade, carbonaceous nanomaterials have re-
ceived considerable attention due to their physical and chemical
properties and their potential industrial applications. This is par-
ticularly the case for carbon nanofibers (CNFs) (see, e.g., the
reviews by De Jong and Geus [1] and Serp et al. [2]), which can
be used as catalyst supports [2–7], electrodes for fuel cells [8]
and lithium ion batteries [9], adsorbents [10,11], polymer addi-
tives [12], and possibly gas storage materials [13–15].
CNFs usually are obtained by catalytic chemical vapor de-
position (CCVD) from the decomposition of gaseous hydro-
carbons (e.g., methane) on third-transition metal particles. For
a long time, the deposition over catalysts of carbon in vari-
*
Corresponding author.
E-mail address: amonzon@unizar.es (A. Monzón).
0021-9517/$ – see front matter © 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2007.06.022

224
L. Dussault et al. / Journal of Catalysis 251 (2007) 223–232
or oblique (i.e., herringbone texture) to the filament axis, and
genuine nanofibers in which graphenes are displayed oblique
(herringbone texture) or perpendicular (platelet texture) to the
fiber axis while leaving no inner cavity. Both CNF type and
size depend on the metal particle features (i.e., size, morphol-
ogy, composition), the carbon source, and the reaction condi-
tions.
as supports, minimizing the sintering effect during calcina-
tion and prereduction and inducing a decrease in metal particle
size.
To improve the stability of the catalysts and minimize
their deactivation with time, we have used smaller propor-
tions of the reducible metals than those reported in the lit-
erature. We prepared the LDH precursors of molecular for-
mula (Ni1−xCuxMgAl)(OH)7(CO2−
0.5, with 0 ꢀ x ꢀ 1, us-
)
Among the different metals commonly used in CNF and H2
production, Ni-based supported catalysts are of special interest
because they show good activity at relatively moderate tem-
peratures [18,22–26]. Indeed, the decomposition temperature
of methane (870–970 K) with these catalysts is about 300 K
lower than that with Fe-based catalysts. The most frequently
used supports are silica, alumina, and magnesia. Given the ben-
eficial effect of using such alloys as Fe–Mo, associating another
metal, such as Cu [27–31], to Ni has been proposed to increase
the activity. Li et al. [29] have thus demonstrated that a small
proportion of Cu promotes the activity of Ni/Al2O3. However,
Ni-based catalysts with high nickel content tend to deactivate
themselves relatively quickly by sintering and formation of en-
capsulating polyaromatic carbon shells. Adding Mg to Ni–Al
catalysts obtained by coprecipitation improves the properties
of these solids [32]. Ni–Mg–Al catalysts are more resistant to
deactivation than Ni–Al catalysts, because the formation of a
spinel phase of MgAl2O4 provides stability to the Ni parti-
cles, avoiding sintering and consequent deactivation. Moreover,
Ni–Mg–Al catalysts have a greater specific surface area, and
thus a higher reactivity per gram of catalyst, than Ni–Al cata-
lysts.
Taking these properties into account, we prepared Ni–Cu–
Mg–Al mixed oxides from lamellar double hydroxides (LDHs),
also known as Feiknecht or hydrotalcite-like compounds. These
precursors belong to a large class of anionic clay minerals
formed by brucite-like [Mg(OH)2] sheets in which the octahe-
dral divalent magnesium is partially isomorphously substituted
by divalent (e.g., Ni, Cu) and trivalent (e.g., Al) cations. The
positive charges resulting from the trivalent cations are com-
pensated by anions (e.g., CO23−, NO32−, Cl−) located in the
interlayer space together with water molecules [33]. In these
compounds, the metallic atoms have a uniform distribution,
so that calcination between 673 and 1073 K transforms the
hydroxides into well-mixed oxides. This maximizes the inter-
metallic interactions, which favors the doping effect of Cu [29]
and the formation of Ni–Cu alloy. Moreover, the irreducible do-
mains (Al2O3, MgO, nickel, and magnesium aluminates) act
ing two different methods: the c3onventional coprecipitation
process and, with the aim of tuning the particle size con-
trol, the colloid mill coprecipitation method. Here we report
the results from the characterization and catalyst activity of
(NiO)1−x(CuO)xMgO(Al2O3)0.5 catalysts obtained from the
LDH precursors as a function of catalyst composition and
preparation procedure. The results include a kinetic study of
the influence of the operating conditions on the rate of CNF
growth. We have presented a detailed CNF characterization
and a discussion of the relationship between catalyst char-
acteristics and nanofiber type in another paper [34]. In what
follows, for the sake of simplicity, we designate the catalysts as
Ni1−xCuxMgAl (e.g., Ni0.8Cu0.2MgAl for x = 0.2).
2. Experimental
2.1. Catalytic precursors
We first prepared two aqueous solutions, one contain-
ing various amounts of Mg(NO3)2·6H2O, Al(NO3)3·9H2O,
Cu(NO3)2·3H2O, and Ni(NO3)2·6H2O, and another containing
NaOH (to maintain the pH constant at 9.5 0.2) and a slight
excess of Na2CO3. LDH materials with various Ni/Cu/Mg/Al
proportions (given in Table 1) were then prepared from the
latter solutions using two different methods, the regular co-
precipitation method [35] and a coprecipitation route using a
colloid mill [36].
In the regular coprecipitation route, the two solutions were
simultaneously added dropwise to a vessel containing stirred,
deionized water, with the pH of the mixture maintained at 9.5.
The mixing process was carried out at room temperature. The
resulting slurry was removed from the vessel and kept as a sus-
pension at 348 K for 15 h under stirring. The precipitate was
then filtered, washed with deionized water, and dried overnight
at 353 K. For the colloid mill method [36], the two solu-
tions were simultaneously added in a colloid mill rotating at
3000 rpm and mixed for 5 min. The resulting slurry was re-
moved from the colloid mill and kept as a suspension at 348 K
Table 1
2−
Atomic ratios of LDH precursors (Ni
Cu MgAl)(OH) (CO
)
0.5
(ICP-AES) and BET surface area values before (393 K) and after (1073 K) calcination
x
7
1−x
3
Nominal
value of x
Mg/Al
Ni/Cu
Ni/Al
Cu/Al
393 K
(m g
1073 K
(m g
2
−1
2
−1
)
(
0.05)
(
0.05)
(
0.05)
(
0.05)
)
x = 0
1
1
1
0.75
1
/
1
/
87
162
147
228
142
130
84
159
177
164
159
150
70
x = 0.05
x = 0.1
x = 0.2
x = 0.5
x = 0.8
x = 1
17.2
8.15
4
1
0.25
0.9
0.9
0.8
0.5
0.2
/
0.05
0.1
0.2
0.5
0.8
1
0.9
0.75
/
65

L. Dussault et al. / Journal of Catalysis 251 (2007) 223–232
225
for 15 h under stirring. The precipitate was then filtered, washed
with deionized water, and dried overnight at 353 K. The com-
position of the resulting LDHs is provided in Table 1.
The Raman spectra were obtained using a Jobin Yvon
T64000 confocal spectrophotometer in air and at room tem-
perature. Spectra were acquired with a 514.5 nm Ar+ ion laser
as an excitation source.
Scanning electron microscopy (SEM) studies were carried
out with a LEO 1530 Vp microscope under ultra-high vacuum,
with voltages of 20–30 kV, an in-lens detector and a 30-µm
aperture.
2.2. Characterization
The X-ray diffraction (XRD) patterns were recorded with an
INEL diffractometer using a curved position-sensitive detector
(INEL CPS 120) calibrated with Na2Ca3Al2F14 as the standard.
The monochromatic radiation applied was CuKα (1.5406 Å)
from a long fine-focus Cu tube operating at 40 kV and 20 mA.
Scans were performed over 2θ angles ranging from 10 to 90◦.
The powder diffraction standards (JCPDS) were used for peak
assignment.
The elemental composition of the samples was determined
by AES-ICP in a Thermo-Electron model 3580 instrument with
a H2SO4/HNO3 solution at 523 K. Surface areas were deter-
mined by the BET method from the adsorption of nitrogen at
77 K using a Micromeritics sorptometer. The samples (0.05 g)
were pretreated at 393 K overnight under vacuum (ca. 16 h,
10−5 mbar) to remove moisture before analysis.
2.3. Synthesis of CNFs
LDH materials from both routes (regular and colloid mill
methods) were calcined at 1073 K under a nitrogen flow after a
5 K min−1 ramp, with three plateaus at 453 K (1 h), 573 K (1 h),
and 1073 K (11 h), resulting in catalytic materials of various
compositions. Approximately 100 mg of each catalyst was then
placed in the quartz basket of a thermal balance (CI Electronics)
operated as a differential reactor. Catalyst reduction was per-
formed in situ at 1023 K for 2 h using a H2 (40%)/N2 mixture,
at a flow of 250 Nml min−1. Reaction conditions were tempera-
ture, 848–973 K; total flow rate, 750 Nml min−1; and feed com-
position (%CH4/%H2/%N2), 2.5/0/97.5–10/10/80. After reac-
tion, the reactor was cooled down to room temperature in a N2
atmosphere.
Thermogravimetric experiments (on ca. 10 mg) were carried
out with a TGA model 2950 coupled with a TA Instruments
mass spectrometer from 303 to 1123 K (5 K min−1) under air
flow.
3. Results and discussion
Before obtaining the temperature-programmed reduction
(TPR) measurements, the precursors (100 mg) were calcined at
1073 K under argon flow with a heating rate of 5 K min−1 and
plateaus at 453 K (1 h), 573 K (1 h), and 1073 K (11 h). TPR
experiments were then carried out at a 4 K min−1 ramp from
303 to 1173 K in a quartz reactor with a flow system (GIRA,
X-sorb model) equipped with a thermal conductivity detector.
A hydrogen–argon mixture (5–95%) was used as the reducing
3.1. From LDH precursors to mixed oxides
The elemental analysis (ICP-AES) (Table 1) found that the
composition of the LDH precursors was very close to that of
the reactive mixture, particularly regarding reducible metals
(Ni and Cu). Only samples with x = 1 and 0.2 showed a slight
loss of magnesium (Mg/Al = 0.75 instead of 1).
atmosphere with a flow rate of 20 ml N min−1
.
The XPS analyses were carried out with a Kratos spectrom-
eter (model Axis Ultra) with focused monochromatic AlKα
radiation (1486.6 eV) under a residual pressure of 10−7 Pa. To
prevent exposure of the samples to air between the calcination
step, the reduction step, and analysis, the XPS spectrometer
was directly coupled to a glove box. The hemispherical an-
alyzer functioned with a constant pass energy of 40 eV for
high-resolution spectra. The area of the samples analyzed was
300 × 700 µm2. Charge neutralization was used for all mea-
surements to compensate for the charge effects. The binding
energy scale was calibrated using the Al2p peak at 74.1 eV (cor-
responding to Al2O3) for calcined samples. The fitting peaks of
the experimental curves were calculated using a combination
of Gaussian (70%) and Lorentzian (30%) functions with a non-
linear Shirley-type background. The XPS quantitative analyses
were performed with the appropriate Scofield factors [37].
Dynamic light scattering (DLS) was used on 1 mg ml−1 of
suspended catalyst materials in glycol ethylene (1 mol l−1) to
measure the average particle size. The apparatus consisted of
a power laser Spectra-Physics Stabilite 2017 (wavelength =
514.5 nm, power = 0.1–2 W) and a Sematech photogoniodif-
fusiometer.
To follow the transformation of the hydroxide precursors
into mixed oxides, diffractograms were obtained before and
after calcination at 1073 K. All of the precursors showed a
hydrotalcite-like structure with a good crystallinity regardless
of the coprecipitation route (not shown here). At 1073 K,
the structure changed drastically. Fig. 1a shows an example
(Ni0.9Cu0.1MgAl, i.e., x = 0.1) of the diffractogram after cal-
cination at 1073 K. As expected, this diffractogram shows the
formation of aluminates, MgAl2O4 (JCPDS file no. 75-1801),
NiAl2O4 (JCPDS file no. 78-1601) with spinel-like structure
(JCPDS file no. 78-0430), NiO (JCPDS file no. 78-0643), and
CuO (JCPDS file no. 80-1916). By TGA and XPS analysis
(not shown here), it was observed that the transformation of
LDHs into mixed oxides was accompanied by the loss of car-
bonate ions and water. If at 723 K, about 30% of carbonates
remain [38], then XPS analysis indicates that practically no car-
bonate remains after calcination at 1073 K. The specific surface
areas of the mixed oxides obtained at 1073 K were relatively
close to those of the LDH precursors (Table 1). It appears that
these areas increased with the nickel proportion and that a small
proportion of Cu improved them slightly (comparison between
x = 0 and 0.05).

226
L. Dussault et al. / Journal of Catalysis 251 (2007) 223–232
Fig. 3. Evolution with time of the carbon growth rate for the Ni Cu MgAl
0.8 0.2
and Ni Cu MgAl catalysts. Influence of operating temperature.
0.9 0.1
and the peak width tend to decrease with decreasing particle
size. The curves of H2 consumption, TPR, reported in Fig. 2 in-
dicate an evolution with a Cu content similar to that of Cu–Zn
supported catalysts [39]. The high-Cu content samples (x be-
tween 1 and 0.5) presented a single band with a FWHM of
about 50 K. For the low-Cu content samples, a double peak ap-
peared with smaller widths (∼20 K), which indicates smaller
particle sizes. This is in agreement with a better dispersion of
Cu cations over the support, as noted by Wang et al. [31], and
can be related to the greater BET areas for the smallest values
of x. The small CuO particles could have more interaction with
the other metals, particularly Ni, which can form Ni–Cu alloy
after reduction. It is indeed known that Ni and Cu easily form
alloys over a wide composition range [30,40]. The absence of a
peak toward 770 K shows that CuAl2O4 was not formed [41].
The second region at 970–1170 K is associated with the re-
duction of the nickel (NiII → Ni0), because magnesium and
aluminum oxides do not reduce below 1170 K. Such tempera-
tures are much higher than those for pure NiO, which is reduced
at 600–700 K [42]. This demonstrates a strong interaction of the
nickel with the neighboring metals [20]. As with copper, the
broad band can be decomposed into two components centered
at 1020 and 1120 K. The former corresponds to NiO particles
in intimate contact with the support, whereas the latter can be
associated with the reduction of nickel aluminate [43–46] ob-
served in the XRD patterns (Fig. 1) and maybe in part with
small NiO particles buried in the MgAl2O4 matrix. The pres-
ence of these different types of nickel was confirmed by XPS
analysis of the samples reduced at 1023 K. In samples with
high Ni content, when the copper was totally reduced (bind-
ing energy Cu2p3/2 at 932.5 eV and Auger band at 919 eV), the
Ni2p3/2 band indicated the presence of about 30% of Ni0 (BE
at 852.5 eV) and 70% of NiII (BE at 855 eV), nonreducible at
this temperature.
Fig. 1. XRD patterns of Ni Cu MgAl after calcination at 1073 K (a), then
0.9 0.1
reduction at 1023 K (b) and after reaction (c).
Fig. 2. TPR (300–1173 K) of the mixed oxides (NiO)
(CuO) MgO-
x
2−
1−x
(Al O )
obtained by calcination of (Ni
Cu MgAl)(OH) (CO
x
)
0.5
7
2
3 0.5
1−x
3
LDHs at 1073 K.
3.2. Characterization of reduced mixed oxides
To be active in the production of CNFs, the catalysts must be
in reduced state. The reducibility of the catalysts was studied by
TPR. The TPR curves (Fig. 2) show two reduction regions, the
first at 443–503 K and the second at 973–1173 K.
The first region corresponds to the reduction of copper
(CuII → Cu0). The one-step reduction of malachite is associ-
ated with a single, quite sharp peak with a maximum at 503 K
[38]. Fierro et al. [39] showed that the reduction temperature
According to TPR and confirmed by XRD and XPS, it seems
that the NiO/NiAl2O4 ratio decreased with increasing Ni con-
tent, which may be related to a stabilizing interaction between
the nickel and copper oxides. However, the total NiO content
seemed to reach its maximum for a composition with a value of

L. Dussault et al. / Journal of Catalysis 251 (2007) 223–232
227
Table 2
Average size and size distribution obtained from light scattering of particles of Ni Cu Mg.Al (x = 0.2) and Ni Cu Mg.Al (x = 0.1) prepared by conventional
0.8 0.2
0.9 0.1
method and with colloid mill at various steps of the preparation process
x
Preparation
method
As-synthesized
Calcined (1073 K)
average size
(nm)
Calcined (1073 K)
reduced (1023 K)
average size (nm)
Average size
(nm)
Size range
(nm)
0.2
0.2
0.1
0.1
Conventional coprecipitation
Colloid mill
Conventional coprecipitation
Colloid mill
155
95
100
100
20–1000
25–500
25–400
25–300
215
160
140
135
175
150
130
140
x of 0.1–0.2, consistent with the activity of these catalysts (vide
infra, Section 3.4).
To summarize, after prereduction at 1023 K, the above re-
sults demonstrate that the catalysts are composed of Cu0, Ni0
(probably as a Ni–Cu alloy), MgO, NiAl2O4, and MgAl2O4.
3.3. Particle size
The particle size of the catalysts (metals and support) and
their distribution depend on the catalyst composition and prepa-
ration method [47–50]. In parallel to the common coprecipita-
tion method, LDHs can be prepared using a colloid mill (see
Section 2), which allows very rapid mixing and nucleation, fol-
lowed by a separate aging step. This method, first proposed by
Zhao et al. [36], gives smaller LDH crystals with a narrow size
distribution (Table 2).
Fig. 4. Evolution with time of the carbon growth rate for the Ni Cu MgAl
and Ni Cu MgAl catalysts. Influence of hydrogen partial pressure.
0.9 0.1
0.8 0.2
The size range and related average values tended toward
lower values when using the colloid mill method instead of
the conventional method and a composition of x = 0.1 instead
of 0.2. This is quite obvious right after synthesis, where size
distributions were 20–1000 nm for x = 0.2 under the copre-
cipitation method (average, 155 nm), 25–500 nm for x = 0.2
using the colloid mill (average, 95 nm), 25–400 nm for x = 0.1
with the coprecipitation method (average, 100 nm), and 25–
300 nm for x = 0.1 using the colloid mill (average, 100 nm).
After calcination and prereduction, however, the values were
close (130–175 nm), regardless of the conditions.
The operating temperature has a significant influence on the
yield (Fig. 3). In our case, the optimal temperature was close
to 923 K, slightly lower than that determined with other simi-
lar systems, such as Ni–Cu–Mg [31] (around 973 K). The feed
composition in these experiments was 5% CH4/95% N2. An
increase in the reaction temperature heightened the kinetics of
methane decomposition and deactivation reactions. At low re-
action temperatures, the carbon formation rate was low, but
deactivation was not observed for the reaction time. The deac-
tivation rate was faster at higher reaction temperatures. When
deactivation occurred, the carbon formation rate curves showed
an initial period of rapid growth until a maximum was reached,
followed by a decrease in rate until a residual constant value
was reached. During the period of activity decay, the deacti-
vation rate exceeded the filament formation rate; consequently,
the catalysts showed an optimal operating temperature at the
maximum carbon content.
3.4. Catalytic activity and CNF growth
Copper is almost inactive for this reaction and has a dilution
effect on the active sites. Thus, an excessive Cu content dra-
matically decreases the carbon yield [29,51]. Nevertheless, the
presence of Cu in the catalyst composition affects carbon dif-
fusion and reaction rates [29,52], and thus promotes catalytic
activity when it is incorporated in small quantities. The maxi-
mum carbon yield is reached for Cu content ranging between
3 wt% [27] and 15 wt% [28]. In agreement with these re-
sults, our catalysts with x ꢁ 0.5 showed very low activity [53].
The maximal yield was obtained with the Ni0.9Cu0.1MgAl and
Ni0.8Cu0.2MgAl catalysts (i.e., 3.8 and 7.6 wt% Cu, respec-
tively). These catalysts showed higher resistance against deac-
tivation than Ni.MgAl2O4 catalyst because Cu favors hydrogen
mobility, inhibiting the formation of encapsulating coke [52].
Moreover, we found that the synthesis method did not affect
the catalyst activity and stability.
The Ni0.9Cu0.1MgAl catalyst showed a higher carbon for-
mation rate than Ni0.8Cu0.2MgAl at the temperatures tested.
Furthermore, Ni0.9Cu0.1MgAl exhibited greater sensitivity to
the reaction temperature, being almost inactive at 700 ◦C. This
catalyst had good yields in an optimum range of operating tem-
peratures and was very active and stable.
Fig. 4 shows the effect of PH at a constant PCH (0.05 atm)
2
4
and operating temperature (650 ◦C). Hydrogen prevents the for-
mation of encapsulating coke, and consequently the formation
of carbon filaments is not hindered by this coke [5]. This fact
explains why deactivation was avoided during the CNFs syn-

228
L. Dussault et al. / Journal of Catalysis 251 (2007) 223–232
Fig. 6. Influence of hydrogen partial pressure on the kinetic parameters r
C
0
and r .
D
Fig. 5. Evolution with time of the carbon growth rate for the Ni Cu MgAl
0.8 0.2
and Ni Cu MgAl catalysts. Influence of methane partial pressure.
0.9 0.1
Table 3
Apparent activation energies of the kinetic parameters
thesis when the PH was approximately 0.02 atm. However, at
2
−1
66.7 1.0
−1
−1
30.8 0.5
−1
−1
Ni Cu MgAl
0.8 0.2
ψ
(min
)
r
(min
)
r
(g
C
0
g
min
min
)
)
D
d
C
cat
high PH , the greater competition between H2 and CH4 for the
2
−1
E (kJ mol
)
164.6 22.7
a
metallic surface sites caused a diminution in the carbon forma-
tion rate.
−1
(g g
C
C
cat
0
−1
−1
Ni Cu MgAl
ψ
(min
)
r
(min
)
r
D
0.9 0.1
d
−1
The effect of PCH on the carbon formation rate is charac-
E (kJ mol
)
–
111.3 6.8
29.6 2.3
a
4
terized in Fig. 5. The operating temperature in these experi-
ments was 650 ◦C (PH = 0 atm). An augmentation of PCH
2
4
increased the carbon formation rate because higher methane
concentrations in the gas phase led to a boost in the diffusion–
precipitation process through the Ni crystallites. Moreover, an
increase in the deactivation rate was not observed for the reac-
tion time. As in the previous studies, the Ni0.9Cu0.1MgAl cat-
alyst showed greater yields than Ni0.8Cu0.2MgAl, with higher
carbon contents and carbon formation rates.
of metallic active sites, CS , by the carbon diffusion coefficient,
0
kC (rC = CS · kC) [54–56]; rD is the intrinsic kinetic rate of
0
0
methane decomposition over the metallic surface, min−1; ψd is
the deactivation kinetic function (in min−1); and ψr is the re-
generation kinetic function (in min−1). These four parameters
include the dependence on reaction temperature (i.e., apparent
activation energies and preexponential factors), the atmospheric
composition (kinetic orders), and the type and composition of
the catalyst (metallic exposed area and intrinsic reactivity).
The term aS represents the residual activity of the catalyst,
which, according to the kinetic model, can be calculated as
With the aim of quantifying the influence of the operating
conditions on the amount and rate of formation of the NCM,
we developed a kinetic model for carbon growth during cat-
alytic hydrocarbon decomposition [54–56]. The mathematical
description of this kinetic model takes into account the main
steps of carbon formation and accumulation over the catalyst:
diffusion, nucleation, filament growth, and catalyst deactivation
aS = (ψr/ψG); ψG = ψd + ψr.
(4)
In the case where the catalyst suffers no deactivation, ψd =
ψr = 0, and thus the residual activity of the catalyst is aS = 1.
In this case, Eqs. (1) and (2) are simplified to
(Fig. 8). According to this model, the concentration of carbon
−1
accumulated over the catalyst, mC (gC g ), and the carbon for-
cat
−1
mation rate, rC (gC g min−1), can be expressed as follows
ꢄ
ꢅ
cat
ꢁ
ꢂ
1
[54–57]:
mC(t) = rC t −
1 − exp(−rDt)
(5)
0
rD
ꢀ
ꢁ
ꢂ
mC(t) = rC aSt + α1 · 1 − exp(−ψGt)
and
0
ꢁ
ꢂꢃ
ꢁ
ꢂ
− α2 · 1 − exp(−rDt) ,
(1)
(2)
rC(t) = rC 1 − exp(−rDt) .
(6)
ꢀ
ꢃ
0
rC(t) = rC aS + ψGα1 exp(−ψGt) − rDα2 exp(−rDt) ,
where α1 and α2 are given by
0
The foregoing expressions can be used directly to fit the exper-
imental data in Figs. 4, 5, and 6 to obtain the dependence of the
kinetic parameters with reaction temperature and partial pres-
sures of H2 and CH4.
rD(1 − aS)
(rD − ψGaS)
rD(rD − ψG)
α1 =
;
α2 =
.
(3)
ψG(rD − ψG)
The values of the apparent activation energies of rD, ψd,
The kinetic parameters of the model are as follows: rC is the
and rC , calculated from the Arrhenius plots, are presented
0
0
rate of carbon formation reached at steady state in the ab-
sence of catalyst deactivation; the units of this parameter are
in Table 3. These values of Ea are similar to those com-
monly reported in the literature [54]. It can be seen that an
(gC g−ca1t min−1); the term rC is the product of the concentration
increase in temperature caused increases in rD, ψd, and rC ,
0
0

L. Dussault et al. / Journal of Catalysis 251 (2007) 223–232
229
Fig. 7. Influence of methane partial pressure on the kinetic parameters r
C
0
and r .
D
that is, increases in the methane decomposition rate, deactiva-
tion rate, and carbon diffusion through the metallic crystallites.
Furthermore, the higher value of the apparent activation ener-
gies also demonstrates the higher activity and sensibility of the
Ni0.9Cu0.1MgAl catalyst with the reaction temperature.
Fig. 8. Scheme with some parameters involved in the reaction kinetics of the
nanocarbonaceous growth mechanism.
respect to H2. This negative dependence is a consequence of the
competition effect of H2 inhibiting the main reaction and caus-
ing an apparent decrease in activity but an increase in catalyst
stability.
Fig. 7 shows that both rD and rC decreased with PH
.
0
2
The decrease in rD was caused by the inhibitory effect of H2
due to the competition with CH4 for the metallic surface sites
[19,57]. In addition, a higher H2 concentration favored the re-
verse methane decomposition reaction, that is, carbon gasifi-
Finally, Fig. 7 shows that rC followed an almost linear
0
increase with the PCH . This increase indicates a first-order
4
cation. The decrease in rC can be attributed to a reduction of
0
dependence of this parameter on methane concentration, in
agreement with previous studies [19,57]. As in the previous
cases, the higher values of the parameters obtained for the
Ni0.9Cu0.1MgAl catalyst indicate the greater activity of this cat-
alyst compared with Ni0.8Cu0.2MgAl catalyst. As in the case of
the study of influence of partial pressure of H2, we also have as-
effective sites caused by the inhibitory effect of hydrogen over
the metallic surface. The values of both parameters were higher
for the Ni0.9Cu0.1MgAl catalyst, according to its higher activ-
ity. To clarify the effect of hydrogen on the parameters, we have
assumed that rD and rC potentially depend on hydrogen pres-
0
sure, according to
sumed the same type of dependence for rD and rC with respect
0
ꢆ
ꢆ
m
m
D,CH
to partial pressure of methane as shown in Eqs. (7) and (8). In
this case, the reference pressure of methane was 0.05 atm. To-
gether with the apparent activation energies, these kinetics dates
allow us to calculate reaction rates whatever the operating con-
ditions within the range studied in this work.
D,H
2
4
P
P
CH4
H2
rD = rD
(7)
(8)
ref
P
H2,ref
P
CH4,ref
and
ꢆ
ꢆ
m
m
C,H
C,CH
4
2
P
H2
P
CH4
rC = rC
.
0
0,ref
P
P
CH4,ref
H2,ref
3.5. Catalyst and carbon characterization after reaction
Here PH was normalized by dividing it by a reference pressure,
2
usually the average value considering all of the experiments
The XRD patterns of the reaction products (Fig. 1c) corre-
spond exactly to the superposition of the diffractogram of the
reduced catalyst (Fig. 1b) and that of polyaromatic carbon typ-
ical of graphene-based materials (002, 10, 004, and 11 reflec-
tions), such as CNFs and multiwall nanotubes. Thus, it seems
conducted. Therefore, rC
represents the preexponential pa-
0,ref
rameter at this reference pressure. For our calculations, we used
a H2 reference pressure of 0.05 atm. The results, given in Ta-
ble 4, indicate a strong negative dependence of rD and rC with
0
Table 4
Numerical values of potentially dependence on r and r parameters with hydrogen and methane pressure. Preexponential parameters and kinetic orders
D
C
0
P
H
P
CH
2
4
Ni Cu MgAl
0.8 0.2
Ni Cu MgAl
0.9 0.1
Ni Cu MgAl
0.8 0.2
Ni Cu MgAl
0.9 0.1
r
0.035
−2.3
0.071
−1.4
0.266
0.38
0.010
0.95
0.293
0.15
0.020
0.96
Dref
m
D
r
Cref
0.007
−0.6
0.008
−0.7
m
C

230
L. Dussault et al. / Journal of Catalysis 251 (2007) 223–232
Fig. 9. SEM image of CNFs produced at 923 K over Ni Cu MgAl catalyst. Feed composition: 5% CH ; 95% N .
0.9 0.1
4
2
Table 5
extent when the colloid mill is used instead of the traditional
mode of LDH preparation. The filaments generally had a diam-
eter of 60–100 nm (with some thicker fibers up to 400 nm) for
Ni0.8Cu0.2MgAl (with the two preparation modes), 30–50 nm
for Ni0.9Cu0.1MgAl (traditional preparation mode), and 20–
50 nm for Ni0.9 Cu0.1MgAl (use of colloid mill).
Structural parameters of the CNFs from XRD: interlayer distance d , average
size of coherently scattering domains along the normal to graphite layers (L )
and in the layer plane (L )
002
c
a
x
0.2
0.2
0.1
0.1
Preparation
method
Conventional
coprecipitation
Colloid
mill
0.335
7.1
Conventional
coprecipitation
0.334
6.1
8.0
Colloid
mill
0.338
5.9
The graphene-based structure of the fibers observed in the
XRD patterns (Fig. 1c) is confirmed by the Raman spectra and
the XPS (C1s) analysis (not shown here). Raman spectroscopy
is a powerful tool for studying carbon materials [58,59]. All
of the Raman spectra in this study showed two intense bands,
the first at 1580 cm−1 (E2g2, usually called the G-band), as-
signed to the in-plane vibrations of carbon atoms in the hexag-
onal sheets, and the second at 1350 cm−1 (usually called the
D-band), associated with disorder (lack of long-range trans-
lational symmetry) [60] and defects in the structure. Other
bands were also associated with this disorder, including one
at ∼1610 cm−1 (Dꢀ-mode), which became a shoulder of the
G-band, and one at ∼2930 cm−1 (2D-mode). The frequency
at about 2690 cm−1 also is associated with graphene perfec-
tion. The band at 1450–1550 cm−1, characteristic of amorphous
saturated carbon [61,62], was missing, indicating the very low
content of this type of carbon in the products.
d
(nm)
0.331
8.5
9.3
002
L (nm)
c
L (nm)
a
7.7
5.4
that the catalysts suffered no degradation during the first hours
of reaction, and the sharpness of the related peaks indicates
that the carbon thus formed had a relatively high nanotexture,
related to relatively large, defect-free graphenes. This is con-
firmed by the structural parameters of the CNFs determined by
XRD (Table 5), that is, the interlayer distance, d002, and the av-
erage size of coherently scattering domains along the normal
graphite layers (Lc) and in the layer plane (La). The average
size decreased with decreasing x for both methods of cata-
lyst synthesis used. The large sizes can be associated with a
graphitic structure, whereas the smallest ones correspond to a
turbostratic structure.
The filamentous morphology of the carbon products is il-
lustrated by the SEM micrograph in Fig. 9. This figure clearly
shows a dense population of entangled CNFs that form nu-
merous mesopores (interstices between interlaced nanofibers).
Thus these products contain both micropores and mesopores.
The BET data for the products (about 165 m2 g−1 with
Ni0.8Cu0.2MgAl and 130 m2 g−1 with Ni0.9Cu0.1MgAl) seem
to indicate a larger surface area for the former, which could
correspond to a higher density of surface defects.
The ratios of the bands G and D (IG/ID: 0.51 and 0.77 for
the products obtained over Ni0.8Cu0.2MgAl0.5 and Ni0.9Cu0.1
-
MgAl0.5, respectively), at a reaction temperature of 923 K and a
feed composition of 5% CH4 and 95% N2, indicate that the car-
bon obtained over the Ni0.8Cu0.2MgAl0.5 catalyst had a more
greatly disordered structure.
These observations agree with the findings of TG/DTG
analyses (Fig. 10). The thermogravimetric curves showed a
weight loss between 670 and 870 K. The differential curves
indicated that the main part of the carbon loss was polyaro-
matic; the maximum rate was measured at about 850 K in all
samples. The DTG peak exhibited a shoulder toward the low
According to the above observations, the size of the cat-
alyst particles, and consequently the diameter of the carbon
filaments, decreased with decreasing Cu content and to a lesser

L. Dussault et al. / Journal of Catalysis 251 (2007) 223–232
231
tion and growth. From this standpoint, the two best catalysts
of the series studied were Ni0.8Cu0.2MgAl (corresponding to
7.6 wt% of Cu) and especially Ni0.9Cu0.1MgAl (correspond-
ing to 3.8 wt% of Cu). According to the XRD and Raman
spectra, the carbon products were typically polyaromatic CNFs
with either a graphitic or a turbostratic structure. MEB images
showed a mesoporosity due to the tangling of the nanofibers.
The structure, texture, and growth mechanism of the CNFs will
be discussed in a forthcoming paper [34].
Acknowledgments
Financial support was provided by the MEC, Madrid (project
CTQ 2004-03973/PPQ). The authors thank J.P. Grenet and
E. Pere (IPREM-ECP) for the Raman spectra and C. Quet
(GRL-Arkema) for the MEB images.
References
[1] K.P. De Jong, J.W. Geus, Catal. Rev. Sci. Eng. 42 (2000) 481.
[2] P. Serp, M. Corrias, P. Kalk, Appl. Catal. A 253 (2003) 337.
[3] F. Salman, C. Park, R.T.K. Baker, Catal. Today 53 (1999) 385.
[4] C. Pham-Huu, N. Keller, G. Ehret, L.J. Charbonniere, R. Ziessel, M.J.
Ledoux, J. Mol. Catal. A 170 (2001) 155.
[5] M.L. Toebes, J.H. Bitter, A.J. Van Dillen, K.P. de Jong, Catal. Today 76
(2002) 33.
[6] M.L. Toebes, F.F. Prinsloo, J.H. Bitter, A.J. Van-Dillen, K.P. de Jong,
J. Catal. 214 (2003) 78.
[7] K. Takehira, T. Ohi, T. Shishido, T. Kawabata, K. Takaki, Appl. Catal.
A 283 (2005) 137.
[8] C.A. Bessel, K. Lauberds, N.M. Rodriguez, R.T.K. Baker, J. Phys. Chem.
B 105 (2001) 1115.
[9] S. Yoon, C. Park, H. Yang, Y. Korai, I. Mochida, R.T.K. Baker, N.M. Ro-
driguez, Carbon 42 (2004) 21.
Fig. 10. TG/DTG analyses of carbon produced over Ni Cu MgAl (a) and
0.8 0.2
Ni Cu MgAl (b) catalysts.
0.9 0.1
temperatures (∼720 K), which could correspond to amorphous
carbon. This shoulder was slightly more intense and the width
of the DTG peak at 850 K was greater for the product obtained
over the Ni0.8Cu0.2MgAl catalyst, possibly corresponding to a
greater variation in size and perhaps also to slight differences in
carbon filament texture or structure.
4. Conclusion
Numerous previous works have demonstrated that mixed
or alloy catalysts have several advantages over single-metal
catalysts for the production of nanostructured carbons. Conse-
quently, we studied a series of catalysts (Ni1−xCuxMgAl with
0 < x < 1) elaborated from LDHs. These precursors have the
advantage of giving well-crystallized mixed oxides with a uni-
form distribution of the metallic elements. The physicochemi-
cal characteristics (e.g., reducibility, particle size) of these cat-
alysts depend on their composition. The different techniques
used (XRD, TPR, XPS) showed that the prereduced active cata-
lysts were formed by particles of Cu0 and Ni0 (probably Ni–Cu
alloy) on MgO, NiAl2O4, and MgAl2O4 supports. The shape
(octahedral and cubo-octahedral) and the size of the catalyst
particles varied with the composition, that is, with the value
of x (decreasing as x decreased). The mode of synthesis of the
LDH precursor also had a slight influence on particle size, espe-
cially on size distribution, which was narrower with the use of a
colloid mill than with the conventional coprecipitation method.
The kinetic study of the carbon growth from the catalytic
decomposition of methane indicates on the one hand good ac-
tivity of the catalysts with low Cu content (x ꢀ 0.2) and on
the other hand good stability over time of the Ni1−xCuxMgAl
catalyst compared with the NiMgAl homologue. The kinetic
model used allows discrimination of the influence of the op-
erating conditions on the successive stages of CNF forma-
[10] V.B. Fenelonov, L.B. Avdeeva, O.V. Goncharova, L.G. Okkel, P.A. Si-
monov, A.Y. Derevyankin, V.A. Likholobov, Stud. Surf. Sci. Catal.
[Preparation of Catalysts VI] 91 (1995) 825.
[11] E.I. Bagrii, I.F. Tretyakov, E.A. Trusova, V.T. Popov, D.I. Slovetsk,
Neftechemia 31 (1991) 236.
[12] K. Lozano, E.V. Barrera, J. Appl. Polym. Sci. 79 (2001) 125.
[13] H.M. Cheng, Q.H. Yang, C. Liu, Carbon 39 (2001) 1447.
[14] G. Tibbetts, G.P. Meisner, C.H. Olk, Carbon 39 (2001) 2291.
[15] A.D. Lueking, R.T. Yang, N.M. Rodriguez, R.T.K. Baker, Langmuir 20
(2004) 714.
[16] J.R. Rostrup-Nielsen, J. Catal. 33 (1974) 184.
[17] R.T.K. Baker, Carbon 27 (1989) 315.
[18] J.C. Rodriguez, E. Romeo, J.L.G. Fierro, J. Santamaría, A. Monzón, Catal.
Today 37 (1997) 255.
[19] J.I. Villacampa, C. Royo, E. Romeo, J.A. Montoya, P. Del Angel, A. Mon-
zon, Appl. Catal. A 252 (2003) 363.
[20] C. Guimon, A. Auroux, E. Romero, A. Monzon, Appl. Catal. A 251 (2003)
199.
[21] N.M. Rodriguez, J. Mater. Res. 8 (1993) 3233.
[22] G.A. Jablonski, F.W.A.H. Geurts, A. Sacco Jr., Carbon 19 (1992) 99.
[23] V. Ivanov, J.B. Nagy, P. Lambin, A. Lucas, X.B. Zhang, X.F. Zhang, D.
Bernaerts, G. Van Tendeloo, S. Amelinckx, J. Van Landuyt, Chem. Phys.
Lett. 223 (1994) 329.
[24] P. Wang, E. Tanabe, K. Ito, J. Jia, H. Morioka, T. Shishido, K. Takehira,
Appl. Catal. A 231 (2002) 35.
[25] S. Takenaka, Y. Shigeta, E. Tanabe, K.J. Otsuka, J. Catal. 220 (2002) 468.
[26] J. Jia, Y. Wang, E. Tanabe, T. Shishido, K. Takehira, Microporous Meso-
porous Mater. 57 (2003) 283.
[27] M.S. Kim, N.M. Rodriguez, R.T.K. Baker, J. Catal. 131 (1991) 60.

232
L. Dussault et al. / Journal of Catalysis 251 (2007) 223–232
[28] L.B. Avdeeva, O.V. Goncharova, D.I. Kochubey, V.I. Zaikovskii, L.M.
Plyasova, B.N. Novgorodov, Sh.K. Shaikhutdinov, Appl. Catal. A 141
(1996) 117.
[29] Y. Li, J. Chen, L. Chang, Y. Qin, J. Catal. 178 (1998) 76.
[30] V.V. Chesnokov, V.I. Zaikovskii, R.A. Buyanov, J. Mol. Catal. A 158
(2000) 267.
[47] D. Chen, K.O. Christensen, E. Ochoa-Fernández, Z. Yu, B. Tøtdal, N. La-
torre, A. Monzón, A. Holmen, J. Catal. 229 (2005) 82.
[48] G.G. Kuvshinov, Yu.I. Mogilnykh, D.G. Kuvshinov, V.I. Zaikovskii, L.B.
Avdeeva, Carbon 36 (1998) 87.
[49] M.A. Ermakova, D.Yu. Ermakov, L.M. Plyasova, G.G. Kuvshinov, Catal.
Lett. 62 (1999) 93.
[31] H. Wang, R.T.J. Baker, Phys. Chem. 108 (2004) 20273.
[32] J.I. Villacampa, N. Latorre, E. Romeo, C. Royo, A. Monzón, J.A. Mon-
toya, in: Proceedings of the XIX Simposio Iberoamericano de Catálisis,
Mérida, Yucatán, México, 5–11 September, 2004.
[50] G.G. Kuvshinov, D.G. Kuvshinov, A.M. Glushenkov, Chem. Sustain. De-
vel. 1 (2003) 135.
[51] T.V. Reshetenko, L.B. Avdeeva, Z.R. Ismagilov, A.L. Chuvilin, Appl.
Catal. A 247 (2003) 51.
[33] F. Cavani, F. Trifiro, A. Vaccari, Catal. Today 11 (1991) 173.
[34] M. Monthioux, L. Noé, L. Dussault, J.C. Dupin, N. Latorre, T. Ubieto,
E. Romeo, C. Royo, A. Monzón, C. Guimon, J. Phys. Chem. B (2007),
submitted for publication.
[35] V. Rives (Ed.), Layered Double Hydroxides: Present and Future, Nova
Science Publishers, New York, 2001.
[52] Y. Nishiyama, Y. Tamai, J. Catal. 33 (1974) 98.
[53] A. Monzón, N. Latorre, T. Ubieto, C. Royo, E. Romeo, J.I. Villacampa,
L. Dussault, J.C. Dupin, C. Guimon, M. Monthioux, Catal. Today 116
(2006) 264.
[54] M. Pérez-Cabero, E. Romeo, C. Royo, A. Monzón, A. Guerrero-Ruiz,
I. Rodríguez-Ramos, J. Catal. 224 (2004) 197.
[55] A. Monzón, N. Latorre, T. Ubieto, A. Valera, E. Romeo, C. Royo, J.I. Vil-
lacampa, in: Proceedings of the AICHE 2005 Annual Meeting, Cincinnati
OH (E.E.U.U.), October 2005. Ref. meeting del AIChE 2005 y Congreso
de desactivacion de 2006.
[56] A. Monzón, N. Latorre, T. Ubieto, A. Valera, E. Romeo, C. Royo, J.I. Vil-
lacampa, in: Proceedings of the 10th Symposium on Catalyst Deactivation,
Berlín (Alemania), February 2006.
[57] J.W. Snoeck, G.F. Froment, M. Fowles, J. Catal. 169 (1997) 240.
[58] T.V. Reshetenko, L.B. Avdeeva, Z.R. Ismagilov, V.V. Pushkarev, S.V.
Cherepanova, A.L. Chuvilin, V.A. Likholobov, Carbon 41 (2003) 1605.
[59] S.K. Doorn, L. Zheng, M.J. O’Connel, Y. Zhu, S. Huang, J. Liu, J. Phys.
Chem. 109 (2005) 3751.
[60] F. Tuinstra, J.L. Koenig, J. Chem. Phys. 53 (1970) 1126.
[61] Y. Wang, D.C. Alsmeyer, R.L. McCreery, Chem. Mater. 2 (1990)
557.
[36] Y. Zhao, F. Li, R. Zhang, D.G. Evans, X. Duan, Chem. Mater. 14 (2002)
4286.
[37] J.H. Scofield, J. Electron. Spectrosc. Relat. Phenom. 8 (1976) 129.
[38] S. Casenave, H. Martinez, C. Guimon, A. Auroux, V. Hulea, A. Cor-
doneanu, E. Dumitriu, Thermochim. Acta 379 (2001) 85.
[39] G. Fierro, M. Lo Jacono, M. Inversi, P. Porta, F. Cioci, R. Lavecchia, Appl.
Catal. A 137 (1996) 327.
[40] K.C. Khulbe, R.S. Mann, Catal. Rev. Sci. Eng. 24 (1982) 311.
[41] M. Ferrandon, E. Björnbom, J. Catal. 200 (2001) 148.
[42] R. Brown, M.E. Cooper, D.A. Whan, Appl. Catal. 3 (1982) 177.
[43] V.R. Choudary, A.M. Rajputa, B. Prabhakar, J. Catal. 13 (1993) 326.
[44] J.A. Peña, J. Herguido, C. Guimon, A. Monzón, J. Santamaría, J. Ca-
tal. 159 (1996) 313.
[45] A. Monzon, E. Romeo, C. Royo, R. Trujillano, F.M. Labajos, V. Rives,
Appl. Catal. A Gen. 185 (1999) 53.
[46] J.C. Rodriguez, A.J. Marchi, A. Borgna, A. Monzon, J. Catal. 171 (1997)
268.
[62] T. Jawhari, A. Roid, J. Casado, Carbon 33 (1995) 1561.