New Ni-Cu-Mg-Al-based catalysts preparation procedures for the synthesis of carbon nanofibers and nanotubes

Article abstract of DOI:10.1016/j.jpcs.2006.01.041

Catalytic decomposition of methane to produce carbon nanofilaments, CNFs (nanofibers and/or nanotubes), and hydrogen was carried out on Ni-Cu-Mg-Al catalysts. The catalysts used as mixed oxides were obtained from the calcination at 1073 K of the corresponding lamellar double hydroxides (LDHs). The subsequent reduction at 1023 K of the calcined LDHs gave rise to Ni⁰ and Cu⁰ (and possibly Ni-Cu alloy) associated to an inorganic substrate with a MgO-MgAl₂O₄-NiAl₂O₄ mixed composition that were able to produce quality CNFs at 923 K. The size of metal particles and the carbon filaments as well as the nanofilament texture depends on the copper content of the catalyst. At a Cu content of 7.6 wt%, the carbon nanofilaments are nanofibers with a platelet texture, and the particles and CNFs sizes are widely distributed (50-400 nm). On the other hand, for a Cu content of 3.8 wt%, the size distribution is narrower and the CNFs are thinner (20-50 nm), nanofibers and nanotubes showing a 'herringbone' texture with twice the angle between the graphene layers and fiber axis, exhibiting values mainly in the range of 40°-60°.

Full text of DOI:10.1016/j.jpcs.2006.01.041

Journal of Physics and Chemistry of Solids 67 (2006) 1162–1167
www.elsevier.com/locate/jpcs
New Ni–Cu–Mg–Al-based catalysts preparation procedures for the synthesis
of carbon nanofibers and nanotubes
a
L. Dussault , J.C. Dupin , N. Latorre , T. Ubieto , L. Noe , M. Monthioux , E. Romeo ,
a
b
b
c
c
b
´
C. Royo , A. Monzon , C. Guimon
b
b
a,
*
´
a LCTPCM-UMR(CNRS) 5624 and FR 2606, 2 Av. P. Angot, 64053 Pau Cedex 9, France
b Dept Chemical and Environmental Engineering, University of Zaragoza, 50009 Zaragoza, Spain
c CEMES, B.P. 94347, 31055 Toulouse Cedex 4, France
Abstract
Catalytic decomposition of methane to produce carbon nanofilaments, CNFs (nanofibers and/or nanotubes), and hydrogen was carried out on
Ni–Cu–Mg–Al catalysts. The catalysts used as mixed oxides were obtained from the calcination at 1073 K of the corresponding lamellar double
hydroxides (LDHs). The subsequent reduction at 1023 K of the calcined LDHs gave rise to Ni0 and Cu0 (and possibly Ni–Cu alloy) associated to
an inorganic substrate with a MgO–MgAl2O4–NiAl2O4 mixed composition that were able to produce quality CNFs at 923 K. The size of metal
particles and the carbon filaments as well as the nanofilament texture depends on the copper content of the catalyst. At a Cu content of 7.6 wt%, the
carbon nanofilaments are nanofibers with a platelet texture, and the particles and CNFs sizes are widely distributed (50–400 nm). On the other
hand, for a Cu content of 3.8 wt%, the size distribution is narrower and the CNFs are thinner (20–50 nm), nanofibers and nanotubes showing a
‘herringbone’ texture with twice the angle between the graphene layers and fiber axis, exhibiting values mainly in the range of 408–608.
q 2006 Elsevier Ltd. All rights reserved.
Keywords: A. Nanostructures; A. Oxides; C. Electron microscopy; C. Photoelectron spectroscopy; C. Raman Spectroscopy
1. Introduction
approximately 3008 lower than that of Fe catalysts. However,
an important drawback is the deactivation of catalyst by
sintering and/or by the deposition of amorphous and/or non-
filamentous graphene-based carbons which encapsulate the
metal particles [6]. Previous studies showed that Ni–Al and
Ni–Mg–Al catalysts suffer such a deactivation at temperatures
higher than 850 K and that Cu enhances the activity and the
stability of Ni–Al catalysts [7]. Whereas the presence of Cu can
has a dilution effect on the metallic phase, thereby diminishing
its reaction rate, the Ni–Cu–Mg–Al catalysts prepared in this
study show increased resistance to deactivation because Cu
favors a higher hydrogen mobility inhibiting the formation of
encapsulating carbon deposit.
Nickel-based catalysts are well known for their activity in the
decomposition of methane to produce hydrogen (alternative
route to the steam reforming) and carbon. Sometimes this process
is called catalytic chemical vapor deposition (CCVD). Among
the numerous types of carbon nanomaterials able to be produced,
nanotubes [1,2] and nanofibers [3,4] present a great interest and
have been the subject of numerous research studies. Carbon
nanofibers have received considerable attention, in particular
because of their excellent adsorbent properties [5] and potential
use as catalyst support [3,4] thanks to their mesoporosity and
their relatively large surface area (between 100 and 300 m2 gK1).
The latter properties result from their size and texture (orientation
of graphene layers) which is a function of both the composition
and structure of the catalysts used in the CCVD process.
In order to maximize the inter-metallic interactions and to
minimize the sintering effects, the catalysts were obtained from
lamellar double hydroxides, LDHs [8]. Two modes of
preparation of LDHs ðNixCu1KxMgAlÞðOHÞ7ðCO23KÞ0:5, with
0%x%1, were used: (i) the conventional method and (ii) the
colloid mill co-precipitation method. After calcination of
LDHs at 1073 K, the resulting (NiO)1Kx(CuO)xMgO(Al2O3)0.5
mixed oxides were reduced in situ at 1023 K and then used in
the decomposition of methane. This paper reports the results of
the characterization of catalysts and carbon nanofilaments and
One of the main advantages of Ni-based catalysts is that
the decomposition temperature of CH4 (873–973 K) is
*
Corresponding author. Tel.: C33 559407620; fax: C33 559407622.
E-mail address: claude.guimon@univ-pau.fr (C. Guimon).
0022-3697/$ - see front matter q 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jpcs.2006.01.041

L. Dussault et al. / Journal of Physics and Chemistry of Solids 67 (2006) 1162–1167
1163
the catalyst activity as a function of catalyst composition and
preparation route.
treatment. The TPR profiles (Fig. 1) show two reduction peaks,
the first at 440–500 K, the second at 970–1170 K.
The first peak corresponds to the reduction of copper
(CuII/Cu0). High Cu content samples (x ranging from 0.5
to 1) present a single band with a FWHM of about 50 K. Lower
Cu content catalysts present a double peak with smaller widths
(w20 K), which indicates smaller particle sizes. The high
temperature component (500 K) can be attributed to segregated
CuO particles, the low temperature component (460 K) to
copper oxide in interaction with the other metals [10], which
can form Ni–Cu alloy. In this alloy, there is an electron
donation from Cu to d orbitals of Ni, which can explain the
easier reducibility of Cu [11]. The absence of a peak towards
770 K shows that CuAl2O4 was not formed [12].
2. Experimental
2.1. Materials
LDH samples with different Ni/Cu/Mg/Al compositions
were prepared using two different methods, the conventional
[8] and the colloid mill [9] co-precipitation methods, with
appropriate amounts of Mg(NO3)2$6H2O, Al(NO3)3$9H2O,
Cu(NO3)2 3H2O, Ni(NO3)2$6H2O, NaOH (to maintain pH
constant at 9.5G0.2) and a slight excess of Na2CO3.
Then, in both cases, the catalyst precursors were calcined at
1073 K under a nitrogen flow during 11 h, then reduced in a
thermobalance (C.I.Electronics) by a N2/H2 mixture (1/1;
150 ml minK1) at 1023 K during 2.8 h. After 150 min at 923 K,
methane (5%) was introduced along with N2 (95%) (total flow
rate of 750 ml minK1) for 3 h. According to a kinetic study (to
be published elsewhere), these reaction conditions were able to
synthesize carbon nanofilaments with the highest carbon yield.
In the following, the catalysts will be noted as Ni1KxCux-
MgAl (for example Ni0.8Cu0.2MgAl for xZ0.2) the for the sake
of simplification.
The second region between 970 and 1170 K is associated
with the reduction of nickel (NiII/Ni0) since magnesium and
aluminum oxides do not reduce below 1170 K. As in the case
of copper peak, the broad band can be decomposed into two
sub-bands centered at 1020 and 1120 K, respectively. The
former should correspond to NiO particles in intimate contact
with the support. The latter can be associated to the reduction
of nickel aluminate. The presence of these types of nickel is
confirmed by the XPS analysis of the samples reduced at
1023 K. For the high Ni content, when the copper is totally
reduced (binding energy Cu2p3/2 at 932.5 eV), the Ni2p3/2 band
indicates the presence of approximately 30% of Ni0 (binding
energy at 852.5 eV) and 70% of NiII (binding energy at
855 eV). It seems that the ratio NiO/NiAl2O4 decreases with
increasing Ni content. However, the total NiO content after the
reduction step seems to reach its maximum for a composition
2.2. Characterization
X-ray diffraction (XRD) patterns were recorded in an INEL
diffractometer using a curved position-sensitive detector and a
˚
monochromatic radiation (Cu Ka, 1.5406 A).
XPS analyses were carried out in a Kratos spectrometer
(model Axis Ultra) using a monochromatic Al Ka radiation
(1486.6 eV), with the entry of its introduction chamber inside a
glove box.
Temperature-Programmed Reduction (TPR) measurements
of the samples were performed in a flow quartz reactor (GIRA,
X-sorb model equipped with a TCD) with a hydrogen–argon
mixture (5–95%) at a flow rate of 20 ml N minK1 and bZ
10 K minK1
.
The Raman spectra were performed using a Jobin Yvon
T64000 confocal spectrophotometer with 514.5 nm ArC ion
laser as excitation source.
Transmission electron microscopy micrographs were
recorded with either a Philips CM12 or CM30 microscope,
LaB6 electron source, operated at 120 and 150 kV,
respectively, so that the irradiation damages are limited.
Specimens for TEM were prepared by gently grinding the
samples, making a sonicated suspension from the resulting
powder, and depositing a droplet of the suspension onto a
copper microgrid coated with a lacey carbon film.
3. Results and discussion
3.1. Characterization of catalysts
The catalysts are active in reduced state. TPR profile can
indicate the degree of metallic phase present after activation
Fig. 1. TPR (293–1173 K) of the mixed oxides (NiO)1Kx(CuO)xMgO(Al2O3)0.5
obtained by calcination of corresponding LDHs at 1073 K.

1164
L. Dussault et al. / Journal of Physics and Chemistry of Solids 67 (2006) 1162–1167
Table 1
with a value of x located in the range 0.1–0.2. According to
TPR and XRD results (not shown here), the reduced catalysts
prepared at 1023 K are composed of Cu0, Ni0, possibly Ni–Cu
alloy, MgO, NiAl2O4 and MgAl2O4.
Structural parameters of carbon nanofilaments: interlayer distance d002, average
size of coherently scattering domains along the normal to graphene layers (Lc)
and in the graphene plane (La) from XRD
x
0.2
0.1
0.1
Preparation method
d002 (nm)
Regular
0.331
8.3
Regular
0.334
5.9
Colloid mill
3.2. Carbon nanofilament growth
0.338
5.6
Lc (XRD) (nm)
La (XRD) (nm)
ID/IG (Raman)
La (Raman) (nm)
In agreement with the literature [7], an excess content of
copper dramatically decreases the carbon yield and the
catalysts with xS0.5 show very little activity. Just as Al, Cu
is relatively non-catalytic and has a dilution effect on the active
sites of Ni. However, it enhances both the carbon diffusion and
reaction rates and thus promotes the catalytic activity when it is
incorporated in small quantities. In agreement with the
literature [7], the maximal carbon yield is reached with
Ni0.9Cu0.1MgAl, i.e. 3.8 wt% of Cu (Ni0.8Cu0.2MgAl, i.e.
7.6 wt% of Cu and Ni0.95Cu0.05MgAl, i.e. 1.9 wt% of Cu, are
slightly less active).
7.3
6.2
4.2
2.4
1.3
1.7
1.8
3.4
2.6
La from Raman spectra is calculated with ID/IGZ4.4/La [15].
In particular, the ratio ID/IG is inversely proportional to the
dimension La, i.e. the average planar size of the perfect
graphenes [15]. The values reported in Table 1 indicate that the
CNFs produced with Ni0.8Cu0.2MgAl have a higher disordered
structure (smaller La) than those obtained with Ni0.9Cu0.1-
MgAl. The discrepancy between La values obtained from XRD
and Raman spectroscopy comes from the underestimation of
values obtained with Raman due to the dominant effect of the
smallest coherent domains [14]. The larger difference between
the results of the two techniques for CNFs from Ni0.8Cu0.2-
MgAl indicates that they present a higher dispersion of
crystallite sizes although the average polyaromatic coherent
domains are larger (following both dimensions La and Lc).
Fig. 2a shows typical TEM images of the CNFs formed over
either Ni0.8Cu0.2MgAl or Ni0.9Cu0.1MgAl. Nanofilaments
actually nanofibers (i.e. not hollowed) are frequently attached
to the catalyst particle where they have grown from (Fig. 2a).
The metallic Ni–Cu particle is detached from the support
possibly because it is pushed up by the fiber during its growth
[4]. Octahedral or cubo-octahedral catalyst particles of
relatively large size (100–400 nm) were found well crystal-
lized, giving rise to sharp and bright dots in electron diffraction
patterns (Fig. 2b, arrows), and to lattice fringes in high-
resolution TEM images (Fig. 2e). The sample appeared quite
homogeneous (a very large majority of the filaments are from
the same type) with only a scarce occurrence of other filament
types (herringbone type, not illustrated). Fig. 2d indicates that
the graphene layers are oriented parallel to the particle surface,
making them perpendicular to the filament axis (‘platelet’
texture). Actually, arrows in Fig. 1a indicate the very location
where the nanofilament was about to split, due to easy gliding
resulting from the weak van der Waals forces between stacked
graphenes. Square-like particles that make up most of the
specimen after the TEM preparation procedure (grinding) are
therefore merely platelet-CNF fragments (not illustrated), most
of them exhibiting the same graphene display. Interestingly,
such a graphene display implies that the fiber surface is actually
made of free-graphene edges, meaning that high surface
reactivity and surface energy are expected. As stated by
numerous Bragg fringes (dark contrasts), the carbon nanofila-
ments decorated with the nanotexture are quite high. Electron
diffraction patterns (Fig. 2b) exhibit rather sharp and dotted
reflections together with multiple order reflections (e.g. 002,
004, 006) which both indicate large coherent domains, which is
The XRD patterns of the reaction products (not shown here)
correspond exactly to the superposition of the diffractogram of
reduced catalyst and of polyaromatic carbon typical of
graphene-based materials such as carbon nanofibers and
multi-walled nanotubes. Thus, it seems that the catalysts do
not suffer any degradation during the first hours of reaction and
sharpness of the related peaks indicates that the carbon formed
presents a relatively high nanotexture, which corresponds to
relatively large defect-free graphenes.
Beside its doping character, copper has a strong effect on the
lifetime of the catalysts. This is shown by the comparison of
carbon growth rates on Ni0.9Cu0.1MgAl (xZ0.1) and NiMgAl
(xZ0). While the latter decreases only after 15 min and is close
to zero after 1 h, the former is quasi-constant during the whole
duration of the test showing a high stability. The benefit of Cu
has already been noted in the case of hydrogenation catalysts.
Indeed, Cu favors a higher hydrogen mobility, it has high
affinity with graphene-based structures, and tends to inhibit the
encapsulation of Ni active sites by polyaromatic carbons [13].
3.3. Nanofilament features
Because of their very low yield, we have not examined the
carbon products formed using the high Cu content catalysts,
but only those obtained with Ni0.8Cu0.2MgAl and Ni0.9Cu0.1-
MgAl synthesized with both methods. In agreement with the
previous observations, the size of the catalyst particle and, in
consequence, the diameters of the nanofilaments are smaller
when a lower Cu content catalyst is used and this effect
increases for the catalysts prepared with the colloid mill
method. The filaments generally show a diameter in the
range 60–100 nm (with some thicker ones up to 400 nm) for
Ni0.8Cu0.2MgAl from both preparation methods, 30–50 nm for
Ni0.9Cu0.1MgAl (traditional preparation mode) and 20–50 nm
for Ni0.9Cu0.1MgAl (from colloid mill).
The graphene-based structure of the nanofilaments observed
in XRD patterns is confirmed by the Raman spectra (not shown
here). The frequency and intensity of the D- and G-Raman
bands provide information about the crystallinity of CNFs [14].

L. Dussault et al. / Journal of Physics and Chemistry of Solids 67 (2006) 1162–1167
1165
Fig. 2. TEM images of carbon nanofiber formed over Ni0.8Cu0.2MgAl. (a) Low magnification. Arrows indicate steps at the CNF surface which result from a slight
gliding; (b) electron diffraction pattern giving rise to the labeled rings, with a very limited contribution of catalyst particles (arrowed spots); (c) enlargement of the
area framed in (b); (d) high magnification image; (e) high magnification image of catalyst particle.
consistent with the high nanotexture quality. The most
outstanding feature is, however, that nanofibers exhibit at
least a partially graphitic structure, i.e. many of the graphene
layers are stacked according to the ABAB stacking sequence
typical of graphite. Such a structural feature is ascertained by
the occurrence of three-indice reflections such as 110 and 112
(Fig. 2c), as opposed to two-indice asymmetric bands (such as
10 and 11) that are found in turbostratic, polyaromatic carbon
materials such as regular multi-wall carbon nanotubes
(MWNTs). All these results are in full agreement with the
previous data, particularly the interplanar distance d002
measured at 0.333 nm in Fig. 2d which matches well the
0.331 nm measured from XRD (Table 1) and is very close to
that of pure graphite (0.335 nm).
Fig. 3a gives a typical TEM image of the products obtained
over the catalyst Ni0.9Cu0.1MgAl (colloid mill). Ni0.9Cu0.1-
MgAl catalyst prepared following the common co-precipitation
method produces CNFs with similar features. As in the case of

1166
L. Dussault et al. / Journal of Physics and Chemistry of Solids 67 (2006) 1162–1167
Fig. 3. TEM images of typical carbon nanotubes formed over Ni0.9Cu0.1MgAl; (b) high magnification of a nanotube from catalysts obtained from colloid mill
method; (c) high magnification of nanotube from catalysts obtained from the conventional co-precipitation method.
xZ0.2, TEM images show a filament growing from single
octahedral or cubo-octahedral crystallites. As already
mentioned, the particle sizes (20–60 nm) and the resulting
CNF diameters (the majority in the range of 25–35 nm) are
much smaller than for the Ni0.8Cu0.2MgAl catalysts. All the
CNFs are herringbone-type nanotubes (i.e. hollowed). The
double angle between the graphene layers and the fiber axis
ranges between w308 and w1508, and most often between 308
and 808 (two examples are shown in Fig. 3b and c). This agrees
with the previous experimental observations [16] and theo-
retical calculations [17]. Statistics, however, show that the
conventional co-precipitation method tends clearly to provide
larger angle values than the colloid mill method (not shown). As
in the case of platelet nanofibers, a gentle grinding can dismantle
the herringbone-nanotubes, which thereby show mechanic
brittleness to shearing stresses. This agrees with a texture that
may be sketched by a stacking of cones rather than by a helical
display of a single graphene layer superimposing to itself at
each turn. The formation of curved graphene on the top of the
cone and the angle between the basal plane (constituted of
hexagonal rings) and the growth axis dependson the number n of
pentagonal carbon rings [18]. However, it is worth noting that as
opposed to nanofibers, the pentagon occurrence is virtual in the
case of nanotubes (as in Fig. 3) due to the hollow core.
According to Zhu et al. [17], in the example of Fig. 3, this
number could be 1 (corresponding to an angle equal to 56.58) for
the CNFs produced over the catalyst Ni0.9Cu0.1MgAl
obtained from the traditional co-precipitation method, and 2
(angle of 42.38) in the case of catalyst obtained from the colloid
mill method.
4. Conclusion
In this work, we have studied a series of catalysts
(Ni1KxCuxMgAl with 0%x%1) obtained by calcination of
lamellar double hydroxides. The physicochemical character-
istics (reducibility, particle size) of these catalysts depend on
their composition. It appears that the reduced catalysts are
formed of particles of Cu0 and Ni0 (with maybe Ni–Cu alloy)
on MgO, NiAl2O4 and MgAl2O4 support. The size of the
octahedral and cubo-octahedral particles decreases with
decreasing Cu amount.
The kinetic study of carbon growth from the catalytic
decomposition of methane indicates a high activity of the
catalysts at low Cu contents (x%0.2) and a high stability in
time of the catalysts Ni1KxCuxMgAl compared to that of
NiMgAl homologue.
The carbon products are typically polyaromatic carbon
nanofibers or nanotubes, with either a graphitic or a
turbostratic structure depending on the texture (either
‘platelet’ or ‘herringbone’). The latter depends on the size
of the catalyst particles, which is related to their composition.
For xZ0.2, there is a wide distribution of catalyst particle
sizes, hence a wide distribution of filament diameters
(between 60 and 400 nm). For xZ0.1, the filament diameters
range between 20 and 50 nm. The thickest filaments are

L. Dussault et al. / Journal of Physics and Chemistry of Solids 67 (2006) 1162–1167
1167
[5] V.B. Fenelonov, L.B. Avdeeva, O.V. Goncharova, L.G. Okkel,
P.A. Simonov, A.Y. Derevyankin, V.A. Likholobov, Stud. Surf. Sci.
Catal. 91 (1995) 825.
nanofibers with a ‘platelet’ texture (graphene layers perpen-
dicular to the fiber axis) with large domains of high
nanotexture quality. On the other hand, filaments with a
diameter smaller than 50 nm are nanotubes with a ‘herring-
bone’ texture, where the graphene layers are oriented
obliquely, with the angle between the graphene planes and
the fiber axis ranging from 308 to 608.
[6] C. Guimon, A. Auroux, E. Romero, A. Monzon, Appl. Catal. A 251
(2003) 199.
[7] Y. Li, J. Chen, L. Chang, Y. Qin, J. Catal. 178 (1998) 76.
[8] F. Cavani, F. Trifiro, A. Vaccari, Catal. Today 11 (1991) 173.
[9] Y. Zhao, F. Li, R. Zhang, D.G. Evans, X. Duan, Chem. Mater. 14 (2002)
4286.
[10] G. Fierro, M. Lo Jacono, M. Inversi, P. Porta, F. Cioci, R. Lavecchia,
Appl. Catal. A 137 (1996) 327.
Acknowledgements
[11] I. Alstrup, M.T. Tavares, J. Catal. 139 (1993) 513.
¨
[12] M. Ferrandon, E. Bjornbom, J. Catal. 200 (2001) 148.
The authors would like to thank E. Pere and J.P. Grenet for
the Raman measurements.
[13] J.S. Zhao, Y.D. Li, F.S. Ma, H.M. Kang, C.L. Zhu, Fuel Chem. Technol.
14 (1986) 201.
[14] T.V. Reshetenko, L.B. Avdeeva, Z.R. Ismagilov, V.V. Pushkarev,
S.V.Cherepanova, A.L. Chuvilin, V.A. Likholobov, Carbon41(2003)1605.
[15] D.S. Knight, W.B. White, J. Mater. Res. 4 (1989) 385.
[16] A. Krishnan, E. Dujardin, M.M.J. Treacy, J. Hugdahl, S. Lynum,
T.W. Ebbesen, Nature (London) 388 (1997) 451.
References
[1] S. Iijima, T. Ichihashi, Nature 363 (1993) 603.
[2] D.S. Bethune, C.H. Klang, M.S.D. De Vries, J. Gorman, Nature 363
(1993) 605.
[17] Y.A. Zhu, Z.J. Sui, T.J. Zhao, Y.Ch. Dai, Zh.M. Cheng, W.K. Yuan,
Carbon. 43 (2005) 1694.
[3] K.P. de Jong, J.W. Geus, Catal. Rev.-Sci. Eng. 42 (2000) 481.
[4] P. Serp, M. Corrias, P. Kalk, Appl. Catal. A 253 (2) (2003) 337.
[18] H. Terrones, T. Hayashi, M. Munoz-Navia, M. Terrones, Y.A. Kim,
N. Grobert, Chem. Phys. Lett. 343 (2001) 241.