Preparation of rhodium catalysts supported on carbon nanotubes by a surface mediated organometallic reaction

Article abstract of DOI:10.1002/ejic.200390083

The dimeric complex [Rh₂Cl₂(CO)₄] was grafted to multiwalled carbon nanotubes (MWNTs) previously oxidised with nitric acid and then treated with sodium carbonate to produce carboxylate groups on their outer surface. The grafting mechanism involves bridge-splitting and substitution of COO for Cl, as evidenced by the presence of NaCl in the samples. A further reduction/decomposition step under dihydrogen at 573 K afforded highly dispersed rhodium nanoparticles (≈1.5-2.5 nm). This novel rhodium-supported carbon material (Rh/MWNT-COONa) was tested as a catalyst for the hydrogenation of trans-cinnamaldehyde and the hydroformylation of hex-1-ene in the liquid phase: in both cases it was found to be very selective, toward C=C double-bond hydrogenation and the production of linear and branched aldehydes, respectively. A comparison was made between its catalytic activity and that of rhodium supported on pristine MWNTs and on nitric acid-oxidised MWNTs. We observed that these latter catalysts have larger particle sizes and lower activities, thus confirming the efficiency of our grafting procedure. Moreover, the better results obtained when using MWNT-COONa as support with respect to carboxylate-containing activated carbon also point to the important role played by the mesoporous nature of the carbon nanotube support, which can ameliorate transfer processes. Wiley-VCH Verlag GmbH & Co. KGaA, 69451 Weinheim, Germany, 2003.

Full text of DOI:10.1002/ejic.200390083

FULL PAPER
Preparation of Rhodium Catalysts Supported on Carbon Nanotubes by a
Surface Mediated Organometallic Reaction
Roberto Giordano,[a] Philippe Serp,*[a] Philippe Kalck,[a] Yolande Kihn,[b]
Joachim Schreiber,[c] Christiane Marhic,[c] and Jean-Luc Duvail[c]
Keywords: Nanotubes / Rhodium / Supported catalysts / Surface chemistry / Heterogeneous catalysis
The dimeric complex [Rh2Cl2(CO)4] was grafted to multi-
walled carbon nanotubes (MWNTs) previously oxidised with
nitric acid and then treated with sodium carbonate to pro-
duce carboxylate groups on their outer surface. The grafting
mechanism involves bridge-splitting and substitution of -
COO for Cl, as evidenced by the presence of NaCl in the
drogenation and the production of linear and branched alde-
hydes, respectively. A comparison was made between its
catalytic activity and that of rhodium supported on pristine
MWNTs and on nitric acid-oxidised MWNTs. We observed
that these latter catalysts have larger particle sizes and lower
activities, thus confirming the efficiency of our grafting pro-
samples. A further reduction/decomposition step under dihy- cedure. Moreover, the better results obtained when using
drogen at 573 K afforded highly dispersed rhodium nanopar- MWNT-COONa as support with respect to carboxylate-con-
ticles (ഠ1.5−2.5 nm). This novel rhodium-supported carbon
material (Rh/MWNT-COONa) was tested as a catalyst for the
taining activated carbon also point to the important role
played by the mesoporous nature of the carbon nanotube
hydrogenation of trans-cinnamaldehyde and the hydrofor- support, which can ameliorate transfer processes.
mylation of hex-1-ene in the liquid phase: in both cases it ( Wiley-VCH Verlag GmbH & Co. KGaA, 69451 Weinheim,
was found to be very selective, toward C=C double-bond hy- Germany, 2003)
Introduction
entations along the fibre axis — the most common struc-
ture being the one in which the planes are stacked up like
a fishbone, with a lot of edges — both single-walled
(SWNT)- and multi-walled (MWNT) carbon nanotubes
show a concentric wall structure consisting of ordered
graphite platelets. The morphology and the size of the
above supports, especially since they present huge lengths
vs. diameters, can play a significant role in catalytic applica-
tions owing to their ability to disperse the active phase.
Their electronic properties are also of primary import-
ance,[4] and their mechanical strength makes them resistant
in view of recycling. A recent comparison between the inter-
action of transition metal atoms with carbon nanotube
walls and with graphite indicates major differences in bond-
ing sites, magnetic moments and charge-transfer direc-
tion.[5]
The most widely used technique to achieve the deposition
of metals on the above materials is the incipient wetness
impregnation one, in which the purified carrier is impreg-
nated with a solution of the metal precursor, then dried,
calcinated and/or reduced so as to obtain metal particles
dispersed on the support.[3,6] Several studies have been per-
formed on graphite-nanofibre-supported Fe-Cu,[7] Ni[8] or
Pd[9] particles for catalytic hydrogenation. Graphite-nanofi-
bre-supported rhodium catalysts were found to be active in
ethene hydroformylation.[10] If we consider the use of car-
bon nanotubes, several examples are worth mentioning: Cu
Carbon materials are attractive supports in heterogen-
eous catalytic processes owing to their ability to be tailored
to meet specific needs. Indeed, activated forms of carbon
are currently employed as catalyst supports because of their
high surface area, their stability at high temperatures under
a nonoxidising atmosphere and the possibility of control-
ling both their porous structure and the chemical nature of
their surface.[1,2] New forms of carbon have appeared dur-
ing the last decade, showing novel interesting potentialities
as supports.[3] Carbon nanofibres, also called graphite nan-
ofibres, and carbon nanotubes, have so far been used suc-
cessfully and shown to present, as final supported materials,
catalytic properties superior to those of catalysts prepared
on activated carbon, soot, or graphite.[2] Whilst carbon nan-
ofibres consist of graphite sheets arranged in various ori-
[a]
`
Laboratoire de Catalyse, Chimie Fine et Polymeres, Ecole
´
´
Nationale Superieure des Ingenieurs en Arts Chimiques Et
Technologiques,
´
118 route de Narbonne, 31077 Toulouse Cedex 4, France
Fax: (internat.) ϩ33-5/6288-5600
E-mail: philippe.serp@ensiacet.fr
CEMES-CNRS no. 8011,
[b]
[c]
2 rue Jeanne Marvig, 31055 Toulouse, France
´
´
Institut des Materiaux Jean Rouxel-UMR 6502, Universite de
Nantes,
`
´
2 rue de le Houssiniere, BP 32229, 44322 Nantes Cedex 3,
France
610
 2003 WILEY-VCH Verlag GmbH & Co. KGaA, 69451 Weinheim 1434Ϫ1948/03/0204Ϫ0610 $ 20.00ϩ.50/0 Eur. J. Inorg. Chem. 2003, 610Ϫ617

Rhodium Catalysts Supported on Carbon Nanotubes
FULL PAPER
nanoparticles were synthesised employing SWNTs as a tem- highly dispersed Rh nanoparticles, after a decomposition/
plate;[11] Ru/SWNTs[12] and alkali-promoted Ru/ reduction step. We also report some preliminary results
MWNTs[13] catalysts were prepared for hydrogenation reac- concerning the catalytic activity of these novel materials
tions and for ammonia synthesis, respectively; Pd, Pt, Ag either in hydrogenation or hydroformylation reactions. Last
and Au nanoparticles were grown on SWNTs, and their size but not least, the catalytic activity of these materials is com-
could be adjusted by changing the metal to carbon ratio;[14] pared to that observed when using rhodium-supported
Rh/MWNTs catalysts for propene hydroformylation were catalysts prepared by reacting [Rh2Cl2(CO)4] with pristine
prepared by incipient wetness impregnation from benzene or HNO3-treated MWNTs, as well as with activated carbon
solutions of [RhH(CO)(PPh3)3] on pristine nanotubes.[15]
Chemical treatment and/or modification of the surface of
carbon nanotubes was found to be a useful tool for control-
ling their hydrophobic or hydrophilic character,[16] and
studies carried out in order to accomplish the deposition of
metals point to a strong interaction between metal atoms
and the groups created by oxidation of the support sur-
face.[17]
bearing sodium carboxylate groups.
Results and Discussion
Surface Modification of MWNTs
High purity samples of MWNTs were employed for this
study (MWNTs 97%, the remaining 3% coming from iron
nanoparticles encapsulated within the nanotubes and gener-
ated by the catalytic production process).[27] Typical TEM
micrographs of this material are shown in Figure 1.
Thus, graphite nanofibre-supported iron FischerϪ
Tropsch catalysts have been prepared by three different
methods on HNO3-treated supports.[18] Important improve-
ments toward the preparation of selective hydrogenation
catalysts were achieved by supporting Pt on SWNTs ox-
idised in dilute HNO3: the bonding between platinum and
the carbon nanotube surface was presumably accomplished
by ion-exchange on the carboxylic acid sites so formed.[19]
Another example of such reactivity is the replacement of
the proton of carboxylic groups with uranyl ions: XPS stud-
ies of oxidised nanotubes revealed that their surface is co-
vered by carboxylic, carbonyl and hydroxyl groups, and, in
particular, the presence of -COOH groups could be used to
initiate nucleation on the support surface, leading to the
selective covering of the outer surface by metallic species.[20]
A Co/MWNTs system showing a good catalytic activity in
dehydrogenation was prepared by oxidising MWNTs in
boiling nitric acid prior to impregnation with a cobalt salt,
followed by drying, calcination and reduction with H2.[21]
Bimetallic Ru/Pt and Ru/Sn nanoparticles have been pre-
pared from molecular heterobimetallic carbonyl clusters de-
posited on MWNTs previously treated with HNO3.[22]
A
heterobimetallic precursor was also used to prepare Pt-Ru
anode catalysts supported on oxidised and nonoxidised car-
bon nanofibres, SWNTs and MWNTs.[23] Coordination of
Pd on HNO3-treated carbon nanofibres[24] and of Pt on
H2SO4/HNO3-treated MWNTs has also been reported.[25]
A suitable technique to achieve the grafting of metal par-
ticles onto carbon nanotubes would therefore consist of
functionalising the outer surface of the tubes and then per-
forming a chemical reaction with a metal complex. This was
initially accomplished by reacting Vaska’s complex
[IrCl(CO)(PPh3)2] with a HNO3-oxidised support: the
tethering reaction occurs by coordination of iridium to the
nanotubes through the oxygen atoms.[26]
Furthermore, a controlled chemical grafting of a metal
species to the surface of carbon nanotubes would very likely
result in a fine dispersion of metallic particles.
Figure 1. Typical TEM micrographs of pristine MWNTs
In this paper we describe a new technique that involves
the organometallic reaction of the rhodium complex
It consists of graphene layers (interlayer spacing d002
ϭ
[Rh2Cl2(CO)4] with a modified MWNT surface bearing so- 0.342 nm) arranged in a roughly parallel way with respect
dium carboxylate groups, which permits the production of to the tube axis. Mean tube external and internal diameters
Eur. J. Inorg. Chem. 2003, 610Ϫ617
611

R. Giordano, P. Serp, P. Kalck, Y. Kihn, J. Schreiber, C. Marhic, J.-L. Duvail
FULL PAPER
are 17 nm and 8 nm, respectively, which corresponds to Furthermore, a slight increase in the surface area is ob-
about 13 walls. These nanotubes are characterised by a served in the case of MWNT-COOH (SBET ϭ 200 m2·gϪ1).
length of several micrometers and a specific surface area of
about 180 m2·gϪ1, with no microporosity. Pores in aggreg-
ated MWNTs can be mainly divided into inner hollow cav-
ities of small diameter (3Ϫ9 nm) and aggregated pores
(15Ϫ25 nm) formed by interaction of isolated MWNTs.[28]
From TEM observations it appears that one end of the tube
is closed by graphene layers, the remaining iron carbide
nanoparticles being located either at the tip of the tubes
(see Figure 1) or in their inner cavities. Moreover, this mat-
erial is highly hydrophobic since it does not present oxygen-
containing groups, and therefore it should be unsuitable for
supporting metallic nanoparticles.[29]
Thus, we envisaged to investigate the effect of chemical
surface modifications on the feasibility of the grafting of
the complex [Rh2Cl2(CO)4] and on the subsequent nano-
particle formation process by reductive treatment under H2.
Bridge-splitting and replacement reactions of rhodium car-
bonyl chloride dimer have been known for many years and
often involve the use of silver carboxylate salts as re-
agents.[30] As it is well-known that oxidation of carbon
nanotubes with nitric acid mainly generates carboxylic
groups on the surface,[17b] we decided to investigate the pos-
sibility of synthesising the corresponding sodium carb-
oxylates and hence of producing a true nanotube salt.
Therefore the surface-mediated organometallic reaction de-
picted in Scheme 1 would be expected to occur on treatment
with the dirhodium precursor, leading to a molecular dis-
persion of Rh carbonyl species and to a high dispersion of
metallic particles after a further decomposition/reduction
step (Scheme 1).
Figure 2. TEM micrograph of HNO3-treated MWNTs
Figure 3. Raman spectra of: (a) pristine MWNTs and (b)
MWNT-COOH
In order to examine the carbon environment on the sur-
face, XPS analyses were performed on the different
samples; the relevant data are given in Table 1.
Table 1. XPS data obtained from MWNT samples
Sample
Binding Energy[a]
(eV)
S[b] O[b] C[b] Na[b]
530.9 532.1 533.2 534.2
MWNT
55
45
44
49
Ϫ
24
Ϫ
Ϫ
9
0.14 0.81 99.05 Ϫ
MWNT-COOH 23
MWNT-COONa 25
Ϫ
3.23 96.77 Ϫ
14
Ϫ
5.29 93.04 1.67
Scheme 1. Surface mediated organometallic reaction between
[Rh2(µ-Cl)2(CO)4] and MWNT-COONa
[a]
[b]
Values given as % of the total amount. Atomic percentage.
From TEM micrographs it appears that the morphology
The ratio of total oxygen to total carbon is a measure of
of MWNT-COOH and MWNT-COONa does not differ the degree of surface oxidation, and reconstruction of the
significantly from that of pristine tubes. Thus, under our O1s peak gives additional information about the surface
experimental conditions the oxidation treatment performed oxygen-containing groups. XPS spectra of all the modified
causes only minor damage to the tips of the nanotubes (Fig- MWNT samples show an increase of the oxygen content.
ure 2). Raman spectra recorded either on pristine tubes or The MWNT-COOH sample exhibits functionalities that
on oxidised samples clearly indicate that oxidation does not contribute to the intensity of the O1s peaks located at
affect the degree of graphitisation of these materials (Fig- 530.9 eV, 532.1 eV, 533.2 eV and 534.2 eV, in agreement
ure 3). In both cases the graphite mode G-band and the with published data for activated carbon treated with nitric
defect mode D-band possess nearly the same intensities. acid.[31] Hence, this type of treatment significantly increases
612
Eur. J. Inorg. Chem. 2003, 610Ϫ617

Rhodium Catalysts Supported on Carbon Nanotubes
FULL PAPER
the amount of surface carboxylic groups. As for MWNT- Rh/MWNTs Catalyst Preparation by [Rh2Cl2(CO)4]
COONa, a new peak appears at 535.4 eV, which can be at- Grafting
tributed to the sodium carboxylate groups[32] (an XPS ana-
The grafting procedure described above involves the reac-
lysis performed on a CH3COONa sample gave a value of
tion between toluene solutions of [Rh2Cl2(CO)4] and vari-
534.9 eV for -COONa). The presence of sodium was con-
ous supports, namely MWNTs, MWNT-COOH, MWNT-
COONa and C*-COONa. It is worth noting that with a
loading of Rh equal to 1% w/w, whatever the carbon nano-
firmed by the Na1s peak at 1071.5 eV.
The formation of sodium carboxylate groups on the
outer surface of MWNTs makes this material highly hydro-
tube support employed, after impregnation and filtration
philic, and it can easily be dispersed in water. Moreover, it
the resulting solution is always colourless, pointing to a
complete adsorption of the rhodium complex by some in-
teraction between the complex and the MWNTs. IR spectra
is interesting to note that a small amount (ca, 5 mg·LϪ1) of
the so formed MWNT sodium salt is dissolved during
washing with water, performed to remove the excess of so-
of these toluene solutions confirmed the absence of any de-
dium carbonate.
tectable carbonyl bands due to rhodium species. Moreover,
infrared and Raman spectra of all the solid samples ob-
tained via the grafting procedure showed no bands attribut-
able to the carbonyl ligands; this could very likely be due
to the low rhodium loading (1Ϫ5% w/w), although the pos-
sibility of a decarbonylation reaction occurring during the
grafting/drying procedure should also be considered. Fur-
thermore, in the case of the reaction between [Rh2Cl2(CO)4]
and MWNT-COONa, the observation by X-ray diffraction
of the presence of NaCl in the samples coming from decom-
position under dihydrogen clearly indicates the occurrence
of the carboxylate group mediated grafting of the complex
through bridge splitting.
After reductive decomposition at 573 K, careful rinsing
with water to remove NaCl, and drying at 383 K under va-
cuum overnight, all of the samples were examined by TEM
(Figure 4 and 5).
In the case of Rh/MWNTs the very large particle size
distribution observed (particle size up to 100 nm has been
measured) points to a weak interaction between the metal
carbonyl precursor and the tube surface, which favours the
migration and agglomeration of the rhodium(0) atoms de-
riving from the decomposition/reduction steps. As for Rh/
MWNT-COOH, a better dispersion is obtained (mean par-
ticle size 2.5Ϫ5 nm), indicating that a reaction has occurred
between the -COOH groups and the rhodium complex.
Figure 4. TEM micrographs of: (a) 1%Rh/MWNTs; (b) and (c)
1%Rh/MWNT-COOH; (d) 1%Rh/MWNT-COONa
Figure 5. TEM micrographs of 9.5%Rh/MWNT-COONa
Eur. J. Inorg. Chem. 2003, 610Ϫ617
613

R. Giordano, P. Serp, P. Kalck, Y. Kihn, J. Schreiber, C. Marhic, J.-L. Duvail
FULL PAPER
However, the results obtained with MWNT-COONa (mean
particle size ഠ1.5Ϫ2.5 nm) clearly show that the grafting to
carboxylate groups is more efficient. Indeed, the reaction of
surface carboxylate groups with the dirhodium precursor,
which gives better results with respect to particle size and
dispersion, is expected to be different from that of the carb-
oxylic groups, owing to the higher nucleophilicity of the
two oxygen atoms of the anionic -COOϪ moiety and to the
symmetric distribution of the charge (the hydrogen atom
having been removed).
The MWNT sodium salt can hence be considered as a
primary ligand that would be prone to react with the dirho-
diumchlorocarbonyl dimeric complex to produce rhodium
species grafted to the surface via splitting of the Cl bridges.
Further decomposition under H2 affords highly dispersed
rhodium-supported particles even with metal loadings up
to 9.5% w/w: a mean particle size of 3 nm has been meas-
ured in the case of 9.5% Rh/MWNT-COONa, with a very
narrow (3 Ϯ 2 nm) particle size distribution (see Figure 5).
Figure 6. TEM micrograph showing graphene layers closing the
inner cavity of a MWNT
Table 2. Catalytic hydrogenation of trans-cinnamaldehyde
Catalyst
Conv (%) Activity[a] Selectivity[b]
Catalytic Experiments
1% Rh/MWNT
Ϫ
Ϫ
Ϫ
In general, we have investigated liquid-phase catalytic re-
actions in order to determine whether the peculiar struc-
tures of MWNTs, and particularly their high mesoporosity,
could improve transfer phenomena with respect to highly
microporous activated carbons. Additionally, we were inter-
ested in the possibility of comparing surface-mediated or-
ganometallic synthesis methods to more classical impregna-
tion techniques involving prior oxidation by nitric acid.
Two reactions were studied: the hydrogenation of trans-cin-
namaldehyde and the hydroformylation of hex-1-ene.
Prior to evaluating the catalytic activity of rhodium-sup-
1% Rh/MWNT-COOH
1% Rh/MWNT-COONa
9.5% Rh/MWNT-COONa 23.2
5.3
15.1
26.8
78.2
11.4
27.1
343.3
100.0
100.0
100.0
100.0
64.5[d]
1% Rh/C*-COONa
4.9
62.8
5% Pd/C*[c]
[a] In g substrate transformed·gϪ1Rh·hϪ1 gϪ1Rh·hϪ1
.
[b] % cinnamal-
[c]
dehyde.
A commercial 5% w/w Pd/C* (Johnson Matthey) was
[d]
used for comparison purposes.
other product of the reaction.
Hydrocinnamyl alcohol was the
ported catalysts, blank experiments were performed with dehyde through CϭC double bond hydrogenation, irre-
MWNTs and MWNT-COOH in order to check the possible spective of the Rh-supported carbon sample employed.
role of the remaining 3% w/w iron nanoparticles, encapsul-
The Rh/MWNTs samples, in accordance with TEM mi-
ated within the nanotubes, in the investigated catalytic sys- crographs in which very large rhodium aggregates can be
tems. For MWNTs no activity was observed either in hydro- observed, possess no catalytic activity. In contrast, Rh/
genation or in hydroformylation; in the case of MWNT- MWNT-COOH and Rh/MWNT-COONa are active and
COOH a 1.7% conversion into hydrocinnamaldehyde was very selective catalysts for hydrocinnamaldehyde formation.
obtained, whilst no hydroformylation product could be de- If one compares the selectivity data, Rh-supported catalysts
tected. Indeed, iron nanoparticles of approx. 5 ϫ 20 nm are more selective than a commercial Pd/C* catalyst, in that
size, located either at the tip, or in inner cavities, of the CϭC double bond hydrogenation is the only reaction oc-
nanotubes, are not accessible to the reactants in the case of curring under our experimental conditions.
MWNTs owing to the presence of graphene layers roughly
As far as catalytic activity and conversion are concerned,
perpendicular to the axis of MWNTs (bamboo-like struc- it appears that the best method to support Rh particles on
ture), as shown on Figure 6. As depicted in Figure 2, the the MWNTs surface is indeed the grafting of the complex
nitric acid treatment induces opening of the tip of the tubes, via surface-mediated organometallic synthesis, followed by
thus making iron particles reachable by the reactants. Since reduction under H2. The good performance of the thus-
iron-based supported catalysts are known to be active in prepared catalysts can hence be attributed to the very small
alkene hydrogenation,[33] it is very likely that the observed particle size obtained after the decomposition/reduction
activity could actually be due to the presence of zero-valent step and to the high purity of the metallic deposit. Higher
iron and/or iron carbide particles.
conversion together with lower activity values for a 9.5%
loading of rhodium on MWNTs could be attributed to the
effect of the larger particle size observed. As for mass-trans-
Selective Hydrogenation of trans-Cinnamaldehyde
The catalytic results of liquid-phase hydrogenation of fer limitations, the complete absence of microporosity, to-
trans-cinnamaldehyde are depicted in Table 2. gether with a relatively high surface area, are crucial to a
One can observe that, except in the case of Rh/MWNTs, good catalytic activity. For instance, if we compare the Rh/
all these catalytic systems selectively afford hydrocinnamal- MWNT-COONa catalyst to Rh/C*-COONa prepared ac-
614
Eur. J. Inorg. Chem. 2003, 610Ϫ617

Rhodium Catalysts Supported on Carbon Nanotubes
FULL PAPER
cording to the same procedure, the former is three times age of using carbon nanotubes as a support in liquid-phase
more active.
catalytic reactions.
Hydroformylation of Hex-1-ene
Liquid-phase hydroformylation runs were carried out Conclusion
with hex-1-ene under relatively mild conditions (2 MPa,
Oxidation of the outer surface of multi-walled carbon
353 K) and afforded n-heptanal and 2-methylhexanal, to-
gether with 2-ethylpentanal. The presence of the last prod-
uct is due to partial isomerization of the substrate to hex-
2-ene, which has been observed in the reaction solutions by
GC. Such a reaction pathway was previously observed un-
der more drastic conditions (10 MPa, 423 K) on Rh/C*
catalysts,[34] leading to C7 aldehyde, hexane and C7 alcohol
formation. The catalytic results are shown in Table 3, and
the selectivities to the products obtained are listed in
Table 4.
It should be noted that the production of linear and
branched aldehydes and a high linear/branched ratio (2.6)
occur only with 1% Rh/MWNT-COOH, a sample charac-
terised by large particle sizes and by the acidic character of
the support. In this case no ethylpentanal and lower
amounts of methylhexanal were detected. The most import-
ant feature of our catalytic system is the high conversion
and activity values found for 1% Rh/MWNT-COONa and
the selectivity of the reaction toward aldehyde formation,
thus confirming, as for trans-cinnamaldehyde hydrogena-
tion, the efficiency of our grafting method and the advant-
nanotubes by nitric acid leads to the formation of carb-
oxylic groups that can be used to anchor rhodium carbonyl
moieties. Better results are obtained upon conversion of
carboxylic into more reactive carboxylate groups. A further
reduction step under dihydrogen affords highly dispersed
(for instance 1.5Ϫ2.5 nm for a 1% w/w loading) rhodium
nanoparticles even for rhodium loadings up to 9.5% w/w.
These Rh-supported catalysts show interesting perform-
ances with respect to liquid-phase hydrogenation and hy-
droformylation reactions. Additionally, the nature of the
MWNT support probably contributes to a better activity,
presumably by ameliorating transfer processes with respect
to what happens on classical activated carbons.
Experimental Section
General Remarks: Nitric acid (Prolabo, 52.5%), Na2CO3·10H2O
(Acros Organics), toluene (Acros Organics, 99%), dioxane (SDS),
trans-cinnamaldehyde (Aldrich, 98%) and hex-1-ene (Janssen, 97%)
were used as received. H2 and CO (Air Liquide, 99.99998 and
99.99%, respectively) were also used as received. The complex
[Rh2Cl2(CO)4] was prepared according to a well established proced-
ure.[36]
Table 3. Catalytic hydroformylation of hex-1-ene
Catalyst
Conv. (%) Activity[a] n/b ratio[b]
Carbon nanotubes were prepared by a catalytic CVD method[27]
on an Fe/Al2O3 catalyst.[35] Dissolution of the alumina support and
some of the iron was performed by acidic treatment (H2SO4, 98%
Prolabo). Typical characterisation data for this material are given
in Table 5.
1% Rh/MWNT
1% Rh/MWNT-COOH
1% Rh/MWNT-COONa
9.5% Rh/MWNT-COONa 65.8
1%Rh/C*-COONa 66.1
Ϫ
Ϫ
Ϫ
52.3
75.8
72.9
197.6
11.6
2.6
1.2
1.2
1.2
105.6
Surface Modification of MWNTs
[a] In g substrate transformed·gϪ1Rh·hϪ1 in gϪ1Rh·hϪ1
isomer; b ϭ branched isomer.
.
[b] n ϭ linear
A mixture of 5 g of MWNTs in 150 mL of HNO3 (52.5%) was
refluxed for 3 hours at 403 K. Then the mixture was filtered and
Table 4. Product distribution for hex-1-ene hydroformylation (as selectivities, %)
Catalyst
2-Ethyl-pentanal
2-Methyl-hexanal
Heptanal
1% Rh/MWNT
Ϫ
Ϫ
12.6
8.1
7.3
Ϫ
Ϫ
1% Rh/MWNT-COOH
1% Rh/MWNT-COONa
9.5% Rh/MWNT-COONa
1%Rh/C*-COONa
27.7
39.8
38.0
37.4
72.3
47.6
53.9
55.3
Table 5. Main characteristics of pristine MWNTs
Internal diameter
(nm)
External diameter
(nm)
Real density
Pore diameter
(nm)
d002
(nm)
SBET
(g·mLϪ1
)
(m2·gϪ1
)
8
17
1.95
20
0.342
180
Eur. J. Inorg. Chem. 2003, 610Ϫ617
615

R. Giordano, P. Serp, P. Kalck, Y. Kihn, J. Schreiber, C. Marhic, J.-L. Duvail
FULL PAPER
rinsed with deionized water until the pH of the water reached 7.
The resulting material was dried overnight at 383 K in a thermo-
statted oven and crushed to a powder by a ball-milling process for
20 min.[37] This material is named MWNT-COOH.
particle-size distributions were determined on the basis of TEM
micrographs. Additional TEM observation were performed on a
Phillips CM12 (120 kV voltage) electron microscope.
Raman spectra were recorded on a T64000 JobinϪYvon spectro-
meter (4 cmϪ1 resolution) equipped with a Spectra Phisycs Argonϩ
Laser; samples were excited at λ ϭ 514 nm. X-ray photoelectron
spectra were obtained on a VG ESCALAB MK II spectrometer.
The measurements were carried out using unmonochromatized
Mg-Kα radiation (λ ϭ 0.989 nm) under a mean pressure of 6 ϫ
10Ϫ11 kPa.
The preparation of the corresponding nanotube sodium carb-
oxylate salt (MWNT-COONa) was accomplished by treatment of
a suspension of 5 g of MWNT-COOH in 100 mL of toluene with
200 mg of Na2CO3·10H2O dissolved in 50 mL of H2O. This suspen-
sion was stirred for 15 min, sonicated for 2 hours at 313 K and
stirred again for 15 min at room temperature. The resulting mixture
was then filtered through a Büchner funnel, and the material
washed with deionized water to fully remove the excess of the so-
dium salt. Drying overnight in a thermostatted oven at 383 K af-
forded MWNT-COONa. Moreover, a C*-COONa sample was pre-
pared for comparison purposes by the same procedure starting
from a commercial activated carbon (Darco-G60, 100 mesh,
SBET ϭ 700 m2·gϪ1, Aldrich).
Acknowledgments
Special thanks to Dr B. Richard for technical assistance and to Mr
J. Y. Mevellec for his help in Raman studies. We would like to thank
also Engelhard-CLAL for a generous loan of RhCl3·3H2O.
[1]
F. Rodriguez-Reinoso, Carbon 1998, 36, 159Ϫ175.
Surface Grafting of [Rh2Cl2(CO)4] and Production of Rh/MWNTs
or Rh/C*-COONa: 100 mg of [Rh2Cl2(CO)4] was added to a sus-
pension of 5.29 g of the carbon support in 100 mL of toluene,
stirred for 3 hours at 333 K under nitrogen and then sonicated for
2 hours at room temperature. Filtration through a Büchner funnel,
washing with n-hexane, drying in vacuo and crushing for 20 min
resulted in the production of the rhodium grafted material. Activa-
tion of the supported complex was done under a continuous flow
(250 sccm) of a N2/H2 gas mixture (80:20 v/v) for three hours at
573 K, then the samples were washed with water to remove the
NaCl formed, dried overnight at 383 K and crushed to powder.
The rhodium loading was so fixed at 1% w/w, if not otherwise
specified in the text.
[2]
E. Auer, A. Freund, J. Pietsch, T. Tacke, Appl. Catal. A 1998,
173, 259Ϫ271.
[3]
B. Coq, J. M. Planeix, V. Brotons, Appl. Catal. A 1998, 173,
175Ϫ183.
J. E. Fischer, A. T. Johnson, Curr. Opin. Solid State Mater. Sci.
1999, 4, 28Ϫ33.
M. Menon, A. N. Andriotis, G. E. Froudakis, Chem. Phys.
[4]
[5]
Lett. 2000, 320, 425Ϫ434.
[6]
K. P. de Jong, Curr. Opin. Solid State Mater. Sci. 1999, 4,
55Ϫ62.
[7]
N. M. Rodriguez, M. Kim, R. T. K. Baker, J. Phys. Chem.
1994, 98, 13108Ϫ13111.
[8] [8a]
F. Salman, C. Park, R. T. K. Baker, Catal. Today 1999, 53,
[8b]
385Ϫ394.
C. Park, R. T. K. Baker, J. Phys. Chem. B 1998,
102, 5168Ϫ5177.
C. Pham-Huu, N. Keller, L. J. Charbonniere, R. Ziessel, M.
J. Ledoux, Chem. Commun. 2000, 1871Ϫ1872.
Huu, N. Keller, G. Erhet, L. J. Charbonniere, R. Ziessel, M. J.
Ledoux, J. Mol. Cat. 2001, 170, 155Ϫ163.
R. Gao, C. D. Tan, R. T. K. Baker, Catal. Today 2001, 65,
Evaluation of Catalyst Activity for trans-Cinnamaldehyde Hydro-
genation and Hex-1-ene Hydroformylation: The selective hydrogena-
tion of cinnamaldehyde was carried out at atmospheric pressure in
a three-necked flask equipped with a reflux condenser and gas in-
let. The mixture containing 100 mg of rhodium-supported catalyst,
2 mL of trans-cinnamaldehyde and 50 mL of dioxane was sonicated
for 15 min at room temperature prior to each catalytic run; nitrogen
gas at room temperature was bubbled through the liquid phase to
purge the system, the temperature was raised to 353 K, then dihy-
drogen gas was let in. Each catalytic run lasted 4 h, during which
time samples were taken at regular intervals for analysis.
[9] [9a]
`
[9b]
C. Pham-
`
[10]
[11]
[12]
19Ϫ29.
P. Chen, X. Wu, J. Lin, K. L. Tan, J. Phys. Chem. 1999, 103,
4559Ϫ4561.
J. M. Planeix, N. Coustel, B. Coq, V. Brotons, P. S. Kumbhar,
R. Dutartre, P. Geneste, P. Bernier, P. M. Ajayan, J. Am. Chem.
Soc. 1994, 116, 7935Ϫ7936.
H. Chen, J. Lin, Y. Cai, X. Wang, J. Yi, J. Wang, G. Wei, Y.
Lin, D. Liao, Appl. Surf. Sci. 2001, 180, 328Ϫ335.
B. Xue, P. Chen, Q. Hong, J. Lin, K. L. Tan, J. Mater. Chem.
2001, 11, 2378Ϫ2381.
[13]
[14]
[15]
[16]
The hydroformylation of hex-1-ene was performed in an autoclave
under the following reaction conditions: 353 K, 2 MPa of CO/H2
(1:1). A catalyst sample (100 mg) was dispersed in a mixture of
40 mL of toluene and 5 mL of hex-1-ene, sonicated in a Schlenk
tube for 15 min and transferred under nitrogen into the autoclave,
which was then pressurized and heated. At the end of the catalytic
runs (24 h) the resulting mixture was cooled to room temperature,
then filtered and analysed.
Y. Zhang, H. Zhang, G. Lin, P. Chen, Y. Yuan, K. R. Tsai,
Appl. Cat. A 1999, 187, 213Ϫ234.
N. M. Rodriguez, J. Mater. Res. 1993, 8, 3233Ϫ3255.
[17] [17a]
S. Nagasawa, M. Yudasaka, K. Hirahara, T. Ichihashi, S.
[17b]
Iijima, Chem. Phys. Lett. 2000, 328, 374Ϫ380.
A. Kuznet-
sova, I. Popova, J. T. Yates, Jr, M. J. Bronikowski, C. B. Huff-
man, J. Lin, R. E. Smalley, H. H. Hwu, J. G. Cheu, J. Am.
Analysis of the reaction products, both from hydrogenation and
hydroformylation runs, was performed on a Carlo Erba HRGC
5160 gas-chromatograph equipped with a capillary column (DB-5
J&W Scientific, 30 m, i.d. O.32 µm, film O.25 µm); a PerkinϪElmer
Q-Mass 910 GC-MS instrument (capillary column DB-1 J&W Sci-
entific, E.I. 70 eV) was used to identify the hydrogenation or hydro-
formylation products and to confirm the nature of the by-products.
[17c]
Chem. Soc. 2001, 123, 10699Ϫ10704.
T. Kyotani, S. Nako-
zaki, W. Xu, A. Tamita, Carbon 2001, 39, 782Ϫ785.
E. van Steen, F. F. Prinsloo, Catal. Today 2002, 71, 327Ϫ334.
V. Lordi, N. Yao, J. Wei, Chem. Mater. 2001, 13, 733Ϫ737.
T. W. Ebbesen, H. Hiura, M. E. Bisher, M. M. J. Treacy,
J. L. Shreeve-Keyer, R. C. Haushalter, Adv. Mater. 1996, 8,
155Ϫ157.
[18]
[19]
[20]
[21]
[22]
[23]
Z. Liu, Z. Yuan, W. Zhou, L. Peng, Z. Xu, Phys. Chem. Chem.
Phys. 2001, 3, 2518Ϫ2521.
S. Hermans, J. Sloan, D. S. Shephard, B. F. J. Johnson, M. L.
H. Green, Chem. Commun. 2002, 276Ϫ277.
E. S. Steigerwalt, G. A. Deluga, C. M. Lukehart, J. Phys.
Chem. 2002, 106, 760Ϫ766.
Characterisation of the Materials: TEM observations were made
with a Hitachi HF2000 (200 kV) microscope equipped with a field
emission gun, a cold cathode and a multiscan camera (Gatan)
which allows the recording and manipulation of high-resolution
images by means of Digital Micrograph software. The rhodium
616
Eur. J. Inorg. Chem. 2003, 610Ϫ617

Rhodium Catalysts Supported on Carbon Nanotubes
FULL PAPER
U. Zielke, K. J. Huttinger, W. P. Hoffman, Carbon 1996, 34,
983Ϫ998.
A. Kazusaka, H. Suzuki, I. Toyoshima, J. Chem. Soc., Chem.
Commun. 1983, 150Ϫ151.
T. A. Kainulainen, M. K. Niemelä, A. O. I. Krause, J. Mol.
Cat. 1997, 122, 39Ϫ49.
P. Serp, R. Feurer, C. Vahlas, P. Kalck, French Patent
01Ϫ13233, 2001.
P. Serp, R. Feurer, R. Morancho, P. Kalck, J. Catal. 1995,
157, 294Ϫ300.
N. Pierard, A. Fonseca, Z. Konya, I. Willems, G. Van Tende-
loo, J. B. Nagy, Chem. Phys. Lett. 2001, 335, 1Ϫ8.
Received June 19, 2002
[24]
[32]
[33]
[34]
[35]
[36]
[37]
B. L. Mojet, M. S. Hoogenraad, A. J. van Dillen, J. W. Geus,
D. C. Koningsberger, J. Chem. Soc., Faraday Trans. 1997, 3,
4371Ϫ4375.
[25]
W. Li, C. Liang, J. Qiu, W. Zhou, H. Han, Z. Wei, G. Sun, Q.
Xin, Carbon 2002, 40, 791Ϫ794.
[26]
S. Banerjee, S. S. Wong, Nano Lett. 2002, 2, 49Ϫ53.
[27]
P. Serp, R. Feurer, C. Vahlas, P. Kalck, French Patent
01Ϫ08511, 2001.
Q. Yang, P. Hou, S. Bai, M. Wang, H. Cheng, Chem. Phys.
[28]
Lett. 2001, 345, 18Ϫ24.
P. Serp, R. Feurer, Y. Kihn, P. Kalck, J. L. Faria, J. L. Figuei-
[29]
redo, J. Phys. IV 2002, 12, 24Ϫ36.
D. N. Lawson, G. Wilkinson, J. Chem. Soc.
1900Ϫ1907.
[30]
A 1965,
[I02335]
[31]
J. L. Figueiredo, M. F. R. Pereira, M. M. A. Freites, J. J. M.
Orfao, Carbon 1996, 37, 1379Ϫ1389.
Eur. J. Inorg. Chem. 2003, 610Ϫ617
617