CNT-supported MoxC catalysts: Effect of loading and carburization parameters

Article abstract of DOI:10.1002/cctc.201300010

Mo_xC/CNT catalysts were prepared through carburization of an oxidic molybdenum precursor impregnated on multiwalled carbon nanotubes (CNTs). The effects of different carburization atmospheres, heating rates, and molybdenum loadings were tested. The catalysts were characterized by using CO temperature-programmed desorption, XRD, N₂ physisorption, SEM, and TEM. The catalytic performance in the steam reforming of methanol was used as a sensitive probe to indicate changes in the catalyst surface during the catalytic action. Contrary to the bulk Mo_xC catalysts, the heating rate during carburization has no effect on the catalysts. Instead, molybdenum loading and carburization atmosphere are the key factors for catalyst structure and performance. The molybdenum-based activity decreases at loadings >10wt% at a constant product selectivity. The CO₂/CH₄ product ratio indicates changes in the catalyst properties at the loadings <20wt%, at which the activity is constant. Carburization in 20% CH₄/H₂ yields 2nm sized crystallites of cubic α-MoC. Carburization in pure H₂ and He yields hexagonal β-Mo₂C with a larger particle size. Both phases show different catalytic performances in terms of activity and CO₂/CH₄ selectivity. Thus, a multiparameter toolbox for fine-tuning of catalyst properties is presented.

Full text of DOI:10.1002/cctc.201300010

CHEMCATCHEM
FULL PAPERS
DOI: 10.1002/cctc.201300010
CNT-Supported Mo C Catalysts: Effect of Loading and
x
Carburization Parameters
Benjamin Frank, Klaus Friedel, Frank Girgsdies, Xing Huang, Robert Schlçgl, and
[a]
Annette Trunschke*
Mo C/CNT catalysts were prepared through carburization of an
carburization atmosphere are the key factors for catalyst struc-
ture and performance. The molybdenum-based activity de-
creases at loadings >10 wt% at a constant product selectivity.
The CO /CH product ratio indicates changes in the catalyst
x
oxidic molybdenum precursor impregnated on multiwalled
carbon nanotubes (CNTs). The effects of different carburization
atmospheres, heating rates, and molybdenum loadings were
tested. The catalysts were characterized by using CO tempera-
2
4
properties at the loadings <20 wt%, at which the activity is
ture-programmed desorption, XRD, N physisorption, SEM, and
constant. Carburization in 20% CH /H yields 2 nm sized crys-
2
4
2
TEM. The catalytic performance in the steam reforming of
methanol was used as a sensitive probe to indicate changes in
the catalyst surface during the catalytic action. Contrary to the
tallites of cubic a-MoC. Carburization in pure H and He yields
2
hexagonal b-Mo C with a larger particle size. Both phases
2
show different catalytic performances in terms of activity and
CO /CH selectivity. Thus, a multiparameter toolbox for fine-
bulk Mo C catalysts, the heating rate during carburization has
x
2
4
no effect on the catalysts. Instead, molybdenum loading and
tuning of catalyst properties is presented.
Introduction
[
6,7]
Group VI transition-metal carbides provide catalytic features
tionalized.
The synthesis of multiwalled carbon nanotube
[
1]
[6]
similar to those of expensive noble metals and thus are per-
manently the focus of intensive research. The high potential in
reactions that involve the transfer of hydrogen, such as alkane
(CNT)-supported b-Mo C has been described by Bao et al.
2
They also highlighted the positive effect of nitric acid treat-
ment of the CNT support before its impregnation with the mo-
lybdenum precursor. The abundance of surface oxygen groups
created here enables a sufficient metal–support interaction
that avoids agglomeration. However, a more systematic study
on the effects of molybdenum loading, carburization condi-
tions, or specific surface properties is still lacking. The variation
in loading in a series of activated carbon and CNT-supported
(
de)hydrogenations or isomerizations, ammonia decomposi-
[2]
tion, and Fischer–Tropsch synthesis, has been documented.
As the conventional metallurgical route to metal carbides
typically provides very low specific surface areas (SSAs), the
first successful synthesis of high-surface area carbides through
[
3]
carburization of an oxidic precursor was a breakthrough for
catalytic applications. However, the synthesis of porous car-
bides is demanding and it seems more appropriate to use
a high-surface area support such as alumina or carbon to stabi-
[10]
Mo C catalysts has been done by Solymosi et al. During eth-
2
anol decomposition at 723 K, the catalysts deactivate within
the initial 7 h time on stream. Unfortunately, neither steady-
state data nor different space velocities are reported to com-
pare the patterns of product selectivities at a similar level of
ethanol conversion. Furthermore, the study overlooks substan-
tial parts of the physicochemical characterization of the sam-
ples, and thus published data are difficult to interpret and do
not support a structure–reactivity correlation with regard to
catalyst optimization. However, a molybdenum loading of up
to 17 wt%, depending on the SSA of the support, has been
[
4–6]
lize highly dispersed carbide (nano)particles.
Carbon-based
support materials can also serve as a carbon source for carburi-
zation. This economic strategy also avoids the formation of
passivating carbon deposits on the catalytically active carbide
surface at the end of the carburization process, which occurs
easily if a gaseous hydrocarbon is used as a carbon source.
[6–9]
Few Mo C/C catalysts have been reported in literature.
x
The carburization in CH /H mixtures typically leads to well-dis-
4
2
persed Mo C nanoparticles if the surface is sufficiently func-
shown to yield finely dispersed b-Mo C nanoparticles. This is in
x
2
good agreement with theoretical considerations: At optimum
[11]
(
maximum) loading, the good wetting ability of MoO3 would
[
a] Dr. B. Frank, K. Friedel, Dr. F. Girgsdies, X. Huang, Prof. Dr. R. Schlçgl,
Dr. A. Trunschke
Department of Inorganic Chemistry
Fritz Haber Institute of the Max Planck Society
Faradayweg 4-6, 14195 Berlin (Germany)
Fax: (+49)3084134405
ideally lead to a homogeneous monolayer coverage of approx-
ꢀ
2 [12]
imately 5.0 atomsMo nm , which is the prerequisite for finely
2
ꢀ1
dispersed carbide particles. Given an SSA of 300 m g of
a carbon support material, this corresponds to a Mo C loading
x
of approximately 20 wt%.
E-mail: trunschke@fhi-berlin.mpg.de
The carburization conditions are discussed predominantly
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cctc.201300010.
[13]
for bulk carbides. A comparison of CH /H and C H /H at-
4
2
2
6
2
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 2296 – 2305 2296

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
mospheres shows that the formation of the carbide phase
[
14]
occurs slower in the presence of CH4, which is the most
stable alkane. However, the higher alkanes tend to decompose
easily and favor coke deposition as an overlayer on the carbide
[
14]
particles. In the same study,
the variation in the heating
rate during carburization reveals that the lower ramping rates
enable the formation of carbides at lower temperatures. Soot
deposit is a potentially highly active form of carbon, and an in-
terfacial reaction incorporating carbon into molybdenum could
explain the kinetic control of carburization. For low-tempera-
ture carburization, molybdenum bronzes are required as inter-
mediates in the topotactic transformation. Such mechanism is
of interest if a carbon-based support material itself, such as
CNTs, is used. Unfortunately, Hanif et al. did not report surface
[
14]
areas in their study ; however, according to the lower con-
centration of the carburization by-product H O, a structural
2
benefit is expected from low heating rates. The phenomenon
of hydrothermal sintering is well known from the reduction
[15]
process of Fe-based catalysts for ammonia synthesis. Both
a lower heating rate and a higher space velocity of the CH /H2
4
flow lower effectively the H O concentration, and at least for
2
bulk systems, a strong structural effect is reported for Mo C
x
[
16,17]
and Mo N systems.
However, adverse effects are also ob-
x
18]
[
served, pointing at the high complexity of involved topotac-
tic transformations, which are not fully understood.
Results and Discussion
Standard parameters and general observations
Figure 1. Profiles of CH and H consumption as well as of CO, CO , H O, and
4
2
2
2
NH
in 20% CH
3
formation during the carburization of 20MoC/oCNT and oCNT samples
The standard MoC/oCNT system serving as the reference
during all the catalyst modifications performed is the medium-
loaded 20MoC/oCNT carburized in CH /H at a heating rate of
ꢀ1
4
/H
2
at 5 Kmin (after drying in CH
4
/H
2
at 1008C for 2 h). The pro-
files in arbitrary units are normalized with respect to the carbon mass of the
sample.
4
2
ꢀ
1
5
Kmin . As described in the Introduction, a high concentra-
tion of well-dispersed molybdenum species is expected here.
Numerous reports confirm the complete carburization of
oxidic molybdenum precursors in this atmosphere at
which is not observed for the oCNT sample. CO and CO are re-
2
leased excessively from the oxidized carbon surface because of
the thermal decomposition of carboxylic acid and anhydride
[
6,14,19]
[20]
7008C,
and thus ramping has been stopped at this
groups in this temperature range. Step 2 is the reduction of
[14]
temperature.
MoO3 to MoO2 at 400–5008C
with both hydrogen and
The gases that evolved during the carburization of 20MoC/
oCNT are shown in Figure 1. Because of the overlap with com-
pounds that are released from the metal-free oxidized carbon
surface, the blank oCNT sample is used as a reference. Accord-
ing to the traces of H and CH consumption as well as those
carbon atoms from the CNT surface. The first reduction is sup-
ported by an enhanced H consumption under H O release;
2
2
the latter reduction is indicated by a sharp peak in both the
CO and CO profiles at 4508C while no CH is consumed. MoO
3
2
4
[21]
is a stochiometric oxidant for elemental carbon. Surprisingly,
instead of CH consumption, a small amount of additional CH
4
2
4
of H O, CO, and CO formation, the following transformations
2
2
4
can be assigned to the features observed: 1) decomposition of
the ammonium heptamolybdate (AHM) precursor into MoO3,
is generated at 400–5008C, which is most likely due to the
high CO concentrations and the simultaneous presence of
x
[22]
2
) reduction of MoO to MoO , 3) partial carburization of MoO
MoO as a moderate CO hydrogenation catalyst. The inter-
2 x
3
2
2
to a nonstochiometric oxycarbide MoO C , and 4) complete car-
mediate formation of an orthorhombic Mo O phase is also
x
y
4
11
13]
[
burization of MoO C to Mo C. These steps are described exten-
suggested to occur in this temperature range. The dissolu-
x
y
x
[
6,14,19]
sively in literature,
in which intermediate phases are iden-
tion of support (CNT) carbon atoms into MoO clusters, which
x
tified by XRD. Step 1 can be located in the temperature range
points at defective carbon sites as the oxycarbide nucleation
[23]
of 250–4008C (Figure 1), in which the formation of NH and
indicates, should also be considered. Step 3 covers the tem-
3
H O is detected. H O also evolves from the blank oCNT surface,
perature range of 500–6508C and is indicated by still heavy
2
2
but to a much lesser extent than from the AHM-loaded
but slowly declining consumption of H under the formation
2
6
+
4+
sample. The (partial) instantaneous reduction of Mo to Mo
of H O. In addition, CO and CO concentrations are still high;
2
2
with H is suggested by a small decrease in the H profile,
however, this is most likely due to the steady defunctionaliza-
2
2
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 2296 – 2305 2297

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
[
20]
tion of the oCNT support. The complete transformation to
the oxycarbide MoO C is indicated by the finished H con-
1) Methanol steam reforming
x
y
2
ꢀ1
CH OH+H O!CO +3H (DH8=49.6 kJmol )
3
2
2
2
sumption and the onset of CH consumption. Finally, step 4 is
4
characterized by the heavy consumption of CH under the for-
2) Methanol reduction
4
mation of CO, CO , and H O, which all decay slowly until the
ꢀ1
2
2
CH OH+H !CH +H O (DH8=ꢀ115.4 kJmol )
3
2
4
2
end of the carburization process after 1 h at 7008C.
Notably, the formation of CO and CO from the blank oCNT
3) Methanol decomposition
CH OH!CO+2H (DH8=90.6 kJmol )
2
sample exceeds temporarily their release from 20MoC/oCNT.
This can be attributed to the strong metal–support interaction
between molybdenum species and the oxygen functionalities
created on the carbon surface to anchor and disperse the
AHM precursor as well as intermediate species. As indicated by
Figure 1, a substantial fraction of oxygen atoms from the
ꢀ1
3
2
4) Reverse water-gas shift
ꢀ1
CO +H !CO+H O (DH8=41.1 kJmol )
2
2
2
5
) CO hydrogenation
x
carbon surface are finally converted to H O.
2
CO +H !alkanes/alkenes+H O
x
2
2
Temperature-programmed desorption (TPD) of CO was per-
formed directly on the freshly carburized catalysts without ex-
position to ambient O and H O. Two main peaks at approxi-
These observations generally agree with previous reports on
[30,31]
supported Mo C catalysts.
However, a detailed comparison
2
2
2
mately 1208C and a shoulder at 2008C are observed (Fig-
ure S1). The comparison with the blank oCNT material confirms
that the CO profile originates exclusively from the interaction
of SRM activity is not possible because of different reaction
conditions used. The metal-free oCNT shows no catalytic activi-
ty under the reaction conditions used.
of CO with the Mo C particles. Yang et al. in their FTIR study of
Three parameters have been selected here to quantify
changes in the catalytic performance of MoC/oCNT catalysts:
the apparent activation energies of CO2 and CH4 formation
and the CO /CH product ratio (Table 1). For 20MoC/oCNT car-
x
CO adsorption on MoN assigned peaks at positions similar to
d+
those of the desorption of CO from Mo (0<d<2) and N
[
24]
sites, respectively. Because of the structurally and catalytical-
ly similar properties between molybdenum carbides and ni-
2
4
ꢀ
1
burized in CH /H at a heating rate of 5 Kmin , the activation
4
2
trides, this assignment is adapted tentatively to Mo C depend-
x
ing on the surface termination by molybdenum or carbon (in-
stead of nitrogen). For b-Mo C(0001), the theory predicts
Table 1. Steam reforming of methanol performance data of MoC/oCNT
catalysts with different loadings carburized at 5 Kmin in CH /H .
4 2
2
ꢀ
1
quite similar stabilities of CO adsorbed on the top positions
[
25]
(
denoted as t1 sites ) on carbon- and molybdenum-terminat-
[a]
Sample
E
a
(CO
2
)
E
a
(CH
4
)
2 4
S(CO )/S(CH )
[
25]
ꢀ
1
ꢀ1
ed surfaces. The total amount of CO desorbed up to 5008C
[kJmol
]
[kJmol ]
ꢀ
1
is 332 mmolg(Mo C) . This value appears rather low; however,
[b]
[b]
[b]
x
oCNT
–
–
–
CO chemisorption on Mo C at ambient temperature measures
5MoC/oCNT
10MoC/oCNT
95ꢁ1
93ꢁ2
91ꢁ1
94ꢁ2
92ꢁ2
111ꢁ1
111ꢁ1
108ꢁ1
108ꢁ2
108ꢁ1
8.4
6.1
4.0
4.1
4.1
x
[26]
only approximately 14% of the total adsorption sites. Thus,
1
2
3
5MoC/oCNT
0MoC/oCNT
0MoC/oCNT
the mean Mo C particle diameter can be estimated as approxi-
x
[
27]
mately 5 nm.
The steam reforming of methanol (SRM) was chosen as
a probe reaction because of a complex selectivity pattern,
which is expected to indicate changes in the catalyst structure.
[
a] 2508C, X(MeOH)=50%; [b] No catalytic activity observed.
[
28]
In the SRM reaction, MeOH reacts to CO or CO, which, in
2
ꢀ
1
the presence of H , reacts further to CH and higher alkanes/al-
energies are 94 and 108 kJmol , respectively, and the product
ratio is 4.1. The CO /CH ratio as a function of MeOH conver-
2
4
kenes through the Fischer–Tropsch synthesis on molybdenum-
2
4
[
29]
based catalysts.
sion (Figure S2b in the Supporting Information) suggests that
both CO and CH are primary products of the reaction; how-
The 20MoC/oCNT catalyst approaches a stable catalytic per-
formance after 2 h time on stream at 2508C and only minor
deactivation is observed in a 24 h test run. The conversion X of
MeOH on this catalyst as a function of contact time is shown
in Figure S2a. The rapid increase in conversion is followed by
a slowdown of the reaction rate after reaching approximately
2
4
ever, secondary methanation of CO also occurs.
2
After SRM tests, the catalyst samples were characterized by
using XRD, N physisorption, and electron microscopy. Because
2
[33]
H O and CO act as mildly oxidizing agents during the SRM,
2
2
no surface passivation by low-concentrated O2 was needed
before exposition to ambient conditions. The XRD analysis con-
firms the pervasive transformation of AHM into face-centered
cubic (fcc) a-MoC (Figure S3a). Although the catalyst has been
used in a catalytic reaction involving potential oxidants such as
H O or CO and even after long-term exposition to ambient
50% conversion. This can be attributed to inhibition by the
main reaction products H and CO as observed on Cu-based
2
2
[
30]
catalysts, although the overall reactant concentration is rela-
tively low. The main carbon-based products of the reaction are
CO and CH , whereas CO, C H , and C H are formed only in
2
4
2
6
2
4
2
2
trace amounts. The following reactions are expected to con-
tribute predominantly to the product pattern observed:
conditions, the XRD patterns show no rise in oxidic bulk
phases. As a reference, the pattern of MoOC is characterized
[34]
by a shift of the fcc pattern to higher angles, which indicates
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 2296 – 2305 2298

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
a lattice contraction due to partial CꢀO substitution. This is not
the case here, in which the detected peaks fall in line with the
[
35]
reference pattern for a-MoC. Peak analysis reveals a crystallite
size of approximately 2 nm. The reason for the formation of
[
35,36]
the metastable
fcc a-MoC instead of hexagonal close
packed (hcp) b-Mo C, which is predominantly reported to form
2
[
3,4,6,13,14,32]
under the synthesis conditions used,
is not fully un-
derstood. Initially, nuclei of sub-carbides or a-MoC can be
formed at low temperatures, at which the thermodynamic
equilibration of the phases is inhibited for diffusion limitation,
and prevail during synthesis. This is evidenced by the pro-
nounced polycrystallinity of the material, as described later.
Han et al. suggested that the Mo C phase can be controlled by
x
the molybdenum loading on an ordered mesoporous carbon
[
8]
support. However, presented XRD diffractograms are difficult
to interpret. It is more likely that the controlled reduction of
the molybdenum precursor is the key factor for a-MoC synthe-
sis. This has been managed by MoO prereduction in a
3
[
37]
H /n-butane mixture or by adding 0.5% Pt to the molybde-
Figure 2. Profiles of H O desorption during the carburization of 20MoC/
oCNT at different heating rates (after drying in CH /H at 1008C for 2 h). The
4 2
m/e=18 traces are normalized for intensity and temperature.
2
2
[
38]
num precursor to facilitate H2 activation.
a-MoC demon-
strated a different performance than b-Mo C in some catalytic
2
[
9,36,38]
reactions
and is prepared classically through the nitrida-
tion of MoO with NH to fcc g-Mo N followed by subsequent
3
3
2
recarburization with CH , which is a time- and resource-de-
A dramatic effect of the heating rate on the carburization
process is revealed by Figure 2. If referred to the temperature,
peaks are sharp and well resolved at the lowest heating rate of
4
[
38]
manding process. In this context, the one-step preparation
of CNT-supported a-MoC is a valuable result.
2
ꢀ1
ꢀ1
The SSA of the catalyst decreases from 290 to 163 m g if
compared to that of the metal-free oCNT (Table 2). This is most
likely due to the plugging of open CNT tips by a-MoC agglom-
erates. The XRD observations are in good agreement with the
a-MoC particle size that is estimated by using CO TPD.
0.2 Kmin . Here, the reaction is completed at approximately
6008C but at the cost of a reaction time of approximately 50 h.
ꢀ
1
Contrarily, at the highest heating rate of 10 Kmin , peaks are
broad and superimposed by each other. The release of H O ap-
2
proaches zero not before 1 h of heat treatment in the CH /H
4
2
atmosphere. Nevertheless, all the features observed at
ꢀ
1
0
.2 Kmin can also be identified here, which indicates carburi-
Table 2. Physicochemical characterization of Mo
x
C/CNT catalysts.
zation via the same intermediate phases in all experiments.
The CO TPD analysis (Figure S1 in the Supporting Informa-
tion) reveals no substantial differences between the 20MoC/
oCNT samples carburized at different heating rates. The high-
temperature shoulder is slightly more intense for the sample
[
a]
Sample
b
Atmosphere
SSA
[m g
CO
ꢀ
1
2
ꢀ1
ꢀ1
[
Kmin
]
]
x
[mmolg(Mo C) ]
oCNT
5
0.2
1
5
10
5
5
5
5
5
CH
CH
CH
CH
CH
CH
CH
CH
4
/H
4
/H
4
/H
4
/H
4
/H
4
/H
4
/H
4
/H
2
2
2
2
2
2
2
2
290
172
169
163
163
245
197
128
170
146
0
332
379
332
346
509
391
229
143
66
2
2
2
2
5
1
3
2
2
0MoC/oCNT
0MoC/oCNT
0MoC/oCNT
0MoC/oCNT
MoC/oCNT
0MoC/oCNT
0MoC/oCNT
0MoC/oCNT
0MoC/oCNT
ꢀ
1
carburized at 1 Kmin . The amount of CO desorbed up to
ꢀ
1
5008C varies in the range of 330–380 mmolg(Mo C) (Table 2)
x
corresponding to a Mo C particle size of 4–5 nm. In addition,
x
the SSAs of the samples are very similar with a very weak tend-
ency to the higher values at the lower heating rates. Partial
[40]
H
2
methanation of the carbon support to create a more defec-
tive surface or improved spreading of the molybdenum precur-
He
[11]
[a] Heating rate during carburization.
sor during the protracted carburization process could be the
reason for this observation.
The similar catalytic performance in the SRM reaction
(
Table S1 in the Supporting Information) suggests only minor
Effect of the heating rate during carburization
differences in the catalyst structure according to the variation
of the heating rate during carburization. The activities for
methanol conversion at 2508C are identical (Figure S2a in the
Supporting Information). The apparent activation energies of
The heating rate during the carburization of the metal carbide
precursor can control the phase transformation temperatures
and the pore structure of the resulting (bulk) carbide. Both ef-
fects are discussed for CNT-supported systems at a nominal
CO and CH4 formation appear unaffected in the ranges of
2
ꢀ
1
Mo C loading of 20 wt%. The heating rates were varied in the
91–95 and 108 kJmol , respectively. Thus, the CO /CH prod-
2 4
x
ꢀ
1
range of 0.2–10 Kmin . The profiles of H O evolution are com-
uct ratio is fixed at approximately 4–5. The catalyst carburized
2
ꢀ
1
pared in Figure 2.
at the lowest heating rate of 0.2 Kmin shows a weakly en-
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 2296 – 2305 2299

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
hanced ratio of 5.1 (Figure S2b), which is in agreement with
a lower apparent activation energy of CO formation.
2
The XRD analysis of the postcatalytic sample reveals the
presence of a-MoC nanocrystallites (ꢂ2 nm). The diffracto-
grams are congruent (Figure S3b) and free of reflexes other
than a-MoC and graphitic CNTs. This finally proves that the
heating rate during carburization in CH /H in the range of
4
2
ꢀ
1
0
.2–10 Kmin has practically no effect on the structure and
catalytic performance of 20MoC/oCNT catalysts. Notably, parti-
cles are polycrystals. These originate from frustrated growth
under the mild conditions used, and this process is most likely
induced by the diffusion of defective carbon atoms from the
[
23]
support material into MoO nuclei. The formation of frustrat-
x
ed high-temperature metastable phases as active particles in
turn could be the cause of activity and stability of the catalysts.
These can clearly survive the redox stress during SRM much
better than, for example, single crystal surfaces or high-tem-
perature phases converting into oxycarbides.
Variation in the metal loading of Mo C/CNT catalysts
x
The metal loading of MoC/oCNT catalysts was varied between
nominal Mo C contents of 5–30 wt% to investigate the limits
x
of surface coverage and agglomeration. AHM-impregnated
samples were carburized in CH /H at a heating rate of
4
2
ꢀ
1
5
Kmin .
The SEM study reveals entangled CNTs without any agglom-
erates, which indicates the dispersion of metal species on the
CNT surface. Even on 30MoC/oCNT, which is the highest load-
ing tested, bulky Mo C particles cannot be visualized by using
x
Figure 3. a) Representative SEM image of 30MoC/oCNTs and b) EDX analysis
SEM (Figure 3a). The energy-dispersive X-ray spectroscopy of
molybdenum and oxygen confirms the nominal metal content
and also confirms the presence of residual oxygen on the sur-
face of all samples (Figure 3b). However, the oxygen content is
similar on all the catalysts and as high as on the blank oCNT
sample, which has been subjected to the carburization treat-
ment and CO TPD, respectively. Thus, the oxygen content can
likely be assigned to the metal-free fractions of the carbon sur-
face. Especially, the CꢀOH, C=O, and CꢀOꢀC groups formed
of MoC/oCNT catalysts. Dashed gray lines represent the parity line (Mo
and the average (O).
x
C)
[
20]
during the initial HNO treatment are very stable and thus
3
can resist high temperatures in a reducing atmosphere.
TEM gives an impression of the catalyst structure (Figure 4a).
The high dispersion of uniformly sized Mo C particles is best
x
observed in the scanning mode at the lowest loading of
5
wt%. A representative high-angle annular dark field image is
Figure 4. Representative a) TEM and b) the corresponding high-angle annu-
lar dark field images of the 5MoC/oCNT catalyst after the catalytic reaction.
Scale bar=20 nm.
shown in Figure 4b. The mean particle size as estimated from
these figures is 1–2 nm. The crystallites agglomerate at higher
loadings (Figure S4 in the Supporting Information). However,
the finely structured agglomerates of small (<5 nm) a-MoC
crystallites are located preferably at the outer CNT surface and
at their tips, which were opened previously by harsh HNO3
treatment. They vary in size but rarely exceed a diameter of
<5008C (Figure 5). If referred to the mass of Mo C, the low
x
loaded 5MoC/oCNT catalyst provides more than twice as much
adsorption sites than the highly loaded 30MoC/oCNT catalyst
10 nm. Notably, the CNT structure is intact after the carburiza-
(Table 2). It indicates that the mean Mo C particle diameter
x
tion process, which indicates that CH rather than the support
within the MoC/oCNT loading series varies in the range of
3–7 nm. The detailed analysis of curve shapes in Figure 5
points at two contributions (as discussed above) for all cata-
lysts. For the loading series, the high-temperature shoulder,
4
serves as the main carbon source for a-MoC formation.
The quantitative analysis of CO TPD reveals a strong effect
of the Mo C loading on the amount of CO desorbed at
x
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 2296 – 2305 2300

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
Figure 5. CO temperature-programmed desorption profiles of differently
loaded MoC/oCNT catalysts.
which is assigned to the CO adsorption on carbon-terminated
surface sites, is more pronounced at higher loadings. The two-
component Gaussian fit proves that the fraction of CO desor-
bed at approximately 2008C increases steadily from 30 to 45%
as the Mo C loading increases from 5 to 30 wt%. With regard
x
to the above assumptions, this could be interpreted as an in-
creased fraction of carbon-terminated Mo C surfaces with the
x
increasing Mo C loading.
x
Regarding SRM catalysis two aspects have to be discussed.
First, the overall activity in terms of MeOH conversion (Fig-
ure 6a) is similarly high for low loaded samples (ꢃ10 wt%
Mo C) if the molybdenum content is considered. The higher
x
metal loadings decrease the effective reaction rate, which
could be due to mass transport limitations. In the high-loaded
Figure 6. a) Conversion of MeOH in the steam reforming of methanol on
MoC/oCNT catalysts as a function of the molybdenum-based contact time.
2 4
b) CO /CH product ratio as a function of MeOH conversion. Reaction condi-
Mo C catalysts, particles are too rich in active sites or larger
x
ꢀ1
2
tions: 167–1000 mg catalyst; 10–100 mLmin of 1 vol% MeOH/1 vol% H O/
Mo C particles have a lower specific (active) surface area. The
x
He; 2508C.
results obtained from CO TPD analysis suggest the latter, al-
though even at low loadings the amount of CO adsorbed dif-
fers remarkably. The second aspect is the CO /CH4 product
crease in temperature in this range, and the effect of conver-
sion on CO /CH is rather low.
2
ratio (Figure 6b). As listed in Table 1, the catalyst with the
2
4
lowest Mo C loading of 5 wt% produces only half of the
The SSAs of the differently loaded catalysts decrease with an
increase in molybdenum loading and finally approach
x
amount of CH that is produced on the highly loaded catalysts.
4
2
ꢀ1
The catalysts obtain similar selectivities at Mo C loadings of
128 m g (Table 2). This result is well explained by the TEM
observation of plugging of open CNT tips with MoC. At higher
loadings, the mass fraction of MoC providing a lower SSA than
that of pure CNTs becomes a significant factor.
x
ꢄ20 wt%. The opposite trend has been reported for SRM on
[
31]
Mo C supported on the active carbon. Here, the CO /CH
4
2
2
product ratios of 2.5 and 4.1 were measured at 50% MeOH
conversion on 5 and 10 wt% loaded Mo C/C catalysts, respec-
The X-ray diffractograms for the loading series are depicted
in Figure 7. The metal-free oCNT identifies the graphitic reflec-
tions originating from the support material. With an increase in
molybdenum loading, the pattern is supplemented by adding
the reflections of fcc a-MoC, which are clearly visible at a MoxC
loading of ꢄ10 wt%. A superposition appears at approximate-
ly 42–438 and 77–788 with the MoC(200)/C(100) and
MoC(400)/C(110) reflections, respectively. For all the samples
2
tively. Contrary to the present study, high conversions have
been achieved with an increase in temperature rather than
with variation in gas hourly space velocity. Thus, higher CH se-
4
lectivities can also be due to the reaction temperature. For
20MoC/oCNT, this assumption is confirmed exemplarily by
plotting the CO /CH ratio as a function of MeOH conversion
2
4
during the stepwise temperature decrease from 250 to 2008C
for the determination of apparent activation energies (Fig-
ure 2b). The CO /CH ratio increases from 4.6 to 6.6 with a de-
with Mo C loadings of ꢄ10 wt%, the pattern analyses reveal
x
an a-MoC crystallite diameter of approximately 2 nm.
2
4
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 2296 – 2305 2301

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
x
Figure 7. X-ray diffractograms of a) oCNT and b)–e) Mo C-loaded oCNT cata-
lysts after the catalytic reaction (top down: oCNT, 5MoC/oCNT, 10MoC/oCNT,
[
34,40]
20MoC/oCNT, and 30MoC/oCNT). f) Reference pattern of a-MoC.
Effect of the carburization atmosphere
The carburization atmosphere can have a significant effect on
the Mo C phase formed during carburization. Typically, in
x
[
6]
Figure 8. Profiles of H
NH formation during the carburization of 20MoC/oCNT in H
5 Kmin
2
consumption as well as of CH
4
, CO, CO
2
, H
and He at
2
O, and
CH /H2 the hcp b-Mo C is formed. However, Han et al.
4
2
3
2
showed that the use of pure H at 7008C leads to the forma-
ꢀ1
2
.
tion of both a-MoC and b-Mo C depending on the molybde-
2
num concentration on the ordered mesoporous carbon sup-
[
8]
port, which serves as a carbon source. They also indicated
that the carburization in inert N at 7008C stops at the partially
2
reduced MoO . Similar carburization conditions are used for
2
20MoC/oCNT catalysts.
The carburization profiles monitored in H and inert He at-
2
mospheres are shown in Figure 8. Similar to the profiles in
CH /H (Figure 1), carburization starts with the decomposition
4
2
of the AHM precursor under the release of NH and H O at
3
2
2
50–3508C. However, for reasons unknown, less H O is de-
2
tected in H . Here, a subsequent decrease in the H profile is
2
2
observed, which is accompanied by steadily increasing H O for-
2
mation and indicates the reduction of MoO to MoO with H .
3
2
2
Carbon oxides are formed only to a relatively low extent. At
008C, the formation of CH is detected, which can originate
7
4
from the methanation of released CO and of the carbon sup-
x
port. Instead, if He is used, the formation of CH is not ob-
4
Figure 9. CO temperature-programmed desorption profiles of 20MoC/oCNT
catalysts carburized in different atmospheres at 5 Kmin .
ꢀ
1
served and CO and CO are the dominating gases produced,
2
6
+
which indicates that the reduction of Mo occurs through
gasification of the carbon support.
makes low-temperature desorption invisible in the setup used.
Thus, the curves shown in Figure 9 present only the high-tem-
perature shoulder of the complete profile. Only 143 and
The CO TPD experiments after carburization in H and He
2
show much less adsorption sites on the catalysts (Table 2 and
Figure 9). This, however, originates from the mode of opera-
tion, that is, TPD starting at ambient temperature. The profile
is shifted clearly to a lower temperature upon alloying, which
ꢀ1
66 mmolgMo can be detected, respectively, as compared to
ꢀ
1
332 mmolgMo after carburization in CH /H . Two peaks are ob-
4
2
served again; however, the low-temperature contribution pro-
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 2296 – 2305 2302

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
vides a much higher fraction of the overall profile. With regard
to the above assumptions, it can be concluded that only 25
and 20% of the CO adsorption sites represent a carbon termi-
nation of the Mo C surface. This is plausible if we consider the
x
hydrocarbon-free carburization atmosphere.
The catalytic performances also depend strongly on the car-
burization atmosphere. The catalyst carburized in pure H is
2
slightly more active than those carburized in CH /H . Instead,
4
2
the carburization in inert He yields a poorly active catalyst. The
microstructural changes are also reflected in the CO /CH prod-
2
4
uct ratios. The catalytic data are assembled in Table 3 and in
Table 3. Steam reforming of methanol performance data of 20MoC/oCNT
ꢀ
1
catalysts carburized at 5 Kmin in different atmospheres.
[
a]
Atmosphere (CO (CH
E
a
2
)
E
a
4
)
2 4
S(CO )/S(CH )
ꢀ1
ꢀ1
[
kJmol
]
[kJmol ]
2
H
He
0% CH
4
/H
2
94ꢁ2
87ꢁ1
88ꢁ1
108ꢁ2
105ꢁ1
128ꢁ1
4.1
6.2
9.2
2
Figure 10. X-ray diffractograms of 20Mo/oCNT catalysts carburized in a) H
and b) He after the catalytic reaction. c) Reference pattern of b-Mo C.
2
2
[40,41]
[a] 2508C, X(MeOH)=50%.
ences in phase composition of the catalysts, the different cata-
lytic performances (Table 3 and Figure S5 in the Supporting In-
formation) are not surprising. The results of the loading series
and the variation in the carburization atmosphere suggest that
the formation of CO and CH could be related to the Mo C
Figure S5. The apparent activation energy of CO formation de-
2
ꢀ
1
creases to 87 and 88 kJmol after carburization in H and He,
2
respectively, whereas the apparent activation energy of CH
4
2
4
x
ꢀ
1
formation changes to 105 and 128 kJmol , respectively. Con-
sequently, for both alternative carburization atmospheres, the
surface terminations by molybdenum and carbon, respectively.
CO /CH4 product ratio increases remarkably and reaches
2
a value as high as 9.2 after carburization in He.
Summary
The reason for changes in the catalytic behavior is poorly ex-
plained by the textural properties: catalysts carburized in H
The preparation, characterization, and catalytic testing of CNT-
2
2
ꢀ1
and He provide SSAs of 170 and 146 m g , which are quite
supported Mo C catalysts are presented. The nature of the
x
2
ꢀ1
close to the reference value of 163 m g obtained through
carburization in CH /H (Table 2). In addition to CO TPD, the
Mo C crystal phase is a sensitive function of the highly defec-
x
tive state of the CNT support structure, which kinetically con-
trols the formation of favorable crystallization nuclei, leading
to a stabilized polycrystalline catalyst. The pervasive carburiza-
tion can be controlled by the carburization atmosphere. The
4
2
most drastic change is observed in the XRD patterns of the
catalysts (Figure 10). Both the H - and He-carburized samples
2
show the typical bands of hcp b-Mo C in addition to the pat-
2
tern of graphitic CNTs. The mean crystallite size of hcp b-Mo2C
carburization at 7008C in CH /H2 yields a-MoC, whereas in
4
is 8 and 10 nm after carburization in H and He, respectively,
pure H a mixture of a-MoC and b-Mo C is formed. In inert He,
2
2
2
which confirms the larger particle sizes estimated by using CO
only the b-Mo C phase is formed; however, temperatures
2
TPD. Moreover, the H carburized sample shows a band at ap-
higher than 7008C are required according to the synthesis
route used here. The heating rate during carburization was
found to be less important for the catalyst structure and reac-
2
proximately 388, which could tentatively be interpreted as
a-MoC or oxycarbide with a crystallite diameter of approxi-
mately 3 nm. If we consider the gases evolved during carburi-
zation, the formation of a-MoC on this sample could be relat-
tivity. The Mo C particle size as estimated by using CO TPD in-
x
creases steadily with the Mo C loading from 5 to 30 wt%. In-
x
ed to the presence of CH at the end of the carburization pro-
stead, the crystallite size as estimated by using XRD remains
4
cess. Instead, on the He-carburized sample, a considerable
constant at approximately 2 nm for a-MoC and <10 nm for
amount of approximately 20% MoO is detected (Figure 10).
b-Mo C (notably volume-average). The CO TPD analysis reveals
2
2
Metallic molybdenum is not identified on any of the XRD
patterns.
two contributions, which are assigned tentatively to molybde-
num- and carbon-terminated surfaces. According to this, the
fraction of carbon termination increases with an increase in
Mo C loading but decreases sharply if a CH -free carburization
The (partial) carburization of AHM in inert He at 7008C is
a surprising result, as the temperatures required for such solid-
x
4
[
8,9]
state reactions are typically much higher.
Taking the large
atmosphere is used.
crystallite size of remaining MoO into account, the problem
The materials synthesized are highly active and stable cata-
lysts in the steam reforming of methanol (SRM) reaction. Here,
2
for incomplete reduction could be the molybdenum dispersion
at elevated temperature. However, with regard to the differ-
the activity and product selectivity depend on the Mo C load-
x
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 2296 – 2305 2303

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
ing and crystal phase. The surface termination by molybdenum
or carbon probably also plays an important role.
Catalytic testing
Catalysts were tested for their catalytic performance in the SRM at
The experiments performed provide a versatile toolbox for
the synthesis of differently performing catalysts, not only for
SRM. It is expected that trends and dependencies presented
here to control the catalyst properties also provide important
parameters to modify the activity and selectivity of CNT-sup-
ꢀ1
2
508C in a stream of 100 mLmin of a CH OH/H O/He mixture
3
2
(2 vol%). After 2 h on stream, the flow rate was varied between
ꢀ1
100 and 10 mLmin , followed by a decrease in temperature in
10 K steps to 2008C at 100 mLmin . Reaction products were
quantified by using GC analysis. Contact of the catalyst with H O
ꢀ1
2
(reactant) and CO (product) ensured the mild passivation of the
ported Mo C catalysts in other reactions, for example, the
2
x
Mo C surface, which is required before the final exposure of the
x
Fischer–Tropsch alcohol synthesis. Thus, the SRM reaction was
used as a probe reaction rather than as the economically/in-
dustrially interesting pathway for H2 generation from liquid
fuels.
catalyst to ambient conditions for its characterization.
Acknowledgements
We thank W. Frandsen, F. Rybicki, G. Weinberg, and Dr. Z.-L. Xie
for experimental assistance. Financial support by the Federal
Ministry of Education and Research (BMBF) within the CarboKat
project (FKZ 03X0204C) of the Inno.CNT alliance is gratefully
acknowledged.
Experimental Section
Synthesis
Commercial multiwalled CNTs (Baytubes C 150 HP) were pretreated
by refluxing in 65% HNO (500 mL per 10 g) for 2 h. The product
3
Keywords: CO TPD · crystal phase · molybdenum carbide ·
was washed with deionized water until neutral pH and dried in air
at 1108C for 1 day (oCNT). Aliquots of oCNT (1 g) were impreg-
nated with differently concentrated aqueous solutions of
nanoparticles
termination
· steam reforming of methanol · surface
(
NH ) Mo O ·4H O (3 mL) to achieve final Mo C loadings between
4 6 7 24 2 x
5
and 30 wt%. Thus, samples were labeled as yMoC/oCNT, in
[
[
1] R. B. Levy, M. Boudart, Science 1973, 181, 547–549.
[43]
which y represented the nominal loading of Mo C in wt%. The
x
2] S. T. Oyama in Handbook of Heterogeneous Catalysis (Eds.: G. Ertl, H.
Knçzinger, F. Schꢁth, J. Weitkamp), Wiley-VCH, Weinheim, 2008,
pp. 342–356.
resulting pastes were kneaded thoroughly in a mortar, followed by
drying in air at 1108C for 1 day. In a standard carburization proce-
dure, a catalyst mass nominally containing 0.5 mmol molybdenum,
for example, 250 mg 20MoC/oCNT, was placed in a quartz tubular
reactor (7 mm inner diameter) in a stream of 37 mLmin of a
CH /H mixture (20 vol% CH ). After elution of H O and O (1 h at
[3] J. S. Lee, S. T. Oyama, M. Boudart, J. Catal. 1987, 106, 125–133.
[4] A. J. Brungs, A. P. E. York, J. B. Claridge, C. Mꢂrquez-Alvarez, M. L. H.
Green, Catal. Lett. 2000, 70, 117–122.
ꢀ1
[
[
5] R. Barthos, A. Szꢃchenyi, F. Solymosi, Catal. Lett. 2008, 120, 161–165.
6] X. Li, D. Ma, L. Chen, X. Bao, Catal. Lett. 2007, 116, 63–69.
4
2
4
2
2
1
5
008C), the reactor temperature was increased linearly by
1
ꢀ
[7] C. Sayag, M. Benkhaleda, S. Suppanb, J. Trawczynskib, G. Djꢃga-Maria-
Kmin up to 7008C and kept for 2 h, followed by cooling to am-
dassoua, Appl. Catal. A 2004, 275, 15–24.
bient temperature in the reducing atmosphere.
[
[
8] J. Han, J. Duan, P. Chen, H. Lou, X. Zheng, H. Hong, ChemSusChem
012, 5, 727–733.
9] H. Wang, A. Wang, X. Wang, T. Zhang, Chem. Commun. 2008, 2565–
567.
2
2
[
[
10] R. Barthos, A. Szꢃchenyi, ꢄ. Koꢅs, F. Solymosi, Appl. Catal. A 2007, 327,
95–105.
11] J. Haber, T. Machej, R. Grabowski, Solid State Ionics 1989, 32–33, Part 2,
Characterization
The TPD of CO was performed immediately after carburization
without passivation or air contact of the catalyst. The reactor was
flushed with 5 vol% CO/Ar for 10 min at ambient temperature and
subsequently flushed with He until no CO was detectable with
Pfeiffer GAM 200 MS and Varian CP-4900 Micro-GC. In a He stream
887–892.
[12] Y.-C. Xie, Y.-Q. Tang, Adv. Catal. 1990, 37, 1–43.
[13] T. Cotter, PhD thesis, Technische Universitꢆt Berlin (Germany), 2011.
[14] A. Hanif, T. Xiao, A. P. E. York, J. Sloan, M. L. H. Green, Chem. Mater. 2002,
1
4, 1009–1015.
ꢀ
1
[15] R. Schlçgl in Handbook of Heterogeneous Catalysis (Eds.: G. Ertl, H. Knç-
zinger, F. Schꢁth, J. Weitkamp), Wiley-VCH, Weiheim, 2008, pp. 2501–
of 30 mLmin , the temperature was increased linearly by
ꢀ
1
1
0 Kmin up to 7008C. Further characterization of the catalysts
2575.
was performed after catalytic testing. Passivated samples were ana-
lyzed after exposition to ambient conditions. TEM was performed
on a Cs-corrected FEI Titan 80-300 microscope (300 kV), and SEM
images were obtained with a Hitachi S-4800 FEG microscope
[
[
[
16] E. J. Markel, J. W. Van Zee, J. Catal. 1990, 126, 643–657.
17] C. H. Jaggers, J. N. Michaels, A. M. Stacy, Chem. Mater. 1990, 2, 150–157.
18] J. G. Choi, J. R. Brenner, L. T. Thompson, J. Catal. 1995, 154, 33–40.
[19] S. T. Oyama, J. C. Schlatter, J. E. Metcalfe, J. M. Lambert, Ind. Eng. Chem.
Res. 1988, 27, 1639–1648.
(
1.5 kV) equipped with an EDAX Genesis EDX detector (15 kV). The
XRD measurements were performed on a Bruker AXS D8 Advance
diffractometer equipped with a secondary graphite monochroma-
tor (CuKa1+2 radiation) and a scintillation detector. Crystallite size
values were extracted by full pattern fitting and are reported as
LVol-IB values (volume-weighted mean column length based on in-
tegral breadth) without further assumptions concerning crystallite
[20] J. L. Figueiredo, M. F. R. Pereira, Catal. Today 2010, 150, 2–7.
[
[
[
21] D. W. McKee, Carbon 1970, 8, 623–635.
22] J.-W. Dun, E. Gulari, B. Streusand, Appl. Catal. 1986, 21, 61–72.
23] A. Rinaldi, J.-P. Tessonnier, M. E. Schuster, R. Blume, F. Girgsdies, Q.
Zhang, T. Jacob, S. B. Abd Hamid, D. S. Su, R. Schlçgl, Angew. Chem.
2
011, 123, 3371–3375; Angew. Chem. Int. Ed. 2011, 50, 3313–3317.
[
24] S. Yang, C. Li, J. Xu, Q. Xin, J. Phys. Chem. B 1998, 102, 6986–6993.
shape or size distribution. N physisorption was performed at 77 K
2
[25] X.-R. Shi, J. Wang, K. Hermann, J. Phys. Chem. C 2010, 114, 13630–
13641.
and P/P =0.05–0.3 after drying the sample in vacuum at 2008C for
0
2
h.
[26] S. Ramanathan, S. T. Oyama, J. Phys. Chem. 1995, 99, 16365–16372.
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 2296 – 2305 2304

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
ꢀ
3
[
27] Assumptions: spherical geometry; Mo
x
C density 8.2 gcm ; CO site
[36] H. O. Pierson in Handbook of Refractory Carbides & Nitrides, Noyes Publi-
cations, New Jersey, 1997, pp. 110–112.
ꢀ
2
density 1015 cm
.
[
28] ꢄ. Mastalir, ꢄ. Patzkꢅ, B. Frank, R. Schomꢆcker, T. Ressler, R. Schlçgl,
Catal. Commun. 2007, 8, 1684–1690.
[37] C. Bouchy, I. Schmidt, J. Anderson, C. J. Jacobsen, E. Derouane, S. Der-
ouane-Abd Hamid, J. Mol. Catal. A: Chem. 2000, 163, 283–296.
[38] J. S. Lee, L. Volpe, F. H. Ribeiro, M. Boudart, J. Catal. 1988, 112, 44–53.
[39] G. S. Ranhotra, A. T. Bell, J. A. Reimer, J. Catal. 1987, 108, 40–49.
[40] J. Liu, J. Shen, X. Gao, L. Lin, J. Therm. Anal. 1993, 40, 1245–1252.
[41] Inorganic Crystal Structure Database, http://icsd.fkf.mpg.de/.
[
29] M. Saito, R. B. Anderson, J. Catal. 1980, 63, 438–446.
30] B. Frank, F. C. Jentoft, H. Soerijanto, J. Krçhnert, R. Schlçgl, R. Schomꢆck-
er, J. Catal. 2007, 246, 177–192.
[
[
31] R. Barthos, F. Solymosi, J. Catal. 2007, 249, 289–299.
32] S. S.-Y. Lin, W. J. Thomson, T. J. Hagensen, S. Y. Ha, Appl. Catal. A 2007,
[
[
42] E. Parthꢃ, V. Sadogopan, Acta Crystallogr. 1963, 16, 202–205.
43] Fritz Haber Institute internal sample numbers in parentheses: 05MoC/
oCNT (13894); 10MoC/oCNT (13896); 20MoC/oCNT (13898 and 13899);
[
3
18, 121–127.
33] W. Wu, Z. Wu, C. Liang, P. Ying, Z. Feng, C. Li, Phys. Chem. Chem. Phys.
004, 6, 5603.
34] I. F. Ferguson, J. B. Ainscough, D. Morse, A. W. Miller, Nature 1964, 202,
327–1328.
35] O. Matsumoto, Y. Yaguchi, Y. Shiota, Y. Kanzaki, High Temp. Sci. 1983, 16,
43–250.
[
[
[
3
2
0MoC/oCNT (13901); 05MoC/oCNT (13894); 10MoC/oCNT (13896);
0MoC/oCNT (13898 and 13899); 30MoC/oCNT (13901).
2
1
Received: January 8, 2013
Published online on April 15, 2013
2
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 2296 – 2305 2305