Catalysis Today 250 (2015) 123133

Contents lists available at ScienceDirect

Catalysis Today
journal homepage: www.elsevier.com/locate/cattod

Synthesis of nanocrystalline molybdenum carbide materials

and their characterization
Gerardo Vitale a,b, , Hctor Guzmn a , Maria L. Frauwallner b , Carlos E. Scott b ,
Pedro Pereira-Almao b
a
Chemistry Department, University of Calgary, Calgary, Alta, Canada T2N 1N4
b
Sculich School of Engineering, University of Calgary, Calgary, Alta, Canada T2N 1N4

a r t i c l e i n f o a b s t r a c t

Article history: A convenient and practical method for the preparation of precursors to form molybdenum carbide
Received 22 January 2014 materials using solutions of ammonium heptamolybdate and sucrose is presented. Depending on the
Received in revised form 6 May 2014 temperature-programmed reaction protocol employed [i.e., low temperature and long times (523 K for
Accepted 10 May 2014
24 h, 598 K for 24 h and 673 K for 72 h) or high temperature and short times (1073 K for 30 min)], the
Available online 28 June 2014
C/Mo ratio of the precursors (i.e., 6.3, 3.6, 2.0, 1.0 and 0.6) and the presence or not of H2 , it is possible
to produce nanocrystalline molybdenum carbide materials with the cubic or hexagonal phases. The two
Keywords:
prepared phases of molybdenum carbide showed activity for the hydrogenation of toluene at 423 K to
Molybdenum carbides
-MoC(1 x)
selectively produce methyl-cyclohexane. In the tested conditions, the hexagonal molybdenum carbide
-Mo2 C phase showed an initial higher activity but the deactivation of the catalyst was more pronounced than
Toluene hydrogenation in the cubic molybdenum carbide catalyst.
TPR Crown Copyright 2014 Published by Elsevier B.V. All rights reserved.

1. Introduction Transition metal carbides are another important group of very

As it has been pointed out by several authors [14], the sustain-
ability and feasibility of many present and future technologies will
depend on the ability to develop novel materials that can be pro-
duced with the most abundant chemical elements on earth. The
replacement of the current scarce and/or rare elements by more
abundant ones to design and produce materials that can fulll the
needs of the developed or developing technologies is a must. Noble
metals like palladium or platinum are well known for their catalytic
properties in many types of reactions. By using the current fuel-
cell technology, an impractically large amount of platinum would
be needed per vehicle, thus, even if using all the mined platinum
for fuel-cell vehicles alone, it only would support the production
of a couple of millions of vehicles per year which is around 5% of
the global worldwide production of automobiles [3]. Attempts to
develop Pt-like materials based on nitrogen-doped graphene have
been carried out for their use in fuel cells [58], however, their
practical commercial application is still far away.
Corresponding author at: Schulich School of Engineering, University of Calgary,
2500 University Drive NW, CCIT Room 008, Calgary, Alta, Canada T2N 1N4. Tel.: +1
403 210 9806; fax: +1 403 210 3973.
E-mail addresses: vitaleg@ucalgary.ca, vitalerojas@gmail.com (G. Vitale).
interesting materials which have been studied for several years.
Levy and Boudart [9] demonstrated in their work, published in
1973, that tungsten carbides behaved similar to platinum for some
types of reactions, and thus, a new era of research in these types
of materials began. Transition metal carbides in bulk or supported
form, as well as promoted or not, have been tested and shown to
be active for several types of reactions that are usually catalyzed
by noble metals [940]. This has led to think those materials as
promising substitutes for the noble metals; however, the possible
commercial applications of these materials as catalysts have been
restricted by the lack of a complete control and understanding of
the relationship between the structure and its physical chemical
properties, as well as a practical and cost-effective way of com-
mercially producing them.
The typical way of synthesizing transition metal carbides is
known as temperature-programmed reaction (TPR). Under the TPR
route, a given amount of the oxide precursor (for instance, MoO3
or WO3 ) is heated while it is exposed to a mixture of hydrogen and
hydrocarbon. During this thermal treatment, the oxide is progres-
sively reduced to lower-valence oxide states further undergoing
carburization to yield the desired carbide. Typical preparations of
molybdenum carbides under the TPR route require temperature
step ramps to obtain the nal product at high temperature, which
is normally the stable and well-known hexagonal molybdenum
carbide phase.

http://dx.doi.org/10.1016/j.cattod.2014.05.011
0920-5861/Crown Copyright 2014 Published by Elsevier B.V. All rights reserved.
124 G. Vitale et al. / Catalysis Today 250 (2015) 123133
The depletion of light crude oil sources and the increase in oil
prices have made commercially viable the extraction of heavy oils
and bitumen sources. However, these types of feedstocks are of
lower quality with large quantities of sulfur, nitrogen and metal
component, and thus, requiring the development of novel catalysts
and processes that can deal with them. Molybdenum carbide seems
to be a promising candidate for developing commercial catalysts
to replace the more expensive and, easier to poison, conventional
noble metal catalysts used in the upgrading of crude oils.
Aiming to develop a most practical and cost-effective way of
producing molybdenum carbide materials, we tested a combina-
tion of our previously developed TPR ramp route [14,15] with a
modication of the carbide precursor, prepared by the solution
route with ammonium heptamolybdate and sucrose [41], to pro-
duce low-temperature molybdenum carbide materials [16]. It was
demonstrated in this work that it is possible to obtain nanocrys-
talline cubic molybdenum carbide materials at 673 K under the
developed synthesis conditions and these solids showed hydro-
genation activity in a model molecule like toluene. Also, our
synchrotron X-ray diffraction (XRD) work carried out in situ with
different precursors indicated that MoO2 seems to be an important
intermediate for the production of molybdenum carbides at high
temperature [42].
The goal of this study is to go deeper into the understanding
of the preparation of molybdenum carbides using the developed
synthesis protocol mentioned above to produce a most viable cost-
efcient catalyst. This work presents the methods of preparation,
the characterization of these materials and their activity for the
hydrogenation of toluene.
2. Experimental methods
2.1. Preparation of Mo2 C precursors
Ammonium heptamolybdate tetrahydrate
[(NH4 )6 Mo7 O24 4H2 O] (AHM), analytical reagent from BDH,
and commercially available household sucrose C12 H22 O11 were
used as starting reagents without further purication or mod-
ication. The methodology of preparation is described in our
previous work [16]. Briey, a sucrose solution prepared with
deionized water was added to the AHM solution, also prepared
with deionized water, under magnetic stirring and homogenized
for 15 min at room temperature, before the solution mixture was
placed in an oven at 393 K for 24 h until it was completely dried.
Five precursors with C/Mo molar ratios of 6.3, 3.6, 2.0, 1.0 and
0.6 were prepared this way and identied as P1, P2, P3, P4 and
P5, respectively. The dried cakes were then crushed into an agate
mortar to obtain ne powders of the precursors for further use.
2.2. Carburization of the precursors with the protocol 523 K
(24 h), 598 K (24 h) and 673 K (72 h)
The production of the cubic molybdenum carbide at 673 K (low
temperature and long-time protocol) requires a ramp-up protocol
which was developed and explained in our previous works [1416]
and described as follows: approximately 1 g of each pulverized pre-
cursor was thermally treated in situ in a steel xed-bed reactor to
form Mo2 C under a H2 ow of 50 scm3 min1 at atmospheric pres-
sure and with the following conditions of low-temperature and
long-time protocol: rst temperature ramp from room tempera-
ture up to 523 K and held there for 24 h, second ramp from 523
to 598 K and held for 24 h and third ramp from 598 to 673 K and
held there for 72 h. The heating rate to reach each temperature in
the protocol was 10 K min1 . After the treatments, the solids were
allowed to cool to room temperature under nitrogen at atmospheric
pressure and with a N2 ow of 180 scm3 min1 . The solids produced
in this way are covered with a thin carbon lm that protects them
from pyrophoric oxidation when exposed to the air, and thus, no
passivation protocol with diluted oxygen is needed. The % yield
was calculated as Y = PR 100/IP, where PR is the amount of prod-
uct recovered after carburization and IP is the amount of the initial
used precursor.
In order to get insights into the importance of the employed
ramp protocol to produce the desired carbides, two additional stud-
ies were conducted. In the rst one, the ramp protocol was carried
out without reaching the last temperature of 673 K, i.e., each 24 h,
ramps at 523 and 598 K were applied and the produced solid was
cool down to room temperature under nitrogen. In the second
study, a direct heating at 10 K min1 from room temperature to
673 K without passing for the 523 and 598 K temperature ramps
was carried out, maintaining this temperature for 72 h. The solid
was cooled down to room temperature under nitrogen atmosphere.
2.3. Carburization of the precursors with the protocol 1073 K
(30 min)
Approximately 1.5 g of each pulverized precursor was thermally
treated in situ in a quartz U-type reactor to form Mo2 C under a
helium ow of 15 scm3 min1 at atmospheric pressure and increas-
ing the temperature at 10 K min1 from room temperature up to
1073 K and maintaining this temperature for 30 min, which is going
to be called the high-temperature and short-time protocol. Finally,
the solids were allowed to cool down to room temperature under
the helium ow. The solids produced in this way are also covered
with a thin carbon lm that protects them from pyrophoric oxi-
dation when exposed to the air, and thus, no passivation protocol
with diluted oxygen was needed in these cases. The % yield was
calculated as Y = PR 100/IP, where PR is the amount of product
recovered after carburization and IP is the amount of the initial
used precursor.
2.4. Characterization of the prepared materials
2.4.1. X-ray powder diffraction
The X-ray analysis of the prepared materials was performed
using a Rigaku ULTIMA III X-ray diffractometer with Cu K radi-
ation as the X-ray source. The scan was done in the range of 1090
2 degrees using a 0.02 step and a counting time of 0.8 per
min to obtain the diffractogram for each prepared material. The
domain crystal sizes were measured using the Scherrer equation,
as implemented in the commercial software JADE [43] (provided
with the diffractometer) by tting the experimental prole to a
pseudo-Voigt prole function, and then calculating the full-width
at half-maximum (FWHM) of the peaks.
The simulated X-ray powder diffraction pattern of the MoO2
structure was calculated with the program POWDER CELL v 2.4
[44] using the obtained experimental data for the MoO2 material
reported for the monoclinic structure [45] and by simulating the
same conditions employed to obtain the experimental XRD pat-
terns with an FWHM of 10 to account for very small crystalline
domains of MoO2 of about 1 nm.
2.4.2. Textural properties
The nitrogen adsorptiondesorption measurements of the sam-
ples were performed at 77 K on a Micromeritics Tristar 3000
analyzer, utilizing BrunauerEmmettTeller (BET) calculations for
surface area. 0.5 g of each sample was used for the analysis, and they
were pretreated on a stream of helium at 423 K overnight before the
measurements.
 G. Vitale et al. / Catalysis Today 250 (2015) 123133
2.4.3. High-resolution transmission electron microscopy
A very small quantity of the powdered carbide sample was dis-
persed in pure ethanol. The powder/ethanol solution was then
agitated until all powder was suspended (approximately 10 min).
A pipette was then used to draw and deposit a drop of the sus-
pension on the high-resolution transmission electron microscopy
(HRTEM) carbon grid sample holder. The drop was then allowed
to dry depositing powder particles on the grid holder. The HRTEM
images were collected on an FEI Tecnai F20 FEG TEM using an accel-
erating voltage of 200 kV.
2.4.4. Infrared spectroscopy (IR)
A Nicolet 6700 FT-IR instrument manufactured by Thermo Elec-
tron Corporation with a smart diffuse reectance attachment to
carry out DRIFTS (diffuse reectance infrared Fourier transform
spectroscopy) analysis was used to study the framework infrared
region of some of the prepared solids. 2.5 mg of the sample was
mixed with 500 mg of KBr and the entire mixture was mounted in
the DRIFTS sample holder. The spectra were recorded from 400 to
4000 cm1 with a resolution of 2 cm1 and 128 scans.
2.4.5. X-ray photoelectron spectroscopy (XPS)
A Physical Electronics PHI VersaProbe 5000-XPS instrument was
used to record XPS spectra. Spectra were taken using monochro-
matic aluminum source, 1486.6 eV, at 50 W, and with a beam
diameter of 200.0 m. Each sample was pressed on indium foil
for XPS analysis. The double neutralization technique, i.e., a low-
energy electron beam and low-energy Ar+ beam, was used during
spectrum acquisition for diminishing the surface charging effect.
The binding energies relative to C 1s, at 284.8 eV, were reported.
Peak tting of the spectra was performed using the Multipak 9.2.0.5
software from Physical Electronics that came with the instrument.
2.4.6. Catalytic testing
As it will be discussed later, two Mo2 C solids prepared at 1073 K
with the AHM-sucrose precursors, and with different crystallo-
graphic phases (cubic and hexagonal), were tested to assess their
hydrogenating activity.
2.4.6.1. Catalyst pretreatment. In order to decrease part of the
amorphous carbon that is covering the samples prepared in this
way [16], a pretreatment in an atmosphere of H2 was applied. The
conditions are as follows: 300 mg of the sample was placed in a
ceramic boat, which in turn, was placed inside a stainless steel hori-
zontal tube (3/4 ) that was heated with a heating tape at 10 K min1
up to 673 K and held at this temperature for 72 h with a H2 ow of
120 scm3 min1 at atmospheric pressure.
2.4.6.2. Catalytic tests. The toluene hydrogenation reaction was
performed in a xed-bed reactor with a continuous ow system.
The unit consisted of a syringe pump (Harvard Apparatus Stan-
dard Infuse/Withdraw PHD Ultra), a hydrogen mass ow controller
(Brooks), a xed-bed reactor with a thermocouple, a back pressure
valve (Swagelok) and a gas ow meter (Brooks). All lines were main-
tained at 458 K during the course of the reaction in order to prevent
any condensation.
One hundred and fty milligrams of the material, previously
treated, was loaded inside the stainless steel 3/8 xed-bed reactor
(length = 15 cm, I.D. = 0.5 cm, O.D. = 0.7 cm). The catalyst was used
without mixing it with any other material and the catalyst bed was
prepared using silicon carbide, which proved unreactive under the
tested conditions, and glass wool in both sides of the catalyst bed.
An inner thermocouple positioned at the height of the catalyst bed
was used to monitor the reaction temperature. Prior to the reaction,
an activation protocol was carried out to the catalyst to remove
surface oxygen that appeared from exposing the Mo2 C phase to
125
air. The following conditions were used: heating at 10 K min1 up
to 673 K and held at this temperature for 3 h with a H2 ow of
120 scm3 min1 at atmospheric pressure. After the activation, the
temperature was decreased to 423 K and the total pressure in the
unit was increased to 4.8 bar. The H2 ow was increased to 300
scm3 min1 to ensure an excess of partial pressure of H2 in the
system and the ow of toluene was 0.2 cm3 h1 . The reaction tem-
perature was 423 K with a weight hourly space velocity (WHSV) of
1.16 h1 . In order to calculate molar conversion of toluene, char-
acterization of gas products was performed. For that purpose, an
HP 6890 Series GC System Hewlett Packard with an incorporated
PONA column (length: 50 m, inner diameter: 200 m, lm thick-
ness: 0.5 m) and a ame ionization detector (FID) was employed
to analyze directly a portion of the product streams.
2.5. Computational studies
Density functional theory (DFT) calculations were performed
using CASTEP [46] as implemented in the Materials Studio suite of
programs v6.1 from Accelrys. In order to get insights into the car-
burization of the MoO2 , as this seems to be the molybdenum oxide
phase that is carburized at high temperature [42], DFT calculations
were applied by replacing each of the two distinct crystallographic
oxygen atoms in the MoO2 structure [41]. Full geometry opti-
mization of the system in the P1 space group, including the cell
parameters, was carried out. In this way, the structural changes
induced by the modication of chemical composition of the solid
could be observed. Geometry optimization was carried out using
the GGA PW91 functional and setting the quality to a cut-off energy
of 750 eV. The total charge of the studied structures was kept con-
stant and equal to zero to avoid problems of non-balanced charge in
the calculations. By doing this, the possibility for the metal cations
to be reduced can be accounted for, as it would be expected to occur
in these types of systems, when the oxygen is replaced by carbon
to form the oxycarbide or carbide materials.
3. Results and discussion
3.1. Mo2 C prepared with the protocol 523 K (24 h), 598 K (24 h)
and 673 K (72 h)
Fig. 1 shows the x-ray diffraction patterns of the materials
prepared with the different precursors. As can be seen in this g-
ure, the precursors prepared with C/Mo ratios of 6.3, 3.6 and 2.0
produced materials with similar X-ray diffraction patterns among
them, which corresponds to the cubic Mo2 C phase. The broaden-
ing of the signals indicates very small crystalline domains and the
estimated crystalline domain size for the samples is 1.4 0.3 nm.
The precursors with C/Mo ratios of 1.0 and 0.6 produced materials
which seem to be a mixture of the cubic carbide phase with other
non-identied phase having also very small crystalline domain
sizes; latter it will be clear that this is a nanocrystalline MoO2 phase.
Table 1 shows the surface area of the precursors and that of the
materials after carburization together with the obtained yield base
on the original amount of reacted precursor. As it can be seen, the
surface areas before or after carburization are low; however, as
the amount of carbon in the precursor is reduced, it seems that
an increase in the surface area is observed concomitantly with a
decrease in the obtained product. These materials are known to
be embedded in an amorphous carbon matrix which explains the
low surface area on them even though the crystalline domain of
the materials is very small [16]. As shown in Table 1, the product
yield increases with the decrease in the C/Mo ratio from 33% for the
C/Mo ratio of 6.3 to 70% for the C/Mo ratio of 0.6 which is understood
as more molybdenum is present in the precursor, the more yield
126 G. Vitale et al. / Catalysis Today 250 (2015) 123133

Table 1
BET specic surface area of the precursors before and after the carburization with the low-temperature and long-time protocol, the obtained yields and the obtained crystalline
phases.

Precursor P1 P2 P3 P4 P5

C/Mo ratio 6.3 3.6 2.0 1.0 0.6

BET area before carburization [m2 g1 ] 0.6 0.5 0.8 1.5 1.3
BET area after carburization [m2 g1 ] 2.1 3.2 4.4 7.8 8.8
Yield [%] 33 4 43 3 53 3 61 2 70 2
Crystalline obtained phase Cubic Mo2 C Cubic Mo2 C Cubic Mo2 C Cubic Mo2 C + MoO2 Cubic Mo2 C + MoO2

is obtained. The obtained XRDs are quite similar for the products
using the precursors P1, P2 and P3; however, subtle differences, of
about 1 m2 g1 , in surface area were observed among them.
Fig. 2 shows the XRD patterns of the materials prepared with
the precursor P3 and with the two partial protocol temperature
ramps. Both obtained materials show similar diffractograms indi-
cating a similar formed structure with very low crystalline domain
sizes. However, the XRDs do not resemble the cubic or hexago-
nal molybdenum carbide phases with small crystalline domains.
As suggested by Guzmn et al. [42], MoO2 seems to be the phase
formed before any carburization starts to occur, thus, a calculation
of the XRD pattern of MoO2 with low crystalline domain sizes was
carried out and the pattern is shown in Fig. 3. Comparison of this
calculated pattern with those obtained experimentally may sug-
gest that indeed a nanocrystalline molybdenum dioxide is formed
when the partial protocol of temperatures is used. This indicates
that the whole temperature ramp protocol is needed to obtain the
cubic carbide in these low-temperature conditions (but with longer
times). If the whole temperature and time protocol is not used, only
a nanocrystalline MoO2 seems to be formed. Fig. 4 shows infrared
spectra of the two samples prepared with the partial protocols and
their comparison with a commercial MoO2 from Aldrich. The struc-
tural region of the infrared spectra (from 400 to about 1700 cm1 )

Fig. 2. X-ray diffraction powder patterns of the prepared materials with the precur-
sor P3 using low-temperature and long-time partial protocol.

Fig. 3. Calculated X-ray diffraction powder pattern of nanocrystalline MoO2 using

the program POWDER CELL [44].

Fig. 4. Infrared spectrum of the commercial MoO2 (Aldrich) and its comparison with
Fig. 1. X-ray diffraction powder patterns of the prepared materials with different the spectra obtained for the two materials using the partial protocols of preparation
C/Mo ratios using the low-temperature and long-time protocol. at low temperature.
 G. Vitale et al. / Catalysis Today 250 (2015) 123133
Table 2
BET specic surface area of the precursors before and after the carburization with the high-temperature and short-time protocol, the obtained yields and the obtained
crystalline phases.
Precursor P1
C/Mo ratio 6.3
BET area before carburization [m2 g1 ] 0.6
BET area after carburization [m2 g1 ] 3.4
Yield [%] 29 3
Crystalline obtained phase Cubic Mo2 C
is also suggesting that a MoO2 phase may be the one present in
these samples as the features observed in this region are similar,
and thus, the ndings of Guzmn et al. [42] may be extrapolated to
low temperature as well. At low temperatures, the formation of the
cubic molybdenum carbides seems to require long times with the
applied protocol of the three ramp temperatures (kinetic effect).
In these low-temperature conditions of reaction, the formed car-
bide phase, prepared with all the different C/Mo ratios, is the cubic
one. It seems that there is not enough available carbon in the lower
C/Mo ratio precursors to induce the complete transformation of the
MoO2 into the carbide, even though H2 is present in the system that
should help to reduce the molybdenum.
3.2. Mo2 C prepared with the protocol 1073 K (30 min)
Fig. 5 shows the XRD patterns of the different products obtained
by the high-temperature protocol treatment (1073 K for 30 min) of
the different precursors. The diffraction patterns indicate that for
Fig. 5. X-ray diffraction powder patterns of the prepared materials with different
C/Mo ratios using the high-temperature and short-time protocol.
127
P2 P3 P4 P5
3.6 2.0 1.0 0.6
0.5 0.8 1.5 1.3
3.9 5.2 7.9 8.4
38 3 47 3 62 2 73 2
Cubic Mo2 C Hexagonal Mo2 C MoO2 + traces MoO2 + traces
hexagonal hexagonal
Mo2 C Mo2 C
the materials prepared with the precursors P1 and P2, the product
is the cubic molybdenum carbide phase having small crystalline
domain sizes with an estimated value of 2.8 0.3 nm. As expected,
the crystalline domain sizes are larger as the temperature increases
from 673 to 1073 K; however, for the precursor P3 (C/Mo ratio of
2.0), the product is not the cubic phase, as it was obtained at low
temperature, but the hexagonal one. As described in the previous
section, at low temperature and in the presence of H2 , the only
carbide phase observed to be produced was the cubic one, even
for the C/Mo = 2.0 precursor. These results suggest the interplay of
temperature, carbon content and H2 presence to favor the cubic or
hexagonal carbide phases. Low temperature (673 K) and H2 atmo-
sphere seemed to favor the formation of the cubic phase even in the
case of low carbon content. On the contrary, lack of H2 , high tem-
perature and carbon content can favor or not the formation of the
hexagonal carbide phase. The patterns of the products with C/Mo
lower than 2.0 produced mainly a well-crystallized MoO2 with
some traces of the hexagonal molybdenum carbide phase. Again,
these results seem to conrm the interplay between temperature,
carbon content and H2 presence as important parameters to selec-
tively produce the cubic or the hexagonal molybdenum carbide
phases with these types of precursors.
Table 2 shows the surface area before and after carburization, at
high temperature, as well as the yields obtained for these prepara-
tions. The results are similar to those obtained at low temperature
in which decreasing the C/Mo ratio increases the surface area and
yield of the products.
Fig. 6 shows the XPS spectra taken on the sample prepared with
the C/Mo ratio of 6.3 at high temperature as received and after
20 min of sputtering to remove the outer most layers of atoms in its
surface. Fig. 6a and d shows the C 1s signal for the material before
and after the sputtering. Before sputtering, only two clear carbon
signals are observed, one at a binding energy of 284.8 eV which
is ascribed to CC bonds that in this case are present because of
the sucrose used as precursor and the adventitious carbon that is
usually found on the surface of most air-exposed samples, the last
one is used as the referenced carbon signal [47]. The second signal
observed is at 287.2 eV which is associated to CO and indicates
the presence of this type of bonds in the surface of the material.
The original carbon source to prepare the precursor was sucrose,
which has CO bonds, and some of them may remain in the nal car-
bon layer, known to be produced under the used reaction protocol
to generate the molybdenum carbides. This carbonaceous mate-
rial is covering the products [16]. No carbon from carbide (around
282.7283 eV [48]) was observed in the surface of the as-received
material. After sputtering the sample with argon ions for 20 min,
three signals are observed in the C 1s spectrum, i.e., 283.4, 284.8
and 287.9 eV. The sputtering allowed going into the bulk of the
material and it is possible to detect the carbide signal at 283.4 eV,
thus conrming the formation of the carbide. However, the CO
signal is still present, indicating that the material has oxygen in the
bulk. This oxygen can be coming from the amorphous carbon that
is covering or mixed with the nanocrystalline material or to the
partial formation of oxycarbides or most probably a mixture of the
last two suggestions.
128 G. Vitale et al. / Catalysis Today 250 (2015) 123133

Fig. 6. XPS spectra of the cubic carbide phase formed at 1073 K, with the P1 precursor, before and after sputtering the surface of the material with argon ions for 20 min. (a)
C 1s signal of the cubic phase before sputtering; (b) O 1s signal of the cubic phase before sputtering; (c) Mo 3d signal of the cubic phase before sputtering; (d) C 1s signal of
the cubic phase after sputtering; (e) O 1s signal of the cubic phase after sputtering; and (f) Mo 3d signal of the cubic phase after sputtering.

Fig. 6b and e illustrates the O 1s signals of the materials before
and after the sputtering. Only two signals are observed in the two
spectra, one with the highest intensity at 530.7 eV and the other
with low intensity at around 532 eV. The signal at 530.7 eV may be
assigned to oxygen bonded to molybdenum as the one reported for
MoO3 (530.6 0.4 eV) or MoO2 (530.5 0.5 eV) [48,49]. The second
signal may be assigned to the oxygen bonded to carbon [49] that
also is detected in the C 1s region. The area ratio of the high to
lower signals is 2.3 for the sample before sputtering and 2.6 for the
sample after sputtering, indicating that both types of oxygen bonds
are more or less homogeneous through the studied sample which is
not the case for carbon. The presence of MoO bonds suggests the
partial oxidation of the sample by its contact with air during the
manipulation or the formation of oxycarbides or most probably a
combination of the two previous suggestions.
Fig. 6c and f shows the Mo 3d signals, which are composed
of doublets 3d5/2 3d3/2 each, before and after sputtering of the
cubic carbide sample prepared at 1073 K with the C/Mo ratio of
6.3. The tting in both spectra indicates the presence of three
oxidation states of molybdenum in this sample. The doublet at
228.5231.6 eV can be assigned to Mo2+ which should be the
one present in the formed carbide. The doublet appearing at
229.4232.6 eV should be assigned to Mo4+ and the doublet at
232.8235.9 eV should be assigned to Mo6+ . Before the sputtering,
the doublet corresponding to Mo6+ is the predominating one with
76% of the total area; which conrms the partial oxidation of the
material by its exposition to the air after the carbide was produced.
After sputtering of the surface of the sample with argon ions for
20 min, the Mo6+ doublet decreases in intensity, and now, the total
area of this signal is only 16%. The predominant doublet is the one
corresponding to Mo4+ with 49% of the total area. Mo2+ is the second
doublet in intensity with a total area of 35%.
In other to get more insights into the prepared material, HRTEM
images were taken for the cubic carbide sample prepared at 1073 K
with the P1 precursor of C/Mo ratio 6.3 and are shown in Fig. 7.
As it can be seen in the images, especially at high magnication,
this material seems to be composed of small crystalline domains
of 510 nm, fused in different directions, with a less crystalline or
 G. Vitale et al. / Catalysis Today 250 (2015) 123133 129

Fig. 7. HRTEM images at different magnications for the cubic carbide material prepared with the C/Mo ratio of 6.3 with the high-temperature and short-time protocol.

amorphous material predominantly visible in the borders of the
particle. This last material should be the carbon that is covering the
material which shows the large C 1s signal (284.8 eV) in the XPS
before sputtering. Also, the low surface area observed in this mate-
rial may be explained by the lack of any porosity in the material as
shown in the HRTEM images.
XPS spectra were taken on the sample prepared at 1073 K with
the precursor P3 having the C/Mo ratio of 2.0 and are shown in
Fig. 8. This material produced the molybdenum carbide phase hav-
ing the hexagonal structure, and thus, it would be important to see
the differences with respect to the cubic phase. Fig. 8a and d shows
the C 1s spectra of the material before and after sputtering with
argon ions for 20 min. Three signals are observed in both spectra:
283.3, 284.8 and 287.8 eV. The rst signal can be assigned to the
carbide carbon, the second to the carbon covering the carbide and
to adventitious carbon, and the last one to carbon bonded to oxygen
130 G. Vitale et al. / Catalysis Today 250 (2015) 123133

Fig. 8. XPS spectra of the hexagonal carbide phase formed at 1073 K, with the precursor P3, before and after sputtering the surface of the material with argon ions for 20 min.
(a) C 1s signal of the hexagonal phase before sputtering; (b) O 1s signal of the hexagonal phase before sputtering; (c) Mo 3d signal of the hexagonal phase before sputtering;
(d) C 1s signal of the hexagonal phase after sputtering; (e) O 1s signal of the hexagonal phase after sputtering; and (f) Mo 3d signal of the hexagonal phase after sputtering.

as it was described for the cubic sample discussed above. However,
in the cubic carbide sample, no carbidic carbon was detected prior
to the sputtering as it is in the case of the hexagonal carbide sample
(less free carbon content). After sputtering, the carbon signal of
the carbidic carbon is the dominant one and the carbon signal at
284.8 eV had an important decrease when compared to the cubic
carbide sample. The signal assigned to CO is still detected after
the sputtering as it happened with the cubic carbide sample. It
seems that the contact with air affected differently the two carbide
samples.
Fig. 8b and e shows the O 1s spectra of the material before and
after sputtering. As it happened with the cubic sample, the hexago-
nal sample shows two clear oxygen signals, one at 530.6 eV and the
other at 532.1 eV and they are assigned to MoO and CO, respec-
tively, as discussed in the case of the cubic sample. The ratio of the
high-to-low intensity signals are 1.5 for the sample before sput-
tering and 2.1 for the sample after sputtering, indicating that a
reduction in the amount of oxygen atoms bonded to carbon atoms
decreased inside the bulk of this sample. The amount of MoO
bonds with respect to the CO bonds seems to be slightly lower in
the hexagonal sample as compared with the cubic one. This may be
due to the lower amount of sucrose used to prepare the hexagonal
sample with respect to the cubic sample.
Fig. 8c and f shows the Mo 3d spectra of the hexagonal car-
bide material. Before and after sputtering, it can be observed that
the tting indicates the presence of three doublets the rst one at
228.3231.2 eV which can be assigned to Mo2+ as in the cubic car-
bide phase, the second one at 229.2232.3 eV assigned to Mo4+ and
the last one at 232.6235.7 eV assigned to Mo6+ . Before the sputter-
ing, the intensity of the Mo2+ signal is important (about 24% of the
total area) which seems to indicate that Mo2+ is a little harder to be
oxidized in the hexagonal phase than in the cubic phase prepared
in this work. This may be because of the differences in carbon and
oxygen content of the two used precursors or to the differences
 G. Vitale et al. / Catalysis Today 250 (2015) 123133
Fig. 9. Structural model of the MoO2 [45] showing the two types of crystallographic
different oxygen atoms in the structure. Blue atoms represent molybdenum and red
atoms represent oxygen.
in crystalline domain sizes of the two produced carbides. The cubic
carbide has very small crystalline domain sizes which may be prone
to an easier oxidation. The amount of Mo6+ in the sample before
the sputtering is 42% of the total area which is lower than the one
observed for the cubic phase (76%). After the sputtering, the Mo2+
signal increased to 58% which is higher than the one observed for
the cubic phase (35%) and the Mo6+ decreases to 10% which is lower
to that observed for the cubic phase (16%). In the cubic sample, after
sputtering, the predominant doublet was the one ascribed to Mo4+
with 49% of the total area but, in the hexagonal sample the amount
of Mo4+ is much lower (28%) and the predominant species is the
Mo2+ . These structural differences should be shown in the activity
test.
3.3. Computational studies
Fig. 9 shows a structural model of MoO2 indicating the two dif-
ferent crystallographic oxygen atoms present in this structure [45].
In this structure, the MoMo distances are alternately shorter and
longer than in the metal and the octahedra are moderately dis-
torted. The short MoMo bond length found in MoO2 is 0.25106 nm
and the MoMo distance in the metal is 0.2725 nm [45]. As sug-
gested in the previous discussion and by Guzmn et al. [42], the
formation of the molybdenum carbide (or oxycarbide) should start
by the partial replacement of the oxygen atoms by carbon atoms
in the lattice of the previously observed molybdenum oxide phase,
which in our preparation methods seems to be MoO2 . The study
presented here is considering the rst oxygen substitution in the
MoO2 structure and observing the differences when the carbon sub-
stitution occurs in the O(1) or in the O(2) sites. The comparison of
total energies of the systems reveals that the structures with one
O(1)-type oxygen center substituted by carbon are the most stable.
Substitution of any of the O(2)-type centers by carbon requires an
additional 47 kJ mol1 . According to this result, it seems that O(1)-
type oxygen atoms are more favored for the carbon replacement,
and thus, some original O(2)-type atoms may be ending up as part
Table 3
Cell parameters of the experimental MoO2 structure and its comparison with the calculated values obtained using CASTEP [46].
a [nm] b [nm]
MoO2 Exp. [45] 0.56109 0.48562
MoO2 Cal. 0.57226 0.50185
MoO2 Cal. C in O(1) 0.59376 0.49770
MoO2 Cal. C in O(2) 0.58423 0.50050
131
of the carbide or, better yet, the oxycarbide formed. This sugges-
tion may explain why in the bulk of the cubic or hexagonal carbide
phases formed experimentally there are oxygen atoms as shown
by the XPS study.
Table 3 shows the comparison of the structural parameters
for the experimentally determined MoO2 structure and the val-
ues obtained in this work for it and for the structures with the
replacement of one oxygen atom O(1) or one O(2) by one carbon
atom. Differences of less than 2% in the structural parameters of the
calculated MoO2 structure, by comparing with the experimental
values [45], were observed which are typical in this type of calcula-
tion. The replacement of one oxygen atom in the O(1) or O(2) sites
produced some slight increase in the a-cell parameter and slight
small decrease in the b- and c-cell parameters when compared
with the calculated MoO2 values. Finally, Fig. 10 shows the total
electronic density of states (TEDOS) plots for the calculated struc-
tures. The TEDOS for the MoO2 structure calculated in this work is
quite similar to that reported by Scanlon et al. [50]; however, the
incorporation of one carbon replacing any of the two types of oxy-
gen produces some slight modication of the TEDOS between the
regions of 8 and 5 eV. New states at around 10 eV are observed
when the carbon replaces one of the two types of oxygen.
3.4. Catalytic tests
It is beyond the scope of the present work to propose a detailed
kinetic mechanism for hydrogenation of toluene on the prepared
carbide materials; however, a comparison between the two differ-
ent carbidic phases prepared with the protocol at 1073 K for 30 min
and, reacting them (under similar conditions for the hydrogenation
of toluene) should show insights into the way the two phases act as
catalysts to hydrogenate toluene. Hydrogenation of toluene under
the tested conditions produced only methyl-cyclohexane as the
only product of reaction. Fig. 11 reports the conversion of toluene
as a function of time on stream for the two carbide phases tested. As
it can be seen in the gure, the hexagonal phase starts with a higher
conversion (around 45%) compared with the cubic phase (around
14%); however, the deactivation of the catalyst with the hexago-
nal carbide phase is very drastic when compared with the cubic
phase under the employed reaction conditions. At around 100 min
on stream, the conversion reached 7% for the hexagonal carbide cat-
alyst, and then it reached a plateau around 5% conversion. The cubic
carbide phase started with a lower conversion, 14%, and there was a
very slow decay in conversion until a plateau was reached at around
3% conversion. Under low H2 pressures, it has been observed that
hexagonal molybdenum carbide materials deactivate with the time
on stream [10]. It has been suggested that coke formation may
be the cause of this deactivation [40]; however, under high pres-
sures of H2 (for instance, 50 bars) deactivation does not seem to
occur with the hexagonal molybdenum carbide phase [40]. Another
explanation has been made in the case of hydrogenation of ben-
zene at atmospheric pressure which suggests that the deactivation
occurs by consumption of the hydrogen that was occluded during
the synthesis of the carbide and, because of the low pressure, no
H2 can diffuse into the bulk structure to be activated, and thus, the
hydrogenation of benzene cannot occur longer [51]. A third possi-
bility for the deactivation of these catalysts, at low H2 pressures, is
c [nm] [ ] [ ]  [ ]
0.56285 90.00 120.95 90.00
0.58370 90.00 120.94 90.00
0.58076 91.07 121.81 89.54
0.58125 90.28 120.95 89.49
132 G. Vitale et al. / Catalysis Today 250 (2015) 123133

Fig. 10. Total electronic density of states (TEDS) plots obtained with CASTEP [46] for the original MoO2 structure and those of the single substitution of one of the two
crystallographic different types of oxygen atoms.

Fig. 11. Conversion of toluene as a function of the time on stream for the two carbide phases (precursor C/Mo = 6.3 cubic phase and precursor C/Mo = 2.0 hexagonal phase)
tested in the hydrogenation reaction at 423 K.

that the aromatic molecules (benzene or toluene, for instance) are
strongly adsorbed. Thus, the sites for the hydrogen adsorption and
splitting are occupied, and then the required adsorption processes
are hard to occur because of the lack of the driving force, which is
the high H2 pressure.
In our previous work [16], the cubic carbide phase prepared at
673 K showed a stable conversion of 44%, for at least up to 72 h
(the maximum time on stream studied), indicating that the high
H2 pressure used (27.6 bar) helped in maintaining the activity in
this phase under the tested conditions.
The preliminary catalytic results, obtained so far in this work,
seem to suggest that the hexagonal phase has a higher activity
than the cubic phase; however, the XPS studies indicated that the
cubic phase has a lower carbidic Mo2+ content and a lower amount
exposed on its surface (carbon covering the particles) which may
be the actual reason why it has a lower activity. Controlling the
conditions of preparation (and the way of handling the materials
after preparation) may increase the carbidic active phase, and thus
enhance its catalytic activity.
4. Conclusions
Under the low-temperature and long-time protocol of carbide
preparation [523 K (24 h), 598 K (24 h) and 673 K (72 h)], the car-
bidic phase that is formed with all the precursors (C/Mo ratios of
6.3, 3.6, 2.0, 1.0 and 0.6) is the cubic one. These materials showed
low surface areas and small crystalline domains sizes which are
embedded in a carbon matrix. This carbon matrix protects the car-
bides from pyrophoric oxidation when exposed to air, and thus only
a mild partial oxidation occurs on their surface when exposed to air.
The lower the C/Mo ratio used, the higher the yield of the material
 G. Vitale et al. / Catalysis Today 250 (2015) 123133 133

produced. The use of partial protocols of preparation, at low tem- [3] E. Nakamura, K. Sato, Nat. Mater. 10 (2011) 158161.
perature [only 523 K (24 h) and 598 K (24 h) or only 673 (72 h)], had [4] Interview, Nat. Mater. 10 (2011) 152163.
[5] J. Ozaki, N. Kimura, T. Anahara, A. Oya, Carbon 45 (2007) 18471853.
a negative effect in the production of the cubic carbide materials. At [6] J. Ozaki, S. Tanifuji, A. Furuichi, K. Yabutsuka, Electrochim. Acta 55 (2010)
low temperature, the formation of the carbide needs longer times 18641871.
and it seems that the formation of nanocrystalline MoO2 is favored [7] Y. Nabae, M. Malon, S.M. Lyth, S. Moriya, K. Matsubayashi, N. Islam, S. Kuroki,
M. Kakimoto, J. Ozaki, S. Miyata, ECS Trans. 25 (2009) 463467.
when the time is shortened, and thus, this indicates that also MoO2 [8] Y. Nabae, M. Malon, S.M. Lyth, S. Moriya, K. Matsubayashi, N. Islam, S. Kuroki,
seems to be a precursor for the formation of the carbides at low M. Kakimoto, J. Ozaki, S. Miyata, 216th ECS Meeting, Abstract MA2009-
temperature. 02836, The Electrochemical Society, http://ma.ecsdl.org/content/MA2009-
02/10/836.abstract%7D (accessed 6.12.13).
Under the high-temperature and short-time protocol of carbide
[9] R.B. Levy, M. Boudart, Science 181 (1973) 547549.
preparation (1073 K for 30 min), two types of carbide phases can be [10] J.S. Lee, M.H. Yeom, K.Y. Park, I.S. Nam, J.S. Chung, Y.G. Kim, S.H. Moon, J. Catal.
produced depending on the C/Mo ratio of the precursors. When the 128 (1991) 126136.
[11] A.S. Rocha, V.L.T. da Silva, A.A. Leito, M.H. Herbst, A.C. Faro Jr., Catal. Today 98
C/Mo ratio is high (6.3 or 3.6), the cubic carbide phase is formed.
(2004) 281288.
When the C/Mo ratio is 2.0, the hexagonal carbide phase is the one [12] A.S. Mamde, J.M. Giraudon, A. Lfberg, L. Leclercq, G. Leclercq, Appl. Catal. A:
formed. Lower C/Mo ratios (1.0 and 0.6) produce mainly MoO2 with Gen. 227 (2002) 7382.
traces of the hexagonal carbide phase. [13] M. Espinoza-Monjardin, J. Cruz-Reyes, M. Del Valle-Granados, E. Flores-Aquino,
M. Avalos-Borja, S. Fuentes-Moyado, Catal. Lett. 120 (2008) 137142.
These results show the interplay of temperature, carbon content [14] M.L. Frauwallner, F. Lpez-Linares, J. Lara-Romero, C.E. Scott, V. Ali, E. Hernn-
and H2 presence in order to favor the cubic or hexagonal carbide dez, P. Pereira-Almao, Appl. Catal. A: Gen. 394 (2011) 6270.
phases. Temperatures up to 673 K and H2 atmosphere seemed to [15] G. Vitale, M.L. Frauwallner, E. Hernandez, C.E. Scott, P. Pereira-Almao, Appl.
Catal. A: Gen. 400 (2011) 221229.
favor the formation of the cubic phase even in the case of low carbon [16] G. Vitale, M.L. Frauwallner, C.E. Scott, P. Pereira-Almao, Appl. Catal. A: Gen. 408
content. On the contrary, lack of H2 , a temperature of 1073 K and (2011) 178186.
carbon content can favor or not the formation of the hexagonal [17] M. Nagai, T. Kurakami, S. Omi, Catal. Today 45 (1998) 235239.
[18] M. Nagai, K. Oshikawa, T. Kurakami, T. Miyao, S. Omi, J. Catal. 180 (1998) 1423.
carbide phase. [19] M.D. Lee, J.F. Lee, C.S. Chang, T.Y. Dong, Appl. Catal. 72 (1991) 267281.
The XPS results indicate that the cubic phase seemed to be more [20] G.P. Khirnova, M.G. Bulygin, E.A. Blyumberg, Petrol. Chem. USSR 21 (1981)
reactive toward partial oxidation when exposed to air than the 4952.
[21] J. Patt, D.J. Moon, C. Phillips, L.T. Thompson, Catal. Lett. 65 (2000) 193195.
hexagonal carbide phase. Both samples contain oxygen atoms in
[22] D.J. Moon, J.W. Ryu, Catal. Lett. 92 (2004) 1724.
the bulk of the material indicating the possible formation of oxycar- [23] P. Liu, J.A. Rodriguez, J. Phys. Chem. B 110 (2006) 1941819425.
bides. The hexagonal phase has more Mo2+ species than the cubic [24] M. Nagai, K. Matsuda, J. Catal. 238 (2006) 489496.
[25] H.C. Woo, K.Y. Park, Y.G. Kim, I.S. Nam, J.S. Chung, J.S. Lee, Appl. Catal. 75 (1991)
phase under the condition of analysis. However, an in situ study
267280.
should be carried out in order to conrm or deny the observation [26] M. Xiang, D. Li, W. Li, B. Zhong, Y. Sun, Fuel 85 (2006) 26622665.
with the samples that were exposed to air before being analyzed [27] J.B.O. Santos, G.P. Valenc, A.J.A.J. Rodrigues, J. Catal. 210 (2002) 16.
by XPS. [28] J.A.J. Rodrigues, G.M. Cruz, G. Bugli, M. Boudart, G. Djega-Mariadassou, Catal.
Lett. 45 (1997) 13.
The computer modeling indicates that the O(1)-type of oxygen [29] D. Treacy, J.R.H. Ross, Prepr. Pap. Am. Chem. Soc., Div. Fuel Chem. 49 (2004)
is more favorable for the carbon substitution than the O(2)-type 643644.
oxygen atoms, and thus, this may suggest that some of the original [30] A.P.E. York, J.B. Claridge, C. Marquez-Alvarez, A.J. Brungs, S.C. Tang, M.L.H.
Green, Stud. Surf. Sci. Catal. 110 (1997) 711720.
O(2)-type atoms may be ending up as part of the carbide or, better [31] J.B. Claridge, A.P.E. York, A.J. Brungs, C. Marquez-Alvarez, J. Sloan, S.C. Tsang,
yet, the oxycarbide formed. This suggestion may explain why in the M.L.H. Green, J. Catal. 180 (1998) 85100.
bulk of the cubic or hexagonal phases formed experimentally there [32] A.J. Brungs, A.P.E. York, M.L.H. Green, Catal. Lett. 57 (1999) 6569.
[33] T.C. Xiao, A.P.E. York, V.C. Williams, H.A. Megren, A. Hanif, X.-Y. Zhou, M.L.H.
are oxygen atoms as shown by the XPS study. Green, Chem. Mater. 12 (2000) 38963905.
Both carbide phases showed activity to hydrogenate toluene [34] J.-G. Choi, J.R. Brenner, L.T. Thompson, J. Catal. 154 (1995) 3340.
to form only methyl-cyclohexane; however, the hexagonal phase [35] P.A. Aegerter, W.W.C. Quigley, G.J. Simpson, D.D. Ziegler, J.W. Logan, K.R.
McCrea, S. Glazier, M.E. Bussell, J. Catal. 164 (1996) 109121.
showed a higher initial activity than the cubic one but the deacti-
[36] J.C. Schlatter, S.T. Oyama, J.E. Metcalfe III, J.M. Lambert Jr., Ind. Eng. Chem. Res.
vation was more pronounced in this phase than in the cubic one. 27 (1988) 16481653.
Thus, a reaction temperature higher than 423 K and/or higher H2 [37] J.S. Lee, M. Boudart, Appl. Catal. 19 (1985) 207210.
[38] S. Ramanathan, S.T. Oyama, J. Phys. Chem. 99 (1995) 1636516372.
pressure might be needed.
[39] G.M. Dolce, P.E. Savage, L.T. Thompson, Energy Fuels 11 (1997) 668675.
[40] B. Dhandapani, T.St. Clair, S.T. Oyama, Appl. Catal. A: Gen. 168 (1998) 219228.
Acknowledgments [41] M. Patel, J. Subrahmanyam, Mater. Res. Bull. 43 (2008) 20362041.
[42] H. Guzmn, W. Xu, D. Stacchiola, G. Vitale, C.E. Scott, J.A. Rodrguez, P. Pereira-
Almao, Can. J. Chem. 91 (2013) 573582.
The authors acknowledge Alberta InnovatesEnergy and Envi- [43] JADE v 7.5.1 XRD Pattern Processing Identication and Quantication Software,
ronment Solutions (AIEES) for the nancial support provided Materials Data, Inc., 2005.
through the Natural Sciences and Engineering Research Council [44] W. Kraus, G. Nolze, J. Appl. Crystallogr. 29 (1996) 301303.
[45] B.G. Brandt, A.C. Skapski, Acta Chem. Scand. 21 (1967) 661672.
of Canada (NSERC) and Nexen Industrial Research Chair in Catal- [46] S.J. Clark, M.D. Segall, C.J. Pickard, P.J. Hasnip, M.J. Probert, K. Refson, M.C. Payne,
ysis for Bitumen Upgrading. A special acknowledge to Dr. Tobias Z. Kristallogr. 220 (2005) 567570.
Frstenhaupt for access to the Microscopy and Imaging Facility [47] T.L. Barr, S. Seal, J. Vac. Sci. Technol. A 13 (1995) 12391246.
[48] X-ray Photoelectron Spectroscopy (XPS) Reference Pages: Carbon:
of the Health Science Center at the University of Calgary, which http://www.xpstting.com/search/label/carbon (accessed 18.12.13); Molyb-
receives support from the Canadian Foundation for Innovation and denum, http://www.xpstting.com/2009/10/molybdenum.html (accessed
the Alberta Science and Research Authority. 18.12.13).
[49] A.V. Naumkin, A. Kraut-Vass, S.W. Gaarenstroom, C.J. Powell, NIST X-
ray photoelectron spectroscopy database, Database 20, Version 4.1.
References http://srdata.nist.gov/xps/ (accessed 18.12.13).
[50] D.O. Scanlon, G.W. Watson, D.J. Payne, G.R. Atkinson, R.G. Egdell, D.S.L. Law, J.
[1] S. Krohns, P. Lunkenheimer, S. Meissner, A. Reller, B. Gleich, A. Rathgeber, T. Phys. Chem. C 114 (2010) 46364645.
Gaugler, H.U. Buhl, D.C. Sinclair, A. Loidl, Nat. Mater. 10 (2011) 899901. [51] R.R. Olivera, A.S. Rocha, V. Teixeira da Silva, A.B. Rocha, Appl. Catal. A: Gen. 469
[2] Editorial, Nat. Mater. 10 (2011) 157. (2014) 139145.