Implications of the Molybdenum Coordination Environment in MFI Zeolites on Methane Dehydroaromatisation Performance

Article abstract of DOI:10.1002/cctc.201901166

The structure and activity of Mo/Silicalite-1 (MFI, Si/Al=∞) were compared to Mo/H-ZSM-5 (MFI, Si/Al=15), a widely studied catalyst for methane dehydroaromatisation (MDA). The anchoring mode of Mo was evaluated by in situ X-ray absorption spectroscopy (XAS) and density functional theory (DFT). The results showed that in Mo/Silicalite-1, calcination leads to dispersion of MoO₃ precursor into tetrahedral Mo-oxo species in close proximity to the microporous framework. A weaker interaction of the Mo-oxo species with the Silicalite-1 was determined by XAS and DFT. While both catalysts are active for MDA, Mo/Silicalite-1 undergoes rapid deactivation which was attributed to a faster sintering of Mo species leading to the accumulation of carbon deposits on the zeolite outer surface. The results shed light onto the nature of the Mo structure(s) while evidencing the importance of framework Al in stabilising active Mo species under MDA conditions.

Full text of DOI:10.1002/cctc.201901166

DOI: 10.1002/cctc.201901166
Full Papers
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
Implications of the Molybdenum Coordination
Environment in MFI Zeolites on Methane
Dehydroaromatisation Performance
[a, b]
[a]
[c]
[b]
Miren Agote-Arán,
Rachel E. Fletcher, Martha Briceno, Anna B. Kroner,
[d]
[a]
[c]
[c]
[c]
Igor V. Sazanovich, Ben Slater, María E. Rivas, Andrew W. J. Smith, Paul Collier,
Inés Lezcano-González,* and Andrew M. Beale*
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
[a, e]
[a, e]
The structure and activity of Mo/Silicalite-1 (MFI, Si/Al=1)
were compared to Mo/H-ZSM-5 (MFI, Si/Al=15), a widely
studied catalyst for methane dehydroaromatisation (MDA). The
anchoring mode of Mo was evaluated by in situ X-ray
absorption spectroscopy (XAS) and density functional theory
weaker interaction of the Mo-oxo species with the Silicalite-1
was determined by XAS and DFT. While both catalysts are active
for MDA, Mo/Silicalite-1 undergoes rapid deactivation which
was attributed to a faster sintering of Mo species leading to the
accumulation of carbon deposits on the zeolite outer surface.
The results shed light onto the nature of the Mo structure(s)
while evidencing the importance of framework Al in stabilising
active Mo species under MDA conditions.
(DFT). The results showed that in Mo/Silicalite-1, calcination
leads to dispersion of MoO precursor into tetrahedral Mo-oxo
3
species in close proximity to the microporous framework. A
Introduction
products over the zeolite Brønsted acid sites (BAS) associated to
framework Al . This mechanism was proposed on the basis of
several activity studies carried out on Mo-based catalysts
3
+
Methane dehydroaromatisation (MDA) is a promising route for
the valorisation of CH into higher value chemicals as it converts
methane directly into aromatics and light hydrocarbons giving
prepared using either non zeolitic supports – such as SiO and
4
2
[11]
TiO2 – or Cs, Ca or Na-exchanged zeolites, with no remaining
BAS. While these materials showed low or no selectivity to
H as co-product. Mo-containing ZSM-5 zeolites are among the
2
[
12,13]
most studied catalysts for this reaction, with the pore
dimensions of the MFI structure being key to provide shape
aromatics,
acidic zeolites led to 100–300 times more
[14]
benzene yield. Furthermore, Mo C and H-ZSM-5 alone were
also reported to be poorly active.
2
[1–6]
[5]
selectivity to benzene (up to 80%).
It has been traditionally accepted that MDA over Mo/
In line with these findings, it has been reported that C H4
2
zeolites occurs via a bifunctional mechanism involving two
conversion on H-ZSM-5 zeolite led to a similar aromatics
distribution than that obtained for CH4 on Mo/H-ZSM-5,
[2,3,7–10]
[15]
different active sites.
According to this mechanism,
molybdenum species constitute the sites responsible for
methane activation, forming H2 as well as C H and C Hx
intermediates, which subsequently transform into aromatic
whereas Marczewski et al. obtained a direct correlation between
[16]
the amount of framework Al and aromatic production, further
supporting the idea that the aromatisation step occurs entirely
on the Brønsted acid sites of the zeolite. In contrast, titration
experiments performed by Tessonier et al. for the quantification
of BAS after Mo exchange, revealed that regardless of the
amount of BAS left, the catalysts had comparable yield to
2
x
3
[
a] Dr. M. Agote-Arán, Dr. R. E. Fletcher, Prof. B. Slater, Dr. I. Lezcano-González,
Prof. A. M. Beale
Chemistry Department, University College of London
Gordon Street, London, WC1H 0AJ (UK)
E-mail: i.lezcano-gonzalez@ucl.ac.uk
[17]
aromatics. The authors proposed that the enhanced activity
in low Si/Al zeolites was mainly due to a better dispersion of
Mo, leading to the formation of highly active isolated Mo
species, and that a very few acid sites must be enough to
perform the aromatisation of the ethylene intermediates.
Interestingly, if the presence of BAS is necessary to produce
aromatics from light hydrocarbons as postulated, their absence
may allow for a catalyst that is selective to methane coupling
products such as ethylene and ethane.
Andrew.Beale@ucl.ac.uk
[b] Dr. M. Agote-Arán, Dr. A. B. Kroner
Diamond Light Source Ltd, Harwell Science and Innovation
Campus, Didcot, OX11 0DE (UK)
[
[
[
c] Dr. M. Briceno, Dr. M. E. Rivas, Dr. A. W. J. Smith, Dr. P. Collier
Johnson Matthey Technology Centre, Blount’s Court, Sonning Common,
Reading RG4 9NH (UK)
d] Dr. I. V. Sazanovich
Central Laser Facility, Research Complex at Harwell, Science and Technology
Facilities Council, Harwell Campus, Didcot OX11 0QX, UK
e] Dr. I. Lezcano-González, Prof. A. M. Beale
Research Complex at Harwell, Rutherford Appleton Laboratory, Harwell
Campus, Didcot, OX11 0FA (UK)
Nonetheless, an alternative monofunctional mechanism has
been suggested on the basis that methane aromatisation can
occur in the absence of BAS. Although with low conversion and
selectivities, aromatisation on MoO /SiO has been previously
Supporting information for this article is available on the WWW under
https://doi.org/10.1002/cctc.201901166
3
2
This publication is part of a Special Collection on “Advanced Microscopy and
Spectroscopy for Catalysis”. Please check the ChemCatChem homepage for
more articles in the collection.
[18]
reported.
More recently, Guo et al. synthesised a novel
Fe@SiO catalyst, wherein isolated iron atoms were embedded
2
ChemCatChem 2019, 11, 1–12
1
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
��
These are not the final page numbers!

Full Papers
[19]
and stabilised in a matrix of amorphous SiO2. The authors
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
showed MDA activity and high yields to benzene at 1000°C. In
agreement, catalytic data reported by Kosinov et al. for Mo-
containing Silicalite-1, the pure siliceous analogue of H-ZSM-5
zeolite, evidenced the formation of aromatics over this catalyst
material, leading to the proposal that Brønsted acid sites may
[20]
not be essential for the aromatisation of methane. The same
group has recently highlighted the involvement of confined
[21,22]
carbon in the reaction mechanism instead.
Our previous synchrotron-based operando studies on Mo/H-
ZSM-5 evidenced the relevance of Mo speciation in MDA
product distribution. X-ray emission showed the selectivity to
different hydrocarbon products to be dependent on Mo
structure (i.e. partially carburised species were observed during
the induction period when light hydrocarbons were produced,
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
[23]
whereas fully carburised sites were selective to aromatics). X-
ray absorption and diffraction experiments allowed to correlate
the Mo species evolution with catalysts deactivation, a major
[24]
handicap for the commercialisation of MDA route. This work
showed that as a result of complete carburisation, Mo detaches
from the zeolite; the subsequent Mo migration from the pores
and loss of shape selectivity leads to the formation of carbon
deposits on the zeolite outer surface. Accumulation of these
deposits block the access of reactants to the active sites
[24]
resulting in catalyst deactivation.
From the works discussed above it is clear that metal ions
such as Mo have a major role in the aromatisation stage than
what has been traditionally postulated in the bifunctional
mechanism. To date, the extensive research carried out to
determine the influence of catalyst support on activity does not
include a thorough characterisation to account for differences
in metal speciation. Therefore, the aim of this research is to
compare the of Mo-containing MFI zeolites in order shed more
Figure 1. Mo K-edge XAS data collected during in situ calcination of Mo/
Silicalite-1 (20% O
(no phase corrected) with vertical lines indicating the peaks assigned to
MoÀ O and MoÀ Mo scattering paths in the MoO
2
in He, 700°C, 5°C/min): a) XANES spectra, b) FT-EXAFS
3
+
light onto the role of Mo speciation, framework Al and BAS.
This is carried out by studying the structure and catalytic
activity of Mo/Silicalite-1 with no Al, in comparison with the
widely studied Mo/H-ZSM-5 with a silicon to aluminium ratio of
3
precursor.
1
5 and comparable Mo loadings of around 4 wt.%. Notably, the
absorption near edge structure (XANES) spectra (Figure1a)
present a pre-edge peak around 20005 eV (dipole-forbidden
1 s!4d electronic transition) and a rising absorption edge at
20015 eV (dipole-allowed 1 s!5p transition). The spectra do
not change significantly up to 600°C showing that the local
study of a pure siliceous zeolite material containing highly
dispersed Mo-oxo species (vide infra) allows us to disentangle
the influence of Mo structure and zeolite acidity on the catalytic
behaviour largely debated over the past decades. The nature of
initial Mo-oxo species is investigated in detail by X-ray
absorption spectroscopy (XAS) and density functional theory
structure of the MoO precursor (comprising interconnected
3
MoO octahedra) remains fairly stable. Above 600°C, the pre-
6
(DFT), while the physicochemical properties of reacted catalysts
edge intensity increases, indicating changes in the symmetry
[
25]
are evaluated by characterisation of samples in the first few
hours of the MDA reaction.
around molybdenum from octahedral to tetrahedral.
In
addition, the post-edge region (>20030 eV) above 600°C loses
its features resulting in a broad peak, suggesting loss of long-
range order and dispersion into isolated Mo-oxo species
[26]
Results and Discussion
attached to the microporous framework.
The Fourier transforms of the X-ray absorption fine structure
(FT-EXAFS) are shown in Figure 1b. At low temperatures the
spectra present features typical of MoO which contains
Study of Mo Speciation within Mo-MFI Structures
XAS during in situ Calcination
3
,
[26,27]
distorted octahedral MoO units.
The scattering from near
6
neighbour O results in two resolved peaks in the FT-EXAFS,
with maxima around 1.3 Å (labelled as MoÀ O1) and 2.0 Å
(MoÀ O2). The third peak observed, between 3 and 4 Å, arises
The Mo K-edge XAS data collected during calcination of the as-
prepared Mo/Silicalite-1 is shown in Figure 1. The X-ray
ChemCatChem 2019, 11, 1–12
www.chemcatchem.org
2
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
��
These are not the final page numbers!

Full Papers
from scattering from neighbouring Mo atoms in agreement
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
[28]
with reported MoÀ Mo bond distance of c.a. 3.5 Å. During
calcination, a gradual decrease in the peak intensity is observed
below 600°C, this decrease is more pronounced for atoms at
longer radial distances and can be attributed to an increasing
thermal disorder which leads to the damping of the EXAFS
oscillations. In line with the XANES results, there are no shifts in
the position of the peaks, suggesting that Mo local structure is
not significantly altered in this temperature range. Above 600°C
however, sudden changes in the FT-EXAFS indicate the
formation of isolated Mo-oxo species attached to the Silicalite-1
framework. The peak at radial distances >3.0 Å disappear,
evidencing loss of MoÀ Mo scattering; the peak with maxima
around 2.0 Å moves to shorter bond distances, probably
corresponding to O atoms bridging to the zeolite, and the peak
at lowest radial distances red-shifts, consistent with the
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
Figure 2. Mo K-edge FT-EXAFS of Mo/Silicalite-1 and Mo/H-ZSM-5 samples at
room temperature after in situ calcination together with FT-EXAFS of
[24]
formation of short terminal Mo=O double bonds.
Fe
the radial distance for the maxima of the Fe
2
(MoO
4
)
3
reference with tetrahedral MoO
4
units. Vertical dashed line marks
(MoO
Silicalite-1 contains no framework Al, although the presence
of Al as an impurity below the detection limit of ICP-OES cannot
be excluded. Unlike Silicalite-1, H-ZSM-5 contains framework Al
atoms (Si/Al=15), which provide ion exchange capacity and
strong anchoring points for the Mo-oxo species. Character-
isation of the acidic properties of the materials here used
2
4 3
) .
In order to refine the two bond distances, a first shell fit of
the EXAFS spectra collected at room temperature after the
in situ calcination was performed. The best fits were obtained
by setting the coordination number 2 for both short and long
(
NH À TPD and FTIR) can be found in Figure S1 in the ESI. In
3
spite of the absence of framework Al, the observed Mo
structure evolution into tetrahedral-like Mo-oxo species on
Silicalite-1 is analogous to previous reports on acidic Mo/H-
molybdenum-oxygen bond distances typical of a (O=) Mo
2
(À OÀ Si) species. Table 1 presents the parameters refined for
2
[23,24,26]
ZSM-5.
For comparison, Figure S2 in the ESI presents the
Mo/Silicalite-1 and Mo/H-ZSM-5, the experimental and simu-
lated spectra are plotted together in Figure S4 in the ESI. The
results suggest similar Mo=O bond distances of 1.69 Å for both
samples while the MoÀ O distance is considerably longer on the
pure siliceous zeolite (i.e. 2.31 Å in Mo/Silicalite-1 vs 1.80 Å in
Mo/H-ZSM-5).
X-ray absorption spectra collected for Mo/H-ZSM-5 during
in situ calcination under identical experimental conditions. In
agreement with the XAFS data, no MoO3 crystallites were
detected by XRD in the calcined Mo/Silicalite-1 (see Figure S3 in
the ESI), indicating that Mo is well dispersed after calcination.
Furthermore, a decrease in the micropore volume (14%) was
also observed upon calcination (Table 2), supporting the
migration of Mo into Silicalite-1 pores. By FTIR a decrease in
intensity of the bands corresponding to the silanol groups in
Silicalite-1 is seen upon calcination of the physical mixture
Density Functional Theory
In order to verify these observations, DFT calculations were
performed, allowing to gain further insight into the interaction
of Mo with the Silicalite-1 support. These were concerned with
the optimisation of isolated di-coordinated Mo-oxo structures
(Figure S1b), suggesting that metal dispersion may partially
occur via interaction of Mo with such groups.
The room temperature FT-EXAFS for calcined Mo/Silicalite-1
and Mo/H-ZSM-5 catalysts are compared in Figure 2. The plot
also includes the spectrum of a Fe (MoO ) reference, contain-
2
4 3
ing tetrahedral MoO units with an average MoÀ O distance of
4
Table 1. EXAFS fitting parameters of the Mo/Silicalite-1 and Mo/H-ZSM-5
~
1.75 Å. Compared to Fe (MoO ) , Mo/zeolites seem to present
2
2
4 3
spectra acquired at RT after calcination: So =0.91, Fit Range: 3<k<12, 1<
a component at lower radial distances, which is consistent with
R<3. Where CN=coordination number, R=bond length of the Absorber-
2
[24]
Scatterer, σ =Mean squared disorder term (sometimes referred to as the
the presence of short Mo=O bonds. Interestingly, the two
MoÀ O distances (i.e. terminal Mo=O and bridging MoÀ O bonds
to the zeolite framework) in the calcined Mo/Silicalite-1 result in
two well resolved peaks, while in contrast, Mo/H-ZSM-5 exhibits
a single broad peak. This could be explained by a different
degree of interaction between Mo and framework oxygen
atoms in both zeolites. Consistent with the absence of frame-
work Al, the interaction of Mo with the pure siliceous Silicalite-1
framework is weaker, resulting in longer MoÀ O distances for the
bridging oxygen atoms and allowing for better resolution of
Mo=O and MoÀ O scattering peaks in the FT-EXAFS.
Debye Waller factor), E =Energy shift, R
o
Factor
=A statistic of the fit, which is
a way of visualising how the misfit is distributed over the fitting range.
2
2
Sample
Schell CN R [Å]
σ [Å ]
E0
RFactor
%]
[
Mo/
Silicalite-1
Mo=O
MoÀ O
Mo=O
MoÀ O
2
2
2
2
1.69
2.31
0.0046
11.93
3.5
(+/À 0.0012) (+/À 2.03)
0.0079
(
+/À 0.03) (+/À 0.0030)
Mo/
H-ZSM-5
1.69
0.0059
À 0.15
3.2
(+/À 0.0043) (+/À 2.61)
1.80
0.0041
(
+/À 0.04) (+/À 0.0016)
ChemCatChem 2019, 11, 1–12
www.chemcatchem.org
3
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
��
These are not the final page numbers!

Full Papers
[
30]
with tetrahedral geometry, as suggested by the above X-ray
absorption results. The fully geometry optimised structures,
shown in Figure 3, comprise pseudo-tetrahedral [MoO2] sites
defects of an amorphous silica support. To evaluate the effect
of silanol defects on the binding of the initial Mo-oxo to the
walls of the zeolite framework, a range of silanol-containing
Mo/Silicalite-1 structures were examined. As geminal sites have
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
2
+
attached to the zeolite walls through two MoÀ O bonds to
F
[
30]
oxygen atoms on oxygens adjacent to the T5 site; they also
been reported to be highly unstable due to their rigidity, only
vicinal and non-vicinal sites are considered.
contain two terminal Mo=O bonds. In the case of Mo/Silicalite-1
(
Figure 3a) the bond lengths obtained are 1.73 Å and 2.33 Å for
As indicated in the experimental section, the presence of
silanol nests were simulated through the removal of a single Si
atom from the T5 sites. The four dangling SiÀ O bonds
generated by the introduction of the vacant site were satisfied
by the presence of hydroxyls. [MoO2] was then di-coordi-
nated across the now vacant T5 site, with the removal of two
protons, simulating binding across two terminal silanol defects
via a condensation reaction. The fully geometry optimised form
of this structure is shown in Figure 3c. This structure contains
Mo=O and MoÀ O respectively. These values are comparable to
F
those refined by EXAFS (1.69 Å and 2.31 Å). Note that the
MoÀ O length could be considered too long to be a real bond;
F
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
2
+
nonetheless, the Mulliken charges on the two framework
oxygens ‘bonded’ to Mo are more negative than charges on
other framework oxygens, (À 0.55 compared to a mean charge
of À 0.45 for the remaining framework oxygens). This indicates
that an electrostatic interaction exists, and that charge transfer
occurs between framework oxygens and Mo.
equivalent Mo=O bond lengths of 1.69 Å; MoÀ O bond lengths
F
For Mo/H-ZSM-5 (Figure 3b), while the Mo=O bond distance
however are shorter than those reported for the Mo-oxo species
supported on silanol-free Silicalite-1 (in Figure 3a) with one
is equivalent to the Mo/Silicalite-1, the MoÀ O bond obtained is
F
significantly shorter with a value of 1.8 Å. This value is
comparable to our EXAFS results in Table 1 as well as previous
bond slightly longer than the other, MoÀ O =1.90 Å and 1.88 Å.
F
2
+
The fully optimised structures of [MoO ] bound across
2
[29]
theoretical calculations on Mo/H-ZSM-5. The shorter distance
compared to Silicalite-1 is presumably due to strong interac-
tions between the Mo-oxo species and the aluminosilicate
framework.
By FTIR a range of silanol defect groups were found to exist
in the Silicalite-1 sample that may serve as anchoring point of
Mo-oxo species. These defects include vicinal, geminal and
large number of silanol nests (see Figure S2 for the assign-
ments). Previous DFT studies have shown variations in the
stability Mo-oxo species anchored through different silanol
vicinal and non-vicinal silanol groups are shown in Figure 3d
and 3e respectively. In these structures, Si has been removed
from the T5 and T9 sites and the resulting dangling bonds are
satisfied by hydroxyls, simulating the presence of two silanol
nests per MFI unit cell. The calculations suggest that bonding to
À 1
vicinal OHs is more stable than non-vicinal by 52 kJmol ,
[30]
contrary to the work by Guesmi et al. The bond lengths in
these structures are Mo=O=1.70 Å and MoÀ O =1.90 Å and
F
1.88 Å for binding across the vicinal site and Mo=O=1.70 Å and
MoÀ O =1.97 Å and 2.08 Å. Mulliken population analysis sug-
F
gests that Mo is interacting with both framework oxygens,
hence it appears that binding at vicinal sites is preferred as a
consequence of stronger, more equal coordination to the
framework. The DFT calculations for Mo attached to Silicalite-1
through silanols result in significantly shorter MoÀ OF bond
lengths (1.88 to 2.08 Å) compared to the ones obtained by
EXAFS fitting (2.31 Å). In contrast, the model with no silanols in
the structure is closer (MoÀ O =2.33 Å) to the fitted values. This
F
suggests that the Mo-oxo species are not bound exclusively
across vacancy sites in the Silicalite-1 framework and that
interaction and charge transfer occurs between framework
oxygen atoms and Mo.
Methane Dehydroaromatisation
Catalytic Activity
To compare the performance of Mo/H-ZSM-5 and Mo/Silicalite-
1
, MDA reactions were carried out for a period of 10 h, with the
outlet gas composition measured by MS. The data correspond-
ing to the main reaction products (i.e. H , C H and C H ) are
2
2
x
6
6
Figure 3. The fully geometry optimised form of the Mo-oxo species bound
to Silicalite-1 (a) and H-ZSM-5 (b). The bonding to Silicalite-1 through silanol
defects is modelled across the nests (c), vicinal (d) and non-vicinal silanol
defects (e). In all structures [MoO2] contains two terminal Mo=O bonds
and is di-coordinated by two short MoÀ OF bonds. The framework is shown
as stick bonds, Mo is represented in purple, oxygen in red and hydrogen in
pink.
presented in Figure 4. Both catalysts exhibit an induction period
during the first minutes of reaction where H formation is seen
2
2
+
but no hydrocarbon production is detected (data from earlier
reaction times are shown in Figures 4a and 4b). This induction
period is coincident with the evolution of CO, CO and H O (see
2
2
ChemCatChem 2019, 11, 1–12
www.chemcatchem.org
4
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
��
These are not the final page numbers!

Full Papers
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
À 1
Figure 4. MS profiles of the products formed during the methane dehydroaromatisation reaction (50% CH
1
4
/Ar, 1500 h , 700°C). First 60 min for Mo/Silicalite-
(a) and Mo/H-ZSM-5 (b) with vertical lines marking the end of the induction period. 10 h reaction data for Mo/Silicalite-1 (c) and Mo/H-ZSM-5 (d).
2
Table 2. N physisorption results and carbon content of Mo/MFI catalysts
Figure S5 for the MS trends of combustion products) and it is
accepted that during this stage tetrahedral Mo-oxo species
undergo reduction and carburisation leading to the formation
of active species for dehydroaromatisation. After the initial
induction period, the aromatisation stage was seen to com-
mence for both catalysts evidenced by the evolution of C H ,
at different reaction times.
Sample
Mo [wt. %]
TGA-Carbon
S
BET
2
V
micr
[cm /g]
3
3
50–600°C
[m /g]
[wt. %]
H-ZSM-5
/
/
/
0.4
3.0
6.3
/
412
344
351
326
280
452
389
409
333
302
0.15
0.12
0.12
0.11
0.09
0.17
0.14
0.14
0.12
0.11
6
6
Mo/H-ZSM-5 calc.
Mo/H-ZSM-5(7 min)
Mo/H-ZSM-5(90 min)
Mo/H-ZSM-5(10 h)
Silicalite-1
Mo/Silicalite-1 calc.
Mo/Silicalite-1(7 min)
Mo/Silicalite-1(90 min)
Mo/Silicalite-1(10 h)
3.80
/
/
/
together with light hydrocarbons and H . In agreement with
2
[20]
Kosinov et al.,
the observed formation of C H on Mo/
6 6
Silicalite-1 suggests that BAS are not required for aromatisation,
pointing out to this being also an intrinsic property of
/
3.62
/
/
/
/
0.5
5.4
7.4
[5]
molybdenum carbide species.
A close analysis of the MS data reveals differences in MDA
performance of both catalysts. Interestingly, a shorter induction
period is observed for Mo/Silicalite-1 (6.0 min) compared with
the Mo/H-ZSM-5 (8.5 min) which suggests a faster carburisation
of initial Mo-oxo species on the pure Si zeolite. Regarding the
aromatisation stage, data over the 10 h of reaction (Figures 4c
and 4d) show that C H and H gradually decreased for Mo/H-
EXAFS and DFT studies above. In the Mo evolution mechanism
proposed from our previous operando studies, during the
induction period the Mo-oxo species present after calcination
are fully carburised; this occurs by the consecutive replacement
6
6
2
ZSM-5 indicating a gradual catalyst deactivation. For Mo/
Silicalite-1 however, C H and H were seen to steeply decrease
[23,24]
of terminal and bridging O atoms by C.
The weaker
6
6
2
within the first ~3 h of reaction, evidencing a significantly faster
deactivation. In addition, the TGA analysis (Table 2) performed
for catalysts reacted for different times revealed increased
amounts of carbon deposits on Mo/Silicalite-1.
These activity differences can be explained by the degree of
interaction of Mo with the zeolite framework as discussed in the
interaction of Mo with Silicalite-1 compared with H-ZSM-5
would potentially facilitate the carburisation of bridging O
therefore shortening the induction period. Such weaker inter-
action would also lead to a faster sintering of Mo species and
their migration to the outer surface under reaction conditions.
In line with the deactivation mechanism previously proposed
ChemCatChem 2019, 11, 1–12
www.chemcatchem.org
5
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
��
These are not the final page numbers!

Full Papers
[24]
by our group, the shape selectivity to aromatics provided by
the MFI pore structure is lost as a consequence of this
migration, favouring the formation of large carbonaceous
deposits in the zeolite outer surface. Hence, a faster sintering of
the active species in Mo/Silicalite-1 would explain the increased
formation of carbon deposits in this catalyst - as observed by
TGA, and thereby the faster deactivation.
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
We note that aromatisation over BAS is believed to occur
via acid catalysed heterolytic reactions (i.e. oligomerisation, β-
scission, hydride transfer, protolysis, alkylation, dealkylation
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
[9,10]
etc),
while a radical-mediated reaction mechanism is more
plausible for MDA over an acid-free catalyst. Variations in
reaction mechanism over Mo/H-ZSM-5 and Mo/Silicalite-1 as a
result of the different BAS content could also possibly
contribute to the differences in the carbon deposition profile
observed.
Figure 5. Raman spectra of Mo/Silicalite-1 and Mo/H-ZSM-5 reacted for
Post Reaction Characterisation
À 1
90 min (50% CH
4
/N
2
, 700°C, 1500 h ).
In order to gain insight into the fast deactivation of Mo/
Silicalite-1, the reacted catalysts (i.e. after 7 min, 90 min and
2
1
0 h of reaction) were further characterised.
plane stretching vibrations of pairs of sp C atoms. This band is
observed in graphitic-type carbon as a result of lattice
Table 2 displays the textural properties and the carbon
[
31–33]
content measured by TGA. Note that the Mo content
determined by ICP for the calcined samples) is similar for both
vibrations.
In both catalysts these bands are shifted to
À 1
(
higher wavenumbers (1611 cm ). This shift has been reported
À 1
catalysts. As compared to the parent zeolites, the micropore
volume of the calcined Mo-containing catalysts was seen to
significantly decrease; this decrease is similar for Mo/Silicalite-1
to be due to a contribution from a second D2 band (1620 cm )
[33,34]
attributed to edges of graphitic crystallites.
Thus, the
observed shift suggests the presence of very small carbon
crystallites with large surface/bulk ratio and therefore a high
(14% decrease) and Mo/H-ZSM-5 (16% decrease) suggesting a
[
33,35]
comparable Mo dispersion. As discussed in the XAFS section
above, the drop in micropore volume is probably driven by the
migration of Mo into the zeolite pores during calcination to
form Mo-oxo species. From 7 min to 10 h of reaction, the
micropore volume was also seen to decrease as a result of
carbon deposition. The increase in the amount of deposited
carbon is clearly evidenced by TGA, responsible for the gradual
catalyst deactivation observed in Figure 4. The derivatives of
the TGA curves at 7, 25 min and 10 h of reaction (Figure S6 in
the ESI) also exhibit a gradual increase of peak intensity and the
combustion temperature with reaction time, in line with
increased carbon deposition as the reaction advances. As
shown in Table 2 the amount of carbon deposited on Mo/
Silicalite-1 was always higher than for Mo/H-ZSM-5 consistent
with the faster deactivation observed when the pure siliceous
zeolite is used as the support.
number of edges.
A weak shoulder can be also observed at
À 1
1200 cm , this Raman shift has been named as a D4 band and
2
3
it is ascribed to the stretching vibrations of sp -sp CÀ C bonds
[36]
in conjugated aliphatic species. The ratio of D1 and G(D2)
bands intensities gives insight regarding the degree of order in
[31,33]
the carbon structure;
increasing I(D1)/I(G) indicates increas-
ing structural disorder. The higher D1/G value for Mo/Silicalite-1
(0.61) than for Mo/H-ZSM-5 (0.52) indicates that the carbon
deposited on the catalyst with no framework Al is more
amorphous in nature. Hence, Raman studies confirm the
formation of significant amounts of disordered graphitic carbon
or small graphite crystallites in the spent samples, with lower
degree of crystallinity for Mo/Silicalite-1. From our Raman or
TGA data it is difficult to discern between carbon deposits
located inside the zeolite channels or on the outer surface. Our
previous PXRD studies on Mo/H-ZSM-5 however, suggests that
most of the carbon is located in the external surface of the
In order to gain knowledge on the nature of carbon
deposits formed during reaction, Raman spectra was acquired
for the catalysts reacted for 90 min. Figure 5 shows the first
order Raman bands for Mo/Silicalite-1 and Mo/H-ZSM-5, which
present two distinct bands typical for carbon compounds. The
[24]
zeolite crystals.
Microscopy studies (TEM and SEM) were also carried out for
reacted catalysts in order to evaluate the Mo sintering process
during MDA. High resolution backscattered electron images for
Mo/Silicalite-1 reacted for 7 min, 90 min and 10 h are shown in
Figure 6 and illustrate the sintering process of Mo species under
reaction conditions. The images contain Silicalite-1 crystals of
around 200 nm, whereas the observed bright spots are, based
À 1
band around 1360 cm is denoted as D1 (disordered) band
2
and it is ascribed to in-plane breathing vibrations of sp -bonded
carbon (rings). The D1 band is usually attributed to amorphous
carbon, carbon nanoparticles or defects in graphitic-type
[
31]
[24]
deposits.
The second band is known as the G (graphitic)
on signal intensity, due to molybdenum carbide. Note that
À 1
band, it usually appears at 1580 cm and corresponds to in-
this analysis was done using a low accelerating voltage of
ChemCatChem 2019, 11, 1–12
www.chemcatchem.org
6
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
��
These are not the final page numbers!

Full Papers
The 90 min reacted sample (Figure 6b) presents an increased
number of molybdenum carbide particles whereas after 10 h
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
(Figure 6c), the number of these particles increases further such
that all zeolite crystals possess sintered molybdenum carbide
on the surface. A similar trend is also observed for Mo/H-ZSM-5,
in line with previous publications (see Figure S7 in the ESI for
[8,23]
equivalent images in Mo/H-ZSM-5).
TEM images were also taken at 90 min of MDA where the
steep deactivation of Silicalite-1 is observed in contrast of more
stable activity of Mo/H-ZSM-5. As shown in Figure 7a Mo/
Silicalite-1 presents large molybdenum carbide particles of
around 15 nm diameter, while Mo/H-ZSM-5 (Figure 7c) exhibits
significantly smaller particles (>5 nm). After 10 h of reaction
extensive sintering was observed for both catalysts. As
expected, the EDX maps (Figure 7b and 7d) reveal Mo to be
associated with C, consistent with the presence of molybdenum
carbide particles. An even distribution of Si and O were found
Figure 6. SEM-ESB (Energy Selected Backscattered) electron images (accel-
erating voltage 1.6 eV) of Mo/Silicalite-1 after 7 min (a), 90 min (b), and 10 h
À 1
(c) of reaction (50% CH
4
/N
2
, 700°C, 1500 h ).
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
1
.6 eV; hence, the Mo distribution observed corresponds to the
outermost surface of the zeolite crystals.
Very few bright spots are present after 7 min of reaction
(Figure 6a) suggesting a low degree of molybdenum carbide
sintering towards the zeolite surface at this stage of reaction.
À 1
Figure 7. STEM-DF image of the catalysts after 90 min of reaction (50% CH
4
/N
2
, 700°C, 1500 h ). Mo/Silicalite-1 (a) and Mo/H-ZSM-5 (c); and 10 h reacted
STEM-DF with the corresponding EDX elemental maps of Mo/Silicalite-1 (b) and Mo/H-ZSM-5 (d).
ChemCatChem 2019, 11, 1–12
www.chemcatchem.org
7
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
��
These are not the final page numbers!

Full Papers
in the zeolite crystals, while the Al distribution in Mo/H-ZSM-5
presents brighter spots probably due to presence of extra
framework aluminium or Al molybdate particles. Interestingly,
in the C elemental map of Mo/Silicalite-1, which contains a
higher amount of C, a more intense carbon signal can be
distinguished on the periphery of the Silicalite-1 particles
nisms open the possibility to optimise active MDA catalysts
based on non-acidic zeolites. These supports exhibit enhanced
thermal stabilities that represent an advantage for the MDA
reaction, which operates >700°C and is equilibrium-limited to
~14% at 700°C, allowing to increase the reaction temperature
for improved yields.
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
(Figure 7b). This indicates a significant accumulation of carbon
While the presence of BAS associated to framework Al may
not required for aromatisation to occur, our research suggest
that Al plays an important role on the stabilisation of Mo
species. The shorter induction period for Mo/Silicalite-1 alludes
to a more rapid carburisation of the initial Mo-oxo species. In
line, a faster Mo sintering under reaction conditions and severe
catalyst deactivation is also seen for this catalyst. These
observations can be attributed to the weaker interaction of Mo
species with the purely siliceous Silicalite-1 evidenced by EXAFS
and DFT. As proposed in our previous studies, such carburisa-
tion and sintering comprise the migration of Mo to the zeolite
outer surface; this leads to the loss of shape electivity to
aromatics provided by the pores and extensive accumulation of
bulky carbon deposits in the zeolite outer surface. While
variations in the reaction mechanism due to different acidic
properties of Mo/H-ZSM-5 and Mo/Sillicalite-1 could also
contribute to the different carbon deposition profile observed,
our data suggests that the instability of Mo active sites on
Silicalite-1 is the main cause of catalyst deactivation. The
characterisation carried out for reacted Mo/MFI samples
suggests a more pronounced sintering of molybdenum on
Silicalite-1. The increased accumulation of carbon deposits is
deposits on the outer surface of the zeolite.
These results are in agreement with the catalytic activity
data and deactivation mechanism discussed in Figure 4. The
weak interaction between initial Mo species and Silicalite-1
observed by EXAFS and DFT is a key factor for the enhanced
sintering of the Mo active species seen by TEM. The migration
of Mo carbides to the outer zeolite surface leads to an increased
carbon deposition on the zeolite periphery and thereby, a more
rapid deactivation of Mo/Silicalite-1.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
Summary and Conclusions
This study compares the properties of Mo/zeolites with MFI
structure using zeolites with and without framework Al (H-ZSM-
5
and Silicalite-1 respectively). The structural refinement of the
initial Mo-oxo species by XAS and DFT, followed by MDA
activity studies and characterisation of reacted catalysts bring
new insights into the role of framework Al on the catalytic
performance.
XAS data collected during in situ calcination of Silicalite-1
and MoO physical mixture suggest that upon calcination MoO
concluded from the N physisorption and TGA; Raman spectra
3
3
2
migrates into zeolite pores, leading to the formation of
tetrahedral Mo-oxo species. The structure of these species (i.e.
with two terminal Mo=O and two bridging MoÀ O groups
attached to the zeolite framework) appears to be analogous to
the ones previously reported for ion-exchanged Mo/H-ZSM-5,
with the BAS acting as anchoring sites for Mo. Longer MoÀ O
distances to the framework oxygens on Silicalite-1 however
allude to a weaker interaction of Mo with this support. DFT
calculations are in agreement with the experimental refinement,
suggesting that Mo in Mo/Silicalite-1 it is not exclusively bound
across silanol defect sites of the Silicalite-1 support and attaches
to the zeolite walls through oxygen atoms via electrostatic
interaction.
indicate the deposits consist of disordered graphitic carbon,
with a lower degree of crystallinity for Mo/Silicalite-1. TEM-EDX
maps on 10 h reacted Mo/Silicalite-1 show high C concentration
in the zeolite outer surface.
The structure and location of Mo species clearly have a
strong impact on catalytic performance. In this regard, frame-
work Al seems to contribute to the stabilisation of Mo species
resulting in a delayed catalyst deactivation. In this sense
promising engineering solutions are being investigated by
several groups to remove carbon deposits (i.e. through catalyst
[
4,15,38–41]
regeneration cycles or by using membrane reactors).
Nonetheless, an alternative approach to prevent deactivation
would comprise the optimisation of the catalyst formulation,
focused on further improving the stability of the active metal
species inside the zeolite pores. In this line, follow up
investigations should focus on enhancing the interaction of the
active metal and the support on MDA catalysts. For instance,
with the use of different metals as the active species in MDA
(i.e. Fe), by the addition of promotors to the catalyst that could
lead to less mobile active species, or by the optimisation of
embedding or encapsulating methods to fix the active species
The formation of analogous tetrahedral Mo-oxo sites upon
calcination catalysts allows us to compare the catalytic activity
of both catalysts and to discriminate between the roles of Mo
structure, framework Al and BAS on MDA reaction. The activity
studies showed benzene formation in both catalysts. In contrast
with the traditionally postulated bifunctional mechanism, the
aromatisation observed for Mo/Silicalite-1 indicates that the
presence of BAS is not essential for the formation of benzene.
Therefore, molybdenum carbide species seem to promote
aromatisation provided that good initial Mo dispersion into the
zeolite pores ensures the shape selectivity to aromatics. These
results are in agreement with Kosinov et al. who proposed that
the conversion of methane to benzene might take place in a
[24,42,43]
and prevent migration and sintering.
[20]
monofunctional manner on Mo carbides in the zeolite micro-
[21,37]
pores, or mediated by carbonaceous species.
Such mecha-
ChemCatChem 2019, 11, 1–12
www.chemcatchem.org
8
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
��
These are not the final page numbers!

Full Papers
Experimental Section
and 2–3 mm working distance. Transmission electron microscopy
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
(
TEM) measurements were performed in a JEM 2800 (Scanning)
with a voltage of 200 kV and aperture of 70 and 40 μm. Bright-field
imaging mode was performed using a CCD in high magnification,
whereas lattice resolution imaging mode was carried out using CCD
Dark-field (Z-contrast) imaging in scanning mode using an off-axis
annular detector. The secondary electron signal was acquired
simultaneously with the other TEM images providing topological
information of the sample. Compositional analysis was performed
by X-ray emission detection in scanning mode.
Catalyst Preparation
Silicalite-1 (MFI structure, Si/Al=1), was prepared by hydrothermal
[
44]
synthesis as described by Lobo et al. and subsequently calcined
at 550°C. The resulting zeolite was then treated with ethylenedi-
[
45]
amine following the procedure reported by Wang et al. in order
to generate silanol groups in the structure by the extraction of
4
+
framework Si . ZSM-5 (also MFI structure, Si/Al=15) was pur-
chased from Zeolyst (CBV3024E) in its ammonium form. The proton
form was obtained by calcination in static air at 550°C for 6 h, the
heating rate used was 2°C/min.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
X-ray Absorption Spectroscopy during in situ Calcination
Mo/Silicalite-1 and Mo/H-ZSM-5 with metal loadings of ~4 wt.%
XAFS studies on Mo/MFI zeolites were carried out at the B18
[48]
were prepared by first mixing the zeolite and MoO
9.95%) powders in an agate mortar for 0.5 h. The resulting
(Sigma,
beamline at Diamond Light Source in Harwell, United Kingdom.
The storage beam energy was 3 GeV and the ring current 300 mA.
Mo K-edge spectra (in the range of 19,797 to 21,000 eV) were
collected in transmission mode using ion chamber detectors with a
fast scanning Si (111) double crystal monochromator, with a Mo
foil placed between It and Iref. X-ray beam dimensions at the
3
9
physical mixtures are referred to as the as-prepared catalysts. The
samples were then calcined in air to 700°C for 0.5 h with a 5°C/min
heating rate.
2
sample position were 1×1 mm , whereas the acquisition of each
Catalyst Characterisation
spectra took ~60 s. In situ experiments were performed using a
[49]
setup developed by A. B. Kroner et al.
For the experiments,
Fourier-transform infrared spectroscopy (FTIR) spectra were re-
corded in a Nicolet iS10 spectrometer. Samples were pressed into
self-supporting wafers with a density of c.a. 10 mg/cm . The wafers
4
0
0 mg of the as-prepared catalysts (sieve fractions: 0.425–
.150 mm) were placed in a 3 mm diameter quartz capillary. The
2
sample was then heated to 700°C for 30 min using a hot gas
were dried prior to the measurements by heating them up to
À 1
blower, under 20% O in He flow (GHSV=3000 h and heating
2
2
85°C for 3 h under 70 ml/min He flow. After dehydration, the
rate=5
°
C/min). After 30 min, the sample was cooled to room
sample was cooled down to 150°C under dry He before spectra
temperature under pure He. XAS data was collected during the
heating ramp as well when back at room temperature after
calcination.
collection. Temperature programmed desorption of ammonia
(NH À TPD) was performed using an AutoChem II 2920 micro-
3
meritics instrument equipped with a moisture trap and a thermo-
conductivity detector. Samples were first preactivated by flowing
XAS data analysis was performed using the Demeter software
[50]
pure N and heating up to 550°C for 30 min (5°C/min). The reactor
package. The X-ray absorption fine structure (EXAFS) fitting to
model spectra was done using an amplitude reduction factor of
0.91, which was obtained by fitting the Mo foil reference to
crystallographic data from the ICSD database. K range values used
2
was then cooled down to 100°C for ammonia adsorption which
was run by flowing 1% NH /N until saturation (~1 h). Next, pure N
2
3
2
was flowed for 2 h to remove physisorbed ammonia on the sample.
Finally, ammonia desorption was carried out by increasing the
À 1
in the fitting were between 3 and 11 Å whereas the R range
temperature up to 1100
°
C with a heating rate of 10°C/min. Powder
spanned 1 to 3 Å were used. The Mo K-edge edge position was
taken as the energy at half-step height and the Fourier transformed
(FT) EXAFS data presented are not phase corrected.
X-ray diffraction (XRD) patterns were recorded using a Rigaku
SmartLab X-Ray Diffractometer fitted with a hemispherical analyser.
The measurements were performed using Cu Kα radiation source
(
λ=1.5406 Å) with a voltage of 40 kV, and a current of 30 mA. The
patterns obtained were compared to crystallographic data in the
reference library (ICSD database). Elemental analysis was carried out
using inductively coupled plasma optical emission spectroscopy
Density Functional Theory
[
24]
Based on the outcome of the EXAFS studies on Mo/zeolites, only
monomeric species were considered in the calculations, which were
performed at 0 K.
(
3
ICP-OES) in a Perkin Elmer Optical Emission Spectrometer Optima
300 RL. Nitrogen physisorption measurements were performed on
a Quadrasorb EVO QDS-30 instrument at 77.3 K. The samples were
Firstly, geometry optimisation (all atom positions and cell parame-
ters were allowed to vary) of a single orthorhombic MFI unit cell
with Si/Al=1 (i.e. Silicalite-1) was undertaken. The initial cell was
outgassed at 350°C overnight under vacuum prior to N sorption.
2
The Brunauer-Emmett-Teller equation was used to calculate the
[51]
specific surface area in the pressure range p/p
=0.0006–0.01. The
obtained from the IZA zeolite structure database, and optimised
[52] [53]
0
micropore volume was calculated from the t-plot curve using the
thickness range between 3.5 and 5.4 Å. Thermogravimetric analysis
using the periodic DFT code CP2K at the PBE level of theory
with a DZVP-MOLOPT basis. To give a reasonable starting geometry
2
À
2+
(TGA) was carried out in a TA Q50 instrument. Samples were heated
for monomeric [MoO ] , molybdate ([MoO2] is known to have a
4
2
À
up to 950°C using a rate of 5°C/min under an air flow of 60 mL/
min. Raman spectra were acquired on ULTRA at the Central Laster
Facility.[46,47] The measurements were carried out using 400 nm
laser to excite the sample (placed in quartz window holders).
Toluene impregnated H-ZSM-5 was used for calibration of the
detected signals. Scanning electron microscopy (SEM) analysis was
performed using a Zeiss ultra 55 Field emission electron micro-
scope. Compositional analysis and low-resolution imaging were
carried out with accelerating voltage of 20 kV, 30–60 μm aperture
and 7–8 mm working distance. High-resolution images were also
taken with an accelerating voltage of 1.6 kV, 20–30 μm aperture
similar OÀ MoÀ O bond angle to [MoO ] ) was geometry optimised
4
as an isolated cluster using the aperiodic Orca code (PBE functional
2
+
and SVP basis). Then, the optimised [MoO2] moiety was grafted
to two framework oxygen atoms (O ) across the T5 site forming
F
2
À
[MoO4] . Several trial positions were optimised and the results
described here relate to the global minimum established in each
case. This structure was geometry optimised at the same level of
theory as the guest-free Silicalite-1 framework (i.e. PBE and the
DZVP-MOLOPT). The interaction of Mo species with different
possible types silanols – i.e. nests, vicinal and geminal, was also
simulated. For the nests, Si was removed from the T5 sites in
ChemCatChem 2019, 11, 1–12
www.chemcatchem.org
9
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
��
These are not the final page numbers!

Full Papers
optimised Silicalite-1 structure charge compensating the four Conflict of Interest
dangling SiÀ O bonds generated with terminating protons. The
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
resultant structure was then geometry optimised at the same level
The authors declare no conflict of interest.
2
+
of theory used throughout. Following this, [MoO2]
was re-
introduced to the same position as before, removing the two
protons from the framework oxygens bound to Mo in order to
mimic the condensation reaction that is proposed to occur when
the initial mobile Mo-oxo species. Vicinal and non-vicinal silanol
defects were simulated by removing Si from the T5 and T9 sites in
both the vicinal and non-vicinal case. [MoO2] was then grafted
across the defects, removing two protons to mimic condensation
reactions. The two structures were geometry optimised at the same
level of theory used throughout these simulations.
Keywords: MDA · Mo/zeolite · in situ XAS · DFT
2
+
[
1] B. M. Weckhuysen, D. Wang, M. P. Rosynek, J. H. Lunsford, J. Catal. 1998,
75, 347–351.
1
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
[2] S. Ma, X. Guo, L. Zhao, S. Scott, X. Bao, J. Energy Chem. 2013, 22, 1–20.
[3] Z. R. Ismagilov, E. V. Matus, L. T. Tsikoza, Energy Environ. Sci. 2008, 1,
526–541.
Finally, the Mo/Silicalite-1 models were compared with that of Mo/
ZSM-5 by introducing a single Al to each of the 12 crystallo-
graphically distinct framework T-site present in orthorhombic MFI.
Brønsted acidic protons were not included in these simulations in
order to mimic the liberation of water which is proposed to occur
on binding of the initial Mo-oxo species to the walls of the
framework. The Mo-oxo species in H-ZSM-5 remained in the same
position, grafted across the T5 site.
[
4] J. J. Spivey, G. Hutchings, Chem. Soc. Rev. 2014, 43, 792–803.
[5] F. Solymosi, J. Cserényi, A. Szöke, T. Bánsági, A. Oszkó, J. Catal. 1997,
165, 150–161.
[6] K. Sun, D. M. Ginosar, T. He, Y. Zhang, M. Fan, R. Chen, Ind. Eng. Chem.
Res. 2018, 57, 1768–1789.
[7] Y. Xu, Y. Shu, S. Liu, J. Huang, X. Guo, Catal. Lett. 1995, 35, 233–243.
[
8] C. H. L. Tempelman, E. J. M. Hensen, Appl. Catal. B 2015, 176–177, 731–
39.
9] C. Karakaya, H. Zhu, R. J. Kee, Chem. Eng. Sci. 2014, 123, 474–486.
7
[
[10] K. S. Wong, J. W. Thybaut, E. Tangstad, M. W. Stöcker, G. B. Marin,
Microporous Mesoporous Mater. 2012, 164, 302–312.
[11] D. Ma, Y. Shu, M. Cheng, Y. Xu, X. Bao, J. Catal. 2000, 194, 105–114.
MDA Catalytic Performance Tests
[
12] J. Shu, A. Adnot, B. P. A. Grandjean, Ind. Eng. Chem. Res. 1999, 38, 3860–
0.6 g of sieved catalyst (150–425 μm sieve fractions) was introduced
3867.
into a tubular quartz reactor (internal diameter=0.7 mm). The
sample was fixed in the isothermal zone of the oven by quartz
wool. A total gas flow of 30 mL/min was fed by means of mass flow
[13] A. Sarıog, A. Mer, T. Savaçı, A. Aye, E.-E. Ae, V. Thu, H. Ae, G. Sapaly, A.
Younès, B. Taârit, Catal. Lett. 2007, 118, 123–128.
14] S. Liu, L. Wang, R. Ohnishi, M. Ichikawa, J. Catal. 1999, 181, 175–188.
[
À 1
[15] H. S. Lacheen, E. Iglesia, Phys. Chem. Chem. Phys. 2005, 7, 538–547.
controllers (GHSV=1500 h ).
[
[
16] M. Marczewski, H. Marczewska, K. Mazowiecka, React. Kinet. Catal. Lett.
1995, 54, 81–86.
17] J. P. Tessonnier, B. Louis, S. Rigolet, M. J. Ledoux, C. Pham-Huu, Appl.
Catal. A 2008, 336, 79–88.
The as-prepared physical mixtures were first calcined under 20%
O /He flow by heating up to 700°C for 30 min using a heating rate
2
of 5°C/min. After flowing pure Ar for 30 min, methane dehydroar-
[
[
18] F. Solymosi, A. Erdöhelyi, A. Szöke, Catal. Lett. 1995, 32, 43–53.
19] X. Guo, G. Fang, G. Li, H. Ma, H. Fan, L. Yu, C. Ma, X. Wu, D. Deng, M.
Wei, D. Tan, R. Si, S. Zhang, J. Li, L. Sun, Z. Tang, X. Pan, X. Bao, Science
2014, 344, 616–619.
omatisation reaction was initiated by flowing 50% CH /Ar. The
4
reaction products were analysed by an online mass spectrometer
(OmniStar GSD 320O1). All MS data intensity presented in this
manuscript are normalised to the Ar signal and are plotted in
logarithmic scale.
[20] N. Kosinov, F. J. A. G. Coumans, E. A. Uslamin, A. S. G. Wijpkema, B.
Mezari, E. J. M. Hensen, ACS Catal. 2017, 7, 520–529.
[
21] N. Kosinov, A. Wijpkema, E. Uslamin, R. Rohling, F. Coumans, B. Mezari,
A. Parastaev, A. Poryvaev, M. Fedin, E. Pidko, E. J. M. Hensen, Angew.
Chem. Int. Ed. 2018, 57, 1016–1020.
To follow the deactivation process, Mo/MFI samples were collected
at different reaction times (7 min, 90 min and 10 h) after cooling
down the reactor under Ar flow. The resulting samples are denoted
by adding the reaction time after the sample code. For example,
Mo/Silicalite-1 sample reacted for 90 min is identified as Mo/
Silicalite-1(90 min).
[22] N. Kosinov, A. E. Uslamin, J. A. G. F. Coumans, S. G. A. Wijpkema, Y. R.
Rohling, J. M. E. Hensen, ACS Catal. 2018, 8, 8459–8467.
[23] I. Lezcano-González, R. Oord, M. Rovezzi, P. Glatzel, S. W. Botchway,
B. M. Weckhuysen, A. M. Beale, Angew. Chem. Int. Ed. 2016, 55, 5215–
5219.
[
[
[
24] M. Agote-Arán, A. B. Kroner, H. U. Islam, W. A. Sławin, D. S. Wragg, I.
Lezcano-González, ChemCatChem 2019, 11, 473–480.
25] S. E. Shadle, B. Hedman, K. Hodgson, E. I. Solomon, Inorg. Chem. 1994,
33, 4235–4244.
Acknowledgements
26] W. Li, G. D. Meitzner, R. W. Borry, E. Iglesia, J. Catal. 2000, 191, 373–383.
[27] C. Brookes, P. P. Wells, G. Cibin, N. Dimitratos, W. Jones, D. J. Morgan, M.
Bowker, ACS Catal. 2014, 4, 243–250.
[28] H. Negishi, S. Negishi, Y. Kuroiwa, N. Sato, S. Aoyagi, Phys. Rev. B 2004,
This study received financial support from EPSRC (MAA, a UCL
Impact PhD award, Diamond Light Source and Johnson Matthey).
EPSRC is also acknowledged for sponsoring the UK Catalysis Hub
which provided resources and support via grant, EP/K014714/1
and for an Early Career Fellowship for AMB (EP/K007467/1). This
research used resources from RCaH and Johnson Matthey
Technology Centre. Diamond Light Source is kindly thanked for
the awarded beamtime (SP11623) whereas Diego Gianolio, Emma
Gibson, Scott Rogers and George Tierney are kindly acknowledged
for the assistance during the synchrotron experiments. Nicolas
Grosjean is also acknowledged for the assistance during catalytic
testing.
69, 064111.
[29] D. Zhou, D. Ma, X. Liu, X. Bao, J. Chem. Phys. 2001, 114, 9125–9129.
[
[
30] H. Guesmi, R. Gryboś, J. Handzlik, F. Tielens, Phys. Chem. Chem. Phys.
2014, 16, 18253–18260.
31] C. A. Johnson, K. M. Thomas, Fuel 1984, 63, 1073–1080.
[32] P. K. Chu, L. Li, Mater. Chem. Phys. 2006, 96, 253–277.
33] A. C. Ferrari, J. Robertson, Phys. Rev. B 2000, 61, 14095–14107.
[34] A. Sadezky, H. Muckenhuber, H. Grothe, R. Niessner, U. Pö Schl, Carbon
005, 43, 1731–1742.
[
2
[35] J. Hoekstra, A. M. Beale, F. Soulimani, M. Versluijs-Helder, J. W. Geus,
L. W. Jenneskens, J. Phys. Chem. 2015, 119, 10653–10661.
[36] B. Dippel, H. Janderb, J. Heintzenberga, Phys. Chem. Chem. Phys. 1999,
1, 4707–4712.
[37] I. Vollmer, G. Li, I. Yarulina, N. Kosinov, E. J. Hensen, K. Houben, D.
Mance, M. Baldus, J. Gascon, F. Kapteijn, Catal. Sci. Technol. 2018, 8,
916–922.
ChemCatChem 2019, 11, 1–12
www.chemcatchem.org
10
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
��
These are not the final page numbers!

Full Papers
[
38] N. Kosinov, F. J. A. G. Coumans, G. Li, E. Uslamin, B. Mezari, A. S. G.
[48] A. J. Dent, G. Cibin, S. Ramos, A. D. Smith, S. M. Scott, L. Varandas, M. R.
Pearson, N. A. Krumpa, C. P. Jones, P. E. Robbins, J. Phys. Conf. Ser. 2009,
190, 012039.
[49] A. B. Kroner, K. M. H. Mohammed, M. Gilbert, G. Duller, L. Cahill, P.
Leicester, R. Woolliscroft, E. J. Shotton, AIP Conf. Proc. 2016, 1741,
030014.
[50] B. Ravel, M. Newville, J. Synchrotron Radiat. 2005, 12, 537–541.
[51] Www.Iza-Structure.Org, n.d., 01, 212–213.
[52] J. VandeVondele, M. Krack, F. Mohamed, M. Parrinello, T. Chassaing, J.
Hutter, Comput. Phys. Commun. 2005, 167, 103–128.
[53] J. P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 1996, 78, 1396–1396.
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
Wijpkema, E. A. Pidko, E. J. M. Hensen, J. Catal. 2017, 346, 125–133.
[39] Y. H. Kim, R. W. Borry, E. Iglesia, Microporous Mesoporous Mater. 2000,
5–36, 495–509.
40] H. Ma, R. Kojima, R. Ohnishi, M. Ichikawa, Appl. Catal. A 2004, 275, 183–
87.
3
[
1
[41] S. H. Morejudo, R. Zanón, S. Escolástico, I. Yuste-Tirados, H. Malerød-
Fjeld, P. K. Vestre, W. G. Coors, A. Martínez, T. Norby, J. M. Serra, C.
Kjølseth, Science 2016, 353, 563–566.
[42] L. Ma, I. Jia, X. Guo, L. Xiang, Chin. J. Catal. 2014, 35, 108–119.
[43] D. Farrusseng, A. Tuel, New J. Chem. 2016, 40, 3933–3949.
[
[
[
[
44] D. D. Kragten, J. M. Fedeyko, K. R. Sawant, J. D. Rimer, D. G. Vlachos, R. F.
Lobo, M. Tsapatsis, J. Phys. Chem. B 2003, 107, 10006–10016.
45] Y. Bu, Y. Wang, Y. Zhang, L. Wanga, Z. Mi, W. Wub, E. Min, S. Fu, Catal.
Commun. 2007, 8, 16–20.
46] “CLF Ultra,” can be found under https://www.clf.stfc.ac.uk/Pages/Ultra.-
aspx
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
Manuscript received: July 1, 2019
Revised manuscript received: September 1, 2019
Accepted manuscript online: September 13, 2019
Version of record online: ■■■, ■■■■
47] P. Matousek, M. Towrie, C. Ma, W. M. Kwok, D. Phillips, W. T. Toner, A. W.
Parker, J. Raman Spectrosc. 2001, 32, 983–988.
ChemCatChem 2019, 11, 1–12
www.chemcatchem.org
11
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
��
These are not the final page numbers!

Full Papers
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
FULL PAPERS
Coordination in Zeolites: X-ray ab-
sorption spectroscopy and density
functional theory calculations were
performed to investigate the Mo
structure on Mo/Silicalite-1 (Si/
Al=1) and Mo/H-ZSM-5 (Si/Al=5)
catalysts. These studies combined
with catalytic activity measurements
bring new insight to unravel the
roles of Mo speciation, framework Al
and Brønsted acid sites on the
methane dehydroaromatisation
reaction.
Dr. M. Agote-Arán, Dr. R. E. Fletcher,
Dr. M. Briceno, Dr. A. B. Kroner, Dr. I. V.
Sazanovich, Prof. B. Slater, Dr. M. E.
Rivas, Dr. A. W. J. Smith, Dr. P. Collier,
Dr. I. Lezcano-González*, Prof. A. M.
Beale*
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
1
– 12
Implications of the Molybdenum
Coordination Environment in MFI
Zeolites on Methane Dehydroaro-
matisation Performance