Active Sites in Olefin Metathesis over Supported Molybdena Catalysts

Article abstract of DOI:10.1002/cctc.201500725

Metathesis of propene to ethene and 2-butenes was studied over a series of MoO_x/SBA-15 catalysts (molybdenum oxide supported on mesoporous silica SBA-15; Mo loading 2.1-13.3 wt %, apparent Mo surface density 0.2-2.5 nm^-2). The catalysts have been prepared by an ion exchange technique. Nitrogen adsorption, ¹H MAS-NMR, Raman, and FTIR spectroscopies were applied to characterize the catalysts. Adsorption of the reactant propene and the probe molecule NH₃ was studied by in situ FTIR spectrometry microcalorimetry and temperature-programmed desorption. Irrespective of the loading, only ≈1 % of the Mo atoms in the MoO_x/SiO₂ catalysts transform into active carbene (Mo=CHR) sites catalyzing propene metathesis. Isolated, distorted molybdenum di-oxo species in close vicinity to two silanol groups have been shown to be the precursor of the active site. Targeted active site creation by pretreatment with methanol resulted in an increase in initial catalytic activity by a factor of 800. Targeted active site creation: Only ≈1 % of isolated distorted Mo di-oxo species that are in close vicinity to two silanol groups have been shown to be the precursor of the active carbene (Mo=CHR) sites on MoO_x/SiO₂ catalysts for propene metathesis. Targeted active site creation resulted in an increase in initial catalytic activity by a factor of 800.

Full text of DOI:10.1002/cctc.201500725

DOI: 10.1002/cctc.201500725
Full Papers
Active Sites in Olefin Metathesis over Supported
Molybdena Catalysts
Kazuhiko Amakawa,[a] Jutta Krçhnert,[a] Sabine Wrabetz,[a] Benjamin Frank,[a]
Felix Hemmann,[b, c] Christian Jäger,[b] Robert Schlçgl,[a] and Annette Trunschke*[a]
Metathesis of propene to ethene and 2-butenes was studied
over a series of MoOx/SBA-15 catalysts (molybdenum oxide
supported on mesoporous silica SBA-15; Mo loading 2.1–
13.3 wt%, apparent Mo surface density 0.2–2.5 nmÀ2). The cat-
alysts have been prepared by an ion exchange technique. Ni-
ture-programmed desorption. Irrespective of the loading, only
ꢀ1% of the Mo atoms in the MoOx/SiO2 catalysts transform
into active carbene (Mo=CHR) sites catalyzing propene meta-
thesis. Isolated, distorted molybdenum di-oxo species in close
vicinity to two silanol groups have been shown to be the pre-
cursor of the active site. Targeted active site creation by pre-
treatment with methanol resulted in an increase in initial cata-
lytic activity by a factor of 800.
1
trogen adsorption, H MAS-NMR, Raman, and FTIR spectroscop-
ies were applied to characterize the catalysts. Adsorption of
the reactant propene and the probe molecule NH3 was studied
by in situ FTIR spectrometry microcalorimetry and tempera-
Introduction
The current yearly production of C2–C4 olefins exceeds
200 million tons. Propene production by cross-metathesis of
ethene and 2-butenes is an economic strategy to satisfy the in-
creasing propene demand,[1] where silica-supported tungsten
oxide catalysts are currently employed at high temperature
(>573 K).[2] It has been considered to substitute the W-based
catalysts by more active molybdenum oxides catalysts operat-
ing at mild conditions.[3] Supported molybdenum oxide cata-
lysts exhibit the additional advantage that regeneration is pos-
sible. Recent studies suggest that Mo oxides supported on
acidic materials (e.g., silica-alumina) are the most promising
candidates.[4]
wherein MoVI exclusively present in the freshly pretreated cata-
lyst is reduced by propene to MoIV, which is followed by oxida-
tive addition of another propene molecule to yield a surface
MoVI=CHR species (as exemplified in Scheme 1a).[6b] The mech-
anism postulates Brønsted acidity and oxidation ability of the
catalyst. Though the general route of carbene generation was
clarified, the structural identification of the relevant surface
molybdenum oxide precursor species remains an elusive chal-
lenge, as it demands discrimination of a small minority (ꢀ1%)
from the majority of spectator species. As for alumina-support-
ed molybdenum oxide catalysts the nature of the precursor
structure has been controversially discussed, favoring both
monomeric,[7] as well as oligomeric[8] surface MoOx species.
Here, we study correlations between surface density of molyb-
denum oxide species, acid-base properties and catalytic prop-
erties taking a series of silica-supported molybdenum oxide
catalysts with different Mo loading as basis. Combined applica-
tion of spectroscopic and functional analyses allows us to fur-
ther elucidate the nature of the active site precursor in olefin
metathesis over supported molybdenum oxide catalysts.
Metal-carbene species (M=CHR) are the active sites in olefin
metathesis[5] and they are generated in situ through surface re-
actions between the supported metal oxide and the olefin
itself. Apparently, only a small fraction of metal atoms (ꢀ1%
in silica-supported molybdena catalysts) are involved in active
site formation.[6] Recently, we proposed a mechanism of the
formation of Mo=CHR sites in silica-supported molybdena,
[a] Dr. K. Amakawa, J. Krçhnert, Dr. S. Wrabetz, Dr. B. Frank,
Prof. Dr. R. Schlçgl, Dr. A. Trunschke
Results and Discussion
Department of Inorganic Chemistry
Fritz-Haber-Institut der Max-Planck-Gesellschaft
Faradayweg 4–6, 14195 Berlin (Germany)
E-mail: trunschke@fhi-berlin.mpg.de
Propene self-metathesis at 323 K over monolayer-type molyb-
dena supported on mesoporous silica SBA-15 (MoOx/SBA-15)
was studied as a model for the desired reverse reaction (i.e.,
propene production). The surface Mo density was varied
changing the Mo content (Table 1) with the aim to identify the
relevant pre-catalyst species that give active carbenes (Mo=
CHR). A detailed structural characterization of the catalyst
series by nitrogen adsorption, XRF, XRD, SEM-EDX, IR, Raman,
UV/vis, O K-edge NEXAFS, Mo K-edge XANES/EXAFS, and DFT
calculations has been reported elsewhere.[9]
[b] Dr. F. Hemmann, Prof. Dr. C. Jäger
BAM Federal Institute for Materials Research and Testing
Richard-Willstätter-Strasse 11, 12489 Berlin (Germany)
[c] Dr. F. Hemmann
Department of Chemistry
Humboldt-Universität zu Berlin
Brook-Taylor-Strasse 2, 12489 Berlin (Germany)
Supporting Information for this article is available on the WWW under
http://dx.doi.org/10.1002/cctc.201500725.
ChemCatChem 2015, 7, 4059 – 4065
4059
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Scheme 1. Suggested mechanisms for the generation of a MoIV-carbene site in propene metathesis a) without pretreatment and b) with methanol pretreat-
ment.
Table 1. Properties of MoOx/SBA-15 catalysts.
[e]
[g]
[h]
Mo
Surface density
As
Am
Vp
dp
loading[a] Mo[b] Isolated SiOH[c,d]
[wt%]
[nmÀ2
]
[nmÀ2
]
[m2 gÀ1
]
[m2 gÀ1
]
[%][f] [mlgÀ1
]
[nm]
0
0
1.6[c]
859
637
554
490
556
332
261
164
127
135
96
36
31
28
28
21
13
1
7.5
7.1
7.1
7.1
7.2
7.4
2.1
5.1
6.6
9.7
13.3
0.21
0.58
0.85
1.09
2.51
1.1[d]
0.79
0.71
0.61
0.78
0.55
0.88[d]
0.68[d]
0.39[d]
0.07[d]
36
[a] by XRF. [b] Mo loading (at%) divided by As. [c] by TG. [d] by IR at the
dehydrated state using relative heights of the silanol peak at 3745 cmÀ1
.
[e] micropore (<0.9 nm of width) surface estimated by t-plot method.
[f] Am divided by AS. [g] at p/p0 =0.95. [g] at the dehydrated state. [h] esti-
mated by NLDFT approach.
Figure 1. Propene metathesis performance of MoOx/SBA-15 at 323 K mea-
sured after 15–21 h time on stream. top) Metathesis rate and surface con-
centration of active carbene sites (Mo=CHR). bottom) Turnover frequency
(TOF). Error bars are estimated by two repeated measurements using fresh
and regenerated catalysts. The catalysts were pretreated or regenerated in
20% O2 at 823 K for 0.5 h.
Textural properties of the MoOx/SBA-15 catalysts are sum-
marized in Table 1. The introduction of surface molybdenum
oxide species at the expense of silanol groups results in a de-
crease of the specific surface area As, while preserving the
large mesopores (ꢀ7 nm) of SBA-15. The decrease in the mi-
cropore surface area Am suggests a preferential anchorage of
surface molybdena in the micropores.
dominant factor for activity. Dividing the rate by the active site
density yields the turnover frequency (TOF; Figure 1, bottom),
which shows a significant increase with increasing Mo density.
The Mo density evidently affects both the probability of active
site formation and the intrinsic metathesis activity (i.e., TOF) of
the Mo=CHR species (Figure 1). Propene adsorption studies
were performed to explain this pattern. The carbene genera-
tion process was traced upon propene adsorption by in situ
FTIR spectroscopy and microcalorimetry (Figure 2).
The surface Mo density shows a great impact on the pro-
pene metathesis activity (Figure 1, top red, line) The time trend
of the fresh and regenerated catalysts that exhibits an increase
in the propene metathesis with time on stream (constant activ-
ity after 15–21 h) is presented in the Supporting Information
(Figure S1)
The surface concentration of active carbene species
(Figure 1, top, blue line) determined by post-reaction titration
by [D4]-ethene metathesis[6b] is at most 1% of all surface Mo
atoms and coarsely follows the trend of the activity (Figure 1,
top, red line), indicating that the number of active sites is the
Bands due to CÀH vibrations (stretching : 2983, 2939,
2880 cmÀ1; deformation: 1465, 1455, 1389, 1375 cmÀ1) and
n(C=O) at 1668 cmÀ1 in the infrared spectra (Figure 2a) are as-
signed to isopropoxide and acetone, respectively.[6b] Formation
of these intermediates indicates the protonation of propene to
ChemCatChem 2015, 7, 4059 – 4065
www.chemcatchem.org
4060
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Figure 2. Adsorption of propene on MoOx/SBA-15 with different Mo loadings
at 323 K. a) FTIR spectra collected after propene dosing at 3 hPa for 18 h
and subsequent evacuation. b) Differential heat of propene adsorption as
a function of coverage determined by microcalorimetry. Surface Mo density
(Moatoms nmÀ2) is indicated close to the data.
Figure 3. Study on acidity and hydroxyl groups in MoOx/SBA-15. a) 1H MAS-
NMR spectra of SBA-15 and MoOx/SBA-15 (1.27 Moatoms nmÀ2). The spectrum
of MoOx/SBA-15 is magnified by factor of 5. b) Density of isolated silanol
groups and ammonia adsorption sites estimated by IR. c) Schematic illustra-
tion of the suggested model for Brønsted acid sites. The original IR spectra
are presented in the Supporting Information (Figure S3). The catalysts were
pretreated in pure O2 at 823 for 0.5 h.
isopropoxide and its subsequent oxidation to acetone, where
Brønsted acidity and oxidation ability of surface molybdena
are involved (Scheme 1a; 1 to 4). This finding agrees with our
previous study that included only a single catalyst with Mo
density of 1.1 MonmÀ2.[6b] However, the intensity of the
observed bands does not follow the Mo surface density.
The quantity and strength of the propene adsorption sites
were evaluated by microcalorimetry (Figure 2b). Heat of ad-
sorption as well as amount of adsorbed propene do also not
follow the trend in surface Mo density, but increase in the
order 0.58Moffi2.5Mo<1.1Mo. A similar tendency is observed
in the overall rate and the number of surface carbene species
(Figure 1, top). Such a positive correlation between the surface
concentration of Mo=CHR species in catalysis and the number
as well as the strength (i.e., heat of adsorption) of propene ad-
sorption sites is also shown in the Supporting Information (Fig-
ure S2). The observation suggests the involvement of the cor-
responding propene adsorption sites in the generation of
active species. The surface concentration of Mo=CHR is signifi-
cantly lower than the overall number of propene adsorption
sites (Figure S2b), confirming that only a fraction of the
adsorbed propene yields the active sites.
cies at the expense of surface silanol groups at all loadings. A
fraction of the residual silanol groups is in hydrogen bonding
with these molybdenum oxide species. The fraction of OH
groups adjacent to molybdenum oxide species becomes small-
er with increasing Mo surface density.
Probing the acid sites by ammonia adsorption monitored by
IR reveals a monotonous increase of the surface concentration
of both Brønsted and Lewis acid sites (coordinatively unsatu-
rated Mo centers) with increasing Mo density, where at most
10% of total Mo atoms serve as Brønsted acid sites (Figure 3b,
and S3). Note that extinction coefficients used for calculating
the surface concentration of Brønsted and Lewis acid sites
have been taken from literature.[11] Consequently, the analysis
represents just a rough estimation.
Bi-functionality of the catalyst in terms of acidity (protona-
tion of propene to isopropoxide) and redox activity (oxidation
of isopropoxide to acetone) is considered to be indispensable
for the carbene generation.[6b] To identify the nature of the
Brønsted acid sites, hydrogen species were characterized by
1H NMR (Figure 3). Although bare SBA-15 possesses virtually
isolated silanol groups only (Figure 3a, 1.75 ppm),[10] the load-
ing of surface molybdenum oxide species leads to an occur-
rence of hydrogen-bonded silanol groups (Figure 3a, broad
band at 2–5 ppm),[10] and a perturbation of isolated silanol (i.e.,
shift to 1.93 ppm). The results are consistent with the structural
analysis by Raman, IR, UV/vis, Mo K-edge XANES/EXFAS, O K-
edge NEXAFS, and DFT calculations,[9] which show the forma-
tion of two-fold anchored tetrahedral di-oxo (SiÀOÀ)2Mo(=O)2
structures as the predominant surface molybdenum oxide spe-
Provided no indication for the presence of molybdenol
groups by 1H NMR (Figure 3a; MoÀOH likely occurs below
1 ppm[12]) and by other spectroscopic characterization,[9] silanol
sites in the vicinity of surface molybdena species are the most
likely and actually the only possible source of Brønsted acidity.
Analogously to Brønsted acid sites in silica-alumina materi-
als,[13] the conjugated Brønsted base ammonia appears to be
protonated in a bonding configuration that is illustrated in Fig-
ure 3c. The hydrogen atom of the silanol group in close vicini-
ty to a surface molybdena species interacts with the nitrogen
atom of ammonia, because it bears a partial positive charge
due to the interaction of the oxygen atom in the OH species
with the adjacent Lewis acidic Mo site. At the same time, the
ammonia molecule undergoes hydrogen-bonding to a terminal
oxygen atom of the surface molybdenum oxide species. Such
ChemCatChem 2015, 7, 4059 – 4065
www.chemcatchem.org
4061
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
a configuration may have an unknown impact on the extinc-
tion coefficient of the ammonium species, and, consequently,
on the quantification of Brønsted acid sites by ammonia ad-
sorption. The uncertainty is reflected by the observation that
the overall acid strength (i.e., including the Lewis sites) esti-
mated by temperature-programmed desorption of ammonia
(Figure S4) appears to be similar for all catalysts, i.e., the trends
in ammonia adsorption seem to be dominated by weak
adsorption of NH3 on Lewis acid sites.
that equip Brønsted acidity and a trapping function for the by-
product acetone. Following the sequential reactions discussed
above, a mono-anchored tetrahedral MoVI-propylidene is ex-
pected to occur (Scheme 1a, 7), which may further undergo
the alkoxide formation (Scheme 1a, 8) and transformation to
MoVI-methylidene by propene metathesis (Scheme 1a, 9). The
strained molybdena species occur at relatively high Mo densi-
ty,[9] while it still needs adjacent silanol sites to form the active
sites. This accounts for the presence of an optimum Mo densi-
ty at sub-monolayer coverage (Figure 1). The low intrinsic ac-
tivity at low Mo density may be due to increased steric hin-
drances by abundant silanol groups surrounding the carbene
center and blocking the access of propene[17] and/or subtle
variations in the geometric configuration (e.g., bond lengths
and angles),[16,17] as suggested by theoretical studies.[18]
1
In summary, H NMR and FTIR indicate that Brønsted acidity
in silica-supported MoOx arises from silanol groups that are lo-
cated in close proximity to surface molybdenum oxide species.
Propene adsorption allows the semi-quantitative estimation of
relevant acidity when taking into account that propene ad-
sorption sites exhibit a heat of adsorption higher than
30 kJmolÀ1. The propene uptake represents at most 3% of
total Mo atoms and shows a non-linear dependence to the Mo
density (Figure S2, and 2b; ꢀ0.03 C3H6 molecules nmÀ2 for
1.1 Moatoms nmÀ2), in which the maximum has been found at
a Mo density of 1.1 atomsnmÀ2, which corresponds to the
most active catalyst.
The population balance of surface molybdenum oxide spe-
cies and surrounding silanol groups is reflected in the n(Mo=O)
stretching frequency in vibrational analysis. The original spec-
tra[9] are presented in the Supporting Information (Figure S6).
With increasing molybdenum loading, the ns(Mo=O) Raman
band at 980–997 cmÀ1 becomes sharper and is blue-shifted
(Figure S6a), which is also observed in the corresponding infra-
red spectra (Figure S6b). The n(MoÀOÀSi) band at 926–
943 cmÀ1, visible in the infrared spectra (Figure S6b), is also
blue-shifted. The blue shift of the bands with increasing Mo
loading indicates the decrease of hydrogen bonding due to
surrounding silanol groups since fewer silanol groups are avail-
able with increasing molybdenum oxide loading. A clear struc-
To transform propene to acetone (Scheme 1a; 3 to 4), the
isopropoxide species formed at a Brønsted acidic silanol site
needs to migrate to a MoVI center that exhibits oxidation abili-
ty. Such migration of surface alkoxide species across silanol
sites and molybdenum oxide sites was observed by in situ IR
using isotope labeling.[14] However, the underlying mechanism
remains unclear. The IR data show an increased acetone forma-
tion at high Mo density (Figure 2a; n(C=O) at 1668 cmÀ1), sug-
gesting a higher probability of the migration, which is in line
with the increased reactivity of the surface molybdenum oxide
species, owing to the increased strain at the anchoring MoÀOÀ
Si bonds that has been observed by O K-edge NEXAFS in com-
bination with DFT calculations.[9] Based on the IR observation
and the structural characterization, we propose that the migra-
tion involves a breaking of a MoÀOÀSi bond and concurrent
formation of a SiÀOÀSi bond (Scheme 1a, 2 to 3), wherein
a high strain at MoÀOÀSi emerging at high Mo density is ex-
pected to be a key driving force of the rearrangement event.
The formed acetone needs to desorb to allow access of an-
other propene molecule to the MoIV center that leads to the
formation of active MoVI=CHR sites (Scheme 1a, 4 to 5). These
processes are expected to occur simultaneously, as it was
show recently for the exchange of pyridine at acid sites.[15] The
desorption of acetone upon heating after catalysis (Figure S5)
suggests a reversible capture of acetone by hydrogen-bonding
at silanol groups[16] in the vicinity to the molybdena sites.[6b]
The low activity at the highest Mo density (Figure 1a) may be,
therefore, related to the low silanol density due to the exten-
sive coverage of the support by molybdenum oxide species
(Figure 3b).
ture-activity relationship exhibiting
a volcano-type profile
(Figure 4) suggests the presence of an optimum degree of hy-
Figure 4. Correlation between the n(Mo=O) frequency detected in Raman/IR
analysis and propene metathesis activity of MoOx/SBA-15 at 323 K and at
15–21 h time on stream. The catalysts were pretreated or regenerated in
20% O2 at 823 K for 0.5 h. The vibrational spectra are presented in the
Supporting Information (Figure S6).
drogen bonding of surface molybdena species, supporting the
hypothesis that molybdena-silanol ensembles are precursors of
the active sites.
Summarizing the structural assignments for the required
functions for carbene generation on silica-supported molybde-
num oxide catalysts, a model for the catalyst precursor is envis-
aged. This model (Scheme 1a, 1) is characterized by a (SiÀOÀ)2-
Mo(=O)2 structure exhibiting high strain at the MoÀOÀSi
bonds[9] surrounded by at least two adjacent silanol groups
Having understood the mechanism of carbene generation,
controlled active site creation is feasible. To enhance the for-
mation of coordinatively unsaturated MoIV sites, the catalyst
was first treated with methanol at 523 K followed by a heat
treatment in Ar at 823 K. The reduction of the abundant di-oxo
(SiÀOÀ)2Mo(=O)2 species by methanol does not require Brønst-
ChemCatChem 2015, 7, 4059 – 4065
www.chemcatchem.org
4062
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Experimental Section
Preparation of the MoOx/SBA-15 catalysts
The MoOx/SBA-15 catalysts (molybdena supported on mesoporous
silica SBA-15; Mo loading 2.1–13.3 wt%) were prepared by an ion
exchange protocol.[22] The details of the preparation were de-
scribed elsewhere.[9] In brief, freshly synthesized metal-free SBA-15
(internal sample ID 8233) was functionalized with propylammoni-
um chloride using (3-aminopropyl)trimethoxysilane followed by
treatment with hydrochloric acid. Then, the functionalized SBA-15
powder was stirred in an aqueous solution containing the desired
amount of ammonium heptamolybdate to perform an anion ex-
change. After washing with water and filtration, the material was
dried and calcined at 823 K in air, yielding supported MoOx/SBA-15
with the actual Mo loadings of 2.1, 5.1, 6.6, 9.7, and 13.3% (internal
sample ID 8442, 8440, 11 054, 8438, 8441, respectively). The proper-
ties of the catalysts are summarized in Table 1. Additionally, a SBA-
15 (BET surface are=833 m2 gÀ1, internal ID 8261) and a MoOx/SBA-
15 (BET surface are=532 m2 gÀ1, 10.8 Mo%, 1.27 MonmÀ2, internal
ID 13578) were prepared using the same procedure but in differ-
ent batches for the 1H NMR study.
Figure 5. Propene metathesis activity of a MoOx/SBA-15 catalyst
(0.85 Moatoms nmÀ2) after different pretreatment procedures. 1) Standard pre-
treatment (20% O2, 823 K, 0.5 h). 2) Methanol pretreatment (4% CH3OH/Ar,
523 K, 0.5 h) and subsequent desorption (Ar, 823 K, 0.5 h) after the standard
pretreatment. The inset shows signal of mass spectrometer for m/e=69
(pentene) and m/e=56 (butenes) at the initial period of the reaction after
methanol pretreatment.
Characterization of the MoOx/SBA-15 catalysts
ed acidity,[19] leaving coordinatively unsaturated MoIV sites after
the desorption of by-products in the post-treatment in Ar
(Scheme 1b, 1 to 4). This procedure increases the initial catalyt-
ic activity by a factor of 800 (Figure 5). The high initial activity
allows detection of the temporal formation of pentene at the
initial period of the reaction (Figure 5, inset), which indicates
the occurrence of MoVI-propylidene species (Scheme 1b, 5)
that corroborates the suggested mechanism of carbene forma-
tion. A continuous deactivation was observed in the case of
the methanol pretreatment, implying that the anticipated two-
fold anchored MoVI-carbene (Scheme 1b, 6) is less stable than
the mono-hapto structure (Scheme 1a, 9). Indeed, the gradual
deactivation of the two-fold anchored tetrahedral MoVI-carbene
was reported previously,[20] whereas mono-anchored Schrock-
type MoVI-alkylidene on dehydroxylated silica, which is similar
to the anticipated structure in the case of the standard pre-
treatment (Scheme 1a, 9), exhibits a stable activity.[21]
Nitrogen adsorption: Nitrogen adsorption was carried out at 77 K
on a Quantachrome Autosorb-6B analyzer. Prior to the measure-
ment, the samples were outgassed in vacuum at 393 K for 16 h.
The data were processed on Autosorb software (Quantachrome).
The specific surface area As was calculated according to the multi-
point Brunauer–Emmett–Teller method (BET) in the pressure range
p/p0 =0.05–0.15 assuming a N2 cross sectional area of 16.2 Š2. The
micropore surface area Am and micropore volume Vm were estimat-
ed using the t-plot method in the statistical thickness t=4.5–6.5 Š
range. The total pore volume Vp was estimated by using the
amount of physisorbed nitrogen at a relative pressure p/p0 =0.95.
The pore size distribution was determined by NLDFT method using
a model based on equilibrated adsorption of N2 on silica assuming
cylindrical pores at 77 K.
1H NMR spectroscopy: Solid-state magic angle spinning (MAS)
NMR experiments were performed on a Bruker DMX 400 spectrom-
eter (400.1 MHz, 9.4 T) at room temperature. Measurements were
run with single pulse excitation using a MAS frequency of 12.5 kHz.
1
The H background signal of the probe was corrected by subtract-
ing the 1H MAS NMR spectrum of an empty rotor. 1H chemical
shifts were referenced versus adamantane. The pretreated samples
were transferred into an air-tight rotor in a glove box to avoid ex-
posure to air. Approximately the same amount of sample was
charged into the rotor for every experiment.
Conclusions
In summary, we specified the structures of the active sites for
propene metathesis in MoOx/SBA-15 by integrating the inputs
from quantification of active sites, probe molecule adsorption
and structural characterization. We propose that ensembles of
strained, two-fold anchored tetrahedral di-oxo (SiÀOÀ)2-
Mo(=O)2 structures and adjacent silanol groups represent the
precursor of the active carbene species. The suggested promi-
nent role of adjacent silanol groups may help to understand
the beneficial effect of using acidic supports.[4] The obtained
insights pave the way for evolution of metathesis catalysts by
rational approaches, as has been exemplified in Figure 5.
FTIR spectroscopy: Diffuse reflectance FT infrared (IR) spectra
were collected at room temperature on a Bruker IFS 66 spectrome-
ter equipped with a liquid nitrogen-cooled MCT detector at a spec-
tral resolution of 4 cmÀ1 and accumulation of 1024 scans. An in situ
cell (Harrick Praying Mantis diffuse reflectance attachment DRP-P72
in combination with a HVC-VUV reaction chamber) was used. KBr
was used as reference material. The spectra were normalized using
the silica band at 1865 cmÀ1
.
Raman spectroscopy: Confocal Raman spectra were collected at
room temperature using a Horiba-Jobin Ybon LabRam instrument
equipped with a red laser excitation (633 nm/1.96 eV, 1.5 mW at
the sample position). Spectral resolutions were better than 2 cmÀ1
.
ChemCatChem 2015, 7, 4059 – 4065
www.chemcatchem.org
4063
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
An in situ cell (a home-made quartz cell) was used to measure the
dehydrated state. Prior to spectroscopic measurements, the sam-
ples were pretreated in dry oxygen (20 kPa, neat or diluted with
a dry inert gas) at 823 K (heating rate 10 KminÀ1) for 0.5 h, then
cooled to room temperatures in the presence of oxygen to achieve
the fully oxidized and dehydrated state of the catalyst.
Adsorption of probe molecules
Prior to the adsorption, the samples were pretreated in dry oxygen
(20 kPa, neat or diluted with a dry inert gas) at 823 K (heating rate
10 KminÀ1) for 0.5 h, then cooled to room temperatures in the
presence of oxygen to achieve the fully oxidized and dehydrated
state of the catalyst.
Microcalorimetry of propene adsorption: Differential heats of
propene adsorption were determined at 323 K using a MS70
Calvet Calorimeter (SETRAM). The calorimeter was combined with
a custom-designed high vacuum and gas dosing apparatus. Pro-
pene was stepwise introduced into the initially evacuated cell (p<
3”10À6 Pa), and the pressure evolution and the heat signal were
recorded for each dosing step. Though the propene was dosed at
the reaction temperature for propene metathesis, possible thermal
and volumetric contribution of the metathesis reaction can be ne-
glected owing to the thermo- and stoichiometric-neutral nature of
the reaction.[6b]
Propene metathesis and post-reaction active site counting
Propene metathesis: The catalytic activity for the self metathesis
of propene to ethene and 2-butenes was measured using a fixed-
bed tube flow reactor at atmospheric pressure. All the gases were
thoroughly dehydrated and deoxygenated (except oxygen) using
trapping filters. The catalysts were pressed under ꢀ135 MPa,
crushed and sieved to a particle size of 250–355 mm. Then, 100 mg
of the catalyst was loaded into a U-shaped quartz reactor with an
inner diameter of 4 mm. Guard beds consisting of silica gel (BET
surface area=428 m2 gÀ1) were placed both immediately above
(100 mg) and below (50 mg) the catalyst bed in order to protect
the catalyst bed from possible contamination by water. The use of
the silica guard beds is essential to obtain a good catalytic per-
formance. A blank test using bare SBA-15 with silica beds con-
firmed inertness of the apparatus and the guard beds. The catalyst
was activated at 823 K (heating rate 10 KminÀ1) for 0.5 h, cooled to
323 K in a 20% O2 in Ar (20 mLminÀ1), and then flushed with
a flow of Ar (20 mLminÀ1) before reaction. A neat propene flow of
8 mLminÀ1 was fed to start the reaction. Inlet and outlet gases
were analyzed by on-line gas chromatography using an Agilent
Technologies 6890 A GC system equipped with a flame ionization
detector. The conversion of propene was kept below 5% to stay in
a differential regime. The selectivity to the metathesis products
(ethane, cis- and trans-butene) was above 99.5%, although trace
amounts of 1-butene and higher hydrocarbons were detected. The
activity is presented as formation rate of the metathesis products
(i.e. sum of ethane, cis- and trans-butene) normalized by the BET
surface area of the catalyst. The catalytic test was repeated after
a regeneration procedure. The regeneration procedure is the same
as the initial activation (823 K (heating rate 10 KminÀ1) for 0.5 h
and cooled to 323 K in a dehydrated 20% O2 in Ar, then flushed
with Ar before starting the reaction). Error bars are estimated by
two repeated measurements using fresh and regenerated catalysts.
In situ IR for adsorption of propene and ammonia: Adsorption of
propene and ammonia was studied by in situ FTIR spectroscopy.
The IR experiments were carried out in transmission mode using
a PerkinElmer 100 FTIR spectrometer equipped with a DTGS detec-
tor at a spectral resolution of 4 cmÀ1 and accumulation of 64 scans.
The samples were pressed (125 MPa) into self-supporting wafers,
which were placed in an in situ IR cell. The IR cell was directly con-
nected to a vacuum system (residual pressure of 3”10À6 Pa)
equipped with a gas dosing line. Propene was dosed at 323 K at
the pressure up to 3 hPa. Ammonia was dosed at 353 K at the
pressure up to 7 hPa. In each experiment, the spectrum taken
before probe dosing was used as background. Contribution of gas
phase species was corrected by subtracting the spectrum without
sample wafer. The spectra shown were normalized by the areal
weight density of the wafer. The concentration of ammonia ad-
sorption sites were estimated using the band at 1614 and
ꢀ1430 cmÀ1 for Lewis acid sites and Brønsted acid sites, respec-
tively. Extinction coefficients of 16 cmmmolÀ1 (Brønsted acid sites)
and 1.46 cmmmolÀ1 (Lewis acid sites) were used.[11] Since the ex-
tinction coefficients have been taken from literature, the quantita-
tive analysis of Lewis and Brønsted acid sites, respectively, repre-
sents a rough estimation.
Temperature-programmed desorption of ammonia (NH3-TPD):
Temperature-programmed desorption of ammonia (NH3-TPD) was
performed using a fixed bed reactor. About 30 mg of catalyst was
used. Adsorption of NH3 was done at 353 K by feeding 1% NH3 in
Ar (40 mLminÀ1) for 0.5 h. After flushing the reactor with He at
353 K for 0.5 h, the bed temperature was raised with a heating rate
of 10 KminÀ1 in He flow (40 mLminÀ1). The desorption of NH3 was
monitored by a quadrupole mass spectrometer (OmniStar GSD301,
Pfeiffer) using the signal of m/e=16. The helium signal (m/e=4)
was used as internal standard.
Active-site counting by post-reaction [D4]-ethene metathesis:
After the metathesis reaction, the reactor was flushed with flowing
Ar (20 mLminÀ1 for 10 min, then 5 mLminÀ1 for 20 min), then the
feed gas was switched to 5 mLminÀ1 of 1%C2D4 in Ar. The forma-
tion of [D2]-propene-1,1 was monitored and quantified with a
quadrupole mass spectrometer (QMS200, Balzer) using the signal
of m/z=43. The formation of [D2]-propene-1,1 was also confirmed
by the simultaneous detection of the molecular ion (m/z=44). The
two-fold amount of the liberated amount of [D2]-propene-1,1 nor-
malized by the BET surface area of the catalyst was assumed as the
active site density.
Acknowledgements
Post-reaction temperature programmed desorption: After the
active counting procedure, the reactor was flushed with flowing Ar
(20 mLminÀ1 for 30 min) followed by a temperature ramp at rate
of 10 KminÀ1. Evolved species are monitored by a mass spectrome-
ter (QMS200, Balzer). Signal of Ar (m/e=40) was used as internal
standard.
We thank M. Hashagen, P. K. Nielsen, G. Lorenz and A. Klein-Hoff-
mann for their experimental assistance, and Prof. I. E. K. Wachs
for fruitful discussion. K. Amakawa is grateful to Mitsubishi Gas
Chemical Co. Inc. for a fellowship.
ChemCatChem 2015, 7, 4059 – 4065
www.chemcatchem.org
4064
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
[12] J. Herrera, J. Kwak, J. Hu, Y. Wang, C. Peden, Top. Catal. 2006, 39, 245–
255.
[13] F. Leydier, C. Chizallet, A. Chaumonnot, M. Digne, E. Soyer, A.-A. Quoi-
neaud, D. Costa, P. Raybaud, J. Catal. 2011, 284, 215–229.
[14] M. Seman, J. N. Kondo, K. Domen, S. T. Oyama, Chem. Lett. 2002, 31,
1082–1083.
Keywords: heterogeneous catalysis · molybdenum · olefin
metathesis · supported catalysts
[1] J. Ding, W. Hua, Chem. Eng. Technol. 2013, 36, 83–90.
[2] J. C. Mol, J. Mol. Catal. A 2004, 213, 39–45.
[15] F. Hemmann, I. Agirrezabal-Telleria, E. Kemnitz, C. Jäger, J. Phys. Chem. C
2013, 117, 14710–14716.
[16] V. Crocellà, G. Cerrato, G. Magnacca, C. Morterra, J. Phys. Chem. C 2009,
113, 16517–16529.
[17] X. Cao, R. Cheng, Z. Liu, L. Wang, Q. Dong, X. He, B. Liu, J. Mol. Catal. A
2010, 321, 50–60.
[18] a) J. Handzlik, J. Phys. Chem. B 2005, 109, 20794–20804; b) J. Handzlik,
J. Phys. Chem. C 2007, 111, 9337–9348.
[3] S. Lwin, I. E. Wachs, ACS Catal. 2014, 4, 2505–2520.
[4] a) X. Zhu, X. Li, S. Xie, S. Liu, G. Xu, W. Xin, S. Huang, L. Xu, Catal. Surv.
Asia 2009, 13, 1–8; b) D. P. Debecker, M. Stoyanova, F. Colbeau-Justin,
U. Rodemerck, C. Boissi›re, E. M. Gaigneaux, C. Sanchez, Angew. Chem.
Int. Ed. 2012, 51, 2129–2131; Angew. Chem. 2012, 124, 2171–2173.
[5] Y. Chauvin, Angew. Chem. Int. Ed. 2006, 45, 3740–3747; Angew. Chem.
2006, 118, 3824–3831.
[6] a) J. Handzlik, J. Ogonowski, Catal. Lett. 2003, 88, 119–122; b) K. Amaka-
wa, S. Wrabetz, J. Krçhnert, G. Tzolova-Müller, R. Schlçgl, A. Trunschke,
J. Am. Chem. Soc. 2012, 134, 11462–11473.
[19] L. J. Gregoriades, J. Dçbler, J. Sauer, J. Phys. Chem. C 2010, 114, 2967–
2979.
[20] K. A. Vikulov, B. N. Shelimov, V. B. Kazansky, J. C. Mol, J. Mol. Catal. 1994,
90, 61–67.
[21] F. Blanc, N. Rendon, R. Berthoud, J.-M. Basset, C. Coperet, Z. J. Tonzetich,
R. R. Schrock, Dalton Trans. 2008, 3156–3158.
[22] J. P. Thielemann, G. Weinberg, C. Hess, ChemCatChem 2011, 3, 1814–
1821.
[7] D. P. Debecker, B. Schimmoeller, M. Stoyanova, C. Poleunis, P. Bertrand,
U. Rodemerck, E. M. Gaigneaux, J. Catal. 2011, 277, 154–163.
[8] T. Hahn, U. Bentrup, M. Armbrüster, E. V. Kondratenko, D. Linke, Chem-
CatChem 2014, 6, 1664–1672.
[9] K. Amakawa, L. Sun, C. Guo, M. Hävecker, P. Kube, I. E. Wachs, S. Lwin,
A. I. Frenkel, A. Patlolla, K. Hermann, R. Schlçgl, A. Trunschke, Angew.
Chem. Int. Ed. 2013, 52, 13553–13557; Angew. Chem. 2013, 125, 13796–
13800.
[10] J. TrØbosc, J. W. Wiench, S. Huh, V. S. Y. Lin, M. Pruski, J. Am. Chem. Soc.
2005, 127, 3057–3068.
Received: June 24, 2015
[11] V. A. Matyshak, O. V. Krylov, Kinet. Catal. 2002, 43, 391–407.
Published online on September 28, 2015
ChemCatChem 2015, 7, 4059 – 4065
www.chemcatchem.org
4065
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim