In situ generation of active sites in olefin metathesis

Article abstract of DOI:10.1021/ja3011989

The depth of our understanding in catalysis is governed by the information we have about the number of active sites and their molecular structure. The nature of an active center on the surface of a working heterogeneous catalyst is, however, extremely difficult to identify and precise quantification of active species is generally missing. In metathesis of propene over dispersed molybdenum oxide supported on silica, only 1.5% of all Mo atoms in the catalyst are captured to form the active centers. Here we combine infrared spectroscopy in operando with microcalorimetry and reactivity studies using isotopic labeling to monitor catalyst formation. We show that the active Mo(VI)-alkylidene moieties are generated in situ by surface reaction of grafted molybdenum oxide precursor species with the substrate molecule itself gaining insight into the pathways limiting the number of active centers on the surface of a heterogeneous catalyst. The active site formation involves sequential steps requiring multiple catalyst functions: protonation of propene to surface Mo(VI)-isopropoxide species driven by surface Bronsted acid sites, subsequent oxidation of isopropoxide to acetone in the adsorbed state owing to the red-ox capability of molybdenum leaving naked Mo(IV) sites after desorption of acetone, and oxidative addition of another propene molecule yielding finally the active Mo(VI)-alkylidene species. This view is quite different from the one-step mechanism, which has been accepted in the community for three decades, however, fully consistent with the empirically recognized importance of acidity, reducibility, and strict dehydration of the catalyst. The knowledge acquired in the present work has been successfully implemented for catalyst improvement. Simple heat treatment after the initial propene adsorption doubled the catalytic activity by accelerating the oxidation and desorption-capturing steps, demonstrating the merit of knowledge-based strategies in heterogeneous catalysis. Molecular structure of active Mo(VI)-alkylidene sites derived from surface molybdena is discussed in the context of similarity to the highly active Schrock-type homogeneous catalysts.

Full text of DOI:10.1021/ja3011989

Article
pubs.acs.org/JACS
In Situ Generation of Active Sites in Olefin Metathesis
Kazuhiko Amakawa, Sabine Wrabetz, Jutta Krohnert, Genka Tzolova-Muller, Robert Schlogl,
̈
̈
̈
and Annette Trunschke*
Department of Inorganic Chemistry, Fritz-Haber-Institut der Max-Planck-Gesellschaft, Faradayweg 4-6, 14195 Berlin, Germany
S
* Supporting Information
ABSTRACT: The depth of our understanding in catalysis is
governed by the information we have about the number of
active sites and their molecular structure. The nature of an
active center on the surface of a working heterogeneous catalyst
is, however, extremely difficult to identify and precise
quantification of active species is generally missing. In
metathesis of propene over dispersed molybdenum oxide
supported on silica, only 1.5% of all Mo atoms in the catalyst
are captured to form the active centers. Here we combine
infrared spectroscopy in operando with microcalorimetry and
reactivity studies using isotopic labeling to monitor catalyst
formation. We show that the active Mo(VI)−alkylidene moieties are generated in situ by surface reaction of grafted molybdenum
oxide precursor species with the substrate molecule itself gaining insight into the pathways limiting the number of active centers
on the surface of a heterogeneous catalyst. The active site formation involves sequential steps requiring multiple catalyst
functions: protonation of propene to surface Mo(VI)−isopropoxide species driven by surface Brønsted acid sites, subsequent
oxidation of isopropoxide to acetone in the adsorbed state owing to the red-ox capability of molybdenum leaving naked Mo(IV)
sites after desorption of acetone, and oxidative addition of another propene molecule yielding finally the active Mo(VI)−
alkylidene species. This view is quite different from the one-step mechanism, which has been accepted in the community for
three decades, however, fully consistent with the empirically recognized importance of acidity, reducibility, and strict dehydration
of the catalyst. The knowledge acquired in the present work has been successfully implemented for catalyst improvement. Simple
heat treatment after the initial propene adsorption doubled the catalytic activity by accelerating the oxidation and desorption-
capturing steps, demonstrating the merit of knowledge-based strategies in heterogeneous catalysis. Molecular structure of active
Mo(VI)−alkylidene sites derived from surface molybdena is discussed in the context of similarity to the highly active Schrock-
type homogeneous catalysts.
possible applying postreaction metathesis using a probe
olefin.5,7,8
1. INTRODUCTION
Despite the vital branch of industrial production that involves
heterogeneous catalysis, underlying research in the entire range
of complexity from model studies over single-crystals to the
investigation of applied catalyst formulations generally lacks in
the most essential information: structure and number of active
sites.1 The termination of the solid surface that represents the
catalytic interface is, even for model catalysts, enormously
diverse2 and changes dynamically under reaction conditions.3,4
Hence, despite significant progress in the field of molecular
surface science,2−4 identification of catalytically relevant surface
structures (i.e., active sites) among the existing variety of
structures and counting of active sites remains extremely
challenging, and the catalyst is still a “Holy Grail”.1
Heterogeneous olefin metathesis is an exceptional case where
the general structural motif of active sites is established and
thus quantification is accessible.5 It is generally accepted that
metal carbene species (MCHR) catalyze the reaction
according to the metallacyclobutane intermediate mechanism
in analogy to homogeneous catalysis.6 Irrespective of the
specific metal carbene structure, quantification of active sites is
Monolayer-type metal (e.g., Mo, W, and Re) oxides
deposited on porous supports (e.g., silica, alumina, and
silica−alumina) are important metathesis catalysts.9−14 The
number of active carbene species in these catalysts is at most
2% of the total metal content.5,7,8 The low abundance is
attributed to multiple geometric configurations, allowing only a
minor fraction of surface metal oxide species to be transformed
into the metal carbene upon contact with the olefin.
Understanding the formation mechanism of the active centers
and their in-depth structural description will provide the criteria
for the relevant precursor structures disclosing access to more
sustainability and improved economics via rational catalyst
design.
The surface reaction between catalyst precursor and olefin is
a demanding process, since the metal oxides need to
accomplish carbene synthesis in one-pot and in situ. Several
surface reactions via different intermediates have been proposed
Received: February 6, 2012
Published: June 15, 2012
© 2012 American Chemical Society
11462
dx.doi.org/10.1021/ja3011989 | J. Am. Chem. Soc. 2012, 134, 11462−11473

Journal of the American Chemical Society
Article
(Scheme 1): (A) formation of a π-complex between the
reacting alkene and a coordinatively unsaturated metal ion
similarities to highly active Schrock-type homogeneous
catalysts.
Scheme 1. Proposed Mechanisms for Initial Carbene
Formation upon Contact of the Catalyst with an Olefin
2. EXPERIMENTAL SECTION
a
The experimental details are presented in the Supporting Information,
SI. Below, the most important information is briefly summarized.
2.1. Preparation of MoOx/SBA-15. Molybdenum oxide was
dispersed on the surface of mesoporous silica SBA-1526 applying an
anion exchange procedure.27 According to this, freshly prepared SBA-
15 (specific surface area (SBET) = 859 m2 g−1, internal sample ID
8233) was at first functionalized with propylammonium chloride using
(3-aminopropyl)trimethoxysilane followed by treatment with hydro-
chloric acid. Then, anion exchange employing an aqueous solution of
ammonium heptamolybdate was performed. After washing and
filtration, the material was dried and calcined at 823 K in air to
remove organic moieties as well as chlorine and to anchor molybdena
species on silica, yielding MoOx/SBA-15 with actual molybdenum
loading of 9.7%, which corresponds to the surface molybdenum
density of 1.1 Mo/nm2 (SBET = 556 m2 g−1, internal sample ID 8438).
The nitrogen adsorption study (Table S1, Figure S1 of the SI)
confirmed that the cylindrical mesopore structure of the SBA-15 was
preserved after the molybdenum loading. Due to the decrease in the
Sμ/SBET ratio (Table S1 of the SI), preferential filling of the micropores
with molybdenum oxides is suggested.
2.2. Physico-Chemical Characterization of MoOx/SBA-15.
Nitrogen adsorption was carried out at 77 K to determine specific
surface area SBET, micropore surface area Sμ, pore volume, and pore
size distribution. X-ray fluorescence (XRF), powder X-ray diffraction
(XRD), ultraviolet−visible (UV−vis) diffuse reflectance spectroscopy
using SBA-15 as reference, Raman spectroscopy applying an excitation
wavelength at 632.8 nm, and laser power of 1.5 mW at the sample
position, and diffuse reflectance infrared Fourier transform (DRIFT)
spectroscopy using KBr as reference were employed to characterize the
catalyst and the support. Prior to Raman, DRIFT, and UV−vis
spectroscopy, the samples were pretreated at 823 K (heating rate 10
K·min−1) for 0.5 h in a dry 20% O2 in Ar flow. To probe acidity,
ammonia adsorption FTIR was performed in transmission mode using
self-supporting wafers. Ammonia was adsorbed at 353 K after
pretreatment of the samples at 823 K in dry oxygen.
2.3. Propene Metathesis. Catalytic activity in metathesis of
propene to ethene and 2-butenes was measured using a fixed-bed
tubular flow reactor at atmospheric pressure. The catalyst bed was
pretreated at 823 K for 0.5 h in a flow of a thoroughly dried gas
mixture of 20% O2 in Ar, cooled to 323 K in the same gas flow, and
then flushed with a flow of Ar before reaction. Dried and deoxygenated
neat propene was fed to the catalyst bed at 323 K at atmospheric
pressure. Inlet and outlet gases were analyzed by online gas
chromatography. The selectivity to the metathesis products (ethene,
cis- and trans-butene) was higher than 99.5%, while 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
ethene, cis- and trans-butene) normalized to the mass of the catalyst.
Regeneration of the catalyst was performed using the same procedure
as the initial activation.
a
M = active metal center for metathesis reaction, S = support element.
followed by a 1,2-hydrogen shift (1,2-hydrogen shift mecha-
nism),15,16 (B) formation of a π-complex followed the
transformation of the π-allyl hydride intermediate into a
metallacyclobutane (allyl mechanism),17−19 (C) formation of a
metal-oxo intermediate mediated by a surface Brønsted acid site
and subsequent hydrogen shift (H-assisted mechanism),20 and
(D) formation of an oxametallacyclobutane and subsequent
elimination of a carbonyl compound (pseudo-Wittig mecha-
nism).7,21−24
All reaction pathways, except the pseudo-Wittig mechanism,
involve an oxidative addition reaction, in which metal centers
are required that can be oxidized by losing two electrons. Since
catalyst pretreatment and regeneration of supported metal
oxides are typically performed by oxidative treatment at
elevated temperature,9,25 the pseudo-Wittig mechanism would
be the only applicable carbene formation route.
In this work, we describe the origin of active carbene species
in propene metathesis over a well-defined model catalyst that
contains dispersed molybdenum oxide species supported on
meso-structured silica. We investigated propene adsorption
using microcalorimetry and in situ IR spectroscopy to trace the
genesis of active carbene sites. Our experimental findings deny
the pseudo-Wittig mechanism, and provide a new picture,
which is fully consistent with the empirical criteria for high-
performing catalysts. The established mechanism allows sorting
out possible molecular structures of surface oxide precursor
species and resulting metal carbene configurations and resulted
in the successful implementation of a new catalyst pretreatment
procedure. The molecular structure of metal carbene species
derived from surface molybdena is discussed in the context of
2.4. Postreaction Ethene-d4 Metathesis for Active Site
Counting. A modified version of the dynamic active site counting
technique originally developed by Handzlik5 was employed, wherein
ethene-d4 (CD2CD2) was used instead of nonlabeled ethene as the
probe olefin to titrate metal carbene centers after the catalysis.
According to the Chauvin mechanism6 (Scheme 2), equal amounts of
molybdenum methylidene (MoCH2) and molybdenum ethylidene
(MoCH−CH3) species are present under steady state conditions of
propene metathesis on the catalyst surface. The active site counting
consists of the quantification of molybdenum-ethylidene (MoCH−
CH3) species by titration with ethene-d4 (CD2CD2), which results
in liberation of propene-1,1-d2 (CD2CH−CH3) via metathesis
reaction according to eq 1 and Scheme 2.
CD2CD2 + MoCH−CH3 → CD2CH−CH3 + MoCD2
(1)
11463
dx.doi.org/10.1021/ja3011989 | J. Am. Chem. Soc. 2012, 134, 11462−11473

Journal of the American Chemical Society
Article
Scheme 2. Reaction Mechanism of Propene Metathesis
a
According to Chauvin1
a
The titration reaction of Mo−ethylidene with ethene-d4 is described
in red color.
Subsequent to the metathesis reaction, the reactor was flushed with
flowing Ar, then the feed gas was switched to 1% C2D4 in Ar at 323 K.
The formation of propene-1,1-d2 was monitored and quantified with a
quadrupole mass spectrometer (QMS200, Balzer) using the signal of
m/z = 43. The active site density corresponds to the 2-fold amount of
liberated propene-1,1-d2 normalized to the weight of the catalyst.
2.5. Microcalorimetry and in Situ IR Spectroscopy of
Propene Adsorption. Microcalorimetry was performed using a
MS70 Calvet Calorimeter (SETRAM). The calorimeter was combined
with a custom-designed high vacuum and gas dosing apparatus
equipped with a volumetric system, which allows measurements of
adsorption isotherms. The IR experiments were carried out in
transmission mode using a Perkin-Elmer 100 FTIR spectrometer
equipped with an in situ cell. In both methods, propene was admitted
at 323 K after pretreatment of the catalyst in 20 kPa of oxygen at 823
K for 0.5 h and subsequent evacuation at 323 K. Adsorption of C2D4
after propene adsorption as well as adsorption of isopropanol and
acetone were also conducted in the IR study.
Figure 1. IR spectra of MoOx/SBA-15 and SBA-15; (A): DRIFT
spectra measured at room temperature after pretreatment in 20% O2 at
823 K for 0.5 h. (B): Transmission IR spectra recorded after
adsorption of ammonia at p = 10 hPa and subsequent evacuation at
353 K. The catalyst was pretreated in 20 kPa of O2 at 823 K for 0.5 h.
The spectrum before ammonia dosing was used as background.
clearly indicates the occurrence of hydrogen-bonded OH
groups when molybdenum is loaded onto the silica SBA-15
surface. The latter band is attributed to a superposition of O−H
stretching vibrations in molybdenol (Mo−OH) groups
involved in hydrogen bonding to neighboring oxygen atoms,
e.g., in OMo−OH moieties or silanol groups that undergo
hydrogen interaction with oxygen atoms of molybdena species
in the vicinity.
3. RESULTS
3.1. Physico-Chemical Characterization of MoOx/SBA-
15. After the oxidative pretreatment, highly dispersed
molybdenum oxide species with molybdenum in the highest
oxidation state exist on the catalyst surface. The presence of
bulk-like MoO3 and H4SiMo12O40 impurities is excluded not
only due to the absence of any related reflection peak in the
XRD patterns (not shown), but, in particular, due to the lack in
The adsorption of ammonia revealed the presence of both
Brønsted and Lewis acid sites. Whereas almost no ammonia
adsorbs on SBA-15 (black line in Figure 1B), the IR spectrum
of MoOx/SBA-15 after ammonia adsorption (blue line in
Figure 1B) shows the presence of ammonium ions formed by
reaction of NH3 with Brønsted acid sites (1678 cm−1
28
bands at 816 cm−1 for MoO3, and 909 cm−1 for H4SiMo12O4
in the Raman spectrum (Figure S2 of the SI). The Raman
bands near 990 and 970 cm−1 due to molybdenum−oxygen
stretching vibrations (νMoO) (Figure S2 of the SI) as well as
the absorption bands in the UV−vis spectrum near 225 and
270 nm (ligand to metal charge transfer from O2− to Mo6+)
(Figure S3 of the SI) are typical fingerprints indicating highly
dispersed molybdenum oxide surface species in the dehydrated
state.28−31 The absence of absorption due to d−d transitions of
molybdenum ions (>500 nm)32 is evidence that molybdenum
is exclusively present as Mo(VI).
The DRIFT spectrum (Figure 1A) of MoOx/SBA-15 shows
significantly weaker intensity of the band due to isolated Si−
OH (silanol) groups33 at 3745 cm−1 than bare SBA-15,
indicating that loading of molybdenum involves consumption
of surface silanol groups to yield anchored surface molybdate
species. Moreover, the DRIFT spectrum of MoOx/SBA-15
features a broad OH stretching mode at 3620 cm−1, which
+
(δsymNH4) and 1431 cm−1 (δasNH4 )) and molecular ammonia
coordinated at Lewis acid sites (1613 cm−1 (δasNH3)).33 The
detection of Brønsted acid sites indicates either the presence of
molybdenol (Mo−OH) groups or the formation of acid Si−
OH groups by interaction of terminal silanol groups with
surface molybdenum oxide species. The broad OH stretching
mode at 3620 cm−1 detected in MoOx/SBA-15 (Figure 1A)
may contain contributions from either of them. Lewis acid sites
are considered as coordinatively unsaturated Mo(VI) centers.
Irrespective of the specific assignment, ammonia adsorption
clearly indicates the generation of acid sites by introduction of
molybdenum oxide species into the pores of SBA-15.
In summary, the MoOx/SBA-15 model catalyst is charac-
terized by extensive coverage of the silica surface with highly
dispersed MoOx species containing Mo in its highest oxidation
11464
dx.doi.org/10.1021/ja3011989 | J. Am. Chem. Soc. 2012, 134, 11462−11473

Journal of the American Chemical Society
Article
state, and comprising Mo oxo (MoO) moieties. The
presence of Brønsted acid sites and coordinatively unsaturated
Mo(VI) species (Lewis acid sites) is evidenced by ammonia
adsorption IR.
3.2. Propene Metathesis and Postreaction Active Site
Counting. Figure 2 shows the propene metathesis activity of
because (1) the heat of reaction is very small (ΔH323K = 1.2
kJ molpropene−1) and, consequently, thermal contributions from
the catalytic reaction can practically be neglected in the
measurement of the heat of adsorption, and (2) the total
number of molecules remains constant through the reaction,
which allows volumetric determination of the amount of
adsorbed propene molecules because the total pressure is not
falsified due to the reaction. This situation opens up the
opportunity to trace thermal events, which are presumably
relevant to the genesis of activity.
Figure 3 A,B reports the differential heat of propene
adsorption on MoOx/SBA-15 (A) and SBA-15 (B) as a
function of propene coverage. While the differential heat
measured on SBA-15 is slightly higher than the heat of propene
condensation and remains constant at 23 1 kJ mol−1 with
increasing coverage (Figure 3B), the MoOx/SBA-15 catalyst
shows significantly higher heats of adsorption up to certain
levels of coverage (Figures 3A). The catalyst MoOx/SBA-15
(Figure 3A) is characterized by a plateau up to 16 μmol g−1,
which corresponds to 1.6% of total molybdenum atoms in the
catalyst. The differential heat of propene adsorption in this
range (78 2 kJ mol−1) is quite high, which indicates strong
adsorption of propene on energetically homogeneous adsorp-
tion sites. Further increase in surface coverage leads to a gradual
decrease in the heat of adsorption up to a coverage of 29 μmol
g−1 that corresponds to 2.9% of total molybdenum atoms in the
catalyst where another plateau is approached. The heat of
adsorption (22 kJ mol−1) is close to the value measured on
SBA-15 (Figure 3B). Notably, the concentration of strong
adsorption sites (16 μmol g−1 within the first plateau) coincides
with the active site density estimated by active site counting
after the catalysis (15 μmol g−1, Table 1). The consistency
strongly suggests the relevance of sites titrated by propene
adsorption in the microcalorimetric experiment to the genesis
of active carbene species.
Figure 2. Propene metathesis activity of MoOx/SBA-15 at 323 K and a
contact time of 0.75 s g mL−1. The catalyst was activated in a 20% O2
flow at 823 K for 0.5 h. Regeneration was performed applying the
same procedure as the initial activation (823 K in 20% O2 for 0.5 h).
MoOx/SBA-15 at 323 K as a function of time on stream. The
metathesis activity gradually develops with time on stream,
reaching 7.5 mmol g−1 h−1 (corresponding to an apparent TOF
of 2.06 mmolpropene molMo−1 s−1) after 15 h. Metathesis activity
and apparent TOF are rather low compared to the recent most
successful silica−alumina supported molybdena catalysts
reported by Debecker et al.11,12 The gradual evolution of the
activity suggests slow generation and continuous accumulation
of active molybdenum carbene centers.34 The catalyst is
regenerable with an excellent reproducibility in the reactivity
(Figure 2).
Strong interaction of propene with MoOx/SBA-15 is also
reflected in the Langmuir constant. The Langmuir adsorption
equation has been formally applied to model the measured
adsorption isotherms (lines in Figure 3C) neglecting that
chemical transformations are probably involved in adsorption
of propene on MoOx/SBA-15 using eq 2,35
Table 1 summarizes the data on the metathesis activity and
the postreaction active site counting as well as the derived
Table 1. Summary of Propene Metathesis Activity and Post-
Reaction Active Site Counting over MoOx/SBA-15
a
b
c
d
catalyst
activity
active sites
active site fraction TOF
Kp
(mmol g−1 h−1
)
(μmol g−1
)
(atom%)
(s−1
)
N
= N
ads
mono 1 + Kp
(2)
fresh
regenerated
7.50
7.44
15.0
14.0
1.5
1.4
0.14
0.15
a
where Nads is the amount of propene adsorbed at equilibrium
pressure p, Nmono is the monolayer adsorption capacity, and K is
the Langmuir constant, which is the ratio of the rate constant of
adsorption and desorption describing the strength of
adsorption. The fit curve for the MoOx/SBA-15 depicted in
Figure 3C deviates from the data points above ∼20 μmol g−1,
which indicates that the propene uptake in this high coverage
region is predominantly due to a different type of adsorption
mechanism. The part of the adsorption isotherm curve of the
catalyst above 20 μmol g−1 is roughly parallel to the curve of
SBA-15 (Figure 3C), suggesting that the same type of
physisorption on silica species and/or inert molybdenum
oxide species becomes predominant at the higher coverage.
This view is also supported by the faster decay of the heat
evolution upon adsorption (vide infra, Figure 4). A Langmuir
fit taking into account data of propene adsorption on MoOx/
SBA-15 up to a coverage of ∼16 μmol g−1, where a quasi-
constant and high heat of adsorption was found, yielded K = 8.6
Propene metathesis activity measured immediately before the
postreaction active site counting procedure (TOS = 15 h) for
MoOx/SBA-15 (see also Figure 2); Metathesis reaction condition: T =
323 K, p = 0.1 MPa, contact time = 0.75 s ml g−1, pretreatment or
regeneration at T = 823 K for 0.5 h under 20% O2−Ar flow. The 2-
fold amount of evolved propene-d2 upon ethene-d4 dosing after the
b
c
propene metathesis. Active site density divided by Mo loading.
d
Metathesis activity divided by active site density.
active site fraction and turn over frequencies. The active site
density corresponds to about 1.5% of total molybdenum atoms
in the catalyst. The value is similar to reported data on MoOx/
Al2O3 and ReOx/Al2O3.7,8
5
3.3. Microcalorimetry of Propene Adsorption at the
Reaction Temperature. To gain insight into the process of
carbene site formation, we studied propene adsorption by
microcalorimetry at the reaction temperature. The metathesis
reaction does not disturb the calorimetric measurement,
11465
dx.doi.org/10.1021/ja3011989 | J. Am. Chem. Soc. 2012, 134, 11462−11473

Journal of the American Chemical Society
Article
Figure 4. Evolution of the heat signals upon propene adsorption at
323 K on MoOx/SBA-15 (pretreated in 20% O2 at 823 K for 0.5 h) at
the cumulative adsorption amount of 6 (A) and 26 μmol g−1 (B).
observation, the integral heat of desorption measured during
the evacuation after the first adsorption steps was much lower
than the integral heat of adsorption. Apparently, irreversible
surface reactions happen, which contribute substantially to the
heat signal measured during the first propene adsorption
experiment. These reactions are most likely related to the
formation of the active sites.
Evaluation of the kinetics of the heat evolution allows further
insight into the adsorption process. The response of the
exothermic signals in the range of the first plateau of propene
adsorption on MoOx/SBA-15 (at coverage of 6 μmol g−1)
shows an extremely slow decay continuing for ∼7 h (Figure
4A). In contrast, the evolution of heat at higher coverage
(Figure 4B), where the differential heat of adsorption is slightly
higher than the heat of propene adsorption measured on SBA-
15 (at a coverage of 26 μmol g−1), is relatively quickly
completed. The initial decay of the heat signals was evaluated
applying an exponential decay model as described in eq 3,
Figure 3. Propene adsorption studied by microcalorimetry at 323 K on
MoOx/SBA-15 and SBA-15 pretreated in O2 at 823 K and at 20 kPa
for 0.5 h; differential heat as a function of the amount of adsorbed
propene on MoOx/SBA-15 (A) and SBA-15 (B); readsorption profiles
were measured after evacuation at ∼10−3 Pa for 2 h; the regeneration
was performed applying the same procedure as the initial pretreatment
(in O2 at 823 K and at 20 kPa for 0.5 h); the adsorption isotherm of
propene over MoOx/SBA-15 and SBA-15 measured at 323 K is shown
in (C); measured data points up to an amount of adsorbed propene of
16 μmol g−1 were used to fit the data of MoOx/SBA-15 based on the
Langmuir equation.
I(t) = I0exp(−t/τ)
(3)
wherein I(t) is the net thermosignal intensity at time t, I0 is the
maximum of the net thermosignal intensity after an increment
of dosing pressure, t is the elapsed time after reaching the I0,
and τ is the time constant. The calculated time constants are
about 3000 and 255 s for 6 and 26 μmol g−1, respectively
(Figure 4). The latter value approximately matches the time
constant of the calorimeter used, indicating weak and reversible
adsorption. The significantly increased time constant at low
coverage indicates that the adsorption is superimposed by slow
surface reactions. The deviation of the fit curve at prolonged
times (Figure 4A) is due to the occurrence of secondary
exothermic processes that are supposed to be consecutive
reactions of adsorbed species. The deviation is less pronounced
for the adsorption at 26 μmol g−1 reflecting the predominance
of reversible adsorption in accordance with the moderate
2.8 hPa−1, whereas the fit for propene adsorption on SBA-15
including the entire data points resulted in a much lower value
of K = 0.039
0.001 hPa−1. These results clearly indicate
strong adsorption of propene on a small fraction of
molybdenum oxide surface sites on MoOx/SBA-15.
Propene adsorption on SBA-15 is fully reversible. In contrast,
the readsorption profile measured on MoOx/SBA-15 after
evacuation (Figure 3A) was almost identical to the adsorption
profile measured on bare SBA-15 (Figure 3B), which evidences
again strong and irreversible adsorption of propene on MoOx
sites during the first adsorption cycle. In accordance with this
11466
dx.doi.org/10.1021/ja3011989 | J. Am. Chem. Soc. 2012, 134, 11462−11473

Journal of the American Chemical Society
Article
differential heat of adsorption (40 kJ mol−1). These
observations indicate that the strong and irreversible adsorption
involves slow and probably consecutive reactions on the
surface, which might be related to the genesis of the
molybdenum carbene sites. The sustaining generation of heat
for ∼7 h bears resemblance to the slow evolution of metathesis
activity over 10−15 h, which are required to reach steady state
activity (Figure 2).
3.4. IR Study of Propene and Subsequent Ethene-d4
Adsorption. Microcalorimetry revealed irreversible and
possibly reactive adsorption of propene on MoOx/SBA-15 up
to a certain level of coverage. The chemical reactions are
presumably associated with the genesis of the active carbene
sites. We employed in situ IR spectroscopy at the same
temperature (323 K) to obtain information about the nature of
the surface species involved in the processes occurring upon
propene adsorption on the MoOx/SBA-15 surface. The
equilibrium pressure of propene was adjusted initially to p =
3 hPa, which corresponds to complete coverage of strong
adsorption sites according to the results of microcalorimetry. A
reference experiment using SBA-15 showed no adsorbed
species that persist after evacuation, which is consistent with
the reversible adsorption found by microcalorimetry.
3.4.1. Progressive Formation of Isopropoxide and Acetone
upon Propene Adsorption. After propene dosing for 0.5 h and
subsequent evacuation, the spectrum (magenta line in Figure
5A) features bands due to the stretching modes of C−H bonds
in methyl groups at 2983, 2939, and 2880 cm−1, and the
corresponding deformation modes at 1465, 1455, 1390, and
1377 cm−1. Notably, characteristic C−H stretching bands of
olefinic methylene species36,37 at ∼3010 and at ∼3090 cm−1 as
well as CC stretching mode at 1610−1650 cm−1 are not
detected, indicating the absence of CC bonds in strongly
adsorbed species that persist evacuation at 323 K. Thus, strong
adsorption involves a reaction, which leads to conversion of the
olefinic CC bond in propene. The detected peak patterns
agree well with the bands of isopropoxy species adsorbed on
the surface of metal oxides.38−41 In fact, adsorption of
isopropanol on MoOx/SBA-15 results in a quite similar
spectrum (Figure S4 (1) of the SI), confirming that
isopropoxide is the major species formed by adsorption of
propene at 323 K on MoOx/SBA-15. The transformation of
propene into isopropoxide happens via protonation of the
olefin by acidic surface OH groups. Accordingly, the detection
of surface isopropoxy species confirms the presence of Brønsted
acid sites on the surface of MoOx/SBA-15 in agreement with
ammonia adsorption (Figure 1B) and clearly indicates
involvement of Brønsted acidity in the surface reactions of
adsorbed propene.
After the 18 h of propene dosing and subsequent evacuation,
(blue line in Figure 5A), all of the bands became considerably
more intense, which evidence progressive slow adsorption as
found by microcalorimetry. Moreover, new bands at 1668 and
1417 cm−1 are clearly visible, which can be assigned to the C
O stretching mode of the carbonyl group and the deformation
mode of methyl groups, respectively, of coordinated
acetone.38,40,42,43 Occurrence of these bands upon acetone
adsorption (Figure S4 (3) of the SI) as well as the development
of a similar carbonyl stretching band at 1673 cm−1 upon
prolonged (18 h) isopropanol adsorption (Figure S4 (2) of the
SI) corroborate the formation of acetone by oxidation of
isopropoxide surface species. The modest intensity of the
carbonyl band at 1668 cm−1 (blue line in Figure 5A) compared
Figure 5. IR spectra recorded after propene adsorption on MoOx/
SBA-15 for 0.5(magenta) and 18 h (blue) and subsequent evacuation
(A). The difference spectrum shown in (B) was obtained by
subsequent ethene-d4 dosing at p = 3 hPa and T = 323 K for 18 h
and evacuation using the blue spectrum in (A) as subtrahend. Propene
was dosed at 323 K and 3.0 hPa after the pretreatment in 20 kPa of O2
at 823 K for 0.5 h. The inset in (B) shows the frequency range of the
C−D stretching.
to the spectrum of acetone adsorption (Figure S4 (3) of the SI)
suggests that only a fraction of the isopropoxide species
undergo oxidation into acetone. Formation of acetone implies
the involvement of surface molybdenum oxide species acting as
oxidizing agent and resulting in partial reduction of
molybdenum on the silica surface. Essentially the same
observation has been reported for alumina-supported molyb-
dena catalysts. The formation of isopropoxide and the
consecutive oxidation to acetone upon propene adsorption
takes place on MoOx/Al2O3 even at room temperature.40,41
Besides the formation of acetone, the broad peaks arising in the
range 1550−1650 cm−1 and below 1490 cm−1 (Figure 5A)
imply the gradual accumulation of various other surface
deposits, such as carboxylates44 or enolates.45
3.4.2. Validation of Metathesis Activity of the Resulting
Surface. The metathesis activity of the resulting surface was
confirmed by testing the reactivity with ethene-d4. After
propene adsorption for 18 h and subsequent evacuation, an
equilibrium pressure of p = 3 hPa ethene-d4 was adjusted at 323
K in the IR cell. Under these conditions, surface Mo−alkylidene
species, if present, should be transformed into deuterium-
labeled Mo−methylidene species (eq 1). The spectrum
presented in Figure 5B was obtained after dosing ethene-d4
followed by evacuation using the spectrum before the ethene-d4
dosing (blue line in Figure 5A) as background. The negative
peaks due to C−H stretching modes at 2984, 2939, 2898, and
11467
dx.doi.org/10.1021/ja3011989 | J. Am. Chem. Soc. 2012, 134, 11462−11473

Journal of the American Chemical Society
Article
Scheme 3. Proposed Route for the Carbene Formation Starting from a Mo(VI) Site upon Interaction with Two Propene
a
Molecules
a
General scheme (A), a scheme assuming a tetrahedral dioxo structure as the pre-catalyst (B) compared to the structure of reference Schrock-type
homogeneous catalysts (C).63
2877 cm−1 and C−H deformation modes at 1468, 1389, and
1375 cm−1, and the appearance of C−D stretching modes at
2100−2300 cm−1 (Figure 5B inset) evidence the exchange of
stable (not removable by evacuation) surface organic species
into ethene-d4 originated surface species most likely due to a
metathesis reaction catalyzed by surface Mo−carbene sites. The
negative peaks due to C−H stretching modes agree well with
the bands of Mo−ethylidene species at 2985, 2910, 2890, and
2850 cm−1,46 indicating the presence of this species before
dosing of ethene-d4. The positions of the observed C−D
stretching modes at 2230 and 2160−2200 cm−1 are close to
those reported for deuterated methylidene (MoCD2) species
detected on MoOx/SiO2 (2245 and 2160 cm−1),47 which
further corroborates the presence of Mo−carbene sites on the
surface of MoOx/SBA-15 and the metathesis activity of the
catalyst surface, which has been generated during the IR
experiment of propene adsorption. The intensities of the
negative peaks due to C−H stretching modes (Figure 5B) are
low compared to the original spectrum (blue line in Figure 5A),
indicating that the concentration of carbene species is low in
contrast to the concentration of other surface species, e.g., the
remaining isopropoxide species.
3.4.3. Summary of the Propene Adsorption IR Study. The
IR study revealed that adsorbed propene is slowly protonated
on the catalyst surface to yield isopropoxide, which undergoes
oxidation to give acetone. These reactions require bifunction-
ality of the catalyst in terms of concurrent abundance of
Brønsted acid sites and redox active molybdena centers. The
evolution of Mo−carbene species evidenced by ethene-d4
adsorption indicates that these processes are related to the
genesis of the active Mo−carbene sites. We relate the very slow
generation of isopropoxide and acetone as observed by IR
spectroscopy to the prolonged heat generation registered in the
microcalorimetric experiment (Figure 3A) and to the retarded
evolution of the metathesis activity (Figure 2).
4. DISCUSSION
4.1. Quantity and Quality of Active Carbene Sites.
Investigation of MoOx/SBA-15 by the postreaction active site
titration detected the expected metathesis product (i.e., CD2
CH−CH3) formed by reaction of surface Mo−ethylidene
species with the probe ethene-d4 (CD2CD2) (Table 1),
providing evidence that the Chauvin mechanism (Scheme 2) is
operative over supported molybdena catalysts.
The titration experiment reveals that only 1.5% of the
molybdenum atoms in the catalyst belong to the pool of active
carbene centers. Accordingly, it is extremely challenging to
establish structure−activity relationships, because structural
characterization generally displays an average picture of the
variety of species usually present on the catalyst surface. A
strategy that allows selective detection of the essential
information is certainly necessary to distinguish catalytically
relevant surface sites from spectator species. The present
approach takes these considerations into account and tackles
the task by applying a combination of microcalorimetry and IR
spectroscopy for investigation of propene adsorption on the
catalyst surface at the reaction temperature.
Quantification of active sites by titrative metathesis allows the
determination of an average intrinsic metathesis activity per
active site, i.e., the measurement of turnover frequencies in the
strict sense of molecular catalysis. This is a rare case in
heterogeneous catalysis due to frequently encountered
difficulties in quantification of active sites, which often lead to
the wrong assumption that all atoms of a specific catalyst
component contribute to the activity. Accordingly, the data in
the literature, which allow a reliable comparison of turnover
frequencies for MoOx/SBA-15 catalysts, are rather limited.
11468
dx.doi.org/10.1021/ja3011989 | J. Am. Chem. Soc. 2012, 134, 11462−11473

Journal of the American Chemical Society
Article
Instead, homogeneous catalysts anchored on silica may
represent suitable reference systems. The optimized Schrock-
type Mo−alkylidene complexes, which are among the most
active propene metathesis catalysts, show, when anchored on
silica, much higher turnover frequencies (e.g., TOF of 1.9 s−1 at
303 K) than MoOx/SBA-15 (∼0.15 s−1, Table 1).48 The
difference probably arises from a different geometric and
electronic structure of the carbene centers. Substantial influence
of the structure on the activity has been demonstrated for
structurally well-defined catalysts prepared by grafting of
organometallic complexes on silica.49 The structure sensitivity
has been confirmed by theoretical studies, in which various
anchoring configurations of oxo-molybdenum carbene sites on
silica50 and alumina were considered.51−55
MoOx/SBA-15 contains molybdenum atoms in a multitude
of different geometric arrangements, which are hardly
distinguishable by spectroscopic techniques. Therefore, we
follow a bottom-up approach starting with the elucidation of
the building mechanism of the active carbene sites, which
comprise only the minor fraction of Mo atoms that are
characterized by a specific and catalytically favorable local
geometric environment.
propene and subsequent ethene-d4 adsorption revealed the
slow formation of isopropoxy species followed by oxidation to
coordinated acetone (Figure 5A) during the genesis of
metathesis activity (Figure 5B). Experimental59 and theoreti-
cal59−61 studies have shown that formation of surface
isopropoxide readily occurs by protonation upon adsorption
of an olefin on Brønsted acidic catalysts with the heat of olefin
adsorption of about 80−100 kJ mol−1. These values agree well
with the differential heats of propene adsorption measured by
microcalorimetry (Figure 3). Having evidence for the presence
of Brønsted acid sites in the catalyst (Figure 1B, Figure 5A), we
consider that the major event upon propene adsorption is the
formation of isopropoxide species through protonation of the
olefin by Brønsted acid sites.
The subsequent event detected by IR spectroscopy during
the genesis of the metathesis activity is the formation of
coordinated acetone (Figure 5A). Although the formation of
carbonyl compounds is in line with the pseudo-Wittig
mechanism (Scheme 1D), it is rather reasonable to consider
that carbonyl compounds are formed through oxidation of the
isopropoxy intermediate on the redox-active MoOx/SBA-15
catalyst. It is well-known that the oxidation of propene to
acetone via the isopropoxy intermediate takes place with acidic
molybdena based mixed oxide catalysts at relatively low (393−
473 K) temperature.62 We consider that the same type of
reaction occurs in the present case. The formation of a carbonyl
compound should involve the reduction of molybdenum
centers, i.e., Mo(VI) to Mo(IV), if the reduction takes place
at a single molybdenum atom.
4.2. Formation Route of Carbene Sites. The genesis of
Mo−carbene sites is a consequence of the reaction between
surface molybdena species and the reactant propene.
The pseudo-Wittig mechanism (Scheme 1D) has persisted to
be the only mechanism for three decades since Rappe and
Goddard proposed it based on purely theoretical consid-
erations. However, the detected high heat of propene
adsorption (78
2 kJ mol−1, Figure 3A) is not in accord
Importantly, coordinatively unsaturated monomeric Mo(IV)
centers are known to be good precursors of Mo−carbene sites
that are formed via an oxidative addition of the propene
molecule to Mo(IV) surface species.15,16,64−66 Actually, most of
highly active molybdenum oxide-based metathesis catalysts are
prepared via efficient creation of coordinatively unsaturated
monomeric Mo(IV) centers by a reductive pretreatment of the
catalyst.46,64 This inspires us to propose a new route for the
carbene formation presented in Scheme 3A.
with this route. The pseudo-Wittig mechanism has estimated
significantly positive Gibbs energies for the carbene formation,
especially in the case of supported molybdena catalysts56−58
(e.g., ΔG300 = +16 kJ mol−1 for the reaction of Cl2MoO2 and
ethene; estimated by an ab initio calculation,21 ΔE = ca. +60 kJ
mol−1 for the reaction of a supported molybdena (Mo(VI)-oxo
centers in a β zeolite cluster with propene; estimated by a DFT
calculation57). Accordingly, the pseudo-Wittig mechanism is a
thermodynamically unfavorable route. In previous experimental
reports concerning fully oxidized supported W24 and Re7
catalysts, the detection of carbonyl compounds during the
genesis of metathesis activity and the dependency of the
reactivity from the type of reacting olefin have been considered
as experimental proofs for the pseudo-Wittig mechanism.7,24
On the basis of our calorimetric and spectroscopic findings
during the genesis of metathesis activity, we propose an
alternative mechanism (Scheme 3), which is not in contra-
diction with the reported experimental results mentioned
above.
On the surface of MoOx/SBA-15, molybdenum occurs in its
highest oxidation state in form of highly dispersed surface
molybdenum oxide species that comprise terminating oxo
(MoO) groups. Brønsted acidity, probably originated by
Mo−OH groups, and Lewis acidity due to coordinatively
unsaturated Mo(VI) centers are found to be present as well
(Section 3.1).
The quantitative agreement between the concentrations of
strong adsorption sites found by microcalorimetry (Figure 3A)
and the postreaction active site titration (Table 1) suggests that
the strong exothermic adsorption of propene on the catalyst is
in relationship with the genesis of the carbene sites.
The first step involves the reduction of Mo(VI) to Mo(IV)
via formation of an isopropoxy species by an acid−base reaction
and the subsequent oxidative dehydrogenation of the
isopropoxide intermediate to acetone (Scheme 3A, 1−3).
The second step includes desorption of acetone followed by an
oxidative addition of another propene molecule to the Mo(IV)
center to form the active Mo(VI)-carbene species (Scheme 3A,
3−5). Consequently, bifunctionality of the catalyst compre-
hending Brønsted acidity and the ability to accomplish
oxidative dehydrogenation is considered as an essential
requirement for the first step. The carbene formation via an
oxidative addition of a propene molecule to Mo(IV) is a well
established event,15,16,64−66 which likely takes place via 1,2-
hydrogen shift mechanism (as shown in Scheme 1A). The
positive effect of acidity in olefin metathesis is well recognized
by the observation that the use of acidic silica−alumina
supports yields relatively active catalysts.67−69 It is also known
that reducibility is relevant to the metathesis activity.70 The
present proposal is quite consistent with these facts and
provides a mechanistic explanation for the role of acidity and
reducibility. The slow development of the metathesis activity
(Figure 2) compared to silica−alumina supported molybdena
catalysts10−12 might be related to the limited acidity of MoOx/
SBA-15. Considering progressive slow formation of isoprop-
oxide and acetone in the in situ IR experiment, these surface
IR spectroscopy provides the explanation for the exothermic
events during the propene adsorption. The IR spectra for
11469
dx.doi.org/10.1021/ja3011989 | J. Am. Chem. Soc. 2012, 134, 11462−11473

Journal of the American Chemical Society
Article
reactions (Scheme 3A 1−3) most likely represent the kinetic
bottleneck of the evolution of active carbene sites. It is noted
that the color of the catalyst bed remained white during the
metathesis reaction, suggesting the absence of reduced
molybdenum centers. Probably, the desorption of acetone
and subsequent oxidative addition of propene (Scheme 3A 3−
5) are much faster than the foregoing steps in the case of
MoOx/SBA-15, which gives rise to a very small concentration
of the Mo(IV) intermediate.
program with a heating rate of 10 K min−1. As a consequence,
we succeeded to double the activity and to eliminate the
induction period (Figure 6). The heating after propene
Besides Brønsted acidity and reducibility, the formation of
Mo−carbene species requires desorption of the coordinated
acetone (Scheme 3A, 3−4) and subsequent oxidative addition
of another propene molecule (Scheme 3A, 4−5). We consider
that strict dehydration of the catalyst and abundant presence of
propene molecules are indispensable requirements to achieve
these steps. The oxidation of the olefin to a carbonyl compound
via the alkoxy intermediate (Scheme 3A, 1−4) is accelerated by
the presence of water;62 whereas, water as well as carbonyl
compounds are known to inhibit the metathesis reaction.63,71,72
Although the presence of water might be favorable for the
oxidation of propene via an alkoxy intermediate (Scheme 3A,
1−4), it is expected that the presence of water or carbonyl
compounds lead to the decomposition of carbene sites, or more
likely, disturb the carbene formation (Scheme 3A, 4−5) by
blocking Mo(IV) sites. In fact, we observed lower catalytic
activity accompanied by the color change of the catalyst from
white to bluish when we use a less purified propene feed (i.e.,
the reaction was done using the same propene feed but without
placing the water-trapping silica guard beds above and below
the catalyst bed (see the SI for experimental details)), while the
catalyst color remained white after 15 h of time on stream in
standard reaction condition. The blue coloration observed in
the experiment using less purified feed indicates the formation
of reduced molybdenum species that are stabilized and blocked
by the adsorption of water ligands, which suggests that the
oxidative addition of propene was disturbed, likely by water
contamination. It is well-known that pretreatment of the
catalyst at elevated temperature (typically 773−873 K) in dry
atmosphere is certainly necessary to obtain the catalytic activity.
We consider that the dehydrated surface acts as an efficient trap
for acetone and any byproduct of the reaction between surface
molybdena with propene allowing the access of the second
propene molecule to Mo(IV) sites, which leads to the
generation of carbene species. It has been established that
surface molybdena73 and silanol groups33 undergo structural
reconstruction during the dehydration at elevated temperature,
leaving surface sites which readily absorb water33 and acetone.74
Probably, the presence of this kind of dehydrated sites on the
catalyst in a dry propene feed facilitates efficient trapping of
acetone and access of the second propene molecule, allowing
the formation of carbene sites (Scheme 3A, 3−5). In fact, we
detected no carbonyl compounds (e.g., acetone) in the reactor
effluent during the propene metathesis by mass spectrometry,
indicating that generated carbonyl compounds are trapped on
the catalyst surface.
Figure 6. Propene metathesis activity of MoOx/SBA-15 (323 K,
contact time = 0.35 s g mL−1) after different regeneration procedures.
(1) Standard oxidative regeneration (20% O2 flow at 823 K for 0.5 h)
was performed. (2) In addition to the standard oxidative regeneration,
the catalyst was treated in neat propene flow at 300 K for 1 h followed
by heat treatment in argon flow at 823 K (heating rate 10 K min−1) for
0.5 h.
adsorption at 300 K accelerates the oxidation of surface
isopropoxide species, forming Mo(IV) sites. During further
temperature rise, desorption of oxidation products (oxygenates
and possibly water) is accelerated, which leaves bare Mo(IV)
sites that are the precursors of the active carbene sites in
metathesis. Thus, the heating has two effects: promotion of
oxidation and desorption of the oxidation products. We
demonstrate here how a knowledge-based approach efficiently
improves catalytic performance. Systematic optimization of the
activation protocol as well as the control of catalyst structure
according to the present knowledge is required to improve the
catalytic performance further.
We further consider the suggested route and criteria of
carbene formation is applicable to other important heteroge-
neous metathesis catalysts, i.e., supported tungsten(VI) oxide
and rhenium(VII) oxide. These oxides are also acidic, reducible,
and deposited on “desiccant-like” high surface area supports.
The reducibility decreases in the order Re(VII) > Mo(VI) >
W(VI) in general; this ranking is in agreement with the activity
in olefin metathesis. These facts are quite consistent with the
present conclusion.
4.3. Molecular Structure of the Carbene Sites. According to
the proposed mechanism, successful precatalyst sites should
fulfill multiple functions, namely, Brønsted acidity, capability to
perform isopropoxide oxidation and trapping capacity with
respect to acetone and water. Apparently, only few grafted
molybdenum oxide species simultaneously meet all of these
criteria. It is, therefore, understandable why only 1.5% of the
existing molybdenum atoms take place in catalysis.
Raman (Figure S1 of the SI) and UV−vis (Figure S2 of the
SI) spectroscopy suggest that monomeric dioxo molybdena
species in tetrahedral geometry are the predominant
species,28,31,75−77 although a certain amount of non-nomomeric
molybdena species probably coexist.12,78,79 Monomeric molyb-
dena has been suggested to be the relevant catalyst
precursors.12 In homogeneous metathesis, four-coordinated
monomeric Mo(VI)−alkylidene species represent the common
general structure of active Mo-based metathesis catalysts.63 In
accordance with this fact, recent DFT calculations52,55,80 as well
To prove these assumptions, we have performed an
additional experiment, which validated the hypothesis estab-
lished here and succeeded in improving the catalytic perform-
ance. Before starting the metathesis reaction, we inserted a
propene adsorption−desorption pretreatment. After the stand-
ard oxidative regeneration, the MoOx/SBA-15 catalyst was
exposed to neat propene at 300 K followed by thermal
treatment in flowing argon at 823 K applying a temperature
11470
dx.doi.org/10.1021/ja3011989 | J. Am. Chem. Soc. 2012, 134, 11462−11473

Journal of the American Chemical Society
Article
as experimental studies16,47 using well-defined catalysts with
low Mo loadings have confirmed the metathesis activity of four-
coordinated monomeric Mo(VI)−alkylidene species in the
class of supported Mo catalysts. Furthermore, if the oxidation of
isopropoxide occurs at nonmonomeric molybdena having Mo−
O−Mo bonding, the two-electron reduction of molybdenum
(i.e., oxidation of isopropoxide to acetone) may possibly result
in the formation of two Mo(V) centers instead of the formation
of a Mo(IV) site that is the precursor of a carbene site.
Monomeric molybdena species are likely surrounded by
reactive silanol groups and Si−O−Si bonds created during
the high-temperature pretreatment step, which provide the
trapping function. Hence, monomeric molybdena species and
resulting monomeric Mo(VI)−alkylidene species are the most
likely candidates for precatalysts and active sites. Taking into
account the presence of Mo−OH groups (i.e., Brønsted acid
sites) and their involvement in the formation of carbene
species, we propose geometrically accessible tetrahedral
monomeric dioxo species with a molybdenol group (Scheme
3B 1) as a candidate for the relevant precatalyst.
Scheme 3B describes the detailed scheme assuming the 1,2-
hydrogen shift mechanism (Scheme 1A) in the oxidative
addition step (Scheme 3B 5−7). Upon the formation of
acetone (Scheme 3B 2−3), the released proton is likely hosted
by the oxo oxygen on the same molybdenum atom in
cooperation with concurrent reduction of the molybdenum
atom, leaving a new molybdenol group (Scheme 3B 3). It is
speculated that this molybdenol further undergoes addition of
another propene molecule to yield an isopropoxide group
owing to its acidic character (Scheme 3B, 7−8). Accordingly,
the final molecular structure of Mo carbene species (Scheme
3B, 8) features an isopropoxide group, an anchoring Mo−O−Si
bond, an oxo group, and an alkylidene group. Actually, this
structure is an analogue of Schrock-type Mo(VI)−alkylidene
complexes (Scheme 3C).63 While Schrock-type complexes
comprise two alkoxy ligands and an imido ligand besides the
alkylidene ligand, the suggested surface Mo(VI)−alkylidene has
an oxo ligand and an anchoring bond instead of an imido ligand
and an alkoxy ligand, respectively. Insight obtained in the
present study allows us to clearly envisage the molecular
structure of active sites in supported transition metal oxide-
derived heterogeneous catalysts, opening a possibility to
address the material gap between homogeneous and heteroge-
neous catalysts. Further information concerning the molecular
structure of the precursor surface molybdenum oxide species
that can readily undergo transformation to Mo−carbene sites
will bring deeper understanding and will allow the rational
design of the catalyst. The specifically low turnover frequency
of the present system indicates that the actual carbene structure
is not optimal and improvements might be possible.
Modification of the electronic structure of the active centers
by appropriate support modification is the design target
mimicking the N-containing ligands of the molecular analogues.
uct capture, and oxidative addition. In the present propene
metathesis over MoOx/SBA-15, only 1.5% of total Mo atoms
accomplish this process. The mechanistic insight into the
genesis of active carbene sites obtained by microcalorimetry
and in situ IR delivers the explanation for the low abundance of
active species, and provides the criteria for successful molecular
structures of the precatalyst, which allows us to envisage
resulting configurations of metal carbenes by applying
established general mechanisms of acid−base and red-ox
chemistry. To our surprise, we find structural similarity between
the anticipated surface Mo(VI)−alkylidene and highly active
Schrock-type catalysts. This encourages us to implement inputs
from homogeneous catalysis in prospective catalyst synthesis
approaches.
We embark two strategies toward catalyst improvement
based on the gained insight: (1) controlling the surface carbene
synthesis, and (2) knowledge-based design of the precatalyst.
Concerning strategy (1), the simple temperature pretreat-
ment (Figure 6) demonstrates that favorable activation
conditions can assist the carbene formation disclosing a great
and, in particular, easy option. Moreover, further control and
optimization of the alkyl ligand in the carbene species should be
possible by extending the choice of activating reagents instead
of using merely the reactant itself. For example, use of
isobutene instead of propene would facilitate the formation of
the alkoxide intermediate due to increased stability of the tert-
butoxide. In addition, the presence of the bulky tert-butoxide
group at the final carbene site would contribute to improve the
stability of the carbene species. Another option would be the
use of alcohols (e.g., methanol, tert-butanol) for efficient
creation of reduced metal centers (i.e., precursor for the
carbene sites). By using alcohols, Brønsted acid sites are not
necessarily required. The knowledge in homogeneous catalysis
and organometallic synthesis will strongly contribute to the
selection of prospective activating agents.
Strategy (2) benefits from the spatial separation of metal
species that can be accomplished on the surface of a
heterogeneous catalyst. Irreversible anchoring prevents self-
condensation of complexes, which is a major difficulty in
homogeneous catalysis. The presence of highly reactive
molybdenol groups accounting for Brønsted acidity can be
implemented within transition metal oxide species anchored on
solid surfaces. Synthesis of isolated surface molybdenum oxide
moieties having molybdenol ligands is a specific target, since
our microcalorimetry results indicate that the amount of
Brønsted acid sites is a limiting factor with respect to the
number of active centers in the final catalyst.
Building bridges in catalysis research on the basis of our case
study propene metathesis, we demonstrate here that the highly
interdisciplinary field of heterogeneous catalysis will find more
solid link to related fields in chemistry through in-depth
understanding.
5. CONCLUSIONS
We contribute to understanding in catalysis through structural
identification of active sites on the surface of a heterogeneous,
meso-structured molybdenum oxide model catalyst for propene
metatheses applying operando techniques on a strictly
quantitative level. The genesis of active sites in such a system
is a demanding in situ and one-pot synthesis of metal carbene
species from surface metal oxides, which involves Brønsted
acid−base chemistry, oxidation−reduction processes, byprod-
ASSOCIATED CONTENT
■
S
* Supporting Information
Nitrogen adsorption data, Raman and UV−vis spectra of the
calcined catalyst and support, IR spectra of isopropanol and
acetone adsorption, as well as experimental details. This
material is available free of charge via the Internet at http://
pubs.acs.org.
11471
dx.doi.org/10.1021/ja3011989 | J. Am. Chem. Soc. 2012, 134, 11462−11473

Journal of the American Chemical Society
Article
(26) Zhao, D.; Feng, J.; Huo, Q.; Melosh, N.; Fredrickson, G. H.;
Chmelka, B. F.; Stucky, G. D. Science 1998, 279, 548−552.
(27) Thielemann, J. P.; Weinberg, G.; Hess, C. ChemCatChem 2011,
3, 1814−1821.
AUTHOR INFORMATION
■
Corresponding Author
trunschke@fhi-berlin.mpg.de
(28) Tian, H.; Roberts, C. A.; Wachs, I. E. J. Phys. Chem. C 2010, 114,
14110−14120.
Notes
The authors declare no competing financial interest.
(29) Hu, H.; Wachs, I. E.; Bare, S. R. J. Phys. Chem. 1995, 99, 10897−
10910.
(30) Vuurman, M. A.; Wachs, I. E. J. Mol. Catal. 1992, 77, 29−39.
(31) Lee, E. L.; Wachs, I. E. J. Phys. Chem. C 2007, 111, 14410−
14425.
(32) Dieterle, M.; Weinberg, G.; Mestl, G. Phys. Chem. Chem. Phys.
2002, 4, 812−821.
(33) Davydov, A. Molecular Spectroscopy of Oxide Catalyst Surface;
John Wiley & Sons Ltd.: Chichester, 2003.
(34) Zhang, B.; Li, Y.; Lin, Q.; Jin, D. J. Mol. Catal. 1988, 46, 229−
241.
(35) Langmuir, I. J. Am. Chem. Soc. 1916, 38, 2221−2295.
(36) Grabowski, R.; Efremov, A.; Davydov, A.; Haber, E. Kinet. Catal.
1981, 22, 794−797.
(37) Efremov, A.; Lokhov, Y.; Davydov, A. Kinet. Catal. 1981, 22,
969−975.
́
(38) Martín, C.; Rives, V.; Sanchez-Escribano, V.; Busca, G.;
Lorenzelli, V.; Ramis, G. Surf. Sci. 1991, 251−252, 825−830.
(39) Finocchio, E.; Busca, G.; Lorenzelli, V.; Willey, R. J. J. Chem.
Soc., Faraday Trans. 1994, 90, 3347.
ACKNOWLEDGMENTS
■
The authors thank M. Hashagen for her help with the IR
experiments, G. Lorenz for measuring N2 physisorption, Dr. F.
Girgsdies, and E. Kitzelmann for performing XRD analysis, Dr.
T. Cotter for Raman experiments, and A. Klein-Hoffmann for
XRF measurements. Dr. B. Frank is acknowledged for valuable
discussion. This work was conducted in the framework of the
COE “UniCat” (www.unicat.tu-berlin.de) of the German
Science Foundation. K.A. is grateful to Mitsubishi Gas
Chemical Co. Inc. for a fellowship.
REFERENCES
■
(1) Dumesic, J. A.; Huber, G. W.; Boudart, M. In Handbook of
Heterogeneous Catalysis, 2nd ed.; Wiley-VCH: Weinheim, 2008; pp 1−
15.
(2) Somorjai, G. A.; Li, Y. Introduction to Surface Chemistry and
Catalysis; 2nd ed.; John Wiley & Sons: Hoboken, NJ, 2010.
(3) Libuda, J.; Freund, H.-J. Surf. Sci. Rep. 2005, 57, 157−298.
(4) Rupprechter, G.; Weilach, C. J. Phys.: Condens. Matter 2008, 20,
184019.
(40) Goncharova, O.; Davydov, A. React. Kinet. Catal. Lett. 1983, 23,
285−289.
(41) Davydov, A.; Efremov, A. Kinet. Catal. 1983, 24, 1214−1220.
(42) Efremov, A.; Davydov, A. React. Kinet. Catal. Lett. 1981, 18,
363−366.
(5) Handzlik, J.; Ogonowski, J. Catal. Lett. 2003, 88, 119−122.
(6) Chauvin, Y. Angew. Chem., Int. Ed. 2006, 45, 3740−3747.
(43) Goncharova, O.; Davydov, A.; Yureva, T. Kinet. Catal. 1984, 25,
124−129.
(7) Salameh, A.; Coper
M. Adv. Synth. Catal. 2007, 349, 238−242.
́
et, C.; Basset, J.-M.; Bohm, V. P. W.; Roper,
̈ ̈
(44) Sanchez Escribano, V.; Busca, G.; Lorenzelli, V. J. Chem. Phys.
1991, 95, 5541−5545.
(8) Chauvin, Y.; Commereuc, D. J. Chem. Soc., Chem. Commun. 1992,
462−464.
(45) Efremov, A. A.; Davydov, A. A. React. Kinet. Catal. Lett. 1981,
18, 353−356.
(9) Mol, J. J. Mol. Catal. A: Chem. 2004, 213, 39−45.
(10) Debecker, D. P.; Stoyanova, M.; Colbeau-Justin, F.; Rodemerck,
(46) Vikulov, K. A.; Shelimov, B. N.; Kazansky, V. B. J. Mol. Catal.
1991, 65, 393−402.
̀
U.; Boissiere, C.; Gaigneaux, E. M.; Sanchez, C. Angew. Chem., Int. Ed.
2012, 51, 2129−2131.
(47) Vikulov, K. A.; Elev, I. V.; Shelimov, B. N.; Kazansky, V. B. J.
Mol. Catal. 1989, 55, 126−145.
(11) Debecker, D. P.; Bouchmella, K.; Stoyanova, M.; Rodemerck,
U.; Gaigneaux, E. M.; Mutin, P. H. Catal. Sci. Technol. 2012, 2, 1157−
1164.
(48) Blanc, F.; Rendon, N.; Berthoud, R.; Basset, J.-M.; Coperet, C.;
Tonzetich, Z. J.; Schrock, R. R. Dalton Trans. 2008, 3156−3158.
(12) Debecker, D. P.; Schimmoeller, B.; Stoyanova, M.; Poleunis, C.;
Bertrand, P.; Rodemerck, U.; Gaigneaux, E. M. J. Catal. 2011, 277,
154−163.
(13) Debecker, D. P.; Stoyanova, M.; Rodemerck, U.; Gaigneaux, E.
M. J. Mol. Catal. A: Chem. 2011, 340, 65−76.
(14) Debecker, D.; Bouchmella, K.; Poleunis, C.; Eloy, P.; Bertrand,
P.; Gaigneaux, E.; Mutin, P. Chem. Mater. 2009, 21, 2817−2824.
(15) Iwasawa, Y.; Hamamura, H. J. Chem. Soc., Chem. Commun. 1983,
130−132.
(16) Iwasawa, Y.; Kubo, H.; Hamamura, H. J. Mol. Catal. 1985, 28,
191−208.
(17) McCoy, J. R.; Farona, M. F. J. Mol. Catal. 1991, 66, 51−58.
(18) Farona, M. F.; Tucker, R. L. J. Mol. Catal. 1980, 8, 85−90.
(19) Grubbs, R. H.; Swetnick, S. J. J. Mol. Catal. 1980, 8, 25−36.
(20) Laverty, D. T.; Rooney, J. J.; Stewart, A. J. Catal. 1976, 45, 110−
113.
(49) Rendon
Basset, J.-M.; Coper
́
, N.; Berthoud, R.; Blanc, F.; Gajan, D.; Maishal, T.;
́
et, C.; Lesage, A.; Emsley, L.; Marinescu, S. C.;
Singh, R.; Schrock, R. R. Chem.Eur. J. 2009, 15, 5083−5089.
(50) Handzlik, J. J. Phys. Chem. B 2005, 109, 20794−20804.
(51) Handzlik, J. J. Catal. 2003, 220, 23−34.
(52) Handzlik, J. Surf. Sci. 2007, 601, 2054−2065.
(53) Handzlik, J.; Ogonowski, J.; Tokarz-Sobieraj, R. Catal. Today
2005, 101, 163−173.
(54) Handzlik, J. Surf. Sci. 2004, 562, 101−112.
(55) Handzlik, J.; Sautet, P. J. Catal. 2008, 256, 1−14.
(56) Li, X.; Guan, J.; Zheng, A.; Zhou, D.; Han, X.; Zhang, W.; Bao,
X. J. Mol. Catal. A: Chem. 2010, 330, 99−106.
(57) Guan, J.; Yang, G.; Zhou, D.; Zhang, W.; Liu, X.; Han, X.; Bao,
X. Catal. Commun. 2008, 9, 2213−2216.
(58) Li, X.; Zheng, A.; Guan, J.; Han, X.; Zhang, W.; Bao, X. Catal.
Lett. 2010, 138, 116−123.
(59) Campbell, K. A.; Janik, M. J.; Davis, R. J.; Neurock, M. Langmuir
2005, 21, 4738−4745.
(60) Kazansky, V. B. Catal. Today 1999, 51, 419−434.
(61) Engelhardt, J.; Hall, W. K. J. Catal. 1995, 151, 1−9.
(62) Moro-oka, Y. Appl. Catal., A 1999, 181, 323−329.
(63) Schrock, R. R.; Hoveyda, A. H. Angew. Chem., Int. Ed. 2003, 42,
4592−4633.
(64) Iwasawa, Y.; Ichinose, H.; Ogasawara, S.; Soma, M. J. Chem. Soc.,
Faraday Trans. 1 1981, 77, 1763−1777.
(21) Rappe, A. K.; Goddard, W. A. J. Am. Chem. Soc. 1982, 104, 448−
456.
(22) Grunert, W.; Stakheev, A. Y.; Feldhaus, R.; Anders, K.; Shpiro,
̈
E. S.; Minachev, K. M. J. Catal. 1992, 135, 287−299.
(23) Chen, X.; Zhang, X.; Chen, P. Angew. Chem., Int. Ed. 2003, 42,
3798−3801.
(24) Basrur, A. G.; Patwardhan, S. R.; Was, S. N. J. Catal. 1991, 127,
86−95.
(25) Howman, E. J.; Mgcrath, B. P.; Williams, K. V. Catalyst
regeneration process. British Patent 1,144,085, March 5, 1969.
11472
dx.doi.org/10.1021/ja3011989 | J. Am. Chem. Soc. 2012, 134, 11462−11473

Journal of the American Chemical Society
Article
(65) Shelimov, B. N.; Elev, I. V.; Kazansky, V. B. J. Mol. Catal. 1988,
46, 187−200.
(66) Elev, I.; Shelimov, B.; Kazansky, V. Kinet. Catal. 1989, 30, 787−
792.
(67) Handzlik, J.; Ogonowski, J.; Stoch, J.; Mikolajczyk, M.;
Michorczyk, P. Appl. Catal., A 2006, 312, 213−219.
(68) Aritani, H.; Fukuda, O.; Yamamoto, T.; Tanaka, T.; Imamura, S.
Chem. Lett. 2000, 66−67.
(69) Debecker, D. P.; Hauwaert, D.; Stoyanova, M.; Barkschat, A.;
Rodemerck, U.; Gaigneaux, E. M. Appl. Catal., A 2011, 391, 78−85.
(70) Thomas, R.; Moulijn, J. A. J. Mol. Catal. 1982, 15, 157−172.
(71) Mol, J. C.; Leeuwen, P. W. N. M. Metathesis of Alkens in
“Handbook of Heterogeneous Catalysis, 2nd ed.; Wiley-VCH: Weinheim,
2008.
(72) Dinger, M. B.; Mol, J. C. Organometallics 2003, 22, 1089−1095.
(73) Boer, M.; Dillen, A. J.; Koningsberger, D. C.; Geus, J. W.;
Vuurman, M. A.; Wachs, I. E. Catal. Lett. 1991, 11, 227−239.
̀
(74) Crocella, V.; Cerrato, G.; Magnacca, G.; Morterra, C. J. Phys.
Chem. C 2009, 113, 16517−16529.
(75) Lee, E. L.; Wachs, I. E. J. Phys. Chem. C 2008, 112, 6487−6498.
(76) Chempath, S.; Zhang, Y.; Bell, A. T. J. Catal. 2007, 111, 1291−
1298.
(77) Guo, C. S.; Hermann, K.; Havecker, M.; Thielemann, J. P.;
̈
Kube, P.; Gregoriades, L. J.; Trunschke, A.; Sauer, J.; Schlogl, R. J.
̈
Phys. Chem. C 2011, 115, 15449−15458.
(78) Thielemann, J. P.; Krohnert, J.; Hess, C. J. Phys. Chem. C 2010,
̈
114, 17092−17098.
(79) Thielemann, J. P.; Ressler, T.; Walter, A.; Tzolova-Muller, G.;
Hess, C. Appl. Catal., A 2011, 399, 28−34.
(80) Handzlik, J. J. Phys. Chem. C 2007, 111, 9337−9348.
̈
11473
dx.doi.org/10.1021/ja3011989 | J. Am. Chem. Soc. 2012, 134, 11462−11473