Insights into the deactivation mechanism of supported tungsten hydride on alumina (W-H/Al2O3) catalyst for the direct conversion of ethylene to propylene

Article abstract of DOI:10.1016/j.molcata.2014.01.025

Tungsten hydride supported on alumina prepared by the surface organometallic chemistry method is an active precursor for the direct conversion of ethylene to propylene at low temperature and pressure. An extensive contact time study revealed that the dimerization of ethylene to 1-butene is the primary and also the rate limiting step. The catalytic cycle further involves isomerization of 1-butene to 2-butene, followed by cross-metathesis of ethylene and 2-butene to yield propylene with high selectivity. The deactivation mechanism of this reaction has been investigated. The used catalyst was extensively examined by DRIFTS, solid-state NMR, EPR, UV-Vis, TGA and DSC techniques. It was found that a large amount of carbonaceous species, which were due to side reaction like olefin polymerization took place with time on stream, significantly hindering the dimerization of ethylene to 1-butene and therefore the production of propylene. Crown Copyright

Full text of DOI:10.1016/j.molcata.2014.01.025

Journal of Molecular Catalysis A: Chemical 385 (2014) 125–132
Contents lists available at ScienceDirect
Journal of Molecular Catalysis A: Chemical
journal homepage: www.elsevier.com/locate/molcata
Insights into the deactivation mechanism of supported tungsten
hydride on alumina (W-H/Al O ) catalyst for the direct conversion of
2
3
ethylene to propylene
a
a
a
a
a
Etienne Mazoyer , Kai C. Szeto , Nicolas Merle , Jean Thivolle-Cazat , Olivier Boyron ,
a,1
b,∗
a,∗
Jean-Marie Basset , Christopher P. Nicholas , Mostafa Taoufik
Université Lyon 1, C2P2, CNRS UMR 5265, ICL, CPE Lyon, 43 Bd du 11 Novembre 1918, 69616 Villeurbanne, France
Exploratory Catalysis Research, UOP LLC, A Honeywell Company, 25 East Algonquin Road, Des Plaines, IL 60017, USA
a
b
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 9 August 2013
Received in revised form 26 January 2014
Accepted 27 January 2014
Available online 3 February 2014
Tungsten hydride supported on alumina prepared by the surface organometallic chemistry method is
an active precursor for the direct conversion of ethylene to propylene at low temperature and pressure.
An extensive contact time study revealed that the dimerization of ethylene to 1-butene is the primary
and also the rate limiting step. The catalytic cycle further involves isomerization of 1-butene to 2-butene,
followed by cross-metathesis of ethylene and 2-butene to yield propylene with high selectivity. The deac-
tivation mechanism of this reaction has been investigated. The used catalyst was extensively examined
by DRIFTS, solid-state NMR, EPR, UV–Vis, TGA and DSC techniques. It was found that a large amount of
carbonaceous species, which were due to side reaction like olefin polymerization took place with time
on stream, significantly hindering the dimerization of ethylene to 1-butene and therefore the production
of propylene.
Keywords:
Heterogeneous catalysis
TGA/DSC
Solid state NMR
Polymerization
Catalyst deactivation
Crown Copyright © 2014 Published by Elsevier B.V. All rights reserved.
1
. Introduction
The consumption of propylene has constantly increased dur-
steam cracking capacity built in the near future will be based on
using ethane as a feedstock, which typically produces only ethylene
as a final product [17]. In this context, a new process that allows the
selective conversion of ethylene to propylene will be economically
relevant.
ing the last past years. The reason is directly linked to the high
demand of propylene-derived products, like polypropylene, pro-
pylene oxide and acrylonitrile [1,2]. Hence, on-purpose propylene
production has become more important in addition to classical
cracking methods. These approaches include propane dehydro-
genation [3–5], olefin metathesis [6–11], methanol to olefins
The conversion of three molecules of ethylene to two molecules
of propylene, as presented in Scheme 1, is highly thermody-
namically favored. Below 900 K the standard Gibbs free energy is
negative, affording a large conversion of ethylene at equilibrium.
Moreover, under 600 K, the conversion of ethylene is complete
at the equilibrium (Fig. S1). Only few examples of the direct
transformation of ethylene into propylene have been reported.
For instance, a deficiency of ethylene formation compared to
butenes has been observed during the metathesis of propyl-
ene catalyzed by [Mo(CO) ]/Al O and attributed to the direct
[
12,13] and methyl halides to olefins [14–16]. Currently, intensive
research efforts are focused on shale gas extraction, with the aim to
utilize this gas as the next generation feedstock for petroleum prod-
ucts. The ethane content of shale gas can be up to 16%, the rest is
mainly methane like natural gas. Therefore, much of the additional
6
2
3
transformation of ethylene into propylene [18]. This reaction has
also been observed, with a short lifetime, during Fischer–Tropsch
synthesis in the presence of small iron nanoclusters formed
by the thermal decomposition of [HFe3(CO)11]/MgO [19]. Other
reported catalysts for the transformation of ethylene to propyl-
ene are based on europium or ytterbium deposited on coal in
combination with titanium compounds (TiCl4, Ti(OiPr)4) [20] as
well as with WO3 on various oxides [21] MoOx/SiO2 [22] or
∗ Corresponding authors at: Université Lyon 1, C2P2, CNRS UMR 5265, ICL, 43 Bd
du 11 Novembre 1918, 69616 Villeurbanne, France; Exploratory Catalysis Research,
UOP LLC, A Honeywell Company, 25 East Algonquin Road, Des Plaines, IL 60017, USA.
Fax: +33 472431795/1 8473911287.
E-mail addresses: Christopher.Nicholas@Honeywell.com (C.P. Nicholas),
mostafa.taoufik@univ-lyon1.fr (M. Taoufik).
1
Current address: KCC, 4700 King Abdullah University of Science and Technology,
Thuwal 23955-6900, Saudi Arabia.
1
381-1169/$ – see front matter. Crown Copyright © 2014 Published by Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.molcata.2014.01.025

1
26
E. Mazoyer et al. / Journal of Molecular Catalysis A: Chemical 385 (2014) 125–132
2. Experimental
Scheme 1. Reaction of direct conversion of ethylene to propylene.
2.1. Catalyst preparation
The catalyst (WH /Al O ) was prepared according to pub-
3−(500)
3
2
lished procedure reported elsewhere [32] and consisted of two
Ru/SiO catalysts [23]. However, the catalytic performances of
all these systems were either not quantified or proven to be
very low as they were only observed in closed recirculation
systems.
steps, grafting and hydrogenolysis. The ␥-alumina (Rhone-Poulenc,
2
−1
◦
2
00 m g , XRD shown in Fig. S2) was dehydroxylated at 500 C
prior to the grafting reaction. Then, W(≡CC(CH ) )(CH C(CH ) )
3
3
2
3 3 3
◦
[
33] was contacted with the ␥-alumina at 66 C under argon to
A significant example of direct conversion of ethylene to pro-
pylene in terms of selectivity and productivity is reported using
zeolites [24]. During their investigation, Oikawa et al. shed light
upon the importance of the pore size and the acidity of the catalyst
to obtain a good selectivity and conversion, respectively. They have
identified SAPO-34 as the most efficient catalyst for direct ethylene
to propylene conversion. Its pore size is similar to the kinetic diame-
ter of the propylene, giving thus a good selectivity in this molecule,
and its acid strength is intermediate, affording the best conver-
obtain a well-defined W(≡CC(CH ) )(CH C(CH ) ) fragment on
3
3
2
3 3 2
the surface. Excess of the molecular complex was washed off with
dry pentane and the solid was dried under high vacuum. The second
◦
step comprised a treatment of the latter solid under H at 150 C.
The catalyst was then stored at −25 C in the glove box.
2
◦
2.2. Characterization
◦
Infrared spectra were recorded in Nicolet FTIR 6700 spectropho-
sion rate. When the reaction is performed at 450 C, after 1 h on
−
1
−1
tometer in diffuse reflectance mode, equipped with a MCT detector.
stream, a propylene production rate of 7.00 mmolC3H6
g
h
−1
Typically 64 scans with a resolution of 4 cm
were applied for
(
79% selectivity) is obtained. Another example of direct conversion
each spectrum. An air-tight cell with CaF2 window was employed.
UV–vis spectra of the samples were recorded in a diffuse reflectance
mode with a Perkin Elmer ␭-1050 spectrometer at room temper-
of ethylene to propylene has been described using nickel catalyst
supported on mesoporous silica [25,26]. In that example the cat-
◦
alyst shows a constant activity at 400 C, during 10 h on stream,
−
1
−
1
−1
ature in the range of 200–900 nm at a rate of 275 nm min and
InGaAs detector. Electron paramagnetic resonance (EPR) experi-
ments were performed on a Bruker EXELSYS spectrometer at room
temperature. A microwave frequency ꢀ of 9.813 GHz and a power
of 30 mW were used. The reactor cell for EPR measurements was
made of quartz (4 mm i.d.) and was normally charged with 0.01 g
samples. Thermogravimetric analysis (TGA) was performed using a
Mettler Toledo instrument TGA/DSC1. About 10 mg of sample was
with a propylene production of 1.85 mmolC3H6
g
h
and a
poor selectivity of 49%. The explanation was given by a mechanism
in which ethylene dimerization on Ni sites affords 1-butene that
can be isomerized to 2-butene on acidic sites. 2-Butene is supposed
to react via cross-metathesis reaction with non-reacted ethylene.
Iwamoto claims that Ni(I) are able to perform metathesis reaction
by the formation of a nickel carbene (Ni(III)) intermediate. The large
pore size of MCM-41 is the major argument presented to exclude an
acidic mechanism such as that presented for SAPO-34. Neverthe-
less, after calcination of impregnated nickel nitrate, it can be agreed
that the surface of the catalyst will exhibit silicates with non-
negligible acidity and the high temperature of reaction does not
exclude cracking of higher olefin to provide propylene as observed
by Oikawa et al. These reported catalysts require high working tem-
◦
accurately weighed in aluminium oxide pans and heated from 35 C
◦
◦
−1
−1
to 800 C at a rate of 20 C min . Air at a flow rate of 30 mL min
was used as the carrier gas. Differential scanning calorimetry (DSC)
analysis was performed using a Mettler Toledo DSC 1 instrument,
equipped with an auto-sampler. The temperature, the heat flow and
the tau lag of the equipment were calibrated with H O, an indium
2
◦
standard and a zinc standard. The sample was accurately weighed
perature (>400 C) to complete the classic acid catalytic cycle and
◦ ◦ ◦ −1
around 20 mg) and heated from 40 C to 160 C at 10 C min with
(
will inevitably undergo rapid catalyst deactivation by formation of
coke. An ultimate system will be a catalyst that works at low tem-
perature and gives high selectivity in propylene. Our laboratory
has been working extensively to develop alternative methods to
produce propylene by using alkene metathesis reactions [27–31].
In this context, we recently discovered a new catalytic reaction
which transforms ethylene directly to propylene at low temper-
an empty aluminum pan as reference. Two successive heating and
cooling were performed and only the second run was considered.
−
1
Dry nitrogen with a flow rate set at 30 mL min was used as the
purge gas. The melting temperature (Tm) was measured at the top
of the endothermic peak. The STARe thermal analysis software was
used to performed calculation on TGA and DSC data. 1 C CP-MAS
solid state NMR spectra were collected on a Bruker NMR AVANCE
3
◦
ature and pressure (150 C, 1 bar) with a selectivity higher than
5
00 spectrometer. The sample was filled in a zirconia impeller of
9
5% [29]. This reaction is catalyzed by tungsten hydride supported
◦
4 mm and then transferred into the probe with a rotation speed
of 10 kHz. Elemental analysis was performed at the Central Analy-
sis Service of the CNRS (Solaize, France) to determine the tungsten
loading (found: 5.5 wt% W).
on alumina dehydroxylated at 500 C, described elsewhere [32].
The initial postulated mechanism for the productivity of propylene
involves: (i) ethylene dimerization to 1-butene; (ii) 1-butene isom-
erization to 2-butenes; (iii) cross metathesis between ethylene and
2
-butenes. In this account, we report in details the performance of
this catalyst for direct conversion of ethylene to propylene. The
complete mechanism of this reaction was studied by identification
of the products released in early stage of catalysts and influence of
contact time. However, the tungsten hydride on alumina catalyst
used here undergoes catalyst deactivation with time on stream.
For further application in industry, the deactivation of the cata-
lyst needs to be understood in order to find ways to improve the
stability of the catalyst. Therefore, on-stream deactivation during
conversion of ethylene to propylene over tungsten hydride catalyst
was examined. TGA, DSC, solid state NMR, DRIFTS, EPR and UV–Vis
techniques were used to characterize the nature of the deactivated
catalyst.
2
.3. Catalyst evaluation
Catalytic performance in ethylene conversion was assessed in a
◦
stainless steel continuous flow reactor (P
flow rate = 4 mL min
fied with a column of molecular sieve and activated Cu O/Al O
3
and controlled by Brooks mass flow controllers. The catalyst was
charged in the glovebox. A 4-way valve allowed isolation of the
charged catalyst in the reactor from the environment and exten-
sive purging of the tubes. The analysis of the products was carried
out on an online gas chromatograph (HP 6890 GC) equipped with
a flame ionization detector (FID). Separation was performed on a
= 1 bar, T = 150 C,
C2H4
−1
−1
or VHSV = 400 h ). The gases were puri-
2
2

E. Mazoyer et al. / Journal of Molecular Catalysis A: Chemical 385 (2014) 125–132
127
Fig. 1. Conversion of ethylene (ꢀ), TON (ꢀ) and selectivity vs. time on stream.
5
0 m KCl/Al O column. Conversion and selectivity are calculated
ethylene in a batch reactor has also confirmed this equimolar evo-
lution of ethane. This finding suggests the formation of a surface
2
3
with respect to the carbon balance.
tungsten ethyl-ethylidene species, [W](CH CH )(=CHCH ) from
2
3
3
tungsten-trisethyl [W](CH CH ) , by the following elementary
steps insertion of three molecules of ethylene into the trishydride,
2
3 3
3
. Results and discussion
elimination of one ethane by ␣-H abstraction to form a carbene
3
.1. Direct conversion of ethylene to propylene over
(Scheme S1). When direct conversion of ethylene to propylene by
WH/Al O
2
3−(500)
WH/Al O is performed in dynamic reactor, it can be noticed
2
3−(500)
that the conversion is very high (37%) in the early stage of the catal-
ysis. The formation of the active species or ethylene adsorption
may also justify this high initial conversion. The mass balance of
the reaction has to be studied in order to discriminate the causes
of this phenomenon. During the course of the reaction, the mass
balance is studied using an ethylene feed with 10% of methane
The catalytic performances of WH/Al O3-(500), i.e. the cat-
2
alyst prepared from ␥-alumina pretreated under vacuum at
◦
5
00 C have been tested for the direct conversion of ethylene
to propylene. Propylene was obtained selectively (95%) by pass-
ing ethylene over WH/Al O
in a continuous-flow reactor
2
3−(500)
◦
−1
(
PC
= 1 bar, T = 150 C, flow rate 4 mL min or volume hourly
2H
4
(inert toward WH/Al O
nal standard. The ratios of ethylene/methane that come in and
out of the reactor toward methane have been plotted (respectively
in these conditions) [34] as inter-
2
3−(500)
−1
space velocity (VHSV) 400 h ). The reaction presents an initial
−1
1
−1
−1
steep maximal conversion rate of 0.6 molC2H4 molW
min
−
before reaching a pseudo plateau of 0.1 molC2H4 molW
min
C2=IN/CH and C
/CH on Fig. S4). In the same manner, the ratio
4
2=OUT
4
with an overall TON of 930 after 120 h on stream. The selec-
tivity for propylene increases rapidly up to a plateau at 95%,
while that of butenes remains below 5% (Fig. 1.). Higher olefins
are present in only trace amounts. These results correspond to
of products (everything but ethylene) that comes out from the reac-
tor has been plotted (ProductsOUT/CH4 on Fig. S4). Finally, in order
to check if the carbon balance is equilibrated, the sum of C2=OUT/CH4
and ProductsOUT/CH has been plotted ((C
+ ProductsOUT)/CH4
4
2=OUT
−1
−1
a productivity of 9.57 mmolC
g
h
at the maximum and
on Fig. S3). The result shows that at any time on stream, the carbon
balance is equilibrated with respect to the analytical equipment,
which implies that at each moment, the mass of gas going in and
coming out of the reactor is virtually equal to the mass of ethylene
coming in the reactor. Regarding these results it is clear that this
high initial activity is due to the catalytic activity and neither to the
formation of the active species nor to ethylene adsorption. A first
insight into the mechanism of the direct conversion of ethylene
to propylene is given by a careful study of the product selectiv-
ity at the early stage of the catalysis. Evolution of ethane shows
that ␣-H abstraction leading to the carbenic species occurs only for
the first 100 min (Fig. S3). At short time on stream, butene isomers
appear in the gas phase. The initial 1-butene/trans-2-butene/cis-2-
butene/isobutene isomeric distribution of 5.2:1:2:3.1 (t = 20 min)
changed over time, tending to the ratio 1.1:3.2:1:0.7 (Fig. S5). This
3H
6
−1
−1
2
.07 molC3H6
g
h
at pseudo plateau.
During the 10 first hours on stream, a fast deactivation rate
−
1
of the catalyst is observed (2.40% h ), before reaching a pseudo-
−
1
plateau with a much lower deactivation rate (0.05% h ). Despite
this phenomenon, the selectivity is constant for all catalytic run.
A short initiation period is observed in the early stage of cataly-
sis, showing that the tungsten hydride is only the precatalyst of
the direct conversion of ethylene to propylene. The initiation step
was elucidated by identifying the products formed, in dynamic
◦
reactor, while heating to the reaction temperature of 150 C. Inter-
estingly a non-negligible amount of ethane is observed during
the 100 first minutes of catalysis (Fig. S3). About one equivalent
of ethane is released per tungsten center after initial contact of
WH/Al O3 with ethylene. The reactivity of WH/Al O toward
2
2
3−(500)
◦
Fig. 2. Global selectivity in propylene and butenes (left) and relative selectivity in butenes (right) vs. inverse space velocity at 46 h and 120 C.

1
28
E. Mazoyer et al. / Journal of Molecular Catalysis A: Chemical 385 (2014) 125–132
Scheme 2. Global mechanism proposed for the direct conversion of ethylene to propylene on WH/Al2O3−(500).
ratio reveals a deficiency of 2-butenes and isobutene compared to
the thermodynamic equilibrium values (1:8:3.84:29.6) [35].
In order to understand the propagation mechanism of this
reaction, a complete contact time study was performed. The obser-
decreases with temperature [27]. To corroborate this hypothesis
a contact time study at lower temperature, i.e. 120 C has been per-
◦
formed. The dimerization and metathesis reaction kinetics do not
have the same dependency on temperature. The global selectivity
−
1
◦
vation of the contact time (VHSV ) influence on the catalyst
performances, i.e. conversion and selectivity, reveals two major
pieces of information on the catalytic process [36,37]. First, if there
is a linear dependence between conversion and contact time that
can be extrapolated to zero at zero contact time, it means that the
catalytic process is not affected by mass transfer limitations (Fig.
S6). The observations can thus be interpreted regarding chemical
steps: it is chemical regime. Then, the extrapolated selectivities
at zero contact time allow a reliable determination of real pri-
mary products. Therefore the reduction of contact time decreases
the period when reagents and active sites can interact, reducing
de facto the formation of secondary products. It has earlier been
observed that the selectivity in propylene is constant with time on
stream, whereas the deactivation rate is high in the early stage of
the catalysis before reaching a pseudo-plateau. To determine if the
mechanism of the reaction is the same during these two periods,
contact time studies were performed for different time on stream.
Thus four experiments of 46 h were carried out at different contact
times (Table S1). Plotting the conversion vs. inverse space velocity
in propylene and butenes vs. inverse space velocity at 46 h at 120 C
◦
and 150 C (Figs. 2 and S7) show that higher selectivity in butenes
◦
is observed when lowering the reaction temperature (42% at 120 C
vs. 16% at 150 C). Moreover, butenes selectivity clearly increases
when contact time decreases (Fig. 2, left). Evidently, at 120 C these
◦
◦
butenes are less consumed by the cross-metathesis reaction than
◦
at 150 C which is less favored by the lower temperature.
These results show that 1-butene is the primary product of the
conversion of ethylene to propylene (Fig. 2 right). Its formation
occurs by ethylene dimerization by the insertion of two ethylene
molecules in the tungsten hydride bond followed by a ␤-H elimi-
nation according to the classical Cosse-Arlman mechanism [38,39].
Then cis- and trans-2-butenes are obtained after reinsertion of the
newly formed 1-butene followed by a second ␤-H elimination of
the sec-butyl species as proposed by Tolman for the isomeriza-
tion of 1-butene to 2-butene by nickel hydride [40]. Regarding
these results, the dimerization seems to be very slow and is the
rate-determining step, compared to other reaction occurring fur-
ther, as it is very difficult to observe 100% selectivity in 1-butene.
According to numerous examples [41], olefin cross-metathesis is
expected to be catalyzed by a tungsten-alkylidene species following
a classical Chauvin mechanism [42]. Furthermore, tungsten alkyl or
hydride species also catalyze the isomerization of 1-butene into 2-
butenes. Thus, [W](CH CH )(=CHCH ), which bears both an alkyl
(
Fig. S6) confirms that the chemical regime is maintained, what-
ever the time on stream, in the range of VHSV studied. At each
time on stream there is a strong linear dependence, which can be
extrapolated to a zero conversion at infinite VHSV, between con-
version and contact time. The selectivities observed during these
contact time studies (Fig. S7) indicate that butenes are primary
products, since the selectivity intercept with the y-axis extrapo-
lates to a non-zero value, and that propene is formed in a reaction
that consumes butenes, since their curves are mirror images at any
flow rate. Furthermore, according to relative selectivity of butene
at 46 h on stream vs. inverse space velocity (Fig. S8), the forma-
tion of 1-butene also appears to precede that of other butenes
since it is the only one for which the selectivity increases with
decreasing inverse space velocity. It is notable that whatever the
time on stream, the global selectivity and butenes partial selectivi-
2
3
3
and an alkylidene fragment, is able to catalyze the dimerization,
isomerization, and metathesis of olefins on the same site. In other
words, it behaves as a trifunctional catalyst (Scheme 2). These mul-
tiple functionalities are observed in other reactions carried out by
this catalyst such as alkane metathesis [32,43], and conversion of
1-butene or 2-butene to propylene [28,44].
Although the supported tungsten hydride precursor catalyzes
an important reaction for propylene production from ethylene feed,
it undergoes a significant decrease of activity with time. The deac-
tivation of this catalyst can have different origins: reduction of the
tungsten center and/or blocking of active sites by coke or polymers,
as reported for different supported olefin metathesis catalysts. Two
strategies will be used. The first one consists in the analysis of by-
products during ethylene conversion in a continuous flow reactor.
The second one is a study of the catalyst at a molecular level by ex
situ studies by solid state NMR spectroscopy, EPR, UV–vis, DRIFT,
TGA and DSC of the used catalyst.
−1
ties vs. VHSV have the same profiles. Indeed, the mechanism of
the conversion of ethylene to propylene stays the same regardless
of deactivation of the catalyst. Note that the selectivity in butenes
◦
is less than 20% at 150 C even at low contact time. In order to
clarify the fact that 1-butene is the primary product, a study at
lower temperature has been performed since the conversion of
ethylene/2-butene cross metathesis reaction on WH/Al O
2
3−(500)

E. Mazoyer et al. / Journal of Molecular Catalysis A: Chemical 385 (2014) 125–132
129
On-stream deactivation during conversion of ethylene to pro-
pylene over tungsten hydride catalyst was examined by online
analysis of the products. Besides the expected products (accord-
ing to the previously proposed mechanism), a small amount of
isobutene (3%) is also observed. As presented on Scheme S3, the
isobutene may have two origins after decomposition of metalla-
cyclobutane by ␤-H transfer [45]: (i) a “productive” path, where
ethylene inserts in the alkyl-hydrido-isobutenyl-tungsten species,
then retrieves the active species by ␣-hydride abstraction; (ii) a
“
destructive” path, eliminating the isobutyl and hydride ligands fol-
(
VI)
(IV)
lowed by reducing the W
to inactive W
as already described
for group 6 imido complexes [45,46]. The number of moles of
isobutene produced per number of moles of tungsten was calcu-
lated and then the cumulated TON isoC4= was plotted vs. time on
stream. Fig. S9 clearly shows that the production of isobutene is
catalytic. After 2 h on stream, the cumulated TONisoC4= exceed 1.
Indeed path (i) must occur during the reaction (Scheme S2). Nev-
ertheless, path (ii) may also occur simultaneously. Spectroscopic
studies (EPR and UV–Vis) may probe a change of the oxidation
state of the used catalyst and thereby provide information about
the evolution of the catalyst. It has been well accepted and proved
that olefin metathesis reaction occurs on the transitional metal cen-
ter and follows the metal carbene mechanism [47]. The oxidation
state of metal species plays an important role in the metathesis
activity [45,48]. Here UV–vis and EPR spectra are applied to detect
the oxidation state changes of aged tungsten catalyst.
Fig. 3. 13C CPMAS NMR spectrum of the catalyst after 120 h on stream.
could be observed in the aromatic and olefinic region indicating
the carbonaceous deposits are mainly paraffins [27].
In order to better understand the mechanism of deactivation
of the catalyst during the direct conversion of ethylene to pro-
pylene four experiments have been done in the same conditions
In fact, at the end of the catalysis no significant color changes
were observed. UV–Vis analysis of the fresh W-H/Al O3(500) catalyst
◦
−1
2
(PC2H4 = 1 bar, T = 150 C, flow rate 4 mL min or volume hourly
precursor shows an absorption band at 222 nm, which corresponds
to typically W(VI) species (Fig. S10) [49]. The used catalyst has a
similar UV–Vis profile with minor red-shift (10 nm), suggesting that
the degree of the oxidation state of tungsten remains essentially
unchanged after the 260 h on stream. Importantly, no bands corre-
sponding to W(IV) or W(V) species were observed [50]. Moreover,
fresh and used catalysts are EPR silent, which further indicate the
absent of reduced tungsten species.
−1
space velocity (VHSV) 400 h ) with different running time (2 h,
2
4 h, 48 h and 264 h). At the end of each experiment, the cata-
lyst was recovered in the glove box to be analyzed under inert
conditions. The study of the aged catalyst was also performed
using DRIFT spectroscopy. The DRIFT spectrum (Fig. 4) of the
−1
fresh catalyst WH/Al O
750–3000 cm , 1360–1465 cm
exhibits bands at 3500–3800 cm
,
2
3−(500)
−1
−1
2
corresponding to ꢀ(OH) from
alumina, ꢀ(CH) and ı(CH) from residual hydrocarbon, respectively.
1
3
C CP/MAS NMR spectroscopy is an efficient tool to study the
−1
Finally the ꢀ(WH) band is present at 1913 cm
[32]. The DRIFT
nature of the carbonaceous species on aged catalysts [51]. 13
C
spectrum of the used catalyst after 2 h on stream shows a com-
plete disappearance of the ꢀ(WH) mode, indicating that all the W-H
have been inserted by alkenes. Simultaneously, three new bands
CP/MAS NMR spectrum in Fig. 3 indicates that two broad peaks
centered around 32 ppm and 30 ppm are observed, correspond-
ing to respectively crystallized and amorphous PE [52]. No peaks
−1
−1
−1
centered at 2923 cm , 2852 cm and 1465 cm corresponding
Fig. 4. DRIFT spectra of the fresh catalyst and of the catalytic bed at different time on stream.

1
30
E. Mazoyer et al. / Journal of Molecular Catalysis A: Chemical 385 (2014) 125–132
Fig. 5. Left: TGA profiles of the catalyst for each time on stream; right: Differential thermo-gravimetric profiles of the catalyst for each time on stream.
◦
to typical ꢀ(CH) and ı(CH) absorptions for long chain alkyl were
growing with time on stream. This spectrum is similar to polyeth-
ylene obtained by polymerization of ethylene on supported metal
hydrides [53].
around 525 C. After 2 h, there is 17% of weight loss when the cat-
alyst is completely calcinated, showing a fast formation of surface
carbonaceous species on the surface of the catalyst, even at short
time on stream. The amount of carbonaceous species gradually
increases to reach 46% after 264 h on stream. It clearly indicates that
carbonaceous species are formed at the surface of the catalyst dur-
ing reaction. It is interesting to note on the DTG profiles that at high
In order to confirm the hypothesis of the formation of PE, the
catalyst was analyzed by TGA and DSC during the direct conver-
sion of ethylene to propylene and compared to the fresh catalyst.
The TG profiles, realized under oxidative condition, of the catalysts
◦
time on stream, 264 h, the first weight loss peak shifts to 433 C indi-
cating that the carbonaceous species burns 50 C higher. In other
◦
◦
after given time on stream at 150 C are presented on Fig. 5. The dif-
ferent weight-loss stages can be seen in the changes in slope of the
TG curve and in the maxima of the differential thermo-gravimetric
word, the longer the time on stream is, the more carbonaceous
deposits are observed, the harder they are to burn.
(
DTG) profiles. Fig. 5 (right) shows that two main different weight
Moreover, DSC profiles for melting and crystallization are pre-
sented (Fig. 6). The melting DSC profile shows that the melting
◦
◦
losses are observed: a first between 383 C and 433 C and a second
Fig. 6. Left: Fusion DSC profiles for each time on stream; right: crystallization DSC profiles for each time on stream.

E. Mazoyer et al. / Journal of Molecular Catalysis A: Chemical 385 (2014) 125–132
131
peak temperature of the formed species increases with the time
References
◦
◦
on stream: starting at 101 C after 2 h and reaching 136 C after
64 h. It is also clear that when the time on stream increases,
[
[
[
1] J.S. Plotkin, Catal. Today 106 (2005) 10.
2] A.H. Tullo, Chem. Eng. News 81 (2003) 15.
3] K. Fukudome, N.-o. Ikenaga, T. Miyake, T. Suzuki, Catal. Sci. Technol. 1 (2011)
987.
2
more than one surface species is formed. This is confirmed by
crystallization DSC profiles which show two crystallization peaks
after 48 h on stream corresponding to low density PE or heavy
PE waxes (i.e. species containing more than 60 carbon atoms)
[
[
4] B.V. Vora, Top. Catal. 55 (2012) 1297.
5] K.C. Szeto, B. Loges, N. Merle, N. Popoff, A. Quadrelli, H. Jia, E. Berrier, A. De
Mallmann, L. Deleyoye, R.M. Gauvin, M. Taoufik, Organometallics 32 (2013)
6452.
[
54].
Furthermore, a comparison between the weight loss due to car-
[
[
[
[
6] P. Amigues, Y. Chauvin, D. Commereuc, C.T. Hong, C.C. Lai, Y.H. Liu, J. Mol. Catal.
65 (1991) 39.
bonaceous deposits on the used catalyst and the conversion of
ethylene (Fig. S11) guide us to the conclusion that the formed car-
bonaceous products are involved in deactivation. Regarding the last
results, it seems that the deactivation of the catalyst is mainly due
to the formation of two types of polymers. The drop of initial activ-
ity, i.e. before 48 h on stream, is accompanied by the formation of
7] T.I. Bhuiyan, P. Arudra, M.N. Akhtar, A.M. Aitani, R.H. Abudawoud, M.A. Al-Yami,
S.S. Al-Khattaf, Appl. Catal., A 467 (2013) 224.
8] D.R. Hua, S.L. Chen, G.M. Yuan, Y.L. Wang, Q.F. Zhao, X.L. Wang, B. Fu, Micro-
porous Mesoporous Mater. 143 (2011) 320.
9] H.J. Liu, L. Zhang, X.J. Li, S.J. Huang, S.L. Liu, W.J. Xin, S.J. Xie, L.Y. Xu, J. Nat. Gas
Chem.V 18 (2009) 331.
[
10] J.C. Mol, J. Mol. Catal. A: Chem. 213 (2004) 39.
◦
[11] D.Z. Zhang, X.J. Li, S.L. Liu, S.J. Huang, X.X. Zhu, F.C. Chen, S.J. Xie, L.Y. Xu, Appl.
Catal., A 439 (2012) 171.
a low melting point polymer (116 C). This polyethylene species
may be formed by multiple insertions of ethylene in the supported
metal–alkyl bond [55]. The resulting species can then not perform
ethylene dimerization anymore, thereby resulting in a deficit of 1-
butene, the primary product in the direct conversion of ethylene to
propylene. With the low quantity of 1-butene, the overall efficiency
of the catalyst drops, therefore leading to the lower conversion
levels observed.
[12] S. Ilias, A. Bhan, ACS Catal. 3 (2013) 18.
[13] D.M. McCann, D. Lesthaeghe, P.W. Kletnieks, D.R. Guenther, M.J. Hayman, V. Van
Speybroeck, M. Waroquier, J.F. Haw, Angew. Chem. Int. Ed. 47 (2008) 5179.
[
14] J. He, T. Xu, Z. Wang, Q. Zhang, W. Deng, Y. Wang, Angew. Chem. Int. Ed. 51
2012) 2438.
[15] M.H. Nilsen, S. Svelle, S. Aravinthan, U. Olsbye, Appl. Catal., A 367 (2009) 23.
(
[
16] U. Olsbye, O.V. Saure, N.B. Muddada, S. Bordiga, C. Lamberti, M.H. Nilsen, K.P.
Lillerud, S. Svelle, Catal. Today 171 (2011) 211.
[
17] A.H. Tullo, J. Johnson, Chem. Eng. News 91 (2013) 9.
The formation of carbonaceous deposits at the surface of the
catalyst during metathesis reaction has been observed many
times [51,56]. These deposits are assumed to be responsible of
the deactivation of the catalyst, such as for WO /SiO , which is
reactivated under air at 500 C to calcine these deposits and to
regenerate the active species [57]. However, contrary to what
[18] P.P. Oneill, J.J. Rooney, J. Am. Chem. Soc. 94 (1972) 4383.
[
19] F. Hugues, B. Besson, J.M. Basset, J. Chem. Soc., Chem. Commun. (1980)
19.
20] H. Imamura, T. Konishi, Lanthanide Actinide Res. 3 (1991) 387.
7
[
3
2
[21] T. Yamaguchi, Y. Tanaka, K. Tanabe, J. Catal. 65 (1980) 442.
[22] T. Suzuki, K. Tanaka, I. Toyoshima, H. Gotoh, Appl. Catal. 50 (1989) 15.
◦
[
[
23] T. Suzuki, React. Kinet. Catal. Lett. 81 (2004) 327.
24] H. Oikawa, Y. Shibata, K. Inazu, Y. Iwase, K. Murai, S. Hyodo, G. Kobayashi, T.
Baba, Appl. Catal., A 312 (2006) 181.
we have shown here for WH/Al O
, the industrial cata-
2
3−(500)
[
[
[
25] M. Iwamoto, Catal. Surv. Asia 12 (2008) 28.
lysts, WO /SiO also suffer from deactivation by reduction of the
3
2
26] M. Iwamoto, Y. Kosugi, J. Phys. Chem. C 111 (2007) 13.
27] E. Mazoyer, K.C. Szeto, N. Merle, S. Norsic, O. Boyron, J.-M. Basset, M. Taoufik,
C.P. Nicholas, J. Catal. 301 (2013) 1.
metal center due to the high temperature required for olefin
metathesis.
[
[
[
[
[
28] E. Mazoyer, K.C. Szeto, J.-M. Basset, C.P. Nicholas, M. Taoufik, Chem. Commun.
4
8 (2012) 3611.
29] M. Taoufik, E. Le Roux, J. Thivolle-Cazat, J.-M. Basset, Angew. Chem. Int. Ed. 46
2007) 7202.
4
. Conclusion
The WH/Al O
(
30] N. Popoff, E. Mazoyer, J. Pelletier, R.M. Gauvin, M. Taoufik, Chem. Soc. Rev. 42
(2013) 9035.
31] K.C. Szeto, E. Mazoyer, N. Merle, S. Norsic, J.M. Basset, C.P. Nicholas, M. Taoufik,
ACS Catal. 3 (2013) 2162.
32] E. Le Roux, M. Taoufik, C. Coperet, A. de Mallmann, J. Thivolle-Cazat, J.M. Basset,
B.M. Maunders, G.J. Sunley, Angew. Chem. Int. Ed. 44 (2005) 6755.
33] D.N. Clark, R.R. Schrock, J. Am. Chem. Soc. 100 (1978) 6774.
34] K.C. Szeto, S. Norsic, L. Hardou, E. Le Roux, S. Chakka, J. Thivolle-Cazat, A. Bau-
douin, C. Papaioannou, J.M. Basset, M. Taoufik, Chem. Commun. 46 (2010)
3985.
catalyst exhibits an outstanding activity in
3−(500)
2
the direct conversion of ethylene to propylene. An extensive con-
tact time study of this reaction shows that 1-butene is the primary
product obtained by dimerization of ethylene, the rate determin-
ing step for this reaction. The ethylene to propylene mechanism
involves a threefold cycle catalyzed by the multifunctional tung-
sten carbene-hydride moiety supported on alumina: dimerization,
isomerization and cross-metathesis. Based on several physical and
spectroscopic methods (DRIFT, solid state NMR, EPR, UV–Vis, TGA
and DSC), we observed that the active species is deactivated by
polymerization with time on stream due to insertion of ethylene
into tungsten alkyl groups (Scheme 2) giving low density PE and
[
[
[
[
35] R.A. Alberty, C.A. Gehrig, J. Phys. Chem. Ref. Data 14 (1985) 803.
36] R.A. van Santen, J.A. Moulijn, P.W.N.M. van Leeuwen, B.A. Averill, Catalysis: An
Integrated Approach, Elsevier, Amsterdam, 1999.
[37] J.M. Basset, C. Coperet, L. Lefort, B.M. Maunders, O. Maury, E. Le Roux, G. Saggio,
S. Soignier, D. Soulivong, G.J. Sunley, M. Taoufik, J. Thivolle-Cazat, J. Am. Chem.
Soc. 127 (2005) 8604.
[
[
38] E.J. Arlman, P. Cossee, J. Catal. 3 (1964) 99.
39] P. Cossee, J. Catal. 3 (1964) 80.
◦
◦
heavy waxes with a melting point of 116 C and 136 C, respectively.
Formation of polyethylene also inhibits or prohibits, for at least a
fraction of the W sites, the dimerization step (according to the pro-
posed mechanism) and thereby disables turnover of the ethylene to
propylene.
[40] C.A. Tolman, J. Am. Chem. Soc. 94 (1972) 2994.
[
41] K.J. Ivin, J.C. Mol, Olefin Metathesis and Metathesis Polymerization, Academic
Press, San Diego, CA, 1997.
[
42] J.L. Herisson, Y. Chauvin, Makromol. Chem. 141 (1971) 161.
[43] K.C. Szeto, L. Hardou, N. Merle, J.-M. Basset, J. Thivolle-Cazat, C. Papaioannou,
M. Taoufik, Catal. Sci. Technol. 2 (2012) 1336.
[
[
[
44] E. Mazoyer, K.C. Szeto, S. Norsic, A. Garron, J.-M. Basset, C.P. Nicholas, M. Taoufik,
ACS Catal. 1 (2011) 1643.
45] J. Robbins, G.C. Bazan, J.S. Murdzek, M.B. Oregan, R.R. Schrock, Organometallics
Acknowledgment
10 (1991) 2902.
46] E. Mazoyer, N. Merle, A. de Mallmann, J.-M. Basset, E. Berrier, L. Delevoye,
J.-F. Paul, C.P. Nicholas, R.M. Gauvin, M. Taoufik, Chem. Commun. 46 (2010)
UOP is acknowledged for generous funding.
8944.
[
47] M. Jezequel, V. Dufaud, M.J. Ruiz-Garcia, F. Carrillo-Hermosilla, U. Neugebauer,
G.P. Niccolai, F. Lefebvre, F. Bayard, J. Corker, S. Fiddy, J. Evans, J.P. Broyer, J.
Malinge, J.M. Basset, J. Am. Chem. Soc. 123 (2001) 3520.
Appendix A. Supplementary data
[48] H. Ahn, T.J. Marks, J. Am. Chem. Soc. 124 (2002) 7103.
[49] E.I. Ross-Medgaarden, I.E. Wachs, J. Phys. Chem. C 111 (2007) 15089.
[50] G.A. Ozin, R.A. Prokopowicz, S. Ozkar, J. Am. Chem. Soc. 114 (1992)
8953.
Supplementary material related to this article can be found,
in the online version, at http://dx.doi.org/10.1016/j.molcata.
2
014.01.025.

1
32
E. Mazoyer et al. / Journal of Molecular Catalysis A: Chemical 385 (2014) 125–132
[
[
51] X. Li, W. Zhang, X. Li, S. Liu, H. Huang, X. Han, L. Xu, X. Bao, J. Phys. Chem. C. 113
2009) 8228.
52] G. Nowaczyk, S. Glowinkowski, S. Jurga, Solid State Nucl. Magn. Reson. 25 (2004)
94.
[55] J.M. Basset, R. Psaro, D. Roberto, R. Ugo, Modern Surface Organometallic Chem-
istry, Wiley-VCH, Weinheim, 2009.
[56] A. Spamer, T.I. Dube, D.J. Moodley, C. van Schalkwyk, J.M. Botha, Appl. Catal., A
255 (2003) 133.
(
1
[
[
53] G. Tosin, C.C. Santini, J.-M. Basset, Top. Catal. 52 (2009) 1203.
54] Y. Chen, H. Zou, M. Liang, P. Liu, J. Appl. Polym. Sci. 129 (2013) 945.
[57] C. van Schalkwyk, A. Spamer, D.J. Moodley, T. Dube, J. Reynhardt, J.M. Botha,
Appl. Catal., A 255 (2003) 121.