On the product formation in 1-Butene metathesis over supported tungsten catalysts

Article abstract of DOI:10.1007/s10562-010-0345-9

1-Butene metathesis was performed over 8 wt% tungsten catalysts supported on silica, silica-alumina and γ-Al₂O₃. Under the applied reaction conditions, 1-butene metathesis yields with a relatively large selectivity iso-butene in addition to the expected metathesis products. The isobutene selectivity is high for catalysts with a relative low activity and decreases with increasing metathesis activity. It is deduced that iso-butene formation involves surface tungsten complexes, whose formation inhibits the metathesis activity. Springer Science+Business Media, LLC 2010.

Full text of DOI:10.1007/s10562-010-0345-9

Catal Lett (2010) 137:123–131
DOI 10.1007/s10562-010-0345-9
On the Product Formation in 1-Butene Metathesis over Supported
Tungsten Catalysts
•
Liesel Harmse Charl van Schalkwyk
•
Eric van Steen
Received: 5 October 2009 / Accepted: 9 April 2010 / Published online: 22 April 2010
Ó Springer Science+Business Media, LLC 2010
Abstract 1-Butene metathesis was performed over
8 wt% tungsten catalysts supported on silica, silica-alu-
mina and c-Al2O3. Under the applied reaction conditions,
1-butene metathesis yields with a relatively large selec-
tivity iso-butene in addition to the expected metathesis
products. The isobutene selectivity is high for catalysts
with a relative low activity and decreases with increasing
metathesis activity. It is deduced that iso-butene formation
involves surface tungsten complexes, whose formation
inhibits the metathesis activity.
metathesis yields ethene and hexenes. The primary
metathesis reaction can be defined as the conversion of
1-butene yielding ethene and n-C6-olefins and the conver-
sion of 1-butene with 2-butene yielding propene and n-C5-
olefins (i.e. assuming a rapid equilibration of the double
bond isomerisation). The metathesis of butenes is equilib-
rium limited (see Fig. 1), and the equilibrium conversion of
the C4-olefins in the primary metathesis reaction increases
slightly with increasing temperature (from ca. 52 mol% at
300 K to 70% at 700 K). The maximum equilibrium yield
of propene via the primary metathesis is obtained in the
temperature range between 500 and 800 K (ca.
25.2 mol%). Higher propene yields can only be obtained
by co-feeding ethene [6] representing a loss in the valuable
ethene.
Keywords 1-Butene metathesis ꢀ Tungsten ꢀ
Silica-alumina
1 Introduction
The higher thermodynamic yield of valuable chemicals
in the metathesis of butenes demands the reaction to be
carried out at relatively high temperature, thus necessitat-
ing the use of a tungsten-based catalyst. The use of WO3/
SiO2 for the metathesis of propene and butene is well
documented [6, 7]. Huang et al. [3] obtained an improved
performance of a supported WO3-catalyst in the cross
metathesis of ethene with 2-butene by changing the support
from c-Al2O3 to c-Al2O3-HY. They attributed the enhanced
performance to the change in the Brønsted acidity of the
catalyst and a modification of the interaction between the
tungsten species and the support. Similar results have been
claimed for supported Re2O7 metathesis catalysts [8, 9].
Higher turn-over frequencies have been obtained in Re2O7-
catalysed metathesis using ordered mesoporous alumina
(OMA) as a support material rather than conventional
aluminas [10–13]. Supporting Re2O7 on mesoporous silica,
such as MCM-41, also results in more active catalysts than
supporting on commercial silica [14]. Bakala et al. [15]
argue that mesoporous materials as support materials for
Metathesis is a useful reaction to shift the composition of a
pool of olefins to meet market demands. For instance, the
Fischer-Tropsch synthesis produces a large amount of
relatively low value butenes [1], which can be converted
into higher value propene [2–4]. In particular, the cross-
metathesis of butenes [4] is of interest, since it negates the
loss of the valuable ethene in the production of propene.
The cross-metathesis of 1-butene with 2-butene yields
propene (with 2-pentene as a co-product) and the 1-butene
L. Harmse ꢀ E. van Steen (&)
Department of Chemical Engineering, Centre for Catalysis
Research, University of Cape Town, Private Bag X3,
Rondebosch 7701, South Africa
e-mail: eric.vansteen@uct.ac.za
L. Harmse ꢀ C. van Schalkwyk
Sasol Technology Research & Development, P.O. Box 1,
Klasie Havenga Road, Sasolburg 1947, South Africa
123

124
L. Harmse et al.
30
20
10
0
Changing the acidity of the support may result in a
change in the product composition obtained in the
metathesis, and yield a better insight in the reactions
occurring during olefin metathesis. In this study the product
distribution in 1-butene metathesis is studied over a series
of supported tungsten-based catalysts with different acidity
with a view to gain a better understanding of the variety of
products formed. Commercially available support materials
were used in this study prepared by similar routes to study
the effect of support acidity in the cross-metathesis of
1-butene, since acidic supports prepared by grafting alu-
mina on mesoporous silica does not seem to enhance the
catalytic activity significantly [15].
Propene
Ethene
300
400
500
600
700
800
Temperature, K
Fig. 1 Equilibrium yield of propene and ethene in the metathesis of
butanes considering double bond isomerisation, cross-metathesis of
1-butene with 2-butene and 1-butene metathesis (calculated with data
from [5])
2 Experimental
Re2O7-based metathesis catalysts do not have a decisive
advantage over commercial oxides, but that the challenge
lies in obtaining a good dispersion of the active site.
The formation of propene via 1-butene metathesis
requires double bond isomerisation activity of the catalyst,
since only the cross metathesis of 2-butene with 1-butene
yields the desired product (with C5-olefins as a co-product).
Brønsted acid sites are thought to be responsible for the
required double bond isomerisation [16], but do not cata-
lyse the metathesis reaction [17, 18].
2.1 Catalyst and Support Synthesis
Silica gel (Davisil grade 646, Aldrich), silica-alumina and
alumina (Sasol Germany GmbH) were used as support
material (see Table 1). The crushed support particles
(dparticle = 300–500 lm) were impregnated with an aque-
ous solution of ammonium metatungstate hydrate (Aldrich;
[(NH4)6(H2W12O40)]initial = 0.2778 mL dm-3; pH: ca. 5)
to yield an 8 wt% WO3-loading on the support [22]. The
slurry was stirred for 2 h at room temperature, after which
the excess water was evaporated under reduced pressure
(100 mbar) at ca. 373 K. The catalysts were dried in an
oven at 383 K for 2 h. The temperature was then increased
to 523 K at a rate of 1 K/min and kept at this temperature
for 2 h. In the final step, the temperature was raised to
873 K at a rate of 3 K/min and maintained at this tem-
perature for 8 h.
The high reaction temperature of olefin metathesis may
result in the formation of coke as was observed in the
1-heptene metathesis over WO3/SiO2 [19]. Coke formation
resulted in a decrease in the BET-surface area and the pore
volume with a minimal loss in activity up to a coke content
of the catalyst up to ca. 45 wt%. Formation of coke on
metal oxide catalysed reactions is thought to be catalysed
by the acid sites present within the catalyst [20]. Both
Lewis acid sites and Brønsted acid sites may catalyse the
formation of coke, the former by strongly interacting with
basic species in the feed, and the latter via the normal
carbocation route for coke formation [21].
2.2 Support and Catalyst Characterization
The specific surface area and pore volume were determined
for all the fresh and spent samples using N2-physiosorption
Table 1 Characteristics of support materials
Support material Supplier SiO2 (wt%) Al2O3 (wt%) SBET (m2/g) VPore (cm3/g) dPore (A) dAl O (A) nacid (mmol/g) xLewis kdb (h-1
)
b
c
d
˚
˚
2
3
c-Al2O3
Siralox-20
Siralox-40
Siralox-75
SiO2
Sasola
Sasola
Sasola
Sasola
0
20
40
75
100
80
60
25
0
242
351
409
339
300
0.80
0.70
0.84
0.31
1.15
133
80
42
39
–
0.16
0.22
0.27
0.25
0.01
100
97
94
72
–
0.005
0.118
0.213
0.143
–
82
37
–
Aldrich 100
149
–
a
Sasol Germany GmbH
b
c
d
Number of acid sites as determined using NH3–TPD
Fraction of Lewis acid sites as determined using pyridine adsorption (FTIR)
First order rate constant for the double bond isomerisation of 1-butene at 373 K
123

On the Product Formation in 1-Butene Metathesis over Supported Tungsten Catalysts
125
on a TRISTAR-Micromeritics apparatus. Approximately
0.25 g of the sample was degassed under N2 flow at 473 K
overnight prior to the measurement at 77 K.
GHSV = 1000 h-1). The gaseous products were collected
in a sample bomb.
The metathesis of 1-butene was performed in a gas-
phase fixed bed reactor [14]. The catalyst (20 mL) was
activated in situ at 773 K for 16 h under a constant nitro-
gen flow of 20 mL (STP)/min. After activation, they were
cooled down to the desired reaction temperature of 723 K.
The gaseous feed (99.38 wt% 1-butene; major impurities:
n-butane: 0.23 wt%, cis ? trans 2-butene: 0.37 wt%) was
led over an alumina bed to remove possible water and
preheated to the reaction temperature over a bed of car-
borundum. The metathesis reaction was performed at
atmospheric pressure (0.85 atm) and a GHSV = 425 mL
(STP)/h/mL. The liquid product was collected in a catch
pot and the gaseous product in the sample bomb after the
reactor. The flow rate of the off-gas was measured using a
wet gas flow meter.
The crystallinity of the samples was measured on an
X’Pert Pro Multi Purpose Diffractometer (MPD). The
diffractometer is equipped with a fast, solid state X’Cel-
erator detector. The X-ray generator was operated at 40 kV
and 40 mA, using a 1.8 kW long fine-focus cobalt tube. A
programmable divergent slit of 1° was used, together with
an anti-scatter slit of 2°, both in fixed mode.
The acidity of the samples was probed by ammonia
desorption and pyridine adsorption. The NH3–TPD mea-
surements were performed on a Micromeritics Autochem
2910. Prior to the NH3 adsorption measurements, approx-
imately 0.150 g of catalyst was activated at 823 K for
60 min under continuous helium flow (8.5 mL (STP)/min).
The catalyst was cooled to 373 K and then saturated with
ammonia in a 5% NH3/He gas mixture at the same flow
rate for a period of 60 min. Physisorbed and gaseous NH3
were removed by purging with helium for 30 min at 373 K.
Thereafter, the sample was heated to 823 K with a heating
rate of 10 K/min under a helium flow. Pyridine adsorption
was used to differentiate between Lewis and Brønsted acid
sites. The samples were ground to a fine powder and
pressed into self-supporting wafers using a pressure of
3 tons. The wafer was dried at 773 K in vacuum for 12 h.
Pyridine was adsorbed at 373 K and allowed to equilibrate
for 12 min. Excess pyridine was evacuated for 1 h prior to
cooling the sample to room temperature. The FTIR-spec-
trum of adsorbed pyridine was recorded on a Bruker Vector
22 spectrometer with air as a background. The relative
amount of Lewis acid sites and Brønsted acid sites was
obtained from the peaks at ca. 1450 and 1540 cm-1 rep-
resenting the Lewis and the Brønsted acid sites respec-
tively, using the reported extinction coefficients [23].
The amount of coke formed on the catalyst was deter-
mined via thermo-gravimetric analysis (TGA) of the spent
catalysts using a SDT 2960 Simultaneous DSC-TGA. The
sample was dried at 373 K under nitrogen flow (120 mL
(STP)/min). The gas environment was changed to air
(120 mL (STP)/min). The sample was initially kept iso-
thermally in this environment for 30 min. Subsequently,
the temperature was raised (10 K/min) up to initially
623 K, at which temperature is was kept for 30 min, before
heating the sample further to 823 K, at which temperature
it was kept for a further 30 min.
The gaseous products were analyzed on a GC-FID
equipped with a PLOT fused silica CP-Al2O3/KCl column
(50 m 9 530 lm 9 10 lm) utilizing a temperature pro-
gram (isothermal at 343 K for 3 min; ramped at a rate of 6 K/
min to 463 K, and kept at this temperature for 10 min). The
liquid products were analyzed on a GC-FID equipped with a
CP Sil PONA column (100 m 9 250 lm 9 0.50 lm) uti-
lizing a temperature program (isothermal at 323 K for 2 min;
ramped at a rate of 12 K/min to 533 K, and kept at this
temperature for 10 min).
3 Results and Discussion
Table 1 gives an overview of the support materials used in
this study. c-Al2O3 consists of small c-Al2O3 crystallites
with an average crystallite diameter (as estimated from
XRD-line width broadening) of ca. 4 nm. Siralox-20, with
a silica content of 20 wt%, consists of c-Al2O3 crystallites
of the same crystal size in addition to X-ray amorphous
material. Support material with a silica content larger than
20 wt% does not show any evidence of crystalline c-Al2O3.
The replacement of aluminium with silicon in the Siralox-
series results in an increase in acidity. The fraction of
Lewis acid sites decreases with increasing aluminium
replacement by silicon. The increased number of acid sites,
and most likely the increase in Brønsted acid sites, results
in a strong increase in the activity of the support for the
double bond isomerisation. Silica as a support does not
show any noticeable evidence of acidity.
2.3 Reaction Studies
The introduction of tungsten via impregnation results in
a reduction in the pore volume and to a lesser extent in the
BET-surface area (cf. Tables 1, 2). The loss in pore volume
must be ascribed to the adsorption of tungsten species on
the support surface. The average crystallite size of c-Al2O3,
in samples with a silica content of less than 20 wt%, was
The acidity of the support materials was further charac-
terized using 1-butene isomerisation. The reaction was
performed at 373 K and atmospheric pressure (0.85 bar) in
a fixed bed reactor (dreactor = 10 mm; Vcatalyst = 2.0 ml;
123

126
L. Harmse et al.
virtually unchanged. Only WO3/SiO2 shows the presence
of WO3 as a crystalline phase with an average diameter
larger than 50 nm. The large crystallite size obtained in
WO3/SiO2 could have been expected [22], since the
impregnation solution (at pH & 5) contains negatively
charged polytungstate anions [24]. Thus, an electrostatic
repulsion is expected between the anions in the impreg-
nation solution and the silica surface (iso-electric point of
silica is expected to be between 1 and 2 [25]). The depo-
sition of relatively large polytungstate anions on the silica
surface during the drying process may lead to the formation
of large WO3-crystals on the silica surface. The alumina
used in this study has a much higher iso-electric point of
ca. 8 [26]. Thus, an electro-statically induced adsorption
process may occur upon impregnating c-Al2O3 with
ammonium metatungstate. Only a limited amount of
tungsten can be adsorbed on an alumina support by
adsorption [26]. A deposition in addition to the adsorption
process [27] is probably taking place in the impregnation of
alumina with ammonium metatungstate. The presence of
pre-adsorbed tungstate species on the alumina surface may
act as nucleation sites for the further deposition tungsten
species leading to the formation of a finely dispersed
tungsten phase. The point of zero-charge of silica-alumina
is expected to shift to lower pH-values with increasing
silica content. However, it should be realized that the
surface of these materials is enriched with aluminium [28]
reducing the shift in the pH value. Jara et al. [28] showed
that a silica-alumina containing 63 wt% SiO2 has a IEP of
4.5. Thus, it can be expected that the electro-static
adsorption of polytungstate ions on silica-alumina is
reduced in comparison to alumina. It might be postulated
that also in this case, the presence of some pre-adsorbed
tungstate on the surface will lead to a disperse tungsten
oxide phase in the catalyst.
and the TPD-profile of the support (see Fig. 2). The
impregnation of ammonium metatungstate on c-Al2O3
results in a decrease in the number of acid sites desorbing
ammonia between 500 and 570 K and an increase in the
number of acid sites desorbing ammonia between 580 and
780 K. Hence, the introduction of tungsten on alumina
leads to the generation of more strongly adsorbing sites.
Similar conclusions were derived for a DFT-study on the
acidity of Mo–O–Al clusters compared to the acidity of
Al–O–Al-clusters [29].
Tungsten supported on silica-alumina also results in an
increase in number of acid sites and in particular to sites
desorbing at temperatures larger than 430 K. Silica does
virtually not adsorb ammonia, whereas WO3/SiO2 shows
acidity [4] through ammonia desorption with a peak
WO3 /SiO2
WO3 /Siralox-75
WO
3 /Siralox-40
WO3 /Siralox-20
WO3/Al2O3
The tungsten supported catalysts contain more acid sites
than the support materials (see Table 2). Furthermore, a
shift towards more Brønsted acid sites is observed. The
change in the acidity due to the introduction of tungsten on
the support is best viewed by looking at the difference
of the NH3–TPD-profile of the supported WO3-catalysts
300
500
700
900
Temperature, K
Fig. 2 Difference of the NH3–TPD-profile of tungsten supported on
alumina, Siralox and silica and their support material
Table 2 Characteristics of supported tungsten catalysts
SBET (m2/g)
VPore (cm3/g)
xLewis
a
nacid (mmol/g)
b
˚
dPore (A)
˚
dAl O (A)
2 3
˚
dWO (A)
3
Catalyst
8 wt% WO3/c-Al2O3
8 wt% WO3/Siralox 20
8 wt% WO3/Siralox 40
8 wt% WO3/Siralox 75
8 wt% WO3/SiO2
a
237
311
328
246
260
0.70
0.63
0.75
0.25
1.02
119
81
47
42
–
–
0.23
0.29
0.33
0.37
0.05
100
90
78
69
99
–
92
–
41
–
–
156
–
[500
Number of acid sites as determined using NH3–TPD
b
Fraction of Lewis acid sites as determined using pyridine adsorption (FTIR)
123

On the Product Formation in 1-Butene Metathesis over Supported Tungsten Catalysts
127
maximum at 490 K. It can thus be concluded that the WO3/
SiO2 has not only less acid sites, but also weaker acid sites,
than WO3/c-Al2O3. The additional sites introduced by
tungsten impregnated on silica-alumina (Siralox) seem to
have an acid strength between WO3/SiO2 and WO3/c-
Al2O3 with peak maximums between 550 and 625 K. This
might be attributed to the direct contact of WO3 with either
alumina or silica type of sites.
Brønsted acid sites (1545 cm-1) are absent for c-Al2O3
and SiO2 and only present in samples containing silica as
well as alumina. The modification of Al2O3 with silica led
to the creation of both highly acidic Lewis and Brønsted
acid sites, as previously observed for silica-alumina [30].
Lewis acid sites are created via the isomorphous substitu-
tion of Si4? by Al3? ions at tetrahedral lattice sites and
Brønsted acid sites via the formation of bridged hydroxyl
groups.
FTIR of adsorbed pyridine was used to study the
Brønsted and Lewis acidity of the range of supports and the
corresponding supported tungsten catalysts (see Fig. 3).
The characteristic bands of pyridine protonated by Brøns-
ted acid sites (pyridinium ions) appearing at *1543 and
1640 cm-1 and bands from pyridine coordinated to Lewis
acid sites appearing at ca. 1453 and 1621 cm-1. The peak
appearing at 1621 cm-1 is associated with Lewis acid sites
in the tetrahedral coordination, while the small peak beside
it at 1612 cm-1 is due to Lewis acid sites in the octahedral
environment [30]. The peak appearing at 1453 cm-1 rep-
resents the total amount of Lewis acid sites in the tetra-
hedral and octahedral environment. The peak associated
with the band appearing at 1543 cm-1, is accompanied by
the vibration of the pyridinium ion resulting from the
protonation of pyridine by Brønsted acid sites.
The FTIR-spectra of the tungsten-impregnated supports
show bands at 1451 cm-1 which corresponds to pyridine
coordinated to Lewis acid sites. This peak represents the
total amount of Lewis acid sites. The peaks appearing at
1548 cm-1 indicate the presence of pyridinium ions orig-
inating from pyridine protonated by Brønsted acid sites.
These bands were only observed with catalysts containing
20, 40, 75 wt% silica. For the pure alumina supported
tungsten, no such peak was observed. The peak at
1491 cm-1, is accompanied by the vibration of the pyrid-
inium ion. Pure silica does not contain any Lewis or
Brønsted acid sites. However, an adsorption band at
1449 cm-1 is observed for WO3/SiO2. This represents the
Lewis acidity of WO3/SiO2 and is thus due to the Lewis
acidity of tungsten impregnated on to the support. Hence,
Fig. 3 FTIR-spectra of the
pyridine adsorbed on the
support materials (left) and on
tungsten supported on the
support materials (right)
γ-Al2O3
WO3 /γ-Al2O3
WO3 /Siralox-20
WO3 /Siralox-40
Siralox-20
Siralox-40
WO3 /Siralox-75
WO3 /SiO2
Siralox-75
SiO2
123

128
L. Harmse et al.
the peaks appearing at 1449 cm-1 for all support materials
must be ascribed due to pyridine coordinated to all Lewis
acid sites, i.e. Lewis sites present on the support and the
Lewis acid sites induced by tungsten present on the sup-
port. The strength of the Lewis acid sites, typically
observed by a shift in peak, has not been noticeably
affected by the introduction of tungsten on to the support.
However, tungsten on the support resulted in an increase in
the total acidity of the catalyst.
support material, i.e. WO3/c-Al2O3 [ WO3/Siralox-
20 [ WO3/Siralox-40 [ WO3/Siralox-75 [ WO3/SiO2.
The decline in catalyst activity cannot be correlated
directly with either the number of acid sites or with the
strength of acid sites or the type of acid sites in the catalyst.
The selectivity of the primary metathesis product, eth-
ene, is typically lower than equilibrium selectivity for the
primary metathesis reaction (see Fig. 5), with the exception
of the selectivity obtained with WO3/SiO2. The ratio of
ethene to C6-olefins decreases with increasing time-on-line,
and is in the product obtained over WO3/SiO2 higher than
expected for the primary metathesis reaction (ca.
1.6 0.1 mol ethene/mol C6-olefins) indicating secondary
metathesis reactions resulting in ethene formation. The
products obtained over WO3/Siralox-40 show a ratio of
ethene to C6-olefins significantly lower than the expected
ratio of the primary metathesis reaction (0.66
0.03 mol ethene/mol C6-olefins). The selectivity for the
other primary metathesis product propene (assuming a
rapid equilibration between n-C4-olefins) is typically
higher than the equilibrium selectivity as predicted for the
primary metathesis reaction. This indicates that other
The primary metathesis reaction is equilibrium limited,
and a maximum conversion of n-C4-olefins of ca. 71% is
achievable at a reaction temperature of 723 K. WO3/SiO2
shows an almost constant activity as a function of time-on-
line close to the equilibrium conversion of the primary
metathesis reaction (see Fig. 4). It should however be
noted that the double bond isomerisation over WO3/SiO2 is
not at equilibrium (in contrast to the double bond isom-
erisation over the other catalysts), since the 2-butene con-
tent in the fraction of linear butenes is only ca. 44 mol% (at
equilibrium a 2-butene content in the fraction of linear
butenes should be ca. 73 mol%). This implies that the
primary metathesis reaction over WO3/SiO2 is not at
equilibrium.
A higher conversion of the n-C4-olefins than expected if
the primary metathesis reaction is equilibrium limited was
obtained over tungsten supported on Siralox-75 and on
Siralox-40 implying that the primary metathesis reaction is
not the sole reaction taking place. WO3/Siralox-20 and
WO3/c-Al2O3 showed a much lower activity. It can be
further noted that all catalysts except WO3/SiO2 show a
decline in activity with time-on-line. The decline in the
catalyst activity, as defined by the apparent first order rate
constant for the conversion of n-C4-olefins after 6 h on line
relative to the apparent first order rate constant after 1 h on
line becomes less with increasing silica content in the
20
WO3/SiO2
15
primary metathesis selectivity
10
WO3/Siralox-75
WO3/Siralox-40
WO3/Siralox-20
5
WO3/γ-Al2O3
0
0
2
4
6
Time on stream, hrs
50
40
30
20
10
0
100
80
WO3/Siralox-40
WO3/SiO2
WO3/Siralox-75
primary metathesis selectivity
60
WO3/Siralox-20
40
n
20
WO3/γ-Al2O3
0
2
4
6
0
Time on stream, hrs
0
2
4
6
Time on stream, hrs
Fig. 5 Molar selectivity for primary metathesis products (top:
ethene; bottom: propene) in the metathesis of n-butenes over
supported tungsten (open circle: WO3/Al2O3; filled triangle: WO3/
Siralox-20; filled square: WO3/Siralox-40; filled circle: WO3/Siralox-
75; open square: WO3/SiO2; dotted line equilibrium selectivity for the
primary metathesis reaction)
Fig. 4 Conversion of n-butenes in the metathesis over supported
tungsten catalysts (open circle: WO3/Al2O3; filled triangle: WO3/
Siralox-20; filled square: WO3/Siralox-40; filled circle: WO3/Siralox-
75; open square: WO3/SiO2)
123

On the Product Formation in 1-Butene Metathesis over Supported Tungsten Catalysts
129
reactions result in the formation of propene as well. Fur-
2.5
%Coke formation
thermore, the molar ratio of propene to C5-olefins is much
larger than 1 ranging from 1.5 to 1.6 for the silica-rich
2
catalysts going down to 1.4 0.1 for the molar ratio of
propene to C5-olefins obtained over WO3/Siralox-20 and
1.5
1.12 0.06 for the ratio obtained over WO3/c-Al2O3. The
higher than expected selectivity for the primary metathesis
1
product propene (and to some extent of ethene) compared
to the selectivity of the corresponding long chain olefins
0.5
might originate from the secondary metathesis reactions
involving butenes yielding even longer chain product
0
compounds. However, it should be noted that the selec-
0
20
40
75
100
tivity for the longer chain products (lumped together as
% Silica
C7?—see Table 3) is rather low (3–13 C%) and can only
partially explain the high propene selectivity.
Fig. 6 Coke formation reported as weight loss in the oxidative
treatment of the spent catalyst in the metathesis of n-butenes over
supported tungsten
Coke formation is typically observed in the metathesis
over supported tungsten catalysts [19, 20]. Coke formation
was observed over all catalyst (with exception of WO3/
SiO2) in the limited run time of these experiments (see
Fig. 6). Coke formation was enhanced over the catalyst
WO3/Siralox-40, and could be related to the acidity of the
support for the silica-alumina supports (Siralox-series) It
should however be noted that the amount of coke formed
could not be correlated directly to the amount of catalyst
deactivation. This implies that at least some of the coke
formed is located on the support and does not interfere with
the metathesis reaction. Coke is thought to be hydrogen
poor, and thus the product effluent should contain more
hydrogen than the feed (see Table 3). This can be in the
form of paraffinic compounds. Some paraffins are found in
each carbon number fraction. The paraffin selectivity passes
a maximum as a function of the silica content in the
support material at a silica content of 20 wt% implying that
the acidity of the support does play a role in the
dehydrogenation of the coke precursor. Coke formation can
be accompanied with methane formation through demeth-
ylation. However, the methane selectivity is typically less
than 0.2 mol%. This is thus not a major reaction pathway
under metathesis reaction conditions. The total selectivity
for C2–C5-paraffins amounted to between 0.1 and 3.5 mol%
implying that the pairing reaction [31] plays some role in
the formation of hydrogen-poor surface species.
The selectivity of branched product compounds, and
especially iso-butene, is rather high (see Table 3), although
much lower than in thermodynamic equilibrium under
these conditions. The formation of branched isomers may
occur through an acid catalysed, skeletal isomerisation, and
it can be best evaluated by considering the amount of
branched, olefinic isomers in the fraction of olefins of a
particular carbon number (see Fig. 7). Their formation was
Table 3 Performance of
WO3/c-Al2O3 WO3/Siralox-20 WO3/Siralox-40 WO3/Siralox-75 WO3/SiO2
supported 8 wt% WO3-catalysts
in the n-butene metathesis
(T = 723 K, p = 0.85 bar;
GHSV = 425 mL (STP)/h/mL)
after 6 h on line
XnꢁC ꢁolefins (mol%) 17.8
(H2/C)product (mol/mol) 2.01
44.2
2.02
58.9
2.01
69.9
2.00
67.7
2.00
4
SCH (C%)
0.1
9.6
0.2
3.8
0.0
2.8
0.0
12.2
6.9
0.0
5.7
4
SC (C%)
7þ
Sethene (mol%)
3.3
4.9
6.8
15.1
0.0
Sethane (mol%)
0.0
0.3
0.1
0.1
Spropene (mol%)
Spropane (mol%)
Sn-butane (mol%)
Siso-butene (mol%)
S1-pentene (mol%)
S2-pentenes (mol%)
Sn-pentane (mol%)
30.2
0.0
42.7
0.8
43.9
0.2
45.3
0.4
43.7
0.0
0.7
1.8
0.0
0.6
0.0
26.8
2.3
10.5
2.3
5.3
6.7
0.3
2.2
2.4
6.0
11.1
0.8
10.7
0.6
15.1
0.1
11.9
0.5
20.9
0.1
SisoꢁC (mol%)
SC (mol%)
15.1
4.2
17.1
5.6
8.9
16.6
7.0
0.4
5
10.6
10.3
6
123

130
L. Harmse et al.
40
30
20
10
0
30
20
10
0
WO3/ -Al2O3
γ
WO3/Siralox-20
WO3/Siralox-75
WO3/Siralox-40
i
i
WO3/SiO2
0
2
4
6
0
20
40
60
80
100
Time on stream, hrs
Xn -C4-olefins, mole-%
25
20
15
10
5
Fig. 8 Molar selectivity for iso-butene as a function of the obtained
conversion level (open circle: WO3/Al2O3; filled triangle: WO3/
Siralox-20; filled square: WO3/Siralox-40; filled circle: WO3/Siralox-
75; open square: WO3/SiO2)
WO3/γ-Al2O3
WO3/Siralox-75
WO3/Siralox-20
lower over catalyst with a high activity and high over cat-
alysts with a low activity (see Fig. 8). The metathesis of 1-
butene is presumably taking place over low coordination
tungsten sites, since they are expected to be more stable than
the high coordination sites [29]. The coordination of an
adsorbed olefin to a low coordination tungsten atom may
result in the formation of a surface alkoxide species leading
to double bond isomerisation. The further interaction of the
low coordination tungsten atom with an olefin may result in
the formation of an iso-C8-alkoxy-species. Intramolecular
skeletal isomerisation of the adsorbed iso-C8-alkoxide-
species and subsequent splitting into C4-fragments may lead
to the formation of iso-butene. Similar reaction schemes are
well known in acid catalysed reactions [32]. The likelihood
for this type of reactions is expected to increase with
increasing strength of the acidity of the hydroxyl-group(s)
on the surface tungsten complex, i.e. with increasing W–O–
Al interaction, but also with increasing size of the surface
tungsten complex [29]. Furthermore, the association of the
surface tungsten site with branched C8-alkoxide species will
result in steric hindrance for the coordination of olefins to
the low coordination tungsten site for further reaction.
Strong acid sites are expected to adsorb olefins more
strongly thus resulting in reduced activity on the tungsten
site due to geometric constraint at the site. Hence, 1-butene
metathesis over WO3/c-Al2O3 results in a relatively large
selectivity to iso-butene and a low activity.
WO3/Siralox-40
WO3/SiO2
i
0
0
2
4
6
Time on stream, hrs
Fig. 7 Molar selectivity for the formation of branched product
compounds (iso-butene: top; iso-pentene: bottom) in the metathesis of
n-butenes over supported tungsten (open circle: WO3/Al2O3; filled
triangle: WO3/Siralox-20; filled square: WO3/Siralox-40; filled
circle: WO3/Siralox-75; open square: WO3/SiO2)
previously related to Brønsted acidity within the catalyst
[22], since they only appear at higher tungsten loadings,
and can be suppressed by poisoning the acid sites with
potassium. All catalysts, with the exception of WO3/SiO2,
show a large extent of skeletal isomerisation, although the
skeletal isomerisation is not at equilibrium. The formation
of iso-pentene seems to be loosely correlated with the
acidity of the catalyst, with catalyst based on the alumina
containing support material showing a higher acidity and a
higher selectivity towards iso-pentene.
This seems however not to be the case, for the selectivity
of iso-butene, although the iso-butene content in the fraction
of C4-olefins follows a similar trend. The highest selectivity
was obtained with the least active catalyst, viz. WO3/c-
Al2O3 (ca. 27–30 mol%), but the relative large amount of n-
C4-olefins present in the effluent results in a low ratio of iso-
butene in the fraction of C4-olefins (0.05 mol/mol). The
obtained ratio of iso-butene in the fraction of C4-olefins over
WO3 supported on silica-alumina is higher (ca. 0.08–0.10
mol/mol), although the selectivity for iso-butene is much
reduced (5–15 mol%). Metathesis of n-butenes over WO3/
SiO2 does not lead noticeably to the formation of iso-butene
(Siso-butene \ 0.3 mol%). The selectivity of iso-butene is
The C5-fraction does not only contain linear C5-products
and their skeletal isomerisation products, but also cyclo-
pentenes. Cyclic compounds are thought to be related to
coke formation. The selectivity for these products is par-
ticularly high on catalysts, with strong acid sites present
due to the incorporation of tungsten on the support mate-
rial. Coke formation correlates with yield of these com-
pounds, i.e. a high activity and a high selectivity are
required for the formation of coke. For instance WO3/c-
123

On the Product Formation in 1-Butene Metathesis over Supported Tungsten Catalysts
131
3. Huang S, Liu S, Xin W, Bai J, Xie S, Wang Q, Xu L (2005) J Mol
Catal A 226:61–68
4. Wang Y, Chen Q, Yang W, Xie Z, Xu W, Huang D (2003) Appl
Catal A 250:25–37
5. Stull DR, Westrum EF, Sinke GC (1969) The chemical thermo-
dynamics of organic compounds. Wiley, New York
6. Zhao Q, Chen S-L, Gao J, Xu C (2009) Transit Met Chem 34:621
7. Ivin KJ, Mol JC (1997) Olefin metathesis and metathesis poly-
merization, vol 2. Academic Press, San Diego, p 25
8. Andreini A, Xu X, Mol JC (1986) Appl Catal 27:31–40
9. Sibeijn M, Mol JC (1991) Appl Catal 67:279–295
Al2O3 shows a high selectivity for these compounds, but
the overall activity is reduced (presumably due to steric
hindrance at the activity sites resulting from the formation
of bulky alkoxy-compounds).
4 Conclusions
The metathesis of n-butenes over supported tungsten cat-
alysts over supported tungsten catalysts yields in addition
to the expected primary metathesis products with relative
selectivity iso-butene. The reaction sequence taking place
on the catalyst surface is more complex that the reaction
scheme following the primary metathesis reaction. It was
shown that the formation of short chain olefins is in
abundance in relation to the formation of the corresponding
long chain product compounds. This might (partially)
result from secondary metathesis reactions taking place
yielding in the formation of long chain olefinic products.
The metathesis of 1-butene yielded a relative large
amount of iso-butene. The selectivity of iso-butene seems
to be inversely related to the activity of the catalysts. This
was ascribed to the formation of bulky iso-alkoxide species
restricting access to the active tungsten species and thus
resulting in a diminished catalyst activity. It is argued that
the formation of the branched iso-alkoxide species requires
strong acid sites as present in tungsten species associated
with alumina.
ˇ
ˇ
´
10. Balcar H, Hamtil R, Zilkova N, Cejka J (2004) Catal Lett 97:25
11. Oikawa T, Ookoshi T, Tanaka T, Yamamoto T, Onanka M (2004)
Microporous Mesoporous Mater 74:93
12. Aguado J, Escola JM, Castro MC, Paredes B (2005) Appl Catal A
284:47
ˇ
13. Hamtil R, Zilkova N, Balcar H, Cejka J (2006) Appl Catal A
ˇ
´
302:193
14. Topka P, Balcar H, Rathousky J, Zilkova N, Verpoort F, Cejka J
ˇ
ˇ
´
(2006) Microporous Mesoporous Mater 96:44
´
15. Bakala PC, Briot E, Millot Y, Piqumal J-Y, Bregeault JM (2008)
J Catal 258:61
16. van Roosmalen AJ, Mol JC (1980) J Chem Soc Chem Commun
704
17. Adreine A, Mol JC (1985) J Chem Soc Faraday Trans I 81:1705
18. van Roosmalen AJ, Kosterm D, Mol JC (1980) J Phys Chem
84:3075–3079
19. Spamer A, Dube TI, Moodley DJ, van Schalkwyk C, Botha JM
(2003) Appl Catal A 255:133–142
20. van Schalkyk C, Spamer A, Moodley DJ, Dube T, Reynhardt J,
Botha JM (2003) Appl Catal A 255:121–131
21. Bartholomew CH (2001) Appl Catal A 212:17–60
22. Spamer A, Dube TI, Moodley DJ, van Schalkwyk C, Botha JM
(2003) Appl Catal A 255:153–167
23. Emeis CA (1993) J Catal 141:347–354
24. Panagiotou GD, Petsi T, Bourikas K, Kordulis C, Lycourghiotis
A (2009) J Catal 262:266–279
25. Parks GA, de Bruyn PL (1962) J Phys Chem 66:967–973
Coke formation requires a relative high activity and a
high selectivity for coke precursors (typically formed by
acid catalysed reactions). A pairing reaction may result in
the dehydrogenation of the coke precursor accompanied by
the formation of paraffinic compounds.
¨
26. Knozinger H, Taglauer E (1993) Catalysis 10:1
27. van Steen E, Viljoen EL, van de Loosdrecht J, Claeys M (2008)
Appl Catal A 335:56–63
28. Jara AA, Goldberg S, Mora ML (2005) J Colloid Interface Sci
292:160–170
29. van Steen E, Viljoen EL, Claeys M (2007) J Mol Catal A
266:254–259
30. Rajagopal S, Marzari JA, Miranda R (1995) J Catal 151:192
References
¨
¨
¨
31. Roger HP, Bohringer W, Moller KP, O’Connor CT (2000) Stud
Surf Sci Catal 130:281
32. Ramani NC, Sullivan DL, Ekerdt JG (1998) J Catal 173:105
1. Claeys M, van Steen E (2004) Stud Surf Sci Catal 152:601–680
2. Mol JC (2004) J Mol Catal A 213:39–45
123