Enhanced metathesis of ethylene and 2-butene on tungsten incorporated ordered mesoporous silicates

Article abstract of DOI:10.1016/j.apcata.2016.10.004

Tungsten-incorporated 3D mesoporous silicates, W-KIT-6, W-KIT-5, and W-SBA-16 catalysts, outperform supported tungsten oxide catalysts (WO₃/SiO₂ and WO₃/KIT-6) for the metathesis of ethylene and 2-butene to propene at 450 °C. All catalysts exhibit steady activity and stability during 7 h runs in a fixed-bed reactor. Furthermore, on all mesoporous catalysts, a maximum propylene yield was obtained at an optimum W loading at which the catalyst acidity is also maximum. Slightly delayed addition of the W source during the one-pot synthesis protocol markedly improves the propene yield on W-KIT-6 catalyst. XPS results conclusively show that this enhancement is due to preferential enrichment of active W species on the catalyst surface. Extended runs lasting five days reveal very little loss of activity even which is easily recovered by calcination in air.

Full text of DOI:10.1016/j.apcata.2016.10.004

Applied Catalysis A: General 528 (2016) 142–149
Contents lists available at ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
Enhanced metathesis of ethylene and 2-butene on tungsten
incorporated ordered mesoporous silicates
Jian-Feng Wua, Anand Ramanathana, William K. Snavelya, Hongda Zhua,b
,
Andrzej Rokickia, Bala Subramaniama,b,∗
a Center for Environmentally Beneficial Catalysis, The University of Kansas, Lawrence, KS 66047, USA
b Department of Chemical and Petroleum Engineering, The University of Kansas, Lawrence, KS 66045, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 5 June 2016
Received in revised form 27 July 2016
Accepted 5 October 2016
Available online 6 October 2016
Tungsten-incorporated 3D mesoporous silicates, W-KIT-6, W-KIT-5, and W-SBA-16 catalysts, outperform
supported tungsten oxide catalysts (WO3/SiO2 and WO3/KIT-6) for the metathesis of ethylene and 2-
butene to propene at 450 ◦C. All catalysts exhibit steady activity and stability during 7 h runs in a fixed-
bed reactor. Furthermore, on all mesoporous catalysts, a maximum propylene yield was obtained at an
optimum W loading at which the catalyst acidity is also maximum. Slightly delayed addition of the W
source during the one-pot synthesis protocol markedly improves the propene yield on W-KIT-6 catalyst.
XPS results conclusively show that this enhancement is due to preferential enrichment of active W species
on the catalyst surface. Extended runs lasting five days reveal very little loss of activity even which is
easily recovered by calcination in air.
Keywords:
Metathesis
Ethylene
2-Butene
© 2016 Elsevier B.V. All rights reserved.
Tungsten
Ordered mesoporous silicates
1. Introduction
structures have been explored. These include WO3/MTS-9 [26]
and MoO3/HMS [13] catalysts synthesized by the impregnation
In the United States, the supply of natural gas liquids (NGLs) has
increased dramatically as a collateral product of shale gas extrac-
tion. While this is expected to provide an oversupply of ethylene,
a shortage of propene is expected as a result of decreased naphtha
availability [1,2]. Hence, the synthesis of propene from ethylene
via olefin metathesis [3–9] is receiving renewed interest. Current
industrial process for the metathesis of ethylene and 1-butene
to propene [the so-called olefin conversion technology or OCT]
employs WO3/SiO2 catalyst, producing a propene yield of ∼54% at
approximately 30–35 bar and >260 ◦C [5]. To develop catalysts that
operate at lower pressures and temperatures, a number of exper-
imental [10–30] and theoretical [31,32] investigations have been
reported. A majority of the investigated catalysts involve Re, Mo,
and W oxides.
method, and W-SBA-15 [11], W-MCM-41 [11], W-KIT-6 [27,35],
W-FDU-12 [28], Mo-SBA-15 [36], Mo-KIT-5 [25] catalysts synthe-
sized by a one-pot method, SiO2-Al2O3-MoO3 [37,38], Re-Si-Al
mixed oxides [29,39], WSi and WSiAl catalysts [40] synthesized by
non-hydrolytic sol-gel method. Although mesoporous silica sup-
ports with different pore structures have been reported for olefin
metathesis, the influence of structural differences on catalytic per-
formance is unclear. In this work, we compared the performance
of three types of W-incorporated ordered mesoporous silicates
with different 3D structures for the metathesis of 2-butene and
ethylene to propene. These include W-KIT-6 (Ia3d) [41], W-KIT-
5 (Fm3m) [42], and W-SBA-16 (Im3m). While W-KIT-6 catalyst had
been investigated previously for olefin metathesis involving 1-
butene and ethylene [12,30], there are no published reports for the
(2-butene + ethylene) metathesis on of W-KIT-6, W-KIT-5 and W-
SBA-16 catalysts.
The activity of supported metathesis catalysts depends on the
catalyst synthesis procedure. Catalysts prepared using the sim-
ple impregnation method typically suffer from low dispersion of
the active W precursors and the generation of inactive bulk WO3
species [10,26,43]. A higher dispersion of W species in the silica
framework is achievable by one-pot methods [11,41,44]. However,
tungsten accessibility could be a problem if the tungsten species are
buried sufficiently deep in the framework. To overcome this prob-
Mesoporous silicas possessing high surface area, large pore
volume, high thermal stability and low diffusion resistance
offer an attractive alternative to amorphous non-ordered sil-
ica supports [33,34]. For olefin metathesis, several W-, Mo-,
and Re-incorporated mesoporous silicas with different pore
∗
Corresponding author at: Department of Chemical and Petroleum Engineering,
University of Kansas, Lawrence, KS 66045 USA.
E-mail address: bsubramaniam@ku.edu (B. Subramaniam).
http://dx.doi.org/10.1016/j.apcata.2016.10.004
0926-860X/© 2016 Elsevier B.V. All rights reserved.

J.-F. Wu et al. / Applied Catalysis A: General 528 (2016) 142–149
143
lem, we investigate herein a modified one-pot synthesis method in
which the addition of the tungsten source is delayed until after par-
tial hydrolysisof the silica source, increasingthe probabilityof more
active tungsten precursors being deposited on the surface. Appro-
priate analytical techniques (XRD, UV/Vis and XPS) are employed
to gain fresh insights into the relative distributions of W species
on the catalysts prepared by various synthesis procedures and to
correlate the relative metathesis activity of such catalysts.
continued for 20 h. The crystallization, the product collection and
calcination procedures are similar to those for W-KIT-6.
For preparing W-SBA-16 catalysts, the synthesis procedure is as
follows. 2.85 g of F127 were dissolved in 135 mL of H2O and 5.6 g
of 37.1 wt% HCl at 45 2 ◦C. Then, 8.65 g of 1-butanol were added
to the above mixture and the stirring was continued for 1 h. Fol-
lowing this step, 13.8 g of TEOS and the desired amount of sodium
tungstate were added and the stirring was continued for 20 h. The
crystallization, the product collection and calcination procedures
are similar to those for W-KIT-6.
2. Experimental
2.2.4. Modified W-KIT-6 synthesis
2.1. Chemicals
The following modified W-KIT-6 synthesis method was adopted
to increase the population of accessible W species on the cata-
lyst surface. The modification involved delaying the addition of the
tungsten source. Following the dissolution of P123 triblock copoly-
mer in conc. HCl solution (∼0.5 mol/L) and the addition of TEOS, the
desired amount of sodium tungstate (dissolved in 5 mL H2O) was
added following a delay of 1, 2, 3, 5 and 8 h. After the addition of
the tungsten source, the mixture was stirred for 24 h. The resulting
mixture was transferred into a Teflon-lined autoclave and heated
at 100 ◦C for 24 h. The solid product was collected and washed with
deionized water, then dried overnight at 100 ◦C and calcined at
550 ◦C for 5 h. The catalyst samples are denoted as W-KIT-6 (t, x)
where t and x represent the delayed addition time of the tungsten
source (in h) and the tungsten weight percent, respectively.
Triblock copolymer Pluronic P123 (EO20-PO70-EO20
, aver-
age MW = 5800), Pluronic F127, tetraethyl orthosilicate (TEOS,
98%), fumed silica (particle size: 0.007 ␮m, surface area:
390 40 m2/g, batch #: 119k0031), ammonium metatungstate
hydrate [(NH4)6H2W12O40·xH2O, ≥85% WO3 basis] and tungsten
oxide were purchased form Sigma-Aldrich. Sodium tungstate (99.0-
101.0%, Na2WO4·2H2O) was purchased from Alfa Aesar. 1-Butanol
(HPLC grade) and hydrochloric acid (certified ACS plus) were pur-
chased from Fisher Scientific. 2-Butene (≥95 wt%, 39% cis-2-butene
and 61% trans-2-butene) and ethylene (≥99.5%) were purchased
from Matheson Tri-Gas Inc. All the reagents were used as received
without further purification.
2.2. Catalyst preparation
2.3. Catalyst characterization
2.2.1. KIT-6 support
The wide-angle XRD patterns were collected on a Bruker X-ray
diffractometer using Mo K˛ radiation (␭ = 0.71073 Å) operated at
50 kV and 35 mA. The small-angle XRD patterns were collected
on a Rigaku system with an S-MAX 3000 instrument operated
at 45 kV and 0.66 mA. The tungsten and silicon contents of the
catalysts were determined by inductively coupled plasma optical
emission spectrometry (ICP-OES) using a Horiba Jobin Yvon JY 2000
instrument. The textural properties (BET surface area, pore vol-
ume and pore size distribution) of the catalysts were measured
on a Quantachrome NOVA 2000e instrument as described else-
where [41,42,47]. The total acidity of the catalysts was measured by
temperature-programmed desorption of ammonia (NH3-TPD) on a
Micromeritics Autochem 2910 instrument equipped with a ther-
mal conductivity detector (TCD) as described elsewhere [41]. The
structure of the surface species was analyzed by diffuse reflectance
UV/Vis (DR UV–vis) spectra with a PerkinElmer Lambda 850 spec-
trometer equipped with diffuse reflectance integrating sphere,
using Spectralon as the reference.
X-ray photoelectron spectroscopy (XPS) was performed using
PHI 5000 Versa Probe II instrument. Monochromated Al K˛ (50 W,
15 kV) X-ray source was used and the X-ray beam diameter was
200 ␮m. A dual-beam charge neutralizer was used to compen-
sate for the charging effect. The spherical capacitance analyzer was
operated with a pass energy of 46.95 eV. All XPS peaks were ref-
erenced to the Si 2p peak at 103.3 eV. To get the information from
the outermost layer, the sample stage was tilted to 20◦. Prior to the
analysis, the surface of the sample was sputtered by Ar+ ion (2.0
keV) for 5 min [48].
KIT-6 support (i.e., without addition of W source solution) was
synthesized according to the procedure described elsewhere [45].
2.2.2. WO3/SiO2 and WO3/KIT-6 catalysts
WO3/SiO2 and WO3/KIT-6 catalysts were prepared by wet
impregnation of the silica supports (fumed SiO2 and KIT-6, respec-
tively) with an aqueous solution of ammonium metatungstate,
followed by sonication for 0.5 h and stirring for 24 h at room tem-
perature [27,46]. The impregnated samples were first dried in a
rotary evaporator at 70 ◦C for 0.5 h, then dried at 100 ◦C overnight
in an oven. The resulting samples were treated in flowing air from
room temperature to 550 ◦C at a ramp rate of 1 ◦C/min, followed
by calcination at 550 ◦C for 5 h and natural cooling to room tem-
perature. The resulting solids are denoted as WO3/SiO2 (x) and
WO3/KIT-6 (x) where x represents the weight percent of tungsten
in these catalysts.
2.2.3. Synthesis of W-KIT-6, W-KIT-5 and W-SBA-16 catalysts
W-KIT-6 [41], W-KIT-5 [42] and W-SBA-16 [47] catalysts were
synthesized as described in the cited publications and the resulting
catalyst samples are denoted as W-KIT-6 (x), W-KIT-5 (x) and W-
SBA-16 (x) where x represents the weight percent of tungsten in
the samples.
For preparing W-KIT-6 catalysts, a typical synthesis procedure is
as follows. 4.5 g of P123 were dissolved in 160 mL of H2O and 8.3 g of
37.1 wt% HCl at 38 2 ◦C. Then, 4.5 g of 1-butanol were added and
the system was stirred for 1 h. Following this step, 9.7 g of TEOS
and the desired amount of sodium tungstate were added and the
stirring continued for 24 h. The resulting mixture was transferred
into a Teflon-lined autoclave and heated to 100 ◦C for 24 h. The solid
product was collected and washed by deionized water, then dried
at 100 ◦C overnight. The resulting sample was calcined in flowing
air at 550 ◦C for 10 h at a ramp rate of 1 ◦C/min.
2.4. Evaluation of metathesis activity
The catalytic activity of various W-incorporated catalysts for
metathesis of ethylene and 2-butene to propene was evaluated in a
continuous flow, stainless steel fixed-bed reactor with i.d. = 9.4 mm.
1.0 g of catalyst (0.425–0.850 mm diameter, precalcined in a muf-
fle furnace at 550 ◦C for 5 h) was loaded in the central portion of
the reactor, held in place by a stainless steel collar and glass wool.
For preparing W-KIT-5 catalysts, the synthesis procedure is as
follows. 2.7 g of F127 were dissolved in 130 mL H2O and 5.3 g
37.1 wt% HCl at 45 2 ◦C. Then, 13.0 g of TEOS and the desired
amount of sodium tungstate were added and the stirring was

144
J.-F. Wu et al. / Applied Catalysis A: General 528 (2016) 142–149
A thermocouple probe was placed in the catalyst bed to measure
the catalyst temperature. After pretreatment at 550 ◦C for 1 h under
N2 (50 std cm3/min) to remove moisture, the reactor was cooled
down to the reaction temperature. After the N2 flow was stopped,
ethylene and 2-butene were introduced into the reactor via mass
flow controllers at predetermined flow rates with the ethylene/2-
butene molar ratio of 3. The metathesis reactions were conducted
at 450 ◦C or 500 ◦C and 1 atm, with a weight hourly space velocity
(WHSV) of 2.0 h−1 based on the total ethylene + 2-butene flow rate.
The reaction products were analyzed by an on-line Varian CP-3800
gas chromatograph equipped with an Agilent GS-gaspro® column
(30 m × 0.320 mm) and a flame ionization detector (FID). The line
between the reactor and the six-way valve was heated to 160 ◦C to
prevent product condensation. A typical gas chromatograph based
on sampling and analysis of the reactor effluent stream is shown in
the supporting information (Fig. S1).
The following equations are used in assessing the performance
of the W-based catalysts and mass balance closure:
Fig. 1. Wide-angle XRD patterns of (a) W-KIT-6 (1 h, 9.3), (b) W-KIT-6 (3 h, 9.2),
(c) W-KIT-6 (5 h, 9.3), and (d) W-KIT-6 (8 h, 9.7) catalysts. The unfilled triangles
indicate the WO3 peaks, while the broad hump between 7- 14◦ (indicated by the
black diamond) suggests the presence of amorphous SiO2.
2 − butene − 2 − butene
[
]
[
]
in
out
X2−butene
=
× 100%
2 − butene
[
]
in
Within the same family of materials, the pore volume and surface
area decrease along with an increase in W loading, also as reported
previously. The wide-angle XRD patterns of catalysts [WO3/SiO2,
WO3/KIT-6, W-KIT-6, W-KIT-5, W-SBA-16, W-KIT-6 (2 h), W-KIT-6
(t) where t = 1, 3, 5, 8 h, and crystalline WO3] showed the character-
istic patterns expected of these materials. These patterns are shown
in Figs. 1 and Figs. S3–S8 in the supporting information. Com-
pared with the WO3-impregnated catalysts [WO3/SiO2 (7.4) and
WO3/KIT-6 (7.6)], the W-incorporated mesoporous silicates [W-
KIT-6, W-KIT-5, W-SBA-16, and W-KIT-6 (2 h)] synthesized using
the one-pot method with or without delayed addition of the W
source show either no or low-intensity crystalline WO3 peaks at
similar W loading. This suggests superior tungsten dispersion in
the one-pot silicates. Comparison of wide-angle XRD patterns for
W-KIT-6 (8.7) and W-KIT-6 (2 h, 9.2) catalysts is provided in Fig.
S7 in the Supporting Information. Clearly, the peak intensity of
crystalline WO3 on W-KIT-6 (2 h, 9.2) catalyst is lower than the cor-
responding peak in the W-KIT-6 (8.7) catalyst, indicating that less
bulk WO3 was formed on the W-KIT-6 (2 h, 9.2) catalyst. For W-
KIT-6 (t) catalysts (t = 1, 3, 5, 8 h) prepared with delayed addition of
the W source, the peak intensity corresponding to crystalline WO3
increases with the delay time from 1 to 5 h, then remains constant
from 5 to 8 h (Fig. 1).
propene
[
]
out
Spropene
=
× 100%
out
2 − butene − 2 − butene
[
]
[
]
in
Ypropene = X2−butene × Spropene × 100%
Total carbon in products
[
]
Mass balance closure =
× 100%
]
Total carbon in feed components
[
where X2-butene denotes 2-butene (the limiting reactant) conver-
sion, Spropene denotes propene selectivity, Ypropene denotes propene
yield, and [A]in or [A]out denotes the molar flow rate of A in the
reactants or products (A = 2-butene or propene). The mass balance
closure at steady state was 100 2.5% which is within experimen-
tal error. The conversions were shown to be free from internal and
external transport limitations by varying the catalyst particle size,
catalyst amount and flow rate while maintaining identical WHSV
in all the runs and by confirming that the variations in conversion
were within experimental error. Details are given in the supporting
Information (Fig. S2).
Given that the actual number of active W species is difficult to
characterize under working conditions, an apparent turnover fre-
quency (TOF) is defined as the moles of propene formed per mole
of W per second [19].
ꢀ
ꢁ
ꢀ
ꢁ
WHSV2−butene g · h−1 /M2−butene g · mol−1 × Yieldpropene
ꢀ
ꢁ
mcatalyst g × W wt%
( )
TOF s−1
=
/
ꢀ
ꢁ
ꢀ
ꢁ
3600 s · h−1
MW g · mol−1
Where WHSV2-butene is the weight hourly space velocity of 2-
butene, M2-butene is the molar mass of 2-butene, mcatalyst is the
weight of catalyst, W wt% is the tungsten loading in wt%, and MW is
the W molar mass. The apparent TOF assumes that all available W
species exhibit the same activity. Thus, variations in apparent TOF
for nearly identical W loading should provide comparative infor-
mation on the number of accessible active W species. Following a
run, the reactor was cooled, and the used catalyst was recovered
and transferred into a glass vial for further characterization.
Small-angle XRD patterns reveal that the long-range structural
ordering of KIT-6 was maintained following wet impregnation of
WO3 (see Fig. S9). The W-KIT-6, W-KIT-5 and W-SBA-16 mate-
rials also show clear long-range structural ordering (Fig. S10), as
demonstrated previously [41,42,47] for these materials. At higher
tungsten loadings (>7.2 wt%), the intensities of the (211) diffraction
bands of W-KIT-6 catalysts become weaker, indicating a decrease in
the structural order (Fig. S11). For a similar W loading, the delayed
addition of the W source also influences the structural order of W-
KIT-6 (t) catalysts (t = 1, 3, 5, 8 h) with the longer delays (5 and 8 h)
yielding a better structural order (Fig. S12).
3. Results and discussion
3.1. Textural and structural properties
3.2. Characterization of surface species on W-KIT-6, WO3/KIT-6,
and WO3/SiO2 catalysts by UV/Vis diffuse reflectance spectroscopy
The composition and textural properties (surface area, pore vol-
ume, and pore size) of W-KIT-6, W-KIT-5 and W-SBA-16 materials
yield values in the range reported previously [41,42,47] (Table S1).
The UV/Vis spectra of fresh W-catalysts (W-KIT-6, WO3/KIT-6,
and WO3/SiO2) were recorded in 200–800 nm region. As shown

J.-F. Wu et al. / Applied Catalysis A: General 528 (2016) 142–149
145
Fig. 2. UV/Vis DRS patterns of fresh W-catalysts: (a) W-KIT-6 (13.7), (b) W-KIT-6
(8.7), (c) W-KIT-6 (5.9), (d) W-KIT-6 (4.1), (e) W-KIT-6 (2.8), (f) WO3/KIT-6 (7.6), and
(g) WO3/SiO2 (7.4) catalysts.
Fig. 3. Propene yield as a function of time on stream for two WO3-impregnated sup-
ports, WO3/SiO2 (7.4) and WO3/KIT-6 (7.6) at T = 450 ◦C, P = 1 atm, WHSV (ethylene
and 2-butene) = 2.0 h−1, n(ethylene)/n(2-butene) = 3/1.
alyst (28.4 0.3% propene yield). Separate 2-butene conversion
and propene selectivity profiles (from which the yield profile was
derived) are shown in Fig. S18.
in Fig. 2, two absorption bands at 225 and 262 nm were observed
for all W-catalysts. An additional absorption band at 400 nm was
recorded for samples with W loading higher than 5.9 wt% in all
W-catalysts. These bands were not observed in the case of pure
KIT-6 (Fig. S13a), indicating that the absorption bands result from
the incorporation of tungsten species. The band at 225 nm is due
to the ligand-to-metal charge transfer, which stems from the iso-
lated [WO4]2− species with tetrahedral configuration [11,18,49].
This assignment was confirmed by the UV/Vis DRS spectroscopy of
Na2WO4 (Fig. S13b). The band at 262 nm is attributed to O2− → W6+
charge transfer, which corresponds to octahedral polytungstate
species [11,18,49]. The absorption band at approximately 400 nm
is due to bulk WO3 species [11,18,49], which was also confirmed
by comparison with the pure WO3 spectra (Fig. S13c). Clearly, the
one-pot W-KIT-6 (5.9) catalyst with lower W loading shows higher
band intensities at 225 and 262 nm compared to the impregnated
WO3/KIT-6 (7.6) and WO3/SiO2 (7.4) catalysts. This indicates that
the one-pot W-KIT-6 (5.9) catalyst has higher concentrations of
tetrahedral and octahedral W species than impregnated WO3/KIT-
6 (7.6) and WO3/SiO2 (7.4) catalysts. Comparative UV/Vis spectra
for W-KIT-6 (8.7) and W-KIT-6 (2 h, 9.2) catalysts is provided in Fig.
S17 in the supporting Information. The W-KIT-6 (2 h, 9.2) catalyst
displays higher band intensity at 262 nm (octahedral polytungstate
species) but lower band intensity at 400 nm (bulk WO3 species)
compared to the W-KIT-6 (8.7) catalyst. This again suggests a higher
population of active W species and a lower population of inactive
bulk WO3 species on the W-KIT-6 (2 h, 9.2) catalyst surface.
On catalysts synthesized by the impregnation method
(WO3/SiO2 and WO3/KIT-6), dispersed surface tungsten species
and/or bulk WO3 species are formed on the surface (Figs. 2
and S3). It is known that bulk WO3 species are not active for
the metathesis reaction of ethylene and 2-butene to propene
[10,26,43]. Experiments performed with commercially purchased
bulk WO3 sample reveal no propene formation supporting this
conclusion. The extent of surface silanol (Si–OH) groups dictates
how many active tungsten precursors can be grafted on the surface.
As shown in Scheme 1a, tungsten loading in excess of the available
silanol groups will result in the formation of inactive bulk WO3
species on the support. In order to further increase the number
of active tungsten precursors, ordered mesoporous silicas (OMS),
in which the tungsten was incorporated into the framework to
generate active precursors, were used as supports to synthesize
the metathesis catalysts (Scheme 1b).
3.4. Metathesis activity of tungsten-incorporated W-KIT-6
The propene yield profile with conventionally prepared W-
KIT-6 catalysts with different W loadings is shown in Fig. 4.
All the W-KIT-6 catalysts show steady yield after the break-in
period (about 1.5 h). The propene yield increases from ∼35% at
2.8 wt% W to ∼46% at 5.9 wt% W, and then decreases to ∼37% at
13.7 wt% W (Fig. 4a and b). The corresponding 2-butene conversion
and propene selectivity profiles for W-KIT-6 catalysts are shown
in Fig. S19. As observed from the UV/Vis DRS patterns (Fig. 2),
while the peak intensity corresponding to bulk WO3 is not evi-
dent at lower W loadings (up to 4.1 wt% W), it becomes increasingly
evident at higher W loadings. Correlating the trend of WO3 peak
intensity with the propene yield, it follows that bulk WO3 species
exhibit little or no metathesis activity. A control experiment (not
shown here) with bulk WO3 displays no metathesis activity, sup-
porting this conclusion. It also follows from the intensity variations
of the bands at 225 and 262 nm at all W loadings (Fig. 2) that these
framework tungsten species in tetrahedral and octahedral config-
urations are the active species. In addition, we observed that the
propene yield also follows the variation of total acidity with W load-
ing (Fig. 4b). The total acidity of the various W-KIT-6 samples are
provided in Fig. S24 in the Supporting Information.
3.3. Benchmarking experiments with WO3-impregnated silicas as
catalysts
Industrial ethylene + 1-butene metathesis catalysts are typi-
cally based on silica-supported WO3 with proprietary additives
to enhance catalyst stability [5]. Benchmarking experiments with
WO3-impregnated silica supports, WO3/SiO2 (7.4) and WO3/KIT-
6 (7.6), were first performed. As can be seen in Fig. 3, there is an
approximately 1.5 h break-in period as reported in previous studies
[43,50,51]. Thomas et al. [43] observed such a period on WO3/SiO2
catalyst, which is attributed to the gradual formation of the active
W-carbene species on the catalyst. Luckner and Wills [52] specu-
lated that the break-in period results from the strong adsorption
of an organic intermediate or fragment from propene dispropor-
tionation (the reverse reaction of the metathesis of ethylene and
2-butene to propene). The propene yield is stable following the
break-in period for the 7 h duration of the experiment. A mod-
erately higher activity (33.2 0.3% propene yield) was observed
on WO3/KIT-6 (7.6) catalyst compared to WO3/SiO2 (7.4) cat-
3.5. Metathesis activities of W-SBA-16 and W-KIT-5 catalysts
Figs. 5a and b shows propene yield and total catalyst acidity
as a function of W loading for W-KIT-5 and W-SBA-16 catalysts,

146
J.-F. Wu et al. / Applied Catalysis A: General 528 (2016) 142–149
Scheme 1. Increasing the amount and accessibility of active W precursors by various synthesis methods. (a) Impregnation method; (b) One pot synthesis of W incorporated
OMS supports; (c) Framework incorporation with enhanced accessibility to active W precursors.
respectively. As observed with W-KIT-6 catalysts, the propene yield
and total catalyst acidity display a maximum at an intermediate W
loading (Fig. 5). Further, the maximum propene yields were sim-
ilar (∼45%) on all W-incorporated mesoporous materials. These
results suggest that the active surface W precursors in the vari-
ous W-incorporated mesoporous silicates are similar. The 2-butene
conversion, propene selectivity, UV/Vis spectra, and total catalyst
acidity results for W-SBA-16 and W-KIT-5 materials are provided
in the supporting information (Figs. S14–S15, S20–S21, S25–S26).
3.6. Metathesis activity of W-KIT-6 prepared by delayed addition
of W source
Considering the relatively thick walls of mesoporous silicates
(∼2–6 nm), it is likely that some of the framework W species are
buried in the walls and hence inaccessible. To minimize such inac-
cessible W species, we attempted a modified synthesis approach in
which the Si and W sources were not added simultaneously. Rather,
the Si source (TEOS) is allowed to hydrolyze partially to first form a
Si-layer near the micelle. After a few hours, the W source (sodium
tungstate) is added which then forms a (W + Si) layer on top of the
Si layer (Scheme 1c).
Compared with the performance of the conventional W-KIT-6
catalysts (Fig. 4b), we find that the propene yield obtained with
the modified W-KIT-6 (2 h) catalysts (Fig. 6) are higher at tung-
sten loadings greater than 7.2 wt%. The best catalytic performance
(59.2 0.5% propene yield) is achieved at 9.2 wt% tungsten loading.
The total acidity of the W-KIT-6 (2 h, 9.2) catalyst is 0.30 mmol/g,
which is higher than the best performing conventional W-KIT-6
(5.9) catalyst with a total acidity of 0.23 mmol/g (Fig. 4). The corre-
sponding 2-butene conversion, propene selectivity, UV/Vis spectra,
Fig. 4. (a) Propene yield as a function of time on stream of W-KIT-6 catalysts with
and total catalyst acidity results are provided in the Supporting
Information (Figs. S16, S22, S27). A comparison of XRD and UV/Vis
data of the conventional W-KIT-6 (5.9) catalyst and the modified
W-KIT-6 (2 h) catalyst with similar W loadings can be found in the
Supporting Information (Figs. S7 and S17).
different loadings (2.8, 4.1, 5.9, 8.7, and 13.7 wt%) at T = 450 ◦C, P = 1 atm, WHSV
(ethylene and 2-butene) = 2.0 h−1, n(ethylene)/n(2-butene) in feed = 3/1. (b) Propene
yield and catalyst total acidity as a function of tungsten loading amount on W-KIT-6
catalysts.

J.-F. Wu et al. / Applied Catalysis A: General 528 (2016) 142–149
147
Table 1
Comparison of X-ray photoelectron spectroscopy (XPS) and ICP analyses of used W-KIT-6 (8.7) and used W-KIT-6 (2 h, 9.2) catalysts.
XPS results
W (%)a
ICP results
Catalyst
Si (%)b
W/Si
atomic ratio
W/Si
W/Si
W/Si
increment (%)c
atomic ratiod
increment (%)c
W-KIT-6 (8.7)
W-KIT-6 (2 h, 9.2)
2.74
3.86
97.3
96.1
0.0282
0.0402
0.03204
0.03421
42.6
6.8
a
Atomic concentration, estimated from the peak area of W 4f.
Atomic concentration, estimated from the peak area of Si 2p.
Normalized with W/Si ratio in catalyst prepared with no delay in W addition [W-KIT-6 (8.7)].
Based on analyses of W and Si in the synthesized catalyst samples.
b
c
d
Table 2
Apparent TOF of W-KIT-6 (x h) catalysts (x = 1, 2, 3, 5, 8) with similar W load-
ing (9.2–9.9 wt%) at T = 450 ◦C, P = 1 atm, WHSV (ethylene and 2-butene) = 2.0 h−1
n(ethylene)/n(2-butene) = 3/1.
,
Catalysts
Time of W addition
(h)
Apparent TOF
(mmolpropene molW−1 s−1
)
W-KIT-6 (1 h, 9.3)
W-KIT-6 (2 h, 9.2)
W-KIT-6 (3 h, 9.9)
W-KIT-6 (5 h, 9.3)
W-KIT-6 (8 h, 9.7)
1
2
3
5
8
3.97 0.01
4.69 0.04
3.96 0.03
3.23 0.02
2.92 0.03
XPS results provide direct evidence that the catalyst surface is
enriched with W species when the addition of the W source is
delayed by 2 h. As seen from Table 1, the surface W/Si atomic ratio
is 42.6% greater in the case of the W-KIT-6 (2 h, 9.2) catalyst com-
pared to the W-KIT-6 (8.7) catalyst. In sharp contrast, the bulk W/Si
atomic ratio, measured by ICP analysis, is only 6.8% greater in W-
KIT-6 (2 h, 9.2) catalyst compared to the W-KIT-6 (8.7) catalyst. The
relevant XPS data are provided in the supporting information (Fig.
S28).
We also compared the performances of modified W-KIT-6 cat-
alysts synthesized by delayed addition of the W source at 1, 2, 3,
5 and 8 h (Table 2 and Fig. S23). The W loading in these catalysts
was similar (9.2–9.9 wt%). The apparent TOF increased as the tung-
sten source addition was delayed from 1 to 2 h, but decreased when
delayed beyond 2 h. This suggests increased availability of active W
precursors when the addition of the W source is delayed by up to
2 h. In contrast, without W addition for 3.5 h, the hydrolysis of the
Si source (TEOS) results in a cotton-like precipitate. Hence, when
the tungsten source addition is delayed beyond 3 h, the tungsten
species are formed mainly on the catalyst surface similar to the
WO3 impregnation method, resulting in a reduced propene yield.
The apparent TOFs of other catalysts are shown in the supporting
information (Table S2).
Fig. 5. Propene yield and total catalyst acidity as a function of tungsten loading
on (a) W-KIT-5 catalysts and (b) W-SBA-16 catalysts at T = 450 ◦C, P = 1 atm, WHSV
(ethylene and 2-butene) = 2.0 h−1, n(ethylene)/n(2-butene) = 3/1.
3.7. Long-term stability of W-KIT-6 (2 h, 9.2) catalyst at 500 ◦C
The stability of W-KIT-6 (2 h, 9.2) catalyst during a 120 h run at
500 ◦C is shown in Fig. 7. Clearly, the propene yield increased signif-
icantly at 500 ◦C compared to the results at 450 ◦C (Figs. 6 and S22).
During a 120 h continuous run, the 2-butene conversion declined
from 84.2 0.1% to 79.4 0.1%), and the propene yield decreased
from 79.2 0.1 to 72.6 0.1% indicating that the W-KIT-6 (2 h, 9.2)
catalyst activity is quite stable. These results show that the W-KIT-6
(2 h, 9.2) catalyst is quite coke resistant. Following regeneration of
the used W-KIT-6 (2 h, 9.2) catalyst in air (100 mL/min, 550 ◦C, 5 h),
its stability was tested during a further 72 h run. The 2-butene con-
version declined from 81.5 0.2% to 80.0 0.3%, and the propene
yield decreased from 76.2 0.2% to 73.5 0.3% indicating that the
regenerated W-KIT-6 (2 h, 9.2) catalyst is still quite stable and can
be reused.
Fig. 6. Propene yield and catalyst total acidity as a function of tungsten loading
on modified W-KIT-6 (2 h) catalysts at T = 450 ◦C, P = 1 atm, WHSV (ethylene and
2-butene) = 2.0 h−1, n(ethylene)/n(2-butene) = 3/1.

148
J.-F. Wu et al. / Applied Catalysis A: General 528 (2016) 142–149
Table 3
Conversion, selectivity and propene yield for various catalysts with P = 1 atm, WHSV (ethylene and 2-butene) = 2.0 h−1, n(ethylene)/n(2-butene) = 3/1, T = 450 ◦C, except where
noted.
Selectivity (%)a
Catalysts
2-Butene conversion (%)a
Propene
1-Butene
Othersb
Propene yield (%)a
WO3/SiO2 (7.4)
WO3/KIT-6 (7.6)
W-KIT-6 (13.7)
W-KIT-6 (8.7)
W-KIT-6 (5.9)
W-KIT-6 (4.1)
W-KIT-6 (2.8)
W-KIT-5 (15.6)
W-KIT-5 (9.1)
W-KIT-5 (5.5)
W-KIT-5 (2.6)
W-SBA-16 (13.1)
W-SBA-16 (10.0)
W-SBA-16 (5.2)
W-SBA-16 (3.0)
W-KIT-6 (2 h, 18.2)
W-KIT-6 (2 h, 9.2)
W-KIT−6 (2 h, 9.2)c
W-KIT-6 (2 h, 7.2)
W-KIT-6 (2 h, 2.8)
W-KIT-6 (1 h, 9.3)
W-KIT-6 (3 h, 9.9)
W-KIT-6 (5 h, 9.3)
W-KIT-6 (8 h, 9.7)
47.0
50.2
46.6
54.9
58.6
54.7
44.3
45.4
57.6
56.1
46.6
53.0
57.5
58.3
48.5
57.8
68.3
84.2
59.8
40.7
61.4
63.9
54.4
52.3
60.5
66.2
78.6
78.1
78.6
78.4
78.7
73.9
77.8
75.3
65.0
76.4
78.2
77.6
68.8
77.8
86.7
94.1
79.6
72.8
82.6
84.2
75.8
74.4
38.6
32.4
20.7
20.9
20.1
20.6
20.5
25.2
20.9
23.3
34.0
22.6
20.6
21.0
30.1
20.9
11.5
4.5
1.0
1.4
0.7
1.0
1.3
1.0
0.7
0.8
1.4
1.4
1.0
0.7
1.2
1.4
1.1
1.4
1.6
1.3
1.5
0.7
1.7
1.6
1.5
1.6
28.4
33.2
36.6
42.9
46.1
42.9
34.9
33.5
44.8
42.3
30.3
40.5
45.0
45.3
33.4
44.9
59.2
79.2
47.6
29.6
50.7
53.8
41.3
38.9
18.9
26.5
15.5
14.0
22.5
23.9
a
Conversion, selectivity and yield values represent averages obtained between 3 and 7 h on stream.
Pentenes, hexenes, heptenes, and octenes (C5 = and C5+).
b
c
Reactions were performed at 500 ◦C.
3.8. Metathesis activity and selectivity
All the W-catalysts were tested in the cross-metathesis of 2-
butene and ethylene at 450 and/or 500 ◦C (Figs. 3, 4 a, 7, and
S18–S23) and the comparative results are summarized in Table 3.
The 2-butene conversion, propene selectivity, and 1-butene selec-
tivity remain stable after an approximately 1.5 h initial break-in
period. The targeted product, propene, is believed to form pri-
marily through two pathways (Scheme 2): the cross-metathesis
of 2-butene and ethylene (Reaction 1) and the cross-metathesis of
2-butene and 1-butene (Reaction 4). The main byproduct on all W-
catalysts tested in this work is 1-butene, which is mainly generated
through the isomerization of 2-butene (Reaction 2). Other byprod-
ucts include pentenes, hexenes, heptenes, and octenes with their
total selectivity being less than 2% (Table 3). 2-Pentene is generated
via the cross-metathesis of propene and 1-butene (Reaction 3) and
the cross-metathesis of 1-butene and 2-butene (Reaction 4).
The cross-metathesis of 2-butene and ethylene to propene
and the isomerization of 2-butene to 1-butene are parallel reac-
tions. Based on the relatively low selectivity of pentenes (Table 3),
the contributions of Reactions 3 and 4 (Scheme 2) are relatively
insignificant. Hence, the propene selectivity depends primarily on
the competition between Reactions 1 and 2. As clearly borne out
by the catalyst evaluation results, propene yield is favored by an
Fig. 7. Long-term stability of W-KIT-6 (2 h, 9.2) catalyst during a 120 h running and
its performance following catalyst regeneration in air after 72 h. Other conditions
are T = 500 ◦C, P = 1 atm, WHSV (ethylene and 2-butene) = 2.0 h−1, n(ethylene)/n(2-
butene) = 3/1.
Scheme 2. Overall reaction network on W-incorporated mesoporous silicates with 2-butene and ethylene as reactants.

J.-F. Wu et al. / Applied Catalysis A: General 528 (2016) 142–149
149
increased amount of active W species on the catalyst surface. This
conclusion is also supported by XPS results.
[7] K. Grela, Olefin Metathesis: Theory and Practice, John Wiley & Sons, 2014.
[8] S. Lwin, I.E. Wachs, ACS Catal. 4 (2014) 2505–2520.
[9] N. Gholampour, M. Yusubov, F. Verpoort, Cat. Rev. Sci. Eng. 58 (2016) 113–156.
[10] A. Spamer, T.I. Dube, D.J. Moodley, C. van Schalkwyk, J.M. Botha, Appl. Catal. A:
Gen. 255 (2003) 153–167.
4. Conclusion
[11] 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: Gen. 467 (2013) 224–234.
[12] B.A. Zhang, N.S. Liu, Q.S. Lin, D. Jin, J. Mol. Catal. 65 (1991) 15–28.
[13] T. Ookoshi, M. Onaka, Chem. Commun. (1998) 2399–2400.
[14] T. Hahn, U. Bentrup, M. Armbruester, E.V. Kondratenko, D. Linke,
ChemCatChem 6 (2014) 1664–1672.
[15] T. Hahn, E.V. Kondratenko, D. Linke, Chem. Commun. 50 (2014) 9060–9063.
[16] S.J. Huang, S.L. Liu, W.J. Xin, J. Bai, S.J. Xie, Q.X. Wang, L.Y. Xu, J. Mol. Catal. A:
Chem. 226 (2005) 61–68.
W-incorporated mesoporous catalysts (W-KIT-6, W-KIT-5, and
W-SBA-16) perform significantly better than WO3-impregnated
catalysts (WO3/SiO2 and WO3/KIT-6) for ethylene and 2-butene
metathesis to propene. The superior performance of the W-
incorporated ordered mesoporous silicates stems from the
higher dispersion of active tungsten precursors compared to
WO3-impregnated catalysts. All the investigated W-incorporated
mesoporous silicates exhibit steady activity and stability at 450 ◦C
during 7 h runs in a fixed-bed reactor. Further, on all these catalysts,
propene yield is similar and maximized at an optimum W load-
ing and catalyst acidity, suggesting that the structural differences
among these catalysts do not play a major role in their performance.
It was demonstrated that optimally delayed addition of the W
source (instead of simultaneous addition with the Si source) results
in a better dispersion of active tungsten precursors on the surface
of the catalyst rather than being buried in the walls. The better
dispersion was manifested in improved metathesis activity as also
supported by NH3-TPD measurements and XPS results. The best
performing W-KIT-6 catalyst (9.2 wt% W and 2 h delayed addition
of W source) exhibited about 79% propene yield at 500 ◦C that was
nearly stable during a 120 h extended run. Even the slight decrease
in activity is easily regenerated by calcination in air.
[17] X.J. Li, W.P. Zhang, S.L. Liu, L.Y. Xu, X.W. Han, X.H. Bao, J. Catal. 250 (2007)
55–66.
[18] Q. Zhao, S.-L. Chen, J. Gao, C. Xu, Transit. Met. Chem. 34 (2009) 621–627.
[19] D.P. Debecker, B. Schimmoeller, M. Stoyanova, C. Poleunis, P. Bertrand, U.
Rodemerck, E.M. Gaigneaux, J. Catal. 277 (2011) 154–163.
[20] D.P. Debecker, M. Stoyanova, F. Colbeau-Justin, U. Rodemerck, C. Boissiere,
E.M. Gaigneaux, C. Sanchez, Angew. Chem. Int. Ed. 51 (2012) 2129–2131.
[21] M. Taoufik, E. Le Roux, J. Thivolle-Cazat, J.M. Basset, Angew. Chem. Int. Ed. 46
(2007) 7202–7205.
[22] E. Mazoyer, K.C. Szeto, N. Merle, S. Norsic, O. Boyron, J.M. Basset, M. Taoufik,
C.P. Nicholas, J. Catal. 301 (2013) 1–7.
[23] A.W. Moses, C. Raab, R.C. Nelson, H.D. Leifeste, N.A. Ramsahye, S.
Chattopadhyay, J. Eckert, B.F. Chmelka, S.L. Scott, J. Am. Chem. Soc. 129 (2007)
8912–8920.
[24] H. Liu, K. Tao, P.P. Zhang, W. Xu, S.H. Zhou, New J. Chem. 39 (2015) 7971–7978.
[25] K. Tao, Q.X. Ma, N. Tsubaki, S.H. Zhou, L. Han, J. Mol. Catal. A: Chem. 416
(2016) 39–46.
[26] D.R. Hua, S.L. Chen, G.M. Yuan, Y.L. Wang, Q.F. Zhao, X.L. Wang, B. Fu,
Microporous Mesoporous Mater. 143 (2011) 320–325.
[27] B. Hu, H. Liu, K. Tao, C.R. Xiong, S.H. Zhou, J. Phys. Chem. C 117 (2013)
26385–26395.
[28] W. Xu, C. Lin, H. Liu, H.B. Yu, K. Tao, S.H. Zhou, RSC Adv. 5 (2015) 23981–23989.
[29] K. Bouchmella, P.H. Mutin, M. Stoyanova, C. Poleunis, P. Eloy, U. Rodemerck,
E.M. Gaigneaux, D.P. Debecker, J. Catal. 301 (2013) 233–241.
[30] R.D. Andrei, M.I. Popa, C. Cammarano, V. Hulea, New J. Chem. 40 (2016)
4146–4152.
Acknowledgments
The authors acknowledge financial support from several
sources: U.S. Department of Agriculture and the National Insti-
tute of Food and Agriculture through Grant No. 2011-10006-30362,
Kansas Board of Regents K*STAR Program, and joint National Sci-
ence Foundation and Environmental Protection Agency program
Networks for Sustainable Material Synthesis and Design (NSF-EPA
1339661). We thank Prof. Franklin (Feng) Tao, Dr. Luan Nguyen
and Dr. Shiran Zhang for helpful discussions about XPS, Dr. Prem S.
Thapa for helping with XPS measurements, and Dr. Victor W. Day
for helping with XRD measurements.
[31] Z. Cheng, C.S. Lo, ACS Catal. 2 (2012) 341–349.
[32] Z. Cheng, C.S. Lo, ACS Catal. 5 (2015) 59–72.
[33] A. Taguchi, F. Schuth, Microporous Mesoporous Mater. 77 (2005) 1–45.
[34] D. Zhao, Y. Wan, W. Zhou, Ordered mesoporous materials, John Wiley & Sons,
2012.
[35] B. Hu, C. Xiong, K. Tao, S. Zhou, J. Porous Mater. 22 (2015) 613–620.
[36] C. Lin, K. Tao, H.B. Yu, D.Y. Hua, S.H. Zhou, Catal. Sci. Technol. 4 (2014)
4010–4019.
[37] D.P. Debecker, K. Bouchmella, C. Poleunis, P. Eloy, P. Bertrand, E.M. Gaigneaux,
P.H. Mutin, Chem. Mater. 21 (2009) 2817–2824.
[38] D.P. Debecker, K. Bouchmella, M. Stoyanova, U. Rodemerck, E.M. Gaigneaux,
P.H. Mutin, Catal. Sci. Technol. 2 (2012) 1157–1164.
[39] K. Bouchmella, M. Stoyanova, U. Rodemerck, D.P. Debecker, P.H. Mutin, Catal.
Commun. 58 (2015) 183–186.
Appendix A. Supplementary data
[40] S. Maksasithorn, P. Praserthdam, K. Suriye, M. Devillers, D.P. Debecker, Appl.
Catal. A: Gen. 488 (2014) 200–207.
[41] A. Ramanathan, B. Subramaniam, D. Badloe, U. Hanefeld, R. Maheswari, J.
Porous Mater. 19 (2012) 961–968.
[42] A. Ramanathan, R. Maheswari, B.P. Grady, D.S. Moore, D.H. Barich, B.
Subramaniam, Microporous Mesoporous Mater. 175 (2013) 43–49.
[43] R. Thomas, J.A. Moulijn, V.H.J. Debeer, J. Medema, J. Mol. Catal. 8 (1980)
161–174.
Supplementary data associated with this article can be found,
in the online version, at http://dx.doi.org/10.1016/j.apcata.2016.10.
004.
References
[44] X.-L. Yang, W.-L. Dai, R. Gao, K. Fan, J. Catal. 249 (2007) 278–288.
[45] T.W. Kim, F. Kleitz, B. Paul, R. Ryoo, J. Am. Chem. Soc. 127 (2005) 7601–7610.
[46] C. van Schalkwyk, A. Spamer, D.J. Moodley, T. Dube, J. Reynhardt, J.M. Botha,
Appl. Catal. A: Gen. 255 (2003) 121–131.
[47] R. Maheswari, M.P. Pachamuthu, A. Ramanathan, B. Subramaniam, Ind. Eng.
Chem. Res. 53 (2014) 18833–18839.
[48] J.J. Shan, W.X. Huang, L. Nguyen, Y. Yu, S.R. Zhang, Y.Y. Li, A.I. Frenkel, F. Tao,
Langmuir 30 (2014) 8558–8569.
[49] E. Briot, J.Y. Piquemal, M. Vennat, J.M. Bregeault, G.A. Chottard, J.M. Manoli, J.
Mater. Chem. 10 (2000) 953–958.
[1] Shale gas: Reshaping the US chemicals industry 2012 PwC, www.pwc.com/
us/en/industrial-products/publications/assets/pwc-shale-gas-chemicals-
industry-potential.pdf.
[2] J. Foster, Can Shale Gale Save the Naphtha Crackers? 2013 http://www.platts.
com/im.platts.content/insightanalysis/industrysolutionpapers/
shalegasreport13.pdf.
[3] Y. Imamoglu, B. Zumreoglu-Karan, A.J. Amass, Olefin Metathesis and
Polymerization Catalysts, Kluwer Academic Publishers, 1990.
[4] K.J. Ivin, J.C. Mol, Olefin Metathesis and Metathesis Polymerization, Academic
Press, 1997.
[5] J.C. Mol, J. Mol. Catal. A: Chem. 213 (2004) 39–45.
[6] N. Popoff, E. Mazoyer, J. Pelletier, R.M. Gauvin, M. Taoufik, Chem. Soc. Rev. 42
(2013) 9035–9054.
[50] S.K. Gangwal, J. Fathikalajahi, G.B. Wills, Ind. Eng. Chem. Prod. Res. 16 (1977)
237–241.
[51] A. Andreini, J.C. Mol, J. Colloid Interface Sci. 84 (1981) 57–65.
[52] R.C. Luckner, G.B. Wills, J. Catal. 28 (1973) 83–91.