Shape selective methanol to olefins over highly thermostable DDR catalysts

Article abstract of DOI:10.1016/j.apcata.2010.07.023

ZSM-58, having a DDR topology, is shown to be a very attractive catalyst for the direct formation of propylene and ethylene via conversion of methanol. A performance similar to the state of the art SAPO-34 catalysts is achieved, while no olefins longer than C₄ are formed. In addition, ZSM-58 has a much higher thermostability than SAPO catalysts. Mainly propylene, ethylene and linear butenes (trans-but-2-ene and butadiene) are formed when materials with the DDR topology are used as catalysts during the MTO process. The ratio propylene/ethylene can be tuned by changing the reaction conditions or the degree of catalyst coking. An optimum in performance, in terms of stability and selectivity, is found for catalysts containing one acid site (one Al) per accessible cavity. Deactivation of the catalysts takes place due to formation of coke and homogeneous blocking of the catalysts porosity. Activity is fully recovered after regeneration in air.

Full text of DOI:10.1016/j.apcata.2010.07.023

Applied Catalysis A: General 391 (2011) 234–243
Contents lists available at ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
Shape selective methanol to olefins over highly thermostable DDR catalysts
Yasukazu Kumitaa,b, Jorge Gascona,∗, Eli Stavitskia,c, Jacob A. Moulijna, Freek Kapteijna
a Catalysis Engineering – Chemical Engineering Department, Delft University of Technology, Julianalaan 136, 2628 BL Delft, The Netherlands
b Global R&D – Processing Development, Kao Corporation 1334, Minato Wakayama 640-8580, Japan
c Inorganic Chemistry and Catalysis Group, Debye Institute for Nanomaterials Science, Utrecht University, Sorbonnelaan 16, 3584 CA Utrecht, The Netherlands
a r t i c l e i n f o
a b s t r a c t
Article history:
ZSM-58, having a DDR topology, is shown to be a very attractive catalyst for the direct formation of
propylene and ethylene via conversion of methanol. A performance similar to the state of the art SAPO-
34 catalysts is achieved, while no olefins longer than C4 are formed. In addition, ZSM-58 has a much
higher thermostability than SAPO catalysts.
Received 22 February 2010
Received in revised form 12 May 2010
Accepted 13 July 2010
Available online 22 July 2010
Mainly propylene, ethylene and linear butenes (trans-but-2-ene and butadiene) are formed when
materials with the DDR topology are used as catalysts during the MTO process. The ratio propy-
lene/ethylene can be tuned by changing the reaction conditions or the degree of catalyst coking. An
optimum in performance, in terms of stability and selectivity, is found for catalysts containing one acid
site (one Al) per accessible cavity. Deactivation of the catalysts takes place due to formation of coke and
homogeneous blocking of the catalysts porosity. Activity is fully recovered after regeneration in air.
© 2010 Elsevier B.V. All rights reserved.
Keywords:
DDR
ZSM-58
MTO
ZSM-5
NH3-TPD
1. Introduction
like SAPO-34, it was discovered that ethylene and propylene can
be selectively produced.
Light olefins, namely propylene and ethylene, are among the
most demanded raw materials in the chemical industry. Their over-
all demand, especially that of propylene is forecasted to keep rising
during the next decades [1]. Traditionally, they have been produced
at very large scale in steam cracking, a non-catalytic process. These
processes are robust but the selectivity is modest. Therefore, it
is not surprising that catalytic processes to supply these chemi-
cals are of high interest for both industry and academia. FCC (fluid
catalytic cracking) processes have been adapted by modifying the
catalysts, increasing the production of lower olefins, in particular
propylene. Several other approaches appear to be promising. Cat-
alytic dehydrogenation of lower paraffins [2], direct or oxidative,
modifications in the catalytic cracking and the direct conversion
of methanol to olefins have been proposed as possible processes.
The latter, the so-called methanol to olefins (MTO) process has
attracted the attention of researchers during the last four decades
[3]. Methanol can be made either from syngas (from natural gas or
coal) or even via fermentation. Subsequently it can be catalytically
processed to gasoline (methanol-to-gasoline, MTG) or to olefins
(MTO). The MTG process was initially designed as an alternative
to Fischer–Tropsch-synthesis, but with the development of new
small pore zeolites, more specifically silico-alumino-phosphates
Most of the open MTO literature deals with the use of either
ZSM-5 or SAPO-34 catalysts, while only recently the use of other
zeolites like ZSM-11, ZSM-22, SAPO-5, SAPO-11, SAPO-18 and
SAPO-35 has been reported [4–6].
It is claimed that the mild acidity of SAPO-34 together with its
CHA topology are responsible for the high selectivity of the mate-
rial towards the formation of ethylene and propylene [3]. This is
probably the reason why mainly silico-alumino-phosphates have
been explored. However, not much is known about materials with
similar topologies containing only Si and Al. From the applica-
tion viewpoint, it would be desirable to produce a catalyst with
a higher thermal stability, since during the MTO process regener-
ation of the catalysts at high-temperatures in the presence of air
is an integral part of the operation and silico-alumino-phosphates
have a limited thermal stability compared to silico-aluminate
systems [7].
A possible candidate is ZSM-58, a silico-aluminate with low
aluminium content, having a DDR topology firstly synthesized
by Exxon Mobil [8,9]. The structure of DDR consists of window-
connected cages, in contrast to zeolite frameworks, which makes
this material a member of the group of clathrasils. The crystal struc-
ture is built by corner-sharing SiO4 tetrahedra that are connected
to pseudohexagonal layers of face-sharing pentagonal dodecahe-
dra (512 cages). These layers are stacked in an ABCABC sequence
and are interconnected by additional SiO4 tetrahedra that form
six-membered rings between the layers. Thus, two new types of
∗
Corresponding author. Tel.: +31 015 2789820.
E-mail address: j.gascon@tudelft.nl (J. Gascon).
0926-860X/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.apcata.2010.07.023

Y. Kumita et al. / Applied Catalysis A: General 391 (2011) 234–243
235
cages arise, a small decahedron, 435661 cage, and a large 19-hedron,
435126183 cage. By connecting the 19-hedra cavities through a
single 8-ring with an aperture of 4.4 A × 3.6 A, a two-dimensional
porous system is formed, which is accessible to small molecules
after thermal treatment at 773 K [10,11].
of each sample was pretreated at 923 K and rapidly cooled to
473 K followed by loading with ammonia applying a flow of 30 mL
(STP)/min for about 10 min. A helium flow of 50 mL/min was
applied for 10 min to remove weakly adsorbed NH3. The adsorp-
tion process was repeated 3 times. After the loading procedure, a
linear temperature program was started (473–973 K at 10 K/min) in
a flow of He (50 ml/min), and the ammonia desorption was followed
by the TCD.
˚
˚
The selective adsorption of propylene over propane and trans-
2-butene and butadiene over other butane and butene isomers
together with the high uptakes of ethylene on the all-silica DD3R
were firstly reported by our group by studying single compo-
nent adsorption using a Tapered Element Oscillating Microbalance
(TEOM) [12–14]. Olson et al. [15] studied the sorption properties
of some 8-ring zeolites, among which ZSM-58, in which the high
selectivity for propylene was also suggested from single component
adsorption. Later we reported the first propane–propylene mixture
separation using DDR in a packed bed adsorption operation [16].
Up to now, little is known about the catalytic properties of ZSM-
58 in the open literature, to the best of our knowledge only its use
in the catalytic cracking of n-octane in its acidic form [17] and the
application of the Mn exchanged form in the catalytic oxidation of
n-hexane [18] have been reported.
In this work we present the application of ZSM-58 for the direct
synthesis of olefins via methanol conversion. The performance of
this DDR-structure type catalyst is comparable in terms of activity
to the best SAPO-34 catalysts, it shows high selectivities to propy-
lene and ethylene, while no product longer than linear butenes is
formed. Moreover, the ratio propylene/ethylene can be tuned either
by changing the reaction temperature or by controlling the coke
level in the catalyst.
2.2. Synthesis of the catalysts
For the synthesis of ZSM-58 [19], first the structure direct-
ing agent (SDA), methyltropinium iodide (MTI), was prepared by
adding 25.1 g methyl iodide (99 wt.%) dropwise to a solution of
25.0 g tropine (98 wt.%) in 100 g ethanol at 273 K under stirring and
keeping the suspension under reflux for 72 h. After cooling and fil-
tration, the resulting crystalline MTI was washed with 100 g ethanol
and dried at 353 K.
The
molar
ratio
of
the
synthesis
mixture
was
MTI:SiO2:Al2O3:Na2O:H2O = 17.5:70:x:11.5:2800. It was pre-
pared as follows: 10 g of Ludox HS-40 was added to a solution of
4.7 g MTI and 8.7 g demineralised H2O and stand overnight while
stirring (solution A). 0.78 g sodium hydroxide and the correspond-
ing amount of sodium aluminate (Al2O3: 50–56, Na2O: 40–45 wt%)
were dissolved in 33.3 g H2O (solution B). The resulting solution
(A + B) is stirred for 30 min, placed in a teflon lined autoclave and
heated under hydrothermal conditions (autogenous pressure) to
433 K for 3–7 days depending on the Al content (the higher the
Al content the longer the synthesis time, i.e. 4 days synthesis for
Si/Al = 392 and 7 days for Si/Al = 80). The as synthesized materials
were calcined (2 K/min) at 973 K for 10 h in a static oven.
2. Experimental
A commercial ZSM-5 from Zeolyst International (CBV28014,
SiO2/Al2O3 = 560) was used as benchmark catalyst.
2.1. General
All chemicals were obtained from Sigma–Aldrich and were used
without further purification. Scanning Electron Microscopy (SEM)
was measured in a JEOL JSM 6500F setup coupled to an Energy
Dispersive Spectrometer (EDS) for micro-analysis.
2.3. Experimental setup
Experiments were carried out with a 1/4 in. stainless steel reac-
tor with a length of 6 cm. 100 mg of catalyst particles (pelleted
zeolite, crushed and sieved to a size of 0.75–1 mm) was placed in
the reactor between quartz wool plugs for packing purpose and gas
distribution. An ISCO liquid pump was used to feed the liquid to the
reactor system.
In a typical experiment, the liquid feed stream (methanol or
methanol + water) was pumped to the reaction section located in
a heated chamber (373 K) to evaporate the liquid while diluting it
with N2. In this chamber all valves and reactor oven were located
to avoid condensation. The temperature of the reactor was varied
between 573 and 723 K in the different experiments. The product
mixture was analyzed online with an Interscience Compact GC over
a 60 m RTX®-1 (1% diphenyl-, 99% dimethylpolysiloxane) column.
Conversion (%) was calculated as follows:
Nitrogen adsorption at 77 K in a Quantachrome Autosorb-6B
unit gas adsorption analyzer was used to determine the textural
properties as BET surface area between 0.05 and 0.15 relative pres-
sures and pore volume at 0.95 relative pressure. The t-plot was
used to calculate the external surface area of the catalyst particles
(calculated as the surface area of pores larger than micropores).
The crystalline structures were analyzed by X-ray diffraction (XRD)
using a Bruker-AXS D5005 with CuK␣ radiation. Thermogravimet-
ric analysis (TGA) was performed by means of a Mettler Toledo
TGA/SDTA851e, on samples of 10 mg under flowing 60 ml/min of
air at a heating rate of 10 K/min up to 873 K.
Elemental analysis was carried out by means of inductively cou-
pled plasma optical emission spectroscopy (ICP-OES). The samples
were digested in duplo in an aqueous mixture of 1% HF and 1.25%
H2SO4 and analyzed with an ICP-OES Perkin Elmer Optima 3000dv
in order to determine the amount of Si and Al present in the struc-
ture.
DRIFT spectra were recorded in a Bruker model IFS66 spectrom-
eter, equipped with a high-temperature cell with CaF2 windows.
The spectra were recorded after accumulation of 128 scans and a
resolution of 4 cm−1. A flow of helium at 10 ml/min was maintained
during the measurements. Before collecting the spectra the differ-
ent samples were pretreated in a helium flow at 393 K for 30 min.
KBr was used as background.
The acid site concentration was determined using temperature-
programmed desorption of ammonia (NH3-TPD). NH3-TPD was
carried out on a Micromeritics TPR/TPD 2900 apparatus equipped
with a thermal conductivity detector (TCD). Approximately 25 mg
NMeOH − NMeOH
out
in
X = 100 ×
NMeOH
in
The overall olefin selectivity (%) was based on the carbon atoms
in the products and calculated as follows:
2NC
+ 3NC
in
+ 4NC
H
H
H
4 8
2
4
6
NMeOH −3NMeOH
S = 100 ×
out
With Ni being the number of moles.
Similarly that for the individual olefins is defined. The yield (%) of
olefin i is then simply the product of conversion and its selectivity:
SiX
Yi =
100

236
Y. Kumita et al. / Applied Catalysis A: General 391 (2011) 234–243
Fig. 1. SEM micrographs (a–d) and XRD patterns (e) of the different ZSM-58 catalysts: SiO2/Al2O3 = (a) 64, (b) 110, (c) 215 and (d) 392.
textural properties before and after 6 reaction hours are sum-
marized. Despite minor differences, every catalyst shows a well
developed crystalline DDR pattern, while the particle size increases
with decreasing the Al content, ranging from 0.5 to 7 ␮m. From
the ICP/MS analyses it is calculated that 2, 1, 0.6 and 0.3 atoms
of Al are present per accessible cavity in the catalyst for the sam-
ples with a SiO2/Al2O3 ratio of 64, 110, 215 and 393, respectively.
3. Experimental results
3.1. Characterization
SEM micrographs of the ZSM-58 catalysts under study are
shown in Fig. 1 together with their XRD patterns. In Table 1 the
Table 1
Textural properties of the different ZSM-58 catalysts.
SiO2/Al2O3
Fresh catalyst
Spent catalyst
SBET (m2/g)
Sexternal (m2/g)
Smicro (m2/g)
SBET (m2/g)
Sexternal (m2/g)
Smicro (m2/g)
Coke (wt.%)
64
110
215
392
307
332
357
396
35
28
19
18
271
305
337
378
23
12
6
21
14
8
2
–
–
–
12.1
9.8
6.5
6
9
6.6

Y. Kumita et al. / Applied Catalysis A: General 391 (2011) 234–243
237
performed at least 2 times with an intermediate regeneration step
in air at 723 K. Activity was completely recovered in every case.
For the sake of clarity and comparison between different opera-
tional conditions, results are presented as a function of the amount
of methanol fed to the reactor per gram of catalyst (‘through-
put’). Main products observed were propylene, ethylene and linear
butenes/butadienes (mostly trans-but-2-ene and buta-1,3-diene).
Formation of coke was observed and quantified by TGA at the end
of every experiment. Selectivity is given as overall selectivity to
C2, C3 and C4 olefins, the balance is assumed to be ‘coke’. In every
case during the first reaction minutes, up to approximately 0.75 g
MeOH/g cat no product breakthrough can be observed. The length
of this period does neither follow any trend with the aluminium
content, nor with the specific surface area of the samples. While the
catalysts with the lowest acidity (SiO2/Al2O3 = 215 and 393) deacti-
vated very fast, the other two (SiO2/Al2O3 = 64 and 110) are in terms
of conversion and selectivity to olefins stable for longer periods: in
both cases after 5 g MeOH fed per gram of catalyst, strong deacti-
vation occurs and conversion drops from 100% to approximately
40%. With respect to the product distribution, at the beginning of
the reaction more propylene than ethylene is formed and during
time-on-stream this ratio changes and ethylene becomes the more
important product. In every case the yield of butenes was lower
than 15% on a molar carbon basis and decreased with time-on-
stream. On the other hand, when conversion dropped below 100%,
dimethyl ether (DME) started to be observed at the exit of the
reactor, becoming the most important product at the end of the
experiments. In view of these screening results, the sample with a
SiO2/Al2O3 = 110 was selected as the optimal one in terms of stabil-
ity (resistance to deactivation) and selectivity to olefins, therefore
further experiments were performed over this catalyst.
Fig. 2. Ammonia TPD analyses of the different ZSM-58 and ZSM-5 catalysts used,
indicated by their SiO2/Al2O3 ratio.
Regarding specific surface area, a clear trend is observed: the lower
the amount of aluminium the higher the BET area; the observed
external surface area correlates well with the mean particle size.
In Fig. 2, the acidic properties of the synthesized catalysts are
compared with that of the studied commercial ZSM-5 catalyst.
Three desorption steps are observed. A major one with a maximum
shifting from 650 to 700 K and increasing with increasing Al con-
tent, a desorption between 800 and 900 K similar for all samples,
and a desorption around 550 K increasing with increasing Al con-
tent. The acidity, i.e. the amount of ammonia desorbing, correlates
with the amount of aluminium in the framework for the ZSM-58
samples, while a lower acidity is obtained for the ZSM-5 sample:
apart from a much higher uptake, the temperature of maximum
ammonia desorption is also higher for the ZSM-58 catalysts. Both
facts point out at a stronger intrinsic acidity of the DDR framework.
Fig. 4 shows the product distribution with time for the optimal
ZSM-58 catalyst at four different temperatures. Temperature plays
a major role, not only on the activity and stability of the catalyst,
but also on the distribution of products. While at low temperatures
propylene is the preferred product (reaching yields above 40%), at
3.2. MTO on ZSM-58
Fig. 3 shows the reaction results for the different catalysts at
673 K feeding methanol using N2 as diluting gas. Experiments were
Fig. 3. MTO reaction results for the different ZSM-58 catalysts: SiO2/Al2O3 = 64, 110, 215 and 392 (from left to right and from top to bottom). Mcat = 100 mg, FMeOH = 0.25 ml/h
(liquid) (WSHV=2 h−1), MeOH:N2 = 1:4, T = 673 K.

238
Y. Kumita et al. / Applied Catalysis A: General 391 (2011) 234–243
Fig. 4. C2–C4 olefin distribution during the MTO reaction over a ZSM-58 catalyst: SiO2/Al2O3 = 110, Mcat = 100 mg, FMeOH = 0.25 ml/h (liquid) (WSHV=2 h−1), MeOH:N2 = 1:4.
(a) T = 573 K, (b) 623 K, (c) 673 K and (d) 723 K.
Fig. 5. Effect of water addition on the MTO reaction results of
a
ZSM-58 catalyst: SiO2/Al2O3 = 110, T = 673 K. Mcat = 100 mg, FMeOH = 0.15 g/h (WSHV=2 h−1), (a)
MeOH:H2O:N2 = 1:5:6 and (b) MeOH:N2 = 1:11.

Y. Kumita et al. / Applied Catalysis A: General 391 (2011) 234–243
239
the highest temperature ethylene is preferentially formed. In gen-
eral, the higher the amount of propylene formed, the higher the
amount ofbutenes generatedand thelowerthe amountofethylene.
With respect to stability, at the lowest and highest temperatures
(573 and 723 K) the catalyst deactivates faster, while at interme-
diate temperatures, and especially at 673 K, the catalyst is much
more stable.
It is generally observed that the addition of water enhances
the catalyst stability during the MTO reaction and that the prod-
uct distribution may change upon varying the partial pressure of
MeOH [3]. The effects of these two variables are presented in Fig. 5,
showing the results of experiments at lower MeOH partial pres-
sure and in the presence or absence of H2O. Nitrogen was used as
diluent, keeping the partial pressure of MeOH constant. The addi-
tion of water, in the case of ZSM-58, increases the selectivity to
lower olefins, approaching 100% during the stage of high methanol
conversions. The percentual formation of the longer butenes is sup-
pressed, while hardly any DME is formed. Further, the stability
is slightly improved, though in both cases conversion drops after
more than 4 g of methanol have been fed per gram of catalyst.
Lower MeOH partial pressures have a negative effect on the cat-
alyst performance when comparing the results presented in Fig. 5b
with those at the same temperature in Figs. 3 and 4: the stability
is slightly lower in terms of methanol throughput, while the yield
to propylene decreases and the overall selectivity to olefins is also
lower, so more coke is formed.
3.3. ZSM-5 as benchmark catalyst
In order to benchmark the catalytic properties of the studied
ZSM-58 catalysts, experiments with a commercial high silica ZSM-
5 catalyst were performed. Differences in terms of topology are
expected to play an important role in the product distribution and
in the stability of the catalyst. In Fig. 6 the experimental results
for the ZSM-5 catalyst at different MeOH partial pressures and
in the presence and absence of water are presented, under simi-
Fig. 6. MTO reaction results on a high Si ZSM-5 catalysts: SiO2/Al2O3 = 140, Mcat = 100 mg, T = 673 K. (a) FMeOH = 0.15 g/h (WSHV=2 h−1), MeOH:N2 = 1:4; (b) FMeOH = 0.15 g/h
(WSHV=2 h−1), MeOH:H2O:N2 = 1:5:6 and (c) FMeOH = 0.15 g/h (WSHV=2 h−1), MeOH:N2 = 1:11.

240
Y. Kumita et al. / Applied Catalysis A: General 391 (2011) 234–243
Fig. 7. Product composition during conversion of ethylene (left) and propylene (right) with time-on-stream on a fresh ZSM-58 catalyst. SiO2/Al2O3 = 110, T = 673 K.
Mcat = 100 mg, Ftotal = 15 ml (STP)/min; 8% olefin in feed. Yield is defined per carbon atom of reactant.
lar conditions as used for the ZSM-58 catalysts. The results when
using ZSM-5 differ quite from the ones obtained for ZSM-58. Main
products in order of decreasing amount are propylene, butenes
and ethylene. The activity of ZSM-5 is fairly stable, while the total
selectivity to olefins is constant: after feeding more than 10 g of
MeOH per gram of catalyst, the activity is still high and the distri-
bution of products remains nearly invariant. The addition of water
seems to improve the stability but has little effect on the product
distribution; the selectivity to olefins is similar, in contrast to the
20 points higher selectivity observed when adding water to the
ZSM-58 system.
It is well known that for consecutive reactions, valuable infor-
mation can be extracted by feeding intermediate reaction products
as starting reactants [20]. In this case, we performed several
experiments with both catalysts feeding ethylene and propy-
lene. Surprisingly, while the ZSM-58 catalysts presented a high
activity, the ZSM-5 sample was hardly active. Results for the
SiO2/Al2O3 = 110 ZSM-58 catalyst are presented in Fig. 7. For the
sake of clarity yields are given on a carbon atom basis (i.e.: for the
experiments feeding ethylene, the yield to propylene is calculated
as: YC
= X3NC
/(2(NC
− NC 4 )), where X is the conversion
H
H
H
H
2
6
3
of eth3ylene). The ev6olution2of4the reactants is plotted in the graph as
(100 − X). At the initial stages of the reaction with ethylene, conver-
sions higher than 60% are obtained and large amounts of propylene
and butenes are observed at the reactor exit. Also a large selectivity
to coke is observed in the first reaction minutes. Although propy-
lene turned out to be less reactive, initial conversions near 40% are
achieved with formation of coke, butenes and ethylene.
3.4. Reactivity of reaction products
From the previous experiments, large differences are observed
upon comparing the performance of the ZSM-58 and ZSM-5 cat-
alysts. These differences might be due to the different topologies
of the samples, but also to the strength of the active sites present
in each catalyst. Ammonia TPD characterization indicates that the
acidity of the ZSM-5 catalyst is the lowest one, as it could be
expected from its high Si/Al ratio. However, while in the case of the
ZSM-58 catalysts such low acidities resulted in a very poor reaction
performance and a fast deactivation, in the case of ZSM-5 this mild
acidity is enough for the conversion of methanol.
3.5. Formation of coke
Fig. 8 shows the evolution of the coke formation and the
decrease in specific surface area for the SiO2/Al2O3 = 110 ZSM-58
catalyst sample upon reaction with MeOH, ethylene or propylene
at different reaction times. The amount of coke formed (wt.%) does
not correspond with the decrease in specific surface area and fol-
Fig. 8. Evolution of the specific surface area (left) and the coke amount on the catalysts (estimated by TGA) (right) with time-on-stream for a ZSM-58 catalyst (SiO2/Al2O3 = 110)
for different reaction feeds. T = 673 K, Mcat = 100 mg. Ftotal = 15 ml (STP)/min.

Y. Kumita et al. / Applied Catalysis A: General 391 (2011) 234–243
241
MTO experiment products breakthrough did not take place until
a time corresponding with a throughput of approximately 0.5–1 g
methanol/g catalyst. This is attributed to the time needed to vapor-
ize the liquid fed together with adsorption of methanol and the
first reaction products. Similar phenomena have been reported for
SAPO-34 catalysts [20]. The fact that in the experiments performed
with ethylene and propylene this delay is not observed is related to
the negligible effect of delay time in the case of gaseous reactants
and to the much lower adsorption coverage of these olefins, more
than 5 times smaller at the studied temperatures in the case of
SAPO-34 [13,20]. From Fig. 3 it is clear that the Si/Al ratio plays
a key role in the stability of the catalyst: an optimum is found
for one aluminium atom per accessible cage: lower amounts of
aluminium result in a prompt deactivation of the catalysts, while
higher amounts do not improve further the performance: even con-
sidering the smaller particle size of the SiO2/Al2O3 = 64, the stability
of the catalyst is not better and a larger amount of ethylene is pro-
duced compared to propane. This fact can be related with a higher
cracking activity when more than one acid site is present per cavity
of the material, leading to a faster production of coke (see Table 1)
and therefore to a faster deactivation. When comparing the effect
of the Si/Al ratio with other catalysts with similar topologies [6],
this effect is much stronger in the case of ZSM-58. NH3-TPD con-
firms the strong acid character of the sites in ZSM-58, as compared
to those of ZSM-5. When comparing the NH3-TPD of the different
ZSM-58 samples, it can be observed that the third desorption peak,
centered at 850 K, hardly changes with increasing the amount of
Al in the framework, while the desorption peak centered at 675 K
increases linearly with the amount of aluminium. The NH3-TPD
results point at the presence of, at least two different acid sites:
the strongest ones (peak at 850 K) would suffer a strong and quick
deactivation, while the softer ones (peak at 675 K) would be respon-
sible for the MTO activity. This fact would also account for the swift
deactivation of the low Al content catalysts.
IR spectroscopy indicates the elevated branching degree of the
aromatics deposited in the catalysts, pointing out the high acidity of
this material. During the period of full conversion a change in the
product distribution is observed. In general production of propy-
lene and butenes decreases with time, while the production of
ethylene increases. Considering that DDR is a shape selective cata-
lysts and sorbent, this trend is attributed to the decreasing available
space in the cavities as the reaction proceeds, allowing only the pro-
duction of smaller olefins by hampering the longer ones to escape
from the catalyst pores [31]. This fact can be used to tune the final
product distribution simply by controlling the total amount of coke
in the catalyst during operation.
The integral behaviour of the plug flow reactor used in this study
is not easy to decouple from the reaction results: while in the initial
part of the bed the methanol is converted into olefins, these may
be converted further downstream in the reactor: the longer the
time-on-stream the smaller the region of olefin conversion.
After deactivation of the catalysts DME becomes the most
important reaction product, but at this point most of the catalyst
porosity is occupied by coke (see Fig. 8). This suggests that this DME
is mainly formed at the external surface of the particles on Lewis
sites that have not been deactivated by the deposition of coke.
The total amount of coke observed correlates better with the
particle size than with the Si/Al ratio, similarly as reported for
SAPO-34 [32]: the larger the particle size the higher the probability
of closing the available porosity. This is actually a very impor-
tant factor that may lead to incorrect conclusions when studying
large zeolite crystals: recently Mores et al. introduced two different
coke formation mechanisms for large ZSM-5 and SAPO-34 crystals
(larger than 20 ␮m) and suggested that SAPO-34 is deactivated due
to formation of coke deposits at the outer regions of the crystals.
Though this might be true for huge crystals, considering the large
Fig. 9. DRIFT spectra of a spent ZSM-58 catalyst after 6 reaction hours under differ-
ent reactant conditions.
lows a clear trend with the partial pressure of the hydrocarbon used
as reactant.
In the case of methanol, already after 180 min (corresponding
with a throughput of 5 g MeOH/g catalyst) the total surface area of
the catalyst is lost, though the amount of coke on a weight basis still
increases further with time-on-stream. In the case of the olefins,
still some surface area of the catalyst is available when no further
reaction is observed. In order to characterize the deposited coke,
DRIFT spectroscopic analysis of the spent catalysts after different
reaction conditions was carried out. Fig. 9 summarizes the most
important results. Absorbance bands at 2961, 2930 and 2869 cm−1
common for the C–H stretching on the ␣ position of alkylben-
zenes [21] can be assigned to the aromatic compounds formed
in the zeolite cavities. The low intensity of the absorption above
3000 cm−1 due to stretching vibration of aromatic C–H bond, sug-
gest that the aromatic species are heavily methylated [22]. A band
at around 2960 cm−1 was previously observed in MOR, FAU and
BEA zeolites, and was assigned to the propylated benzenes [21];
large pores of 12MR zeolites allow for the formation of larger alkyl
substituents. In the case of ZSM-58 the accessible cavities of this
material would also be able to host large alkyl substituted ben-
zenes. A strong absorption around 1600 cm−1 (so-called “coke”
band) arises from a combination of ı(C–H) modes of a mixture of
carbonaceous hydrogen-deficient deposits [23–25] and aromatic
ꢀ(C–C) modes [26]. The intensity of the latter band has been used
as a measure of the amount of coke deposited in zeolite channels
[23]. Finally, the adsorption at 1567 cm−1 can be ascribed to alkyl-
naphtalenes or polyphenylene structures [23,27,28]. No IR bands
around 1500 cm−1 are observed which are characteristic of highly
condensed aromatic deposits [26]. The IR band intensities of coke
in the case of methanol as reactant are noticeably higher than in
the other cases.
4. Discussion
As inferred from the similar topology and the 8MR window cage
size of ZSM-58 as SAPO-34 this catalyst has a good performance
in the MTO reaction. Furthermore, mainly propylene, ethylene
and linear butenes (trans-but-2-ene and butadiene) are formed, in
agreement with our adsorption results on DD3R [12–14,16,29,30],
where only these components entered this structure. In every

242
Y. Kumita et al. / Applied Catalysis A: General 391 (2011) 234–243
amounts of coke found after reaction in smaller crystals, (in gen-
eral more than 10 wt.%, see Table 1 and Refs. [32,33]) it is clear that
when dealing with micrometer sized crystallites most of the cat-
alyst porosity is occupied by coke deposits, especially in the case
of the high Al content catalysts (smallest particles). In the case of
the ZSM-58 catalysts, the accessible [435126183] cavity has a free
volume of about 0.35 nm3 and a free cross sectional diameter of
0.875 nm by assuming a sphere-like cage. There are six of these
cavities present per unit cell, and considering that every large cav-
ity can be filled with two stacked benzene molecules at maximum
(or by one heavily branched benzene molecule), this value would
correspond to a 13 wt.% of coke in the spent catalysts, almost the
value calculated by TGA analysis of the spent catalysts containing
one acid site per large cavity. This clearly demonstrates that full
blocking of the pores takes place.
The addition of water, even at low concentrations, improves to
a large extent the performance of the ZSM-58 catalyst. Though the
period of stable operation does not change, almost 100% selectiv-
ity is found to propylene and ethylene. This fact could be related
with a faster desorption of the olefins in the presence of water,
that might compete for the same (acid) adsorption sites, avoiding
further consecutive reaction of these products.
resulting in a 2D pore system, in the case of CHA the cavities com-
˚
˚
municate through 3.8 A × 3.8 A windows in a 3D pore structure. This
difference in topology has an impact on diffusion. Olson et al. [15]
reported much faster diffusion of both propane and propylene in
CHA than in DDR, demonstrating that even a small difference of
˚
0.2 A together with a different porous system (2D vs. 3D) makes a
large difference in the behaviour of adsorbates with kinetic diam-
eters similar to the size of the pores. This fact obviously accounts
for the different product distributions in ZSM-58 and CHA.
Summarizing, ZSM-58 is shown to be a very attractive cata-
lyst for the selective formation of olefins from methanol, with a
selective formation of propylene and ethylene and the possibility
of tuning this ratio. The stability to deactivation, similar in terms of
throughput to the one reported for SAPO-34 catalysts [31], should
be overcome by a proper reactor design. In this sense FCC- or two-
zone fluidized bed reactors [37,38] would be the most appropriate
ones to maintain the catalyst at a certain level of deactivation (coke
content), enabling even to tune the ratio of olefins on demand. In
comparison with other possible candidates for this kind of reac-
tor technology, ZSM-58 presents as important additional advantage
the higher thermal stability of alumino-silicates (higher than 973 K
for ZSM-58) in comparison with silico-alumino-phosphates (not
higher than 773 K).
Methanol conversion already produces water, so this influence
is already present during reaction. At lower methanol pressure
this intrinsic water influence may be less effective, explaining the
slightly faster deactivation of the catalyst. The presence of added
water suppresses the DME formation over the deactivated catalysts
due to thermodynamic limitations.
5. Conclusions
ZSM-58 is a very attractive catalyst for the direct formation of
propylene and ethylene via conversion of methanol. Mainly propy-
lene, ethylene and linear butenes (trans-but-2-ene and butadiene)
are formed when materials with the DDR topology are used as cat-
alysts for the MTO process. Moreover, the ratio propylene/ethylene
can be tuned by changing the reaction conditions or controlling the
coke level if an appropriate reactor is used. An optimum in perfor-
mance, in terms of stability and selectivity is found for catalysts
containing one acid site per accessible cavity. Water increases the
olefin selectivity due to competitive adsorption for the strongly acid
sites and suppressing the consecutive reaction of the olefins. Deac-
tivation of the catalysts takes place due to formation of coke and
homogeneous blocking of the catalysts porosity. Activity is fully
recovered after coke combustion.
Upon comparing the results obtained for ZSM-58 with the ones
obtained for ZSM-5, it is clear that the behaviour of both cata-
lysts is very different and that physical constraints in terms of
topology dominate product distribution and catalyst stability. First
important difference is that the needed acidity as estimated from
ammonia TPD in the case of ZSM-5 is far less than for ZSM-58. The
lower reactivity of propylene and ethylene on the MFI together
with the lower acidity demonstrate that the oligomerization and
coke formation activity of this catalyst are also lower, accounting
for the higher stability of the ZSM-5 catalyst. With respect to the
distribution of products similar amounts of propylene are formed
in both catalysts, but far less ethylene is produced over MFI, while
larger amounts of longer chain olefins are formed, illustrating the
shape selective character of ZSM-58. The addition of water in the
case of MFI clearly has a different effect than in the case of ZSM-58.
The affinity of DDR materials for short chain olefins is stronger than
that of ZSM-5 [34,35], so the coverage of propylene and especially
of ethylene [13] on the ZSM-58 catalysts is much more lowered
by the competition with water. Hence, the consecutive reaction of
these olefins over the strong acid sites is suppressed.
The comparison in terms of performance of the best ZSM-
58 catalysts with results published in literature for other
8-membered ring (8MR) alumino-silicates and silico-alumino-
phosphates [3–6,20,31,33,36] yields an interesting observation:
when using ZSM-58 the amount of butenes formed is lower than
reported for SAPO-34, state of the art 8MR MTO-catalyst, while no
hydrocarbons longer than C4 are detected in the case of DDR in con-
trast to catalysts with the CHA topology [31], and their stabilities
(in terms of deactivation) are similar under comparable reaction
conditions. Considering that the main target of the MTO process
is the selective formation of propylene, followed by ethylene, the
lower selectivity to butenes is an advantage in the case of ZSM-58.
Moreover, by varying the temperatures the yield of propylene may
be double that to ethylene (see Fig. 4b).
The similar performance of ZSM-58 with respect to the state of
the art SAPO-34 catalysts together with the higher thermal stability
of pure alumino-silicates may form the basis for the development
of more robust catalysts for the MTO process.
Acknowledgments
KAO company is gratefully acknowledged for financial sup-
port. J.G. gratefully acknowledges the Netherlands National Science
Foundation (NWO) for his personal VENI grant. The X-ray facilities
of the Department of Materials Science and Engineering of the Delft
University of Technology is acknowledged for the XRD analyses
References
[1] J.S. Plotkin, Catal. Today 106 (2005) 10–14.
[2] K.J. Caspary, H. Gehrke, M. Heinritz-Adrian, M. Schwefer, in: G. Erlt, H.
Knözinger, F. Schüth, J. Weitkamp (Eds.), Handbook of Heterogeneous Catalysis,
Wiley–VCH Verlag GmbH & Co., Weinheim, 2008, pp. 3207–3227.
[3] M. Stöcker, Micropor. Mesopor. Mater. 29 (1999) 3–48.
[4] Z.M. Cui, Q. Liu, W.G. Song, L.J. Wan, Angew. Chem., Int. Ed. 45 (2006)
6512–6515.
[5] A.T. Aguayo, A.G. Gayubo, R. Vivanco, M. Olazar, J. Bilbao, Appl. Catal. A: Gen.
283 (2005) 197–207.
[6] Z. Zhu, M. Hartmann, L. Kevan, Chem. Mater. 12 (2000) 2781–2787.
[7] B.V. Vora, T.L. Marker, P.T. Barger, H.R. Nilsen, S. Kvisle, T. Fuglerud, Nat. Gas
Convers. IV 107 (1997) 87–98.
[8] P.G. Rodweald, E.W. Valyocisk, Conversion of Oxygenate(s) to Hydrocarbon(s)
using ZSM-58 Zeolite catalyst, Mobil Oil Corp., 1987.
When comparing both zeotypes several differences are found in
terms of topology that we believe are responsible for the different
selectivities: DDR is built up of cavities communicating through
small ellipsoidal windows with a pore opening of 3.6 A × 4.4 A,
˚
˚

Y. Kumita et al. / Applied Catalysis A: General 391 (2011) 234–243
243
[9] E.W. Valyocsik, New Zeolite ZSM-58 Prepd. using Methyl-tropinium Cation as
Directing Agent, Mobil Oil Corp., 1987.
[25] M. Rozwadowski, M. Lezanska, J. Wloch, K. Erdmann, R. Golembiewski, J. Kor-
natowski, Chem. Mater. 13 (2001) 1609–1616.
[10] H. Gies, J. Inclusion Phenom. Macrocycl. Chem. 2 (1984) 275–278.
[11] H. Gies, Zeitschrift Für Kristallographie 175 (1986) 93–104.
[12] W. Zhu, F. Kapteijn, J.A. Moulijn, Chem. Commun. (1999) 2453–2454.
[13] W. Zhu, F. Kapteijn, J.A. Moulijn, M.C. den Exter, J.C. Jansen, Langmuir 16 (2000)
3322–3329.
[14] W. Zhu, F. Kapteijn, J.A. Moulijn, J.C. Jansen, Phys. Chem. Chem. Phys. 2 (2000)
1773–1779.
[15] D.H. Olson, M.A. Camblor, L.A. Villaescusa, G.H. Kuehl, Micropor. Mesopor.
Mater. 67 (2004) 27–33.
[16] J. Gascon, W. Blom, A. van Miltenburg, A. Ferreira, R. Berger, F. Kapteijn, Micro-
por. Mesopor. Mater. 115 (2008) 585–593.
[17] S. Altwasser, C. Welker, Y. Traa, J. Weitkamp, Micropor. Mesopor. Mater. 83
(2005) 345–356.
[18] B.Z. Zhan, B. Moden, J. Dakka, J.G. Santiesteban, E. Iglesia, J. Catal. 245 (2007)
314–323.
[19] S. Ernst, J. Weitkamp, Chem. Ing. Tech. 63 (1991) 748–750.
[20] D. Chen, H.P. Rebo, K. Moljord, A. Holmen, Ind. Eng. Chem. Res. 38 (1999)
4241–4249.
[26] H.S. Cerqueira, P. Ayrault, J. Datka, M. Guisnet, Micropor. Mesopor. Mater. 38
(2000) 197–205.
[27] J.W. Park, S.J. Kim, M. Seo, S.Y. Kim, Y. Sugi, G. Seo, Appl. Catal A: Gen. 349 (2008)
76–85.
[28] S. van Donk, E. Bus, A. Broersma, J.H. Bitter, K.P. de Jong, Appl. Catal. A: Gen. 237
(2002) 149–159.
[29] M.J. den Exter, J.C. Jansen, H. van Bekkum, Zeolites and Related Microporous
Materials: State of the Art, 1994, pp. 1159–1166.
[30] M.J. den Exter, J.C. Jansen, H. van Bekkum, A. Zikanova, Zeolites 19 (1997)
353–358.
[31] B.P.C. Hereijgers, F. Bleken, M.H. Nilsen, S. Svelle, K.-P. Lillerud, M. Björgen, B.M.
Weckhuysen, U. Olsbye, J. Catal. 264 (2009) 77–87.
[32] D. Chen, K. Moljord, T. Fuglerud, A. Holmen, Micropor. Mesopor. Mater. 29
(1999) 191–203.
[33] I.M. Dahl, R. Wendelbo, A. Andersen, D. Akporiaye, H. Mostad, T. Fuglerud,
Micropor. Mesopor. Mater. 29 (1999) 159–171.
[34] R. Krishna, J.M. van Baten, J. Phys. Chem. B 109 (2005) 6386–6396.
[35] R. Krishna, J.M. van Baten, Chem. Phys. Lett. 446 (2007) 344–349.
[36] Weiguo Song, James F. Haw, Angew. Chem. 115 (2003) 920–922.
[37] J. Gascon, C. Tellez, J. Herguido, A. Menendez, Chem. Eng. J. 106 (2005)
91–96.
[21] J.W. Park, G. Seo, Appl. Catal. A: Gen. 356 (2009) 180–188.
[22] L. Palumbo, F. Bonino, P. Beato, M. Björgen, A. Zecchina, S. Bordiga, J. Phys. Chem.
C 112 (2008) 9710–9716.
[23] H.G. Karge, W. Niessen, H. Bludau, Appl. Catal A: Gen. 146 (1996) 339–349.
[24] D. Meloni, D. Martin, P. Ayrault, M. Guisnet, Catal. Lett. 71 (2001) 213–217.
[38] J. Gascon, C. Tellez, J. Herguido, M. Menendez, Ind. Eng. Chem. Res. 44 (2005)
8945–8951.