Coking behavior of a submicron MFI catalyst during ethanol dehydration to ethylene in a pilot-scale fixed-bed reactor

Article abstract of DOI:10.1016/j.apcata.2010.11.036

Coking behavior of a submicron MFI catalyst for ethanol to ethylene was studied in a pilot scale tubular reactor. The experimental results showed that the submicron MFI catalysts possessed high activity and stability. Ethanol conversion and ethylene selectivity over the fresh/regenerated catalyst were around 97.9%/96.2% and 99.1%/98.9%, respectively. The fresh catalyst illustrated an active duration of 1606 h while the regenerated of 2008 h, which implied the regeneration can enhance the stability. In the reactor, the coke mainly consisted of adsorbed substances at the forepart and of alkyl aromatics at the end, and the coke contents presented an approximate U-shape trend along the tube length. Further tests revealed that the loss of acidity rather than the pore volume should be responsible for the catalyst deactivation.

Full text of DOI:10.1016/j.apcata.2010.11.036

Applied Catalysis A: General 393 (2011) 161–170
Contents lists available at ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
Coking behavior of a submicron MFI catalyst during ethanol dehydration to
ethylene in a pilot-scale fixed-bed reactor
Fei Wanga, Man Luob, Wende Xiaob,∗, Xiaowei Chengc, Yingcai Longc
a State Key Laboratory of Chemical Engineering, East China University of Science and Technology, Shanghai 200237, China
b School of Chemistry and Chemical Engineering, Shanghai Jiao Tong University, Shanghai 200240, China
c Shanghai Key Laboratory of Molecular Catalysis and Innovative Materials, Department of Chemistry, Fudan University, Shanghai 200433, China
a r t i c l e i n f o
a b s t r a c t
Article history:
Coking behavior of a submicron MFI catalyst for ethanol to ethylene was studied in a pilot scale tubular
reactor. The experimental results showed that the submicron MFI catalysts possessed high activity and
stability. Ethanol conversion and ethylene selectivity over the fresh/regenerated catalyst were around
97.9%/96.2% and 99.1%/98.9%, respectively. The fresh catalyst illustrated an active duration of 1606 h while
the regenerated of 2008 h, which implied the regeneration can enhance the stability. In the reactor, the
coke mainly consisted of adsorbed substances at the forepart and of alkyl aromatics at the end, and the
coke contents presented an approximate U-shape trend along the tube length. Further tests revealed that
the loss of acidity rather than the pore volume should be responsible for the catalyst deactivation.
© 2010 Elsevier B.V. All rights reserved.
Received 26 July 2010
Received in revised form
18 November 2010
Accepted 23 November 2010
Available online 27 November 2010
Keywords:
Ethanol dehydration
Submicron MFI zeolite
Coke deposition
Pilot plant
1. Introduction
zeolite catalysts can overcome the disadvantages of the alumina
catalysts with lower reaction temperature and higher activity, and
Increasing efforts for sustainable development have activated
researches on the alternative processes to the fossil resources
derived routes. Among them, bioprocesses based on the abundant
biomass have being considered as a feasible option for commodity
chemicals production. Recently, one can find that many contribu-
tions are devoted to bio-ethanol conversion to ethylene [1–5], H2
[6–8], ethylene oxide [9], olefins (propylene and butylene) [10–13],
and hydrocarbons [14,15]. The bio-ethylene, which can be directly
formed by catalytic dehydration of bio-ethanol, was considered as
one of the most promising biochemicals.
Generally, the bioethylene process consists of three steps:
(1) biomass fermentation to produce a dilute aqueous ethanol,
(2) distillation dehydration to purify the ethanol, and (3) cat-
alytic dehydration of ethanol to generate ethylene. Many previous
works were engaged in substituting foodstuff crops by agricultural
residues to reduce the raw material cost [16,17], adding adsorbent
into the conventional distillation process to save energy [18], and
replacing alumina with MFI zeolite to employ the hydrous rather
than the anhydrous ethanol for further energy saving [19,20].
The conventional alumina catalysts are low in activity, high in
reaction temperature and low in ethylene selectivity [21]. The MFI
especially is able to employ the hydrous ethanol, but is apt to
deactivation due to coke deposition. Some works focused on the
modification of MFI to improve its hydrothermal stability, anti-
coking ability and molecular diffusion ability by acid treatment
[22,23], hydrothermal dealumination [24], metal loading [4], phos-
phorus loading [25–28], and other reagent loading [12,13]. These
methods are post-synthesis treatments and may have limitations
of structure damage of the framework, acidity deterioration and
low reproducibility [29]. It has been found that the adjustment of
crystalline size during the synthesis of MFI zeolite is more effective,
which can improve the mesoporosity and modulate acid strength
distribution [30]. This method of MFI zeolite synthesis can achieve
a crystalline size in the submicron (100–1000 nm) rather than in
the nanometer level [31,32].
In the present work, the performance and coking behavior of a
submicron H-MFI catalyst for ethanol dehydration to ethylene in
a pilot plant were studied with emphasis on the activity, stability
and regenerability of the catalyst.
2. Experimental
2.1. Catalyst preparation
∗
The raw H-MFI powder of submicron zeolite crystallites (Shang-
hai Fuxu molecular sieve Co., Ltd.) with a SiO2/Al2O3 ratio (SAR)
Corresponding author. Tel.: +86 21 3420 3788; fax: +86 21 3420 3788.
E-mail address: wdxiao@sjtu.edu.cn (W. Xiao).
0926-860X/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.apcata.2010.11.036

162
F. Wang et al. / Applied Catalysis A: General 393 (2011) 161–170
2.2. Pilot plant set-up
The purpose of the pilot-scale reactor was basically to ver-
ify the feasibility of the submicron catalyst and to demonstrate
catalytic performance and operability for a projected commercial-
scale plant. The design basis for the pilot reactor was a simplified
version of the commercial one. The diagram of the pilot plant is dis-
played in Fig. 2. The reactor tube is a 32 mm ID × 3 m stainless steel
tube. An iron-constantan thermocouple of 1.5 mm OD is inserted
in a 3 mm OD stainless steel tube, which is fixed in the centre of
the reactor to measure the temperature of the catalyst bed. The
thermocouple can be moved up and down manually to determine
the axial temperature distribution in the reactor. Six electrother-
mal wires were surrounded the outside of the reactor tube with six
corresponding thermocouples (whose sensors contact the outside
of the tube) connected for the automatic temperature control. Once
the conversion of ethanol showed an evident decline (or lower than
96%), the controlled temperatures outside the reactor tube would
be raised by 2–5 ◦C synchronously.
Fig. 3 shows the core temperature distributions within the reac-
tor during blank test with ceramic rings as flow distributor. One can
see that the fixed-bed reactor showed a good isothermal behav-
ior under different temperatures. The core temperatures along the
tube length were approximately identical except the two ends.
Nevertheless, the average core temperature is about 20 ◦C lower
than the controlled temperatures outside the reactor tube. The
blank test also showed no conversion of ethanol.
The catalysts were packed in the tubular reactor with ceramic
rings (around 100 g) at each end as shown in Fig. 2. Before con-
tacting with the reactant, the catalyst bed was treated with steam
at 230 ◦C for 12 h. Hydrous ethanol was fed into a preheater
(32 mm ID × 2 m, filled with ceramic rings) by a measuring pump
(2PB80C, Beijing Satellite Manufacturing Facility) and the gaseous
mixture went through the catalyst bed from the top of the reac-
tor. The product mixture was cooled down by a water condenser
and was separated by a gas–liquid separator. Then the gas phase
products were analyzed on-line by a GC (Gas Chromatography, GC-
7890, Shanghai Techcomp Ltd.) equipped with a packed column
(Porapak-Q, 2 mm ID × 3 m) connected to a thermal conductivity
detector. Meanwhile, the flow rate of the gas phase products was
measured by a high-precision rotameter. The liquid products were
analyzed by the same GC off-line.
After the first run, the catalyst bed was cooled down to room
temperature under pure nitrogen (200 L/h). Then the catalysts were
unloaded and sampled (5 g/150 g) sequentially. The rest catalysts
were sequentiallyrepackedin the reactoraccording to their original
positions. Then pure N2 and O2 were employed to regenerate the
catalysts. Furthermore, the catalysts were unloaded and sampled
in the same way after the second run.
Fig. 1. SEM image (a) and pore distribution (b) of fresh submicron catalyst.
of 30 and a crystal size of 300–800 nm was mixed with the
silica sol binder at a binder content of 20 wt.%. Then the mix-
ture was extruded into cylindrical pellets 2.8–3.2 mm in diameter
and 5–10 mm in length. The pellets were then hydrothermally
treated in an autoclave under an atmosphere of mixed vapor
of organic amine template and water at 180 ◦C and 1.1 MPa for
36 h according to the procedure as described [33]. During the
hydrothermal treatment, the binder in the pellets was transformed
to MFI, and the so-called binder-free zeolite catalyst was synthe-
sized. The transformation can increase the relative crystallinity
of the zeolite by about 25%, the specific surface area by 75% and
pore volume by 18%, while remain its XRD patterns [33]. The
further post-synthesis treatments before packed in the reactor
include the ion-exchange with dilute hydrochloric acid for fur-
ther de-sodium, drying at 120 ◦C, and calcination at 550 ◦C for
3 h.
Fig. 1 shows the SEM image and the pore distribution of fresh
catalyst. It can be seen from Fig. 1a that the crystal size of sub-
micron MFI catalyst is in the range of 300–1000 nm. The white
particles in the image are the crystallized binders. The fresh cat-
alyst exhibits a mesopore distribution centred at 3.8 nm (Fig. 1b).
The adsorption and desorption isothermals (Fig. 1b, inset) also con-
firmed the existence of mesopores. Besides micropores (pore size
<2 nm) and mesopores (2 nm < pore size < 50 nm), the macropores
(pore size >50 nm) are only in a tiny minority. The micropore vol-
ume and the total pore volume of fresh catalyst are 0.103 and
0.228 cm3/g, respectively, indicating that the mesopores account
for a large proportion of the total.
Ethanol conversion and ethylene selectivity were calculated as
follows:
FEt,in − FEt,out
XEt
=
(1)
(2)
FEt,in
FE,out
SE
=
FEt,in − FEt,out
where FEt,in, FEt,out and FE,out are molar flow rates of injected ethanol,
unreacted ethanol and ethylene.
2.3. Catalyst characterization
2.3.1. Thermal gravimetric analysis
Thermal gravimetric analysis (TGA) was carried out on a
Q5000IR instrument (TA instruments, USA) based on a previous
method [34]. The temperature was raised from 40 to 200 ◦C and

F. Wang et al. / Applied Catalysis A: General 393 (2011) 161–170
163
Fig. 2. Diagram of the pilot plant. NV, needle value; R, rotameter; TIC, temperature indicating controller; TI, temperature indicator; PI, pressure indicator.
200 to 550 ◦C under N2 dynamic flow at a heating rate of 20 ◦C/min.
The temperature was hold for 1 min at 200 and 550 ◦C, respec-
tively. Thereafter, the temperature was raised from 550 to 800 ◦C
under air dynamic flow at the same heating rate. The weight loss in
200–550 ◦C was attributed to the desorption of light coke, while the
weight loss in 550–800 ◦C was attributed to the burning of heavy
coke. Thus, the coke contents of the sample can be calculated from
the following formulas:
where m200, m550, and mf denote the weight at 200 ◦C, the weight
at 550 ◦C and the final residual weight. The procedure of calculating
light coke/heavy coke is illustrated in Fig. 4.
2.3.2. Nitrogen adsorption–desorption technique
The pore volumes were measured by N2 adsorption–desorption
isotherms at −196 ◦C using an ASAP 2000 instrument (Micromerit-
ics Instrument Corporation, USA). Prior to the measurements, all
samples were degassed at 150 ◦C until a stable vacuum of 1 mT was
reached. Since the outer surface area only account for very small
part (0.07%) of total surface area, it is assumed that all the coke was
deposited in the pores of the catalyst. Therefore, the average density
(apparent density) of coke could be calculated from the following
m200 − m550
Light coke content =
Heavy coke content =
(3)
(4)
mf
m550 − mf
mf
Fig. 3. Axial temperature distributions in the reactor during blank test with ceramic
Fig. 4. Light coke and heavy coke of a coked sample during thermogravimetric
rings.
analysis.

164
F. Wang et al. / Applied Catalysis A: General 393 (2011) 161–170
formula:
coke content
VP0 − VPC
Average density =
(5)
where VP0 denotes the pore volume of fresh catalyst, VPC the pore
volume of spent catalyst (based on pure catalyst excluding coke).
2.3.3. X-ray diffraction
XRD patterns of the samples were recorded by a D/max-2200/PC
(Rigaku, Japan) diffractometer using CuK␣ radiation at 40 kV and
20 mA with a scanning speed of 6◦/min in the 2ꢀ range of 5–35◦.
2.3.4. Nuclear magnetic resonance
Solid state CP/MAS NMR (cross polarization/magic angle spin-
ning nuclear magnetic resonance) spectra were obtained at room
temperature on a Bruker MSL-300 NMR instrument. 27Al NMR spec-
tra were obtained at a resonance frequency of 78.205 MHz with a
0.5 s recycle time and 3.0 kHz spinning rate. 27Al chemical shifts
were referenced to AlCl3·6H2O. 13C NMR spectra were recorded at
a resonance frequency of 78.486 MHz with a 2 s recycle time, 2 ms
contact time, and 4.0 kHz spinning rate. 13C chemical shifts were
referenced to tetramethylsilane.
Fig. 6. Axial temperature distribution in the reactor over fresh catalyst.
10 ◦C lower than the controlled temperatures outside the reac-
tor tube. The lower temperature difference may be attributed to
the low voidage of the catalysts compared with the ceramic rings.
There was a low temperature region in the middle of the reactor,
which was in accordance with the endothermic reaction of ethanol
dehydration to ethylene.
The activity of fresh catalyst was maintained for about 1606 h
at temperatures lower than 290 ◦C as shown in Fig. 7a. During the
stable period 40–1607 h, ethanol conversion and ethylene selectiv-
ity were around 97.9% and 99.1%, respectively. The main byproduct
was DEE, while propylene and butylene were negligible. After the
pilot plant was shut down at 1607 h due to an appliance’s problem,
ethanol conversion and ethylene selectivity decreased dramatically
2.3.5. Fourier transform infrared spectroscopy
The chemical nature of surface acid sites on fresh and spent
catalysts was determined on a Nexus-470 (Nicolet, USA) spec-
trophotometer using pyridine as probe molecule. Sample wafers
were formed by pressing 10–20 mg of the catalyst at 5 tonnes/cm2
and transferred to a sample holder made of quartz. Sample wafers
were evacuated 4 h at 300 ◦C and then the catalyst spectrum was
recorded after cooling the sample to room temperature. After that,
the adsorption of pyridine at room temperature was conducted,
followed by sequential evacuation at 25, 100, 150, 200, 250, and
300 ◦C. The spectra of the adsorbed species were obtained by sub-
tracting the catalyst spectrum before pyridine adsorption.
3. Results and discussion
3.1. Performance of the fresh catalyst
The evolution of ethanol conversion and product distribution as
a function of space-time are shown in Fig. 5. Ethanol conversion
and ethylene selectivity increased while diethyl ether (DEE) selec-
tivity declined with increasing space-time. The pilot plant should be
operated at higher space-time to obtain higher ethylene selectivity.
Fig. 6 shows the core temperature distribution in the reactor
during ethanol dehydration. The average temperature was about
Fig. 7. (a) Ethanol conversion and product distribution and (b) reaction temperature
and pressure drop over fresh catalyst. Reaction conditions: catalyst, 1505 g; ethanol
concentration, 95% (v/v); WHSV, 1.25 h−1. The temperature was measured at 1 m
form the top of the reactor.
Fig. 5. Evolution of ethanol conversion and product distribution as a function of
space time.

F. Wang et al. / Applied Catalysis A: General 393 (2011) 161–170
165
Fig. 8. CO2 flow rate and regeneration conditions as a function of time.
from 98.6% and 99.3% to 75.6% and 77.5%, respectively. After this
stop, pure nitrogen (200 L/h) was introduced attempting to protect
the catalyst from deactivation. However, this operation was proved
ineffective.
The pilot plant was shut down for other two times at 625 h
(249 ◦C) and 1389 h (271 ◦C) due to power failures. The reactor was
naturally cooled down without any sweeping after these two shut-
downs. Once the power was restored, the reactor was heated to the
previous temperature and the feed was introduced into the reac-
tor as before. The first shutdown caused little apparent effect on
the dehydration while the second shutdown caused an increase in
the rate of temperature increase. Compared with the third shut-
down as mentioned above, these two shutdowns had no apparent
effects on catalyst activity due to the lower temperatures and, more
importantly, to no purge gas. This deactivation may be attributed
to the evolution of coke precursors during the nitrogen aging. The
coke precursors may undergo pyrolysis or physical stripping under
nitrogen aging [35]. Gaseous olefins or paraffins were formed and
the H/C ratio of the coke was reduced, leading to hard coke and
causing evident loss in catalyst activity.
Fig. 9. (a) Ethanol conversion and product distribution and (b) reaction temperature
and pressure drop over regenerated catalyst. Reaction conditions: catalyst, 1468 g;
ethanol concentration, 95% (v/v); WHSV, 1.25 h−1. The temperature was measured
as before.
maintained around 96.2% and 98.9%, respectively. Meanwhile, the
reaction temperature was gradually raised from 245 to 290 ◦C. The
temperature rise in 2000 h was 45 ◦C. Unfortunately, the apparatus
has to be shut down due to a leak in the water condenser at about
2008 h. In comparison to fresh catalyst, the regenerated catalyst
exhibits a little lower ethanol conversion, which indicated a slight
loss of catalyst activity. On the other hand, the regenerated catalyst
showed extraordinary stability. Overall, the excellent performance
of the regenerated catalyst implied a successful in situ regeneration.
The pressure drop of the reactor increased slowly with time on
stream during the second run. Compared with the pressure drop
in the first run, the pressure drop in the second run was a lit-
tle higher, which might be related to a small quantity of catalyst
powder caused by the regeneration process.
The pressure drop of the reactor increased with time on stream
especially after 1607 h as shown in Fig. 7b. This result might be
related to the deposition of coke. The coke deposits covered the
surface of the catalyst and blocked the pores of the catalyst thus
caused the increase of pressure drop.
3.2. Regeneration of the first-cycle used catalyst
During the regeneration process, N2/O2 mixture was preheated
and went through from the top of the tubular reactor. The coke
deposits were burning off gradually and the outlet gas mixture was
analyzed to monitor the CO2 content. At the beginning, the catalyst
bed was aged in nitrogen flow for 5 h at 300 ◦C. The temperature
was raised from 300 to 550 ◦C and the oxygen content in the N2/O2
mixture was raised from 0 to 17.1% gradually as shown in Fig. 8.
During the last 6 h (27–33 h), the flow rate of nitrogen was reduced
from 400 L/h to 300 L/h.
It can be seen from Fig. 8 that the flow rate of outlet CO2
decreased over time at constant temperature, and it temporarily
rose with the increasing temperature or oxygen content. A great
part of the coke deposits were burnt at temperatures lower than
550 ◦C. Eventually, only a little CO2 was produced at 550 ◦C, imply-
ing the end of regeneration.
3.4. Properties of catalyst and coke deposit
3.4.1. Coke content
Fig. 10 exhibits that the coke contents of spent catalysts after
the second run were much lower, which indicates that the coke
deposition was alleviated over the regenerated catalyst. Along the
tube length of the reactor (from the top), the heavy coke content
increased after each run, particularly in the second half of the reac-
tor; while the light coke content decreased a little in the first half of
the reactor and then kept approximate constant in the second half.
Considering the component distribution in the reactor, it seems
that the heavy coke was attributed to olefin polymers and the light
coke was attributed to dehydration intermediates. Along the tube
length of the reactor, ethanol concentration decreased while ethy-
lene concentration increased gradually. The oligomerization rate
increased with increasing ethylene concentration, thus caused the
3.3. Performance of the regenerated catalyst
It can be seen from Fig. 9 that the regenerated catalyst performed
excellent stability in dehydration of ethanol to ethylene. After the
initial period, the ethanol conversion and ethylene selectivity were

166
F. Wang et al. / Applied Catalysis A: General 393 (2011) 161–170
Fig. 11. Pore volumes of spent catalysts: (a) after the first run and (b) after the
Fig. 10. Coke contents of spent catalysts: (a) after the first run and (b) after the
second run.
second run.
spent catalysts in 2–3 m, the two peaks (2ꢀ = 23.0◦, 23.3◦) merged
into one peak at 2ꢀ = 23.1◦. The absence of the peak and the change
of crystal parameters indicate that the catalyst has transformed
from orthorhombic (lattice parameters a =/ b) to tetragonal (a = b)
during ethanol dehydration. Lee et al. [36] and Parise et al. [37]
found similar phase transformations. The former suggested that
the orthorhombic phase partly transformed into tetragonal phase
upon NO uptake. The latter concluded that the apparent symme-
try of MFI changes to tetragonal upon adsorption of two molecules
of stilbene per unit cell, but the true symmetry is orthorhombic
with a very small difference between a and b (lattice constant).
However, Vasiliev et al. [38] and Tuan et al. [39] attributed the
diffracted intensity changes to the mutual transformation between
orthorhombic phase and monoclinic phase.
increase in heavy coke content. Ethanol dehydration occurred in
the whole reactor (ethanol conversion <100%), thus caused a broad
distribution of light coke. Furthermore, the overall coke contents
show an approximate U-shape trend along the tube length. The
lower total coke content in the middle of the reactor may imply that
the main part of coke deposits has transferred from dehydration
intermediates to olefin polymers.
3.4.2. Catalyst porosity
It can be seen from Fig. 11 that more than 70% of the pores
(micropores and mesopores) were maintained except the two ends
of the reactor after each run. The micropore volumes and mesopore
volumes changed simultaneously in the same trend. In comparison
to the catalysts after the first run, the catalysts after the second run
maintained more pores, which were in consist with the lower coke
contents and the higher catalyst activity at the end of this run.
Fig. 12 reveals that the average densities of coke become much
bigger after the second run. This result means that the coke deposits
were much condensed during the second run, which may be
attributed to a lower temperature and a longer time on stream.
Moreover, the average densities changed in the same way as the
heavy coke contents.
3.4.3. Catalyst crystal morphology
Fig. 13 exhibits the XRD patterns of fresh, spent and regenerated
catalysts. XRD patterns of all the catalysts are shown the MFI crystal
phase. However, detailed observation shows that there are differ-
ences in the peak intensities in the 2ꢀ range of 22–25◦. Along the
tube length of the reactor, in the XRD patterns of the spent catalysts,
the second peak at 2ꢀ = 23.3◦ gradually approached to the first peak
at 2ꢀ = 23.0◦ before 2 m. Meanwhile, the diffracted intensities at
2ꢀ = 7.9◦ and 8.8◦ kept increasing. Finally, in the XRD patterns of the
Fig. 12. Average densities of coke deposits after each run.

F. Wang et al. / Applied Catalysis A: General 393 (2011) 161–170
167
ated to pentacoordinated aluminum [40] or distorted tetrahedral
aluminum [41].
The contents of non-framework aluminum (NF Al) were calcu-
lated based on 27Al MAS NMR analysis. Meanwhile, the framework
SiO2/Al2O3 ratio (SAR) was obtained from the element analysis
(Bruker-AXS S4 Explorer, Germany). Fig. 14c and d reveals that
the catalyst dealumination was severer at the end of the reactor.
Since the tetrahedral aluminum is mainly responsible for the sur-
face acidity of zeolite [27], the catalyst at the end of the reactor
suffered more loss of activity. Moreover, the catalyst dealumi-
nation after the second run was much more serious with lower
NF Al contents and higher framework SAR. Since the dehydra-
tion of ethanol occurred at a lower temperature (no more than
350 ◦C), the serious dealumination might happen in the regenera-
tion process (up to 550 ◦C). Furthermore, the dealumination effect
should be partly responsible for the activity loss of the regenerated
catalyst.
3.4.5. Chemical nature of coke deposit
The 13C NMR spectra of four samples after the first run and
the second run are shown in Fig. 15a and b. Approximate peak
assignments for the spectra are as follows: alkyl, 0–50 ppm; O-
alkyl, 50–90 ppm; aromatic or olefinic C C, 90–170 ppm; carbonyl
C
O (ketones, quinones, and aldehydes), 170–230 ppm [42–47].
Moreover, the carbon species were relatively quantified by the
integrated areas of the corresponding regions in 13C NMR profiles.
In this case, the content of unsaturated carbon of coke or aro-
maticity [42] was defined as [the integrated area of unsaturated
carbon]/[total integrated area].
It can be seen from Fig. 15a that most bands are located in low
chemical shift region 0–70 ppm at the forepart of the reactor. The
bands at 17.3 and 58.9 ppm are very close to those bands of ethanol
(18.1 and 57.8 ppm), indicating that these bands may be attributed
to adsorbed ethanol. The bands at 32.0 and 177.0 are close to those
bands of acetaldehyde (30.9 and 199.9 ppm), implying the possible
presence of acetaldehyde in the catalysts. The rarity of signals above
70 ppm indicates that contributions from unsaturated carbons are
negligible. Along the tube length, the intensities of the bands in high
chemical shift region (above 50 ppm) increased gradually while
the intensities of acetaldehyde bands disappeared gradually. The
spectra of the catalysts at 1.9 m and 2.7 m are similar to each other
except for the differences in intensity. These two spectra indicated
that the coke deposits consisted of alkyl carbon, O-alkyl carbon,
aromatic or olefinic carbon, and carbonyl carbon.
Fig. 13. XRD patterns of catalyst: (a) fresh catalyst and spent catalysts after first run
and (b) regenerated catalyst and spent catalysts after the second run.
The evolutions of the spectra after the second run (Fig. 15b) are
similar to those after the first run. However, the bands located in
50–90 ppm (O-alkyl) and 170–230 ppm (carbonyl C O) apparently
decreased or even disappeared from the spectra of the catalysts
after the second run. This phenomenon indicated that oxygenated
chemicals were negligible in the coke. The coke deposits might
change from adsorbed ethanol to alkyl aromatics along the tube
length.
Fig. 15c and d shows that the unsaturated carbon content
increased along the tube length after each run. These trends imply
that the coke deposits become more unsaturated along the tube
length, which are in agreement with TG results. The coke deposits
after the second run are a bit more unsaturated especially at the end
ofthe reactor. Nevertheless, the biggest unsaturatedcarbon content
of the coke was less than 60%, indicating that the coke deposits were
relatively light even after 2008 h on stream.
The XRD pattern of regenerated catalyst is very similar to that
of fresh catalyst as shown in Fig. 13b. It seems that the regen-
erated catalysts transformed back to orthorhombic phase after
regeneration. This phase transformation may be attributed to coke
deposition. The coke deposits may replace the hydrogen in hydrox-
yls and thus cause the change in the length of crystal axis. The
hydroxyls would be recovered by regeneration and the catalyst
would transform back to orthorhombic phase. Nevertheless, the
crystal system transformation may have no direct effect on catalyst
activity; still it may be an indicator of catalyst deactivation.
3.4.4. Status of aluminum in catalyst
It can be seen from the 27Al MAS NMR spectra (Fig. 14a and
b) that most of the aluminum atoms were tetrahedral framework
aluminum (51–53 ppm). However, along the tube length, the inten-
sities of tetrahedral framework aluminum decreased a little and
the peak shifted to low chemical shift region. At the same time,
the octahedral aluminum species (∼0 ppm) in an extra framework
position increased gradually. It seems that the catalyst dealumina-
tion became much severer along the tube length. On the other hand,
all spectra show no bands at ∼30 ppm that are commonly associ-
3.4.6. Catalyst acidity
The FT-IR spectra of pyridine adsorbed on fresh, regener-
ated and spent catalysts after evacuation at 150 ◦C in the region
1400–1600 cm−1 are shown in Fig. 16. There are three sharp bands
on fresh catalyst (Fig. 16a) due to the C–C stretching vibrations of

168
F. Wang et al. / Applied Catalysis A: General 393 (2011) 161–170
Fig. 14. Analyses of the status of aluminum in the catalysts. (a) 27Al NMR profiles of the catalysts after the first run, (b) 27Al NMR profiles of the catalysts after the second
run, (c) contents of non-framework (NF) Al of the catalysts, and (d) framework SiO2/Al2O3 of the catalysts.
pyridine [48]. The strong band at 1491 cm−1 is due to the pyri-
dinium ion formed on very strong Brönsted or Lewis acid sites [49],
while bands at 1540 and 1448 cm−1 are due to protonation of the
pyridine molecule by Brönsted acid sites and pyridine coordinated
on Lewis acid sites [50], respectively. The B/L (Brönsted/Lewis) acid
site ratio (calculated from the ratio of integrated areas at 1540 and
1448 cm−1 with an extinction coefficient ratio of 1.5 [51]) was 1.2.
The B/L on the fresh catalyst increased from 1.2 to 2.6 as the evacua-
Fig. 15. 13C NMR analyses of spent catalysts. (a) 13C NMR profiles of the catalysts after the first run, (b) 13C NMR profiles of the catalysts after the second run, (c) carbon
distributions of the catalysts after the first run, and (d) carbon distributions of the catalysts after the second run. C–, saturated carbon; C , unsaturated carbon.

F. Wang et al. / Applied Catalysis A: General 393 (2011) 161–170
169
period, respectively. In situ regeneration took about 33 h after
a severe deactivation due to inappropriate nitrogen purging.
The regenerated catalyst exhibited excellent performance with
an ethanol conversion of 96.2% and an ethylene selectivity of
98.9%. The temperature rise was only 45 ◦C in an active duration
of 2008 h, which implied the regeneration could enhance the
catalyst stability.
(2) In the reactor, the coke mainly consisted of adsorbed substances
at the forepart and of alkyl aromatics at the end, and the coke
contents profile presented an approximate U-shape trend along
the tube length. The coke deposits in the second run were much
more condensed and unsaturated.
(3) Possibly, due to a high content of mesopore, the submicron
MFI catalyst retained more than 70% of the pores after each
run. The catalyst has undergone reversible phase transforma-
tion from orthorhombic to tetragonal upon coke deposition.
Along the tube length, the catalyst suffered severer dealumi-
nation. The loss of acidity rather than the pore volume should
be responsible for the catalyst deactivation in the pilot plant.
References
[1] H. Chiang, A. Bhan, J. Catal. 271 (2010) 251–261.
[2] I. Takahara, M. Saito, M. Inaba, K. Murata, Catal. Lett. 105 (2005) 249–252.
[3] K. Inoue, M. Inaba, I. Takahara, K. Murata, Catal. Lett. 136 (2010) 14–19.
[4] J. Ouyang, F. Kong, G. Su, Y. Hu, Q. Song, Catal. Lett. 132 (2009) 64–74.
[5] D. Varisli, T. Dogu, G. Dogu, Chem. Eng. Sci. 65 (2010) 153–159.
[6] A. Kazama, Y. Sekine, K. Oyama, M. Matsukata, E. Kikuchi, Appl. Catal. A: Gen.
(2010), doi:10.1016/j.apcata.2010.05.031.
[7] S. Mostafa, J.R. Croy, H. Heinrich, B.R. Cuenya, Appl. Catal. A: Gen. 366 (2009)
353–362.
[8] H. Song, B. Mirkelamoglu, U.S. Ozkan, Appl. Catal. A: Gen. 382 (2010) 58–64.
[9] M.J.
Lippits,
B.E.
Nieuwenhuys,
Catal.
Today
(2010),
doi:10.1016/j.cattod.2010.03.019.
[10] Z.X. Song, A. Takahashi, I. Nakamura, T. Fujitani, Appl. Catal. A: Gen. (2010),
doi:10.1016/j.apcata.2010.06.035.
[11] Z.X. Song, A. Takahashi, N. Mimura, T. Fujitani, Catal. Lett. 131 (2009) 364–369.
[12] A.G. Gayubo, A. Alonso, B. Valle, A.T. Aguayo, J. Bilbao, Ind. Eng. Chem. Res.
(2010), doi:10.1021/ie100407d.
[13] A.G. Gayubo, A. Alonso, B. Valle, A.T. Aguayo, J. Bilbao, App. Catal. B: Environ.
97 (2010) 299–306.
Fig. 16. FT-IR spectra of pyridine adsorbed over catalysts after evacuation at 150 ◦C:
(a) fresh catalyst and spent catalysts after first run and (b) regenerated catalyst and
spent catalysts after the second run.
[14] A.T. Aguayo, A.G. Gayubo, A. Atutxa, M. Olazar, J. Bilbao, Ind. Eng. Chem. Res. 41
(2002) 4216–4224.
[15] A.K. Talukdar, K.G. Bhattacharyya, S. Sivasanker, Appl. Catal. A: Gen. 148 (1997)
357–371.
[16] B. Hahn-Hägerdal, M. Galbe, M.F. Gorwa-Grauslund, G. Lidén, G. Zacchi, Trends
Biotechnol. 24 (2006) 549–556.
[17] Y. Chen, R.R. Sharma-Shivappa, D. Keshwani, C. Chen, Appl. Biochem. Biotech-
nol. 142 (2007) 276–290.
tion temperature increased from 150 to 300 ◦C, thereby indicating
that Brönsted acid sites are more strongly acidic than the Lewis
acid sites and retain pyridine up to higher evacuation tempera-
tures.
After each run, the spent catalyst at 0.3 m only exhibits a weak
band at ∼1440 cm−1 and a negligible band at ∼1540 cm−1, while
the spent catalyst at 2.7 m hardly exhibits any evident bands in
these regions. It is obvious that the spent catalyst at 0.3 m has only
a few Lewis acid sites and no Brönsted acid sites left, while the
spent catalyst at 2.7 m has almost no acid sites left. It seems that
the coke deposits take responsible for the acidity loss and the coke
deposits deplete Brönsted acid sites preferentially. Fig. 16b reveals
that a majority of acidity was recovered by the in situ regeneration.
More than 75%/60% of acid sites were regained for the catalysts at
0.3 m/2.7 m, implying the catalysts at the end of the reactor were
more difficult to be regenerated. In a word, based on the intensities
of all the bands present in the spectra of samples, it suggests that
the acidity of spent catalyst at 0.3 m was a bit stronger than that
of 2.7 m after each run, but all spent catalysts exhibited extensive
weaker acidities than fresh catalyst, which was in accordance with
the activity of the catalysts.
[18] E. Lalik, R. Mirek, J. Rakoczy, A. Groszek, Catal. Today 114 (2006) 242–247.
[19] R. Le Van Mao, T.M. Nguyen, J. Yao, Appl. Catal. 61 (1990) 161–173.
[20] X. Zhang, R. Wang, X. Yang, F. Zhang, Micropor. Mesopor. Mater. 116 (2008)
210–215.
[21] H. Huang, in: G.Q. Chen (Ed.), Plastic from Bacteria, Springer, Berlin/Heidelberg,
2010, pp. 390–404.
[22] N. Danilina, F. Krumeich, S.A. Castelanelli, J.A. van Bokhoven, J. Phys. Chem. C
114 (2010) 6640–6645.
[23] R. Caicedo-Realpe, J. Pérez-Ramírez, Micropor. Mesopor. Mater. 128 (2010)
91–100.
[24] L.J. Jin, H.Q. Hu, S.W. Zhu, B. Ma, Catal. Today 149 (2010) 207–211.
[25] N. Zhan, Y. Hu, H. Li, D. Yu, Y. Han, H. Huang, Catal. Commun. 11 (2010) 633–637.
[26] D. Zhang, R. Wang, X. Yang, Catal. Lett. 124 (2008) 384–391.
[27] K. Ramesh, L.M. Hui, Y.F. Han, A. Borgna, Catal. Commun. 10 (2009) 567–571.
[28] K. Ramesh, C. Jie, Y. Han, A. Borgna, Ind. Eng. Chem. Res. 49 (2010) 4080–4090.
[29] N. Viswanadham, R. Kamble, M. Singh, M. Kumar, G.M. Dhar, Catal. Today 141
(2009) 182–186.
[30] C.S. Triantafillidis, N.P. Evmiridis, Ind. Eng. Chem. Res. 38 (1999) 916–927.
[31] N. Ren, Z.J. Yang, X.C. Lv, J. Shi, Y.H. Zhang, Y. Tang, Micropor. Mesopor. Mater.
131 (2010) 103–114.
[32] J. Bi, X. Guo, M. Liu, X. Wang, Catal. Today 149 (2010) 143–147.
[33] X.W. Cheng, J. Wang, J. Guo, Y.C. Long, Chin. J. Chem. 28 (2010) 183–188.
[34] B. Wang, G. Manos, J. Catal. 250 (2007) 121–127.
[35] A.T. Aguayo, A.G. Gayubo, J. Erea, A. Atutxa, J. Bilbao, Ind. Eng. Chem. Res. 42
(2003) 3914–3921.
4. Conclusions
[36] K.Y. Lee, K. Watanabe, M. Misono, Appl. Catal. B: Environ. 13 (1997) 241–249.
[37] J.B. Parise, J.A. Hriljac, D.E. Cox, D.R. Corbin, V. Ramamurthy, Chem. Commun.
(1993) 226–228.
[38] P.O. Vasiliev, A. Ojuva, J. Grins, L. Bergström, J. Eur. Ceram. Soc. (2010),
doi:10.1016/j.jeurceramsoc.2010.02.007.
(1) The fresh submicron MFI zeolite catalyst showed high activity
and good stability in the pilot plant. Ethanol conversion and
ethylene selectivity were around 97.9% and 99.1% in the stable

170
F. Wang et al. / Applied Catalysis A: General 393 (2011) 161–170
[39] H.T. Tuan, I.K. Bae, Y.N. Jang, S.C. Chae, Y.B. Chae, D.S. Suhr, J. Ceram, Process.
Res. 11 (2010) 204–208.
[40] A. Martínez, M.A. Arribas, M. Derewinski, A. Burkat-Dulak, Appl. Catal. A: Gen.
379 (2010) 188–197.
[45] A. Hauser, A. Stanislaus, A. Marafi, A. Al-Adwani, Fuel 84 (2005) 259–269.
[46] A. Watanabe, N. Fujitake, Anal. Chim. Acta 618 (2008) 110–115.
[47] J.D. Mao, A. Schimmelmann, M. Mastalerz, P.G. Hatcher, Y. Li, Energy Fuel 24
(2010) 2536–2544.
[41] Q. Huo, T. Dou, Z. Zhao, H. Pan, Appl. Catal. A: Gen. 381 (2010) 101–108.
[42] B. Paweewan, P.J. Barrie, L.F. Gladden, Appl. Catal. A: Gen. 185 (1999) 259–
268.
[43] M.A. Callejas, M.T. Martinez, T. Blasco, E. Sastre, Appl. Catal. A: Gen. 218 (2001)
181–188.
[48] J. Bedia, R.R. Rosas, J.R. Mirasol, T. Cordero, J. Catal. 271 (2010) 33–42.
[49] C. Tagusagawa, A. Takagaki, A. Iguchi, K. Takanabe, J.N. Kondo, K. Ebitani, S.
Hayashi, T. Tatsumi, K. Domen, Angew. Chem. Int. Ed. 49 (2010) 1128–1132.
[50] C. Morterra, A. Chiorino, G. Ghiotti, E. Garrone, J. Chem. Soc., Faraday Trans. 1
75 (1979) 271–288.
[44] A. Feller, J. Barth, A. Guzman, I. Zuazo, J.A. Lercher, J. Catal. 220 (2003) 192–206.
[51] Z. Xie, Q. Chen, C. Zhang, J. Bao, Y. Cao, J. Phys. Chem. B 104 (2000) 2853–2859.