An investigation into cyclohexanone ammoximation over Ti-MWW in a continuous slurry reactor

Article abstract of DOI:10.1016/j.apcata.2010.10.037

In the present study, the liquid-phase ammoximation of cyclohexanone with ammonia and hydrogen peroxide was studied using a MWW-type titanosilicate (Ti-MWW) catalyst in a continuous slurry reactor to develop a clean process for producing cyclohexanone oxime. The reaction parameters, which governed the cyclohexanone conversion, oxime selectivity and catalyst deactivation, were investigated by simulating the operating conditions of an industrial process. Under optimized reaction conditions, Ti-MWW produced a cyclohexanone conversion and oxime selectivity over 96% and 99%, respectively. Moreover, Ti-MWW was extremely robust and showed a longer lifetime than the conventional titanium silicalite-1 catalyst. The causes of deactivation were elucidated to be the coke deposition and partial dissolution of the zeolite framework of Ti-MWW during ammoximation. The deactivated Ti-MWW catalyst was regenerated effectively by a combination of acid treatment and cyclic amine-assisted structural rearrangement.

Full text of DOI:10.1016/j.apcata.2010.10.037

Applied Catalysis A: General 394 (2011) 1–8
Contents lists available at ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
An investigation into cyclohexanone ammoximation over Ti-MWW in
a continuous slurry reactor
Song Zhao, Wei Xie, Junxia Yang, Yueming Liu, Yingtian Zhang, Biliang Xu, Jin-gang Jiang,
Mingyuan He, Peng Wu∗
Shanghai Key Laboratory of Green Chemistry and Chemical Processes, Department of Chemistry, East China Normal University,
North Zhongshan Rd. 3663, Shanghai 200062, PR China
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 16 July 2010
Received in revised form 8 October 2010
Accepted 17 October 2010
Available online 30 December 2010
In the present study, the liquid-phase ammoximation of cyclohexanone with ammonia and hydrogen per-
oxide was studied using a MWW-type titanosilicate (Ti-MWW) catalyst in a continuous slurry reactor to
develop a clean process for producing cyclohexanone oxime. The reaction parameters, which governed
the cyclohexanone conversion, oxime selectivity and catalyst deactivation, were investigated by simu-
lating the operating conditions of an industrial process. Under optimized reaction conditions, Ti-MWW
produced a cyclohexanone conversion and oxime selectivity over 96% and 99%, respectively. Moreover,
Ti-MWW was extremely robust and showed a longer lifetime than the conventional titanium silicalite-1
catalyst. The causes of deactivation were elucidated to be the coke deposition and partial dissolution of
the zeolite framework of Ti-MWW during ammoximation. The deactivated Ti-MWW catalyst was regen-
erated effectively by a combination of acid treatment and cyclic amine-assisted structural rearrangement.
Keywords:
Ti-MWW
Ammoximation
Cyclohexanone oxime
Slurry reactor
Regeneration
© 2010 Elsevier B.V. All rights reserved.
1. Introduction
because TS-1 is capable of catalyzing a variety of oxidation
reactions with aqueous hydrogen peroxide as an oxidant and
Cyclohexanone oxime is an important captive intermediate with
a constantly increasing presence on the world market because
it is readily transformed through the Beckmann rearrangement
to ␧-caprolactam (CPL), a core ingredient for the nylon-6 man-
ufacturing industry. Starting from benzene or toluene as a raw
material, the conventional manufacturing processes of CPL include
noncatalytic oximation of cyclohexanone with hydroxylamine and
liquid-phase Beckmann rearrangement of the oxime in oleum.
These processes are encountering serious disadvantages because
they require multiple reaction steps, use harmful oximation agents
such as hydroxylamine and sulfuric acid catalyst, and in particu-
lar, coproduce a large quantity of less valuable byproducts such
as ammonium sulfate or phosphate [1,2]. These drawbacks not
only bring about environmental pollution, but also decrease the
profitability of the CPL process. Hence, an environmentally benign,
material-saving and zero-emission CPL process, which avoids using
hazardous reagents and generating valueless ammonium salts, is
urgently required.
water as the sole byproduct. Enichem first reported that TS-1
is an efficient catalyst for the ammoximation of cyclohexanone
[2]. The direct route combines in situ formation of hydroxy-
lamine intermediate and subsequent oximation of cyclohexanone,
avoiding co-production of ammonium salts and reducing the
capital investment of establishing a CPL process. In 2003, Sum-
itomo Chemical Co. and Sinopec commercialized successfully
this innovative technology of cyclohexanone oxime processes
with the capability to upscale worldwide [3]. Moreover, Sum-
itomo has further combined the TS-1-catalyzed liquid-phase
ammoximation and the silicalite-1-catalyzed gas-phase Beck-
mann rearrangement, leading to an extremely clean route for
CPL production. These new processes exclude the dangerous
reagents and the co-production of massive valueless ammonium
salts.
The successes achieved with TS-1 have spurred intensive
research on new titanosilicate catalysts to obtain novel materials
which possess a better catalytic performance and are capable of
overcoming the drawbacks encountered with TS-1. Being a typical
of medium pore zeolite of 10-membered ring (MR), TS-1 suffers
constraints and mass transfer limitations. The manufacturing cost
of TS-1 is generally high because of using expensive tetrapropy-
lammonium hydroxide as a structure-directing agent. In addition,
a relatively high concentration of TS-1 catalyst is needed to achieve
a reasonable catalytic activity level. Moreover, a high performance
The discovery of titanium silicalite-1 (TS-1) opened up new
possibilities of developing zeolite-based green catalytic processes
∗
Corresponding authors. Tel.: +86 21 62232292; fax: +86 21 62232292.
E-mail addresses: ymliu@chem.ecnu.edu.cn (Y. Liu), pwu@chem.ecnu.edu.cn
(P. Wu).
0926-860X/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.apcata.2010.10.037

2
S. Zhao et al. / Applied Catalysis A: General 394 (2011) 1–8
TS-1 catalyst made up of small particles leads to difficult membrane
separation for a slurry-bed reactor.
2.2. Catalytic reactions
With great efforts made in past decades, a number of new
titanosilicates with different zeolite structures have been devel-
oped, i.e., Ti-MWW [4–6], TS-2 [7], Ti-MOR [8] and Ti-Beta [9]. In
particular, by possessing a pore system that consists of two inde-
pendent interlayer and intralayer 10-MR channels as well as 12-MR
supercages and side cups, Ti-MWW shows a unique solvent effect
and high catalytic activities for the epoxidation of various alkenes
with H2O2 [10–13]. We have previously investigated the catalytic
properties of Ti-MWW in the ammoximation of ketones using a
batch-type reactor [14–16]. Ti-MWW was capable of catalyzing the
ammoximation of cyclohexanone and methyl ethyl ketone in the
presence of water producing a ketone conversion and oxime selec-
tivity both as high as 99%. This implies that Ti-MWW is a promising
catalyst for the clean synthesis of oximes. Nevertheless, the pro-
duction of bulk chemicals like cyclohexanone oxime on a large
scale favors a continuous slurry-bed reactor, which is equipped
with a membrane separator to save the labor found in batchwise
processes.
In this study, by simulating the operating conditions in a
commercial processing procedure, we applied Ti-MWW to the
ammoximation of cyclohexanone in a continuous slurry reactor,
and investigated the effects of reaction parameters, deactivation
behavior and catalyst regeneration by comparing them with those
from conventional TS-1 under the same operating conditions. Ti-
MWW was expected to exhibit a higher catalytic activity than TS-1
because of its structural advantages. The present results lay a solid
foundation for the industrial use of Ti-MWW, and are also help-
ful for clarifying the chemical engineering issues involved in the
liquid-phase ammoximation of cyclohexanone.
The liquid-phase ammoximation of cyclohexanone was carried
out continuously in a 160 mL glass slurry reactor equipped with
a glass sand filter and a magnetic stirrer. For a typical reaction, a
desirable amount of catalyst powder (0.8–3.2 g) and 120 mL of tert-
butanol aqueous solution (85 wt%) were added in the reactor and
heated under stirring at a determined temperature (328–353 K).
The mixture of cyclohexanone and 85 wt% t-BuOH aqueous solu-
tion (weight ratio of 1:2.94) and 27 wt% H2O2 were then fed into
the reactor separately with a micro-pump. The feeding rate of the
mixture of cyclohexanone and t-BuOH was always kept constant at
96 mL h−1. Meanwhile, ammonia gas (99.9%) was charged into the
reaction system with a mass flowmeter at a constant rate. The molar
ratios of H2O2/ketone and NH3/ketone were varied in the range of
0.95–1.2 and 1.0–2.2, respectively. With the reaction proceeding,
the reaction mixture overflowed from the outlet filter and the cat-
alyst powder remained in the reactor. The ammonia unconverted
and not soluble in the reaction mixture was exhausted through a
condenser vent. The organic products were analyzed on a gas chro-
matograph (Shimadzu 14B) with a flame ionization detector and a
HP-5 capillary column to calculate the conversion of cyclohexanone
and the selectivity of oxime. The content of unconverted H2O2 was
determined by iodometric titration.
The used catalyst was gathered by filtration and dried at 353 K.
Three regeneration ways have been attempted on the deactivated
Ti-MWW: (i) directly calcination in air at 823 K for 10 h; (ii) first
calcined as method (i), then structural rearrangement in an auto-
clave with an aqueous solution of PI (6.5 wt%) at 443 K for 24 h, and
further calcined at 823 K for 10 h; (iii) first washed with 0.3 M HNO3
at 353 K for 4 h, and then regenerated as method (ii).
3. Results and discussion
2. Experimental
3.1. Characterization of titanosilicate catalysts
2.1. Catalyst preparation and characterization methods
The Si/Ti atomic ratio was 38 for Ti-MWW and 30 for TS-
1. The XRD patterns confirmed that both titanosilicates had the
expected crystalline structures (data not shown). In the UV–visible
and IR spectra, the samples showed characteristic adsorption bands
at 220 nm and 960 cm−1, respectively, which are assigned to the
tetrahedral Ti species isolated in the zeolite framework [1]. The
specific surface areas (Langmuir) determined from N2 adsorption at
77 K were 552 m2 g−1 for Ti-MWW and 530 m2 g−1 for TS-1. These
physicochemical properties indicate that these titanosilicates were
qualified as liquid-phase oxidation catalysts with aqueous H2O2 as
an oxidant.
Ti-MWW was prepared following the procedures reported
previously [5]. Ti-containing MWW lamellar precursor was first
hydrothermally crystallized in boric acid system with piperidine
(PI) as a structure-directing agent (SDA). The synthetic gel had a
molar composition of 1SiO2:0.67B2O3:0.05TiO2:1.4PI:19H2O. The
precursor was refluxed with 2 M HNO3 at a solid to liquid weight
ratio of 1:50 to remove the extraframework Ti species. For control
experiments, conventional TS-1 catalyst (Si/Ti = 30) was hydrother-
mally synthesized following the procedures widely adopted [17].
Both catalysts were activated by calcination in air at 773 K for 10 h.
The samples were characterized by powder X-ray diffraction
(XRD) on a Bruker D8 ADVANCE diffractometer (Cu-K␣ radiation).
Fourier-transformed infrared (FT-IR) were recorded on a Nico-
let Fourier transform infrared spectrometer (NEXUS 670) using
KBr technique. The inductively coupled plasma (ICP) elemental
analysis was completed on a Thermo IRIS Intrepid II XSP atomic
emission spectrometer to determine the amounts of Si, Ti, and B
species in Ti-MWW and TS-1. UV–visible spectra were recorded
on a Shimadzu 2400PC spectrophotometer with BaSO4 as a ref-
erence. BET surface area was measured by N2 adsorption on an
Autosorb Quancachrome 02108-KR-1 instrument. The amounts
of coke after ammoximation were measured by combustion in
a TGA/SDTA851 thermo-gravimetric analyzer. The samples were
heated from 298 K to 1073 K at 10 K min−1 in the flow of air. The
weight loss above 473 K was attributed to coke. The solid-state
29Si MAS NMR spectra were performed at ambient tempera-
ture on a VARIAN VNMRS-400WB spectrometer at 79.5 MHz by
using a spinning rate of 3.0 kHz. The chemical shift was referred
to an external standard of 2,2-dimethyl-2-silapentane-5-sulfonic
acid.
3.2. Effects of reaction parameters on the continuous
ammoximation of cyclohexanone using Ti-MWW as a catalyst
3.2.1. Effect of reaction temperature
The reaction temperature greatly affected the ammoximation
of cyclohexanone over Ti-MWW in a continuous slurry reactor.
When the reaction was carried out at ratios of H2O2/cyclohexanone
at 1.1 and NH3/cyclohexanone at 1.7, the conversion of cyclohex-
anone increased with rising temperature and almost leveled off at
335–348 K, while the selectivity to produce oxime was maintained
over 99% (Fig. 1). The conversion and the selectivity decreased
sharply to 20% and 80%, respectively, when the temperature was
further raised to 353 K, which was mainly attributed to the faster
vaporization and decomposition rates of the reactants. Partic-
ularly, H2O2 easily underwent nonproductive decomposition at
higher temperature in basic media. The above results indicated that
Ti-MWW possessed a relatively wide window of operating tem-
perature for producing a cyclohexanone conversion of over 96% at

S. Zhao et al. / Applied Catalysis A: General 394 (2011) 1–8
3
100
100
80
60
40
20
Oxime sel.
Ketone conv.
95
Ketone conv.
Oxime sel.
90
85
80
330
340
350
0.95
1.00
1.05
1.10
1.15
1.20
Reaction temp. /K
H2O2/cyclohexanone molar ratio
Fig. 1. Effect of reaction temperature on the ammoximation of cyclohexanone on
Fig. 3. Effect of H2O2/cyclohexanone ratio on the ammoximation of cyclohexanone
on Ti-MWW. Other conditions, see Fig. 1.
Ti-MWW. Ammoximation conditions: Ti-MWW, 3.2 g; H2O2/cyclohexanone = 1.1:1;
WHSV for cyclohexanone, 6.3 h−1
(85%).
; NH3/cyclohexanone = 1.7:1; solvent, t-BuOH
ment this loss. Moreover, this is in agreement with a previous
report that suggests ammoximation prefers a basic environ-
ment created by an excess of NH3 [18]. Nevertheless, no further
changes were observed for either the cyclohexanone conver-
sion or the oxime selectivity at NH3/cyclohexanone ratios above
1.7, because the excess NH3 was no longer being dissolved in
the reaction mixture which would be evaporated easily into the
atmosphere.
99% oxime selectivity. Nevertheless, the ammoximation proceeded
most effectively at an optimum reaction temperature of 343 K.
3.2.2. Effect of NH3/cyclohexanone and H2O2/cyclohexanone
ratios
The effect of NH3/cyclohexanone molar ratio on ammoxima-
tion was investigated at 343 K and H2O2/cyclohexanone of 1.1
(Fig. 2). The cyclohexanone ammoximation requires stoichio-
metrically equivalent moles of NH3 and cyclohexanone, but the
cyclohexanone conversion was only 83% at a NH3/cyclohexanone
ratio of 1.0. Nevertheless, the cyclohexanone conversion and
oxime selectivity reached ca. 96% and 100%, respectively when
the NH3/cyclohexanone ratio was over 1.7. The lower cyclo-
hexanone conversion obtained at an equivalent ratio of NH3
is probably because gaseous NH3 partially evaporates and dis-
solves inefficiently in the liquid-phase because the reaction
is occurring in an open reactor system at normal pressure.
Thus, the reaction needs an excess amount of NH3 to supple-
Fig. 3 shows the effect of the H2O2/cyclohexanone ratio on
the cyclohexanone ammoximation. In contrast to the fact that
the remainingNH3 is recycled in an actual industrial process, the
excessive H2O2 may explode during separation. Thus, it is prefer-
able to use the stoichiometric amount of H2O2 whenever possible.
By increasing H2O2/cyclohexanone ratio, the cyclohexanone con-
version and oxime selectivity increased simultaneously. Because of
the nonproductive decomposition of H2O2, ammoximation needed
slightly more H2O2 than cyclohexanone to proceed to its highest
possible level. The conversion of cyclohexanone reached the maxi-
mum (>96%) at a H2O2/cyclohexanone ratio of 1.1, while the oxime
selectivity was maintained at over 99%. However, cyclohexanone
conversion decreased to 86% when the H2O2/cyclohexanone ratio
was higher than 1.2, and oxime selectivity also declined slightly.
The excess H2O2 in the reaction system would accelerate the deep
oxidation of oxime to organic by-products such as nitrocyclohex-
ane [19], and also induce successive oxidation of hydroxylamine
intermediates. Moreover, H2O2 may decompose more easily at
higher concentrations. These side reactions could decrease oxime
selectivity.
100
Oxime sel.
95
Ketone conv.
90
3.2.3. Effect of catalyst amount
The effect of catalyst amount on ammoximation was investi-
gated by keeping the feeding rates of cyclohexanone constant, NH3,
H2O2 and solvent, and varying the amount of Ti-MWW catalyst
between 1.0 g and 3.2 g. The initial cyclohexanone conversion and
oxime selectivity in these cases all reached about 96% and greater
than 99% respectively, even when the catalyst amount was only
1.0 g (Fig. 4). Thus, a relatively low concentration of Ti-MWW cat-
alyst is sufficient to achieve high cyclohexanone conversion and
oxime selectivity. However, the ammoximation lifetime of the Ti-
MWW catalyst depended greatly on the loading. The amount of
cyclohexanone converted was almost proportional to the amount
of Ti-MWW used in the ammoximation process. When the amount
85
80
0.8
1.2
1.6
2.0
2.4
NH3 /cyclohexanone molar ratio
Fig. 2. Effect of NH3/cyclohexanone ratio on the ammoximation of cyclohexanone
on Ti-MWW. Other conditions, see Fig. 1.

4
S. Zhao et al. / Applied Catalysis A: General 394 (2011) 1–8
100
90
80
70
60
50
40
30
100
80
60
H2O2 conv.
H2O2 eff.
c
d
b
40
a
4
2
Remaining H2O2
0
0
50
100
150
200
0
50
100
150
200
Time on stream /h
Time on stream /h
Fig. 4. The ammoximation on Ti-MWW with catalyst loading of 3.2 g (a), 2.0 g (b),
1.5 g (c), and 1.0 g (d). Ammoximation conditions: temp, 343 K; others, see Fig. 1.
Fig. 6. H2O2 conversion, efficiency and unconverted H2O2 in the cyclohexanone
ammoximation on Ti-MWW.
of catalyst was nearly doubled from 1.5 g to 3.2 g, the ammoxima-
tion lifetime was prolonged from 91 h to 214 h.
reactor suggests that it is an important catalyst for cyclohexanone
ammoximation from the viewpoint of industrialization.
3.3. Comparison of the ammoximation lifetime between Ti-MWW
and TS-1
3.4. Deactivation behavior of Ti-MWW catalyst
The cyclohexanone conversion and oxime selectivity of the Ti-
MWW catalyst were constant during the 214 h of time on stream
(TOS), but declined sharply within 2 h (Fig. 5). TS-1 also showed
a very similar deactivation behavior, although it was consider-
ably shorter in catalytic life. The comparison was carried out at
optimized reaction temperature for each catalyst while keeping
other conditions the same. As far as the conversion of H2O2 was
concerned, Ti-MWW was fairly stable during the whole process.
When the catalyst began deactivating, the conversion of ketone
decreased, and the amount of H2O2 consumed in producing cyclo-
hexanone oxime also decreased, leading to a higher concentration
of unconverted H2O2 in the reaction mixture (Fig. 6). At the same
time, the utilization efficiency of H2O2 decreased significantly
because of nonproductive decomposition. Excessive H2O2 concen-
tration in the reaction system may promote the deep oxidation of
cyclohexanone oxime, which could account for the lowered oxime
selectivity (Fig. 3). To clarify the main reasons for the deactivation of
Ti-MWW during ammoximation, we have evaluated the reactions
at catalyst loading of 1.5 g, which exhibited a lifetime of about 93 h
(Fig. 4c). This allowed us to follow the ammoximation performance
of Ti-MWW within a relatively shortened time.
The XRD patterns and UV–visible spectra of fresh and used cat-
alysts at TOSs of 30 h, 60 h, 90 h, and 93 h (deactivated catalyst) are
shown in Figs. 7 and 8, respectively. The XRD patterns verified that
the used catalysts still possessed the typical MWW structure, but its
crystallinity decreased to a certain extent with prolonged reaction
time (Fig. 7a–e). The used catalysts and deactivated catalyst showed
very similar UV–visible spectra to the fresh catalyst, exhibiting the
characteristic band because of the tetrahedral Ti species at 220 nm
(Fig. 8) and indicating that most of the Ti species still occupied
the tetrahedral sites in the zeolite framework after ammoxima-
tion. However, a shoulder around 275 nm appeared in the spectra
of the used catalysts. This band became more intensive with reac-
tion time, in particular it was clearly observed for the deactivated
catalyst (Fig. 8d). The band is generally attributed to the octahedral
Ti species developed by the coordination of adsorbate molecules
to the tetrahedral Ti species. This UV–visible band has been con-
sidered a clear indication of the propensity of Ti species to increase
In addition to high catalytic activity and oxime selectivity, the
real applicability of titanosilicates depends also on their lifetime
and reusability. Fig. 5 shows that Ti-MWW exhibited a much
longer ammoximation lifetime than the conventional TS-1 cata-
lyst. The ammoximation runs were carried out under optimized
conditions where the conversion and oxime selectivity reached
>96% and ca. 100%, respectively, but at the same weight hourly
space velocity (WHSV). Ti-MWW remained active for 214 h in the
cyclohexanone ammoximation, whereas TS-1 deactivated at 120 h.
Clearly, Ti-MWW was superior to TS-1 in terms of catalyst lifetime.
We previously reported that Ti-MWW was a promising catalyst for
ammoximation according to the results observed in a batchwise
reactor within an extremely short reaction time of 2 h [14]. The
outstanding duration that Ti-MWW showed here in a continuous
100
Ti-MWW
90
80
70
60
TS-1
0
50
100
150
200
Time on stream /h
Fig. 5. The ketone conversion (ꢀ, ꢁ) and the oxime selectivity (᭹, ꢀ) of the
ammoximation of cyclohexanone on Ti-MWW and TS-1. Ammoximation conditions:
Ti-MWW, 343 K; TS-1, 349 K; others, see Fig. 1.

S. Zhao et al. / Applied Catalysis A: General 394 (2011) 1–8
5
1.2
f
e
1.0
d
0.8
e
c
0.6
d
c
b
0.4
b
0.2
0.0
a
a
5
10
15
20
25
30
3000
2000
1000
2Theta /degree
Wavenumber /cm-1
Fig. 7. The XRD patterns of fresh Ti-MWW (a), after used for 30 h (b), 60 h (c), 90 h
(d), and 93 h (e) (deactivated catalyst). Ammoximation conditions: Ti-MWW, 1.5 g;
others, see Fig. 1.
Fig. 9. The IR spectra of fresh Ti-MWW (a), after used for 30 h (b), 60 h (c), 90 h
(d), 93 h (e) (deactivated catalyst), and deactivated catalyst after calcination (f).
Ammoximation conditions: Ti-MWW, 1.5 g; others, see Fig. 1.
coordination number in TS-1 in the presence of ligands such as NH3,
H2O or H2O2 [20]. When the deactivated catalysts were calcined to
remove any adsorbed inorganic and organic species, a spectrum
nearly same as that of the fresh catalyst was obtained, but a weak
shoulder was still present around 275 nm (Fig. 8e).
The N2 adsorption investigation verified that both the surface
area and the pore volume of Ti-MWW decreased greatly during
ammoximation, especially within the first 30 h (Table 1, No. 2). A
further decrease was observed for the deactivated catalyst at a TOS
of 93 h (Table 1, No. 5). The surface area and pore volume of the
deactivated catalyst were recovered significantly by regeneration
treatments (Table 1, Nos. 6 and 7).
TD-DTA profiles were recorded to obtain the information con-
cerning the coke deposition in Ti-MWW (Fig. S1). The weight loss
below 473 K could be assigned to physically adsorbed water, while
that at higher temperatures is presumed to be because of dehydrox-
ylation and coke desorption. The weight loss increased with the
reaction time, indicating the formation and deposition of organic
by-products with high molecular weights on the surface or inside
the pores of Ti-MWW. In particular, coke deposition increased from
7.8% at 90 h (not yet deactivated) to 17.7% at 93 h (deactivated cat-
alyst) (Table 1, Nos. 4 and 5). This implied that the channels of
Ti-MWW became severely blocked and the Ti active sites were cov-
ered by the by-product molecules within a reaction period of only
3 h. Although the absolute amount of coke was extremely low, we
used GC–MS to detect the by-products after dissolving the deacti-
vated Ti-MWW catalyst in HF solution. Table 2 lists the molecular
formula and molecular weight of the representative organic by-
product species found.
To investigate the effects of coke formation during the deac-
tivation of Ti-MWW, we compared the reactions among different
catalyst loadings. When the amount of Ti-MWW employed was
decreased from 1.5 g to 1.0 g, the cyclohexanone conversion and
oxime selectivity still reached 96% and 99% respectively, but the
ammoximation lifetime was shortened from 93 h to 21 h (Fig. 4c
and d). The lifetime was 6 h at a catalyst loading of 0.8 g. The
TG-DTA investigation indicated that the Ti-MWW catalyst that
deactivated at a TOS of 6 h showed a weight loss (also considered
as the amount of organic species) of 10.1%, which was compara-
ble with that shown by the catalyst deactivating at a TOS of 21 h
(Fig. S1g and h). The results suggested that the coke formation and
deposition took place easily when the amount of catalyst was low,
because of higher weight ratio of by-products to catalyst. This then
induces a rapid deactivation of the catalyst.
Fig. 9 shows the FT-IR spectra of the Ti-MWW samples. The
main absorption bands were observed at 1400, 1300–1000, and
960 cm−1, which were in agreement with the IR spectra of Ti-MWW
reported previously [12]. The band at 960 cm−1 is closely related
to the stretching mode of the [SiO4] tetrahedrally bonded with
the Ti ions and can be considered the fingerprint of the Ti frame-
work. As the ammoximation proceeded with TOS, bands in the
range of 2900–3000 cm−1, assigned to the stretching of saturated
C–H bonds, were developed in the used catalysts. Furthermore, the
intensity of these bands became stronger with prolonged TOS. This
revealed that some organic substances with high boiling temper-
atures were deposited gradually inside the channels of Ti-MWW
(Fig. 9c and d), especially for the deactivated catalyst (Fig. 9e). These
bands disappeared completely when the deactivated catalyst was
calcined at 773 K in air to burn off the organic species (Fig. 9f).
220
1.0
275
0.8
d
c
0.6
b
0.4
a
0.2
e
0.0
200
300
400
500
Wavelength /nm
As the catalyst was always immersed in basic media with
the presence of excess NH3, a partial dissolution of the silicalite
framework was unavoidable. However, the desilication degree is
Fig. 8. The UV–visible spectra of fresh Ti-MWW (a), after used for 60 h (b), 90 h
(c), 93 h (d) (deactivated catalyst), and deactivated catalyst after calcination (e).
Ammoximation conditions: Ti-MWW, 1.5 g; others, see Fig. 1.

6
S. Zhao et al. / Applied Catalysis A: General 394 (2011) 1–8
Table 1
Characterization results of Ti-MWW catalysts.
No.
Samples
Surface area (m2 g−1
)
Pore volume (cm3 g−1
)
Weight lossa (%)
Si/Bb
Si/Tib
Q2 (%)
1.5
Q3 (%)
8.0
1
2
3
4
5
6
7
Fresh
TOS 30 h
TOS 60 h
TOS 90 h
552
297
291
263
157
486
530
0.27
0.17
0.17
0.16
0.13
0.26
0.26
2.0
5.1
6.7
7.8
17.7
1.0
30
76
95
114
114
26
24
22
20
1.7
TOS 93 hc
Regenerated 1d
Regenerated 2e
20
12.6
7.4
1.3
a
Given by TG-DTG analysis at 473–1073 K.
Given by ICP analysis.
Deactivated Ti-MWW sample.
Regenerated by calcination at 773 K in air.
Regenerated by acid treatment first (0.3N HNO3, solid to liquid ratio of 1:10), and then rearrangement with PI at 343 K for 24 h and further calcination at 773 K.
b
c
d
e
reasonably lower at a TOS of 6 h in the case of a 0.8 g catalyst load-
ing than at 93 h with a 1.5 g loading of Ti-MWW, suggesting that
the coke formation on the internal surface, rather than the struc-
tural degradation, was the main reason for catalyst deactivation.
The coke formation is a common phenomenon in zeolite-catalyzed
reactions, and an important factor governing catalyst lifetime.
Although the deactivation was mainly ascribed to pore blocking
by coke, it was not able to restore the original catalytic activity
when the deactivated Ti-MWW catalyst was regenerated by air
calcination at 773 K (not shown).
The ammoximation lifetime of the Ti-MWW catalyst was greatly
prolonged by increasing the catalyst amount. The lifetime of Ti-
MWW was nearly proportional to the catalyst loading in the range
1.5–3.2 g (Fig. 4a–c). However, the lifetime was greatly shortened
to 8 h at 1.0 g of catalyst loading (Fig. 4d). When the amount of cat-
alyst was low, a mass of by-product molecules could deposit on
the surface, or even enter the channels. This would close off the
active sites of Ti-MWW, leading to catalyst deactivation within a
short time. However, the pore blocking could be postponed or tem-
porarily suppressed when the catalyst was high enough for splitting
the adsorption or deposition of by-products with a high molecular
weight in the reaction system.
In addition to coke formation, decreased crystallinity also
occurred in the ammoximation as a result of framework desilication
(Fig. 7). The desilication made the apparent Si/Ti ratio of Ti-MWW
decrease from 26 to 20 after ammoximation was conducted for
93 h to deactivation (Table 1), suggesting that the content of Ti
increased relatively to that of Si. The change of the silica matrix,
that is, the scaffolds where the Ti active sites exist, is presumed to
vary the coordination states of the Ti species or to leach some Ti
molecules out of the framework. In fact, the deactivation of TS-1
in the ammoximation of cyclohexanone is likely to be attributed to
another reason, that is, dissolution of framework [21].
Besides the catalytic activity and selectivity, the most important
property that titanosilicate should have is that it is robust against
Ti leaching, a notable problem commonly encountered in liquid-
Table 2
Major by-products detected in cyclohexanone ammoximation by GC–MS.
No.
Molecular weight
Compound
Formula
OH
1
100
Cyclohexanol
O
2
127
Cyclohexanecarboxamide
NH2
O
3
4
180
179
2-Cyclohexylcyclohexanone
N-Cyclohexylidenecyclohexanamine
N
O
5
113
ε-Caprolactam
NH

S. Zhao et al. / Applied Catalysis A: General 394 (2011) 1–8
7
100
Q3
Q2
90
a
80
a
b
70
b
60
0
40
80
120
Time on stream /h
Fig. 11. Effect of silica species addition on the ammoximation over Ti-MWW.
Ammoximation reaction without (a) and with (b) 5 ppm colloidal silica addition.
Reaction conditions: Ti-MWW, 1.5 g; others, see Fig. 1.
mation of cyclohexanone, forming defect sites such as hydroxyl
nests.
c
-80
-100
-120
-140
3.5. Enhancing the ammoximation lifetime and regeneration of
deactivated Ti-MWW
Chemical shift /ppm
Fig. 10. 29Si MAS NMR spectra of fresh (a), deactivated at TOS of 93 h (b) and regen-
erated (c) Ti-MWW catalysts. The regeneration involved acid washing first and then
PI-assisted structural rearrangement.
As shown above, the framework desilication took place readily
in the ammoximation process. It is possible to restrain this desil-
ication by controlling its dynamic equilibrium. We have tried to
suppress Si leaching by feeding 5 ppm colloidal silica solution (30%)
continuously into the reaction mixture. Interestingly, the oxime
selectivity was maintained at over 99% irrespective of silica addi-
tion. As shown in Fig. 11, the addition of a small amount of silica gel
postponed the deactivation of Ti-MWW effectively, and prolonged
the ammoximation lifetime from 93 h to 105 h. In contrast, these
results suggested that the dissolution of the framework restricted
the catalytic activity of Ti-MWW in a basic environment.
On the basis of the above results, the main reasons for cata-
lyst deactivation are assumed to be the coke formation as well as
the destruction of the –Ti–O–Si– sites accompanying framework
desilication. We first tried to regenerate the deactivated catalyst
by removing any high boiling point organic substances deposited
inside the pores. After calcination in air at 773 K for 10 h, the color of
the deactivated catalyst changed from light yellow to snow-white.
TG-DTA analysis showed that a complete removal of heavy organic
species was achieved. However, the catalyst that was regener-
ated was almost inactive for the ammoximation of cyclohexanone,
indicating a simple calcination in air was insufficient for recov-
ering the initial reactivity of Ti-MWW. The deactivated Ti-MWW
catalyst after calcination was then subjected to a secondary treat-
ment with an aqueous solution of PI at molar ratios of PI/Si = 0.3
and H2O/Si = 15 at 443 K for 1 day. This treatment would cause a
reversible structural rearrangement for the MWW structure, par-
tially mend the defect vacancies and enhance the hydrophobicity of
the framework [13,27]. This would be a useful technique in regener-
ating the deactivated catalyst of Ti-MWW. The treated sample was
applied to the cyclohexanone ammoximation under the same con-
ditions after burning off the organic species of PI at 773 K in air. The
regenerated Ti-MWW restored the initial activity completely, pro-
ducing a conversion of cyclohexanone greater than 96% (Fig. 12c).
However, its lifetime was ca. 18 h, which was much shorter than
that of the fresh Ti-MWW.
phase reactions. The state of Ti in TS-1 has been intensively studied
and considered to be Ti(OSi)4, which has an important role in oxi-
dation reactions [22–25]. The Ti species are generally believed to
be stable enough in the liquid-phase oxidation of simple alkenes;
however, TS-1 slowly dissolves in an alkaline solution of ketone
ammoximation. The Ti species are then continuously extracted
from the zeolitic framework, and changes to a less active species.
Similarly, Ti-MWW is stable against Ti leaching in the liquid-phase
oxidation of alkenes [10,11], but it would also be subject to the
leaching and coordination state change of the Ti species, especially
when the dissolution of the framework occurs after a long run in
the cyclohexanone ammoximation process.
Because Ti-MWW was synthesized in the presence of boric acid,
it contained a relatively high amount of tetrahedral boron species,
corresponding to an Si/B ratio of 30 (Table 1, No. 1). The fresh Ti-
MWW catalyst developed a strong Si–O–B stretching band around
1400 cm−1 [5]. The 1400 cm−1 band almost disappeared for the
deactivated catalyst (Fig. S2). The B species leached gradually in
the ammoximation of cyclohexanone because its ionic radius is
relatively tiny compared with that of Si, and produced a Si/B ratio
of 114 for the deactivated catalyst (Table 1, No. 5). The loss of B
may decrease the framework electronegativity and vary the elec-
tron density in the vicinity of the Ti sites, enhancing hydrophobicity.
Thus, the leaching of B could not be a decisive factor for the deac-
tivation of Ti-MWW.
Fig. 10 shows the 29Si MAS NMR spectra of the Ti-MWW sam-
ples. The resonances at −94 ppm and −101 ppm are attributed to
the Si(OSi)2(OH)2 (Q2) and Si(OSi)3OH (Q3) groups [26]. In com-
parison with fresh Ti-MWW, the Q2 and Q3 groups increased
clearly in the spectra of the deactivated catalyst (Fig. 10a
and b). These results also verified that the desilication and
deboronation of the Si framework occurred during the ammoxi-

8
S. Zhao et al. / Applied Catalysis A: General 394 (2011) 1–8
combination of acid treatment and PI-assisted structural arrange-
100
90
80
70
60
ment was effective for restoring the conversion, selectivity, and
even the lifetime of regenerated Ti-MWW. Ti-MWW is thus con-
sidered to be a promising catalyst for developing a much cleaner
process of cyclohexanone ammoximation.
Acknowledgements
b
We gratefully acknowledge the NSFC of China (20890124,
20925310, 20873043), the Science and Technology Commis-
sion of Shanghai Municipality (09XD1401500, 08JC1408700),
973 Program (2006CB202508), 863 Program (2007AA03Z342,
2008AA030801), and the Shanghai Leading Academic Discipline
Project (B409).
c
a
Appendix A. Supplementary data
0
20
40
60
80
100
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.apcata.2010.10.037.
Time on stream /h
Fig. 12. Aammoximation on regenerated Ti-MWW catalysts. Regenerated by cal-
cination only and then structurally rearranged (a), regenerated by acid treatment
in advance and then structurally rearranged (b), and fresh Ti-MWW (c). Reaction
conditions: Ti-MWW, 1.5 g; others, see Fig. 1.
References
[1] G. Bellussi, M.S. Rigguto, Stud. Surf. Sci. Catal. 137 (2001) 911–920.
[2] H. Ichihashi, H. Sato, Appl. Catal. A 221 (2001) 359–366.
[3] H. Ichihashi, Catalysts Catal. 47 (2005) 190–195.
Nevertheless, when the deactivated Ti-MWW catalyst was first
treated with 0.3 M HNO3 for 2 h at 353 K, then structurally rear-
ranged with PI and calcined, the ammoximation lifetime was
recovered effectively. The regenerated catalyst showed a lifetime of
about 67 h (Fig. 12b), corresponding to 70% of that of the fresh cat-
alyst. The acid treatment in advance would remove selectively the
extra-framework Ti species as well as organic substances, which
are both formed during ammoximation. The consequent structural
rearrangement assisted with PI then removed some of the defect
sites, while avoiding the deposition of inactive Ti species on the
zeolites. Also, the regenerated catalyst showed a reduced number
of Q2 and Q3 groups in the 29Si MAS NMR spectrum when compared
with that of the deactivated catalyst (Fig. 10c and Table 1, No. 7).
[4] P. Wu, T. Tatsumi, T. Komatsu, T. Yashima, Chem. Lett. (2000) 774–776.
[5] P. Wu, T. Tatsumi, T. Komatsu, T. Yashima, J. Phys. Chem. C 105 (2001)
2897–2905.
[6] P. Wu, T. Tatsumi, Chem. Commun. (2001) 897–898.
[7] J.S. Reddy, S. Sivasanker, P. Ratnasamy, J. Mol. Catal. 69 (1991) 383–395.
[8] P. Wu, T. Komatsu, T. Yashima, J. Catal. 168 (1997) 400–406.
[9] J. Le Bars, J. Dakka, R.A. Sheldon, Appl. Catal. A 136 (1996) 69–80.
[10] P. Wu, T. Tatsumi, J. Catal. 214 (2003) 317–326.
[11] P. Wu, Y. Liu, M. He, T. Tatsumi, J. Catal. 228 (2004) 183–191.
[12] P. Wu, T. Tatsumi, J. Phys. Chem. B 106 (2002) 748–753.
[13] L. Wang, Y. Liu, W. Xie, H. Zhang, H. Wu, Y. Jiang, M. He, P. Wu, J. Catal. 246
(2007) 205–214.
[14] F. Song, Y. Liu, H. Wu, M. He, P. Wu, T. Tatsumi, J. Catal. 237 (2006) 359–367.
[15] F. Song, Y. Liu, L. Wang, H. Zhang, M. He, P. Wu, Appl. Catal. A 327 (2007) 22–31.
[16] F. Song, Y. Liu, L. Wang, H. Zhang, H. Wu, P. Wu, M. He, Chin. J. Catal. 27 (2006)
562–566.
[17] T. Taramasso, G. Perego, B. Notari, U.S. patent 4410501, 1983.
[18] M. Oikawa, M. Fukao, U.S. patent 7067699, 2006.
[19] A. Cesana, M.A. Mantegazza, M. Pastori, J. Mol. Catal. A: Chem. 117 (1997)
367–373.
4. Conclusion
[20] A. Zecchina, S. Bordiga, C. Lamberti, Catal. Today 32 (1996) 97–105.
[21] G. Petrini, A. Cesana, G. De Alberti, F. Genoni, G. Leofanti, M. Padovan, G.
Paparatto, P. Roffia, Stud. Surf. Sci. Catal. 69 (1991) 761–766.
[22] T. Blasco, M.A. Camblor, A. Corma, J. Pérez-Pariente, J. Am. Chem. Soc. 115 (1993)
11806–11816.
[23] E. Astorino, J.B. Peri, R.J. Willey, G. Busca, J. Catal. 157 (1995) 482–500.
[24] A. Thangaraj, R. Kumar, S.P. Mirajkar, P. Ratnasamy, J. Catal. 130 (1990) 1–8.
[25] T. Ito, H. Kanai, T. Nakai, S. Imamura, React. Kinet. Catal. Lett. 52 (1994) 421–427.
[26] P. Wu, D. Nuntasri, J. Ruan, Y. Liu, M. He, W. Fan, O. Terasali, T. Tatsumi, J. Phys.
Chem. B 108 (2004) 19126–19131.
Ti-MWW was capable of effectively catalyzing the ammoxima-
tion of cyclohexanone in a continuous slurry reactor, producing
a cyclohexanone conversion of over 96% and an oxime selectivity
of over 99%. The unique catalytic properties of Ti-MWW as het-
erogeneous catalyst led to a considerably longer ammoximation
lifetime than TS-1 for the production of cyclohexanone oxime. The
ammoximation behavior of Ti-MWW implied that the main rea-
sons for catalyst deactivation were the coke deposition inside the
micropores and the partial destruction of the –Ti–O–Si– sites. A
[27] L. Wang, Y. Liu, X. Wei, H. Wu, M. He, P. Wu, J. Phys. Chem. C 112 (2008)
6132–6138.