An insight into crystal morphology-dependent catalytic properties of MOR-type titanosilicate in liquid-phase selective oxidation

Article abstract of DOI:10.1016/j.jcat.2015.03.001

Mordenite zeolites, with controllable crystal morphologies grown orientedly along the c-axis, that is, the running direction of the 12-membered ring main channels, were hydrothermally synthesized. Corresponding Ti-containing mordenite (Ti-MOR) catalysts were then prepared by a secondary isomorphous substation technique, which combined together acid-assisted dealumination and sequential TiCl₄ vapor treatment. The diffusion properties of this obtained Ti-MOR catalysts with various morphologies were measured using n-butane as a probe molecule, and their catalytic properties were investigated in the liquid-phase oxidation reactions including ketone ammoximation and aromatics hydroxylation. The mordenite zeolites with smaller crystals, especially consisting of nanoparticles, were in favor of forming tetrahedrally coordinated Ti in TiCl₄ vapor treatment. The physicochemical properties and catalytic activity of Ti-MOR exhibited regular changes with an increasing crystal length along the c-axis. The aromatic hydroxylation was more sensitive than ketone ammoximation to the variety of crystal length. In particular, the adsorption capacity (Q_-), apparent diffusivity (D/L²), turnover number of ammoximation at 2.5 h and hydroxylation at 7 h decreased from 0.78 mmol g^-1, 22.7 × 10^-3 s^-1, 1783 mol (Ti-mol)^-1 and 70.5 mol (Ti-mol)^-1 to 0.59 mmol g^-1, 4.2 × 10^-3 s^-1, 500 mol (Ti-mol)^-1 and 3.4 mol (Ti-mol)^-1, respectively, when the crystal size of Ti-MOR was enlarged from 110 to 5160 nm. The adsorption properties and oxidation activity of Ti-MOR were found to exhibit a distinct change around a crystal length demarcation of ca. 1 μm. Further studies indicated that the oxidation activity of Ti-MOR was proportional to its apparent diffusivity.

Full text of DOI:10.1016/j.jcat.2015.03.001

Journal of Catalysis 325 (2015) 101–110
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
An insight into crystal morphology-dependent catalytic properties
of MOR-type titanosilicate in liquid-phase selective oxidation
a
a
b
b
a,
⇑
Yulin Yang , Jianghong Ding , Chunhui Xu , Weidong Zhu , Peng Wu
a
Shanghai Key Laboratory of Green Chemistry and Chemical Processes, Department of Chemistry, East China Normal University, North Zhongshan Rd. 3663, Shanghai 200062,
PR China
Key Laboratory of the Ministry of Education for Advanced Catalysis Materials, Institute of Physical Chemistry, Zhejiang Normal University, 321004 Jinhua, PR China
b
a r t i c l e i n f o
a b s t r a c t
Article history:
Mordenite zeolites, with controllable crystal morphologies grown orientedly along the c-axis, that is, the
running direction of the 12-membered ring main channels, were hydrothermally synthesized.
Corresponding Ti-containing mordenite (Ti-MOR) catalysts were then prepared by a secondary isomor-
phous substation technique, which combined together acid-assisted dealumination and sequential
Received 7 February 2015
Revised 27 February 2015
Accepted 2 March 2015
4
TiCl vapor treatment. The diffusion properties of this obtained Ti-MOR catalysts with various morpholo-
gies were measured using n-butane as a probe molecule, and their catalytic properties were investigated
in the liquid-phase oxidation reactions including ketone ammoximation and aromatics hydroxylation.
The mordenite zeolites with smaller crystals, especially consisting of nanoparticles, were in favor of form-
ing tetrahedrally coordinated Ti in TiCl vapor treatment. The physicochemical properties and catalytic
4
activity of Ti-MOR exhibited regular changes with an increasing crystal length along the c-axis. The aro-
Keywords:
Titanosilicate
Ti-MOR
Crystal morphology
Adsorption
Liquid-phase oxidation
Ammoximation
Hydroxylation
matic hydroxylation was more sensitive than ketone ammoximation to the variety of crystal length. In
2
particular, the adsorption capacity (Q ), apparent diffusivity (D/L ), turnover number of ammoximation
1
ꢀ
1
ꢀ3 ꢀ1
ꢀ1
ꢀ1
at 2.5 h and hydroxylation at 7 h decreased from 0.78 mmol g , 22.7 ꢁ 10
s
, 1783 mol (Ti-mol)
ꢀ1
ꢀ1
ꢀ3 ꢀ1
ꢀ1
and 70.5 mol (Ti-mol)
to 0.59 mmol g , 4.2 ꢁ 10
s
, 500 mol (Ti-mol)
and 3.4 mol (Ti-mol) ,
respectively, when the crystal size of Ti-MOR was enlarged from 110 to 5160 nm. The adsorption proper-
ties and oxidation activity of Ti-MOR were found to exhibit a distinct change around a crystal length
demarcation of ca. 1
lm. Further studies indicated that the oxidation activity of Ti-MOR was proportional
to its apparent diffusivity.
Ó 2015 Elsevier Inc. All rights reserved.
1
. Introduction
Titanosilicates, typically represented by TS-1, have attracted
Among the titanosilicates developed so far, Ti-MOR cannot yet
be prepared directly by hydrothermal synthesis, but it is post-syn-
thesized readily by isomorphous substitution of Ti ions into
enormous research interests because of their outstanding catalytic
properties in liquid-phase selective oxidation reactions using
hydrogen peroxide as a mild oxidant [1–3]. In the past decades,
in addition to TS-1, dozens of Ti-containing porous materials with
different structures and pore sizes have been developed, for exam-
ple, Ti-MWW [4,5], Ti-Beta [6], Ti-MOR [7], Ti-MCM-41 [8], and Ti-
SBA-15 [9]. Based on extensive fundamental researches, some
titanosilicates have also achieved great successes in environmen-
tally benign petrochemical processes in industry, for example,
TS-1 for phenol hydroxylation [10], cyclohexanone ammoximation
dealuminated mordenite through solid–gas reaction with TiCl
4
vapor [7]. With this secondary synthesis technique, Ti-MOR is pre-
pared without using expensive organic structure-directing agents.
Moreover, its large porosity of 12-membered ring (MR) endows Ti-
MOR with attractive catalytic performances in ketone (or acetalde-
hyde) ammoximation and aromatics hydroxylation [16–21].
Previous researches focused mainly on the oxidation reactions
themselves or the chemistry of Ti active sites. However, the rele-
vance of diffusion properties and oxidation activity to the mor-
phology and channel system of Ti-MOR has never been
investigated in detail.
[
11–13] and propylene epoxidation [14], and Ti-MWW for the
ammoximation of methyl ethyl ketone [15].
MOR framework is consisting of parallel 12-MR (6.5 ꢁ 7.0 Å)
and 8-MR (2.6 ꢁ 5.7 Å) channels along c-axis, which are intercon-
nected by 8-MR windows (3.4 ꢁ 4.8 Å) along b-axis [22,23].
Because of the small pore size of 8-MR channels and their bad
⇑
Corresponding author. Fax: +86 21 6223 2292.
E-mail address: pwu@chem.ecnu.edu.cn (P. Wu).
http://dx.doi.org/10.1016/j.jcat.2015.03.001
021-9517/Ó 2015 Elsevier Inc. All rights reserved.
0

1
02
Y. Yang et al. / Journal of Catalysis 325 (2015) 101–110
accessibility along c-axis, the MOR structure is commonly counted
as one-dimensional 12-MR channel zeolite, which may induce dif-
fusion limitations in its applications [23–25]. Up to now, the MOR
aluminosilicates with different crystal morphologies and change-
able crystal sizes have been synthesized, such as sphere-like,
disk-like, needle-like, and nanosheet crystals [23,26,27]. All
reported MOR zeolites seem to show an oriented growth along c-
axis, the running direction of 12-MR channels.
for 2 h, forming gel A with a molar composition of 1.0 SiO
2
:0.9
PI:0.05 Na O:0.025 Al :20 H O. The gel A was tumbled in a
2
2
O
3
2
Teflon-lined stainless steel autoclave at 60 rpm and at 423 K for
24 h. Solution B was prepared by dissolving 1.06 g sodium hydrox-
ide and 0.96 g sodium aluminate in 36 g distilled water. After the
crystallization of solution A for 24 h, the autoclave was cooled,
and solution B was added under vigorous stirring. Then 0.5 wt.%
2
M-3, relative to the weight of SiO in gel A, was added as the seed.
Previous studies reveal that the properties and performances of
zeolites in the applications of adsorption, separation, ion exchange,
and heterogeneous catalysis strongly depended on their crystal
morphology, including the shape and size [28,29]. The crystal mor-
phology is anticipated to be closely related to the micropore
accessibility and diffusion. As for titanosilicates, the diffusion and
mass transfer are particularly important issues in the liquid-phase
reactions using granular catalysts. TS-1 with a fine crystallite
around 300 nm exhibited the best catalytic activity, but those with
After stirring for 2 h at room temperature, the resultant gels were
crystallized in an autoclave at 443 K for 48 h under tumbling
(60 rpm).
The procedures and gel composition for synthesizing M-5, a
sample with a fiber-like crystal shape, were the same as M-1
except that the silicon source was replaced by silica gel
2 2
(95.0 wt.% SiO , 5.0 wt.% H O, Qingdao Haiyang Chemical Co.,
Ltd.), and the synthesis was carried out free of boric acid.
After completion of the crystallization, the autoclave was cooled
and the solid product was obtained by filtrating, washing with
water, and drying in air at 353 K for 12 h. All Na-mordenite zeolites
were ion-exchanged with 1.0 M ammonium chloride at a solid-to-
liquid ratio of 1:40 at 353 K for 2 h. The ion exchange repeated
twice. The sample was subsequently washed with deionized water
and calcined at 973 K for 10 h, giving rise to proton-type zeolites,
a crystal size exceeded 1 lm or longer showed a significantly
retarded activity because of diffusion problem [30,31]. It is thus
preferable to synthesize nanoscaled titanosilicates in the pursuit
of high performance.
Different to TS-1 equipped with a three-dimensional channel
structure, the sole useful pore system of Ti-MOR is the 12-MR
(
6.5 ꢁ 7.0 Å) channel in c-axis direction. In the preparation of Ti-
MOR, the TiCl (6.7 Å) vapor molecules need to diffuse into the
2-MR channels firstly, and then they are incorporated into the
3
H-MOR. The H-MOR samples were refluxed in 6 M HNO solution
4
at a solid-to-liquid ratio of 1:50 for 10 h. The acid-treated samples
were subsequently filtrated and washed with deionized water
until the pH value of the filtrate was over 5. These procedures were
repeated once more, giving rise to dealuminated zeolites, Del-MOR.
1
framework. As for liquid-phase oxidation, the reactant molecules
must approach to the Ti active sites located inside channels.
Therefore, the crystal morphology of Ti-MOR may alter a double
superimposition effect on the catalytic activity from the TiCl
4
vapor
2.2. Preparation of Ti-MOR titanosilicates
treatment and the catalytic reaction. Nevertheless, how the crystal
morphology influences the catalytic performance of Ti-MOR has
not yet been investigated systematically.
Ti-containing mordenite was post-synthesized through a solid–
gas reaction between dealuminated MOR and TiCl vapor at ele-
4
In this study, a series of mordenite zeolites with different crys-
tal morphologies, ranging from nanometer scale to micrometer,
have been hydrothermally synthesized, and corresponding Ti-
MOR catalysts were then post-synthesized. The crystal morphol-
ogy-dependent physicochemical properties and catalytic perfor-
mances have been investigated. The catalytic activity of Ti-MOR
in liquid-phase oxidations was correlated to its crystal morphol-
ogy, length as well as apparent diffusivity.
vated temperature following previous procedures [7,18]. For a
typical preparation, a Del-M sample (2.0 g) placed in a quartz tube
reactor (£ 3 cm) was pretreated at 673 K for 2 h in a dry N
2
stream
vapor was brought into the reactor by
flow, treating the Del-M sample at 673 K for 2 h. The sample
was then purged with pure N at the same temperature for 1 h
to remove any residual TiCl from the zeolite powder. After cooling
to room temperature in N atmosphere, the treated samples were
ꢀ1
(50 mL min ). Then, TiCl
4
N
2
2
4
2
washed with deionized water and dried in air at 353 K overnight,
resulting in Ti-M-x (x = 1–5).
2
. Experimental
2.3. Characterization methods
2.1. Synthesis of MOR zeolites and dealuminated MOR with different
morphologies
Powder X-ray diffraction (XRD) patterns were collected on a
Rigaku Ultima IV diffractometer using CuK
a
radiation
A couple of commercially available Na-mordenite zeolites with
prism-like morphologies and different crystal lengths, denoted as
M-2 (230 nm) and M-3 (630 nm), were used as received.
(k = 1.5405 Å) at 35 kV and 25 mA in the 2h angle range of 5–35°.
Scanning electron microscopy (SEM) was performed on a Hitachi
S-4800 microscope to determine the morphology. The 29Si MAS
NMR spectra were measured on a VARIAN VNMRS 400WB NMR
Na-mordenite M-1 sample with granular-like crystal shape was
synthesized from colloidal silica (30.0 wt.% SiO
9.7 wt.% H O), sodium hydroxide, sodium aluminate (44.39 wt.%
Al , 37.50 wt.% Na O), and boric acid. These chemicals were
2
, 0.3 wt.% Na
2
O,
spectrometer using single-pulse method at
79.43 MHz, a spinning rate of 3 kHz, and a recycling delay of
60 s. The chemical shift was referred to Q ([(CH SiO ]SiO12).
a frequency of
6
2
2
O
3
2
M
8 8
3
)
3
8
mixed in distilled water under vigorous stirring, forming a
homogeneous gel. Then, 0.5 wt.% M-3 zeolite, relative to the weight
The Si, Ti, and Al contents were determined by inductively coupled
plasma emission spectrometry (ICP) on a thermo IRIS Intrepid II
XSP atomic emission spectrometer after dissolving the samples in
HF solution. The specific surface area, pore volume, and pore size
of SiO
2
in gel, was added as the seed. The gel had a molar com-
3
:0.24 Na O:0.067 Al :22 H O:0.2 H BO .
position of 1.0 SiO
2
2
2
O
3
2
3
After being stirred at room temperature for 2 h, the resultant gels
were transferred into a Teflon-lined stainless steel autoclave,
which was tumbled at 60 rpm and at 443 K for 24 h.
M-4 sample with needle-like shape was synthesized by two
steps. In a typical run, 7.65 g piperidine (PI), 0.15 g sodium hydrox-
ide, 0.48 g sodium aluminate, and 20 g colloidal silica were firstly
mixed in order into 22 g distilled water under vigorous stirring
2
distribution were measured by N adsorption at 77 K on a
BELSORP–MAX instrument after the samples were activated in
advance at 573 K under vacuum for at least 4 h. The UV–visible dif-
fuse reflectance spectra were recorded on a Shimadzu UV-2550
spectrophotometer using BaSO
4
as a reference. The IR spectra were
collected on Nicolet Nexus 670 FT-IR spectrometer in absorbance
ꢀ1
mode at a spectral resolution of 2 cm
using KBr technique

Y. Yang et al. / Journal of Catalysis 325 (2015) 101–110
103
(
£ 1.5 cm wafer, 3 wt.% diluted in KBr). Thermogravimetry (TG)
and M-5. The M-4 and M-5 samples also showed the phenomena
of peak broadening in XRD patterns (Fig. 1d and e). This is probably
resulted from the decreased crystal sizes along a-axis and b-axis
directions, or the increased one along c-axis direction [21], as the
reflections of MOR crystalline structure are mostly independent
of the c-axis.
SEM micrographs shown in Fig. 2 exhibit that all MOR samples
differed greatly in morphologies with each other. As summarized
in Table 1, the crystal length was in the order of M-1
and derivative thermogravity (DTG) analyses were carried out on
a Mettler TGA/SDTA 851 instrument with a ramping rate of
1
on various Ti-M samples was performed on a NETZSCH STA 449C
thermogravimetric analyzer. The samples were pretreated at
6
of adsorption amount on time was then recorded under a n-butane
partial pressure of 10 kPa at 198 K.
e
ꢀ1
ꢀ1
0 K min in an air flow of 40 mL min . The sorption of n-butane
ꢀ1
23 K for 5 h in a helium stream (36 mL min ). The dependence
(
110 nm) < M-2 (230 nm) < M-3 (630 nm) < M-4 (2450 nm) < M-5
2
.4. Catalytic reactions
(5160 nm). The aspect ratio, that is, length/width ratio, also varied
significantly between different Na-MOR samples. In addition, the
Si/Al ratio decreased with the crystal length enlarging.
The ammoximation of cyclohexanone was carried out in a
batchwise reactor using water as a solvent. The mixture of catalyst
25 mg), 10 mmol cyclohexanone, and 6 g H O was heated in a 50-
mL flask to 333 K, where H (30 wt.%) and NH O were added
ꢂH
to initiate the reaction for a desirable period of time (0.5–2.5 h).
The molar ratio of NH :H :cyclohexanone was 1.5:1.2:1. The
Combining the information given by XRD patterns (Fig. 1) and
SEM micrographs (Fig. 2), we can determine that the crystal length
is along with the c-axis direction of MOR topology. Videlicet, the
main 12-MR channels of MOR increase in length with the direction
of crystal growth, as displayed in Scheme 1.
(
2
2
O
2
3
2
3
2 2
O
reusability of Ti-M samples in cyclohexanone ammoximation was
carried out with an enlarged catalyst loading of 0.1 g, and used
catalyst was washed with water and acetone, and dried at 353 K.
3.2. Preparation and characterization of Ti-MOR with different
morphologies
The ratio of catalyst:H
2
O:ketone:H
2
O
2
:NH
3
was kept at
0
.1 g:6 g:10 mmol:12 mmol:15 mmol and reacted at 333 K for 2 h
The structure of these mordenite samples after dealumination
and Ti incorporation was first investigated with XRD
(Supplementary Material, Fig. S1). All Ti-M samples were still char-
acteristic of a typical MOR topology and possessed the diffractions
very comparable to corresponding pristine Na-MOR in intensity.
This indicated that there was no destruction of the crystalline
structure.
in a 50-mL flask. All catalysts were used five cycles.
In a typical run of toluene hydroxylation, the mixture of 50 mg
catalyst, 10 g toluene, and 1 g H
2
O was stirred vigorously at 363 K.
Then, 10 mmol H (30 wt.%) was added into the reactor to start
2 2
O
the hydroxylation. After the reaction was carried out for a desirable
period of time (1–7 h), 20 mL ethanol was added to homogenize
the organic phase and the aqueous phase.
2
To evaluate the porosity of Ti-M samples, the N adsorption
After removing the catalyst solid, the products of two reactions
were analyzed on a gas chromatograph (Shimadzu GC-2014) and
quantified using chlorobenzene as an internal standard.
measurements were performed at 77 K. They exhibited type I
adsorption–desorption isotherms and similar adsorption behavior
at low relative pressure irrespective of the crystal morphology
(
Fig. 3). The specific surfaces calculated by Langmuir method were
2
ꢀ1
in a similar range of 555–595 m g (Table 2), which confirmed
that these samples were of highly crystalline. The differences
observed in XRD intensity simply resulted from the oriented crys-
tal growth along specific crystalline planes, that is, c-axis direction.
Nevertheless, with respect to the adsorption behavior in the region
of higher relative pressure, there were slight differences among the
Ti-M samples. Compared to prism-like samples (Ti-M-2 and Ti-M-
3
. Results and discussion
3.1. Synthesis and characterization of Na-MOR zeolite with different
morphologies
Fig. 1 shows the XRD patterns of five Na-MOR samples with dif-
ferent morphologies. They were all of pure crystalline phases with
the MOR topology, with other impurity unobserved. The intensity
of the [002] reflection, related to the c-axis, showed no obvious
difference for all samples. However, compared to the XRD patterns
of M-1, M-2 and M-3, the intensities of [110], [020], [200], [310],
3
), the other three samples all showed slightly enhanced N
2
adsorption at P/P = 0.9–0.99 (Fig. 3). This is probably due to the
0
nitrogen adsorption inside the intercrystal pores that existed in
Ti-M-1 with a much smaller crystals (110 nm), as well as Ti-M-4
and Ti-M-5 with much longer crystals [2450–5160 nm].
Following previously well-established procedures [7], we con-
ducted the incorporation of Ti ions into the defect sites such as
hydroxyl nests in the MOR framework by atom-planting method
[
330], [150], and [350] reflections decreased distinctly for M-4
4
using TiCl vapor. The existing states of the Ti species introduced
were first investigated by IR spectroscopy. Titanosilicates are char-
e
ꢀ1
acteristic of an IR band at approximately 960 cm irrespective of
crystalline structure, which is taken as an important evidence for
the presence of tetrahedral Ti species [2,3]. As shown in Fig. 4A,
d
ꢀ1
all Ti-MOR samples showed the typical band at 963 cm , which
is attributed to the Si–O–Ti stretching vibration. The band intensity
decreased with the crystal length increasing, especially when the
crystal reached micrometer scale range, indicating the amount of
incorporated Ti was reduced.
c
b
a
The level of dealumination was investigated by chemical analy-
sis. As depicted in Table 2, when the crystal was within nanometer
size, the Si/Al ratio was all over 200. However, with the crystal size
up to micrometer, the Si/Al ratio was decreased to 112 and 156 for
Ti-M-4 and Ti-M-5, respectively, implying that the dealumination
process was affected by the crystal morphology. As the crystal
length became larger, namely the length of 12-MR channels was
5
10
15
2
20
25
30
35
Theta (deg.)
Fig. 1. XRD patterns of Na-MOR samples. (a) M-1, (b) M-2, (c) M-3, (d) M-4, and (e)
M-5.

1
04
Y. Yang et al. / Journal of Catalysis 325 (2015) 101–110
(
a)
4
3
2
1
0
0
0
0
0
50
70 90 110 130 150 170 190
Crystal length (nm)
1
1
1
1
2
0 μm
0 μm
0 μm
0 μm
0 μm
1 μm
2 μm
2 μm
2 μm
2 μm
(
b)
3
3
2
2
1
1
5
0
5
0
5
0
5
0
8
5
135 185 235 285 335 385 435
Crystal length (nm)
(
c)
25
20
15
10
5
0
2
50 350 450 550 650 750 850 950
Crystal length (nm)
(
d)
30
25
20
15
10
5
0
1
.3 1.7 2.1 2.5 2.9 3.3 3.7 4.1
Crystal length μm
(
)
(
e)
50
40
30
20
10
0
4
.0 4.3 4.6 4.9 5.2 5.5 5.8 6.1
Crystal length μm
(
)
Fig. 2. SEM micrographs of Na-MOR samples. (a) M-1 (granular-like), (b) M-2 (prism-like), (c) M-3 (prism-like), (d) M-4 (needle-like), and (e) M-5 (fiber-like).
enlarged, the acid molecules needed to take a longer diffusion dis-
tance to enter into the 12-MR channels. This was same for the alu-
minum species diffusing out of the channels. Exceptionally, the
crystals of Ti-M-5 were larger than that of Ti-M-4, but the former
had a higher Si/Al ratio, that is, a relatively lower Al content. This
could be a result of the different crystal aggregation degrees
between these two materials. Compared with the M-5 crystals
loosely piled up together (Fig. 2e), the M-4 particles were tightly
bound together forming sarciniform aggregates (Fig. 2d), which
induced a slightly more diffusing limitations than the former.
The saturated amount of incorporated Ti, represented by Si/Ti ratio,
varied obviously with the crystal length. The Si/Ti ratio was at a

Y. Yang et al. / Journal of Catalysis 325 (2015) 101–110
105
Table 1
comparable level of 73–77, for Ti-M-1, Ti-M-2, and Ti-M-3 (Table 2,
Nos. 1–3). However, the Si/Ti ratio increased to 138 and 116 for Ti-
M-4 and Ti-M-5 both with the micrometer-sized crystals in length.
Characterizations of various Na-MOR zeolites.
No. Sample Crystal lengtha
Crystal widtha
(a ꢃ b, nm)
Aspect ratio
(c/b)
Si/
Al
b
(c, nm)
4
The TiCl vapor molecules with a relatively bulky dimension (6.7 Å)
1
2
3
4
5
M-1
M-2
M-3
M-4
M-5
110
230
630
2450
5160
70
150
350
100
100
1.6
1.5
1.8
24.5
51.6
6.1
6.5
6.6
5.9
4.7
may suffer diffusion difficulty when entering into the 12-MR chan-
nels of mordenite. This kind of limitation is considered to be
enhanced distinctly inside the channels with longer diffusion dis-
tance. Similar to the Si/Al ratio for Ti-M-4 and Ti-M-5, the Si/Ti
ratio of Ti-M-4 was a little higher than Ti-M-5, of which the phe-
nomenon was also caused by the aggregation of crystals.
UV–visible spectroscopy was employed to characterize further
the coordination states of the Ti species incorporated. All Ti-M
samples revealed a dominated band at approximately 210 nm
a
The crystal length and width in this table are an average of 50 appropriate
crystals from SEM images.
b
Molar ratio given by ICP analysis.
(
Fig. 4B), which is assigned to the ligand-to-metal charge transfer
2
ꢀ
4+
from O to Ti . This band is usually observed for a variety of Ti-
substituted zeolites, and it is characteristic of tetrahedrally coordi-
nated Ti highly dispersed in the framework [32]. Obviously, the
intensity of 210 nm band decreased along with the crystal particle
growing, in agreement with IR spectra. On the contrary, the inten-
sity of 260 nm, assigned to hexa-coordinated Ti outside the frame-
work, increased accompanied by the crystal particle expanding.
Moreover, no obvious band was observed around 330 nm due to
anatase phase for Ti-M-1, Ti-M-2, and Ti-M-3. The 330 nm band
was observed visibly when the crystal length reached or exceeded
1
lm for Ti-M-4 and Ti-M-5. This is because that when suffering a
stronger diffusion resistance in longer 12-MR channels, the TiCl
4
vapor molecules tended to form easily the hexa-coordinated Ti
species and anatase phase in the extra-framework positions.
Based on the results of IR spectroscopy, UV–visible spec-
troscopy, and ICP analysis, it is concluded that the amount of tetra-
hedrally coordinated Ti dispersed in framework decreased with the
increase of crystal length.
Scheme 1. Crystal morphology, channel network, and pore sizes of MOR zeolite.
With respect to the Ti incorporation mechanism, we once
addressed that the Ti species were inserted into the hydroxyl nests
through the reaction of one TiCl
4
molecule with four internal sila-
1
000
Ti-M-5
+
+
800
600
1
8
nol groups on the basis of isotope exchange with C
O
2
[7]. Here,
Si MAS NMR spectra were measured to take insight into this
issue. As shown in Fig. 5, two main resonances were observed at
2
9
8
6
4
2
00
00
00
00
0
Ti-M-4
Ti-M-3
4
ꢀ
113 ppm and ꢀ115 ppm, which are attributed to the Q groups
in Si(SiO) configuration [33]. With comparable intensities, they
were caused by overlap of four crystallographically non-equivalent
tetrahedral Si(SiO) sites. The acid treatments removed the Al ions
+400
200
4
4
+
Ti-M-2
Ti-M-1
from the framework extensively and developed a broad resonance
3
around ꢀ103 ppm, assigned to the Q groups of silicon-bearing OH
3
groups, Si(OH)(SiO) [34]. According to the titanation mechanism
3
investigated previously [7], the quantity of Q groups remaining
after TiCl treatment would reflect the level of titanation.
4
0
.0
0.2
0.4
0.6
0.8
1.0
Compared to dealuminated samples (Del-M) (Fig. 5A), the reso-
P/P0
3
nance intensity of Q groups at ꢀ103 ppm was decreased for
corresponding Ti-M samples (Fig. 5B), especially for the Ti-M-1,
Ti-M-2, and Ti-M-3 materials with nanosized crystals. A cal-
2
Fig. 3. N adsorption–desorption isotherms of various Ti-M samples at 77 K.
3
culation of deconvoluted spectra showed that the Q percentages
were decreased by TiCl
.3%, and 15.8% of Del-M-1, Del-M-2, Del-M-3, Del-M-4, and Del-
M-5 to 1.7%, 0.5%, 1.5%, 6.3%, and 12.7% of Ti-M-1, Ti-M-2, Ti-M-
, Ti-M-4, and Ti-M-5, respectively (Table 2). It is worth noting
4
vapor treatment from 8.9%, 3.1%, 4.1%,
9
Table 2
Physico-chemical properties of different Ti-M samples.
3
Si/Ala
Si/Tia
SA (m2
ꢀ1)b
3
3
4
c
3
No.
Sample
g
Q /(Q + Q ) (%)
1.7 (8.9)d
that the Ti-M-1 had the biggest drop of Q groups (7%), while
the decline values of other four samples were about 3%.
Moreover, even for the micrometer-scaled materials, Ti-M-4
1
2
3
4
5
Ti-M-1
Ti-M-2
Ti-M-3
Ti-M-4
Ti-M-5
429
416
234
112
156
73
75
77
138
116
595
591
582
587
555
d
0.5 (3.1)
1.5 (4.1)d
3
and Ti-M-5, contained a much higher content of Q than that
d
6.3 (9.3)
12.7 (15.8)d
4
of other samples, the TiCl vapor treatment could only remove
a small part of them. The result again confirmed that for the
micrometer-sized crystals, the Ti species were difficult to be
incorporated in the framework according to the titanation
a
b
c
Molar ratio given by ICP analysis.
2
Specific surface area (Langmuir) given by N adsorption at 77 K.
29
Determined by Si MAS NMR investigation.
d
The figure between brackets represents the Q3 content of corresponding
mechanism through the reaction between the TiCl
cules and the Si–OH groups.
vapor mole-
4
dealuminated MOR sample (Del-M) respectively.

1
06
Y. Yang et al. / Journal of Catalysis 325 (2015) 101–110
3.3. Molecular diffusion behaviors of various Ti-MOR zeolites
samples. The maximum uptake of Ti-M-1 was achieved after
min, roughly the same as Ti-M-2 and Ti-M-3. However, at least
5
The diffusion properties of zeolites are frequently investigated
10 min was required to reach the maximum uptake for Ti-M-4
and Ti-M-5. And, the adsorption capacity of Ti-M-1, Ti-M-2, and
by employing hydrocarbons as probe molecules [35]. Existing
researches on the adsorption properties of mordenite are mainly
focused on the lower availability of micropore volume resulted
ꢀ1
Ti-M-3 was 0.78, 0.77, and 0.76 mmol g , respectively (Table 3),
difference of which was little. Nevertheless, the values of Ti-M-4
+
ꢀ1
from the lodged Na cations [36,37] or octahedral non-framework
and Ti-M-5 decreased to 0.71 and 0.59 mmol g , respectively
alumina [24]. The influence of morphology on the diffusion perfor-
mance of MOR-type zeolite has seldom been studied up to date.
Fig. 6A shows the uptake curves of n-butane on different Ti-M
(Table 3). Thus, the adsorption capacity of Ti-M was negatively
related to its crystal length in the order of Ti-M-1 > Ti-M-2 > Ti-
M-3 > Ti-M-4 > Ti-M-5.
2
1
1
0
0
.0
.5
.0
.5
.0
Ti-M-1
Ti-M-2
Ti-M-3
Ti-M-4
Ti-M-5
210
Ti-M-1
Ti-M-2
Ti-M-3
Ti-M-4
Ti-M-5
(
A)
8
6
4
2
0
(B)
260
963
330
1
050 1000 950 900 850 800 750 700
200
300
400
500
-1
Wavelength (nm)
Wavenumber (cm )
Fig. 4. IR spectra in the framework vibration region (A) and UV–visible spectra (B) of various Ti-M samples with different morphologies.
Del-M-1
Del-M-2
Del-M-3
Del-M-4
Del-M-5
Ti-M-1
Ti-M-2
Ti-M-3
Ti-M-4
Ti-M-5
(
A)
(B)
3
3
Q
Q
-
80 -90 -100 -110 -120 -130 -140
Chemical shift (ppm)
-80 -90 -100 -110 -120 -130 -140
Chemical shift (ppm)
Fig. 5. 29Si MAS NMR spectra of various Del-M (A) and Ti-M (B) samples.
0
0
0
0
0
0
0
0
0
.8
1.0
(
A)
(B)
.7
.6
.5
.4
.3
.2
.1
.0
0
0
.8
.6
Ti-M-1
Ti-M-2
Ti-M-3
Ti-M-4
Ti-M-5
0.4
.2
0.0
Ti-M-1
Ti-M-2
Ti-M-3
Ti-M-4
Ti-M-5
0
0
5
10
15
20
0
5
10 15 20 25 30 35
1/2
1/2
Time (min)
t
(s )
Fig. 6. n-Butane uptake curves (A) and relative adsorption uptake of n-butane (B) in various Ti-M samples. Sorption at 298 K and a n-butane partial pressure of 10 kPa.

Y. Yang et al. / Journal of Catalysis 325 (2015) 101–110
107
Table 3
Adsorption capacity (Q
different Ti-M samples.
Table 4
2
A comparison of cyclohexanone ammoximation on different Ti-M samples.a
1
) and apparent diffusivity (D/L ) of n-butane adsorbed in
No. Sample Cyclohexanone
conv. (%)
Initial reaction rate
Oxime sel.
(%)
ꢀ
1
ꢀ1
Sample
Ti-M-1
0.78
22.7
Ti-M-2
Ti-M-3
Ti-M-4
Ti-M-5
(mmol g
h
)
(mmol g 1)
ꢀ
0.77
20.1
0.76
17.7
0.71
5.8
0.59
4.2
1
2
3
4
5
Ti-M-1 94.4
Ti-M-2 87.3
Ti-M-3 80.0
Ti-M-4 27.0
Ti-M-5 16.6
344.4
316.8
296.0
70.4
>98
>98
>98
>98
>98
Q
1
2
ꢀ3 ꢀ1
D/L (10
s )
40.0
Fig. 6B shows the dependence of the relative adsorption uptake
on square root of time, where Q and Q denote the mass adsorbed
at time t and infinite time, respectively. The Q value was taken
a
For ammoximation conditions, see Fig. 8; reaction time, 2.5 h.
t
1
1
from the saturated amount in each uptake curve. The diffusion
rates of n-butane on various Ti-M samples also gave inverse cor-
relations with their particle size. During the initial stage of uptake,
the following appropriate solution applies [35]:
Fig. 7A shows the time course of specific activity per Ti site, that
is, turnover number (TON). The TON of Ti-M-1, Ti-M-2, and Ti-M-3
increased rapidly with reaction time at initial stage, while the dis-
parity among them was slight, and the initial reaction rate of each
sffiffiffiffiffi
ꢀ1 ꢀ1
sample was 344.4, 316.8, and 296.0 mmol g
h , respectively
pffiffi
Qt
2
pffiffiffi
p
D
(
Table 4, Nos. 1–3). As for Ti-M-4 and Ti-M-5, the TON increased
¼
t
2
Q
L
1
slowly along with reaction time. Their initial reaction rate was
ꢀ1
ꢀ1
lowered to 70.4 and 40.0 mmol g
h , respectively (Table 4,
where Q
t
/Q
1
is the normalized loading, D is the diffusivity, L is the
Nos. 4 and 5). The difference in catalytic activity of TON was dis-
tinct between these two micrometer-sized samples and the other
three nanosized ones. After 2.5 h, the reaction approximated the
equilibrium, giving a cyclohexanone conversion of 94.4%, 87.3%,
0.0%, 27.0%, and 16.6% over Ti-M-1, Ti-M-2, Ti-M-3, Ti-M-4, and
Ti-M-5, respectively, while with a high oxime selectivity over
8% for all samples (Table 4). Obviously, the catalysts of nanosized
crystals (Ti-M-1, Ti-M-2, and Ti-M-3) were benefit for the ammox-
imation, showing a narrowed disparity in TON and reaction rate at
initial stage. Accompanied by reaction time stretching, the gap of
TON became larger, as well as the ketone conversion. More explic-
itly, when crystal particles reached the micrometer scale (Ti-M-4
and Ti-M-5), the TON, initial reaction rate, and also the equilibrium
conversion decreased sharply compared to nanoparticles.
Therefore, the catalytic activity of Ti-MOR decreases along with
the particle length increasing.
2
characteristic diffusion length, D/L is the apparent diffusivity, and t
is time. This fitting adequately describes the experimental data
covering the uptake curve from Q
The D/L ratio of various Ti-MOR samples obtained according to
t
/Q
1
ꢄ 0.1 up to Q
t
/Q
1
ꢄ 0.8.
2
8
the equation is compared in Table 3. It is obvious that the apparent
2
diffusivity (D/L ) did not decrease significantly when the particle
9
size of adsorbent (Ti-M) expanded within nanometer scale.
However, when the particle size of Ti-M enlarged up to micrometer
2
scale, the apparent diffusivity (D/L ) would fall sharply to
ꢀ3
ꢀ1
ꢀ3 ꢀ1
4
1
of 1
.2 ꢁ 10
10 nm), with a decreasing degree of about fivefold. Thus, the value
m for crystal length is a distinguished demarcation line deter-
mining the adsorption and diffusion properties of Ti-MOR zeolite.
s
(Ti-M-5, 5160 nm) from 22.7 ꢁ 10
s
(Ti-M-1,
l
3.4. Catalytic properties of various Ti-M zeolites in liquid-phase
The ammoximation is believed to occur in two steps, that is,
oxidations
in situ formation of hydroxylamine via catalytic oxidation of NH
by H on Ti sites and oximation of ketone with hydroxylamine
3
2 2
O
3
.4.1. Ammoximation of cyclohexanone with ammonia and hydrogen
peroxide
The influence of morphology on Ti-M catalytic performance was
to corresponding oxime [13,16]. Since the second step of oximation
is a non-catalytic reaction, it can occur either inside the 12-MR
channels of Ti-MOR (Scheme 2A, reaction route I), or in the liquid
phase after the hydroxylamine intermediate diffuses out of the
zeolite channels (Scheme 2A, reaction route II). Moreover, the
intermediate of hydroxylamine is relatively unstable and may
first demonstrated in the ammoximation of cyclohexanone with
ammonia and hydrogen peroxide. The cyclohexanone oxime pro-
duct is the raw material for producing e-caprolactam, an important
intermediate for nylon-6 synthesis. In addition to TS-1 [38] and Ti-
MWW [39], Ti-MOR also proves to be an efficient catalyst for this
reaction [16].
x
undergo decomposition to NO (Scheme 2A, reaction route III).
The reaction route I is more favorable than route II for preventing
8
7
60
5
4
3
2
1
0
0
2
1
1
000
600
200
Ti-M-1
Ti-M-2
Ti-M-3
Ti-M-4
Ti-M-5
(A)
Ti-M-1
Ti-M-2
Ti-M-3
Ti-M-4
Ti-M-5
(B)
0
0
0
0
0
8
4
00
00
0
3
0
0.0
0.5
1.0
1.5
2.0
2.5
0
1
2
3
4
5
6
7
Reaction time (h)
Reaction time (h)
Fig. 7. The catalytic activity of cyclohexanone ammoximation (A) and toluene hydroxylation (B) over various Ti-M samples. Ammoximation conditions: cat., 25 mg;
cyclohexanone, 10 mmol; ketone:H :NH = 1:1.2:1.5 (molar ratio); water, 6 g; temp., 333 K. Hydroxylation conditions: cat., 50 mg; toluene, 10 g; water, 1 g; H
30 wt.%), 10 mmol; temp., 363 K.
2
O
2
3
2 2
O
(

1
08
Y. Yang et al. / Journal of Catalysis 325 (2015) 101–110
Scheme 2. Cyclohexanone ammoximation (A) and toluene hydroxylation (B) over the 12-MR channel of Ti-MOR.
the non-productive decomposition of hydroxylamine, because the
cyclohexanone molecules diffusing into the pores would react with
hydroxylamine formed therein immediately. Consequently, if the
crystal particles become larger, the diffusion distance would be far-
ther for reactants to enter into the 12-MR channels as well as for
the intermediates and the products to diffuse out of the channels,
and the possible non-productive decomposition for hydroxylamine
molecules would increase.
1
1
1
400
200
000
2
1
1
5
0
0
5
0
a
8
6
00
00
b
3.4.2. Hydroxylation of toluene
400
200
0
The effect of morphology on the catalytic properties of Ti-M was
further verified in the hydroxylation of toluene, a reaction needing
open spaces. The Ti-M catalyst has already been revealed to be
more active than TS-1 zeolite in the hydroxylation of aromatics
because of a large porosity of mordenite zeolite [17]. The oxida-
tions are supposed to take place only when the aromatic molecules
reach the active sites of Ti peroxo species (Ti–OOH) formed inside
zeolite pores.
0
5
10
15
20
25
-
3
-1
Apparent diffusivity (10 s )
2
Fig. 8. The correlation between apparent diffusivity (D/L ) and TON of cyclohex-
anone ammoximation (a), and that of toluene hydroxylation (b). The TON was
measured at 1 h of reaction time.
Fig. 7B exhibits the time course of TON for various Ti-M samples
formed. The main products of toluene hydroxylation were the mix-
ture of o-cresol, m-cresol, and p-cresol isomers. The TON of Ti-M-1
ꢀ
1
(
110 nm) was 20 mol (Ti-mol) at initial 1 h, but the value gradu-
within nanometer scale, whereas the activity gap is sharply
enlarged when crystals reach micrometer range.
ꢀ1
ally downgraded to 1 mol (Ti-mol) for Ti-M-5 (5160 nm) along
with the crystal length enlarging. Prolonging the reaction time to
h, the TON reached 70.5, 54.7, 44.7, 4.9, and 3.4 mol (Ti-mol)
ꢀ
1
7
3.5. Correlation between oxidation activity and morphology of Ti-MOR
for Ti-M-1, Ti-M-2, Ti-M-3, Ti-M-4, and Ti-M-5, respectively.
Thus, the gap of oxidation activity between different Ti-M samples
became more remarkable in the wake of reaction time stretching.
The regularity was similar to the abovementioned ammoximation,
but the hydroxylation TON disparity among Ti-M-1, Ti-M-2, and Ti-
M-3 was larger than that of ammoximation. The 12-MR channels
available for catalytic reactions are of one dimensional along the
crystal growing direction (c-axis). A smaller particle size would
shorten the diffusion path for toluene molecules approaching to
the Ti sites inside the pores, improving the catalytic performance,
as the hydroxylation reaction only occurs over the Ti sites located
in the 12-MR channels (Scheme 2B). In addition, the crystal length
The dependence of relative activity on the crystal length has
been investigated (Supplementary Material, Fig. S2). The specific
activity (TON) of Ti-M-1 measured at 1 h is assumed to be 100%.
The relative activity of cyclohexanone ammoximation and toluene
hydroxylation both decreased rapidly with the crystal length
becoming longer, especially when the crystal length reached
micrometer scale. Another phenomenon observed was that the
falling range of relative activity for toluene hydroxylation was lar-
ger than that for cyclohexanone ammoximation.
Furthermore, the TON of ammoximation was found to be
almost proportional to the apparent diffusivity calculated from
the n-butane diffusion (Fig. 8a). The same dependence was also
observed for the TON of hydroxylation (Fig. 8b). Similar to the
of ca. 1
lm seems to be a demarcation line determining the oxida-
tion activity of Ti-M zeolites: the disparity of activity is small

Y. Yang et al. / Journal of Catalysis 325 (2015) 101–110
109
4
.0 wt.% heavy species inside channels (Supplementary Material,
Fig. S3). Thus, the nanosized crystals are helpful to minimize diffu-
sion limitation and suppress coke formation efficiently, which then
leads to a longer catalytic life.
4
. Conclusions
A series of Ti-MOR with different morphologies have been pre-
pared by post-synthesis method, from corresponding aluminosili-
cates with different degrees of oriented crystal growth along the
c-axis, the running direction of 12-MR main channels. Smaller par-
ticles, especially nanoparticles, are in favor of forming fine tetra-
hedrally coordinated Ti for shortening the diffusion distance of
4
TiCl vapor molecule inside zeolite micropores. As well, the
physicochemical properties and catalytic activity of Ti-MOR zeo-
lites exhibit regular changes with crystal length along the c-axis
increasing. The hydroxylation of toluene is more sensitive than
the ammoximation of cyclohexanone to the variety of crystal
length. The crystal length of ca. 1 lm is a remarkable demarcation
line governing the adsorption properties and oxidation activity of
Fig. 9. Reusability of Ti-M samples with different morphologies toward the
ammoximation
of
cyclohexanone
in
water.
Reaction
conditions:
temp.,
catalyst:H O:ketone:H
2
2
O
2
:NH = 0.1 g:6 g:10 mmol:12 mmol:15 mmol;
3
3
33 K; 2 h. Regeneration: used catalyst was washed with water and acetone, and
Ti-MOR. Besides, the catalytic activity of Ti-MOR, which receives
double superimposition effects by TiCl vapor treatment and reac-
4
dried at 353 K.
tion itself, behaves a linear dependence on the apparent diffusivity,
for both ammoximation and hydroxylation.
result shown in Fig. S2, the TON of toluene hydroxylation
decreased more rapidly than ammoximation with the apparent
diffusivity decreasing. That is to say hydroxylation is more sensi-
tive than ammoximation to the changes of crystal length. This is
because that the oximation could occur either in the 12-MR chan-
nels or in liquid phase outside the pores (Scheme 2A); however, the
hydroxylation has no option but to occur over the Ti sites in the 12-
MR channels (Scheme 2B).
Acknowledgments
We gratefully acknowledge the National Natural Science
Foundation of China (21373089), the PhD Programs Foundation
of the Ministry of Education (2012007613000), and the Shanghai
Leading Academic Discipline Project (B409).
Appendix A. Supplementary material
3.6. Stability and reusability of Ti-MOR with different morphologies
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jcat.2015.03.001.
In addition to catalytic activity, the applicability of titanosili-
cates depends also on their lifetime and reusability. The reusability
of five Ti-M samples has been checked in the ammoximation of
cyclohexanone independently. The experiments were initiated
with an enlarged catalyst loading of 0.1 g. The used catalyst was
washed with water and acetone, and dried at 353 K. Then, it was
subjected to a repeated reaction at a constant ratio of catalyst–sub-
strate–solvent. After five recycles of reaction, nanometer-scaled Ti-
M catalysts, namely Ti-M-1, Ti-M-2, and Ti-M-3, maintained the
cyclohexanone conversion about 95.0% (Fig. 9). On the contrary,
the cyclohexanone conversion decreased to 62.0% from 96.9% and
decreased to 32.0% from 70.7% for Ti-M-4 and Ti-M-5, respectively
Fig. 9). Thus, the nanosized Ti-M catalysts were not only more
active, but were also more stable than micrometer-scaled ones in
batchwise reactions.
The deactivation of titanosilicates in ammoximation has been
previously investigated to be caused mainly by coke formation
References
[1] M. Tramasso, G. Perego, B. Notari, US Patent 4410501, 1983.
[2] G. Bellussi, M.S. Rigguto, Stud. Surf. Sci. Catal. 137 (2001) 911–920.
[
[
[
3] P. Ratnasamy, D. Srinivas, H. Knözinger, Adv. Catal. 48 (2004) 1–169.
4] P. Wu, T. Tatsumi, T. Komatsu, T. Yashima, J. Catal. 202 (2001) 245–255.
5] P. Wu, T. Tatsumi, Chem. Commun. (2002) 1026–1027.
[6] M.A. Camblor, A. Corma, A. Martínez, J. Pérez-Pariente, J. Chem. Soc. Chem.
Commun. (1992) 589–590.
[
[
7] P. Wu, T. Komastu, T. Yashima, J. Phys. Chem. 100 (1996) 10316–10322.
8] A. Corma, M.T. Navarro, J. Pérez-Pariente, J. Chem. Soc. Chem. Commun. (1994)
147–148.
[
9] P. Wu, T. Tatsumi, T. Komatsu, T. Yashima, Chem. Mater. 14 (2002) 1657–1664.
(
[
10] G. Bellussi, C. Perego, in: G. Ertl, H. Knozinger, J. Weiakamp (Eds.), Handbook
of Heterogeneous Catalysis, vol. 5, Wiley, Poiters, 1997, p. 2329.
[11] Eur. Chem. News 6 (1995) 23.
[
[
12] P.R. Saronno, M.P. Milan, et al., US Patent 4745221, 1988.
13] P. Roffia, G. Leofanti, A. Cesana, M. Mantegazza, M. Padovan, G. Petrini, S. Tonti,
P. Gervasutti, Stud. Surf. Sci. Catal. 55 (1990) 43–52.
[
40]. Thus, we have characterized the used catalysts. The heavy
[14] C. Neri, B. Anfossi, et al., US Patent 4833260, 1989.
[
15] J.X. Yang, M.K. Yao, S. Zhao, Y.M. Liu, P. Wu, M.Y. He, Chin. J. Catal. 31 (2010)
5–99.
16] P. Wu, T. Komatsu, T. Yashima, J. Catal. 168 (1997) 400–411.
by-products accumulated gradually with process time and
deposited inside the channels of titanosilicates, which may block
the Ti active sites wherein. Actually, the TG analysis indicated that
the used Ti-M-1 (110 nm) catalysts contained about 5.9 wt.%
organic species, which showed a weight loss in a wide temperature
range (440–1000 K), including 2.4 wt.% organic species inside
channels (Supplementary Material, Fig. S3). We have demonstrated
that they corresponded to the heavy by-products formed in
ammoximation system, including cyclohexanecarboxamide,
9
[
[17] P. Wu, T. Komatsu, T. Yashima, J. Phys. Chem. B 102 (1998) 9297–9303.
[
18] H. Xu, Y.T. Zhang, H.H. Wu, Y.M. Liu, X.H. Li, J.G. Jiang, M.Y. He, P. Wu, J. Catal.
81 (2011) 263–272.
2
[
19] Y.L. Yang, J.H. Ding, B.S. Wang, J. Wu, C. Zhao, G.H. Gao, P. Wu, J. Catal. 320
(2014) 160–169.
[20] J.H. Ding, L. Xu, Y.J. Yu, H.H. Wu, S.J. Huang, Y.L. Yang, J. Wu, P. Wu, Catal. Sci.
Technol. 3 (2013) 2587–2595.
[
21] J.H. Ding, L. Xu, H. Xu, H.H. Wu, Y.M. Liu, P. Wu, Chin. J. Catal. 34 (2013) 243–
50.
[22] W.M. Meier, Z. Kristallogr. 115 (1961) 439–450.
2
e
-caprolactam, N-cyclohexylidenecyclohexanamine, and 2-cyclo-
[
23] L. Zhang, A.N.C. van Laak, P.E. de Jongh, K.P. de Jong, Microporous Mesoporous
Mater. 126 (2009) 115–124.
hexylcyclohexanone [18,40]. On the other hand, the weight loss
due to organic species reached 7.2 wt.% on the washed Ti-M-5
(
[
24] S. van Donk, J.H. Bitter, A. Verberckmoes, M. Versluijs-Helder, A. Broersma, K.P.
de Jong, Angew. Chem. Int. Ed. 44 (2005) 1360–1363.
5160 nm) catalyst after being used for 5 times, including

1
10
Y. Yang et al. / Journal of Catalysis 325 (2015) 101–110
[
[
25] H. Liu, G.D. Lei, W.M.H. Sachtler, Appl. Catal. A 137 (1996) 167–177.
26] L. Zhang, S.J. Xie, W.J. Xin, X.J. Li, S.L. Liu, L.Y. Xu, Mater. Res. Bull. 46 (2011)
[34] J.T. Miller, P.D. Hopkins, B.L. Meyers, G.J. Ray, R.T. Roginski, G.W. Zajac, N.H.
Rosenbaum, J. Catal. 138 (1992) 115–128.
8
94–900.
27] C. Jo, J. Jung, H.S. Shin, J. Kim, R. Ryoo, Angew. Chem. Int. Ed. 52 (2013) 10014–
0017.
28] L. Bonetto, M.A. Camblor, A. Corma, J. Pérez-Pariente, Appl. Catal. A 82 (1992)
7–50.
[35] J.C. Groen, W.D. Zhu, S. Brouwer, S.J. Huynink, F. Kapteijn, J.A. Moulijn, J. Pérez-
Ramírez, J. Am. Chem. Soc. 129 (2007) 355–360.
[36] C.E. Webster, A. Cottone, R.S. Drago, J. Am. Chem. Soc. 121 (1999) 12127–
12139.
[37] D. Lozano-Castello, W. Zhu, A. Linares-Solano, F. Kapteijn, J.A. Moulijn,
Microporous Mesoporous Mater. 92 (2006) 145–153.
[38] A. Thangaraj, S. Sivasanker, P. Ratnasamy, J. Catal. 131 (1991) 394–400.
[39] F. Song, Y.M. Liu, H.H. Wu, M.Y. He, P. Wu, T. Tatsumi, J. Catal. 237 (2006) 359–
367.
[
[
1
3
[
[
29] L.Y. Hu, Z.K. Zhang, S.J. Xie, S.L. Liu, L.Y. Xu, Catal. Commun. 10 (2009) 900–904.
30] A.J.H.P. van der Pol, A.J. Verduyn, J.H.C. van Hooff, Appl. Catal. A 92 (1992) 113–
1
30.
[
[
[
31] T. Tatsumi, N. Jappar, J. Catal. 161 (1996) 570–576.
32] G. Bellussi, M.S. Rigguto, Stud. Surf. Sci. Catal. 85 (1994) 177–213.
33] K. Segawa, H. Tachibana, J. Catal. 131 (1991) 482–490.
[40] S. Zhao, W. Xie, J.X. Yang, Y.M. Liu, Y.T. Zhang, B.L. Xu, J.G. Jiang, M.Y. He, P. Wu,
Appl. Catal. A 394 (2011) 1–8.